duckdb-simple 0.3.0.0 → 1.0.0.0
raw patch · 80 files changed
+33539/−11399 lines, 80 filesdep +filepathdep +hs-bindgen-runtimedep ~basedep ~bytestringdep ~duckdb-ffiPVP ok
version bump matches the API change (PVP)
Dependencies added: filepath, hs-bindgen-runtime
Dependency ranges changed: base, bytestring, duckdb-ffi, text, time
API changes (from Hackage documentation)
- Database.DuckDB.Simple: [connectionGeometryCRS] :: ConnectionOptions -> [Text]
- Database.DuckDB.Simple: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.ConnectionOptions
- Database.DuckDB.Simple: instance GHC.Internal.Show.Show Database.DuckDB.Simple.ConnectionOptions
- Database.DuckDB.Simple.Copy: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Copy.CopyBindInfo
- Database.DuckDB.Simple.Copy: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Copy.CopyBindInfo
- Database.DuckDB.Simple.Deprecated.Streaming: fold :: (FromRow row, ToRow params) => Connection -> Query -> params -> a -> (a -> row -> IO a) -> IO a
- Database.DuckDB.Simple.Deprecated.Streaming: foldArrow :: ToRow params => Connection -> Query -> params -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a
- Database.DuckDB.Simple.Deprecated.Streaming: foldArrow_ :: Connection -> Query -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a
- Database.DuckDB.Simple.Deprecated.Streaming: foldNamed :: FromRow row => Connection -> Query -> [NamedParam] -> a -> (a -> row -> IO a) -> IO a
- Database.DuckDB.Simple.Deprecated.Streaming: fold_ :: FromRow row => Connection -> Query -> a -> (a -> row -> IO a) -> IO a
- Database.DuckDB.Simple.Deprecated.Streaming: nextRow :: FromRow r => Statement -> IO (Maybe r)
- Database.DuckDB.Simple.Deprecated.Streaming: nextRowWith :: RowParser r -> Statement -> IO (Maybe r)
- Database.DuckDB.Simple.FromField: [timeWithZoneZone] :: TimeWithZone -> TimeZone
- Database.DuckDB.Simple.FromField: instance (GHC.Internal.Data.Typeable.Internal.Typeable a, Database.DuckDB.Simple.FromField.FromField a) => Database.DuckDB.Simple.FromField.FromField (GHC.Internal.Arr.Array GHC.Internal.Types.Int a)
- Database.DuckDB.Simple.FromField: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.FromField.Field
- Database.DuckDB.Simple.FromField: instance GHC.Internal.Show.Show Database.DuckDB.Simple.FromField.Field
- Database.DuckDB.Simple.Function: class FunctionArg a
- Database.DuckDB.Simple.Function: data ScalarType
- Database.DuckDB.Simple.Function: instance (Database.DuckDB.Simple.FromField.FromField a, Database.DuckDB.Simple.Function.FunctionArg a, Database.DuckDB.Simple.Function.Function r) => Database.DuckDB.Simple.Function.Function (a -> r)
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg Data.Text.Internal.Text
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Base.String
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Int.Int16
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Int.Int32
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Int.Int64
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Types.Bool
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Types.Double
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Types.Float
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Types.Int
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Types.Word
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Word.Word16
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Word.Word32
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg GHC.Internal.Word.Word64
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionArg a => Database.DuckDB.Simple.Function.FunctionArg (GHC.Internal.Maybe.Maybe a)
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult Data.Text.Internal.Text
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Base.String
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Int.Int16
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Int.Int32
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Int.Int64
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Types.Bool
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Types.Double
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Types.Float
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Types.Int
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Types.Word
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Word.Word16
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Word.Word32
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult GHC.Internal.Word.Word64
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult a => Database.DuckDB.Simple.Function.Function (GHC.Internal.Types.IO a)
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult a => Database.DuckDB.Simple.Function.Function a
- Database.DuckDB.Simple.Function: instance Database.DuckDB.Simple.Function.FunctionResult a => Database.DuckDB.Simple.Function.FunctionResult (GHC.Internal.Maybe.Maybe a)
- Database.DuckDB.Simple.Generic: instance Database.DuckDB.Simple.Generic.DuckValue a => Database.DuckDB.Simple.Generic.DuckValue (GHC.Internal.Arr.Array GHC.Internal.Types.Int a)
- Database.DuckDB.Simple.Internal: [connectionDatabase] :: ConnectionState -> DuckDBDatabase
- Database.DuckDB.Simple.Internal: [connectionTypeCache] :: ConnectionState -> TypeCache
- Database.DuckDB.Simple.Internal: [statementStreamMode] :: StatementStream -> ResultMode
- Database.DuckDB.Simple.Internal: destroyClientContext :: DuckDBClientContext -> IO ()
- Database.DuckDB.Simple.Internal: fetchPrepareError :: Text -> DuckDBPreparedStatement -> IO Text
- Database.DuckDB.Simple.Internal: fetchResultError :: Ptr DuckDBResult -> IO (Text, Maybe DuckDBErrorType)
- Database.DuckDB.Simple.Internal: instance GHC.Internal.Base.Semigroup Database.DuckDB.Simple.Internal.Query
- Database.DuckDB.Simple.Internal: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Internal.Query
- Database.DuckDB.Simple.Internal: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Internal.SQLError
- Database.DuckDB.Simple.Internal: instance GHC.Internal.Classes.Ord Database.DuckDB.Simple.Internal.Query
- Database.DuckDB.Simple.Internal: instance GHC.Internal.Data.String.IsString Database.DuckDB.Simple.Internal.Query
- Database.DuckDB.Simple.Internal: instance GHC.Internal.Exception.Type.Exception Database.DuckDB.Simple.Internal.SQLError
- Database.DuckDB.Simple.Internal: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Internal.Query
- Database.DuckDB.Simple.Internal: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Internal.SQLError
- Database.DuckDB.Simple.Internal: mkExecuteError :: Query -> Text -> Maybe DuckDBErrorType -> SQLError
- Database.DuckDB.Simple.Internal: newtype Connection
- Database.DuckDB.Simple.Internal: throwRegistrationError :: String -> IO a
- Database.DuckDB.Simple.Internal: throwResultError :: Query -> Ptr DuckDBResult -> IO ()
- Database.DuckDB.Simple.Internal: toSQLError :: Exception e => e -> SQLError
- Database.DuckDB.Simple.Internal: withClientContext :: Connection -> (DuckDBClientContext -> IO a) -> IO a
- Database.DuckDB.Simple.Internal: withTypeCache :: Connection -> (TypeCache -> IO a) -> IO a
- Database.DuckDB.Simple.Logging: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Logging.LogEntry
- Database.DuckDB.Simple.Logging: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Logging.LogEntry
- Database.DuckDB.Simple.LogicalRep: destroyLogicalType :: DuckDBLogicalType -> IO ()
- Database.DuckDB.Simple.LogicalRep: logicalTypeFromRep :: LogicalTypeRep -> IO DuckDBLogicalType
- Database.DuckDB.Simple.LogicalRep: logicalTypeFromRepWith :: (LogicalTypeRep -> IO DuckDBLogicalType) -> LogicalTypeRep -> IO DuckDBLogicalType
- Database.DuckDB.Simple.LogicalRep: logicalTypeToRep :: DuckDBLogicalType -> IO LogicalTypeRep
- Database.DuckDB.Simple.ToField: instance (Database.DuckDB.Simple.ToField.DuckDBColumnType a, Database.DuckDB.Simple.ToField.ToDuckValue a) => Database.DuckDB.Simple.ToField.ToDuckValue (GHC.Internal.Arr.Array GHC.Internal.Types.Int a)
- Database.DuckDB.Simple.ToField: instance (Database.DuckDB.Simple.ToField.DuckDBColumnType a, Database.DuckDB.Simple.ToField.ToField a) => Database.DuckDB.Simple.ToField.ToField (GHC.Internal.Arr.Array GHC.Internal.Types.Int a)
- Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.DuckDBColumnType a => Database.DuckDB.Simple.ToField.DuckDBColumnType (GHC.Internal.Arr.Array GHC.Internal.Types.Int a)
+ Database.DuckDB.Simple: Aggregate :: state -> (state -> args -> IO state) -> (state -> state -> IO state) -> (state -> IO result) -> Aggregate state args result
+ Database.DuckDB.Simple: ErrorApi :: ErrorType
+ Database.DuckDB.Simple: ErrorConfigurationInvalid :: ErrorType
+ Database.DuckDB.Simple: ErrorConfigurationPermission :: ErrorType
+ Database.DuckDB.Simple: ErrorConfigurationSettings :: ErrorType
+ Database.DuckDB.Simple: ErrorDatabaseCatalog :: ErrorType
+ Database.DuckDB.Simple: ErrorDatabaseConstraint :: ErrorType
+ Database.DuckDB.Simple: ErrorDatabaseIndex :: ErrorType
+ Database.DuckDB.Simple: ErrorDatabaseSequence :: ErrorType
+ Database.DuckDB.Simple: ErrorDatabaseSerialization :: ErrorType
+ Database.DuckDB.Simple: ErrorDatabaseStatistics :: ErrorType
+ Database.DuckDB.Simple: ErrorDatabaseTransaction :: ErrorType
+ Database.DuckDB.Simple: ErrorIOEof :: ErrorType
+ Database.DuckDB.Simple: ErrorIOFileNotFound :: ErrorType
+ Database.DuckDB.Simple: ErrorIOGeneral :: ErrorType
+ Database.DuckDB.Simple: ErrorIOHTTP :: ErrorType
+ Database.DuckDB.Simple: ErrorIONetwork :: ErrorType
+ Database.DuckDB.Simple: ErrorIOReadFailure :: ErrorType
+ Database.DuckDB.Simple: ErrorInputInvalid :: ErrorType
+ Database.DuckDB.Simple: ErrorInputMetadataDeferred :: ErrorType
+ Database.DuckDB.Simple: ErrorInputObjectSize :: ErrorType
+ Database.DuckDB.Simple: ErrorInputOutOfRange :: ErrorType
+ Database.DuckDB.Simple: ErrorInputParameterInvalid :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryBinder :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryExecutor :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryExpression :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryNotImplemented :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryOptimizer :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryParameterNotAllowed :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryParameterNotResolved :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryParser :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryPlanner :: ErrorType
+ Database.DuckDB.Simple: ErrorQueryScheduler :: ErrorType
+ Database.DuckDB.Simple: ErrorQuerySyntax :: ErrorType
+ Database.DuckDB.Simple: ErrorResourceAutoload :: ErrorType
+ Database.DuckDB.Simple: ErrorResourceConnection :: ErrorType
+ Database.DuckDB.Simple: ErrorResourceDependency :: ErrorType
+ Database.DuckDB.Simple: ErrorResourceInUse :: ErrorType
+ Database.DuckDB.Simple: ErrorResourceMissingExtension :: ErrorType
+ Database.DuckDB.Simple: ErrorResourceOutOfMemory :: ErrorType
+ Database.DuckDB.Simple: ErrorRuntimeFatal :: ErrorType
+ Database.DuckDB.Simple: ErrorRuntimeInternal :: ErrorType
+ Database.DuckDB.Simple: ErrorRuntimeInterrupt :: ErrorType
+ Database.DuckDB.Simple: ErrorRuntimeNullPointer :: ErrorType
+ Database.DuckDB.Simple: ErrorTypeConversion :: ErrorType
+ Database.DuckDB.Simple: ErrorTypeDecimal :: ErrorType
+ Database.DuckDB.Simple: ErrorTypeDivideByZero :: ErrorType
+ Database.DuckDB.Simple: ErrorTypeInvalid :: ErrorType
+ Database.DuckDB.Simple: ErrorTypeMismatch :: ErrorType
+ Database.DuckDB.Simple: ErrorTypeUnknown :: ErrorType
+ Database.DuckDB.Simple: LogicalTypeIdAny :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdArray :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdBigInt :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdBigNum :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdBit :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdBlob :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdBoolean :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdDate :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdDecimal :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdDouble :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdEnum :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdFloat :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdGeometry :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdInteger :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdInterval :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdInvalid :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdList :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdMap :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdSQLNull :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdStruct :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTime :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTimeNs :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTimeTz :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTimestamp :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTimestampMs :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTimestampNs :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTimestampSec :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTimestampTz :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTimestampTzNs :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdTuple :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdType :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdUBigInt :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdUHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdUInteger :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdUSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdUTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdUUID :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdUnion :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdUnknown :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdVarchar :: LogicalTypeId
+ Database.DuckDB.Simple: LogicalTypeIdVariant :: LogicalTypeId
+ Database.DuckDB.Simple: [aggregateCombine] :: Aggregate state args result -> state -> state -> IO state
+ Database.DuckDB.Simple: [aggregateInitial] :: Aggregate state args result -> state
+ Database.DuckDB.Simple: [aggregateResult] :: Aggregate state args result -> state -> IO result
+ Database.DuckDB.Simple: [aggregateStep] :: Aggregate state args result -> state -> args -> IO state
+ Database.DuckDB.Simple: [connectionAggregateStripes] :: ConnectionOptions -> Int
+ Database.DuckDB.Simple: [connectionArrowBatchSize] :: ConnectionOptions -> Int
+ Database.DuckDB.Simple: [connectionCallbackLockWait] :: ConnectionOptions -> Maybe Int
+ Database.DuckDB.Simple: [sqlErrorBody] :: SQLError -> Maybe Text
+ Database.DuckDB.Simple: attachDatabase :: Instance -> FilePath -> Text -> [(Text, Text)] -> IO ()
+ Database.DuckDB.Simple: class TypedRow r
+ Database.DuckDB.Simple: closeEnvironment :: Environment -> IO ()
+ Database.DuckDB.Simple: closeInstance :: Instance -> IO ()
+ Database.DuckDB.Simple: connect :: Instance -> IO Connection
+ Database.DuckDB.Simple: connectWithOptions :: Instance -> ConnectionOptions -> IO Connection
+ Database.DuckDB.Simple: createAggregate :: (FromRow args, TypedRow args, FunctionResult result) => Connection -> Text -> Aggregate state args result -> IO ()
+ Database.DuckDB.Simple: createEnvironment :: IO Environment
+ Database.DuckDB.Simple: createInstance :: Environment -> [(Text, Text)] -> IO Instance
+ Database.DuckDB.Simple: createTableFunction :: (FromRow params, TypedRow params, TypedRow row) => Connection -> Text -> [Text] -> (params -> IO [row]) -> IO ()
+ Database.DuckDB.Simple: data Aggregate state args result
+ Database.DuckDB.Simple: data Conversion a
+ Database.DuckDB.Simple: data Environment
+ Database.DuckDB.Simple: data ErrorType
+ Database.DuckDB.Simple: data Instance
+ Database.DuckDB.Simple: data LogicalTypeId
+ Database.DuckDB.Simple: data SizedArray (n :: Nat) a
+ Database.DuckDB.Simple: defaultAggregateStripes :: Int
+ Database.DuckDB.Simple: defaultArrowBatchSize :: Int
+ Database.DuckDB.Simple: defaultCallbackLockWait :: Int
+ Database.DuckDB.Simple: detachDatabase :: Instance -> Text -> IO ()
+ Database.DuckDB.Simple: duckdbTypeDeclaration :: DuckDBColumnType a => Proxy a -> Either Text TypeDeclaration
+ Database.DuckDB.Simple: liftConversion :: IO a -> Conversion a
+ Database.DuckDB.Simple: runConversion :: Conversion a -> IO (Ok a)
+ Database.DuckDB.Simple: setDefaultDatabase :: Instance -> Text -> IO ()
+ Database.DuckDB.Simple: sizedArray :: forall (n :: Nat) a. KnownNat n => Array Int a -> Maybe (SizedArray n a)
+ Database.DuckDB.Simple: sizedArrayFromList :: forall (n :: Nat) a. KnownNat n => [a] -> Maybe (SizedArray n a)
+ Database.DuckDB.Simple: sizedArrayToArray :: forall (n :: Nat) a. SizedArray n a -> Array Int a
+ Database.DuckDB.Simple: withConnectionOn :: Instance -> (Connection -> IO a) -> IO a
+ Database.DuckDB.Simple: withEnvironment :: (Environment -> IO a) -> IO a
+ Database.DuckDB.Simple: withInstance :: Environment -> [(Text, Text)] -> (Instance -> IO a) -> IO a
+ Database.DuckDB.Simple.Arrow: releaseArrowArray :: Ptr ArrowArray -> IO ()
+ Database.DuckDB.Simple.Arrow: releaseArrowSchema :: Ptr ArrowSchema -> IO ()
+ Database.DuckDB.Simple.Arrow: releaseArrowStream :: Ptr ArrowArrayStream -> IO ()
+ Database.DuckDB.Simple.Catalog: CatalogIndex :: CatalogEntryType
+ Database.DuckDB.Simple.Catalog: CatalogSequence :: CatalogEntryType
+ Database.DuckDB.Simple.Catalog: CatalogTable :: CatalogEntryType
+ Database.DuckDB.Simple.Catalog: CatalogType :: CatalogEntryType
+ Database.DuckDB.Simple.Catalog: CatalogView :: CatalogEntryType
+ Database.DuckDB.Simple.Catalog: ColumnDescription :: Text -> LogicalTypeRep -> Bool -> Bool -> ColumnDescription
+ Database.DuckDB.Simple.Catalog: TableDescription :: [Text] -> Bool -> [ColumnDescription] -> TableDescription
+ Database.DuckDB.Simple.Catalog: [columnDescriptionGenerated] :: ColumnDescription -> Bool
+ Database.DuckDB.Simple.Catalog: [columnDescriptionHasDefault] :: ColumnDescription -> Bool
+ Database.DuckDB.Simple.Catalog: [columnDescriptionName] :: ColumnDescription -> Text
+ Database.DuckDB.Simple.Catalog: [columnDescriptionType] :: ColumnDescription -> LogicalTypeRep
+ Database.DuckDB.Simple.Catalog: [tableDescriptionColumns] :: TableDescription -> [ColumnDescription]
+ Database.DuckDB.Simple.Catalog: [tableDescriptionName] :: TableDescription -> [Text]
+ Database.DuckDB.Simple.Catalog: [tableDescriptionReadonly] :: TableDescription -> Bool
+ Database.DuckDB.Simple.Catalog: data CatalogEntryType
+ Database.DuckDB.Simple.Catalog: data ColumnDescription
+ Database.DuckDB.Simple.Catalog: data TableDescription
+ Database.DuckDB.Simple.Catalog: describeTable :: Connection -> Text -> IO TableDescription
+ Database.DuckDB.Simple.Catalog: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Catalog.CatalogEntryType
+ Database.DuckDB.Simple.Catalog: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Catalog.ColumnDescription
+ Database.DuckDB.Simple.Catalog: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Catalog.TableDescription
+ Database.DuckDB.Simple.Catalog: instance GHC.Internal.Classes.Ord Database.DuckDB.Simple.Catalog.CatalogEntryType
+ Database.DuckDB.Simple.Catalog: instance GHC.Internal.Read.Read Database.DuckDB.Simple.Catalog.CatalogEntryType
+ Database.DuckDB.Simple.Catalog: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Catalog.CatalogEntryType
+ Database.DuckDB.Simple.Catalog: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Catalog.ColumnDescription
+ Database.DuckDB.Simple.Catalog: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Catalog.TableDescription
+ Database.DuckDB.Simple.Config: ConfigOption :: Text -> Text -> Maybe Text -> Maybe Text -> SettingScope -> [SettingScope] -> [Text] -> ConfigOption
+ Database.DuckDB.Simple.Config: CustomOption :: Text -> TypeDeclaration -> Maybe DefaultValue -> Text -> SettingScope -> [SettingScope] -> Maybe (FieldValue -> IO ()) -> CustomOption
+ Database.DuckDB.Simple.Config: SettingScopeDefault :: SettingScope
+ Database.DuckDB.Simple.Config: SettingScopeGlobal :: SettingScope
+ Database.DuckDB.Simple.Config: SettingScopeSession :: SettingScope
+ Database.DuckDB.Simple.Config: [configOptionAliases] :: ConfigOption -> [Text]
+ Database.DuckDB.Simple.Config: [configOptionDefault] :: ConfigOption -> Maybe Text
+ Database.DuckDB.Simple.Config: [configOptionDescription] :: ConfigOption -> Text
+ Database.DuckDB.Simple.Config: [configOptionName] :: ConfigOption -> Text
+ Database.DuckDB.Simple.Config: [configOptionScope] :: ConfigOption -> SettingScope
+ Database.DuckDB.Simple.Config: [configOptionScopes] :: ConfigOption -> [SettingScope]
+ Database.DuckDB.Simple.Config: [configOptionSetting] :: ConfigOption -> Maybe Text
+ Database.DuckDB.Simple.Config: [customOptionDefaultScope] :: CustomOption -> SettingScope
+ Database.DuckDB.Simple.Config: [customOptionDefault] :: CustomOption -> Maybe DefaultValue
+ Database.DuckDB.Simple.Config: [customOptionDescription] :: CustomOption -> Text
+ Database.DuckDB.Simple.Config: [customOptionName] :: CustomOption -> Text
+ Database.DuckDB.Simple.Config: [customOptionScopes] :: CustomOption -> [SettingScope]
+ Database.DuckDB.Simple.Config: [customOptionType] :: CustomOption -> TypeDeclaration
+ Database.DuckDB.Simple.Config: [customOptionUpdate] :: CustomOption -> Maybe (FieldValue -> IO ())
+ Database.DuckDB.Simple.Config: customOption :: Text -> TypeDeclaration -> CustomOption
+ Database.DuckDB.Simple.Config: data ConfigOption
+ Database.DuckDB.Simple.Config: data CustomOption
+ Database.DuckDB.Simple.Config: data SettingScope
+ Database.DuckDB.Simple.Config: getConfigDefaultValue :: Connection -> Text -> IO FieldValue
+ Database.DuckDB.Simple.Config: getConfigValue :: Connection -> Text -> IO (FieldValue, SettingScope)
+ Database.DuckDB.Simple.Config: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Config.ConfigOption
+ Database.DuckDB.Simple.Config: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Config.ConfigOption
+ Database.DuckDB.Simple.Config: listConfigOptions :: Connection -> IO [ConfigOption]
+ Database.DuckDB.Simple.Config: listInstanceOptions :: Instance -> IO [ConfigOption]
+ Database.DuckDB.Simple.Config: registerCustomOption :: Connection -> CustomOption -> IO ()
+ Database.DuckDB.Simple.Config: setConfigOption :: Connection -> SettingScope -> Text -> Text -> IO ()
+ Database.DuckDB.Simple.Config: setConfigValue :: ToDuckValue a => Connection -> SettingScope -> Text -> a -> IO ()
+ Database.DuckDB.Simple.Conversion: conversionError :: Exception e => e -> Conversion a
+ Database.DuckDB.Simple.Conversion: conversionErrorWith :: Exception e => Conversion e -> Conversion a
+ Database.DuckDB.Simple.Conversion: data Conversion a
+ Database.DuckDB.Simple.Conversion: instance GHC.Internal.Base.Alternative Database.DuckDB.Simple.Conversion.Conversion
+ Database.DuckDB.Simple.Conversion: instance GHC.Internal.Base.Applicative Database.DuckDB.Simple.Conversion.Conversion
+ Database.DuckDB.Simple.Conversion: instance GHC.Internal.Base.Functor Database.DuckDB.Simple.Conversion.Conversion
+ Database.DuckDB.Simple.Conversion: instance GHC.Internal.Base.Monad Database.DuckDB.Simple.Conversion.Conversion
+ Database.DuckDB.Simple.Conversion: instance GHC.Internal.Base.MonadPlus Database.DuckDB.Simple.Conversion.Conversion
+ Database.DuckDB.Simple.Conversion: instance GHC.Internal.Control.Monad.Fail.MonadFail Database.DuckDB.Simple.Conversion.Conversion
+ Database.DuckDB.Simple.Conversion: liftConversion :: IO a -> Conversion a
+ Database.DuckDB.Simple.Conversion: mapConversionErrors :: ([SomeException] -> [SomeException]) -> Conversion a -> Conversion a
+ Database.DuckDB.Simple.Conversion: mapConversionErrorsWith :: ([SomeException] -> Conversion [SomeException]) -> Conversion a -> Conversion a
+ Database.DuckDB.Simple.Conversion: runConversion :: Conversion a -> IO (Ok a)
+ Database.DuckDB.Simple.Copy: CopyFromInfo :: FilePath -> [Text] -> [LogicalTypeId] -> CopyFromInfo
+ Database.DuckDB.Simple.Copy: CopyTo :: (CopyBindInfo -> IO bindState) -> (CopyInitInfo bindState -> IO globalState) -> (CopySinkInfo bindState globalState -> [[Field]] -> IO ()) -> (CopyFinalizeInfo bindState globalState -> IO ()) -> CopyTo bindState globalState
+ Database.DuckDB.Simple.Copy: LogicalTypeIdAny :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdArray :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdBigNum :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdBit :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdBlob :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdBoolean :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdDate :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdDecimal :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdDouble :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdEnum :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdFloat :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdGeometry :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdInteger :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdInterval :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdInvalid :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdList :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdMap :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdSQLNull :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdStruct :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTime :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTimeNs :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTimeTz :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTimestamp :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTimestampMs :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTimestampNs :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTimestampSec :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTimestampTz :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTimestampTzNs :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdTuple :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdType :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdUBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdUHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdUInteger :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdUSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdUTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdUUID :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdUnion :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdUnknown :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdVarchar :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: LogicalTypeIdVariant :: LogicalTypeId
+ Database.DuckDB.Simple.Copy: MultiFileOptions :: Maybe Text -> Maybe Text -> MultiFileOptions
+ Database.DuckDB.Simple.Copy: [copyFromColumnNames] :: CopyFromInfo -> [Text]
+ Database.DuckDB.Simple.Copy: [copyFromColumnTypes] :: CopyFromInfo -> [LogicalTypeId]
+ Database.DuckDB.Simple.Copy: [copyFromPath] :: CopyFromInfo -> FilePath
+ Database.DuckDB.Simple.Copy: [copyToBind] :: CopyTo bindState globalState -> CopyBindInfo -> IO bindState
+ Database.DuckDB.Simple.Copy: [copyToFinalize] :: CopyTo bindState globalState -> CopyFinalizeInfo bindState globalState -> IO ()
+ Database.DuckDB.Simple.Copy: [copyToInit] :: CopyTo bindState globalState -> CopyInitInfo bindState -> IO globalState
+ Database.DuckDB.Simple.Copy: [copyToSink] :: CopyTo bindState globalState -> CopySinkInfo bindState globalState -> [[Field]] -> IO ()
+ Database.DuckDB.Simple.Copy: [multiFileExtension] :: MultiFileOptions -> Maybe Text
+ Database.DuckDB.Simple.Copy: [multiFileReaderType] :: MultiFileOptions -> Maybe Text
+ Database.DuckDB.Simple.Copy: class TypedRow r
+ Database.DuckDB.Simple.Copy: data CopyFromInfo
+ Database.DuckDB.Simple.Copy: data CopyTo bindState globalState
+ Database.DuckDB.Simple.Copy: data LogicalTypeId
+ Database.DuckDB.Simple.Copy: data MultiFileOptions
+ Database.DuckDB.Simple.Copy: defaultMultiFileOptions :: MultiFileOptions
+ Database.DuckDB.Simple.Copy: registerCopyFromFunction :: TypedRow row => Connection -> Text -> (CopyFromInfo -> IO [row]) -> IO ()
+ Database.DuckDB.Simple.Copy: registerCopyFunction :: TypedRow row => Connection -> Text -> CopyTo bindState globalState -> (CopyFromInfo -> IO [row]) -> IO ()
+ Database.DuckDB.Simple.Copy: registerMultiFileFunction :: Connection -> Text -> Text -> MultiFileOptions -> IO ()
+ Database.DuckDB.Simple.Extension: MacroFunctionTypeScalar :: MacroFunctionType
+ Database.DuckDB.Simple.Extension: MacroFunctionTypeTable :: MacroFunctionType
+ Database.DuckDB.Simple.Extension: createAggregate :: (FromRow args, TypedRow args, FunctionResult result) => Extension -> Text -> Aggregate state args result -> IO ()
+ Database.DuckDB.Simple.Extension: createFunction :: Function f => Extension -> Text -> f -> IO ()
+ Database.DuckDB.Simple.Extension: createFunctionWithState :: Function f => Extension -> Text -> IO s -> (s -> f) -> IO ()
+ Database.DuckDB.Simple.Extension: createTableFunction :: (FromRow params, TypedRow params, TypedRow row) => Extension -> Text -> [Text] -> (params -> IO [row]) -> IO ()
+ Database.DuckDB.Simple.Extension: data Extension
+ Database.DuckDB.Simple.Extension: data MacroFunctionType
+ Database.DuckDB.Simple.Extension: extensionTarget :: Extension -> RegistrationTarget
+ Database.DuckDB.Simple.Extension: registerCopyFromFunction :: TypedRow row => Extension -> Text -> (CopyFromInfo -> IO [row]) -> IO ()
+ Database.DuckDB.Simple.Extension: registerCopyFunction :: TypedRow row => Extension -> Text -> CopyTo bindState globalState -> (CopyFromInfo -> IO [row]) -> IO ()
+ Database.DuckDB.Simple.Extension: registerCopyToFunction :: Extension -> Text -> (CopyBindInfo -> IO bindState) -> (CopyInitInfo bindState -> IO globalState) -> (CopySinkInfo bindState globalState -> [[Field]] -> IO ()) -> (CopyFinalizeInfo bindState globalState -> IO ()) -> IO ()
+ Database.DuckDB.Simple.Extension: registerCustomOption :: Extension -> CustomOption -> IO ()
+ Database.DuckDB.Simple.Extension: registerCustomType :: Extension -> Text -> TypeDeclaration -> IO ()
+ Database.DuckDB.Simple.Extension: registerLogSink :: Extension -> Text -> (LogEntry -> IO ()) -> IO () -> IO ()
+ Database.DuckDB.Simple.Extension: registerMacroFunction :: Extension -> Text -> MacroFunctionType -> [FunctionParameter] -> Text -> IO ()
+ Database.DuckDB.Simple.Extension: registerMultiFileFunction :: Extension -> Text -> Text -> MultiFileOptions -> IO ()
+ Database.DuckDB.Simple.Extension: registerPragmaFunction :: FromRow args => Extension -> Text -> [FunctionParameter] -> PragmaCallback args -> IO ()
+ Database.DuckDB.Simple.Extension: registerReplacementScan :: Extension -> ([Text] -> IO (Maybe Replacement)) -> IO ()
+ Database.DuckDB.Simple.Extension: runExtension :: (Extension -> IO ()) -> ExtensionEntry
+ Database.DuckDB.Simple.Extension: type ExtensionEntry = Ptr () -> Ptr () -> Ptr () -> IO ()
+ Database.DuckDB.Simple.FileSystem: FileFlagAppend :: FileFlag
+ Database.DuckDB.Simple.FileSystem: FileFlagCreate :: FileFlag
+ Database.DuckDB.Simple.FileSystem: FileFlagCreateNew :: FileFlag
+ Database.DuckDB.Simple.FileSystem: FileFlagExclusiveCreate :: FileFlag
+ Database.DuckDB.Simple.FileSystem: FileFlagExternalFileCache :: FileFlag
+ Database.DuckDB.Simple.FileSystem: FileFlagParallelAccess :: FileFlag
+ Database.DuckDB.Simple.FileSystem: FileFlagRead :: FileFlag
+ Database.DuckDB.Simple.FileSystem: FileFlagWrite :: FileFlag
+ Database.DuckDB.Simple.FileSystem: data FileFlag
+ Database.DuckDB.Simple.FileSystem: data FileHandle
+ Database.DuckDB.Simple.FileSystem: expandPath :: Connection -> FilePath -> IO FilePath
+ Database.DuckDB.Simple.FileSystem: globPaths :: Connection -> FilePath -> IO [FilePath]
+ Database.DuckDB.Simple.FileSystem: movePath :: Connection -> FilePath -> FilePath -> IO ()
+ Database.DuckDB.Simple.FileSystem: pathExists :: Connection -> FilePath -> IO Bool
+ Database.DuckDB.Simple.FileSystem: pathIsRemote :: Connection -> FilePath -> IO Bool
+ Database.DuckDB.Simple.FileSystem: readFileHandleAt :: FileHandle -> Int64 -> Int64 -> IO ByteString
+ Database.DuckDB.Simple.FileSystem: removePath :: Connection -> FilePath -> IO Bool
+ Database.DuckDB.Simple.FileSystem: writeFileHandleAt :: FileHandle -> Int64 -> ByteString -> IO ()
+ Database.DuckDB.Simple.FromField: FieldTuple :: TupleValue FieldValue -> FieldValue
+ Database.DuckDB.Simple.FromField: LogicalTypeAlias :: Text -> LogicalTypeRep -> LogicalTypeRep
+ Database.DuckDB.Simple.FromField: LogicalTypeCollatedVarchar :: Text -> LogicalTypeRep
+ Database.DuckDB.Simple.FromField: LogicalTypeIdAny :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdArray :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdBigNum :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdBit :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdBlob :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdBoolean :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdDate :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdDecimal :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdDouble :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdEnum :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdFloat :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdGeometry :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdInteger :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdInterval :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdInvalid :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdList :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdMap :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdSQLNull :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdStruct :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTime :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTimeNs :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTimeTz :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTimestamp :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTimestampMs :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTimestampNs :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTimestampSec :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTimestampTz :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTimestampTzNs :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdTuple :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdType :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdUBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdUHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdUInteger :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdUSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdUTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdUUID :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdUnion :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdUnknown :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdVarchar :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeIdVariant :: LogicalTypeId
+ Database.DuckDB.Simple.FromField: LogicalTypeTuple :: Array Int LogicalTypeRep -> LogicalTypeRep
+ Database.DuckDB.Simple.FromField: TupleValue :: Array Int a -> Array Int LogicalTypeRep -> TupleValue a
+ Database.DuckDB.Simple.FromField: [fieldType] :: Field -> Conversion LogicalTypeRep
+ Database.DuckDB.Simple.FromField: [timeWithZoneOffsetSeconds] :: TimeWithZone -> Int32
+ Database.DuckDB.Simple.FromField: [tupleValueFields] :: TupleValue a -> Array Int a
+ Database.DuckDB.Simple.FromField: [tupleValueTypes] :: TupleValue a -> Array Int LogicalTypeRep
+ Database.DuckDB.Simple.FromField: data LogicalTypeId
+ Database.DuckDB.Simple.FromField: data SizedArray (n :: Nat) a
+ Database.DuckDB.Simple.FromField: data TupleValue a
+ Database.DuckDB.Simple.FromField: fieldTypeName :: LogicalTypeRep -> FieldValue -> Text
+ Database.DuckDB.Simple.FromField: instance (GHC.Internal.TypeNats.KnownNat n, GHC.Internal.Data.Typeable.Internal.Typeable a, Database.DuckDB.Simple.FromField.FromField a) => Database.DuckDB.Simple.FromField.FromField (Database.DuckDB.Simple.FromField.SizedArray n a)
+ Database.DuckDB.Simple.FromField: instance Database.DuckDB.Simple.FromField.FromField (Database.DuckDB.Simple.LogicalRep.TupleValue Database.DuckDB.Simple.FromField.FieldValue)
+ Database.DuckDB.Simple.FromField: instance GHC.Internal.Base.Functor (Database.DuckDB.Simple.FromField.SizedArray n)
+ Database.DuckDB.Simple.FromField: instance GHC.Internal.Classes.Eq a => GHC.Internal.Classes.Eq (Database.DuckDB.Simple.FromField.SizedArray n a)
+ Database.DuckDB.Simple.FromField: instance GHC.Internal.Classes.Ord a => GHC.Internal.Classes.Ord (Database.DuckDB.Simple.FromField.SizedArray n a)
+ Database.DuckDB.Simple.FromField: instance GHC.Internal.Data.Foldable.Foldable (Database.DuckDB.Simple.FromField.SizedArray n)
+ Database.DuckDB.Simple.FromField: instance GHC.Internal.Data.Traversable.Traversable (Database.DuckDB.Simple.FromField.SizedArray n)
+ Database.DuckDB.Simple.FromField: instance GHC.Internal.Show.Show a => GHC.Internal.Show.Show (Database.DuckDB.Simple.FromField.SizedArray n a)
+ Database.DuckDB.Simple.FromField: sizedArray :: forall (n :: Nat) a. KnownNat n => Array Int a -> Maybe (SizedArray n a)
+ Database.DuckDB.Simple.FromField: sizedArrayFromList :: forall (n :: Nat) a. KnownNat n => [a] -> Maybe (SizedArray n a)
+ Database.DuckDB.Simple.FromField: sizedArrayToArray :: forall (n :: Nat) a. SizedArray n a -> Array Int a
+ Database.DuckDB.Simple.FromField: timeWithZone :: TimeOfDay -> TimeZone -> TimeWithZone
+ Database.DuckDB.Simple.FromField: timeWithZoneZone :: TimeWithZone -> TimeZone
+ Database.DuckDB.Simple.Function: Aggregate :: state -> (state -> args -> IO state) -> (state -> state -> IO state) -> (state -> IO result) -> Aggregate state args result
+ Database.DuckDB.Simple.Function: LogicalTypeIdAny :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdArray :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdBigNum :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdBit :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdBlob :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdBoolean :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdDate :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdDecimal :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdDouble :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdEnum :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdFloat :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdGeometry :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdInteger :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdInterval :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdInvalid :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdList :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdMap :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdSQLNull :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdStruct :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTime :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTimeNs :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTimeTz :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTimestamp :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTimestampMs :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTimestampNs :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTimestampSec :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTimestampTz :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTimestampTzNs :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdTuple :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdType :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdUBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdUHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdUInteger :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdUSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdUTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdUUID :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdUnion :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdUnknown :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdVarchar :: LogicalTypeId
+ Database.DuckDB.Simple.Function: LogicalTypeIdVariant :: LogicalTypeId
+ Database.DuckDB.Simple.Function: StructuredType :: LogicalTypeRep -> TypeDeclaration
+ Database.DuckDB.Simple.Function: TypeExpression :: Text -> TypeDeclaration
+ Database.DuckDB.Simple.Function: [aggregateCombine] :: Aggregate state args result -> state -> state -> IO state
+ Database.DuckDB.Simple.Function: [aggregateInitial] :: Aggregate state args result -> state
+ Database.DuckDB.Simple.Function: [aggregateResult] :: Aggregate state args result -> state -> IO result
+ Database.DuckDB.Simple.Function: [aggregateStep] :: Aggregate state args result -> state -> args -> IO state
+ Database.DuckDB.Simple.Function: class TypedRow r
+ Database.DuckDB.Simple.Function: createAggregate :: (FromRow args, TypedRow args, FunctionResult result) => Connection -> Text -> Aggregate state args result -> IO ()
+ Database.DuckDB.Simple.Function: createTableFunction :: (FromRow params, TypedRow params, TypedRow row) => Connection -> Text -> [Text] -> (params -> IO [row]) -> IO ()
+ Database.DuckDB.Simple.Function: data Aggregate state args result
+ Database.DuckDB.Simple.Function: data LogicalTypeId
+ Database.DuckDB.Simple.Function: data TypeDeclaration
+ Database.DuckDB.Simple.Generic: duckFromTypedField :: DuckValue a => LogicalTypeRep -> FieldValue -> Conversion a
+ Database.DuckDB.Simple.Generic: instance (GHC.Internal.Generics.Generic a, Database.DuckDB.Simple.Generic.GToField (GHC.Internal.Generics.Rep a)) => Database.DuckDB.Simple.ToField.ToDuckValue (Database.DuckDB.Simple.Generic.ViaDuckDB a)
+ Database.DuckDB.Simple.Generic: instance (GHC.Internal.TypeNats.KnownNat n, Database.DuckDB.Simple.Generic.DuckValue a) => Database.DuckDB.Simple.Generic.DuckValue (Database.DuckDB.Simple.FromField.SizedArray n a)
+ Database.DuckDB.Simple.Generic: typedFromField :: FromField a => LogicalTypeRep -> FieldValue -> Conversion a
+ Database.DuckDB.Simple.Internal: ArrowArray :: Int64 -> Int64 -> Int64 -> Int64 -> Int64 -> Ptr (PtrConst Void) -> Ptr (Ptr ArrowArray) -> Ptr ArrowArray -> FunPtr (Ptr ArrowArray -> IO ()) -> Ptr Void -> ArrowArray
+ Database.DuckDB.Simple.Internal: ArrowArrayStream :: FunPtr (Ptr ArrowArrayStream -> Ptr ArrowSchema -> IO CInt) -> FunPtr (Ptr ArrowArrayStream -> Ptr ArrowArray -> IO CInt) -> FunPtr (Ptr ArrowArrayStream -> IO (PtrConst CChar)) -> FunPtr (Ptr ArrowArrayStream -> IO ()) -> Ptr Void -> ArrowArrayStream
+ Database.DuckDB.Simple.Internal: ArrowSchema :: PtrConst CChar -> PtrConst CChar -> PtrConst CChar -> Int64 -> Int64 -> Ptr (Ptr ArrowSchema) -> Ptr ArrowSchema -> FunPtr (Ptr ArrowSchema -> IO ()) -> Ptr Void -> ArrowSchema
+ Database.DuckDB.Simple.Internal: ConnectionOptions :: [(Text, Text)] -> Maybe Int -> Int -> Int -> ConnectionOptions
+ Database.DuckDB.Simple.Internal: DuckDBIdx :: Word64 -> DuckDBIdx
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionBindCallbackFn :: FunPtr DuckDBV2AggregateFunctionBindCallbackFn_Aux -> DuckDBV2AggregateFunctionBindCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionBindCallbackFn_Aux :: (DuckDBV2FunctionBindInfoHandle -> DuckDBV2AggregateFunctionBindInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2AggregateFunctionBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionBindInfoHandle :: Ptr DuckDBV2AggregateFunctionBindInfoStruct -> DuckDBV2AggregateFunctionBindInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionCombineCallbackFn :: FunPtr DuckDBV2AggregateFunctionCombineCallbackFn_Aux -> DuckDBV2AggregateFunctionCombineCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionCombineCallbackFn_Aux :: (DuckDBV2AggregateFunctionCombineInfoHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2AggregateFunctionCombineCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionCombineInfoHandle :: Ptr DuckDBV2AggregateFunctionCombineInfoStruct -> DuckDBV2AggregateFunctionCombineInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionDestroyCallbackFn :: FunPtr DuckDBV2AggregateFunctionDestroyCallbackFn_Aux -> DuckDBV2AggregateFunctionDestroyCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionDestroyCallbackFn_Aux :: (DuckDBV2AggregateFunctionDestroyInfoHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2AggregateFunctionDestroyCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionDestroyInfoHandle :: Ptr DuckDBV2AggregateFunctionDestroyInfoStruct -> DuckDBV2AggregateFunctionDestroyInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionFinalizeCallbackFn :: FunPtr DuckDBV2AggregateFunctionFinalizeCallbackFn_Aux -> DuckDBV2AggregateFunctionFinalizeCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionFinalizeCallbackFn_Aux :: (DuckDBV2AggregateFunctionFinalizeInfoHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2AggregateFunctionFinalizeCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionFinalizeInfoHandle :: Ptr DuckDBV2AggregateFunctionFinalizeInfoStruct -> DuckDBV2AggregateFunctionFinalizeInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionHandle :: Ptr DuckDBV2AggregateFunctionStruct -> DuckDBV2AggregateFunctionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionInitCallbackFn :: FunPtr DuckDBV2AggregateFunctionInitCallbackFn_Aux -> DuckDBV2AggregateFunctionInitCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionInitCallbackFn_Aux :: (DuckDBV2AggregateFunctionInitInfoHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2AggregateFunctionInitCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionInitInfoHandle :: Ptr DuckDBV2AggregateFunctionInitInfoStruct -> DuckDBV2AggregateFunctionInitInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionSizeCallbackFn :: FunPtr DuckDBV2AggregateFunctionSizeCallbackFn_Aux -> DuckDBV2AggregateFunctionSizeCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionSizeCallbackFn_Aux :: (DuckDBV2AggregateFunctionSizeInfoHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2AggregateFunctionSizeCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionSizeInfoHandle :: Ptr DuckDBV2AggregateFunctionSizeInfoStruct -> DuckDBV2AggregateFunctionSizeInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionUpdateCallbackFn :: FunPtr DuckDBV2AggregateFunctionUpdateCallbackFn_Aux -> DuckDBV2AggregateFunctionUpdateCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionUpdateCallbackFn_Aux :: (DuckDBV2AggregateFunctionUpdateInfoHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2AggregateFunctionUpdateCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2AggregateFunctionUpdateInfoHandle :: Ptr DuckDBV2AggregateFunctionUpdateInfoStruct -> DuckDBV2AggregateFunctionUpdateInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ArenaHandle :: Ptr DuckDBV2ArenaStruct -> DuckDBV2ArenaHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ArrowExporterHandle :: Ptr DuckDBV2ArrowExporterStruct -> DuckDBV2ArrowExporterHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ArrowImporterHandle :: Ptr DuckDBV2ArrowImporterStruct -> DuckDBV2ArrowImporterHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ArrowResultHandle :: Ptr DuckDBV2ArrowResultStruct -> DuckDBV2ArrowResultHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2AttachOptionsHandle :: Ptr DuckDBV2AttachOptionsStruct -> DuckDBV2AttachOptionsHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2BignumT :: DuckDBV2Bytes -> DuckDBV2BignumT
+ Database.DuckDB.Simple.Internal: DuckDBV2BitT :: DuckDBV2Bytes -> DuckDBV2BitT
+ Database.DuckDB.Simple.Internal: DuckDBV2BlobT :: DuckDBV2Bytes -> DuckDBV2BlobT
+ Database.DuckDB.Simple.Internal: DuckDBV2Bytes :: Duckdb_v2_bytes_value -> DuckDBV2Bytes
+ Database.DuckDB.Simple.Internal: DuckDBV2CastFunctionExecCallbackFn :: FunPtr DuckDBV2CastFunctionExecCallbackFn_Aux -> DuckDBV2CastFunctionExecCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CastFunctionExecCallbackFn_Aux :: (DuckDBV2CastFunctionExecInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CastFunctionExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CastFunctionExecInfoHandle :: Ptr DuckDBV2CastFunctionExecInfoStruct -> DuckDBV2CastFunctionExecInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CastFunctionHandle :: Ptr DuckDBV2CastFunctionStruct -> DuckDBV2CastFunctionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CastMode :: CUInt -> DuckDBV2CastMode
+ Database.DuckDB.Simple.Internal: DuckDBV2ColumnDataCollectionAppendStateHandle :: Ptr DuckDBV2ColumnDataCollectionAppendStateStruct -> DuckDBV2ColumnDataCollectionAppendStateHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ColumnDataCollectionHandle :: Ptr DuckDBV2ColumnDataCollectionStruct -> DuckDBV2ColumnDataCollectionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ColumnDataCollectionSharedScanStateHandle :: Ptr DuckDBV2ColumnDataCollectionSharedScanStateStruct -> DuckDBV2ColumnDataCollectionSharedScanStateHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ColumnDataCollectionWorkerScanStateHandle :: Ptr DuckDBV2ColumnDataCollectionWorkerScanStateStruct -> DuckDBV2ColumnDataCollectionWorkerScanStateHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ColumnDescriptionHandle :: Ptr DuckDBV2ColumnDescriptionStruct -> DuckDBV2ColumnDescriptionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ConfigHandle :: Ptr DuckDBV2ConfigStruct -> DuckDBV2ConfigHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ConnectionHandle :: Ptr DuckDBV2ConnectionStruct -> DuckDBV2ConnectionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ContextHandle :: Ptr DuckDBV2ContextStruct -> DuckDBV2ContextHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromBindCallbackFn :: FunPtr DuckDBV2CopyFromBindCallbackFn_Aux -> DuckDBV2CopyFromBindCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromBindCallbackFn_Aux :: (DuckDBV2CopyFromBindInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyFromBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromBindInfoHandle :: Ptr DuckDBV2CopyFromBindInfoStruct -> DuckDBV2CopyFromBindInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromExecCallbackFn :: FunPtr DuckDBV2CopyFromExecCallbackFn_Aux -> DuckDBV2CopyFromExecCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromExecCallbackFn_Aux :: (DuckDBV2CopyFromExecInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyFromExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromExecInfoHandle :: Ptr DuckDBV2CopyFromExecInfoStruct -> DuckDBV2CopyFromExecInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromInitGlobalCallbackFn :: FunPtr DuckDBV2CopyFromInitGlobalCallbackFn_Aux -> DuckDBV2CopyFromInitGlobalCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromInitGlobalCallbackFn_Aux :: (DuckDBV2CopyFromInitGlobalInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyFromInitGlobalCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromInitGlobalInfoHandle :: Ptr DuckDBV2CopyFromInitGlobalInfoStruct -> DuckDBV2CopyFromInitGlobalInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromInitLocalCallbackFn :: FunPtr DuckDBV2CopyFromInitLocalCallbackFn_Aux -> DuckDBV2CopyFromInitLocalCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromInitLocalCallbackFn_Aux :: (DuckDBV2CopyFromInitLocalInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyFromInitLocalCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromInitLocalInfoHandle :: Ptr DuckDBV2CopyFromInitLocalInfoStruct -> DuckDBV2CopyFromInitLocalInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromProgressCallbackFn :: FunPtr DuckDBV2CopyFromProgressCallbackFn_Aux -> DuckDBV2CopyFromProgressCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromProgressCallbackFn_Aux :: (DuckDBV2CopyFromProgressInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyFromProgressCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFromProgressInfoHandle :: Ptr DuckDBV2CopyFromProgressInfoStruct -> DuckDBV2CopyFromProgressInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyFunctionHandle :: Ptr DuckDBV2CopyFunctionStruct -> DuckDBV2CopyFunctionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBatchCallbackFn :: FunPtr DuckDBV2CopyToBatchCallbackFn_Aux -> DuckDBV2CopyToBatchCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBatchCallbackFn_Aux :: (DuckDBV2CopyToBatchInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyToBatchCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBatchInfoHandle :: Ptr DuckDBV2CopyToBatchInfoStruct -> DuckDBV2CopyToBatchInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBatchSizeCallbackFn :: FunPtr DuckDBV2CopyToBatchSizeCallbackFn_Aux -> DuckDBV2CopyToBatchSizeCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBatchSizeCallbackFn_Aux :: (DuckDBV2CopyToBatchSizeInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyToBatchSizeCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBatchSizeInfoHandle :: Ptr DuckDBV2CopyToBatchSizeInfoStruct -> DuckDBV2CopyToBatchSizeInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBindCallbackFn :: FunPtr DuckDBV2CopyToBindCallbackFn_Aux -> DuckDBV2CopyToBindCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBindCallbackFn_Aux :: (DuckDBV2CopyToBindInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyToBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToBindInfoHandle :: Ptr DuckDBV2CopyToBindInfoStruct -> DuckDBV2CopyToBindInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToFinalizeCallbackFn :: FunPtr DuckDBV2CopyToFinalizeCallbackFn_Aux -> DuckDBV2CopyToFinalizeCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToFinalizeCallbackFn_Aux :: (DuckDBV2CopyToFinalizeInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyToFinalizeCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToFinalizeInfoHandle :: Ptr DuckDBV2CopyToFinalizeInfoStruct -> DuckDBV2CopyToFinalizeInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToFlushCallbackFn :: FunPtr DuckDBV2CopyToFlushCallbackFn_Aux -> DuckDBV2CopyToFlushCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToFlushCallbackFn_Aux :: (DuckDBV2CopyToFlushInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyToFlushCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToFlushInfoHandle :: Ptr DuckDBV2CopyToFlushInfoStruct -> DuckDBV2CopyToFlushInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToInitCallbackFn :: FunPtr DuckDBV2CopyToInitCallbackFn_Aux -> DuckDBV2CopyToInitCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToInitCallbackFn_Aux :: (DuckDBV2CopyToInitInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyToInitCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToInitInfoHandle :: Ptr DuckDBV2CopyToInitInfoStruct -> DuckDBV2CopyToInitInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToStatisticsCallbackFn :: FunPtr DuckDBV2CopyToStatisticsCallbackFn_Aux -> DuckDBV2CopyToStatisticsCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToStatisticsCallbackFn_Aux :: (DuckDBV2CopyToStatisticsInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CopyToStatisticsCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CopyToStatisticsInfoHandle :: Ptr DuckDBV2CopyToStatisticsInfoStruct -> DuckDBV2CopyToStatisticsInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CustomOptionHandle :: Ptr DuckDBV2CustomOptionStruct -> DuckDBV2CustomOptionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CustomOptionUpdateCallbackFn :: FunPtr DuckDBV2CustomOptionUpdateCallbackFn_Aux -> DuckDBV2CustomOptionUpdateCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2CustomOptionUpdateCallbackFn_Aux :: (DuckDBV2CustomOptionUpdateInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2CustomOptionUpdateCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2CustomOptionUpdateInfoHandle :: Ptr DuckDBV2CustomOptionUpdateInfoStruct -> DuckDBV2CustomOptionUpdateInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2CustomTypeHandle :: Ptr DuckDBV2CustomTypeStruct -> DuckDBV2CustomTypeHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2DataChunkHandle :: Ptr DuckDBV2DataChunkStruct -> DuckDBV2DataChunkHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2EnvironmentHandle :: Ptr DuckDBV2EnvironmentStruct -> DuckDBV2EnvironmentHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2Error :: CUInt -> DuckDBV2Error
+ Database.DuckDB.Simple.Internal: DuckDBV2ErrorInfoHandle :: Ptr DuckDBV2ErrorInfoStruct -> DuckDBV2ErrorInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ExpressionHandle :: Ptr DuckDBV2ExpressionStruct -> DuckDBV2ExpressionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ExpressionType :: CUInt -> DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: DuckDBV2ExtensionGetApiFn :: FunPtr DuckDBV2ExtensionGetApiFn_Aux -> DuckDBV2ExtensionGetApiFn
+ Database.DuckDB.Simple.Internal: DuckDBV2ExtensionGetApiFn_Aux :: (DuckDBV2ExtensionHandle -> PtrConst CChar -> IO (PtrConst Void)) -> DuckDBV2ExtensionGetApiFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2ExtensionHandle :: Ptr DuckDBExtensionInfoStruct -> DuckDBV2ExtensionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ExtensionInput :: DuckDBV2ExtensionGetApiFn -> DuckDBV2ExtensionHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> DuckDBV2ExtensionInput
+ Database.DuckDB.Simple.Internal: DuckDBV2FactoryHandle :: Ptr DuckDBV2FactoryStruct -> DuckDBV2FactoryHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2FileFlag :: CUInt -> DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: DuckDBV2FileHandle :: Ptr DuckDBV2FileStruct -> DuckDBV2FileHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2FileOpenOptionsHandle :: Ptr DuckDBV2FileOpenOptionsStruct -> DuckDBV2FileOpenOptionsHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2FileSystemHandle :: Ptr DuckDBV2FileSystemStruct -> DuckDBV2FileSystemHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2FileSystemPathGlobCallbackFn :: FunPtr DuckDBV2FileSystemPathGlobCallbackFn_Aux -> DuckDBV2FileSystemPathGlobCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2FileSystemPathGlobCallbackFn_Aux :: (PtrConst DuckDBV2Str -> Ptr Void -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2FileSystemPathGlobCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2FunctionBindInfoHandle :: Ptr DuckDBV2FunctionBindInfoStruct -> DuckDBV2FunctionBindInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2FunctionDocsHandle :: Ptr DuckDBV2FunctionDocsStruct -> DuckDBV2FunctionDocsHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2FunctionParameterKind :: CUInt -> DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.Internal: DuckDBV2FunctionPropertyKey :: CUInt -> DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: DuckDBV2FunctionPropertyValue :: CUInt -> DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: DuckDBV2FunctionSignatureHandle :: Ptr DuckDBV2FunctionSignatureStruct -> DuckDBV2FunctionSignatureHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2HugeintT :: Word64 -> Int64 -> DuckDBV2HugeintT
+ Database.DuckDB.Simple.Internal: DuckDBV2IdentifierT :: DuckDBV2Str -> DuckDBV2IdentifierT
+ Database.DuckDB.Simple.Internal: DuckDBV2InstanceHandle :: Ptr DuckDBV2InstanceStruct -> DuckDBV2InstanceHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2IntervalT :: Int32 -> Int32 -> Int64 -> DuckDBV2IntervalT
+ Database.DuckDB.Simple.Internal: DuckDBV2ListEntry :: DuckDBIdx -> DuckDBIdx -> DuckDBV2ListEntry
+ Database.DuckDB.Simple.Internal: DuckDBV2LogLevel :: CUInt -> DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: DuckDBV2LogScope :: CUInt -> DuckDBV2LogScope
+ Database.DuckDB.Simple.Internal: DuckDBV2LogSinkFlushCallbackFn :: FunPtr DuckDBV2LogSinkFlushCallbackFn_Aux -> DuckDBV2LogSinkFlushCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2LogSinkFlushCallbackFn_Aux :: (DuckDBV2LogSinkFlushInfoHandle -> Ptr Void -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2LogSinkFlushCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2LogSinkFlushInfoHandle :: Ptr DuckDBV2LogSinkFlushInfoStruct -> DuckDBV2LogSinkFlushInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2LogSinkHandle :: Ptr DuckDBV2LogSinkStruct -> DuckDBV2LogSinkHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2LogSinkWriteCallbackFn :: FunPtr DuckDBV2LogSinkWriteCallbackFn_Aux -> DuckDBV2LogSinkWriteCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2LogSinkWriteCallbackFn_Aux :: (DuckDBV2LogSinkWriteInfoHandle -> Ptr Void -> Int64 -> DuckDBV2LogLevel -> DuckDBV2LogScope -> PtrConst DuckDBV2Str -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2LogSinkWriteCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2LogSinkWriteInfoHandle :: Ptr DuckDBV2LogSinkWriteInfoStruct -> DuckDBV2LogSinkWriteInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2LogicalTypeHandle :: Ptr DuckDBV2LogicalTypeStruct -> DuckDBV2LogicalTypeHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2LogicalTypeId :: CUInt -> DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: DuckDBV2MacroFunctionHandle :: Ptr DuckDBV2MacroFunctionStruct -> DuckDBV2MacroFunctionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2MacroFunctionType :: CUInt -> DuckDBV2MacroFunctionType
+ Database.DuckDB.Simple.Internal: DuckDBV2MultiFileFunctionHandle :: Ptr DuckDBV2MultiFileFunctionStruct -> DuckDBV2MultiFileFunctionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2Opaque :: Ptr Void -> DuckDBV2OpaqueDestroyFn -> DuckDBV2OpaqueEqualsFn -> DuckDBV2Opaque
+ Database.DuckDB.Simple.Internal: DuckDBV2OpaqueDestroyFn :: FunPtr DuckDBV2OpaqueDestroyFn_Aux -> DuckDBV2OpaqueDestroyFn
+ Database.DuckDB.Simple.Internal: DuckDBV2OpaqueDestroyFn_Aux :: (Ptr Void -> IO ()) -> DuckDBV2OpaqueDestroyFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2OpaqueEqualsFn :: FunPtr DuckDBV2OpaqueEqualsFn_Aux -> DuckDBV2OpaqueEqualsFn
+ Database.DuckDB.Simple.Internal: DuckDBV2OpaqueEqualsFn_Aux :: (Ptr Void -> Ptr Void -> IO CBool) -> DuckDBV2OpaqueEqualsFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2OptionHandle :: Ptr DuckDBV2OptionStruct -> DuckDBV2OptionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2OrderPreservation :: CUInt -> DuckDBV2OrderPreservation
+ Database.DuckDB.Simple.Internal: DuckDBV2PragmaFunctionExecCallbackFn :: FunPtr DuckDBV2PragmaFunctionExecCallbackFn_Aux -> DuckDBV2PragmaFunctionExecCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2PragmaFunctionExecCallbackFn_Aux :: (DuckDBV2FunctionBindInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2PragmaFunctionExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2PragmaFunctionHandle :: Ptr DuckDBV2PragmaFunctionStruct -> DuckDBV2PragmaFunctionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2PragmaFunctionQueryCallbackFn :: FunPtr DuckDBV2PragmaFunctionQueryCallbackFn_Aux -> DuckDBV2PragmaFunctionQueryCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2PragmaFunctionQueryCallbackFn_Aux :: (DuckDBV2FunctionBindInfoHandle -> DuckDBV2PragmaFunctionQueryInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2PragmaFunctionQueryCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2PragmaFunctionQueryInfoHandle :: Ptr DuckDBV2PragmaFunctionQueryInfoStruct -> DuckDBV2PragmaFunctionQueryInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2PreparedStatementHandle :: Ptr DuckDBV2PreparedStatementStruct -> DuckDBV2PreparedStatementHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2QnameHandle :: Ptr DuckDBV2QnameStruct -> DuckDBV2QnameHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2QueryMetricsHandle :: Ptr DuckDBV2QueryMetricsStruct -> DuckDBV2QueryMetricsHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2QueryMetricsNodeHandle :: Ptr DuckDBV2QueryMetricsNodeStruct -> DuckDBV2QueryMetricsNodeHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ReplacementScanCallbackFn :: FunPtr DuckDBV2ReplacementScanCallbackFn_Aux -> DuckDBV2ReplacementScanCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2ReplacementScanCallbackFn_Aux :: (DuckDBV2ReplacementScanInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2ReplacementScanCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2ReplacementScanHandle :: Ptr DuckDBV2ReplacementScanStruct -> DuckDBV2ReplacementScanHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ReplacementScanInfoHandle :: Ptr DuckDBV2ReplacementScanInfoStruct -> DuckDBV2ReplacementScanInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ResultHandle :: Ptr DuckDBV2ResultStruct -> DuckDBV2ResultHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ResultStepStatus :: CUInt -> DuckDBV2ResultStepStatus
+ Database.DuckDB.Simple.Internal: DuckDBV2ResultType :: CUInt -> DuckDBV2ResultType
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionBindCallbackFn :: FunPtr DuckDBV2ScalarFunctionBindCallbackFn_Aux -> DuckDBV2ScalarFunctionBindCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionBindCallbackFn_Aux :: (DuckDBV2FunctionBindInfoHandle -> DuckDBV2ScalarFunctionBindInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2ScalarFunctionBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionBindInfoHandle :: Ptr DuckDBV2ScalarFunctionBindInfoStruct -> DuckDBV2ScalarFunctionBindInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionExecCallbackFn :: FunPtr DuckDBV2ScalarFunctionExecCallbackFn_Aux -> DuckDBV2ScalarFunctionExecCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionExecCallbackFn_Aux :: (DuckDBV2ScalarFunctionExecInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2ScalarFunctionExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionExecInfoHandle :: Ptr DuckDBV2ScalarFunctionExecInfoStruct -> DuckDBV2ScalarFunctionExecInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionHandle :: Ptr DuckDBV2ScalarFunctionStruct -> DuckDBV2ScalarFunctionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionInitCallbackFn :: FunPtr DuckDBV2ScalarFunctionInitCallbackFn_Aux -> DuckDBV2ScalarFunctionInitCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionInitCallbackFn_Aux :: (DuckDBV2ScalarFunctionInitInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2ScalarFunctionInitCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionInitInfoHandle :: Ptr DuckDBV2ScalarFunctionInitInfoStruct -> DuckDBV2ScalarFunctionInitInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionResolveTypesCallbackFn :: FunPtr DuckDBV2ScalarFunctionResolveTypesCallbackFn_Aux -> DuckDBV2ScalarFunctionResolveTypesCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionResolveTypesCallbackFn_Aux :: (DuckDBV2FunctionBindInfoHandle -> DuckDBV2ScalarFunctionResolveTypesInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2ScalarFunctionResolveTypesCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2ScalarFunctionResolveTypesInfoHandle :: Ptr DuckDBV2ScalarFunctionResolveTypesInfoStruct -> DuckDBV2ScalarFunctionResolveTypesInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2SchemaHandle :: Ptr DuckDBV2SchemaStruct -> DuckDBV2SchemaHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2SelT :: Word32 -> DuckDBV2SelT
+ Database.DuckDB.Simple.Internal: DuckDBV2SettingScope :: CUInt -> DuckDBV2SettingScope
+ Database.DuckDB.Simple.Internal: DuckDBV2SqlStatementHandle :: Ptr DuckDBV2SqlStatementStruct -> DuckDBV2SqlStatementHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2StatementIteratorHandle :: Ptr DuckDBV2StatementIteratorStruct -> DuckDBV2StatementIteratorHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2StatementType :: CUInt -> DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: DuckDBV2Str :: PtrConst CChar -> DuckDBIdx -> DuckDBV2Str
+ Database.DuckDB.Simple.Internal: DuckDBV2TableDescriptionHandle :: Ptr DuckDBV2TableDescriptionStruct -> DuckDBV2TableDescriptionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionBindCallbackFn :: FunPtr DuckDBV2TableFunctionBindCallbackFn_Aux -> DuckDBV2TableFunctionBindCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionBindCallbackFn_Aux :: (DuckDBV2FunctionBindInfoHandle -> DuckDBV2TableFunctionBindInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionBindInfoHandle :: Ptr DuckDBV2TableFunctionBindInfoStruct -> DuckDBV2TableFunctionBindInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionClaimBatchCallbackFn :: FunPtr DuckDBV2TableFunctionClaimBatchCallbackFn_Aux -> DuckDBV2TableFunctionClaimBatchCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionClaimBatchCallbackFn_Aux :: (DuckDBV2TableFunctionClaimBatchInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionClaimBatchCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionClaimBatchInfoHandle :: Ptr DuckDBV2TableFunctionClaimBatchInfoStruct -> DuckDBV2TableFunctionClaimBatchInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionExecCallbackFn :: FunPtr DuckDBV2TableFunctionExecCallbackFn_Aux -> DuckDBV2TableFunctionExecCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionExecCallbackFn_Aux :: (DuckDBV2TableFunctionExecInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionExecInfoHandle :: Ptr DuckDBV2TableFunctionExecInfoStruct -> DuckDBV2TableFunctionExecInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionFilterPushdownCallbackFn :: FunPtr DuckDBV2TableFunctionFilterPushdownCallbackFn_Aux -> DuckDBV2TableFunctionFilterPushdownCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionFilterPushdownCallbackFn_Aux :: (DuckDBV2TableFunctionFilterPushdownInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionFilterPushdownCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionFilterPushdownInfoHandle :: Ptr DuckDBV2TableFunctionFilterPushdownInfoStruct -> DuckDBV2TableFunctionFilterPushdownInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionGetBindInfoCallbackFn :: FunPtr DuckDBV2TableFunctionGetBindInfoCallbackFn_Aux -> DuckDBV2TableFunctionGetBindInfoCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionGetBindInfoCallbackFn_Aux :: (DuckDBV2TableFunctionGetBindInfoInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionGetBindInfoCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionGetBindInfoInfoHandle :: Ptr DuckDBV2TableFunctionGetBindInfoInfoStruct -> DuckDBV2TableFunctionGetBindInfoInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionHandle :: Ptr DuckDBV2TableFunctionStruct -> DuckDBV2TableFunctionHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionInitGlobalCallbackFn :: FunPtr DuckDBV2TableFunctionInitGlobalCallbackFn_Aux -> DuckDBV2TableFunctionInitGlobalCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionInitGlobalCallbackFn_Aux :: (DuckDBV2TableFunctionInitGlobalInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionInitGlobalCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionInitGlobalInfoHandle :: Ptr DuckDBV2TableFunctionInitGlobalInfoStruct -> DuckDBV2TableFunctionInitGlobalInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionInitLocalCallbackFn :: FunPtr DuckDBV2TableFunctionInitLocalCallbackFn_Aux -> DuckDBV2TableFunctionInitLocalCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionInitLocalCallbackFn_Aux :: (DuckDBV2TableFunctionInitLocalInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionInitLocalCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionInitLocalInfoHandle :: Ptr DuckDBV2TableFunctionInitLocalInfoStruct -> DuckDBV2TableFunctionInitLocalInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionPartitionDataCallbackFn :: FunPtr DuckDBV2TableFunctionPartitionDataCallbackFn_Aux -> DuckDBV2TableFunctionPartitionDataCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionPartitionDataCallbackFn_Aux :: (DuckDBV2TableFunctionPartitionDataInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionPartitionDataCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionPartitionDataInfoHandle :: Ptr DuckDBV2TableFunctionPartitionDataInfoStruct -> DuckDBV2TableFunctionPartitionDataInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionPartitioningCallbackFn :: FunPtr DuckDBV2TableFunctionPartitioningCallbackFn_Aux -> DuckDBV2TableFunctionPartitioningCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionPartitioningCallbackFn_Aux :: (DuckDBV2TableFunctionPartitioningInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionPartitioningCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionPartitioningInfoHandle :: Ptr DuckDBV2TableFunctionPartitioningInfoStruct -> DuckDBV2TableFunctionPartitioningInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionProgressCallbackFn :: FunPtr DuckDBV2TableFunctionProgressCallbackFn_Aux -> DuckDBV2TableFunctionProgressCallbackFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionProgressCallbackFn_Aux :: (DuckDBV2TableFunctionProgressInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TableFunctionProgressCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TableFunctionProgressInfoHandle :: Ptr DuckDBV2TableFunctionProgressInfoStruct -> DuckDBV2TableFunctionProgressInfoHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TablePartitionInfo :: CUInt -> DuckDBV2TablePartitionInfo
+ Database.DuckDB.Simple.Internal: DuckDBV2TextSinkFn :: FunPtr DuckDBV2TextSinkFn_Aux -> DuckDBV2TextSinkFn
+ Database.DuckDB.Simple.Internal: DuckDBV2TextSinkFn_Aux :: (PtrConst DuckDBV2Str -> Ptr Void -> Ptr DuckDBV2ErrorInfoHandle -> IO ()) -> DuckDBV2TextSinkFn_Aux
+ Database.DuckDB.Simple.Internal: DuckDBV2TokenIteratorHandle :: Ptr DuckDBV2TokenIteratorStruct -> DuckDBV2TokenIteratorHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2TokenType :: CUInt -> DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: DuckDBV2UhugeintT :: Word64 -> Word64 -> DuckDBV2UhugeintT
+ Database.DuckDB.Simple.Internal: DuckDBV2ValueHandle :: Ptr DuckDBV2ValueStruct -> DuckDBV2ValueHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2VarcharT :: DuckDBV2Bytes -> DuckDBV2VarcharT
+ Database.DuckDB.Simple.Internal: DuckDBV2VectorHandle :: Ptr DuckDBV2VectorStruct -> DuckDBV2VectorHandle
+ Database.DuckDB.Simple.Internal: DuckDBV2VectorType :: CUInt -> DuckDBV2VectorType
+ Database.DuckDB.Simple.Internal: DuckDBV2VectorView :: PtrConst Void -> PtrConst Word64 -> PtrConst DuckDBV2SelT -> DuckDBIdx -> DuckDBV2VectorView
+ Database.DuckDB.Simple.Internal: Duckdb_v2_bytes_value :: ByteArray -> Duckdb_v2_bytes_value
+ Database.DuckDB.Simple.Internal: Duckdb_v2_bytes_value_inlined :: Word32 -> ConstantArray 12 CChar -> Duckdb_v2_bytes_value_inlined
+ Database.DuckDB.Simple.Internal: Duckdb_v2_bytes_value_pointer :: Word32 -> ConstantArray 4 CChar -> Ptr CChar -> Duckdb_v2_bytes_value_pointer
+ Database.DuckDB.Simple.Internal: Environment :: IORef EnvironmentState -> Environment
+ Database.DuckDB.Simple.Internal: EnvironmentClosed :: EnvironmentState
+ Database.DuckDB.Simple.Internal: EnvironmentOpen :: DuckDBV2EnvironmentHandle -> Children InstanceState -> EnvironmentState
+ Database.DuckDB.Simple.Internal: ErrorApi :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorConfigurationInvalid :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorConfigurationPermission :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorConfigurationSettings :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorDatabaseCatalog :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorDatabaseConstraint :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorDatabaseIndex :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorDatabaseSequence :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorDatabaseSerialization :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorDatabaseStatistics :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorDatabaseTransaction :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorIOEof :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorIOFileNotFound :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorIOGeneral :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorIOHTTP :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorIONetwork :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorIOReadFailure :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorInputInvalid :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorInputMetadataDeferred :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorInputObjectSize :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorInputOutOfRange :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorInputParameterInvalid :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryBinder :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryExecutor :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryExpression :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryNotImplemented :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryOptimizer :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryParameterNotAllowed :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryParameterNotResolved :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryParser :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryPlanner :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQueryScheduler :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorQuerySyntax :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorResourceAutoload :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorResourceConnection :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorResourceDependency :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorResourceInUse :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorResourceMissingExtension :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorResourceOutOfMemory :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorRuntimeFatal :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorRuntimeInternal :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorRuntimeInterrupt :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorRuntimeNullPointer :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorTypeConversion :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorTypeDecimal :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorTypeDivideByZero :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorTypeInvalid :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorTypeMismatch :: ErrorType
+ Database.DuckDB.Simple.Internal: ErrorTypeUnknown :: ErrorType
+ Database.DuckDB.Simple.Internal: FileFlagAppend :: FileFlag
+ Database.DuckDB.Simple.Internal: FileFlagCreate :: FileFlag
+ Database.DuckDB.Simple.Internal: FileFlagCreateNew :: FileFlag
+ Database.DuckDB.Simple.Internal: FileFlagExclusiveCreate :: FileFlag
+ Database.DuckDB.Simple.Internal: FileFlagExternalFileCache :: FileFlag
+ Database.DuckDB.Simple.Internal: FileFlagParallelAccess :: FileFlag
+ Database.DuckDB.Simple.Internal: FileFlagRead :: FileFlag
+ Database.DuckDB.Simple.Internal: FileFlagWrite :: FileFlag
+ Database.DuckDB.Simple.Internal: Instance :: IORef InstanceState -> Environment -> Instance
+ Database.DuckDB.Simple.Internal: InstanceClosed :: InstanceState
+ Database.DuckDB.Simple.Internal: InstanceOpen :: DuckDBV2InstanceHandle -> DuckDBV2FactoryHandle -> Children ConnectionState -> InstanceState
+ Database.DuckDB.Simple.Internal: LogLevelDebug :: LogLevel
+ Database.DuckDB.Simple.Internal: LogLevelError :: LogLevel
+ Database.DuckDB.Simple.Internal: LogLevelFatal :: LogLevel
+ Database.DuckDB.Simple.Internal: LogLevelInfo :: LogLevel
+ Database.DuckDB.Simple.Internal: LogLevelTrace :: LogLevel
+ Database.DuckDB.Simple.Internal: LogLevelWarning :: LogLevel
+ Database.DuckDB.Simple.Internal: LogScopeConnection :: LogScope
+ Database.DuckDB.Simple.Internal: LogScopeDatabase :: LogScope
+ Database.DuckDB.Simple.Internal: LogScopeThread :: LogScope
+ Database.DuckDB.Simple.Internal: LogicalTypeIdAny :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdArray :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdBigNum :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdBit :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdBlob :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdBoolean :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdDate :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdDecimal :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdDouble :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdEnum :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdFloat :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdGeometry :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdInteger :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdInterval :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdInvalid :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdList :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdMap :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdSQLNull :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdStruct :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTime :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTimeNs :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTimeTz :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTimestamp :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTimestampMs :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTimestampNs :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTimestampSec :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTimestampTz :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTimestampTzNs :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdTuple :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdType :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdUBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdUHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdUInteger :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdUSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdUTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdUUID :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdUnion :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdUnknown :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdVarchar :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: LogicalTypeIdVariant :: LogicalTypeId
+ Database.DuckDB.Simple.Internal: MacroFunctionTypeScalar :: MacroFunctionType
+ Database.DuckDB.Simple.Internal: MacroFunctionTypeTable :: MacroFunctionType
+ Database.DuckDB.Simple.Internal: SettingScopeDefault :: SettingScope
+ Database.DuckDB.Simple.Internal: SettingScopeGlobal :: SettingScope
+ Database.DuckDB.Simple.Internal: SettingScopeSession :: SettingScope
+ Database.DuckDB.Simple.Internal: TokenTypeComment :: TokenType
+ Database.DuckDB.Simple.Internal: TokenTypeEndOfInput :: TokenType
+ Database.DuckDB.Simple.Internal: TokenTypeIdentifier :: TokenType
+ Database.DuckDB.Simple.Internal: TokenTypeInvalid :: TokenType
+ Database.DuckDB.Simple.Internal: TokenTypeKeyword :: TokenType
+ Database.DuckDB.Simple.Internal: TokenTypeNumberLiteral :: TokenType
+ Database.DuckDB.Simple.Internal: TokenTypeOperator :: TokenType
+ Database.DuckDB.Simple.Internal: TokenTypeStringLiteral :: TokenType
+ Database.DuckDB.Simple.Internal: TokenTypeTerminator :: TokenType
+ Database.DuckDB.Simple.Internal: [buffers] :: ArrowArray -> Ptr (PtrConst Void)
+ Database.DuckDB.Simple.Internal: [children] :: ArrowSchema -> Ptr (Ptr ArrowSchema)
+ Database.DuckDB.Simple.Internal: [connectionAggregateStripes] :: ConnectionOptions -> Int
+ Database.DuckDB.Simple.Internal: [connectionArrowBatchSize] :: ConnectionOptions -> Int
+ Database.DuckDB.Simple.Internal: [connectionCallbackLockWait] :: ConnectionOptions -> Maybe Int
+ Database.DuckDB.Simple.Internal: [connectionConfig] :: ConnectionOptions -> [(Text, Text)]
+ Database.DuckDB.Simple.Internal: [connectionDecoder] :: ConnectionState -> MVar (Maybe VariantState)
+ Database.DuckDB.Simple.Internal: [connectionInstance] :: Connection -> Instance
+ Database.DuckDB.Simple.Internal: [connectionLock] :: Connection -> MVar ()
+ Database.DuckDB.Simple.Internal: [connectionOptions] :: Connection -> ConnectionOptions
+ Database.DuckDB.Simple.Internal: [connectionReleaseParent] :: ConnectionState -> IO ()
+ Database.DuckDB.Simple.Internal: [connectionStatements] :: ConnectionState -> Children StatementState
+ Database.DuckDB.Simple.Internal: [connectionWindow] :: Connection -> IORef (Maybe ThreadId)
+ Database.DuckDB.Simple.Internal: [context] :: DuckDBV2ExtensionInput -> DuckDBV2ContextHandle
+ Database.DuckDB.Simple.Internal: [count] :: DuckDBV2VectorView -> DuckDBIdx
+ Database.DuckDB.Simple.Internal: [data'] :: DuckDBV2VectorView -> PtrConst Void
+ Database.DuckDB.Simple.Internal: [days] :: DuckDBV2IntervalT -> Int32
+ Database.DuckDB.Simple.Internal: [destroy] :: DuckDBV2Opaque -> DuckDBV2OpaqueDestroyFn
+ Database.DuckDB.Simple.Internal: [dictionary] :: ArrowSchema -> Ptr ArrowSchema
+ Database.DuckDB.Simple.Internal: [environmentState] :: Environment -> IORef EnvironmentState
+ Database.DuckDB.Simple.Internal: [equals] :: DuckDBV2Opaque -> DuckDBV2OpaqueEqualsFn
+ Database.DuckDB.Simple.Internal: [err] :: DuckDBV2ExtensionInput -> Ptr DuckDBV2ErrorInfoHandle
+ Database.DuckDB.Simple.Internal: [extension] :: DuckDBV2ExtensionInput -> DuckDBV2ExtensionHandle
+ Database.DuckDB.Simple.Internal: [flags] :: ArrowSchema -> Int64
+ Database.DuckDB.Simple.Internal: [format] :: ArrowSchema -> PtrConst CChar
+ Database.DuckDB.Simple.Internal: [get_api] :: DuckDBV2ExtensionInput -> DuckDBV2ExtensionGetApiFn
+ Database.DuckDB.Simple.Internal: [get_last_error] :: ArrowArrayStream -> FunPtr (Ptr ArrowArrayStream -> IO (PtrConst CChar))
+ Database.DuckDB.Simple.Internal: [get_next] :: ArrowArrayStream -> FunPtr (Ptr ArrowArrayStream -> Ptr ArrowArray -> IO CInt)
+ Database.DuckDB.Simple.Internal: [get_schema] :: ArrowArrayStream -> FunPtr (Ptr ArrowArrayStream -> Ptr ArrowSchema -> IO CInt)
+ Database.DuckDB.Simple.Internal: [inlined] :: Duckdb_v2_bytes_value_inlined -> ConstantArray 12 CChar
+ Database.DuckDB.Simple.Internal: [instanceEnvironment] :: Instance -> Environment
+ Database.DuckDB.Simple.Internal: [instanceState] :: Instance -> IORef InstanceState
+ Database.DuckDB.Simple.Internal: [len] :: DuckDBV2Str -> DuckDBIdx
+ Database.DuckDB.Simple.Internal: [length] :: Duckdb_v2_bytes_value_pointer -> Word32
+ Database.DuckDB.Simple.Internal: [lower] :: DuckDBV2UhugeintT -> Word64
+ Database.DuckDB.Simple.Internal: [metadata] :: ArrowSchema -> PtrConst CChar
+ Database.DuckDB.Simple.Internal: [micros] :: DuckDBV2IntervalT -> Int64
+ Database.DuckDB.Simple.Internal: [months] :: DuckDBV2IntervalT -> Int32
+ Database.DuckDB.Simple.Internal: [n_buffers] :: ArrowArray -> Int64
+ Database.DuckDB.Simple.Internal: [n_children] :: ArrowSchema -> Int64
+ Database.DuckDB.Simple.Internal: [name] :: ArrowSchema -> PtrConst CChar
+ Database.DuckDB.Simple.Internal: [null_count] :: ArrowArray -> Int64
+ Database.DuckDB.Simple.Internal: [offset] :: DuckDBV2ListEntry -> DuckDBIdx
+ Database.DuckDB.Simple.Internal: [prefix] :: Duckdb_v2_bytes_value_pointer -> ConstantArray 4 CChar
+ Database.DuckDB.Simple.Internal: [private_data] :: ArrowSchema -> Ptr Void
+ Database.DuckDB.Simple.Internal: [ptr] :: Duckdb_v2_bytes_value_pointer -> Ptr CChar
+ Database.DuckDB.Simple.Internal: [release] :: ArrowSchema -> FunPtr (Ptr ArrowSchema -> IO ())
+ Database.DuckDB.Simple.Internal: [sel] :: DuckDBV2VectorView -> PtrConst DuckDBV2SelT
+ Database.DuckDB.Simple.Internal: [sqlErrorBody] :: SQLError -> Maybe Text
+ Database.DuckDB.Simple.Internal: [statementArrowResult] :: Statement -> IORef DuckDBV2ArrowResultHandle
+ Database.DuckDB.Simple.Internal: [statementBindings] :: Statement -> IORef (IntMap DuckDBV2ValueHandle)
+ Database.DuckDB.Simple.Internal: [statementParameterNames] :: Statement -> Array Int Text
+ Database.DuckDB.Simple.Internal: [statementParameterTypes] :: Statement -> IORef (Maybe (IntMap (Maybe LogicalTypeRep)))
+ Database.DuckDB.Simple.Internal: [statementParsed] :: StatementState -> DuckDBV2SqlStatementHandle
+ Database.DuckDB.Simple.Internal: [statementSchemaCache] :: StatementState -> IORef (Maybe (DuckDBV2SchemaHandle, DuckDBV2SchemaHandle))
+ Database.DuckDB.Simple.Internal: [statementStreamColumnLabels] :: StatementStreamColumn -> IORef (IntMap Text)
+ Database.DuckDB.Simple.Internal: [statementStreamColumnTree] :: StatementStreamColumn -> IORef (Maybe LogicalTypeRep)
+ Database.DuckDB.Simple.Internal: [statementStreamPendingChunks] :: StatementStream -> [DuckDBV2DataChunkHandle]
+ Database.DuckDB.Simple.Internal: [statementStreamTypes] :: StatementStream -> TypeScope
+ Database.DuckDB.Simple.Internal: [unwrap] :: Duckdb_v2_bytes_value -> ByteArray
+ Database.DuckDB.Simple.Internal: [upper] :: DuckDBV2UhugeintT -> Word64
+ Database.DuckDB.Simple.Internal: [validity] :: DuckDBV2VectorView -> PtrConst Word64
+ Database.DuckDB.Simple.Internal: [value] :: DuckDBV2Bytes -> Duckdb_v2_bytes_value
+ Database.DuckDB.Simple.Internal: aRROW_FLAG_DICTIONARY_ORDERED :: CInt
+ Database.DuckDB.Simple.Internal: aRROW_FLAG_MAP_KEYS_SORTED :: CInt
+ Database.DuckDB.Simple.Internal: aRROW_FLAG_NULLABLE :: CInt
+ Database.DuckDB.Simple.Internal: acquireItems :: (a -> IO ()) -> [IO a] -> IO [a]
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_bind_set_return_type :: DuckDBV2AggregateFunctionBindInfoHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_combine_get_bind_data :: DuckDBV2AggregateFunctionCombineInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_combine_get_sources :: DuckDBV2AggregateFunctionCombineInfoHandle -> Ptr (Ptr (Ptr Void)) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_combine_get_state_count :: DuckDBV2AggregateFunctionCombineInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_combine_get_targets :: DuckDBV2AggregateFunctionCombineInfoHandle -> Ptr (Ptr (Ptr Void)) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_combine_get_user_data :: DuckDBV2AggregateFunctionCombineInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2AggregateFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_destroy :: Ptr DuckDBV2AggregateFunctionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_destroy_get_bind_data :: DuckDBV2AggregateFunctionDestroyInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_destroy_get_state_count :: DuckDBV2AggregateFunctionDestroyInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_destroy_get_states :: DuckDBV2AggregateFunctionDestroyInfoHandle -> Ptr (Ptr (Ptr Void)) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_destroy_get_user_data :: DuckDBV2AggregateFunctionDestroyInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_finalize_get_bind_data :: DuckDBV2AggregateFunctionFinalizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_finalize_get_result :: DuckDBV2AggregateFunctionFinalizeInfoHandle -> Ptr DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_finalize_get_result_offset :: DuckDBV2AggregateFunctionFinalizeInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_finalize_get_state_count :: DuckDBV2AggregateFunctionFinalizeInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_finalize_get_states :: DuckDBV2AggregateFunctionFinalizeInfoHandle -> Ptr (Ptr (Ptr Void)) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_finalize_get_user_data :: DuckDBV2AggregateFunctionFinalizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_get_docs :: DuckDBV2AggregateFunctionHandle -> Ptr DuckDBV2FunctionDocsHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_get_signature :: DuckDBV2AggregateFunctionHandle -> Ptr DuckDBV2FunctionSignatureHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_init_get_bind_data :: DuckDBV2AggregateFunctionInitInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_init_get_state_count :: DuckDBV2AggregateFunctionInitInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_init_get_states :: DuckDBV2AggregateFunctionInitInfoHandle -> Ptr (Ptr (Ptr Void)) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_init_get_user_data :: DuckDBV2AggregateFunctionInitInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_bind_callback :: DuckDBV2AggregateFunctionHandle -> DuckDBV2AggregateFunctionBindCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_combine_callback :: DuckDBV2AggregateFunctionHandle -> DuckDBV2AggregateFunctionCombineCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_destroy_callback :: DuckDBV2AggregateFunctionHandle -> DuckDBV2AggregateFunctionDestroyCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_finalize_callback :: DuckDBV2AggregateFunctionHandle -> DuckDBV2AggregateFunctionFinalizeCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_init_callback :: DuckDBV2AggregateFunctionHandle -> DuckDBV2AggregateFunctionInitCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_name :: DuckDBV2AggregateFunctionHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_property :: DuckDBV2AggregateFunctionHandle -> DuckDBV2FunctionPropertyKey -> DuckDBV2FunctionPropertyValue -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_size_callback :: DuckDBV2AggregateFunctionHandle -> DuckDBV2AggregateFunctionSizeCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_update_callback :: DuckDBV2AggregateFunctionHandle -> DuckDBV2AggregateFunctionUpdateCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_set_user_data :: DuckDBV2AggregateFunctionHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_size_get_bind_data :: DuckDBV2AggregateFunctionSizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_size_get_user_data :: DuckDBV2AggregateFunctionSizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_size_set_state_size :: DuckDBV2AggregateFunctionSizeInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_update_get_arg :: DuckDBV2AggregateFunctionUpdateInfoHandle -> DuckDBIdx -> Ptr DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_update_get_arg_count :: DuckDBV2AggregateFunctionUpdateInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_update_get_bind_data :: DuckDBV2AggregateFunctionUpdateInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_update_get_row_count :: DuckDBV2AggregateFunctionUpdateInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_update_get_states :: DuckDBV2AggregateFunctionUpdateInfoHandle -> Ptr (Ptr (Ptr Void)) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_aggregate_function_update_get_user_data :: DuckDBV2AggregateFunctionUpdateInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arena_allocate :: DuckDBV2ArenaHandle -> DuckDBIdx -> Ptr (Ptr Word8) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_exporter_append :: DuckDBV2ArrowExporterHandle -> Ptr DuckDBV2DataChunkHandle -> CBool -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_exporter_create :: DuckDBV2ContextHandle -> PtrConst DuckDBV2LogicalTypeHandle -> PtrConst DuckDBV2Str -> DuckDBIdx -> DuckDBIdx -> Ptr DuckDBV2ArrowExporterHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_exporter_destroy :: Ptr DuckDBV2ArrowExporterHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_exporter_get_schema :: DuckDBV2ArrowExporterHandle -> Ptr ArrowSchema -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_exporter_next_array :: DuckDBV2ArrowExporterHandle -> Ptr ArrowArray -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_importer_append :: DuckDBV2ArrowImporterHandle -> Ptr ArrowArray -> CBool -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_importer_create :: DuckDBV2ContextHandle -> Ptr ArrowSchema -> DuckDBIdx -> Ptr DuckDBV2ArrowImporterHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_importer_destroy :: Ptr DuckDBV2ArrowImporterHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_importer_get_schema :: DuckDBV2ArrowImporterHandle -> Ptr DuckDBV2SchemaHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_importer_next_chunk :: DuckDBV2ArrowImporterHandle -> Ptr DuckDBV2DataChunkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_destroy :: Ptr DuckDBV2ArrowResultHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_drain :: DuckDBV2ArrowResultHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_fetch_array :: DuckDBV2ArrowResultHandle -> Ptr ArrowArray -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_get_result_type :: DuckDBV2ArrowResultHandle -> Ptr DuckDBV2ResultType -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_get_schema :: DuckDBV2ArrowResultHandle -> Ptr ArrowSchema -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_get_statement_type :: DuckDBV2ArrowResultHandle -> Ptr DuckDBV2StatementType -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_step :: DuckDBV2ArrowResultHandle -> Ptr ArrowArray -> Ptr DuckDBV2ResultStepStatus -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_to_arrow_c_stream :: Ptr DuckDBV2ArrowResultHandle -> Ptr ArrowArrayStream -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_arrow_result_wait :: DuckDBV2ArrowResultHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_attach_options_create :: DuckDBV2InstanceHandle -> Ptr DuckDBV2AttachOptionsHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_attach_options_destroy :: Ptr DuckDBV2AttachOptionsHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_attach_options_set :: DuckDBV2AttachOptionsHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_bignum_decode :: PtrConst Word8 -> DuckDBIdx -> Ptr Word8 -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_bignum_encode :: PtrConst Word8 -> DuckDBIdx -> CBool -> Ptr Word8 -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2CastFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_destroy :: Ptr DuckDBV2CastFunctionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_exec_get_input :: DuckDBV2CastFunctionExecInfoHandle -> Ptr DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_exec_get_mode :: DuckDBV2CastFunctionExecInfoHandle -> Ptr DuckDBV2CastMode -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_exec_get_output :: DuckDBV2CastFunctionExecInfoHandle -> Ptr DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_exec_get_row_count :: DuckDBV2CastFunctionExecInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_exec_get_user_data :: DuckDBV2CastFunctionExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_set_exec_callback :: DuckDBV2CastFunctionHandle -> DuckDBV2CastFunctionExecCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_set_implicit_cast_cost :: DuckDBV2CastFunctionHandle -> Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_set_source_type :: DuckDBV2CastFunctionHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_set_target_type :: DuckDBV2CastFunctionHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_cast_function_set_user_data :: DuckDBV2CastFunctionHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_append :: DuckDBV2ColumnDataCollectionHandle -> DuckDBV2ColumnDataCollectionAppendStateHandle -> DuckDBV2DataChunkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_append_state_create :: DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ColumnDataCollectionAppendStateHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_append_state_destroy :: Ptr DuckDBV2ColumnDataCollectionAppendStateHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_clear :: DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_combine :: DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_create :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2LogicalTypeHandle -> DuckDBIdx -> Ptr DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_destroy :: Ptr DuckDBV2ColumnDataCollectionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_reset :: DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_row_count :: DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_scan :: DuckDBV2ColumnDataCollectionHandle -> DuckDBV2ColumnDataCollectionSharedScanStateHandle -> DuckDBV2ColumnDataCollectionWorkerScanStateHandle -> DuckDBV2DataChunkHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_shared_scan_state_create :: DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ColumnDataCollectionSharedScanStateHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_shared_scan_state_destroy :: Ptr DuckDBV2ColumnDataCollectionSharedScanStateHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_worker_scan_state_create :: DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ColumnDataCollectionWorkerScanStateHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_data_collection_worker_scan_state_destroy :: Ptr DuckDBV2ColumnDataCollectionWorkerScanStateHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_description_destroy :: Ptr DuckDBV2ColumnDescriptionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_description_get_name :: DuckDBV2ColumnDescriptionHandle -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_description_get_type :: DuckDBV2ColumnDescriptionHandle -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_description_has_default :: DuckDBV2ColumnDescriptionHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_column_description_has_generated :: DuckDBV2ColumnDescriptionHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_config_get_option_by_index :: DuckDBV2ConfigHandle -> DuckDBIdx -> Ptr DuckDBV2OptionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_config_get_option_by_name :: DuckDBV2ConfigHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2OptionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_config_get_option_count :: DuckDBV2ConfigHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_config_get_option_value :: DuckDBV2ConfigHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2SettingScope -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_config_set_option_text :: DuckDBV2ConfigHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2Str -> DuckDBV2SettingScope -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_config_set_option_value :: DuckDBV2ConfigHandle -> PtrConst DuckDBV2IdentifierT -> DuckDBV2ValueHandle -> DuckDBV2SettingScope -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_create :: DuckDBV2InstanceHandle -> Ptr DuckDBV2ConnectionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_describe_table :: DuckDBV2ConnectionHandle -> DuckDBV2QnameHandle -> Ptr DuckDBV2TableDescriptionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_destroy :: Ptr DuckDBV2ConnectionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_enable_profiling :: DuckDBV2ConnectionHandle -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_get_config :: DuckDBV2ConnectionHandle -> Ptr DuckDBV2ConfigHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_get_factory :: DuckDBV2ConnectionHandle -> Ptr DuckDBV2FactoryHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_get_file_system :: DuckDBV2ConnectionHandle -> Ptr DuckDBV2FileSystemHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_get_query_metrics :: DuckDBV2ConnectionHandle -> Ptr DuckDBV2QueryMetricsHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_interrupt :: DuckDBV2ConnectionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_log :: DuckDBV2ConnectionHandle -> DuckDBV2LogLevel -> PtrConst DuckDBV2Str -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_progress_get :: DuckDBV2ConnectionHandle -> Ptr CDouble -> Ptr Word64 -> Ptr Word64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_aggregate_function :: DuckDBV2ConnectionHandle -> DuckDBV2AggregateFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_cast_function :: DuckDBV2ConnectionHandle -> DuckDBV2CastFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_copy_function :: DuckDBV2ConnectionHandle -> DuckDBV2CopyFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_custom_option :: DuckDBV2ConnectionHandle -> DuckDBV2CustomOptionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_custom_type :: DuckDBV2ConnectionHandle -> DuckDBV2CustomTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_multi_file_function :: DuckDBV2ConnectionHandle -> DuckDBV2MultiFileFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_pragma_function :: DuckDBV2ConnectionHandle -> DuckDBV2PragmaFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_replacement_scan :: DuckDBV2ConnectionHandle -> DuckDBV2ReplacementScanHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_scalar_function :: DuckDBV2ConnectionHandle -> DuckDBV2ScalarFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_connection_register_table_function :: DuckDBV2ConnectionHandle -> DuckDBV2TableFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_context_get_config :: DuckDBV2ContextHandle -> Ptr DuckDBV2ConfigHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_context_get_factory :: DuckDBV2ContextHandle -> Ptr DuckDBV2FactoryHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_context_get_file_system :: DuckDBV2ContextHandle -> Ptr DuckDBV2FileSystemHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_context_log :: DuckDBV2ContextHandle -> DuckDBV2LogLevel -> PtrConst DuckDBV2Str -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_get_column_count :: DuckDBV2CopyFromBindInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_get_column_name :: DuckDBV2CopyFromBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_get_column_type :: DuckDBV2CopyFromBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_get_file_path :: DuckDBV2CopyFromBindInfoHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_get_option_count :: DuckDBV2CopyFromBindInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_get_option_name :: DuckDBV2CopyFromBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_get_option_value :: DuckDBV2CopyFromBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_get_user_data :: DuckDBV2CopyFromBindInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_set_bind_data :: DuckDBV2CopyFromBindInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_bind_set_cardinality :: DuckDBV2CopyFromBindInfoHandle -> DuckDBIdx -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_exec_get_bind_data :: DuckDBV2CopyFromExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_exec_get_global_state :: DuckDBV2CopyFromExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_exec_get_local_state :: DuckDBV2CopyFromExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_exec_get_output_chunk :: DuckDBV2CopyFromExecInfoHandle -> Ptr DuckDBV2DataChunkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_exec_get_user_data :: DuckDBV2CopyFromExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_init_global_get_bind_data :: DuckDBV2CopyFromInitGlobalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_init_global_get_user_data :: DuckDBV2CopyFromInitGlobalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_init_global_set_global_state :: DuckDBV2CopyFromInitGlobalInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_init_global_set_max_threads :: DuckDBV2CopyFromInitGlobalInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_init_local_get_bind_data :: DuckDBV2CopyFromInitLocalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_init_local_get_global_state :: DuckDBV2CopyFromInitLocalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_init_local_get_user_data :: DuckDBV2CopyFromInitLocalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_init_local_set_local_state :: DuckDBV2CopyFromInitLocalInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_progress_get_bind_data :: DuckDBV2CopyFromProgressInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_progress_get_global_state :: DuckDBV2CopyFromProgressInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_progress_get_user_data :: DuckDBV2CopyFromProgressInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_progress_set_progress :: DuckDBV2CopyFromProgressInfoHandle -> CDouble -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_set_bind_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyFromBindCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_set_exec_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyFromExecCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_set_init_global_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyFromInitGlobalCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_set_init_local_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyFromInitLocalCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_from_set_progress_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyFromProgressCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_function_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2CopyFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_function_destroy :: Ptr DuckDBV2CopyFunctionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_function_set_file_extension :: DuckDBV2CopyFunctionHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_function_set_name :: DuckDBV2CopyFunctionHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_function_set_user_data :: DuckDBV2CopyFunctionHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_batch_get_bind_data :: DuckDBV2CopyToBatchInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_batch_get_init_data :: DuckDBV2CopyToBatchInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_batch_get_user_data :: DuckDBV2CopyToBatchInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_batch_set_batch_data :: DuckDBV2CopyToBatchInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_batch_size_get_bind_data :: DuckDBV2CopyToBatchSizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_batch_size_get_user_data :: DuckDBV2CopyToBatchSizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_batch_size_set_target :: DuckDBV2CopyToBatchSizeInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_batch_take_input :: DuckDBV2CopyToBatchInfoHandle -> Ptr DuckDBV2ColumnDataCollectionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_get_column_count :: DuckDBV2CopyToBindInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_get_column_name :: DuckDBV2CopyToBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_get_column_type :: DuckDBV2CopyToBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_get_file_path :: DuckDBV2CopyToBindInfoHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_get_option_count :: DuckDBV2CopyToBindInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_get_option_name :: DuckDBV2CopyToBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_get_option_value :: DuckDBV2CopyToBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_get_user_data :: DuckDBV2CopyToBindInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_bind_set_bind_data :: DuckDBV2CopyToBindInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_finalize_get_bind_data :: DuckDBV2CopyToFinalizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_finalize_get_init_data :: DuckDBV2CopyToFinalizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_finalize_get_user_data :: DuckDBV2CopyToFinalizeInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_flush_get_batch_data :: DuckDBV2CopyToFlushInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_flush_get_bind_data :: DuckDBV2CopyToFlushInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_flush_get_init_data :: DuckDBV2CopyToFlushInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_flush_get_user_data :: DuckDBV2CopyToFlushInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_init_get_bind_data :: DuckDBV2CopyToInitInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_init_get_file_path :: DuckDBV2CopyToInitInfoHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_init_get_user_data :: DuckDBV2CopyToInitInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_init_set_init_data :: DuckDBV2CopyToInitInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_set_batch_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyToBatchCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_set_batch_size_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyToBatchSizeCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_set_bind_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyToBindCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_set_finalize_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyToFinalizeCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_set_flush_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyToFlushCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_set_init_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyToInitCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_set_statistics_callback :: DuckDBV2CopyFunctionHandle -> DuckDBV2CopyToStatisticsCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_statistics_get_bind_data :: DuckDBV2CopyToStatisticsInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_statistics_get_init_data :: DuckDBV2CopyToStatisticsInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_statistics_get_user_data :: DuckDBV2CopyToStatisticsInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_statistics_set_file_size :: DuckDBV2CopyToStatisticsInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_copy_to_statistics_set_row_count :: DuckDBV2CopyToStatisticsInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2CustomOptionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_destroy :: Ptr DuckDBV2CustomOptionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_set_default_scope :: DuckDBV2CustomOptionHandle -> DuckDBV2SettingScope -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_set_default_value :: DuckDBV2CustomOptionHandle -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_set_description :: DuckDBV2CustomOptionHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_set_name :: DuckDBV2CustomOptionHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_set_supports_scope :: DuckDBV2CustomOptionHandle -> DuckDBV2SettingScope -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_set_type :: DuckDBV2CustomOptionHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_set_update_callback :: DuckDBV2CustomOptionHandle -> DuckDBV2CustomOptionUpdateCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_set_user_data :: DuckDBV2CustomOptionHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_update_get_scope :: DuckDBV2CustomOptionUpdateInfoHandle -> Ptr DuckDBV2SettingScope -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_update_get_user_data :: DuckDBV2CustomOptionUpdateInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_update_get_value :: DuckDBV2CustomOptionUpdateInfoHandle -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_update_is_reset :: DuckDBV2CustomOptionUpdateInfoHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_option_update_set_value :: DuckDBV2CustomOptionUpdateInfoHandle -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_type_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2CustomTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_type_destroy :: Ptr DuckDBV2CustomTypeHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_type_set_base_type :: DuckDBV2CustomTypeHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_custom_type_set_name :: DuckDBV2CustomTypeHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_data_chunk_clear :: DuckDBV2DataChunkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_data_chunk_copy :: DuckDBV2FactoryHandle -> DuckDBV2DataChunkHandle -> Ptr DuckDBV2DataChunkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_data_chunk_create :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2LogicalTypeHandle -> DuckDBIdx -> Ptr DuckDBV2DataChunkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_data_chunk_destroy :: Ptr DuckDBV2DataChunkHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_data_chunk_get_capacity :: DuckDBV2DataChunkHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_data_chunk_get_size :: DuckDBV2DataChunkHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_data_chunk_get_vector :: DuckDBV2DataChunkHandle -> DuckDBIdx -> Ptr DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_data_chunk_get_vector_count :: DuckDBV2DataChunkHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_environment_create :: Ptr DuckDBV2EnvironmentHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_environment_destroy :: Ptr DuckDBV2EnvironmentHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_environment_get_instance_count :: DuckDBV2EnvironmentHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_error_info_destroy :: Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_error_info_get_code :: DuckDBV2ErrorInfoHandle -> Ptr DuckDBV2Error -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_error_info_get_raw_text :: DuckDBV2ErrorInfoHandle -> Ptr DuckDBV2Str -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_error_info_get_text :: DuckDBV2ErrorInfoHandle -> Ptr DuckDBV2Str -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_error_info_set_code :: DuckDBV2ErrorInfoHandle -> DuckDBV2Error -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_error_info_set_text :: DuckDBV2ErrorInfoHandle -> PtrConst DuckDBV2Str -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_cast_get_mode :: DuckDBV2ExpressionHandle -> Ptr DuckDBV2CastMode -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_column_ref_get_index :: DuckDBV2ExpressionHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_constant_get_value :: DuckDBV2ExpressionHandle -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_function_get_name :: DuckDBV2ExpressionHandle -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_function_get_qname :: DuckDBV2ExpressionHandle -> Ptr DuckDBV2QnameHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_get_child :: DuckDBV2ExpressionHandle -> DuckDBIdx -> Ptr DuckDBV2ExpressionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_get_child_count :: DuckDBV2ExpressionHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_get_return_type :: DuckDBV2ExpressionHandle -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_expression_get_type :: DuckDBV2ExpressionHandle -> Ptr DuckDBV2ExpressionType -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_aggregate_function :: DuckDBV2ExtensionHandle -> DuckDBV2AggregateFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_cast_function :: DuckDBV2ExtensionHandle -> DuckDBV2CastFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_copy_function :: DuckDBV2ExtensionHandle -> DuckDBV2CopyFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_custom_option :: DuckDBV2ExtensionHandle -> DuckDBV2CustomOptionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_custom_type :: DuckDBV2ExtensionHandle -> DuckDBV2CustomTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_log_sink :: DuckDBV2ExtensionHandle -> DuckDBV2LogSinkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_macro_function :: DuckDBV2ExtensionHandle -> DuckDBV2MacroFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_multi_file_function :: DuckDBV2ExtensionHandle -> DuckDBV2MultiFileFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_pragma_function :: DuckDBV2ExtensionHandle -> DuckDBV2PragmaFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_replacement_scan :: DuckDBV2ExtensionHandle -> DuckDBV2ReplacementScanHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_scalar_function :: DuckDBV2ExtensionHandle -> DuckDBV2ScalarFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_extension_register_table_function :: DuckDBV2ExtensionHandle -> DuckDBV2TableFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_factory_create_common_type :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2LogicalTypeHandle -> DuckDBIdx -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_factory_create_type_from_id :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeId -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_factory_create_type_from_name :: DuckDBV2FactoryHandle -> DuckDBV2QnameHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_factory_create_type_from_text :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_factory_create_type_with_alias :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_close :: DuckDBV2FileHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_destroy :: Ptr DuckDBV2FileHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_open_options_create :: DuckDBV2FileSystemHandle -> Ptr DuckDBV2FileOpenOptionsHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_open_options_destroy :: Ptr DuckDBV2FileOpenOptionsHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_open_options_set_flag :: DuckDBV2FileOpenOptionsHandle -> DuckDBV2FileFlag -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_open_options_set_value :: DuckDBV2FileOpenOptionsHandle -> PtrConst DuckDBV2Str -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_read :: DuckDBV2FileHandle -> Ptr Void -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_read_at :: DuckDBV2FileHandle -> Ptr Void -> DuckDBIdx -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_seek :: DuckDBV2FileHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_size :: DuckDBV2FileHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_sync :: DuckDBV2FileHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_system_open :: DuckDBV2FileSystemHandle -> PtrConst DuckDBV2Str -> DuckDBV2FileOpenOptionsHandle -> Ptr DuckDBV2FileHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_system_path_exists :: DuckDBV2FileSystemHandle -> PtrConst DuckDBV2Str -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_system_path_expand :: DuckDBV2FileSystemHandle -> PtrConst DuckDBV2Str -> DuckDBV2TextSinkFn -> Ptr Void -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_system_path_glob :: DuckDBV2FileSystemHandle -> PtrConst DuckDBV2Str -> DuckDBV2FileSystemPathGlobCallbackFn -> Ptr Void -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_system_path_is_remote :: DuckDBV2FileSystemHandle -> PtrConst DuckDBV2Str -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_system_path_move :: DuckDBV2FileSystemHandle -> PtrConst DuckDBV2Str -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_system_path_remove :: DuckDBV2FileSystemHandle -> PtrConst DuckDBV2Str -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_tell :: DuckDBV2FileHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_write :: DuckDBV2FileHandle -> PtrConst Void -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_file_write_at :: DuckDBV2FileHandle -> PtrConst Void -> DuckDBIdx -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_bind_get_arg_count :: DuckDBV2FunctionBindInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_bind_get_arg_index :: DuckDBV2FunctionBindInfoHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBIdx -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_bind_get_arg_name :: DuckDBV2FunctionBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_bind_get_arg_type :: DuckDBV2FunctionBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_bind_get_arg_value :: DuckDBV2FunctionBindInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_bind_get_user_data :: DuckDBV2FunctionBindInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_bind_set_bind_data :: DuckDBV2FunctionBindInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_docs_add_category :: DuckDBV2FunctionDocsHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_docs_add_example :: DuckDBV2FunctionDocsHandle -> PtrConst DuckDBV2Str -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_docs_set_description :: DuckDBV2FunctionDocsHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_signature_add_parameter :: DuckDBV2FunctionSignatureHandle -> PtrConst DuckDBV2IdentifierT -> DuckDBV2LogicalTypeHandle -> DuckDBV2ValueHandle -> DuckDBV2FunctionParameterKind -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_function_signature_set_return_type :: DuckDBV2FunctionSignatureHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_identifier_render_quoted :: PtrConst DuckDBV2IdentifierT -> Ptr CChar -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_attach :: DuckDBV2InstanceHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2Str -> DuckDBV2AttachOptionsHandle -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_create :: DuckDBV2EnvironmentHandle -> Ptr DuckDBV2InstanceHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_create_with_options :: DuckDBV2EnvironmentHandle -> PtrConst DuckDBV2Str -> PtrConst DuckDBV2Str -> DuckDBIdx -> Ptr DuckDBV2InstanceHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_destroy :: Ptr DuckDBV2InstanceHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_detach :: DuckDBV2InstanceHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_get_config :: DuckDBV2InstanceHandle -> Ptr DuckDBV2ConfigHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_get_factory :: DuckDBV2InstanceHandle -> Ptr DuckDBV2FactoryHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_get_file_system :: DuckDBV2InstanceHandle -> Ptr DuckDBV2FileSystemHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_log :: DuckDBV2InstanceHandle -> DuckDBV2LogLevel -> PtrConst DuckDBV2Str -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_register_log_sink :: DuckDBV2InstanceHandle -> DuckDBV2LogSinkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_register_replacement_scan :: DuckDBV2InstanceHandle -> DuckDBV2ReplacementScanHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_instance_set_default :: DuckDBV2InstanceHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_library_version :: Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2LogSinkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_destroy :: Ptr DuckDBV2LogSinkHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_set_flush_callback :: DuckDBV2LogSinkHandle -> DuckDBV2LogSinkFlushCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_set_name :: DuckDBV2LogSinkHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_set_user_data :: DuckDBV2LogSinkHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_set_write_callback :: DuckDBV2LogSinkHandle -> DuckDBV2LogSinkWriteCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_write_info_get_connection_id :: DuckDBV2LogSinkWriteInfoHandle -> Ptr DuckDBIdx -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_write_info_get_query_id :: DuckDBV2LogSinkWriteInfoHandle -> Ptr DuckDBIdx -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_write_info_get_thread_id :: DuckDBV2LogSinkWriteInfoHandle -> Ptr DuckDBIdx -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_log_sink_write_info_get_transaction_id :: DuckDBV2LogSinkWriteInfoHandle -> Ptr DuckDBIdx -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_logical_type_copy :: DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_logical_type_destroy :: Ptr DuckDBV2LogicalTypeHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_logical_type_get_id :: DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2LogicalTypeId -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_logical_type_get_name :: DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_logical_type_get_param :: DuckDBV2LogicalTypeHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_logical_type_get_param_count :: DuckDBV2LogicalTypeHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_logical_type_is_equal :: DuckDBV2LogicalTypeHandle -> DuckDBV2LogicalTypeHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_logical_type_to_text :: DuckDBV2LogicalTypeHandle -> Ptr CChar -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_macro_function_create :: DuckDBV2FactoryHandle -> DuckDBV2MacroFunctionType -> Ptr DuckDBV2MacroFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_macro_function_destroy :: Ptr DuckDBV2MacroFunctionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_macro_function_get_docs :: DuckDBV2MacroFunctionHandle -> Ptr DuckDBV2FunctionDocsHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_macro_function_get_signature :: DuckDBV2MacroFunctionHandle -> Ptr DuckDBV2FunctionSignatureHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_macro_function_set_body :: DuckDBV2MacroFunctionHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_macro_function_set_name :: DuckDBV2MacroFunctionHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_multi_file_function_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2MultiFileFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_multi_file_function_destroy :: Ptr DuckDBV2MultiFileFunctionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_multi_file_function_set_file_extension :: DuckDBV2MultiFileFunctionHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_multi_file_function_set_name :: DuckDBV2MultiFileFunctionHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_multi_file_function_set_reader_type :: DuckDBV2MultiFileFunctionHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_multi_file_function_set_single_file_function :: DuckDBV2MultiFileFunctionHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_option_destroy :: Ptr DuckDBV2OptionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_option_get_alias :: DuckDBV2OptionHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_option_get_alias_count :: DuckDBV2OptionHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_option_get_default_scope :: DuckDBV2OptionHandle -> Ptr DuckDBV2SettingScope -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_option_get_default_value :: DuckDBV2OptionHandle -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_option_get_description :: DuckDBV2OptionHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_option_get_name :: DuckDBV2OptionHandle -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_option_supports_scope :: DuckDBV2OptionHandle -> DuckDBV2SettingScope -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_parse_sql :: DuckDBV2ConnectionHandle -> PtrConst CChar -> Ptr DuckDBV2StatementIteratorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2PragmaFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_destroy :: Ptr DuckDBV2PragmaFunctionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_get_docs :: DuckDBV2PragmaFunctionHandle -> Ptr DuckDBV2FunctionDocsHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_get_signature :: DuckDBV2PragmaFunctionHandle -> Ptr DuckDBV2FunctionSignatureHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_query_set_query :: DuckDBV2PragmaFunctionQueryInfoHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_set_exec_callback :: DuckDBV2PragmaFunctionHandle -> DuckDBV2PragmaFunctionExecCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_set_name :: DuckDBV2PragmaFunctionHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_set_query_callback :: DuckDBV2PragmaFunctionHandle -> DuckDBV2PragmaFunctionQueryCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_pragma_function_set_user_data :: DuckDBV2PragmaFunctionHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_prepared_statement_create :: DuckDBV2ConnectionHandle -> DuckDBV2SqlStatementHandle -> CBool -> Ptr DuckDBV2PreparedStatementHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_prepared_statement_destroy :: Ptr DuckDBV2PreparedStatementHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_prepared_statement_execute :: DuckDBV2PreparedStatementHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ResultHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_prepared_statement_execute_arrow :: DuckDBV2PreparedStatementHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> DuckDBIdx -> Ptr DuckDBV2ArrowResultHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_prepared_statement_reuses_plan :: DuckDBV2PreparedStatementHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_qname_create :: PtrConst DuckDBV2IdentifierT -> DuckDBIdx -> Ptr DuckDBV2QnameHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_qname_destroy :: Ptr DuckDBV2QnameHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_qname_equals :: DuckDBV2QnameHandle -> DuckDBV2QnameHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_qname_get_part :: DuckDBV2QnameHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_qname_get_part_count :: DuckDBV2QnameHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_qname_hash :: DuckDBV2QnameHandle -> Ptr Word64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_qname_parse :: PtrConst DuckDBV2Str -> Ptr DuckDBV2QnameHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_qname_render :: DuckDBV2QnameHandle -> Ptr CChar -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_destroy :: Ptr DuckDBV2QueryMetricsHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_get_root :: DuckDBV2QueryMetricsHandle -> Ptr DuckDBV2QueryMetricsNodeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_node_get_child :: DuckDBV2QueryMetricsNodeHandle -> DuckDBIdx -> Ptr DuckDBV2QueryMetricsNodeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_node_get_child_count :: DuckDBV2QueryMetricsNodeHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_node_get_metric :: DuckDBV2QueryMetricsNodeHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ValueHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_node_get_metric_count :: DuckDBV2QueryMetricsNodeHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_node_get_metric_name :: DuckDBV2QueryMetricsNodeHandle -> DuckDBIdx -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_node_get_metric_value :: DuckDBV2QueryMetricsNodeHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_query_metrics_node_get_name :: DuckDBV2QueryMetricsNodeHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_add_argument :: DuckDBV2ReplacementScanInfoHandle -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_add_named_argument :: DuckDBV2ReplacementScanInfoHandle -> PtrConst DuckDBV2IdentifierT -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2ReplacementScanHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_destroy :: Ptr DuckDBV2ReplacementScanHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_get_name :: DuckDBV2ReplacementScanInfoHandle -> Ptr DuckDBV2QnameHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_get_user_data :: DuckDBV2ReplacementScanInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_set_alias :: DuckDBV2ReplacementScanInfoHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_set_callback :: DuckDBV2ReplacementScanHandle -> DuckDBV2ReplacementScanCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_set_collection :: DuckDBV2ReplacementScanInfoHandle -> DuckDBV2ColumnDataCollectionHandle -> PtrConst DuckDBV2IdentifierT -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_set_function_name :: DuckDBV2ReplacementScanInfoHandle -> DuckDBV2QnameHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_set_subquery :: DuckDBV2ReplacementScanInfoHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_replacement_scan_set_user_data :: DuckDBV2ReplacementScanHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_destroy :: Ptr DuckDBV2ResultHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_drain :: DuckDBV2ResultHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_fetch_chunk :: DuckDBV2ResultHandle -> Ptr DuckDBV2DataChunkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_get_result_type :: DuckDBV2ResultHandle -> Ptr DuckDBV2ResultType -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_get_schema :: DuckDBV2ResultHandle -> Ptr DuckDBV2SchemaHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_get_statement_type :: DuckDBV2ResultHandle -> Ptr DuckDBV2StatementType -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_render_box :: Ptr DuckDBV2ResultHandle -> DuckDBIdx -> DuckDBIdx -> DuckDBIdx -> PtrConst DuckDBV2Str -> DuckDBIdx -> DuckDBIdx -> DuckDBV2TextSinkFn -> Ptr Void -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_step :: DuckDBV2ResultHandle -> Ptr DuckDBV2DataChunkHandle -> Ptr DuckDBV2ResultStepStatus -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_result_wait :: DuckDBV2ResultHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_bind_set_bind_data :: DuckDBV2ScalarFunctionBindInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2ScalarFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_destroy :: Ptr DuckDBV2ScalarFunctionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_exec_get_arg :: DuckDBV2ScalarFunctionExecInfoHandle -> DuckDBIdx -> Ptr DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_exec_get_arg_count :: DuckDBV2ScalarFunctionExecInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_exec_get_bind_data :: DuckDBV2ScalarFunctionExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_exec_get_init_data :: DuckDBV2ScalarFunctionExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_exec_get_result :: DuckDBV2ScalarFunctionExecInfoHandle -> Ptr DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_exec_get_row_count :: DuckDBV2ScalarFunctionExecInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_exec_get_user_data :: DuckDBV2ScalarFunctionExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_get_docs :: DuckDBV2ScalarFunctionHandle -> Ptr DuckDBV2FunctionDocsHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_get_signature :: DuckDBV2ScalarFunctionHandle -> Ptr DuckDBV2FunctionSignatureHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_init_get_bind_data :: DuckDBV2ScalarFunctionInitInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_init_get_user_data :: DuckDBV2ScalarFunctionInitInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_init_set_init_data :: DuckDBV2ScalarFunctionInitInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_resolve_types_set_return_type :: DuckDBV2ScalarFunctionResolveTypesInfoHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_set_bind_callback :: DuckDBV2ScalarFunctionHandle -> DuckDBV2ScalarFunctionBindCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_set_exec_callback :: DuckDBV2ScalarFunctionHandle -> DuckDBV2ScalarFunctionExecCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_set_init_callback :: DuckDBV2ScalarFunctionHandle -> DuckDBV2ScalarFunctionInitCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_set_name :: DuckDBV2ScalarFunctionHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_set_property :: DuckDBV2ScalarFunctionHandle -> DuckDBV2FunctionPropertyKey -> DuckDBV2FunctionPropertyValue -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_set_resolve_types_callback :: DuckDBV2ScalarFunctionHandle -> DuckDBV2ScalarFunctionResolveTypesCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_scalar_function_set_user_data :: DuckDBV2ScalarFunctionHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_schema_destroy :: Ptr DuckDBV2SchemaHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_schema_get_count :: DuckDBV2SchemaHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_schema_get_field :: DuckDBV2SchemaHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_sql_statement_destroy :: Ptr DuckDBV2SqlStatementHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_sql_statement_get_parameter_count :: DuckDBV2SqlStatementHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_sql_statement_get_parameter_name :: DuckDBV2SqlStatementHandle -> DuckDBIdx -> Ptr DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_sql_statement_get_text :: DuckDBV2SqlStatementHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_sql_statement_get_type :: DuckDBV2SqlStatementHandle -> Ptr DuckDBV2StatementType -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_statement_add_collection :: DuckDBV2SqlStatementHandle -> PtrConst DuckDBV2IdentifierT -> DuckDBV2ColumnDataCollectionHandle -> PtrConst DuckDBV2IdentifierT -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_statement_bind :: DuckDBV2ConnectionHandle -> DuckDBV2SqlStatementHandle -> Ptr DuckDBV2SchemaHandle -> Ptr DuckDBV2SchemaHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_statement_execute :: DuckDBV2ConnectionHandle -> DuckDBV2SqlStatementHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ResultHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_statement_execute_arrow :: DuckDBV2ConnectionHandle -> DuckDBV2SqlStatementHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> DuckDBIdx -> Ptr DuckDBV2ArrowResultHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_statement_iterator_destroy :: Ptr DuckDBV2StatementIteratorHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_statement_iterator_next :: DuckDBV2StatementIteratorHandle -> Ptr DuckDBV2SqlStatementHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_description_destroy :: Ptr DuckDBV2TableDescriptionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_description_get_column :: DuckDBV2TableDescriptionHandle -> DuckDBIdx -> Ptr DuckDBV2ColumnDescriptionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_description_get_column_count :: DuckDBV2TableDescriptionHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_description_get_qname :: DuckDBV2TableDescriptionHandle -> Ptr DuckDBV2QnameHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_description_is_readonly :: DuckDBV2TableDescriptionHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_bind_add_result_column :: DuckDBV2TableFunctionBindInfoHandle -> PtrConst DuckDBV2IdentifierT -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_bind_set_cardinality :: DuckDBV2TableFunctionBindInfoHandle -> DuckDBIdx -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_bind_set_order_preservation :: DuckDBV2TableFunctionBindInfoHandle -> DuckDBV2OrderPreservation -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_claim_batch_get_bind_data :: DuckDBV2TableFunctionClaimBatchInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_claim_batch_get_global_state :: DuckDBV2TableFunctionClaimBatchInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_claim_batch_get_local_state :: DuckDBV2TableFunctionClaimBatchInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_claim_batch_get_user_data :: DuckDBV2TableFunctionClaimBatchInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_claim_batch_set_claimed :: DuckDBV2TableFunctionClaimBatchInfoHandle -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_create :: DuckDBV2FactoryHandle -> Ptr DuckDBV2TableFunctionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_destroy :: Ptr DuckDBV2TableFunctionHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_exec_get_bind_data :: DuckDBV2TableFunctionExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_exec_get_column_count :: DuckDBV2TableFunctionExecInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_exec_get_column_index :: DuckDBV2TableFunctionExecInfoHandle -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_exec_get_global_state :: DuckDBV2TableFunctionExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_exec_get_local_state :: DuckDBV2TableFunctionExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_exec_get_output_chunk :: DuckDBV2TableFunctionExecInfoHandle -> Ptr DuckDBV2DataChunkHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_exec_get_user_data :: DuckDBV2TableFunctionExecInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_filter_pushdown_accept :: DuckDBV2TableFunctionFilterPushdownInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_filter_pushdown_get_bind_data :: DuckDBV2TableFunctionFilterPushdownInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_filter_pushdown_get_column_count :: DuckDBV2TableFunctionFilterPushdownInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_filter_pushdown_get_column_index :: DuckDBV2TableFunctionFilterPushdownInfoHandle -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_filter_pushdown_get_filter :: DuckDBV2TableFunctionFilterPushdownInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ExpressionHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_filter_pushdown_get_filter_count :: DuckDBV2TableFunctionFilterPushdownInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_filter_pushdown_get_user_data :: DuckDBV2TableFunctionFilterPushdownInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_get_bind_info_get_bind_data :: DuckDBV2TableFunctionGetBindInfoInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_get_bind_info_get_user_data :: DuckDBV2TableFunctionGetBindInfoInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_get_bind_info_set_column_identifier :: DuckDBV2TableFunctionGetBindInfoInfoHandle -> DuckDBIdx -> PtrConst DuckDBIdx -> DuckDBIdx -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_get_bind_info_set_option :: DuckDBV2TableFunctionGetBindInfoInfoHandle -> Ptr DuckDBV2Str -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_get_docs :: DuckDBV2TableFunctionHandle -> Ptr DuckDBV2FunctionDocsHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_get_signature :: DuckDBV2TableFunctionHandle -> Ptr DuckDBV2FunctionSignatureHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_global_get_bind_data :: DuckDBV2TableFunctionInitGlobalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_global_get_column_count :: DuckDBV2TableFunctionInitGlobalInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_global_get_column_index :: DuckDBV2TableFunctionInitGlobalInfoHandle -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_global_get_user_data :: DuckDBV2TableFunctionInitGlobalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_global_set_global_state :: DuckDBV2TableFunctionInitGlobalInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_global_set_max_threads :: DuckDBV2TableFunctionInitGlobalInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_local_get_bind_data :: DuckDBV2TableFunctionInitLocalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_local_get_column_count :: DuckDBV2TableFunctionInitLocalInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_local_get_column_index :: DuckDBV2TableFunctionInitLocalInfoHandle -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_local_get_global_state :: DuckDBV2TableFunctionInitLocalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_local_get_user_data :: DuckDBV2TableFunctionInitLocalInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_init_local_set_local_state :: DuckDBV2TableFunctionInitLocalInfoHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_get_bind_data :: DuckDBV2TableFunctionPartitionDataInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_get_global_state :: DuckDBV2TableFunctionPartitionDataInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_get_local_state :: DuckDBV2TableFunctionPartitionDataInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_get_partition_column_count :: DuckDBV2TableFunctionPartitionDataInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_get_partition_column_index :: DuckDBV2TableFunctionPartitionDataInfoHandle -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_get_user_data :: DuckDBV2TableFunctionPartitionDataInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_requires_batch_index :: DuckDBV2TableFunctionPartitionDataInfoHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_requires_partition_columns :: DuckDBV2TableFunctionPartitionDataInfoHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_set_batch_index :: DuckDBV2TableFunctionPartitionDataInfoHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partition_data_set_partition_value :: DuckDBV2TableFunctionPartitionDataInfoHandle -> DuckDBIdx -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partitioning_get_bind_data :: DuckDBV2TableFunctionPartitioningInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partitioning_get_partition_column_count :: DuckDBV2TableFunctionPartitioningInfoHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partitioning_get_partition_column_index :: DuckDBV2TableFunctionPartitioningInfoHandle -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partitioning_get_user_data :: DuckDBV2TableFunctionPartitioningInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_partitioning_set_partition_info :: DuckDBV2TableFunctionPartitioningInfoHandle -> DuckDBV2TablePartitionInfo -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_progress_get_bind_data :: DuckDBV2TableFunctionProgressInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_progress_get_global_state :: DuckDBV2TableFunctionProgressInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_progress_get_user_data :: DuckDBV2TableFunctionProgressInfoHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_progress_set_progress :: DuckDBV2TableFunctionProgressInfoHandle -> CDouble -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_bind_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionBindCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_claim_batch_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionClaimBatchCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_exec_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionExecCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_filter_pushdown_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionFilterPushdownCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_get_bind_info_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionGetBindInfoCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_init_global_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionInitGlobalCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_init_local_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionInitLocalCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_name :: DuckDBV2TableFunctionHandle -> PtrConst DuckDBV2IdentifierT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_partition_data_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionPartitionDataCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_partitioning_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionPartitioningCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_progress_callback :: DuckDBV2TableFunctionHandle -> DuckDBV2TableFunctionProgressCallbackFn -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_projection_pushdown :: DuckDBV2TableFunctionHandle -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_table_function_set_user_data :: DuckDBV2TableFunctionHandle -> Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_token_iterator_destroy :: Ptr DuckDBV2TokenIteratorHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_token_iterator_ends_unterminated :: DuckDBV2TokenIteratorHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_token_iterator_next :: DuckDBV2TokenIteratorHandle -> Ptr DuckDBV2TokenType -> Ptr DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_tokenize_sql :: DuckDBV2ConnectionHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2TokenIteratorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_validate_utf8 :: PtrConst DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_cast :: DuckDBV2FactoryHandle -> DuckDBV2ValueHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_array :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_bigint :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_bignum :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_bit :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_blob :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_bool :: DuckDBV2FactoryHandle -> CBool -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_date :: DuckDBV2FactoryHandle -> Int32 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_decimal :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2HugeintT -> Word8 -> Word8 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_double :: DuckDBV2FactoryHandle -> CDouble -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_float :: DuckDBV2FactoryHandle -> CFloat -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_hugeint :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2HugeintT -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_int :: DuckDBV2FactoryHandle -> Int32 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_interval :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2IntervalT -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_list :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_map :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> DuckDBV2LogicalTypeHandle -> PtrConst DuckDBV2ValueHandle -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_null :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_smallint :: DuckDBV2FactoryHandle -> Int16 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_struct :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2IdentifierT -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_time :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_time_ns :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_time_tz :: DuckDBV2FactoryHandle -> Word64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_timestamp :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_timestamp_ms :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_timestamp_ns :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_timestamp_sec :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_timestamp_tz :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_timestamp_tz_ns :: DuckDBV2FactoryHandle -> Int64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_tinyint :: DuckDBV2FactoryHandle -> Int8 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_tuple :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_type :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_ubigint :: DuckDBV2FactoryHandle -> Word64 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_uhugeint :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2UhugeintT -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_uint :: DuckDBV2FactoryHandle -> Word32 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_usmallint :: DuckDBV2FactoryHandle -> Word16 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_utinyint :: DuckDBV2FactoryHandle -> Word8 -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_uuid :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2HugeintT -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_create_varchar :: DuckDBV2FactoryHandle -> PtrConst DuckDBV2Str -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_destroy :: Ptr DuckDBV2ValueHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_bigint :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_blob :: DuckDBV2ValueHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_bool :: DuckDBV2ValueHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_child :: DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_child_count :: DuckDBV2ValueHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_date :: DuckDBV2ValueHandle -> Ptr Int32 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_decimal :: DuckDBV2ValueHandle -> Ptr DuckDBV2HugeintT -> Ptr Word8 -> Ptr Word8 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_double :: DuckDBV2ValueHandle -> Ptr CDouble -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_float :: DuckDBV2ValueHandle -> Ptr CFloat -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_hugeint :: DuckDBV2ValueHandle -> Ptr DuckDBV2HugeintT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_int :: DuckDBV2ValueHandle -> Ptr Int32 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_interval :: DuckDBV2ValueHandle -> Ptr DuckDBV2IntervalT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_logical_type :: DuckDBV2ValueHandle -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_smallint :: DuckDBV2ValueHandle -> Ptr Int16 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_time :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_time_ns :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_time_tz :: DuckDBV2ValueHandle -> Ptr Word64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_timestamp :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_timestamp_ms :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_timestamp_ns :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_timestamp_sec :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_timestamp_tz :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_timestamp_tz_ns :: DuckDBV2ValueHandle -> Ptr Int64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_tinyint :: DuckDBV2ValueHandle -> Ptr Int8 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_type :: DuckDBV2ValueHandle -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_ubigint :: DuckDBV2ValueHandle -> Ptr Word64 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_uhugeint :: DuckDBV2ValueHandle -> Ptr DuckDBV2UhugeintT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_uint :: DuckDBV2ValueHandle -> Ptr Word32 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_usmallint :: DuckDBV2ValueHandle -> Ptr Word16 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_utinyint :: DuckDBV2ValueHandle -> Ptr Word8 -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_uuid :: DuckDBV2ValueHandle -> Ptr DuckDBV2HugeintT -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_get_varchar :: DuckDBV2ValueHandle -> Ptr DuckDBV2Str -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_is_null :: DuckDBV2ValueHandle -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_value_to_string :: DuckDBV2ValueHandle -> Ptr CChar -> DuckDBIdx -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_cast :: DuckDBV2FactoryHandle -> DuckDBV2VectorHandle -> DuckDBV2VectorHandle -> DuckDBV2CastMode -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_constant_set_valid :: DuckDBV2VectorHandle -> CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_flat_get_validity_mutable :: DuckDBV2VectorHandle -> Ptr (Ptr Word64) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_flatten :: DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_arena :: DuckDBV2VectorHandle -> Ptr DuckDBV2ArenaHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_child :: DuckDBV2VectorHandle -> DuckDBIdx -> Ptr DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_child_count :: DuckDBV2VectorHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_data_mutable :: DuckDBV2VectorHandle -> Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_logical_type :: DuckDBV2VectorHandle -> Ptr DuckDBV2LogicalTypeHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_size :: DuckDBV2VectorHandle -> Ptr DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_value :: DuckDBV2VectorHandle -> DuckDBIdx -> Ptr DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_vector_type :: DuckDBV2VectorHandle -> Ptr DuckDBV2VectorType -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_get_view :: DuckDBV2VectorHandle -> Ptr DuckDBV2VectorView -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_make_constant :: DuckDBV2VectorHandle -> DuckDBV2ValueHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_make_sequence :: DuckDBV2VectorHandle -> Int64 -> Int64 -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_reference :: DuckDBV2VectorHandle -> DuckDBV2VectorHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_set_null :: DuckDBV2VectorHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_set_size :: DuckDBV2VectorHandle -> DuckDBIdx -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: c_duckdb_v2_vector_set_value :: DuckDBV2VectorHandle -> DuckDBIdx -> DuckDBV2ValueHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error
+ Database.DuckDB.Simple.Internal: checkCode :: DuckDBV2Error -> IO ()
+ Database.DuckDB.Simple.Internal: checked :: (Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO ()
+ Database.DuckDB.Simple.Internal: cleanupStatementStreamRef :: IORef StatementStreamState -> IO ()
+ Database.DuckDB.Simple.Internal: clearArrowResult :: Statement -> IO ()
+ Database.DuckDB.Simple.Internal: clearBoundValues :: IORef (IntMap DuckDBV2ValueHandle) -> IO ()
+ Database.DuckDB.Simple.Internal: clearStatementSchemas :: Statement -> IO ()
+ Database.DuckDB.Simple.Internal: closeConnection :: Connection -> IO ()
+ Database.DuckDB.Simple.Internal: closeEnvironment :: Environment -> IO ()
+ Database.DuckDB.Simple.Internal: closeInstance :: Instance -> IO ()
+ Database.DuckDB.Simple.Internal: closeStatementState :: Statement -> IO ()
+ Database.DuckDB.Simple.Internal: createConnection :: Instance -> ConnectionOptions -> IO () -> IO Connection
+ Database.DuckDB.Simple.Internal: createEnvironment :: IO Environment
+ Database.DuckDB.Simple.Internal: createInstance :: Environment -> [(Text, Text)] -> IO Instance
+ Database.DuckDB.Simple.Internal: createStatementFromParsed :: Connection -> Query -> DuckDBV2SqlStatementHandle -> IO Statement
+ Database.DuckDB.Simple.Internal: createStatementFromParsedWithCollections :: Connection -> Query -> DuckDBV2SqlStatementHandle -> [(Text, DuckDBV2ColumnDataCollectionHandle, [Text])] -> IO Statement
+ Database.DuckDB.Simple.Internal: createStatementWithCollections :: Connection -> Query -> [(Text, DuckDBV2ColumnDataCollectionHandle, [Text])] -> IO Statement
+ Database.DuckDB.Simple.Internal: dUCKDB_V2_API_ALLOW_DEPRECATED :: CInt
+ Database.DuckDB.Simple.Internal: dUCKDB_V2_API_VERSION_MAJOR :: CInt
+ Database.DuckDB.Simple.Internal: dUCKDB_V2_API_VERSION_MINOR :: CInt
+ Database.DuckDB.Simple.Internal: dUCKDB_V2_API_VERSION_PATCH :: CInt
+ Database.DuckDB.Simple.Internal: dUCKDB_V2_BYTES_INLINE_LENGTH :: CInt
+ Database.DuckDB.Simple.Internal: data ArrowArray
+ Database.DuckDB.Simple.Internal: data ArrowArrayStream
+ Database.DuckDB.Simple.Internal: data ArrowSchema
+ Database.DuckDB.Simple.Internal: data Connection
+ Database.DuckDB.Simple.Internal: data ConnectionOptions
+ Database.DuckDB.Simple.Internal: data DuckDBExtensionInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AggregateFunctionBindInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AggregateFunctionCombineInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AggregateFunctionDestroyInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AggregateFunctionFinalizeInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AggregateFunctionInitInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AggregateFunctionSizeInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AggregateFunctionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AggregateFunctionUpdateInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ArenaStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ArrowExporterStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ArrowImporterStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ArrowResultStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2AttachOptionsStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2Bytes
+ Database.DuckDB.Simple.Internal: data DuckDBV2CastFunctionExecInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CastFunctionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ColumnDataCollectionAppendStateStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ColumnDataCollectionSharedScanStateStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ColumnDataCollectionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ColumnDataCollectionWorkerScanStateStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ColumnDescriptionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ConfigStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ConnectionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ContextStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyFromBindInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyFromExecInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyFromInitGlobalInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyFromInitLocalInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyFromProgressInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyFunctionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyToBatchInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyToBatchSizeInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyToBindInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyToFinalizeInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyToFlushInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyToInitInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CopyToStatisticsInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CustomOptionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CustomOptionUpdateInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2CustomTypeStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2DataChunkStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2EnvironmentStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ErrorInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ExpressionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ExtensionInput
+ Database.DuckDB.Simple.Internal: data DuckDBV2FactoryStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2FileOpenOptionsStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2FileStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2FileSystemStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2FunctionBindInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2FunctionDocsStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2FunctionSignatureStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2HugeintT
+ Database.DuckDB.Simple.Internal: data DuckDBV2InstanceStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2IntervalT
+ Database.DuckDB.Simple.Internal: data DuckDBV2ListEntry
+ Database.DuckDB.Simple.Internal: data DuckDBV2LogSinkFlushInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2LogSinkStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2LogSinkWriteInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2LogicalTypeStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2MacroFunctionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2MultiFileFunctionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2Opaque
+ Database.DuckDB.Simple.Internal: data DuckDBV2OptionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2PragmaFunctionQueryInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2PragmaFunctionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2PreparedStatementStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2QnameStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2QueryMetricsNodeStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2QueryMetricsStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ReplacementScanInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ReplacementScanStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ResultStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ScalarFunctionBindInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ScalarFunctionExecInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ScalarFunctionInitInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ScalarFunctionResolveTypesInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2ScalarFunctionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2SchemaStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2SqlStatementStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2StatementIteratorStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2Str
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableDescriptionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionBindInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionClaimBatchInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionExecInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionFilterPushdownInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionGetBindInfoInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionInitGlobalInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionInitLocalInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionPartitionDataInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionPartitioningInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionProgressInfoStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TableFunctionStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2TokenIteratorStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2UhugeintT
+ Database.DuckDB.Simple.Internal: data DuckDBV2ValueStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2VectorStruct
+ Database.DuckDB.Simple.Internal: data DuckDBV2VectorView
+ Database.DuckDB.Simple.Internal: data Duckdb_v2_bytes_value_inlined
+ Database.DuckDB.Simple.Internal: data Duckdb_v2_bytes_value_pointer
+ Database.DuckDB.Simple.Internal: data EnvironmentState
+ Database.DuckDB.Simple.Internal: data ErrorType
+ Database.DuckDB.Simple.Internal: data FileFlag
+ Database.DuckDB.Simple.Internal: data Instance
+ Database.DuckDB.Simple.Internal: data InstanceState
+ Database.DuckDB.Simple.Internal: data LogLevel
+ Database.DuckDB.Simple.Internal: data LogScope
+ Database.DuckDB.Simple.Internal: data LogicalTypeId
+ Database.DuckDB.Simple.Internal: data MacroFunctionType
+ Database.DuckDB.Simple.Internal: data SettingScope
+ Database.DuckDB.Simple.Internal: data TokenType
+ Database.DuckDB.Simple.Internal: defaultAggregateStripes :: Int
+ Database.DuckDB.Simple.Internal: defaultArrowBatchSize :: Int
+ Database.DuckDB.Simple.Internal: defaultCallbackLockWait :: Int
+ Database.DuckDB.Simple.Internal: defaultConnectionOptions :: ConnectionOptions
+ Database.DuckDB.Simple.Internal: destroy :: Storable a => (Ptr a -> IO DuckDBV2Error) -> a -> IO ()
+ Database.DuckDB.Simple.Internal: errorTypeCodes :: [(ErrorType, DuckDBV2Error)]
+ Database.DuckDB.Simple.Internal: errorTypeFromNative :: DuckDBV2Error -> Maybe ErrorType
+ Database.DuckDB.Simple.Internal: exceptionText :: SomeException -> String
+ Database.DuckDB.Simple.Internal: executeParsedResult :: Connection -> Query -> DuckDBV2SqlStatementHandle -> IO DuckDBV2ResultHandle
+ Database.DuckDB.Simple.Internal: executeParsedStatement :: Connection -> Query -> DuckDBV2SqlStatementHandle -> IO Int
+ Database.DuckDB.Simple.Internal: executePreparedArrowResult :: Statement -> IO DuckDBV2ArrowResultHandle
+ Database.DuckDB.Simple.Internal: factoryOutput :: Storable a => DuckDBV2ConnectionHandle -> (DuckDBV2FactoryHandle -> Ptr a -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a
+ Database.DuckDB.Simple.Internal: fileFlagToNative :: FileFlag -> DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: finalizeChunk :: StatementStreamChunk -> IO ()
+ Database.DuckDB.Simple.Internal: finalizeStream :: StatementStream -> IO ()
+ Database.DuckDB.Simple.Internal: foldIdentifier :: Text -> Text
+ Database.DuckDB.Simple.Internal: indices :: DuckDBIdx -> [DuckDBIdx]
+ Database.DuckDB.Simple.Internal: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Internal.ConnectionOptions
+ Database.DuckDB.Simple.Internal: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Internal.ResultMode
+ Database.DuckDB.Simple.Internal: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Internal.ConnectionOptions
+ Database.DuckDB.Simple.Internal: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Internal.ResultMode
+ Database.DuckDB.Simple.Internal: instanceCastFactory :: Instance -> IO DuckDBV2FactoryHandle
+ Database.DuckDB.Simple.Internal: logLevelFromNative :: DuckDBV2LogLevel -> IO LogLevel
+ Database.DuckDB.Simple.Internal: logLevelToNative :: LogLevel -> DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: logScopeFromNative :: DuckDBV2LogScope -> IO LogScope
+ Database.DuckDB.Simple.Internal: logicalTypeIdFromNative :: DuckDBV2LogicalTypeId -> IO LogicalTypeId
+ Database.DuckDB.Simple.Internal: logicalTypeIdName :: LogicalTypeId -> Text
+ Database.DuckDB.Simple.Internal: logicalTypeIdToNative :: LogicalTypeId -> DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: macroFunctionTypeToNative :: MacroFunctionType -> DuckDBV2MacroFunctionType
+ Database.DuckDB.Simple.Internal: newtype DuckDBIdx
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionBindCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionBindInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionCombineCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionCombineCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionCombineInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionDestroyCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionDestroyCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionDestroyInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionFinalizeCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionFinalizeCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionFinalizeInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionInitCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionInitCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionInitInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionSizeCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionSizeCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionSizeInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionUpdateCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionUpdateCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AggregateFunctionUpdateInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ArenaHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ArrowExporterHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ArrowImporterHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ArrowResultHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2AttachOptionsHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2BignumT
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2BitT
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2BlobT
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CastFunctionExecCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CastFunctionExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CastFunctionExecInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CastFunctionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CastMode
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ColumnDataCollectionAppendStateHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ColumnDataCollectionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ColumnDataCollectionSharedScanStateHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ColumnDataCollectionWorkerScanStateHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ColumnDescriptionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ConfigHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ConnectionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ContextHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromBindCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromBindInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromExecCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromExecInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromInitGlobalCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromInitGlobalCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromInitGlobalInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromInitLocalCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromInitLocalCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromInitLocalInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromProgressCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromProgressCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFromProgressInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyFunctionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBatchCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBatchCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBatchInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBatchSizeCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBatchSizeCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBatchSizeInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBindCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToBindInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToFinalizeCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToFinalizeCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToFinalizeInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToFlushCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToFlushCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToFlushInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToInitCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToInitCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToInitInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToStatisticsCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToStatisticsCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CopyToStatisticsInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CustomOptionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CustomOptionUpdateCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CustomOptionUpdateCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CustomOptionUpdateInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2CustomTypeHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2DataChunkHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2EnvironmentHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2Error
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ErrorInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ExpressionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ExtensionGetApiFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ExtensionGetApiFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ExtensionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FactoryHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FileHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FileOpenOptionsHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FileSystemHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FileSystemPathGlobCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FileSystemPathGlobCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FunctionBindInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FunctionDocsHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2FunctionSignatureHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2IdentifierT
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2InstanceHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogScope
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogSinkFlushCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogSinkFlushCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogSinkFlushInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogSinkHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogSinkWriteCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogSinkWriteCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogSinkWriteInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogicalTypeHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2MacroFunctionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2MacroFunctionType
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2MultiFileFunctionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2OpaqueDestroyFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2OpaqueDestroyFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2OpaqueEqualsFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2OpaqueEqualsFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2OptionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2OrderPreservation
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2PragmaFunctionExecCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2PragmaFunctionExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2PragmaFunctionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2PragmaFunctionQueryCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2PragmaFunctionQueryCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2PragmaFunctionQueryInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2PreparedStatementHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2QnameHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2QueryMetricsHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2QueryMetricsNodeHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ReplacementScanCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ReplacementScanCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ReplacementScanHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ReplacementScanInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ResultHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ResultStepStatus
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ResultType
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionBindCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionBindInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionExecCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionExecInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionInitCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionInitCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionInitInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionResolveTypesCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionResolveTypesCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ScalarFunctionResolveTypesInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2SchemaHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2SelT
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2SettingScope
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2SqlStatementHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2StatementIteratorHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableDescriptionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionBindCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionBindCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionBindInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionClaimBatchCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionClaimBatchCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionClaimBatchInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionExecCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionExecCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionExecInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionFilterPushdownCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionFilterPushdownCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionFilterPushdownInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionGetBindInfoCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionGetBindInfoCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionGetBindInfoInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionInitGlobalCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionInitGlobalCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionInitGlobalInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionInitLocalCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionInitLocalCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionInitLocalInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionPartitionDataCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionPartitionDataCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionPartitionDataInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionPartitioningCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionPartitioningCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionPartitioningInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionProgressCallbackFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionProgressCallbackFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TableFunctionProgressInfoHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TablePartitionInfo
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TextSinkFn
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TextSinkFn_Aux
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TokenIteratorHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2ValueHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2VarcharT
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2VectorHandle
+ Database.DuckDB.Simple.Internal: newtype DuckDBV2VectorType
+ Database.DuckDB.Simple.Internal: newtype Duckdb_v2_bytes_value
+ Database.DuckDB.Simple.Internal: newtype Environment
+ Database.DuckDB.Simple.Internal: nullHandle :: Coercible (Ptr ()) a => a
+ Database.DuckDB.Simple.Internal: output :: Storable a => (Ptr a -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2CastModeMaxEnum :: DuckDBV2CastMode
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2CastModeNormal :: DuckDBV2CastMode
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2CastModeTry :: DuckDBV2CastMode
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorApi :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorConfigurationInvalid :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorConfigurationPermission :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorConfigurationSettings :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorDatabaseCatalog :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorDatabaseConstraint :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorDatabaseIndex :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorDatabaseSequence :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorDatabaseSerialization :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorDatabaseStatistics :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorDatabaseTransaction :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorIOEof :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorIOFileNotFound :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorIOGeneral :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorIOHTTP :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorIONetwork :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorIOReadFailure :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorInputInvalid :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorInputMetadataDeferred :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorInputObjectSize :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorInputOutOfRange :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorInputParameterInvalid :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorMaxEnum :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorNone :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryBinder :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryExecutor :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryExpression :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryNotImplemented :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryOptimizer :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryParameterNotAllowed :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryParameterNotResolved :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryParser :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryPlanner :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQueryScheduler :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorQuerySyntax :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorResourceAutoload :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorResourceConnection :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorResourceDependency :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorResourceInUse :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorResourceMissingExtension :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorResourceOutOfMemory :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorRuntimeFatal :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorRuntimeInternal :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorRuntimeInterrupt :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorRuntimeNullPointer :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorTypeConversion :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorTypeDecimal :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorTypeDivideByZero :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorTypeInvalid :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorTypeMismatch :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ErrorTypeUnknown :: DuckDBV2Error
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeBoundColumnRef :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeBoundFunction :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCaseExpr :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareBetween :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareDistinctFrom :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareEqual :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareGreaterthan :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareGreaterthanorequalto :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareIn :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareLessthan :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareLessthanorequalto :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareNotDistinctFrom :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareNotIn :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeCompareNotequal :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeConjunctionAnd :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeConjunctionOr :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeInvalid :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeMaxEnum :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeOperatorCast :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeOperatorCoalesce :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeOperatorIsNotNull :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeOperatorIsNull :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeOperatorNot :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeValueConstant :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ExpressionTypeValueParameter :: DuckDBV2ExpressionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagAppend :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagCreate :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagCreateNew :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagExclusiveCreate :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagExternalFileCache :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagInvalid :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagMaxEnum :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagParallelAccess :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagRead :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FileFlagWrite :: DuckDBV2FileFlag
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionParameterKindMaxEnum :: DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionParameterKindNamedOnly :: DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionParameterKindNamedVariadic :: DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionParameterKindPositionalOnly :: DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionParameterKindPositionalVariadic :: DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionParameterKindStandard :: DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyAggDistinctDependent :: DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyAggDistinctDependentNo :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyAggDistinctDependentYes :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyAggOrderDependent :: DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyAggOrderDependentNo :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyAggOrderDependentYes :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyCollationHandling :: DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyCollationHandlingIgnore :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyCollationHandlingPropagate :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyCollationHandlingPushCombinable :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyFallibility :: DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyFallibilityFallible :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyFallibilityInfallible :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyKeyMaxEnum :: DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyNullHandling :: DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyNullHandlingDefault :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyNullHandlingSpecial :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyStability :: DuckDBV2FunctionPropertyKey
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyStabilityConsistent :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyStabilityConsistentWithinQuery :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyStabilityVolatile :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2FunctionPropertyValueMaxEnum :: DuckDBV2FunctionPropertyValue
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogLevelDebug :: DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogLevelError :: DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogLevelFatal :: DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogLevelInfo :: DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogLevelMaxEnum :: DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogLevelTrace :: DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogLevelWarning :: DuckDBV2LogLevel
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogScopeConnection :: DuckDBV2LogScope
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogScopeDatabase :: DuckDBV2LogScope
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogScopeMaxEnum :: DuckDBV2LogScope
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogScopeThread :: DuckDBV2LogScope
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdAny :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdArray :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdBigInt :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdBigNum :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdBit :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdBlob :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdBoolean :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdDate :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdDecimal :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdDouble :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdEnum :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdFloat :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdGeometry :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdHugeInt :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdInteger :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdInterval :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdInvalid :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdList :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdMap :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdMaxEnum :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdSQLNull :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdSmallInt :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdStruct :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTime :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTimeNs :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTimeTz :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTimestamp :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTimestampMs :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTimestampNs :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTimestampSec :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTimestampTz :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTimestampTzNs :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTinyInt :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdTuple :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdType :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdUBigInt :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdUHugeInt :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdUInteger :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdUSmallInt :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdUTinyInt :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdUUID :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdUnion :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdUnknown :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdVarchar :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2LogicalTypeIdVariant :: DuckDBV2LogicalTypeId
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2MacroFunctionTypeMaxEnum :: DuckDBV2MacroFunctionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2MacroFunctionTypeScalar :: DuckDBV2MacroFunctionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2MacroFunctionTypeTable :: DuckDBV2MacroFunctionType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2OrderPreservationFixedOrder :: DuckDBV2OrderPreservation
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2OrderPreservationInsertionOrder :: DuckDBV2OrderPreservation
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2OrderPreservationMaxEnum :: DuckDBV2OrderPreservation
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2OrderPreservationNoOrder :: DuckDBV2OrderPreservation
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultStepStatusCancelled :: DuckDBV2ResultStepStatus
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultStepStatusChunk :: DuckDBV2ResultStepStatus
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultStepStatusFinished :: DuckDBV2ResultStepStatus
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultStepStatusMaxEnum :: DuckDBV2ResultStepStatus
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultStepStatusWaiting :: DuckDBV2ResultStepStatus
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultTypeChangedRows :: DuckDBV2ResultType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultTypeMaxEnum :: DuckDBV2ResultType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultTypeNothing :: DuckDBV2ResultType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2ResultTypeQueryResult :: DuckDBV2ResultType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2SettingScopeDefault :: DuckDBV2SettingScope
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2SettingScopeGlobal :: DuckDBV2SettingScope
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2SettingScopeMaxEnum :: DuckDBV2SettingScope
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2SettingScopeSession :: DuckDBV2SettingScope
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeAlter :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeAnalyze :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeAttach :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeCall :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeConnect :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeCopy :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeCopyDatabase :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeCreate :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeCreateFunc :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeDelete :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeDetach :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeDisconnect :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeDrop :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeExecute :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeExplain :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeExport :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeExtension :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeExternalResource :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeInsert :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeInvalid :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeLoad :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeLogicalPlan :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeMaxEnum :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeMergeInto :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeMulti :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypePassthrough :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypePragma :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypePrepare :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeRelation :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeSelect :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeSet :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeTransaction :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeUpdate :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeUpdateExtensions :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeVacuum :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2StatementTypeVariableSet :: DuckDBV2StatementType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TablePartitionInfoDisjointPartitions :: DuckDBV2TablePartitionInfo
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TablePartitionInfoMaxEnum :: DuckDBV2TablePartitionInfo
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TablePartitionInfoNotPartitioned :: DuckDBV2TablePartitionInfo
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TablePartitionInfoOverlappingPartitions :: DuckDBV2TablePartitionInfo
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TablePartitionInfoSingleValuePartitions :: DuckDBV2TablePartitionInfo
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeComment :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeEndOfInput :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeIdentifier :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeInvalid :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeKeyword :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeMaxEnum :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeNumberLiteral :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeOperator :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeStringLiteral :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2TokenTypeTerminator :: DuckDBV2TokenType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2VectorTypeConstant :: DuckDBV2VectorType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2VectorTypeDictionary :: DuckDBV2VectorType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2VectorTypeFlat :: DuckDBV2VectorType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2VectorTypeMaxEnum :: DuckDBV2VectorType
+ Database.DuckDB.Simple.Internal: pattern DuckDBV2VectorTypeOther :: DuckDBV2VectorType
+ Database.DuckDB.Simple.Internal: preparedParameterType :: Statement -> DuckDBIdx -> IO (Maybe LogicalTypeRep)
+ Database.DuckDB.Simple.Internal: quoteIdentifier :: Text -> Text
+ Database.DuckDB.Simple.Internal: readBytes :: DuckDBV2Str -> IO ByteString
+ Database.DuckDB.Simple.Internal: readConnectionVariant :: Connection -> DuckDBV2VectorHandle -> IO (Int -> IO FieldValue)
+ Database.DuckDB.Simple.Internal: readIdentifier :: DuckDBV2IdentifierT -> IO Text
+ Database.DuckDB.Simple.Internal: readText :: DuckDBV2Str -> IO Text
+ Database.DuckDB.Simple.Internal: releaseItems :: (a -> IO ()) -> [a] -> IO ()
+ Database.DuckDB.Simple.Internal: settingScopeFromNative :: DuckDBV2SettingScope -> IO SettingScope
+ Database.DuckDB.Simple.Internal: settingScopeToNative :: SettingScope -> DuckDBV2SettingScope
+ Database.DuckDB.Simple.Internal: statementSchemas :: Statement -> IO (DuckDBV2SchemaHandle, DuckDBV2SchemaHandle)
+ Database.DuckDB.Simple.Internal: storeBoundValue :: Statement -> DuckDBIdx -> DuckDBV2ValueHandle -> IO ()
+ Database.DuckDB.Simple.Internal: throwConversionError :: Text -> IO a
+ Database.DuckDB.Simple.Internal: throwInternalError :: Text -> IO a
+ Database.DuckDB.Simple.Internal: throwInvalidArgument :: Text -> IO a
+ Database.DuckDB.Simple.Internal: tokenTypeFromNative :: DuckDBV2TokenType -> IO TokenType
+ Database.DuckDB.Simple.Internal: type NativeValue = DuckDBV2ValueHandle
+ Database.DuckDB.Simple.Internal: type ValueFactory = DuckDBV2FactoryHandle
+ Database.DuckDB.Simple.Internal: valueText :: DuckDBV2ValueHandle -> IO Text
+ Database.DuckDB.Simple.Internal: withArrowResult :: Statement -> (DuckDBV2ArrowResultHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withBytes :: ByteString -> (ConstPtr DuckDBV2Str -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withCollectionStatement :: Connection -> Query -> [(Text, DuckDBV2ColumnDataCollectionHandle, [Text])] -> (Statement -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withConnectionFactory :: DuckDBV2ConnectionHandle -> (DuckDBV2FactoryHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withConnectionLock :: Connection -> IO a -> IO a
+ Database.DuckDB.Simple.Internal: withContextFactory :: DuckDBV2ContextHandle -> (DuckDBV2FactoryHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withCreatedValues :: [IO DuckDBV2ValueHandle] -> ([DuckDBV2ValueHandle] -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withDefaultCastVector :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> DuckDBIdx -> (DuckDBIdx -> (DuckDBV2ValueHandle -> IO ()) -> IO ()) -> (DuckDBV2VectorHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withIdentifier :: Text -> (ConstPtr DuckDBV2IdentifierT -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withIdentifierArray :: [Text] -> (Ptr DuckDBV2IdentifierT -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withInstanceFactory :: DuckDBV2InstanceHandle -> (DuckDBV2FactoryHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withInstanceHandle :: Instance -> (DuckDBV2InstanceHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withKeptParameterTypes :: Statement -> IO a -> IO a
+ Database.DuckDB.Simple.Internal: withOwnedVector :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> (DuckDBV2VectorHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withParsedStatements :: Connection -> Query -> ([DuckDBV2SqlStatementHandle] -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withQueryError :: Query -> IO a -> IO a
+ Database.DuckDB.Simple.Internal: withResultWindow :: Connection -> (a -> IO ()) -> IO a -> IO a
+ Database.DuckDB.Simple.Internal: withSQLStatement :: DuckDBV2ConnectionHandle -> Text -> (DuckDBV2SqlStatementHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withStatementSchemas :: Statement -> (DuckDBV2SchemaHandle -> DuckDBV2SchemaHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withText :: Text -> (ConstPtr DuckDBV2Str -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withTextArray :: [Text] -> (Ptr DuckDBV2Str -> IO a) -> IO a
+ Database.DuckDB.Simple.Logging: LogLevelDebug :: LogLevel
+ Database.DuckDB.Simple.Logging: LogLevelError :: LogLevel
+ Database.DuckDB.Simple.Logging: LogLevelFatal :: LogLevel
+ Database.DuckDB.Simple.Logging: LogLevelInfo :: LogLevel
+ Database.DuckDB.Simple.Logging: LogLevelTrace :: LogLevel
+ Database.DuckDB.Simple.Logging: LogLevelWarning :: LogLevel
+ Database.DuckDB.Simple.Logging: LogScopeConnection :: LogScope
+ Database.DuckDB.Simple.Logging: LogScopeDatabase :: LogScope
+ Database.DuckDB.Simple.Logging: LogScopeThread :: LogScope
+ Database.DuckDB.Simple.Logging: [logEntryConnectionId] :: LogEntry -> Maybe Word64
+ Database.DuckDB.Simple.Logging: [logEntryQueryId] :: LogEntry -> Maybe Word64
+ Database.DuckDB.Simple.Logging: [logEntryScope] :: LogEntry -> LogScope
+ Database.DuckDB.Simple.Logging: [logEntrySeverity] :: LogEntry -> LogLevel
+ Database.DuckDB.Simple.Logging: [logEntryThreadId] :: LogEntry -> Maybe Word64
+ Database.DuckDB.Simple.Logging: [logEntryTransactionId] :: LogEntry -> Maybe Word64
+ Database.DuckDB.Simple.Logging: data LogLevel
+ Database.DuckDB.Simple.Logging: data LogScope
+ Database.DuckDB.Simple.Logging: logConnectionMessage :: Connection -> LogLevel -> Text -> Text -> IO ()
+ Database.DuckDB.Simple.Logging: logInstanceMessage :: Instance -> LogLevel -> Text -> Text -> IO ()
+ Database.DuckDB.Simple.Logging: registerInstanceLogSink :: Instance -> Text -> (LogEntry -> IO ()) -> IO () -> IO ()
+ Database.DuckDB.Simple.Logging: registerLogSink :: Connection -> Text -> (LogEntry -> IO ()) -> IO () -> IO ()
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeAlias :: Text -> LogicalTypeRep -> LogicalTypeRep
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeCollatedVarchar :: Text -> LogicalTypeRep
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdAny :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdArray :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdBigNum :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdBit :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdBlob :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdBoolean :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdDate :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdDecimal :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdDouble :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdEnum :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdFloat :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdGeometry :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdInteger :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdInterval :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdInvalid :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdList :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdMap :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdSQLNull :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdStruct :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTime :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTimeNs :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTimeTz :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTimestamp :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTimestampMs :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTimestampNs :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTimestampSec :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTimestampTz :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTimestampTzNs :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdTuple :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdType :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdUBigInt :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdUHugeInt :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdUInteger :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdUSmallInt :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdUTinyInt :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdUUID :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdUnion :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdUnknown :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdVarchar :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeIdVariant :: LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeTuple :: Array Int LogicalTypeRep -> LogicalTypeRep
+ Database.DuckDB.Simple.LogicalRep: StructuredType :: LogicalTypeRep -> TypeDeclaration
+ Database.DuckDB.Simple.LogicalRep: TupleValue :: Array Int a -> Array Int LogicalTypeRep -> TupleValue a
+ Database.DuckDB.Simple.LogicalRep: TypeExpression :: Text -> TypeDeclaration
+ Database.DuckDB.Simple.LogicalRep: [tupleValueFields] :: TupleValue a -> Array Int a
+ Database.DuckDB.Simple.LogicalRep: [tupleValueTypes] :: TupleValue a -> Array Int LogicalTypeRep
+ Database.DuckDB.Simple.LogicalRep: aliasBase :: LogicalTypeRep -> LogicalTypeRep
+ Database.DuckDB.Simple.LogicalRep: data LogicalTypeId
+ Database.DuckDB.Simple.LogicalRep: data TupleValue a
+ Database.DuckDB.Simple.LogicalRep: data TypeDeclaration
+ Database.DuckDB.Simple.LogicalRep: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.LogicalRep.TypeDeclaration
+ Database.DuckDB.Simple.LogicalRep: instance GHC.Internal.Classes.Eq a => GHC.Internal.Classes.Eq (Database.DuckDB.Simple.LogicalRep.TupleValue a)
+ Database.DuckDB.Simple.LogicalRep: instance GHC.Internal.Show.Show Database.DuckDB.Simple.LogicalRep.TypeDeclaration
+ Database.DuckDB.Simple.LogicalRep: instance GHC.Internal.Show.Show a => GHC.Internal.Show.Show (Database.DuckDB.Simple.LogicalRep.TupleValue a)
+ Database.DuckDB.Simple.LogicalRep: listElementType :: LogicalTypeRep -> LogicalTypeRep
+ Database.DuckDB.Simple.LogicalRep: logicalTypeSQL :: LogicalTypeRep -> Text
+ Database.DuckDB.Simple.LogicalRep: mapEntryTypes :: LogicalTypeRep -> (LogicalTypeRep, LogicalTypeRep)
+ Database.DuckDB.Simple.LogicalRep: tupleValueTypeRep :: TupleValue a -> LogicalTypeRep
+ Database.DuckDB.Simple.LogicalRep: withoutAliases :: LogicalTypeRep -> LogicalTypeRep
+ Database.DuckDB.Simple.Pragma: ExecPragma :: (args -> IO ()) -> PragmaCallback args
+ Database.DuckDB.Simple.Pragma: QueryPragma :: (args -> IO Text) -> PragmaCallback args
+ Database.DuckDB.Simple.Pragma: data PragmaCallback args
+ Database.DuckDB.Simple.Pragma: registerPragmaFunction :: FromRow args => Connection -> Text -> [FunctionParameter] -> PragmaCallback args -> IO ()
+ Database.DuckDB.Simple.QueryMetrics: QueryMetrics :: Text -> [(Text, FieldValue)] -> [QueryMetrics] -> QueryMetrics
+ Database.DuckDB.Simple.QueryMetrics: [metricsChildren] :: QueryMetrics -> [QueryMetrics]
+ Database.DuckDB.Simple.QueryMetrics: [metricsNodeName] :: QueryMetrics -> Text
+ Database.DuckDB.Simple.QueryMetrics: [metricsValues] :: QueryMetrics -> [(Text, FieldValue)]
+ Database.DuckDB.Simple.QueryMetrics: data QueryMetrics
+ Database.DuckDB.Simple.QueryMetrics: enableProfiling :: Connection -> Bool -> IO ()
+ Database.DuckDB.Simple.QueryMetrics: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.QueryMetrics.QueryMetrics
+ Database.DuckDB.Simple.QueryMetrics: instance GHC.Internal.Show.Show Database.DuckDB.Simple.QueryMetrics.QueryMetrics
+ Database.DuckDB.Simple.QueryMetrics: queryMetrics :: Connection -> IO QueryMetrics
+ Database.DuckDB.Simple.Registration: ReplaceWithQuery :: Text -> Replacement
+ Database.DuckDB.Simple.Registration: StructuredType :: LogicalTypeRep -> TypeDeclaration
+ Database.DuckDB.Simple.Registration: TypeExpression :: Text -> TypeDeclaration
+ Database.DuckDB.Simple.Registration: data DefaultValue
+ Database.DuckDB.Simple.Registration: data FunctionParameter
+ Database.DuckDB.Simple.Registration: data Replacement
+ Database.DuckDB.Simple.Registration: data TypeDeclaration
+ Database.DuckDB.Simple.Registration: defaultValue :: ToDuckValue a => a -> DefaultValue
+ Database.DuckDB.Simple.Registration: parameter :: Text -> TypeDeclaration -> FunctionParameter
+ Database.DuckDB.Simple.Registration: parameterDefault :: FunctionParameter -> Maybe DefaultValue
+ Database.DuckDB.Simple.Registration: parameterName :: FunctionParameter -> Text
+ Database.DuckDB.Simple.Registration: parameterType :: FunctionParameter -> TypeDeclaration
+ Database.DuckDB.Simple.Registration: registerCustomType :: Connection -> Text -> TypeDeclaration -> IO ()
+ Database.DuckDB.Simple.Registration: registerInstanceReplacementScan :: Instance -> ([Text] -> IO (Maybe Replacement)) -> IO ()
+ Database.DuckDB.Simple.Registration: registerReplacementScan :: Connection -> ([Text] -> IO (Maybe Replacement)) -> IO ()
+ Database.DuckDB.Simple.Streaming: fold :: (FromRow row, ToRow params) => Connection -> Query -> params -> a -> (a -> row -> IO a) -> IO a
+ Database.DuckDB.Simple.Streaming: foldArrow :: ToRow params => Connection -> Query -> params -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a
+ Database.DuckDB.Simple.Streaming: foldArrow_ :: Connection -> Query -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a
+ Database.DuckDB.Simple.Streaming: foldNamed :: FromRow row => Connection -> Query -> [NamedParam] -> a -> (a -> row -> IO a) -> IO a
+ Database.DuckDB.Simple.Streaming: fold_ :: FromRow row => Connection -> Query -> a -> (a -> row -> IO a) -> IO a
+ Database.DuckDB.Simple.Streaming: nextRow :: FromRow r => Statement -> IO (Maybe r)
+ Database.DuckDB.Simple.Streaming: nextRowWith :: RowParser r -> Statement -> IO (Maybe r)
+ Database.DuckDB.Simple.ToField: duckdbTypeDeclaration :: DuckDBColumnType a => Proxy a -> Either Text TypeDeclaration
+ Database.DuckDB.Simple.ToField: instance (Database.DuckDB.Simple.ToField.DuckDBColumnType a, Database.DuckDB.Simple.ToField.ToDuckValue a) => Database.DuckDB.Simple.ToField.ToDuckValue [a]
+ Database.DuckDB.Simple.ToField: instance (Database.DuckDB.Simple.ToField.DuckDBColumnType a, Database.DuckDB.Simple.ToField.ToField a) => Database.DuckDB.Simple.ToField.ToField (Database.DuckDB.Simple.FromField.SizedArray n a)
+ Database.DuckDB.Simple.ToField: instance (Database.DuckDB.Simple.ToField.DuckDBColumnType a, Database.DuckDB.Simple.ToField.ToField a) => Database.DuckDB.Simple.ToField.ToField [a]
+ Database.DuckDB.Simple.ToField: instance (Database.DuckDB.Simple.ToField.DuckDBColumnType k, Database.DuckDB.Simple.ToField.DuckDBColumnType v) => Database.DuckDB.Simple.ToField.DuckDBColumnType (Data.Map.Internal.Map k v)
+ Database.DuckDB.Simple.ToField: instance (Database.DuckDB.Simple.ToField.DuckDBColumnType k, Database.DuckDB.Simple.ToField.DuckDBColumnType v, Database.DuckDB.Simple.ToField.ToDuckValue k, Database.DuckDB.Simple.ToField.ToDuckValue v) => Database.DuckDB.Simple.ToField.ToDuckValue (Data.Map.Internal.Map k v)
+ Database.DuckDB.Simple.ToField: instance (Database.DuckDB.Simple.ToField.DuckDBColumnType k, Database.DuckDB.Simple.ToField.DuckDBColumnType v, Database.DuckDB.Simple.ToField.ToField k, Database.DuckDB.Simple.ToField.ToField v) => Database.DuckDB.Simple.ToField.ToField (Data.Map.Internal.Map k v)
+ Database.DuckDB.Simple.ToField: instance (GHC.Internal.TypeNats.KnownNat n, Database.DuckDB.Simple.ToField.DuckDBColumnType a) => Database.DuckDB.Simple.ToField.DuckDBColumnType (Database.DuckDB.Simple.FromField.SizedArray n a)
+ Database.DuckDB.Simple.ToField: instance (GHC.Internal.TypeNats.KnownNat n, Database.DuckDB.Simple.ToField.DuckDBColumnType a, Database.DuckDB.Simple.ToField.ToDuckValue a) => Database.DuckDB.Simple.ToField.ToDuckValue (Database.DuckDB.Simple.FromField.SizedArray n a)
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.DuckDBColumnType (Database.DuckDB.Simple.LogicalRep.TupleValue Database.DuckDB.Simple.FromField.FieldValue)
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.DuckDBColumnType Database.DuckDB.Simple.FromField.IntervalValue
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.DuckDBColumnType Database.DuckDB.Simple.FromField.TimeWithZone
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.DuckDBColumnType a => Database.DuckDB.Simple.ToField.DuckDBColumnType [a]
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToDuckValue (Database.DuckDB.Simple.LogicalRep.TupleValue Database.DuckDB.Simple.FromField.FieldValue)
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToDuckValue Database.DuckDB.Simple.FromField.IntervalValue
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToDuckValue Database.DuckDB.Simple.FromField.TimeWithZone
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToDuckValue Database.DuckDB.Simple.Variant.Variant
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToField (Database.DuckDB.Simple.LogicalRep.TupleValue Database.DuckDB.Simple.FromField.FieldValue)
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToField Database.DuckDB.Simple.FromField.DecimalValue
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToField Database.DuckDB.Simple.FromField.FieldValue
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToField Database.DuckDB.Simple.FromField.IntervalValue
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToField Database.DuckDB.Simple.FromField.TimeWithZone
+ Database.DuckDB.Simple.ToField: instance Database.DuckDB.Simple.ToField.ToField Database.DuckDB.Simple.Geometry.RawGeometry
+ Database.DuckDB.Simple.ToField: type NativeValue = DuckDBV2ValueHandle
+ Database.DuckDB.Simple.ToField: type ValueFactory = DuckDBV2FactoryHandle
+ Database.DuckDB.Simple.Tokenizer: Token :: TokenType -> Int -> ByteString -> Token
+ Database.DuckDB.Simple.Tokenizer: TokenTypeComment :: TokenType
+ Database.DuckDB.Simple.Tokenizer: TokenTypeEndOfInput :: TokenType
+ Database.DuckDB.Simple.Tokenizer: TokenTypeIdentifier :: TokenType
+ Database.DuckDB.Simple.Tokenizer: TokenTypeInvalid :: TokenType
+ Database.DuckDB.Simple.Tokenizer: TokenTypeKeyword :: TokenType
+ Database.DuckDB.Simple.Tokenizer: TokenTypeNumberLiteral :: TokenType
+ Database.DuckDB.Simple.Tokenizer: TokenTypeOperator :: TokenType
+ Database.DuckDB.Simple.Tokenizer: TokenTypeStringLiteral :: TokenType
+ Database.DuckDB.Simple.Tokenizer: TokenTypeTerminator :: TokenType
+ Database.DuckDB.Simple.Tokenizer: Tokenization :: [Token] -> Bool -> Tokenization
+ Database.DuckDB.Simple.Tokenizer: [tokenBytes] :: Token -> ByteString
+ Database.DuckDB.Simple.Tokenizer: [tokenOffset] :: Token -> Int
+ Database.DuckDB.Simple.Tokenizer: [tokenType] :: Token -> TokenType
+ Database.DuckDB.Simple.Tokenizer: [tokenizationTokens] :: Tokenization -> [Token]
+ Database.DuckDB.Simple.Tokenizer: [tokenizationUnterminated] :: Tokenization -> Bool
+ Database.DuckDB.Simple.Tokenizer: data Token
+ Database.DuckDB.Simple.Tokenizer: data TokenType
+ Database.DuckDB.Simple.Tokenizer: data Tokenization
+ Database.DuckDB.Simple.Tokenizer: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Tokenizer.Token
+ Database.DuckDB.Simple.Tokenizer: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.Tokenizer.Tokenization
+ Database.DuckDB.Simple.Tokenizer: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Tokenizer.Token
+ Database.DuckDB.Simple.Tokenizer: instance GHC.Internal.Show.Show Database.DuckDB.Simple.Tokenizer.Tokenization
+ Database.DuckDB.Simple.Tokenizer: tokenize :: Connection -> Text -> IO Tokenization
+ Database.DuckDB.Simple.Types: ErrorApi :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorConfigurationInvalid :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorConfigurationPermission :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorConfigurationSettings :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorDatabaseCatalog :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorDatabaseConstraint :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorDatabaseIndex :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorDatabaseSequence :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorDatabaseSerialization :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorDatabaseStatistics :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorDatabaseTransaction :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorIOEof :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorIOFileNotFound :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorIOGeneral :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorIOHTTP :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorIONetwork :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorIOReadFailure :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorInputInvalid :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorInputMetadataDeferred :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorInputObjectSize :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorInputOutOfRange :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorInputParameterInvalid :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryBinder :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryExecutor :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryExpression :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryNotImplemented :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryOptimizer :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryParameterNotAllowed :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryParameterNotResolved :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryParser :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryPlanner :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQueryScheduler :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorQuerySyntax :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorResourceAutoload :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorResourceConnection :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorResourceDependency :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorResourceInUse :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorResourceMissingExtension :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorResourceOutOfMemory :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorRuntimeFatal :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorRuntimeInternal :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorRuntimeInterrupt :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorRuntimeNullPointer :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorTypeConversion :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorTypeDecimal :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorTypeDivideByZero :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorTypeInvalid :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorTypeMismatch :: ErrorType
+ Database.DuckDB.Simple.Types: ErrorTypeUnknown :: ErrorType
+ Database.DuckDB.Simple.Types: [sqlErrorBody] :: SQLError -> Maybe Text
+ Database.DuckDB.Simple.Types: data ErrorType
+ Database.DuckDB.Simple.V2.Config: registerCustomOptionWith :: RegistrationTarget -> (DuckDBV2CustomOptionHandle -> IO ()) -> Maybe (ContextCallback DuckDBV2CustomOptionUpdateInfoHandle) -> IO ()
+ Database.DuckDB.Simple.V2.Copy: ConnectionTarget :: Connection -> RegistrationTarget
+ Database.DuckDB.Simple.V2.Copy: CopyFromCallbacks :: ContextCallback DuckDBV2CopyFromBindInfoHandle -> Maybe (ContextCallback DuckDBV2CopyFromInitGlobalInfoHandle) -> Maybe (ContextCallback DuckDBV2CopyFromInitLocalInfoHandle) -> ContextCallback DuckDBV2CopyFromExecInfoHandle -> Maybe (ContextCallback DuckDBV2CopyFromProgressInfoHandle) -> CopyFromCallbacks
+ Database.DuckDB.Simple.V2.Copy: CopyToCallbacks :: Maybe (ContextCallback DuckDBV2CopyToBindInfoHandle) -> Maybe (ContextCallback DuckDBV2CopyToBatchSizeInfoHandle) -> Maybe (ContextCallback DuckDBV2CopyToInitInfoHandle) -> ContextCallback DuckDBV2CopyToBatchInfoHandle -> ContextCallback DuckDBV2CopyToFlushInfoHandle -> Maybe (ContextCallback DuckDBV2CopyToFinalizeInfoHandle) -> Maybe (ContextCallback DuckDBV2CopyToStatisticsInfoHandle) -> CopyToCallbacks
+ Database.DuckDB.Simple.V2.Copy: ExtensionTarget :: DuckDBV2ExtensionHandle -> DuckDBV2ContextHandle -> RegistrationTarget
+ Database.DuckDB.Simple.V2.Copy: [copyFromBindCallback] :: CopyFromCallbacks -> ContextCallback DuckDBV2CopyFromBindInfoHandle
+ Database.DuckDB.Simple.V2.Copy: [copyFromExecCallback] :: CopyFromCallbacks -> ContextCallback DuckDBV2CopyFromExecInfoHandle
+ Database.DuckDB.Simple.V2.Copy: [copyFromInitGlobalCallback] :: CopyFromCallbacks -> Maybe (ContextCallback DuckDBV2CopyFromInitGlobalInfoHandle)
+ Database.DuckDB.Simple.V2.Copy: [copyFromInitLocalCallback] :: CopyFromCallbacks -> Maybe (ContextCallback DuckDBV2CopyFromInitLocalInfoHandle)
+ Database.DuckDB.Simple.V2.Copy: [copyFromProgressCallback] :: CopyFromCallbacks -> Maybe (ContextCallback DuckDBV2CopyFromProgressInfoHandle)
+ Database.DuckDB.Simple.V2.Copy: [copyToBatchCallback] :: CopyToCallbacks -> ContextCallback DuckDBV2CopyToBatchInfoHandle
+ Database.DuckDB.Simple.V2.Copy: [copyToBatchSizeCallback] :: CopyToCallbacks -> Maybe (ContextCallback DuckDBV2CopyToBatchSizeInfoHandle)
+ Database.DuckDB.Simple.V2.Copy: [copyToBindCallback] :: CopyToCallbacks -> Maybe (ContextCallback DuckDBV2CopyToBindInfoHandle)
+ Database.DuckDB.Simple.V2.Copy: [copyToFinalizeCallback] :: CopyToCallbacks -> Maybe (ContextCallback DuckDBV2CopyToFinalizeInfoHandle)
+ Database.DuckDB.Simple.V2.Copy: [copyToFlushCallback] :: CopyToCallbacks -> ContextCallback DuckDBV2CopyToFlushInfoHandle
+ Database.DuckDB.Simple.V2.Copy: [copyToInitCallback] :: CopyToCallbacks -> Maybe (ContextCallback DuckDBV2CopyToInitInfoHandle)
+ Database.DuckDB.Simple.V2.Copy: [copyToStatisticsCallback] :: CopyToCallbacks -> Maybe (ContextCallback DuckDBV2CopyToStatisticsInfoHandle)
+ Database.DuckDB.Simple.V2.Copy: copyFromCallbacks :: ContextCallback DuckDBV2CopyFromBindInfoHandle -> ContextCallback DuckDBV2CopyFromExecInfoHandle -> CopyFromCallbacks
+ Database.DuckDB.Simple.V2.Copy: copyToCallbacks :: ContextCallback DuckDBV2CopyToBatchInfoHandle -> ContextCallback DuckDBV2CopyToFlushInfoHandle -> CopyToCallbacks
+ Database.DuckDB.Simple.V2.Copy: data CopyFromCallbacks
+ Database.DuckDB.Simple.V2.Copy: data CopyToCallbacks
+ Database.DuckDB.Simple.V2.Copy: data RegistrationTarget
+ Database.DuckDB.Simple.V2.Copy: getCallbackState :: Typeable a => (Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a
+ Database.DuckDB.Simple.V2.Copy: registerCopyFromCallbacks :: Connection -> Text -> CopyFromCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Copy: registerCopyFunction :: Connection -> Text -> Maybe CopyToCallbacks -> Maybe CopyFromCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Copy: registerCopyFunctionWith :: RegistrationTarget -> (DuckDBV2CopyFunctionHandle -> IO ()) -> Maybe CopyToCallbacks -> Maybe CopyFromCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Copy: registerCopyToCallbacks :: Connection -> Text -> CopyToCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Copy: registerMultiFileFunctionWith :: RegistrationTarget -> (DuckDBV2MultiFileFunctionHandle -> IO ()) -> IO ()
+ Database.DuckDB.Simple.V2.Copy: setCallbackState :: Typeable a => (Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> a -> IO ()
+ Database.DuckDB.Simple.V2.Copy: type ContextCallback info = info -> DuckDBV2ContextHandle -> IO ()
+ Database.DuckDB.Simple.V2.FunctionDocs: FunctionDocs :: Maybe Text -> [(Text, Text)] -> [Text] -> FunctionDocs
+ Database.DuckDB.Simple.V2.FunctionDocs: [functionCategories] :: FunctionDocs -> [Text]
+ Database.DuckDB.Simple.V2.FunctionDocs: [functionDescription] :: FunctionDocs -> Maybe Text
+ Database.DuckDB.Simple.V2.FunctionDocs: [functionExamples] :: FunctionDocs -> [(Text, Text)]
+ Database.DuckDB.Simple.V2.FunctionDocs: data FunctionDocs
+ Database.DuckDB.Simple.V2.FunctionDocs: defaultFunctionDocs :: FunctionDocs
+ Database.DuckDB.Simple.V2.FunctionDocs: instance GHC.Internal.Classes.Eq Database.DuckDB.Simple.V2.FunctionDocs.FunctionDocs
+ Database.DuckDB.Simple.V2.FunctionDocs: instance GHC.Internal.Read.Read Database.DuckDB.Simple.V2.FunctionDocs.FunctionDocs
+ Database.DuckDB.Simple.V2.FunctionDocs: instance GHC.Internal.Show.Show Database.DuckDB.Simple.V2.FunctionDocs.FunctionDocs
+ Database.DuckDB.Simple.V2.FunctionDocs: setFunctionDocs :: DuckDBV2FunctionDocsHandle -> FunctionDocs -> IO ()
+ Database.DuckDB.Simple.V2.Logging: logMessage :: DuckDBV2ContextHandle -> LogLevel -> Text -> Text -> IO ()
+ Database.DuckDB.Simple.V2.LogicalRep: bindingLogicalType :: DuckDBV2FactoryHandle -> DuckDBV2InstanceHandle -> LogicalTypeRep -> IO DuckDBV2LogicalTypeHandle
+ Database.DuckDB.Simple.V2.LogicalRep: destroyLogicalType :: DuckDBV2LogicalTypeHandle -> IO ()
+ Database.DuckDB.Simple.V2.LogicalRep: logicalTypeFromRep :: DuckDBV2FactoryHandle -> LogicalTypeRep -> IO DuckDBV2LogicalTypeHandle
+ Database.DuckDB.Simple.V2.LogicalRep: logicalTypeFromRepWith :: DuckDBV2FactoryHandle -> (LogicalTypeRep -> IO (Maybe DuckDBV2LogicalTypeHandle)) -> LogicalTypeRep -> IO DuckDBV2LogicalTypeHandle
+ Database.DuckDB.Simple.V2.LogicalRep: logicalTypeToBaseRep :: DuckDBV2LogicalTypeHandle -> IO LogicalTypeRep
+ Database.DuckDB.Simple.V2.LogicalRep: logicalTypeToRep :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> IO LogicalTypeRep
+ Database.DuckDB.Simple.V2.LogicalRep: valuesHaveType :: DuckDBV2LogicalTypeHandle -> [DuckDBV2ValueHandle] -> IO Bool
+ Database.DuckDB.Simple.V2.Pragma: ExecPragma :: ContextCallback DuckDBV2FunctionBindInfoHandle -> PragmaCallback
+ Database.DuckDB.Simple.V2.Pragma: QueryPragma :: (DuckDBV2FunctionBindInfoHandle -> DuckDBV2ContextHandle -> IO Text) -> PragmaCallback
+ Database.DuckDB.Simple.V2.Pragma: data PragmaCallback
+ Database.DuckDB.Simple.V2.Pragma: registerPragmaFunctionOn :: RegistrationTarget -> Text -> [FunctionParameter] -> PragmaCallback -> IO ()
+ Database.DuckDB.Simple.V2.Pragma: registerPragmaFunctionWith :: RegistrationTarget -> (DuckDBV2PragmaFunctionHandle -> IO ()) -> PragmaCallback -> IO ()
+ Database.DuckDB.Simple.V2.Registration: AggregateCallbacks :: Maybe (BindCallback DuckDBV2AggregateFunctionBindInfoHandle) -> (DuckDBV2AggregateFunctionSizeInfoHandle -> IO ()) -> (DuckDBV2AggregateFunctionInitInfoHandle -> IO ()) -> (DuckDBV2AggregateFunctionUpdateInfoHandle -> IO ()) -> (DuckDBV2AggregateFunctionCombineInfoHandle -> IO ()) -> (DuckDBV2AggregateFunctionFinalizeInfoHandle -> IO ()) -> Maybe (DuckDBV2AggregateFunctionDestroyInfoHandle -> IO ()) -> AggregateCallbacks
+ Database.DuckDB.Simple.V2.Registration: ConnectionTarget :: Connection -> RegistrationTarget
+ Database.DuckDB.Simple.V2.Registration: DefaultValue :: (DuckDBV2FactoryHandle -> IO DuckDBV2ValueHandle) -> DefaultValue
+ Database.DuckDB.Simple.V2.Registration: ExtensionTarget :: DuckDBV2ExtensionHandle -> DuckDBV2ContextHandle -> RegistrationTarget
+ Database.DuckDB.Simple.V2.Registration: FunctionParameter :: Text -> TypeDeclaration -> Maybe DefaultValue -> DuckDBV2FunctionParameterKind -> FunctionParameter
+ Database.DuckDB.Simple.V2.Registration: FunctionSignature :: [FunctionParameter] -> Maybe TypeDeclaration -> FunctionSignature
+ Database.DuckDB.Simple.V2.Registration: ScalarCallbacks :: Maybe (BindCallback DuckDBV2ScalarFunctionResolveTypesInfoHandle) -> Maybe (BindCallback DuckDBV2ScalarFunctionBindInfoHandle) -> Maybe (ContextCallback DuckDBV2ScalarFunctionInitInfoHandle) -> ContextCallback DuckDBV2ScalarFunctionExecInfoHandle -> ScalarCallbacks
+ Database.DuckDB.Simple.V2.Registration: StructuredType :: LogicalTypeRep -> TypeDeclaration
+ Database.DuckDB.Simple.V2.Registration: TableCallbacks :: BindCallback DuckDBV2TableFunctionBindInfoHandle -> Maybe (ContextCallback DuckDBV2TableFunctionInitGlobalInfoHandle) -> Maybe (ContextCallback DuckDBV2TableFunctionInitLocalInfoHandle) -> ContextCallback DuckDBV2TableFunctionExecInfoHandle -> Maybe (ContextCallback DuckDBV2TableFunctionProgressInfoHandle) -> Maybe (ContextCallback DuckDBV2TableFunctionFilterPushdownInfoHandle) -> Maybe (ContextCallback DuckDBV2TableFunctionPartitionDataInfoHandle) -> Maybe (ContextCallback DuckDBV2TableFunctionPartitioningInfoHandle) -> Maybe (ContextCallback DuckDBV2TableFunctionClaimBatchInfoHandle) -> Maybe (ContextCallback DuckDBV2TableFunctionGetBindInfoInfoHandle) -> Bool -> TableCallbacks
+ Database.DuckDB.Simple.V2.Registration: TypeExpression :: Text -> TypeDeclaration
+ Database.DuckDB.Simple.V2.Registration: [aggregateBindCallback] :: AggregateCallbacks -> Maybe (BindCallback DuckDBV2AggregateFunctionBindInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [aggregateCombineCallback] :: AggregateCallbacks -> DuckDBV2AggregateFunctionCombineInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: [aggregateDestroyCallback] :: AggregateCallbacks -> Maybe (DuckDBV2AggregateFunctionDestroyInfoHandle -> IO ())
+ Database.DuckDB.Simple.V2.Registration: [aggregateFinalizeCallback] :: AggregateCallbacks -> DuckDBV2AggregateFunctionFinalizeInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: [aggregateInitCallback] :: AggregateCallbacks -> DuckDBV2AggregateFunctionInitInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: [aggregateSizeCallback] :: AggregateCallbacks -> DuckDBV2AggregateFunctionSizeInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: [aggregateUpdateCallback] :: AggregateCallbacks -> DuckDBV2AggregateFunctionUpdateInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: [parameterDefault] :: FunctionParameter -> Maybe DefaultValue
+ Database.DuckDB.Simple.V2.Registration: [parameterKind] :: FunctionParameter -> DuckDBV2FunctionParameterKind
+ Database.DuckDB.Simple.V2.Registration: [parameterName] :: FunctionParameter -> Text
+ Database.DuckDB.Simple.V2.Registration: [parameterType] :: FunctionParameter -> TypeDeclaration
+ Database.DuckDB.Simple.V2.Registration: [scalarBindCallback] :: ScalarCallbacks -> Maybe (BindCallback DuckDBV2ScalarFunctionBindInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [scalarExecCallback] :: ScalarCallbacks -> ContextCallback DuckDBV2ScalarFunctionExecInfoHandle
+ Database.DuckDB.Simple.V2.Registration: [scalarInitCallback] :: ScalarCallbacks -> Maybe (ContextCallback DuckDBV2ScalarFunctionInitInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [scalarResolveTypesCallback] :: ScalarCallbacks -> Maybe (BindCallback DuckDBV2ScalarFunctionResolveTypesInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [signatureParameters] :: FunctionSignature -> [FunctionParameter]
+ Database.DuckDB.Simple.V2.Registration: [signatureReturnType] :: FunctionSignature -> Maybe TypeDeclaration
+ Database.DuckDB.Simple.V2.Registration: [tableBindCallback] :: TableCallbacks -> BindCallback DuckDBV2TableFunctionBindInfoHandle
+ Database.DuckDB.Simple.V2.Registration: [tableClaimBatchCallback] :: TableCallbacks -> Maybe (ContextCallback DuckDBV2TableFunctionClaimBatchInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [tableExecCallback] :: TableCallbacks -> ContextCallback DuckDBV2TableFunctionExecInfoHandle
+ Database.DuckDB.Simple.V2.Registration: [tableFilterPushdownCallback] :: TableCallbacks -> Maybe (ContextCallback DuckDBV2TableFunctionFilterPushdownInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [tableGetBindInfoCallback] :: TableCallbacks -> Maybe (ContextCallback DuckDBV2TableFunctionGetBindInfoInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [tableInitGlobalCallback] :: TableCallbacks -> Maybe (ContextCallback DuckDBV2TableFunctionInitGlobalInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [tableInitLocalCallback] :: TableCallbacks -> Maybe (ContextCallback DuckDBV2TableFunctionInitLocalInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [tablePartitionDataCallback] :: TableCallbacks -> Maybe (ContextCallback DuckDBV2TableFunctionPartitionDataInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [tablePartitioningCallback] :: TableCallbacks -> Maybe (ContextCallback DuckDBV2TableFunctionPartitioningInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [tableProgressCallback] :: TableCallbacks -> Maybe (ContextCallback DuckDBV2TableFunctionProgressInfoHandle)
+ Database.DuckDB.Simple.V2.Registration: [tableProjectionPushdown] :: TableCallbacks -> Bool
+ Database.DuckDB.Simple.V2.Registration: aggregateCallbacks :: (DuckDBV2AggregateFunctionSizeInfoHandle -> IO ()) -> (DuckDBV2AggregateFunctionInitInfoHandle -> IO ()) -> (DuckDBV2AggregateFunctionUpdateInfoHandle -> IO ()) -> (DuckDBV2AggregateFunctionCombineInfoHandle -> IO ()) -> (DuckDBV2AggregateFunctionFinalizeInfoHandle -> IO ()) -> AggregateCallbacks
+ Database.DuckDB.Simple.V2.Registration: checked :: (Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO ()
+ Database.DuckDB.Simple.V2.Registration: configureFunctionSignature :: DuckDBV2FactoryHandle -> FunctionSignature -> DuckDBV2FunctionSignatureHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: data AggregateCallbacks
+ Database.DuckDB.Simple.V2.Registration: data FunctionParameter
+ Database.DuckDB.Simple.V2.Registration: data FunctionSignature
+ Database.DuckDB.Simple.V2.Registration: data RegistrationTarget
+ Database.DuckDB.Simple.V2.Registration: data ScalarCallbacks
+ Database.DuckDB.Simple.V2.Registration: data TableCallbacks
+ Database.DuckDB.Simple.V2.Registration: data TypeDeclaration
+ Database.DuckDB.Simple.V2.Registration: defaultValue :: ToDuckValue a => a -> DefaultValue
+ Database.DuckDB.Simple.V2.Registration: destroy :: Storable a => (Ptr a -> IO DuckDBV2Error) -> a -> IO ()
+ Database.DuckDB.Simple.V2.Registration: getCallbackState :: Typeable a => (Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a
+ Database.DuckDB.Simple.V2.Registration: newtype DefaultValue
+ Database.DuckDB.Simple.V2.Registration: nullHandle :: Coercible (Ptr ()) a => a
+ Database.DuckDB.Simple.V2.Registration: output :: Storable a => (Ptr a -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a
+ Database.DuckDB.Simple.V2.Registration: parameter :: Text -> TypeDeclaration -> FunctionParameter
+ Database.DuckDB.Simple.V2.Registration: readIdentifier :: DuckDBV2IdentifierT -> IO Text
+ Database.DuckDB.Simple.V2.Registration: readText :: DuckDBV2Str -> IO Text
+ Database.DuckDB.Simple.V2.Registration: registerAggregateFunction :: Connection -> Text -> FunctionSignature -> FunctionProperties -> AggregateCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerAggregateFunctionWith :: RegistrationTarget -> (DuckDBV2AggregateFunctionHandle -> IO ()) -> AggregateCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerAggregateFunctionWithCleanup :: RegistrationTarget -> IO () -> (DuckDBV2AggregateFunctionHandle -> IO ()) -> AggregateCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerCast :: Connection -> TypeDeclaration -> TypeDeclaration -> Maybe Int64 -> ContextCallback DuckDBV2CastFunctionExecInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerCastWith :: RegistrationTarget -> (DuckDBV2CastFunctionHandle -> IO ()) -> ContextCallback DuckDBV2CastFunctionExecInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerCustomType :: Connection -> Text -> TypeDeclaration -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerCustomTypeOn :: RegistrationTarget -> Text -> TypeDeclaration -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerCustomTypeWith :: RegistrationTarget -> (DuckDBV2CustomTypeHandle -> IO ()) -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerInstanceReplacementScan :: Instance -> ContextCallback DuckDBV2ReplacementScanInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerMacroFunctionWith :: RegistrationTarget -> MacroFunctionType -> (DuckDBV2MacroFunctionHandle -> IO ()) -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerReplacementScan :: Connection -> ContextCallback DuckDBV2ReplacementScanInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerReplacementScanOn :: RegistrationTarget -> ContextCallback DuckDBV2ReplacementScanInfoHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerScalarFunction :: Connection -> Text -> FunctionSignature -> FunctionProperties -> ScalarCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerScalarFunctionWith :: RegistrationTarget -> (DuckDBV2ScalarFunctionHandle -> IO ()) -> ScalarCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerTableFunction :: Connection -> Text -> [FunctionParameter] -> TableCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Registration: registerTableFunctionWith :: RegistrationTarget -> (DuckDBV2TableFunctionHandle -> IO ()) -> TableCallbacks -> IO ()
+ Database.DuckDB.Simple.V2.Registration: scalarCallbacks :: ContextCallback DuckDBV2ScalarFunctionExecInfoHandle -> ScalarCallbacks
+ Database.DuckDB.Simple.V2.Registration: setCallbackState :: Typeable a => (Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> a -> IO ()
+ Database.DuckDB.Simple.V2.Registration: tableCallbacks :: BindCallback DuckDBV2TableFunctionBindInfoHandle -> ContextCallback DuckDBV2TableFunctionExecInfoHandle -> TableCallbacks
+ Database.DuckDB.Simple.V2.Registration: type BindCallback result = DuckDBV2FunctionBindInfoHandle -> result -> DuckDBV2ContextHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: type ContextCallback info = info -> DuckDBV2ContextHandle -> IO ()
+ Database.DuckDB.Simple.V2.Registration: type FunctionProperties = [(DuckDBV2FunctionPropertyKey, DuckDBV2FunctionPropertyValue)]
+ Database.DuckDB.Simple.V2.Registration: withIdentifier :: Text -> (ConstPtr DuckDBV2IdentifierT -> IO a) -> IO a
+ Database.DuckDB.Simple.V2.Registration: withRegistrationFactory :: RegistrationTarget -> (DuckDBV2FactoryHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.V2.Registration: withText :: Text -> (ConstPtr DuckDBV2Str -> IO a) -> IO a
+ Database.DuckDB.Simple.V2.Registration: withTypeDeclaration :: DuckDBV2FactoryHandle -> TypeDeclaration -> (DuckDBV2LogicalTypeHandle -> IO a) -> IO a
- Database.DuckDB.Simple: ConnectionOptions :: [(Text, Text)] -> [Text] -> ConnectionOptions
+ Database.DuckDB.Simple: ConnectionOptions :: [(Text, Text)] -> Maybe Int -> Int -> Int -> ConnectionOptions
- Database.DuckDB.Simple: SQLError :: Text -> Maybe DuckDBErrorType -> Maybe Query -> SQLError
+ Database.DuckDB.Simple: SQLError :: Text -> Maybe ErrorType -> Maybe Query -> Maybe Text -> SQLError
- Database.DuckDB.Simple: [sqlErrorType] :: SQLError -> Maybe DuckDBErrorType
+ Database.DuckDB.Simple: [sqlErrorType] :: SQLError -> Maybe ErrorType
- Database.DuckDB.Simple: argumentTypes :: Function a => Proxy a -> [DuckDBType]
+ Database.DuckDB.Simple: argumentTypes :: Function a => Proxy a -> [Either Text TypeDeclaration]
- Database.DuckDB.Simple: createFunctionWithState :: forall s f. Function f => Connection -> Text -> IO s -> (s -> f) -> IO ()
+ Database.DuckDB.Simple: createFunctionWithState :: Function f => Connection -> Text -> IO s -> (s -> f) -> IO ()
- Database.DuckDB.Simple: returnType :: Function a => Proxy a -> ScalarType
+ Database.DuckDB.Simple: returnType :: Function a => Proxy a -> Either Text TypeDeclaration
- Database.DuckDB.Simple: type FieldParser a = Field -> Ok a
+ Database.DuckDB.Simple: type FieldParser a = Field -> Conversion a
- Database.DuckDB.Simple.Catalog: CatalogEntry :: Text -> DuckDBCatalogEntryType -> CatalogEntry
+ Database.DuckDB.Simple.Catalog: CatalogEntry :: Text -> CatalogEntryType -> CatalogEntry
- Database.DuckDB.Simple.Catalog: [catalogEntryType] :: CatalogEntry -> DuckDBCatalogEntryType
+ Database.DuckDB.Simple.Catalog: [catalogEntryType] :: CatalogEntry -> CatalogEntryType
- Database.DuckDB.Simple.Catalog: lookupCatalogEntry :: Connection -> Text -> Text -> Text -> DuckDBCatalogEntryType -> IO (Maybe CatalogEntry)
+ Database.DuckDB.Simple.Catalog: lookupCatalogEntry :: Connection -> Text -> Text -> Text -> CatalogEntryType -> IO (Maybe CatalogEntry)
- Database.DuckDB.Simple.Config: ConfigValue :: Text -> Maybe DuckDBConfigOptionScope -> ConfigValue
+ Database.DuckDB.Simple.Config: ConfigValue :: Maybe Text -> SettingScope -> ConfigValue
- Database.DuckDB.Simple.Config: [configValueScope] :: ConfigValue -> Maybe DuckDBConfigOptionScope
+ Database.DuckDB.Simple.Config: [configValueScope] :: ConfigValue -> SettingScope
- Database.DuckDB.Simple.Config: [configValueText] :: ConfigValue -> Text
+ Database.DuckDB.Simple.Config: [configValueText] :: ConfigValue -> Maybe Text
- Database.DuckDB.Simple.Config: listConfigFlags :: IO [ConfigFlag]
+ Database.DuckDB.Simple.Config: listConfigFlags :: Connection -> IO [ConfigFlag]
- Database.DuckDB.Simple.Copy: CopyBindInfo :: [DuckDBType] -> CopyBindInfo
+ Database.DuckDB.Simple.Copy: CopyBindInfo :: [LogicalTypeId] -> CopyBindInfo
- Database.DuckDB.Simple.Copy: [copyBindColumnTypes] :: CopyBindInfo -> [DuckDBType]
+ Database.DuckDB.Simple.Copy: [copyBindColumnTypes] :: CopyBindInfo -> [LogicalTypeId]
- Database.DuckDB.Simple.FileSystem: fileHandleSeek :: DuckDBFileHandle -> Int64 -> IO ()
+ Database.DuckDB.Simple.FileSystem: fileHandleSeek :: FileHandle -> Int64 -> IO ()
- Database.DuckDB.Simple.FileSystem: fileHandleSize :: DuckDBFileHandle -> IO Int64
+ Database.DuckDB.Simple.FileSystem: fileHandleSize :: FileHandle -> IO Int64
- Database.DuckDB.Simple.FileSystem: fileHandleSync :: DuckDBFileHandle -> IO ()
+ Database.DuckDB.Simple.FileSystem: fileHandleSync :: FileHandle -> IO ()
- Database.DuckDB.Simple.FileSystem: fileHandleTell :: DuckDBFileHandle -> IO Int64
+ Database.DuckDB.Simple.FileSystem: fileHandleTell :: FileHandle -> IO Int64
- Database.DuckDB.Simple.FileSystem: readFileHandleChunk :: DuckDBFileHandle -> Int64 -> IO ByteString
+ Database.DuckDB.Simple.FileSystem: readFileHandleChunk :: FileHandle -> Int64 -> IO ByteString
- Database.DuckDB.Simple.FileSystem: withFileHandle :: Connection -> FilePath -> [DuckDBFileFlag] -> (DuckDBFileHandle -> IO a) -> IO a
+ Database.DuckDB.Simple.FileSystem: withFileHandle :: Connection -> FilePath -> [FileFlag] -> (FileHandle -> IO a) -> IO a
- Database.DuckDB.Simple.FileSystem: writeFileHandleBytes :: DuckDBFileHandle -> ByteString -> IO Int64
+ Database.DuckDB.Simple.FileSystem: writeFileHandleBytes :: FileHandle -> ByteString -> IO Int64
- Database.DuckDB.Simple.FromField: Field :: Text -> Int -> FieldValue -> Field
+ Database.DuckDB.Simple.FromField: Field :: Text -> Int -> FieldValue -> Conversion LogicalTypeRep -> Field
- Database.DuckDB.Simple.FromField: FieldEnum :: Word32 -> FieldValue
+ Database.DuckDB.Simple.FromField: FieldEnum :: Word32 -> Text -> FieldValue
- Database.DuckDB.Simple.FromField: LogicalTypeScalar :: DuckDBType -> LogicalTypeRep
+ Database.DuckDB.Simple.FromField: LogicalTypeScalar :: LogicalTypeId -> LogicalTypeRep
- Database.DuckDB.Simple.FromField: TimeWithZone :: TimeOfDay -> TimeZone -> TimeWithZone
+ Database.DuckDB.Simple.FromField: TimeWithZone :: TimeOfDay -> Int32 -> TimeWithZone
- Database.DuckDB.Simple.FromField: returnError :: Typeable b => (Text -> Text -> Text -> Text -> ResultError) -> Field -> Text -> Ok b
+ Database.DuckDB.Simple.FromField: returnError :: Typeable b => (Text -> Text -> Text -> Text -> ResultError) -> Field -> Text -> Conversion b
- Database.DuckDB.Simple.FromField: type FieldParser a = Field -> Ok a
+ Database.DuckDB.Simple.FromField: type FieldParser a = Field -> Conversion a
- Database.DuckDB.Simple.FromRow: RowParser :: ReaderT RowParseRO (StateT (Int, [Field]) Ok) a -> RowParser a
+ Database.DuckDB.Simple.FromRow: RowParser :: ReaderT RowParseRO (StateT (Int, [Field]) Conversion) a -> RowParser a
- Database.DuckDB.Simple.FromRow: [runRowParser] :: RowParser a -> ReaderT RowParseRO (StateT (Int, [Field]) Ok) a
+ Database.DuckDB.Simple.FromRow: [runRowParser] :: RowParser a -> ReaderT RowParseRO (StateT (Int, [Field]) Conversion) a
- Database.DuckDB.Simple.FromRow: parseRow :: RowParser a -> [Field] -> Ok a
+ Database.DuckDB.Simple.FromRow: parseRow :: RowParser a -> [Field] -> Conversion a
- Database.DuckDB.Simple.Function: argumentTypes :: Function a => Proxy a -> [DuckDBType]
+ Database.DuckDB.Simple.Function: argumentTypes :: Function a => Proxy a -> [Either Text TypeDeclaration]
- Database.DuckDB.Simple.Function: class FunctionResult a
+ Database.DuckDB.Simple.Function: class ToDuckValue a => FunctionResult a
- Database.DuckDB.Simple.Function: createFunctionWithState :: forall s f. Function f => Connection -> Text -> IO s -> (s -> f) -> IO ()
+ Database.DuckDB.Simple.Function: createFunctionWithState :: Function f => Connection -> Text -> IO s -> (s -> f) -> IO ()
- Database.DuckDB.Simple.Function: returnType :: Function a => Proxy a -> ScalarType
+ Database.DuckDB.Simple.Function: returnType :: Function a => Proxy a -> Either Text TypeDeclaration
- Database.DuckDB.Simple.Generic: ($dmduckFromField) :: (DuckValue a, FromField a, Show a) => FieldValue -> Either String a
+ Database.DuckDB.Simple.Generic: ($dmduckFromField) :: (DuckValue a, FromField a) => FieldValue -> Conversion a
- Database.DuckDB.Simple.Generic: duckFromField :: DuckValue a => FieldValue -> Either String a
+ Database.DuckDB.Simple.Generic: duckFromField :: DuckValue a => FieldValue -> Conversion a
- Database.DuckDB.Simple.Generic: genericFromFieldValue :: (Generic a, GFromField (Rep a)) => FieldValue -> Either String a
+ Database.DuckDB.Simple.Generic: genericFromFieldValue :: (Generic a, GFromField (Rep a)) => FieldValue -> Conversion a
- Database.DuckDB.Simple.Internal: Connection :: IORef ConnectionState -> Connection
+ Database.DuckDB.Simple.Internal: Connection :: IORef ConnectionState -> Instance -> MVar () -> IORef (Maybe ThreadId) -> ConnectionOptions -> Connection
- Database.DuckDB.Simple.Internal: ConnectionOpen :: DuckDBDatabase -> DuckDBConnection -> TypeCache -> ConnectionState
+ Database.DuckDB.Simple.Internal: ConnectionOpen :: DuckDBV2ConnectionHandle -> MVar (Maybe VariantState) -> Children StatementState -> IO () -> ConnectionState
- Database.DuckDB.Simple.Internal: SQLError :: Text -> Maybe DuckDBErrorType -> Maybe Query -> SQLError
+ Database.DuckDB.Simple.Internal: SQLError :: Text -> Maybe ErrorType -> Maybe Query -> Maybe Text -> SQLError
- Database.DuckDB.Simple.Internal: Statement :: IORef StatementState -> Connection -> Query -> IORef StatementStreamState -> Statement
+ Database.DuckDB.Simple.Internal: Statement :: IORef StatementState -> Connection -> Query -> IORef StatementStreamState -> IORef DuckDBV2ArrowResultHandle -> IORef (IntMap DuckDBV2ValueHandle) -> Array Int Text -> IORef (Maybe (IntMap (Maybe LogicalTypeRep))) -> Statement
- Database.DuckDB.Simple.Internal: StatementOpen :: DuckDBPreparedStatement -> StatementState
+ Database.DuckDB.Simple.Internal: StatementOpen :: DuckDBV2PreparedStatementHandle -> DuckDBV2SqlStatementHandle -> IORef (Maybe (DuckDBV2SchemaHandle, DuckDBV2SchemaHandle)) -> StatementState
- Database.DuckDB.Simple.Internal: StatementStream :: Ptr DuckDBResult -> [StatementStreamColumn] -> Maybe StatementStreamChunk -> ResultMode -> StatementStream
+ Database.DuckDB.Simple.Internal: StatementStream :: DuckDBV2ResultHandle -> TypeScope -> [StatementStreamColumn] -> [DuckDBV2DataChunkHandle] -> Maybe StatementStreamChunk -> StatementStream
- Database.DuckDB.Simple.Internal: StatementStreamChunk :: DuckDBDataChunk -> Int -> Int -> [Int -> IO FieldValue] -> StatementStreamChunk
+ Database.DuckDB.Simple.Internal: StatementStreamChunk :: DuckDBV2DataChunkHandle -> Int -> Int -> [Int -> IO FieldValue] -> StatementStreamChunk
- Database.DuckDB.Simple.Internal: StatementStreamColumn :: Int -> Text -> DuckDBType -> StatementStreamColumn
+ Database.DuckDB.Simple.Internal: StatementStreamColumn :: Int -> Text -> Conversion LogicalTypeRep -> IORef (Maybe LogicalTypeRep) -> IORef (IntMap Text) -> StatementStreamColumn
- Database.DuckDB.Simple.Internal: [connectionHandle] :: ConnectionState -> DuckDBConnection
+ Database.DuckDB.Simple.Internal: [connectionHandle] :: ConnectionState -> DuckDBV2ConnectionHandle
- Database.DuckDB.Simple.Internal: [sqlErrorType] :: SQLError -> Maybe DuckDBErrorType
+ Database.DuckDB.Simple.Internal: [sqlErrorType] :: SQLError -> Maybe ErrorType
- Database.DuckDB.Simple.Internal: [statementHandle] :: StatementState -> DuckDBPreparedStatement
+ Database.DuckDB.Simple.Internal: [statementHandle] :: StatementState -> DuckDBV2PreparedStatementHandle
- Database.DuckDB.Simple.Internal: [statementStreamChunkPtr] :: StatementStreamChunk -> DuckDBDataChunk
+ Database.DuckDB.Simple.Internal: [statementStreamChunkPtr] :: StatementStreamChunk -> DuckDBV2DataChunkHandle
- Database.DuckDB.Simple.Internal: [statementStreamColumnType] :: StatementStreamColumn -> DuckDBType
+ Database.DuckDB.Simple.Internal: [statementStreamColumnType] :: StatementStreamColumn -> Conversion LogicalTypeRep
- Database.DuckDB.Simple.Internal: [statementStreamResult] :: StatementStream -> Ptr DuckDBResult
+ Database.DuckDB.Simple.Internal: [statementStreamResult] :: StatementStream -> DuckDBV2ResultHandle
- Database.DuckDB.Simple.Internal: destroyDataChunk :: DuckDBDataChunk -> IO ()
+ Database.DuckDB.Simple.Internal: destroyDataChunk :: DuckDBV2DataChunkHandle -> IO ()
- Database.DuckDB.Simple.Internal: destroyLogicalType :: DuckDBLogicalType -> IO ()
+ Database.DuckDB.Simple.Internal: destroyLogicalType :: DuckDBV2LogicalTypeHandle -> IO ()
- Database.DuckDB.Simple.Internal: destroyValue :: DuckDBValue -> IO ()
+ Database.DuckDB.Simple.Internal: destroyValue :: DuckDBV2ValueHandle -> IO ()
- Database.DuckDB.Simple.Internal: duckDBTypeFromName :: Text -> Maybe DuckDBType
+ Database.DuckDB.Simple.Internal: duckDBTypeFromName :: Text -> Maybe LogicalTypeId
- Database.DuckDB.Simple.Internal: duckDBTypeToName :: DuckDBType -> Text
+ Database.DuckDB.Simple.Internal: duckDBTypeToName :: LogicalTypeId -> Text
- Database.DuckDB.Simple.Internal: executePreparedResult :: ResultMode -> DuckDBPreparedStatement -> Ptr DuckDBResult -> IO DuckDBState
+ Database.DuckDB.Simple.Internal: executePreparedResult :: Statement -> IO DuckDBV2ResultHandle
- Database.DuckDB.Simple.Internal: fetchResultChunk :: ResultMode -> Connection -> Ptr DuckDBResult -> IO DuckDBDataChunk
+ Database.DuckDB.Simple.Internal: fetchResultChunk :: Connection -> DuckDBV2ResultHandle -> IO DuckDBV2DataChunkHandle
- Database.DuckDB.Simple.Internal: peekUtf8CString :: CString -> IO Text
+ Database.DuckDB.Simple.Internal: peekUtf8CString :: Coercible a CString => a -> IO Text
- Database.DuckDB.Simple.Internal: runInterruptibleQuery :: Connection -> IO DuckDBState -> IO DuckDBState
+ Database.DuckDB.Simple.Internal: runInterruptibleQuery :: Connection -> IO a -> IO a
- Database.DuckDB.Simple.Internal: withConnectionHandle :: Connection -> (DuckDBConnection -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withConnectionHandle :: Connection -> (DuckDBV2ConnectionHandle -> IO a) -> IO a
- Database.DuckDB.Simple.Internal: withDatabaseHandle :: Connection -> (DuckDBDatabase -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withDatabaseHandle :: Connection -> (DuckDBV2InstanceHandle -> IO a) -> IO a
- Database.DuckDB.Simple.Internal: withQueryCString :: Query -> (CString -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withQueryCString :: Query -> (ConstPtr CChar -> IO a) -> IO a
- Database.DuckDB.Simple.Internal: withResult :: Connection -> Query -> (Ptr DuckDBResult -> IO DuckDBState) -> (Ptr DuckDBResult -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withResult :: Statement -> (DuckDBV2ResultHandle -> IO a) -> IO a
- Database.DuckDB.Simple.Internal: withStatementHandle :: Statement -> (DuckDBPreparedStatement -> IO a) -> IO a
+ Database.DuckDB.Simple.Internal: withStatementHandle :: Statement -> (DuckDBV2PreparedStatementHandle -> IO a) -> IO a
- Database.DuckDB.Simple.Logging: LogEntry :: Maybe UTCTime -> Text -> Text -> Text -> LogEntry
+ Database.DuckDB.Simple.Logging: LogEntry :: Maybe UTCTime -> Text -> Text -> Text -> LogLevel -> LogScope -> Maybe Word64 -> Maybe Word64 -> Maybe Word64 -> Maybe Word64 -> LogEntry
- Database.DuckDB.Simple.LogicalRep: LogicalTypeScalar :: DuckDBType -> LogicalTypeRep
+ Database.DuckDB.Simple.LogicalRep: LogicalTypeScalar :: LogicalTypeId -> LogicalTypeRep
- Database.DuckDB.Simple.ToField: toDuckValue :: ToDuckValue a => a -> IO DuckDBValue
+ Database.DuckDB.Simple.ToField: toDuckValue :: ToDuckValue a => ValueFactory -> a -> IO NativeValue
- Database.DuckDB.Simple.Types: SQLError :: Text -> Maybe DuckDBErrorType -> Maybe Query -> SQLError
+ Database.DuckDB.Simple.Types: SQLError :: Text -> Maybe ErrorType -> Maybe Query -> Maybe Text -> SQLError
- Database.DuckDB.Simple.Types: [sqlErrorType] :: SQLError -> Maybe DuckDBErrorType
+ Database.DuckDB.Simple.Types: [sqlErrorType] :: SQLError -> Maybe ErrorType
Files
- CHANGELOG.md +594/−2
- README.md +1765/−595
- bench/Main.hs +155/−70
- dataframe-test/Main.hs +65/−58
- duckdb-simple.cabal +110/−46
- leaktest/Main.hs +480/−263
- nonthreaded-test/Main.hs +50/−0
- src/Database/DuckDB/Simple.hs +643/−698
- src/Database/DuckDB/Simple/Arrow.hs +37/−38
- src/Database/DuckDB/Simple/Arrow/Internal.hs +113/−77
- src/Database/DuckDB/Simple/Callback.hs +454/−64
- src/Database/DuckDB/Simple/CallbackMark.hs +131/−0
- src/Database/DuckDB/Simple/Catalog.hs +121/−73
- src/Database/DuckDB/Simple/Config.hs +209/−74
- src/Database/DuckDB/Simple/Config/Internal.hs +86/−0
- src/Database/DuckDB/Simple/Conversion.hs +145/−0
- src/Database/DuckDB/Simple/Copy.hs +39/−200
- src/Database/DuckDB/Simple/Copy/Internal.hs +394/−0
- src/Database/DuckDB/Simple/Deprecated/Streaming.hs +0/−87
- src/Database/DuckDB/Simple/Element.hs +263/−175
- src/Database/DuckDB/Simple/Extension.hs +260/−0
- src/Database/DuckDB/Simple/FileSystem.hs +164/−149
- src/Database/DuckDB/Simple/FromField.hs +1249/−813
- src/Database/DuckDB/Simple/FromRow.hs +157/−184
- src/Database/DuckDB/Simple/Function.hs +227/−437
- src/Database/DuckDB/Simple/Function/Internal.hs +523/−0
- src/Database/DuckDB/Simple/Generic.hs +874/−733
- src/Database/DuckDB/Simple/Geometry.hs +38/−25
- src/Database/DuckDB/Simple/Internal.hs +1403/−453
- src/Database/DuckDB/Simple/Logging.hs +21/−92
- src/Database/DuckDB/Simple/Logging/Internal.hs +178/−0
- src/Database/DuckDB/Simple/LogicalRep.hs +283/−220
- src/Database/DuckDB/Simple/Materialize.hs +705/−264
- src/Database/DuckDB/Simple/Native.hs +982/−0
- src/Database/DuckDB/Simple/Ok.hs +34/−35
- src/Database/DuckDB/Simple/Pragma.hs +13/−0
- src/Database/DuckDB/Simple/Pragma/Internal.hs +103/−0
- src/Database/DuckDB/Simple/QueryMetrics.hs +68/−0
- src/Database/DuckDB/Simple/Registration.hs +84/−0
- src/Database/DuckDB/Simple/Result.hs +248/−185
- src/Database/DuckDB/Simple/Streaming.hs +91/−0
- src/Database/DuckDB/Simple/Time.hs +12/−9
- src/Database/DuckDB/Simple/ToField.hs +2305/−1006
- src/Database/DuckDB/Simple/ToRow.hs +26/−27
- src/Database/DuckDB/Simple/Tokenizer.hs +57/−0
- src/Database/DuckDB/Simple/TypeCache.hs +0/−174
- src/Database/DuckDB/Simple/TypeScope.hs +99/−0
- src/Database/DuckDB/Simple/Typed.hs +249/−0
- src/Database/DuckDB/Simple/Typed/Aggregate.hs +402/−0
- src/Database/DuckDB/Simple/Typed/Table.hs +220/−0
- src/Database/DuckDB/Simple/Types.hs +40/−40
- src/Database/DuckDB/Simple/V2/Config.hs +56/−0
- src/Database/DuckDB/Simple/V2/Copy.hs +295/−0
- src/Database/DuckDB/Simple/V2/FunctionDocs.hs +51/−0
- src/Database/DuckDB/Simple/V2/Logging.hs +27/−0
- src/Database/DuckDB/Simple/V2/LogicalRep.hs +304/−0
- src/Database/DuckDB/Simple/V2/Pragma.hs +121/−0
- src/Database/DuckDB/Simple/V2/Registration.hs +962/−0
- src/Database/DuckDB/Simple/Variant.hs +83/−26
- src/Database/DuckDB/Simple/VariantCodec.hs +454/−407
- test-codec/Main.hs +0/−88
- test-quack/Main.hs +98/−0
- test/ArrowTests.hs +291/−192
- test/CancellationTests.hs +403/−139
- test/CopyTests.hs +426/−0
- test/CoreRegressionTests.hs +1027/−153
- test/ExtensionRegressionTests.hs +1015/−300
- test/GeometryRegressionTests.hs +625/−358
- test/Properties.hs +317/−258
- test/RegistrationTests.hs +705/−0
- test/SizedArrayTests.hs +209/−0
- test/Spec.hs +1732/−1257
- test/SqlFeatureTests.hs +240/−0
- test/StreamingTests.hs +261/−135
- test/TestUtils.hs +239/−7
- test/TimeTests.hs +832/−63
- test/TypedRegistrationTests.hs +2023/−0
- test/UserCastTests.hs +350/−0
- test/ValueRegressionTests.hs +2770/−362
- test/VariantRegressionTests.hs +654/−288
CHANGELOG.md view
@@ -1,5 +1,597 @@ # Changelog +See [MIGRATION.md](https://github.com/Tritlo/duckdb-haskell/blob/main/MIGRATION.md)+for the upgrade instructions.++## 1.0.0.0++- Target DuckDB 2.0 with `duckdb-ffi` 2.0.0.x. The backend uses the+ generated V2 C API. Upstream marks the V2 API as unstable ("The C API v2 is+ not yet stable. Its interface may change in upcoming releases."), so this+ release requires exactly DuckDB 2.0.0. The dependency bound excludes a+ `duckdb-ffi` release for a later DuckDB version. Creating an+ environment raises `SQLError` for another library version. Use the earlier+ release with DuckDB 1.5.+- Require `base` 4.18 or later, that is GHC 9.6 or later. Version 0.3 accepted+ `base` 4.14.+- Raw handles, callback handles, and the error codes of raw calls use V2+ types. `sqlErrorType` is `Maybe ErrorType`. See MIGRATION.md.+- Add `Database.DuckDB.Simple.Extension` for loadable Haskell extensions.+ Enable the `extension` flags of both packages. See `dev/extensions/`.+- `SQLError` has a fourth field, `sqlErrorBody`: the text of a native error+ without the prefix of its type, and `Nothing` for an error of the library.+ Code that builds or matches an `SQLError` by position does not compile. See+ MIGRATION.md.+- An `SQLError` with a type keeps that type when it leaves a Haskell callback.+ The outer statement fails with that type, and its text has one prefix, for example+ `Conversion Error: ...` for a failed query in a scalar function. A failed cast+ gives `ErrorTypeMismatch`. Sixteen types give `ErrorInputInvalid` and the text+ `Invalid Input Error: ` followed by the exception message. An `SQLError`+ with no type and each other exception give the same result. The "Errors"+ section of README.md lists those types and the reasons. When a callback error+ fails a statement in an explicit transaction, DuckDB invalidates that transaction, and+ COMMIT then stores none of its rows, also under the DuckDB setting+ `SYNTACTIC_ERRORS_DO_NOT_INVALIDATE`.+- Add shared environments and instances, attachment management, option+ discovery, table descriptions, SQL tokenization, and positioned file I/O.+- `listConfigFlags` takes a `Connection`. It lists that connection's database+ options, including custom options. See MIGRATION.md.+- `deleteFunction` reads the name with the rules of DuckDB for a qualified+ name, so a part in double quotes can contain a dot. See MIGRATION.md.+- `ConfigValue.configValueText` is `Maybe Text`, and `Nothing` for an option+ with a NULL value. Version 0.3 stopped the process for such an option. See+ MIGRATION.md.+- A `Day` outside the range of DuckDB, from -5877641-06-25 to 5881580-07-10, and+ a `LocalTime` or a `UTCTime` outside the range from -290308-12-22 00:00:00 to+ 294247-01-10 04:00:54.775806, in the notation of `Data.Time`, raise an+ `SQLError` of the type `ErrorTypeConversion` when they bind. For a target of+ the type TIMESTAMP_NS or TIMESTAMPTZ_NS, a value outside the smaller range of+ that type gets the same type, from the library or from DuckDB by its position.+ Version 0.3 bound some of these values and raised an `IOException` for the+ others. See MIGRATION.md.+- Add typed configuration values and defaults, managed log sinks, macro+ registration, function documentation, query metrics, and file path+ operations. Expose table bind metadata for multi-file readers.+- Add typed registrations. Their signatures have no native handle.+ `Database.DuckDB.Simple.Function` has `createAggregate`,+ `createTableFunction`, and the row class `TypedRow`. `Database.DuckDB.Simple`+ exports the two functions, `Aggregate`, and `TypedRow`. There is no typed+ cast: the README section "Typed registration" tells why.+- A cast that `registerCast` registers does not change a value that the+ library reads or binds, on a connection and in the callbacks of an extension+ (see the README). An exception in the callback of `registerCast` gives an+ `SQLError` of the type `ErrorTypeMismatch`, and `TRY(CAST(...))` raises the+ error. The section Errors of the README tells why.+- `createAggregate` registers a Haskell fold with initial state, step,+ combine, and result functions. Its result needs a static SQL declaration+ and a `ToDuckValue` encoder that works with an instance factory. Examples+ include nullable GEOMETRY, VARIANT, and records through `ViaDuckDB`.+- `V2.Registration.registerAggregateFunctionWithCleanup` runs a cleanup+ action when DuckDB releases the last reference to the aggregate's callbacks.+- `createTableFunction` registers a Haskell function that gives the rows of a+ table function as a list. DuckDB reads the list one chunk at a time. It reads+ only the columns that the query uses, and the first column for a query such+ as `SELECT count(*)`. The library does not evaluate the other columns of a+ row, but it evaluates each strict field of each row that it writes.+- `DuckDBColumnType` has the method `duckdbTypeDeclaration`. It gives the SQL+ type that a typed table function and a typed aggregate function declare. A+ record or a sum type gets `ToDuckValue` and a STRUCT or UNION declaration+ through `ViaDuckDB`. `StructValue`, `TupleValue`, and `UnionValue` have no+ declaration, and the registration raises an error for them. A+ `SizedArray n a` declares the ARRAY of size `n`. DuckDB casts a LIST of the+ same length to it with no error.+ `Database.DuckDB.Simple.LogicalRep` exports `TypeDeclaration`.+- `SizedArray n a` is new, and it replaces the instances of `Array Int a`: an+ `Array Int a` of exactly `n` values, with the bounds `(0, n - 1)`.+ `sizedArray` makes one from an `Array` with any bounds, `sizedArrayFromList`+ from a list, and `sizedArrayToArray` gives the `Array`. It binds as an ARRAY+ of size `n`, a typed registration declares that type, also for a field of a+ record through `ViaDuckDB`, and it reads an ARRAY and a LIST of `n` elements.+ A value of another length gives an `Incompatible` conversion error,+ reported as `SQLError` by the query APIs. `Array Int a` has no+ `ToField`, `FromField`, `DuckDBColumnType`, `ToDuckValue`, or `DuckValue`+ instance. A list reads an ARRAY of each size, and a `FieldArray` is an ARRAY+ value that a program builds at run time. See MIGRATION.md.+- An ARRAY parameter with more than 100000 elements, the largest ARRAY of+ DuckDB, raises an `SQLError` of the type `ErrorInputInvalid`, as an empty+ ARRAY parameter does.+- An argument of `createFunction` can have each type with `FromField` and+ `DuckDBColumnType`, for example a record through `ViaDuckDB`, a list, a `Map`,+ or a newtype with its own column type name. `FunctionArg` is gone.+ Arguments use the same conversion as query results, including nested values.+- A result of `createFunction` can have each type with `ToDuckValue`. Its SQL+ type is the `duckdbTypeDeclaration` of the type: an `Int16` result is+ SMALLINT, a `Word16` result is USMALLINT, and a `Float` result is FLOAT, in+ place of BIGINT and DOUBLE. `ScalarType` is gone, and `FunctionResult` has an+ instance for each type with `ToDuckValue`.+- `Database.DuckDB.Simple.Registration` has `TypeDeclaration`, `parameter`,+ `defaultValue`, `registerCustomType`, and typed replacement scans.+ `registerReplacementScan` and `registerInstanceReplacementScan` give the+ parts of an unknown table name to a Haskell function. The function can return+ a SELECT statement that DuckDB reads instead.+- Add `registerCopyFromFunction`, `registerCopyFunction` with the record+ `CopyTo`, `CopyFromInfo`, and `registerMultiFileFunction` to+ `Database.DuckDB.Simple.Copy`. `registerCopyFromFunction` registers a COPY+ FROM side that gives a list of typed rows. `registerCopyFunction` registers a+ COPY TO side and a COPY FROM side under one name. DuckDB keeps the first+ registration of a format name and ignores a later one. Thus a format with two+ directions needs `registerCopyFunction`.+- Add `Database.DuckDB.Simple.Pragma`. `registerPragmaFunction` takes a typed+ callback `PragmaCallback args`. The callback gets the arguments of the call+ as a `FromRow` row.+- Add custom options to `Database.DuckDB.Simple.Config`. `customOptionUpdate`+ is a function that gets the new value as a `FieldValue`. An exception in the+ function rejects the value.+- `Database.DuckDB.Simple.Extension` has the same typed registrations for a+ loadable extension: `createAggregate`, `createTableFunction`,+ `registerReplacementScan`, `registerCustomType`,+ `registerMultiFileFunction`, `registerCopyToFunction`,+ `registerCopyFromFunction`, `registerCopyFunction`,+ `registerPragmaFunction`, and `registerCustomOption`.+- A typed registration has more overhead for each row than raw vector access.+ Use the V2 modules for speed, projection and filter pushdown,+ parallel table scans, and scalar functions with a bind callback or variadic+ parameters.+- Add `SettingScope`, `LogLevel`, `LogScope`, `TokenType`, `MacroFunctionType`,+ `ErrorType`, and `LogicalTypeId`. The configuration, logging, tokenizer,+ macro, error, and logical type APIs use these types instead of the generated+ V2 enum types, so a later C API version does not change these APIs. The file+ system API uses `FileFlag` and the opaque `FileHandle` instead of the+ generated V2 types. `ToDuckValue.toDuckValue` uses the synonyms+ `ValueFactory` and `NativeValue` for the two native handles in its signature.+- The functions with native handles are in modules with `V2` in the name.+ `Database.DuckDB.Simple.V2.Registration` has the scalar, aggregate, table,+ and cast registrations and the replacement scans with native callbacks. It+ also has `registerMacroFunctionWith`, which gives the native macro builder to+ an action. It takes the `RegistrationTarget` from+ `Extension.extensionTarget`. `Database.DuckDB.Simple.V2.Copy` has the native+ COPY callback interface: `CopyToCallbacks`, `CopyFromCallbacks`, the raw+ `registerCopyFunction`, `registerCopyFunctionWith`,+ `registerCopyToCallbacks`, `registerCopyFromCallbacks`, and+ `registerMultiFileFunctionWith`. `Database.DuckDB.Simple.V2.Pragma` has the+ pragma callbacks with native handles. `Database.DuckDB.Simple.V2.Config` has+ `registerCustomOptionWith`. `Database.DuckDB.Simple.V2.FunctionDocs` has+ `setFunctionDocs`. `Database.DuckDB.Simple.V2.Logging` has `logMessage`,+ which writes a message through the context of a native callback.+ `Database.DuckDB.Simple.V2.LogicalRep` has `logicalTypeToRep`,+ `logicalTypeFromRep`, `logicalTypeFromRepWith`, and `destroyLogicalType`.+- Keep statements alive while reading cached schemas for column metadata and+ parameter types. `Internal.withStatementSchemas` scopes these borrowed handles.+- Release acquired native handles if evaluation of a lazy input list fails.+- Mask native parameter cleanup so cancellation cannot skip releases.+- Read LIST, ARRAY, and MAP elements with a constant stack. Keep native reads in+ element order.+- Read VARIANT metadata from its result vector. Construct VARIANT types by name.+ Read singleton INT64 and UINT64 cast payloads with native getters.+- `Database.DuckDB.Simple.Internal` changes. It re-exports the native helper+ layer. It no longer exports `toSQLError`, `fetchPrepareError`,+ `fetchResultError`, `throwResultError`, `mkExecuteError`,+ `throwRegistrationError`, `withClientContext`, and `destroyClientContext`.+ `Connection` adds `connectionInstance`, `connectionLock`,+ `connectionWindow`, and `connectionOptions`. `ConnectionOpen` has no+ `connectionDatabase` and adds `connectionStatements` and+ `connectionReleaseParent`. `Statement` adds `statementArrowResult`,+ `statementBindings`, `statementParameterNames`, and+ `statementParameterTypes`.+ `StatementOpen` adds `statementParsed` and `statementSchemaCache`.+ `StatementStream` has no `statementStreamMode` and adds+ `statementStreamPendingChunks`. `StatementStreamColumn` has no+ `statementStreamColumnType` and adds `statementStreamColumnTree`.+- The SQL error of a failed Haskell callback has the plain message of the+ exception. Version 0.3 gave `user error (...)` for a `userError` and the+ Haskell record for an `SQLError`.+ The error of an argument that does not convert names the argument (`arg0`,+ `arg1`, and so on). Its text has one form for scalar functions, table+ functions, and aggregate functions, for example+ `cannot convert argument arg0 from VARCHAR to Int64`. A table function and an+ aggregate function put their name before it:+ `duckdb-simple: NAME: cannot convert argument ...`.+- Add scoped named collection inputs and prepared dynamic PIVOT with a deferred+ output schema. Prepared dynamic PIVOT cannot have parameters.+- Rename `Database.DuckDB.Simple.Deprecated.Streaming` to+ `Database.DuckDB.Simple.Streaming`.+- Export Arrow release helpers from `Database.DuckDB.Simple.Arrow`.+- A DuckDB callback that waits for the lock of a connection waits at most the+ time limit of that connection. Add `connectionCallbackLockWait` to+ `ConnectionOptions`, in microseconds, with the default+ `defaultCallbackLockWait` (10 seconds). `Nothing` sets no time limit, and the+ open rejects a negative time limit. A statement of a callback on the+ connection that runs the callback thus raises an `SQLError` of the type+ `ErrorResourceInUse` after the time limit. For most callbacks, the outer+ statement then fails. The README gives the cases that the time limit does+ not cover.+- The README and the module header of `Database.DuckDB.Simple.Function` say that+ a program that registers a callback must be linked with `-threaded`. DuckDB+ can call a callback from several threads at the same time, and without+ `-threaded` such a program can crash. Version 0.3 needed `-threaded` in the+ same way, and its README named it only for prompt cancellation. A program+ that is not linked with `-threaded` now gets an `SQLError` of the type+ `ErrorInputInvalid` when it registers a callback for a database with the+ DuckDB setting `threads` above 1. COPY format callbacks and log sinks+ require `-threaded` even when `threads` is 1: DuckDB can call them on its+ asynchronous workers. For other registrations the library reads the setting+ at registration and cannot detect a later `SET threads`.+- Add `connectionAggregateStripes` to `ConnectionOptions`. It sets the number+ of stripes in the state table of each typed aggregate that `createAggregate`+ registers through the connection. The+ default is `defaultAggregateStripes` (64). The open rejects a number that is+ not positive. A registration on an extension uses the default.+- Add `connectionArrowBatchSize` to `ConnectionOptions`. It sets the largest+ number of rows in one Arrow array of an Arrow fold on the connection. The+ default is `defaultArrowBatchSize` (2048 rows). The open rejects a value that+ is not positive.+- The library decodes the message of a native error with replacement+ characters. Thus `show` of such an `SQLError` does not raise for a text that+ is not valid UTF-8. Version 0.3 decoded the text with `decodeUtf8`, which+ raises for such a text.+- While an explicit stream is incomplete, DuckDB rejects other statements on+ its connection. Closing an incomplete stream aborts its statement.+- Each connection serializes statement preparation, its open native result,+ and each fetch of an explicit stream. Threads that share a connection wait+ for each other in the default query functions. Serialize transactions and+ explicit streams in the application.+- `query_` and `execute_` parse all statements before they run any.+- An INSERT, UPDATE, or DELETE without RETURNING gives one row with the number+ of changed rows in each function that reads rows, also in `fold`, `nextRow`,+ and the same functions of `Database.DuckDB.Simple.Streaming`.+- Decode and bind TUPLE values with `TupleValue`, `FieldTuple`, and+ `LogicalTypeTuple`.+- Preserve composite VARCHAR collations with `LogicalTypeCollatedVarchar`.+- Keep the alias of a type with `LogicalTypeAlias`, at each depth. A+ `ColumnDescription` of a JSON column gives+ `LogicalTypeAlias "JSON" (LogicalTypeScalar LogicalTypeIdVarchar)`, and a+ column of a type from `registerCustomType` gives the name of the type.+ `logicalTypeFromRep` creates the alias type, so a `DuckValue` field can+ declare JSON in a typed registration and in a parameter. The type fields of+ a STRUCT, TUPLE, or UNION value keep the alias, so a value that a query reads+ binds again with its type. The readers and the binding of parameters convert+ with the base type. A column of a type from `CREATE TYPE name AS base` has the+ base type and no alias.+- `logicalTypeToRep` takes a factory before the logical type handle. Keep the+ factory valid throughout the conversion. The returned `LogicalTypeRep` is+ pure and remains valid after the factory closes. The function does not+ destroy the logical type handle. See MIGRATION.md.+- Decoders and field metadata use `Conversion a` with explicit IO. `Field`+ loses its `Eq` and `Show` instances. Native field metadata expires with its+ query result or callback scope. See the+ [source migration](https://github.com/Tritlo/duckdb-haskell/blob/main/MIGRATION.md#duckdb-simple-03-to-10)+ for changed signatures and instances, and+ [Decoding rows](https://github.com/Tritlo/duckdb-haskell/blob/main/duckdb-simple/README.md#decoding-rows) for conversion and metadata rules.+- Read errors name the type of the column with `logicalTypeSQL`, the type+ text of DuckDB, for example DECIMAL(18,2), TIMESTAMP_NS, ENUM('a', 'b'), or+ JSON. For a VARIANT column, they name the type of the stored value in the+ same text, for example TIMESTAMP WITH TIME ZONE. The bind errors+ `cannot encode a value of type ... as ...` and+ `unsupported scalar conversion for ... from a value of type ...` name each+ type in the same text.+- Decode nanosecond timezone timestamps in results and VARIANT values.+- Preserve the digits below a microsecond of a `TimeOfDay`, a `LocalTime`, and+ a `UTCTime`, and of their `Unbounded` forms, for a nanosecond parameter+ (`TIME_NS`, `TIMESTAMP_NS`, or `TIMESTAMPTZ_NS`), also in collections,+ STRUCTs, MAPs, and UNIONs. For a `VARIANT` parameter, one such value keeps+ them when it fits the range of the nanosecond type, but the elements of a+ list, a `SizedArray`, or a `Map` of such values get the microsecond value (see+ the entry on `VARIANT` parameters below). Only a value with such digits+ reads the parameter type. Each other parameter gets the microsecond value,+ as for the SQL literal. An explicit cast such as `?::TIMESTAMP_NS` makes a+ nanosecond parameter, so for a `TIMESTAMP` column DuckDB rounds the value to+ the microsecond, where a plain `?` cuts the digits. Version 0.3 bound the+ microsecond value for each parameter.+- Decode VARIANT values through public V2 operations. Object keys use DuckDB's+ lexicographic order.+- Allow a VARIANT object to contain an empty key with other named keys.+ Version 0.3 rejected such an object.+- A `Variant` parameter with a UNION, a MAP, an ENUM, or a GEOMETRY with a CRS+ in its payload, at any depth, raises `SQLError` with the type+ `ErrorInputInvalid`. The error gives the reason: a VARIANT keeps only the+ member value of a UNION, stores a MAP as a list of key and value objects,+ stores an ENUM as its text, and does not keep the CRS of a GEOMETRY. Version+ 0.3 bound a UNION payload with no error.+- Read GEOMETRY results through a native cast to BLOB. DuckDB 2.0 rejects WKB+ with mixed Z/M member layouts.+- `ToDuckValue.toDuckValue`, `logicalTypeFromRep`, and `logicalTypeFromRepWith`+ take a V2 factory. Composite parameters keep GEOMETRY CRS metadata, including+ unknown custom labels resolved through SQL on a helper connection.+- `logicalTypeFromRep` creates a VARIANT type, and a GEOMETRY type with a CRS+ that DuckDB recognizes. Version 0.3 raised an error for VARIANT and created+ GEOMETRY without the CRS.+- Add a `ToDuckValue` instance for `Variant`. Version 0.3 had none.+- A composite parameter that the native value constructor rejects raises an+ `SQLError`. For example, a STRUCT parameter with a MAP field that has a NULL+ key raises an `SQLError` with the type `ErrorInputInvalid`. Version 0.3+ raised an `IOException`.+- Remove `connectionGeometryCRS` from `ConnectionOptions` and `withTypeCache`+ from `Database.DuckDB.Simple.Internal`. Remove the field+ from record construction and updates, or its argument from positional+ construction and patterns. Parameters construct recognized CRS names and full+ definitions directly. Unknown custom labels use a helper SQL query on demand,+ including in nested types, typed NULLs, empty collections, and inactive UNION+ members. The library does not keep a permanent cache of custom CRS types.+ Blank labels and labels with NUL fail when used by a parameter, not when the+ connection opens. Query reads retain custom labels. Typed registrations and+ direct `logicalTypeFromRep` calls still require a recognized CRS or a full+ definition.+- The library builds a UNION parameter in a one-row UNION vector, because the+ V2 API has no UNION value constructor. This requires flat child vectors.+ DuckDB 2.0.0 supplies them in a new vector, but its header does not guarantee+ this.+- A STRUCT, TUPLE, or UNION parameter binds a `FieldGeometry` value as a+ GEOMETRY with the same WKB bytes and CRS, through DuckDB's cast from BLOB.+ A value whose CRS is not the CRS of its declared type raises `SQLError` with+ the type `ErrorInputInvalid`. A `Variant` parameter binds a `FieldGeometry`+ value without a CRS. One with a CRS raises `SQLError`, because a VARIANT does+ not keep the CRS. Version 0.3 raised an error for each such value, because+ DuckDB 1.5 has no cast from BLOB to GEOMETRY.+- Add a `ToField` instance for `RawGeometry`. It binds as a GEOMETRY with the+ same WKB bytes and its own CRS. DuckDB casts it to the type of the parameter,+ as in SQL text. A value without a CRS takes the CRS of the target. A target+ without a CRS removes the CRS. A target with another CRS raises `SQLError`.+ Version 0.3 had only a `FromField` instance for `RawGeometry`. A+ `SizedArray`, a list, or a typed row cannot hold a `RawGeometry`.+- An empty ARRAY parameter, a `SizedArray` of size 0 or a `FieldArray` with no+ element, raises `SQLError`. Version 0.3 bound an empty `Array` of scalar+ elements. Bind an empty list for an empty LIST.+- `TimeWithZone` holds the offset in seconds east of UTC, as a TIMETZ stores it.+ The field `timeWithZoneZone :: TimeZone` of version 0.3 is now+ `timeWithZoneOffsetSeconds :: Int32`. `timeWithZone` and `timeWithZoneZone`+ convert from and to a `TimeZone`. A TIMETZ with an offset that has seconds,+ for example `12:00:00+01:00:30`, now reads. Version 0.3 raised an error for+ it. Update code that uses the old field. See MIGRATION.md.+- Add `ToField` instances for `IntervalValue`, `TimeWithZone`, and+ `DecimalValue`, and `ToDuckValue` instances for `IntervalValue` and+ `TimeWithZone`. An `IntervalValue` binds as an INTERVAL and a `TimeWithZone`+ as a TIMETZ, also in a `SizedArray`, a list, or a typed row. A `TimeWithZone`+ reads back equal to the bound value, except that a time loses its digits+ below one microsecond. An offset outside -15:59:59 to +15:59:59 raises an+ error. A `DecimalValue` binds as a DECIMAL with its own width and scale. A+ `SizedArray`, a list, or a typed row cannot hold a `DecimalValue`. Version 0.3+ had only `FromField` instances for these types.+- Add `ToField` and `ToDuckValue` instances for `Map`. A `Map` binds as a MAP+ value with its pairs in ascending key order. The keys and the values follow+ the element rules of a list, and an empty `Map` binds as an empty MAP of its+ key and value types. Version 0.3 had only a `FromField` instance for `Map`.+- Add `ToField` and `ToDuckValue` instances for lists. A list binds as a LIST+ value with the element rules of a `SizedArray`, and an empty list binds as an+ empty LIST of its element type (for an empty list of `Integer` or `Natural`+ values, see the entry on `Integer` parameters). A `String` stays one VARCHAR+ parameter, and a list as a row is still a row of parameters, so a list+ parameter is `Only [..]`. Version 0.3 had no `ToField` instance for lists.+- Each input check of the library raises `SQLError` with the type+ `ErrorInputInvalid`: a parameter value that does not fit its type (for example+ a `BitString` with bad padding, a `UnionValue` with a tag out of range, or an+ empty `SizedArray`), a logical type with a NUL in a name or a bad DECIMAL+ width, a CRS name that DuckDB does not keep, a NUL in a database path, an+ alias, an attach option, a configuration, a query, or a parameter name, a NUL+ in a file path, a negative file offset or read size, a file range outside+ `Int64`, a bad `ConnectionOptions` field, SQL text with no statement or with+ more than one statement, a parameter index out of bounds, a column index of+ `columnName` out of bounds, a bad name or argument of a registration, a+ catalog lookup, a configuration option, or a log sink, and the column count of a+ COPY FROM table. An error about a statement has+ the query of the statement, and each other error has no query. A value outside+ the range of its SQL type raises `SQLError` with the type+ `ErrorTypeConversion` when it binds, the type of the "Conversion Error" that+ DuckDB raises for the same value in SQL: a date or a timestamp outside the+ range of DuckDB, an integer outside the range of its parameter type or its+ declared type, a nanosecond timestamp outside its storage or at an infinity+ value as a STRUCT field, a TUPLE element, a UNION member, or an element of a+ list, a `SizedArray`, or a `Map`, a `FieldHugeInt` or a `FieldUHugeInt`+ outside its type, a TIMETZ offset outside -57599 to 57599 seconds, a time of+ day that is not a TIME value (a field outside its own range, or a time after+ 24:00:00), an ENUM label that is not in the dictionary of its type, and a+ `DecimalValue` that its width cannot hold. For such a nanosecond timestamp in+ the range of DuckDB, for example in the year 2300, as a TIMESTAMP_NS or+ TIMESTAMPTZ_NS parameter, DuckDB raises its own error, of the same type. A+ `LocalTime` or a `UTCTime` (in UTC) from 2262-04-11 23:47:16.854775807 to+ 23:47:16.854775999 gets the error of the library in each position. A value of+ the wrong shape or type keeps `ErrorInputInvalid`, also where DuckDB gives+ another type for the same input, for example a UNION tag that is not a member,+ a leaf built from another type, a STRUCT field that does not match its type,+ or a blank GEOMETRY CRS label. A parameter count or a parameter name that does not match+ the query keeps its `FormatError`. A check in a callback fails the outer+ statement with the type of the check: such a value as the result of a typed+ function or a row of a typed table function gives `ErrorTypeConversion` and+ the text `Conversion Error: duckdb-simple: ...`. Version 0.3 raised an+ `IOException` for the checks of parameter values, logical types, and CRS+ names. The error for a scalar value that does not match its declared type+ names the SQL type, for example `"NULL"` in place of `LogicalTypeIdSQLNull`.+- A check of the state of the library, or of a vector, a type, or a value that+ DuckDB gave, raises `SQLError` with the type `ErrorRuntimeInternal`. Such an+ error means a defect of the library or of DuckDB. When a statement raises it,+ it has the query of the statement, as each other `SQLError` of a statement+ has. Inside a callback, the statement that runs the callback fails instead,+ with an `SQLError` of the type `ErrorInputInvalid` and the text "Invalid Input+ Error: duckdb-simple: ...", because a callback does not pass the type+ `ErrorRuntimeInternal` (a cast gives `ErrorTypeMismatch`). The use of a closed+ environment, instance, connection, or statement, a DuckDB library+ of another version, and a callback state read with the wrong type are not+ internal checks; they raise an `SQLError` with no type. Version 0.3 raised an+ `IOException` (a `userError`) or an `ErrorCall` for the internal checks, for+ example for a UNION tag outside the members of a result vector.+- A `TimeOfDay`, a `LocalTime`, or a `UTCTime` with digits below a microsecond,+ and a `FieldTime`, a `FieldTimestamp`, or a `FieldTimestampTZ` with such+ digits, binds with the nanosecond type for a nanosecond parameter, and for a+ `VARIANT` parameter when it fits the range of that type, so the VARIANT keeps+ the digits. A value with no such digits binds as before, with no read of the+ parameter type. These values get the microsecond value: a value outside the+ range of the nanosecond type, an element of a list, a `SizedArray`, or a `Map`+ for a `VARIANT` parameter, and an element of a `VARIANT[]` or a+ `MAP(..., VARIANT)` parameter. In a `Variant` payload, each time leaf, also+ an element of a list payload and a field of an object payload, binds with the+ nanosecond type when it has such digits and fits the range.+- An empty `SizedArray` and an empty `FieldArray` raise the same `SQLError`, of+ the type `ErrorInputInvalid`.+- Add a parameter instance for `FieldValue`, so a row read as `[FieldValue]`+ binds back. A scalar constructor binds as the DuckDB type of the+ constructor, and a `FieldList`, a `FieldArray`, or a `FieldMap` takes the+ element type that its elements give, with the rule of a SQL list literal.+ A `FieldEnum` binds its label as a VARCHAR. An empty `FieldArray` raises+ `SQLError` with the type `ErrorInputInvalid`. A `FieldTimestamp`, a+ `FieldTime`, and a `FieldTimestampTZ` bind with the microsecond type. A value+ with digits below a microsecond binds with the nanosecond type for a+ nanosecond parameter, and for a `VARIANT` parameter when it fits the range of+ that type.+- A `SizedArray`, a list, or a `Map` of STRUCT values, UNION values, generic+ records, lists, or arrays with different element types gets the type that+ DuckDB gives a SQL list literal of the elements, for example+ `STRUCT(a INTEGER, b INTEGER)` for `{'a': 1}` and `{'b': 2}`. Elements with+ no common type raise the `SQLError` of DuckDB. Version 0.3 raised an error+ for each `SizedArray` with different element types.+- A list or a `Map` of such elements that has no element, and a `SizedArray`, a+ list, or a `Map` with only NULL elements, take the element type of the+ parameter when the parameter has a known type, for example for an `INSERT`+ into a typed column. An empty `SizedArray` raises an error, as an entry above+ says. For a parameter with no known type, such a list of generic records+ through `ViaDuckDB` takes the STRUCT or UNION type of the record. Each other+ such list or `Map` with only NULL elements takes the type of the SQL list+ literal, such as `"NULL"[]`, and each other empty list or `Map` takes the+ type NULL, as `[]` and `MAP {}` have. Version 0.3 raised an error for an+ `SizedArray` of NULL elements.+- Some parameters bind for the type that DuckDB gives the parameter: a time+ value with digits below a microsecond, a `SizedArray`, a list, a `Map`, a+ manual STRUCT, TUPLE, or UNION value, and an `Integer` or a `Natural` outside+ the HUGEINT range. `executeMany` reads these types one time for each call. A+ column type that changes during the call does not change how the later rows of+ that call bind.+- An `Integer` or a `Natural` parameter binds as a HUGEINT value when the value+ is in the HUGEINT range. Before, it bound as a BIGNUM value in each case. In+ DuckDB 2.0.0 the casts from BIGNUM to the integer types have three defects. A+ cast to HUGEINT, UHUGEINT, or DECIMAL fails. A cast of a negative value to an+ unsigned type gives a wrapped value. A cast of a value of 2^128 or more gives+ the value modulo 2^128. DuckDB checks the range of a HUGEINT value for its+ target. A value outside the HUGEINT range binds as BIGNUM, or as UHUGEINT for+ a UHUGEINT parameter that can hold it. For each other integer or DECIMAL+ parameter, such a value raises an `SQLError` of the type+ `ErrorTypeConversion`. `typeof(?)` is thus HUGEINT for an `Integer` in the+ HUGEINT range. Bind a `BigNum` to bind a BIGNUM value in each case. A+ `SizedArray` of `Integer` or `Natural` values takes the element type of its+ target when that type is an integer type or DECIMAL, so DuckDB checks the+ range of each element. Before, a negative element for an unsigned array column+ was stored as a wrapped value. For each other target, a `SizedArray`, a list,+ or a `Map` of such values takes the type that its elements give, as a SQL list+ literal of the HUGEINT and BIGNUM values: `[1, 2^64]` is `HUGEINT[]`,+ `[1, 2^200]` is `BIGNUM[]`, and an empty list or a list of only NULL elements+ is `"NULL"[]`. The rule follows the column type name BIGNUM, so an all-NULL+ list of `BigNum` values with no target is also `"NULL"[]` in place of+ `BIGNUM[]`, and a type whose column type name is BIGNUM follows it too.+- `toDuckValue` for an `Integer` or a `Natural` gives a HUGEINT value when the+ value is in the HUGEINT range, and a `Natural` up to 2^128 - 1 gives a+ UHUGEINT value. Each other value raises an `SQLError` of the type+ `ErrorTypeConversion`; use `BigNum` for a larger value. Before, it gave a+ BIGNUM value in each case. DuckDB checks the range when it casts a HUGEINT+ value. Thus `setConfigValue` with -5 for an unsigned option raises an error,+ and a default value from `defaultValue` that its parameter cannot hold raises+ an error at a call that omits the argument. A newtype over `Integer` with the+ default `toField` binds through `toDuckValue`, so its parameter outside the+ HUGEINT range also raises this error.+- An `Integer` declares HUGEINT and a `Natural` declares UHUGEINT as a column of+ a typed row, an argument, and a result, as for a field of a generic record. A+ list or a `Map` of them in a typed row has HUGEINT or UHUGEINT elements. The+ column type name (`duckdbColumnTypeFor`) still gives BIGNUM for them, the type+ of a parameter that holds each value.+- A product type with no fields has one encoding: the STRUCT without fields,+ which DuckDB shows as `{}`. `genericToFieldValue` gives this STRUCT, as the+ `ViaDuckDB` binding and `genericToStructValue` did before. A constructor with+ no fields in a sum type has this STRUCT as its payload. This includes each+ constructor of an enum-like sum type, such as `data Colour = Red | Blue`.+ Version 0.3 gave `FieldNull` in these two places, so `Just x` and `Nothing`+ of such a type were the same value in a field. The decoders accept+ `FieldNull` and the STRUCT without fields.+- A value that is not NULL with the declared type SQLNULL raises a conversion+ error. Before, it was bound as NULL with no error. This applies to a manual+ `StructValue`, `TupleValue`, or `UnionValue`, to a LIST of that type, and to+ a `Variant` that holds such a value.+- `errSQLType` of a `ResultError`, and the `SQLError` message of a failed+ conversion, give the DuckDB type names `TIMESTAMPTZ` and `TIMETZ`. Earlier+ versions gave `TIMESTAMP_TZ` and `TIME_TZ`, which DuckDB does not accept as+ type names.+- `errSQLType` of a `ResultError`, and the `SQLError` message of a failed+ conversion, give the type names of DuckDB's `typeof`: `TINYINT`, `SMALLINT`,+ `INTEGER`, `BIGINT`, and `VARCHAR`. Earlier versions gave `INT1`, `INT2`,+ `INT4`, `INT8`, and `TEXT`. The errors of DuckDB use `INT8` for the 8-bit+ type, not for `BIGINT`.+- `duckdbColumnType` gives `VARCHAR` for `Text` and `String`. Earlier versions+ gave `TEXT`.+- A DECIMAL read as `Double` or `Float`, and a BIGINT read as `Float`, give+ the value of the cast of DuckDB to DOUBLE or FLOAT. Earlier versions rounded+ two times, so the value could differ from the cast in the last bit. For+ example, `2706284552926637.70::DECIMAL(18,2)` read as `2706284552926638.0`,+ and the cast gives `2706284552926637.5`. For scale 0 and a width of 19 to 38,+ the cast of DuckDB does not always give the nearest value, and the library+ gives the value of the cast. As seen in CI, the DuckDB 2.0.0 build for macOS+ on Apple Silicon rounds some BIGINT values above 2^53 two times in its cast+ to FLOAT, and the library gives the nearest value there.+- A BOOLEAN read as `Text` or `String` gives `true` or `false`, as the cast to+ VARCHAR does. Earlier versions gave `1` or `0`.+- A `FromField` instance reads another SQL type only when it gives the value+ of the cast of DuckDB, or an error. The Haddock of each instance gives the+ SQL types that it reads. MIGRATION.md gives the SQL to write in place of each+ read that the next three entries remove.+- A TIMESTAMP, TIMESTAMP_S, TIMESTAMP_MS, TIMESTAMP_NS, or DATE read as+ `UTCTime`, and a TIMESTAMPTZ or TIMESTAMPTZ_NS read as `LocalTime`, raise the+ error of an incompatible column, also in the `Unbounded` forms. The casts of+ DuckDB use the TimeZone setting of the session. Earlier versions used UTC.+- A FLOAT or DOUBLE read as `Text` or `String` raises the error of an+ incompatible column. Earlier versions gave the text of `show`, for example+ `1.0e-2` for 0.01, where the cast gives `0.01`.+- A VARCHAR read as `ByteString` raises the error of an incompatible column.+ Earlier versions gave the UTF-8 bytes, where the cast to BLOB reads escape+ sequences.+- `FieldEnum` holds the index and the label of an ENUM value:+ `FieldEnum Word32 Text`. An ENUM value reads as `Text` or `String` with its+ label, as the cast to VARCHAR gives it. Earlier versions raised an error. An+ ENUM value read as an integer type raises the error of an incompatible+ column. Earlier versions gave the index, where the cast of DuckDB parses the+ label. The reader reads a label through the dictionary of the type and does+ not cast. A result column reads each label that it needs one time for each+ execution. The readers of a MAP key or value, of an argument, and of a COPY+ TO row read the labels again for each vector. A child of a STRUCT, a TUPLE,+ or a UNION reads the whole dictionary one time for each execution.+- A `FieldEnum` inside a value of a declared ENUM type binds its label, not its+ index. A label that is not in the dictionary raises `SQLError` with the type+ `ErrorTypeConversion`, as the cast of DuckDB gives for that label.+- `displayException` of a `ResultError` gives the text of the error, with the+ column and the two types, as the `SQLError` of a column. Earlier versions+ gave the derived `show`. The error of a field of a `ViaDuckDB` record names+ the field, for example+ `duckdb-simple: field stampedAt: a value has type TIMESTAMP but expected UTCTime`.+- A list field of a generic record reads an ARRAY, as a top-level list does.+ Earlier versions read only a LIST in a field.+- The errors of the generic decoders name the SQL type of a value, for+ example+ `duckdb-simple: field xs: a value has type MAP(VARCHAR, BIGINT) but expected LIST or ARRAY`.+ Earlier versions printed the derived `show` of the whole value.+- Two bind errors name the SQL type of the value in the same way:+ `cannot encode a value of type INTEGER as ...` and+ `unsupported scalar conversion for BIGINT from a value of type VARCHAR`.+ Earlier versions printed the derived `show` of the whole `FieldValue`.+- A `Variant` read accepts only a value that a VARIANT can hold. A column that+ holds an ENUM, a MAP, a UNION, an ARRAY, a TUPLE, a STRUCT that is not in the+ form of a VARIANT object (each field of the type VARIANT, the keys in+ ascending order), or a GEOMETRY with a CRS, at any depth, raises the error of+ an incompatible column. Earlier versions wrapped the value, which DuckDB's cast+ to VARIANT changes. A type that is or contains JSON, at any depth, also raises+ that error, because DuckDB casts JSON to VARIANT by parsing the text.+- Preserve the previous set of bindings when positional or named parameter+ validation or conversion fails.+- Allow session variables to supply omitted named parameters. Too few named+ parameters no longer raise `FormatError`. Too many named parameters still+ raise it.+- Reject duplicate named parameters with DuckDB's ASCII case rules. Keep+ distinct non-ASCII names separate. Preserve sparse numeric parameter names+ such as `$3` and distinguish quoted parameter names that start with+ `$`, `:`, or `@`.+- Resolve configuration names and aliases case-insensitively. Use DuckDB's+ ASCII identifier rules for catalog names and top-level schemas.+- `deleteFunction` reports the error from DuckDB. DuckDB does not permit+ deletion of functions registered through the C API, so the call raises+ `SQLError` for a function made with `createFunction`. Version 0.3 ignored+ this error. A `WARNING` pragma on `deleteFunction` causes a compile-time+ warning at each use. A build with `-Werror` must allow that warning, for+ example with `-Wno-error=deprecations`.+- Add the optional `quack-tests` suite for authenticated remote queries,+ parameter binding, writes, streaming, and nanosecond timestamps.+ ## 0.3.0.0 - Use native DuckDB 1.5.6 by default. Keep native support for DuckDB >= 1.5.3@@ -12,8 +604,8 @@ depth limit for acyclic values. - Add `Variant`, a `FieldValue` that binds as a VARIANT, and `variantObject`, which builds an object payload. Scalars keep their native type. Object- parameters reject duplicate, empty, and NUL-containing keys because the- native constructors cannot represent all of these names.+ parameters reject duplicate and NUL-containing keys, and objects with only+ an empty key, because the native constructors cannot represent these names. `Variant` has no `ToDuckValue` instance, and `logicalTypeFromRep` raises an error for VARIANT. - A VARIANT timestamp payload outside the microsecond range binds as a
README.md view
@@ -1,597 +1,1767 @@ # duckdb-simple -`duckdb-simple` provides a high-level Haskell interface to DuckDB inspired by-the APIs of [`sqlite-simple`](https://hackage.haskell.org/package/sqlite-simple) and-[`postgresql-simple`](https://hackage.haskell.org/package/postgresql-simple).-It builds on the low-level bindings exposed by [`duckdb-ffi`](../duckdb-ffi) and-provides a focused API for opening connections, running queries, binding-parameters, and decoding typed results—including the full set of DuckDB scalar-types (signed/unsigned integers, decimals, hugeints, intervals, precise and-timezone-aware temporals, blobs, enums, bit strings, and bignums).--## Getting Started--```haskell-{-# LANGUAGE OverloadedStrings #-}--import Database.DuckDB.Simple-import Database.DuckDB.Simple.Types (Only (..))--main :: IO ()-main =- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE items (id INTEGER, name TEXT)"- _ <- execute conn "INSERT INTO items VALUES (?, ?)" (1 :: Int, "banana" :: String)- rows <- query_ conn "SELECT id, name FROM items ORDER BY id"- mapM_ print (rows :: [(Int, String)])-```--### Key Modules--- `Database.DuckDB.Simple` – connections, prepared statements, execution,- queries, metadata, and error handling.-- `Database.DuckDB.Simple.ToField` / `ToRow` – typeclasses and helpers for- preparing positional or named parameters.-- `Database.DuckDB.Simple.FromField` / `FromRow` – typeclasses for decoding- query results, with generic deriving support for product types.-- `Database.DuckDB.Simple.Generic` – automatic encoding/decoding of Haskell- ADTs as DuckDB STRUCTs and UNIONs via GHC generics and the `ViaDuckDB`- deriving-via helper.-- `Database.DuckDB.Simple.LogicalRep` – structured value types (`StructValue`,- `UnionValue`) for working with DuckDB's composite types.-- `Database.DuckDB.Simple.Types` – shared types (`Query`, `Null`, `Only`,- `(:.)`, `SQLError`).-- `Database.DuckDB.Simple.Function` – register scalar Haskell functions that- can be invoked directly from SQL.--## Querying Data--```haskell-import Database.DuckDB.Simple-import Database.DuckDB.Simple.Types (Only (..))--fetchNames :: Connection -> IO [Maybe String]-fetchNames conn = do- _ <- execute_ conn "CREATE TABLE names (value TEXT)"- _ <- executeMany conn "INSERT INTO names VALUES (?)"- [Only (Just "Alice"), Only (Nothing :: Maybe String)]- fmap fromOnly <$> query_ conn "SELECT value FROM names ORDER BY value IS NULL, value"-```--The execution helpers return the number of affected rows (`Int`) so callers can-assert on data changes when needed.--## Named Parameters--duckdb-simple supports both positional (`?`) and named parameters. Named-parameters are bound with the `(:=)` helper exported from-`Database.DuckDB.Simple.ToField`.--```haskell-import Database.DuckDB.Simple-import Database.DuckDB.Simple.ToField (NamedParam ((:=)))--insertNamed :: Connection -> IO Int-insertNamed conn =- executeNamed conn- "INSERT INTO events VALUES ($kind, $payload)"- ["$kind" := ("metric" :: String), "$payload" := ("ok" :: String)]-```--DuckDB does not allow mixing positional and named placeholders within the same-SQL statement; the library preserves DuckDB’s error message in that situation.-DuckDB does not support savepoints. This library does not provide `withSavepoint`.--If the number of supplied parameters does not match the statement’s declared-placeholders—or if you attempt to bind named arguments to a positional-only-statement—`duckdb-simple` raises a `FormatError` before executing the query.--### Decoding rows--`FromRow` is powered by a `RowParser`, which means instances can be written in a-monadic/Applicative style and even derived generically for product types:--```haskell-{-# LANGUAGE DeriveAnyClass #-}-{-# LANGUAGE DeriveGeneric #-}--import Database.DuckDB.Simple-import GHC.Generics (Generic)--data Person = Person- { personId :: Int- , personName :: Text- }- deriving stock (Show, Generic)- deriving anyclass (FromRow)--fetchPeople :: Connection -> IO [Person]-fetchPeople conn = query_ conn "SELECT id, name FROM person ORDER BY id"-```--Helper combinators such as `field`, `fieldWith`, and `numFieldsRemaining` are-available when a custom instance needs fine-grained control.--## Generic Encoding with ViaDuckDB--The `Database.DuckDB.Simple.Generic` module provides automatic encoding and-decoding of Haskell algebraic data types as DuckDB STRUCTs and UNIONs via-GHC generics.--### Product Types as STRUCTs--Product types (records) are automatically encoded as DuckDB STRUCT values:--```haskell-{-# LANGUAGE DeriveGeneric #-}-{-# LANGUAGE DerivingVia #-}--import Data.Int (Int64)-import Data.Text (Text)-import Database.DuckDB.Simple-import Database.DuckDB.Simple.Generic (ViaDuckDB (..))-import GHC.Generics (Generic)--data User = User- { userId :: Int64- , userName :: Text- }- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB User)---- Round-trip through the database-storeAndFetchUser :: Connection -> User -> IO [User]-storeAndFetchUser conn user = do- _ <- execute_ conn "CREATE TABLE users (data STRUCT(userId BIGINT, userName TEXT))"- _ <- execute conn "INSERT INTO users VALUES (?)" (Only user)- fmap fromOnly <$> query_ conn "SELECT data FROM users"-```--### Sum Types as UNIONs--Sum types are encoded as DuckDB UNION values, with each constructor becoming a-union member:--```haskell-data Shape- = Circle Double- | Rectangle Double Double- | Point- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB Shape)---- Store and retrieve shape data-storeShape :: Connection -> Shape -> IO [Shape]-storeShape conn shape = do- _ <- execute_ conn- "CREATE TABLE shapes (s UNION(Circle STRUCT(field1 DOUBLE), \- \Rectangle STRUCT(field1 DOUBLE, field2 DOUBLE), Point STRUCT()))"- _ <- execute conn "INSERT INTO shapes VALUES (?)" (Only shape)- fmap fromOnly <$> query_ conn "SELECT s FROM shapes"-```--Nullary constructors (like `Point`) are encoded with a null payload.-Non-record constructors use positional field names (`field1`, `field2`, etc.).--### Arrays and Lists--DuckDB arrays (fixed-length) and lists (variable-length) are also supported:--```haskell-import Data.Array (Array, listArray)--storeArray :: Connection -> IO [Array Int Int]-storeArray conn = do- _ <- execute_ conn "CREATE TABLE arrays (vals INTEGER[3])"- let arr = listArray (0, 2) [1, 2, 3] :: Array Int Int- _ <- execute conn "INSERT INTO arrays VALUES (?)" (Only arr)- fmap fromOnly <$> query_ conn "SELECT vals FROM arrays"--storeList :: Connection -> IO [[Int]]-storeList conn = do- _ <- execute_ conn "CREATE TABLE lists (vals INTEGER[])"- let arr = listArray (0, 2) [1, 2, 3] :: Array Int Int- _ <- execute conn "INSERT INTO lists VALUES (?)" (Only arr)- fmap fromOnly <$> query_ conn "SELECT vals FROM lists"-```--A Haskell list reads a LIST result. Lists have no parameter instance, so bind-an `Array`. DuckDB casts the array to the LIST column type.--An array parameter with scalar elements takes its element type from the-column type name of the element, so an empty array keeps its type. An array-of STRUCT values, UNION values, generic records, or arrays takes the type of-its first element that is not NULL. All other non-NULL elements must have the-same type, including field names, decimal precision and scale, and nested-types. Different types raise an error before DuckDB can cast away fields or-round values. An empty or all-NULL array of such elements raises an error.--### Infinite dates and timestamps--Use `Database.DuckDB.Simple.Time` when a column can contain temporal infinity.-Its `Date`, `LocalTimestamp`, and `UTCTimestamp` types wrap `Day`, `LocalTime`,-and `UTCTime` in `Unbounded`: `NegInfinity`, `Finite value`, or `PosInfinity`.-These types support parameters, results, and fields in generic composites.-`UTCTimestamp` binds as TIMESTAMPTZ.--```haskell-import Database.DuckDB.Simple.Time--infiniteDates :: Connection -> IO [Only Date]-infiniteDates conn = query conn "SELECT ?::DATE" (Only (PosInfinity :: Date))-```--The ordinary `Day`, `LocalTime`, and `UTCTime` instances reject infinity with-a conversion error. Use `Maybe Date` to distinguish SQL NULL from infinity.-Floating-point NaN and infinities remain valid `Float` and `Double` values.--### Manual STRUCT and UNION Handling--Temporal fields retain their SQL units when composite values are rebound.-The `FieldDate`, `FieldTimestamp`, and `FieldTimestampTZ` constructors hold-`Unbounded` values. Wrap finite payloads in `Finite` when constructing them.-For TIMESTAMP_S or TIMESTAMP_MS values outside the TIMESTAMP range, use an-explicit parameter cast, such as `SELECT ?::STRUCT(value TIMESTAMP_S)`.-DuckDB otherwise attempts to convert these parameters to microseconds.--For more control, you can work directly with `StructValue` and `UnionValue`-from `Database.DuckDB.Simple.LogicalRep`:--```haskell-import Database.DuckDB.Simple.LogicalRep (StructValue (..), UnionValue (..))-import Database.DuckDB.Simple.FromField (FieldValue (..))--manualStruct :: Connection -> IO [(StructValue FieldValue, UnionValue FieldValue)]-manualStruct conn = do- _ <- execute_ conn- "CREATE TABLE composite (s STRUCT(a INT, b INT), \- \u UNION(x INT, y VARCHAR))"- [(s, u)] <- query_ conn- "SELECT {'a': 1, 'b': 2}, \- \CAST(union_value(x := 42) AS UNION(x INT, y VARCHAR))"- :: IO [(StructValue FieldValue, UnionValue FieldValue)]- _ <- execute conn "INSERT INTO composite VALUES (?, ?)" (s, u)- query_ conn "SELECT s, u FROM composite"-```--### Resource Management--- `withConnection` and `withStatement` wrap the open/close lifecycle and guard- against exceptions; use them whenever possible to avoid leaking C handles.-- All intermediate DuckDB objects (results, prepared statements, values) are- released immediately after use. Query helpers return a Haskell list of all- rows. Folds and cursors decode rows incrementally, while DuckDB retains the- materialized native result until it is exhausted, reset, or closed.-- `execute`/`query` variants reset statement bindings each run so prepared- statements can be reused safely.--For concurrent workers, use a separate connection per worker. A connection-can move between threads or be shared when the application serializes access.-Hold that lock for the whole transaction or cursor lifetime, including `close`.-DuckDB serializes native query calls, but this does not protect the Haskell-handle state or prevent another call from interfering with an active cursor.-Statements and connections do not provide their own lock. A callback must not-execute another query on its active connection or close that connection.--Link your executable with `ghc-options: -threaded` to allow prompt cancellation-of native queries, including Ctrl-C. On cancellation, the library interrupts-DuckDB and waits for the native call to return before it releases resources-and propagates the exception. Cancellation is cooperative: native code and-Haskell callbacks must return before cleanup can finish.--Close or reset an abandoned statement to release its result. DuckDB can retain-native result buffers until that result is destroyed.-Use `withStatement` for manual iteration. `fold` releases the result after-success or an exception. The accumulator determines Haskell memory use.--### Metadata helpers--- `columnCount` and `columnName` expose prepared-statement metadata so you can- inspect result shapes before executing a query.-- `execute` returns the number of affected rows. Use SQL `RETURNING` clauses- when you need generated identifiers.-### Cursors and folds--`fold`, `fold_`, and `foldNamed` decode one row at a time from DuckDB's result-chunks. DuckDB 1.5 materializes the native result before the first row is-returned. These functions avoid a complete Haskell row list, but native memory-use still depends on the result size. The native API for starting a streaming-result is deprecated; the default interface uses the supported execution API.--```haskell-import Database.DuckDB.Simple.Types (Only (..))--sumValues :: Connection -> IO Int-sumValues conn =- fold_ conn "SELECT n FROM stream_fold ORDER BY n" 0 $ \acc (Only n) ->- pure (acc + n)-```--For manual cursor-style iteration, use `nextRow`/`nextRowWith` on an open-`Statement` to pull rows one at a time and decide when to stop.--Cursors support the same column types as eager queries, including STRUCT-and UNION values with nested collections and NULLs. VARIANT and GEOMETRY-work with eager queries, cursors, and folds.--#### Optional native streaming--`Database.DuckDB.Simple.Deprecated.Streaming` provides `fold`, `fold_`,-`foldNamed`, `nextRow`, and `nextRowWith` with native streaming enabled.-Import it qualified:--```haskell-import qualified Database.DuckDB.Simple.Deprecated.Streaming as Streaming--streamSum :: Connection -> IO Int-streamSum conn =- Streaming.fold_ conn "SELECT i FROM range(1000000) t(i)" 0 $ \acc (Only n) ->- pure (acc + n)-```--The import emits a deprecation warning because DuckDB has deprecated the-execution entry point. DuckDB can still materialize some queries. Streaming-does not bound the memory used by query operators.--The first cursor fetch selects the execution mode until an explicit reset.-Switching between default and streaming `nextRow` calls retains that mode.-Keep the connection dedicated to the active stream; another query on that-connection can invalidate it. Cancellation interrupts native chunk fetching-as well as execution.--This module also provides `foldArrow` and `foldArrow_` for streaming Arrow-batches. They use the same supported Arrow conversion and scoped ownership-as `Database.DuckDB.Simple.Arrow`.--### Arrow batches--`Database.DuckDB.Simple.Arrow` provides `foldArrow` and `foldArrow_` for clients-that consume the Arrow C Data Interface. Each callback receives a separate-schema and array batch. It can read them or pass them to an Arrow consumer-that releases or moves them. The fold releases any remaining contents on-success, failure, or cancellation. The consumer owns any contents it moves.--The original pointers are valid only during the callback. To retain contents,-a consumer must move the root structs into its own storage and set the source-release fields to NULL. Moved contents remain valid after the query and-connection close. Empty results do not produce a callback.--For example, the `dataframe-arrow-bridge` package can copy each batch into a-Haskell `DataFrame` and release the Arrow objects:--```haskell-import qualified DataFrame.IO.Arrow as DataFrame-import qualified Database.DuckDB.Simple.Arrow as Arrow-import Foreign.Ptr (castPtr)--frames <- Arrow.foldArrow_ conn "SELECT id::BIGINT, name::VARCHAR FROM people" [] $ \acc schema array -> do- frame <- DataFrame.arrowToDataframe (castPtr schema) (castPtr array)- pure (frame : acc)--- Reverse frames to recover the query's batch order.-```--The bridge currently imports signed 32-bit and 64-bit integers, Float, Double,-and text columns. It is a test dependency of this repository; applications-that use it must declare their own dependency on `dataframe-arrow-bridge`.--Arrow export uses DuckDB's schema and chunk conversion API. DuckDB materializes-the native result before callbacks start, so its memory use depends on the-result size. The older Arrow query and scan bindings remain available through-`Database.DuckDB.FFI.Deprecated` and emit deprecation warnings.--### VARIANT--A VARIANT result decodes to the `FieldValue` of the stored value, with its-native type. The usual `FromField` instances read VARIANT columns, for example-as `Int64`, `Text`, `[a]`, or a generic record. Arrays decode to `FieldList`.-Objects decode to `FieldStruct` values whose fields have the VARIANT type, in-entry order. `variantObject` builds such a payload from a list of entries.-SQL NULL decodes to `FieldNull`.--`Variant` from `Database.DuckDB.Simple.Variant` wraps a `FieldValue`. Its-`FromField` instance reads any column. Its `ToField` instance binds the-payload as a VARIANT, so `SELECT ?` returns a VARIANT:--```haskell-query conn "SELECT ?" (Only (Variant (FieldList [FieldInt8 1, FieldText "two"])))-```--A scalar payload keeps its native type, such as `TINYINT` or `DECIMAL(4,2)`.-Time and timestamp payloads bind as microsecond types, or as nanosecond types-when they have sub-microsecond digits. So a `TIMESTAMP_S` result binds back as-a `TIMESTAMP` when its value fits. Wider timestamp values use milliseconds or-seconds without losing digits. Lists, arrays, and STRUCT fields bind as-VARIANT values. MAP and ENUM payloads raise an error. Object parameters reject-duplicate keys, empty keys, and keys that contain NUL. You can also bind a-plain value and cast it in SQL, as in `?::VARIANT`.--The C API cannot create a usable VARIANT type-([#27](https://github.com/Tritlo/duckdb-haskell/issues/27)). Each connection-reads the VARIANT type the first time a parameter needs it, and its-parameters use that type. The connection reads the type with a query on a-separate connection. So the query does not run in your transaction, and it-does not end a streaming result.-`Variant` has no `ToDuckValue` instance, and `logicalTypeFromRep` raises an-error for VARIANT, because neither has a connection.--A GEOMETRY payload decodes to `FieldGeometry` with raw WKB and no CRS. Import-these bytes with `ST_GeomFromWKB(?)::VARIANT`. A `Variant` parameter that-contains `FieldGeometry` raises an error, also inside arrays and objects. A-TIMETZ payload with an offset in seconds raises an error, as a TIMETZ column-does. Results can contain empty or NUL object keys. Text and blob payloads can-contain NUL.--DuckDB 1.5 has no C API for reading VARIANT values. The decoder checks the-native version and physical schema before it reads the internal representation.-It checks payload bounds and rejects unknown tags. This format dependency is-limited to the supported DuckDB 1.5 line.--Persistent VARIANT columns require storage format `v1.5.0` or later. For a new-database, pass `[("storage_compatibility_version", "v1.5.0")]` to-`openWithConfig` or `withConnectionWithConfig`. The library does not change an-existing database's storage compatibility setting.--### GEOMETRY--Use `Geometry` from `Data.Geometry` for decoded shapes. The `geometry-simple`-package supplies the type and unboxed coordinate vectors. `duckdb-simple`-supplies its parameter and result instances:--```haskell-import qualified Data.Geometry as G-import qualified Data.Vector.Unboxed as U-import Database.DuckDB.Simple--let line = G.LineString (G.CoordinatesXY (U.fromList [G.XY 0 0, G.XY 1 2, G.XY 3 4]))-rows <- query conn "SELECT ?" (Only line) :: IO [Only G.Geometry]-```--Points and coordinate sequences store their XY, XYZ, XYM, or XYZM layout.-`G.EmptyPoint G.DimXYZ` is an empty XYZ point. Use `Maybe G.Geometry` for SQL NULL.-Use `decodeWKT` from `Data.Geometry.WKT` to parse text without a database.--`G.Geometry` has no CRS metadata. Reading it returns the coordinates without-their CRS label. Binding it creates a `GEOMETRY` with no CRS. To attach a label,-use `ST_SetCRS(?, ?)` with a shape and CRS text. This does not transform coordinates.--Use `RawGeometry` from `Database.DuckDB.Simple.Geometry` to read WKB and CRS-without decoding coordinates. Its `rawGeometryWKB` and `rawGeometryCRS` fields-own their data. They remain usable after the connection closes. Existing-`ByteString` result decoding also returns WKB.--`RawGeometry` has no `ToField` instance. Import its bytes explicitly:--```haskell-import Database.DuckDB.Simple.Geometry--[Only raw] <- query_ conn "SELECT 'POINT Z (1 2 3)'::GEOMETRY('OGC:CRS84')"-rows <- (case rawGeometryCRS raw of- Nothing -> query conn- "SELECT system.main.ST_GeomFromWKB(?)"- (Only (rawGeometryWKB raw))- Just crs -> query conn- "SELECT system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?)"- (rawGeometryWKB raw, crs)- ) :: IO [Only RawGeometry]-```--Use `Nothing` for no CRS and omit `ST_SetCRS` in that case. Passing SQL NULL-to `ST_SetCRS` returns a NULL geometry. CRS text can contain `OGC:CRS84`,-a custom name, or a full WKT2/PROJJSON definition. DuckDB can reduce a known-CRS definition to its registered identifier. It can also normalize WKB byte-order. The qualified function names select DuckDB's built-ins even if a user-macro has the same name.--`fromRawGeometry` decodes the shape and drops CRS metadata. `toRawGeometry`-encodes a shape with no CRS. These helpers follow the geometry-simple contracts.-NaN in both WKB point X and Y denotes an empty point. Empty multi-geometries-and collections have no stored layout tag in the decoded representation.-Keep the raw bytes when these details must survive.--Structured parameters use `encodeWKT` and DuckDB's native cast. DuckDB 1.5-has no WKB value constructor in its C API. The WKT writer combines layouts-in multi-geometries and polygon rings. It fills absent Z or M with NaN.-DuckDB's WKT parser limits nesting to 16 levels and rejects empty polygon rings-and mixed collection layouts. Explicit WKB import preserves mixed member-layouts and native NaN payload bits that WKT cannot retain.--`FieldGeometry` contains a raw result. Generic STRUCT and UNION parameters-that contain non-NULL raw geometry raise an error. Construct those values in-SQL with explicit WKB import. Their result decoding retains WKB and CRS,-including inside LIST, ARRAY, MAP, STRUCT, and UNION values.--`LogicalTypeGeometry` describes CRS metadata. The C API cannot create a-GEOMETRY type with a CRS. So each connection reads a GEOMETRY type for each-CRS in a list, together with the VARIANT type, the first time a parameter-needs one of them. The default list holds-`OGC:CRS84`. Composite parameters use these types, so typed NULLs, empty-collections, and inactive UNION members keep their CRS. Set the list in-`ConnectionOptions`:--```haskell-let options = defaultConnectionOptions{connectionGeometryCRS = ["OGC:CRS84", "EPSG:3857"]}-withConnectionWithOptions "shapes.duckdb" options \conn -> ...-```--A list entry can be an identifier, a custom name, or a full WKT2 or PROJJSON-definition. A composite parameter with a CRS that is not in the list binds its-geometry members without a CRS. `logicalTypeFromRep` has no connection, so it-creates `GEOMETRY` without a CRS. To apply a CRS in these cases, insert the-value into a column with that CRS, or cast it in SQL:--```haskell-query conn "SELECT ?::UNION(number BIGINT, shape GEOMETRY('OGC:CRS84'))" (Only value)-```--The CRS in a cast must be a constant. DuckDB rejects a parameter as a type-modifier, so `?::GEOMETRY(?)` is not valid. To use a CRS that is known only at-run time, write it into the query text as a SQL string literal, and double each-single quote. A cast accepts a CRS that DuckDB recognizes, such as `OGC:CRS84`,-or a full WKT2 or PROJJSON definition. Unless an extension recognizes it,-DuckDB rejects other identifiers, such as `EPSG:4326`, and custom names.-`ST_SetCRS` also accepts custom names, but it returns `GEOMETRY` with no CRS-for a NULL input.--A bound composite without a CRS also changes the type of expressions that-combine it with CRS data. `UNION ALL` and `COALESCE` of `GEOMETRY` and-`GEOMETRY('OGC:CRS84')` give `GEOMETRY` with no CRS for all rows. Cast the-parameter before you combine it with other values.--See [geometry-simple](https://github.com/Tritlo/geometry-simple) for the seven-supported families, construction checks, and codec normalization rules.--### Feature Coverage--- Connections, prepared statements, positional/named parameter binding.-- High-level execution (`execute*`) and eager queries (`query*`, `queryNamed`).-- Cursor and fold helpers (`fold`, `foldNamed`, `fold_`, `nextRow`) that decode- native result chunks one row at a time.-- Comprehensive scalar type support: signed/unsigned integers, HUGEINT/UHUGEINT,- decimals (with width/scale), intervals, precise and timezone-aware temporals,- enums, bit strings, blobs, bignums, and UUIDs.-- Composite types: STRUCTs, UNIONs, LISTs, fixed-length ARRAYs, and MAPs with- typed parameters and results.-- Generic encoding/decoding: automatic STRUCT/UNION mapping for Haskell ADTs via- GHC generics and the `ViaDuckDB` deriving-via helper.-- Row decoding via `FromField`/`FromRow`, with generic deriving for product types.-- User-defined scalar functions backed by Haskell functions (including IO and- nullable arguments).-- Transaction helpers (`withTransaction`) and metadata accessors (`columnCount`,- `columnName`).--## User-Defined Functions--Scalar Haskell functions can be registered with DuckDB connections and used in-SQL expressions. Argument and result types reuse the existing `FromField` and-`FunctionResult` machinery, so `Maybe` values and `IO` actions work out of the-box.--```haskell-import Data.Int (Int64)-import Database.DuckDB.Simple-import Database.DuckDB.Simple.Function (createFunction, deleteFunction)-import Database.DuckDB.Simple.Types (Only (..))--registerAndUse :: Connection -> IO [Only Int64]-registerAndUse conn = do- createFunction conn "hs_times_two" (\(x :: Int64) -> x * 2)- result <- query_ conn "SELECT hs_times_two(21)" :: IO [Only Int64]- deleteFunction conn "hs_times_two"- pure result-```--Exceptions raised while the function executes are propagated back to DuckDB as-`SQLError` values, and `deleteFunction` issues a `DROP FUNCTION IF EXISTS`-statement to remove the registration. DuckDB registers C API scalar functions-as internal entries; attempting to drop them this way will yield an error, which-the library surfaces as an `SQLError`.--## Tests--The test suite is built with [tasty](https://hackage.haskell.org/package/tasty)-and covers connection management, statement lifecycle, parameter binding, and-query execution.--```-cabal test duckdb-simple-test --test-show-details=direct-```+`duckdb-simple` is a Haskell interface to DuckDB. Its API is like the APIs of+[`sqlite-simple`](https://hackage.haskell.org/package/sqlite-simple) and+[`postgresql-simple`](https://hackage.haskell.org/package/postgresql-simple).+It uses the [`duckdb-ffi`](https://hackage.haskell.org/package/duckdb-ffi) bindings.+It opens connections, runs queries, binds parameters, and decodes results into+Haskell types. It supports DuckDB scalar types, for example signed and unsigned integers,+decimals, HUGEINT and UHUGEINT, intervals, dates, times, and timestamps of each+precision and with a time zone, blobs, enums, bit strings, and bignums.++The backend uses the generated DuckDB V2 API. Upstream marks the complete V2+API as unstable: "The C API v2 is not yet stable. Its interface may change in+upcoming releases." This release therefore requires exactly DuckDB 2.0.0.+Creating an environment raises `SQLError` for another library version.++The examples use the `OverloadedStrings` extension for `Query` and `Text`+literals.++## Shared environments and instances++Use an explicit instance when several connections must share catalogs and+configuration. Each connection has at most one open native result. Use a+separate connection for each concurrent worker.+Wait for child creation and all worker operations to finish before calling+`closeInstance` or `closeEnvironment`. Closing a parent also closes its children.++```haskell+withEnvironment $ \environment ->+ withInstance environment [("threads", "2")] $ \instance_ -> do+ attachDatabase instance_ ":memory:" "shared" []+ setDefaultDatabase instance_ "shared"+ withConnectionOn instance_ $ \writer -> do+ execute_ writer "CREATE TABLE items(id BIGINT)"+ execute writer "INSERT INTO items VALUES (?)" (Only (42 :: Int))+ withConnectionOn instance_ $ \reader ->+ query_ reader "SELECT id FROM items" :: IO [Only Int]+```++Closing an environment closes its instances, connections, and statements.+DuckDB can keep an instance alive that `createInstance` did not create. In+that case, `closeEnvironment` raises an `SQLError` and keeps the environment open.+Closing an instance closes its connections and statements. Each close operation+is idempotent. `open` and `withConnection` continue to create private parents.++## V2 facilities++- `Config` lists options with effective values, typed defaults, aliases, and+ supported scopes. It reads and writes typed values and registers custom+ options. The update function of a custom option gets the new value as a+ `FieldValue`. `V2.Config` registers custom options with native builders.+- `Catalog.describeTable` copies resolved names, column types, default flags,+ generated-column flags, and catalog read-only status. A column type keeps the+ alias of DuckDB at each depth as a `LogicalTypeAlias`, for example JSON and+ the name of a type from `registerCustomType`. A column of a type from+ `CREATE TYPE name AS base` has the base type, and an ENUM from `CREATE TYPE`+ has no name.+- `V2.LogicalRep.logicalTypeToRep factory logical` copies a native logical type+ into a pure `LogicalTypeRep`. Keep the factory valid throughout the+ conversion. The function does not destroy the logical type handle.+- `Tokenizer.tokenize` returns token kinds, UTF-8 byte offsets, token bytes, and+ the native indication of an unterminated final token.+- `Function` registers scalar, aggregate, and table functions with+ Haskell values. `Registration` registers custom types and replacement scans,+ and has the type and parameter declarations. See+ [Typed registration](#typed-registration).+- `V2.Registration` supports scalar, aggregate, table, cast, and custom-type+ registration, replacement scans, function properties, and parameter kinds,+ names, and defaults. Haskell callbacks use typed V2 info handles. The library+ retains their function pointers and contains callback exceptions. The advanced+ extension builders accept a borrowed native extension handle and context.+ Scalar type resolution runs before binding. Table callbacks can report+ multi-file column identifiers and metadata after binding.+- `Copy` registers COPY formats and multi-file table functions with Haskell+ values. `V2.Copy` has the native COPY callbacks, with batch transfer, flush,+ statistics, progress, and multi-file setup.+- `FileSystem` supports sequential and positioned reads and writes, file size,+ position, seek, sync, path tests, expansion, globbing, movement, and removal.+- `Logging` registers native log sinks with managed write and flush callbacks.+ It copies timestamps, severity, scope, and available context identifiers.+ Callbacks can run concurrently. They must not log or change logging settings.+ `V2.Logging` writes a message through the context of a native callback.+- `Pragma` registers query and execution callbacks. A callback gets the+ arguments of the call as a `FromRow` row. `V2.Pragma` has the callbacks that+ get native handles.+- `Extension` has the typed registrations for a loadable extension. It also+ registers scalar and table macros during extension initialization. The+ native macro builder has no connection target; use SQL `CREATE MACRO` on a+ connection. `V2.Registration.registerMacroFunctionWith` gives the native+ macro builder to a configuration action.+- `V2.FunctionDocs` sets descriptions, examples, and categories on function+ builders.+- `QueryMetrics` enables profiling and copies query and operator metric trees.+- `Internal.withCollectionStatement` binds native column collections as named+ table inputs for one scoped prepared-statement action. The caller retains the+ collections.++The pinned V2 API does not expose custom file-system registration or collation+lookup. DuckDB 2.0 rejects an ARRAY type with zero elements. An empty LIST is+valid. The direct `query_` and `execute_` paths support statements that expand+during parsing, including dynamic PIVOT. Prepared dynamic PIVOT supports+repeated execution with a changing schema. It does not support parameters or a+cacheable plan. Read its schema after execution starts. VARIANT decoding uses+SQL metadata and native casts on a helper connection. An unknown custom+GEOMETRY CRS label uses a helper SQL query when a parameter needs its type.+GEOMETRY results use a native cast to BLOB. Native GEOMETRY collections require+the same Z/M dimensions on every member, including empty members. The standalone WKB codec+retains mixed layouts.++## Getting started++```haskell+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}++import Database.DuckDB.Simple++main :: IO ()+main =+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE items (id INTEGER, name TEXT)"+ _ <- execute conn "INSERT INTO items VALUES (?, ?)" (1 :: Int, "banana" :: String)+ rows <- query_ conn "SELECT id, name FROM items ORDER BY id"+ mapM_ print (rows :: [(Int, String)])+```++### Key modules++- `Database.DuckDB.Simple`: connections, prepared statements, execution,+ queries, metadata, and error handling.+- `Database.DuckDB.Simple.ToField` / `ToRow`: typeclasses and helpers for+ preparing positional or named parameters.+- `Database.DuckDB.Simple.FromField` / `FromRow`: typeclasses for decoding+ query results, with generic deriving support for product types.+- `Database.DuckDB.Simple.Conversion`: explicit IO and conversion errors for+ custom decoders. `runConversion` returns `IO (Ok a)`.+- `Database.DuckDB.Simple.Generic`: automatic encoding/decoding of Haskell+ ADTs as DuckDB STRUCTs and UNIONs via GHC generics and the `ViaDuckDB`+ deriving-via helper.+- `Database.DuckDB.Simple.LogicalRep`: structured value types (`StructValue`,+ `TupleValue`, `UnionValue`) for working with DuckDB's composite types.+- `Database.DuckDB.Simple.Types`: shared types (`Query`, `Null`, `Only`,+ `(:.)`, `SQLError`).+- `Database.DuckDB.Simple.Function`: register scalar Haskell functions that+ SQL can call, and casts, aggregate functions, and table functions.+- `Database.DuckDB.Simple.Registration`, `Copy`, `Pragma`, and `Config`:+ register custom types, replacement scans, COPY formats, pragmas, and custom+ options with Haskell values.++[Types](#types) lists the SQL type of each Haskell type.+[Errors](#errors) describes the exceptions that a program can catch.++## Querying data++```haskell+import Database.DuckDB.Simple++fetchNames :: Connection -> IO [Maybe String]+fetchNames conn = do+ _ <- execute_ conn "CREATE TABLE names (value TEXT)"+ _ <- executeMany conn "INSERT INTO names VALUES (?)"+ [Only (Just "Alice"), Only (Nothing :: Maybe String)]+ fmap fromOnly <$> query_ conn "SELECT value FROM names ORDER BY value IS NULL, value"+```++The execution helpers return the number of affected rows (`Int`). Callers can+use this count to check data changes.++An `Integer` or a `Natural` parameter binds as a HUGEINT, a UHUGEINT, or a+BIGNUM value, or raises an error, depending on its size and target type:+note 1 of [Types](#types) gives the rule.++In DuckDB 2.0.0 the casts from BIGNUM to the integer types have defects: some+fail, and some store a wrong value with no error ([issue+41](https://github.com/Tritlo/duckdb-haskell/issues/41)). A `SizedArray` or a+list of `Integer` or `Natural` values takes the element type of its target when+that type is an integer type or DECIMAL, so DuckDB checks each element. A+`BigNum` value, and a `SizedArray` or a list of `BigNum` values, still have+these defects for an integer target. For each other target, and for a parameter+with no known type, a `SizedArray`, a list, or a `Map` of `Integer` or `Natural`+values takes the type that its elements give, as a scalar `Integer` and a SQL+list literal of the HUGEINT and BIGNUM values do: `[1, 2^64]` binds as+`HUGEINT[]`, and `[1, 2^200]` as `BIGNUM[]`. An empty list and a list of only+NULL elements get the type `"NULL"[]`, as `[]` and `[NULL]` in SQL, because an+`Integer` has no fixed SQL type. The rule follows the column type name BIGNUM,+so an all-NULL list of `BigNum` values with no target also gets `"NULL"[]`.++An `IntervalValue` binds as an INTERVAL, and DuckDB keeps its months, days, and+microseconds. DuckDB compares intervals with 30 days for each month, so+`INTERVAL 30 DAY = INTERVAL 1 MONTH`. A `TimeWithZone` binds as a TIMETZ with+its time and its offset in seconds east of UTC (`timeWithZoneOffsetSeconds`).+`timeWithZone` makes a value from a `Data.Time` `TimeZone`, and+`timeWithZoneZone` gives the zone of a value, with the seconds of its offset+dropped. DuckDB stores an offset from -15:59:59 to +15:59:59, that is from+-57599 to 57599 seconds, and a value reads back equal to the bound value. A time+with digits below one microsecond loses these digits, as in SQL text. A+`DecimalValue` binds as a DECIMAL with its own width and scale. DuckDB casts it+to the type of its target: a smaller scale rounds half away from zero, and a+value that the target cannot hold raises an error. A `SizedArray`, a list, or a+typed row (a row type with `TypedRow`) of `DecimalValue` does not compile: note+14 of [Types](#types).++`query_` and `execute_` parse all input statements before execution. A syntax+error prevents every statement in that input from running. An execution error+can leave effects from earlier statements. Use `withTransaction` when those+effects must roll back together.++## Named parameters++duckdb-simple supports both positional (`?`) and named parameters. Bind a+named parameter with `(:=)` from `Database.DuckDB.Simple`.++```haskell+import Database.DuckDB.Simple++insertNamed :: Connection -> IO Int+insertNamed conn =+ executeNamed conn+ "INSERT INTO events VALUES ($kind, $payload)"+ ["$kind" := ("metric" :: String), "$payload" := ("ok" :: String)]+```++DuckDB does not allow positional and named placeholders in the same SQL+statement. The library preserves DuckDB's error message in that case.+DuckDB does not support savepoints. This library does not provide `withSavepoint`.++Parameter names use DuckDB's ASCII case rules. For example, `$amount` and+`$AMOUNT` refer to the same parameter. Supplying both raises a `FormatError`.+Non-ASCII characters retain their case. Session variables+can supply omitted named parameters. DuckDB rejects an omitted parameter+that has no variable value.++`duckdb-simple` raises a `FormatError` before it executes the query in these+cases:++- The number of positional parameters is not the number of placeholders.+- You bind positional parameters to a statement with named placeholders.+- You supply more named parameters than the statement declares.+- You supply a name that the statement does not declare.+- You supply one parameter two times, for example as `$amount` and `$AMOUNT`.+- You bind named parameters to a statement with only positional placeholders.++`bind` and `bindNamed` reset the statement cursor before validation. If+validation or value conversion fails, the previous set of bindings remains.+A successful `bindNamed` replaces the explicit bindings. Current session+variables can supply omitted parameters when the statement executes.++### Decoding rows++The `fromRow` method of a `FromRow` instance is a `RowParser`. Write it with+`Applicative` or `Monad` functions, or derive the instance with `Generic` for a+product type:++```haskell+{-# LANGUAGE DeriveAnyClass #-}+{-# LANGUAGE DeriveGeneric #-}+{-# LANGUAGE DerivingStrategies #-}+{-# LANGUAGE OverloadedStrings #-}++import Data.Text (Text)+import Database.DuckDB.Simple+import GHC.Generics (Generic)++data Person = Person+ { personId :: Int+ , personName :: Text+ }+ deriving stock (Show, Generic)+ deriving anyclass (FromRow)++fetchPeople :: Connection -> IO [Person]+fetchPeople conn = query_ conn "SELECT id, name FROM person ORDER BY id"+```++Use `field`, `fieldWith`, and `numFieldsRemaining` for more control in a custom+instance.++A `FromField` instance reads its Haskell type's SQL type, for example BIGINT+for `Int64`. It also reads other SQL types when it can return the value of+DuckDB's cast. The "Reads" column and notes 3 to 5 of [Types](#types) list+these types.++Decoders return `Conversion a`. This follows the `postgresql-simple` style:+a conversion can run IO and report conversion errors. This library's+`runConversion` takes no connection argument. The query and callback APIs run+conversions for you. The decoder types are:++```haskell+type FieldParser a = Field -> Conversion a++fromField :: FromField a => Field -> Conversion a+parseRow :: RowParser a -> [Field] -> Conversion a+runConversion :: Conversion a -> IO (Ok a)+```++For example, `runConversion (fromField field :: Conversion Int64)` returns+`Ok value` or `Errors errors` in IO. Import the runner from+`Database.DuckDB.Simple.Conversion` and the result constructors from+`Database.DuckDB.Simple.Ok`. Use `liftConversion` to run IO inside a custom+decoder. IO exceptions, including cancellation, propagate to the caller.+Sequential and applicative conversions stop at the first conversion failure.+`Alternative` tries the second branch if the first fails. If both fail, it+combines their diagnostics.++A decoder for a registered function's arguments runs inside its callback.+Do not run another query on that callback's connection from the decoder. Use+another connection. See [User-defined functions](#user-defined-functions)+for the connection rules.+`fieldType` reads its scoped logical type without taking the connection lock.++`fieldType :: Field -> Conversion LogicalTypeRep` reads the SQL type of a+column, or of an element, key, or value in a LIST, ARRAY, or MAP. It preserves+aliases such as JSON. A query result or callback owns the native metadata.+The library converts a column's type to `LogicalTypeRep` when a decoder+requests `fieldType` and caches a successful result within that scope. This+metadata action does no tree traversal or ENUM dictionary copy until requested.+Structured value readers can also build type trees to decode their values.+If cancellation interrupts a metadata read while the scope remains open,+another call can retry it. The cache does not store exceptions.++Native metadata is available while a query parser runs. A cursor retains its+metadata until exhaustion, reset, or statement closure. Callback metadata is+available during that callback invocation. An escaped native `fieldType`+action raises `SQLError` with the message+`duckdb-simple: field type scope is closed`, including when its result was+cached before closure. Read a pure+`LogicalTypeRep` snapshot inside the scope if you need the type later.+Snapshots remain usable after the statement, connection, instance, and+environment close.++Metadata scopes release their native types when they close. They use no+metadata finalizers. Existing statement and connection finalizers are+unchanged. `Field` has no `Eq` or `Show` instance. A value in a VARIANT has the+type VARIANT, and so does a `Field` made from a `FieldValue` alone. In those+cases, the stored value determines the type. `logicalTypeSQL` gives the SQL+type text, for example DECIMAL(18,3). A custom decoder can check the type:++```haskell+newtype Cents = Cents DecimalValue++instance FromField Cents where+ fromField field = do+ logical <- fieldType field+ case logical of+ LogicalTypeDecimal _ 2 -> Cents <$> fromField field+ _ -> returnError Incompatible field "expected a DECIMAL with the scale 2"+```++A field of a `ViaDuckDB` record reads with its type when its `DuckValue`+instance sets `duckFromTypedField = typedFromField`. The library's instances+do this. Generic decoders have these types:++```haskell+duckFromField :: DuckValue a => FieldValue -> Conversion a+duckFromTypedField :: DuckValue a => LogicalTypeRep -> FieldValue -> Conversion a+typedFromField :: FromField a => LogicalTypeRep -> FieldValue -> Conversion a+genericFromFieldValue :: (Generic a, GFromField (Rep a)) => FieldValue -> Conversion a+```++Encoding and logical type declarations remain pure.++## Types++Each row gives a Haskell type and these facts:++- Binds as: the SQL type of a parameter of this type.+- Declared: the SQL type that a typed registration declares, if it differs+ from "Binds as". This applies to columns and parameters of+ `createTableFunction`, and arguments and results of `createAggregate` and+ `createFunction`. "none" means that the type has no declaration.+ `registerCopyFromFunction` declares no type. The COPY FROM table supplies the+ column types. The row type must have the same number of columns. DuckDB casts+ each row value to its column's type.+- Reads: the SQL types that the `FromField` instance reads.+- Function: the use of the type in `createFunction`, as an argument, a result,+ or both. See the paragraph after the notes.++| Haskell type | Binds as | Declared | Reads | Function |+| --- | --- | --- | --- | --- |+| `Bool` | BOOLEAN | | BOOLEAN, TINYINT to BIGINT (2) | argument, result |+| `Int8`, `Int16`, `Int32`, `Int64`, `Int` | TINYINT, SMALLINT, INTEGER, BIGINT, BIGINT | | TINYINT to BIGINT, HUGEINT, UHUGEINT (3) | argument, result |+| `Word8`, `Word16`, `Word32`, `Word64`, `Word` | UTINYINT, USMALLINT, UINTEGER, UBIGINT, UBIGINT | | each integer type (3) | argument, result |+| `Integer` | HUGEINT (1) | | each integer type, BIGNUM (3) | argument, result |+| `Natural` | HUGEINT (1) | UHUGEINT (10) | each integer type, BIGNUM (3) | argument, result |+| `BigNum` | BIGNUM | | BIGNUM | argument, result |+| `Double` | DOUBLE | | DOUBLE, FLOAT, TINYINT to BIGINT, DECIMAL (4) | argument, result |+| `Float` | FLOAT | | as `Double` (4) | argument, result |+| `DecimalValue` | DECIMAL of the value | none (14) | DECIMAL | none |+| `Text`, `String` | VARCHAR | | VARCHAR, TINYINT to BIGINT, BOOLEAN, ENUM (5) | argument, result |+| `ByteString` | BLOB | | BLOB, BIT, GEOMETRY (5) | argument, result |+| `BitString` | BIT | | BIT | argument, result |+| `UUID` | UUID | | UUID, VARCHAR (5) | argument, result |+| `Day`, `Unbounded Day` | DATE (7) | | DATE, each timestamp type (6) | argument, result |+| `TimeOfDay` | TIME (7) | | TIME, TIME_NS, each timestamp type (6) | argument, result |+| `LocalTime`, `Unbounded LocalTime` | TIMESTAMP (7) | | each timestamp type (6), DATE (5) | argument, result |+| `UTCTime`, `Unbounded UTCTime` | TIMESTAMPTZ (7) | | TIMESTAMPTZ, TIMESTAMPTZ_NS (5) | argument, result |+| `TimeWithZone` | TIMETZ | | TIMETZ | argument, result |+| `IntervalValue` | INTERVAL | | INTERVAL | argument, result |+| `[a]` | LIST | | LIST, ARRAY | argument, result |+| `SizedArray n a` | ARRAY of size `n` (8) | | ARRAY, LIST (8) | argument, result |+| `Map k v` | MAP | | MAP | argument, result |+| `StructValue FieldValue` | STRUCT (9) | none | STRUCT | none |+| `TupleValue FieldValue` | TUPLE (9) | none | TUPLE | none |+| `UnionValue FieldValue` | UNION (9) | none | UNION | none |+| `ViaDuckDB a` | STRUCT or UNION (10) | | STRUCT or UNION | argument, result |+| `Maybe a` | as `a`, or NULL | as `a` | NULL, or as `a` | as `a` |+| `G.Geometry` | GEOMETRY without a CRS | | GEOMETRY | argument, result |+| `RawGeometry` | GEOMETRY with its CRS | none (14) | GEOMETRY | none |+| `Variant` | VARIANT | | VARIANT, and the values that a VARIANT can hold (13) | argument, result |+| `FieldValue` | the type of its constructor (11) | none | each type | none |+| `Null` | `"NULL"` (12) | | NULL | argument, result (12) |++Notes:++1. A value from -2^127 to 2^127 - 1 binds as a HUGEINT. DuckDB then checks that+ the value fits its target: for example, a negative value for an unsigned+ column raises an error. A value outside that range binds as a UHUGEINT for a+ UHUGEINT parameter that can hold it. It binds as a BIGNUM for a parameter+ that has no integer or DECIMAL type. For each other integer or DECIMAL+ parameter, it raises an `SQLError` of the type `ErrorTypeConversion`. Bind a+ `BigNum` to bind a BIGNUM value in each case. [Querying Data](#querying-data)+ gives the defects of the casts from BIGNUM in DuckDB 2.0.0, and the rule for+ a list, a `SizedArray`, or a `Map` of `Integer` values.+2. The value 0 reads as `False`, and each other value reads as `True`.+3. "Each integer type" is TINYINT to UHUGEINT. A value that the Haskell type+ cannot hold raises a library conversion error: an `SQLError` with no type.+ This includes a negative value for an unsigned type or for `Natural`. An ENUM+ value does not read as an integer type. Write `enum_code(c)::BIGINT` in SQL+ for its index, or `c::INTEGER` for the value of its label.+4. A read returns the value of DuckDB's `c::DOUBLE` or `c::FLOAT` cast. This+ includes a DECIMAL, and a BIGINT read as `Float`. As seen in CI, the DuckDB+ 2.0.0 build for macOS on Apple Silicon rounds some BIGINT values above 2^53 two+ times in its cast to FLOAT, and the library gives the nearest value there. A+ finite value that `Float` cannot hold raises a conversion error of the+ library, an `SQLError` with no type.+5. A read of another SQL type returns the value of DuckDB's cast to the Haskell+ type's SQL type, or an error if the value does not fit. For+ example, `Int8` reads a BIGINT that fits in a TINYINT, and `Double` reads a+ DECIMAL with the value of `c::DOUBLE`. Thus `Text` reads a BOOLEAN as `true`+ or `false`, and `Text` and `String` read an ENUM as its label, the text of+ `c::VARCHAR`; a `FieldEnum` holds the index and the label. `ByteString` reads+ a BIT or a GEOMETRY as the bytes of `c::BLOB`. `UUID` reads a VARCHAR in the+ form `a0eebc99-9c0b-4ef8-bb6d-6bb9bd380a11`, and another form that the cast+ reads, for example with braces, raises an error. A read that needs the+ TimeZone setting of the session, or another text than the cast gives, raises+ the error of an incompatible column. For each such read, write the SQL in the+ parentheses in place of the column `c` (note 3 gives the SQL for an ENUM read+ as an integer type):+ - `Text` from FLOAT or DOUBLE (`c::VARCHAR`).+ - `ByteString` from VARCHAR (`encode(c)`).+ - `LocalTime` from TIMESTAMPTZ or TIMESTAMPTZ_NS (`c AT TIME ZONE 'UTC'`).+ - `UTCTime` from DATE, TIMESTAMP_S, TIMESTAMP_MS, or TIMESTAMP+ (`c::TIMESTAMP AT TIME ZONE 'UTC'`), and from TIMESTAMP_NS+ (`c AT TIME ZONE 'UTC'`).++ The `Unbounded` forms of `LocalTime` and `UTCTime` follow the same rule. The+ Haddock of each `FromField` instance gives the SQL types that it reads.+6. "Each timestamp type" is TIMESTAMP_S, TIMESTAMP_MS, TIMESTAMP, and+ TIMESTAMP_NS. TIMESTAMP alone is the type with microseconds. `Day` reads the+ date of a timestamp, and `TimeOfDay` reads its time. A nanosecond type keeps+ its nanoseconds. An infinite value raises a conversion error of the library,+ an `SQLError` with no type, for `Day`, `TimeOfDay`, `LocalTime`, and+ `UTCTime`. The `Unbounded` forms read it. Use `Maybe Date` to distinguish SQL+ NULL from infinity. `Date`, `LocalTimestamp`, and `UTCTimestamp` from+ `Database.DuckDB.Simple.Time` are the `Unbounded` forms.+7. A value with digits below a microsecond keeps them for a nanosecond+ parameter, and as a scalar value that fits the range of the nanosecond type+ for a `VARIANT` parameter: [Infinite dates and+ timestamps](#infinite-dates-and-timestamps) gives the rule. The range of+ `Day` is 5877642-06-25 (BC), the year -5877641 of `Data.Time`, to+ 5881580-07-10. The range of `LocalTime` and `UTCTime` is 290309-12-22 (BC)+ 00:00:00 to 294247-01-10 04:00:54.775806. A value outside its range raises an+ `SQLError` of the type `ErrorTypeConversion`. A target of the type+ TIMESTAMP_NS or TIMESTAMPTZ_NS has a smaller range, about the years 1677 to+ 2262, and a value outside it gets the same type in each position. A+ `TimeOfDay` with 60 seconds binds as the next second, as `make_time` of+ DuckDB gives it: 23:59:60 binds as 24:00:00. A later time, for example+ 23:59:60.5, and a field outside its own range, for example 61 minutes, raise+ an `SQLError` of the type `ErrorTypeConversion`. [Infinite dates and+ timestamps](#infinite-dates-and-timestamps) gives the full rule.+8. A DuckDB ARRAY type has a size, for example `BIGINT[3]`. A `SizedArray n a`+ is the Haskell type of that ARRAY: an `Array Int a` of exactly `n` values,+ with the bounds `(0, n - 1)`. `sizedArray` makes one from an `Array` with+ any bounds and gives `Nothing` for an `Array` of another count. The value+ has the elements in their order, and `sizedArrayToArray` gives its `Array`.+ `sizedArrayFromList` makes one from a list, and `toList` gives the values.+ A `SizedArray`+ binds as an ARRAY of size `n`, and a typed registration declares that type+ in each position, also as a field of a record through `ViaDuckDB`. It reads+ an ARRAY and a LIST of `n` elements, as the cast of DuckDB to the ARRAY+ does. A query that reads a value of another length raises an `SQLError`.+ Its message names the column type, the `SizedArray` type, and the length.+ In a typed registration, DuckDB rejects an ARRAY of another size when it+ binds the call. It casts a LIST of the same length to the ARRAY with no+ error, by its own implicit cast, and a LIST of another length raises an+ `SQLError` when the statement runs. Use one type for each need:+ - An ARRAY of a fixed size: a `SizedArray n a`.+ - An ARRAY of a size that varies: a list. A list reads an ARRAY of each+ size, and DuckDB casts a bound list to an ARRAY column of the same+ length.+ - An ARRAY value that the program builds at run time: a `FieldArray`.++ A DuckDB ARRAY has 1 to 100000 elements. A `SizedArray` or a `FieldArray`+ parameter with no element or with more than 100000 elements raises an+ `SQLError` of the type `ErrorInputInvalid`, and a typed registration raises+ the same error for a `SizedArray` of such a size. Bind an empty list for an+ empty LIST.+9. A STRUCT, TUPLE, or UNION parameter binds with the types of its own fields+ or members. These types keep an alias, so a value that a query reads from a+ `STRUCT(k JSON)` column binds as `STRUCT(k JSON)`. These values have no+ static field types, so a typed registration with such a type raises an+ error. Use a record with `ViaDuckDB`.+10. A record binds as a STRUCT, and a sum type binds as a UNION. A typed+ registration declares the STRUCT or UNION type of a record or a sum type+ that derives `DuckDBColumnType` and `ToDuckValue` via `ViaDuckDB`, also in a+ list or a MAP. Each field type comes from `DuckValue`. An `Integer` declares+ HUGEINT and a `Natural` declares UHUGEINT, as a field, a column of a typed+ row, an argument, and a result. A value that this type cannot hold raises an+ `SQLError` of the type `ErrorTypeConversion`: use `BigNum`, which declares+ BIGNUM, for a larger value. DuckDB computes with these types: HUGEINT+ arithmetic stays HUGEINT and raises an error on overflow. DuckDB casts a+ non-negative integer literal to a UHUGEINT argument, but no signed column or+ typed value without an explicit cast. `sum()` over a UHUGEINT column gives a+ DOUBLE value, and so do UHUGEINT + INTEGER and HUGEINT + UHUGEINT for+ columns and typed values. A `SizedArray` field declares the ARRAY of its+ size (see note 8). A `duckLogicalType`+ that gives a `LogicalTypeAlias`, for example+ `LogicalTypeAlias "JSON" (LogicalTypeScalar LogicalTypeIdVarchar)`, declares+ that alias type in a typed registration and in a parameter, and a value of+ the field reads as its base type. See+ [Generic Encoding with ViaDuckDB](#generic-encoding-with-viaduckdb).+11. Each constructor binds as its SQL type, with these exceptions:+ - A `FieldEnum` binds its label as a VARCHAR.+ - A `FieldTimestamp` binds as TIMESTAMP, also when it came from another+ timestamp type. [Infinite dates and+ timestamps](#infinite-dates-and-timestamps) gives the rule for digits+ below a microsecond.+ - An empty `FieldArray` raises an `SQLError` of the type+ `ErrorInputInvalid`, because a DuckDB ARRAY has at least one element.+ - An empty `FieldList` or `FieldMap`, and a `FieldList`, a `FieldArray`, or+ a `FieldMap` with only `FieldNull` elements, bind as the SQL list literal+ does: they take the element type of a parameter with a known type. For a+ parameter with no known type, `[]` and `[NULL, NULL]` have the type+ `"NULL"[]`, and `MAP {}` has the type `MAP("NULL", "NULL")`.++ See [FieldValue parameters](#fieldvalue-parameters).+12. `typeof` gives `"NULL"`, with the quotes, for a `Null` parameter and for a+ `Null` declaration. A `Null` argument takes only a NULL value. A call with+ another value is a Binder Error. A record cannot have a `Null` field.+13. A `Variant` reads a value that a VARIANT can hold: a VARIANT column, or a+ column of another type with the same shape. That is a value of each scalar+ type other than ENUM, for example an INTEGER or a GEOMETRY with no CRS, a+ LIST of such values, and a STRUCT in the form of a VARIANT object: each+ field of the type VARIANT, the keys in ascending order. At each depth, it+ refuses an ENUM, an ARRAY, a MAP, a TUPLE, a UNION, another STRUCT, and a+ GEOMETRY with a CRS, with the error of an incompatible column, because the+ cast of DuckDB to VARIANT changes these values. Write `c::VARIANT` in SQL to+ read such a column. It also refuses a type that is or contains JSON, at+ each depth, because DuckDB casts JSON to VARIANT by parsing the text: the+ error says `DuckDB casts JSON to VARIANT by parsing the text; write+ c::VARIANT`, and `c::VARIANT` gives the parsed value.+14. `DecimalValue` and `RawGeometry` have no `DuckDBColumnType` and no+ `ToDuckValue` instance, so a `SizedArray`, a list, or a typed row of either+ does not compile: GHC reports+ `No instance for 'DuckDBColumnType DecimalValue'` (or `RawGeometry`), or the+ same for `ToDuckValue` in a typed row. One column type cannot give the width+ and scale of each `DecimalValue`. A declared GEOMETRY type has no CRS, so it+ would remove the CRS of each `RawGeometry`. A composite parameter holds a+ `RawGeometry` as a `FieldGeometry`.++A value of a VARIANT column reads as its stored value, so the row of the stored+type applies. A value in a LIST, ARRAY, MAP, STRUCT, or UNION reads with the+same rules, and so does a field of a `ViaDuckDB` record. A SQL type that is not+in "Reads" raises the error of an incompatible column. See [Errors](#errors).++A function registered with `createFunction` or `createAggregate` declares the "Declared" type, or+the "Binds as" type if "Declared" is empty, for each argument and its result.+An argument type needs `FromField` and a declaration. A result+type needs `ToDuckValue` and a declaration. Some result types write each value+directly into the data of the result vector: `Bool` (BOOLEAN), an integer of 8+to 64 bits (TINYINT to BIGINT, UTINYINT to UBIGINT), `Float` (FLOAT), `Double`+(DOUBLE), `Text` and `String` (VARCHAR), `Maybe` of one of them, and a newtype+that derives `FunctionResult` from one of them+(`deriving newtype (FunctionResult)`). Each other result type, for example a+record or a list, requires a native value for each row. DuckDB uses its default+cast to convert that value to the declared type. For example, it rounds a+DECIMAL to the declared scale and raises an error if the value does not fit.++An aggregate result encoder receives an instance factory. This factory resolves+built-in types. DuckDB resolves the result declaration on the registration+host, so a custom encoder can return the underlying value of a catalog type.+See [User-Defined Functions](#user-defined-functions).++## Errors++A program can catch these exceptions:++- `SQLError`: an error of DuckDB, an input check of the library, a failed+ internal check of the library, or a result value that does not convert.+- `FormatError`: parameters that do not agree with the placeholders of the+ statement.+- The exception of a cancellation, for example from `timeout`.++A failed check of the library state, or of a vector, type, or value from+DuckDB, raises an `SQLError` of the type `ErrorRuntimeInternal`. Such an error+indicates a defect in the library or DuckDB, not an input error. Report it.+Inside a callback, the statement that runs the callback fails with+the type `ErrorInputInvalid` and the text+`Invalid Input Error: duckdb-simple: ...`, because a callback does not pass the+type `ErrorRuntimeInternal`; a cast gives `ErrorTypeMismatch` (see the errors of+a callback below). Using a closed environment, instance, connection, or+statement raises an `SQLError` with no type. A DuckDB version mismatch and+reading callback state with the wrong type also raise untyped `SQLError`s.++`SQLError` has four fields:++- `sqlErrorMessage` is the text of the error. For an error of DuckDB, it is+ the text of DuckDB, for example `Binder Error: ...`.+- `sqlErrorType` is the error category, as `Just` an `ErrorType`. For a DuckDB+ error, it is DuckDB's error code. The library assigns a type to an input+ check, a lock wait, or a failed internal check.+ `ErrorType` has one constructor for each error code of the V2 C API, for+ example `ErrorQueryBinder` and `ErrorTypeConversion`.+- `sqlErrorQuery` is the SQL text of the statement for an error during its+ preparation or execution, and for a result value that does not convert. It+ is `Nothing` for an error outside a statement, for example when a connection+ opens. An error of a parameter value can also have `Nothing`.+- `sqlErrorBody` is DuckDB's error text without the type prefix. It is+ `Nothing` for a library error.++An input check of the library raises an `SQLError`. A check for a value outside+the range of its SQL type gives the type `ErrorTypeConversion`, the type of the+Conversion Error that DuckDB gives the same value in SQL: an `Integer` that its+parameter or its declared type cannot hold, a date or a timestamp outside the+range of DATE or TIMESTAMP of DuckDB (note 7), a nanosecond timestamp outside+its storage, a time of day that is not a TIME value, a TIMETZ offset outside+-57599 to 57599 seconds, an `Integer` or `Natural` field of a record outside HUGEINT or+UHUGEINT, a `DecimalValue` that its width cannot hold, and an ENUM label that is+not in the dictionary of a declared ENUM type. Each other input check gives the+type `ErrorInputInvalid`. This includes a value of the wrong shape or type, also+where DuckDB gives another type for the same input, for example a UNION tag that+is not a member, a leaf built from another type, a STRUCT field that does not+match its type, or a blank GEOMETRY CRS label. Examples are a VARIANT object with two equal+keys, a registration with an argument type that the library cannot read, and a+DECIMAL width above 38.++A statement in a callback that waits for a connection lock longer than+`connectionCallbackLockWait` raises an `SQLError` of the type+`ErrorResourceInUse` ([User-Defined Functions](#user-defined-functions) gives+the rule). A statement on a connection that streams a result, for example in the+step of a fold of `Database.DuckDB.Simple.Streaming`, also gives+`ErrorResourceInUse`, and so does a close of an environment while an instance+that `createInstance` did not create is alive.++A result value that its Haskell type cannot read raises an `SQLError` with no+type. The error includes the query. Its message describes the first field that+fails:++- The error of an incompatible column, for a SQL type that the instance does+ not read: `duckdb-simple: column NAME has type TYPE but expected HASKELLTYPE`.+ The message can end with `: ` and a detail, for example `: expected NULL`+ for `Null`. TYPE is the type of the column in the text of DuckDB, as+ `typeof` gives it, for example DECIMAL(18,2), TIMESTAMP_NS, TIMESTAMP WITH+ TIME ZONE, ENUM('a', 'b'), INTEGER[], or JSON. `logicalTypeSQL` gives that+ text. For a VARIANT column, TYPE is the type of the stored value in the same+ text, for example DOUBLE or TIMESTAMP WITH TIME ZONE. A value that does not+ hold its full type, for example a LIST or a STRUCT in a VARIANT, gives the+ name of its kind: LIST or STRUCT. The library reads the diagnostic type+ only when the conversion's failure is selected.+ HASKELLTYPE is the type of the instance that reads, for example `Text` for+ `String`.+- `duckdb-simple: column NAME is NULL but expected HASKELLTYPE` for NULL in a+ type without `Maybe`.+- `duckdb-simple: field FIELD: ` and the error of the value for a field of a+ `ViaDuckDB` record, for example+ `duckdb-simple: field pairValue: a value has type DOUBLE but expected Int64`.+- A conversion error of the library, for a value that the instance cannot+ convert: the message of the instance, for example `integer value out of+ bounds`.++The library creates a `ResultError` for each failure and puts its text in the+`SQLError`. It does not raise the `ResultError`. A custom `FromField` instance+reports a failure with `returnError` and a constructor of `ResultError`.+`returnError` defers its metadata read until the final conversion fails. If an+`Alternative` branch succeeds, the library discards the other branch's error+without reading its diagnostic metadata. A completed `ResultError` contains+pure data. Its rendering and comparison perform no native IO.+`query` and `query_` raise the error and give no rows. A fold raises it at the+row that fails. A row type with more or fewer fields than the result has+columns raises an `SQLError` with the text `column index N out of bounds`.++An exception in a registered Haskell function, for example one registered with+`createFunction`, fails the statement that runs the function. That statement+raises an `SQLError`. An `SQLError` with a type keeps its type, and its text+gets one prefix. For example, a function that runs a failing query on another+connection fails the outer statement with `ErrorTypeConversion` and the text+`Conversion Error: ...` of the inner query, and a check of the library in a+result, for example a time of day after 24:00:00, keeps its type+`ErrorTypeConversion`. A failed cast is the exception: it gives+`ErrorTypeMismatch` (see below). Each other exception, an `SQLError` with no+type, and an `SQLError` of one of the 16 types below give `ErrorInputInvalid`+and the text `Invalid Input Error: ` followed by the message of the exception:+the message of an `SQLError`, the text of a `userError`, or the+`displayException` text of another exception. An argument that does not convert+gives an `SQLError` of the type `ErrorInputInvalid` with a text that names the+argument, for example+`duckdb-simple: argument arg0 is NULL, but Int64 cannot hold NULL` for a scalar+function. For an aggregate or a table function, the text names the function+after the prefix, for example `duckdb-simple: tf_int: argument arg0 ...`. DuckDB+ignores an exception of a log sink.++These 16 types do not pass through a callback:++- `ErrorQueryBinder`, `ErrorDatabaseCatalog`, `ErrorResourceConnection`,+ `ErrorQueryParameterNotAllowed`, `ErrorQueryParser`,+ `ErrorConfigurationPermission`, and `ErrorResourceInUse`. DuckDB expects these+ types before the execution of a statement. Under the DuckDB setting+ `SYNTACTIC_ERRORS_DO_NOT_INVALIDATE`, it does not invalidate an explicit+ transaction after them, and a callback raises during the execution.+- `ErrorRuntimeInternal` and `ErrorRuntimeFatal`. DuckDB can invalidate the+ database after them.+- `ErrorApi`, `ErrorIOFileNotFound`, `ErrorIOReadFailure`, `ErrorIOEof`,+ `ErrorInputParameterInvalid`, `ErrorInputMetadataDeferred`, and+ `ErrorIONetwork`. DuckDB 2.0.0 cannot raise them again with their own type.++When an error of a callback fails a statement in an explicit transaction, DuckDB+aborts that transaction, also under the setting+`SYNTACTIC_ERRORS_DO_NOT_INVALIDATE`: [User-Defined+Functions](#user-defined-functions) gives what this means for a `COMMIT`.++An exception in the callback of `registerCast` in+`Database.DuckDB.Simple.V2.Registration` fails the cast with an `SQLError` of+the type `ErrorTypeMismatch`, also an `SQLError` with another type. Its message+is `Mismatch Type Error: ` and the text of the exception. When DuckDB casts a+constant through the cast while it binds the query, for example the INTEGER+literal of `range(10)` with a cast from INTEGER to BIGINT, the error has the+type `ErrorInputInvalid` and the text `Failed to cast value: ` before the text+of the exception, and the database stays usable. The type `ErrorTypeMismatch`+is the same for each pair of types, because of a property of DuckDB 2.0.0.+DuckDB plans casts between some pairs of types as casts that cannot fail,+including registered casts. Examples are casts to VARCHAR, including+from an ENUM, a LIST, or a STRUCT, and a widening cast such as INTEGER to+BIGINT or FLOAT to DOUBLE.+For such a cast, DuckDB changes an error of the type `ErrorInputInvalid`,+`ErrorInputOutOfRange`, or `ErrorTypeConversion` to an INTERNAL error. It does+the same for an error type that it does not map, for example `ErrorApi`. An+INTERNAL error during the execution of a query invalidates the database. Then+each statement on each connection of the database fails until a new instance+opens the database. An error of the type `ErrorTypeMismatch` keeps the database+usable, and it aborts an explicit transaction. In `TRY_CAST`, DuckDB discards+the error and keeps the rows that the callback wrote, so the callback must write+each row, for example NULL for a failed row. `TRY(CAST(...))` raises the error,+because `TRY` catches only the types `ErrorInputInvalid`,+`ErrorInputOutOfRange`, and `ErrorTypeConversion`. DuckDB still plans such a+cast as one that cannot fail, so it can run the cast for a row that another+condition of the query excludes, for example in+`WHERE c < 2500 AND CAST(c AS VARCHAR) <> 'x'`. In DuckDB 2.0.0, only a guard in+a CASE expression, for example+`CASE WHEN d <> 0 THEN CAST(c AS VARCHAR) <> 'x' ELSE false END`, kept the cast+from the excluded rows. A cast that a program registers with the native API of+`duckdb-ffi` also gets the INTERNAL error for such a pair and such an error+type. A V2 cast can report the code `DUCKDB_V2_ERROR_TYPE_MISMATCH` in their+place.++An asynchronous exception, for example from `timeout` or Ctrl-C, cancels a+running query. The library interrupts DuckDB, waits until the native call+returns, and raises the same exception again. Catch that exception, not+`SQLError`. See [Resource Management](#resource-management) for `-threaded`.++`FormatError` has the fields `formatErrorMessage`, `formatErrorQuery`, and+`formatErrorParams` (the parameter values as text). The library raises it+before it executes the statement, in the cases that+[Named Parameters](#named-parameters) lists. A `FormatError` is not an+`SQLError`, so catch it separately.++This function gives `Nothing` when the table `items` has no column `value`:++```haskell+import Control.Exception (throwIO, try)+import Database.DuckDB.Simple++countValues :: Connection -> IO (Maybe Int)+countValues conn = do+ result <- try (query_ conn "SELECT count(value) FROM items")+ case result of+ Right [Only n] -> pure (Just n)+ Right _ -> pure Nothing+ Left err+ | sqlErrorType err == Just ErrorQueryBinder -> pure Nothing+ | otherwise -> throwIO err+```++## Generic encoding with ViaDuckDB++The `Database.DuckDB.Simple.Generic` module provides automatic encoding and+decoding of Haskell algebraic data types as DuckDB STRUCTs and UNIONs via+GHC generics.++### Product types as STRUCTs++The generic encoder converts product types (records) to DuckDB STRUCT values:++```haskell+{-# LANGUAGE DeriveGeneric #-}+{-# LANGUAGE DerivingVia #-}++import Data.Int (Int64)+import Data.Text (Text)+import Database.DuckDB.Simple+import Database.DuckDB.Simple.Generic (ViaDuckDB (..))+import GHC.Generics (Generic)++data User = User+ { userId :: Int64+ , userName :: Text+ }+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB User)++-- Round-trip through the database+storeAndFetchUser :: Connection -> User -> IO [User]+storeAndFetchUser conn user = do+ _ <- execute_ conn "CREATE TABLE users (data STRUCT(userId BIGINT, userName TEXT))"+ _ <- execute conn "INSERT INTO users VALUES (?)" (Only user)+ fmap fromOnly <$> query_ conn "SELECT data FROM users"+```++### Sum types as UNIONs++The generic encoder converts sum types to DuckDB UNION values. Each constructor+becomes a union member:++```haskell+data Shape+ = Circle Double+ | Rectangle Double Double+ | Point+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB Shape)++-- Store and retrieve shape data+storeShape :: Connection -> Shape -> IO [Shape]+storeShape conn shape = do+ _ <- execute_ conn+ "CREATE TABLE shapes (s UNION(Circle STRUCT(field1 DOUBLE), \+ \Rectangle STRUCT(field1 DOUBLE, field2 DOUBLE), Point STRUCT))"+ _ <- execute conn "INSERT INTO shapes VALUES (?)" (Only shape)+ fmap fromOnly <$> query_ conn "SELECT s FROM shapes"+```++A constructor without fields (like `Point`) has an empty STRUCT as+its payload. DuckDB shows this payload as `{}`, and it is not NULL.+Non-record constructors use positional field names (`field1`, `field2`, etc.).++### Arrays and lists++The library supports fixed-length DuckDB arrays and variable-length lists:++```haskell+{-# LANGUAGE DataKinds #-}++import Control.Monad (forM_)+import Data.Array (listArray)++storeArray :: Connection -> IO [SizedArray 3 Int]+storeArray conn = do+ _ <- execute_ conn "CREATE TABLE arrays (vals INTEGER[3])"+ forM_ (sizedArray (listArray (0, 2) [1, 2, 3]) :: Maybe (SizedArray 3 Int)) $ \arr ->+ execute conn "INSERT INTO arrays VALUES (?)" (Only arr)+ fmap fromOnly <$> query_ conn "SELECT vals FROM arrays"++storeList :: Connection -> IO [[Int]]+storeList conn = do+ _ <- execute_ conn "CREATE TABLE lists (vals INTEGER[])"+ _ <- execute conn "INSERT INTO lists VALUES (?)" (Only ([1, 2, 3] :: [Int]))+ _ <- execute conn "INSERT INTO lists VALUES (?)" (Only ([] :: [Int]))+ fmap fromOnly <$> query_ conn "SELECT vals FROM lists"+```++A Haskell list binds as a LIST parameter and reads a LIST result. Put a list+parameter in `Only`, because a list as a row is a row of parameters. A+`String` stays one VARCHAR parameter. A `SizedArray` binds as an ARRAY of its+size (note 8 of [Types](#types)). DuckDB casts a LIST or an ARRAY to the type+of a LIST column. It casts a LIST to an ARRAY column only when the lengths are+equal.++A list or an array parameter with scalar elements takes its element type from+the column type name (`duckdbColumnTypeFor`) of the element. Some elements take+the element type of the target instead: `Integer`, `Natural`, and `BigNum`+elements for an integer or DECIMAL element type, and otherwise the type of their+values (see above), and `TimeOfDay`, `LocalTime`, and `UTCTime` elements for a+nanosecond element type (see [Infinite dates and+timestamps](#infinite-dates-and-timestamps)). An empty list binds as an empty+LIST of its element type, except an empty list of `Integer` or `Natural` values+with no integer or DECIMAL target (see above). An empty array raises an error:+note 8 of [Types](#types).++A list or an array of STRUCT values, UNION values, generic records, lists, or+arrays gets the type that DuckDB gives a SQL list literal of the same elements.+When the elements that are not NULL have one type, the list has that type.+When their types are different, DuckDB uses their common type: `{'a': 1}` and+`{'b': 2}` give `STRUCT(a INTEGER, b INTEGER)[]`, with NULL for a field that an+element does not have. A `BIGINT` field and a `DOUBLE` field give a `DOUBLE`+field, so `9007199254740993` becomes `9007199254740992.0`, as in SQL. Elements+with no common type, such as `{'a': 1}` and `{'a': 'x'}`, raise the `SQLError`+of DuckDB. The column type of an `INSERT` does not change the type of the+list: DuckDB casts the list to the column type after it gives the list its+type. A list or an array of such elements that are all NULL, and an empty+list, take the element type of the parameter when the parameter has a known+type, as for `INSERT INTO t VALUES ([])` and `([NULL, NULL])` in SQL. For a+parameter with no known type, an empty list or a list of NULL elements gets+the type of the SQL literal, `"NULL"[]`, and an empty `Map` of such values gets+the type NULL for its values, as `MAP {}` does. A generic record through+`ViaDuckDB` has a declared type, so for a parameter with no known type an+empty list or a list of NULL records gets the STRUCT or UNION type of the+record.++A `Map` from `Data.Map` binds as a MAP parameter and reads a MAP result. The+pairs go in ascending key order. The keys and the values follow the element+rules of a list, and an empty `Map` binds as an empty MAP of its key and value+types. DuckDB compares two MAP values pair by pair in their order, so+`WHERE m = ?` finds a stored MAP only when it has the same pairs in ascending+key order. DuckDB rejects a NULL key and two equal keys, for example two NaN+keys of a `Map Double`. When reading a MAP, a decoder rejects keys that become+equal after conversion. For example, the integer keys 1 and 2 both decode as+`True` in a `Map Bool v`.++### Infinite dates and timestamps++Use `Database.DuckDB.Simple.Time` when a column can contain temporal infinity.+Its `Date`, `LocalTimestamp`, and `UTCTimestamp` types wrap `Day`, `LocalTime`,+and `UTCTime` in `Unbounded`: `NegInfinity`, `Finite value`, or `PosInfinity`.+These types support parameters, results, and fields in generic composites.+`UTCTimestamp` binds as TIMESTAMPTZ.++```haskell+import Database.DuckDB.Simple+import Database.DuckDB.Simple.Time++infiniteDates :: Connection -> IO [Only Date]+infiniteDates conn = query conn "SELECT ?::DATE" (Only (PosInfinity :: Date))+```++The ordinary `Day`, `LocalTime`, and `UTCTime` instances reject infinity: note 6+of [Types](#types). Floating-point NaN and infinities remain valid `Float` and+`Double` values.++A `TimeOfDay`, a `LocalTime`, or a `UTCTime` parameter, and an `Unbounded`+form, keeps its digits below a microsecond for a nanosecond parameter:+`TIME_NS`, `TIMESTAMP_NS`, or `TIMESTAMPTZ_NS`. Use an explicit cast, such as+`SELECT ?::TIMESTAMP_NS`, or a column with that type. A `VARIANT` parameter+also gets the nanosecond value, when the value fits the range of the nanosecond+type, because a VARIANT holds `TIME_NS`, `TIMESTAMP_NS`, and `TIMESTAMPTZ_NS`+values. This applies only to a scalar value: a list, a `SizedArray`, or a `Map` of+time values for a `VARIANT` parameter binds microsecond elements, and so does+an element of a `VARIANT[]` list or a value of a MAP with `VARIANT` values,+because one list has one element type for all its elements. Each other+parameter, and a parameter with no known type, gets the microsecond value: the+digits below a microsecond are discarded, as for the SQL literal. Only a value+with such digits reads the parameter type. An explicit cast such as+`?::TIMESTAMP_NS` makes a nanosecond parameter. For a `TIMESTAMP` column, DuckDB+then rounds the value to the microsecond. A plain `?` truncates the digits.+Values inside collections, STRUCTs, MAPs, and UNIONs follow their target types.+The library refreshes prepared metadata when it resets or+rebinds a statement.+One `executeMany` call reads the parameter types once. A column type that+changes during the call does not change how its later rows bind.+Use separate parameters when one expression requires microseconds and another+requires nanoseconds.++A date or a timestamp outside the range of DuckDB raises an `SQLError` of the+type `ErrorTypeConversion` when it binds, also as an element of a list, a+`SizedArray`, or a `Map`, inside a STRUCT, a TUPLE, or a UNION, and in a VARIANT.+For a `Day`, that range is the range of DATE, from -5877641-06-25 to 5881580-07-10.+For a `LocalTime` or a `UTCTime` (in UTC), it is the range of TIMESTAMP,+TIMESTAMPTZ, TIMESTAMP_S, and TIMESTAMP_MS, from -290308-12-22 00:00:00 to+294247-01-10 04:00:54.775806. The dates are in the notation of `Data.Time`. A+TIMESTAMP_NS or TIMESTAMPTZ_NS target has a smaller range, about the years+1677 to 2262. A value outside that range also gets the type+`ErrorTypeConversion` in each position. The position of the value determines+whether the library or DuckDB raises the error:++- A parameter: the error of the library for a value outside the range of DuckDB,+ and the Conversion Error of DuckDB for each other such value, for example in+ the year 2300.+- An element of a list, a `SizedArray`, or a `Map`: the error of the library.+ For a value in the range of DuckDB, its text is "TIMESTAMP value out of finite+ range".+- A field of a STRUCT, an element of a TUPLE, or a member of a UNION: the error+ of the library, "TIMESTAMP value out of finite range", also for a value+ outside the range of DuckDB.+- In each position, a `LocalTime` or a `UTCTime` (in UTC) from 2262-04-11+ 23:47:16.854775807 to 23:47:16.854775999 gets that error of the library.++In the notation of `Data.Time`, the year -5877641 is 5877642 BC and the year+-290308 is 290309 BC. The rule holds also for a finite `Unbounded` value. DuckDB+does not accept a value outside its range: its SQL literals reject it, and a+stored value can read as an infinity or make DuckDB raise an INTERNAL error. A+leap second, a `TimeOfDay` with 60 seconds, binds as the next second, as+`make_time` and `make_timestamp` of DuckDB give it. The SQL literal `'23:59:60'`+raises an error. A time after 24:00:00, for example 23:59:60.5, and a field+outside its own range, for example 61 minutes, raise an `SQLError` of the type+`ErrorTypeConversion`, also as the result of a typed function or a row of a+table function.++### Manual STRUCT, TUPLE, and UNION handling++SQL tuples, such as `row(42, 'text')`, decode as `TupleValue`. Read their+positional fields in `tupleValueFields` and types in `tupleValueTypes`.+`FieldTuple` and `LogicalTypeTuple` preserve empty tuples independently of+empty STRUCT values. STRUCT fields retain their names in `StructValue`.+`LogicalTypeCollatedVarchar` preserves a VARCHAR field's collation in composite+parameters. Its comparisons use that collation after a round trip.++Temporal fields retain their SQL units when composite values are rebound.+The `FieldDate`, `FieldTimestamp`, and `FieldTimestampTZ` constructors hold+`Unbounded` values. Wrap finite payloads in `Finite` when constructing them.+TIMESTAMP_S and TIMESTAMP_MS fields use the finite range described in+[Infinite dates and timestamps](#infinite-dates-and-timestamps). An explicit+parameter cast does not extend that range.++For more control, you can work directly with `StructValue` and `UnionValue`+from `Database.DuckDB.Simple.LogicalRep`:++```haskell+import Database.DuckDB.Simple+import Database.DuckDB.Simple.LogicalRep (StructValue (..), UnionValue (..))+import Database.DuckDB.Simple.FromField (FieldValue (..))++manualStruct :: Connection -> IO [(StructValue FieldValue, UnionValue FieldValue)]+manualStruct conn = do+ _ <- execute_ conn+ "CREATE TABLE composite (s STRUCT(a INT, b INT), \+ \u UNION(x INT, y VARCHAR))"+ [(s, u)] <- query_ conn+ "SELECT {'a': 1, 'b': 2}, \+ \CAST(union_value(x := 42) AS UNION(x INT, y VARCHAR))"+ :: IO [(StructValue FieldValue, UnionValue FieldValue)]+ _ <- execute conn "INSERT INTO composite VALUES (?, ?)" (s, u)+ query_ conn "SELECT s, u FROM composite"+```++Low-level `ToDuckValue.toDuckValue` takes a `ValueFactory`, which is the V2+factory handle, and `logicalTypeFromRep` takes the same handle.+`Database.DuckDB.Simple.V2.LogicalRep` has `logicalTypeFromRep`.+Get the factory of a connection with `withConnectionHandle` and+`withConnectionFactory` from `Database.DuckDB.Simple.Internal`. Keep owned+values and logical types within the lifetime of the factory's owner.++### FieldValue parameters++A `FieldValue` binds as a parameter, so a row that a query reads as+`[FieldValue]` binds back as a row. A scalar constructor binds as the DuckDB+type of the constructor, with the parameter instance of its payload.+`FieldStruct`, `FieldTuple`, and `FieldUnion` bind with their own types, and+`FieldGeometry` binds with its CRS. `FieldList`, `FieldArray`, and `FieldMap`+take the element type that their elements give, with the rule of a SQL list+literal (see [Arrays and Lists](#arrays-and-lists)). They read the parameter+type, so that a list read from a `TIME_NS[]` column binds back with its+nanoseconds.++TIME aliases keep their alias and exact microsecond value when they bind+inside a composite. The library uses public vector storage because the+default TIME-to-alias cast fails in DuckDB 2.0.0. Other aliases require native+cast support.++Some values bind with another type than the column that they came from. A+`FieldTimestamp` binds as a `TIMESTAMP`, or as a `TIMESTAMP_NS` when it has+digits below a microsecond and the parameter has the type `TIMESTAMP_NS` or+`VARIANT` (see [Infinite dates and timestamps](#infinite-dates-and-timestamps)).+The same holds for `FieldTime` and `FieldTimestampTZ`. A `FieldText` binds as a+`VARCHAR` with no collation, also when it came from a `JSON` value. A value+from a `VARIANT` column binds as its payload. A `FieldEnum` binds its label as+a `VARCHAR`. `FieldNull` binds as an untyped NULL. DuckDB casts each of these+values to the type of a column with no loss, with one exception: a time value+with digits below a microsecond in a `FieldList` or a `FieldMap` for a+`VARIANT` parameter gets the microsecond value, so a `VARIANT` array of+`TIMESTAMP_NS` values loses these digits when it binds back. A `VARIANT` object+reads as a `FieldStruct` with `VARIANT` fields and keeps them.++An empty or all-NULL `FieldList`, `FieldArray`, or `FieldMap` follows note 11 of+[Types](#types). Inside a `FieldStruct`, a `FieldTuple`, or a `FieldUnion`, a+`FieldEnum` binds its label to the declared ENUM type, and a label that is not+in the dictionary raises an `SQLError` of the type `ErrorTypeConversion`, as the+cast of DuckDB gives for that label. `FieldValue` has no `DuckDBColumnType` and+no `ToDuckValue` instance, because no one DuckDB type holds each of its values.++### Resource management++- `withConnection` and `withStatement` open and close resources, including when+ an exception occurs. Use them whenever possible to avoid leaking C handles.+- The query functions release their results, prepared statements, and values+ after use, and return a Haskell list of all rows. Folds and cursors decode+ rows incrementally from owned native chunks. The materialized functions of+ `Database.DuckDB.Simple` drain and close the native result before row+ callbacks; the streaming functions of `Database.DuckDB.Simple.Streaming` do+ not.+- `execute`/`query` variants reset statement bindings each run so prepared+ statements can be reused safely.++For concurrent workers, use a separate connection per worker. Threads can also+share one connection. DuckDB permits one open result for each connection, so+each connection has a lock. `query`, `execute`, and the materialized folds hold+this lock from the execution of a statement until its native result closes.+Statement preparation holds it only while DuckDB prepares the statement.+Metadata binding holds it only while DuckDB binds the statement. A call from+another thread waits. The lock does not cover these cases. Serialize them in+the application:++- A transaction. Statements of other threads run inside it.+- One `Statement` that two threads use.+- An explicit stream from `Database.DuckDB.Simple.Streaming`. Each fetch of+ the stream holds the lock, but its result stays open between calls, and+ another query on that connection fails until the stream is complete or+ closed.+- `close`. Wait for all operations on the connection before you close it.++A callback must not run a statement on the connection that runs the callback,+or close that connection: [User-Defined Functions](#user-defined-functions)+gives the rule, its time limit, and its limits.++Link your executable with `ghc-options: -threaded`. A program that registers a+callback needs it, because DuckDB can call a callback from several threads at+the same time: [User-Defined Functions](#user-defined-functions) gives the rule.+It also allows prompt cancellation of native queries, including Ctrl-C. On+cancellation, the library interrupts DuckDB and waits for the native call to+return before it releases resources and propagates the exception. Cancellation+is cooperative: native code and Haskell callbacks must return before cleanup can+finish.++Native work that can take a long time runs in a helper thread so that an+asynchronous exception can interrupt it. The materialized functions run the+native calls of one result in one helper thread. A bound thread, such as the+main thread, changes operating system threads to wait for a helper thread.+Thus a small query takes longer from a bound thread than from an unbound+thread. Run code with many small queries in a thread made with `forkIO`, or+wrap the main action in `Control.Concurrent.runInUnboundThread`.++Call `closeStatement` or `clearStatementBindings` on an abandoned statement to+release its result. DuckDB can retain native result buffers until that result+is destroyed.+Use `withStatement` for manual iteration. `fold` releases the result after+success or an exception. The accumulator determines Haskell memory use.++### Metadata helpers++- `columnCount` and `columnName` expose prepared-statement metadata so you can+ inspect result shapes before executing a query.+- `execute` returns the number of affected rows. Use SQL `RETURNING` clauses+ when you need generated identifiers.++### Cursors and folds++`fold`, `fold_`, and `foldNamed` decode one row at a time from DuckDB's result+chunks. This API materializes the native result before the first row is+returned. These functions avoid a complete Haskell row list, but native memory+use still depends on the result size. The materialized functions close the+native result before row delivery. A row callback can run another query on the+same connection.++```haskell+import Database.DuckDB.Simple++sumValues :: Connection -> IO Int+sumValues conn =+ fold_ conn "SELECT n FROM stream_fold ORDER BY n" 0 $ \acc (Only n) ->+ pure (acc + n)+```++For manual cursor-style iteration, use `nextRow`/`nextRowWith` on an open+`Statement` to pull rows one at a time and decide when to stop.++Cursors support the same column types as the query functions, including STRUCT+and UNION values with nested collections and NULLs. GEOMETRY and VARIANT work+with the query functions, cursors, and folds.++#### Optional native streaming++`Database.DuckDB.Simple.Streaming` provides `fold`, `fold_`,+`foldNamed`, `nextRow`, and `nextRowWith` with native streaming enabled.+Import it qualified:++```haskell+import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Streaming as Streaming++streamSum :: Connection -> IO Int+streamSum conn =+ Streaming.fold_ conn "SELECT i FROM range(1000000) t(i)" 0 $ \acc (Only n) ->+ pure (acc + n)+```++This module uses V2 execution and incremental chunk fetching. DuckDB can still+materialize some queries. Streaming does not bound the memory used by query+operators.++The first cursor fetch selects the execution mode. The mode stays until `bind`,+`bindNamed`, `clearStatementBindings`, or `executeStatement` releases the+result. Switching between materialized and streaming `nextRow` calls retains+that mode. Keep the connection dedicated to the explicit stream. Until the+stream is complete, DuckDB rejects other statements on that connection with the+`ErrorResourceInUse` error of DuckDB. If you close an incomplete stream, DuckDB+aborts its statement. In autocommit mode, DuckDB rolls back the writes of that+statement. Cancellation interrupts native chunk fetching as well as execution.++This module also provides `foldArrow` and `foldArrow_` for streaming Arrow+batches. They use the same supported Arrow conversion and scoped ownership+as `Database.DuckDB.Simple.Arrow`.++### Arrow batches++`Database.DuckDB.Simple.Arrow` provides `foldArrow` and `foldArrow_` for clients+that consume the Arrow C Data Interface. Each callback receives a separate+schema and array batch. It can read them or pass them to an Arrow consumer+that releases or moves them. The fold releases any remaining contents on+success, failure, or cancellation. The consumer owns any contents it moves.++Each array has at most `connectionArrowBatchSize` rows. Set this option in+`ConnectionOptions`. The default is `defaultArrowBatchSize`, 2048 rows. DuckDB+can give shorter arrays. Opening a connection rejects a value that is not+positive. For DuckDB's batch size, set 131072.++The original pointers are valid only during the callback. To retain contents,+a consumer must move the root structs into its own storage and set the source+release fields to NULL. Moved contents remain valid after the query and+connection close. Empty results do not produce a callback.++For example, the `dataframe-arrow-bridge` package can copy each batch into a+Haskell `DataFrame` and release the Arrow objects:++```haskell+import qualified DataFrame.IO.Arrow as DataFrame+import qualified Database.DuckDB.Simple.Arrow as Arrow+import Foreign.Ptr (castPtr)++frames <- Arrow.foldArrow_ conn "SELECT id::BIGINT, name::VARCHAR FROM people" [] $ \acc schema array -> do+ frame <- DataFrame.arrowToDataframe (castPtr schema) (castPtr array)+ pure (frame : acc)+-- Reverse frames to recover the query's batch order.+```++The bridge currently imports signed 32-bit and 64-bit integers, Float, Double,+and text columns. It is a test dependency of this repository; applications+that use it must declare their own dependency on `dataframe-arrow-bridge`.++Arrow export uses V2 Arrow results. The materialized `foldArrow` retains owned+batches and closes the native result before callbacks start. Its memory use+depends on the result size. The streaming `foldArrow` delivers each batch while+the native result remains open.++### VARIANT++VARIANT results decode to the `FieldValue` of the stored value. Existing+`FromField` instances read scalar payloads. Objects decode to `FieldStruct`+values with VARIANT fields in DuckDB's native key order (lexicographic).+Arrays decode to `FieldList` and preserve element order.+The decoder uses SQL type metadata and native casts. It preserves scalar types,+precision, and nested values without reading DuckDB's private payload format.++```haskell+query_ conn "SELECT 42::BIGINT::VARIANT" :: IO [Only Int64]+```++SQL NULL VARIANT values decode to `FieldNull`. `Variant` from+`Database.DuckDB.Simple.Variant` wraps a `FieldValue`. Its `ToField` instance+binds the payload as VARIANT. `variantObject` builds an object from named fields.+It can read the bound value back directly:++```haskell+query conn "SELECT ?"+ (Only (Variant (FieldList [FieldInt64 1, FieldInt64 2])))+ :: IO [Only Variant]+```++A scalar payload keeps its native type, such as `TINYINT` or `DECIMAL(4,2)`.+Time and timestamp payloads bind as microsecond types, or as nanosecond types+when they have sub-microsecond digits and fit the nanosecond range. So a+`TIMESTAMP_S` result binds back as a `TIMESTAMP` when its value fits. Each+element of a list payload and each field of an object payload follow the same+rule, so one list payload can hold both types. A Haskell list, a `SizedArray`,+or a `Map` of time values for a `VARIANT` parameter is not a payload: it binds+microsecond elements (see [Infinite dates and+timestamps](#infinite-dates-and-timestamps)). A value outside the range of+DuckDB raises an `SQLError` of the type `ErrorTypeConversion`. Lists, arrays,+and STRUCT fields bind as VARIANT values. A VARIANT does not keep a UNION, a+MAP, an ENUM, or the CRS of a GEOMETRY, so these payloads raise an error at any+depth: a VARIANT keeps only the member value of a UNION, stores a MAP as a list+of key and value objects, and stores an ENUM as its text. Object parameters+reject duplicate keys and keys that contain NUL. An object with only an empty+key also raises an error because DuckDB treats its STRUCT type as a tuple.+DuckDB 2.0 permits an empty key with other named keys. You can also bind a plain+value and cast it in SQL, as in `?::VARIANT`.++A `Variant` result reads only a value that a VARIANT can hold: note 13 of+[Types](#types) gives the shapes, and the refusal of a JSON type.++For query results the library uses a separate helper connection for VARIANT decoder metadata.+It caches one prepared VARIANT metadata statement and the types for VARIANT+casts. It batches VARIANT metadata queries. It constructs VARIANT types+directly through the native factory. This preserves an explicit stream on the+caller's connection. The helper connection has the settings of the database,+not the caller's session settings. Metadata preparation needs an expression+depth of 11. A GLOBAL `max_expression_depth` below 11 fails preparation with a+Binder Error, including preparation for a connection's COPY TO. Extension COPY TO+and typed callback arguments use an independent instance with default expression+depth. A prepared plan on a query helper can remain+usable after the limit is lowered. The query for unusual object keys still+binds on each execution and can fail at a lower limit.++New DuckDB 2.0 databases use storage format 2.0, which supports VARIANT columns.+The library does not change an existing database's storage compatibility+setting. VARIANT columns require storage format `v1.5.0` or later.++### GEOMETRY++Use `Geometry` from `Data.Geometry` for decoded shapes. The `geometry-simple`+package supplies the type and unboxed coordinate vectors. `duckdb-simple`+supplies its `ToField` and `FromField` instances:++```haskell+import qualified Data.Geometry as G+import qualified Data.Vector.Unboxed as U+import Database.DuckDB.Simple++let line = G.LineString (G.CoordinatesXY (U.fromList [G.XY 0 0, G.XY 1 2, G.XY 3 4]))+rows <- query conn "SELECT ?" (Only line) :: IO [Only G.Geometry]+```++Points and coordinate sequences store their XY, XYZ, XYM, or XYZM layout.+`G.EmptyPoint G.DimXYZ` is an empty XYZ point. Use `Maybe G.Geometry` for SQL NULL.+Use `decodeWKT` from `Data.Geometry.WKT` to parse text without a database.++`G.Geometry` has no CRS metadata. Reading it returns the coordinates without+their CRS label. Binding it creates a `GEOMETRY` with no CRS. To attach a label,+use `ST_SetCRS(?, ?)` with a shape and CRS text. This does not transform coordinates.++Use `RawGeometry` from `Database.DuckDB.Simple.Geometry` to read WKB and CRS+without decoding coordinates. Its `rawGeometryWKB` and `rawGeometryCRS` fields+own their data. They remain usable after the connection closes. Existing+`ByteString` result decoding also returns WKB.++A `RawGeometry` parameter binds as a `GEOMETRY` value with the same WKB bytes+and its own CRS. DuckDB then casts it to the type of the parameter, as it casts+a `GEOMETRY` value in SQL text. A value without a CRS takes the CRS of the+target. A target without a CRS removes the CRS of the value, with no error. A+target with another CRS raises `SQLError`. Use `Maybe RawGeometry` for SQL NULL.+A `SizedArray`, a list, and a typed row cannot hold a `RawGeometry` (note 14 of+[Types](#types)). A composite parameter holds it as `FieldGeometry` (see below):++```haskell+import Database.DuckDB.Simple.Geometry++[Only raw] <- query_ conn "SELECT 'POINT Z (1 2 3)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+rows <- query conn "SELECT ?" (Only raw) :: IO [Only RawGeometry]+```++For explicit SQL import, use `system.main.ST_GeomFromWKB(?)` with the WKB+bytes. To attach a CRS, use+`system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?)` with the bytes and CRS+text. Omit `ST_SetCRS` for no CRS: a NULL CRS argument returns NULL geometry.+CRS text can be a recognized name, a custom label, or WKT2/PROJJSON text.+DuckDB can normalize WKB byte order and reduce a known CRS definition to its+registered identifier. The qualified names select the built-ins even if a+user macro has the same name.++`fromRawGeometry` decodes the shape and drops CRS metadata. `toRawGeometry`+encodes a shape with no CRS. These helpers follow the geometry-simple contracts.+NaN in both WKB point X and Y denotes an empty point. Empty multi-geometries+and collections have no stored layout tag in the decoded representation.+Keep the raw bytes when these details must survive.++Structured parameters use `encodeWKT` and DuckDB's native cast. The WKT writer+combines layouts in multi-geometries and polygon rings. It fills absent Z or M+with NaN.+DuckDB's WKT parser limits nesting to 16 levels and rejects empty polygon rings+and mixed collection layouts. DuckDB 2.0 also rejects WKB with mixed Z/M member+layouts. Explicit WKB import preserves native NaN payload bits that WKT cannot+retain.++`FieldGeometry` contains a raw result. Result decoding retains WKB and CRS,+including inside LIST, ARRAY, MAP, STRUCT, and UNION values. A STRUCT, TUPLE,+or UNION parameter binds each `FieldGeometry` as a `GEOMETRY` with the same+WKB bytes and CRS, also inside LIST, ARRAY, and MAP values. Thus a value that+a query returns binds again without a change. The CRS of the value must be+equal to the CRS of its declared type. A mismatch raises `SQLError` with the+type `ErrorInputInvalid`, for example for a value with the CRS `OGC:CRS84` in+a field of type `GEOMETRY`. A `Variant` parameter binds a `FieldGeometry`+without a CRS as a `GEOMETRY` payload. A VARIANT does not keep the CRS of a+GEOMETRY, so a `FieldGeometry` with a CRS in a `Variant` raises `SQLError`.++`LogicalTypeGeometry` describes CRS metadata. Composite parameters retain it.+Recognized names, such as `OGC:CRS84`, and full WKT2/PROJJSON definitions use+direct V2 type construction. For an unknown custom label, parameter binding+reads a SQL-derived type with `ST_SetCRS` on a helper connection. This happens+on demand, including for nested types, typed NULLs, empty collections, and+inactive UNION members. Binding the same custom label again repeats the lookup.+A blank label or a label with NUL raises `SQLError` at bind time.++Query results retain a custom CRS label. Typed registrations and direct+low-level `logicalTypeFromRep` calls use native type construction. They require+a recognized name or a full definition. A custom label in a declared argument,+result, cast source or target, pragma parameter, or table-function column fails+at registration. DuckDB casts a plain `GEOMETRY` argument to its declared type,+which removes the CRS. Direct type construction raises an error if DuckDB+drops the CRS under `ignore_unknown_crs`.++The CRS in a cast must be a constant. DuckDB rejects a parameter as a type+modifier, so `?::GEOMETRY(?)` is not valid. To use a CRS that is known only at+run time, write it into the query text as a SQL string literal, and double each+single quote. A cast accepts a CRS that DuckDB recognizes, such as `OGC:CRS84`,+or a full WKT2 or PROJJSON definition. Unless an extension recognizes it,+DuckDB rejects other identifiers, such as `EPSG:4326`, and custom names.+`ST_SetCRS` also accepts custom names, but it returns `GEOMETRY` with no CRS+for a NULL input.++A bound composite without a CRS also changes the type of expressions that+combine it with CRS data. `UNION ALL` and `COALESCE` of `GEOMETRY` and+`GEOMETRY('OGC:CRS84')` give `GEOMETRY` with no CRS for all rows. Cast the+parameter before you combine it with other values.++See [geometry-simple](https://github.com/Tritlo/geometry-simple) for the seven+supported families, construction checks, and codec normalization rules.++### Feature coverage++- Connections, prepared statements, positional/named parameter binding.+- High-level execution (`execute*`) and the query functions (`query*`,+ `queryNamed`).+- Cursor and fold helpers (`fold`, `foldNamed`, `fold_`, `nextRow`) that decode+ native result chunks one row at a time.+- Scalar types: signed/unsigned integers, HUGEINT/UHUGEINT,+ decimals (with width/scale), intervals, precise and timezone-aware temporals,+ enums, bit strings, blobs, bignums, and UUIDs.+- Composite types: STRUCTs, UNIONs, LISTs, fixed-length ARRAYs, and MAPs with+ typed parameters and results.+- Generic encoding/decoding: automatic STRUCT/UNION mapping for Haskell ADTs via+ GHC generics and the `ViaDuckDB` deriving-via helper.+- Row decoding via `FromField`/`FromRow`, with generic deriving for product types.+- User-defined scalar functions backed by Haskell functions (including IO and+ nullable arguments).+- Aggregate functions, table functions, replacement scans, COPY formats, and+ pragmas. See [Typed registration](#typed-registration).+- Transaction helpers (`withTransaction`) and metadata accessors (`columnCount`,+ `columnName`).++## User-defined functions++Register a scalar Haskell function on a connection, and call it in SQL+expressions. The arguments use `FromField`, and the result uses+`ToDuckValue`. A function can accept and return `Maybe` values, and it can+run an `IO` action.++An argument type needs `FromField` and `DuckDBColumnType` instances. Its SQL+type is the `duckdbTypeDeclaration` of the type, as for a column of a typed row.+A newtype with its own column type name (`duckdbColumnTypeFor`), for example+`DECIMAL(18,3)`, gives that text. Arguments use the same conversion as query+results, including nested values. Registered casts do not change how the+library reads GEOMETRY arguments. See [Typed registration](#typed-registration).++The SQL type of the result is the `duckdbTypeDeclaration` of the result type:+an `Int16` result is SMALLINT, and a `Float` result is FLOAT. DuckDB computes+with that type. Thus `hs_small(x) * hs_small(y)` of an `Int16` function raises+an Out of Range Error when the product does not fit in SMALLINT, as+`30000::SMALLINT * 30000::SMALLINT` does. Cast the result to a wider type+first. Some result types write directly into the result vector. Each other+result type requires a native value for each row. [Types](#types) lists them.++```haskell+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}++import Data.Int (Int64)+import Database.DuckDB.Simple++registerAndUse :: Connection -> IO [Only Int64]+registerAndUse conn = do+ createFunction conn "hs_times_two" (\(x :: Int64) -> x * 2)+ query_ conn "SELECT hs_times_two(21)"+```++An exception in the function fails the statement with an `SQLError`:+[Errors](#errors) gives its text. `deleteFunction` issues a+`DROP FUNCTION IF EXISTS` statement and reports native errors. DuckDB registers+C API scalar functions as internal entries and rejects their deletion. The+function remains registered after this `SQLError`. Deleting an absent name+succeeds. To replace a function, register the name again. `deleteFunction` has a+GHC `WARNING` pragma, so each use causes a compile-time warning. A build with+`-Werror` must allow that warning, for example with `-Wno-error=deprecations`.++A program that registers a callback, for example of a scalar function, a cast,+an aggregate function, a table function, a COPY format, a replacement scan, a+pragma, or a log sink, must be linked with `ghc-options: -threaded`. DuckDB can+call a callback from several of its threads at the same time. Without+`-threaded`, the GHC runtime does not support calls from several operating+system threads at the same time, so the program can crash.++Without `-threaded`, registering a COPY format with callbacks or a log sink+raises `ErrorInputInvalid` regardless of the settings. DuckDB can open a COPY+file on an asynchronous worker, which can also write to the log sink.+For other callbacks, registration raises `ErrorInputInvalid` when the DuckDB+setting `threads` is above 1. This check only reads the setting at registration.+It cannot prevent another connection from raising the thread count later.++With `-threaded`, each call of a callback needs a capability of the GHC runtime.+With one capability, which is the default, a callback from a second DuckDB+thread waits until the capability is free, and the operating system switches+threads for each such call. A query that makes many callbacks on several DuckDB+threads is then much slower. Give the program as many capabilities as DuckDB has+threads, for example with `ghc-options: -threaded -rtsopts "-with-rtsopts=-N"`,+or set the DuckDB setting `threads` to 1.++A callback must not run a statement on the connection that runs the callback,+and it must not close that connection. Use another connection of the same+instance. This rule applies to each registered callback, for example of a scalar+function, a cast, an aggregate function, a table function, a COPY format, a+replacement scan, a pragma, or a log sink. `query`, `execute`, the materialized+folds, statement preparation, and each fetch of an explicit stream hold the lock+of the connection while DuckDB runs these callbacks. The lock is not reentrant,+so a statement in the callback waits for the lock. With more than one DuckDB+thread, DuckDB can also run a callback of an explicit stream between two+fetches, while no fetch holds the lock. Then a statement in the callback on the+same connection fails at once with the `ErrorResourceInUse` error of DuckDB,+because the stream has an open result.++That wait has a time limit. `connectionCallbackLockWait` in `ConnectionOptions`+sets it in microseconds. The default is `defaultCallbackLockWait`, 10 seconds,+and `Nothing` sets no time limit. When the time limit ends, the statement in the+callback raises an `SQLError` of the type `ErrorResourceInUse`, and its text+gives the rule. If the callback does not catch the error, the outer statement+fails with the type `ErrorInputInvalid`, because a callback does not pass the+type `ErrorResourceInUse` ([Errors](#errors) gives the list; a cast gives+`ErrorTypeMismatch`), and its message contains the text of that error. The+connection can run the next statement. DuckDB ignores the error of a log sink.+The time limit applies to each wait of a callback for the lock of a connection.+It applies also to another connection that a different statement uses. If+callbacks share a connection that runs slow statements, set a larger time limit+or `Nothing` for that connection.++In these cases the time limit does not apply, and a statement in the callback+can wait indefinitely:++- DuckDB runs the callback during a call that does not hold the lock of the+ connection, for example a log sink for a message of `logConnectionMessage`.+ Then the statement goes into DuckDB on the same connection.+- The statement runs in a thread that the callback starts. The library marks+ only the thread of the callback.+- The statement runs inside `uninterruptibleMask`.+- The statement runs in the flush function of a log sink when DuckDB destroys+ the sink.++A statement in a callback on another connection of the same instance runs in its+own transaction, also when the outer statement runs in an explicit transaction.+It does not see rows that the outer transaction has not committed. A write to+a row that the outer transaction changed fails immediately with DuckDB's+`TransactionContext Error: Conflict on update!` error. The outer statement+fails with that text. That failure aborts the outer transaction, as each error+during the execution of a statement does in DuckDB:+each later statement in the transaction fails, and a `COMMIT` returns with no+error but stores no change of the transaction.++A callback that refers to a connection keeps that connection and its instance+reachable while DuckDB keeps the callback, possibly until the database closes.+The garbage collector cannot close them while they remain reachable. The+program must close them.++## Typed registration++These functions register SQL objects with Haskell values. Their signatures have+no native handle. `Database.DuckDB.Simple.Extension` has the same functions for+a loadable extension. An exception in a Haskell function gives an SQL error with+the message of the exception (see [Errors](#errors)). The examples use these+imports:++```haskell+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}++import Data.Int (Int64)+import Data.Text (Text)+import qualified Data.Text as Text+import qualified Data.Text.IO as TextIO+import Database.DuckDB.Simple+import Database.DuckDB.Simple.Copy+import Database.DuckDB.Simple.FromField (Field (..), FieldValue (..))+import Database.DuckDB.Simple.Pragma (PragmaCallback (..), registerPragmaFunction)+import Database.DuckDB.Simple.Registration+```++The library has no typed cast in this release. A registered cast can give+wrong rows with no error, because the optimizer of DuckDB 2.0.0 decides from+the two types of a cast alone whether it can move the cast away from a column+in a filter or a join. In a run, it also gave min and max from the statistics+of the column without the cast. The cast function then does not run on that+column, or, in a comparison of two columns, it runs on one side only. In runs+with DuckDB 2.0.0, a cast from INTEGER to BIGINT, and a cast between BIGINT and+a custom type with the base type BIGINT, gave wrong rows when its function+changed the value. `registerCast` in+`Database.DuckDB.Simple.V2.Registration` registers a native cast, and its+Haddock gives the warning. For a conversion that changes the value, register a+scalar function with `createFunction` (see+[User-Defined Functions](#user-defined-functions)).++A registered cast does not change a value that the library reads or binds, on+a connection and in the callbacks of an extension. The library reads GEOMETRY+and VARIANT values with the built-in casts, and it builds GEOMETRY, ENUM, and+VARIANT values with the default cast of DuckDB. For example, a cast from+GEOMETRY to BLOB changes the result of `CAST(g AS BLOB)`, but not a GEOMETRY+value that a query gives or that a callback reads. Of the casts that the library+performs itself, only `setConfigValue` and `setConfigOption` use registered+casts. DuckDB casts the value to the type of the option, as SQL `SET` does.++`createAggregate` registers an aggregate function as a fold. Its result needs+`ToDuckValue` and a static `duckdbTypeDeclaration`, as a scalar function result+does. Lists, maps, and records or sum types through `ViaDuckDB` can be results.+A `Maybe a` result declares the type of `a`, and `Nothing` gives NULL. Types+with no static declaration, such as `StructValue`, are errors at registration.+See [Types](#types) for the encoding rules. Aggregates support `GROUP BY`,+`DISTINCT`, `FILTER`, `ORDER BY`, and window functions.++```haskell+aggregateExample :: Connection -> IO [Only Int64]+aggregateExample conn = do+ createAggregate conn "haskell_sum" Aggregate+ { aggregateInitial = 0 :: Int64+ , aggregateStep = \total (Only value) -> pure (total + value)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }+ query_ conn "SELECT haskell_sum(i) FROM range(10) t(i)" -- [Only 45]+```++The library keeps the live aggregate states in a table with stripes. Two+DuckDB threads wait for each other only when they change the same stripe.+`connectionAggregateStripes` in `ConnectionOptions` sets the number of stripes+for each aggregate that `createAggregate` registers through the connection.+The default is `defaultAggregateStripes`, 64. No result depends on the number.++Scalar and aggregate registrations create a private helper instance when an+argument needs VARIANT decoding. In a loadable extension, this helper also+supplies the factory for type metadata, GEOMETRY reads, and aggregate result+encoders. The registration shares the instance+across queries until the last native callback owner releases it. Each VARIANT+vector uses a separate connection and decoder state. The helper uses+`threads=1` and `async_threads=0`, with no background worker pool. A registration+on a connection uses its existing instance for type metadata and GEOMETRY reads.++`createTableFunction` registers a table function that gives its rows as a+list. DuckDB reads only the columns that the query uses, also in `WHERE` and+`ORDER BY`. For a query that uses no column, such as `SELECT count(*)`, DuckDB+reads the first column. The library does not evaluate the other columns of a+row, so an exception in such a column does not fail the query. The library+evaluates each row that it writes to its constructor. This evaluates each+strict field of the row (a field with `!`, or each field with `StrictData`).++```haskell+tableExample :: Connection -> IO [(Int64, Int64)]+tableExample conn = do+ createTableFunction conn "squares" ["n", "square"] \(Only count) ->+ pure [(n, n * n) | n <- [0 .. count - 1 :: Int64]]+ query_ conn "SELECT n, square FROM squares(3)" -- [(0, 0), (1, 1), (2, 4)]+```++The type of a column, a parameter, or an aggregate argument comes from+`duckdbTypeDeclaration`: notes 8, 9, and 10 of [Types](#types) give the rule+for a `SizedArray`, records, sum types, and `StructValue`, `TupleValue`, and+`UnionValue`. Arguments use the same conversion as query results.++The decoder uses SQL to inspect VARIANT values. For callback arguments it+copies the rows into a private DuckDB instance. The helper has no access to+the host's catalog, settings, or registered casts. A concrete SQL argument type+avoids this work when all rows have a known type. A decoding error fails the+calling statement under the usual callback error rules.++`registerReplacementScan` gives the parts of an unknown table name to the+function. The function can give a SELECT statement that DuckDB reads instead.++```haskell+scanExample :: Connection -> IO [Only Int64]+scanExample conn = do+ registerReplacementScan conn \parts ->+ pure (if parts == ["numbers"] then Just (ReplaceWithQuery "SELECT i FROM range(3) t(i)") else Nothing)+ query_ conn "SELECT i FROM numbers ORDER BY i" -- [Only 0, Only 1, Only 2]+```++`registerCopyFunction` registers a COPY format with a COPY TO side and a COPY+FROM side. DuckDB keeps the first registration of a format name. Thus register+a format with two directions in one call.++```haskell+copyExample :: Connection -> Text -> IO [Only Text]+copyExample conn path = do+ registerCopyFunction conn "haskell_lines" linesTo \info ->+ map Only . Text.lines <$> TextIO.readFile (copyFromPath info)+ _ <- execute conn "COPY (SELECT i::VARCHAR AS line FROM range(3) t(i)) TO ? (FORMAT haskell_lines)" (Only path)+ _ <- execute_ conn "CREATE TABLE copied (line VARCHAR)"+ _ <- execute conn "COPY copied FROM ? (FORMAT haskell_lines)" (Only path)+ query_ conn "SELECT line FROM copied ORDER BY line" -- [Only "0", Only "1", Only "2"]++linesTo :: CopyTo () FilePath+linesTo =+ CopyTo+ { copyToBind = \_ -> pure ()+ , copyToInit = \info -> TextIO.writeFile (copyInitFilePath info) "" >> pure (copyInitFilePath info)+ , copyToSink = \info rows -> TextIO.appendFile (copySinkGlobalState info) (Text.unlines (map line rows))+ , copyToFinalize = \_ -> pure ()+ }+ where+ line [Field{fieldValue = FieldText text}] = text+ line _ = error "haskell_lines: each row must have one VARCHAR value"+```++`registerPragmaFunction` gives the arguments of the call to the callback as a+`FromRow` row. A `QueryPragma` callback gives SQL that DuckDB runs in place of+the PRAGMA statement.++```haskell+pragmaExample :: Connection -> IO [Only Int64]+pragmaExample conn = do+ registerPragmaFunction conn "repeat_answer" [parameter "times" (TypeExpression "BIGINT")] $+ QueryPragma \(Only (times :: Int64)) -> pure ("SELECT 42::BIGINT FROM range(" <> Text.pack (show times) <> ")")+ query_ conn "PRAGMA repeat_answer(2)" -- [Only 42, Only 42]+```++A typed registration has more overhead for each row than raw vector access. Use+`Database.DuckDB.Simple.V2.Registration`, `V2.Copy`, `V2.Pragma`, and+`V2.Config` for speed, and for filter pushdown, parallel table scans, and+scalar functions with a bind callback or variadic parameters. A typed table+function has projection pushdown (see `createTableFunction` above), but no+filter pushdown.++## Loadable extensions++`Database.DuckDB.Simple.Extension` registers the typed functions of this package+from a loadable DuckDB extension. Build the extension with the `extension` flag+of `duckdb-ffi` and of `duckdb-simple`. The module's Haddock describes the+entry point, typed registrations, and resource lifetimes. See+`dev/extensions/README.md` in this repo. CI builds and tests the example.++## Tests++The package has five test suites. Each suite tests the library through a+database.++- `duckdb-simple-test` is the main suite, built with+ [tasty](https://hackage.haskell.org/package/tasty). It covers connections,+ statements, parameters, results, streaming, Arrow batches, cancellation,+ COPY functions, VARIANT and GEOMETRY values, typed registrations, and+ callbacks.+- `duckdb-simple-leak-test` counts native threads and resident memory through+ repeated queries, failures, and cancellation.+- `duckdb-simple-nonthreaded-test` is linked without `-threaded`. It checks the+ error for the registration of a callback in such a program.+- `duckdb-simple-dataframe-test` tests the Arrow export with the dataframe+ library. It needs the flag `dataframe-tests`.+- `duckdb-simple-quack-test` tests authenticated local Quack connections. It+ needs the flag `quack-tests`, and the variables `DUCKDB_HTTPFS_EXTENSION` and+ `DUCKDB_QUACK_EXTENSION` must name the extension files.++```+cabal test all --test-show-details=direct+cabal test duckdb-simple-test --test-show-details=direct+cabal test duckdb-simple-dataframe-test -fdataframe-tests+```++`cabal test all` runs the first three suites. The repository's CONTRIBUTING.md+gives the rules for new tests and the variables for a test run.
bench/Main.hs view
@@ -1,23 +1,30 @@+{-# LANGUAGE BangPatterns #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-} --- | Repeatable end-to-end benchmarks with checked results.+-- |+-- Module : Main+-- Description : Repeatable end-to-end benchmarks with checked results. module Main (main) where import Control.Exception (evaluate)-import Control.Monad (forM_, replicateM, unless)+import Control.Monad (forM_, unless)+import Data.Array (Array, elems, listArray, (!)) import qualified Data.ByteString as BS+import qualified Data.Foldable as Foldable import qualified Data.Geometry as G import Data.Int (Int64) import qualified Data.List as List+import qualified Data.Map.Strict as Map import qualified Data.Text as Text-import Data.Time.Clock.POSIX (utcTimeToPOSIXSeconds)+import Data.Time.Clock.POSIX (posixSecondsToUTCTime, utcTimeToPOSIXSeconds) import Data.Time.LocalTime (LocalTime, localTimeToUTC, utc) import Database.DuckDB.Simple-import Database.DuckDB.Simple.FromField (FieldValue (..))+import Database.DuckDB.Simple.FromField (BigNum (..), FieldValue (..), StructField (..), StructValue (..)) import Database.DuckDB.Simple.Geometry (RawGeometry (..)) import Database.DuckDB.Simple.Variant (Variant (..)) import GHC.Clock (getMonotonicTimeNSec)+import GHC.Stats (RTSStats (allocated_bytes), getRTSStats, getRTSStatsEnabled) import System.Environment (getArgs) import System.Mem (performMajorGC) import Text.Printf (printf)@@ -25,76 +32,154 @@ -- | Run one workload seven times after a warm-up run. main :: IO () main = do- args <- getArgs- let workload = case args of name : _ -> name; _ -> "eager"- count = case args of _ : n : _ -> read n; _ -> 100000 :: Int64- expected = count * (count - 1) `div` 2- withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn -> do- createFunction conn "bench_identity" (id :: Int64 -> Int64)- [Only geometry] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]- let typedGeometry = G.PointGeometry (G.PointXY (G.XY 1 2))- let sql = Query ("SELECT i FROM range(" <> Text.pack (show count) <> ") t(i)")- action = case workload of- "eager" -> do- rows <- query_ conn sql- evaluate (List.foldl' (\acc (Only n) -> acc + n) 0 rows)- "fold" -> fold_ conn sql 0 (\acc (Only n) -> pure (acc + n))- "scalar" -> do- rows <- query_ conn (Query ("SELECT bench_identity(i) FROM range(" <> Text.pack (show count) <> ") t(i)"))- evaluate (List.foldl' (\acc (Only n) -> acc + n) 0 rows)- "text" -> do- rows <- query_ conn (Query ("SELECT repeat('duckdb λ text', 4) FROM range(" <> Text.pack (show count) <> ")"))- evaluate (List.foldl' (\acc (Only value) -> acc + fromIntegral (Text.length value)) 0 rows)- "timestamp" -> do- rows <- query_ conn (Query ("SELECT TIMESTAMP '2000-01-01' + i * INTERVAL 1 SECOND FROM range(" <> Text.pack (show count) <> ") t(i)"))- evaluate (List.foldl' (\acc (Only (value :: LocalTime)) -> acc + floor (utcTimeToPOSIXSeconds (localTimeToUTC utc value))) 0 rows)- "parameters" -> do- rows <- replicateM (fromIntegral count) (query conn "SELECT ?::BIGINT" (Only (1 :: Int64)))- evaluate (sum [n | [Only n] <- rows])- "variant" -> do- rows <- query_ conn (Query ("SELECT i::VARIANT FROM range(" <> Text.pack (show count) <> ") t(i)"))- evaluate (List.foldl' (\acc (Only value) -> acc + variantNumber value) 0 rows)- "variant-parameters" -> do- rows <- replicateM (fromIntegral count) (query conn "SELECT ?" (Only (Variant (FieldInt64 1))))- evaluate (sum [variantNumber value | [Only value] <- rows])- "geometry" -> do- rows <- query_ conn (Query ("SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84') FROM range(" <> Text.pack (show count) <> ")"))- evaluate (sum [fromIntegral (BS.length (rawGeometryWKB value)) | Only value <- rows])- "geometry-parameters" -> do- rows <- replicateM (fromIntegral count) (query conn "SELECT system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?)" (rawGeometryWKB geometry, rawGeometryCRS geometry))- evaluate (sum [fromIntegral (BS.length (rawGeometryWKB value)) | [Only value] <- rows])- "geometry-typed" -> do- rows <- query_ conn (Query ("SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84') FROM range(" <> Text.pack (show count) <> ")"))- evaluate (sum [pointChecksum value | Only value <- rows])- "geometry-typed-parameters" -> do- rows <- replicateM (fromIntegral count) (query conn "SELECT ?" (Only typedGeometry))- evaluate (sum [pointChecksum value | [Only value] <- rows])- _ -> fail "Expected eager, fold, scalar, text, timestamp, parameters, variant, variant-parameters, geometry, geometry-parameters, geometry-typed, or geometry-typed-parameters"- expectedResult = case workload of- "parameters" -> count- "variant-parameters" -> count- "geometry" -> count * 21- "geometry-parameters" -> count * 21- "geometry-typed" -> count * 3- "geometry-typed-parameters" -> count * 3- "text" -> count * fromIntegral (Text.length (Text.replicate 4 "duckdb λ text"))- "timestamp" -> count * 946684800 + expected- _ -> expected- check actual = unless (actual == expectedResult) (fail "benchmark result mismatch")- action >>= check- putStrLn "workload,rows,run,milliseconds,checksum"- forM_ [1 .. 7 :: Int] $ \run -> do- performMajorGC- start <- getMonotonicTimeNSec- result <- action- end <- getMonotonicTimeNSec- check result- printf "%s,%d,%d,%.3f,%d\n" workload count run (fromIntegral (end - start) / 1000000 :: Double) result+ statsEnabled <- getRTSStatsEnabled+ unless statsEnabled (fail "Run the benchmark with +RTS -T")+ args <- getArgs+ let workload = case args of name : _ -> name; _ -> "eager"+ count = case args of _ : n : _ -> read n; _ -> 100000 :: Int64+ expected = count * (count - 1) `div` 2+ collectionQuery expression = Query ("SELECT " <> expression <> " FROM range(" <> Text.pack (show count) <> ") t(i)")+ objectKeys = listArray (0, 63) ["k" <> Text.pack (show index) | index <- [0 .. 63 :: Int]]+ withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn -> do+ createFunction conn "bench_identity" (id :: Int64 -> Int64)+ [Only geometry] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ let typedGeometry = G.PointGeometry (G.PointXY (G.XY 1 2))+ let sql = Query ("SELECT i FROM range(" <> Text.pack (show count) <> ") t(i)")+ action = case workload of+ "eager" -> do+ rows <- query_ conn sql+ evaluate (List.foldl' (\acc (Only n) -> acc + n) 0 rows)+ "fold" -> fold_ conn sql 0 (\acc (Only n) -> pure (acc + n))+ "scalar" -> do+ rows <- query_ conn (Query ("SELECT bench_identity(i) FROM range(" <> Text.pack (show count) <> ") t(i)"))+ evaluate (List.foldl' (\acc (Only n) -> acc + n) 0 rows)+ "text" -> do+ rows <- query_ conn (Query ("SELECT repeat('duckdb λ text', 4) FROM range(" <> Text.pack (show count) <> ")"))+ evaluate (List.foldl' (\acc (Only value) -> acc + fromIntegral (Text.length value)) 0 rows)+ "timestamp" -> do+ rows <- query_ conn (Query ("SELECT TIMESTAMP '2000-01-01' + i * INTERVAL 1 SECOND FROM range(" <> Text.pack (show count) <> ") t(i)"))+ evaluate (List.foldl' (\acc (Only (value :: LocalTime)) -> acc + floor (utcTimeToPOSIXSeconds (localTimeToUTC utc value))) 0 rows)+ "parameters" -> do+ repeatChecksum count $ do+ [Only value] <- query conn "SELECT ?::BIGINT" (Only (1 :: Int64))+ pure value+ "timestamp-parameters" -> do+ repeatChecksum count $ do+ [Only value] <- query conn "SELECT 1::BIGINT WHERE ? IS NOT NULL" (Only (posixSecondsToUTCTime 0))+ pure value+ "list-bigint" -> do+ [Only values] <- query_ conn (collectionQuery "list(i::BIGINT)") :: IO [Only [Int64]]+ collectionChecksum count values+ "list-bignum" -> do+ [Only values] <- query_ conn (collectionQuery "list(i::BIGNUM)") :: IO [Only [Integer]]+ collectionChecksum count values+ "array-bignum" -> do+ -- The array size comes from the command line, so read its FieldValue.+ [Only (FieldArray values)] <-+ query_ conn (collectionQuery ("list(i::BIGNUM)::BIGNUM[" <> Text.pack (show count) <> "]")) :: IO [Only FieldValue]+ collectionChecksum count [value | FieldBigNum (BigNum value) <- elems values]+ "map-bignum" -> do+ [Only values] <- query_ conn (collectionQuery "map(list(i::BIGINT), list(i::BIGNUM))") :: IO [Only (Map.Map Int64 Integer)]+ collectionChecksum count values+ "variant" -> do+ rows <- query_ conn (Query ("SELECT i::VARIANT FROM range(" <> Text.pack (show count) <> ") t(i)"))+ evaluate (List.foldl' (\acc (Only value) -> acc + variantNumber value) 0 rows)+ "variant-list" -> do+ [Only (Variant value)] <- query_ conn (collectionQuery "list(i)::VARIANT")+ case value of+ FieldList values -> collectionChecksum count (map (variantNumber . Variant) values)+ _ -> fail "unexpected VARIANT list benchmark shape"+ "variant-parameters" -> do+ repeatChecksum count $ do+ [Only value] <- query conn "SELECT ?" (Only (Variant (FieldInt64 1)))+ pure (variantNumber value)+ -- Objects with 64 different key sets.+ "variant-objects" -> do+ rows <- query_ conn (Query ("SELECT json_object('k' || (i % 64)::VARCHAR, i)::VARIANT FROM range(" <> Text.pack (show count) <> ") t(i)"))+ collectionChecksum count (map (\(Only value) -> variantObjectNumber objectKeys value) rows)+ "geometry" -> do+ rows <- query_ conn (Query ("SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84') FROM range(" <> Text.pack (show count) <> ")"))+ evaluate (sum [fromIntegral (BS.length (rawGeometryWKB value)) | Only value <- rows])+ "geometry-parameters" -> do+ repeatChecksum count $ do+ [Only value] <-+ query conn "SELECT system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?)" (rawGeometryWKB geometry, rawGeometryCRS geometry)+ pure (fromIntegral (BS.length (rawGeometryWKB value)))+ "geometry-typed" -> do+ rows <- query_ conn (Query ("SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84') FROM range(" <> Text.pack (show count) <> ")"))+ evaluate (sum [pointChecksum value | Only value <- rows])+ "geometry-typed-parameters" -> do+ repeatChecksum count $ do+ [Only value] <- query conn "SELECT ?" (Only typedGeometry)+ pure (pointChecksum value)+ _ ->+ fail+ "Expected eager, fold, scalar, text, timestamp, timestamp-parameters, parameters, list-bigint, list-bignum, array-bignum, map-bignum, variant, variant-list, variant-parameters, variant-objects, geometry, geometry-parameters, geometry-typed, or geometry-typed-parameters"+ expectedResult = case workload of+ "parameters" -> count+ "timestamp-parameters" -> count+ "variant-parameters" -> count+ "geometry" -> count * 21+ "geometry-parameters" -> count * 21+ "geometry-typed" -> count * 3+ "geometry-typed-parameters" -> count * 3+ "text" -> count * fromIntegral (Text.length (Text.replicate 4 "duckdb λ text"))+ "timestamp" -> count * 946684800 + expected+ _ -> expected+ check actual = unless (actual == expectedResult) (fail "benchmark result mismatch")+ action >>= check+ putStrLn "workload,rows,run,milliseconds,allocated_bytes,checksum"+ forM_ [1 .. 7 :: Int] $ \run -> do+ performMajorGC+ before <- getRTSStats+ start <- getMonotonicTimeNSec+ result <- action >>= evaluate+ end <- getMonotonicTimeNSec+ performMajorGC+ after <- getRTSStats+ check result+ printf+ "%s,%d,%d,%.3f,%d,%d\n"+ workload+ count+ run+ (fromIntegral (end - start) / 1000000 :: Double)+ (allocated_bytes after - allocated_bytes before)+ result --- | Fail if the VARIANT benchmark changes its scalar type.+-- | Repeat an action and add its results with a constant stack.+repeatChecksum :: Int64 -> IO Int64 -> IO Int64+repeatChecksum count action = go count 0+ where+ go remaining !total+ | remaining <= 0 = pure total+ | otherwise = do+ value <- action+ go (remaining - 1) (total + value)++-- | Check the element count and force the checksum of a collection.+collectionChecksum :: (Foldable f, Integral a) => Int64 -> f a -> IO Int64+collectionChecksum count values = do+ let (!actualCount, !checksum) = Foldable.foldl' (\(!size, !total) value -> (size + 1, total + fromIntegral value)) (0, 0) values+ unless (actualCount == count) (fail "benchmark collection size mismatch")+ pure checksum++-- | Read the integer in a scalar VARIANT. variantNumber :: Variant -> Int64 variantNumber (Variant (FieldInt64 value)) = value+variantNumber (Variant (FieldWord64 value))+ | value <= fromIntegral (maxBound :: Int64) = fromIntegral value variantNumber value = error ("unexpected VARIANT benchmark value: " <> show value)++-- | Check one object field and its key, then read its integer payload.+variantObjectNumber :: Array Int Text.Text -> Variant -> Int64+variantObjectNumber keys (Variant (FieldStruct value)) = case elems (structValueFields value) of+ [StructField key payload]+ | let number = variantNumber (Variant payload)+ , key == keys ! fromIntegral (number `mod` 64) ->+ number+ _ -> error "unexpected VARIANT object benchmark fields"+variantObjectNumber _ _ = error "unexpected VARIANT object benchmark shape" -- | Force coordinate decoding and check the point shape. pointChecksum :: G.Geometry -> Int64
dataframe-test/Main.hs view
@@ -1,23 +1,30 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE DuplicateRecordFields #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-}-{-# OPTIONS_GHC -Wno-deprecations #-}+{-# LANGUAGE TypeApplications #-} --- | Consume DuckDB Arrow batches with the dataframe library.+-- |+-- Module : Main+-- Description : Consume DuckDB Arrow batches with the dataframe library. module Main (main) where import Control.Exception (ErrorCall, displayException, try) import Control.Monad (forM_)+import Data.Coerce (coerce) import Data.Int (Int64) import qualified Data.Text as Text+import Data.Void (Void) import qualified DataFrame.Core as DataFrame import qualified DataFrame.IO.Arrow as DataFrameArrow-import Database.DuckDB.FFI (ArrowArray (..), ArrowSchema (..))+import Database.DuckDB.FFI.V2 (ArrowArray (..), ArrowSchema (..)) import Database.DuckDB.Simple import qualified Database.DuckDB.Simple.Arrow as Arrow-import qualified Database.DuckDB.Simple.Deprecated.Streaming as Streaming+import qualified Database.DuckDB.Simple.Streaming as Streaming import Foreign.Ptr (Ptr, castPtr, nullFunPtr) import Foreign.Storable (peek)+import GHC.Records (getField) import Test.Tasty (TestTree, defaultMain, testGroup) import Test.Tasty.HUnit @@ -28,45 +35,45 @@ -- | Check data and ownership across successful and failed imports. dataframeTests :: Bool -> TestTree dataframeTests streaming =- testGroup- (if streaming then "deprecated streaming" else "materialized")- [ testCase "several batches remain readable after connection close" do- (rowCount, batches) <- withDb \conn ->- foldArrow- conn- "SELECT i::INTEGER AS id, CASE WHEN i % 3 = 0 THEN NULL ELSE i - 2500 END AS \"íslenska_λ\", (i::FLOAT / 4)::FLOAT AS small, CASE WHEN i % 5 = 0 THEN NULL ELSE i::DOUBLE / 4 END AS real, CASE WHEN i % 7 = 0 THEN NULL WHEN i % 7 = 1 THEN '' ELSE ? || i::VARCHAR END AS label, NULL::BIGINT AS missing FROM range(?::BIGINT) t(i)"- ("λ\0雪" :: Text.Text, 5000 :: Int64)- (0 :: Int, [])- \(offset, acc) schema array -> do- assertLiveSchema schema- count <- fromIntegral . arrowArrayLength <$> peek array- frame <- DataFrameArrow.arrowToDataframe (castPtr schema) (castPtr array)- assertConsumed schema array- pure (offset + count, (offset, count, frame) : acc)- rowCount @?= 5000- assertBool "multiple native batches" (length batches > 1)- forM_ batches \(offset, count, frame) -> do- let expected = expectedFrame [offset .. offset + count - 1]- DataFrame.columnNames frame @?= DataFrame.columnNames expected- frame @?= expected- , testCase "an unsupported type releases the failed export" $- withDb \conn -> do- -- Import one supported column before the unsupported column fails.- result <- try $ foldArrow conn "SELECT 42::BIGINT AS n, true AS unsupported" () () \() schema array -> do- _ <- DataFrameArrow.arrowToDataframe (castPtr schema) (castPtr array)- pure ()- case result of- Left (err :: ErrorCall) ->- assertBool "DataFrame format error" ("unsupported format" `Text.isInfixOf` Text.pack (displayException err))- Right () -> assertFailure "expected an unsupported Arrow type"- query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])- , testCase "an empty result does not call the consumer" $- withDb \conn -> do- count <- foldArrow conn "SELECT 1::BIGINT AS n WHERE false" () (0 :: Int) \_ schema array -> do- _ <- DataFrameArrow.arrowToDataframe (castPtr schema) (castPtr array)- assertFailure "unexpected batch"- count @?= 0- ]+ testGroup+ (if streaming then "native streaming" else "materialized")+ [ testCase "several batches remain readable after connection close" do+ (rowCount, batches) <- withDb \conn ->+ foldArrow+ conn+ "SELECT i::INTEGER AS id, CASE WHEN i % 3 = 0 THEN NULL ELSE i - 2500 END AS \"íslenska_λ\", (i::FLOAT / 4)::FLOAT AS small, CASE WHEN i % 5 = 0 THEN NULL ELSE i::DOUBLE / 4 END AS real, CASE WHEN i % 7 = 0 THEN NULL WHEN i % 7 = 1 THEN '' ELSE ? || i::VARCHAR END AS label, NULL::BIGINT AS missing FROM range(?::BIGINT) t(i)"+ ("λ\0雪" :: Text.Text, 5000 :: Int64)+ (0 :: Int, [])+ \(offset, acc) schema array -> do+ assertLiveSchema schema+ count <- fromIntegral . (getField @"length") <$> peek array+ frame <- DataFrameArrow.arrowToDataframe (castPtr schema) (castPtr array)+ assertConsumed schema array+ pure (offset + count, (offset, count, frame) : acc)+ rowCount @?= 5000+ assertBool "multiple native batches" (length batches > 1)+ forM_ batches \(offset, count, frame) -> do+ let expected = expectedFrame [offset .. offset + count - 1]+ DataFrame.columnNames frame @?= DataFrame.columnNames expected+ frame @?= expected+ , testCase "an unsupported type releases the failed export" $+ withDb \conn -> do+ -- Import one supported column before the unsupported column fails.+ result <- try $ foldArrow conn "SELECT 42::BIGINT AS n, true AS unsupported" () () \() schema array -> do+ _ <- DataFrameArrow.arrowToDataframe (castPtr schema) (castPtr array)+ pure ()+ case result of+ Left (err :: ErrorCall) ->+ assertBool "DataFrame format error" ("unsupported format" `Text.isInfixOf` Text.pack (displayException err))+ Right () -> assertFailure "expected an unsupported Arrow type"+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "an empty result does not call the consumer" $+ withDb \conn -> do+ count <- foldArrow conn "SELECT 1::BIGINT AS n WHERE false" () (0 :: Int) \_ schema array -> do+ _ <- DataFrameArrow.arrowToDataframe (castPtr schema) (castPtr array)+ assertFailure "unexpected batch"+ count @?= 0+ ] where foldArrow :: (ToRow q) => Connection -> Query -> q -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a foldArrow = if streaming then Streaming.foldArrow else Arrow.foldArrow@@ -74,31 +81,31 @@ -- | Construct the expected values independently of the Arrow buffer layout. expectedFrame :: [Int] -> DataFrame.DataFrame expectedFrame rows =- DataFrame.fromNamedColumns- [ ("id", DataFrame.fromList rows)- , ("íslenska_λ", DataFrame.fromList [if i `rem` 3 == 0 then Nothing else Just (i - 2500) | i <- rows])- , ("small", DataFrame.fromList [fromIntegral i / 4 :: Double | i <- rows])- , ("real", DataFrame.fromList [if i `rem` 5 == 0 then Nothing else Just (fromIntegral i / 4 :: Double) | i <- rows])- , ("label", DataFrame.fromList [label i | i <- rows])- , ("missing", DataFrame.fromList [Nothing :: Maybe Int | _ <- rows])- ]+ DataFrame.fromNamedColumns+ [ ("id", DataFrame.fromList rows)+ , ("íslenska_λ", DataFrame.fromList [if i `rem` 3 == 0 then Nothing else Just (i - 2500) | i <- rows])+ , ("small", DataFrame.fromList [fromIntegral i / 4 :: Double | i <- rows])+ , ("real", DataFrame.fromList [if i `rem` 5 == 0 then Nothing else Just (fromIntegral i / 4 :: Double) | i <- rows])+ , ("label", DataFrame.fromList [label i | i <- rows])+ , ("missing", DataFrame.fromList [Nothing :: Maybe Int | _ <- rows])+ ] where label i- | i `rem` 7 == 0 = Nothing- | i `rem` 7 == 1 = Just ""- | otherwise = Just ("λ\0雪" <> Text.pack (show i))+ | i `rem` 7 == 0 = Nothing+ | i `rem` 7 == 1 = Just ""+ | otherwise = Just ("λ\0雪" <> Text.pack (show i)) -- | Fail before the consumer can dereference a previously released schema. assertLiveSchema :: Ptr ArrowSchema -> Assertion assertLiveSchema ptr = do- schema <- peek ptr- assertBool "each batch needs an unreleased schema" (arrowSchemaRelease schema /= nullFunPtr)+ schema <- peek ptr+ assertBool "each batch needs an unreleased schema" ((getField @"release") schema /= (coerce (nullFunPtr @Void))) -- | The DataFrame importer releases both objects after copying their contents. assertConsumed :: Ptr ArrowSchema -> Ptr ArrowArray -> Assertion assertConsumed schema array = do- arrowSchemaRelease <$> peek schema >>= (@?= nullFunPtr)- arrowArrayRelease <$> peek array >>= (@?= nullFunPtr)+ (getField @"release") <$> peek schema >>= (@?= (coerce (nullFunPtr @Void)))+ (getField @"release") <$> peek array >>= (@?= (coerce (nullFunPtr @Void))) -- | Keep native batch ordering deterministic in both execution modes. withDb :: (Connection -> IO a) -> IO a
duckdb-simple.cabal view
@@ -1,6 +1,6 @@ cabal-version: 3.4 name: duckdb-simple-version: 0.3.0.0+version: 1.0.0.0 license: MPL-2.0 license-file: LICENSE author: Matthias Pall Gissurarson@@ -21,9 +21,12 @@ A high-level DuckDB interface with typed parameters and results, prepared statements, chunked folds, transactions, and Haskell scalar functions. The API follows the style of sqlite-simple and postgresql-simple.- .- Supports native DuckDB >= 1.5.3 and < 1.6. Tested with versions 1.5.3 through 1.5.6. + The backend uses the generated V2 API from duckdb-ffi. Upstream marks+ that API as unstable, so this release requires exactly DuckDB 2.0.0.+ It provides shared environments and instances, function registration,+ native streaming, and Arrow batches.+ extra-doc-files: CHANGELOG.md README.md@@ -38,14 +41,25 @@ default: False manual: True +flag extension+ description: Build loadable extensions with private C callback helpers.+ default: False+ manual: True++flag quack-tests+ description: Test authenticated local Quack connections with supplied extensions.+ default: False+ manual: True+ library exposed-modules: Database.DuckDB.Simple Database.DuckDB.Simple.Arrow Database.DuckDB.Simple.Catalog Database.DuckDB.Simple.Config+ Database.DuckDB.Simple.Conversion Database.DuckDB.Simple.Copy- Database.DuckDB.Simple.Deprecated.Streaming+ Database.DuckDB.Simple.Extension Database.DuckDB.Simple.FileSystem Database.DuckDB.Simple.FromField Database.DuckDB.Simple.FromRow@@ -56,49 +70,91 @@ Database.DuckDB.Simple.Logging Database.DuckDB.Simple.LogicalRep Database.DuckDB.Simple.Ok+ Database.DuckDB.Simple.Pragma+ Database.DuckDB.Simple.QueryMetrics+ Database.DuckDB.Simple.Registration+ Database.DuckDB.Simple.Streaming Database.DuckDB.Simple.Time Database.DuckDB.Simple.ToField+ Database.DuckDB.Simple.Tokenizer Database.DuckDB.Simple.ToRow Database.DuckDB.Simple.Types+ Database.DuckDB.Simple.V2.Config+ Database.DuckDB.Simple.V2.Copy+ Database.DuckDB.Simple.V2.FunctionDocs+ Database.DuckDB.Simple.V2.Logging+ Database.DuckDB.Simple.V2.LogicalRep+ Database.DuckDB.Simple.V2.Pragma+ Database.DuckDB.Simple.V2.Registration Database.DuckDB.Simple.Variant other-modules: Database.DuckDB.Simple.Arrow.Internal Database.DuckDB.Simple.Callback+ Database.DuckDB.Simple.CallbackMark+ Database.DuckDB.Simple.Config.Internal+ Database.DuckDB.Simple.Copy.Internal Database.DuckDB.Simple.Element+ Database.DuckDB.Simple.Function.Internal+ Database.DuckDB.Simple.Logging.Internal Database.DuckDB.Simple.Materialize+ Database.DuckDB.Simple.Native+ Database.DuckDB.Simple.Pragma.Internal Database.DuckDB.Simple.Result- Database.DuckDB.Simple.TypeCache+ Database.DuckDB.Simple.Typed+ Database.DuckDB.Simple.Typed.Aggregate+ Database.DuckDB.Simple.Typed.Table+ Database.DuckDB.Simple.TypeScope Database.DuckDB.Simple.VariantCodec hs-source-dirs: src default-language: Haskell2010++ if flag(extension)+ if os(linux) || os(osx)+ -- Keep the C stubs private to this shared library. Omit unfoldings, so+ -- optimized clients cannot inline references to the hidden stubs.+ cc-options: -fvisibility=hidden+ ghc-options:+ -fomit-interface-pragmas+ -optc-fvisibility=hidden+ else+ buildable: False build-depends: array >=0.5 && <0.6,- base >=4.14 && <5,+ base >=4.18 && <5, bytestring >=0.11 && <0.13, containers >=0.6 && <0.9,- duckdb-ffi >=1.5.6.0 && <1.6,+ duckdb-ffi >=2.0.0.0 && <2.0.1, geometry-simple >=0.1.1.0 && <0.2,+ hs-bindgen-runtime >=1.0.0.0 && <1.1, text >=2.0 && <2.2, time >=1.12 && <1.16, transformers >=0.6 && <0.7, uuid >=1.3 && <1.4, test-suite duckdb-simple-test+ if flag(extension)+ buildable: False type: exitcode-stdio-1.0 hs-source-dirs: test main-is: Spec.hs other-modules: ArrowTests CancellationTests+ CopyTests CoreRegressionTests ExtensionRegressionTests GeometryRegressionTests Properties+ RegistrationTests+ SizedArrayTests+ SqlFeatureTests StreamingTests TestUtils TimeTests+ TypedRegistrationTests+ UserCastTests ValueRegressionTests VariantRegressionTests @@ -109,13 +165,15 @@ build-depends: array >=0.5 && <0.6,- base >=4.14 && <5,+ base >=4.18 && <5, bytestring, containers >=0.6 && <0.9, directory >=1.3 && <1.4,- duckdb-ffi >=1.5.6.0 && <1.6,+ duckdb-ffi >=2.0.0.0 && <2.0.1, duckdb-simple,+ filepath >=1.4 && <1.6, geometry-simple >=0.1.1.0 && <0.2,+ hs-bindgen-runtime >=1.0.0.0 && <1.1, QuickCheck >=2.14 && <2.18, tasty >=1.4 && <1.6, tasty-expected-failure >=0.12 && <0.13,@@ -126,51 +184,34 @@ uuid >=1.3 && <1.4, vector >=0.13 && <0.14, -test-suite duckdb-simple-codec-test+test-suite duckdb-simple-leak-test+ if flag(extension)+ buildable: False type: exitcode-stdio-1.0- hs-source-dirs:- test-codec- src-+ hs-source-dirs: leaktest main-is: Main.hs- other-modules:- Database.DuckDB.Simple.Element- Database.DuckDB.Simple.FromField- Database.DuckDB.Simple.Geometry- Database.DuckDB.Simple.Internal- Database.DuckDB.Simple.LogicalRep- Database.DuckDB.Simple.Ok- Database.DuckDB.Simple.Time- Database.DuckDB.Simple.TypeCache- Database.DuckDB.Simple.Variant- Database.DuckDB.Simple.VariantCodec- default-language: Haskell2010+ ghc-options: -threaded build-depends:- array >=0.5 && <0.6,- base >=4.14 && <5,- bytestring >=0.11 && <0.13,+ base >=4.18 && <5,+ bytestring, containers >=0.6 && <0.9,- duckdb-ffi >=1.5.6.0 && <1.6,+ duckdb-simple, geometry-simple >=0.1.1.0 && <0.2,- tasty >=1.4 && <1.6,- tasty-hunit >=0.10 && <0.12,- text >=2.0 && <2.2,- time >=1.12 && <1.16,- transformers >=0.6 && <0.7,- uuid >=1.3 && <1.4,+ text, -test-suite duckdb-simple-leak-test+-- This suite has no -threaded. It checks the guard for such a program.+test-suite duckdb-simple-nonthreaded-test+ if flag(extension)+ buildable: False type: exitcode-stdio-1.0- hs-source-dirs: leaktest+ hs-source-dirs: nonthreaded-test main-is: Main.hs default-language: Haskell2010- ghc-options: -threaded build-depends:- base >=4.14 && <5,- bytestring,+ base >=4.18 && <5, duckdb-simple,- geometry-simple >=0.1.1.0 && <0.2,+ text, test-suite duckdb-simple-dataframe-test type: exitcode-stdio-1.0@@ -179,19 +220,40 @@ default-language: Haskell2010 ghc-options: -threaded - if !flag(dataframe-tests)+ if flag(extension) || !flag(dataframe-tests) buildable: False build-depends:- base >=4.14 && <5,+ base >=4.18 && <5, dataframe-arrow-bridge >=1.0 && <1.1, dataframe-core >=2.5 && <2.6,- duckdb-ffi >=1.5.6.0 && <1.6,+ duckdb-ffi >=2.0.0.0 && <2.0.1, duckdb-simple,+ hs-bindgen-runtime >=1.0.0.0 && <1.1, tasty >=1.4 && <1.6, tasty-hunit >=0.10 && <0.12, text, +test-suite duckdb-simple-quack-test+ type: exitcode-stdio-1.0+ hs-source-dirs: test-quack+ main-is: Main.hs+ default-language: Haskell2010+ ghc-options: -threaded++ if flag(extension) || !flag(quack-tests)+ buildable: False+ build-depends:+ base >=4.18 && <5,+ directory >=1.3 && <1.4,+ duckdb-simple,+ tasty >=1.4 && <1.6,+ tasty-hunit >=0.10 && <0.12,+ text,+ time,+ benchmark duckdb-simple-bench+ if flag(extension)+ buildable: False type: exitcode-stdio-1.0 hs-source-dirs: bench main-is: Main.hs@@ -202,8 +264,10 @@ -rtsopts build-depends:- base >=4.14 && <5,+ array >=0.5 && <0.6,+ base >=4.18 && <5, bytestring,+ containers >=0.6 && <0.9, duckdb-simple, geometry-simple >=0.1.1.0 && <0.2, text,
leaktest/Main.hs view
@@ -1,38 +1,56 @@ {-# LANGUAGE BlockArguments #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-}-{-# OPTIONS_GHC -Wno-deprecations #-} -{- | Regression check for leaked DuckDB handles.--The handles live in C memory, so the GHC heap statistics cannot see them.-The open/close check counts native threads and resident memory. The sustained-checks keep one connection open through repeated queries, failures, cursor-resets, callback replacement, and cancellation. Weak references check callback-state release independently of resident memory.--Each check runs in this dedicated process. On systems without-@/proc/self/status@, the native resource counters are unavailable. The-functional checks and callback collection checks still run.--}+-- |+-- Module : Main+-- Description : Regression check for leaked DuckDB handles.+--+-- The handles live in C memory, so the GHC heap statistics cannot see them. The+-- open/close check counts native threads and resident memory. The sustained+-- checks keep one connection open through repeated queries, failures, cursor+-- resets, callback replacement, and cancellation. Weak references check callback+-- state release independently of resident memory.+--+-- Each check runs in this dedicated process. On systems without+-- @/proc/self/status@, the native resource counters are unavailable. The+-- functional checks and callback collection checks still run. module Main (main) where import Control.Concurrent (forkFinally, killThread, newEmptyMVar, putMVar, takeMVar, threadDelay, tryPutMVar)-import Control.Exception (AsyncException (ThreadKilled), IOException, SomeException, evaluate, fromException, try)-import Control.Monad (forM_, replicateM, unless, void, when)+import Control.Exception (+ AsyncException (ThreadKilled),+ BlockedIndefinitelyOnMVar (..),+ IOException,+ SomeException,+ evaluate,+ finally,+ fromException,+ throwIO,+ try,+ )+import Control.Monad (forM, forM_, replicateM, replicateM_, unless, void, when) import qualified Data.ByteString as BS import qualified Data.Geometry as G import Data.IORef (atomicModifyIORef', atomicWriteIORef, mkWeakIORef, newIORef, readIORef) import Data.Int (Int64)-import Data.List (isPrefixOf)-import Data.Maybe (catMaybes, listToMaybe)+import Data.List (isPrefixOf, minimumBy)+import qualified Data.Map.Strict as Map+import Data.Maybe (catMaybes, isJust, listToMaybe)+import Data.Ord (comparing)+import Data.Text (Text)+import qualified Data.Text as Text import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Arrow as Arrow import qualified Database.DuckDB.Simple.Copy as Copy-import qualified Database.DuckDB.Simple.Deprecated.Streaming as Streaming-import Database.DuckDB.Simple.FromField (DecimalValue (..), FieldValue (..))+import Database.DuckDB.Simple.FromField (BigNum (..), DecimalValue (..), FieldValue (..), StructValue, UnionValue) import Database.DuckDB.Simple.Geometry (RawGeometry (..), fromRawGeometry) import qualified Database.DuckDB.Simple.Logging as Logging-import Database.DuckDB.Simple.Variant (Variant (..))+import Database.DuckDB.Simple.Registration (TypeDeclaration (..), registerCustomType)+import qualified Database.DuckDB.Simple.Streaming as Streaming+import qualified Database.DuckDB.Simple.V2.Registration as R+import Database.DuckDB.Simple.Variant (Variant (..), variantObject)+import GHC.Weak.Finalize (getFinalizerExceptionHandler, setFinalizerExceptionHandler) import System.Environment (getArgs, lookupEnv) import System.Exit (exitFailure) import System.Mem (performMajorGC)@@ -41,10 +59,10 @@ -- | The process-global resource counters that a leaked handle increases. data Usage = Usage- { usageThreads :: Int- , usageRssKb :: Int- }- deriving (Show)+ { usageThreads :: Int+ , usageRssKb :: Int+ }+ deriving (Show) -- | The number of open\/close cycles that the check runs. cycles :: Int@@ -58,263 +76,406 @@ rssSlackKb :: Int rssSlackKb = 32 * 1024 -{- | Read the current resource counters. The result is 'Nothing' if the system-has no @\/proc\/self\/status@.--}+-- | Read the current resource counters. The result is 'Nothing' if the system+-- has no @\/proc\/self\/status@. readUsage :: IO (Maybe Usage) readUsage = do- result <- try (readFile "/proc/self/status") :: IO (Either IOException String)- case result of- Left _ -> pure Nothing- Right contents -> do- _ <- evaluate (length contents)- pure (Usage <$> statusField "Threads" contents <*> statusField "VmRSS" contents)+ result <- try (readFile "/proc/self/status") :: IO (Either IOException String)+ case result of+ Left _ -> pure Nothing+ Right contents -> do+ _ <- evaluate (length contents)+ pure (Usage <$> statusField "Threads" contents <*> statusField "VmRSS" contents) where statusField name contents =- listToMaybe- [ value- | line <- lines contents- , (name <> ":") `isPrefixOf` line- , (value, _) <- reads (drop (length name + 1) line)- ]+ listToMaybe+ [ value+ | line <- lines contents+ , (name <> ":") `isPrefixOf` line+ , (value, _) <- reads (drop (length name + 1) line)+ ] -- | Open a connection, use a statement, then close both. openCloseCycle :: IO () openCloseCycle = do- conn <- open ":memory:"- stmt <- openStatement conn "SELECT 42"- closeStatement stmt- _ <- query_ conn "SELECT 42" :: IO [Only Int64]- -- The first VARIANT parameter fills the type cache. Close must destroy it.- _ <- query conn "SELECT ?" (Only (Variant (FieldInt64 42))) :: IO [Only Variant]- -- A TIMETZ offset with seconds fails during result materialization.- -- The result and connection must still be destroyed.- rejected <- try (query_ conn "SELECT {'xs': [i, i + 1, i + 2, i + 3], 'bad': '12:00:00+01:23:45'::TIMETZ} FROM range(100000) t(i)") :: IO (Either SomeException [Only FieldValue])- case rejected of- Left _ -> pure ()- Right _ -> fail "expected TIMETZ offset conversion failure"- close conn+ conn <- open ":memory:"+ stmt <- openStatement conn "SELECT 42"+ closeStatement stmt+ _ <- query_ conn "SELECT 42" :: IO [Only Int64]+ values <- query conn "SELECT ?" (Only (Variant (FieldInt64 42)))+ unless (values == [Only (Variant (FieldInt64 42))]) (fail "wrong VARIANT payload")+ -- The reader of a VARIANT value fails while it reads the chunk. See+ -- breakVariantReads. The result and connection must still be destroyed.+ breakVariantReads conn+ rejected <- try (query_ conn rejectedQuery) :: IO (Either SomeException [Only FieldValue])+ case rejected of+ Left _ -> pure ()+ Right _ -> fail "expected the failure of the VARIANT reader"+ close conn main :: IO () main = do- args <- getArgs- let (mode, batches) = case args of- [] -> ("all", 10)- [name] -> (name, 10)- [name, n] | Just count <- readMaybe n, count > 0 -> (name, count)- _ -> ("invalid", 0)- unless (mode `elem` ["all", "open-close", "long-lived", "callbacks", "cancel", "decode-failure", "new-types"]) $- fail "Expected [all|open-close|long-lived|callbacks|cancel|decode-failure|new-types] [positive batch count]"- when (mode `elem` ["all", "open-close"]) checkOpenClose- when (mode `elem` ["all", "long-lived"]) (checkLongLived batches)- when (mode `elem` ["all", "callbacks"]) (checkCallbacks batches)- when (mode `elem` ["all", "cancel"]) (checkCancellation batches)- when (mode == "decode-failure") checkDecodeFailure- when (mode `elem` ["all", "new-types"]) (checkNewTypes batches)+ args <- getArgs+ let (mode, batches) = case args of+ [] -> ("all", 10)+ [name] -> (name, 10)+ [name, n] | Just count <- readMaybe n, count > 0 -> (name, count)+ _ -> ("invalid", 0)+ unless (mode `elem` ["all", "open-close", "dropped-handles", "long-lived", "callbacks", "cancel", "decode-failure", "new-types", "typed"]) $+ fail "Expected [all|open-close|dropped-handles|long-lived|callbacks|cancel|decode-failure|new-types|typed] [positive batch count]"+ when (mode `elem` ["all", "open-close"]) checkOpenClose+ when (mode `elem` ["all", "dropped-handles"]) checkDroppedHandles+ when (mode `elem` ["all", "long-lived"]) (checkLongLived batches)+ when (mode `elem` ["all", "callbacks"]) (checkCallbacks batches)+ when (mode `elem` ["all", "cancel"]) (checkCancellation batches)+ when (mode == "decode-failure") checkDecodeFailure+ when (mode `elem` ["all", "new-types"]) (checkNewTypes batches)+ when (mode `elem` ["all", "typed"]) (checkTyped batches) +-- | Drop a connection and one of its statements without closing them. The+-- statement is younger than the connection, so the collector can run the+-- finalizer of the connection first. Cleanup must still release the database,+-- and no finalizer may wait for a finalizer that runs after it.+checkDroppedHandles :: IO ()+checkDroppedHandles = do+ blocked <- newIORef (0 :: Int)+ previous <- getFinalizerExceptionHandler+ let record err = do+ case fromException err of+ Just BlockedIndefinitelyOnMVar -> atomicModifyIORef' blocked (\count -> (count + 1, ()))+ Nothing -> pure ()+ previous err+ -- The database owns the callback, and the callback retains the token.+ allocate before after = do+ token <- newIORef (42 :: Int64)+ weak <- mkWeakIORef token (pure ())+ conn <- open ":memory:"+ createFunction conn "dropped_token" (readIORef token)+ -- Decoding a VARIANT opens the helper that the finalizer must release.+ values <- query conn "SELECT ?" (Only (Variant (FieldInt64 42)))+ unless (values == [Only (Variant (FieldInt64 42))]) (fail "wrong VARIANT payload before the drop")+ replicateM_ before performMajorGC+ stmt <- openStatement conn "SELECT dropped_token()"+ replicateM_ after performMajorGC+ row <- nextRow stmt+ unless (row == Just (Only (42 :: Int64))) (fail "wrong row before the drop")+ pure weak+ released weak attempts+ | attempts == (0 :: Int) = fail "a dropped connection kept its database open"+ | otherwise = do+ performMajorGC+ token <- deRefWeak weak+ when (isJust token) (threadDelay 10000 >> released weak (attempts - 1))+ flip finally (setFinalizerExceptionHandler previous) do+ setFinalizerExceptionHandler record+ forM_ [(before, after) | before <- [0 .. 3 :: Int], after <- [0, 1 :: Int]] \(before, after) ->+ allocate before after >>= \weak -> released weak 500+ count <- readIORef blocked+ unless (count == 0) (fail ("a finalizer waited for a later finalizer " <> show count <> " time(s)"))+ -- | Import raw WKB and release native values after failed nested construction. checkNewTypes :: Int -> IO () checkNewTypes batches =- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- [Only geometry] <- query_ conn "SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]- typed <- either fail pure (fromRawGeometry geometry) :: IO G.Geometry- let variant = Variant (FieldList [FieldInt64 42, FieldList [FieldNull, FieldText "before\0after"]])- batch = forM_ [1 .. 100 :: Int] \_ -> do- rows <- query conn "SELECT ?, system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?), ?" (variant, rawGeometryWKB geometry, rawGeometryCRS geometry, typed)- unless (rows == [(variant, geometry, typed)]) (fail "new type round trip failed")- expectFailure (query conn "SELECT ?, system.main.ST_GeomFromWKB(?)" (typed, BS.pack [1, 1, 0, 0, 0]) :: IO [(G.Geometry, RawGeometry)])- expectFailure (query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (rawGeometryWKB geometry)) :: IO [Only Int64])- expectFailure (query conn "SELECT ?" (Only (Variant (FieldList [variantPayload variant, FieldDecimal (DecimalValue 0 0 0)]))) :: IO [Only Variant])- expectFailure (query conn "SELECT ?" (Only variant) :: IO [Only Int64])- batch- performMajorGC- before <- readUsage- forM_ [1 .. batches] \n -> do- batch- performMajorGC- after <- readUsage- reportOptional ("open connection, " <> show (n * 100) <> " new type cycles") before after+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ [Only geometry] <- query_ conn "SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ typed <- either fail pure (fromRawGeometry geometry) :: IO G.Geometry+ -- These values come from DuckDB, so that they have the native types.+ [Only collated] <- query_ conn "SELECT {'text': 'A' COLLATE nocase, 'items': ['B' COLLATE nocase]}" :: IO [Only (StructValue FieldValue)]+ [Only union] <- query_ conn "SELECT union_value(t := 'λ')::UNION(n BIGINT, t VARCHAR)" :: IO [Only (UnionValue FieldValue)]+ let emptyKey = Variant (variantObject [("", FieldWord64 1), ("n", FieldWord64 2)])+ foldedKeys = Variant (variantObject [("A", FieldWord64 1), ("a", FieldWord64 2), ("b", FieldWord64 3)])+ -- The encoded value of this number has more than 100 bytes.+ large = 2 ^ (1000 :: Int) + 12345 :: Integer+ roundTrip :: (ToField a, FromField a, Eq a) => String -> a -> IO ()+ roundTrip label value = query conn "SELECT ?" (Only value) >>= \rows -> unless (rows == [Only value]) (fail (label <> " binding failed"))+ let variant = Variant (FieldList [FieldInt64 42, FieldList [FieldNull, FieldText "before\0after"]])+ custom = geometry{rawGeometryCRS = Just "local' íslenska λ"}+ batch = forM_ [1 .. 100 :: Int] \_ -> do+ roundTrip "custom CRS" custom+ expectFailure (query conn "SELECT ?" (Only geometry{rawGeometryCRS = Just " "}) :: IO [Only RawGeometry])+ rows <-+ query+ conn+ "SELECT ?, system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?), ?"+ (variant, rawGeometryWKB geometry, rawGeometryCRS geometry, typed)+ unless (rows == [(variant, geometry, typed)]) (fail "new type round trip failed")+ expectFailure (query conn "SELECT ?, system.main.ST_GeomFromWKB(?)" (typed, BS.pack [1, 1, 0, 0, 0]) :: IO [(G.Geometry, RawGeometry)])+ expectFailure (query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (rawGeometryWKB geometry)) :: IO [Only Int64])+ expectFailure+ (query conn "SELECT ?" (Only (Variant (FieldList [variantPayload variant, FieldDecimal (DecimalValue 0 0 0)]))) :: IO [Only Variant])+ expectFailure (query conn "SELECT ?" (Only variant) :: IO [Only Int64])+ -- STRUCT field names cannot be an empty key or two keys that differ+ -- only in case. Thus these objects use the extraction query. The two+ -- key counts replace the kept extraction statement.+ objects <- query_ conn "SELECT json_object('', 1, 'n', 2)::VARIANT, json_object('A', 1, 'a', 2, 'b', 3)::VARIANT"+ unless (objects == [(emptyKey, foldedKeys)]) (fail "VARIANT object read failed")+ roundTrip "collated STRUCT" collated+ roundTrip "UNION" union+ roundTrip "MAP" (Map.fromList [(1 :: Int64, [10, 11 :: Int64]), (2, [])])+ roundTrip "LIST" [[1, 2], [3 :: Int64]]+ query conn "SELECT ?::VARCHAR" (Only (BigNum large)) >>= \text ->+ unless (text == [Only (show large)]) (fail "BIGNUM binding failed")+ batch+ performMajorGC+ before <- readUsage+ forM_ [1 .. batches] \n -> do+ batch+ performMajorGC+ after <- readUsage+ reportOptional ("open connection, " <> show (n * 100) <> " new type cycles") before after -- | Check native results and database handles across connection lifetimes. checkOpenClose :: IO () checkOpenClose = do- -- The first cycle also does the one-time initialization, which must not- -- count as growth.- openCloseCycle- before <- readUsage- case before of- Nothing -> putStrLn "duckdb-simple leak check: /proc/self/status is unavailable; skipped."- Just baseline -> do- forM_ [1 .. cycles] \_ -> openCloseCycle- after <- readUsage- case after of- Nothing -> putStrLn "duckdb-simple leak check: /proc/self/status disappeared; skipped."- Just final -> report ("open/close, " <> show cycles <> " cycles") baseline final+ -- The first cycle also does the one-time initialization, which must not+ -- count as growth.+ openCloseCycle+ before <- readUsage+ case before of+ Nothing -> putStrLn "duckdb-simple leak check: /proc/self/status is unavailable; skipped."+ Just baseline -> do+ forM_ [1 .. cycles] \_ -> openCloseCycle+ after <- readUsage+ case after of+ Nothing -> putStrLn "duckdb-simple leak check: /proc/self/status disappeared; skipped."+ Just final -> report ("open/close, " <> show cycles <> " cycles") baseline final -- | Reuse one connection and statement through successful and failed operations. checkLongLived :: Int -> IO () checkLongLived batches =- withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- withStatement conn "SELECT ?::BIGINT FROM range(10000)" \stmt -> do- let cycleQuery = do- rows <- query_ conn "SELECT sum(i)::BIGINT FROM range(10000) t(i)"- unless (rows == [Only (49995000 :: Int64)]) (fail "wrong query result")- expectFailure (query_ conn "SELECT {'xs': [i, i + 1, i + 2, i + 3], 'bad': '12:00:00+01:23:45'::TIMETZ} FROM range(10000) t(i)" :: IO [Only FieldValue])- expectFailure (query_ conn "SELECT CAST('bad' AS BIGINT)" :: IO [Only Int64])- bind stmt [toField (42 :: Int64)]- nextRow stmt >>= \row -> unless (row == Just (Only (42 :: Int64))) (fail "wrong cursor result")- clearStatementBindings stmt- expectFailure (nextRow stmt :: IO (Maybe (Only Int64)))- batch = forM_ [1 .. 100 :: Int] (const cycleQuery)- batch- performMajorGC- before <- readUsage- forM_ [1 .. batches] \n -> do- batch- performMajorGC- after <- readUsage- reportOptional ("open connection, " <> show (n * 100) <> " cycles") before after+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ breakVariantReads conn+ withStatement conn "SELECT ?::BIGINT FROM range(10000)" \stmt -> do+ let cycleQuery = do+ rows <- query_ conn "SELECT sum(i)::BIGINT FROM range(10000) t(i)"+ unless (rows == [Only (49995000 :: Int64)]) (fail "wrong query result")+ expectFailure+ (query_ conn "SELECT {'xs': [i, i + 1, i + 2, i + 3], 'bad': 42::BIGINT::VARIANT} FROM range(10000) t(i)" :: IO [Only FieldValue])+ expectFailure (query_ conn "SELECT CAST('bad' AS BIGINT)" :: IO [Only Int64])+ bind stmt [toField (42 :: Int64)]+ nextRow stmt >>= \row -> unless (row == Just (Only (42 :: Int64))) (fail "wrong cursor result")+ clearStatementBindings stmt+ expectFailure (nextRow stmt :: IO (Maybe (Only Int64)))+ -- Close a statement while its cursor holds the other chunks of a+ -- materialized result, and then while its cursor holds a native stream.+ withStatement conn "SELECT i FROM range(100000) t(i)" \cursor -> void (nextRow cursor :: IO (Maybe (Only Int64)))+ withStatement conn "SELECT i FROM range(100000) t(i)" \cursor -> void (Streaming.nextRow cursor :: IO (Maybe (Only Int64)))+ -- The step fails at the first batch of a materialized Arrow fold.+ -- The fold must release each batch that it did not deliver.+ expectFailure (Arrow.foldArrow_ conn "SELECT i FROM range(100000) t(i)" () (\() _ _ -> fail "step"))+ batch = forM_ [1 .. 100 :: Int] (const cycleQuery)+ batch+ performMajorGC+ before <- readUsage+ forM_ [1 .. batches] \n -> do+ batch+ performMajorGC+ after <- readUsage+ reportOptional ("open connection, " <> show (n * 100) <> " cycles") before after -- | Isolate failed result cleanup for comparison with earlier package versions. checkDecodeFailure :: IO () checkDecodeFailure =- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- let rejected = expectFailure (query_ conn "SELECT {'xs': [i, i + 1, i + 2, i + 3], 'bad': '12:00:00+01:23:45'::TIMETZ} FROM range(100000) t(i)" :: IO [Only FieldValue])- rejected- performMajorGC- before <- readUsage- forM_ [1 .. cycles] (const rejected)- performMajorGC- after <- readUsage- reportOptional ("open connection, " <> show cycles <> " decode failures") before after+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ breakVariantReads conn+ let rejected = expectFailure (query_ conn rejectedQuery :: IO [Only FieldValue])+ rejected+ performMajorGC+ before <- readUsage+ forM_ [1 .. cycles] (const rejected)+ performMajorGC+ after <- readUsage+ reportOptional ("open connection, " <> show cycles <> " failed VARIANT reads") before after -- | Check that callback closures and query state are released before close. checkCallbacks :: Int -> IO () checkCallbacks batches =- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- Logging.registerLogStorage conn "leak_log" (\_ -> pure ())- copyStates <- newIORef []- let newCopyState = do+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ copyStates <- newIORef []+ let newCopyState = do+ state <- newIORef (7 :: Int64)+ weak <- mkWeakIORef state (pure ())+ atomicModifyIORef' copyStates (\old -> (weak : old, ()))+ pure state+ Copy.registerCopyToFunction conn "leak_copy" (\_ -> newCopyState) (\_ -> newCopyState) (\_ _ -> pure ()) (\_ -> pure ())+ Copy.registerCopyToFunction+ conn+ "leak_copy_failure"+ (\_ -> newCopyState)+ (\_ -> fail "COPY init failure" :: IO ())+ (\_ _ -> pure ())+ (\_ -> pure ())+ -- Register leak_log once, so that the registration in each cycle fails and+ -- the test can check that the rejected callback is released.+ Logging.registerLogStorage conn "leak_log" (\_ -> pure ())+ let cycleCallback = do+ retained <- newIORef (42 :: Int64)+ weak <- mkWeakIORef retained (pure ())+ createFunction conn "leak_scalar" (readIORef retained)+ query_ conn "SELECT leak_scalar()" >>= \rows -> unless (rows == [Only (42 :: Int64)]) (fail "wrong callback result")+ states <- newIORef []+ createFunctionWithState+ conn+ "leak_state"+ ( do state <- newIORef (7 :: Int64)- weak <- mkWeakIORef state (pure ())- atomicModifyIORef' copyStates (\old -> (weak : old, ()))+ stateWeak <- mkWeakIORef state (pure ())+ -- The initializer returns each weak reference to the test.+ atomicModifyIORef' states (\old -> (stateWeak : old, ())) pure state- Copy.registerCopyToFunction conn "leak_copy" (\_ -> newCopyState) (\_ -> newCopyState) (\_ _ -> pure ()) (\_ -> pure ())- Copy.registerCopyToFunction conn "leak_copy_failure" (\_ -> newCopyState) (\_ -> fail "COPY init failure" :: IO ()) (\_ _ -> pure ()) (\_ -> pure ())- let cycleCallback = do- retained <- newIORef (42 :: Int64)- weak <- mkWeakIORef retained (pure ())- createFunction conn "leak_scalar" (readIORef retained)- query_ conn "SELECT leak_scalar()" >>= \rows -> unless (rows == [Only (42 :: Int64)]) (fail "wrong callback result")- states <- newIORef []- createFunctionWithState- conn- "leak_state"- ( do- state <- newIORef (7 :: Int64)- stateWeak <- mkWeakIORef state (pure ())- -- The initializer returns each weak reference to the test.- atomicModifyIORef' states (\old -> (stateWeak : old, ()))- pure state- )- readIORef- query_ conn "SELECT leak_state()" >>= \rows -> unless (rows == [Only (7 :: Int64)]) (fail "wrong state result")- rejectedLog <- newIORef (0 :: Int64)- logWeak <- mkWeakIORef rejectedLog (pure ())- expectFailure (Logging.registerLogStorage conn "leak_log" (\_ -> void (readIORef rejectedLog)))- void (execute_ conn "COPY (SELECT i FROM range(100) t(i)) TO 'unused' (FORMAT leak_copy)")- expectFailure (execute_ conn "COPY (SELECT 1) TO 'unused' (FORMAT leak_copy_failure)")- copyRefs <- atomicModifyIORef' copyStates (\refs -> ([], refs))- stateRefs <- readIORef states- pure (weak : logWeak : stateRefs <> copyRefs)- batch = concat <$> replicateM 100 cycleCallback- checkReleased refs = do- -- Replace the last closure, then run a new transaction.- createFunction conn "leak_scalar" (0 :: Int64)- createFunction conn "leak_state" (0 :: Int64)- void (query_ conn "SELECT 1" :: IO [Only Int64])- performMajorGC- performMajorGC- live <- length . catMaybes <$> mapM deRefWeak refs- unless (live == 0) (fail (show live <> " callback values remain live after replacement"))- batch >>= checkReleased- before <- readUsage- forM_ [1 .. batches] \n -> do- batch >>= checkReleased- after <- readUsage- reportOptional ("open connection, " <> show (n * 100) <> " callback cycles") before after+ )+ readIORef+ query_ conn "SELECT leak_state()" >>= \rows -> unless (rows == [Only (7 :: Int64)]) (fail "wrong state result")+ rejectedLog <- newIORef (0 :: Int64)+ logWeak <- mkWeakIORef rejectedLog (pure ())+ expectFailure (Logging.registerLogStorage conn "leak_log" (\_ -> void (readIORef rejectedLog)))+ void (execute_ conn "COPY (SELECT i FROM range(100) t(i)) TO 'unused' (FORMAT leak_copy)")+ expectFailure (execute_ conn "COPY (SELECT 1) TO 'unused' (FORMAT leak_copy_failure)")+ copyRefs <- atomicModifyIORef' copyStates (\refs -> ([], refs))+ stateRefs <- readIORef states+ pure (weak : logWeak : stateRefs <> copyRefs)+ batch = concat <$> replicateM 100 cycleCallback+ checkReleased refs = do+ -- Replace the last closure, then run a new transaction.+ createFunction conn "leak_scalar" (0 :: Int64)+ createFunction conn "leak_state" (0 :: Int64)+ void (query_ conn "SELECT 1" :: IO [Only Int64])+ performMajorGC+ performMajorGC+ live <- length . catMaybes <$> mapM deRefWeak refs+ unless (live == 0) (fail (show live <> " callback values remain live after replacement"))+ batch >>= checkReleased+ before <- readUsage+ forM_ [1 .. batches] \n -> do+ batch >>= checkReleased+ after <- readUsage+ reportOptional ("open connection, " <> show (n * 100) <> " callback cycles") before after +-- | Make VARIANT metadata preparation and object extraction fail. A prepared+-- metadata plan can survive a lower expression depth. 'rejectedQuery' uses an+-- unusual key, so its extraction query still binds on each execution and fails+-- at depth 3. Set the caller's SESSION depth first, so its control statements+-- can still bind after the GLOBAL depth changes.+breakVariantReads :: Connection -> IO ()+breakVariantReads conn = do+ void (execute_ conn "SET SESSION max_expression_depth = 1000")+ void (execute_ conn "SET GLOBAL max_expression_depth = 3")++-- | A query whose result fails in the VARIANT reader after 'breakVariantReads'.+rejectedQuery :: Query+rejectedQuery = "SELECT {'xs': [i, i + 1, i + 2, i + 3], 'bad': json_object('', i)::VARIANT} FROM range(100000) t(i)"+ -- | Require an error without retaining the exception or its native resources. expectFailure :: IO a -> IO () expectFailure action = do- outcome <- try (void action) :: IO (Either SomeException ())- case outcome of- Left _ -> pure ()- Right () -> fail "expected operation to fail"+ outcome <- try (void action) :: IO (Either SomeException ())+ case outcome of+ Left _ -> pure ()+ Right () -> fail "expected operation to fail" -- | Reuse a connection after cancelling native execution and row decoding.+--+-- Each batch reports its resident memory and fails if the thread count grew+-- too much. The rule for the resident memory comes after the last batch. The+-- resident memory rises in the first batches, and later single samples go up+-- and down. A leak never returns memory, so it raises each later sample, also+-- the lowest sample. A single high sample does not raise the lowest sample.+-- Thus the rule compares the lowest sample of the second half of the batches+-- with the lowest sample of batch 3 to the half. Batches 1 and 2 are not in the+-- rule, because the memory can still rise in them. The rule needs at least 6+-- batches. With fewer batches, the check does not compare the memory. checkCancellation :: Int -> IO () checkCancellation batches =- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- void (execute_ conn "SET streaming_buffer_size = '64KB'")- let cancel action = do- started <- newEmptyMVar- done <- newEmptyMVar- let signal = void (tryPutMVar started ())- tid <- forkFinally (action signal) (\outcome -> putMVar done outcome >> signal)- takeMVar started- threadDelay 1000- killThread tid- outcome <- takeMVar done- case outcome of- Left err | fromException err == Just ThreadKilled -> pure ()- Left err -> fail ("unexpected cancellation error: " <> show err)- Right () -> fail "query completed before cancellation"- rows <- query_ conn "SELECT 42"- unless (rows == [Only (42 :: Int64)]) (fail "connection failed after cancellation")- batch = forM_ [1 .. 10 :: Int] \_ -> do- cancel \signal -> do- createFunction conn "leak_cancel_started" (signal >> pure (1 :: Int64))- void- ( query_- conn- "WITH started AS MATERIALIZED (SELECT leak_cancel_started() AS seed) \- \SELECT sum(sin((a.i + b.j + started.seed)::DOUBLE)) \- \FROM started, range(1000000) a(i), range(1000000) b(j)" ::- IO [Only Double]- )- cancel \signal -> do- signal- void (query_ conn "SELECT {'x': i, 'values': [i, i + 1]} FROM range(100000) t(i)" :: IO [Only FieldValue])- cancel \signal -> do- blocked <- newEmptyMVar- -- Keep the first chunk live until the worker is cancelled.- void (fold_ conn "SELECT {'x': i, 'values': [i, NULL]} FROM range(100000) t(i)" (0 :: Int64) (\n (Only (_ :: FieldValue)) -> signal >> takeMVar blocked >> pure (n + 1)))- forM_ [False, True] \arrow -> cancel \signal -> do- filtering <- newIORef False- createFunction conn "leak_stream_filter" \(_ :: Int64) -> do- active <- readIORef filtering- when active signal- pure active- let sql = "SELECT i FROM range(1000000000000) t(i) WHERE NOT leak_stream_filter(i)"- delivered = atomicWriteIORef filtering True- if arrow- then Streaming.foldArrow_ conn sql () (\() _ _ -> delivered)- else Streaming.fold_ conn sql () (\() (Only (_ :: Int64)) -> delivered)- cancel \signal -> do- signal- void (query_ conn "SELECT {'x': i, 'values': [i, i + 1]}::VARIANT FROM range(100000) t(i)" :: IO [Only Variant])- cancel \signal -> do- blocked <- newEmptyMVar- void (fold_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84') FROM range(100000)" (0 :: Int64) (\n (Only (_ :: RawGeometry)) -> signal >> takeMVar blocked >> pure (n + 1)))- batch- performMajorGC- before <- readUsage- forM_ [1 .. batches] \n -> do- batch- performMajorGC- after <- readUsage- reportOptional ("open connection, " <> show (n * 70) <> " cancellations") before after+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ void (execute_ conn "SET streaming_buffer_size = '64KB'")+ let run mayFinish action = do+ started <- newEmptyMVar+ done <- newEmptyMVar+ let signal = void (tryPutMVar started ())+ tid <- forkFinally (action signal) (\outcome -> putMVar done outcome >> signal)+ takeMVar started+ threadDelay 1000+ killThread tid+ outcome <- takeMVar done+ case outcome of+ Left err | fromException err == Just ThreadKilled -> pure ()+ Left err -> fail ("unexpected cancellation error: " <> show err)+ Right ()+ | mayFinish -> pure ()+ | otherwise -> fail "query completed before cancellation"+ rows <- query_ conn "SELECT 42"+ unless (rows == [Only (42 :: Int64)]) (fail "connection failed after cancellation")+ -- These queries block or run until the thread is killed.+ cancel = run False+ -- These finite queries can finish before the interrupt reaches them.+ interrupt = run True+ batch = forM_ [1 .. 10 :: Int] \_ -> do+ cancel \signal -> do+ createFunction conn "leak_cancel_started" (signal >> pure (1 :: Int64))+ void+ ( query_+ conn+ "WITH started AS MATERIALIZED (SELECT leak_cancel_started() AS seed) \+ \SELECT sum(sin((a.i + b.j + started.seed)::DOUBLE)) \+ \FROM started, range(1000000) a(i), range(1000000) b(j)"+ :: IO [Only Double]+ )+ interrupt \signal -> do+ signal+ void (query_ conn "SELECT {'x': i, 'values': [i, i + 1]} FROM range(100000) t(i)" :: IO [Only FieldValue])+ cancel \signal -> do+ blocked <- newEmptyMVar+ -- Keep the first chunk live until the worker is cancelled.+ void+ ( fold_+ conn+ "SELECT {'x': i, 'values': [i, NULL]} FROM range(100000) t(i)"+ (0 :: Int64)+ (\n (Only (_ :: FieldValue)) -> signal >> takeMVar blocked >> pure (n + 1))+ )+ forM_ [False, True] \arrow -> cancel \signal -> do+ filtering <- newIORef False+ createFunction conn "leak_stream_filter" \(_ :: Int64) -> do+ active <- readIORef filtering+ when active signal+ pure active+ let sql = "SELECT i FROM range(1000000000000) t(i) WHERE NOT leak_stream_filter(i)"+ delivered = atomicWriteIORef filtering True+ if arrow+ then Streaming.foldArrow_ conn sql () (\() _ _ -> delivered)+ else Streaming.fold_ conn sql () (\() (Only (_ :: Int64)) -> delivered)+ interrupt \signal -> do+ signal+ void (query_ conn "SELECT {'x': i, 'values': [i, i + 1]}::VARIANT FROM range(100000) t(i)" :: IO [Only FieldValue])+ cancel \signal -> do+ blocked <- newEmptyMVar+ void+ ( fold_+ conn+ "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84') FROM range(100000)"+ (0 :: Int64)+ (\n (Only (_ :: RawGeometry)) -> signal >> takeMVar blocked >> pure (n + 1))+ )+ batch+ performMajorGC+ before <- readUsage+ samples <- forM [1 .. batches] \n -> do+ batch+ performMajorGC+ after <- readUsage+ let label = "open connection, " <> show (n * 70) <> " cancellation attempts"+ case (before, after) of+ (Just first, Just current) -> reportWith False label first current+ _ -> reportOptional label before after+ pure after+ case sequence samples of+ Just usages | batches >= 6 -> do+ let half = batches `div` 2+ (early, late) = splitAt half usages+ lowest = minimumBy (comparing usageRssKb)+ label = "open connection, lowest rss of cancellation batches 3 to " <> show half <> " -> batches " <> show (half + 1) <> " to " <> show batches+ report label (lowest (drop 2 early)) (lowest late)+ _ -> pure () -- | Report native counters when the operating system provides them. reportOptional :: String -> Maybe Usage -> Maybe Usage -> IO ()@@ -323,29 +484,85 @@ -- | Print both measurements and fail if either one grew too much. report :: String -> Usage -> Usage -> IO ()-report label before after = do- checkRss <- (/= Just "0") <$> lookupEnv "DUCKDB_LEAK_RSS_CHECK"- putStrLn label- putStrLn $- "threads: "- <> show (usageThreads before)- <> " -> "- <> show (usageThreads after)- <> " (slack "- <> show threadSlack- <> ")"- putStrLn $- "rss kB: "- <> show (usageRssKb before)- <> " -> "- <> show (usageRssKb after)- <> " (slack "- <> show rssSlackKb- <> ")"- unless checkRss (putStrLn "RSS assertion disabled for native memory instrumentation")- when (threadGrowth > threadSlack || (checkRss && rssGrowth > rssSlackKb)) do- putStrLn ("FAIL: " <> label <> " retains DuckDB resources")- exitFailure+report = reportWith True++-- | Print both measurements. Fail if the thread count grew too much. If the+-- first argument is 'True', also fail if the resident memory grew too much.+reportWith :: Bool -> String -> Usage -> Usage -> IO ()+reportWith rssRule label before after = do+ enabled <- (/= Just "0") <$> lookupEnv "DUCKDB_LEAK_RSS_CHECK"+ let checkRss = enabled && rssRule+ putStrLn label+ putStrLn $+ "threads: "+ <> show (usageThreads before)+ <> " -> "+ <> show (usageThreads after)+ <> " (slack "+ <> show threadSlack+ <> ")"+ putStrLn $+ "rss kB: "+ <> show (usageRssKb before)+ <> " -> "+ <> show (usageRssKb after)+ <> " (slack "+ <> show rssSlackKb+ <> ")"+ unless enabled (putStrLn "RSS assertion disabled for native memory instrumentation")+ when (threadGrowth > threadSlack || (checkRss && rssGrowth > rssSlackKb)) do+ putStrLn ("FAIL: " <> label <> " retains DuckDB resources")+ exitFailure where threadGrowth = usageThreads after - usageThreads before rssGrowth = usageRssKb after - usageRssKb before++-- | Run the typed registrations through the public API on one connection: a+-- scalar function with VARCHAR and BIGINT results over 5000 rows, a grouped+-- aggregate with VARCHAR results, a typed table function, and a cast that+-- fails.+-- Each VARCHAR value is longer than 12 bytes, so DuckDB stores it outside the+-- string slot.+checkTyped :: Int -> IO ()+checkTyped batches =+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ let label prefix n = Text.pack (prefix <> "-" <> show n <> "-with-a-long-suffix")+ numbers = [0 .. 4999] :: [Int64]+ labelLength prefix = sum (map (fromIntegral . Text.length . label prefix) numbers) :: Int64+ groups = [(group, label "total" (sum [n | n <- numbers, n `mod` 50 == group])) | group <- [0 .. 49]]+ createFunction conn "leak_text" (label "row" :: Int64 -> Text)+ createFunction conn "leak_double" ((* 2) :: Int64 -> Int64)+ createAggregate+ conn+ "leak_total"+ Aggregate+ { aggregateInitial = 0 :: Int64+ , aggregateStep = \total (Only n) -> pure (total + n)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure . label "total"+ }+ createTableFunction conn "leak_rows" ["n", "label"] \(Only count) ->+ pure [(n, label "row" n) | n <- [0 .. count - 1 :: Int64]]+ registerCustomType conn "leak_tag" (TypeExpression "VARCHAR")+ -- Each cycle fails the cast, so that the error path of the callback runs.+ R.registerCast conn (TypeExpression "BIGINT") (TypeExpression "leak_tag") Nothing \_ _ -> throwIO (userError "leak cast refused the value")+ let cycleTyped = do+ scalar <- query_ conn "SELECT sum(length(leak_text(i)))::BIGINT, sum(leak_double(i))::BIGINT FROM range(5000) t(i)"+ unless (scalar == [(labelLength "row", 2 * sum numbers)]) (fail "wrong typed scalar result")+ grouped <- query_ conn "SELECT i % 50 AS g, leak_total(i) FROM range(5000) t(i) GROUP BY g ORDER BY g"+ unless (grouped == groups) (fail "wrong typed aggregate result")+ table <- query_ conn "SELECT count(*), sum(n)::BIGINT, sum(length(label))::BIGINT FROM leak_rows(5000)"+ unless (table == [(5000 :: Int64, sum numbers, labelLength "row")]) (fail "wrong typed table function result")+ cast <- try (query_ conn "SELECT CAST(i AS leak_tag) FROM range(5000) t(i)" :: IO [Only Text])+ case cast of+ Left (err :: SQLError) | sqlErrorType err == Just ErrorTypeMismatch -> pure ()+ _ -> fail "wrong cast result"+ batch = forM_ [1 .. 20 :: Int] (const cycleTyped)+ batch+ performMajorGC+ before <- readUsage+ forM_ [1 .. batches] \n -> do+ batch+ performMajorGC+ after <- readUsage+ reportOptional ("open connection, " <> show (n * 20) <> " typed registration cycles") before after
+ nonthreaded-test/Main.hs view
@@ -0,0 +1,50 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}++-- | Check the guard for a program that is not linked with @-threaded@. This+-- suite has no @-threaded@ in its @ghc-options@. See Note [Callbacks without+-- the threaded runtime] in "Database.DuckDB.Simple.Callback".+module Main (main) where++import Control.Exception (try)+import Control.Monad (unless, void)+import Data.IORef (modifyIORef', newIORef, readIORef)+import Data.Int (Int64)+import qualified Data.Text as Text+import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Copy as Copy+import qualified Database.DuckDB.Simple.Logging as Logging+import qualified Database.DuckDB.Simple.V2.Copy as V2Copy+import System.Exit (exitFailure)++main :: IO ()+main = do+ -- With the setting threads above 1, the registration raises the input error.+ withConnection ":memory:" \conn -> do+ void (execute_ conn "SET threads = 4")+ result <- try (createFunction conn "plus_one" ((+ 1) :: Int64 -> Int64))+ case result of+ Left SQLError{sqlErrorType = Just ErrorInputInvalid, sqlErrorMessage}+ | "not linked with -threaded" `Text.isInfixOf` sqlErrorMessage -> pure ()+ other -> failWith ("threads = 4: expected the error of the guard, got " <> show other)+ -- With the setting threads at 1, a query of a registered function runs.+ withConnection ":memory:" \conn -> do+ void (execute_ conn "SET threads = 1")+ createFunction conn "plus_one" ((+ 1) :: Int64 -> Int64)+ rows <- query_ conn "SELECT sum(plus_one(i))::BIGINT FROM range(100000) t(i)" :: IO [Only Int64]+ unless (rows == [Only 5000050000]) (failWith ("threads = 1: expected 5000050000, got " <> show rows))+ checkGuard "COPY" $ Copy.registerCopyFromFunction conn "lines" (\_ -> pure [Only (1 :: Int64)])+ let noop _ _ = pure ()+ checkGuard "COPY" $ V2Copy.registerCopyToCallbacks conn "raw_lines" (V2Copy.copyToCallbacks noop noop)+ flushCount <- newIORef (0 :: Int)+ checkGuard "log sink" $ Logging.registerLogSink conn "events" (const (pure ())) (modifyIORef' flushCount (+ 1))+ flushed <- readIORef flushCount+ unless (flushed == 1) (failWith ("rejected log sink: expected one flush, got " <> show flushed))+ where+ checkGuard kind action =+ try action >>= \result -> case result of+ Left SQLError{sqlErrorType = Just ErrorInputInvalid, sqlErrorMessage}+ | (kind <> " callbacks require a program linked with -threaded") `Text.isInfixOf` sqlErrorMessage -> pure ()+ other -> failWith ("threads = 1: expected the " <> Text.unpack kind <> " guard, got " <> show (other :: Either SQLError ()))+ failWith message = putStrLn message >> exitFailure
src/Database/DuckDB/Simple.hs view
@@ -3,701 +3,646 @@ {-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE TupleSections #-} -{- |-Module : Database.DuckDB.Simple-Description : High-level DuckDB API in the duckdb-simple style.--The API mirrors the ergonomics of @sqlite-simple@ while being backed by the-DuckDB C API. It supports connection management, parameter binding, execution,-and typed result decoding. See @README.md@ for usage examples.--}-module Database.DuckDB.Simple (- -- * Connections- Connection,- open,- openWithConfig,- ConnectionOptions (..),- defaultConnectionOptions,- openWithOptions,- close,- withConnection,- withConnectionWithConfig,- withConnectionWithOptions,-- -- * Queries and statements- Query (..),- Statement,- openStatement,- closeStatement,- withStatement,- clearStatementBindings,- namedParameterIndex,- columnCount,- columnName,- executeStatement,- execute,- executeMany,- execute_,- bind,- bindNamed,- executeNamed,- queryNamed,- fold,- fold_,- foldNamed,- withTransaction,- query,- queryWith,- query_,- queryWith_,- nextRow,- nextRowWith,-- -- * Errors and conversions- SQLError (..),- FormatError (..),- ResultError (..),- FieldParser,- FromField (..),- FromRow (..),- RowParser,- field,- fieldWith,- numFieldsRemaining,- -- Re-export parameter helper types.- ToField (..),- ToRow (..),- FieldBinding,- NamedParam (..),- DuckDBColumnType (..),- duckdbColumnType,- Null (..),- Only (..),- (:.) (..),-- -- * User-defined scalar functions- Function (..),- createFunction,- createFunctionWithState,- deleteFunction,-) where--import Control.Exception (SomeException, bracket, finally, mask, mask_, onException, throwIO, try)-import Control.Monad (forM, forM_, join, void, when, zipWithM_)-import Data.IORef (atomicModifyIORef', mkWeakIORef, newIORef)-import Data.Maybe (isJust, isNothing)-import qualified Data.Set as Set-import Data.Text (Text)-import qualified Data.Text as Text-import qualified Data.Text.Foreign as TextForeign-import Database.DuckDB.FFI-import Database.DuckDB.Simple.FromField (- Field (..),- FieldParser,- FromField (..),- ResultError (..),- )-import Database.DuckDB.Simple.FromRow (- FromRow (..),- RowParser,- field,- fieldWith,- numFieldsRemaining,- parseRow,- rowErrorsToSqlError,- )-import Database.DuckDB.Simple.Function (Function (..), createFunction, createFunctionWithState, deleteFunction)-import Database.DuckDB.Simple.Internal (- Connection (..),- ConnectionState (..),- Query (..),- ResultMode (..),- SQLError (..),- Statement (..),- StatementState (..),- StatementStreamState (..),- fetchPrepareError,- keepAlive,- peekUtf8CString,- runInterruptibleQuery,- withConnectionHandle,- withQueryCString,- withResult,- withStatementHandle,- )-import Database.DuckDB.Simple.Ok (Ok (..))-import Database.DuckDB.Simple.Result (cleanupStatementStreamRef, collectRows, resetStatementStream)-import qualified Database.DuckDB.Simple.Result as Result-import Database.DuckDB.Simple.ToField (DuckDBColumnType (..), FieldBinding, NamedParam (..), ToField (..), bindFieldBinding, duckdbColumnType, renderFieldBinding)-import Database.DuckDB.Simple.ToRow (ToRow (..))-import Database.DuckDB.Simple.TypeCache (TypeCache, createTypeCache, defaultGeometryCRS, destroyTypeCache)-import Database.DuckDB.Simple.Types (FormatError (..), Null (..), Only (..), (:.) (..))-import Foreign.C.String (CString)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (Ptr, castPtr, nullPtr)-import Foreign.Storable (peek, poke)---- | Open a DuckDB database located at the supplied path.-open :: FilePath -> IO Connection-open path = openWithConfig path []---- | Open a DuckDB database with configuration flags applied before startup.-openWithConfig :: FilePath -> [(Text, Text)] -> IO Connection-openWithConfig path settings = openWithOptions path defaultConnectionOptions{connectionConfig = settings}---- | Settings that duckdb-simple applies when it opens a connection.-data ConnectionOptions = ConnectionOptions- { connectionConfig :: [(Text, Text)]- -- ^ DuckDB configuration flags, as for 'openWithConfig'.- , connectionGeometryCRS :: [Text]- {- ^ CRS definitions. The connection reads a GEOMETRY type for each CRS- the first time a parameter needs a VARIANT type or a GEOMETRY type with a- CRS. Parameters can then contain GEOMETRY types with these CRSs.- -}- }- deriving (Eq, Show)---- | No configuration flags, and the CRS @OGC:CRS84@.-defaultConnectionOptions :: ConnectionOptions-defaultConnectionOptions =- ConnectionOptions- { connectionConfig = []- , connectionGeometryCRS = defaultGeometryCRS- }--{- | Open a DuckDB database with options. The connection reads the VARIANT type-and the GEOMETRY types for the configured CRSs the first time a parameter-needs one of them. It reads them with one query on a separate connection.--}-openWithOptions :: FilePath -> ConnectionOptions -> IO Connection-openWithOptions path ConnectionOptions{connectionConfig, connectionGeometryCRS} =- mask_ do- db <- openDatabaseWithConfig path connectionConfig- conn <-- connectDatabase db- `onException` closeDatabaseHandle db- let closeBoth = closeConnectionHandle conn >> closeDatabaseHandle db- cache <-- createTypeCache db connectionGeometryCRS- `onException` closeBoth- createConnection db conn cache- `onException` (destroyTypeCache cache >> closeBoth)---- | Close a connection. The operation is idempotent.-close :: Connection -> IO ()-close Connection{connectionState} =- mask_ $- join $- atomicModifyIORef' connectionState \case- ConnectionClosed -> (ConnectionClosed, pure ())- openState@(ConnectionOpen{}) ->- (ConnectionClosed, closeHandles openState)---- | Run an action with a freshly opened connection, closing it afterwards.-withConnection :: FilePath -> (Connection -> IO a) -> IO a-withConnection path = bracket (open path) close---- | Run an action with a freshly opened configured connection, closing it afterwards.-withConnectionWithConfig :: FilePath -> [(Text, Text)] -> (Connection -> IO a) -> IO a-withConnectionWithConfig path settings = bracket (openWithConfig path settings) close---- | Run an action with a connection opened with options, closing it afterwards.-withConnectionWithOptions :: FilePath -> ConnectionOptions -> (Connection -> IO a) -> IO a-withConnectionWithOptions path options = bracket (openWithOptions path options) close---- | Prepare a SQL statement for execution.-openStatement :: Connection -> Query -> IO Statement-openStatement conn queryText =- mask_ do- handle <-- withConnectionHandle conn \connPtr ->- withQueryCString queryText \sql ->- alloca \stmtPtr -> do- poke stmtPtr nullPtr- flip onException (c_duckdb_destroy_prepare stmtPtr) do- rc <- runInterruptibleQuery conn (c_duckdb_prepare connPtr sql stmtPtr)- stmt <- peek stmtPtr- if rc == DuckDBSuccess- then pure stmt- else do- errMsg <- fetchPrepareError (Text.pack "duckdb-simple: prepare failed") stmt- throwIO $ mkPrepareError queryText errMsg- createStatement conn handle queryText- `onException` destroyPrepared handle---- | Finalise a prepared statement. The operation is idempotent.-closeStatement :: Statement -> IO ()-closeStatement stmt@Statement{statementState} = mask_ do- resetStatementStream stmt- finish <- atomicModifyIORef' statementState \case- StatementClosed -> (StatementClosed, pure ())- StatementOpen{statementHandle} ->- (StatementClosed, destroyPrepared statementHandle)- finish---- | Run an action with a prepared statement, closing it afterwards.-withStatement :: Connection -> Query -> (Statement -> IO a) -> IO a-withStatement conn sql = bracket (openStatement conn sql) closeStatement---- | Bind positional parameters to a prepared statement, replacing any previous bindings.-bind :: Statement -> [FieldBinding] -> IO ()-bind stmt fields = do- resetStatementStream stmt- withStatementHandle stmt \handle -> do- let actual = length fields- expected <- fmap fromIntegral (c_duckdb_nparams handle)- when (actual /= expected) $- throwFormatErrorBindings stmt (parameterCountMessage expected actual) fields- parameterNames <- fetchParameterNames handle expected- when (any isJust parameterNames) $- throwFormatErrorBindings stmt (Text.pack "duckdb-simple: statement defines named parameters; use executeNamed or bindNamed") fields- clearStatementBindings stmt- zipWithM_ apply [1 ..] fields- where- parameterCountMessage expected actual =- Text.pack $- "duckdb-simple: SQL query contains "- <> show expected- <> " parameter(s), but "- <> show actual- <> " argument(s) were supplied"-- apply :: Int -> FieldBinding -> IO ()- apply idx = bindFieldBinding stmt (fromIntegral idx :: DuckDBIdx)---- | Bind named parameters to a prepared statement, preserving any positional bindings.-bindNamed :: Statement -> [NamedParam] -> IO ()-bindNamed stmt params =- let bindings = fmap (\(name := value) -> (name, toField value)) params- parameterCountMessage expected actual =- Text.pack $- "duckdb-simple: SQL query contains "- <> show expected- <> " named parameter(s), but "- <> show actual- <> " argument(s) were supplied"- unknownNameMessage name =- Text.concat- [ Text.pack "duckdb-simple: unknown named parameter "- , name- ]- apply (name, binding) = do- mIdx <- namedParameterIndex stmt name- case mIdx of- Nothing ->- throwFormatErrorNamed stmt (unknownNameMessage name) bindings- Just idx -> bindFieldBinding stmt (fromIntegral idx :: DuckDBIdx) binding- in do- resetStatementStream stmt- let names = map (normalizeName . fst) bindings- when (Set.size (Set.fromList names) /= length names) $- throwFormatErrorNamed stmt (Text.pack "duckdb-simple: duplicate named parameter") bindings- withStatementHandle stmt \handle -> do- let actual = length bindings- expected <- fmap fromIntegral (c_duckdb_nparams handle)- when (actual /= expected) $- throwFormatErrorNamed stmt (parameterCountMessage expected actual) bindings- parameterNames <- fetchParameterNames handle expected- when (all isNothing parameterNames && expected > 0) $- throwFormatErrorNamed stmt (Text.pack "duckdb-simple: statement does not define named parameters; use positional bindings or adjust the SQL") bindings- clearStatementBindings stmt- mapM_ apply bindings--fetchParameterNames :: DuckDBPreparedStatement -> Int -> IO [Maybe Text]-fetchParameterNames handle count =- forM [1 .. count] \idx ->- bracket (c_duckdb_parameter_name handle (fromIntegral idx)) (c_duckdb_free . castPtr) \namePtr ->- if namePtr == nullPtr- then pure Nothing- else do- name <- peekUtf8CString namePtr- let normalized = normalizeName name- if normalized == Text.pack (show idx)- then pure Nothing- else pure (Just name)---- | Remove all parameter bindings associated with a prepared statement.-clearStatementBindings :: Statement -> IO ()-clearStatementBindings stmt =- withStatementHandle stmt \handle -> do- resetStatementStream stmt- rc <- c_duckdb_clear_bindings handle- when (rc /= DuckDBSuccess) $ do- err <- fetchPrepareError (Text.pack "duckdb-simple: prepare failed") handle- throwIO $ mkPrepareError (statementQuery stmt) err---- | Look up the 1-based index of a named placeholder.-namedParameterIndex :: Statement -> Text -> IO (Maybe Int)-namedParameterIndex stmt name =- withStatementHandle stmt \handle -> do- when (Text.any (== '\0') name) $- throwIO (mkPrepareError (statementQuery stmt) (Text.pack "duckdb-simple: parameter name contains NUL"))- let normalized = normalizeName name- TextForeign.withCString normalized \cName ->- alloca \idxPtr -> do- rc <- c_duckdb_bind_parameter_index handle idxPtr cName- if rc == DuckDBSuccess- then do- idx <- peek idxPtr- if idx == 0- then pure Nothing- else pure (Just (fromIntegral idx))- else pure Nothing---- | Retrieve the number of columns produced by the supplied prepared statement.-columnCount :: Statement -> IO Int-columnCount stmt =- withStatementHandle stmt $ \handle ->- fmap fromIntegral (c_duckdb_prepared_statement_column_count handle)---- | Look up the zero-based column name exposed by a prepared statement result.-columnName :: Statement -> Int -> IO Text-columnName stmt columnIndex- | columnIndex < 0 = throwIO (columnIndexError stmt columnIndex Nothing)- | otherwise =- withStatementHandle stmt \handle -> do- total <- fmap fromIntegral (c_duckdb_prepared_statement_column_count handle)- when (columnIndex >= total) $- throwIO (columnIndexError stmt columnIndex (Just total))- bracket (c_duckdb_prepared_statement_column_name handle (fromIntegral columnIndex)) (c_duckdb_free . castPtr) \namePtr ->- if namePtr == nullPtr- then throwIO (columnNameUnavailableError stmt columnIndex)- else peekUtf8CString namePtr--{- | Execute a prepared statement and return the number of affected rows.- Resets any active result stream before running and raises an @SQLError@- if DuckDB reports a failure.--}-executeStatement :: Statement -> IO Int-executeStatement stmt =- withStatementHandle stmt \handle -> do- resetStatementStream stmt- withResult (statementConnection stmt) (statementQuery stmt) (c_duckdb_execute_prepared handle) resultRowsChanged---- | Execute a query with positional parameters and return the affected row count.-execute :: (ToRow q) => Connection -> Query -> q -> IO Int-execute conn queryText params =- withStatement conn queryText \stmt -> do- bind stmt (toRow params)- executeStatement stmt---- | Execute the same query multiple times with different parameter sets.-executeMany :: (ToRow q) => Connection -> Query -> [q] -> IO Int-executeMany conn queryText rows =- withStatement conn queryText \stmt -> do- sum <$> mapM (\row -> bind stmt (toRow row) >> executeStatement stmt) rows---- | Execute an ad-hoc query without parameters and return the affected row count.-execute_ :: Connection -> Query -> IO Int-execute_ conn queryText =- withConnectionHandle conn \connPtr ->- withQueryCString queryText \sql ->- withResult conn queryText (c_duckdb_query connPtr sql) resultRowsChanged---- | Execute a query that uses named parameters.-executeNamed :: Connection -> Query -> [NamedParam] -> IO Int-executeNamed conn queryText params =- withStatement conn queryText \stmt -> do- bindNamed stmt params- executeStatement stmt---- | Run a parameterised query and decode every resulting row eagerly.-query :: (ToRow q, FromRow r) => Connection -> Query -> q -> IO [r]-query = queryWith fromRow---- | Run a parameterised query with a custom row parser.-queryWith :: (ToRow q) => RowParser r -> Connection -> Query -> q -> IO [r]-queryWith parser conn queryText params =- withStatement conn queryText \stmt -> do- bind stmt (toRow params)- withStatementHandle stmt \handle ->- withResult conn queryText (c_duckdb_execute_prepared handle) \resPtr ->- collectRows queryText resPtr >>= convertRowsWith parser queryText---- | Run a query that uses named parameters and decode all rows eagerly.-queryNamed :: (FromRow r) => Connection -> Query -> [NamedParam] -> IO [r]-queryNamed conn queryText params =- withStatement conn queryText \stmt -> do- bindNamed stmt params- withStatementHandle stmt \handle ->- withResult conn queryText (c_duckdb_execute_prepared handle) \resPtr ->- collectRows queryText resPtr >>= convertRows queryText---- | Run a query without supplying parameters and decode all rows eagerly.-query_ :: (FromRow r) => Connection -> Query -> IO [r]-query_ = queryWith_ fromRow---- | Run a query without parameters using a custom row parser.-queryWith_ :: RowParser r -> Connection -> Query -> IO [r]-queryWith_ parser conn queryText =- withConnectionHandle conn \connPtr ->- withQueryCString queryText \sql ->- withResult conn queryText (c_duckdb_query connPtr sql) \resPtr ->- collectRows queryText resPtr >>= convertRowsWith parser queryText---- Cursors and folds -----------------------------------------------------------{- | Fold a parameterised result without constructing a complete Haskell row list.- DuckDB materializes the native result before decoding starts. Native memory- use depends on the result size; the step function controls Haskell memory use.--}-fold :: (FromRow row, ToRow params) => Connection -> Query -> params -> a -> (a -> row -> IO a) -> IO a-fold conn queryText params initial step =- withStatement conn queryText \stmt -> do- resetStatementStream stmt- bind stmt (toRow params)- Result.foldStatementWith MaterializedResult fromRow stmt initial step---- | Fold a parameterless result. Native materialization follows 'fold'.-fold_ :: (FromRow row) => Connection -> Query -> a -> (a -> row -> IO a) -> IO a-fold_ conn queryText initial step =- withStatement conn queryText \stmt -> do- resetStatementStream stmt- Result.foldStatementWith MaterializedResult fromRow stmt initial step---- | Fold a result with named parameters. Native materialization follows 'fold'.-foldNamed :: (FromRow row) => Connection -> Query -> [NamedParam] -> a -> (a -> row -> IO a) -> IO a-foldNamed conn queryText params initial step =- withStatement conn queryText \stmt -> do- resetStatementStream stmt- bindNamed stmt params- Result.foldStatementWith MaterializedResult fromRow stmt initial step---- | Fetch the next row. The first call materializes the native result.-nextRow :: (FromRow r) => Statement -> IO (Maybe r)-nextRow = nextRowWith fromRow---- | Fetch the next row using a custom parser, returning @Nothing@ once exhausted.-nextRowWith :: RowParser r -> Statement -> IO (Maybe r)-nextRowWith = Result.nextRowWith MaterializedResult---- | Run an action inside a transaction.-withTransaction :: Connection -> IO a -> IO a-withTransaction conn action =- mask \restore -> do- void (execute_ conn begin)- let rollbackAction = void (try (execute_ conn rollback) :: IO (Either SomeException Int))- result <- restore action `onException` rollbackAction- void (execute_ conn commit) `onException` rollbackAction- pure result- where- begin = Query (Text.pack "BEGIN TRANSACTION")- commit = Query (Text.pack "COMMIT")- rollback = Query (Text.pack "ROLLBACK")---- Internal helpers -------------------------------------------------------------createConnection :: DuckDBDatabase -> DuckDBConnection -> TypeCache -> IO Connection-createConnection db conn cache = do- ref <- newIORef (ConnectionOpen db conn cache)- _ <-- mkWeakIORef ref $- join $- atomicModifyIORef' ref \case- ConnectionClosed -> (ConnectionClosed, pure ())- openState@(ConnectionOpen{}) ->- (ConnectionClosed, closeHandles openState)- pure Connection{connectionState = ref}--createStatement :: Connection -> DuckDBPreparedStatement -> Query -> IO Statement-createStatement parent handle queryText = do- ref <- newIORef (StatementOpen handle)- streamRef <- newIORef StatementStreamIdle- _ <-- mkWeakIORef ref $- keepAlive parent do- join $- atomicModifyIORef' ref $ \case- StatementClosed -> (StatementClosed, pure ())- StatementOpen{statementHandle} ->- ( StatementClosed- , do- cleanupStatementStreamRef streamRef- destroyPrepared statementHandle- )- pure- Statement- { statementState = ref- , statementConnection = parent- , statementQuery = queryText- , statementStream = streamRef- }--openDatabaseWithConfig :: FilePath -> [(Text, Text)] -> IO DuckDBDatabase-openDatabaseWithConfig path settings = do- when ('\0' `elem` path || any (\(name, value) -> Text.any (== '\0') name || Text.any (== '\0') value) settings) $- throwIO (mkOpenError (Text.pack "duckdb-simple: database path or configuration contains NUL"))- alloca \dbPtr ->- alloca \configPtr ->- alloca \errPtr -> do- rcConfig <- c_duckdb_create_config configPtr- when (rcConfig /= DuckDBSuccess) $- throwIO (mkOpenError (Text.pack "duckdb-simple: failed to allocate DuckDB config"))- config <- peek configPtr- poke errPtr nullPtr- let destroyConfig =- when (config /= nullPtr) $- alloca \cfgPtr -> poke cfgPtr config >> c_duckdb_destroy_config cfgPtr- flip finally destroyConfig $- do- forM_ settings \(name, value) ->- TextForeign.withCString name \cName ->- TextForeign.withCString value \cValue -> do- rcSet <- c_duckdb_set_config config cName cValue- when (rcSet /= DuckDBSuccess)- $ throwIO- $ mkOpenError- $ Text.concat- [ Text.pack "duckdb-simple: failed to set config option "- , name- ]- TextForeign.withCString (Text.pack path) \cPath -> do- rc <- c_duckdb_open_ext cPath dbPtr config errPtr- if rc == DuckDBSuccess- then do- db <- peek dbPtr- maybeFreeErr errPtr- pure db- else do- errMsg <- peekError errPtr- maybeFreeErr errPtr- throwIO $ mkOpenError errMsg--connectDatabase :: DuckDBDatabase -> IO DuckDBConnection-connectDatabase db =- alloca \connPtr -> do- rc <- c_duckdb_connect db connPtr- if rc == DuckDBSuccess- then peek connPtr- else throwIO mkConnectError--closeHandles :: ConnectionState -> IO ()-closeHandles ConnectionClosed = pure ()-closeHandles ConnectionOpen{connectionDatabase, connectionHandle, connectionTypeCache} = do- destroyTypeCache connectionTypeCache- closeConnectionHandle connectionHandle- closeDatabaseHandle connectionDatabase--closeConnectionHandle :: DuckDBConnection -> IO ()-closeConnectionHandle conn =- alloca \ptr -> poke ptr conn >> c_duckdb_disconnect ptr--closeDatabaseHandle :: DuckDBDatabase -> IO ()-closeDatabaseHandle db =- alloca \ptr -> poke ptr db >> c_duckdb_close ptr--destroyPrepared :: DuckDBPreparedStatement -> IO ()-destroyPrepared stmt =- alloca \ptr -> poke ptr stmt >> c_duckdb_destroy_prepare ptr--mkOpenError :: Text -> SQLError-mkOpenError msg =- SQLError- { sqlErrorMessage = msg- , sqlErrorType = Nothing- , sqlErrorQuery = Nothing- }--mkConnectError :: SQLError-mkConnectError =- SQLError- { sqlErrorMessage = Text.pack "duckdb-simple: failed to create connection handle"- , sqlErrorType = Nothing- , sqlErrorQuery = Nothing- }--mkPrepareError :: Query -> Text -> SQLError-mkPrepareError queryText msg =- SQLError- { sqlErrorMessage = msg- , sqlErrorType = Nothing- , sqlErrorQuery = Just queryText- }--throwFormatError :: Statement -> Text -> [String] -> IO a-throwFormatError Statement{statementQuery} message params =- throwIO- FormatError- { formatErrorMessage = message- , formatErrorQuery = statementQuery- , formatErrorParams = params- }--throwFormatErrorBindings :: Statement -> Text -> [FieldBinding] -> IO a-throwFormatErrorBindings stmt message bindings =- throwFormatError stmt message (map renderFieldBinding bindings)--throwFormatErrorNamed :: Statement -> Text -> [(Text, FieldBinding)] -> IO a-throwFormatErrorNamed stmt message bindings =- throwFormatError stmt message (map renderNamed bindings)- where- renderNamed (name, binding) =- Text.unpack name <> " := " <> renderFieldBinding binding--columnIndexError :: Statement -> Int -> Maybe Int -> SQLError-columnIndexError stmt idx total =- let base =- Text.concat- [ Text.pack "duckdb-simple: column index "- , Text.pack (show idx)- , Text.pack " out of bounds"- ]- message =- case total of- Nothing -> base- Just count ->- Text.concat- [ base- , Text.pack " (column count: "- , Text.pack (show count)- , Text.pack ")"- ]- in SQLError- { sqlErrorMessage = message- , sqlErrorType = Nothing- , sqlErrorQuery = Just (statementQuery stmt)- }--columnNameUnavailableError :: Statement -> Int -> SQLError-columnNameUnavailableError stmt idx =- SQLError- { sqlErrorMessage =- Text.concat- [ Text.pack "duckdb-simple: column name unavailable for index "- , Text.pack (show idx)- ]- , sqlErrorType = Nothing- , sqlErrorQuery = Just (statementQuery stmt)- }--normalizeName :: Text -> Text-normalizeName name =- case Text.uncons name of- Just (prefix, rest)- | prefix == ':' || prefix == '$' || prefix == '@' -> rest- _ -> name--resultRowsChanged :: Ptr DuckDBResult -> IO Int-resultRowsChanged resPtr = fromIntegral <$> c_duckdb_rows_changed resPtr--convertRows :: (FromRow r) => Query -> [[Field]] -> IO [r]-convertRows = convertRowsWith fromRow--convertRowsWith :: RowParser r -> Query -> [[Field]] -> IO [r]-convertRowsWith parser queryText rows =- case traverse (parseRow parser) rows of- Errors err -> throwIO (rowErrorsToSqlError queryText err)- Ok ok -> pure ok--peekError :: Ptr CString -> IO Text-peekError ptr = do- errPtr <- peek ptr- if errPtr == nullPtr- then pure (Text.pack "duckdb-simple: failed to open database")- else do- peekUtf8CString errPtr--maybeFreeErr :: Ptr CString -> IO ()-maybeFreeErr ptr = do- errPtr <- peek ptr- when (errPtr /= nullPtr) $ c_duckdb_free (castPtr errPtr)+-- |+-- Module : Database.DuckDB.Simple+-- Description : High-level DuckDB API in the duckdb-simple style.+--+-- The API mirrors the ergonomics of @sqlite-simple@ while being backed by the+-- DuckDB C API. It supports connection management, parameter binding, execution,+-- and typed result decoding. See @README.md@ for usage examples.+module Database.DuckDB.Simple (+ -- * Connections+ Environment,+ createEnvironment,+ closeEnvironment,+ withEnvironment,+ Instance,+ createInstance,+ closeInstance,+ withInstance,+ attachDatabase,+ detachDatabase,+ setDefaultDatabase,+ connect,+ connectWithOptions,+ withConnectionOn,+ Connection,+ open,+ openWithConfig,+ ConnectionOptions (..),+ defaultConnectionOptions,+ defaultCallbackLockWait,+ defaultAggregateStripes,+ defaultArrowBatchSize,+ openWithOptions,+ close,+ withConnection,+ withConnectionWithConfig,+ withConnectionWithOptions,++ -- * Queries and statements+ Query (..),+ Statement,+ openStatement,+ closeStatement,+ withStatement,+ clearStatementBindings,+ namedParameterIndex,+ columnCount,+ columnName,+ executeStatement,+ execute,+ executeMany,+ execute_,+ bind,+ bindNamed,+ executeNamed,+ queryNamed,+ fold,+ fold_,+ foldNamed,+ withTransaction,+ query,+ queryWith,+ query_,+ queryWith_,+ nextRow,+ nextRowWith,++ -- * Errors and conversions+ SQLError (..),+ ErrorType (..),+ FormatError (..),+ ResultError (..),+ FieldParser,+ FromField (..),+ Conversion,+ runConversion,+ liftConversion,+ FromRow (..),+ RowParser,+ field,+ fieldWith,+ numFieldsRemaining,+ -- Re-export parameter helper types.+ ToField (..),+ ToRow (..),+ FieldBinding,+ NamedParam (..),+ DuckDBColumnType (..),+ duckdbColumnType,+ SizedArray,+ sizedArray,+ sizedArrayFromList,+ sizedArrayToArray,+ Null (..),+ Only (..),+ (:.) (..),++ -- * User-defined scalar functions+ Function (..),+ LogicalTypeId (..),+ createFunction,+ createFunctionWithState,+ deleteFunction,++ -- * User-defined table functions+ TypedRow,+ createTableFunction,++ -- * User-defined aggregate functions+ Aggregate (..),+ createAggregate,+) where++import Control.Exception (SomeException, bracket, evaluate, finally, mask, mask_, onException, throwIO, try)+import Control.Monad (foldM, forM_, void, when, zipWithM_)+import Data.Array (elems)+import Data.IORef (atomicModifyIORef', newIORef, readIORef, writeIORef)+import qualified Data.IntMap.Strict as IntMap+import Data.List (findIndex)+import qualified Data.Map.Strict as Map+import qualified Data.Set as Set+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple.Conversion (Conversion, liftConversion, runConversion)+import Database.DuckDB.Simple.FromField (+ Field (..),+ FieldParser,+ FromField (..),+ ResultError (..),+ SizedArray,+ sizedArray,+ sizedArrayFromList,+ sizedArrayToArray,+ )+import Database.DuckDB.Simple.FromRow (FromRow (..), RowParser, field, fieldWith, numFieldsRemaining, parseRow, rowErrorsToSqlError)+import Database.DuckDB.Simple.Function (+ Aggregate (..),+ Function (..),+ TypedRow,+ createAggregate,+ createFunction,+ createFunctionWithState,+ createTableFunction,+ deleteFunction,+ )+import Database.DuckDB.Simple.Internal+import Database.DuckDB.Simple.Ok (Ok (..))+import Database.DuckDB.Simple.Result (resetStatementStream, withRows)+import qualified Database.DuckDB.Simple.Result as Result+import Database.DuckDB.Simple.ToField (+ DuckDBColumnType (..),+ FieldBinding,+ NamedParam (..),+ ToField (..),+ bindFieldBinding,+ duckdbColumnType,+ renderFieldBinding,+ )+import Database.DuckDB.Simple.ToRow (ToRow (..))+import Database.DuckDB.Simple.Types (FormatError (..), Null (..), Only (..), (:.) (..))+import Foreign.Marshal.Alloc (alloca)+import Foreign.Storable (peek)++-- | Run an action with a shared native environment.+withEnvironment :: (Environment -> IO a) -> IO a+withEnvironment = bracket createEnvironment closeEnvironment++-- | Run an action with a configured instance in a shared environment.+withInstance :: Environment -> [(Text, Text)] -> (Instance -> IO a) -> IO a+withInstance environment settings = bracket (createInstance environment settings) closeInstance++-- | Attach a catalog. Use 'setDefaultDatabase' to select it for new connections.+attachDatabase :: Instance -> FilePath -> Text -> [(Text, Text)] -> IO ()+attachDatabase instance_ path alias settings = attachCatalog instance_ path alias settings False++-- | Attach a catalog and optionally select it in the same native call.+attachCatalog :: Instance -> FilePath -> Text -> [(Text, Text)] -> Bool -> IO ()+attachCatalog instance_ path alias settings makeDefault = withInstanceHandle instance_ \raw -> do+ when ('\0' `elem` path || Text.any (== '\0') alias) $+ throwInvalidArgument (Text.pack "database path or alias contains NUL")+ when (any (\(key, value) -> Text.any (== '\0') key || Text.any (== '\0') value) settings) $+ throwInvalidArgument (Text.pack "attach option contains NUL")+ bracket (output (c_duckdb_v2_attach_options_create raw)) (destroy c_duckdb_v2_attach_options_destroy) \options -> do+ forM_ settings \(key, value) -> withIdentifier key \name -> withText value \text ->+ checked (c_duckdb_v2_attach_options_set options name text)+ withText (Text.pack path) \file -> withIdentifier alias \name ->+ checked (c_duckdb_v2_instance_attach raw name file options (if makeDefault then 1 else 0))++-- | Detach a catalog from an instance.+detachDatabase :: Instance -> Text -> IO ()+detachDatabase instance_ alias = withInstanceHandle instance_ \raw -> do+ when (Text.any (== '\0') alias) $+ throwInvalidArgument (Text.pack "database alias contains NUL")+ withIdentifier alias \name -> checked (c_duckdb_v2_instance_detach raw name)++-- | Select the default catalog for new connections to an instance.+setDefaultDatabase :: Instance -> Text -> IO ()+setDefaultDatabase instance_ alias = withInstanceHandle instance_ \raw -> do+ when (Text.any (== '\0') alias) $+ throwInvalidArgument (Text.pack "database alias contains NUL")+ withIdentifier alias \name -> checked (c_duckdb_v2_instance_set_default raw name)++-- | Open a connection to a shared instance.+connect :: Instance -> IO Connection+connect instance_ = connectWithOptions instance_ defaultConnectionOptions++-- | Open a connection with options. The function writes each option with its+-- default scope, as a bare SQL @SET@ does. An option whose default scope is+-- GLOBAL thus changes the instance. This change applies to all connections of+-- the instance.+connectWithOptions :: Instance -> ConnectionOptions -> IO Connection+connectWithOptions instance_ options@ConnectionOptions{connectionConfig} = mask_ do+ when (any (\(name, value) -> Text.any (== '\0') name || Text.any (== '\0') value) connectionConfig) $+ throwInvalidArgument (Text.pack "configuration contains NUL")+ checked' <- checkConnectionOptions options+ conn <- createConnection instance_ checked' (pure ())+ flip onException (close conn) do+ withConnectionHandle conn \raw -> do+ config <- output (c_duckdb_v2_connection_get_config raw)+ forM_ connectionConfig \(key, value) ->+ withIdentifier key \name -> withText value \text ->+ checked (c_duckdb_v2_config_set_option_text config name text DuckDBV2SettingScopeDefault)+ pure conn++-- | Run an action with a connection to a shared instance.+withConnectionOn :: Instance -> (Connection -> IO a) -> IO a+withConnectionOn instance_ = bracket (connect instance_) close++-- | Open a DuckDB database located at the supplied path.+open :: FilePath -> IO Connection+open path = openWithConfig path []++-- | Open a DuckDB database with configuration flags applied before startup.+openWithConfig :: FilePath -> [(Text, Text)] -> IO Connection+openWithConfig path settings = openWithOptions path defaultConnectionOptions{connectionConfig = settings}++-- | Check the option bounds before an open creates a native handle.+-- Return the record with the checked values.+checkConnectionOptions :: ConnectionOptions -> IO ConnectionOptions+checkConnectionOptions options@ConnectionOptions{connectionCallbackLockWait, connectionAggregateStripes, connectionArrowBatchSize} = do+ callbackLockWait <- checkCallbackLockWait connectionCallbackLockWait+ stripes <- checkAggregateStripes connectionAggregateStripes+ arrowBatch <- checkArrowBatchSize connectionArrowBatchSize+ pure+ options+ { connectionCallbackLockWait = callbackLockWait+ , connectionAggregateStripes = stripes+ , connectionArrowBatchSize = arrowBatch+ }++-- | Evaluate the bound of 'connectionCallbackLockWait' and reject a negative+-- bound before an open creates a native handle.+checkCallbackLockWait :: Maybe Int -> IO (Maybe Int)+checkCallbackLockWait bound = case bound of+ Just micros+ | micros < 0 -> throwInvalidArgument (Text.pack "connectionCallbackLockWait is negative")+ _ -> evaluate bound++-- | Reject a number of 'connectionAggregateStripes' that is not positive. The+-- check evaluates the field, so an error in the field occurs here and not in a+-- later registration.+checkAggregateStripes :: Int -> IO Int+checkAggregateStripes stripes+ | stripes <= 0 = throwInvalidArgument (Text.pack "connectionAggregateStripes is not positive")+ | otherwise = pure stripes++-- | Reject a 'connectionArrowBatchSize' that is not positive. The check+-- evaluates the field, so an error in the field occurs here and not in a later+-- fold.+checkArrowBatchSize :: Int -> IO Int+checkArrowBatchSize size+ | size <= 0 = throwInvalidArgument (Text.pack "connectionArrowBatchSize is not positive")+ | otherwise = pure size++-- | Open a database with a private environment and instance.+openWithOptions :: FilePath -> ConnectionOptions -> IO Connection+openWithOptions path options = mask_ do+ checked' <- checkConnectionOptions options+ environment <- createEnvironment+ instance_ <- createInstance environment (connectionConfig checked') `onException` closeEnvironment environment+ -- See Note [Native handle ownership] in "Database.DuckDB.Simple.Internal".+ let release = closeInstance instance_ `finally` closeEnvironment environment+ flip onException release do+ -- An empty alias lets DuckDB derive the catalog name from the path.+ attachCatalog instance_ (if null path then ":memory:" else path) Text.empty [] True+ createConnection instance_ checked' release++-- | Close a connection and its statements. The operation is idempotent.+close :: Connection -> IO ()+close = closeConnection++-- | Run an action with a freshly opened connection, closing it afterwards.+withConnection :: FilePath -> (Connection -> IO a) -> IO a+withConnection path = bracket (open path) close++-- | Run an action with a freshly opened configured connection, closing it afterwards.+withConnectionWithConfig :: FilePath -> [(Text, Text)] -> (Connection -> IO a) -> IO a+withConnectionWithConfig path settings = bracket (openWithConfig path settings) close++-- | Run an action with a connection opened with options, closing it afterwards.+withConnectionWithOptions :: FilePath -> ConnectionOptions -> (Connection -> IO a) -> IO a+withConnectionWithOptions path options = bracket (openWithOptions path options) close++-- | Prepare a SQL statement for execution.+openStatement :: Connection -> Query -> IO Statement+openStatement conn queryText = createStatementWithCollections conn queryText []++-- | Finalize a prepared statement. The operation is idempotent.+closeStatement :: Statement -> IO ()+closeStatement = closeStatementState++-- | Run an action with a prepared statement, closing it afterwards.+withStatement :: Connection -> Query -> (Statement -> IO a) -> IO a+withStatement conn sql = bracket (openStatement conn sql) closeStatement++-- | Bind positional parameters to a prepared statement, replacing any previous bindings.+-- The call resets the cursor. If validation or value conversion fails, the+-- previous set of bindings remains.+bind :: Statement -> [FieldBinding] -> IO ()+bind stmt fields = do+ resetStatementStream stmt+ withStatementHandle stmt \_ -> do+ -- The handle is not used. withStatementHandle raises the error for a closed statement or connection.+ let actual = length fields+ expected = length (statementParameterNames stmt)+ when (actual /= expected) $+ throwFormatErrorBindings stmt (parameterCountMessage expected actual) fields+ when (hasNamedParameters stmt) $+ throwFormatErrorBindings stmt (Text.pack "duckdb-simple: statement defines named parameters; use executeNamed or bindNamed") fields+ replaceStatementBindings stmt \staged ->+ zipWithM_ (\idx -> bindFieldBinding staged (fromIntegral idx)) [1 :: Int ..] fields+ where+ parameterCountMessage expected actual =+ Text.pack $+ "duckdb-simple: SQL query contains "+ <> show expected+ <> " parameter(s), but "+ <> show actual+ <> " argument(s) were supplied"++-- | Bind named parameters to a prepared statement.+-- The cursor reset and failure guarantees of 'bind' also apply.+-- A binding name can start with one optional @$@, @:@, or @\@@ prefix. To bind+-- the SQL parameter @$"$foo"@, supply @"$$foo"@.+-- A successful call replaces the explicit bindings. Current session variables+-- can supply omitted parameters when the statement executes.+bindNamed :: Statement -> [NamedParam] -> IO ()+bindNamed stmt params =+ let bindings = fmap (\(name := value) -> (name, toField value)) params+ parameterCountMessage expected actual =+ Text.pack $+ "duckdb-simple: SQL query contains "+ <> show expected+ <> " named parameter(s), but "+ <> show actual+ <> " argument(s) were supplied"+ unknownNameMessage name =+ Text.concat+ [ Text.pack "duckdb-simple: unknown named parameter "+ , name+ ]+ -- One lookup table serves all bindings of this call.+ positions = Map.fromList (zip (map foldIdentifier (elems (statementParameterNames stmt))) [1 :: Int ..])+ apply target (name, binding) = do+ when (Text.any (== '\0') name) $+ withQueryError (statementQuery stmt) (throwInvalidArgument (Text.pack "parameter name contains NUL"))+ case Map.lookup (bindingName name) positions of+ Nothing ->+ throwFormatErrorNamed stmt (unknownNameMessage name) bindings+ Just idx -> bindFieldBinding target (fromIntegral idx :: DuckDBIdx) binding+ in do+ resetStatementStream stmt+ let names = map (bindingName . fst) bindings+ when (Set.size (Set.fromList names) /= length names) $+ throwFormatErrorNamed stmt (Text.pack "duckdb-simple: duplicate named parameter") bindings+ withStatementHandle stmt \_ -> do+ -- The handle is not used. withStatementHandle raises the error for a closed statement or connection.+ let actual = length bindings+ expected = length (statementParameterNames stmt)+ when (actual > expected) $+ throwFormatErrorNamed stmt (parameterCountMessage expected actual) bindings+ when (expected > 0 && not (hasNamedParameters stmt)) $+ throwFormatErrorNamed+ stmt+ (Text.pack "duckdb-simple: statement does not define named parameters; use positional bindings or adjust the SQL")+ bindings+ replaceStatementBindings stmt \staged -> mapM_ (apply staged) bindings++{- Note [Atomic parameter bindings]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+A statement owns its parameter values in a Haskell map. A bind constructs the+replacement map separately. A failed bind releases only the new values.+A successful bind transfers the complete map and moves the old values into the+temporary map. The same cleanup releases that map after success or failure.+Binding resets the cursor before it validates the new parameters.+-}++-- | Replace all bindings after the action constructs every value successfully.+-- See Note [Atomic parameter bindings].+replaceStatementBindings :: Statement -> (Statement -> IO ()) -> IO ()+replaceStatementBindings stmt action =+ bracket (newIORef IntMap.empty) clearBoundValues \staged -> do+ action stmt{statementBindings = staged}+ mask_ do+ replacement <- readIORef staged+ previous <- atomicModifyIORef' (statementBindings stmt) (\values -> (replacement, values))+ writeIORef staged previous++-- | Dense numeric keys in binding order are positional parameters.+hasNamedParameters :: Statement -> Bool+hasNamedParameters stmt =+ or (zipWith (\idx name -> name /= Text.pack (show idx)) [1 :: Int ..] (elems (statementParameterNames stmt)))++-- | Remove all parameter bindings and release an active result.+clearStatementBindings :: Statement -> IO ()+clearStatementBindings stmt = withStatementHandle stmt \_ -> do+ -- The handle is not used. withStatementHandle raises the error for a closed statement or connection.+ resetStatementStream stmt+ clearBoundValues (statementBindings stmt)++-- | Look up the 1-based index of a named placeholder.+namedParameterIndex :: Statement -> Text -> IO (Maybe Int)+namedParameterIndex stmt name = withStatementHandle stmt \_ -> do+ -- The handle is not used. withStatementHandle raises the error for a closed statement or connection.+ when (Text.any (== '\0') name) $+ withQueryError (statementQuery stmt) (throwInvalidArgument (Text.pack "parameter name contains NUL"))+ pure ((+ 1) <$> findIndex ((== bindingName name) . foldIdentifier) (elems (statementParameterNames stmt)))++-- | Retrieve the number of columns produced by the supplied prepared statement.+columnCount :: Statement -> IO Int+columnCount stmt = withStatementSchemas stmt \schema _ ->+ fromIntegral <$> output (c_duckdb_v2_schema_get_count schema)++-- | Look up the zero-based column name of a prepared statement result.+columnName :: Statement -> Int -> IO Text+columnName stmt columnIndex+ | columnIndex < 0 = columnIndexError stmt columnIndex Nothing+ | otherwise = do+ total <- columnCount stmt+ when (columnIndex >= total) $ columnIndexError stmt columnIndex (Just total)+ withStatementSchemas stmt \schema _ -> alloca \name -> do+ _ <- output (c_duckdb_v2_schema_get_field schema (fromIntegral columnIndex) name)+ peek name >>= readIdentifier++-- | Execute a prepared statement and return the number of affected rows.+-- Resets any active result stream before running and raises an @SQLError@+-- if DuckDB reports a failure.+executeStatement :: Statement -> IO Int+executeStatement stmt = do+ resetStatementStream stmt+ -- See Note [Result window] in "Database.DuckDB.Simple.Internal".+ withQueryError (statementQuery stmt) . withResultWindow (statementConnection stmt) (const (pure ())) $ withResult stmt \result ->+ fromIntegral <$> runInterruptibleQuery (statementConnection stmt) (output (c_duckdb_v2_result_drain result))++-- | Execute a query with positional parameters and return the affected row count.+execute :: (ToRow q) => Connection -> Query -> q -> IO Int+execute conn queryText params =+ withStatement conn queryText \stmt -> do+ bind stmt (toRow params)+ executeStatement stmt++-- | Execute the same query multiple times with different parameter sets.+-- The call reads the parameter types one time, for the first row that needs+-- them. A column type that changes during the call does not change how the+-- later rows bind. See Note [Statement schemas] in+-- "Database.DuckDB.Simple.Internal".+executeMany :: (ToRow q) => Connection -> Query -> [q] -> IO Int+executeMany conn queryText rows =+ withStatement conn queryText \stmt -> withKeptParameterTypes stmt do+ foldM+ ( \total row -> do+ bind stmt (toRow row)+ affected <- executeStatement stmt+ pure $! total + affected+ )+ 0+ rows++-- | Execute an ad-hoc query without parameters and return the affected row count.+execute_ :: Connection -> Query -> IO Int+execute_ conn queryText = withAdHocStatement conn queryText (executeParsedStatement conn queryText) (pure 0)++-- | Execute a query that uses named parameters.+executeNamed :: Connection -> Query -> [NamedParam] -> IO Int+executeNamed conn queryText params =+ withStatement conn queryText \stmt -> do+ bindNamed stmt params+ executeStatement stmt++-- | Run a parameterised query and decode every resulting row eagerly.+-- An INSERT, UPDATE, or DELETE without RETURNING gives one row with one BIGINT+-- column: the number of changed rows.+query :: (ToRow q, FromRow r) => Connection -> Query -> q -> IO [r]+query = queryWith fromRow++-- | Run a parameterised query with a custom row parser.+queryWith :: (ToRow q) => RowParser r -> Connection -> Query -> q -> IO [r]+queryWith parser conn queryText params =+ withStatement conn queryText \stmt -> do+ bind stmt (toRow params)+ withRows stmt (convertRowsWith parser queryText)++-- | Run a query that uses named parameters and decode all rows eagerly.+queryNamed :: (FromRow r) => Connection -> Query -> [NamedParam] -> IO [r]+queryNamed conn queryText params =+ withStatement conn queryText \stmt -> do+ bindNamed stmt params+ withRows stmt (convertRows queryText)++-- | Run a query without supplying parameters and decode all rows eagerly.+query_ :: (FromRow r) => Connection -> Query -> IO [r]+query_ = queryWith_ fromRow++-- | Run a query without parameters using a custom row parser.+queryWith_ :: RowParser r -> Connection -> Query -> IO [r]+queryWith_ parser conn queryText =+ withAdHocStatement+ conn+ queryText+ (\parsed -> Result.withParsedRows conn queryText parsed (convertRowsWith parser queryText))+ (pure [])++-- | Execute preceding statements before returning the final statement's result.+withAdHocStatement :: Connection -> Query -> (DuckDBV2SqlStatementHandle -> IO a) -> IO a -> IO a+withAdHocStatement conn queryText action empty = withParsedStatements conn queryText \statements ->+ let run [] = empty+ run [parsed] = action parsed+ run (parsed : rest) = do+ void (executeParsedStatement conn queryText parsed)+ run rest+ in run statements++-- Cursors and folds ---------------------------------------------------------++-- | Fold a parameterised result without constructing a complete Haskell row list.+-- The fold collects all owned native chunks before decoding starts. Native+-- memory use depends on the result size. The step function controls Haskell memory use.+-- An INSERT, UPDATE, or DELETE without RETURNING gives one row with one BIGINT+-- column: the number of changed rows.+fold :: (FromRow row, ToRow params) => Connection -> Query -> params -> a -> (a -> row -> IO a) -> IO a+fold conn queryText params initial step =+ withStatement conn queryText \stmt -> do+ bind stmt (toRow params)+ Result.foldStatementWith MaterializedResult fromRow stmt initial step++-- | Fold a parameterless result. Native materialization follows 'fold'.+fold_ :: (FromRow row) => Connection -> Query -> a -> (a -> row -> IO a) -> IO a+fold_ conn queryText initial step =+ withStatement conn queryText \stmt ->+ Result.foldStatementWith MaterializedResult fromRow stmt initial step++-- | Fold a result with named parameters. Native materialization follows 'fold'.+foldNamed :: (FromRow row) => Connection -> Query -> [NamedParam] -> a -> (a -> row -> IO a) -> IO a+foldNamed conn queryText params initial step =+ withStatement conn queryText \stmt -> do+ bindNamed stmt params+ Result.foldStatementWith MaterializedResult fromRow stmt initial step++-- | Fetch the next row. The first call collects all owned native chunks.+-- An INSERT, UPDATE, or DELETE without RETURNING gives one row with one BIGINT+-- column: the number of changed rows.+nextRow :: (FromRow r) => Statement -> IO (Maybe r)+nextRow = nextRowWith fromRow++-- | Fetch the next row using a custom parser, returning @Nothing@ once exhausted.+nextRowWith :: RowParser r -> Statement -> IO (Maybe r)+nextRowWith = Result.nextRowWith MaterializedResult++-- | Run an action inside a transaction.+withTransaction :: Connection -> IO a -> IO a+withTransaction conn action =+ mask \restore -> do+ void (execute_ conn begin)+ let rollbackAction = void (try (execute_ conn rollback) :: IO (Either SomeException Int))+ result <- restore action `onException` rollbackAction+ void (execute_ conn commit) `onException` rollbackAction+ pure result+ where+ -- See Note [SQL text of the library] in "Database.DuckDB.Simple.Native".+ begin = Query (Text.pack "BEGIN TRANSACTION")+ commit = Query (Text.pack "COMMIT")+ rollback = Query (Text.pack "ROLLBACK")++-- Internal helpers -----------------------------------------------------------++throwFormatError :: Statement -> Text -> [String] -> IO a+throwFormatError Statement{statementQuery} message params =+ throwIO+ FormatError+ { formatErrorMessage = message+ , formatErrorQuery = statementQuery+ , formatErrorParams = params+ }++throwFormatErrorBindings :: Statement -> Text -> [FieldBinding] -> IO a+throwFormatErrorBindings stmt message bindings =+ throwFormatError stmt message (map renderFieldBinding bindings)++throwFormatErrorNamed :: Statement -> Text -> [(Text, FieldBinding)] -> IO a+throwFormatErrorNamed stmt message bindings =+ throwFormatError stmt message (map renderNamed bindings)+ where+ renderNamed (name, binding) =+ Text.unpack name <> " := " <> renderFieldBinding binding++-- | Raise the error for a column index out of bounds, with the query of the+-- statement.+columnIndexError :: Statement -> Int -> Maybe Int -> IO a+columnIndexError stmt idx total =+ let base =+ Text.concat+ [ Text.pack "column index "+ , Text.pack (show idx)+ , Text.pack " out of bounds"+ ]+ message =+ case total of+ Nothing -> base+ Just count ->+ Text.concat+ [ base+ , Text.pack " (column count: "+ , Text.pack (show count)+ , Text.pack ")"+ ]+ in withQueryError (statementQuery stmt) (throwInvalidArgument message)++-- | Strip one optional prefix from a supplied binding and fold ASCII case.+bindingName :: Text -> Text+bindingName name =+ foldIdentifier $+ case Text.uncons name of+ Just (prefix, rest)+ | prefix == ':' || prefix == '$' || prefix == '@' -> rest+ _ -> name++convertRows :: (FromRow r) => Query -> [[Field]] -> IO [r]+convertRows = convertRowsWith fromRow++convertRowsWith :: RowParser r -> Query -> [[Field]] -> IO [r]+convertRowsWith parser queryText rows =+ runConversion (traverse (parseRow parser) rows) >>= \case+ Errors err -> throwIO (rowErrorsToSqlError queryText err)+ Ok ok -> pure ok
src/Database/DuckDB/Simple/Arrow.hs view
@@ -1,48 +1,47 @@-{- |-Module : Database.DuckDB.Simple.Arrow-Description : Scoped Arrow export through the DuckDB C Data Interface.--These functions execute a query and visit its Arrow batches. DuckDB materializes-the native result before the first callback. The Haskell code converts and-releases one batch at a time. Native result memory can grow with the query size.--Each callback receives a separate schema and array. It may read them or pass-them to a consumer that releases or moves them under the Arrow C Data Interface.-The fold releases objects that the consumer has not released or moved, including-when the callback throws. The consumer must release any contents it moves.--The pointers themselves are valid during the callback only. To retain contents,-the consumer must move the root structs into its own storage and set the source-release fields to NULL. Do not retain the original pointers or move children-separately. Moved contents remain valid after the query and connection close.-Haskell consumers must mask asynchronous exceptions while moving a root and-register its cleanup before restoring exceptions.--This module uses the schema and chunk conversion API. The older query and scan-functions in @Database.DuckDB.FFI.Deprecated@ are deprecated by DuckDB.--}+-- |+-- Module : Database.DuckDB.Simple.Arrow+-- Description : Scoped Arrow export through the Arrow C Data Interface.+--+-- These functions execute a query with V2 and visit its Arrow batches. The+-- default fold collects owned batches and closes the native cursor before the+-- first callback.+-- It releases each batch after its callback. Memory use can grow with the query size.+--+-- Each callback receives a separate schema and array. It may read them or pass+-- them to a consumer that releases or moves them under the Arrow C Data Interface.+-- The fold releases objects that the consumer has not released or moved, including+-- when the callback throws. The consumer must release any contents it moves.+--+-- The pointers themselves are valid during the callback only. To retain contents,+-- the consumer must move the root structs into its own storage and set the source+-- release fields to NULL. Do not retain the original pointers or move children+-- separately. Moved contents remain valid after the query and connection close.+-- Haskell consumers must mask asynchronous exceptions while moving a root and+-- register its cleanup before restoring exceptions. module Database.DuckDB.Simple.Arrow (- foldArrow,- foldArrow_,+ foldArrow,+ foldArrow_,+ releaseArrowSchema,+ releaseArrowArray,+ releaseArrowStream, ) where -import Database.DuckDB.FFI (ArrowArray, ArrowSchema)-import Database.DuckDB.Simple.Arrow.Internal (foldArrowWith)+import Database.DuckDB.FFI.V2 (ArrowArray, ArrowSchema)+import Database.DuckDB.Simple.Arrow.Internal (foldArrowWith, releaseArrowArray, releaseArrowSchema, releaseArrowStream) import Database.DuckDB.Simple.Internal (Connection, Query, ResultMode (MaterializedResult)) import Database.DuckDB.Simple.ToRow (ToRow) import Foreign.Ptr (Ptr) -{- | Execute a parameterized query and fold over Arrow batches.--The schema includes the executed result's column names and types. The callback-runs once per nonempty batch and never runs for an empty result. The schema and-array pointers are valid only during the callback. A consumer may release or-move their contents as described in the module documentation. The accumulator-is evaluated to weak head normal form after each callback.--DuckDB materializes the native result before this fold starts. This function-does not provide bounded-memory query execution.--}+-- | Execute a parameterized query and fold over Arrow batches.+--+-- The schema includes the executed result's column names and types. The callback+-- runs once per nonempty batch and never runs for an empty result. The schema and+-- array pointers are valid only during the callback. A consumer may release or+-- move their contents as described in the module documentation. The accumulator+-- is evaluated to weak head normal form after each callback.+--+-- The fold collects all owned batches and closes the native cursor before it+-- calls the step function. It does not provide bounded-memory query execution. foldArrow :: (ToRow q) => Connection -> Query -> q -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a foldArrow = foldArrowWith MaterializedResult
src/Database/DuckDB/Simple/Arrow/Internal.hs view
@@ -1,91 +1,127 @@ {-# LANGUAGE BlockArguments #-}-{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE MultiWayIf #-}+{-# LANGUAGE OverloadedRecordDot #-}+{-# LANGUAGE TypeApplications #-} -{- |-Module : Database.DuckDB.Simple.Arrow.Internal-Description : Shared ownership and conversion for Arrow result batches.--}+-- |+-- Module : Database.DuckDB.Simple.Arrow.Internal+-- Description : V2 Arrow execution and ownership of exported schemas and arrays. module Database.DuckDB.Simple.Arrow.Internal (- foldArrowWith,+ foldArrowWith,+ releaseArrowSchema,+ releaseArrowArray,+ releaseArrowStream, ) where -import Control.Exception (bracket, bracket_, throwIO)-import Control.Monad (forM, when)-import Database.DuckDB.FFI+import Control.Exception (bracket, finally, mask, mask_, onException)+import Control.Monad (when)+import Data.IORef (atomicModifyIORef', newIORef, readIORef)+import Data.Proxy (Proxy (..)) import Database.DuckDB.Simple (bind, withStatement)-import Database.DuckDB.Simple.Internal (Connection, Query, ResultMode, SQLError (..), destroyDataChunk, destroyLogicalType, executePreparedResult, fetchResultChunk, peekUtf8CString, throwResultError, withResult, withStatementHandle)+import Database.DuckDB.Simple.Internal import Database.DuckDB.Simple.ToRow (ToRow (..))-import Foreign.Marshal.Alloc (alloca)-import Foreign.Marshal.Array (withArray)-import Foreign.Marshal.Utils (fillBytes, withMany)-import Foreign.Ptr (Ptr, nullPtr)-import Foreign.Storable (Storable (sizeOf), poke)+import Foreign.Marshal.Alloc (free, malloc)+import Foreign.Marshal.Utils (fillBytes)+import Foreign.Ptr (Ptr, nullFunPtr)+import Foreign.Storable (peek)+import HsBindgen.Runtime.Marshal (staticSizeOf)+import HsBindgen.Runtime.Support.FunPtr (fromFunPtr) --- | Fold over Arrow batches with the selected native execution mode.+-- | Owned root storage for an Arrow schema and array.+type ArrowBatch = (Ptr ArrowSchema, Ptr ArrowArray)++-- | Fold over owned Arrow batches with materialized or incremental execution. foldArrowWith :: (ToRow q) => ResultMode -> Connection -> Query -> q -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a-foldArrowWith mode conn queryText params initial step =- withStatement conn queryText \stmt -> do- bind stmt (toRow params)- withStatementHandle stmt \handle ->- withResult conn queryText (executePreparedResult mode handle) \result ->- bracket (c_duckdb_result_get_arrow_options result) destroyArrowOptions \options ->- loop result options initial- where- loop result options acc = do- next <- bracket (fetchResultChunk mode conn result) destroyDataChunk \chunk ->- if chunk == nullPtr- then do- throwResultError queryText result- pure Nothing- else do- rowCount <- c_duckdb_data_chunk_get_size chunk- if rowCount == 0- then pure (Just acc)- else withResultSchema queryText result options \schema ->- withArrowArray \array -> do- checkArrowError queryText (c_duckdb_data_chunk_to_arrow options chunk array)- nextAcc <- step acc schema array- nextAcc `seq` pure (Just nextAcc)- case next of- Nothing -> pure acc- Just nextAcc -> loop result options nextAcc+foldArrowWith mode conn queryText params initial step = withStatement conn queryText \stmt -> do+ bind stmt (toRow params)+ mask \restore -> do+ -- The native calls of the fold attach its query to an error. The step runs+ -- outside them, so that an error of the step keeps its own query, as in+ -- the row folds.+ let native = withQueryError queryText+ deliver acc (schema, array) = do+ -- A step that closes the connection stops the later steps.+ withStatementHandle stmt (\_ -> pure ())+ next <- restore (step acc schema array)+ next `seq` pure next+ case mode of+ StreamingResult ->+ -- Serialize the execution call. The stream stays open after it. See+ -- Note [Connection lock] in "Database.DuckDB.Simple.Internal".+ bracket (native (withConnectionLock conn (executePreparedArrowResult stmt))) (const (native (clearArrowResult stmt))) \result ->+ -- Recurse outside the handle scope, so that the stack does not grow for each batch.+ -- Each fetch holds the connection lock. The step runs after the+ -- fetch, without the lock.+ let stream acc =+ withStatementHandle stmt (\_ -> withConnectionLock conn (fetchBatch stmt result)) >>= \case+ Nothing -> pure acc+ Just batch -> (deliver acc batch `finally` releaseBatch batch) >>= stream+ in stream initial+ MaterializedResult -> do+ owned <- newIORef []+ let gather result =+ fetchBatch stmt result >>= \case+ Nothing -> pure ()+ Just batch -> atomicModifyIORef' owned (\batches -> (batch : batches, ())) >> gather result+ consume acc =+ atomicModifyIORef' owned (\case [] -> ([], Nothing); batch : rest -> (rest, Just batch)) >>= \case+ Nothing -> pure acc+ Just batch -> do+ acc' <- deliver acc batch `finally` releaseBatch batch+ consume acc'+ release = readIORef owned >>= mapM_ releaseBatch+ flip finally release do+ -- Hold the lock until the native result closes. The callbacks run after that.+ -- The batches are in storage that this thread releases. See+ -- Note [Result window] in "Database.DuckDB.Simple.Internal".+ native (withResultWindow conn (const (pure ())) (withArrowResult stmt gather))+ atomicModifyIORef' owned (\batches -> (reverse batches, ()))+ consume initial --- | Convert the executed result's schema and release it after the action.-withResultSchema :: Query -> Ptr DuckDBResult -> DuckDBArrowOptions -> (Ptr ArrowSchema -> IO a) -> IO a-withResultSchema queryText result options action = do- count <- c_duckdb_column_count result- let indices = if count == 0 then [] else [0 .. count - 1]- withMany (\idx -> bracket (c_duckdb_column_logical_type result idx) destroyLogicalType) indices \types -> do- names <- forM indices (c_duckdb_column_name result)- withArray types \typeArray ->- withArray names \nameArray ->- alloca \schema ->- bracket_ (fillBytes schema 0 (sizeOf (undefined :: ArrowSchema))) (releaseArrowSchema schema) do- checkArrowError queryText (c_duckdb_to_arrow_schema options typeArray nameArray count schema)- action schema+-- | Fetch one owned nonempty batch and its independent schema. See+-- Note [Interruptible native calls] in "Database.DuckDB.Simple.Internal".+fetchBatch :: Statement -> DuckDBV2ArrowResultHandle -> IO (Maybe ArrowBatch)+fetchBatch stmt result = mask_ do+ schema <- malloc+ fillBytes schema 0 (staticSizeOf (Proxy @ArrowSchema))+ array <- malloc `onException` free schema+ fillBytes array 0 (staticSizeOf (Proxy @ArrowArray))+ let batch = (schema, array)+ exported <- flip onException (releaseBatch batch) do+ runInterruptibleQuery (statementConnection stmt) (checked (c_duckdb_v2_arrow_result_fetch_array result array))+ exported <- peek array+ when (exported.release /= nullFunPtr && exported.length /= 0) $+ checked (c_duckdb_v2_arrow_result_get_schema result schema)+ pure exported+ if+ | exported.release == nullFunPtr -> releaseBatch batch >> pure Nothing+ | exported.length == 0 -> releaseBatch batch >> fetchBatch stmt result+ | otherwise -> pure (Just batch) --- | Allocate an empty Arrow array and release its contents after the action.-withArrowArray :: (Ptr ArrowArray -> IO a) -> IO a-withArrowArray action =- alloca \array ->- bracket_ (fillBytes array 0 (sizeOf (undefined :: ArrowArray))) (releaseArrowArray array) (action array)+-- | Release both Arrow objects and free their root storage.+releaseBatch :: ArrowBatch -> IO ()+releaseBatch (schema, array) =+ mask_ $+ (releaseArrowArray array `finally` free array) `finally` (releaseArrowSchema schema `finally` free schema) --- | Convert and release an owned Arrow conversion error.-checkArrowError :: Query -> IO DuckDBErrorData -> IO ()-checkArrowError queryText makeError =- bracket makeError destroyError \err ->- when (err /= nullPtr) do- failed <- c_duckdb_error_data_has_error err- when (failed /= 0) do- messagePtr <- c_duckdb_error_data_message err- message <- if messagePtr == nullPtr then pure "DuckDB Arrow conversion failed" else peekUtf8CString messagePtr- errorType <- c_duckdb_error_data_error_type err- throwIO (SQLError message (Just errorType) (Just queryText))+-- | Release an initialized schema if its release callback is present.+-- The caller still owns the storage for the structure.+releaseArrowSchema :: Ptr ArrowSchema -> IO ()+releaseArrowSchema ptr = mask_ do+ schema <- peek ptr+ when (schema.release /= nullFunPtr) $ fromFunPtr schema.release ptr --- | Destroy the result's Arrow conversion options.-destroyArrowOptions :: DuckDBArrowOptions -> IO ()-destroyArrowOptions options = alloca \ptr -> poke ptr options >> c_duckdb_destroy_arrow_options ptr+-- | Release an initialized array if its release callback is present.+-- The caller still owns the storage for the structure.+releaseArrowArray :: Ptr ArrowArray -> IO ()+releaseArrowArray ptr = mask_ do+ array <- peek ptr+ when (array.release /= nullFunPtr) $ fromFunPtr array.release ptr --- | Destroy an Arrow conversion error handle.-destroyError :: DuckDBErrorData -> IO ()-destroyError err = alloca \ptr -> poke ptr err >> c_duckdb_destroy_error_data ptr+-- | Release an initialized stream if its release callback is present.+-- The caller still owns the storage for the structure.+releaseArrowStream :: Ptr ArrowArrayStream -> IO ()+releaseArrowStream ptr = mask_ do+ stream <- peek ptr+ when (stream.release /= nullFunPtr) $ fromFunPtr stream.release ptr
src/Database/DuckDB/Simple/Callback.hs view
@@ -3,95 +3,485 @@ {-# LANGUAGE RankNTypes #-} {-# LANGUAGE ScopedTypeVariables #-} --- | Manage callback resources at the DuckDB ownership boundary.+-- |+-- Module : Database.DuckDB.Simple.Callback+-- Description : Manage callback resources and registration builders at the DuckDB ownership boundary. module Database.DuckDB.Simple.Callback (- withCallbackResources,- transferCallbackState,- runCallback,- ignoreCallbackExceptions,+ RegistrationTarget (..),+ withRegistrationFactory,+ targetCastFactory,+ checkTargetThreads,+ checkQueryThreads,+ requireWorkerCallbackRuntime,+ withTarget,+ Allocator,+ install,+ installOptional,+ installAll,+ guardContext,+ withCallbackResources,+ withCallbackResourcesFinally,+ transferCallbackState,+ readCallbackState,+ runCallback,+ runCallbackWith,+ ignoreCallbackExceptions,+ attachOpaque, ) where -import Control.Exception (SomeException, catch, displayException, finally, mask, mask_, onException, try)+import Control.Concurrent (rtsSupportsBoundThreads)+import Control.Exception (SomeException, bracket, catch, finally, fromException, mask, mask_, onException, try)+import Control.Monad (unless, when) import Data.IORef (modifyIORef', newIORef, readIORef)+import Data.Maybe (fromMaybe)+import Data.Text (Text) import qualified Data.Text as Text-import qualified Data.Text.Foreign as TextForeign-import Database.DuckDB.FFI (DuckDBDeleteCallback)-import Foreign.C.String (CString, withCString)-import Foreign.Ptr (FunPtr, Ptr, freeHaskellFunPtr, nullPtr)+import Data.Void (Void)+import Database.DuckDB.Simple.CallbackMark (withCallbackMark)+import Database.DuckDB.Simple.Internal (Connection, connectionInstance, instanceCastFactory, withConnectionHandle)+import Database.DuckDB.Simple.Native+import Foreign.Marshal.Alloc (alloca)+import Foreign.Marshal.Utils (with)+import Foreign.Ptr (FunPtr, Ptr, castFunPtr, castPtr, freeHaskellFunPtr, nullFunPtr, nullPtr) import Foreign.StablePtr (StablePtr, castPtrToStablePtr, castStablePtrToPtr, deRefStablePtr, freeStablePtr, newStablePtr)+import Foreign.Storable (Storable, peek)+import HsBindgen.Runtime.Support.FunPtr (ToFunPtr, toFunPtr)+import System.IO.Error (ioeGetErrorString, isUserError) -{- | Acquire callbacks and transfer their cleanup to a DuckDB object.-The object must be non-null. Its destructor must run after the action.+-- | A registration target. Extension handles are borrowed from an extension load+-- callback and must not escape it. Configure extension types and values with+-- the context supplied to that callback.+data RegistrationTarget = ConnectionTarget Connection | ExtensionTarget DuckDBV2ExtensionHandle DuckDBV2ContextHandle++-- | Borrow the factory of a registration target. The factory must not escape+-- this action. Use it to construct signature types and values in a configuration action.+withRegistrationFactory :: RegistrationTarget -> (DuckDBV2FactoryHandle -> IO a) -> IO a+withRegistrationFactory target action = case target of+ ConnectionTarget connection -> withConnectionHandle connection $ \native -> withConnectionFactory native action+ ExtensionTarget _ context -> withContextFactory context action++-- | The factory for the casts that the library runs in the callbacks of a+-- target. A connection target gives the factory of its instance. An extension+-- target gives 'Nothing', because an extension has no instance. A callback with+-- a context can use the default per-row cast. Scalar and aggregate functions+-- use a private instance. See Note [Function argument readers] in+-- "Database.DuckDB.Simple.Typed". A registration+-- takes the factory one time, and its callbacks keep it. The factory is valid+-- until the instance handle is destroyed. The library destroys that handle only+-- after it closes each child of the instance (Note [Native handle ownership] in+-- "Database.DuckDB.Simple.Internal"), and a callback runs only in a statement on+-- a connection of the instance. A callback keeps the factory and not the connection, for the+-- reason that withVariantReader in "Database.DuckDB.Simple.Copy.Internal" gives.+-- See Note [User casts and the casts of the library] in+-- "Database.DuckDB.Simple.Materialize".+targetCastFactory :: RegistrationTarget -> IO (Maybe DuckDBV2FactoryHandle)+targetCastFactory target = case target of+ ConnectionTarget connection -> withConnectionHandle connection \_ -> Just <$> instanceCastFactory (connectionInstance connection)+ ExtensionTarget _ _ -> pure Nothing++-- | Check the query thread count of a connection or extension registration.+checkTargetThreads :: RegistrationTarget -> IO ()+checkTargetThreads target = checkQueryThreads $ case target of+ ConnectionTarget connection -> withConnectionHandle connection (output . c_duckdb_v2_connection_get_config)+ ExtensionTarget _ context -> output (c_duckdb_v2_context_get_config context)++-- | Keep the target alive while borrowing its factory and registration builder.+withTarget+ :: (Storable handle)+ => RegistrationTarget+ -> (DuckDBV2FactoryHandle -> Ptr handle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error)+ -> (Ptr handle -> IO DuckDBV2Error)+ -> (DuckDBV2ConnectionHandle -> handle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error)+ -> (DuckDBV2ExtensionHandle -> handle -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error)+ -> (handle -> IO () -> IO a)+ -> IO a+withTarget target create release registerConnection registerExtension action = case target of+ ConnectionTarget connection -> withConnectionHandle connection $ \native ->+ withConnectionFactory native $ \factory ->+ bracket (output (create factory)) (destroy release) $ \handle -> action handle (checked (registerConnection native handle))+ ExtensionTarget extension context -> withContextFactory context $ \factory ->+ bracket (output (create factory)) (destroy release) $ \handle -> action handle (checked (registerExtension extension handle))++-- | Allocate a callback pointer tracked by the native registration owner.+type Allocator = forall a. IO (FunPtr a) -> IO (FunPtr a)++-- | Allocate one generated callback and defer its checked setter.+install+ :: (ToFunPtr auxiliary)+ => Allocator -> (FunPtr auxiliary -> callback) -> auxiliary -> (callback -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO (IO ())+install allocate wrap callback setter = do+ pointer <- wrap <$> allocate (toFunPtr callback)+ pure (checked (setter pointer))++-- | Allocate an optional callback when it is present.+installOptional :: Maybe a -> (a -> IO (IO ())) -> IO (IO ())+installOptional value action = maybe (pure (pure ())) action value++-- | Run each install action now, and return one action that runs their setters+-- in order. A registration has two phases. First the install actions allocate+-- each function pointer, in the acquire step of 'withCallbackResources', so+-- that its cleanup frees each pointer. Then the setters run after the+-- configuration action of the registration, so a callback of the library+-- replaces a callback that the configuration action set.+installAll :: [IO (IO ())] -> IO (IO ())+installAll installers = sequence_ <$> sequence installers++-- | Translate a callback exception through its native error channel.+guardContext :: (info -> DuckDBV2ContextHandle -> IO ()) -> info -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()+guardContext callback info context err = runCallback err (callback info context)++{- Note [Callback ownership]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~+The library gives two kinds of Haskell values to DuckDB. DuckDB holds each+value as a stable pointer with a destructor, in an opaque structure+(attachOpaque). After a setter takes the pointer, DuckDB owns it, and only the+destructor frees it.++- Callback resources. withCallbackResourcesFinally collects the cleanup of the+ function pointers of one registration, and its final action, in one IO+ action. It gives a stable pointer to this action to the user data of the+ registered object, with the destructor+ duckdb_simple_release_callback_resources. This destructor runs the action and+ frees the stable pointer.++- Callback state. transferCallbackState gives a stable pointer to one value to+ a state slot, for example the bind data, the init data, or the global state.+ The destructor duckdb_simple_release_callback_state frees only the stable+ pointer. readCallbackState reads the slot and dereferences the pointer. The+ caller supplies the type, so it must be the type of the value in the slot.++These two rules are the behaviour of DuckDB 2.0.0. The header does not state+them.++- A setter that fails has not taken the pointer. Thus the handlers of+ withCallbackResourcesFinally and transferCallbackState free it.++- A setter for init data, batch data, global state, or local state does not+ release a value that it replaces. Thus give one value to each slot.++The two destructors are foreign exports, and DuckDB gets their static+addresses. A static address needs no release. A function pointer of a wrapper+import must be freed with freeHaskellFunPtr after its last call. The destructor+is the last call for its object, so no later call could free such a pointer.+The destructors mask asynchronous exceptions (mask_). They have no error+channel, so they contain each exception (ignoreCallbackExceptions). Other+modules attach destructors of their own with attachOpaque, for example+duckdb_simple_release_aggregate_states in+"Database.DuckDB.Simple.Typed.Aggregate" and duckdb_simple_release_copy_binding+in "Database.DuckDB.Simple.Copy.Internal". -}-withCallbackResources ::- ((forall a. IO (FunPtr a) -> IO (FunPtr a)) -> IO r) ->- (Ptr () -> DuckDBDeleteCallback -> IO ()) ->- (r -> IO b) ->- IO b-withCallbackResources acquire attach action = mask \restore -> do- cleanups <- newIORef []- let cleanup = readIORef cleanups >>= sequence_- allocate make = do- ptr <- make- modifyIORef' cleanups (freeHaskellFunPtr ptr :)- `onException` freeHaskellFunPtr ptr- pure ptr- resources <- acquire allocate `onException` cleanup- stable <- newStablePtr cleanup `onException` cleanup- attach (castStablePtrToPtr stable) callbackResourcesDestructor- `onException` (freeStablePtr stable >> cleanup)- restore (action resources) +-- | Acquire callbacks and transfer their cleanup to a DuckDB object.+-- The object must be non-null. Its destructor must run after the action. The+-- first argument checks whether the runtime supports the callbacks. See+-- Note [Callback ownership] and Note [Callbacks without the threaded runtime].+withCallbackResources+ :: IO ()+ -> (Allocator -> IO r)+ -> (Ptr Void -> DuckDBV2OpaqueDestroyFn -> IO ())+ -> (r -> IO b)+ -> IO b+withCallbackResources checkRuntime = withCallbackResourcesFinally checkRuntime (pure ())++-- | Run a final action before the callback resources are released.+-- This includes acquisition and registration failure. Contain final-action errors.+-- A registration that acquires a callback checks the runtime before it+-- registers. See Note [Callbacks without the threaded runtime].+withCallbackResourcesFinally+ :: IO ()+ -> IO ()+ -> (Allocator -> IO r)+ -> (Ptr Void -> DuckDBV2OpaqueDestroyFn -> IO ())+ -> (r -> IO b)+ -> IO b+withCallbackResourcesFinally checkRuntime finalizer acquire attach action = mask \restore -> do+ cleanups <- newIORef []+ let cleanup = ignoreCallbackExceptions finalizer `finally` (readIORef cleanups >>= sequence_)+ allocate make = do+ ptr <- make+ modifyIORef' cleanups (freeHaskellFunPtr ptr :)+ `onException` freeHaskellFunPtr ptr+ pure ptr+ resources <- acquire allocate `onException` cleanup+ acquired <- readIORef cleanups+ unless (null acquired) checkRuntime `onException` cleanup+ stable <- newStablePtr cleanup `onException` cleanup+ -- A setter that fails has not taken the pointer, so this handler frees it.+ -- See Note [Callback ownership].+ attach (castPtr (castStablePtrToPtr stable)) callbackResourcesDestructor+ `onException` (freeStablePtr stable >> cleanup)+ restore (action resources)++{- Note [Callbacks without the threaded runtime]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+DuckDB calls a callback from its own threads. The GHC runtime without -threaded+cannot take a call from a foreign thread. The GHC User's Guide says: "In order+to use the FFI in a multi-threaded setting, you must use the -threaded option."+A program without -threaded crashed in 20 of 20 runs when a scalar function ran+over 2,000,000 rows with the setting threads = 4 ("schedule: re-entered+unsafely", a segmentation fault, or a heap corruption). With threads = 1, or+with -threaded, it crashed in none of 20 runs.++COPY can open an output file on an asynchronous worker even with threads = 1.+A log sink can receive the file-open log entry on that worker. Both kinds of+registration require -threaded regardless of the current settings. Another+connection can change async_threads after registration.++Each other registration that acquires a callback calls checkQueryThreads.+With the threaded runtime the check is one Bool test and no native call.+Without it, the check reads the+setting threads of the database that runs the callbacks (four native calls: the+config, the option value, its BIGINT, and the release of the value) and raises+the input error of the library when the setting is above 1. A registration with+no callback, for example a custom type, a macro, or a custom option with no+update function, does not check. The setting threads has only the GLOBAL scope+("cannot be set locally" in DuckDB 2.0.0), so each connection of an instance and+the instance itself give the same value.++The check cannot see a later change. A statement SET threads = 4 after the+registration, also on another connection of the instance, raises the count+again, and the library does not read the SQL text of a statement. The check is+a guard for the common mistake, not a proof that no callback runs on a second+thread.+-}++-- | Raise the input error of the library when the GHC runtime is not threaded+-- and the database that the config belongs to runs more than one thread. See+-- Note [Callbacks without the threaded runtime].+checkQueryThreads :: IO DuckDBV2ConfigHandle -> IO ()+checkQueryThreads config = unless rtsSupportsBoundThreads do+ handle <- config+ threads <- withIdentifier (Text.pack "threads") \key -> alloca \scope ->+ bracket+ (output (\value -> c_duckdb_v2_config_get_option_value handle key value scope))+ destroyValue+ (output . c_duckdb_v2_value_get_bigint)+ when (threads > 1) $+ throwInvalidArgument+ ( Text.pack+ ( "this program is not linked with -threaded, and the database runs "+ <> show threads+ <> " threads: link the program with -threaded, or set the DuckDB setting threads to 1 before you register a callback"+ )+ )++-- | Reject registrations whose callbacks can run on independent native workers.+-- See Note [Callbacks without the threaded runtime].+requireWorkerCallbackRuntime :: Text -> IO ()+requireWorkerCallbackRuntime kind =+ unless rtsSupportsBoundThreads $+ throwInvalidArgument (kind <> Text.pack " callbacks require a program linked with -threaded")+ -- | Transfer one state value to a DuckDB callback state slot.-transferCallbackState :: (Ptr () -> DuckDBDeleteCallback -> IO ()) -> a -> IO ()+-- Use each slot once. See Note [Callback ownership].+transferCallbackState :: (Ptr Void -> DuckDBV2OpaqueDestroyFn -> IO ()) -> a -> IO () transferCallbackState attach state = mask_ do- stable <- newStablePtr state- attach (castStablePtrToPtr stable) callbackStateDestructor- `onException` freeStablePtr stable+ stable <- newStablePtr state+ -- A setter that fails has not taken the pointer, so this handler frees it.+ -- See Note [Callback ownership].+ attach (castPtr (castStablePtrToPtr stable)) callbackStateDestructor+ `onException` freeStablePtr stable --- | Convert callback exceptions to DuckDB errors.-runCallback :: (CString -> IO ()) -> IO () -> IO ()-runCallback setError action = mask \restore -> do- outcome <- try (restore action)- case outcome of- Right () -> pure ()- Left (err :: SomeException) ->- TextForeign.withCString (Text.pack (displayException err)) setError- `catch` \(_ :: SomeException) ->- ignoreCallbackExceptions $- withCString "duckdb-simple: Haskell callback failed" setError+-- | Read state stored with 'transferCallbackState'. The caller supplies its type.+-- See Note [Callback ownership].+readCallbackState :: (Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a+readCallbackState getter = do+ ptr <- output getter+ when (ptr == nullPtr) $ throwInternalError (Text.pack "callback state is not initialized")+ deRefStablePtr (castPtrToStablePtr (castPtr ptr)) +-- | Convert callback exceptions to native V2 diagnostics. Mark the thread as a+-- thread inside a DuckDB callback for the time of the action, so that its+-- waits for a connection lock have the bound of the connection. See+-- 'Database.DuckDB.Simple.Internal.withConnectionLock' and+-- Note [Callback marks] in "Database.DuckDB.Simple.CallbackMark". Each DuckDB+-- callback with user code runs through this function or 'runCallbackWith',+-- except the flush function of a log sink when DuckDB destroys the sink. See+-- Note [Errors that leave a callback] for the code and the text of an error.+runCallback :: Ptr DuckDBV2ErrorInfoHandle -> IO () -> IO ()+runCallback = runCallbackAs DuckDBV2ErrorApi callbackError++-- | 'runCallback' with another native error code for each failure. The code+-- replaces the type of each exception, also the type of an t'SQLError', and the+-- text is the 'failureText' of the exception. A cast callback uses it. See+-- Note [The error code of a failed cast] in+-- "Database.DuckDB.Simple.V2.Registration".+runCallbackWith :: DuckDBV2Error -> Ptr DuckDBV2ErrorInfoHandle -> IO () -> IO ()+runCallbackWith code = runCallbackAs code \failure -> (code, Text.pack (failureText failure))++-- | Run a callback action. For an exception, write the code and the text that+-- the description gives into the error slot. Use the fallback code if the+-- description throws.+runCallbackAs :: DuckDBV2Error -> (SomeException -> (DuckDBV2Error, Text)) -> Ptr DuckDBV2ErrorInfoHandle -> IO () -> IO ()+runCallbackAs fallback describe err action = mask \restore -> do+ outcome <- try (withCallbackMark (restore action))+ case outcome of+ Right () -> pure ()+ Left (failure :: SomeException) ->+ ignoreCallbackExceptions $+ if err == nullPtr+ then pure ()+ else do+ info <- peek err+ if info == nullHandle+ then pure ()+ else do+ c_duckdb_v2_error_info_set_code info fallback >>= checkCode+ withText (Text.pack "Haskell callback failed") $ \message ->+ c_duckdb_v2_error_info_set_text info message >>= checkCode+ let (code, text) = describe failure+ when (code /= fallback) $ c_duckdb_v2_error_info_set_code info code >>= checkCode+ withText text $ \message ->+ c_duckdb_v2_error_info_set_text info message >>= checkCode++{- Note [Errors that leave a callback]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+An SQLError with a type leaves a callback with its own type, except for the ten+types of typesWithoutOwnCode. An SQLError of one of these ten types leaves the+callback as a plain Invalid Input error, because DuckDB would do something+wrong or unsafe with its own code. There are three groups:++- Seven types keep a transaction open: ErrorQueryBinder, ErrorDatabaseCatalog,+ ErrorResourceConnection, ErrorQueryParameterNotAllowed, ErrorQueryParser,+ ErrorConfigurationPermission, and ErrorResourceInUse. DuckDB expects these+ types before the execution of a statement. Under the DuckDB setting+ SYNTACTIC_ERRORS_DO_NOT_INVALIDATE of default_transaction_invalidation_policy,+ DuckDB keeps the transaction after an error of these types+ (Exception::InvalidatesTransaction in exception.cpp). A callback raises during+ the execution, for example after an INSERT has written rows. With the code of+ such a type, these rows would stay in the transaction, and COMMIT would store+ them. With DuckDBV2ErrorApi, DuckDB invalidates the explicit transaction, and+ COMMIT stores none of its rows.+- ErrorRuntimeInternal and ErrorRuntimeFatal invalidate the database. For these+ codes, DuckDB 2.0.0 invalidates the database of the outer statement, also+ when the error came from another database.+- ErrorIONetwork has no name. Its exception type has no name in DuckDB 2.0.0+ (EXCEPTION_MAP in exception.cpp). With its code, the outer statement fails+ with ErrorApi and the prefix "Invalid Error: " of a type with no name. With+ DuckDBV2ErrorApi, it fails with ErrorInputInvalid, as for an SQLError without+ a type.++The section "Errors" of the README lists these types for users.++runCallback writes an exception of a callback into the error slot of DuckDB as+a code and a text. The statement that runs the callback then fails with an+error of the type of the code. DuckDB puts the prefix of that type before the+text, for example "Conversion Error: " for the code of ErrorTypeConversion, and+"Invalid Input Error: " for DuckDBV2ErrorApi.++- An SQLError with another type writes the code of its type. The text is the+ body of the error (sqlErrorBody), the text of DuckDB without the prefix. An+ error of duckdb-simple has no body, so the text is its message. Thus the error+ of a statement in the callback reaches the outer statement with its type and+ with one prefix. An error that passes through two callbacks keeps its type+ and its body in the same way.+- An SQLError of a type of typesWithoutOwnCode writes the code DuckDBV2ErrorApi+ and its message.+- DuckDB 2.0.0 cannot raise some codes again with their own type. Six codes+ have no exception type in the map of the C API (TryGetExceptionTypeFromErrorCode+ in capi_v2.cpp): the codes of ErrorApi, ErrorIOFileNotFound,+ ErrorIOReadFailure, ErrorIOEof, ErrorInputParameterInvalid, and+ ErrorInputMetadataDeferred. DuckDB raises them as Invalid Input, so the outer+ statement fails with ErrorInputInvalid and "Invalid Input Error: ".+- Each other exception writes the code DuckDBV2ErrorApi and failureText: an+ SQLError without a type, a userError, or another exception.+- A cast callback runs through runCallbackWith and not through this rule. It+ writes the code DuckDBV2ErrorTypeMismatch and failureText for each+ exception, also for an SQLError with a type. See+ Note [The error code of a failed cast] in+ "Database.DuckDB.Simple.V2.Registration".+-}++-- | The code and the text that a callback writes into its error slot for an+-- exception. See Note [Errors that leave a callback].+callbackError :: SomeException -> (DuckDBV2Error, Text)+callbackError failure = case fromException failure of+ Just SQLError{sqlErrorType = Just errorType, sqlErrorMessage = message, sqlErrorBody = body}+ | errorType `notElem` typesWithoutOwnCode+ , Just code <- lookup errorType errorTypeCodes ->+ (code, fromMaybe message body)+ _ -> (DuckDBV2ErrorApi, Text.pack (failureText failure))++-- | The error types whose own code a callback must not write. An t'SQLError' of+-- one of these types leaves the callback as a plain Invalid Input error,+-- because DuckDB would do something wrong or unsafe with its own code.+-- Note [Errors that leave a callback] gives the reason for each type.+typesWithoutOwnCode :: [ErrorType]+typesWithoutOwnCode =+ [ ErrorQueryBinder+ , ErrorDatabaseCatalog+ , ErrorResourceConnection+ , ErrorQueryParameterNotAllowed+ , ErrorQueryParser+ , ErrorConfigurationPermission+ , ErrorResourceInUse+ , ErrorIONetwork+ , ErrorRuntimeInternal+ , ErrorRuntimeFatal+ ]++-- | The text of a callback exception in the SQL error. An t'SQLError' gives its+-- message, and a 'userError' gives its text. Thus the SQL error does not show a+-- Haskell record or the words \"user error\". Each other exception gives its+-- 'exceptionText'.+failureText :: SomeException -> String+failureText failure+ | Just sqlError <- fromException failure = Text.unpack (sqlErrorMessage sqlError)+ | Just ioError' <- fromException failure, isUserError ioError' = ioeGetErrorString ioError'+ | otherwise = exceptionText failure++-- | Transfer an opaque resource to a checked V2 setter. See+-- Note [Callback ownership].+attachOpaque+ :: (Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error)+ -> Ptr Void+ -> DuckDBV2OpaqueDestroyFn+ -> IO ()+attachOpaque setter ptr release =+ with (DuckDBV2Opaque ptr release (DuckDBV2OpaqueEqualsFn nullFunPtr)) $ \opaque ->+ checked (setter opaque)+ -- | Contain exceptions in callbacks which have no error channel. ignoreCallbackExceptions :: IO () -> IO () ignoreCallbackExceptions action = mask \restore ->- restore action `catch` \(_ :: SomeException) -> pure ()+ restore action `catch` \(_ :: SomeException) -> pure () --- | Release object-owned callbacks through a static C entry point.-releaseCallbackResources :: Ptr () -> IO ()+-- | Release object-owned callbacks through a static C entry point. See+-- Note [Callback ownership].+releaseCallbackResources :: Ptr Void -> IO () releaseCallbackResources raw =- mask_- $ ignoreCallbackExceptions- $ if raw == nullPtr- then pure ()- else do- let stable = castPtrToStablePtr raw :: StablePtr (IO ())- (deRefStablePtr stable >>= id) `finally` freeStablePtr stable+ mask_+ $ ignoreCallbackExceptions+ $ if raw == nullPtr+ then pure ()+ else do+ let stable = castPtrToStablePtr (castPtr raw) :: StablePtr (IO ())+ (deRefStablePtr stable >>= id) `finally` freeStablePtr stable --- | Release callback state independently of the registered function.-releaseCallbackState :: Ptr () -> IO ()+-- | Release callback state independently of the registered function. See+-- Note [Callback ownership].+releaseCallbackState :: Ptr Void -> IO () releaseCallbackState raw =- mask_- $ ignoreCallbackExceptions- $ if raw == nullPtr then pure () else freeStablePtr (castPtrToStablePtr raw)+ mask_+ $ ignoreCallbackExceptions+ $ if raw == nullPtr then pure () else freeStablePtr (castPtrToStablePtr (castPtr raw)) foreign export ccall "duckdb_simple_release_callback_resources"- releaseCallbackResources :: Ptr () -> IO ()+ releaseCallbackResources :: Ptr Void -> IO () foreign import ccall "&duckdb_simple_release_callback_resources"- callbackResourcesDestructor :: DuckDBDeleteCallback+ callbackResourcesDestructorAddress :: FunPtr (Ptr Void -> IO ()) foreign export ccall "duckdb_simple_release_callback_state"- releaseCallbackState :: Ptr () -> IO ()+ releaseCallbackState :: Ptr Void -> IO () foreign import ccall "&duckdb_simple_release_callback_state"- callbackStateDestructor :: DuckDBDeleteCallback+ callbackStateDestructorAddress :: FunPtr (Ptr Void -> IO ())++-- | Use the generated callback type for the static destructor address.+callbackResourcesDestructor :: DuckDBV2OpaqueDestroyFn+callbackResourcesDestructor = DuckDBV2OpaqueDestroyFn (castFunPtr callbackResourcesDestructorAddress)++-- | Use the generated callback type for the static state destructor address.+callbackStateDestructor :: DuckDBV2OpaqueDestroyFn+callbackStateDestructor = DuckDBV2OpaqueDestroyFn (castFunPtr callbackStateDestructorAddress)
+ src/Database/DuckDB/Simple/CallbackMark.hs view
@@ -0,0 +1,131 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE CPP #-}+{-# LANGUAGE MagicHash #-}+{-# LANGUAGE UnboxedTuples #-}+{-# LANGUAGE UnliftedFFITypes #-}++-- |+-- Module : Database.DuckDB.Simple.CallbackMark+-- Description : Mark the threads that are inside a DuckDB callback.+--+-- 'withCallbackMark' marks the current thread for the time of an action.+-- 'hasCallbackMark' tells if the current thread has a mark. See+-- Note [Callback marks].+module Database.DuckDB.Simple.CallbackMark (+ withCallbackMark,+ hasCallbackMark,+) where++import Control.Concurrent (myThreadId)+import Control.Exception (evaluate, finally, mask)+import Control.Monad (unless)+import Data.IORef (newIORef, readIORef)+import qualified Data.IntMap.Strict as IntMap+import GHC.Exts (casMutVar#)+import GHC.IO (IO (..))+import GHC.IORef (IORef (..))+import GHC.STRef (STRef (..))+import System.IO.Unsafe (unsafePerformIO)+#if MIN_VERSION_base(4,19,0)+import GHC.Conc.Sync (ThreadId, fromThreadId)+#else+import Foreign.C.Types (CULLong (..))+import GHC.Conc.Sync (ThreadId (..))+import GHC.Exts (ThreadId#)+#endif++{- Note [Callback marks]+~~~~~~~~~~~~~~~~~~~~~~~~+DuckDB runs a callback while a native call of the library holds the lock of a+connection. A statement that the callback runs on that connection waits for+the lock, and the holder of the lock waits for the callback. Thus+withConnectionLock gives a bound to the wait of a thread that is inside a+callback. A mark tells it which threads these are. runCallback marks its thread+with withCallbackMark, and withConnectionLock asks hasCallbackMark.++The mark of a thread is an entry in callbackMarks, one table for the process.+These are the reasons for its form.++- The table is a top-level value. A callback does not always have a connection.+ For example, an aggregate callback gets no context. Thus the table cannot be+ a field of a connection.++- The mark is not a thread label. In GHC 9.6 to 9.14.1, threadLabel gives a+ null pointer for a thread without a label, and the garbage collector can+ crash on that pointer (GHC issue 27618). DuckDB calls each callback in a new+ thread without a label.++- The key is the number of the thread, not its ThreadId. A ThreadId in the+ table keeps its thread reachable. Then the GHC runtime does not raise+ BlockedIndefinitelyOnMVar in a callback that waits for a value that no+ thread can supply, and the statement does not return. The runtime gives each+ thread a different number and does not use a number again.++- The value is a count of marks. A thread that already has a mark thus keeps+ it until the outer withCallbackMark ends.++- A change is a compare-and-swap (changeCallbackMarks). atomicModifyIORef'+ stores a value that is not evaluated, and it allocates more. If the runtime+ stops a thread during that evaluation, each other thread that needs the value+ waits. With one capability, each such wait of a callback thread is a context+ switch of the operating system. With compare-and-swap, the table holds only+ evaluated maps, and no thread waits for another thread.++- The table is one IORef. Only the threads that are inside a callback at the+ same time have an entry, so the map is small. Two threads that change the+ table at the same time cost one more attempt for one of them.+-}++-- | The threads inside a DuckDB callback, with the number of marks of each+-- thread. The key is the number of the thread. See Note [Callback marks].+callbackMarks :: IORef (IntMap.IntMap Int)+callbackMarks = unsafePerformIO (newIORef IntMap.empty)+{-# NOINLINE callbackMarks #-}++-- | The number of a Haskell thread. See Note [Callback marks].+threadNumber :: ThreadId -> Int+#if MIN_VERSION_base(4,19,0)+threadNumber = fromIntegral . fromThreadId+#else+-- base 4.18 has no fromThreadId. This is the import that base 4.19 uses for it.+threadNumber (ThreadId thread) = fromIntegral (rts_getThreadId thread)++foreign import ccall unsafe "rts_getThreadId" rts_getThreadId :: ThreadId# -> CULLong+#endif++-- | Mark the current thread as a thread inside a DuckDB callback for the time+-- of the action. The entry adds 1 to the marks of the thread, and each exit+-- subtracts 1. A thread that the action starts has no mark. See+-- Note [Callback marks].+withCallbackMark :: IO a -> IO a+withCallbackMark action = mask \restore -> do+ key <- threadNumber <$> myThreadId+ changeCallbackMarks (IntMap.insertWith (+) key 1)+ restore action `finally` changeCallbackMarks (IntMap.update unmark key)+ where+ unmark :: Int -> Maybe Int+ unmark count = if count > 1 then Just (count - 1) else Nothing++-- | Change 'callbackMarks' with compare-and-swap. Read the map, evaluate the+-- new map, and store it if the table still holds the map that was read. If+-- another thread changed the table, do the change again. See+-- Note [Callback marks].+--+-- The swap compares pointers. A map is a value of a type with more than one+-- constructor, so GHC keeps its pointer and does not make a new box for it.+changeCallbackMarks :: (IntMap.IntMap Int -> IntMap.IntMap Int) -> IO ()+changeCallbackMarks change = do+ old <- readIORef callbackMarks+ new <- evaluate (change old)+ swapped <- case callbackMarks of+ IORef (STRef var) -> IO \s -> case casMutVar# var old new s of+ (# s', 0#, _ #) -> (# s', True #)+ (# s', _, _ #) -> (# s', False #)+ unless swapped (changeCallbackMarks change)++-- | Tell if 'withCallbackMark' marks the current thread. Only the thread+-- itself adds and removes its key, so the answer is exact without a lock.+hasCallbackMark :: IO Bool+hasCallbackMark = do+ key <- threadNumber <$> myThreadId+ IntMap.member key <$> readIORef callbackMarks
src/Database/DuckDB/Simple/Catalog.hs view
@@ -1,88 +1,136 @@-{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE StrictData #-} -{- |-Module : Database.DuckDB.Simple.Catalog-Description : High-level helpers for DuckDB catalog inspection.--}+-- |+-- Module : Database.DuckDB.Simple.Catalog+-- Description : Catalog inspection through V2 queries and native table descriptions.+--+-- 'catalogTypeName' and 'lookupCatalogEntry' run SQL on the V2 connection.+-- The SQL reads the @duckdb_*()@ table functions, which are the metadata+-- interface of DuckDB. The C API V2 of DuckDB 2.0.0 has no call that looks up+-- a catalog entry. The V1 C API has such calls, but it has no supported way+-- to use a V2 connection. In a reply of 2026-10-20 to an extension author,+-- DuckDB wrote: "we'll add a catalog access interface eventually"+-- (<https://github.com/duckdb/duckdb/discussions/27525 DuckDB discussion 27525>). module Database.DuckDB.Simple.Catalog (- CatalogEntry (..),- catalogTypeName,- lookupCatalogEntry,+ CatalogEntryType (..),+ CatalogEntry (..),+ catalogTypeName,+ lookupCatalogEntry,+ TableDescription (..),+ ColumnDescription (..),+ describeTable, ) where import Control.Exception (bracket)+import Control.Monad (forM, when) import Data.Text (Text) import qualified Data.Text as Text-import qualified Data.Text.Foreign as TextForeign-import Database.DuckDB.FFI-import Database.DuckDB.Simple.Internal (Connection, peekUtf8CString, throwRegistrationError, withClientContext)-import Foreign.C.String (CString)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (nullPtr)-import Foreign.Storable (poke)+import Database.DuckDB.Simple (Only (..), query)+import Database.DuckDB.Simple.Internal (Connection, withConnectionHandle)+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeToRep) --- | A simplified view of a catalog entry returned by DuckDB.+-- | Entry kinds that DuckDB exposes through catalog metadata. There is no+-- collation kind: @pragma_collations()@ gives the name of each collation, but+-- not its catalog and its schema.+data CatalogEntryType = CatalogTable | CatalogView | CatalogIndex | CatalogSequence | CatalogType+ deriving (Eq, Ord, Show, Read)++-- | A copied catalog entry name and kind. data CatalogEntry = CatalogEntry- { catalogEntryName :: !Text- , catalogEntryType :: !DuckDBCatalogEntryType- }- deriving (Eq, Show)+ { catalogEntryName :: Text+ , catalogEntryType :: CatalogEntryType+ }+ deriving (Eq, Show) --- | Look up the backend type name of a named catalog.-catalogTypeName :: Connection -> Text -> IO (Maybe Text)-catalogTypeName conn catalogName- | Text.any (== '\0') catalogName = throwRegistrationError "catalog name contains NUL"- | otherwise =- withClientContext conn \ctx ->- TextForeign.withCString catalogName \cName ->- withMaybeCatalog ctx cName \catalog -> do- namePtr <- c_duckdb_catalog_get_type_name catalog- if namePtr == nullPtr- then pure Nothing- else Just <$> peekUtf8CString namePtr+-- | A table's qualified name, access mode, and columns.+data TableDescription = TableDescription+ { tableDescriptionName :: [Text]+ , tableDescriptionReadonly :: Bool+ , tableDescriptionColumns :: [ColumnDescription]+ }+ deriving (Eq, Show) --- | Look up a catalog entry by catalog, schema, name, and expected entry kind.-lookupCatalogEntry :: Connection -> Text -> Text -> Text -> DuckDBCatalogEntryType -> IO (Maybe CatalogEntry)-lookupCatalogEntry conn catalogName schemaName entryName entryType- | any (Text.any (== '\0')) [catalogName, schemaName, entryName] = throwRegistrationError "catalog lookup name contains NUL"- | entryType `notElem` [DuckDBCatalogEntryTypeTable, DuckDBCatalogEntryTypeView, DuckDBCatalogEntryTypeIndex, DuckDBCatalogEntryTypeSequence, DuckDBCatalogEntryTypeCollation, DuckDBCatalogEntryTypeType] =- throwRegistrationError "unsupported catalog entry type"- | otherwise =- withClientContext conn \ctx ->- TextForeign.withCString catalogName \cCatalog ->- TextForeign.withCString schemaName \cSchema ->- TextForeign.withCString entryName \cEntry ->- withMaybeCatalog ctx cCatalog \catalog ->- withMaybeCatalogEntry catalog ctx entryType cSchema cEntry \entry -> do- typ <- c_duckdb_catalog_entry_get_type entry- namePtr <- c_duckdb_catalog_entry_get_name entry- if namePtr == nullPtr- then pure Nothing- else do- name <- peekUtf8CString namePtr- pure (Just CatalogEntry{catalogEntryName = name, catalogEntryType = typ})+-- | A copied column description. Native handles remain inside the call. The+-- column type keeps the alias of DuckDB at each depth, for example JSON. See+-- Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+data ColumnDescription = ColumnDescription+ { columnDescriptionName :: Text+ , columnDescriptionType :: LogicalTypeRep+ , columnDescriptionHasDefault :: Bool+ , columnDescriptionGenerated :: Bool+ }+ deriving (Eq, Show) -destroyCatalog :: DuckDBCatalog -> IO ()-destroyCatalog catalog =- alloca \ptr -> poke ptr catalog >> c_duckdb_destroy_catalog ptr+-- | Read the backend type of an attached database, for example @duckdb@.+-- DuckDB gives a type for each attached database, so the result is 'Nothing'+-- only for a name that no attached database has.+catalogTypeName :: Connection -> Text -> IO (Maybe Text)+catalogTypeName conn name = do+ validateNames [name]+ -- See Note [SQL text of the library] in "Database.DuckDB.Simple.Native".+ rows <-+ query conn (Query ("SELECT type::VARCHAR FROM system.main.duckdb_databases() WHERE " <> identifierMatches "database_name")) (Only name)+ pure (case rows of Only value : _ -> Just value; [] -> Nothing) -destroyCatalogEntry :: DuckDBCatalogEntry -> IO ()-destroyCatalogEntry entry =- alloca \ptr -> poke ptr entry >> c_duckdb_destroy_catalog_entry ptr+-- | Look up an entry in a top-level schema. Names use DuckDB's identifier rules.+lookupCatalogEntry :: Connection -> Text -> Text -> Text -> CatalogEntryType -> IO (Maybe CatalogEntry)+lookupCatalogEntry conn catalog schema name kind = do+ validateNames [catalog, schema, name]+ (functionName, columnName) <- case kind of+ CatalogTable -> pure ("duckdb_tables()", "table_name")+ CatalogView -> pure ("duckdb_views()", "view_name")+ CatalogIndex -> pure ("duckdb_indexes()", "index_name")+ CatalogSequence -> pure ("duckdb_sequences()", "sequence_name")+ CatalogType -> pure ("duckdb_types()", "type_name")+ -- See Note [SQL text of the library] in "Database.DuckDB.Simple.Native".+ let sql =+ Query+ ( "SELECT e."+ <> columnName+ <> "::VARCHAR FROM system.main."+ <> functionName+ <> " e"+ <> " JOIN system.main.duckdb_schemas() s ON e.database_oid = s.database_oid AND e.schema_oid = s.oid"+ <> " WHERE s.parent_schema_oid IS NULL AND "+ <> identifierMatches "e.database_name"+ <> " AND "+ <> identifierMatches "e.schema_name"+ <> " AND "+ <> identifierMatches ("e." <> columnName)+ <> " LIMIT 1"+ )+ rows <- query conn sql (catalog, schema, name)+ pure (case rows of Only value : _ -> Just (CatalogEntry value kind); [] -> Nothing) -withMaybeCatalog :: DuckDBClientContext -> CString -> (DuckDBCatalog -> IO (Maybe a)) -> IO (Maybe a)-withMaybeCatalog ctx name action =- bracket (c_duckdb_client_context_get_catalog ctx name) destroyCatalog \catalog ->- if catalog == nullPtr then pure Nothing else action catalog+-- | Compare a trusted column with a parameter using ASCII case folding.+-- Binary comparison keeps the result independent of the SQL collation.+identifierMatches :: Text -> Text+identifierMatches column = normalize column <> " = " <> normalize "?"+ where+ normalize expression = "system.main.encode(system.main.translate(" <> expression <> ", 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz'))" -withMaybeCatalogEntry ::- DuckDBCatalog ->- DuckDBClientContext ->- DuckDBCatalogEntryType ->- CString ->- CString ->- (DuckDBCatalogEntry -> IO (Maybe a)) ->- IO (Maybe a)-withMaybeCatalogEntry catalog ctx entryType schemaName entryName action =- bracket (c_duckdb_catalog_get_entry catalog ctx entryType schemaName entryName) destroyCatalogEntry \entry ->- if entry == nullPtr then pure Nothing else action entry+-- | Copy a native base-table description using a parsed qualified name.+describeTable :: Connection -> Text -> IO TableDescription+describeTable conn name = withConnectionHandle conn $ \handle -> withConnectionFactory handle $ \factory ->+ withText name $ \text -> bracket (output (c_duckdb_v2_qname_parse text)) (destroy c_duckdb_v2_qname_destroy) $ \qname ->+ bracket (output (c_duckdb_v2_connection_describe_table handle qname)) (destroy c_duckdb_v2_table_description_destroy) $ \table -> do+ copiedName <- bracket (output (c_duckdb_v2_table_description_get_qname table)) (destroy c_duckdb_v2_qname_destroy) $ \resolved -> do+ parts <- output (c_duckdb_v2_qname_get_part_count resolved)+ forM (indices parts) $ \index -> output (c_duckdb_v2_qname_get_part resolved index) >>= readIdentifier+ readonly <- output (c_duckdb_v2_table_description_is_readonly table)+ count <- output (c_duckdb_v2_table_description_get_column_count table)+ columns <- forM (indices count) $ \index ->+ bracket (output (c_duckdb_v2_table_description_get_column table index)) (destroy c_duckdb_v2_column_description_destroy) $ \column -> do+ columnName <- output (c_duckdb_v2_column_description_get_name column) >>= readIdentifier+ columnType <- output (c_duckdb_v2_column_description_get_type column) >>= logicalTypeToRep factory+ defaultValue <- output (c_duckdb_v2_column_description_has_default column)+ generated <- output (c_duckdb_v2_column_description_has_generated column)+ pure (ColumnDescription columnName columnType (defaultValue /= 0) (generated /= 0))+ pure (TableDescription copiedName (readonly /= 0) columns)++-- | Reject names that contain NUL.+validateNames :: [Text] -> IO ()+validateNames names = when (any (Text.any (== '\0')) names) $ throwInvalidArgument "catalog lookup name contains NUL"
src/Database/DuckDB/Simple/Config.hs view
@@ -1,89 +1,224 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE StrictData #-} -{- |-Module : Database.DuckDB.Simple.Config-Description : High-level helpers for DuckDB 1.5 configuration inspection.--}+-- |+-- Module : Database.DuckDB.Simple.Config+-- Description : V2 configuration values, declared scopes, defaults, and aliases. module Database.DuckDB.Simple.Config (- ConfigFlag (..),- ConfigValue (..),- listConfigFlags,- getConfigOption,+ ConfigFlag (..),+ ConfigValue (..),+ ConfigOption (..),+ SettingScope (..),+ listConfigFlags,+ listConfigOptions,+ listInstanceOptions,+ getConfigOption,+ setConfigOption,+ getConfigValue,+ getConfigDefaultValue,+ setConfigValue,+ CustomOption (..),+ customOption,+ registerCustomOption, ) where -import Control.Exception (bracket)+import Control.Exception (bracket, catch, throwIO)+import Control.Monad (filterM, forM, when) import Data.Text (Text) import qualified Data.Text as Text-import qualified Data.Text.Foreign as TextForeign-import Database.DuckDB.FFI-import Database.DuckDB.Simple.Internal (Connection, destroyValue, peekUtf8CString, throwRegistrationError, withClientContext)+import Database.DuckDB.Simple.Config.Internal (CustomOption (..), registerCustomOptionOn)+import Database.DuckDB.Simple.FromField (FieldValue)+import Database.DuckDB.Simple.Internal (+ Connection,+ Instance,+ connectionInstance,+ instanceCastFactory,+ readConnectionVariant,+ withConnectionHandle,+ withInstanceHandle,+ )+import Database.DuckDB.Simple.Materialize (ReadFactory (..), materializeValue)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.ToField (ToDuckValue (..))+import Database.DuckDB.Simple.V2.Registration (RegistrationTarget (..), TypeDeclaration) import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (castPtr, nullPtr)-import Foreign.Storable (peek, poke)+import Foreign.Storable (peek) --- | Static metadata describing a known DuckDB configuration flag.+-- | The canonical name and description of a configuration option. data ConfigFlag = ConfigFlag- { configFlagName :: !Text- , configFlagDescription :: !Text- }- deriving (Eq, Show)+ { configFlagName :: Text+ , configFlagDescription :: Text+ }+ deriving (Eq, Show) --- | The current value of a configuration option plus its scope, when available.+-- | An effective option value and the scope that supplied it.+-- 'configValueText' is 'Nothing' for a NULL value, as 'configOptionSetting'. data ConfigValue = ConfigValue- { configValueText :: !Text- , configValueScope :: !(Maybe DuckDBConfigOptionScope)- }- deriving (Eq, Show)+ { configValueText :: Maybe Text+ , configValueScope :: SettingScope+ }+ deriving (Eq, Show) --- | List all configuration flags known to the linked DuckDB runtime.-listConfigFlags :: IO [ConfigFlag]-listConfigFlags = do- count <- c_duckdb_config_count- let indices = [0 .. fromIntegral count - 1] :: [Int]- mapM fetchFlag indices- where- fetchFlag idx =- alloca \namePtr ->- alloca \descPtr -> do- rc <- c_duckdb_get_config_flag (fromIntegral idx) namePtr descPtr- if rc /= DuckDBSuccess- then pure ConfigFlag{configFlagName = Text.pack (show idx), configFlagDescription = Text.pack ""}- else do- name <- peek namePtr >>= peekUtf8CString- description <- peek descPtr >>= peekUtf8CString- pure ConfigFlag{configFlagName = name, configFlagDescription = description}+-- | A copied option descriptor. The list of aliases can be empty.+-- 'configOptionSetting' is the effective value. 'Nothing' means a NULL value,+-- which the @value@ column of @duckdb_settings()@ also gives as NULL.+-- 'configOptionDefault' is the static default of the declaration. 'Nothing'+-- means that the declaration has no default. The value after @RESET@ can+-- differ from the static default: @progress_bar_time@ has no static default,+-- and DuckDB 2.0.0 gives 2000 after @RESET@.+data ConfigOption = ConfigOption+ { configOptionName :: Text+ , configOptionDescription :: Text+ , configOptionSetting :: Maybe Text+ , configOptionDefault :: Maybe Text+ , configOptionScope :: SettingScope+ , configOptionScopes :: [SettingScope]+ , configOptionAliases :: [Text]+ }+ deriving (Eq, Show) --- | Read a configuration option from a live connection's client context.+-- | Declare a global option without a default value or an update function.+customOption :: Text -> TypeDeclaration -> CustomOption+customOption name dtype = CustomOption name dtype Nothing "" SettingScopeGlobal [SettingScopeGlobal] Nothing++-- | Register a typed custom option on the connection's database.+-- The update function must not write options.+registerCustomOption :: Connection -> CustomOption -> IO ()+registerCustomOption conn = registerCustomOptionOn (ConnectionTarget conn)++-- | List the name and the description of each option of the database of the+-- connection, for example a custom option of the database.+-- 'listConfigOptions' gives the same options with their values.+listConfigFlags :: Connection -> IO [ConfigFlag]+listConfigFlags conn = withConnectionHandle conn \handle -> do+ config <- output (c_duckdb_v2_connection_get_config handle)+ copyOptions config copyFlag++-- | List options with effective values from a live connection.+listConfigOptions :: Connection -> IO [ConfigOption]+listConfigOptions conn = withConnectionHandle conn \handle -> do+ config <- output (c_duckdb_v2_connection_get_config handle)+ copyOptions config (copyOption config)++-- | List options and global values from an instance.+listInstanceOptions :: Instance -> IO [ConfigOption]+listInstanceOptions instance_ = withInstanceHandle instance_ \handle -> do+ config <- output (c_duckdb_v2_instance_get_config handle)+ copyOptions config (copyOption config)++-- | Read an effective option by its canonical name or an alias. getConfigOption :: Connection -> Text -> IO (Maybe ConfigValue)-getConfigOption conn name- | Text.any (== '\0') name = throwRegistrationError "config option name contains NUL"- | otherwise =- withClientContext conn \ctx ->- TextForeign.withCString name \cName ->- alloca \scopePtr -> do- poke scopePtr DuckDBConfigOptionScopeInvalid- bracket- (c_duckdb_client_context_get_config_option ctx cName scopePtr)- destroyValue- \duckValue ->- if duckValue == nullPtr- then pure Nothing- else do- rendered <-- bracket- (c_duckdb_get_varchar duckValue)- (c_duckdb_free . castPtr)- \strPtr ->- if strPtr == nullPtr- then pure Text.empty- else peekUtf8CString strPtr- scope <- peek scopePtr- pure $- Just- ConfigValue- { configValueText = rendered- , configValueScope =- if scope == DuckDBConfigOptionScopeInvalid- then Nothing- else Just scope- }+getConfigOption conn name = do+ checkOptionName name+ (withConfigOptionValue conn name \value scope -> Just . (`ConfigValue` scope) <$> nullableText value)+ `catch` \err ->+ if sqlErrorType err == Just ErrorInputInvalid+ then pure Nothing+ else throwIO err++-- | Write an option at its default, global, or session scope.+setConfigOption :: Connection -> SettingScope -> Text -> Text -> IO ()+setConfigOption conn scope name value = do+ checkOptionName name+ withConnectionHandle conn \handle -> do+ config <- output (c_duckdb_v2_connection_get_config handle)+ withIdentifier name \key -> withText value \setting ->+ checked (c_duckdb_v2_config_set_option_text config key setting (settingScopeToNative scope))++-- | Read a typed option value during one action. The action borrows the value.+withConfigOptionValue :: Connection -> Text -> (DuckDBV2ValueHandle -> SettingScope -> IO a) -> IO a+withConfigOptionValue conn name action = withConnectionHandle conn \handle -> do+ config <- output (c_duckdb_v2_connection_get_config handle)+ withOptionValue config name action++-- | Copy the effective typed value of an option and the scope of that value.+-- An unknown option name raises t'SQLError'. A VARIANT value gives the value+-- that it holds, as in a result. A GEOMETRY value reads with the factory of the+-- instance. See Note [User casts and the casts of the library] in+-- <src/Database.DuckDB.Simple.Materialize.html Database.DuckDB.Simple.Materialize>. The function releases all native+-- handles before it returns.+getConfigValue :: Connection -> Text -> IO (FieldValue, SettingScope)+getConfigValue conn name = do+ checkOptionName name+ withConnectionHandle conn \handle -> do+ config <- output (c_duckdb_v2_connection_get_config handle)+ casts <- instanceCastFactory (connectionInstance conn)+ withOptionValue config name \value scope -> do+ copied <- materializeValue (readConnectionVariant conn) (InstanceFactory casts) value+ pure (copied, scope)++-- | Copy the typed static default of an option. An unknown option name raises+-- t'SQLError'. A VARIANT value gives the value that it holds, as in a result. A+-- GEOMETRY value reads as 'getConfigValue' reads it. The function releases all+-- native handles before it returns.+getConfigDefaultValue :: Connection -> Text -> IO FieldValue+getConfigDefaultValue conn name = do+ checkOptionName name+ withConnectionHandle conn \handle -> do+ config <- output (c_duckdb_v2_connection_get_config handle)+ casts <- instanceCastFactory (connectionInstance conn)+ withIdentifier name \key ->+ bracket (output (c_duckdb_v2_config_get_option_by_name config key)) (destroy c_duckdb_v2_option_destroy) \option ->+ bracket+ (output (c_duckdb_v2_option_get_default_value option))+ (destroy c_duckdb_v2_value_destroy)+ (materializeValue (readConnectionVariant conn) (InstanceFactory casts))++-- | Set an option from a typed Haskell value. Native code casts it to the option type.+setConfigValue :: (ToDuckValue a) => Connection -> SettingScope -> Text -> a -> IO ()+setConfigValue conn scope name value = do+ checkOptionName name+ withConnectionHandle conn \handle -> do+ config <- output (c_duckdb_v2_connection_get_config handle)+ withConnectionFactory handle \factory ->+ bracket (toDuckValue factory value) (destroy c_duckdb_v2_value_destroy) \native ->+ withIdentifier name \key -> checked (c_duckdb_v2_config_set_option_value config key native (settingScopeToNative scope))++-- | Reject an option name that contains NUL, before a native call. Each+-- function of this module that takes an option name calls it first.+checkOptionName :: Text -> IO ()+checkOptionName name = when (Text.any (== '\0') name) $ throwInvalidArgument "config option name contains NUL"++-- | Copy the options visible through a borrowed config.+copyOptions :: DuckDBV2ConfigHandle -> (DuckDBV2OptionHandle -> IO a) -> IO [a]+copyOptions config copy = do+ count <- output (c_duckdb_v2_config_get_option_count config)+ forM (indices count) \index ->+ bracket (output (c_duckdb_v2_config_get_option_by_index config index)) (destroy c_duckdb_v2_option_destroy) copy++-- | Copy the name and description without reading the option value.+copyFlag :: DuckDBV2OptionHandle -> IO ConfigFlag+copyFlag handle =+ ConfigFlag+ <$> (output (c_duckdb_v2_option_get_name handle) >>= readIdentifier)+ <*> (output (c_duckdb_v2_option_get_description handle) >>= readText)++-- | Copy a descriptor while its native strings are alive.+copyOption :: DuckDBV2ConfigHandle -> DuckDBV2OptionHandle -> IO ConfigOption+copyOption config handle = do+ ConfigFlag name description <- copyFlag handle+ setting <- withOptionValue config name \value _ -> nullableText value+ defaultSetting <- bracket (output (c_duckdb_v2_option_get_default_value handle)) (destroy c_duckdb_v2_value_destroy) nullableText+ scope <- output (c_duckdb_v2_option_get_default_scope handle) >>= settingScopeFromNative+ scopes <-+ filterM+ (\candidate -> (/= 0) <$> output (c_duckdb_v2_option_supports_scope handle (settingScopeToNative candidate)))+ [SettingScopeGlobal, SettingScopeSession]+ count <- output (c_duckdb_v2_option_get_alias_count handle)+ aliases <- forM (indices count) \index -> output (c_duckdb_v2_option_get_alias handle index) >>= readIdentifier+ pure (ConfigOption name description setting defaultSetting scope scopes aliases)++-- | Copy the text of a value, or 'Nothing' for a NULL value.+nullableText :: DuckDBV2ValueHandle -> IO (Maybe Text)+nullableText value = do+ isNull <- output (c_duckdb_v2_value_is_null value)+ if isNull /= 0 then pure Nothing else Just <$> valueText value++-- | Release each owned option value after its action.+withOptionValue :: DuckDBV2ConfigHandle -> Text -> (DuckDBV2ValueHandle -> SettingScope -> IO a) -> IO a+withOptionValue config name action = withIdentifier name \key -> alloca \scope ->+ bracket+ (output (\value -> c_duckdb_v2_config_get_option_value config key value scope))+ (destroy c_duckdb_v2_value_destroy)+ \value -> peek scope >>= settingScopeFromNative >>= action value
+ src/Database/DuckDB/Simple/Config/Internal.hs view
@@ -0,0 +1,86 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}++-- |+-- Module : Database.DuckDB.Simple.Config.Internal+-- Description : Share the typed custom option registration for connections and extensions.+--+-- The cabal file lists this hidden module in @other-modules@.+-- 'registerCustomOptionOn' takes a registration target.+-- "Database.DuckDB.Simple.Config" exports the connection form.+-- "Database.DuckDB.Simple.Extension" exports the extension form.+module Database.DuckDB.Simple.Config.Internal (+ CustomOption (..),+ registerCustomOptionOn,+) where++import Control.Exception (bracket)+import Control.Monad (forM, forM_, when)+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple.Callback (targetCastFactory)+import Database.DuckDB.Simple.FromField (FieldValue)+import Database.DuckDB.Simple.Materialize (callbackReadFactory, materializeValue, readCallbackVariant)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.V2.Config (registerCustomOptionWith)+import Database.DuckDB.Simple.V2.Registration (+ ContextCallback,+ DefaultValue (..),+ RegistrationTarget (..),+ TypeDeclaration,+ withRegistrationFactory,+ withTypeDeclaration,+ )++-- | A custom setting. The optional update function gets each new value of the+-- option before DuckDB stores it. The value has the type of the option. An+-- exception in the function rejects the value. The text of the exception is the+-- SQL error. A RESET gives the default value to the function, or+-- 'Database.DuckDB.Simple.FromField.FieldNull' for an option without a default.+-- Thus the function must accept the default value. A VARIANT gives its+-- contained value, as in 'Database.DuckDB.Simple.Config.getConfigValue'.+data CustomOption = CustomOption+ { customOptionName :: !Text+ , customOptionType :: !TypeDeclaration+ , customOptionDefault :: Maybe DefaultValue+ , customOptionDescription :: !Text+ , customOptionDefaultScope :: !SettingScope+ , customOptionScopes :: ![SettingScope]+ , customOptionUpdate :: Maybe (FieldValue -> IO ())+ }++-- | Register a declared option on a connection or loading extension.+registerCustomOptionOn :: RegistrationTarget -> CustomOption -> IO ()+registerCustomOptionOn target option = do+ when (Text.null (customOptionName option) || Text.any (== '\0') (customOptionName option)) $+ throwInvalidArgument "custom option name must be nonempty and must not contain NUL"+ update <-+ forM (customOptionUpdate option) \callback -> do+ casts <- targetCastFactory target+ pure (updateWithValue casts callback)+ registerCustomOptionWith+ target+ ( \handle -> withRegistrationFactory target \factory -> do+ withIdentifier (customOptionName option) $ checked . c_duckdb_v2_custom_option_set_name handle+ withTypeDeclaration factory (customOptionType option) $ checked . c_duckdb_v2_custom_option_set_type handle+ forM_ (customOptionDefault option) \(DefaultValue make) ->+ bracket (make factory) destroyValue \native -> checked (c_duckdb_v2_custom_option_set_default_value handle native)+ withText (customOptionDescription option) $ checked . c_duckdb_v2_custom_option_set_description handle+ mapM_+ ( \scope ->+ checked+ (c_duckdb_v2_custom_option_set_supports_scope handle (settingScopeToNative scope) (if scope `elem` customOptionScopes option then 1 else 0))+ )+ [SettingScopeGlobal, SettingScopeSession]+ checked (c_duckdb_v2_custom_option_set_default_scope handle (settingScopeToNative (customOptionDefaultScope option)))+ )+ update++-- | Copy the new value of an option update. Then give it to the update function.+updateWithValue :: Maybe DuckDBV2FactoryHandle -> (FieldValue -> IO ()) -> ContextCallback DuckDBV2CustomOptionUpdateInfoHandle+updateWithValue casts update info context = withContextFactory context \factory ->+ bracket+ (output (c_duckdb_v2_custom_option_update_get_value info))+ destroyValue+ (materializeValue readCallbackVariant (callbackReadFactory casts factory))+ >>= update
+ src/Database/DuckDB/Simple/Conversion.hs view
@@ -0,0 +1,145 @@+-- |+-- Module : Database.DuckDB.Simple.Conversion+-- Description : Effectful conversion from DuckDB values.+module Database.DuckDB.Simple.Conversion (+ Conversion,+ runConversion,+ liftConversion,+ conversionError,+ conversionErrorWith,+ mapConversionErrors,+ mapConversionErrorsWith,+) where++import Control.Applicative (Alternative (..))+import Control.Exception (Exception, SomeException, evaluate, toException)+import Control.Monad (MonadPlus (..), ap)+import qualified Control.Monad.Fail as Fail+import Database.DuckDB.Simple.Ok (Ok (..))++-- | A conversion with explicit IO and deferred conversion errors.+-- See Note [Conversion effects].+newtype Conversion a = Conversion (IO (Either (IO [SomeException]) a))++{- Note [Conversion effects]+~~~~~~~~~~~~~~~~~~~~~~~~~~~+A conversion can read native metadata through liftConversion. These reads run+only when the conversion demands them. IO exceptions, including cancellation,+propagate to the caller. They are not conversion errors.++A failed conversion stores an action that makes its diagnostics. Alternative+discards this action when the other branch succeeds. runConversion runs the+actions only after all branches fail. Thus a discarded error does not read+metadata, such as an ENUM dictionary, just to name the SQL type.++Field metadata uses Conversion too. Its native owner and successful-result+cache belong to its result or callback scope. Native type actions fail after+that scope closes. See Note [Scoped field types] in+"Database.DuckDB.Simple.TypeScope". A child field maps its type projection+over the parent's conversion. No pure thunk performs IO or stores an IO+exception.++Each successful step forces its value to weak head normal form, as Ok does.+The force runs in IO with the caller's masking state. Thus a value nested in+the final row cannot defer its field conversion until after runConversion.+-}++instance Functor Conversion where+ {-# INLINE fmap #-}+ fmap f (Conversion action) = Conversion $ fmap mapResult action+ where+ mapResult (Left errors) = Left errors+ mapResult (Right value) = strictRight (f value)++instance Applicative Conversion where+ {-# INLINE pure #-}+ pure value = Conversion (pure (strictRight value))+ {-# INLINE (<*>) #-}+ (<*>) = ap++instance Monad Conversion where+ {-# INLINE (>>=) #-}+ Conversion action >>= next = Conversion $ do+ result <- action+ case result of+ Left errors -> pure (Left errors)+ Right value -> case next value of+ Conversion following -> following++instance Alternative Conversion where+ empty = Conversion (pure (Left (pure [])))+ Conversion first <|> Conversion second = Conversion $ do+ result <- first+ case result of+ Right value -> pure (Right value)+ Left firstErrors -> do+ alternative <- second+ pure $ case alternative of+ Right value -> Right value+ Left secondErrors -> Left ((<>) <$> firstErrors <*> secondErrors)++instance MonadPlus Conversion where+ mzero = empty+ mplus = (<|>)++instance Fail.MonadFail Conversion where+ fail = conversionError . userError++-- | Run a conversion and resolve the diagnostics if it fails.+-- Force a successful value to weak head normal form. IO exceptions propagate+-- to the caller.+runConversion :: Conversion a -> IO (Ok a)+{-# INLINE runConversion #-}+runConversion (Conversion action) = do+ result <- action+ case result of+ Left errors -> Errors <$> errors+ Right value -> evaluate (Ok value)++-- | Run an IO action when the conversion demands its result.+-- A decoder in a registered callback must use another connection for nested+-- SQL. Reads of the owned field type need no connection lock.+liftConversion :: IO a -> Conversion a+{-# INLINE liftConversion #-}+liftConversion action = Conversion (strictRight <$> action)++-- | Construct a success that forces its value when its result tag is read.+strictRight :: a -> Either e a+{-# INLINE strictRight #-}+strictRight value = value `seq` Right value++-- | Fail with a conversion error.+conversionError :: (Exception e) => e -> Conversion a+conversionError err = Conversion (pure (Left (pure [toException err])))++-- | Make an error only if the final conversion fails.+-- See Note [Conversion effects].+conversionErrorWith :: (Exception e) => Conversion e -> Conversion a+conversionErrorWith (Conversion diagnostic) = Conversion $ pure $ Left $ do+ result <- diagnostic+ case result of+ Left errors -> errors+ Right err -> pure [toException err]++-- | Change deferred diagnostics without running them.+mapConversionErrors :: ([SomeException] -> [SomeException]) -> Conversion a -> Conversion a+mapConversionErrors f (Conversion action) = Conversion $ do+ result <- action+ pure $ case result of+ Left errors -> Left (f <$> errors)+ Right value -> Right value++-- | Change the diagnostics with a conversion when the final conversion fails.+mapConversionErrorsWith :: ([SomeException] -> Conversion [SomeException]) -> Conversion a -> Conversion a+mapConversionErrorsWith f (Conversion action) = Conversion $ do+ result <- action+ pure $ case result of+ Right value -> Right value+ Left errors -> Left $ do+ original <- errors+ case f original of+ Conversion following -> do+ mapped <- following+ case mapped of+ Left failures -> failures+ Right replacement -> pure replacement
src/Database/DuckDB/Simple/Copy.hs view
@@ -1,203 +1,42 @@-{-# LANGUAGE BlockArguments #-}-{-# LANGUAGE NamedFieldPuns #-}-{-# LANGUAGE ScopedTypeVariables #-}--{- |-Module : Database.DuckDB.Simple.Copy-Description : High-level wrappers for DuckDB custom COPY functions.--}+-- |+-- Module : Database.DuckDB.Simple.Copy+-- Description : Register COPY formats and multi-file table functions with Haskell functions.+--+-- COPY format callbacks require the threaded GHC runtime (@-threaded@).+-- The functions get and give Haskell values. Batch preparation can run+-- concurrently. The sink calls for one output file run one at a time. A+-- multi-file function uses the callbacks of a registered single-file table+-- function. "Database.DuckDB.Simple.V2.Copy" has the native callback interface. module Database.DuckDB.Simple.Copy (- CopyBindInfo (..),- CopyInitInfo (..),- CopySinkInfo (..),- CopyFinalizeInfo (..),- registerCopyToFunction,+ CopyBindInfo (..),+ LogicalTypeId (..),+ CopyInitInfo (..),+ CopySinkInfo (..),+ CopyFinalizeInfo (..),+ registerCopyToFunction,+ CopyFromInfo (..),+ registerCopyFromFunction,+ CopyTo (..),+ registerCopyFunction,+ TypedRow,+ MultiFileOptions (..),+ defaultMultiFileOptions,+ registerMultiFileFunction, ) where -import Control.Exception (bracket)-import Control.Monad (forM, when)-import Data.Text (Text)-import qualified Data.Text as Text-import qualified Data.Text.Foreign as TextForeign-import Database.DuckDB.FFI-import Database.DuckDB.Simple.Callback (runCallback, transferCallbackState, withCallbackResources)-import Database.DuckDB.Simple.FromField (Field (..))-import Database.DuckDB.Simple.Internal (Connection, destroyLogicalType, peekUtf8CString, throwRegistrationError, withConnectionHandle)-import Database.DuckDB.Simple.Materialize (prepareVectorReader)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (Ptr, nullPtr)-import Foreign.StablePtr (StablePtr, castPtrToStablePtr, deRefStablePtr)-import Foreign.Storable (poke)---- | Bind-phase metadata for a custom `COPY ... TO` function.-data CopyBindInfo = CopyBindInfo- { copyBindColumnTypes :: ![DuckDBType]- }- deriving (Eq, Show)---- | Init-phase inputs for a custom `COPY ... TO` function.-data CopyInitInfo bindState = CopyInitInfo- { copyInitBindState :: !bindState- , copyInitFilePath :: !FilePath- }---- | Sink-phase inputs for a custom `COPY ... TO` function.-data CopySinkInfo bindState globalState = CopySinkInfo- { copySinkBindState :: !bindState- , copySinkGlobalState :: !globalState- }---- | Finalize-phase inputs for a custom `COPY ... TO` function.-data CopyFinalizeInfo bindState globalState = CopyFinalizeInfo- { copyFinalizeBindState :: !bindState- , copyFinalizeGlobalState :: !globalState- }--data CopyFunctionResources = CopyFunctionResources- { copyBindPtr :: !DuckDBCopyFunctionBindFun- , copyInitPtr :: !DuckDBCopyFunctionGlobalInitFun- , copySinkPtr :: !DuckDBCopyFunctionSinkFun- , copyFinalizePtr :: !DuckDBCopyFunctionFinalizeFun- }---- | Register a custom `COPY ... TO` implementation backed by Haskell callbacks.-registerCopyToFunction ::- forall bindState globalState.- Connection ->- Text ->- (CopyBindInfo -> IO bindState) ->- (CopyInitInfo bindState -> IO globalState) ->- (CopySinkInfo bindState globalState -> [[Field]] -> IO ()) ->- (CopyFinalizeInfo bindState globalState -> IO ()) ->- IO ()-registerCopyToFunction conn name bindFn initFn sinkFn finalizeFn = do- when (Text.null name || Text.any (== '\0') name) $- throwRegistrationError "invalid copy function name"- bracket c_duckdb_create_copy_function destroyCopyFunction \copyFun -> do- when (copyFun == nullPtr) $ throwRegistrationError "allocate copy function"- withCallbackResources- ( \allocate -> do- copyBindPtr <- allocate (mkCopyBindFun (copyBindHandler bindFn))- copyInitPtr <- allocate (mkCopyGlobalInitFun (copyGlobalInitHandler initFn))- copySinkPtr <- allocate (mkCopySinkFun (copySinkHandler sinkFn))- copyFinalizePtr <- allocate (mkCopyFinalizeFun (copyFinalizeHandler finalizeFn))- pure CopyFunctionResources{copyBindPtr, copyInitPtr, copySinkPtr, copyFinalizePtr}- )- (c_duckdb_copy_function_set_extra_info copyFun)- \CopyFunctionResources{copyBindPtr, copyInitPtr, copySinkPtr, copyFinalizePtr} -> do- TextForeign.withCString name $ c_duckdb_copy_function_set_name copyFun- c_duckdb_copy_function_set_bind copyFun copyBindPtr- c_duckdb_copy_function_set_global_init copyFun copyInitPtr- c_duckdb_copy_function_set_sink copyFun copySinkPtr- c_duckdb_copy_function_set_finalize copyFun copyFinalizePtr- withConnectionHandle conn \connPtr -> do- rc <- c_duckdb_register_copy_function connPtr copyFun- when (rc /= DuckDBSuccess) $ throwRegistrationError "register copy function"--copyBindHandler ::- forall bindState.- (CopyBindInfo -> IO bindState) ->- DuckDBCopyFunctionBindInfo ->- IO ()-copyBindHandler bindFn info =- runCallback (c_duckdb_copy_function_bind_set_error info) do- copyBindColumnTypes <- fetchColumnTypes info- bindState <- bindFn CopyBindInfo{copyBindColumnTypes}- transferCallbackState (c_duckdb_copy_function_bind_set_bind_data info) bindState--copyGlobalInitHandler ::- forall bindState globalState.- (CopyInitInfo bindState -> IO globalState) ->- DuckDBCopyFunctionGlobalInitInfo ->- IO ()-copyGlobalInitHandler initFn info =- runCallback (c_duckdb_copy_function_global_init_set_error info) do- rawBindState <- c_duckdb_copy_function_global_init_get_bind_data info- when (rawBindState == nullPtr) $- throwRegistrationError "missing copy bind state"- bindState <- deRefStablePtr (castPtrToStablePtr rawBindState :: StablePtr bindState)- pathPtr <- c_duckdb_copy_function_global_init_get_file_path info- filePath <-- if pathPtr == nullPtr- then pure ""- else Text.unpack <$> peekUtf8CString pathPtr- globalState <- initFn CopyInitInfo{copyInitBindState = bindState, copyInitFilePath = filePath}- transferCallbackState (c_duckdb_copy_function_global_init_set_global_state info) globalState--copySinkHandler ::- forall bindState globalState.- (CopySinkInfo bindState globalState -> [[Field]] -> IO ()) ->- DuckDBCopyFunctionSinkInfo ->- DuckDBDataChunk ->- IO ()-copySinkHandler sinkFn info chunk =- runCallback (c_duckdb_copy_function_sink_set_error info) do- bindState <- readStablePtrState c_duckdb_copy_function_sink_get_bind_data info- globalState <- readStablePtrState c_duckdb_copy_function_sink_get_global_state info- rows <- materializeChunkRows chunk- sinkFn CopySinkInfo{copySinkBindState = bindState, copySinkGlobalState = globalState} rows--copyFinalizeHandler ::- forall bindState globalState.- (CopyFinalizeInfo bindState globalState -> IO ()) ->- DuckDBCopyFunctionFinalizeInfo ->- IO ()-copyFinalizeHandler finalizeFn info =- runCallback (c_duckdb_copy_function_finalize_set_error info) do- bindState <- readStablePtrState c_duckdb_copy_function_finalize_get_bind_data info- globalState <- readStablePtrState c_duckdb_copy_function_finalize_get_global_state info- finalizeFn CopyFinalizeInfo{copyFinalizeBindState = bindState, copyFinalizeGlobalState = globalState}--fetchColumnTypes :: DuckDBCopyFunctionBindInfo -> IO [DuckDBType]-fetchColumnTypes info = do- count <- c_duckdb_copy_function_bind_get_column_count info- let indices = [0 .. fromIntegral count - 1] :: [Int]- forM indices \idx -> do- bracket- (c_duckdb_copy_function_bind_get_column_type info (fromIntegral idx))- destroyLogicalType- c_duckdb_get_type_id--readStablePtrState :: forall a i. (i -> IO (Ptr ())) -> i -> IO a-readStablePtrState getter info = do- rawPtr <- getter info- if rawPtr == nullPtr- then throwRegistrationError "missing copy callback state"- else deRefStablePtr (castPtrToStablePtr rawPtr :: StablePtr a)--materializeChunkRows :: DuckDBDataChunk -> IO [[Field]]-materializeChunkRows chunk = do- rawColumnCount <- c_duckdb_data_chunk_get_column_count chunk- let columnCount = fromIntegral rawColumnCount :: Int- readers <- mapM (makeColumnReader chunk) [0 .. columnCount - 1]- rawRowCount <- c_duckdb_data_chunk_get_size chunk- let rowCount = fromIntegral rawRowCount :: Int- forM [0 .. rowCount - 1] \row ->- forM readers \reader ->- reader (fromIntegral row)--type ColumnReader = DuckDBIdx -> IO Field--makeColumnReader :: DuckDBDataChunk -> Int -> IO ColumnReader-makeColumnReader chunk columnIndex = do- readValue <- c_duckdb_data_chunk_get_vector chunk (fromIntegral columnIndex) >>= prepareVectorReader- let name = Text.pack ("column" <> show columnIndex)- pure \rowIdx -> do- fieldValue <- readValue (fromIntegral rowIdx)- pure Field{fieldName = name, fieldIndex = columnIndex, fieldValue}--destroyCopyFunction :: DuckDBCopyFunction -> IO ()-destroyCopyFunction copyFun =- alloca \ptr -> poke ptr copyFun >> c_duckdb_destroy_copy_function ptr--foreign import ccall "wrapper"- mkCopyBindFun :: (DuckDBCopyFunctionBindInfo -> IO ()) -> IO DuckDBCopyFunctionBindFun--foreign import ccall "wrapper"- mkCopyGlobalInitFun :: (DuckDBCopyFunctionGlobalInitInfo -> IO ()) -> IO DuckDBCopyFunctionGlobalInitFun--foreign import ccall "wrapper"- mkCopySinkFun :: (DuckDBCopyFunctionSinkInfo -> DuckDBDataChunk -> IO ()) -> IO DuckDBCopyFunctionSinkFun--foreign import ccall "wrapper"- mkCopyFinalizeFun :: (DuckDBCopyFunctionFinalizeInfo -> IO ()) -> IO DuckDBCopyFunctionFinalizeFun+import Database.DuckDB.Simple.Copy.Internal (+ CopyBindInfo (..),+ CopyFinalizeInfo (..),+ CopyFromInfo (..),+ CopyInitInfo (..),+ CopySinkInfo (..),+ CopyTo (..),+ LogicalTypeId (..),+ MultiFileOptions (..),+ TypedRow,+ defaultMultiFileOptions,+ registerCopyFromFunction,+ registerCopyFunction,+ registerCopyToFunction,+ registerMultiFileFunction,+ )
+ src/Database/DuckDB/Simple/Copy/Internal.hs view
@@ -0,0 +1,394 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE ForeignFunctionInterface #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : Database.DuckDB.Simple.Copy.Internal+-- Description : Share the typed COPY registrations for connections and extensions.+--+-- The cabal file lists this hidden module in @other-modules@.+-- The functions with the suffix @On@ take a registration target.+-- "Database.DuckDB.Simple.Copy" exports the connection forms.+-- "Database.DuckDB.Simple.Extension" exports the extension forms.+module Database.DuckDB.Simple.Copy.Internal (+ CopyBindInfo (..),+ LogicalTypeId (..),+ CopyInitInfo (..),+ CopySinkInfo (..),+ CopyFinalizeInfo (..),+ registerCopyToFunction,+ registerCopyToFunctionOn,+ CopyFromInfo (..),+ registerCopyFromFunction,+ registerCopyFromFunctionOn,+ CopyTo (..),+ registerCopyFunction,+ registerCopyFunctionOn,+ TypedRow,+ MultiFileOptions (..),+ defaultMultiFileOptions,+ registerMultiFileFunction,+ registerMultiFileFunctionOn,+) where++import Control.Concurrent.MVar (MVar, modifyMVar_, newMVar)+import Control.Exception (bracket, finally, mask, mask_, onException)+import Control.Monad (forM, forM_, when)+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Void (Void)+import Database.DuckDB.Simple.Callback (attachOpaque, ignoreCallbackExceptions, readCallbackState, targetCastFactory, transferCallbackState)+import Database.DuckDB.Simple.FromField (Field (..), FieldValue)+import Database.DuckDB.Simple.Internal (Connection, withDatabaseHandle)+import Database.DuckDB.Simple.Materialize (+ FallbackReaders (..),+ ReadFactory,+ VectorReader,+ callbackReadFactory,+ prepareGeometryReader,+ prepareVectorReader,+ readCallbackVariant,+ readFactoryHandle,+ readInstanceVariant,+ )+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.TypeScope (scopedTypeRep, withTypeScope)+import Database.DuckDB.Simple.Typed (TypedRow (..))+import Database.DuckDB.Simple.Typed.Table (writeRows)+import Database.DuckDB.Simple.V2.Copy (+ CopyFromCallbacks (..),+ CopyToCallbacks (..),+ RegistrationTarget (..),+ copyFromCallbacks,+ copyToCallbacks,+ registerCopyFunctionWith,+ registerMultiFileFunctionWith,+ )+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeToRep)+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Marshal.Array (withArray)+import Foreign.Ptr (FunPtr, Ptr, castFunPtr, castPtr, nullPtr)+import Foreign.StablePtr (castPtrToStablePtr, castStablePtrToPtr, deRefStablePtr, freeStablePtr, newStablePtr)++-- | Bind-phase metadata for a custom COPY TO function.+data CopyBindInfo = CopyBindInfo+ { copyBindColumnTypes :: ![LogicalTypeId]+ }+ deriving (Eq, Show)++-- | Init-phase inputs for one output file or partition.+data CopyInitInfo bindState = CopyInitInfo+ { copyInitBindState :: !bindState+ , copyInitFilePath :: !FilePath+ }++-- | Flush-phase state passed to the row-oriented sink callback.+data CopySinkInfo bindState globalState = CopySinkInfo+ { copySinkBindState :: !bindState+ , copySinkGlobalState :: !globalState+ }++-- | Finalize-phase inputs for one output file or partition.+data CopyFinalizeInfo bindState globalState = CopyFinalizeInfo+ { copyFinalizeBindState :: !bindState+ , copyFinalizeGlobalState :: !globalState+ }++-- | Register row-oriented COPY TO callbacks. Native batch buffers are copied.+-- Use 'registerCopyFunction' for a format with a COPY FROM side too. DuckDB keeps+-- the first registration of a format name. A later registration of the same name+-- has no effect and gives no error.+registerCopyToFunction+ :: Connection+ -> Text+ -> (CopyBindInfo -> IO bindState)+ -> (CopyInitInfo bindState -> IO globalState)+ -> (CopySinkInfo bindState globalState -> [[Field]] -> IO ())+ -> (CopyFinalizeInfo bindState globalState -> IO ())+ -> IO ()+registerCopyToFunction = registerCopyToFunctionOn . ConnectionTarget++-- | Register row-oriented COPY TO callbacks on a connection or a loading+-- extension. The rules of 'registerCopyToFunction' apply.+registerCopyToFunctionOn+ :: RegistrationTarget+ -> Text+ -> (CopyBindInfo -> IO bindState)+ -> (CopyInitInfo bindState -> IO globalState)+ -> (CopySinkInfo bindState globalState -> [[Field]] -> IO ())+ -> (CopyFinalizeInfo bindState globalState -> IO ())+ -> IO ()+registerCopyToFunctionOn target name bindFn initFn sinkFn finalizeFn = do+ checkCopyName name+ casts <- targetCastFactory target+ withVariantReader target \readVariant ->+ registerCopyOn target name (Just (copyToSide casts readVariant bindFn initFn sinkFn finalizeFn)) Nothing++-- | Check the format name and register the native callbacks of each side.+registerCopyOn :: RegistrationTarget -> Text -> Maybe CopyToCallbacks -> Maybe CopyFromCallbacks -> IO ()+registerCopyOn target name to from = do+ checkCopyName name+ registerCopyFunctionWith target (\function -> withIdentifier name (checked . c_duckdb_v2_copy_function_set_name function)) to from++-- | Raise the input error for a format name that is empty or contains NUL. A+-- COPY TO side checks the name before the read of the state of the target.+checkCopyName :: Text -> IO ()+checkCopyName name = when (Text.null name || Text.any (== '\0') name) $ throwInvalidArgument "invalid copy function name"++-- | Use the host instance for a connection's COPY TO and a private one for an extension.+withVariantReader :: RegistrationTarget -> (VectorReader -> IO a) -> IO a+withVariantReader target action = case target of+ -- A catalog-owned callback must not keep the managed instance alive.+ ConnectionTarget connection -> withDatabaseHandle connection (action . readInstanceVariant)+ ExtensionTarget _ _ -> action readCallbackVariant++-- | The native COPY TO callbacks of the row-oriented functions. The first+-- argument is the factory of 'targetCastFactory'.+copyToSide+ :: forall bindState globalState+ . Maybe DuckDBV2FactoryHandle+ -> VectorReader+ -> (CopyBindInfo -> IO bindState)+ -> (CopyInitInfo bindState -> IO globalState)+ -> (CopySinkInfo bindState globalState -> [[Field]] -> IO ())+ -> (CopyFinalizeInfo bindState globalState -> IO ())+ -> CopyToCallbacks+copyToSide casts readVariant bindFn initFn sinkFn finalizeFn =+ (copyToCallbacks batch flush)+ { copyToBindCallback = Just bind+ , copyToInitCallback = Just initialize+ , copyToFinalizeCallback = Just finalize+ }+ where+ bind info _ = mask \restore -> do+ count <- output (c_duckdb_v2_copy_to_bind_get_column_count info)+ types <- acquireItems destroyLogicalType [output (c_duckdb_v2_copy_to_bind_get_column_type info index) | index <- indices count]+ state <-+ restore (mapM (\logical -> output (c_duckdb_v2_logical_type_get_id logical) >>= logicalTypeIdFromNative) types >>= bindFn . CopyBindInfo)+ `onException` releaseItems destroyLogicalType types+ stable <- newStablePtr (state, types) `onException` releaseItems destroyLogicalType types+ attachOpaque (c_duckdb_v2_copy_to_bind_set_bind_data info) (castPtr (castStablePtrToPtr stable)) copyBindingDestructor+ `onException` (freeStablePtr stable >> releaseItems destroyLogicalType types)+ initialize info _ = do+ (state, _) <- readCallbackState (c_duckdb_v2_copy_to_init_get_bind_data info)+ path <- output (c_duckdb_v2_copy_to_init_get_file_path info) >>= readText+ global <- initFn (CopyInitInfo state (Text.unpack path))+ transferCallbackState (attachOpaque (c_duckdb_v2_copy_to_init_set_init_data info)) global+ batch info context = withContextFactory context $ \factory -> do+ (_, types) <- readCallbackState (c_duckdb_v2_copy_to_batch_get_bind_data info) :: IO (bindState, [DuckDBV2LogicalTypeHandle])+ rows <-+ bracket+ (output (c_duckdb_v2_copy_to_batch_take_input info))+ (destroy c_duckdb_v2_column_data_collection_destroy)+ (materializeCollection (callbackReadFactory casts factory) readVariant types)+ transferCallbackState (attachOpaque (c_duckdb_v2_copy_to_batch_set_batch_data info)) rows+ flush info context = withTypeScope \scope -> withContextFactory context \factory -> do+ (state, types) <- readCallbackState (c_duckdb_v2_copy_to_flush_get_bind_data info)+ global <- readCallbackState (c_duckdb_v2_copy_to_flush_get_init_data info)+ rows <- readCallbackState (c_duckdb_v2_copy_to_flush_get_batch_data info)+ -- Bind data owns the types through this callback. Copy them into the+ -- flush scope. See Note [Scoped field types] in+ -- "Database.DuckDB.Simple.TypeScope".+ columnTypes <- mapM (scopedTypeRep scope (pure factory) . output . c_duckdb_v2_logical_type_copy) types+ let fields = zipWith3 (\index logical value -> Field (Text.pack ("column" <> show index)) index value logical) [0 ..] columnTypes+ sinkFn (CopySinkInfo state global) (map fields rows)+ finalize info _ = do+ (state, _) <- readCallbackState (c_duckdb_v2_copy_to_finalize_get_bind_data info)+ global <- readCallbackState (c_duckdb_v2_copy_to_finalize_get_init_data info)+ finalizeFn (CopyFinalizeInfo state global)++-- | Release native bind descriptors before their database closes. See+-- Note [Callback ownership] in "Database.DuckDB.Simple.Callback" for the rules+-- of a destructor.+releaseCopyBinding :: Ptr Void -> IO ()+releaseCopyBinding raw =+ mask_+ $ ignoreCallbackExceptions+ $ if raw == nullPtr+ then pure ()+ else do+ let stable = castPtrToStablePtr (castPtr raw)+ (deRefStablePtr stable >>= \(_, types) -> releaseItems destroyLogicalType types)+ `finally` freeStablePtr stable++foreign export ccall "duckdb_simple_release_copy_binding" releaseCopyBinding :: Ptr Void -> IO ()++foreign import ccall "&duckdb_simple_release_copy_binding" copyBindingDestructorAddress :: FunPtr (Ptr Void -> IO ())++-- | Use a static destructor address for native bind-state ownership.+copyBindingDestructor :: DuckDBV2OpaqueDestroyFn+copyBindingDestructor = DuckDBV2OpaqueDestroyFn (castFunPtr copyBindingDestructorAddress)++-- | Copy each scanned chunk before the scan invalidates its borrowed buffers.+-- The factory makes the chunk and reads GEOMETRY.+materializeCollection+ :: ReadFactory -> VectorReader -> [DuckDBV2LogicalTypeHandle] -> DuckDBV2ColumnDataCollectionHandle -> IO [[FieldValue]]+materializeCollection readFactory readVariant types collection =+ withArray types \typePtr ->+ bracket+ (output (c_duckdb_v2_data_chunk_create (readFactoryHandle readFactory) (ConstPtr typePtr) (fromIntegral (length types))))+ (destroy c_duckdb_v2_data_chunk_destroy)+ \chunk ->+ bracket+ (output (c_duckdb_v2_column_data_collection_shared_scan_state_create collection))+ (destroy c_duckdb_v2_column_data_collection_shared_scan_state_destroy)+ \shared ->+ bracket+ (output (c_duckdb_v2_column_data_collection_worker_scan_state_create collection))+ (destroy c_duckdb_v2_column_data_collection_worker_scan_state_destroy)+ \worker ->+ let loop groups = do+ produced <- output (c_duckdb_v2_column_data_collection_scan collection shared worker chunk)+ if produced == 0+ then pure (concat (reverse groups))+ else do+ rows <-+ materializeChunkRows+ (FallbackReaders (logicalTypeToRep (readFactoryHandle readFactory)) readVariant (prepareGeometryReader readFactory))+ chunk+ loop (rows : groups)+ in loop []++-- | Copy the values of a complete chunk before its collection scan advances.+-- The flush callback adds the fields and their scoped type actions.+materializeChunkRows :: FallbackReaders -> DuckDBV2DataChunkHandle -> IO [[FieldValue]]+materializeChunkRows readers chunk = do+ columns <- output (c_duckdb_v2_data_chunk_get_vector_count chunk)+ columnReaders <- forM [0 .. fromIntegral columns - 1] \index -> do+ vector <- output (c_duckdb_v2_data_chunk_get_vector chunk (fromIntegral (index :: Int)))+ prepareVectorReader readers vector+ rows <- output (c_duckdb_v2_data_chunk_get_size chunk)+ forM [0 .. fromIntegral rows - 1] \row -> mapM ($ row) columnReaders++-- | The inputs of the COPY FROM side for one statement. The path is the text+-- of the statement. DuckDB does not expand a glob pattern in it. The names and+-- the types are the columns of the target table, in order.+data CopyFromInfo = CopyFromInfo+ { copyFromPath :: !FilePath+ , copyFromColumnNames :: ![Text]+ , copyFromColumnTypes :: ![LogicalTypeId]+ }+ deriving (Eq, Show)++-- | Register the COPY FROM side of a format. Use 'registerCopyFunction' for a+-- format with the two sides. DuckDB keeps the first registration of a format+-- name. A later registration of the same name has no effect and gives no error.+--+-- The function gets the path and the columns of the target table. It gives the+-- rows to insert. It runs one time for each run of a COPY FROM statement. DuckDB+-- takes the rows from the list one chunk at a time, so the list can be lazy. A+-- row must have one column for each column of the table, in the same order. An+-- exception from the function or from a row fails the statement.+--+-- > registerCopyFromFunction conn "lines" \info -> map Only . Text.lines <$> TextIO.readFile (copyFromPath info)+registerCopyFromFunction :: (TypedRow row) => Connection -> Text -> (CopyFromInfo -> IO [row]) -> IO ()+registerCopyFromFunction = registerCopyFromFunctionOn . ConnectionTarget++-- | Register the COPY FROM side of a format on a connection or a loading+-- extension. The rules of 'registerCopyFromFunction' apply.+registerCopyFromFunctionOn :: (TypedRow row) => RegistrationTarget -> Text -> (CopyFromInfo -> IO [row]) -> IO ()+registerCopyFromFunctionOn target name readRows = registerCopyOn target name Nothing (Just (copyFromSide name readRows))++-- | The native COPY FROM callbacks of a typed reader. The bind callback reads+-- the path and the columns of the target table. See+-- Note [Row lists of typed scans] in "Database.DuckDB.Simple.Typed.Table".+copyFromSide :: forall row. (TypedRow row) => Text -> (CopyFromInfo -> IO [row]) -> CopyFromCallbacks+copyFromSide name readRows = (copyFromCallbacks bind execute){copyFromInitGlobalCallback = Just initialize}+ where+ width = length (rowColumnDeclarations (Proxy @row))+ bind info _ = do+ path <- output (c_duckdb_v2_copy_from_bind_get_file_path info) >>= readText+ count <- output (c_duckdb_v2_copy_from_bind_get_column_count info)+ when (fromIntegral count /= width) $+ throwInvalidArgument+ ( "the table of COPY FROM has "+ <> Text.pack (show (fromIntegral count :: Int))+ <> " columns, but a row of "+ <> name+ <> " has "+ <> Text.pack (show width)+ )+ let columns = takeWhile (< count) [0 ..]+ names <- forM columns \index -> output (c_duckdb_v2_copy_from_bind_get_column_name info index) >>= readIdentifier+ types <- forM columns \index ->+ bracket+ (output (c_duckdb_v2_copy_from_bind_get_column_type info index))+ destroyLogicalType+ (\logical -> output (c_duckdb_v2_logical_type_get_id logical) >>= logicalTypeIdFromNative)+ transferCallbackState+ (attachOpaque (c_duckdb_v2_copy_from_bind_set_bind_data info))+ (readRows (CopyFromInfo (Text.unpack path) names types))+ initialize info _ = do+ scan <- readCallbackState (c_duckdb_v2_copy_from_init_global_get_bind_data info)+ remaining <- scan >>= newMVar :: IO (MVar [row])+ transferCallbackState (attachOpaque (c_duckdb_v2_copy_from_init_global_set_global_state info)) remaining+ execute info context = withContextFactory context \factory -> do+ remaining <- readCallbackState (c_duckdb_v2_copy_from_exec_get_global_state info) :: IO (MVar [row])+ chunk <- output (c_duckdb_v2_copy_from_exec_get_output_chunk info)+ -- The lock gives each row to one call only. See+ -- Note [Row lists of typed scans] in "Database.DuckDB.Simple.Typed.Table".+ modifyMVar_ remaining (writeRows factory chunk (take width [0 ..]))++-- | The COPY TO side of a format. The four functions have the roles of the+-- four functions of 'registerCopyToFunction'.+data CopyTo bindState globalState = CopyTo+ { copyToBind :: CopyBindInfo -> IO bindState+ , copyToInit :: CopyInitInfo bindState -> IO globalState+ , copyToSink :: CopySinkInfo bindState globalState -> [[Field]] -> IO ()+ , copyToFinalize :: CopyFinalizeInfo bindState globalState -> IO ()+ }++-- | Register a format with a COPY TO side and a COPY FROM side under one+-- name. The rules of 'registerCopyToFunction' apply to the COPY TO side. The+-- rules of 'registerCopyFromFunction' apply to the COPY FROM side. Two+-- registrations of one name do not give a format with the two sides, because+-- DuckDB keeps the first registration.+registerCopyFunction :: (TypedRow row) => Connection -> Text -> CopyTo bindState globalState -> (CopyFromInfo -> IO [row]) -> IO ()+registerCopyFunction = registerCopyFunctionOn . ConnectionTarget++-- | Register a format with the two sides on a connection or a loading+-- extension. The rules of 'registerCopyFunction' apply.+registerCopyFunctionOn+ :: (TypedRow row) => RegistrationTarget -> Text -> CopyTo bindState globalState -> (CopyFromInfo -> IO [row]) -> IO ()+registerCopyFunctionOn target name (CopyTo bindFn initFn sinkFn finalizeFn) readRows = do+ checkCopyName name+ casts <- targetCastFactory target+ withVariantReader target \readVariant ->+ registerCopyOn target name (Just (copyToSide casts readVariant bindFn initFn sinkFn finalizeFn)) (Just (copyFromSide name readRows))++-- | Options supplied to DuckDB's multi-file wrapper.+data MultiFileOptions = MultiFileOptions+ { multiFileReaderType :: Maybe Text+ , multiFileExtension :: Maybe Text+ }+ deriving (Eq, Show)++-- | Use native defaults for reader descriptions and directory expansion.+defaultMultiFileOptions :: MultiFileOptions+defaultMultiFileOptions = MultiFileOptions Nothing Nothing++-- | Register globbing, file lists, filename columns, and hive partition+-- support for a registered single-file table function. The second name is that+-- table function. DuckDB can run the exec callback of the single-file function on+-- more than one thread at the same time for one scan. A function from+-- 'Database.DuckDB.Simple.Function.createTableFunction' permits that. A table+-- function with native callbacks must synchronize its scan state.+registerMultiFileFunction :: Connection -> Text -> Text -> MultiFileOptions -> IO ()+registerMultiFileFunction connection = registerMultiFileFunctionOn (ConnectionTarget connection)++-- | Register a multi-file wrapper on a connection or a loading extension. The+-- second name is the registered single-file table function. Both names must not+-- be empty and must not contain NUL.+registerMultiFileFunctionOn :: RegistrationTarget -> Text -> Text -> MultiFileOptions -> IO ()+registerMultiFileFunctionOn target name singleFile MultiFileOptions{multiFileReaderType, multiFileExtension} = do+ when (any (\value -> Text.null value || Text.any (== '\0') value) [name, singleFile]) $+ throwInvalidArgument "invalid multi-file function name"+ registerMultiFileFunctionWith target \function -> do+ withText name (\(ConstPtr view) -> checked (c_duckdb_v2_multi_file_function_set_name function view))+ withText singleFile (\(ConstPtr view) -> checked (c_duckdb_v2_multi_file_function_set_single_file_function function view))+ forM_ multiFileReaderType \value -> withText value (\(ConstPtr view) -> checked (c_duckdb_v2_multi_file_function_set_reader_type function view))+ forM_ multiFileExtension \value -> withText value (\(ConstPtr view) -> checked (c_duckdb_v2_multi_file_function_set_file_extension function view))
− src/Database/DuckDB/Simple/Deprecated/Streaming.hs
@@ -1,87 +0,0 @@-{-# LANGUAGE BlockArguments #-}--{- |-Module : Database.DuckDB.Simple.Deprecated.Streaming-Description : Optional native streaming through DuckDB's deprecated execution API.--Import this module qualified to request native streaming. DuckDB can still-materialize a result, and operators such as sorting can buffer data. Streaming-reduces result storage when the query supports it; it does not bound all query-memory. The default @Database.DuckDB.Simple@ API materializes native results.--Rows and Arrow batches use the same decoders and cleanup as the default API.-Cancellation also interrupts native chunk fetching. Prompt interruption requires-@-threaded@ and waits for native code and Haskell callbacks to return.--Serialize connection use for the entire cursor or fold. Another query on the-same connection can invalidate an active stream. Close or reset an abandoned-statement to release its result.--}-module Database.DuckDB.Simple.Deprecated.Streaming- {-# DEPRECATED "Uses DuckDB's deprecated streaming execution API. Prefer Database.DuckDB.Simple or Database.DuckDB.Simple.Arrow when materialization is acceptable." #-} (- fold,- fold_,- foldNamed,- nextRow,- nextRowWith,- foldArrow,- foldArrow_,-) where--import Database.DuckDB.FFI (ArrowArray, ArrowSchema)-import Database.DuckDB.Simple (bind, bindNamed, withStatement)-import qualified Database.DuckDB.Simple.Arrow.Internal as Arrow-import Database.DuckDB.Simple.FromRow (FromRow (..), RowParser)-import Database.DuckDB.Simple.Internal (Connection, Query, ResultMode (StreamingResult), Statement)-import qualified Database.DuckDB.Simple.Result as Result-import Database.DuckDB.Simple.ToField (NamedParam)-import Database.DuckDB.Simple.ToRow (ToRow (..))-import Foreign.Ptr (Ptr)---- | Fold a parameterized query, requesting native streaming.-fold :: (FromRow row, ToRow params) => Connection -> Query -> params -> a -> (a -> row -> IO a) -> IO a-fold conn sql params initial step =- withStatement conn sql \stmt -> do- bind stmt (toRow params)- Result.foldStatementWith StreamingResult fromRow stmt initial step---- | Fold a query without parameters, requesting native streaming.-fold_ :: (FromRow row) => Connection -> Query -> a -> (a -> row -> IO a) -> IO a-fold_ conn sql initial step =- withStatement conn sql \stmt ->- Result.foldStatementWith StreamingResult fromRow stmt initial step---- | Fold a query with named parameters, requesting native streaming.-foldNamed :: (FromRow row) => Connection -> Query -> [NamedParam] -> a -> (a -> row -> IO a) -> IO a-foldNamed conn sql params initial step =- withStatement conn sql \stmt -> do- bindNamed stmt params- Result.foldStatementWith StreamingResult fromRow stmt initial step---- | Fetch the next row, requesting native streaming on the first fetch.-nextRow :: (FromRow r) => Statement -> IO (Maybe r)-nextRow = nextRowWith fromRow--{- | Fetch a row with a custom parser, requesting streaming on the first fetch.- The active result retains its execution mode until reset. Switching between- this module and the default @Database.DuckDB.Simple.nextRow@ does not restart- or change an active result. EOF remains exhausted until an explicit reset.--}-nextRowWith :: RowParser r -> Statement -> IO (Maybe r)-nextRowWith = Result.nextRowWith StreamingResult--{- | Fold Arrow batches, requesting native streaming.- Each batch has a separate schema and array. The callback may read, release,- or move them under the Arrow C Data Interface. The pointers themselves are- valid during the callback only. See @Database.DuckDB.Simple.Arrow@ for the- ownership rules.- Arrow conversion uses the supported API; query execution uses the deprecated- streaming entry point. The fold releases contents that the consumer has not- released or moved. The consumer owns any moved contents.--}-foldArrow :: (ToRow params) => Connection -> Query -> params -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a-foldArrow = Arrow.foldArrowWith StreamingResult---- | Fold Arrow batches from a query without parameters.-foldArrow_ :: Connection -> Query -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a-foldArrow_ conn sql = foldArrow conn sql ()
src/Database/DuckDB/Simple/Element.hs view
@@ -1,228 +1,316 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE MagicHash #-} {-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE TypeApplications #-} -{- | Decoders for one native DuckDB element in vector memory. Result decoding-uses them for the types whose element needs no logical type metadata.--}+-- |+-- Module : Database.DuckDB.Simple.Element+-- Description : Decoders for one native DuckDB element in vector memory.+--+-- Result decoding uses them for the types whose element needs no logical type+-- metadata. module Database.DuckDB.Simple.Element (- chunkIsRowValid,- bitStringFromBytes,- chunkDecodeBlob,- decodeElement,- duckDBHugeIntToInteger,- vectorElementType,- withVectorType,+ chunkIsRowValid,+ containsType,+ decodeElement,+ duckDBHugeIntToInteger,+ timeTzMaxOffsetSeconds,+ timeTzOffsetBits,+ valueTypeName,+ vectorElementType,+ withVectorType, ) where -import Control.Exception (bracket, throwIO)-import Control.Monad (when)-import Data.Bits (clearBit, shiftL, xor, (.|.))+import Control.Exception (bracket, throwIO, try)+import Data.Bits (clearBit, shiftL, shiftR, testBit, xor, (.&.), (.|.)) import qualified Data.ByteString as BS-import Data.Int (Int64)+import Data.Coerce (coerce)+import Data.Int (Int32, Int64) import Data.Ratio ((%))+import Data.Text (Text)+import qualified Data.Text as Text import qualified Data.Text.Encoding as TextEncoding import Data.Time.Calendar (addDays, fromGregorian) import Data.Time.Clock.POSIX (posixSecondsToUTCTime)-import Data.Time.LocalTime (TimeOfDay (..), minutesToTimeZone, utc, utcToLocalTime)+import Data.Time.LocalTime (TimeOfDay (..), utc, utcToLocalTime) import qualified Data.UUID as UUID+import Data.Void (Void) import Data.Word (Word64, Word8)-import Database.DuckDB.FFI import Database.DuckDB.Simple.FromField (- BigNum (..),- BitString (..),- FieldValue (..),- IntervalValue (..),- TimeWithZone (..),- fromBigNumBytes,+ BigNum (..),+ BitString (..),+ FieldValue (..),+ IntervalValue (..),+ TimeWithZone (..),+ fieldTypeName, )-import Database.DuckDB.Simple.LogicalRep (destroyLogicalType)+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructField (..), UnionMemberType (..))+import Database.DuckDB.Simple.Native import Database.DuckDB.Simple.Time (Date, LocalTimestamp, UTCTimestamp, Unbounded (..))-import Foreign.C.Types (CBool (..))-import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (Ptr, castPtr, nullPtr, plusPtr)+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Marshal.Alloc (alloca, allocaBytes)+import Foreign.Marshal.Array (advancePtr)+import Foreign.Ptr (Ptr, castPtr, nullPtr) import Foreign.Storable (Storable (..), peekElemOff)+import GHC.Exts (Ptr (..), Word (W#))+import GHC.Num.Integer (integerFromAddr)+import GHC.Records (getField) -chunkIsRowValid :: Ptr Word64 -> DuckDBIdx -> IO Bool-chunkIsRowValid validity rowIdx- | validity == nullPtr = pure True- | otherwise = do- CBool flag <- c_duckdb_validity_row_is_valid validity rowIdx- pure (flag /= 0)+{- Note [Vector memory layouts]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+Some readers and writers use the memory of a vector directly. These are the+layouts that they use. -chunkDecodeBlob :: Ptr () -> DuckDBIdx -> IO BS.ByteString-chunkDecodeBlob dataPtr rowIdx = do- let base = castPtr dataPtr :: Ptr Word8- offset = fromIntegral rowIdx * duckdbStringTSize- stringPtr = castPtr (base `plusPtr` offset) :: Ptr DuckDBStringT- len <- c_duckdb_string_t_length stringPtr- if len == 0- then pure BS.empty- else do- ptr <- c_duckdb_string_t_data stringPtr- BS.packCStringLen (ptr, fromIntegral len)+The validity mask. Each uint64_t word covers 64 rows: bit N of word W is row+W*64+N, and a set bit means valid. The header gives this layout in the comment+of duckdb_v2_vector_flat_get_validity_mutable. -duckdbStringTSize :: Int-duckdbStringTSize = 16+- Read (chunkIsRowValid). The V2 API has no call that reads the validity of+ one row, so the reader reads the mask memory from duckdb_v2_vector_get_view.+ The header does not say what a null mask pointer means. In DuckDB 2.0.0 a+ null pointer means that the vector has no mask, and then every row is valid. -{- | Decode DuckDB's BIT bytes: a padding count, then the data bytes. Clear the-unused high bits of the first data byte, which DuckDB sets.+- Write (markValid and markInvalid). The writer sets or clears one bit of the+ mask from duckdb_v2_vector_flat_get_validity_mutable. The header says that+ this call allocates the mask on first use, so the pointer is never null. The+ header also says that a raw mask write marks only the rows of this vector,+ and that a NULL STRUCT or ARRAY row needs duckdb_v2_vector_set_null. The+ writer writes a mask bit only for a raw result type. A NULL of each other+ result type goes through duckdb_v2_vector_set_value. See+ Note [Result writers] in "Database.DuckDB.Simple.Function.Internal".++The bytes slot. A VARCHAR, BLOB, BIT, or BIGNUM element is a 16-byte+duckdb_v2_bytes value. The byte length is a uint32_t at byte 0. A value of at+most DUCKDB_V2_BYTES_INLINE_LENGTH (12) bytes follows the length, at byte 4.+For a longer value, the slot holds the first 4 bytes of the value at byte 4 and+the address of the value at byte 8. The header defines the slot in the comment+of struct duckdb_v2_bytes.++- Read (chunkDecodeBlob). The reader reads the generated fields of the union.++- Write (writeArenaText). The writer needs no factory. It writes the length and+ the bytes through the generated field pointers, as the reader reads them. A+ longer value goes to the arena of the vector.++The storage of a DECIMAL or ENUM element depends on the type (decimalStorage and+enumStorage). The header of duckdb_v2_vector_get_view gives both mappings. -}++-- | Read a validity bit of a mask from @duckdb_v2_vector_get_view@. A null+-- mask pointer gives 'True' for each row. See Note [Vector memory layouts].+chunkIsRowValid :: Ptr Word64 -> DuckDBIdx -> IO Bool+chunkIsRowValid validity row+ | validity == nullPtr = pure True+ | otherwise = (\word -> testBit word (fromIntegral (row `rem` 64))) <$> peekElemOff validity (fromIntegral (row `div` 64))++-- | Copy one generated V2 byte value, including interior NUL bytes. See+-- Note [Vector memory layouts].+chunkDecodeBlob :: Ptr Void -> DuckDBIdx -> IO BS.ByteString+chunkDecodeBlob dataPtr row = do+ -- Read each field through its generated field pointer. A getField on the+ -- union value calls the bindings package. That call copies the union.+ let ptr = advancePtr @DuckDBV2Bytes (castPtr dataPtr) (fromIntegral row)+ value = getField @"value" ptr+ len <- peek (getField @"length" (getField @"inlined" value))+ source <-+ if len <= fromIntegral dUCKDB_V2_BYTES_INLINE_LENGTH+ then pure (coerce (getField @"inlined" (getField @"inlined" value)))+ else peek (getField @"ptr" (getField @"pointer" value))+ if len == 0 then pure BS.empty else BS.packCStringLen (source, fromIntegral len)++-- | Decode DuckDB's BIT bytes: a padding count, then the data bytes. Clear the+-- unused high bits of the first data byte, which DuckDB sets. bitStringFromBytes :: BS.ByteString -> BitString bitStringFromBytes bytes = case BS.unpack bytes of- [] -> BitString 0 BS.empty- [padding] -> BitString padding BS.empty- padding : first : rest ->- BitString padding (BS.pack (foldl clearBit first [8 - fromIntegral padding .. 7] : rest))+ [] -> BitString 0 BS.empty+ [padding] -> BitString padding BS.empty+ padding : first : rest ->+ BitString padding (BS.pack (foldl clearBit first [8 - fromIntegral padding .. 7] : rest)) -- | Borrow the logical type of a vector. The type is destroyed after the action.-withVectorType :: DuckDBVector -> (DuckDBLogicalType -> IO a) -> IO a-withVectorType vector = bracket (c_duckdb_vector_get_column_type vector) destroyLogicalType+withVectorType :: DuckDBV2VectorHandle -> (DuckDBV2LogicalTypeHandle -> IO a) -> IO a+withVectorType vector = bracket (output (c_duckdb_v2_vector_get_logical_type vector)) destroyLogicalType -vectorElementType :: DuckDBVector -> IO DuckDBType-vectorElementType vec =- bracket (c_duckdb_vector_get_column_type vec) destroyLogicalType c_duckdb_get_type_id+-- | Read the native logical type id of a vector.+vectorElementType :: DuckDBV2VectorHandle -> IO DuckDBV2LogicalTypeId+vectorElementType vector = withVectorType vector (output . c_duckdb_v2_logical_type_get_id) -{- | Decode the element at an index of a vector's data. The type must not need-logical type metadata: DECIMAL, ENUM, GEOMETRY, and nested types use their own-decoders. The caller checks validity.--}-decodeElement :: DuckDBType -> Ptr () -> Int -> IO FieldValue+-- | Decode the element at an index of a vector's data. The type must not need+-- logical type metadata: DECIMAL, ENUM, GEOMETRY, and nested types use their own+-- decoders. The caller checks validity.+decodeElement :: DuckDBV2LogicalTypeId -> Ptr Void -> Int -> IO FieldValue decodeElement dtype dataPtr rowIdx = case dtype of- DuckDBTypeBoolean -> do- raw <- peekElemOff (castPtr dataPtr :: Ptr Word8) rowIdx- pure (FieldBool (raw /= 0))- DuckDBTypeTinyInt -> FieldInt8 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeSmallInt -> FieldInt16 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeInteger -> FieldInt32 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeBigInt -> FieldInt64 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeUTinyInt -> FieldWord8 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeUSmallInt -> FieldWord16 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeUInteger -> FieldWord32 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeUBigInt -> FieldWord64 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeFloat -> FieldFloat <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeDouble -> FieldDouble <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeVarchar -> FieldText . TextEncoding.decodeUtf8 <$> chunkDecodeBlob dataPtr index- DuckDBTypeStringLiteral -> FieldText . TextEncoding.decodeUtf8 <$> chunkDecodeBlob dataPtr index- DuckDBTypeBlob -> FieldBlob <$> chunkDecodeBlob dataPtr index- DuckDBTypeUUID -> do- DuckDBUHugeInt lower upperBiased <- peekElemOff (castPtr dataPtr :: Ptr DuckDBUHugeInt) rowIdx- let upper = upperBiased `xor` (0x8000000000000000 :: Word64)- pure (FieldUUID (UUID.fromWords64 (fromIntegral upper) lower))- DuckDBTypeDate -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldDate . decodeDuckDBDate- DuckDBTypeTime -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTime . decodeDuckDBTime- DuckDBTypeTimeNs -> FieldTime . decodeDuckDBTimeNs <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeTimeTz -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimeTZ . decodeDuckDBTimeTz- DuckDBTypeTimestamp -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestamp . decodeDuckDBTimestamp- DuckDBTypeTimestampS -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestamp . decodeDuckDBTimestampSeconds- DuckDBTypeTimestampMs -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestamp . decodeDuckDBTimestampMilliseconds- DuckDBTypeTimestampNs -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestamp . decodeDuckDBTimestampNanoseconds- DuckDBTypeTimestampTz -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestampTZ . decodeDuckDBTimestampUTCTime- DuckDBTypeInterval -> FieldInterval . intervalValueFromDuckDB <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeHugeInt -> FieldHugeInt . duckDBHugeIntToInteger <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeUHugeInt -> FieldUHugeInt . duckDBUHugeIntToInteger <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeBit -> FieldBit . bitStringFromBytes <$> chunkDecodeBlob dataPtr index- DuckDBTypeBigNum -> do- bytes <- chunkDecodeBlob dataPtr index- pure (FieldBigNum (BigNum (if BS.length bytes < 3 then 0 else fromBigNumBytes (BS.unpack bytes))))- DuckDBTypeIntegerLiteral -> FieldInt64 <$> peekElemOff (castPtr dataPtr) rowIdx- DuckDBTypeInvalid ->- error "duckdb-simple: INVALID type in eager result"- DuckDBTypeAny ->- error "duckdb-simple: ANY columns should not appear in results"- other ->- error ("duckdb-simple: UNKNOWN type in eager result: " <> show other)+ DuckDBV2LogicalTypeIdBoolean -> do+ raw <- peekElemOff (castPtr dataPtr :: Ptr Word8) rowIdx+ pure (FieldBool (raw /= 0))+ DuckDBV2LogicalTypeIdTinyInt -> FieldInt8 <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdSmallInt -> FieldInt16 <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdInteger -> FieldInt32 <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdBigInt -> FieldInt64 <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdUTinyInt -> FieldWord8 <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdUSmallInt -> FieldWord16 <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdUInteger -> FieldWord32 <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdUBigInt -> FieldWord64 <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdFloat -> FieldFloat <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdDouble -> FieldDouble <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdVarchar -> FieldText . TextEncoding.decodeUtf8 <$> chunkDecodeBlob dataPtr index+ DuckDBV2LogicalTypeIdBlob -> FieldBlob <$> chunkDecodeBlob dataPtr index+ DuckDBV2LogicalTypeIdUUID -> do+ DuckDBV2HugeintT lower upperBiased <- peekElemOff (castPtr dataPtr :: Ptr DuckDBV2HugeintT) rowIdx+ let upper = fromIntegral upperBiased `xor` (0x8000000000000000 :: Word64)+ pure (FieldUUID (UUID.fromWords64 (fromIntegral upper) lower))+ DuckDBV2LogicalTypeIdDate -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldDate . decodeDuckDBDate+ DuckDBV2LogicalTypeIdTime -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTime . decodeDuckDBTime+ DuckDBV2LogicalTypeIdTimeNs -> FieldTime . decodeDuckDBTimeNs <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdTimeTz -> FieldTimeTZ . decodeDuckDBTimeTz <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdTimestamp -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestamp . decodeDuckDBTimestamp+ DuckDBV2LogicalTypeIdTimestampSec -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestamp . decodeDuckDBTimestampSeconds+ DuckDBV2LogicalTypeIdTimestampMs -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestamp . decodeDuckDBTimestampMilliseconds+ DuckDBV2LogicalTypeIdTimestampNs -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestamp . decodeDuckDBTimestampNanoseconds+ DuckDBV2LogicalTypeIdTimestampTz -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestampTZ . decodeDuckDBTimestampUTCTime+ DuckDBV2LogicalTypeIdTimestampTzNs -> peekElemOff (castPtr dataPtr) rowIdx >>= fmap FieldTimestampTZ . decodeDuckDBTimestampNsUTCTime+ DuckDBV2LogicalTypeIdInterval -> FieldInterval . intervalValueFromDuckDB <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdHugeInt -> FieldHugeInt . duckDBHugeIntToInteger <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdUHugeInt -> FieldUHugeInt . duckDBUHugeIntToInteger <$> peekElemOff (castPtr dataPtr) rowIdx+ DuckDBV2LogicalTypeIdBit -> FieldBit . bitStringFromBytes <$> chunkDecodeBlob dataPtr index+ DuckDBV2LogicalTypeIdBigNum -> do+ bytes <- chunkDecodeBlob dataPtr index+ FieldBigNum . BigNum <$> decodeBigNum bytes+ DuckDBV2LogicalTypeIdInvalid ->+ throwInternalError (Text.pack "INVALID type in eager result")+ DuckDBV2LogicalTypeIdAny ->+ throwInternalError (Text.pack "ANY columns should not appear in results")+ other ->+ throwInternalError (Text.pack ("UNKNOWN type in eager result: " <> show other)) where index = fromIntegral rowIdx :: DuckDBIdx -- | Decode dates with exact epoch arithmetic and preserve infinity.-decodeDuckDBDate :: DuckDBDate -> IO Date-decodeDuckDBDate (DuckDBDate days) =- pure (decodeUnbounded (\value -> addDays (toInteger value) (fromGregorian 1970 1 1)) days)+decodeDuckDBDate :: Int32 -> IO Date+decodeDuckDBDate = pure . decodeUnbounded (\value -> addDays (toInteger value) (fromGregorian 1970 1 1)) -decodeDuckDBTime :: DuckDBTime -> IO TimeOfDay-decodeDuckDBTime raw =- alloca $ \ptr -> do- c_duckdb_from_time raw ptr- timeStruct <- peek ptr- pure (timeStructToTimeOfDay timeStruct)+decodeDuckDBTime :: Int64 -> IO TimeOfDay+decodeDuckDBTime = pure . decodeTimeUnits 1000000 -decodeDuckDBTimestamp :: DuckDBTimestamp -> IO LocalTimestamp-decodeDuckDBTimestamp (DuckDBTimestamp micros) = decodeTimestampUnits 1000000 micros+decodeDuckDBTimestamp :: Int64 -> IO LocalTimestamp+decodeDuckDBTimestamp = decodeTimestampUnits 1000000 -- | Interpret native infinity sentinels before converting a finite payload. decodeUnbounded :: (Integral a, Bounded a) => (a -> b) -> a -> Unbounded b decodeUnbounded decode value- | value == maxBound = PosInfinity- | value == negate maxBound = NegInfinity- | otherwise = Finite (decode value)+ | value == maxBound = PosInfinity+ | value == negate maxBound = NegInfinity+ | otherwise = Finite (decode value) -- | Decode timestamp units without overflowing an intermediate Int64. decodeTimestampUnits :: Integer -> Int64 -> IO LocalTimestamp-decodeTimestampUnits units =- pure . decodeUnbounded (utcToLocalTime utc . posixSecondsToUTCTime . fromRational . (% units) . toInteger)+decodeTimestampUnits units = pure . decodeUnbounded (utcToLocalTime utc . posixSecondsToUTCTime . fromRational . (% units) . toInteger) -decodeDuckDBTimeNs :: DuckDBTimeNs -> TimeOfDay-decodeDuckDBTimeNs (DuckDBTimeNs nanos) =- let (hours, remainderHours) = nanos `divMod` (60 * 60 * 1000000000)- (minutes, remainderMinutes) = remainderHours `divMod` (60 * 1000000000)- (seconds, fractionalNanos) = remainderMinutes `divMod` 1000000000- fractional = fromRational (toInteger fractionalNanos % 1000000000)- totalSeconds = fromIntegral seconds + fractional- in TimeOfDay- (fromIntegral hours)- (fromIntegral minutes)- totalSeconds+-- | Decode raw time units with exact fractional seconds.+decodeTimeUnits :: Int64 -> Int64 -> TimeOfDay+decodeTimeUnits units value =+ let (hours, remHours) = value `divMod` (3600 * units)+ (minutes, remMinutes) = remHours `divMod` (60 * units)+ in TimeOfDay (fromIntegral hours) (fromIntegral minutes) (fromRational (toInteger remMinutes % toInteger units)) -decodeDuckDBTimeTz :: DuckDBTimeTz -> IO TimeWithZone+decodeDuckDBTimeNs :: Int64 -> TimeOfDay+decodeDuckDBTimeNs = decodeTimeUnits 1000000000++-- | The largest UTC offset of a TIMETZ value in seconds: 15:59:59. DuckDB 2.0.0+-- defines it as @dtime_tz_t::MAX_OFFSET@ in+-- @src/include/duckdb/common/types/datetime.hpp@. The low 24 bits of a TIMETZ+-- value hold this maximum minus the offset (@encode_offset@ and+-- @decode_offset@ in that file). The V2 header says only "micros in the high+-- 40 bits, a biased offset in the low 24" (the comment of+-- @duckdb_v2_value_get_time_tz@). The header does not give the value of the+-- bias.+timeTzMaxOffsetSeconds :: Int+timeTzMaxOffsetSeconds = 16 * 60 * 60 - 1++-- | The number of low bits of a TIMETZ value that hold the biased offset. The+-- high 40 bits hold the time of day in microseconds. DuckDB 2.0.0 names the two+-- widths @dtime_tz_t::OFFSET_BITS@ and @dtime_tz_t::TIME_BITS@.+timeTzOffsetBits :: Int+timeTzOffsetBits = 24++-- | Decode the time of day and the offset in seconds of a TIMETZ value.+decodeDuckDBTimeTz :: Word64 -> TimeWithZone decodeDuckDBTimeTz raw =- alloca $ \ptr -> do- c_duckdb_from_time_tz raw ptr- DuckDBTimeTzStruct{duckDBTimeTzStructTime = timeStruct, duckDBTimeTzStructOffset = offset} <- peek ptr- when (offset `rem` 60 /= 0) $- throwIO (userError "duckdb-simple: TIMETZ offset cannot be represented in whole minutes")- let timeOfDay = timeStructToTimeOfDay timeStruct- minutes = fromIntegral offset `div` 60- zone = minutesToTimeZone minutes- pure TimeWithZone{timeWithZoneTime = timeOfDay, timeWithZoneZone = zone}+ TimeWithZone+ { timeWithZoneTime = decodeTimeUnits 1000000 (fromIntegral (raw `shiftR` timeTzOffsetBits))+ , timeWithZoneOffsetSeconds = fromIntegral (timeTzMaxOffsetSeconds - fromIntegral (raw .&. (1 `shiftL` timeTzOffsetBits - 1)))+ } -decodeDuckDBTimestampSeconds :: DuckDBTimestampS -> IO LocalTimestamp-decodeDuckDBTimestampSeconds (DuckDBTimestampS seconds) =- decodeTimestampUnits 1 seconds+decodeDuckDBTimestampSeconds :: Int64 -> IO LocalTimestamp+decodeDuckDBTimestampSeconds = decodeTimestampUnits 1 -decodeDuckDBTimestampMilliseconds :: DuckDBTimestampMs -> IO LocalTimestamp-decodeDuckDBTimestampMilliseconds (DuckDBTimestampMs millis) =- decodeTimestampUnits 1000 millis+decodeDuckDBTimestampMilliseconds :: Int64 -> IO LocalTimestamp+decodeDuckDBTimestampMilliseconds = decodeTimestampUnits 1000 -decodeDuckDBTimestampNanoseconds :: DuckDBTimestampNs -> IO LocalTimestamp-decodeDuckDBTimestampNanoseconds (DuckDBTimestampNs nanos) = decodeTimestampUnits 1000000000 nanos+decodeDuckDBTimestampNanoseconds :: Int64 -> IO LocalTimestamp+decodeDuckDBTimestampNanoseconds = decodeTimestampUnits 1000000000 -decodeDuckDBTimestampUTCTime :: DuckDBTimestamp -> IO UTCTimestamp-decodeDuckDBTimestampUTCTime (DuckDBTimestamp micros) =- pure (decodeUnbounded (posixSecondsToUTCTime . fromRational . (% 1000000) . toInteger) micros)+decodeDuckDBTimestampUTCTime :: Int64 -> IO UTCTimestamp+decodeDuckDBTimestampUTCTime = pure . decodeUnbounded (posixSecondsToUTCTime . fromRational . (% 1000000) . toInteger) -intervalValueFromDuckDB :: DuckDBInterval -> IntervalValue-intervalValueFromDuckDB DuckDBInterval{duckDBIntervalMonths, duckDBIntervalDays, duckDBIntervalMicros} =- IntervalValue- { intervalMonths = duckDBIntervalMonths- , intervalDays = duckDBIntervalDays- , intervalMicros = duckDBIntervalMicros- }+-- | Decode UTC nanoseconds and preserve infinity.+decodeDuckDBTimestampNsUTCTime :: Int64 -> IO UTCTimestamp+decodeDuckDBTimestampNsUTCTime = pure . decodeUnbounded (posixSecondsToUTCTime . fromRational . (% 1000000000) . toInteger) -duckDBHugeIntToInteger :: DuckDBHugeInt -> Integer-duckDBHugeIntToInteger DuckDBHugeInt{duckDBHugeIntLower, duckDBHugeIntUpper} =- (fromIntegral duckDBHugeIntUpper `shiftL` 64) .|. fromIntegral duckDBHugeIntLower+intervalValueFromDuckDB :: DuckDBV2IntervalT -> IntervalValue+intervalValueFromDuckDB (DuckDBV2IntervalT months days micros) = IntervalValue months days micros -duckDBUHugeIntToInteger :: DuckDBUHugeInt -> Integer-duckDBUHugeIntToInteger DuckDBUHugeInt{duckDBUHugeIntLower, duckDBUHugeIntUpper} =- (fromIntegral duckDBUHugeIntUpper `shiftL` 64) .|. fromIntegral duckDBUHugeIntLower+duckDBHugeIntToInteger :: DuckDBV2HugeintT -> Integer+duckDBHugeIntToInteger (DuckDBV2HugeintT lower upper) = (fromIntegral upper `shiftL` 64) .|. fromIntegral lower -timeStructToTimeOfDay :: DuckDBTimeStruct -> TimeOfDay-timeStructToTimeOfDay DuckDBTimeStruct{duckDBTimeStructHour, duckDBTimeStructMinute, duckDBTimeStructSecond, duckDBTimeStructMicros} =- let secondsInt = fromIntegral duckDBTimeStructSecond :: Integer- micros = fromIntegral duckDBTimeStructMicros :: Integer- fractional = fromRational (micros % 1000000)- totalSeconds = fromInteger secondsInt + fractional- in TimeOfDay- (fromIntegral duckDBTimeStructHour)- (fromIntegral duckDBTimeStructMinute)- totalSeconds+duckDBUHugeIntToInteger :: DuckDBV2UhugeintT -> Integer+duckDBUHugeIntToInteger (DuckDBV2UhugeintT lower upper) = (fromIntegral upper `shiftL` 64) .|. fromIntegral lower++-- | Decode opaque BIGNUM storage through the V2 magnitude interface.+decodeBigNum :: BS.ByteString -> IO Integer+decodeBigNum bytes = BS.useAsCStringLen bytes \(raw, len) -> alloca \lengthSlot -> alloca \signSlot -> do+ let decode ptr capacity = checked (c_duckdb_v2_bignum_decode (ConstPtr (castPtr raw)) (fromIntegral len) ptr capacity lengthSlot signSlot)+ finish magnitude = do+ count <- peek lengthSlot+ negative <- peek signSlot+ value <- bigEndianInteger magnitude count+ pure $! if negative == 0 then value else negate value+ -- A buffer of the storage size usually holds the magnitude. If the buffer is+ -- too short, DuckDB gives the required size, and a second call decodes the value.+ allocaBytes len \first -> do+ outcome <- try (decode first (fromIntegral len))+ case outcome of+ Right () -> finish first+ Left err+ | sqlErrorType err == Just ErrorInputObjectSize -> do+ count <- peek lengthSlot+ allocaBytes (fromIntegral count) \magnitude -> decode magnitude count >> finish magnitude+ | otherwise -> throwIO err++-- | Read an unsigned big-endian magnitude in linear time.+bigEndianInteger :: Ptr Word8 -> DuckDBIdx -> IO Integer+bigEndianInteger (Ptr addr) count = case fromIntegral count of W# size -> integerFromAddr size addr 1#++-- | The SQL type name of a decoded value. Errors name the type because a+-- value can be large.+valueTypeName :: FieldValue -> Text+valueTypeName = fieldTypeName (LogicalTypeScalar LogicalTypeIdVariant)++-- | Test whether a type is or contains a type that the predicate accepts. The+-- test visits the element type of a LIST and an ARRAY, the key type and the+-- value type of a MAP, the field types of a STRUCT and a TUPLE, the member+-- types of a UNION, and the base type of an alias. See Note [Type aliases] in+-- "Database.DuckDB.Simple.LogicalRep".+containsType :: (LogicalTypeRep -> Bool) -> LogicalTypeRep -> Bool+containsType accepts rep =+ accepts rep || case rep of+ LogicalTypeAlias _ base -> containsType accepts base+ LogicalTypeList child -> containsType accepts child+ LogicalTypeArray child _ -> containsType accepts child+ LogicalTypeMap key value -> containsType accepts key || containsType accepts value+ LogicalTypeStruct fields -> any (containsType accepts . structFieldValue) fields+ LogicalTypeTuple fields -> any (containsType accepts) fields+ LogicalTypeUnion members -> any (containsType accepts . unionMemberType) members+ _ -> False
+ src/Database/DuckDB/Simple/Extension.hs view
@@ -0,0 +1,260 @@+{-# LANGUAGE BlockArguments #-}++-- |+-- Module : Database.DuckDB.Simple.Extension+-- Description : Entry point and typed registrations for a loadable Haskell extension.+--+-- A loadable extension is a shared library that DuckDB loads into its own process with @LOAD@. DuckDB calls the+-- entry point of the extension one time in each database instance. A second @LOAD@ of the same name in that+-- instance does not call it again.+-- The entry point registers SQL objects, and DuckDB keeps them in the database after the entry point returns. This+-- module gives the entry point and the typed registrations of "Database.DuckDB.Simple.Function" and of the other+-- typed modules for an extension.+--+-- An extension needs these items:+--+-- - The @extension@ flag of duckdb-ffi and of duckdb-simple. Enable the two flags together. A flag applies to the+-- complete build plan, so one project cannot build a DuckDB host and an extension. With the flag, the two+-- packages build only on Linux and macOS.+-- - A C entry point @NAME_init_c_api_v2@ that calls @duckdb_haskell_extension_load@ with the foreign export of an+-- 'ExtensionEntry' and does nothing else. DuckDB finds the entry point from the file name @NAME.duckdb_extension@.+-- - The threaded runtime. Link the extension with @-threaded@. Otherwise the load fails with an error.+--+-- See @dev\/extensions\/README.md@ in this repo.+--+-- The Haskell side of a small extension, from @dev\/extensions\/example\/src\/Demo.hs@:+--+-- > {-# LANGUAGE ForeignFunctionInterface #-}+-- > {-# LANGUAGE OverloadedStrings #-}+-- >+-- > module Demo where+-- >+-- > import Data.Int (Int64)+-- > import Data.Text (Text)+-- > import Database.DuckDB.Simple.Extension (Extension, ExtensionEntry)+-- > import qualified Database.DuckDB.Simple.Extension as Extension+-- >+-- > initialize :: ExtensionEntry+-- > initialize = Extension.runExtension register+-- >+-- > foreign export ccall "duckdb_haskell_demo_register" initialize :: ExtensionEntry+-- >+-- > register :: Extension -> IO ()+-- > register extension = do+-- > Extension.createFunction extension "haskell_answer" (42 :: Int64)+-- > Extension.createFunction extension "haskell_greet" (("Hello " <>) :: Text -> Text)+-- > Extension.registerMacroFunction extension "haskell_table" Extension.MacroFunctionTypeTable [] "SELECT 42::BIGINT AS answer"+--+-- The C entry point of the same extension, from @dev\/extensions\/example\/cbits\/entry.c@. The packaged file must+-- be @haskell_demo.duckdb_extension@:+--+-- > #include "duckdb_haskell_extension.h"+-- >+-- > extern void duckdb_haskell_demo_register(void *extension, void *context, void *err);+-- >+-- > /* DuckDB derives this name from the file name of the packaged extension. */+-- > DUCKDB_EXTENSION_API void haskell_demo_init_c_api_v2(struct duckdb_v2_extension_input *input) {+-- > duckdb_haskell_extension_load(input, duckdb_haskell_demo_register);+-- > }+--+-- The t'Extension' value is valid only while 'runExtension' runs. DuckDB gives a borrowed extension handle and a+-- borrowed context to the entry point. They are not valid after the entry point returns. Thus do not keep the+-- t'Extension' value, or the target of 'extensionTarget', in a registered callback or in other state. A registered+-- function keeps its callback resources after the entry point returns.+--+-- The README of the example gives these limits:+--+-- - The runtime has one capability (@-N1@). The callbacks of all DuckDB worker threads share it. You cannot change+-- the runtime options.+-- - Haskell extensions built with the same GHC share one runtime in the process, and the options of the first+-- loaded extension apply. Extensions built with different GHC versions in one process are not tested. The runtime+-- stays until the process ends. Do not call @hs_exit@.+-- - The process supports one DuckDB implementation.+-- - The extension loads only into the DuckDB release of the pinned header, 2.0.0.+-- - In extension mode, duckdb-ffi exposes only "Database.DuckDB.FFI.V2".+-- - A replacement scan of an extension belongs to the instance, not to one connection.+-- - The example links the GHC and package libraries dynamically. It loads only on a machine where its run paths+-- exist.+--+-- The complete example is in the directory @dev\/extensions\/example\/@ of the repository.+-- Its file @src\/Advanced.hs@ registers the other kinds of SQL objects.+module Database.DuckDB.Simple.Extension (+ Extension,+ ExtensionEntry,+ runExtension,+ extensionTarget,+ createFunction,+ createFunctionWithState,+ registerReplacementScan,+ registerCustomType,+ registerMultiFileFunction,+ createTableFunction,+ createAggregate,+ MacroFunctionType (..),+ registerMacroFunction,+ registerPragmaFunction,+ registerCustomOption,+ registerLogSink,+ registerCopyToFunction,+ registerCopyFromFunction,+ registerCopyFunction,+) where++import Data.Text (Text)+import Database.DuckDB.Simple.Callback (runCallback)+import Database.DuckDB.Simple.Config.Internal (CustomOption)+import qualified Database.DuckDB.Simple.Config.Internal as Config+import Database.DuckDB.Simple.Copy.Internal (+ CopyBindInfo,+ CopyFinalizeInfo,+ CopyFromInfo,+ CopyInitInfo,+ CopySinkInfo,+ CopyTo,+ MultiFileOptions,+ registerMultiFileFunctionOn,+ )+import qualified Database.DuckDB.Simple.Copy.Internal as Copy+import Database.DuckDB.Simple.FromField (Field)+import Database.DuckDB.Simple.FromRow (FromRow)+import Database.DuckDB.Simple.Function.Internal (Function, FunctionResult, createFunctionOn, createFunctionWithStateOn)+import Database.DuckDB.Simple.Logging.Internal (LogEntry)+import qualified Database.DuckDB.Simple.Logging.Internal as Logging+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.Pragma.Internal (PragmaCallback)+import qualified Database.DuckDB.Simple.Pragma.Internal as Pragma+import Database.DuckDB.Simple.Typed (Replacement, TypedRow, replacementScanCallback)+import Database.DuckDB.Simple.Typed.Aggregate (Aggregate, createAggregateOn)+import Database.DuckDB.Simple.Typed.Table (createTableFunctionOn)+import Database.DuckDB.Simple.V2.Registration (+ FunctionParameter,+ FunctionSignature (..),+ RegistrationTarget (ExtensionTarget),+ TypeDeclaration,+ configureFunctionSignature,+ registerCustomTypeOn,+ registerMacroFunctionWith,+ registerReplacementScanOn,+ )+import Foreign.Ptr (Ptr, castPtr)++-- | The borrowed registration target and context for one extension load.+-- Use this value only inside the action passed to 'runExtension'. Do not retain it+-- in registered callbacks or application state.+data Extension = Extension DuckDBV2ExtensionHandle DuckDBV2ContextHandle++-- | The Haskell entry that @duckdb_haskell_extension_load@ calls. Its arguments+-- are the extension handle, the borrowed context, and the error slot of the load.+-- The plain pointers let a @foreign export@ use this type with no other import.+type ExtensionEntry = Ptr () -> Ptr () -> Ptr () -> IO ()++-- | Run registration and report exceptions through the borrowed load error slot.+-- The extension is valid only during this call. Registered functions own their+-- callback resources after this call returns.+runExtension :: (Extension -> IO ()) -> ExtensionEntry+runExtension initialize extension context err =+ runCallback (castPtr err) (initialize (Extension (DuckDBV2ExtensionHandle (castPtr extension)) (DuckDBV2ContextHandle (castPtr context))))++-- | The registration target of the loading extension. Pass it to the+-- registration functions of the V2 modules, which take a 'RegistrationTarget'.+-- The typed functions of this module do not need it. The target is valid only+-- during 'runExtension'.+extensionTarget :: Extension -> RegistrationTarget+extensionTarget (Extension extension context) = ExtensionTarget extension context++-- | Register a typed Haskell function with the loading DuckDB host. The rules+-- of 'Database.DuckDB.Simple.Function.createFunction' apply.+createFunction :: (Function f) => Extension -> Text -> f -> IO ()+createFunction = createFunctionOn . extensionTarget++-- | Register a typed function with per-worker state owned by DuckDB. The rules+-- of 'Database.DuckDB.Simple.Function.createFunctionWithState' apply.+createFunctionWithState :: (Function f) => Extension -> Text -> IO s -> (s -> f) -> IO ()+createFunctionWithState = createFunctionWithStateOn . extensionTarget++-- | Register a replacement scan with the loading DuckDB host. The scan belongs+-- to the instance. Each connection of the instance uses it until the instance+-- closes. No function removes a scan. DuckDB asks the scans of a connection+-- before the scans of the instance. It asks the scans of the instance in the+-- order of registration. The function can run on several connections at the same+-- time. 'Database.DuckDB.Simple.Registration.registerReplacementScan' gives the+-- rules for the function.+registerReplacementScan :: Extension -> ([Text] -> IO (Maybe Replacement)) -> IO ()+registerReplacementScan extension = registerReplacementScanOn (extensionTarget extension) . replacementScanCallback++-- | Register a custom type with the loading DuckDB host. The rules of+-- 'Database.DuckDB.Simple.Registration.registerCustomType' apply.+registerCustomType :: Extension -> Text -> TypeDeclaration -> IO ()+registerCustomType = registerCustomTypeOn . extensionTarget++-- | Register a multi-file table function with the loading DuckDB host. The+-- rules of 'Database.DuckDB.Simple.Copy.registerMultiFileFunction' apply.+registerMultiFileFunction :: Extension -> Text -> Text -> MultiFileOptions -> IO ()+registerMultiFileFunction = registerMultiFileFunctionOn . extensionTarget++-- | Register a typed table function with the loading DuckDB host. The rules of+-- 'Database.DuckDB.Simple.Function.createTableFunction' apply.+createTableFunction :: (FromRow params, TypedRow params, TypedRow row) => Extension -> Text -> [Text] -> (params -> IO [row]) -> IO ()+createTableFunction = createTableFunctionOn . extensionTarget++-- | Register an aggregate function with the loading DuckDB host. The rules of+-- 'Database.DuckDB.Simple.Function.createAggregate' apply. SQL type declarations+-- resolve on the loading host. Result encoders use an instance factory+-- for built-in types, with no catalog lookup on the host.+createAggregate :: (FromRow args, TypedRow args, FunctionResult result) => Extension -> Text -> Aggregate state args result -> IO ()+createAggregate = createAggregateOn . extensionTarget++-- | Register a scalar expression or a table SELECT as a macro of the loading+-- extension. The V2 API registers a macro only for an extension. Use SQL+-- @CREATE MACRO@ on a connection. Parameters must use the standard parameter+-- kind. Use @StructuredType (LogicalTypeScalar LogicalTypeIdAny)@ for an+-- untyped parameter and 'Database.DuckDB.Simple.Registration.defaultValue' for a+-- default. DuckDB validates the body and the signature. It resolves the body+-- when the macro runs.+registerMacroFunction :: Extension -> Text -> MacroFunctionType -> [FunctionParameter] -> Text -> IO ()+registerMacroFunction extension@(Extension _ context) name kind parameters body =+ withContextFactory context \factory ->+ registerMacroFunctionWith (extensionTarget extension) kind \function -> do+ withIdentifier name (checked . c_duckdb_v2_macro_function_set_name function)+ withText body (checked . c_duckdb_v2_macro_function_set_body function)+ signature <- output (c_duckdb_v2_macro_function_get_signature function)+ configureFunctionSignature factory (FunctionSignature parameters Nothing) signature++-- | Register a typed PRAGMA callback on the loading extension. The rules of+-- 'Database.DuckDB.Simple.Pragma.registerPragmaFunction' apply.+registerPragmaFunction :: (FromRow args) => Extension -> Text -> [FunctionParameter] -> PragmaCallback args -> IO ()+registerPragmaFunction = Pragma.registerPragmaFunctionOn . extensionTarget++-- | Register a typed option during extension initialization. The rules of+-- 'Database.DuckDB.Simple.Config.registerCustomOption' apply.+registerCustomOption :: Extension -> CustomOption -> IO ()+registerCustomOption = Config.registerCustomOptionOn . extensionTarget++-- | Register a log sink with callback resources owned by DuckDB.+-- Callbacks can run concurrently. They must not log or change logging settings.+-- The rules of 'Database.DuckDB.Simple.Logging.registerLogSink' apply.+registerLogSink :: Extension -> Text -> (LogEntry -> IO ()) -> IO () -> IO ()+registerLogSink = Logging.registerLogSinkOn . extensionTarget++-- | Register row-oriented COPY TO callbacks with the loading DuckDB host. The+-- rules of 'Database.DuckDB.Simple.Copy.registerCopyToFunction' apply.+registerCopyToFunction+ :: Extension+ -> Text+ -> (CopyBindInfo -> IO bindState)+ -> (CopyInitInfo bindState -> IO globalState)+ -> (CopySinkInfo bindState globalState -> [[Field]] -> IO ())+ -> (CopyFinalizeInfo bindState globalState -> IO ())+ -> IO ()+registerCopyToFunction = Copy.registerCopyToFunctionOn . extensionTarget++-- | Register the COPY FROM side of a format with the loading DuckDB host. The+-- rules of 'Database.DuckDB.Simple.Copy.registerCopyFromFunction' apply.+registerCopyFromFunction :: (TypedRow row) => Extension -> Text -> (CopyFromInfo -> IO [row]) -> IO ()+registerCopyFromFunction = Copy.registerCopyFromFunctionOn . extensionTarget++-- | Register a format with a COPY TO side and a COPY FROM side with the+-- loading DuckDB host. The rules of+-- 'Database.DuckDB.Simple.Copy.registerCopyFunction' apply.+registerCopyFunction :: (TypedRow row) => Extension -> Text -> CopyTo bindState globalState -> (CopyFromInfo -> IO [row]) -> IO ()+registerCopyFunction = Copy.registerCopyFunctionOn . extensionTarget
src/Database/DuckDB/Simple/FileSystem.hs view
@@ -1,175 +1,190 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-} -{- |-Module : Database.DuckDB.Simple.FileSystem-Description : High-level wrappers around DuckDB's file-system API.--}+-- |+-- Module : Database.DuckDB.Simple.FileSystem+-- Description : Checked file operations through the V2 file system. module Database.DuckDB.Simple.FileSystem (- withFileHandle,- readFileHandleChunk,- writeFileHandleBytes,- fileHandleTell,- fileHandleSize,- fileHandleSeek,- fileHandleSync,+ FileFlag (..),+ FileHandle,+ withFileHandle,+ readFileHandleChunk,+ writeFileHandleBytes,+ readFileHandleAt,+ writeFileHandleAt,+ fileHandleTell,+ fileHandleSize,+ fileHandleSeek,+ fileHandleSync,+ pathExists,+ pathIsRemote,+ expandPath,+ globPaths,+ movePath,+ removePath, ) where -import Control.Exception (bracket, mask_, throwIO)+import Control.Exception (bracket)+import Control.Monad (when) import qualified Data.ByteString as BS+import qualified Data.ByteString.Internal as BSI+import Data.IORef (modifyIORef', newIORef, readIORef) import Data.Int (Int64)-import Data.Text (Text) import qualified Data.Text as Text-import qualified Data.Text.Foreign as TextForeign-import Database.DuckDB.FFI-import Database.DuckDB.Simple.Internal (Connection, SQLError (..), peekUtf8CString, withClientContext)-import Foreign.Marshal.Alloc (alloca, free, mallocBytes)-import Foreign.Ptr (Ptr, castPtr, nullPtr)-import Foreign.Storable (peek, poke)+import Database.DuckDB.Simple.Callback (runCallback)+import Database.DuckDB.Simple.Internal (Connection, withConnectionHandle)+import Database.DuckDB.Simple.Native+import Foreign.C.ConstPtr (ConstPtr (..), unConstPtr)+import Foreign.Ptr (castPtr, freeHaskellFunPtr, nullPtr)+import Foreign.Storable (peek)+import HsBindgen.Runtime.Support.FunPtr (toFunPtr) --- | Open a file through DuckDB's file-system layer for the duration of an action.-withFileHandle :: Connection -> FilePath -> [DuckDBFileFlag] -> (DuckDBFileHandle -> IO a) -> IO a+-- | An open file. 'withFileHandle' gives it to one action. Do not use it after that action returns.+newtype FileHandle = FileHandle DuckDBV2FileHandle++-- | Open a file for one action. Release its handle on every exit path.+withFileHandle :: Connection -> FilePath -> [FileFlag] -> (FileHandle -> IO a) -> IO a withFileHandle conn path flags action- | '\0' `elem` path = throwIO (SQLError (Text.pack "duckdb-simple: file path contains NUL") Nothing Nothing)- | otherwise =- withFileSystem conn \fs ->- bracket- c_duckdb_create_file_open_options- destroyFileOpenOptions- \opts -> do- whenNull opts "allocate file-open options"- mapM_ (\flag -> expectState "set file-open flag" (c_duckdb_file_open_options_set_flag opts flag 1)) flags- TextForeign.withCString (Text.pack path) \cPath ->- bracket- ( alloca \filePtr -> do- poke filePtr nullPtr- rc <- c_duckdb_file_system_open fs cPath opts filePtr- if rc /= DuckDBSuccess- then throwFileSystemError fs path- else do- handle <- peek filePtr- whenNull handle "allocate file handle"- pure handle- )- destroyFileHandle- action+ | '\0' `elem` path = throwInvalidArgument "file path contains NUL"+ | otherwise = withConnectionHandle conn \handle -> do+ fs <- output (c_duckdb_v2_connection_get_file_system handle)+ bracket (output (c_duckdb_v2_file_open_options_create fs)) (destroy c_duckdb_v2_file_open_options_destroy) \options -> do+ mapM_ (checked . c_duckdb_v2_file_open_options_set_flag options . fileFlagToNative) flags+ withText (Text.pack path) \name ->+ bracket (output (c_duckdb_v2_file_system_open fs name options)) (destroy c_duckdb_v2_file_destroy) (action . FileHandle) --- | Read up to the requested number of bytes from a file handle.-readFileHandleChunk :: DuckDBFileHandle -> Int64 -> IO BS.ByteString-readFileHandleChunk handle requested- | requested <= 0 = pure BS.empty- | toInteger requested > toInteger (maxBound :: Int) =- throwIO (SQLError (Text.pack "duckdb-simple: file read size exceeds Int range") Nothing Nothing)- | otherwise =- bracket (mallocBytes (fromIntegral requested)) free \raw -> do- bytesRead <- c_duckdb_file_handle_read handle raw requested- if bytesRead < 0- then throwFileHandleError handle (Text.pack "read failed")- else- if bytesRead > requested- then throwFileHandleError handle (Text.pack "read size exceeds buffer size")- else BS.packCStringLen (castPtr raw, fromIntegral bytesRead)+-- | Read up to the requested byte count and advance the file position. The+-- buffer and the size that the native call gets come from one value. A count+-- above that size raises an error. The library cannot check that the native+-- call writes no more than that size: the header of DuckDB states it as the+-- contract of the call.+readFileHandleChunk :: FileHandle -> Int64 -> IO BS.ByteString+readFileHandleChunk (FileHandle handle) requested+ | requested <= 0 = pure BS.empty+ | toInteger requested > toInteger (maxBound :: Int) = throwInvalidArgument "file read size exceeds Int range"+ | otherwise = BSI.createAndTrim size \buffer -> do+ count <- output (c_duckdb_v2_file_read handle (castPtr buffer) (fromIntegral size))+ when (toInteger count > toInteger size) $ throwInternalError "native file read exceeds buffer size"+ pure (fromIntegral count)+ where+ size = fromIntegral requested :: Int --- | Write an entire bytestring to a file handle.-writeFileHandleBytes :: DuckDBFileHandle -> BS.ByteString -> IO Int64-writeFileHandleBytes handle bytes =- BS.useAsCStringLen bytes \(ptr, len) -> do- written <- c_duckdb_file_handle_write handle (castPtr ptr) (fromIntegral len)- if written < 0- then throwFileHandleError handle (Text.pack "write failed")- else pure written+-- | Write bytes and return the byte count. Advance the file position.+writeFileHandleBytes :: FileHandle -> BS.ByteString -> IO Int64+writeFileHandleBytes (FileHandle handle) bytes = BS.useAsCStringLen bytes \(ptr, count) ->+ fromIntegral <$> output (c_duckdb_v2_file_write handle (ConstPtr (castPtr ptr)) (fromIntegral count)) --- | Return the current file position.-fileHandleTell :: DuckDBFileHandle -> IO Int64-fileHandleTell handle = do- pos <- c_duckdb_file_handle_tell handle- if pos < 0 then throwFileHandleError handle (Text.pack "tell failed") else pure pos+-- | Read exactly the requested bytes without changing the file position. A+-- range whose end exceeds the 'Int64' range raises an error before the native+-- call. DuckDB checks the range against the size of the file, and a read past+-- the end raises the error of DuckDB. The function allocates the buffer before+-- that check. The buffer and the size that the native call gets come from one+-- value. The library cannot check that the native call writes no more than+-- that size: the header of DuckDB states it as the contract of the call.+readFileHandleAt :: FileHandle -> Int64 -> Int64 -> IO BS.ByteString+readFileHandleAt (FileHandle handle) offset requested+ | offset < 0 = throwInvalidArgument "file offset is negative"+ | requested < 0 = throwInvalidArgument "file read size is negative"+ | toInteger requested > toInteger (maxBound :: Int) = throwInvalidArgument "file read size exceeds Int range"+ | toInteger offset + toInteger requested > toInteger (maxBound :: Int64) = throwInvalidArgument "file read range exceeds Int64 range"+ | otherwise = BSI.create size \buffer -> checked (c_duckdb_v2_file_read_at handle (castPtr buffer) (fromIntegral size) (fromIntegral offset))+ where+ size = fromIntegral requested :: Int --- | Return the current file size in bytes.-fileHandleSize :: DuckDBFileHandle -> IO Int64-fileHandleSize handle = do- size <- c_duckdb_file_handle_size handle- if size < 0 then throwFileHandleError handle (Text.pack "size failed") else pure size+-- | Write all bytes at an offset without changing the file position. A range+-- whose end exceeds the 'Int64' range raises an error.+writeFileHandleAt :: FileHandle -> Int64 -> BS.ByteString -> IO ()+writeFileHandleAt (FileHandle handle) offset bytes+ | offset < 0 = throwInvalidArgument "file offset is negative"+ | toInteger offset + toInteger (BS.length bytes) > toInteger (maxBound :: Int64) =+ throwInvalidArgument "file write range exceeds Int64 range"+ | otherwise = BS.useAsCStringLen bytes \(ptr, count) ->+ checked (c_duckdb_v2_file_write_at handle (ConstPtr (castPtr ptr)) (fromIntegral count) (fromIntegral offset)) --- | Seek to an absolute byte offset.-fileHandleSeek :: DuckDBFileHandle -> Int64 -> IO ()-fileHandleSeek handle pos = do- rc <- c_duckdb_file_handle_seek handle pos- if rc == DuckDBSuccess- then pure ()- else throwFileHandleError handle (Text.pack "seek failed")+-- | Return the current byte position.+fileHandleTell :: FileHandle -> IO Int64+fileHandleTell (FileHandle handle) = output (c_duckdb_v2_file_tell handle) >>= checkedPosition --- | Flush file-handle writes to stable storage.-fileHandleSync :: DuckDBFileHandle -> IO ()-fileHandleSync handle = do- rc <- c_duckdb_file_handle_sync handle- if rc == DuckDBSuccess- then pure ()- else throwFileHandleError handle (Text.pack "sync failed")+-- | Return the file size in bytes.+fileHandleSize :: FileHandle -> IO Int64+fileHandleSize (FileHandle handle) = output (c_duckdb_v2_file_size handle) >>= checkedPosition -withFileSystem :: Connection -> (DuckDBFileSystem -> IO a) -> IO a-withFileSystem conn action =- withClientContext conn \ctx ->- bracket- (c_duckdb_client_context_get_file_system ctx)- destroyFileSystem- (\fs -> whenNull fs "allocate file system" >> action fs)+-- | Set the absolute byte position.+fileHandleSeek :: FileHandle -> Int64 -> IO ()+fileHandleSeek (FileHandle handle) position+ | position < 0 = throwInvalidArgument "file offset is negative"+ | otherwise = checked (c_duckdb_v2_file_seek handle (fromIntegral position)) -destroyFileSystem :: DuckDBFileSystem -> IO ()-destroyFileSystem fs =- alloca \ptr -> poke ptr fs >> c_duckdb_destroy_file_system ptr+-- | Flush pending file writes to storage.+fileHandleSync :: FileHandle -> IO ()+fileHandleSync (FileHandle handle) = checked (c_duckdb_v2_file_sync handle) -destroyFileOpenOptions :: DuckDBFileOpenOptions -> IO ()-destroyFileOpenOptions opts =- alloca \ptr -> poke ptr opts >> c_duckdb_destroy_file_open_options ptr+-- | Test whether a file exists in DuckDB's file system. A directory gives 'False'.+pathExists :: Connection -> FilePath -> IO Bool+pathExists conn path = withPath conn path \fs name ->+ (/= 0) <$> output (c_duckdb_v2_file_system_path_exists fs name) -destroyFileHandle :: DuckDBFileHandle -> IO ()-destroyFileHandle handle =- alloca \ptr -> poke ptr handle >> c_duckdb_destroy_file_handle ptr+-- | Test whether DuckDB routes a path to a remote file system.+pathIsRemote :: Connection -> FilePath -> IO Bool+pathIsRemote conn path = withPath conn path \fs name ->+ (/= 0) <$> output (c_duckdb_v2_file_system_path_is_remote fs name) -throwFileSystemError :: DuckDBFileSystem -> FilePath -> IO a-throwFileSystemError fs path = mask_ do- err <- c_duckdb_file_system_error_data fs- throwErrorData err (Text.concat [Text.pack "duckdb-simple: failed to open file ", Text.pack path])+-- | Expand a path with DuckDB's file-system rules.+expandPath :: Connection -> FilePath -> IO FilePath+expandPath conn path = withPath conn path \fs name -> do+ pieces <- newIORef []+ bracket+ ( toFunPtr+ ( DuckDBV2TextSinkFn_Aux+ ( \text _ err -> runCallback err do+ value <- peek (unConstPtr text) >>= readText+ modifyIORef' pieces (value :)+ )+ )+ )+ freeHaskellFunPtr+ \callback -> checked (c_duckdb_v2_file_system_path_expand fs name (DuckDBV2TextSinkFn callback) nullPtr)+ Text.unpack . Text.concat . reverse <$> readIORef pieces -throwFileHandleError :: DuckDBFileHandle -> Text -> IO a-throwFileHandleError handle fallback = mask_ do- err <- c_duckdb_file_handle_error_data handle- throwErrorData err fallback+-- | Collect glob matches in the order supplied by DuckDB.+globPaths :: Connection -> FilePath -> IO [FilePath]+globPaths conn pattern_ = withPath conn pattern_ \fs name -> do+ paths <- newIORef []+ bracket+ ( toFunPtr+ ( DuckDBV2FileSystemPathGlobCallbackFn_Aux+ ( \path _ err -> runCallback err do+ value <- peek (unConstPtr path) >>= readText+ modifyIORef' paths (Text.unpack value :)+ )+ )+ )+ freeHaskellFunPtr+ \callback -> checked (c_duckdb_v2_file_system_path_glob fs name (DuckDBV2FileSystemPathGlobCallbackFn callback) nullPtr)+ reverse <$> readIORef paths -throwErrorData :: DuckDBErrorData -> Text -> IO a-throwErrorData err fallback =- bracket (pure err) destroyErrorData \errData -> do- msgPtr <- c_duckdb_error_data_message errData- errType <- c_duckdb_error_data_error_type errData- message <-- if msgPtr == nullPtr- then pure fallback- else peekUtf8CString msgPtr- throwIO- SQLError- { sqlErrorMessage = message- , sqlErrorType = Just errType- , sqlErrorQuery = Nothing- }+-- | Move a path through DuckDB's file system.+movePath :: Connection -> FilePath -> FilePath -> IO ()+movePath conn source target+ | '\0' `elem` target = throwInvalidArgument "file path contains NUL"+ | otherwise = withPath conn source \fs name -> withText (Text.pack target) \destination ->+ checked (c_duckdb_v2_file_system_path_move fs name destination) -destroyErrorData :: DuckDBErrorData -> IO ()-destroyErrorData err =- alloca \ptr -> poke ptr err >> c_duckdb_destroy_error_data ptr+-- | Remove a file. Return whether DuckDB removed it.+removePath :: Connection -> FilePath -> IO Bool+removePath conn path = withPath conn path \fs name ->+ (/= 0) <$> output (c_duckdb_v2_file_system_path_remove fs name) -expectState :: String -> IO DuckDBState -> IO ()-expectState label action = do- rc <- action- if rc == DuckDBSuccess- then pure ()- else- throwIO $- SQLError- { sqlErrorMessage = Text.pack ("duckdb-simple: " <> label <> " failed")- , sqlErrorType = Nothing- , sqlErrorQuery = Nothing- }+-- | Borrow the connection's file system and a checked path.+withPath :: Connection -> FilePath -> (DuckDBV2FileSystemHandle -> ConstPtr DuckDBV2Str -> IO a) -> IO a+withPath conn path action+ | '\0' `elem` path = throwInvalidArgument "file path contains NUL"+ | otherwise = withConnectionHandle conn \handle -> do+ fs <- output (c_duckdb_v2_connection_get_file_system handle)+ withText (Text.pack path) (action fs) --- | Reject a null file-system handle before use.-whenNull :: Ptr a -> String -> IO ()-whenNull ptr label =- if ptr == nullPtr then expectState label (pure DuckDBError) else pure ()+-- | Check the public signed range before converting a native byte position.+checkedPosition :: DuckDBIdx -> IO Int64+checkedPosition value+ | toInteger value > toInteger (maxBound :: Int64) = throwInternalError "file position exceeds Int64 range"+ | otherwise = pure (fromIntegral value)
src/Database/DuckDB/Simple/FromField.hs view
@@ -1,813 +1,1249 @@-{-# LANGUAGE DerivingVia #-}-{-# LANGUAGE FlexibleInstances #-}-{-# LANGUAGE LambdaCase #-}-{-# LANGUAGE NamedFieldPuns #-}-{-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE PatternSynonyms #-}-{-# LANGUAGE ScopedTypeVariables #-}-{-# LANGUAGE StrictData #-}-{-# LANGUAGE ViewPatterns #-}--{- |-Module : Database.DuckDB.Simple.FromField-Description : Conversion from DuckDB column values to Haskell types.--}-module Database.DuckDB.Simple.FromField (- Field (..),- FieldValue (..),- RawGeometry (..),- StructField (..),- StructValue (..),- UnionMemberType (..),- UnionValue (..),- LogicalTypeRep (..),- BitString (..),- bsFromBool,- BigNum (..),- fromBigNumBytes,- toBigNumBytes,- DecimalValue (..),- IntervalValue (..),- TimeWithZone (..),- ResultError (..),- FieldParser,- FromField (..),- returnError,-) where--import Control.Exception (Exception, SomeException (..))-import Data.Array (Array, assocs, bounds, listArray, range, (!))-import Data.Bits (Bits (..), finiteBitSize)-import qualified Data.ByteString as BS-import Data.Data (Typeable, typeRep)-import qualified Data.Geometry as G-import Data.Int (Int16, Int32, Int64, Int8)-import Data.Map.Strict (Map)-import qualified Data.Map.Strict as Map-import Data.Proxy (Proxy (..))-import Data.Text (Text)-import qualified Data.Text as Text-import qualified Data.Text.Encoding as TextEncoding-import Data.Time.Calendar (Day)-import Data.Time.Clock (UTCTime (..))-import Data.Time.LocalTime (- LocalTime (..),- TimeOfDay (..),- TimeZone (..),- localTimeToUTC,- utc,- utcToLocalTime,- )-import qualified Data.UUID as UUID-import Data.Word (Word16, Word32, Word64, Word8)-import Database.DuckDB.Simple.Geometry (RawGeometry (..), fromRawGeometry)-import Database.DuckDB.Simple.LogicalRep (- LogicalTypeRep (..),- StructField (..),- StructValue (..),- UnionMemberType (..),- UnionValue (..),- )-import Database.DuckDB.Simple.Ok-import Database.DuckDB.Simple.Time (Date, LocalTimestamp, UTCTimestamp, Unbounded (..))-import GHC.Float (double2Float, float2Double)-import GHC.Num.Integer (integerFromWordList)-import Numeric.Natural (Natural)--{- | Internal representation of a column value.-'FieldGeometry' contains raw WKB and CRS metadata. Generic parameters that-contain a non-NULL geometry require explicit SQL import.--}-data FieldValue- = FieldNull- | FieldInt8 Int8- | FieldInt16 Int16- | FieldInt32 Int32- | FieldInt64 Int64- | FieldWord8 Word8- | FieldWord16 Word16- | FieldWord32 Word32- | FieldWord64 Word64- | FieldUUID UUID.UUID- | FieldFloat Float- | FieldDouble Double- | FieldText Text- | FieldBool Bool- | FieldBlob BS.ByteString- | FieldGeometry RawGeometry- | FieldDate Date- | FieldTime TimeOfDay- | FieldTimestamp LocalTimestamp- | FieldInterval IntervalValue- | FieldHugeInt Integer- | FieldUHugeInt Integer- | FieldDecimal DecimalValue- | FieldTimestampTZ UTCTimestamp- | FieldTimeTZ TimeWithZone- | FieldBit BitString- | FieldBigNum BigNum- | FieldEnum Word32- | FieldArray (Array Int FieldValue)- | FieldList [FieldValue]- | FieldMap [(FieldValue, FieldValue)]- | FieldStruct (StructValue FieldValue)- | FieldUnion (UnionValue FieldValue)- deriving (Eq, Show)---- | Exact-width decimal payload plus its declared width and scale.-data DecimalValue = DecimalValue- { decimalWidth :: !Word8- , decimalScale :: !Word8- , decimalInteger :: !Integer- }- deriving (Eq, Show)---- | DuckDB interval payload split into months, days, and microseconds.-data IntervalValue = IntervalValue- { intervalMonths :: !Int32- , intervalDays :: !Int32- , intervalMicros :: !Int64- }- deriving (Eq, Show)---- | Arbitrary-precision integer wrapper used for DuckDB's BIGNUM type.-newtype BigNum = BigNum Integer- deriving stock (Eq, Show)- deriving (Num) via Integer--{- | Decode DuckDB’s BIGNUM blob (3-byte header + big-endian payload where negative magnitudes are-bitwise complemented) back into a Haskell @Integer@. We undo the complement when needed, then chunk-the remaining bytes into machine-word limbs (MSB chunk first) for @integerFromWordList@.--}-fromBigNumBytes :: [Word8] -> Integer-fromBigNumBytes bytes =- let header = take 3 bytes- payloadRaw = drop 3 bytes- isNeg = case header of- (h : _) -> h .&. 0x80 == 0- [] -> False- payload = if isNeg then map complement payloadRaw else payloadRaw- bytesPerWord = finiteBitSize (0 :: Word) `div` 8 -- 8 on 64-bit, 4 on 32-bit- len = length payload- (firstChunk, rest) = splitAt (len `mod` bytesPerWord) payload-- chunkWords [] = []- chunkWords xs =- let (chunk, remainder) = splitAt bytesPerWord xs- in chunk : chunkWords remainder-- toWord = foldl (\acc b -> (acc `shiftL` 8) .|. fromIntegral b) 0- limbs = (if null firstChunk then id else (firstChunk :)) (chunkWords rest)- in integerFromWordList isNeg (map toWord limbs)--{- | Encode an @Integer@ into DuckDB’s BIGNUM blob layout: emit the 3-byte header- (sign bit plus payload length) followed by the magnitude bytes in the same- big-endian / complemented-on-negative form that DuckDB stores internally.--}-toBigNumBytes :: Integer -> [Word8]-toBigNumBytes value =- let isNeg = value < 0- magnitude = if isNeg then negate value else value- payloadBE- | magnitude == 0 = [0]- | otherwise = go magnitude []- where- go 0 acc = acc- go n acc =- let (q, r) = quotRem n 256- in go q (fromIntegral r : acc)-- headerBase :: Word32- headerBase = (fromIntegral (length payloadBE) .|. 0x00800000)- headerVal :: Word32- headerVal = if isNeg then complement headerBase else headerBase- headerMasked = headerVal .&. 0x00FFFFFF- headerBytes =- [ fromIntegral ((headerMasked `shiftR` 16) .&. 0xFF)- , fromIntegral ((headerMasked `shiftR` 8) .&. 0xFF)- , fromIntegral (headerMasked .&. 0xFF)- ]- payloadBytes = if isNeg then map complement payloadBE else payloadBE- in headerBytes <> payloadBytes---- | DuckDB BIT value represented as raw bytes plus left-padding bit count.-data BitString = BitString- { padding :: !Word8- , bits :: !BS.ByteString- }- deriving stock (Eq)--instance Show BitString where- show (BitString padding bits) =- drop (fromIntegral padding) $ concatMap word8ToString $ BS.unpack bits- where- word8ToString :: Word8 -> String- word8ToString w = map (\n -> if testBit w n then '1' else '0') [7, 6 .. 0]---- | Construct a @BitString@ from a list of @Bool@s, where the first element-bsFromBool :: [Bool] -> BitString-bsFromBool bits =- let totalBits = length bits- padding = (8 - (totalBits `mod` 8)) `mod` 8- paddedBits = replicate padding False ++ bits- byteChunks = chunk 8 paddedBits- byteValues = map bitsToWord8 byteChunks- in BitString (fromIntegral padding) (BS.pack byteValues)- where- chunk _ [] = []- chunk n xs = take n xs : chunk n (drop n xs)-- bitsToWord8 :: [Bool] -> Word8- bitsToWord8 bs = foldl (\acc (b, i) -> if b then setBit acc i else acc) 0 (zip bs [7, 6 .. 0])---- | Time-of-day paired with its associated timezone offset.-data TimeWithZone = TimeWithZone- { timeWithZoneTime :: !TimeOfDay- , timeWithZoneZone :: !TimeZone- }- deriving (Eq, Show)---- | Match signed values without narrowing to the machine word size.-pattern FieldInt :: Int64 -> FieldValue-pattern FieldInt i <- (fieldValueToInt -> Just i)- where- FieldInt i = FieldInt64 i--fieldValueToInt :: FieldValue -> Maybe Int64-fieldValueToInt (FieldInt8 i) = Just (fromIntegral i)-fieldValueToInt (FieldInt16 i) = Just (fromIntegral i)-fieldValueToInt (FieldInt32 i) = Just (fromIntegral i)-fieldValueToInt (FieldInt64 i) = Just i-fieldValueToInt _ = Nothing---- | Match unsigned values without narrowing to the machine word size.-pattern FieldWord :: Word64 -> FieldValue-pattern FieldWord i <- (fieldValueToWord -> Just i)- where- FieldWord i = FieldWord64 i--fieldValueToWord :: FieldValue -> Maybe Word64-fieldValueToWord (FieldWord8 i) = Just (fromIntegral i)-fieldValueToWord (FieldWord16 i) = Just (fromIntegral i)-fieldValueToWord (FieldWord32 i) = Just (fromIntegral i)-fieldValueToWord (FieldWord64 i) = Just i-fieldValueToWord _ = Nothing---- | Metadata for a single column in a row.-data Field = Field- { fieldName :: Text- , fieldIndex :: Int- , fieldValue :: FieldValue- }- deriving (Eq, Show)--{- | Exception thrown if conversion from a SQL value to a Haskell-value fails.--}-data ResultError- = -- | The SQL and Haskell types are not compatible.- Incompatible- { errSQLType :: Text- , errSQLField :: Text- , errHaskellType :: Text- , errMessage :: Text- }- | {- | A SQL @NULL@ was encountered when the Haskell- type did not permit it.- -}- UnexpectedNull- { errSQLType :: Text- , errSQLField :: Text- , errHaskellType :: Text- , errMessage :: Text- }- | {- | The SQL value could not be parsed, or could not- be represented as a valid Haskell value, or an- unexpected low-level error occurred (e.g. mismatch- between metadata and actual data in a row).- -}- ConversionFailed- { errSQLType :: Text- , errSQLField :: Text- , errHaskellType :: Text- , errMessage :: Text- }- deriving (Eq, Show)--instance Exception ResultError--{- | Parser used by @FromField@ instances and utilities such as-@Database.DuckDB.Simple.FromRow.fieldWith@. The supplied @Field@ contains-column metadata and an already-decoded @FieldValue@; callers should return-@Ok@ on success or @Errors@ (typically wrapping a @ResultError@) when the-conversion fails.--}-type FieldParser a = Field -> Ok a---- | Types that can be constructed from a DuckDB column.-class FromField a where- fromField :: FieldParser a--instance FromField FieldValue where- fromField Field{fieldValue} = Ok fieldValue--instance FromField RawGeometry where- fromField f@Field{fieldValue} =- case fieldValue of- FieldGeometry value -> Ok value- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f "expected GEOMETRY"---- | Decode the shape without CRS metadata. Use t'RawGeometry' to retain the CRS.-instance FromField G.Geometry where- fromField f@Field{fieldValue} = case fieldValue of- FieldGeometry raw -> either (returnError ConversionFailed f . Text.pack) pure (fromRawGeometry raw)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f "expected GEOMETRY"--instance FromField (StructValue FieldValue) where- fromField f@Field{fieldValue} =- case fieldValue of- FieldStruct structVal -> Ok structVal- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f "expected STRUCT"--instance FromField (UnionValue FieldValue) where- fromField f@Field{fieldValue} =- case fieldValue of- FieldUnion unionVal -> Ok unionVal- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f "expected UNION"--instance FromField UUID.UUID where- fromField f@Field{fieldValue} =- case fieldValue of- FieldText t ->- case UUID.fromText t of- Just uuid -> Ok uuid- Nothing -> returnError ConversionFailed f "invalid UUID format"- FieldUUID uuid -> Ok uuid- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Bool where- fromField f@Field{fieldValue} =- case fieldValue of- FieldBool b -> Ok b- FieldInt i -> Ok (i /= 0)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Int8 where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i -> boundedIntegral f i- FieldHugeInt value -> boundedFromInteger f value- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> boundedFromInteger f (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Int64 where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i -> Ok i- FieldHugeInt value -> boundedFromInteger f value- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> Ok (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Int32 where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i -> boundedIntegral f i- FieldHugeInt value -> boundedFromInteger f value- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> boundedFromInteger f (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Int16 where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i -> boundedIntegral f i- FieldHugeInt value -> boundedFromInteger f value- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> boundedFromInteger f (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Int where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i -> boundedIntegral f i- FieldHugeInt value -> boundedFromInteger f value- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> boundedFromInteger f (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Integer where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i -> Ok (fromIntegral i)- FieldWord w -> Ok (fromIntegral w)- FieldBigNum (BigNum big) -> Ok big- FieldHugeInt value -> Ok value- FieldUHugeInt value -> Ok value- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Natural where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i- | i >= 0 -> Ok (fromIntegral i)- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to Natural"- FieldWord w -> Ok (fromIntegral w)- FieldHugeInt value- | value >= 0 -> Ok (fromIntegral value)- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to Natural"- FieldUHugeInt value -> Ok (fromIntegral value)- FieldBigNum (BigNum big) ->- if big >= 0- then Ok $ fromIntegral big- else returnError ConversionFailed f "negative value cannot be converted to Natural"- FieldEnum value -> Ok (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Word64 where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i- | i >= 0 -> Ok (fromIntegral i)- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldWord w -> Ok w- FieldHugeInt value- | value >= 0 -> boundedFromInteger f value- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> Ok (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Word32 where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i- | i >= 0 -> boundedIntegral f i- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldWord w -> boundedFromInteger f (toInteger w)- FieldHugeInt value- | value >= 0 -> boundedFromInteger f value- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> Ok value- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Word16 where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i- | i >= 0 -> boundedIntegral f i- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldWord w -> boundedFromInteger f (toInteger w)- FieldHugeInt value- | value >= 0 -> boundedFromInteger f value- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> boundedFromInteger f (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Word8 where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i- | i >= 0 -> boundedIntegral f i- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldWord w -> boundedFromInteger f (toInteger w)- FieldHugeInt value- | value >= 0 -> boundedFromInteger f value- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> boundedFromInteger f (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Word where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInt i- | i >= 0 -> boundedFromInteger f (fromIntegral i)- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldWord w -> boundedFromInteger f (toInteger w)- FieldHugeInt value- | value >= 0 -> boundedFromInteger f value- | otherwise ->- returnError ConversionFailed f "negative value cannot be converted to unsigned integer"- FieldUHugeInt value -> boundedFromInteger f value- FieldEnum value -> boundedFromInteger f (fromIntegral value)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Double where- fromField f@Field{fieldValue} =- case fieldValue of- FieldDouble d -> Ok d- FieldFloat value -> Ok (float2Double value)- FieldInt i -> Ok (fromIntegral i)- FieldDecimal DecimalValue{decimalInteger, decimalScale} ->- Ok (realToFrac decimalInteger / 10 ^ decimalScale)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Float where- fromField field =- case (fromField field :: Ok Double) of- Errors err -> Errors err- Ok d- | not (isInfinite d || isNaN d) && isInfinite (double2Float d) ->- returnError ConversionFailed field "floating-point value out of bounds"- | otherwise -> Ok (double2Float d)--instance FromField Text where- fromField f@Field{fieldValue} =- case fieldValue of- FieldText t -> Ok t- FieldInt i -> Ok (Text.pack (show i))- FieldFloat value -> Ok (Text.pack (show value))- FieldDouble d -> Ok (Text.pack (show d))- FieldBool b -> Ok (if b then Text.pack "1" else Text.pack "0")- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField String where- fromField field = Text.unpack <$> fromField field--instance FromField BS.ByteString where- fromField f@Field{fieldValue} =- case fieldValue of- FieldBlob bs -> Ok bs- FieldGeometry RawGeometry{rawGeometryWKB} -> Ok rawGeometryWKB- FieldText t -> Ok (TextEncoding.encodeUtf8 t)- FieldBit (BitString _ bits) -> Ok bits- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField BitString where- fromField f@Field{fieldValue} =- case fieldValue of- FieldBit b -> Ok b- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField BigNum where- fromField f@Field{fieldValue} =- case fieldValue of- FieldBigNum big -> Ok big- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance {-# OVERLAPPABLE #-} (Typeable a, FromField a) => FromField [a] where- fromField f@Field{fieldValue} =- case fieldValue of- FieldList entries ->- traverse (uncurry parseElement) (zip [0 :: Int ..] entries)- FieldArray arr ->- traverse (uncurry parseElement) (assocs arr)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""- where- parseElement idx value =- fromField- Field- { fieldName = fieldNameWithIndex idx- , fieldIndex = fieldIndex f- , fieldValue = value- }- fieldNameWithIndex idx =- let base = fieldName f- idxText = Text.pack (show idx)- in base <> Text.pack "[" <> idxText <> Text.pack "]"--instance (Typeable a, FromField a) => FromField (Array Int a) where- fromField f@Field{fieldValue} =- case fieldValue of- FieldArray arr -> do- converted <- traverse (traverseElement arr) (range (bounds arr))- pure (listArray (bounds arr) converted)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""- where- traverseElement arr idx =- fromField- Field- { fieldName = fieldNameWithIndex idx- , fieldIndex = fieldIndex f- , fieldValue = arr ! idx- }- fieldNameWithIndex idx =- let idxText = Text.pack (show idx)- in fieldName f <> Text.pack "[" <> idxText <> Text.pack "]"--instance (Ord k, Typeable k, Typeable v, FromField k, FromField v) => FromField (Map k v) where- fromField f@Field{fieldValue} =- case fieldValue of- FieldMap pairs -> Map.fromList <$> traverse (uncurry parsePair) (zip [0 :: Int ..] pairs)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""- where- parsePair idx (keyValue, valValue) = do- key <-- fromField- Field- { fieldName = fieldName f <> suffix idx ".key"- , fieldIndex = fieldIndex f- , fieldValue = keyValue- }- value <-- fromField- Field- { fieldName = fieldName f <> suffix idx ".value"- , fieldIndex = fieldIndex f- , fieldValue = valValue- }- pure (key, value)- suffix idx label =- let idxText = Text.pack (show idx)- in Text.pack "[" <> idxText <> Text.pack "]" <> Text.pack label--instance FromField DecimalValue where- fromField f@Field{fieldValue} =- case fieldValue of- FieldDecimal dec -> Ok dec- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField Day where- fromField f@Field{fieldValue} =- case fieldValue of- FieldDate day -> finiteField f day- FieldTimestamp timestamp -> finiteField f (localDay <$> timestamp)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField (Unbounded Day) where- fromField f@Field{fieldValue} =- case fieldValue of- FieldDate day -> Ok day- FieldTimestamp timestamp -> Ok (localDay <$> timestamp)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField TimeOfDay where- fromField f@Field{fieldValue} =- case fieldValue of- FieldTime tod -> Ok tod- FieldTimestamp timestamp -> finiteField f (localTimeOfDay <$> timestamp)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField TimeWithZone where- fromField f@Field{fieldValue} =- case fieldValue of- FieldTimeTZ tz -> Ok tz- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField LocalTime where- fromField f@Field{fieldValue} =- case fieldValue of- FieldTimestamp ts -> finiteField f ts- FieldDate day -> finiteField f ((\value -> LocalTime value midnight) <$> day)- FieldTimestampTZ utcTime -> finiteField f (utcToLocalTime utc <$> utcTime)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""- where- midnight = TimeOfDay 0 0 0--instance FromField (Unbounded LocalTime) where- fromField f@Field{fieldValue} =- case fieldValue of- FieldTimestamp ts -> Ok ts- FieldDate day -> Ok ((\value -> LocalTime value midnight) <$> day)- FieldTimestampTZ utcTime -> Ok (utcToLocalTime utc <$> utcTime)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""- where- midnight = TimeOfDay 0 0 0--instance FromField IntervalValue where- fromField f@Field{fieldValue} =- case fieldValue of- FieldInterval interval -> Ok interval- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""--instance FromField UTCTime where- fromField f@Field{fieldValue} =- case fieldValue of- FieldTimestamp ts -> finiteField f (localTimeToUTC utc <$> ts)- FieldTimestampTZ utcTime -> finiteField f utcTime- FieldDate day -> finiteField f ((\value -> localTimeToUTC utc (LocalTime value midnight)) <$> day)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""- where- midnight = TimeOfDay 0 0 0--instance FromField (Unbounded UTCTime) where- fromField f@Field{fieldValue} =- case fieldValue of- FieldTimestamp ts -> Ok (localTimeToUTC utc <$> ts)- FieldTimestampTZ utcTime -> Ok utcTime- FieldDate day -> Ok ((\value -> localTimeToUTC utc (LocalTime value midnight)) <$> day)- FieldNull -> returnError UnexpectedNull f ""- _ -> returnError Incompatible f ""- where- midnight = TimeOfDay 0 0 0---- | Reject infinity when the requested Haskell type holds only finite values.-finiteField :: (Typeable a) => Field -> Unbounded a -> Ok a-finiteField _ (Finite value) = Ok value-finiteField field _ = returnError ConversionFailed field "infinity requires an Unbounded date or timestamp"--instance (FromField a) => FromField (Maybe a) where- fromField Field{fieldValue = FieldNull} = Ok Nothing- fromField field = Just <$> fromField field---- | Helper for bounded integral conversions.-boundedIntegral :: forall a. (Integral a, Bounded a, Typeable a) => Field -> Int64 -> Ok a-boundedIntegral f = boundedFromInteger f . toInteger--boundedFromInteger :: forall a. (Integral a, Bounded a, Typeable a) => Field -> Integer -> Ok a-boundedFromInteger f@Field{} value- | value < toInteger (minBound :: a) =- returnError ConversionFailed f "integer value out of bounds"- | value > toInteger (maxBound :: a) =- returnError ConversionFailed f "integer value out of bounds"- | otherwise = Ok (fromInteger value)--{- | Helper to construct a ResultError with field context.-based on postgresql-simple's implementation--}-returnError :: forall b. (Typeable b) => (Text -> Text -> Text -> Text -> ResultError) -> Field -> Text -> Ok b-returnError mkError Field{fieldValue, fieldName} msg =- Errors- [ SomeException $- mkError- (fieldValueTypeName fieldValue)- fieldName- (Text.pack $ show (typeRep (Proxy :: Proxy b)))- msg- ]--fieldValueTypeName :: FieldValue -> Text-fieldValueTypeName = \case- FieldNull -> "NULL"- FieldInt8{} -> "INT1"- FieldInt16{} -> "INT2"- FieldInt32{} -> "INT4"- FieldInt64{} -> "INT8"- FieldWord8{} -> "UTINYINT"- FieldWord16{} -> "USMALLINT"- FieldWord32{} -> "UINTEGER"- FieldWord64{} -> "UBIGINT"- FieldFloat{} -> "FLOAT"- FieldDouble{} -> "DOUBLE"- FieldText{} -> "TEXT"- FieldBool{} -> "BOOLEAN"- FieldBlob{} -> "BLOB"- FieldGeometry{} -> "GEOMETRY"- FieldDate{} -> "DATE"- FieldTime{} -> "TIME"- FieldTimestamp{} -> "TIMESTAMP"- FieldInterval{} -> "INTERVAL"- FieldHugeInt{} -> "HUGEINT"- FieldUHugeInt{} -> "UHUGEINT"- FieldDecimal{} -> "DECIMAL"- FieldTimestampTZ{} -> "TIMESTAMP_TZ"- FieldTimeTZ{} -> "TIME_TZ"- FieldBit{} -> "BIT"- FieldBigNum{} -> "BIGNUM"- FieldEnum{} -> "ENUM"- FieldArray{} -> "ARRAY"- FieldList{} -> "LIST"- FieldMap{} -> "MAP"- FieldUUID{} -> "UUID"- FieldStruct{} -> "STRUCT"- FieldUnion{} -> "UNION"+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE DeriveTraversable #-}+{-# LANGUAGE DerivingVia #-}+{-# LANGUAGE FlexibleInstances #-}+{-# LANGUAGE KindSignatures #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE PatternSynonyms #-}+{-# LANGUAGE RoleAnnotations #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE StrictData #-}+{-# LANGUAGE TypeApplications #-}+{-# LANGUAGE ViewPatterns #-}++-- |+-- Module : Database.DuckDB.Simple.FromField+-- Description : Conversion from DuckDB column values to Haskell types.+module Database.DuckDB.Simple.FromField (+ Field (..),+ FieldValue (..),+ RawGeometry (..),+ StructField (..),+ StructValue (..),+ TupleValue (..),+ UnionMemberType (..),+ UnionValue (..),+ LogicalTypeRep (..),+ LogicalTypeId (..),+ BitString (..),+ bsFromBool,+ BigNum (..),+ fromBigNumBytes,+ toBigNumBytes,+ DecimalValue (..),+ IntervalValue (..),+ TimeWithZone (..),+ timeWithZone,+ timeWithZoneZone,+ SizedArray,+ sizedArray,+ sizedArrayFromList,+ sizedArrayToArray,+ ResultError (..),+ FieldParser,+ FromField (..),+ returnError,+ fieldTypeName,+) where++import Control.Exception (Exception (..))+import Data.Array (Array, assocs, bounds, elems, listArray, rangeSize)+import Data.Bits (Bits (..), finiteBitSize)+import qualified Data.ByteString as BS+import Data.Data (Typeable, typeRep)+import qualified Data.Geometry as G+import Data.Int (Int16, Int32, Int64, Int8)+import Data.Map.Strict (Map)+import qualified Data.Map.Strict as Map+import Data.Ord (clamp)+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Time.Calendar (Day, addDays, diffDays, fromGregorian)+import Data.Time.Clock (UTCTime (..), diffTimeToPicoseconds)+import Data.Time.LocalTime (+ LocalTime (..),+ TimeOfDay (..),+ TimeZone,+ minutesToTimeZone,+ timeOfDayToTime,+ timeZoneMinutes,+ )+import qualified Data.UUID as UUID+import Data.Word (Word16, Word32, Word64, Word8)+import Database.DuckDB.Simple.Conversion (Conversion, conversionErrorWith)+import Database.DuckDB.Simple.Geometry (RawGeometry (..), fromRawGeometry)+import Database.DuckDB.Simple.LogicalRep (+ LogicalTypeId (..),+ LogicalTypeRep (..),+ StructField (..),+ StructValue (..),+ TupleValue (..),+ UnionMemberType (..),+ UnionValue (..),+ aliasBase,+ listElementType,+ logicalTypeSQL,+ mapEntryTypes,+ )+import Database.DuckDB.Simple.Native (arraySizeReason)+import Database.DuckDB.Simple.Time (Date, LocalTimestamp, UTCTimestamp, Unbounded (..))+import GHC.Float (double2Float, float2Double, int2Double, int2Float, rationalToDouble, rationalToFloat, word2Double)+import GHC.Num.Integer (integerFromWordList)+import GHC.TypeLits (KnownNat, Nat, natVal)+import Numeric.Natural (Natural)++-- | Internal representation of a column value.+-- 'FieldGeometry' contains raw WKB and CRS metadata. A STRUCT, TUPLE, or UNION+-- parameter binds it as a GEOMETRY with the same CRS. A VARIANT parameter+-- binds it only when it has no CRS. A top-level t'RawGeometry' parameter binds+-- as a GEOMETRY with its own CRS. See Note [Raw GEOMETRY parameters] in+-- "Database.DuckDB.Simple.ToField".+data FieldValue+ = FieldNull+ | FieldInt8 Int8+ | FieldInt16 Int16+ | FieldInt32 Int32+ | FieldInt64 Int64+ | FieldWord8 Word8+ | FieldWord16 Word16+ | FieldWord32 Word32+ | FieldWord64 Word64+ | FieldUUID UUID.UUID+ | FieldFloat Float+ | FieldDouble Double+ | FieldText Text+ | FieldBool Bool+ | FieldBlob BS.ByteString+ | FieldGeometry RawGeometry+ | FieldDate Date+ | FieldTime TimeOfDay+ | FieldTimestamp LocalTimestamp+ | FieldInterval IntervalValue+ | FieldHugeInt Integer+ | FieldUHugeInt Integer+ | FieldDecimal DecimalValue+ | FieldTimestampTZ UTCTimestamp+ | FieldTimeTZ TimeWithZone+ | FieldBit BitString+ | FieldBigNum BigNum+ | FieldEnum Word32 Text+ | FieldArray (Array Int FieldValue)+ | FieldList [FieldValue]+ | FieldMap [(FieldValue, FieldValue)]+ | FieldStruct (StructValue FieldValue)+ | FieldTuple (TupleValue FieldValue)+ | FieldUnion (UnionValue FieldValue)+ deriving (Eq, Show)++-- | Exact-width decimal payload plus its declared width and scale. A parameter+-- of this type binds as a DECIMAL with this width and scale. A @SizedArray@, a+-- list, or a typed row cannot hold it, because it has no @DuckDBColumnType@+-- instance: one column type cannot give the width and scale of each value.+data DecimalValue = DecimalValue+ { decimalWidth :: Word8+ , decimalScale :: Word8+ , decimalInteger :: Integer+ }+ deriving (Eq, Show)++-- | DuckDB interval payload split into months, days, and microseconds. A+-- parameter of this type binds as an INTERVAL, and DuckDB keeps the three+-- fields.+data IntervalValue = IntervalValue+ { intervalMonths :: Int32+ , intervalDays :: Int32+ , intervalMicros :: Int64+ }+ deriving (Eq, Show)++-- | Arbitrary-precision integer wrapper used for DuckDB's BIGNUM type.+newtype BigNum = BigNum Integer+ deriving stock (Eq, Show)+ deriving (Num) via Integer++-- | Decode a DuckDB BIGNUM blob as a Haskell @Integer@. The blob has a 3-byte+-- header and a big-endian payload. Negative values have a complemented payload.+-- Restore that payload, then split it into machine-word limbs for+-- @integerFromWordList@, with the most significant limb first.+-- This is the BIGNUM storage of DuckDB 2.0.0. The V2 API does not commit to this+-- layout, and this library does not use this function.+fromBigNumBytes :: [Word8] -> Integer+fromBigNumBytes bytes =+ let header = take 3 bytes+ payloadRaw = drop 3 bytes+ isNeg = case header of+ (h : _) -> h .&. 0x80 == 0+ [] -> False+ payload = if isNeg then map complement payloadRaw else payloadRaw+ bytesPerWord = finiteBitSize (0 :: Word) `div` 8 -- 8 on 64-bit, 4 on 32-bit+ len = length payload+ (firstChunk, rest) = splitAt (len `mod` bytesPerWord) payload++ chunkWords [] = []+ chunkWords xs =+ let (chunk, remainder) = splitAt bytesPerWord xs+ in chunk : chunkWords remainder++ toWord = foldl (\acc b -> (acc `shiftL` 8) .|. fromIntegral b) 0+ limbs = (if null firstChunk then id else (firstChunk :)) (chunkWords rest)+ in integerFromWordList isNeg (map toWord limbs)++-- | Encode an @Integer@ as a DuckDB BIGNUM blob. Write the 3-byte header+-- (sign bit and payload length), then the magnitude bytes in big-endian order.+-- Complement the header and the payload for a negative value.+-- This is the BIGNUM storage of DuckDB 2.0.0. The V2 API does not commit to+-- this layout, and this library does not use this function.+toBigNumBytes :: Integer -> [Word8]+toBigNumBytes value =+ let isNeg = value < 0+ magnitude = if isNeg then negate value else value+ payloadBE+ | magnitude == 0 = [0]+ | otherwise = go magnitude []+ where+ go 0 acc = acc+ go n acc =+ let (q, r) = quotRem n 256+ in go q (fromIntegral r : acc)++ headerBase :: Word32+ headerBase = (fromIntegral (length payloadBE) .|. 0x00800000)+ headerVal :: Word32+ headerVal = if isNeg then complement headerBase else headerBase+ headerMasked = headerVal .&. 0x00FFFFFF+ headerBytes =+ [ fromIntegral ((headerMasked `shiftR` 16) .&. 0xFF)+ , fromIntegral ((headerMasked `shiftR` 8) .&. 0xFF)+ , fromIntegral (headerMasked .&. 0xFF)+ ]+ payloadBytes = if isNeg then map complement payloadBE else payloadBE+ in headerBytes <> payloadBytes++-- | DuckDB BIT value represented as raw bytes plus left-padding bit count.+data BitString = BitString+ { padding :: Word8+ , bits :: BS.ByteString+ }+ deriving stock (Eq)++instance Show BitString where+ show (BitString padding bits) =+ drop (fromIntegral padding) $ concatMap word8ToString $ BS.unpack bits+ where+ word8ToString :: Word8 -> String+ word8ToString w = map (\n -> if testBit w n then '1' else '0') [7, 6 .. 0]++-- | Construct a @BitString@ from a list of @Bool@ values. The first element is+-- the first bit, and 'True' is a 1 bit. The function puts 0 bits in front of the+-- first element to fill the first byte, and stores their count as the padding.+bsFromBool :: [Bool] -> BitString+bsFromBool bits =+ let totalBits = length bits+ padding = (8 - (totalBits `mod` 8)) `mod` 8+ paddedBits = replicate padding False ++ bits+ byteChunks = chunk 8 paddedBits+ byteValues = map bitsToWord8 byteChunks+ in BitString (fromIntegral padding) (BS.pack byteValues)+ where+ chunk _ [] = []+ chunk n xs = take n xs : chunk n (drop n xs)++ bitsToWord8 :: [Bool] -> Word8+ bitsToWord8 bs = foldl (\acc (b, i) -> if b then setBit acc i else acc) 0 (zip bs [7, 6 .. 0])++-- | Time of day with its offset from UTC in seconds, as a TIMETZ stores it. A+-- positive offset is east of UTC: @12:00:00+02@ has the offset 7200. DuckDB+-- stores an offset from -57599 to 57599 seconds (-15:59:59 to +15:59:59) and the+-- time in microseconds. A value with such an offset reads back equal to the+-- bound value. A time with digits below one microsecond loses these digits, as+-- in SQL text. 'timeWithZone' makes a value from a "Data.Time" zone, and+-- 'timeWithZoneZone' gives the zone of a value.+data TimeWithZone = TimeWithZone+ { timeWithZoneTime :: TimeOfDay+ , timeWithZoneOffsetSeconds :: Int32+ }+ deriving (Eq, Show)++-- | Make a t'TimeWithZone' from a time of day and a "Data.Time" zone. The offset+-- is the minutes of the zone times 60. If the seconds exceed the 'Int32'+-- range, keep the nearest bound. Binding rejects either bound as an invalid+-- TIMETZ offset. DuckDB does not store the zone name or the summer flag.+timeWithZone :: TimeOfDay -> TimeZone -> TimeWithZone+timeWithZone time zone = TimeWithZone time (fromInteger (clamp bounds32 seconds))+ where+ seconds = toInteger (timeZoneMinutes zone) * 60+ bounds32 = (toInteger (minBound @Int32), toInteger (maxBound @Int32))++-- | Give the "Data.Time" zone of the offset of a t'TimeWithZone'. A 'TimeZone'+-- holds whole minutes, so the seconds of an offset are dropped: the offset+-- 3630 seconds of @12:00:00+01:00:30@ gives the zone @+0100@. The zone has no+-- name.+timeWithZoneZone :: TimeWithZone -> TimeZone+timeWithZoneZone value = minutesToTimeZone (fromIntegral (timeWithZoneOffsetSeconds value `quot` 60))++type role SizedArray nominal representational++-- | An 'Array' of exactly @n@ values, with the bounds @(0, n - 1)@. A+-- parameter of this type binds as the ARRAY type @T[n]@, and a typed+-- registration declares that type. See Note [Sized arrays] in+-- "Database.DuckDB.Simple.ToField". Make a value with 'sizedArray' or+-- 'sizedArrayFromList'.+newtype SizedArray (n :: Nat) a = SizedArray (Array Int a)+ deriving stock (Eq, Ord, Show, Functor, Foldable, Traversable)++-- | Make a t'SizedArray' from an 'Array' with any bounds. The result is+-- 'Nothing' when the array does not have exactly @n@ values. The result has+-- the values in their order, with the bounds @(0, n - 1)@.+sizedArray :: forall n a. (KnownNat n) => Array Int a -> Maybe (SizedArray n a)+sizedArray values+ | toInteger count == natVal (Proxy @n) = Just (SizedArray (listArray (0, count - 1) (elems values)))+ | otherwise = Nothing+ where+ count = rangeSize (bounds values)++-- | Make a t'SizedArray' from a list. The result is 'Nothing' when the list+-- does not have exactly @n@ values. The count is checked before the 'Array'+-- is built, and the check reads at most @n + 1@ elements.+-- A size above @maxBound :: Int@ gives 'Nothing' without reading the list.+sizedArrayFromList :: forall n a. (KnownNat n) => [a] -> Maybe (SizedArray n a)+sizedArrayFromList values+ | size > toInteger (maxBound :: Int) = Nothing+ | hasSize size values = Just (SizedArray (listArray (0, fromInteger size - 1) values))+ | otherwise = Nothing+ where+ size = natVal (Proxy @n)+ hasSize 0 [] = True+ hasSize 0 (_ : _) = False+ hasSize _ [] = False+ hasSize remaining (_ : rest) = hasSize (remaining - 1) rest++-- | Give the 'Array' of a t'SizedArray', with the bounds @(0, n - 1)@.+sizedArrayToArray :: SizedArray n a -> Array Int a+sizedArrayToArray (SizedArray values) = values++-- | Match signed values without narrowing to the machine word size.+pattern FieldInt :: Int64 -> FieldValue+pattern FieldInt i <- (fieldValueToInt -> Just i)+ where+ FieldInt i = FieldInt64 i++fieldValueToInt :: FieldValue -> Maybe Int64+fieldValueToInt (FieldInt8 i) = Just (fromIntegral i)+fieldValueToInt (FieldInt16 i) = Just (fromIntegral i)+fieldValueToInt (FieldInt32 i) = Just (fromIntegral i)+fieldValueToInt (FieldInt64 i) = Just i+fieldValueToInt _ = Nothing++-- | Match unsigned values without narrowing to the machine word size.+pattern FieldWord :: Word64 -> FieldValue+pattern FieldWord i <- (fieldValueToWord -> Just i)+ where+ FieldWord i = FieldWord64 i++fieldValueToWord :: FieldValue -> Maybe Word64+fieldValueToWord (FieldWord8 i) = Just (fromIntegral i)+fieldValueToWord (FieldWord16 i) = Just (fromIntegral i)+fieldValueToWord (FieldWord32 i) = Just (fromIntegral i)+fieldValueToWord (FieldWord64 i) = Just i+fieldValueToWord _ = Nothing++-- | Metadata for a single column in a row. The SQL type is an explicit+-- conversion, so a field has no equality or display instance.+data Field = Field+ { fieldName :: Text+ , fieldIndex :: Int+ , fieldValue :: FieldValue+ , fieldType :: Conversion LogicalTypeRep+ -- ^ Read the SQL type, with each alias. VARIANT means that the type of the+ -- stored value decides. See Note [Column types].+ }++{- Note [Column types]+~~~~~~~~~~~~~~~~~~~~~~+fieldType reads the SQL type of a Field, with each alias (see+Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep"). Each place that+makes a Field supplies a conversion:++- A result column. The query result owns its native logical type. A query+ parser runs before this metadata scope closes. A cursor retains its scope+ until exhaustion, reset, or statement closure.+- A callback argument. The callback owns its native logical type. Its scope+ closes when the callback invocation returns.+- A LIST or ARRAY element, or a MAP key or value. The reader maps the type+ projection over the parent's conversion with listElementType or mapEntryTypes.+- A synthetic field made from a FieldValue alone. Its conversion returns VARIANT.++VARIANT means that the stored value determines its type. An element of a+VARIANT column also has type VARIANT. The fieldType action converts no type+tree or ENUM dictionary until requested. Structured value readers can also+build type trees. See Note [Type trees] in+"Database.DuckDB.Simple.Materialize".++The scope caches successful type reads. An interrupted read can retry while+the scope remains open. Closure releases native types and clears the cache.+An escaped native conversion raises SQLError with the message+"duckdb-simple: field type scope is closed", including a cached conversion.+A LogicalTypeRep snapshot already returned to the caller is pure and remains+valid after closure. Completed conversion diagnostics are also pure. The+library resolves them before closing the scope. See Note [Conversion effects]+in "Database.DuckDB.Simple.Conversion".+-}++-- | Exception thrown if conversion from a SQL value to a Haskell+-- value fails.+data ResultError+ = -- | The SQL and Haskell types are not compatible.+ Incompatible+ { errSQLType :: Text+ , errSQLField :: Text+ , errHaskellType :: Text+ , errMessage :: Text+ }+ | -- | A SQL @NULL@ was encountered when the Haskell+ -- type did not permit it.+ UnexpectedNull+ { errSQLType :: Text+ , errSQLField :: Text+ , errHaskellType :: Text+ , errMessage :: Text+ }+ | -- | The SQL value could not be parsed, or could not+ -- be represented as a valid Haskell value, or an+ -- unexpected low-level error occurred (e.g. mismatch+ -- between metadata and actual data in a row).+ ConversionFailed+ { errSQLType :: Text+ , errSQLField :: Text+ , errHaskellType :: Text+ , errMessage :: Text+ }+ deriving (Eq, Show)++-- | The text of the error names the column and the two types.+instance Exception ResultError where+ displayException = Text.unpack . renderResultError++-- | Render a conversion error. An error without a column name, for example one+-- of a field of a generic record, names "a value".+renderResultError :: ResultError -> Text+renderResultError = \case+ Incompatible{errSQLType, errSQLField, errHaskellType, errMessage} ->+ Text.concat ["duckdb-simple: ", subject errSQLField, " has type ", errSQLType, " but expected ", errHaskellType, detail errMessage]+ UnexpectedNull{errSQLField, errHaskellType, errMessage} ->+ Text.concat ["duckdb-simple: ", subject errSQLField, " is NULL but expected ", errHaskellType, detail errMessage]+ ConversionFailed{errMessage} -> errMessage+ where+ subject name = if Text.null name then Text.pack "a value" else Text.pack "column " <> name+ detail message = if Text.null message then Text.empty else Text.pack ": " <> message++-- | Parser used by @FromField@ instances and utilities such as+-- @Database.DuckDB.Simple.FromRow.fieldWith@. The supplied @Field@ contains+-- column metadata and an already-decoded @FieldValue@. Use 'pure' for success+-- or 'returnError' for a conversion failure.+type FieldParser a = Field -> Conversion a++-- | Types that can be constructed from a DuckDB column. Each instance says+-- which SQL types it reads. See Note [Reads of another SQL type].+--+-- An instance that cannot read a value fails with a 'ResultError', for example+-- 'ConversionFailed' or 'Incompatible'. The library does not raise the+-- 'ResultError'. It puts the text of the 'ResultError' into the @SQLError@+-- that the statement raises. The section \"Errors\" of the README gives the+-- type of that @SQLError@.+class FromField a where+ fromField :: FieldParser a++{- Note [Reads of another SQL type]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+Each FromField instance reads the SQL type of its Haskell type, for example+BIGINT for Int64 and TIMESTAMP for LocalTime. An instance also reads another+SQL type only when it gives, for each value, the value of the cast of DuckDB to+the SQL type of the Haskell type, or an error. For example, Int8 reads the+BIGINT value 5 as 5, which is the value of 5::BIGINT::TINYINT, and it raises+ConversionFailed for 300, where the cast raises an error. When the Haskell type+holds more digits than its SQL type, the comparison is with the cast to the+narrowest SQL type that holds the value exactly. Thus LocalTime reads a+TIMESTAMP_NS value with its nanoseconds, as the cast to TIMESTAMP_NS keeps+them.++The instances do not read a pair whose cast needs data that the decoded value+does not have, or whose value the library cannot give exactly. A read of such a+pair raises Incompatible, as for each other pair that an instance does not+read. The error names the SQL type of the column in the text of DuckDB+(fieldTypeName), for example TIMESTAMP_NS or DECIMAL(18,3). For a column of the+type VARIANT, it names the type of the stored value in the same text+(fieldValueTypeName), and the error of a String names Text. Error diagnostics+read type metadata only when the final conversion fails. These pairs are:++- UTCTime from TIMESTAMP, TIMESTAMP_S, TIMESTAMP_MS, TIMESTAMP_NS, and DATE, and+ LocalTime from TIMESTAMPTZ and TIMESTAMPTZ_NS. These casts use the TimeZone+ setting of the session. To read the value as the other type in UTC, write+ c::TIMESTAMP AT TIME ZONE 'UTC' in SQL for a TIMESTAMP, TIMESTAMP_S,+ TIMESTAMP_MS, or DATE value, and c AT TIME ZONE 'UTC' for each other type.+ Without the cast to TIMESTAMP, DuckDB converts a DATE, TIMESTAMP_S, or+ TIMESTAMP_MS value to nanoseconds, which fails outside the years 1677 to+ 2262. The cast rounds a TIMESTAMP_NS value to the microsecond.+- Text and String from FLOAT and DOUBLE. The text of the cast differs from the+ text of show, for example 0.01 and 1.0e-2. Read a Double, or write c::VARCHAR.+- ByteString from VARCHAR. The cast to BLOB reads escape sequences such as \x41.+ Read a Text, or write encode(c) for the UTF-8 bytes.+- Each integer type from ENUM. The cast to an integer type parses the label,+ for example '20' gives 20. Write c::INTEGER for that value, or+ enum_code(c)::BIGINT for the index. enum_code gives an unsigned type, which+ the signed integer types do not read.+- Variant from a value that a VARIANT cannot hold (variantShape in+ "Database.DuckDB.Simple.Variant"): a FieldArray, a FieldTuple, a FieldStruct+ that is not in the form of a VARIANT object (a field of a type other than+ VARIANT, or keys out of ascending order), a FieldEnum, a FieldMap, a+ FieldUnion, or a GEOMETRY with a CRS, at any depth. The cast to VARIANT gives+ another value. Write c::VARIANT. A VARIANT column decodes only to values of+ the VARIANT shape. A JSON type at any depth raises Incompatible, because+ DuckDB parses its text when it casts to VARIANT. Write c::VARIANT for the+ parsed value. The bind side+ (variantPayloadType in "Database.DuckDB.Simple.ToField") accepts more: it+ binds a FieldArray and a FieldTuple as a VARIANT list, and the fields of each+ FieldStruct as VARIANT values.++Text and String read the label of an ENUM value, which is the text of the cast+to VARCHAR. See Note [ENUM labels] in "Database.DuckDB.Simple.Materialize".++The rule holds at each depth. A field of a generic record reads the SQL types+that the FromField instance of its type reads: a scalar field uses that+instance, and a list field reads a LIST and an ARRAY (the DuckValue instances+in "Database.DuckDB.Simple.Generic").++See Note [DECIMAL as DOUBLE and FLOAT] for the floating-point value of a DECIMAL.+-}++-- | Reads each SQL type, and does not convert the value.+instance FromField FieldValue where+ fromField Field{fieldValue} = pure fieldValue++-- | Reads GEOMETRY, with its CRS.+instance FromField RawGeometry where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldGeometry value -> pure value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f "expected GEOMETRY"++-- | Reads GEOMETRY. The shape has no CRS metadata. Use t'RawGeometry' to retain+-- the CRS.+instance FromField G.Geometry where+ fromField f@Field{fieldValue} = case fieldValue of+ FieldGeometry raw -> either (returnError ConversionFailed f . Text.pack) pure (fromRawGeometry raw)+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f "expected GEOMETRY"++-- | Reads STRUCT.+instance FromField (StructValue FieldValue) where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldStruct structVal -> pure structVal+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f "expected STRUCT"++-- | Reads UNION.+instance FromField (UnionValue FieldValue) where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldUnion unionVal -> pure unionVal+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f "expected UNION"++-- | Reads UUID, and VARCHAR in the form of 36 characters with dashes. Each+-- other text fails with 'ConversionFailed'.+instance FromField UUID.UUID where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldText t ->+ case UUID.fromText t of+ Just uuid -> pure uuid+ Nothing -> returnError ConversionFailed f "invalid UUID format"+ FieldUUID uuid -> pure uuid+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads TUPLE.+instance FromField (TupleValue FieldValue) where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldTuple tuple -> pure tuple+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f "expected TUPLE"++-- | Reads BOOLEAN, and TINYINT, SMALLINT, INTEGER, and BIGINT. An integer other+-- than 0 is True, as in the cast to BOOLEAN.+instance FromField Bool where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldBool b -> pure b+ FieldInt i -> pure (i /= 0)+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads TINYINT, SMALLINT, INTEGER, BIGINT, HUGEINT, and UHUGEINT. A value+-- out of range fails with 'ConversionFailed'. It does not read ENUM.+instance FromField Int8 where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i -> boundedIntegral f i+ FieldHugeInt value -> boundedFromInteger f value+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads BIGINT, TINYINT, SMALLINT, INTEGER, HUGEINT, and UHUGEINT. A value+-- out of range fails with 'ConversionFailed'. It does not read ENUM.+instance FromField Int64 where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i -> pure i+ FieldHugeInt value -> boundedFromInteger f value+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads INTEGER, TINYINT, SMALLINT, BIGINT, HUGEINT, and UHUGEINT. A value+-- out of range fails with 'ConversionFailed'. It does not read ENUM.+instance FromField Int32 where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i -> boundedIntegral f i+ FieldHugeInt value -> boundedFromInteger f value+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads SMALLINT, TINYINT, INTEGER, BIGINT, HUGEINT, and UHUGEINT. A value+-- out of range fails with 'ConversionFailed'. It does not read ENUM.+instance FromField Int16 where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i -> boundedIntegral f i+ FieldHugeInt value -> boundedFromInteger f value+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads BIGINT, TINYINT, SMALLINT, INTEGER, HUGEINT, and UHUGEINT. A value+-- out of range fails with 'ConversionFailed'. It does not read ENUM.+instance FromField Int where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i -> boundedIntegral f i+ FieldHugeInt value -> boundedFromInteger f value+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads BIGNUM, TINYINT to BIGINT, UTINYINT to UBIGINT, HUGEINT, and UHUGEINT.+instance FromField Integer where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i -> pure (fromIntegral i)+ FieldWord w -> pure (fromIntegral w)+ FieldBigNum (BigNum big) -> pure big+ FieldHugeInt value -> pure value+ FieldUHugeInt value -> pure value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads BIGNUM, TINYINT to BIGINT, UTINYINT to UBIGINT, HUGEINT, and+-- UHUGEINT. A negative value fails with 'ConversionFailed'. It does not read+-- ENUM.+instance FromField Natural where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i+ | i >= 0 -> pure (fromIntegral i)+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to Natural"+ FieldWord w -> pure (fromIntegral w)+ FieldHugeInt value+ | value >= 0 -> pure (fromIntegral value)+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to Natural"+ FieldUHugeInt value -> pure (fromIntegral value)+ FieldBigNum (BigNum big) ->+ if big >= 0+ then pure $ fromIntegral big+ else returnError ConversionFailed f "negative value cannot be converted to Natural"+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads UTINYINT to UBIGINT, TINYINT to BIGINT, HUGEINT, and UHUGEINT. A+-- negative value or a value out of range fails with 'ConversionFailed'. It does+-- not read ENUM.+instance FromField Word64 where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i+ | i >= 0 -> pure (fromIntegral i)+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldWord w -> pure w+ FieldHugeInt value+ | value >= 0 -> boundedFromInteger f value+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads UTINYINT to UBIGINT, TINYINT to BIGINT, HUGEINT, and UHUGEINT. A+-- negative value or a value out of range fails with 'ConversionFailed'. It does+-- not read ENUM.+instance FromField Word32 where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i+ | i >= 0 -> boundedIntegral f i+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldWord w -> boundedFromInteger f (toInteger w)+ FieldHugeInt value+ | value >= 0 -> boundedFromInteger f value+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads UTINYINT to UBIGINT, TINYINT to BIGINT, HUGEINT, and UHUGEINT. A+-- negative value or a value out of range fails with 'ConversionFailed'. It does+-- not read ENUM.+instance FromField Word16 where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i+ | i >= 0 -> boundedIntegral f i+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldWord w -> boundedFromInteger f (toInteger w)+ FieldHugeInt value+ | value >= 0 -> boundedFromInteger f value+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads UTINYINT to UBIGINT, TINYINT to BIGINT, HUGEINT, and UHUGEINT. A+-- negative value or a value out of range fails with 'ConversionFailed'. It does+-- not read ENUM.+instance FromField Word8 where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i+ | i >= 0 -> boundedIntegral f i+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldWord w -> boundedFromInteger f (toInteger w)+ FieldHugeInt value+ | value >= 0 -> boundedFromInteger f value+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads UTINYINT to UBIGINT, TINYINT to BIGINT, HUGEINT, and UHUGEINT. A+-- negative value or a value out of range fails with 'ConversionFailed'. It does+-- not read ENUM.+instance FromField Word where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInt i+ | i >= 0 -> boundedFromInteger f (fromIntegral i)+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldWord w -> boundedFromInteger f (toInteger w)+ FieldHugeInt value+ | value >= 0 -> boundedFromInteger f value+ | otherwise ->+ returnError ConversionFailed f "negative value cannot be converted to unsigned integer"+ FieldUHugeInt value -> boundedFromInteger f value+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads DOUBLE, FLOAT, TINYINT to BIGINT, and DECIMAL. A BIGINT above 2^53+-- rounds as the cast to DOUBLE. See Note [DECIMAL as DOUBLE and FLOAT].+instance FromField Double where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldDouble d -> pure d+ FieldFloat value -> pure (float2Double value)+ FieldInt i -> pure (fromIntegral i)+ FieldDecimal value -> pure (decimalToDouble value)+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads FLOAT, DOUBLE, TINYINT to BIGINT, and DECIMAL. A finite DOUBLE out of+-- the range of FLOAT fails with 'ConversionFailed'. See+-- Note [DECIMAL as DOUBLE and FLOAT].+instance FromField Float where+ fromField field@Field{fieldValue} =+ case fieldValue of+ -- DuckDB casts an integer to FLOAT with static_cast<float>, which+ -- rounds one time to the nearest value, ties to even. int2Float rounds+ -- in the same way. As seen in CI, the DuckDB 2.0.0 build for macOS on+ -- Apple Silicon rounds some BIGINT values above 2^53 two times in its+ -- cast to FLOAT, and the library gives the nearest value there.+ FieldInt i -> pure (int2Float (fromIntegral i))+ FieldDecimal value -> pure (decimalToFloat value)+ _ -> do+ d <- fromField field :: Conversion Double+ if not (isInfinite d || isNaN d) && isInfinite (double2Float d)+ then returnError ConversionFailed field "floating-point value out of bounds"+ else pure (double2Float d)++{- Note [DECIMAL as DOUBLE and FLOAT]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+A DECIMAL that a Double or a Float reads gives the value of the cast of DuckDB+2.0.0 to DOUBLE or FLOAT. The cast is TryCastDecimalToFloatingPoint in+src/common/operator/cast_operators.cpp of DuckDB. decimalToDouble and+decimalToFloat use the same three paths:++1. Scale 0 and width 19 to 38. DuckDB stores the value as a HUGEINT and casts+ it with CastBigintToFloating in src/common/types/hugeint.cpp:++ case -1: result = -REAL_T(NumericLimits<uint64_t>::Maximum() - input.lower) - 1;+ default: result = REAL_T(input.lower) + REAL_T(input.upper) * (REAL_T(NumericLimits<uint64_t>::Maximum()) + 1);++ The case is the value of input.upper. A FLOAT result is this DOUBLE value+ cast to float (Hugeint::TryCast for float). hugeIntToDouble copies the+ formula. The formula rounds more than one time, so it does not always give+ the nearest value. DuckDB casts -9007199254740994::DECIMAL(19,0) to+ -9007199254740992.0, and 3356951675583485076575379654993209529::DECIMAL(38,0)+ to 3.356951675583485e36. The nearest values are -9007199254740994.0 and+ 3.3569516755834854e36. This path exists only to give the value of the cast.+2. Scale 0 with a width of at most 18, or an integer of at most 2^53 (2^24+ for FLOAT) with a scale of at most 22 (10 for FLOAT). DuckDB converts the+ integer to the floating-point type and divides it by the power of ten in+ that type. Under the limits, the integer and the power of ten are exact in+ the type, so only the division rounds. With scale 0, the division by 1 is+ exact, so only the conversion of the integer rounds. Each step rounds one+ time to the nearest value.+3. Each other value. DuckDB converts the decimal digits with fast_float, which+ gives the nearest value.++Thus paths 2 and 3 give the nearest value, with ties to even. In path 2 the+library does the same steps. With scale 0 and a width of at most 18, the+integer is an Int64 value, and fromInteger converts it with int2Double or+int2Float, which round one time as the C++ conversion of int64_t does. In path+3 the library uses rationalToDouble or rationalToFloat. fromRational uses these+functions too. They give the nearest value with ties to even, and they do not+reduce the fraction first.++Path 3 is slower than path 2. It applies to a Double with a scale above 0 when+the unscaled integer is above 2^53 or the scale is above 22: at scale 2 for an+absolute value above 9.0e13, and at scale 10 above 9.0e5. It applies to a Float+with a scale above 0 when the unscaled integer is above 2^24 or the scale is+above 10.+-}++-- | The value of the cast of DuckDB from a DECIMAL to DOUBLE.+-- See Note [DECIMAL as DOUBLE and FLOAT].+decimalToDouble :: DecimalValue -> Double+decimalToDouble DecimalValue{decimalWidth, decimalScale, decimalInteger}+ | decimalScale == 0 && decimalWidth > 18 = hugeIntToDouble decimalInteger+ | decimalScale == 0 = fromInteger decimalInteger+ | decimalScale <= 22 && abs decimalInteger <= 9007199254740992 = fromInteger decimalInteger / 10 ^ decimalScale+ | otherwise = rationalToDouble decimalInteger (10 ^ decimalScale)++-- | The value of the cast of DuckDB from a DECIMAL to FLOAT.+-- See Note [DECIMAL as DOUBLE and FLOAT].+decimalToFloat :: DecimalValue -> Float+decimalToFloat DecimalValue{decimalWidth, decimalScale, decimalInteger}+ | decimalScale == 0 && decimalWidth > 18 = double2Float (hugeIntToDouble decimalInteger)+ | decimalScale == 0 = fromInteger decimalInteger+ | decimalScale <= 10 && abs decimalInteger <= 16777216 = fromInteger decimalInteger / 10 ^ decimalScale+ | otherwise = rationalToFloat decimalInteger (10 ^ decimalScale)++-- | The cast of DuckDB from a HUGEINT to DOUBLE. It is not always the nearest+-- value. See Note [DECIMAL as DOUBLE and FLOAT].+hugeIntToDouble :: Integer -> Double+hugeIntToDouble value+ | upper == -1 = negate (word2Double (maxBound - lower)) - 1+ | otherwise = word2Double lower + int2Double upper * 18446744073709551616+ where+ -- The two 64-bit halves of the HUGEINT in two's complement. word2Double+ -- and int2Double round one time to the nearest value, ties to even, as+ -- the C++ conversions of uint64_t and int64_t to double do. The C+++ -- factor is 2^64 after its rounding, so the product is exact. The sum or+ -- the difference rounds one time, as in C++.+ upper = fromInteger (value `shiftR` 64) :: Int+ lower = fromInteger value :: Word++-- | Reads VARCHAR, TINYINT to BIGINT as decimal digits, BOOLEAN as @true@ or+-- @false@, and the label of an ENUM value. It does not read FLOAT or DOUBLE. See+-- Note [Reads of another SQL type].+instance FromField Text where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldText t -> pure t+ FieldInt i -> pure (Text.pack (show i))+ FieldBool b -> pure (if b then Text.pack "true" else Text.pack "false")+ FieldEnum _ label -> pure label+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads the SQL types that 'Text' reads.+instance FromField String where+ fromField field = Text.unpack <$> fromField field++-- | Reads BLOB, the WKB bytes of a GEOMETRY, and the data bytes of a BIT, as the+-- casts to BLOB give them. It does not read VARCHAR. See+-- Note [Reads of another SQL type].+instance FromField BS.ByteString where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldBlob bs -> pure bs+ FieldGeometry RawGeometry{rawGeometryWKB} -> pure rawGeometryWKB+ FieldBit (BitString _ bits) -> pure bits+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads BIT.+instance FromField BitString where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldBit b -> pure b+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads BIGNUM.+instance FromField BigNum where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldBigNum big -> pure big+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads LIST and ARRAY. Each element reads as @a@.+instance {-# OVERLAPPABLE #-} (Typeable a, FromField a) => FromField [a] where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldList entries -> traverse fromField (elementFields f (zip [0 ..] entries))+ FieldArray arr -> traverse fromField (elementFields f (assocs arr))+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Give the fields of the elements of a LIST or an ARRAY field, from the index+-- and the value of each element. An element has the name @name[index]@ and the+-- element type of the field.+elementFields :: Field -> [(Int, FieldValue)] -> [Field]+elementFields f = map element+ where+ element (idx, value) =+ Field+ { fieldName = fieldName f <> Text.pack "[" <> Text.pack (show idx) <> Text.pack "]"+ , fieldIndex = fieldIndex f+ , fieldValue = value+ , fieldType = elementType+ }+ -- The elements share the parent type conversion.+ elementType = listElementType <$> fieldType f++-- | Reads ARRAY and LIST, as a list does. A size @n@ with no DuckDB ARRAY type+-- or a value with another count raises 'Incompatible'. See Note [Sized arrays] in+-- "Database.DuckDB.Simple.ToField".+instance (KnownNat n, Typeable a, FromField a) => FromField (SizedArray n a) where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldList entries -> sized (zip [0 ..] entries)+ FieldArray arr -> sized (assocs arr)+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""+ where+ size = natVal (Proxy @n)+ sized elements+ | Just reason <- arraySizeReason size = returnError Incompatible f reason+ | toInteger count == size = SizedArray . listArray (0, count - 1) <$> traverse fromField (elementFields f elements)+ | otherwise = returnError Incompatible f (Text.pack ("the value has " <> show count <> " elements"))+ where+ count = length elements++-- | Reads MAP. Each key reads as @k@, and each value reads as @v@.+-- Keys must remain distinct after conversion.+instance (Ord k, Typeable k, Typeable v, FromField k, FromField v) => FromField (Map k v) where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldMap pairs -> do+ result <- Map.fromList <$> traverse (uncurry parsePair) (zip [0 :: Int ..] pairs)+ if Map.size result == length pairs+ then pure result+ else returnError ConversionFailed f "duplicate MAP key after conversion"+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""+ where+ parsePair idx (keyValue, valValue) = do+ key <-+ fromField+ Field+ { fieldName = fieldName f <> suffix idx ".key"+ , fieldIndex = fieldIndex f+ , fieldValue = keyValue+ , fieldType = keyType+ }+ value <-+ fromField+ Field+ { fieldName = fieldName f <> suffix idx ".value"+ , fieldIndex = fieldIndex f+ , fieldValue = valValue+ , fieldType = valueType+ }+ pure (key, value)+ entryTypes = mapEntryTypes <$> fieldType f+ keyType = fst <$> entryTypes+ valueType = snd <$> entryTypes+ suffix idx label =+ let idxText = Text.pack (show idx)+ in Text.pack "[" <> idxText <> Text.pack "]" <> Text.pack label++-- | Reads DECIMAL.+instance FromField DecimalValue where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldDecimal dec -> pure dec+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++{- Note [Temporal casts]+~~~~~~~~~~~~~~~~~~~~~+DuckDB converts TIMESTAMP_S, TIMESTAMP_MS, and TIMESTAMP_NS to microseconds+before a cast to DATE. The S/MS conversion can overflow even when the date+fits DATE. The NS conversion rounds to the nearest microsecond, with ties away+from the epoch. Both Day readers use this intermediate. A DATE-to-timestamp+cast checks midnight in the same microsecond range. Both LocalTime readers+check this range for a DATE input. For a timestamp input, they keep the wider+range and the nanoseconds. Unbounded readers also keep infinity.++A TIMESTAMP_S or TIMESTAMP_MS cast to TIME checks the same microsecond range.+The TimeOfDay reader checks this range for a timestamp input. It reads+TIMESTAMP_NS at nanosecond precision, as a cast to TIME_NS.+-}++-- | Get the checked microsecond intermediate of a temporal cast.+-- Round to the nearest microsecond, with ties away from the epoch. See+-- Note [Temporal casts].+localTimeMicros :: LocalTime -> Maybe Integer+localTimeMicros LocalTime{localDay, localTimeOfDay} =+ let epoch = fromGregorian 1970 1 1+ picoseconds =+ diffDays localDay epoch * 86400000000000000+ + diffTimeToPicoseconds (timeOfDayToTime localTimeOfDay)+ micros+ | picoseconds < 0 = negate ((negate picoseconds + 500000) `div` 1000000)+ | otherwise = (picoseconds + 500000) `div` 1000000+ in if micros < toInteger (minBound :: Int64) || micros > toInteger (maxBound :: Int64)+ then Nothing+ else Just micros++-- | Convert a timestamp as the native cast to DATE does. See+-- Note [Temporal casts].+timestampDate :: LocalTimestamp -> Maybe Date+timestampDate (Finite timestamp) = do+ micros <- localTimeMicros timestamp+ pure (Finite (addDays (micros `div` 86400000000) (fromGregorian 1970 1 1)))+timestampDate NegInfinity = Just NegInfinity+timestampDate PosInfinity = Just PosInfinity++-- | Convert a date to midnight in the TIMESTAMP range. See+-- Note [Temporal casts].+dateTimestamp :: Date -> Maybe LocalTimestamp+dateTimestamp (Finite day) =+ let timestamp = LocalTime day (TimeOfDay 0 0 0)+ in Finite timestamp <$ localTimeMicros timestamp+dateTimestamp NegInfinity = Just NegInfinity+dateTimestamp PosInfinity = Just PosInfinity++-- | Reads DATE, and the date of TIMESTAMP, TIMESTAMP_S, TIMESTAMP_MS, and+-- TIMESTAMP_NS. Infinity and a timestamp outside the range of a DATE cast fail+-- with 'ConversionFailed'. See Note [Temporal casts].+instance FromField Day where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldDate day -> finiteField f day+ FieldTimestamp timestamp -> case timestampDate timestamp of+ Just day -> finiteField f day+ Nothing -> returnError ConversionFailed f "timestamp is outside the range of a DATE cast"+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads the SQL types that 'Day' reads, also infinity. See+-- Note [Temporal casts].+instance FromField (Unbounded Day) where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldDate day -> pure day+ FieldTimestamp timestamp -> case timestampDate timestamp of+ Just day -> pure day+ Nothing -> returnError ConversionFailed f "timestamp is outside the range of a DATE cast"+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads TIME, TIME_NS, and the time of TIMESTAMP, TIMESTAMP_S, TIMESTAMP_MS,+-- and TIMESTAMP_NS. Infinity and a timestamp outside the range of a TIME cast+-- fail with 'ConversionFailed'. See Note [Temporal casts].+instance FromField TimeOfDay where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldTime tod -> pure tod+ FieldTimestamp (Finite timestamp)+ | Nothing <- localTimeMicros timestamp ->+ returnError ConversionFailed f "timestamp is outside the range of a TIME cast"+ FieldTimestamp timestamp -> finiteField f (localTimeOfDay <$> timestamp)+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads TIMETZ.+instance FromField TimeWithZone where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldTimeTZ tz -> pure tz+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads TIMESTAMP, TIMESTAMP_S, TIMESTAMP_MS, TIMESTAMP_NS, and DATE at+-- midnight. It does not read TIMESTAMPTZ. Infinity and a DATE outside the+-- TIMESTAMP range fail with 'ConversionFailed'. See Note [Temporal casts]+-- and Note [Reads of another SQL type].+instance FromField LocalTime where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldTimestamp ts -> finiteField f ts+ FieldDate day -> case dateTimestamp day of+ Just timestamp -> finiteField f timestamp+ Nothing -> returnError ConversionFailed f "date is outside the TIMESTAMP range"+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads the SQL types that t'LocalTime' reads, also infinity. See+-- Note [Temporal casts].+instance FromField (Unbounded LocalTime) where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldTimestamp ts -> pure ts+ FieldDate day -> case dateTimestamp day of+ Just timestamp -> pure timestamp+ Nothing -> returnError ConversionFailed f "date is outside the TIMESTAMP range"+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads INTERVAL.+instance FromField IntervalValue where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldInterval interval -> pure interval+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads TIMESTAMPTZ and TIMESTAMPTZ_NS. It does not read TIMESTAMP or DATE.+-- Infinity fails with 'ConversionFailed'. See Note [Reads of another SQL type].+instance FromField UTCTime where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldTimestampTZ utcTime -> finiteField f utcTime+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reads the SQL types that t'UTCTime' reads, also infinity.+instance FromField (Unbounded UTCTime) where+ fromField f@Field{fieldValue} =+ case fieldValue of+ FieldTimestampTZ utcTime -> pure utcTime+ FieldNull -> returnError UnexpectedNull f ""+ _ -> returnError Incompatible f ""++-- | Reject infinity when the requested Haskell type holds only finite values.+finiteField :: (Typeable a) => Field -> Unbounded a -> Conversion a+finiteField _ (Finite value) = pure value+finiteField field _ = returnError ConversionFailed field "infinity requires an Unbounded date or timestamp"++-- | Reads NULL as Nothing, and each other value as @a@.+instance (FromField a) => FromField (Maybe a) where+ fromField Field{fieldValue = FieldNull} = pure Nothing+ fromField field = Just <$> fromField field++-- | Helper for bounded integral conversions.+boundedIntegral :: forall a. (Integral a, Bounded a, Typeable a) => Field -> Int64 -> Conversion a+boundedIntegral f = boundedFromInteger f . toInteger++boundedFromInteger :: forall a. (Integral a, Bounded a, Typeable a) => Field -> Integer -> Conversion a+boundedFromInteger f@Field{} value+ | value < toInteger (minBound :: a) =+ returnError ConversionFailed f "integer value out of bounds"+ | value > toInteger (maxBound :: a) =+ returnError ConversionFailed f "integer value out of bounds"+ | otherwise = pure (fromInteger value)++-- | Construct a conversion failure with field context when diagnostics are needed.+-- See Note [Conversion effects] in "Database.DuckDB.Simple.Conversion".+returnError :: forall b. (Typeable b) => (Text -> Text -> Text -> Text -> ResultError) -> Field -> Text -> Conversion b+returnError mkError Field{fieldValue, fieldName, fieldType} msg =+ conversionErrorWith $ do+ logical <- fieldType+ pure $+ mkError+ (fieldTypeName logical fieldValue)+ fieldName+ (Text.pack $ show (typeRep (Proxy @b)))+ msg++-- | The SQL type name of a field in an error: the text of its type, or the type+-- of the stored value when the type is VARIANT. See+-- Note [Reads of another SQL type].+fieldTypeName :: LogicalTypeRep -> FieldValue -> Text+fieldTypeName logical value = case aliasBase logical of+ LogicalTypeScalar LogicalTypeIdVariant -> fieldValueTypeName value+ _ -> logicalTypeSQL logical++-- | The SQL type name of a decoded value, in the type text of DuckDB+-- (logicalTypeSQL), for example TIMESTAMP WITH TIME ZONE or DECIMAL(18,3). A+-- value that does not hold its full type (a LIST, an ARRAY, a MAP, an ENUM, a+-- STRUCT, a TUPLE, or a UNION) gives the name of its kind. See+-- Note [Reads of another SQL type].+fieldValueTypeName :: FieldValue -> Text+fieldValueTypeName = \case+ FieldNull -> scalar LogicalTypeIdSQLNull+ FieldInt8{} -> scalar LogicalTypeIdTinyInt+ FieldInt16{} -> scalar LogicalTypeIdSmallInt+ FieldInt32{} -> scalar LogicalTypeIdInteger+ FieldInt64{} -> scalar LogicalTypeIdBigInt+ FieldWord8{} -> scalar LogicalTypeIdUTinyInt+ FieldWord16{} -> scalar LogicalTypeIdUSmallInt+ FieldWord32{} -> scalar LogicalTypeIdUInteger+ FieldWord64{} -> scalar LogicalTypeIdUBigInt+ FieldFloat{} -> scalar LogicalTypeIdFloat+ FieldDouble{} -> scalar LogicalTypeIdDouble+ FieldText{} -> scalar LogicalTypeIdVarchar+ FieldBool{} -> scalar LogicalTypeIdBoolean+ FieldBlob{} -> scalar LogicalTypeIdBlob+ FieldGeometry RawGeometry{rawGeometryCRS} -> logicalTypeSQL (LogicalTypeGeometry rawGeometryCRS)+ FieldDate{} -> scalar LogicalTypeIdDate+ FieldTime{} -> scalar LogicalTypeIdTime+ FieldTimestamp{} -> scalar LogicalTypeIdTimestamp+ FieldInterval{} -> scalar LogicalTypeIdInterval+ FieldHugeInt{} -> scalar LogicalTypeIdHugeInt+ FieldUHugeInt{} -> scalar LogicalTypeIdUHugeInt+ FieldDecimal DecimalValue{decimalWidth, decimalScale} -> logicalTypeSQL (LogicalTypeDecimal decimalWidth decimalScale)+ FieldTimestampTZ{} -> scalar LogicalTypeIdTimestampTz+ FieldTimeTZ{} -> scalar LogicalTypeIdTimeTz+ FieldBit{} -> scalar LogicalTypeIdBit+ FieldBigNum{} -> scalar LogicalTypeIdBigNum+ FieldEnum{} -> "ENUM"+ FieldArray{} -> "ARRAY"+ FieldList{} -> "LIST"+ FieldMap{} -> "MAP"+ FieldUUID{} -> scalar LogicalTypeIdUUID+ FieldStruct{} -> "STRUCT"+ FieldTuple{} -> "TUPLE"+ FieldUnion{} -> "UNION"+ where+ scalar = logicalTypeSQL . LogicalTypeScalar
src/Database/DuckDB/Simple/FromRow.hs view
@@ -3,278 +3,251 @@ {-# LANGUAGE DerivingStrategies #-} {-# LANGUAGE FlexibleContexts #-} {-# LANGUAGE GeneralizedNewtypeDeriving #-}-{-# LANGUAGE LambdaCase #-} {-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE TypeOperators #-} -{- |-Module : Database.DuckDB.Simple.FromRow-Description : Convert rows of @Field@s into Haskell values using a parser-style interface.--}+-- |+-- Module : Database.DuckDB.Simple.FromRow+-- Description : Parse rows of @Field@s as Haskell values. module Database.DuckDB.Simple.FromRow (- -- * Row parsing- RowParser (..),- field,- fieldWith,- numFieldsRemaining,- parseRow,+ -- * Row parsing+ RowParser (..),+ field,+ fieldWith,+ numFieldsRemaining,+ parseRow, - -- * Generic derivation- GFromRow (..),- FromRow (..),+ -- * Generic derivation+ GFromRow (..),+ FromRow (..), - -- * Error conversion- resultErrorToSqlError,- rowErrorsToSqlError,+ -- * Error conversion+ resultErrorToSqlError,+ rowErrorsToSqlError, ) where import Control.Applicative (Alternative (..))-import Control.Exception (Exception, SomeException (SomeException), fromException, toException)+import Control.Exception (Exception (..), SomeException, fromException) import Control.Monad (MonadPlus, replicateM) import Control.Monad.Trans.Class (lift) import Control.Monad.Trans.Reader (ReaderT, ask, runReaderT) import Control.Monad.Trans.State.Strict (StateT, get, put, runStateT) import Data.Maybe (listToMaybe, mapMaybe)-import Data.Text (Text) import qualified Data.Text as Text-import GHC.Generics-+import Database.DuckDB.Simple.Conversion (Conversion, conversionError) import Database.DuckDB.Simple.FromField import Database.DuckDB.Simple.Internal (SQLError (..))-import Database.DuckDB.Simple.Ok (Ok (..)) import Database.DuckDB.Simple.Types (Only (..), Query, (:.) (..))+import GHC.Generics -- | Row parsing environment (read-only data available to the parser). newtype RowParseRO = RowParseRO- { rowParseColumnCount :: Int- }+ { rowParseColumnCount :: Int+ } -- | Column-out-of-bounds sentinel used internally to map parser failures. newtype ColumnOutOfBounds = ColumnOutOfBounds {columnOutOfBoundsIndex :: Int}- deriving stock (Eq, Show)+ deriving stock (Eq, Show) instance Exception ColumnOutOfBounds -- | Parser used by @FromRow@ implementations. newtype RowParser a = RowParser- { runRowParser :: ReaderT RowParseRO (StateT (Int, [Field]) Ok) a- }- deriving stock (Functor)- deriving newtype (Applicative, Alternative, Monad, MonadPlus)+ { runRowParser :: ReaderT RowParseRO (StateT (Int, [Field]) Conversion) a+ }+ deriving stock (Functor)+ deriving newtype (Applicative, Alternative, Monad, MonadPlus) -- | Generic derivation helper mirroring @sqlite-simple@. class GFromRow f where- gFromRow :: RowParser (f p)+ gFromRow :: RowParser (f p) instance GFromRow U1 where- gFromRow = pure U1+ gFromRow = pure U1 instance (GFromRow a) => GFromRow (M1 i c a) where- gFromRow = M1 <$> gFromRow+ gFromRow = M1 <$> gFromRow instance (FromField a) => GFromRow (K1 i a) where- gFromRow = K1 <$> field+ gFromRow = K1 <$> field instance (GFromRow a, GFromRow b) => GFromRow (a :*: b) where- gFromRow = (:*:) <$> gFromRow <*> gFromRow+ gFromRow = (:*:) <$> gFromRow <*> gFromRow -- | Types that can be constructed from database rows. class FromRow a where- fromRow :: RowParser a- default fromRow :: (Generic a, GFromRow (Rep a)) => RowParser a- fromRow = to <$> gFromRow+ fromRow :: RowParser a+ default fromRow :: (Generic a, GFromRow (Rep a)) => RowParser a+ fromRow = to <$> gFromRow --- | Pull the next field using the provided @FieldParser@.+-- | Read the next field with the supplied @FieldParser@. fieldWith :: FieldParser a -> RowParser a+{-# INLINE fieldWith #-} fieldWith fieldParser = RowParser $ do- RowParseRO{rowParseColumnCount} <- ask- (columnIndex, remaining) <- lift get- case remaining of- [] -> do- lift (put (columnIndex + 1, []))- lift (lift (Errors [toException (ColumnOutOfBounds (columnIndex + 1))]))- (f : rest) -> do- lift (put (columnIndex + 1, rest))- if columnIndex >= rowParseColumnCount- then lift (lift (Errors [toException (ColumnOutOfBounds (columnIndex + 1))]))- else case fieldParser f of- Errors err -> lift $ lift $ Errors err- Ok value -> pure value+ RowParseRO{rowParseColumnCount} <- ask+ (columnIndex, remaining) <- lift get+ case remaining of+ [] -> do+ lift (put (columnIndex + 1, []))+ lift (lift (conversionError (ColumnOutOfBounds (columnIndex + 1))))+ (f : rest) -> do+ lift (put (columnIndex + 1, rest))+ if columnIndex >= rowParseColumnCount+ then lift (lift (conversionError (ColumnOutOfBounds (columnIndex + 1))))+ else lift (lift (fieldParser f)) --- | Pull the next field and parse it using its @FromField@ instance.+-- | Read the next field with its @FromField@ instance. field :: (FromField a) => RowParser a+{-# INLINE field #-} field = fieldWith fromField -- | Report how many columns remain unread in the current row. numFieldsRemaining :: RowParser Int numFieldsRemaining = RowParser $ do- RowParseRO{rowParseColumnCount} <- ask- (columnIndex, _) <- lift get- pure (rowParseColumnCount - columnIndex)+ RowParseRO{rowParseColumnCount} <- ask+ (columnIndex, _) <- lift get+ pure (rowParseColumnCount - columnIndex) -- | Execute a @RowParser@ against the provided row.-parseRow :: RowParser a -> [Field] -> Ok a+parseRow :: RowParser a -> [Field] -> Conversion a+{-# INLINE parseRow #-} parseRow parser fields =- let context = RowParseRO (length fields)- initialState = (0, fields)- in case runStateT (runReaderT (runRowParser parser) context) initialState of- Ok (value, (columnCount, _))- | columnCount == length fields -> Ok value- | otherwise -> Errors [SomeException $ ColumnOutOfBounds (columnCount + 1)]- Errors errs -> Errors errs+ let context = RowParseRO (length fields)+ initialState = (0, fields)+ in do+ (value, (columnCount, _)) <- runStateT (runReaderT (runRowParser parser) context) initialState+ if columnCount == length fields+ then pure value+ else conversionError (ColumnOutOfBounds (columnCount + 1)) instance FromRow () where- fromRow = pure ()+ fromRow = pure () instance (FromField a) => FromRow (Only a) where- fromRow = Only <$> field+ fromRow = Only <$> field instance (FromRow a, FromRow b) => FromRow (a :. b) where- fromRow = (:.) <$> fromRow <*> fromRow+ fromRow = (:.) <$> fromRow <*> fromRow instance (FromField a, FromField b) => FromRow (a, b) where- fromRow = (,) <$> field <*> field+ fromRow = (,) <$> field <*> field instance (FromField a, FromField b, FromField c) => FromRow (a, b, c) where- fromRow = (,,) <$> field <*> field <*> field+ fromRow = (,,) <$> field <*> field <*> field instance (FromField a, FromField b, FromField c, FromField d) => FromRow (a, b, c, d) where- fromRow = (,,,) <$> field <*> field <*> field <*> field+ fromRow = (,,,) <$> field <*> field <*> field <*> field instance (FromField a, FromField b, FromField c, FromField d, FromField e) => FromRow (a, b, c, d, e) where- fromRow = (,,,,) <$> field <*> field <*> field <*> field <*> field+ fromRow = (,,,,) <$> field <*> field <*> field <*> field <*> field instance (FromField a, FromField b, FromField c, FromField d, FromField e, FromField f) => FromRow (a, b, c, d, e, f) where- fromRow = (,,,,,) <$> field <*> field <*> field <*> field <*> field <*> field+ fromRow = (,,,,,) <$> field <*> field <*> field <*> field <*> field <*> field instance- (FromField a, FromField b, FromField c, FromField d, FromField e, FromField f, FromField g) =>- FromRow (a, b, c, d, e, f, g)- where- fromRow = (,,,,,,) <$> field <*> field <*> field <*> field <*> field <*> field <*> field+ (FromField a, FromField b, FromField c, FromField d, FromField e, FromField f, FromField g)+ => FromRow (a, b, c, d, e, f, g)+ where+ fromRow = (,,,,,,) <$> field <*> field <*> field <*> field <*> field <*> field <*> field instance- (FromField a, FromField b, FromField c, FromField d, FromField e, FromField f, FromField g, FromField h) =>- FromRow (a, b, c, d, e, f, g, h)- where- fromRow = (,,,,,,,) <$> field <*> field <*> field <*> field <*> field <*> field <*> field <*> field+ (FromField a, FromField b, FromField c, FromField d, FromField e, FromField f, FromField g, FromField h)+ => FromRow (a, b, c, d, e, f, g, h)+ where+ fromRow = (,,,,,,,) <$> field <*> field <*> field <*> field <*> field <*> field <*> field <*> field instance- ( FromField a- , FromField b- , FromField c- , FromField d- , FromField e- , FromField f- , FromField g- , FromField h- , FromField i- ) =>- FromRow (a, b, c, d, e, f, g, h, i)- where- fromRow =- (,,,,,,,,)- <$> field- <*> field- <*> field- <*> field- <*> field- <*> field- <*> field- <*> field- <*> field+ ( FromField a+ , FromField b+ , FromField c+ , FromField d+ , FromField e+ , FromField f+ , FromField g+ , FromField h+ , FromField i+ )+ => FromRow (a, b, c, d, e, f, g, h, i)+ where+ fromRow =+ (,,,,,,,,)+ <$> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field instance- ( FromField a- , FromField b- , FromField c- , FromField d- , FromField e- , FromField f- , FromField g- , FromField h- , FromField i- , FromField j- ) =>- FromRow (a, b, c, d, e, f, g, h, i, j)- where- fromRow =- (,,,,,,,,,)- <$> field- <*> field- <*> field- <*> field- <*> field- <*> field- <*> field- <*> field- <*> field- <*> field+ ( FromField a+ , FromField b+ , FromField c+ , FromField d+ , FromField e+ , FromField f+ , FromField g+ , FromField h+ , FromField i+ , FromField j+ )+ => FromRow (a, b, c, d, e, f, g, h, i, j)+ where+ fromRow =+ (,,,,,,,,,)+ <$> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field+ <*> field instance (FromField a) => FromRow [a] where- fromRow = do- remaining <- numFieldsRemaining- replicateM remaining field+ fromRow = do+ remaining <- numFieldsRemaining+ replicateM remaining field -- | Convert a @ResultError@ into a user-facing @SQLError@. resultErrorToSqlError :: Query -> ResultError -> SQLError resultErrorToSqlError query err =- SQLError- { sqlErrorMessage = renderError err- , sqlErrorType = Nothing- , sqlErrorQuery = Just query- }+ SQLError+ { sqlErrorMessage = Text.pack (displayException err)+ , sqlErrorType = Nothing+ , sqlErrorQuery = Just query+ , sqlErrorBody = Nothing+ } --- | Collapse parser failure diagnostics into an @SQLError@ while preserving the query.+-- | Convert parser errors to an @SQLError@ that retains the query. rowErrorsToSqlError :: Query -> [SomeException] -> SQLError rowErrorsToSqlError query errs =- case listToMaybe (mapMaybe (fromException :: SomeException -> Maybe ResultError) errs) of- Just resultErr -> resultErrorToSqlError query resultErr+ case listToMaybe (mapMaybe (fromException :: SomeException -> Maybe ResultError) errs) of+ Just resultErr -> resultErrorToSqlError query resultErr+ Nothing ->+ case listToMaybe (mapMaybe (fromException :: SomeException -> Maybe ColumnOutOfBounds) errs) of+ Just (ColumnOutOfBounds idx) ->+ SQLError+ { sqlErrorMessage =+ Text.concat+ [ "duckdb-simple: column index "+ , Text.pack (show idx)+ , " out of bounds"+ ]+ , sqlErrorType = Nothing+ , sqlErrorQuery = Just query+ , sqlErrorBody = Nothing+ } Nothing ->- case listToMaybe (mapMaybe (fromException :: SomeException -> Maybe ColumnOutOfBounds) errs) of- Just (ColumnOutOfBounds idx) ->- SQLError- { sqlErrorMessage =- Text.concat- [ "duckdb-simple: column index "- , Text.pack (show idx)- , " out of bounds"- ]- , sqlErrorType = Nothing- , sqlErrorQuery = Just query- }- Nothing ->- SQLError- { sqlErrorMessage =- Text.pack $ "duckdb-simple: row-parsing failed:" <> show errs- , sqlErrorType = Nothing- , sqlErrorQuery = Just query- }--renderError :: ResultError -> Text-renderError = \case- Incompatible{errSQLType, errSQLField, errHaskellType, errMessage} ->- Text.concat- [ "duckdb-simple: column "- , errSQLField- , " has type "- , errSQLType- , " but expected "- , errHaskellType- , if Text.null errMessage- then ""- else ": " <> errMessage- ]- UnexpectedNull{errHaskellType, errSQLField, errMessage} ->- Text.concat- [ "duckdb-simple: column "- , errSQLField- , " is NULL but expected "- , errHaskellType- , if Text.null errMessage- then ""- else ": " <> errMessage- ]- ConversionFailed{errMessage} ->- errMessage+ SQLError+ { sqlErrorMessage =+ Text.pack $ "duckdb-simple: row-parsing failed:" <> show errs+ , sqlErrorType = Nothing+ , sqlErrorQuery = Just query+ , sqlErrorBody = Nothing+ }
src/Database/DuckDB/Simple/Function.hs view
@@ -1,456 +1,246 @@ {-# LANGUAGE BlockArguments #-}-{-# LANGUAGE FlexibleInstances #-}-{-# LANGUAGE LambdaCase #-}-{-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE RankNTypes #-} {-# LANGUAGE ScopedTypeVariables #-}-{-# LANGUAGE UndecidableInstances #-} -{- |-Module : Database.DuckDB.Simple.Function-Description : Register scalar Haskell functions with DuckDB connections.--This module mirrors the high-level API provided by @sqlite-simple@ for user-defined functions, adapted to DuckDB's chunked execution model. It allows-pure and 'IO'-based Haskell functions to be exposed to SQL while reusing the-existing field-decoding and result-marshalling machinery for arguments and-return values.--}+-- |+-- Module : Database.DuckDB.Simple.Function+-- Description : Register Haskell scalar functions, table functions, and aggregate functions with DuckDB connections.+--+-- Register pure and IO scalar functions with the managed argument and result+-- conversions. Register table functions and aggregate functions that convert+-- their arguments with 'FromRow'.+--+-- A program that registers a callback must be linked with @-threaded@. DuckDB+-- can call a callback from several of its threads at the same time, and the+-- runtime of GHC without @-threaded@ does not support that: the program can+-- crash. Without @-threaded@, the registration of a function in this module+-- raises an t'Database.DuckDB.Simple.SQLError' of+-- the type 'Database.DuckDB.Simple.ErrorInputInvalid' when the setting+-- @threads@ is above 1. The section \"User-Defined Functions\" of the README+-- gives the rule.+--+-- A callback must not run a statement on the connection that runs the callback.+-- Use another connection of the same instance. This rule applies to each+-- registered callback, for example of a scalar function, a cast, an aggregate+-- function, a table function, a COPY format, a replacement scan, a pragma, or a+-- log sink. The library holds the lock of a connection while it prepares a+-- statement, while 'Database.DuckDB.Simple.query',+-- 'Database.DuckDB.Simple.execute', or a default fold runs a statement, and+-- while it fetches a chunk of an explicit stream. DuckDB can run a callback in+-- each of these steps. The lock is not reentrant. Thus a statement in the+-- callback waits for the lock. With more than one DuckDB thread, DuckDB can also+-- run a callback of an explicit stream between two fetches, while no fetch holds+-- the lock. Then a statement in the callback on the same connection fails at+-- once with the error of DuckDB for a connection with an open result.+--+-- The wait has the bound 'Database.DuckDB.Simple.connectionCallbackLockWait' of+-- the connection: 10 seconds by default, and no bound for 'Nothing'. After the+-- bound, the statement raises an t'Database.DuckDB.Simple.SQLError' of the type+-- 'Database.DuckDB.Simple.ErrorResourceInUse'. For most callbacks, the outer+-- statement then fails. DuckDB ignores the error of a log sink. The bound+-- applies to each wait of a callback for a connection lock, also for a second+-- connection that another statement uses. The bound does not apply to+-- a callback that DuckDB runs while the library holds no lock of the+-- connection, for example a log sink for a message of+-- 'Database.DuckDB.Simple.Logging.logConnectionMessage'. It also does not apply+-- to a statement in a thread that the callback starts, to a statement inside+-- 'Control.Exception.uninterruptibleMask', and to a statement in the flush+-- function of a log sink when DuckDB destroys the sink.+--+-- A callback that refers to a connection keeps that connection and its instance+-- reachable while DuckDB keeps the callback, possibly until the database closes.+-- Thus the garbage collector does not close them, and the program must close+-- them. module Database.DuckDB.Simple.Function (- ScalarType,- ScalarValue,- FunctionArg (),- FunctionResult (),- Function (..),- createFunction,- createFunctionWithState,- deleteFunction,+ ScalarValue,+ FunctionResult (),+ Function (..),+ LogicalTypeId (..),+ createFunction,+ createFunctionWithState,+ deleteFunction,+ TypeDeclaration (..),+ TypedRow,+ createTableFunction,+ Aggregate (..),+ createAggregate, ) where -import Control.Exception (- SomeException,- bracket,- throwIO,- try,- )-import Control.Monad (forM, forM_, when)-import Data.Int (Int16, Int32, Int64)-import Data.Proxy (Proxy (..))+import Control.Exception (bracket)+import Control.Monad (forM) import Data.Text (Text) import qualified Data.Text as Text-import qualified Data.Text.Foreign as TextForeign-import Data.Word (Word16, Word32, Word64, Word8)-import Database.DuckDB.FFI-import Database.DuckDB.Simple.Callback (runCallback, transferCallbackState, withCallbackResources)-import Database.DuckDB.Simple.FromField (- Field (..),- FromField (..),+import Database.DuckDB.Simple.Callback (RegistrationTarget (ConnectionTarget))+import Database.DuckDB.Simple.FromRow (FromRow)+import Database.DuckDB.Simple.Function.Internal (+ Function (..),+ FunctionResult,+ ScalarValue,+ createFunctionOn,+ createFunctionWithStateOn, ) import Database.DuckDB.Simple.Internal (- Connection,- Query (..),- SQLError (..),- destroyLogicalType,- withConnectionHandle,- withQueryCString,- withResult,+ Connection,+ executeParsedStatement,+ withParsedStatements, )-import Database.DuckDB.Simple.Materialize (prepareVectorReader)-import Database.DuckDB.Simple.Ok (Ok (..))-import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (FunPtr, Ptr, castPtr, nullPtr)-import Foreign.StablePtr (StablePtr, castPtrToStablePtr, deRefStablePtr)-import Foreign.Storable (poke, pokeElemOff)-import GHC.Float (float2Double)--data ScalarFunctionResources = ScalarFunctionResources- { scalarFunctionExecPtr :: !DuckDBScalarFunctionFun- , scalarFunctionInitPtr :: !(Maybe DuckDBScalarFunctionInitFun)- }---- | Tag DuckDB logical types we support for scalar return values.-data ScalarType- = ScalarTypeBoolean- | ScalarTypeBigInt- | ScalarTypeUBigInt- | ScalarTypeDouble- | ScalarTypeVarchar---- | Runtime representation of values returned to DuckDB.-data ScalarValue- = ScalarNull- | ScalarBoolean !Bool- | ScalarInteger !Int64- | ScalarUnsigned !Word64- | ScalarDouble !Double- | ScalarText !Text---- | Class of scalar results that can be produced by user-defined functions.-class FunctionResult a where- scalarReturnType :: Proxy a -> ScalarType- toScalarValue :: a -> IO ScalarValue--instance FunctionResult Int where- scalarReturnType _ = ScalarTypeBigInt- toScalarValue value = pure (ScalarInteger (fromIntegral value))--instance FunctionResult Int16 where- scalarReturnType _ = ScalarTypeBigInt- toScalarValue value = pure (ScalarInteger (fromIntegral value))--instance FunctionResult Int32 where- scalarReturnType _ = ScalarTypeBigInt- toScalarValue value = pure (ScalarInteger (fromIntegral value))--instance FunctionResult Int64 where- scalarReturnType _ = ScalarTypeBigInt- toScalarValue value = pure (ScalarInteger value)--instance FunctionResult Word where- scalarReturnType _ = ScalarTypeUBigInt- toScalarValue value = pure (ScalarUnsigned (fromIntegral value))--instance FunctionResult Word16 where- scalarReturnType _ = ScalarTypeBigInt- toScalarValue value = pure (ScalarInteger (fromIntegral value))--instance FunctionResult Word32 where- scalarReturnType _ = ScalarTypeBigInt- toScalarValue value = pure (ScalarInteger (fromIntegral value))--instance FunctionResult Word64 where- scalarReturnType _ = ScalarTypeUBigInt- toScalarValue value = pure (ScalarUnsigned (fromIntegral value))--instance FunctionResult Double where- scalarReturnType _ = ScalarTypeDouble- toScalarValue value = pure (ScalarDouble value)--instance FunctionResult Float where- scalarReturnType _ = ScalarTypeDouble- toScalarValue value = pure (ScalarDouble (float2Double value))--instance FunctionResult Bool where- scalarReturnType _ = ScalarTypeBoolean- toScalarValue value = pure (ScalarBoolean value)--instance FunctionResult Text where- scalarReturnType _ = ScalarTypeVarchar- toScalarValue value = pure (ScalarText value)--instance FunctionResult String where- scalarReturnType _ = ScalarTypeVarchar- toScalarValue value = pure (ScalarText (Text.pack value))--instance (FunctionResult a) => FunctionResult (Maybe a) where- scalarReturnType _ = scalarReturnType (Proxy :: Proxy a)- toScalarValue Nothing = pure ScalarNull- toScalarValue (Just value) = toScalarValue value---- | Argument types supported by the scalar function machinery.-class FunctionArg a where- argumentType :: Proxy a -> DuckDBType--instance FunctionArg Int where- argumentType _ = DuckDBTypeBigInt--instance FunctionArg Int16 where- argumentType _ = DuckDBTypeSmallInt--instance FunctionArg Int32 where- argumentType _ = DuckDBTypeInteger--instance FunctionArg Int64 where- argumentType _ = DuckDBTypeBigInt--instance FunctionArg Word where- argumentType _ = DuckDBTypeUBigInt--instance FunctionArg Word16 where- argumentType _ = DuckDBTypeUSmallInt--instance FunctionArg Word32 where- argumentType _ = DuckDBTypeUInteger--instance FunctionArg Word64 where- argumentType _ = DuckDBTypeUBigInt--instance FunctionArg Double where- argumentType _ = DuckDBTypeDouble--instance FunctionArg Float where- argumentType _ = DuckDBTypeFloat--instance FunctionArg Bool where- argumentType _ = DuckDBTypeBoolean--instance FunctionArg Text where- argumentType _ = DuckDBTypeVarchar--instance FunctionArg String where- argumentType _ = DuckDBTypeVarchar--instance (FunctionArg a) => FunctionArg (Maybe a) where- argumentType _ = argumentType (Proxy :: Proxy a)---- | Typeclass describing Haskell functions that can be exposed to DuckDB.-class Function a where- argumentTypes :: Proxy a -> [DuckDBType]- returnType :: Proxy a -> ScalarType- isVolatile :: Proxy a -> Bool- applyFunction :: [Field] -> a -> IO ScalarValue--instance {-# OVERLAPPABLE #-} (FunctionResult a) => Function a where- argumentTypes _ = []- returnType _ = scalarReturnType (Proxy :: Proxy a)- isVolatile _ = False- applyFunction [] value = toScalarValue value- applyFunction _ _ = throwIO (functionInvocationError (Text.pack "unexpected arguments supplied"))--instance {-# OVERLAPPING #-} (FunctionResult a) => Function (IO a) where- argumentTypes _ = []- returnType _ = scalarReturnType (Proxy :: Proxy a)- isVolatile _ = True- applyFunction [] action = action >>= toScalarValue- applyFunction _ _ = throwIO (functionInvocationError (Text.pack "unexpected arguments supplied"))--instance {-# OVERLAPPABLE #-} (FromField a, FunctionArg a, Function r) => Function (a -> r) where- argumentTypes _ = argumentType (Proxy :: Proxy a) : argumentTypes (Proxy :: Proxy r)- returnType _ = returnType (Proxy :: Proxy r)- isVolatile _ = isVolatile (Proxy :: Proxy r)- applyFunction [] _ =- throwIO (functionInvocationError (Text.pack "insufficient arguments supplied"))- applyFunction (field : rest) fn =- case fromField field of- Errors err -> throwIO (argumentConversionError (fieldIndex field) err)- Ok value -> applyFunction rest (fn value)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.Typed (TypedRow)+import Database.DuckDB.Simple.Typed.Aggregate (Aggregate (..), createAggregateOn)+import Database.DuckDB.Simple.Typed.Table (createTableFunctionOn)+import Database.DuckDB.Simple.V2.Registration (TypeDeclaration (..)) -- | Register a Haskell function under the supplied name.-createFunction :: forall f. (Function f) => Connection -> Text -> f -> IO ()-createFunction conn name fn =- registerScalarFunction conn name (Proxy :: Proxy f) \allocate -> do- scalarFunctionExecPtr <- allocate (mkScalarFun (scalarFunctionHandler fn))- pure ScalarFunctionResources{scalarFunctionExecPtr, scalarFunctionInitPtr = Nothing}---- | Register a scalar function with per-worker thread-local state.-createFunctionWithState :: forall s f. (Function f) => Connection -> Text -> IO s -> (s -> f) -> IO ()-createFunctionWithState conn name initState mkFn =- registerScalarFunction conn name (Proxy :: Proxy f) \allocate -> do- scalarFunctionExecPtr <- allocate (mkScalarFun (scalarFunctionHandlerWithState mkFn))- initPtr <- allocate (mkScalarInitFun (scalarFunctionInitHandler initState))- pure ScalarFunctionResources{scalarFunctionExecPtr, scalarFunctionInitPtr = Just initPtr}---- | Configure a scalar function and transfer its callbacks to DuckDB.-registerScalarFunction :: (Function f) => Connection -> Text -> Proxy f -> ((forall a. IO (FunPtr a) -> IO (FunPtr a)) -> IO ScalarFunctionResources) -> IO ()-registerScalarFunction conn name proxy acquire = do- when (Text.null name || Text.any (== '\0') name) $- throwIO (functionInvocationError "duckdb-simple: invalid scalar function name")- bracket c_duckdb_create_scalar_function cleanupScalarFunction \scalarFun -> do- when (scalarFun == nullPtr) $- throwIO (functionInvocationError "duckdb-simple: failed to allocate scalar function")- withCallbackResources- acquire- (c_duckdb_scalar_function_set_extra_info scalarFun)- \ScalarFunctionResources{scalarFunctionExecPtr, scalarFunctionInitPtr} -> do- TextForeign.withCString name $ c_duckdb_scalar_function_set_name scalarFun- forM_ (argumentTypes proxy) \dtype ->- withLogicalType dtype $ c_duckdb_scalar_function_add_parameter scalarFun- withLogicalType (duckTypeForScalar (returnType proxy)) $ c_duckdb_scalar_function_set_return_type scalarFun- when (isVolatile proxy) $ c_duckdb_scalar_function_set_volatile scalarFun- c_duckdb_scalar_function_set_special_handling scalarFun- c_duckdb_scalar_function_set_function scalarFun scalarFunctionExecPtr- forM_ scalarFunctionInitPtr $ c_duckdb_scalar_function_set_init scalarFun- withConnectionHandle conn \connPtr -> do- rc <- c_duckdb_register_scalar_function connPtr scalarFun- when (rc /= DuckDBSuccess) $- throwIO (functionInvocationError "duckdb-simple: registering function failed")---- | Drop a previously registered scalar function by issuing a DROP FUNCTION statement.-deleteFunction :: Connection -> Text -> IO ()-deleteFunction conn name =- do- outcome <-- try $- withConnectionHandle conn \connPtr -> do- let dropQuery =- Query $- Text.concat- [ Text.pack "DROP FUNCTION IF EXISTS "- , qualifyIdentifier name- ]- withQueryCString dropQuery \sql ->- withResult conn dropQuery (c_duckdb_query connPtr sql) (const (pure ()))- case outcome of- Right () -> pure ()- Left err- -- DuckDB does not allow dropping scalar functions registered via the C API,- -- so we ignore that specific error here.- -- TODO: Update this when DuckDB adds support for dropping such functions.- | Text.isInfixOf (Text.pack "Cannot drop internal catalog entry") (sqlErrorMessage err) -> return ()- | otherwise -> throwIO err--cleanupScalarFunction :: DuckDBScalarFunction -> IO ()-cleanupScalarFunction scalarFun =- alloca \ptr -> do- poke ptr scalarFun- c_duckdb_destroy_scalar_function ptr--withLogicalType :: DuckDBType -> (DuckDBLogicalType -> IO a) -> IO a-withLogicalType dtype =- bracket- ( do- logical <- c_duckdb_create_logical_type dtype- when (logical == nullPtr)- $ throwIO- $ functionInvocationError (Text.pack "duckdb-simple: failed to allocate logical type")- pure logical- )- destroyLogicalType--duckTypeForScalar :: ScalarType -> DuckDBType-duckTypeForScalar = \case- ScalarTypeBoolean -> DuckDBTypeBoolean- ScalarTypeBigInt -> DuckDBTypeBigInt- ScalarTypeUBigInt -> DuckDBTypeUBigInt- ScalarTypeDouble -> DuckDBTypeDouble- ScalarTypeVarchar -> DuckDBTypeVarchar--scalarFunctionHandler :: forall f. (Function f) => f -> DuckDBFunctionInfo -> DuckDBDataChunk -> DuckDBVector -> IO ()-scalarFunctionHandler fn info chunk outVec =- runCallback (c_duckdb_scalar_function_set_error info) do- rawColumnCount <- c_duckdb_data_chunk_get_column_count chunk- let columnCount = fromIntegral rawColumnCount :: Int- expected = length (argumentTypes (Proxy :: Proxy f))- when (columnCount /= expected)- $ throwIO- $ functionInvocationError- $ Text.concat- [ Text.pack "duckdb-simple: function expected "- , Text.pack (show expected)- , Text.pack " arguments but received "- , Text.pack (show columnCount)- ]- rawRowCount <- c_duckdb_data_chunk_get_size chunk- let rowCount = fromIntegral rawRowCount :: Int- readers <- mapM (makeColumnReader chunk) [0 .. expected - 1]- rows <-- forM [0 .. rowCount - 1] \row ->- forM readers \reader ->- reader (fromIntegral row)- results <- mapM (`applyFunction` fn) rows- writeResults (returnType (Proxy :: Proxy f)) results outVec--scalarFunctionHandlerWithState :: forall s f. (Function f) => (s -> f) -> DuckDBFunctionInfo -> DuckDBDataChunk -> DuckDBVector -> IO ()-scalarFunctionHandlerWithState mkFn info chunk outVec =- runCallback (c_duckdb_scalar_function_set_error info) do- statePtr <- c_duckdb_scalar_function_get_state info- when (statePtr == nullPtr) $- throwIO (functionInvocationError "duckdb-simple: scalar function state was not initialised")- state <- deRefStablePtr (castPtrToStablePtr statePtr :: StablePtr s)- scalarFunctionHandler (mkFn state) info chunk outVec--scalarFunctionInitHandler :: IO s -> DuckDBInitInfo -> IO ()-scalarFunctionInitHandler initState info =- runCallback (c_duckdb_scalar_function_init_set_error info) do- state <- initState- transferCallbackState (c_duckdb_scalar_function_init_set_state info) state--type ColumnReader = DuckDBIdx -> IO Field--makeColumnReader :: DuckDBDataChunk -> Int -> IO ColumnReader-makeColumnReader chunk columnIndex = do- readValue <- c_duckdb_data_chunk_get_vector chunk (fromIntegral columnIndex) >>= prepareVectorReader- let name = Text.pack ("arg" <> show columnIndex)- pure \rowIdx -> do- value <- readValue (fromIntegral rowIdx)- pure- Field- { fieldName = name- , fieldIndex = columnIndex- , fieldValue = value- }-writeResults :: ScalarType -> [ScalarValue] -> DuckDBVector -> IO ()-writeResults resultType values outVec = do- let hasNulls = any isNullValue values- when hasNulls $- c_duckdb_vector_ensure_validity_writable outVec- dataPtr <- c_duckdb_vector_get_data outVec- validityPtr <- c_duckdb_vector_get_validity outVec- forM_ (zip [0 ..] values) \(idx, val) ->- case (resultType, val) of- (_, ScalarNull) ->- markInvalid validityPtr idx- (ScalarTypeBoolean, ScalarBoolean flag) -> do- markValid validityPtr idx- pokeElemOff (castPtr dataPtr :: Ptr Word8) idx (if flag then 1 else 0)- (ScalarTypeBigInt, ScalarInteger intval) -> do- markValid validityPtr idx- pokeElemOff (castPtr dataPtr :: Ptr Int64) idx intval- (ScalarTypeUBigInt, ScalarUnsigned intval) -> do- markValid validityPtr idx- pokeElemOff (castPtr dataPtr :: Ptr Word64) idx intval- (ScalarTypeDouble, ScalarDouble dbl) -> do- markValid validityPtr idx- pokeElemOff (castPtr dataPtr :: Ptr Double) idx dbl- (ScalarTypeVarchar, ScalarText txt) -> do- markValid validityPtr idx- TextForeign.withCStringLen txt \(ptr, len) ->- c_duckdb_vector_assign_string_element_len outVec (fromIntegral idx) ptr (fromIntegral len)- _ ->- throwIO- $ functionInvocationError- $ Text.pack "duckdb-simple: result type mismatch when materialising scalar function output"--markInvalid :: Ptr Word64 -> Int -> IO ()-markInvalid validity idx- | validity == nullPtr = pure ()- | otherwise = c_duckdb_validity_set_row_invalid validity (fromIntegral idx)--markValid :: Ptr Word64 -> Int -> IO ()-markValid validity idx- | validity == nullPtr = pure ()- | otherwise = c_duckdb_validity_set_row_valid validity (fromIntegral idx)--isNullValue :: ScalarValue -> Bool-isNullValue = \case- ScalarNull -> True- _ -> False--argumentConversionError :: Int -> [SomeException] -> SQLError-argumentConversionError idx err =- let message =- Text.concat- [ Text.pack "duckdb-simple: unable to convert argument #"- , Text.pack (show (idx + 1))- , Text.pack ": "- , Text.pack (show err)- ]- in functionInvocationError message+--+-- The result type needs a t'Database.DuckDB.Simple.ToField.ToDuckValue'+-- instance, and the SQL type of the result is its+-- 'Database.DuckDB.Simple.ToField.duckdbTypeDeclaration'. Thus an+-- 'Data.Int.Int16' result is SMALLINT and a 'Float' result is FLOAT. DuckDB+-- computes with that type: the product of two SMALLINT values is SMALLINT, so+-- @f(x) * f(y)@ of an 'Data.Int.Int16' function raises an Out of Range Error+-- when the product does not fit. Cast a result to a wider type first, for+-- example @f(x)::INTEGER * f(y)@. A 'Bool', an integer of 8 to 64 bits, a+-- 'Float', a 'Double', a 'Text', or a 'String' result goes to the memory of the+-- result vector. Each other result type gives a native value for each row+-- through the context of the call. DuckDB casts that value to the declared+-- type, as a SQL @CAST@ does: it rounds a value to the scale of a DECIMAL type+-- text, and it raises an error for a value that does not fit.+--+-- Each argument type needs t'Database.DuckDB.Simple.FromField.FromField' and+-- t'Database.DuckDB.Simple.ToField.DuckDBColumnType' instances. The SQL type of+-- an argument is the 'Database.DuckDB.Simple.ToField.duckdbTypeDeclaration' of+-- its type. An argument type with no declaration raises an @SQLError@ at+-- registration. Arguments use the same conversion as query results, including+-- nested values.+createFunction :: (Function f) => Connection -> Text -> f -> IO ()+createFunction conn = createFunctionOn (ConnectionTarget conn) -functionInvocationError :: Text -> SQLError-functionInvocationError message =- SQLError- { sqlErrorMessage = message- , sqlErrorType = Nothing- , sqlErrorQuery = Nothing- }+-- | Register a scalar function with per-worker thread-local state. The+-- arguments have the rules of 'createFunction'.+createFunctionWithState :: (Function f) => Connection -> Text -> IO s -> (s -> f) -> IO ()+createFunctionWithState conn = createFunctionWithStateOn (ConnectionTarget conn) -qualifyIdentifier :: Text -> Text-qualifyIdentifier rawName =- let parts = Text.splitOn "." rawName- in Text.intercalate (Text.pack ".") (map quoteIdent parts)+-- | Register a table function. The function gives the rows of one scan as a+-- list. The list of names gives the names of the result columns. 'TypedRow' of+-- @row@ gives the column types. A different number of names and columns, or a+-- row with no column, is an error at registration. A @Maybe@ column gives NULL+-- for 'Nothing'. The SQL type of each parameter and each column is the+-- 'Database.DuckDB.Simple.ToField.duckdbTypeDeclaration' of its type. A record+-- or a sum type with 'Database.DuckDB.Simple.Generic.ViaDuckDB' instances is a+-- STRUCT or a UNION column. A type with no declaration, for example+-- 'Database.DuckDB.Simple.LogicalRep.StructValue', is an error at registration.+--+-- 'TypedRow' of @params@ gives the parameter types. Use @()@ for no parameter.+-- The parameters are positional only. DuckDB casts each argument to its+-- parameter type. Then 'FromRow' converts the arguments. An argument that does+-- not convert is an SQL error that names the function. A NULL argument converts+-- only to a @Maybe@ field.+--+-- DuckDB converts the arguments when it plans the query. Thus an argument that+-- does not convert also fails the preparation of a statement. The function runs+-- one time for each scan, when the execution of the statement starts. It does not+-- run at preparation. A query that reads the function two times runs it two+-- times. A prepared statement runs it at each execution.+--+-- DuckDB reads the list one chunk at a time. It can run two exec callbacks of+-- one scan at the same time, for example for a multi-file function. Each exec+-- callback takes its rows from the list under a lock, so two exec callbacks never+-- read the list at the same time. DuckDB stops when the query needs no more rows,+-- so the list can be long or infinite, for example with @LIMIT@. It reads a full+-- chunk, and it can read more chunks than the query needs. Thus it can read rows+-- after the @LIMIT@. DuckDB asks only for the columns that the query uses, also+-- in @WHERE@ and @ORDER BY@. When the query uses no column, as in+-- @SELECT count(*)@, DuckDB asks for the first column. The library evaluates+-- only the columns that DuckDB asks for, so an exception in another column does+-- not fail the query. But the library evaluates each row that it writes to its+-- constructor, and this evaluates each strict field of the row (a field with+-- @!@, or each field with @StrictData@). Thus an exception in a strict field+-- fails the query, also when the query does not use the column. An exception+-- in the function, in the list, in a row, or in a column that the query uses+-- is an SQL error. DuckDB releases the list when the scan ends.+--+-- > createTableFunction conn "haskell_squares" ["n", "square"] \(Only count) ->+-- > pure [(n, n * n) | n <- [0 .. count - 1 :: Int64]]+createTableFunction :: (FromRow params, TypedRow params, TypedRow row) => Connection -> Text -> [Text] -> (params -> IO [row]) -> IO ()+createTableFunction conn = createTableFunctionOn (ConnectionTarget conn) -quoteIdent :: Text -> Text-quoteIdent ident =- Text.concat- [ Text.pack "\""- , Text.replace (Text.pack "\"") (Text.pack "\"\"") ident- , Text.pack "\""- ]+-- | Register an aggregate function. The function folds the rows of each group+-- into a state.+--+-- 'aggregateStep' adds one row to a state. The step does not run for a row that+-- has NULL in an argument. A group with no other rows gets the result of+-- 'aggregateInitial'. DuckDB can split a group into partial states, for example+-- over threads or for a window frame. 'aggregateCombine' merges two partial+-- states. DuckDB also merges a partial state into a new state, so+-- 'aggregateCombine' must give the other state when one state is+-- 'aggregateInitial'. Each new state is evaluated to weak head normal form.+--+-- The arguments decode with t'Database.DuckDB.Simple.FromRow.FromRow'. The SQL+-- type of each argument is the+-- 'Database.DuckDB.Simple.ToField.duckdbTypeDeclaration' of its column in+-- 'TypedRow'. A type with no declaration is an error at registration, as for+-- 'createTableFunction'. Arguments use the same conversion as query results,+-- also when DuckDB runs the update callback on several threads. The SQL return+-- type is the+-- 'Database.DuckDB.Simple.ToField.duckdbTypeDeclaration' of the result type,+-- for example SMALLINT for 'Data.Int.Int16' and FLOAT for 'Float'. A 'Maybe'+-- result has the SQL type of its value, and 'Nothing' gives NULL.+--+-- The result needs a static SQL type declaration. Lists, maps, and records+-- through 'Database.DuckDB.Simple.Generic.ViaDuckDB' can supply one.+-- DuckDB resolves the declaration on the database that registers the function.+--+-- 'Bool', 8- to 64-bit integers, 'Float', 'Double', 'Text', 'String', and+-- 'Maybe' of these types write directly into the result vector. A newtype+-- can inherit that writer with @deriving newtype (FunctionResult)@.+-- For other results, 'Database.DuckDB.Simple.ToField.toDuckValue' receives+-- an instance factory that can construct built-in types. For a catalog type,+-- the encoder can return its underlying value. DuckDB casts that value to+-- the declared result type.+--+-- An exception in a function fails the query. DuckDB can run the functions on+-- more than one thread at one time, but not on the same state. The aggregate+-- function works with GROUP BY, with DISTINCT, FILTER, and ORDER BY, and as a+-- window function.+--+-- > createAggregate conn "haskell_sum" Aggregate+-- > { aggregateInitial = 0 :: Int64+-- > , aggregateStep = \total (Only value) -> pure (total + value)+-- > , aggregateCombine = \a b -> pure (a + b)+-- > , aggregateResult = pure+-- > }+createAggregate :: (FromRow args, TypedRow args, FunctionResult result) => Connection -> Text -> Aggregate state args result -> IO ()+createAggregate conn = createAggregateOn (ConnectionTarget conn) -foreign import ccall "wrapper"- mkScalarFun :: (DuckDBFunctionInfo -> DuckDBDataChunk -> DuckDBVector -> IO ()) -> IO DuckDBScalarFunctionFun+-- | Run @DROP FUNCTION IF EXISTS@ and report native errors.+-- DuckDB does not permit deletion of functions registered through the C API.+-- For a function from 'createFunction', this call raises t'SQLError' and the+-- function stays registered. Register the name again to replace the function.+-- GHC reports a warning at each use of this function. A build with @-Werror@+-- must allow that warning, for example with @-Wno-error=deprecations@.+--+-- The name follows the rules of DuckDB for a qualified name, as in+-- 'Database.DuckDB.Simple.Catalog.describeTable'. A dot separates two parts,+-- and a part in double quotes can contain a dot. Thus @\"a.b\"@ names the+-- function @a.b@, and @s.f@ names the function @f@ in the schema @s@. A name+-- with a quote inside a part that has no quotes, with an unterminated quote, or+-- with more than three parts raises the t'SQLError' of the parser.+deleteFunction :: Connection -> Text -> IO ()+deleteFunction conn name = do+ qualified <- qualifyIdentifier name+ -- See Note [SQL text of the library] in "Database.DuckDB.Simple.Native".+ let dropQuery = Query ("DROP FUNCTION IF EXISTS " <> qualified)+ withParsedStatements conn dropQuery (mapM_ (executeParsedStatement conn dropQuery))+{-# WARNING+ deleteFunction+ "DuckDB cannot drop a function that its C API registered. deleteFunction raises SQLError for a function from createFunction. Register the name again to replace the function."+ #-} -foreign import ccall "wrapper"- mkScalarInitFun :: (DuckDBInitInfo -> IO ()) -> IO DuckDBScalarFunctionInitFun+-- | Parse a qualified function name with @duckdb_v2_qname_parse@, and quote+-- each part with 'quoteIdentifier'.+qualifyIdentifier :: Text -> IO Text+qualifyIdentifier name = withText name \text ->+ bracket (output (c_duckdb_v2_qname_parse text)) (destroy c_duckdb_v2_qname_destroy) \qname -> do+ count <- output (c_duckdb_v2_qname_get_part_count qname)+ parts <- forM (indices count) \index -> output (c_duckdb_v2_qname_get_part qname index) >>= readIdentifier+ pure (Text.intercalate "." (map quoteIdentifier parts))
+ src/Database/DuckDB/Simple/Function/Internal.hs view
@@ -0,0 +1,523 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE FlexibleInstances #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE RankNTypes #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE StrictData #-}+{-# LANGUAGE TypeApplications #-}+{-# LANGUAGE UndecidableInstances #-}++-- |+-- Module : Database.DuckDB.Simple.Function.Internal+-- Description : Share typed scalar registration for connections and extensions.+--+-- The cabal file lists this hidden module in @other-modules@.+-- "Database.DuckDB.Simple.Function" exports the forms for a connection.+-- The forms that take a registration target,+-- 'createFunctionOn' and 'createFunctionWithStateOn', are for the module of+-- the loadable extensions. Both forms use the same argument conversion, result+-- conversion, and callback resources.+module Database.DuckDB.Simple.Function.Internal (+ ScalarValue,+ FunctionResult (..),+ Function (..),+ createFunctionOn,+ createFunctionWithStateOn,+ prepareAggregateResultWriter,+) where++import Control.Exception (+ SomeException,+ bracket,+ throwIO,+ )+import Control.Monad (forM, forM_, when)+import Data.Bits (clearBit, setBit)+import Data.Int (Int16, Int32, Int64, Int8)+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import qualified Data.Text.Foreign as TextForeign+import Data.Void (Void)+import Data.Word (Word16, Word32, Word64, Word8)+import Database.DuckDB.Simple.Callback (+ Allocator,+ RegistrationTarget,+ attachOpaque,+ checkTargetThreads,+ install,+ installAll,+ runCallback,+ transferCallbackState,+ withCallbackResourcesFinally,+ withRegistrationFactory,+ withTarget,+ )+import Database.DuckDB.Simple.Conversion (runConversion)+import Database.DuckDB.Simple.Element (vectorElementType)+import Database.DuckDB.Simple.FromField (+ Field,+ FromField (..),+ )+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), TypeDeclaration, logicalTypeSQL)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.Ok (Ok (..))+import Database.DuckDB.Simple.ToField (DuckDBColumnType (..), ToDuckValue (..))+import Database.DuckDB.Simple.TypeScope (withTypeScope)+import Database.DuckDB.Simple.Typed (+ ArgumentReader,+ argumentName,+ conversionMessage,+ newArgumentReader,+ positionalSignature,+ rowDeclarations,+ )+import Database.DuckDB.Simple.V2.Registration (configureFunctionSignature)+import Foreign.Marshal.Array (advancePtr, allocaArray, peekArray)+import Foreign.Marshal.Utils (copyBytes, fillBytes)+import Foreign.Ptr (Ptr, castPtr, nullPtr)+import Foreign.StablePtr (StablePtr, castPtrToStablePtr, deRefStablePtr)+import Foreign.Storable (Storable, peekElemOff, poke, pokeElemOff)+import GHC.Records (getField)+import HsBindgen.Runtime.Marshal (staticSizeOf)++{- Note [Result writers]+~~~~~~~~~~~~~~~~~~~~~~~~+The FunctionResult instance of a scalar or aggregate result type selects one+of two paths.++- A raw result (rawResult gives Just): Bool, Int, Int8 to Int64, Word, Word8 to+ Word64, Float, Double, Text, String, and Maybe of one of them. The writer+ writes the value to the memory of the vector, with no native call. A VARCHAR+ value of more than 12 bytes needs 2 native calls for the arena of the vector+ (writeArenaText).+- For other results, toDuckValue constructs an owned value and vector_set_value+ copies it into the output vector. DuckDB casts it to the declared result+ type. A scalar callback uses its context factory. An aggregate finalize+ callback has no context and borrows the registration's instance factory.+ This factory resolves built-in types. The registration declares the SQL+ result type on the host, where catalog types are visible.++A raw writer checks the logical type before writing a value so the width of+the write matches the vector. It writes at the offset supplied by DuckDB.+An ordered window can supply a result vector with logical size zero, even+when there are rows to finalize. vector_set_value checks this size. Before+writing native values, grow the vector to cover offset + count. Keep the+existing size when it is larger, since other groups can occupy later rows.+The native writer borrows no data or validity pointer across a resize.+-}++-- | The value of one result row. See Note [Result writers].+data ScalarValue+ = ScalarNull+ | ScalarBoolean Bool+ | ScalarInt8 Int8+ | ScalarInt16 Int16+ | ScalarInt32 Int32+ | ScalarInt64 Int64+ | ScalarWord8 Word8+ | ScalarWord16 Word16+ | ScalarWord32 Word32+ | ScalarWord64 Word64+ | ScalarFloat Float+ | ScalarDouble Double+ | ScalarText Text+ | ScalarNative (DuckDBV2FactoryHandle -> IO DuckDBV2ValueHandle)++-- | A result type of a scalar function or of a typed aggregate function. Each+-- type with a 'ToDuckValue' instance is a result type. Its SQL type is its+-- 'duckdbTypeDeclaration'. A newtype over a type with a raw writer gets that+-- writer with @deriving newtype (FunctionResult)@. See Note [Result writers].+class (ToDuckValue a) => FunctionResult a where+ -- | The raw value of a result, or 'Nothing' when 'toDuckValue' writes the+ -- result.+ rawResult :: Maybe (a -> ScalarValue)+ rawResult = Nothing++-- | A result that 'toDuckValue' writes.+instance {-# OVERLAPPABLE #-} (ToDuckValue a) => FunctionResult a++instance FunctionResult Bool where+ rawResult = Just ScalarBoolean++instance FunctionResult Int where+ rawResult = Just (ScalarInt64 . fromIntegral)++instance FunctionResult Int8 where+ rawResult = Just ScalarInt8++instance FunctionResult Int16 where+ rawResult = Just ScalarInt16++instance FunctionResult Int32 where+ rawResult = Just ScalarInt32++instance FunctionResult Int64 where+ rawResult = Just ScalarInt64++instance FunctionResult Word where+ rawResult = Just (ScalarWord64 . fromIntegral)++instance FunctionResult Word8 where+ rawResult = Just ScalarWord8++instance FunctionResult Word16 where+ rawResult = Just ScalarWord16++instance FunctionResult Word32 where+ rawResult = Just ScalarWord32++instance FunctionResult Word64 where+ rawResult = Just ScalarWord64++instance FunctionResult Float where+ rawResult = Just ScalarFloat++instance FunctionResult Double where+ rawResult = Just ScalarDouble++instance FunctionResult Text where+ rawResult = Just ScalarText++instance FunctionResult String where+ rawResult = Just (ScalarText . Text.pack)++instance (FunctionResult a) => FunctionResult (Maybe a) where+ rawResult = maybe ScalarNull <$> (rawResult :: Maybe (a -> ScalarValue))++-- | Give the value of one result row. See Note [Result writers].+scalarValue :: (FunctionResult a) => a -> ScalarValue+scalarValue value = case rawResult of+ Just raw -> raw value+ Nothing -> ScalarNative (`toDuckValue` value)++-- | A Haskell function that DuckDB can call as a scalar function: a result+-- value, an 'IO' action that gives a result, or a function of one more+-- argument. 'ScalarValue' is abstract, so an instance outside the library can+-- only use 'applyFunction' of another instance.+class Function a where+ -- | The SQL type of each argument, in order, or the reason why an argument+ -- type has no SQL type. See Note [Type declarations] in+ -- "Database.DuckDB.Simple.ToField".+ argumentTypes :: Proxy a -> [Either Text TypeDeclaration]++ -- | The SQL type of the result, or the reason why the result type has no SQL+ -- type.+ returnType :: Proxy a -> Either Text TypeDeclaration++ -- | 'True' for an 'IO' result. The registration then sets the stability of+ -- the function to VOLATILE: the result can be different for each row, and+ -- DuckDB calls the function one time for each row. 'False' keeps the+ -- default stability CONSISTENT: the function always gives the same result+ -- for the same arguments. DuckDB then calls a function with constant+ -- arguments one time for each call site, when it prepares the statement. It+ -- calls the function one time for each row in each other case, also for an+ -- argument with few different values.+ isVolatile :: Proxy a -> Bool++ -- | Convert the fields of one row to the arguments, and give the result of+ -- the function. A field that does not convert raises an error that names+ -- the argument.+ applyFunction :: [Field] -> a -> IO ScalarValue++-- | A result value: a function with no argument. It is not volatile.+instance {-# OVERLAPPABLE #-} (FunctionResult a) => Function a where+ argumentTypes _ = []+ returnType _ = duckdbTypeDeclaration (Proxy @a)+ isVolatile _ = False+ applyFunction [] value = pure (scalarValue value)+ applyFunction _ _ = throwInternalError (Text.pack "unexpected arguments supplied")++-- | An 'IO' action that gives a result: a function with no argument. It is+-- volatile, so DuckDB runs the action one time for each row.+instance {-# OVERLAPPING #-} (FunctionResult a) => Function (IO a) where+ argumentTypes _ = []+ returnType _ = duckdbTypeDeclaration (Proxy @a)+ isVolatile _ = True+ applyFunction [] action = scalarValue <$> action+ applyFunction _ _ = throwInternalError (Text.pack "unexpected arguments supplied")++-- | A function of one more argument. The argument has the SQL type of+-- 'duckdbTypeDeclaration'. The function is volatile when its result is+-- volatile. See Note [Type declarations] in "Database.DuckDB.Simple.ToField".+instance {-# OVERLAPPABLE #-} (FromField a, DuckDBColumnType a, Function r) => Function (a -> r) where+ argumentTypes _ = duckdbTypeDeclaration (Proxy @a) : argumentTypes (Proxy @r)+ returnType _ = returnType (Proxy @r)+ isVolatile _ = isVolatile (Proxy @r)+ applyFunction [] _ =+ throwInternalError (Text.pack "insufficient arguments supplied")+ applyFunction (field : rest) fn =+ runConversion (fromField field) >>= \case+ Errors err -> throwIO (argumentConversionError err)+ Ok value -> applyFunction rest (fn value)++-- | Register a typed function on a connection or a loading extension.+createFunctionOn :: forall f. (Function f) => RegistrationTarget -> Text -> f -> IO ()+createFunctionOn target name fn =+ registerScalarFunction target name (Proxy @f) \readArgument allocate function ->+ [ install+ allocate+ DuckDBV2ScalarFunctionExecCallbackFn+ (DuckDBV2ScalarFunctionExecCallbackFn_Aux (scalarFunctionHandler readArgument fn))+ (c_duckdb_v2_scalar_function_set_exec_callback function)+ ]++-- | Register a typed function with per-worker state on a connection or extension.+createFunctionWithStateOn :: forall s f. (Function f) => RegistrationTarget -> Text -> IO s -> (s -> f) -> IO ()+createFunctionWithStateOn target name initState mkFn =+ registerScalarFunction target name (Proxy @f) \readArgument allocate function ->+ [ install+ allocate+ DuckDBV2ScalarFunctionExecCallbackFn+ (DuckDBV2ScalarFunctionExecCallbackFn_Aux (scalarFunctionHandlerWithState readArgument mkFn))+ (c_duckdb_v2_scalar_function_set_exec_callback function)+ , install+ allocate+ DuckDBV2ScalarFunctionInitCallbackFn+ (DuckDBV2ScalarFunctionInitCallbackFn_Aux (scalarFunctionInitHandler initState))+ (c_duckdb_v2_scalar_function_set_init_callback function)+ ]++-- | Configure a scalar function and transfer its callbacks to DuckDB. An+-- argument type or a result type with no declaration raises an error at registration. The check of the+-- name and the declarations of the arguments and the result come before the+-- read of the state of the target, so their input errors win over a closed+-- connection. The last argument gives the install actions of the+-- callbacks for the builder, with the reader for this registration. Their+-- setters run after the properties. See 'installAll'.+registerScalarFunction+ :: (Function f)+ => RegistrationTarget+ -> Text+ -> Proxy f+ -> (ArgumentReader -> Allocator -> DuckDBV2ScalarFunctionHandle -> [IO (IO ())])+ -> IO ()+registerScalarFunction target name proxy installers = do+ when (Text.null name || Text.any (== '\0') name) $+ throwInvalidArgument "invalid scalar function name"+ -- See Note [Type declarations] in "Database.DuckDB.Simple.ToField".+ parameters <- rowDeclarations name ["argument " <> argumentName index | index <- [0 ..]] (argumentTypes proxy)+ [result] <- rowDeclarations name ["result"] [returnType proxy]+ (readArgument, _, closeReader) <- newArgumentReader target+ -- withTarget keeps its factory private. Borrow another one for the signature types.+ withRegistrationFactory target \factory ->+ withTarget+ target+ c_duckdb_v2_scalar_function_create+ c_duckdb_v2_scalar_function_destroy+ c_duckdb_v2_connection_register_scalar_function+ c_duckdb_v2_extension_register_scalar_function+ \scalarFun register ->+ withCallbackResourcesFinally+ (checkTargetThreads target)+ closeReader+ (\allocate -> installAll (installers readArgument allocate scalarFun))+ (attachOpaque (c_duckdb_v2_scalar_function_set_user_data scalarFun))+ \installCallbacks -> do+ withIdentifier name $ checked . c_duckdb_v2_scalar_function_set_name scalarFun+ signature <- output (c_duckdb_v2_scalar_function_get_signature scalarFun)+ configureFunctionSignature factory (positionalSignature parameters (Just result)) signature+ when (isVolatile proxy) $+ checked (c_duckdb_v2_scalar_function_set_property scalarFun DuckDBV2FunctionPropertyStability DuckDBV2FunctionPropertyStabilityVolatile)+ checked+ (c_duckdb_v2_scalar_function_set_property scalarFun DuckDBV2FunctionPropertyNullHandling DuckDBV2FunctionPropertyNullHandlingSpecial)+ checked (c_duckdb_v2_scalar_function_set_property scalarFun DuckDBV2FunctionPropertyFallibility DuckDBV2FunctionPropertyFallibilityFallible)+ installCallbacks+ register++scalarFunctionHandler+ :: forall f+ . (Function f)+ => ArgumentReader -> f -> DuckDBV2ScalarFunctionExecInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()+scalarFunctionHandler readArgument fn info context err = runCallback err (executeScalarFunction readArgument fn info context)++-- | Run one call of a scalar function.+executeScalarFunction+ :: forall f. (Function f) => ArgumentReader -> f -> DuckDBV2ScalarFunctionExecInfoHandle -> DuckDBV2ContextHandle -> IO ()+executeScalarFunction readArgument fn info context = withTypeScope \scope -> do+ let expected = length (argumentTypes (Proxy @f))+ columnCount <- allocaArray 4 $ \counts -> do+ checked (c_duckdb_v2_scalar_function_exec_get_arg_count info counts (advancePtr counts 1) (advancePtr counts 2) (advancePtr counts 3))+ fromIntegral . sum <$> peekArray 4 counts+ when (columnCount /= expected) $+ throwInternalError (Text.pack ("function expected " <> show expected <> " arguments but received " <> show columnCount))+ rawRowCount <- output (c_duckdb_v2_scalar_function_exec_get_row_count info)+ let rowCount = fromIntegral rawRowCount :: Int+ readers <- forM [0 .. expected - 1] \index ->+ output (c_duckdb_v2_scalar_function_exec_get_arg info (fromIntegral index)) >>= readArgument scope index+ rows <- forM [0 .. rowCount - 1] $ \row -> forM readers ($ row)+ results <- mapM (`applyFunction` fn) rows+ outVec <- output (c_duckdb_v2_scalar_function_exec_get_result info)+ writeResults context results outVec++scalarFunctionHandlerWithState+ :: forall s f+ . (Function f)+ => ArgumentReader+ -> (s -> f)+ -> DuckDBV2ScalarFunctionExecInfoHandle+ -> DuckDBV2ContextHandle+ -> Ptr DuckDBV2ErrorInfoHandle+ -> IO ()+scalarFunctionHandlerWithState readArgument mkFn info context err = runCallback err do+ statePtr <- output (c_duckdb_v2_scalar_function_exec_get_init_data info)+ when (statePtr == nullPtr) $+ throwInternalError "scalar function state was not initialised"+ -- The cast has no type check, and it is safe. createFunctionWithStateOn has+ -- one signature for initState :: IO s and mkFn :: s -> f. Thus the init+ -- callback stores a StablePtr of the same type s that this cast gives.+ state <- deRefStablePtr (castPtrToStablePtr (castPtr statePtr) :: StablePtr s)+ executeScalarFunction readArgument (mkFn state) info context++scalarFunctionInitHandler :: IO s -> DuckDBV2ScalarFunctionInitInfoHandle -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()+scalarFunctionInitHandler initState info _context err = runCallback err do+ state <- initState+ transferCallbackState (attachOpaque (c_duckdb_v2_scalar_function_init_set_init_data info)) state++-- | Write the results of one call to the result vector. Set the size, then+-- flatten, then read the pointers. A row of a result that is not raw borrows+-- the factory of the context. See Note [Vector flatten] in+-- "Database.DuckDB.Simple.Materialize" and Note [Result writers].+writeResults :: DuckDBV2ContextHandle -> [ScalarValue] -> DuckDBV2VectorHandle -> IO ()+writeResults context values outVec = do+ checked (c_duckdb_v2_vector_set_size outVec (fromIntegral (length values)))+ checked (c_duckdb_v2_vector_flatten outVec)+ let native = \case+ ScalarNative _ -> True+ _ -> False+ withFactory action+ | any native values = withContextFactory context (action . Just)+ | otherwise = action Nothing+ withFactory \factory -> do+ write <- prepareResultWriter factory outVec+ forM_ (zip [0 ..] values) (uncurry write)++-- | Write aggregate results into a borrowed vector. The end is the exclusive+-- upper bound of this finalize call. See Note [Result writers].+prepareAggregateResultWriter+ :: (FunctionResult a) => IO DuckDBV2FactoryHandle -> DuckDBV2VectorHandle -> DuckDBIdx -> IO (Int -> a -> IO ())+prepareAggregateResultWriter acquireFactory vector end = case rawResult of+ Just raw -> do+ write <- prepareResultWriter Nothing vector+ pure (\row -> write row . raw)+ Nothing -> do+ size <- output (c_duckdb_v2_vector_get_size vector)+ when (size < end) (checked (c_duckdb_v2_vector_set_size vector end))+ factory <- acquireFactory+ pure (\row -> writeNativeResult vector row . toDuckValue factory)++-- | Copy an owned value into a result row, then release it even if the cast fails.+writeNativeResult :: DuckDBV2VectorHandle -> Int -> IO DuckDBV2ValueHandle -> IO ()+writeNativeResult vector row create = bracket create destroyValue (checked . c_duckdb_v2_vector_set_value vector (fromIntegral row))++-- | Prepare a writer for the rows of a flat result vector. The writer does not+-- change the size of the vector. It reads the logical type of the vector one+-- time, and it raises an error for a raw value of another type. It writes a+-- value of 'ScalarNative' with the factory, and raises an error without a+-- factory. A vector that is not flat raises an error. See+-- Note [Result writers] and Note [Vector flatten] in+-- "Database.DuckDB.Simple.Materialize".+prepareResultWriter :: Maybe DuckDBV2FactoryHandle -> DuckDBV2VectorHandle -> IO (Int -> ScalarValue -> IO ())+prepareResultWriter factory vector = do+ kind <- output (c_duckdb_v2_vector_get_vector_type vector)+ when (kind /= DuckDBV2VectorTypeFlat) $+ throwInternalError (Text.pack ("the result vector is not flat: " <> show kind))+ vectorType <- vectorElementType vector+ dataPtr <- output (c_duckdb_v2_vector_get_data_mutable vector)+ validityPtr <- output (c_duckdb_v2_vector_flat_get_validity_mutable vector)+ pure (writeScalarValue factory vector vectorType dataPtr validityPtr)++-- | Write one value at a row of a flat vector. The data and validity pointers+-- belong to the vector. A raw value must have the logical type of the vector.+-- See Note [Result writers].+writeScalarValue+ :: Maybe DuckDBV2FactoryHandle -> DuckDBV2VectorHandle -> DuckDBV2LogicalTypeId -> Ptr Void -> Ptr Word64 -> Int -> ScalarValue -> IO ()+writeScalarValue factory vector vectorType dataPtr validityPtr idx = \case+ ScalarNull -> markInvalid validityPtr idx+ -- A BOOLEAN value has one byte.+ ScalarBoolean flag -> raw DuckDBV2LogicalTypeIdBoolean (if flag then 1 else 0 :: Word8)+ ScalarInt8 value -> raw DuckDBV2LogicalTypeIdTinyInt value+ ScalarInt16 value -> raw DuckDBV2LogicalTypeIdSmallInt value+ ScalarInt32 value -> raw DuckDBV2LogicalTypeIdInteger value+ ScalarInt64 value -> raw DuckDBV2LogicalTypeIdBigInt value+ ScalarWord8 value -> raw DuckDBV2LogicalTypeIdUTinyInt value+ ScalarWord16 value -> raw DuckDBV2LogicalTypeIdUSmallInt value+ ScalarWord32 value -> raw DuckDBV2LogicalTypeIdUInteger value+ ScalarWord64 value -> raw DuckDBV2LogicalTypeIdUBigInt value+ ScalarFloat value -> raw DuckDBV2LogicalTypeIdFloat value+ ScalarDouble value -> raw DuckDBV2LogicalTypeIdDouble value+ ScalarText text -> do+ expect DuckDBV2LogicalTypeIdVarchar+ markValid validityPtr idx+ writeArenaText vector dataPtr idx text+ ScalarNative make -> case factory of+ Just valueFactory -> writeNativeResult vector idx (make valueFactory)+ Nothing -> throwInternalError "the result writer has no factory for a value of this result type"+ where+ raw :: (Storable s) => DuckDBV2LogicalTypeId -> s -> IO ()+ raw expected value = do+ expect expected+ markValid validityPtr idx+ pokeElemOff (castPtr dataPtr) idx value+ expect expected = when (vectorType /= expected) do+ actualName <- logicalTypeSQL . LogicalTypeScalar <$> logicalTypeIdFromNative vectorType+ expectedName <- logicalTypeSQL . LogicalTypeScalar <$> logicalTypeIdFromNative expected+ throwInternalError ("the result vector has the type " <> actualName <> ", but the result value has the type " <> expectedName)++-- | Write a VARCHAR value at a row of a flat vector without a factory. A value+-- of more than @DUCKDB_V2_BYTES_INLINE_LENGTH@ (12) bytes goes to the arena of+-- the vector. See Note [Vector memory layouts] in "Database.DuckDB.Simple.Element".+writeArenaText :: DuckDBV2VectorHandle -> Ptr Void -> Int -> Text -> IO ()+writeArenaText vector dataPtr idx txt = do+ -- Write each field through its generated field pointer, as chunkDecodeBlob+ -- in "Database.DuckDB.Simple.Element" reads it.+ let len = TextForeign.lengthWord8 txt+ slot = advancePtr @DuckDBV2Bytes (castPtr dataPtr) idx+ inlined = getField @"inlined" (getField @"value" slot)+ pointer = getField @"pointer" (getField @"value" slot)+ when (len > fromIntegral (maxBound :: Word32)) $+ throwIO (functionInvocationError "duckdb-simple: a VARCHAR result is longer than 4 GiB")+ -- DuckDB compares the unused inline bytes. They must be zero.+ fillBytes slot 0 (staticSizeOf (Proxy @DuckDBV2Bytes))+ if len <= fromIntegral dUCKDB_V2_BYTES_INLINE_LENGTH+ then do+ poke (getField @"length" inlined) (fromIntegral len)+ TextForeign.unsafeCopyToPtr txt (castPtr (getField @"inlined" inlined))+ else do+ arena <- output (c_duckdb_v2_vector_get_arena vector)+ bytes <- output (c_duckdb_v2_arena_allocate arena (fromIntegral len))+ TextForeign.unsafeCopyToPtr txt bytes+ poke (getField @"length" pointer) (fromIntegral len)+ copyBytes (castPtr (getField @"prefix" pointer)) bytes 4+ poke (getField @"ptr" pointer) (castPtr bytes)++-- | Clear the validity bit of a row. See Note [Vector memory layouts] in+-- "Database.DuckDB.Simple.Element".+markInvalid :: Ptr Word64 -> Int -> IO ()+markInvalid validity idx = do+ let (wordIndex, bitIndex) = idx `divMod` 64+ word <- peekElemOff validity wordIndex+ pokeElemOff validity wordIndex (clearBit word bitIndex)++-- | Set the validity bit of a row. See Note [Vector memory layouts] in+-- "Database.DuckDB.Simple.Element".+markValid :: Ptr Word64 -> Int -> IO ()+markValid validity idx = do+ let (wordIndex, bitIndex) = idx `divMod` 64+ word <- peekElemOff validity wordIndex+ pokeElemOff validity wordIndex (setBit word bitIndex)++argumentConversionError :: [SomeException] -> SQLError+argumentConversionError err = functionInvocationError (Text.pack ("duckdb-simple: " <> conversionMessage err))++functionInvocationError :: Text -> SQLError+functionInvocationError message =+ SQLError+ { sqlErrorMessage = message+ , sqlErrorType = Nothing+ , sqlErrorQuery = Nothing+ , sqlErrorBody = Nothing+ }
src/Database/DuckDB/Simple/Generic.hs view
@@ -8,736 +8,877 @@ {-# LANGUAGE MultiParamTypeClasses #-} {-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE ScopedTypeVariables #-}-{-# LANGUAGE TypeFamilies #-}-{-# LANGUAGE TypeOperators #-}-{-# LANGUAGE UndecidableInstances #-}--{- |-Module : Database.DuckDB.Simple.Generic-Description : Generic helpers for encoding Haskell ADTs as DuckDB structs/unions.--This module provides the glue needed to reuse the existing @ToField@/@FromField@-machinery with algebraic data types via GHC generics. The supported mapping is-currently intentionally conservative:--* /Product types/ (records or tuples) whose fields already satisfy @DuckValue@- are encoded as STRUCT values. Record fields retain their selector name;- positional products fall back to @field1@, @field2@, …-* /Sum types/ (:+:) become UNION values. Each constructor becomes a union- member; payloads are encoded as structs (or @NULL@ for nullary constructors).--Recursive types are supported as long as every payload is itself encodable-through @DuckValue@. Note that sum types must have constructor fields that are-structural products (i.e. we do not yet expose mixed union/record nesting for-non-record constructors).--Typical usage looks like:--> data User = User { userId :: Int64, userName :: Text }-> deriving stock (Generic)->-> instance DuckValue User->-> toField (genericToFieldValue user) -- Struct {"userId" := ..., ...}--For sum types:--> data Shape-> = Circle Double-> | Rectangle Double Double-> | Origin-> deriving stock (Generic)->-> instance DuckValue Shape--Constructors are turned into a union with members @Circle{radius}@,-@Rectangle{width,height}@, and @Origin@ (null payload).--You can also lean on @DerivingVia@ using the exported @ViaDuckDB@ newtype:--> data User = User { userId :: Int64, userName :: Text }-> deriving stock (Generic)-> deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB User)--The derived instances automatically encode/decode via STRUCT/UNION representations.--=== Extending this module--The rest of this file is organised so that each building block is reusable:--* @DuckValue@ covers leaf-level conversions between Haskell values, DuckDB- @FieldValue@s, and logical type metadata.-* @GToField@ and friends walk the generic representation to assemble structs or- unions and carry around the logical type information we later need when- binding parameters.-* @ViaDuckDB@ wires everything together for deriving via.--When adding new features, mimic the structure used here (and document new-classes the way the existing ones are documented) so other backends can take-inspiration from this implementation.--}-module Database.DuckDB.Simple.Generic (- -- * Field-level primitives- DuckValue (..),- GToField (),- GFromField (),-- -- * Generic encoding/decoding for ADTs- genericToFieldValue,- genericFromFieldValue,- genericLogicalType,- genericToStructValue,- genericToUnionValue,-- -- * DerivingVia helper- ViaDuckDB (..),-) where--import Control.Exception (displayException)-import Control.Monad (unless)-import Data.Array (Array, elems, listArray)-import qualified Data.ByteString as BS-import Data.Int (Int16, Int32, Int64, Int8)-import qualified Data.Map.Strict as Map-import Data.Proxy (Proxy (..))-import Data.Text (Text)-import qualified Data.Text as Text-import Data.Time (UTCTime)-import Data.Time.Calendar (Day)-import Data.Time.LocalTime (LocalTime, TimeOfDay)-import Data.Typeable (Typeable)-import qualified Data.UUID as UUID-import Data.Word (Word16, Word32, Word64, Word8)-import GHC.Generics-import Numeric.Natural (Natural)--import Database.DuckDB.FFI-import Database.DuckDB.Simple.FromField (- Field (..),- FieldValue (..),- FromField (..),- IntervalValue (..),- ResultError (..),- TimeWithZone (..),- returnError,- )-import Database.DuckDB.Simple.Internal (duckDBTypeToName)-import Database.DuckDB.Simple.LogicalRep (- LogicalTypeRep (..),- StructField (..),- StructValue (..),- UnionMemberType (..),- UnionValue (..),- )-import Database.DuckDB.Simple.Ok (Ok (..))-import Database.DuckDB.Simple.Time (Unbounded (..))-import Database.DuckDB.Simple.ToField (DuckDBColumnType (..), ToField (..))-import Database.DuckDB.Simple.Variant (Variant (..))------------------------------------------------------------------------------------- DuckValue: bridge between Haskell scalars and FieldValue/LogicalTypeRep--{- | Types that can appear inside generated structs/unions.--A @DuckValue@ instance must provide:--* encoding to @FieldValue@-* logical type metadata (@duckLogicalType@)-* decoding from @FieldValue@--The primitive instances below are the canonical source for how scalar types-should be represented; both the generic implementation and the manual-`ToField`/`FromField` instances rely on them.--}-class DuckValue a where- duckToField :: a -> FieldValue- duckFromField :: FieldValue -> Either String a- duckLogicalType :: Proxy a -> LogicalTypeRep-- default duckFromField :: (FromField a, Show a) => FieldValue -> Either String a- duckFromField fv =- case fromField Field{fieldName = Text.empty, fieldIndex = 0, fieldValue = fv} of- Ok x -> Right x- Errors errs -> Left (unlines (map displayException errs))--instance DuckValue Variant where- duckToField = variantPayload- duckLogicalType _ = LogicalTypeScalar DuckDBTypeVariant--instance DuckValue Bool where- duckToField = FieldBool- duckLogicalType _ = LogicalTypeScalar DuckDBTypeBoolean--instance DuckValue Int where- duckToField = FieldInt64 . fromIntegral- duckLogicalType _ = LogicalTypeScalar DuckDBTypeBigInt--instance DuckValue Int8 where- duckToField = FieldInt8- duckLogicalType _ = LogicalTypeScalar DuckDBTypeTinyInt--instance DuckValue Int16 where- duckToField = FieldInt16- duckLogicalType _ = LogicalTypeScalar DuckDBTypeSmallInt--instance DuckValue Int32 where- duckToField = FieldInt32- duckLogicalType _ = LogicalTypeScalar DuckDBTypeInteger--instance DuckValue Int64 where- duckToField = FieldInt64- duckLogicalType _ = LogicalTypeScalar DuckDBTypeBigInt--instance DuckValue Integer where- duckToField = FieldHugeInt- duckLogicalType _ = LogicalTypeScalar DuckDBTypeHugeInt--instance DuckValue Natural where- duckToField = FieldUHugeInt . toInteger- duckLogicalType _ = LogicalTypeScalar DuckDBTypeUHugeInt--instance DuckValue Word where- duckToField = FieldWord64 . fromIntegral- duckLogicalType _ = LogicalTypeScalar DuckDBTypeUBigInt--instance DuckValue Word8 where- duckToField = FieldWord8- duckLogicalType _ = LogicalTypeScalar DuckDBTypeUTinyInt--instance DuckValue Word16 where- duckToField = FieldWord16- duckLogicalType _ = LogicalTypeScalar DuckDBTypeUSmallInt--instance DuckValue Word32 where- duckToField = FieldWord32- duckLogicalType _ = LogicalTypeScalar DuckDBTypeUInteger--instance DuckValue Word64 where- duckToField = FieldWord64- duckLogicalType _ = LogicalTypeScalar DuckDBTypeUBigInt--instance DuckValue Float where- duckToField = FieldFloat- duckLogicalType _ = LogicalTypeScalar DuckDBTypeFloat--instance DuckValue Double where- duckToField = FieldDouble- duckLogicalType _ = LogicalTypeScalar DuckDBTypeDouble--instance DuckValue Text where- duckToField = FieldText- duckLogicalType _ = LogicalTypeScalar DuckDBTypeVarchar--instance DuckValue String where- duckToField = FieldText . Text.pack- duckLogicalType _ = LogicalTypeScalar DuckDBTypeVarchar- duckFromField fv = Text.unpack <$> duckFromField fv--instance DuckValue BS.ByteString where- duckToField = FieldBlob- duckLogicalType _ = LogicalTypeScalar DuckDBTypeBlob--instance DuckValue Day where- duckToField = FieldDate . Finite- duckLogicalType _ = LogicalTypeScalar DuckDBTypeDate--instance DuckValue TimeOfDay where- duckToField = FieldTime- duckLogicalType _ = LogicalTypeScalar DuckDBTypeTime--instance DuckValue LocalTime where- duckToField = FieldTimestamp . Finite- duckLogicalType _ = LogicalTypeScalar DuckDBTypeTimestamp--instance DuckValue UTCTime where- duckToField = FieldTimestampTZ . Finite- duckLogicalType _ = LogicalTypeScalar DuckDBTypeTimestampTz--instance DuckValue (Unbounded Day) where- duckToField = FieldDate- duckLogicalType _ = LogicalTypeScalar DuckDBTypeDate--instance DuckValue (Unbounded LocalTime) where- duckToField = FieldTimestamp- duckLogicalType _ = LogicalTypeScalar DuckDBTypeTimestamp--instance DuckValue (Unbounded UTCTime) where- duckToField = FieldTimestampTZ- duckLogicalType _ = LogicalTypeScalar DuckDBTypeTimestampTz--instance DuckValue UUID.UUID where- duckToField = FieldUUID- duckLogicalType _ = LogicalTypeScalar DuckDBTypeUUID--instance DuckValue IntervalValue where- duckToField = FieldInterval- duckLogicalType _ = LogicalTypeScalar DuckDBTypeInterval--instance DuckValue TimeWithZone where- duckToField = FieldTimeTZ- duckLogicalType _ = LogicalTypeScalar DuckDBTypeTimeTz--instance (DuckValue a) => DuckValue (Maybe a) where- duckToField (Just x) = duckToField x- duckToField Nothing = FieldNull- duckLogicalType _ = duckLogicalType (Proxy :: Proxy a)- duckFromField FieldNull = Right Nothing- duckFromField other = Just <$> duckFromField other---- | List values encode as DuckDB LIST (variable-length).-instance (DuckValue a) => DuckValue [a] where- duckToField xs = FieldList (map duckToField xs)- duckLogicalType _ = LogicalTypeList (duckLogicalType (Proxy :: Proxy a))- duckFromField (FieldList fvs) = traverse duckFromField fvs- duckFromField other = Left ("duckdb-simple: expected LIST, got " <> show other)--{- | Array values encode as DuckDB ARRAY (fixed-length).-Note: Arrays must have consistent bounds to work correctly with DuckDB.--}-instance (DuckValue a) => DuckValue (Array Int a) where- duckToField arr =- let values = elems arr- (low, high) = (0, length values - 1)- in FieldArray (listArray (low, high) (map duckToField values))- duckLogicalType _ =- -- We can't determine array size at the type level, so this is approximate.- -- The actual size will be determined at runtime from the array bounds.- LogicalTypeArray (duckLogicalType (Proxy :: Proxy a)) 0- duckFromField (FieldArray arr) = do- let values = elems arr- decoded <- traverse duckFromField values- let (low, high) = (0, length decoded - 1)- pure (listArray (low, high) decoded)- duckFromField other = Left ("duckdb-simple: expected ARRAY, got " <> show other)---- | Map values encode as DuckDB MAP.-instance (Ord k, DuckValue k, DuckValue v) => DuckValue (Map.Map k v) where- duckToField m =- let pairs = Map.toList m- in FieldMap [(duckToField k, duckToField v) | (k, v) <- pairs]- duckLogicalType _ =- LogicalTypeMap- (duckLogicalType (Proxy :: Proxy k))- (duckLogicalType (Proxy :: Proxy v))- duckFromField (FieldMap pairs) = do- decodedPairs <- traverse decodePair pairs- pure (Map.fromList decodedPairs)- where- decodePair (kfv, vfv) = do- k <- duckFromField kfv- v <- duckFromField vfv- pure (k, v)- duckFromField other = Left ("duckdb-simple: expected MAP, got " <> show other)------------------------------------------------------------------------------------- Generic machinery--{- | Internal representation used while traversing the generic structure. We-keep both the encoded value and its logical type so we can re-use the same-traversal when generating metadata (@genericLogicalType@) and when producing-concrete values (@genericToFieldValue@).--}-data Encoded- = EncodedStruct (StructValue FieldValue) (Array Int (StructField LogicalTypeRep))- | EncodedUnion (UnionValue FieldValue)- | EncodedNull--encodedValue :: Encoded -> FieldValue-encodedValue = \case- EncodedStruct sv _ -> FieldStruct sv- EncodedUnion uv -> FieldUnion uv- EncodedNull -> FieldNull---- | Convert a Haskell value (using its generic representation) into a DuckDB @FieldValue@.-genericToFieldValue :: forall a. (Generic a, GToField (Rep a)) => a -> FieldValue-genericToFieldValue = encodedValue . gToField . from---- | Extract the logical DuckDB type corresponding to a Haskell value.-genericLogicalType :: forall a. (Generic a, GToField (Rep a)) => Proxy a -> LogicalTypeRep-genericLogicalType _ = gLogicalType (Proxy :: Proxy (Rep a ()))---- | Decode a DuckDB @FieldValue@ back into a Haskell value using its generic representation.-genericFromFieldValue :: forall a. (Generic a, GFromField (Rep a)) => FieldValue -> Either String a-genericFromFieldValue fv = to <$> gFromField fv---- | Convenience helpers that project out structured values directly.-genericToStructValue :: forall a. (Generic a, GToField (Rep a)) => a -> Maybe (StructValue FieldValue)-genericToStructValue value =- case gToField (from value) of- EncodedStruct sv _ -> Just sv- EncodedNull -> Just emptyStruct- _ -> Nothing- where- emptyStruct =- StructValue- { structValueFields = listArray (0, -1) []- , structValueTypes = listArray (0, -1) []- , structValueIndex = Map.empty- }---- | Extract a UNION-shaped generic encoding directly when one is produced.-genericToUnionValue :: forall a. (Generic a, GToField (Rep a)) => a -> Maybe (UnionValue FieldValue)-genericToUnionValue value =- case gToField (from value) of- EncodedUnion uv -> Just uv- _ -> Nothing--{- | Wrapper for deriving-via so that @instance ToField (ViaDuckDB a)@ picks up-the generic encoding provided by this module.--}-newtype ViaDuckDB a = ViaDuckDB {getViaDuckDB :: a}---- Type family to decide whether a representation is a sum.--{- | Type family evaluating to @'True@ for sum-of-constructors generic-representations. We use this to select the appropriate encoding strategy.--}-type family IsSum f :: Bool where- IsSum (f :+: g) = 'True- IsSum (M1 D _ f) = IsSum f- IsSum (M1 C _ f) = IsSum f- IsSum _ = 'False--{- | Generic encoding to the intermediate @Encoded@ representation. Every-instance must also supply the corresponding logical type description.--}-class GToField f where- gToField :: f p -> Encoded- gLogicalType :: Proxy (f p) -> LogicalTypeRep--instance (GToField' (IsSum f) f) => GToField f where- gToField = gToField' (Proxy :: Proxy (IsSum f))- gLogicalType _ = gLogicalType' (Proxy :: Proxy (IsSum f)) (Proxy :: Proxy f)--{- | Helper class that splits the product and sum handling using the @IsSum@-type family. We specialise on products (@'False@) and sums (@'True@) to keep-the core logic small and easy to reason about.--}-class GToField' (isSum :: Bool) f where- gToField' :: Proxy isSum -> f p -> Encoded- gLogicalType' :: Proxy isSum -> Proxy f -> LogicalTypeRep---- Products (single constructor records)---- | Product encoding: single-constructor datatypes become STRUCT values.-instance (GStruct f) => GToField' 'False (M1 D meta (M1 C c f)) where- gToField' _ (M1 (M1 inner)) =- let comps = gStructValues inner- typeComps = gStructTypes (Proxy :: Proxy (f p))- in case comps of- [] -> EncodedNull- _ ->- let names = resolveNames (zip progIndices (map fcName comps))- valueArray = listArrayFrom names (map fcValue comps)- typeArray = listArrayFrom names (map fcValue typeComps)- indexMap = Map.fromList (zip names [0 ..])- in EncodedStruct- StructValue- { structValueFields = valueArray- , structValueTypes = typeArray- , structValueIndex = indexMap- }- typeArray- where- progIndices = [0 :: Int ..]- gLogicalType' _ _ =- let typeComps = gStructTypes (Proxy :: Proxy (f p))- names = resolveNames (zip [0 :: Int ..] (map fcName typeComps))- typeArray = listArrayFrom names (map fcValue typeComps)- in LogicalTypeStruct typeArray---- Sums (encode as union)---- | Sum encoding: multi-constructor datatypes become UNION values.-instance (GSum f) => GToField' 'True (M1 D meta f) where- gToField' _ (M1 value) =- let members = gSumMembers (Proxy :: Proxy (f p))- membersArray = listArray (0, length members - 1) members- (idx, payload) = gSumEncode value- label = unionMemberName (members !! idx)- in EncodedUnion- UnionValue- { unionValueIndex = fromIntegral idx- , unionValueLabel = label- , unionValuePayload = payload- , unionValueMembers = membersArray- }- gLogicalType' _ _ =- let members = gSumMembers (Proxy :: Proxy (f p))- membersArray = listArray (0, length members - 1) members- in LogicalTypeUnion membersArray------------------------------------------------------------------------------------- GStruct: products--data FieldComponent a = FieldComponent- { fcName :: Maybe Text- , fcValue :: a- }--resolveNames :: [(Int, Maybe Text)] -> [Text]--{- | Assign canonical names to struct fields. We preserve any selector names-provided by GHC.Generics and fall back to @fieldN@ for positional products.--}-resolveNames =- map pick- where- pick (_, Just n) = n- pick (idx, Nothing) = Text.pack ("field" <> show (idx + 1))--{- | Helper that builds an @Array@ of struct fields from parallel lists of names-and payloads.--}-listArrayFrom :: [Text] -> [b] -> Array Int (StructField b)-listArrayFrom names values =- listArray- (0, length values - 1)- (zipWith (\n v -> StructField{structFieldName = n, structFieldValue = v}) names values)--{- | Collect the components (values and types) of a product. Implementations-produce parallel lists so we can zip them during encoding and decoding.--}-class GStruct f where- gStructValues :: f p -> [FieldComponent FieldValue]- gStructTypes :: Proxy (f p) -> [FieldComponent LogicalTypeRep]--instance GStruct U1 where- gStructValues _ = []- gStructTypes _ = []--instance (GStruct a, GStruct b) => GStruct (a :*: b) where- gStructValues (a :*: b) = gStructValues a ++ gStructValues b- gStructTypes _ = gStructTypes (Proxy :: Proxy (a p)) ++ gStructTypes (Proxy :: Proxy (b p))--instance (Selector s, DuckValue a) => GStruct (M1 S s (K1 i a)) where- gStructValues m@(M1 (K1 x)) =- let name = toMaybe (selName m)- in [FieldComponent name (duckToField x)]- gStructTypes _ =- let raw = selName (undefined :: M1 S s (K1 i a) ())- name = toMaybe raw- in [FieldComponent name (duckLogicalType (Proxy :: Proxy a))]--instance (GStruct f) => GStruct (M1 C c f) where- gStructValues (M1 x) = gStructValues x- gStructTypes _ = gStructTypes (Proxy :: Proxy (f p))--toMaybe :: String -> Maybe Text-toMaybe name- | null name = Nothing- | otherwise = Just (Text.pack name)------------------------------------------------------------------------------------- Sums (unions)--{- | Sum type encoding. We gather the metadata (@gSumMembers@), convert a value-to its discriminant and payload (@gSumEncode@), and provide the inverse-(@gSumDecode@).--}-class GSum f where- gSumMembers :: Proxy (f p) -> [UnionMemberType]- gSumEncode :: f p -> (Int, FieldValue)- gSumDecode :: Int -> FieldValue -> Either String (f p)--instance (GSum a, GSum b) => GSum (a :+: b) where- gSumMembers _ = gSumMembers (Proxy :: Proxy (a p)) ++ gSumMembers (Proxy :: Proxy (b p))- gSumEncode (L1 x) = gSumEncode x- gSumEncode (R1 x) =- let leftCount = length (gSumMembers (Proxy :: Proxy (a p)))- (idx, payload) = gSumEncode x- in (idx + leftCount, payload)- gSumDecode idx payload =- let leftCount = length (gSumMembers (Proxy :: Proxy (a p)))- in if idx < leftCount- then L1 <$> gSumDecode idx payload- else R1 <$> gSumDecode (idx - leftCount) payload--instance (Constructor c, GStruct f, GStructDecode f) => GSum (M1 C c f) where- gSumMembers _ =- [ UnionMemberType- { unionMemberName = Text.pack (conName (undefined :: M1 C c f p))- , unionMemberType =- let typeComps = gStructTypes (Proxy :: Proxy (f p))- names = resolveNames (zip [0 :: Int ..] (map fcName typeComps))- in LogicalTypeStruct (listArrayFrom names (map fcValue typeComps))- }- ]- gSumEncode (M1 x) =- case gStructValues x of- [] -> (0, FieldNull)- comps ->- let typeComps = gStructTypes (Proxy :: Proxy (f p))- names = resolveNames (zip [0 :: Int ..] (map fcName comps))- valueArray = listArrayFrom names (map fcValue comps)- typeArray = listArrayFrom names (map fcValue typeComps)- indexMap = Map.fromList (zip names [0 ..])- in (0, FieldStruct StructValue{structValueFields = valueArray, structValueTypes = typeArray, structValueIndex = indexMap})- gSumDecode idx payload- | idx /= 0 = Left ("duckdb-simple: union tag mismatch (expected 0, got " <> show idx <> ")")- | otherwise =- case payload of- FieldNull -> M1 . fst <$> gStructDecodeList (Proxy :: Proxy (f p)) []- FieldStruct structVal -> do- ordered <- orderStructFields (gStructTypes (Proxy :: Proxy (f p))) structVal- M1 <$> gStructDecodeStruct (Proxy :: Proxy (f p)) ordered- other -> Left ("duckdb-simple: expected STRUCT payload for union member, got " <> show other)---- | Match decoded fields to their selector names before product decoding.-orderStructFields :: [FieldComponent LogicalTypeRep] -> StructValue FieldValue -> Either String (StructValue FieldValue)-orderStructFields components sv = do- let names = resolveNames (zip [0 ..] (map fcName components))- fields = elems (structValueFields sv)- byName = Map.fromList [(structFieldName field, field) | field <- fields]- unless (length fields <= length names) $- Left "duckdb-simple: extra fields when decoding struct"- unless (length fields == length names && Map.size byName == length names) $- Left "duckdb-simple: struct field names or count mismatch"- ordered <- traverse (\name -> maybe (Left ("duckdb-simple: missing struct field " <> Text.unpack name)) Right (Map.lookup name byName)) names- pure sv{structValueFields = listArray (0, length ordered - 1) ordered}------------------------------------------------------------------------------------- GStructDecode: inverse of GStruct for decoding---- | Inverse of @GStruct@: decode struct payloads back into a generic product.-class GStructDecode f where- gStructDecodeStruct :: Proxy (f p) -> StructValue FieldValue -> Either String (f p)-- {- | Consume a prefix of fields from left to right while decoding, returning- the reconstructed value and any remaining fields.- -}- gStructDecodeList :: Proxy (f p) -> [FieldValue] -> Either String (f p, [FieldValue])--instance GStructDecode U1 where- gStructDecodeStruct _ structVal =- if null (elems (structValueFields structVal))- then Right U1- else Left ("duckdb-simple: expected empty struct, but got " <> show (length (elems (structValueFields structVal))) <> " field(s)")- gStructDecodeList _ xs = Right (U1, xs)--instance (GStructDecode a, GStructDecode b) => GStructDecode (a :*: b) where- gStructDecodeStruct _ structVal = do- let values = map structFieldValue (elems (structValueFields structVal))- (leftVal, rest) <- gStructDecodeList (Proxy :: Proxy (a p)) values- (rightVal, rest') <- gStructDecodeList (Proxy :: Proxy (b p)) rest- unless (null rest') $- Left ("duckdb-simple: extra " <> show (length rest') <> " field(s) when decoding struct (too many fields provided)")- pure (leftVal :*: rightVal)- gStructDecodeList _ xs = do- (leftVal, rest) <- gStructDecodeList (Proxy :: Proxy (a p)) xs- (rightVal, rest') <- gStructDecodeList (Proxy :: Proxy (b p)) rest- pure (leftVal :*: rightVal, rest')--instance (Selector s, DuckValue a) => GStructDecode (M1 S s (K1 i a)) where- gStructDecodeStruct _ structVal =- case map structFieldValue (elems (structValueFields structVal)) of- [fv] -> M1 . K1 <$> duckFromField fv- [] -> Left "duckdb-simple: missing struct field (expected 1, got 0)"- xs -> Left ("duckdb-simple: expected single field struct, but got " <> show (length xs) <> " fields")-- gStructDecodeList _ [] = Left "duckdb-simple: missing struct field (expected field but list is empty)"- gStructDecodeList _ (fv : rest) = do- val <- duckFromField fv- pure (M1 (K1 val), rest)--instance (GStructDecode f) => GStructDecode (M1 C c f) where- gStructDecodeStruct _ structVal = M1 <$> gStructDecodeStruct (Proxy :: Proxy (f p)) structVal- gStructDecodeList _ values = do- (inner, rest) <- gStructDecodeList (Proxy :: Proxy (f p)) values- pure (M1 inner, rest)------------------------------------------------------------------------------------- GFromField (inverse generic)--{- | Generic decoding entry point mirroring @GToField@. This is used both by-@genericFromFieldValue@ and the @Generically@ deriving helper.--}-class GFromField f where- gFromField :: FieldValue -> Either String (f p)--instance (GFromField' (IsSum f) f) => GFromField f where- gFromField = gFromField' (Proxy :: Proxy (IsSum f))--class GFromField' (isSum :: Bool) f where- gFromField' :: Proxy isSum -> FieldValue -> Either String (f p)--instance (GStruct f, GStructDecode f) => GFromField' 'False (M1 D meta (M1 C c f)) where- gFromField' _ = \case- FieldNull -> M1 . M1 . fst <$> gStructDecodeList (Proxy :: Proxy (f p)) []- FieldStruct sv -> do- ordered <- orderStructFields (gStructTypes (Proxy :: Proxy (f p))) sv- M1 . M1 <$> gStructDecodeStruct (Proxy :: Proxy (f p)) ordered- other -> Left ("duckdb-simple: expected STRUCT value for product type, got " <> show other)--instance (GSum f) => GFromField' 'True (M1 D meta f) where- gFromField' _ = \case- FieldUnion uv -> do- let actualMembers = elems (unionValueMembers uv)- actualIndex = fromIntegral (unionValueIndex uv)- unless (actualIndex < length actualMembers) $- Left "duckdb-simple: union tag out of range"- unless (unionMemberName (actualMembers !! actualIndex) == unionValueLabel uv) $- Left "duckdb-simple: union tag and member name mismatch"- let members = gSumMembers (Proxy :: Proxy (f p))- indexed = zip [0 ..] members- matching = [idx | (idx, member) <- indexed, unionMemberName member == unionValueLabel uv]- case matching of- [idx] -> M1 <$> gSumDecode idx (unionValuePayload uv)- _ -> Left "duckdb-simple: unknown union member name"- other -> Left ("duckdb-simple: expected UNION value for sum type, got " <> show other)--instance GFromField' 'False (M1 D meta U1) where- gFromField' _ _ = Right (M1 U1)------------------------------------------------------------------------------------- ViaDuckDB instances--{- | The deriving-via version of @DuckDBColumnType@. We look at the generic-logical type and map it back to a textual name. The textual names are only-used for diagnostics (errors and column metadata).--}-instance (Generic a, GToField (Rep a)) => DuckDBColumnType (ViaDuckDB a) where- duckdbColumnTypeFor _ =- case genericLogicalType (Proxy :: Proxy a) of- LogicalTypeStruct{} -> Text.pack "STRUCT"- LogicalTypeUnion{} -> Text.pack "UNION"- LogicalTypeList{} -> Text.pack "LIST"- LogicalTypeArray{} -> Text.pack "ARRAY"- LogicalTypeMap{} -> Text.pack "MAP"- LogicalTypeScalar dtype -> duckDBTypeToName dtype- LogicalTypeDecimal{} -> Text.pack "DECIMAL"- LogicalTypeEnum{} -> Text.pack "ENUM"- LogicalTypeGeometry{} -> Text.pack "GEOMETRY"--{- | Deriving-via @ToField@ instance. We reuse the helpers above to decide-whether the top-level representation is a union, struct, or scalar and then-delegate to the existing @ToField@ instances for those composite types.--}-instance (Generic a, GToField (Rep a)) => ToField (ViaDuckDB a) where- toField (ViaDuckDB x) =- case genericToUnionValue x of- Just unionVal -> toField unionVal- Nothing ->- case genericToStructValue x of- Just structVal -> toField structVal- Nothing ->- case genericToFieldValue x of- FieldUnion uv -> toField uv- FieldStruct sv -> toField sv- FieldNull -> toField (Nothing :: Maybe Int)- other -> error ("duckdb-simple: unsupported generic encoding " <> show other)--{- | Deriving-via @FromField@ instance. Errors are rewrapped using the existing-@returnError@ helper so callers receive a proper @ResultError@.--}-instance (Generic a, GFromField (Rep a), Typeable a) => FromField (ViaDuckDB a) where- fromField f@Field{fieldValue} =- case genericFromFieldValue fieldValue of- Right value -> pure (ViaDuckDB value)- Left err ->- returnError ConversionFailed f (Text.pack err)------------------------------------------------------------------------------------- DuckDB type constructors (re-exported patterns)---- These pattern synonyms come from duckdb-ffi; re-exporting to avoid users having to import it.+{-# LANGUAGE TypeApplications #-}+{-# LANGUAGE TypeFamilies #-}+{-# LANGUAGE TypeOperators #-}+{-# LANGUAGE UndecidableInstances #-}++-- |+-- Module : Database.DuckDB.Simple.Generic+-- Description : Generic helpers for encoding Haskell ADTs as DuckDB structs/unions.+--+-- Encode and decode algebraic data types through @ToField@/@FromField@ with+-- GHC generics. The mapping is:+--+-- * /Product types/ (records or tuples) whose fields already satisfy @DuckValue@+-- are encoded as STRUCT values. Record fields retain their selector name;+-- positional products fall back to @field1@, @field2@, …+-- * /Sum types/ (:+:) become UNION values. Each constructor becomes a union+-- member; payloads are encoded as structs (a STRUCT without fields for a+-- constructor without fields).+--+-- Each field needs a @DuckValue@ instance with a finite logical type. A nested+-- user-defined type needs its own instance.+--+-- Encode a product:+--+-- > data User = User { userId :: Int64, userName :: Text }+-- > deriving stock (Generic)+-- >+-- > genericToFieldValue (User 1 "alice")+-- > -- A FieldStruct with the fields userId = FieldInt64 1 and userName = FieldText "alice".+--+-- For sum types:+--+-- > data Shape+-- > = Circle Double+-- > | Rectangle Double Double+-- > | Origin+-- > deriving stock (Generic)+--+-- Constructors are turned into a union with the members @Circle@ (a STRUCT with+-- the field @field1@), @Rectangle@ (a STRUCT with the fields @field1@ and+-- @field2@), and @Origin@ (a STRUCT without fields).+--+-- Derive instances with @DerivingVia@ and the @ViaDuckDB@ newtype:+--+-- > data User = User { userId :: Int64, userName :: Text }+-- > deriving stock (Generic)+-- > deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB User)+--+-- The derived instances encode and decode STRUCT/UNION representations.+-- Also derive @ToDuckValue@ to use the type as a column of a typed table function,+-- as an argument or result of a typed aggregate function, or as the result of+-- a scalar function.+--+-- === Extending this module+--+-- The implementation has these parts:+--+-- * @DuckValue@ covers leaf-level conversions between Haskell values, DuckDB+-- @FieldValue@s, and logical type metadata.+-- * @GToField@ and the related classes convert the generic representation to+-- structs or unions. They keep the logical type information for parameter+-- binding.+-- * @ViaDuckDB@ supplies the deriving-via instances.+--+-- Follow this structure when you add a feature, and document each new class.+module Database.DuckDB.Simple.Generic (+ -- * Field-level primitives+ DuckValue (..),+ typedFromField,+ GToField (),+ GFromField (),++ -- * Generic encoding/decoding for ADTs+ genericToFieldValue,+ genericFromFieldValue,+ genericLogicalType,+ genericToStructValue,+ genericToUnionValue,++ -- * DerivingVia helper+ ViaDuckDB (..),+) where++import Control.Exception (toException)+import Control.Monad (unless)+import Data.Array (Array, elems, listArray)+import qualified Data.ByteString as BS+import Data.Int (Int16, Int32, Int64, Int8)+import Data.List (intercalate, stripPrefix)+import qualified Data.Map.Strict as Map+import Data.Maybe (fromMaybe, isJust)+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Time (UTCTime)+import Data.Time.Calendar (Day)+import Data.Time.LocalTime (LocalTime, TimeOfDay)+import Data.Typeable (Typeable)+import qualified Data.UUID as UUID+import Data.Word (Word16, Word32, Word64, Word8)+import Database.DuckDB.Simple.Conversion (+ Conversion,+ conversionError,+ mapConversionErrors,+ mapConversionErrorsWith,+ )+import Database.DuckDB.Simple.Element (containsType, valueTypeName)+import Database.DuckDB.Simple.FromField (+ Field (..),+ FieldValue (..),+ FromField (..),+ IntervalValue (..),+ ResultError (..),+ SizedArray,+ TimeWithZone (..),+ fieldTypeName,+ returnError,+ sizedArrayFromList,+ sizedArrayToArray,+ )+import Database.DuckDB.Simple.Internal (duckDBTypeToName)+import Database.DuckDB.Simple.LogicalRep (+ LogicalTypeRep (..),+ StructField (..),+ StructValue (..),+ TypeDeclaration (..),+ UnionMemberType (..),+ UnionValue (..),+ listElementType,+ mapEntryTypes,+ )+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.Time (Unbounded (..))+import Database.DuckDB.Simple.ToField (DuckDBColumnType (..), ToDuckValue (..), ToField (..))+import Database.DuckDB.Simple.Variant (Variant (..))+import GHC.Generics+import GHC.TypeLits (KnownNat, natVal)+import Numeric.Natural (Natural)++--------------------------------------------------------------------------------+-- DuckValue: bridge between Haskell scalars and FieldValue/LogicalTypeRep++-- | Types that can appear inside generated structs/unions.+--+-- A @DuckValue@ instance must provide:+--+-- * encoding to @FieldValue@+-- * logical type metadata (@duckLogicalType@)+-- * decoding from @FieldValue@+--+-- The primitive instances define the scalar representations for generic+-- conversion. The default decoder uses the @FromField@ instance of the type.+-- See Note [Generic decoding].+class DuckValue a where+ -- | Encode a field value.+ duckToField :: a -> FieldValue++ -- | Decode a field value. Run it with @runConversion@.+ duckFromField :: FieldValue -> Conversion a++ -- | Give the SQL type used to encode the field.+ duckLogicalType :: Proxy a -> LogicalTypeRep++ -- | Decode a value with the SQL type of its place, for example the type of a+ -- field of a STRUCT or of an element of a LIST. The default runs+ -- 'duckFromField' and ignores the supplied type. An instance that reads+ -- through 'FromField' can use+ -- 'typedFromField', so that 'fieldType' returns the type. See+ -- Note [Column types] in "Database.DuckDB.Simple.FromField".+ duckFromTypedField :: LogicalTypeRep -> FieldValue -> Conversion a+ duckFromTypedField _ = duckFromField++ default duckFromField :: (FromField a) => FieldValue -> Conversion a+ duckFromField = typedFromField (LogicalTypeScalar LogicalTypeIdVariant)++{- Note [Generic decoding]+~~~~~~~~~~~~~~~~~~~~~~~~~~+Generic decoding uses Conversion because a FromField instance can request SQL+metadata. Encoding and logical type declarations remain pure. Error context is+added only when a conversion's failure is selected. A discarded Alternative+branch does not request metadata for its diagnostic.+-}++-- | Decode a value through 'FromField', with a t'Field' of the given SQL type.+-- See Note [Generic decoding] and Note [Column types] in+-- "Database.DuckDB.Simple.FromField".+typedFromField :: (FromField a) => LogicalTypeRep -> FieldValue -> Conversion a+typedFromField logical fv =+ fromField Field{fieldName = Text.empty, fieldIndex = 0, fieldValue = fv, fieldType = pure logical}++instance DuckValue Variant where+ duckToField = variantPayload+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdVariant+ duckFromTypedField = typedFromField++instance DuckValue Bool where+ duckToField = FieldBool+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdBoolean+ duckFromTypedField = typedFromField++instance DuckValue Int where+ duckToField = FieldInt64 . fromIntegral+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdBigInt+ duckFromTypedField = typedFromField++instance DuckValue Int8 where+ duckToField = FieldInt8+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdTinyInt+ duckFromTypedField = typedFromField++instance DuckValue Int16 where+ duckToField = FieldInt16+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdSmallInt+ duckFromTypedField = typedFromField++instance DuckValue Int32 where+ duckToField = FieldInt32+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdInteger+ duckFromTypedField = typedFromField++instance DuckValue Int64 where+ duckToField = FieldInt64+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdBigInt+ duckFromTypedField = typedFromField++-- | A field of a generic record has the type HUGEINT. A value outside that range+-- raises an error. A top-level parameter of this type also binds as HUGEINT+-- when the value is in that range. See Note [Integer parameters] in+-- "Database.DuckDB.Simple.ToField".+instance DuckValue Integer where+ duckToField = FieldHugeInt+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdHugeInt+ duckFromTypedField = typedFromField++-- | A field of a generic record has the type UHUGEINT. A value outside that range+-- raises an error. A top-level parameter of this type binds as HUGEINT when the+-- value is in the HUGEINT range. See Note [Integer parameters] in+-- "Database.DuckDB.Simple.ToField".+instance DuckValue Natural where+ duckToField = FieldUHugeInt . toInteger+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdUHugeInt+ duckFromTypedField = typedFromField++instance DuckValue Word where+ duckToField = FieldWord64 . fromIntegral+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdUBigInt+ duckFromTypedField = typedFromField++instance DuckValue Word8 where+ duckToField = FieldWord8+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdUTinyInt+ duckFromTypedField = typedFromField++instance DuckValue Word16 where+ duckToField = FieldWord16+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdUSmallInt+ duckFromTypedField = typedFromField++instance DuckValue Word32 where+ duckToField = FieldWord32+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdUInteger+ duckFromTypedField = typedFromField++instance DuckValue Word64 where+ duckToField = FieldWord64+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdUBigInt+ duckFromTypedField = typedFromField++instance DuckValue Float where+ duckToField = FieldFloat+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdFloat+ duckFromTypedField = typedFromField++instance DuckValue Double where+ duckToField = FieldDouble+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdDouble+ duckFromTypedField = typedFromField++instance DuckValue Text where+ duckToField = FieldText+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdVarchar+ duckFromTypedField = typedFromField++instance DuckValue String where+ duckToField = FieldText . Text.pack+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdVarchar+ duckFromField fv = Text.unpack <$> duckFromField fv+ duckFromTypedField logical fv = Text.unpack <$> duckFromTypedField logical fv++instance DuckValue BS.ByteString where+ duckToField = FieldBlob+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdBlob+ duckFromTypedField = typedFromField++instance DuckValue Day where+ duckToField = FieldDate . Finite+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdDate+ duckFromTypedField = typedFromField++instance DuckValue TimeOfDay where+ duckToField = FieldTime+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdTime+ duckFromTypedField = typedFromField++instance DuckValue LocalTime where+ duckToField = FieldTimestamp . Finite+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdTimestamp+ duckFromTypedField = typedFromField++instance DuckValue UTCTime where+ duckToField = FieldTimestampTZ . Finite+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdTimestampTz+ duckFromTypedField = typedFromField++instance DuckValue (Unbounded Day) where+ duckToField = FieldDate+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdDate+ duckFromTypedField = typedFromField++instance DuckValue (Unbounded LocalTime) where+ duckToField = FieldTimestamp+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdTimestamp+ duckFromTypedField = typedFromField++instance DuckValue (Unbounded UTCTime) where+ duckToField = FieldTimestampTZ+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdTimestampTz+ duckFromTypedField = typedFromField++instance DuckValue UUID.UUID where+ duckToField = FieldUUID+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdUUID+ duckFromTypedField = typedFromField++instance DuckValue IntervalValue where+ duckToField = FieldInterval+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdInterval+ duckFromTypedField = typedFromField++instance DuckValue TimeWithZone where+ duckToField = FieldTimeTZ+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdTimeTz+ duckFromTypedField = typedFromField++instance (DuckValue a) => DuckValue (Maybe a) where+ duckToField (Just x) = duckToField x+ duckToField Nothing = FieldNull+ duckLogicalType _ = duckLogicalType (Proxy @a)+ duckFromField FieldNull = pure Nothing+ duckFromField other = Just <$> duckFromField other+ duckFromTypedField _ FieldNull = pure Nothing+ duckFromTypedField logical other = Just <$> duckFromTypedField logical other++-- | List values encode as DuckDB LIST (variable-length). A field reads a LIST+-- and an ARRAY, as the top-level @FromField@ instance of a list does. 'String'+-- keeps its own instance.+instance {-# OVERLAPPABLE #-} (DuckValue a) => DuckValue [a] where+ duckToField xs = FieldList (map duckToField xs)+ duckLogicalType _ = LogicalTypeList (duckLogicalType (Proxy @a))+ duckFromField = duckFromTypedField (LogicalTypeScalar LogicalTypeIdVariant)+ duckFromTypedField logical (FieldList fvs) = traverse (duckFromTypedField (listElementType logical)) fvs+ duckFromTypedField logical (FieldArray arr) = traverse (duckFromTypedField (listElementType logical)) (elems arr)+ duckFromTypedField logical other = unexpectedTypedValue logical "LIST or ARRAY" other++-- | A 'SizedArray' field is an ARRAY of size @n@. It reads an ARRAY and a LIST of+-- @n@ elements, as the 'FromField' instance does. See Note [Sized arrays] in+-- "Database.DuckDB.Simple.ToField".+instance (KnownNat n, DuckValue a) => DuckValue (SizedArray n a) where+ duckToField values = FieldArray (fmap duckToField (sizedArrayToArray values))++ -- See Note [Sized ARRAY bounds] in "Database.DuckDB.Simple.Native".+ duckLogicalType _ = LogicalTypeArray (duckLogicalType (Proxy @a)) (fromInteger (min (natVal (Proxy @n)) (toInteger (maxBound :: Word64))))+ duckFromField = duckFromTypedField (LogicalTypeScalar LogicalTypeIdVariant)+ duckFromTypedField logical value = case value of+ FieldArray arr -> sized (elems arr)+ FieldList entries -> sized entries+ other -> unexpectedTypedValue logical "ARRAY" other+ where+ size = natVal (Proxy @n)+ sized entries+ | Just reason <- arraySizeReason size = conversionError (sizeError reason)+ | otherwise = case sizedArrayFromList entries of+ Nothing -> conversionError (sizeError (Text.pack ("the value has " <> show (length entries) <> " elements")))+ Just values -> traverse (duckFromTypedField (listElementType logical)) values+ sizeError reason =+ Incompatible+ (fieldTypeName logical value)+ Text.empty+ (Text.pack ("SizedArray " <> show size))+ reason++-- | Map values encode as DuckDB MAP.+instance (Ord k, DuckValue k, DuckValue v) => DuckValue (Map.Map k v) where+ duckToField m =+ let pairs = Map.toList m+ in FieldMap [(duckToField k, duckToField v) | (k, v) <- pairs]+ duckLogicalType _ =+ LogicalTypeMap+ (duckLogicalType (Proxy @k))+ (duckLogicalType (Proxy @v))+ duckFromField = duckFromTypedField (LogicalTypeScalar LogicalTypeIdVariant)+ duckFromTypedField logical (FieldMap pairs) = do+ result <- Map.fromList <$> traverse decodePair pairs+ if Map.size result == length pairs+ then pure result+ else decodingError "duplicate MAP key after conversion"+ where+ (keyType, valueType) = mapEntryTypes logical+ decodePair (kfv, vfv) = do+ k <- duckFromTypedField keyType kfv+ v <- duckFromTypedField valueType vfv+ pure (k, v)+ duckFromTypedField logical other = unexpectedTypedValue logical "MAP" other++-- | The error of a decoder that does not read a value. The text has the form of+-- each other read error: it names the SQL type of the value, not the value,+-- because a value can be large.+unexpectedValue :: String -> FieldValue -> Conversion a+unexpectedValue expected value = conversionError (Incompatible (valueTypeName value) Text.empty (Text.pack expected) Text.empty)++-- | The error of a decoder that cannot read a value with the given SQL type.+-- It names the type as 'returnError' does.+unexpectedTypedValue :: LogicalTypeRep -> String -> FieldValue -> Conversion a+unexpectedTypedValue logical expected value = conversionError (Incompatible (fieldTypeName logical value) Text.empty (Text.pack expected) Text.empty)++--------------------------------------------------------------------------------+-- Generic machinery++-- | A STRUCT or UNION value with its logical type metadata, produced by the+-- generic traversal for @genericToFieldValue@. @genericLogicalType@ produces+-- the type metadata without a value.+data Encoded+ = EncodedStruct (StructValue FieldValue)+ | EncodedUnion (UnionValue FieldValue)++encodedValue :: Encoded -> FieldValue+encodedValue = \case+ EncodedStruct sv -> FieldStruct sv+ EncodedUnion uv -> FieldUnion uv++-- | Convert a Haskell value (using its generic representation) into a DuckDB @FieldValue@.+genericToFieldValue :: forall a. (Generic a, GToField (Rep a)) => a -> FieldValue+genericToFieldValue = encodedValue . gToField . from++-- | Give the logical DuckDB type of a Haskell type.+genericLogicalType :: forall a. (Generic a, GToField (Rep a)) => Proxy a -> LogicalTypeRep+genericLogicalType _ = gLogicalType (Proxy @(Rep a ()))++-- | Decode a DuckDB @FieldValue@ with its generic representation. Run the+-- result with @runConversion@. See Note [Generic decoding].+genericFromFieldValue :: forall a. (Generic a, GFromField (Rep a)) => FieldValue -> Conversion a+genericFromFieldValue fv = to <$> gFromField fv++-- | Extract the STRUCT encoding when the value produces one.+genericToStructValue :: forall a. (Generic a, GToField (Rep a)) => a -> Maybe (StructValue FieldValue)+genericToStructValue value =+ case gToField (from value) of+ EncodedStruct sv -> Just sv+ _ -> Nothing++-- | Extract a UNION-shaped generic encoding directly when one is produced.+genericToUnionValue :: forall a. (Generic a, GToField (Rep a)) => a -> Maybe (UnionValue FieldValue)+genericToUnionValue value =+ case gToField (from value) of+ EncodedUnion uv -> Just uv+ _ -> Nothing++-- | Wrapper for deriving-via so that @instance ToField (ViaDuckDB a)@ picks up+-- the generic encoding provided by this module.+newtype ViaDuckDB a = ViaDuckDB {getViaDuckDB :: a}++-- Type family to decide whether a representation is a sum.++-- | Select sum encoding with @'True@ and product encoding with @'False@.+type family IsSum f :: Bool where+ IsSum (f :+: g) = 'True+ IsSum (M1 D _ f) = IsSum f+ IsSum (M1 C _ f) = IsSum f+ IsSum _ = 'False++-- | Generic encoding to the intermediate @Encoded@ representation. Every+-- instance must also supply the corresponding logical type description.+class GToField f where+ gToField :: f p -> Encoded+ gLogicalType :: Proxy (f p) -> LogicalTypeRep++instance (GToField' (IsSum f) f) => GToField f where+ gToField = gToField' (Proxy @(IsSum f))+ gLogicalType _ = gLogicalType' (Proxy @(IsSum f)) (Proxy @f)++-- | Select product (@'False@) or sum (@'True@) encoding with the @IsSum@ type+-- family.+class GToField' (isSum :: Bool) f where+ gToField' :: Proxy isSum -> f p -> Encoded+ gLogicalType' :: Proxy isSum -> Proxy f -> LogicalTypeRep++-- Products (single constructor records)++{- Note [Types without fields]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+A product type with no fields encodes as the STRUCT without fields. The payload+of a constructor with no fields in a sum type is also the STRUCT without fields,+and the type of its UNION member is the STRUCT type without fields. DuckDB shows+this value as {}. The value is not NULL.++- Why not NULL. NULL is the encoding of Nothing in a field of the type Maybe a.+ If a value without fields encodes as NULL, then Just x and Nothing are the+ same value in a field.++- The decoders also accept NULL for a product type with no fields and for the+ payload of a constructor with no fields. NULL decodes as the value with+ no fields.+-}++-- | Product encoding: single-constructor datatypes become STRUCT values. See+-- Note [Types without fields].+instance (GStruct f) => GToField' 'False (M1 D meta (M1 C c f)) where+ gToField' _ (M1 (M1 inner)) =+ let comps = gStructValues inner+ typeComps = gStructTypes (Proxy @(f _))+ names = resolveNames (zip progIndices (map fcName comps))+ valueArray = listArrayFrom names (map fcValue comps)+ typeArray = listArrayFrom names (map fcValue typeComps)+ indexMap = Map.fromList (zip names [0 ..])+ in EncodedStruct+ StructValue+ { structValueFields = valueArray+ , structValueTypes = typeArray+ , structValueIndex = indexMap+ }+ where+ progIndices = [0 :: Int ..]+ gLogicalType' _ _ =+ let typeComps = gStructTypes (Proxy @(f _))+ names = resolveNames (zip [0 :: Int ..] (map fcName typeComps))+ typeArray = listArrayFrom names (map fcValue typeComps)+ in LogicalTypeStruct typeArray++-- Sums (encode as union)++-- | Sum encoding: multi-constructor datatypes become UNION values.+instance (GSum f) => GToField' 'True (M1 D meta f) where+ gToField' _ (M1 value) =+ let members = gSumMembers (Proxy @(f _))+ membersArray = listArray (0, length members - 1) members+ (idx, payload) = gSumEncode value+ label = unionMemberName (members !! idx)+ in EncodedUnion+ UnionValue+ { unionValueIndex = fromIntegral idx+ , unionValueLabel = label+ , unionValuePayload = payload+ , unionValueMembers = membersArray+ }+ gLogicalType' _ _ =+ let members = gSumMembers (Proxy @(f _))+ membersArray = listArray (0, length members - 1) members+ in LogicalTypeUnion membersArray++--------------------------------------------------------------------------------+-- GStruct: products++data FieldComponent a = FieldComponent+ { fcName :: Maybe Text+ , fcValue :: a+ }++-- | Keep the selector names from GHC.Generics. Use @fieldN@ for positional+-- products.+resolveNames :: [(Int, Maybe Text)] -> [Text]+resolveNames =+ map pick+ where+ pick (_, Just n) = n+ pick (idx, Nothing) = Text.pack ("field" <> show (idx + 1))++-- | Helper that builds an @Array@ of struct fields from parallel lists of names+-- and payloads.+listArrayFrom :: [Text] -> [b] -> Array Int (StructField b)+listArrayFrom names values =+ listArray+ (0, length values - 1)+ (zipWith (\n v -> StructField{structFieldName = n, structFieldValue = v}) names values)++-- | Collect the values and types of a product in parallel lists for encoding+-- and decoding.+class GStruct f where+ gStructValues :: f p -> [FieldComponent FieldValue]+ gStructTypes :: Proxy (f p) -> [FieldComponent LogicalTypeRep]++instance GStruct U1 where+ gStructValues _ = []+ gStructTypes _ = []++instance (GStruct a, GStruct b) => GStruct (a :*: b) where+ gStructValues (a :*: b) = gStructValues a ++ gStructValues b+ gStructTypes _ = gStructTypes (Proxy @(a _)) ++ gStructTypes (Proxy @(b _))++instance (Selector s, DuckValue a) => GStruct (M1 S s (K1 i a)) where+ gStructValues m@(M1 (K1 x)) =+ let name = toMaybe (selName m)+ in [FieldComponent name (duckToField x)]+ gStructTypes _ =+ let raw = selName (undefined :: M1 S s (K1 i a) ())+ name = toMaybe raw+ in [FieldComponent name (duckLogicalType (Proxy @a))]++instance (GStruct f) => GStruct (M1 C c f) where+ gStructValues (M1 x) = gStructValues x+ gStructTypes _ = gStructTypes (Proxy @(f _))++toMaybe :: String -> Maybe Text+toMaybe name+ | null name = Nothing+ | otherwise = Just (Text.pack name)++--------------------------------------------------------------------------------+-- Sums (unions)++-- | Describe the members of a sum type (@gSumMembers@). Encode a value as its+-- tag and payload (@gSumEncode@), or decode them (@gSumDecode@).+class GSum f where+ gSumMembers :: Proxy (f p) -> [UnionMemberType]+ gSumEncode :: f p -> (Int, FieldValue)+ gSumDecode :: Int -> FieldValue -> Conversion (f p)++instance (GSum a, GSum b) => GSum (a :+: b) where+ gSumMembers _ = gSumMembers (Proxy @(a _)) ++ gSumMembers (Proxy @(b _))+ gSumEncode (L1 x) = gSumEncode x+ gSumEncode (R1 x) =+ let leftCount = length (gSumMembers (Proxy @(a _)))+ (idx, payload) = gSumEncode x+ in (idx + leftCount, payload)+ gSumDecode idx payload =+ let leftCount = length (gSumMembers (Proxy @(a _)))+ in if idx < leftCount+ then L1 <$> gSumDecode idx payload+ else R1 <$> gSumDecode (idx - leftCount) payload++instance (Constructor c, GStruct f, GStructDecode f) => GSum (M1 C c f) where+ gSumMembers _ =+ [ UnionMemberType+ { unionMemberName = Text.pack (conName (undefined :: M1 C c f p))+ , unionMemberType =+ let typeComps = gStructTypes (Proxy @(f _))+ names = resolveNames (zip [0 :: Int ..] (map fcName typeComps))+ in LogicalTypeStruct (listArrayFrom names (map fcValue typeComps))+ }+ ]++ -- See Note [Types without fields].+ gSumEncode (M1 x) =+ let comps = gStructValues x+ typeComps = gStructTypes (Proxy @(f _))+ names = resolveNames (zip [0 :: Int ..] (map fcName comps))+ valueArray = listArrayFrom names (map fcValue comps)+ typeArray = listArrayFrom names (map fcValue typeComps)+ indexMap = Map.fromList (zip names [0 ..])+ in (0, FieldStruct StructValue{structValueFields = valueArray, structValueTypes = typeArray, structValueIndex = indexMap})+ gSumDecode idx payload+ | idx /= 0 = decodingError ("duckdb-simple: union tag mismatch (expected 0, got " <> show idx <> ")")+ | otherwise =+ case payload of+ -- NULL also decodes as a value with no fields. See Note [Types without fields].+ FieldNull -> M1 . fst <$> gStructDecodeList (Proxy @(f _)) []+ FieldStruct structVal -> do+ ordered <- either decodingError pure (orderStructFields (gStructTypes (Proxy @(f _))) structVal)+ M1 <$> gStructDecodeStruct (Proxy @(f _)) ordered+ other -> unexpectedValue "STRUCT for a member of a sum type" other++-- | Match decoded fields to their selector names before product decoding.+orderStructFields :: [FieldComponent LogicalTypeRep] -> StructValue FieldValue -> Either String (StructValue FieldValue)+orderStructFields components sv+ -- The fields have the order of the selectors, so the names are unique.+ | map structFieldName fields == names = Right sv+ | otherwise = reorder+ where+ names = resolveNames (zip [0 ..] (map fcName components))+ fields = elems (structValueFields sv)+ reorder = do+ let byName = Map.fromList [(structFieldName field, field) | field <- fields]+ unless (length fields <= length names) $+ Left "duckdb-simple: extra fields when decoding struct"+ unless (length fields == length names && Map.size byName == length names) $+ Left "duckdb-simple: struct field names or count mismatch"+ ordered <-+ traverse (\name -> maybe (Left ("duckdb-simple: missing struct field " <> Text.unpack name)) Right (Map.lookup name byName)) names+ -- The types keep the order of the fields. See Note [Column types] in+ -- "Database.DuckDB.Simple.FromField".+ let typeByName = Map.fromList [(structFieldName field, field) | field <- elems (structValueTypes sv)]+ variant name = StructField name (LogicalTypeScalar LogicalTypeIdVariant)+ orderedTypes = [fromMaybe (variant name) (Map.lookup name typeByName) | name <- names]+ pure+ sv{structValueFields = listArray (0, length ordered - 1) ordered, structValueTypes = listArray (0, length orderedTypes - 1) orderedTypes}++--------------------------------------------------------------------------------+-- GStructDecode: inverse of GStruct for decoding++-- | Inverse of @GStruct@: decode struct payloads back into a generic product.+class GStructDecode f where+ gStructDecodeStruct :: Proxy (f p) -> StructValue FieldValue -> Conversion (f p)++ -- | Consume a prefix of fields from left to right while decoding, returning+ -- the reconstructed value and any remaining fields. Each field has its+ -- type and its value.+ gStructDecodeList :: Proxy (f p) -> [(LogicalTypeRep, FieldValue)] -> Conversion (f p, [(LogicalTypeRep, FieldValue)])++instance GStructDecode U1 where+ gStructDecodeStruct _ structVal =+ if null (elems (structValueFields structVal))+ then pure U1+ else decodingError ("duckdb-simple: expected empty struct, but got " <> show (length (elems (structValueFields structVal))) <> " field(s)")+ gStructDecodeList _ xs = pure (U1, xs)++instance (GStructDecode a, GStructDecode b) => GStructDecode (a :*: b) where+ gStructDecodeStruct proxy structVal = do+ (decoded, rest) <- gStructDecodeList proxy (typedFields structVal)+ unless (null rest) $+ decodingError ("duckdb-simple: extra " <> show (length rest) <> " field(s) when decoding struct (too many fields provided)")+ pure decoded+ gStructDecodeList _ xs = do+ (leftVal, rest) <- gStructDecodeList (Proxy @(a _)) xs+ (rightVal, rest') <- gStructDecodeList (Proxy @(b _)) rest+ pure (leftVal :*: rightVal, rest')++instance (Selector s, DuckValue a) => GStructDecode (M1 S s (K1 i a)) where+ gStructDecodeStruct _ structVal =+ case typedFields structVal of+ [(logical, fv)] -> M1 . K1 <$> namedField (selName (undefined :: M1 S s (K1 i a) ())) (duckFromTypedField logical fv)+ [] -> decodingError "duckdb-simple: missing struct field (expected 1, got 0)"+ xs -> decodingError ("duckdb-simple: expected single field struct, but got " <> show (length xs) <> " fields")++ gStructDecodeList _ [] = decodingError "duckdb-simple: missing struct field (expected field but list is empty)"+ gStructDecodeList _ ((logical, fv) : rest) = do+ val <- namedField (selName (undefined :: M1 S s (K1 i a) ())) (duckFromTypedField logical fv)+ pure (M1 (K1 val), rest)++-- | The type and the value of each field of a STRUCT value, in order. A field+-- with no type in the value has the type VARIANT. See Note [Column types] in+-- "Database.DuckDB.Simple.FromField".+typedFields :: StructValue FieldValue -> [(LogicalTypeRep, FieldValue)]+typedFields StructValue{structValueFields, structValueTypes} =+ zip+ (map structFieldValue (elems structValueTypes) <> repeat (LogicalTypeScalar LogicalTypeIdVariant))+ (map structFieldValue (elems structValueFields))++-- | Name the record field in the error of its decoder, after the prefix of the+-- library. A positional field has no name. See Note [Generic decoding].+namedField :: String -> Conversion a -> Conversion a+namedField name+ | null name = id+ | otherwise = mapConversionErrors \errors ->+ [ toException+ ( ConversionFailed+ Text.empty+ Text.empty+ Text.empty+ (Text.pack ("duckdb-simple: field " <> name <> ": " <> withoutPrefix (intercalate "; " (map exceptionText errors))))+ )+ ]+ where+ withoutPrefix message = fromMaybe message (stripPrefix "duckdb-simple: " message)++-- | Fail a generic decoder with the supplied diagnostic text.+decodingError :: String -> Conversion a+decodingError = conversionError . ConversionFailed Text.empty Text.empty Text.empty . Text.pack++instance (GStructDecode f) => GStructDecode (M1 C c f) where+ gStructDecodeStruct _ structVal = M1 <$> gStructDecodeStruct (Proxy @(f _)) structVal+ gStructDecodeList _ values = do+ (inner, rest) <- gStructDecodeList (Proxy @(f _)) values+ pure (M1 inner, rest)++--------------------------------------------------------------------------------+-- GFromField (inverse generic)++-- | Generic decoding entry point mirroring @GToField@. This is used both by+-- @genericFromFieldValue@ and the @ViaDuckDB@ deriving helper.+class GFromField f where+ gFromField :: FieldValue -> Conversion (f p)++instance (GFromField' (IsSum f) f) => GFromField f where+ gFromField = gFromField' (Proxy @(IsSum f))++class GFromField' (isSum :: Bool) f where+ gFromField' :: Proxy isSum -> FieldValue -> Conversion (f p)++instance (GStruct f, GStructDecode f) => GFromField' 'False (M1 D meta (M1 C c f)) where+ gFromField' _ = \case+ -- NULL also decodes as a value with no fields. See Note [Types without fields].+ FieldNull -> M1 . M1 . fst <$> gStructDecodeList (Proxy @(f _)) []+ FieldStruct sv -> do+ ordered <- either decodingError pure (orderStructFields (gStructTypes (Proxy @(f _))) sv)+ M1 . M1 <$> gStructDecodeStruct (Proxy @(f _)) ordered+ other -> unexpectedValue "STRUCT for a product type" other++instance (GSum f) => GFromField' 'True (M1 D meta f) where+ gFromField' _ = \case+ FieldUnion uv -> do+ let actualMembers = elems (unionValueMembers uv)+ actualIndex = fromIntegral (unionValueIndex uv)+ unless (actualIndex < length actualMembers) $+ decodingError "duckdb-simple: union tag out of range"+ unless (unionMemberName (actualMembers !! actualIndex) == unionValueLabel uv) $+ decodingError "duckdb-simple: union tag and member name mismatch"+ let members = gSumMembers (Proxy @(f _))+ indexed = zip [0 ..] members+ matching = [idx | (idx, member) <- indexed, unionMemberName member == unionValueLabel uv]+ case matching of+ [idx] -> M1 <$> gSumDecode idx (unionValuePayload uv)+ _ -> decodingError "duckdb-simple: unknown union member name"+ other -> unexpectedValue "UNION for a sum type" other++instance GFromField' 'False (M1 D meta U1) where+ gFromField' _ _ = pure (M1 U1)++--------------------------------------------------------------------------------+-- ViaDuckDB instances++-- | The deriving-via version of @DuckDBColumnType@. The text is the bare name+-- of the generic logical type, for diagnostics. The declaration is the+-- structured generic logical type. Each field type comes from 'DuckValue'.+-- See Note [Type declarations] in "Database.DuckDB.Simple.ToField".+-- See Note [Sized ARRAY bounds] in+-- <src/Database.DuckDB.Simple.Native.html Database.DuckDB.Simple.Native>.+instance (Generic a, GToField (Rep a)) => DuckDBColumnType (ViaDuckDB a) where+ duckdbTypeDeclaration _+ | containsType invalidArray logical =+ Left (Text.pack ("a generic field has no ARRAY type, because a DuckDB ARRAY must have 1 to " <> show maxArraySize <> " elements"))+ | otherwise = Right (StructuredType logical)+ where+ logical = genericLogicalType (Proxy @a)+ invalidArray (LogicalTypeArray _ size) = isJust (arraySizeReason (toInteger size))+ invalidArray _ = False+ duckdbColumnTypeFor _ =+ case genericLogicalType (Proxy @a) of+ LogicalTypeStruct{} -> Text.pack "STRUCT"+ LogicalTypeTuple{} -> Text.pack "TUPLE"+ LogicalTypeUnion{} -> Text.pack "UNION"+ LogicalTypeList{} -> Text.pack "LIST"+ LogicalTypeArray{} -> Text.pack "ARRAY"+ LogicalTypeMap{} -> Text.pack "MAP"+ LogicalTypeScalar dtype -> duckDBTypeToName dtype+ LogicalTypeDecimal{} -> Text.pack "DECIMAL"+ LogicalTypeEnum{} -> Text.pack "ENUM"+ LogicalTypeGeometry{} -> Text.pack "GEOMETRY"+ LogicalTypeCollatedVarchar{} -> Text.pack "VARCHAR"+ LogicalTypeAlias name _ -> name++-- | Deriving-via @ToField@ instance. A sum type binds as a UNION and a product+-- type as a STRUCT, through the @ToField@ instances of those values. A product+-- type with no fields binds as a STRUCT without fields. See+-- Note [Types without fields].+instance (Generic a, GToField (Rep a)) => ToField (ViaDuckDB a) where+ toField (ViaDuckDB x) = case gToField (from x) of+ EncodedUnion unionVal -> toField unionVal+ EncodedStruct structVal -> toField structVal++-- | Deriving-via @ToDuckValue@ instance. A sum type gives a UNION value and a+-- product type a STRUCT value, through the @ToDuckValue@ instances of those+-- values. See Note [Types without fields].+instance (Generic a, GToField (Rep a)) => ToDuckValue (ViaDuckDB a) where+ toDuckValue factory (ViaDuckDB x) = case gToField (from x) of+ EncodedUnion unionVal -> toDuckValue factory unionVal+ EncodedStruct structVal -> toDuckValue factory structVal++-- | Deriving-via @FromField@ instance. It reads a STRUCT for a product type and+-- a UNION for a sum type. Convert decoder errors to @ResultError@ with+-- @returnError@. See Note [Generic decoding].+instance (Generic a, GFromField (Rep a), Typeable a) => FromField (ViaDuckDB a) where+ fromField f@Field{fieldValue} =+ mapConversionErrorsWith+ (\errors -> [] <$ returnError @(ViaDuckDB a) ConversionFailed f (Text.pack (intercalate "; " (map exceptionText errors))))+ (ViaDuckDB <$> genericFromFieldValue fieldValue)
src/Database/DuckDB/Simple/Geometry.hs view
@@ -1,8 +1,12 @@--- | Raw geometry values and conversion to 'G.Geometry' from @geometry-simple@.+{-# LANGUAGE StrictData #-}++-- |+-- Module : Database.DuckDB.Simple.Geometry+-- Description : Raw geometry values and conversion to 'G.Geometry' from @geometry-simple@. module Database.DuckDB.Simple.Geometry (- RawGeometry (..),- fromRawGeometry,- toRawGeometry,+ RawGeometry (..),+ fromRawGeometry,+ toRawGeometry, ) where import Data.ByteString (ByteString)@@ -10,31 +14,40 @@ import Data.Geometry.WKB (decodeWKB, encodeWKB) import Data.Text (Text) -{- | ISO WKB bytes with an optional CRS definition.-The definition can be an identifier, a custom name, or WKT2\/PROJJSON text.-Reading this type copies the bytes without decoding the coordinates.-This type has no @ToField@ instance. Bind 'rawGeometryWKB' with-@ST_GeomFromWKB(?)@. Apply 'rawGeometryCRS' with @ST_SetCRS(..., ?)@.-Omit @ST_SetCRS@ when the CRS is 'Nothing'. SQL NULL propagates to the geometry.-DuckDB can normalize the byte order and CRS definition during import.-Derived equality compares the bytes and CRS, not spatial equivalence.--}+-- | ISO WKB bytes with an optional CRS definition.+-- The definition can be an identifier, a custom name, or WKT2\/PROJJSON text.+-- Reading this type copies the bytes without decoding the coordinates.+-- A parameter of this type binds as a GEOMETRY value with the CRS+-- 'rawGeometryCRS'. DuckDB then casts it to the type of the parameter. A value+-- without a CRS takes the CRS of the target. A target without a CRS removes the+-- CRS of the value. A target with another CRS raises an error. Use+-- @Maybe RawGeometry@ for SQL NULL. A @SizedArray@, a list, or a typed row cannot+-- hold this type. A STRUCT, TUPLE, or UNION parameter binds a @FieldGeometry@+-- value with its CRS. A VARIANT parameter binds one only when it has no CRS. See+-- Note [Raw GEOMETRY parameters] in "Database.DuckDB.Simple.ToField".+-- Parameters accept custom CRS labels, including in composite types. Binding a+-- blank label or a label containing NUL raises an @SQLError@. Typed registrations+-- require a recognized name or a full definition.+-- DuckDB can normalize the byte order and CRS definition during import.+-- The pinned native API requires the same Z/M dimensions for all collection+-- members, including empty members. Native import rejects mixed member layouts.+-- The standalone WKB codec preserves these layouts.+-- Derived equality compares the bytes and CRS, not spatial equivalence. data RawGeometry = RawGeometry- { rawGeometryWKB :: !ByteString- , rawGeometryCRS :: !(Maybe Text)- }- deriving (Eq, Show, Read)+ { rawGeometryWKB :: ByteString+ , rawGeometryCRS :: Maybe Text+ }+ deriving (Eq, Show, Read) -{- | Decode owned WKB with the standalone codec. The shape does not store CRS-metadata. Empty multi-geometries and collections have no layout tag in the-decoded representation. Keep the raw value when these details are needed.--}+-- | Decode owned WKB with the standalone codec. The shape does not store CRS+-- metadata. Empty multi-geometries and collections have no layout tag in the+-- decoded representation. Keep the raw value when these details are needed. fromRawGeometry :: RawGeometry -> Either String G.Geometry fromRawGeometry = decodeWKB . rawGeometryWKB -{- | Encode a shape as ISO WKB with no CRS. Set 'rawGeometryCRS' on the result-to supply a CRS definition for explicit SQL import. This does not transform-the coordinates.--}+-- | Encode a shape as ISO WKB with no CRS. Set 'rawGeometryCRS' on the result+-- to give the value a CRS definition. This does not transform+-- the coordinates. The standalone codec can encode mixed collection layouts+-- that the pinned native API rejects during import. toRawGeometry :: G.Geometry -> Either String RawGeometry toRawGeometry shape = (`RawGeometry` Nothing) <$> encodeWKB shape
src/Database/DuckDB/Simple/Internal.hs view
@@ -1,453 +1,1403 @@-{-# LANGUAGE DerivingStrategies #-}-{-# LANGUAGE MagicHash #-}-{-# LANGUAGE NamedFieldPuns #-}-{-# LANGUAGE PatternSynonyms #-}-{-# LANGUAGE StrictData #-}-{-# OPTIONS_GHC -Wno-deprecations #-}--{- |-Module : Database.DuckDB.Simple.Internal-Description : Internal machinery backing the duckdb-simple API surface.--This module provides access to the opaque data constructors and helper-utilities required by the high-level API. It is not part of the supported-public interface; consumers should depend on @Database.DuckDB.Simple@ instead.--}-module Database.DuckDB.Simple.Internal (- -- * Data constructors (internal use only)- Query (..),- Connection (..),- ConnectionState (..),- Statement (..),- StatementState (..),- StatementStreamState (..),- StatementStream (..),- ResultMode (..),- StatementStreamColumn (..),- StatementStreamChunk (..),- SQLError (..),- toSQLError,-- -- * Helpers- connectionClosedError,- statementClosedError,- keepAlive,- withDatabaseHandle,- withConnectionHandle,- withTypeCache,- withStatementHandle,- withQueryCString,- peekUtf8CString,- fetchPrepareError,- duckDBTypeFromName,- duckDBTypeToName,- withResult,- runInterruptibleQuery,- executePreparedResult,- fetchResultChunk,- destroyDataChunk,- fetchResultError,- throwResultError,- mkExecuteError,- withClientContext,- destroyClientContext,- destroyValue,- destroyLogicalType,- throwRegistrationError,-) where--import Control.Concurrent (forkIO, newEmptyMVar, putMVar, readMVar, threadDelay, tryReadMVar)-import Control.Exception (Exception, SomeException, bracket, bracket_, mask, mask_, onException, throwIO, try, uninterruptibleMask_)-import Control.Monad (when)-import qualified Data.ByteString as BS-import Data.IORef (IORef, readIORef)-import Data.List (find)-import Data.String (IsString (..))-import Data.Text (Text)-import qualified Data.Text as Text-import qualified Data.Text.Encoding as TextEncoding-import qualified Data.Text.Foreign as TextForeign-import Database.DuckDB.FFI (- DuckDBClientContext,- DuckDBConnection,- DuckDBDataChunk,- DuckDBDatabase,- DuckDBErrorType,- DuckDBPreparedStatement,- DuckDBResult,- DuckDBState,- DuckDBType,- DuckDBValue,- c_duckdb_connection_get_client_context,- c_duckdb_destroy_client_context,- c_duckdb_destroy_data_chunk,- c_duckdb_destroy_result,- c_duckdb_destroy_value,- c_duckdb_execute_prepared,- c_duckdb_fetch_chunk,- c_duckdb_interrupt,- c_duckdb_prepare_error,- c_duckdb_result_error,- c_duckdb_result_error_type,- pattern DuckDBErrorInvalid,- pattern DuckDBSuccess,- )-import qualified Database.DuckDB.FFI as FFI-import Database.DuckDB.FFI.Deprecated (c_duckdb_execute_prepared_streaming)-import Database.DuckDB.Simple.FromField (FieldValue)-import Database.DuckDB.Simple.LogicalRep (destroyLogicalType)-import Database.DuckDB.Simple.TypeCache (TypeCache)-import Foreign.C.String (CString)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Marshal.Utils (fillBytes)-import Foreign.Ptr (Ptr, nullPtr)-import Foreign.Storable (peek, poke, sizeOf)-import GHC.Exts (keepAlive#)-import GHC.IO (IO (..))---- | Represents a textual SQL query with UTF-8 encoding semantics.-newtype Query = Query- { fromQuery :: Text- -- ^ Extract the underlying textual representation of the query.- }- deriving stock (Eq, Ord, Show)--instance Semigroup Query where- Query a <> Query b = Query (a <> b)--instance IsString Query where- fromString = Query . Text.pack---- | Tracks the lifetime of a DuckDB database and connection pair.-newtype Connection = Connection {connectionState :: IORef ConnectionState}---- | Internal connection lifecycle state.-data ConnectionState- = ConnectionClosed- | ConnectionOpen- { connectionDatabase :: DuckDBDatabase- , connectionHandle :: DuckDBConnection- , connectionTypeCache :: TypeCache- }---- | Tracks the lifetime of a prepared statement.-data Statement = Statement- { statementState :: IORef StatementState- , statementConnection :: Connection- , statementQuery :: Query- , statementStream :: IORef StatementStreamState- }---- | Internal statement lifecycle state.-data StatementState- = StatementClosed- | StatementOpen- { statementHandle :: DuckDBPreparedStatement- }---- | Streaming execution state for prepared statements.-data StatementStreamState- = StatementStreamIdle- | StatementStreamExhausted- | StatementStreamActive !StatementStream---- | Streaming cursor backing an active result set.-data StatementStream = StatementStream- { statementStreamResult :: Ptr DuckDBResult- , statementStreamColumns :: [StatementStreamColumn]- , statementStreamChunk :: Maybe StatementStreamChunk- , statementStreamMode :: ResultMode- }---- | Select native execution. DuckDB can materialize a streaming request.-data ResultMode = MaterializedResult | StreamingResult---- | Metadata describing a result column surfaced through streaming.-data StatementStreamColumn = StatementStreamColumn- { statementStreamColumnIndex :: Int- , statementStreamColumnName :: Text- , statementStreamColumnType :: DuckDBType- }---- | Currently loaded data chunk plus iteration cursor.-data StatementStreamChunk = StatementStreamChunk- { statementStreamChunkPtr :: DuckDBDataChunk- , statementStreamChunkSize :: Int- , statementStreamChunkIndex :: Int- , statementStreamChunkReaders :: [Int -> IO FieldValue]- -- ^ One reader for each column. A reader must not outlive its chunk.- }---- | Represents an error reported by DuckDB or by duckdb-simple itself.-data SQLError = SQLError- { sqlErrorMessage :: Text- , sqlErrorType :: Maybe DuckDBErrorType- , sqlErrorQuery :: Maybe Query- }- deriving stock (Eq, Show)--instance Exception SQLError---- | Convert an arbitrary exception into an untyped @SQLError@.-toSQLError :: (Exception e) => e -> SQLError-toSQLError ex =- SQLError- { sqlErrorMessage = Text.pack (show ex)- , sqlErrorType = Nothing- , sqlErrorQuery = Nothing- }---- | Shared error value used when an operation targets a closed connection.-connectionClosedError :: SQLError-connectionClosedError =- SQLError- { sqlErrorMessage = Text.pack "duckdb-simple: connection is closed"- , sqlErrorType = Nothing- , sqlErrorQuery = Nothing- }---- | Shared error value used when an operation targets a closed statement.-statementClosedError :: Statement -> SQLError-statementClosedError Statement{statementQuery} =- SQLError- { sqlErrorMessage = Text.pack "duckdb-simple: statement is closed"- , sqlErrorType = Nothing- , sqlErrorQuery = Just statementQuery- }---- | Provide a UTF-8 encoded C string view of the query text.-withQueryCString :: Query -> (CString -> IO a) -> IO a-withQueryCString query@(Query txt) action- | Text.any (== '\0') txt =- throwIO (SQLError (Text.pack "duckdb-simple: SQL contains NUL") Nothing (Just query))- | otherwise = TextForeign.withCString txt action---- | Copy a NUL-terminated UTF-8 string from DuckDB.-peekUtf8CString :: CString -> IO Text-peekUtf8CString ptr = TextEncoding.decodeUtf8 <$> BS.packCString ptr---- | Execute a query and destroy its result after success or failure.-withResult :: Connection -> Query -> (Ptr DuckDBResult -> IO DuckDBState) -> (Ptr DuckDBResult -> IO a) -> IO a-withResult conn queryText executeResult action =- alloca $ \resPtr ->- bracket_- (fillBytes resPtr 0 (sizeOf (undefined :: DuckDBResult)))- (c_duckdb_destroy_result resPtr)- $ do- rc <- runInterruptibleQuery conn (executeResult resPtr)- when (rc /= DuckDBSuccess) $ do- (message, errorType) <- fetchResultError resPtr- throwIO (mkExecuteError queryText message errorType)- action resPtr---- | Execute a prepared statement with the requested native result mode.-executePreparedResult :: ResultMode -> DuckDBPreparedStatement -> Ptr DuckDBResult -> IO DuckDBState-executePreparedResult MaterializedResult = c_duckdb_execute_prepared-executePreparedResult StreamingResult = c_duckdb_execute_prepared_streaming--{- | Fetch an owned chunk. The caller must remain masked until it records ownership.- Streaming fetches can execute SQL. Keep their output in a caller-owned slot- so cancellation can destroy a chunk that the worker has already fetched.--}-fetchResultChunk :: ResultMode -> Connection -> Ptr DuckDBResult -> IO DuckDBDataChunk-fetchResultChunk MaterializedResult conn result =- withConnectionHandle conn $ \_ -> c_duckdb_fetch_chunk result-fetchResultChunk StreamingResult conn result =- alloca $ \chunkPtr -> mask_ $ do- poke chunkPtr nullPtr- let fetch = do- chunk <- c_duckdb_fetch_chunk result- poke chunkPtr chunk- pure DuckDBSuccess- (runInterruptibleQuery conn fetch >> peek chunkPtr)- `onException` c_duckdb_destroy_data_chunk chunkPtr---- | Destroy an owned native data chunk. A null chunk is allowed.-destroyDataChunk :: DuckDBDataChunk -> IO ()-destroyDataChunk chunk = alloca $ \ptr -> poke ptr chunk >> c_duckdb_destroy_data_chunk ptr--{- | Run a native query while the caller can receive asynchronous exceptions.- Prompt cancellation requires the threaded RTS. Cleanup interrupts DuckDB and- waits for the worker before the caller can release native storage.--}-runInterruptibleQuery :: Connection -> IO DuckDBState -> IO DuckDBState-runInterruptibleQuery conn action =- withConnectionHandle conn $ \handle ->- mask $ \restore -> do- done <- newEmptyMVar- _ <- forkIO $ do- outcome <- try action :: IO (Either SomeException DuckDBState)- putMVar done outcome- let cancelAndWait = do- finished <- tryReadMVar done- case finished of- Just _ -> pure ()- Nothing -> do- -- DuckDB clears the interrupt flag at query entry.- -- Repeat the interrupt to cover cancellation before entry.- c_duckdb_interrupt handle- threadDelay interruptRetryDelayMicros- cancelAndWait- -- A second exception must not release storage still used by the worker.- -- Native code and Haskell callbacks must return before cleanup can finish.- outcome <- restore (readMVar done) `onException` uninterruptibleMask_ cancelAndWait- either throwIO pure outcome--{- | Retry cancellation at most 100 times per second while native work finishes.-DuckDB can clear an earlier interrupt at query entry. This interval avoids-busy-spinning and can add up to one interval to completion detection.--}-interruptRetryDelayMicros :: Int-interruptRetryDelayMicros = 10 * 1000--{- | The column type names that 'Database.DuckDB.Simple.ToField.DuckDBColumnType'-instances use, and the types they denote. Each type has one name.--}-duckDBTypeNames :: [(Text, DuckDBType)]-duckDBTypeNames =- map- (\(name, dtype) -> (Text.pack name, dtype))- [ ("BOOLEAN", FFI.DuckDBTypeBoolean)- , ("TINYINT", FFI.DuckDBTypeTinyInt)- , ("SMALLINT", FFI.DuckDBTypeSmallInt)- , ("INTEGER", FFI.DuckDBTypeInteger)- , ("BIGINT", FFI.DuckDBTypeBigInt)- , ("HUGEINT", FFI.DuckDBTypeHugeInt)- , ("UTINYINT", FFI.DuckDBTypeUTinyInt)- , ("USMALLINT", FFI.DuckDBTypeUSmallInt)- , ("UINTEGER", FFI.DuckDBTypeUInteger)- , ("UBIGINT", FFI.DuckDBTypeUBigInt)- , ("UHUGEINT", FFI.DuckDBTypeUHugeInt)- , ("FLOAT", FFI.DuckDBTypeFloat)- , ("DOUBLE", FFI.DuckDBTypeDouble)- , ("DATE", FFI.DuckDBTypeDate)- , ("TIME", FFI.DuckDBTypeTime)- , ("TIMETZ", FFI.DuckDBTypeTimeTz)- , ("TIMESTAMP", FFI.DuckDBTypeTimestamp)- , ("TIMESTAMPTZ", FFI.DuckDBTypeTimestampTz)- , ("INTERVAL", FFI.DuckDBTypeInterval)- , ("TEXT", FFI.DuckDBTypeVarchar)- , ("BLOB", FFI.DuckDBTypeBlob)- , ("GEOMETRY", FFI.DuckDBTypeGeometry)- , ("VARIANT", FFI.DuckDBTypeVariant)- , ("UUID", FFI.DuckDBTypeUUID)- , ("BIT", FFI.DuckDBTypeBit)- , ("BIGNUM", FFI.DuckDBTypeBigNum)- , -- NULL gives an element type to Maybe values without data.- ("NULL", FFI.DuckDBTypeSQLNull)- ]---- | Find the type that a column type name denotes.-duckDBTypeFromName :: Text -> Maybe DuckDBType-duckDBTypeFromName name = lookup name duckDBTypeNames---- | Find the column type name of a type. Other types use their 'Show' text.-duckDBTypeToName :: DuckDBType -> Text-duckDBTypeToName dtype =- maybe (Text.pack (show dtype)) fst (find ((== dtype) . snd) duckDBTypeNames)--{- | Read the error message of a prepared statement as UTF-8. Use the fallback-when DuckDB reports no message.--}-fetchPrepareError :: Text -> DuckDBPreparedStatement -> IO Text-fetchPrepareError fallback statement = do- messagePtr <- c_duckdb_prepare_error statement- if messagePtr == nullPtr then pure fallback else peekUtf8CString messagePtr---- | Copy a result error while its native result remains alive.-fetchResultError :: Ptr DuckDBResult -> IO (Text, Maybe DuckDBErrorType)-fetchResultError resultPtr = do- msgPtr <- c_duckdb_result_error resultPtr- message <-- if msgPtr == nullPtr- then pure (Text.pack "duckdb-simple: query failed")- else peekUtf8CString msgPtr- errorType <- c_duckdb_result_error_type resultPtr- pure (message, if errorType == DuckDBErrorInvalid then Nothing else Just errorType)---- | Attach the query and native error category to an execution error.-mkExecuteError :: Query -> Text -> Maybe DuckDBErrorType -> SQLError-mkExecuteError queryText message errorType =- SQLError- { sqlErrorMessage = message- , sqlErrorType = errorType- , sqlErrorQuery = Just queryText- }---- | Report a fetch failure before treating a null chunk as end of input.-throwResultError :: Query -> Ptr DuckDBResult -> IO ()-throwResultError queryText resPtr = do- errorPtr <- c_duckdb_result_error resPtr- when (errorPtr /= nullPtr) $ do- (message, errorType) <- fetchResultError resPtr- throwIO (mkExecuteError queryText message errorType)---- | Keep the weak-finalizer owner alive until the native operation returns.-keepAlive :: a -> IO b -> IO b-keepAlive owner (IO action) = IO (\state -> keepAlive# owner state action)---- | Internal helper for safely accessing the underlying prepared statement.-withStatementHandle :: Statement -> (DuckDBPreparedStatement -> IO a) -> IO a-withStatementHandle stmt@Statement{statementState, statementConnection} action =- keepAlive stmt $- withConnectionHandle statementConnection $ \_ -> do- state <- readIORef statementState- case state of- StatementClosed -> throwIO (statementClosedError stmt)- StatementOpen{statementHandle} -> action statementHandle---- | Internal helper for safely accessing the underlying connection handle.-withConnectionHandle :: Connection -> (DuckDBConnection -> IO a) -> IO a-withConnectionHandle conn@Connection{connectionState} action =- keepAlive conn $ do- state <- readIORef connectionState- case state of- ConnectionClosed -> throwIO connectionClosedError- ConnectionOpen{connectionHandle} -> action connectionHandle---- | Borrow the type cache of an open connection.-withTypeCache :: Connection -> (TypeCache -> IO a) -> IO a-withTypeCache conn@Connection{connectionState} action =- keepAlive conn $ do- state <- readIORef connectionState- case state of- ConnectionClosed -> throwIO connectionClosedError- ConnectionOpen{connectionTypeCache} -> action connectionTypeCache---- | Internal helper for safely accessing the underlying database handle.-withDatabaseHandle :: Connection -> (DuckDBDatabase -> IO a) -> IO a-withDatabaseHandle conn@Connection{connectionState} action =- keepAlive conn $ do- state <- readIORef connectionState- case state of- ConnectionClosed -> throwIO connectionClosedError- ConnectionOpen{connectionDatabase} -> action connectionDatabase---- | Acquire the client context for the connection, destroying it after the action.-withClientContext :: Connection -> (DuckDBClientContext -> IO a) -> IO a-withClientContext conn action =- withConnectionHandle conn $ \connPtr ->- bracket- (alloca $ \ctxPtr -> c_duckdb_connection_get_client_context connPtr ctxPtr >> peek ctxPtr)- destroyClientContext- action---- | Destroy a client context handle.-destroyClientContext :: DuckDBClientContext -> IO ()-destroyClientContext ctx =- alloca $ \ptr -> poke ptr ctx >> c_duckdb_destroy_client_context ptr---- | Destroy a value handle.-destroyValue :: DuckDBValue -> IO ()-destroyValue value =- alloca $ \ptr -> poke ptr value >> c_duckdb_destroy_value ptr---- | Throw a standardised registration error.-throwRegistrationError :: String -> IO a-throwRegistrationError label =- throwIO- SQLError- { sqlErrorMessage = Text.pack ("duckdb-simple: " <> label <> " failed")- , sqlErrorType = Nothing- , sqlErrorQuery = Nothing- }+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE FlexibleContexts #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE MagicHash #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE StrictData #-}+{-# LANGUAGE TupleSections #-}++-- |+-- Module : Database.DuckDB.Simple.Internal+-- Description : Native V2 ownership and execution used by the public API.+module Database.DuckDB.Simple.Internal (+ module Database.DuckDB.Simple.Native,+ Environment (..),+ EnvironmentState (..),+ Instance (..),+ InstanceState (..),+ Connection (..),+ ConnectionOptions (..),+ defaultConnectionOptions,+ defaultCallbackLockWait,+ defaultAggregateStripes,+ defaultArrowBatchSize,+ ConnectionState (..),+ Statement (..),+ StatementState (..),+ StatementStreamState (..),+ StatementStream (..),+ ResultMode (..),+ StatementStreamColumn (..),+ StatementStreamChunk (..),+ createEnvironment,+ closeEnvironment,+ createInstance,+ closeInstance,+ createConnection,+ closeConnection,+ closeStatementState,+ createStatementWithCollections,+ createStatementFromParsed,+ createStatementFromParsedWithCollections,+ withParsedStatements,+ withCollectionStatement,+ statementSchemas,+ withStatementSchemas,+ clearStatementSchemas,+ withKeptParameterTypes,+ preparedParameterType,+ storeBoundValue,+ clearBoundValues,+ cleanupStatementStreamRef,+ finalizeStream,+ finalizeChunk,+ connectionClosedError,+ statementClosedError,+ keepAlive,+ withInstanceHandle,+ withDatabaseHandle,+ withConnectionHandle,+ withConnectionLock,+ withResultWindow,+ instanceCastFactory,+ readConnectionVariant,+ withStatementHandle,+ withQueryCString,+ withQueryError,+ peekUtf8CString,+ duckDBTypeFromName,+ duckDBTypeToName,+ withResult,+ runInterruptibleQuery,+ executePreparedResult,+ executeParsedResult,+ executeParsedStatement,+ executePreparedArrowResult,+ withArrowResult,+ clearArrowResult,+ fetchResultChunk,+ destroyDataChunk,+) where++import Control.Concurrent (+ MVar,+ ThreadId,+ forkIO,+ modifyMVarMasked,+ myThreadId,+ newEmptyMVar,+ newMVar,+ putMVar,+ readMVar,+ swapMVar,+ takeMVar,+ threadDelay,+ tryReadMVar,+ tryTakeMVar,+ withMVar,+ )+import Control.Exception (+ SomeException,+ allowInterrupt,+ bracket,+ bracketOnError,+ bracket_,+ catch,+ finally,+ mask,+ mask_,+ onException,+ throwIO,+ try,+ uninterruptibleMask_,+ )+import Control.Monad (foldM, forM, forM_, join, unless, void, when)+import Data.Array (Array, bounds, inRange, listArray, (!))+import qualified Data.ByteString as BS+import Data.Coerce (Coercible, coerce)+import Data.IORef (IORef, atomicModifyIORef', atomicWriteIORef, mkWeakIORef, modifyIORef', newIORef, readIORef, writeIORef)+import qualified Data.IntMap.Strict as IntMap+import Data.List (dropWhileEnd, find, intercalate, partition)+import Data.Maybe (fromMaybe, isJust)+import Data.Text (Text)+import qualified Data.Text as Text+import qualified Data.Text.Encoding as Text+import qualified Data.Text.Foreign as TextForeign+import Database.DuckDB.Simple.CallbackMark (hasCallbackMark)+import Database.DuckDB.Simple.Conversion (Conversion)+import Database.DuckDB.Simple.FromField (FieldValue)+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep)+import Database.DuckDB.Simple.Materialize (prepareVariantReader)+import Database.DuckDB.Simple.Native hiding (arraySizeReason, maxArraySize)+import Database.DuckDB.Simple.TypeScope (TypeScope, closeTypeScope)+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeToBaseRep)+import Database.DuckDB.Simple.VariantCodec (VariantState, destroyVariantState, newVariantState, variantConnection)+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.C.String (CString)+import Foreign.C.Types (CChar)+import Foreign.Marshal.Alloc (alloca)+import Foreign.Marshal.Array (withArray)+import Foreign.Ptr (nullPtr)+import Foreign.Storable (peek, poke)+import GHC.Exts (keepAlive#)+import GHC.IO (IO (..))+import GHC.Weak.Finalize (getFinalizerExceptionHandler)+import System.Mem.Weak (Weak, deRefWeak, finalize)+import System.Timeout (timeout)++{- Note [Native handle ownership]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+The native handles form a tree. An environment owns its instances, an instance+owns its connections, and a connection owns its prepared statements. The open+state of each parent holds a registry of its children (Children). Each child+value also holds its parent: instanceEnvironment, connectionInstance, and+statementConnection.++A registry entry is a weak pointer and an MVar. The key of the weak pointer is+the state IORef of the child, so the registry does not keep a live child+reachable. The finalizer of the weak pointer runs the cleanup of the child, and+then it puts the MVar (registerChild). The cleanup closes the state of the+child, and it keeps the parent alive until the cleanup ends (keepAlive). A+child thus closes in one of three ways: by an explicit close, by the close of+its parent, or by the finalizer after the child becomes unreachable. A close+first replaces the state with the closed state+(joinCleanup). A close of a closed state has no effect. Thus a finalizer after+an explicit close has no effect.++A parent closes its children before it destroys its own native handle+(closeChildren). Thus the close order is statement, connection, instance, and+environment. finishChild runs the finalizer of one child, and it waits for the+MVar of that child. The MVar tells that the cleanup is complete, also when the+cleanup runs in another thread. A parent marks its state closed before it+closes its children. A create function raises an error when the state of its+parent is closed. A close must not start while the parent creates a child (see+closeEnvironment and closeInstance).++The garbage collector releases a child first. The finalizer of a parent runs+after the references to the parent drop. The collector runs the finalizers of+one collection in sequence in one thread, and a parent waits for its children.+A cleanup in that thread can wait for a child finalizer that is behind it.+Thus the finalizers of instances and connections run in new threads+(forkFinalizer). A statement has no children, so its finalizer runs in the+thread of the collector.++The environment has no parent and no registry entry. createEnvironment gives+its state a weak pointer of its own, with a finalizer in a new thread.++openWithOptions makes a private environment and instance for one connection.+The connection keeps an action that closes them (connectionReleaseParent). An+explicit close of the connection runs this action, and the finalizer of the+connection does not. Then the finalizers of the private instance and+environment close them.++A function that borrows a native handle, for example withConnectionHandle, runs+its action under keepAlive of the owner. Thus the finalizer of the owner does+not destroy the handle while the action uses it.+-}++-- | Track child finalizers without retaining live children. See+-- Note [Native handle ownership].+newtype Children a = Children (IORef [(Weak (IORef a), MVar ())])++-- | Create an empty child registry.+newChildren :: IO (Children a)+newChildren = Children <$> newIORef []++-- | Register a child and remove completed entries. The first argument runs+-- the finalizer of the child. Use 'forkFinalizer' for a child that has children.+-- See Note [Native handle ownership].+registerChild :: (IO () -> IO ()) -> (a -> Bool) -> Children a -> IORef a -> IO () -> IO ()+registerChild runFinalizer isOpen (Children children) child cleanup = mask_ do+ previous <- readIORef children+ finished <- fmap concat $ forM previous \entry@(weak, done) ->+ deRefWeak weak >>= \case+ Nothing -> finishChild entry >> pure [done]+ Just ref -> do+ state <- readIORef ref+ if isOpen state then pure [] else finishChild entry >> pure [done]+ -- Finish stale cleanup before transferring ownership of the new child.+ done <- newEmptyMVar+ weak <- mkWeakIORef child (runFinalizer (cleanup `finally` putMVar done ()))+ -- Remove only finished snapshot entries. Keep concurrent registrations.+ atomicModifyIORef' children $ \current ->+ ((weak, done) : filter (\(_, completed) -> completed `notElem` finished) current, ())++-- | Run a finalizer in a new thread. Report an exception as the collector+-- reports it for a finalizer. See Note [Native handle ownership].+forkFinalizer :: IO () -> IO ()+forkFinalizer cleanup =+ void . forkIO $+ cleanup `catch` \(err :: SomeException) -> getFinalizerExceptionHandler >>= ($ err)++-- | Run a child finalizer and wait for its completion. See+-- Note [Native handle ownership].+finishChild :: (Weak (IORef a), MVar ()) -> IO ()+finishChild (weak, done) = uninterruptibleMask_ (finalize weak `finally` readMVar done)++-- | Complete child finalizers before their parent closes. See+-- Note [Native handle ownership].+closeChildren :: Children a -> IO ()+closeChildren (Children children) = atomicModifyIORef' children ([],) >>= cleanupAll . map finishChild++-- | Run every cleanup action in order, even when an earlier action throws.+cleanupAll :: [IO ()] -> IO ()+cleanupAll = foldr finally (pure ())++-- | A native environment shared by one or more instances.+newtype Environment = Environment {environmentState :: IORef EnvironmentState}++-- | Internal environment state.+data EnvironmentState = EnvironmentClosed | EnvironmentOpen DuckDBV2EnvironmentHandle (Children InstanceState)++-- | An instance with catalogs and configuration shared by its connections.+data Instance = Instance {instanceState :: IORef InstanceState, instanceEnvironment :: Environment}++-- | Internal instance state.+data InstanceState = InstanceClosed | InstanceOpen DuckDBV2InstanceHandle DuckDBV2FactoryHandle (Children ConnectionState)++-- | A connection that retains its instance until it closes.+data Connection = Connection+ { connectionState :: IORef ConnectionState+ , connectionInstance :: Instance+ , connectionLock :: MVar ()+ -- ^ Held while a native result of this connection runs, and for each fetch+ -- of an explicit stream. A DuckDB callback waits for it at most+ -- 'connectionCallbackLockWait'. See 'withConnectionLock' and+ -- Note [Connection lock].+ , connectionWindow :: IORef (Maybe ThreadId)+ -- ^ The helper thread that runs the current result. See 'withResultWindow'+ -- and Note [Result window].+ , connectionOptions :: ConnectionOptions+ -- ^ The checked options of the open. 'withConnectionLock' reads+ -- 'connectionCallbackLockWait', a typed aggregate reads+ -- 'connectionAggregateStripes', and 'executePreparedArrowResult' reads+ -- 'connectionArrowBatchSize'.+ }++-- | Settings that duckdb-simple applies when it opens a connection.+data ConnectionOptions = ConnectionOptions+ { connectionConfig :: [(Text, Text)]+ -- ^ DuckDB configuration flags. 'Database.DuckDB.Simple.openWithOptions'+ -- applies them when it creates the instance, as+ -- 'Database.DuckDB.Simple.openWithConfig' does.+ -- 'Database.DuckDB.Simple.connectWithOptions' applies them to a new+ -- connection of a running instance, so options that DuckDB accepts only+ -- at startup fail there. 'Database.DuckDB.Simple.connectWithOptions'+ -- writes each option with its default scope. An option whose default+ -- scope is GLOBAL thus changes the instance. This change applies to all+ -- connections of the instance.+ , connectionCallbackLockWait :: Maybe Int+ -- ^ The longest time in microseconds that a DuckDB callback waits for the+ -- lock of this connection. A statement of a callback on the connection+ -- that runs the callback waits for this lock. After the bound, the+ -- statement raises an t'SQLError' of the type 'ErrorResourceInUse'. For+ -- most callbacks, the outer statement then fails. The bound applies+ -- also to a callback that uses this connection while another statement+ -- runs on it. 'Nothing' sets no bound. The default is+ -- 'defaultCallbackLockWait'. The open rejects a negative value.+ -- "Database.DuckDB.Simple.Function" gives the cases that the bound does+ -- not cover.+ , connectionAggregateStripes :: Int+ -- ^ The number of stripes in the state table of each typed aggregate+ -- function that 'Database.DuckDB.Simple.createAggregate' registers+ -- through this connection. The registration keeps the number, also when+ -- another connection calls the aggregate. The table holds the live+ -- aggregate states. Two DuckDB threads that change the table wait for+ -- each other only when they change the same stripe. No result depends on+ -- the number. The default is 'defaultAggregateStripes'. The open rejects+ -- a number that is not positive. The library sets no upper limit. Each+ -- bind of each call site (a call of the aggregate in a query) makes one+ -- IORef for each stripe.+ , connectionArrowBatchSize :: Int+ -- ^ The largest number of rows in one Arrow array of an Arrow fold on this+ -- connection ("Database.DuckDB.Simple.Arrow" and+ -- "Database.DuckDB.Simple.Streaming"). DuckDB can give shorter arrays.+ -- The default is 'defaultArrowBatchSize'. The open rejects a value that is+ -- not positive. For the batch size of DuckDB, write 131072.+ }+ deriving (Eq, Show)++-- | Open without configuration flags and use the default callback lock wait,+-- aggregate stripe count, and Arrow batch size.+defaultConnectionOptions :: ConnectionOptions+defaultConnectionOptions =+ ConnectionOptions+ { connectionConfig = []+ , connectionCallbackLockWait = Just defaultCallbackLockWait+ , connectionAggregateStripes = defaultAggregateStripes+ , connectionArrowBatchSize = defaultArrowBatchSize+ }++-- | The default of 'connectionCallbackLockWait': 10 seconds, in microseconds.+defaultCallbackLockWait :: Int+defaultCallbackLockWait = 10000000++-- | The default of 'connectionAggregateStripes': 64 stripes. See+-- Note [State table stripes] in+-- <src/Database.DuckDB.Simple.Typed.Aggregate.html Database.DuckDB.Simple.Typed.Aggregate>.+defaultAggregateStripes :: Int+defaultAggregateStripes = 64++-- | The default of 'connectionArrowBatchSize': 2048 rows.+defaultArrowBatchSize :: Int+defaultArrowBatchSize = 2048++-- | Internal connection state.+data ConnectionState+ = ConnectionClosed+ | ConnectionOpen+ { connectionHandle :: DuckDBV2ConnectionHandle+ , connectionDecoder :: MVar (Maybe VariantState)+ , connectionStatements :: Children StatementState+ , connectionReleaseParent :: IO ()+ }++-- | A prepared statement and its owned bindings and cursor.+data Statement = Statement+ { statementState :: IORef StatementState+ , statementConnection :: Connection+ , statementQuery :: Query+ , statementStream :: IORef StatementStreamState+ , statementArrowResult :: IORef DuckDBV2ArrowResultHandle+ , statementBindings :: IORef (IntMap.IntMap DuckDBV2ValueHandle)+ , statementParameterNames :: Array Int Text+ -- ^ The parameter names in binding order. The first index is 1.+ , statementParameterTypes :: IORef (Maybe (IntMap.IntMap (Maybe LogicalTypeRep)))+ -- ^ The kept parameter types while 'withKeptParameterTypes' runs. Otherwise+ -- Nothing. See Note [Statement schemas].+ }++-- | Internal prepared statement state.+data StatementState+ = StatementClosed+ | StatementOpen+ { statementHandle :: DuckDBV2PreparedStatementHandle+ , statementParsed :: DuckDBV2SqlStatementHandle+ , statementSchemaCache :: IORef (Maybe (DuckDBV2SchemaHandle, DuckDBV2SchemaHandle))+ }++{- Note [Statement cursor]+~~~~~~~~~~~~~~~~~~~~~~~~~~+The cursor of a statement (statementStream) is in one of three states.++- StatementStreamIdle. The statement has no result. The next fetch+ (nextRowWith in "Database.DuckDB.Simple.Result") executes the statement in+ the mode of that fetch, and the cursor becomes active. A result that is not a+ query result or a count of changed rows gives no rows, and the cursor becomes+ exhausted.++- StatementStreamActive. The cursor owns a result. Each next fetch reads from+ this result in the mode of the result, also when the fetch asks for the+ other mode. A materialized result has all its chunks in+ statementStreamPendingChunks, and its result handle is null. A streaming+ result keeps its result handle open, and it fetches each next chunk from+ DuckDB.++- StatementStreamExhausted. The cursor has no result, and each fetch gives no+ row. A fetch that raises an error while it reads from an active cursor also+ makes the cursor exhausted.++Only a reset makes the cursor idle again: resetStatementStream, or the close of+the statement (closeStatementState). Both release an active result with+cleanupStatementStreamRef.++The cursor owns each native chunk that it holds. fetchChunk writes a new chunk+to the cursor before it reads the metadata of the chunk or makes the readers.+Thus a release of the cursor also releases the chunk. finalizeStream closes the metadata scope, then releases+the current chunk, the pending chunks, and the result.++An effectful decoder can fetch again, reset the statement, or close it.+consumeStream copies the field values and advances the cursor before it runs+the decoder. It releases the chunk at its last row. Field metadata belongs to+the result, not the chunk, and a reset or close invalidates that metadata.+The decoder can also start a new execution and then fail. Error cleanup+compares the metadata scopes by identity: each execution owns a distinct+scope, so cleanup cannot release a replacement result.+-}++-- | Cursor state. Exhaustion persists until the caller resets the statement.+-- See Note [Statement cursor].+data StatementStreamState = StatementStreamIdle | StatementStreamExhausted | StatementStreamActive StatementStream++-- | An execution cursor or an owned collection of materialized chunks. See+-- Note [Statement cursor].+data StatementStream = StatementStream+ { statementStreamResult :: DuckDBV2ResultHandle+ , statementStreamTypes :: TypeScope+ , statementStreamColumns :: [StatementStreamColumn]+ , statementStreamPendingChunks :: [DuckDBV2DataChunkHandle]+ , statementStreamChunk :: Maybe StatementStreamChunk+ }++-- | Select materialization before decoding or incremental native execution.+data ResultMode = MaterializedResult | StreamingResult deriving (Eq, Show)++-- | Metadata for one result column.+data StatementStreamColumn = StatementStreamColumn+ { statementStreamColumnIndex :: Int+ , statementStreamColumnName :: Text+ , statementStreamColumnType :: Conversion LogicalTypeRep+ -- ^ The scoped type of the column, converted on demand through+ -- 'Database.DuckDB.Simple.FromField.fieldType'.+ -- See Note [Column types] in "Database.DuckDB.Simple.FromField".+ , statementStreamColumnTree :: IORef (Maybe LogicalTypeRep)+ -- ^ The type tree of the column, converted from the vector of the first chunk+ -- that needs it. Each execution makes new columns, so the tree lives for one+ -- execution. See Note [Type trees] in+ -- <src/Database.DuckDB.Simple.Materialize.html Database.DuckDB.Simple.Materialize>.+ , statementStreamColumnLabels :: IORef (IntMap.IntMap Text)+ -- ^ The ENUM labels that the readers of the column read, by index. They live+ -- for one execution, as the tree. See Note [ENUM labels] in+ -- <src/Database.DuckDB.Simple.Materialize.html Database.DuckDB.Simple.Materialize>.+ }++-- | One owned chunk and the readers that borrow its vectors.+data StatementStreamChunk = StatementStreamChunk+ { statementStreamChunkPtr :: DuckDBV2DataChunkHandle+ , statementStreamChunkSize :: Int+ , statementStreamChunkIndex :: Int+ , statementStreamChunkReaders :: [Int -> IO FieldValue]+ }++-- | Create an environment. Close it after all related work finishes. See+-- Note [Native handle ownership].+createEnvironment :: IO Environment+createEnvironment = mask_ do+ checkLibraryVersion+ raw <- output c_duckdb_v2_environment_create+ children <- newChildren+ ref <- newIORef (EnvironmentOpen raw children)+ _ <- mkWeakIORef ref (forkFinalizer (closeEnvironmentState ref))+ pure (Environment ref)++-- | Require the DuckDB release that the V2 bindings describe. Upstream marks+-- the V2 API as unstable, so another release can change its functions.+checkLibraryVersion :: IO ()+checkLibraryVersion = do+ actual <- output c_duckdb_v2_library_version >>= readText+ unless (actual == expected) $+ throwIO+ ( SQLError+ ( "duckdb-simple: the V2 API requires DuckDB "+ <> expected+ <> ", but the loaded library is "+ <> actual+ <> ". DuckDB marks the V2 API as unstable, so each DuckDB release needs its own release of duckdb-ffi and duckdb-simple."+ )+ Nothing+ Nothing+ Nothing+ )+ where+ expected = Text.pack ('v' : intercalate "." (map show [dUCKDB_V2_API_VERSION_MAJOR, dUCKDB_V2_API_VERSION_MINOR, dUCKDB_V2_API_VERSION_PATCH]))++-- | Close an environment and its instances and connections.+-- Wait for child creation and all descendant operations to finish before closing.+-- DuckDB refuses to destroy the environment while an instance is alive that+-- 'createInstance' did not create. In this case, this function closes the+-- instances from 'createInstance' and their connections, raises an t'SQLError'+-- with the type 'ErrorResourceInUse', and keeps the environment open.+-- Destroy the other instance, then close the environment again.+-- After a successful close, a second close has no effect.+closeEnvironment :: Environment -> IO ()+closeEnvironment Environment{environmentState} = closeEnvironmentState environmentState++-- | Mark the environment closed before releasing its native state, so that a+-- 'createInstance' that starts after the mark raises an error. A+-- 'createInstance' that read the open state before the mark can still register+-- its instance. Thus the caller must wait for child creation before the close+-- (see 'closeEnvironment'). If the native destroy refuses with+-- @ERROR_RESOURCE_IN_USE@, the native environment is still alive. In this case,+-- put back the open state with the same child registry, and raise the error.+-- 'closeChildren' empties the registry, and a later 'createInstance' can register+-- in it. After a different error, the state stays closed, because the header does+-- not say that the handle stays valid.+closeEnvironmentState :: IORef EnvironmentState -> IO ()+closeEnvironmentState ref = mask_ $ joinCleanup ref \case+ EnvironmentClosed -> (EnvironmentClosed, pure ())+ EnvironmentOpen raw children ->+ ( EnvironmentClosed+ , cleanupAll+ [ closeChildren children+ , destroy c_duckdb_v2_environment_destroy raw `catch` \(err :: SQLError) -> do+ when (sqlErrorType err == Just ErrorResourceInUse) (atomicWriteIORef ref (EnvironmentOpen raw children))+ throwIO err+ ]+ )++-- | Create an instance with configuration applied before its first connection.+createInstance :: Environment -> [(Text, Text)] -> IO Instance+createInstance environment@Environment{environmentState} settings = keepAlive environment $ mask_ do+ (rawEnvironment, children) <-+ readIORef environmentState >>= \case+ EnvironmentClosed -> throwIO (SQLError "duckdb-simple: environment is closed" Nothing Nothing Nothing)+ EnvironmentOpen raw children -> pure (raw, children)+ when (any (\(name, value) -> Text.any (== '\0') name || Text.any (== '\0') value) settings) $+ throwInvalidArgument "configuration contains NUL"+ let hasWhitelist = any ((`elem` ["allowed_directories", "allowed_paths"]) . Text.toLower . fst) settings+ needsInstance (name, _) = Text.toLower name `elem` runtimeOptions || (hasWhitelist && Text.toLower name == "enable_external_access")+ (deferredSettings, startupSettings) = partition needsInstance settings+ (runtimeSettings, externalAccessSettings) = partition ((/= "enable_external_access") . Text.toLower . fst) deferredSettings+ raw <- withTextArray (map fst startupSettings) \names -> withTextArray (map snd startupSettings) \values ->+ output+ ( c_duckdb_v2_instance_create_with_options+ rawEnvironment+ (ConstPtr (if null startupSettings then nullPtr else names))+ (ConstPtr (if null startupSettings then nullPtr else values))+ (fromIntegral (length startupSettings))+ )+ flip onException (destroy c_duckdb_v2_instance_destroy raw) do+ config <- output (c_duckdb_v2_instance_get_config raw)+ -- Set the whitelist before external access can be disabled.+ forM_ (runtimeSettings ++ externalAccessSettings) \(name, value) -> withIdentifier name \key -> withText value \text ->+ checked (c_duckdb_v2_config_set_option_text config key text DuckDBV2SettingScopeDefault)+ connections <- newChildren+ casts <- withInstanceFactory raw pure+ ref <- newIORef (InstanceOpen raw casts connections)+ registerChild forkFinalizer (\case InstanceOpen{} -> True; _ -> False) children ref (keepAlive environment (closeInstanceState ref))+ pure (Instance ref environment)+ where+ -- These options require the native instance and its file system or logger.+ runtimeOptions =+ [ "enable_logging"+ , "logging_mode"+ , "logging_level"+ , "logging_storage"+ , "enabled_log_types"+ , "disabled_log_types"+ , "allowed_directories"+ , "allowed_paths"+ , "disabled_filesystems"+ ]++-- | Close an instance and all its connections. This is idempotent.+-- Wait for connection creation and all descendant operations to finish before closing.+closeInstance :: Instance -> IO ()+closeInstance Instance{instanceState} = closeInstanceState instanceState++-- | Mark the instance closed before releasing its native state.+closeInstanceState :: IORef InstanceState -> IO ()+closeInstanceState ref = mask_ $ joinCleanup ref \case+ InstanceClosed -> (InstanceClosed, pure ())+ InstanceOpen raw _ children -> (InstanceClosed, cleanupAll [closeChildren children, destroy c_duckdb_v2_instance_destroy raw])++-- | Create a connection with checked options. The final action releases the+-- private instance and environment used by the convenience open functions.+createConnection :: Instance -> ConnectionOptions -> IO () -> IO Connection+createConnection parent@Instance{instanceState} options releaseParent = keepAlive parent $ mask_ do+ (rawInstance, children) <-+ readIORef instanceState >>= \case+ InstanceClosed -> throwIO (SQLError "duckdb-simple: instance is closed" Nothing Nothing Nothing)+ InstanceOpen raw _ children -> pure (raw, children)+ raw <- output (c_duckdb_v2_connection_create rawInstance)+ flip onException (destroy c_duckdb_v2_connection_destroy raw) do+ decoder <- newMVar Nothing+ statements <- newChildren+ ref <- newIORef (ConnectionOpen raw decoder statements releaseParent)+ registerChild forkFinalizer (\case ConnectionOpen{} -> True; _ -> False) children ref (keepAlive parent (closeConnectionState False ref))+ Connection ref parent <$> newMVar () <*> newIORef Nothing <*> pure options++-- | Close a connection and every prepared statement that belongs to it.+closeConnection :: Connection -> IO ()+closeConnection Connection{connectionState} = closeConnectionState True connectionState++-- | Close the state of a connection. The first argument tells whether to run+-- 'connectionReleaseParent'. An explicit close gives 'True', and the finalizer+-- gives 'False'. See Note [Native handle ownership].+closeConnectionState :: Bool -> IORef ConnectionState -> IO ()+closeConnectionState releaseParents ref = mask_ $ joinCleanup ref \case+ ConnectionClosed -> (ConnectionClosed, pure ())+ ConnectionOpen{connectionHandle, connectionDecoder, connectionStatements, connectionReleaseParent} ->+ ( ConnectionClosed+ , cleanupAll+ [ closeChildren connectionStatements+ , uninterruptibleMask_ do+ decoder <- swapMVar connectionDecoder Nothing+ forM_ decoder \state -> destroyVariantState state `finally` destroy c_duckdb_v2_connection_destroy (variantConnection state)+ , destroy c_duckdb_v2_connection_destroy connectionHandle+ , when releaseParents connectionReleaseParent+ ]+ )++-- | Replace lifecycle state before running its cleanup action. See+-- Note [Native handle ownership].+joinCleanup :: IORef a -> (a -> (a, IO ())) -> IO ()+joinCleanup ref action = join (atomicModifyIORef' ref action)++-- | Close all native state owned by one statement.+closeStatementState :: Statement -> IO ()+closeStatementState Statement{statementState, statementStream, statementArrowResult, statementBindings} = mask_ $ joinCleanup statementState \case+ StatementClosed -> (StatementClosed, pure ())+ StatementOpen{statementHandle, statementParsed, statementSchemaCache} ->+ ( StatementClosed+ , cleanupAll+ [ cleanupStatementStreamRef statementStream+ , clearArrowResultRef statementArrowResult+ , clearBoundValues statementBindings+ , releaseSchemas statementSchemaCache+ , destroy c_duckdb_v2_prepared_statement_destroy statementHandle+ , destroy c_duckdb_v2_sql_statement_destroy statementParsed+ ]+ )++-- | Release both owned statement schemas.+releaseSchemas :: IORef (Maybe (DuckDBV2SchemaHandle, DuckDBV2SchemaHandle)) -> IO ()+releaseSchemas ref =+ atomicModifyIORef' ref (Nothing,) >>= mapM_ \(result, parameters) ->+ destroy c_duckdb_v2_schema_destroy result `finally` destroy c_duckdb_v2_schema_destroy parameters++-- | Release cached metadata before binding or executing the statement again.+-- See Note [Statement schemas].+clearStatementSchemas :: Statement -> IO ()+clearStatementSchemas Statement{statementState} = mask_ do+ readIORef statementState >>= \case+ StatementClosed -> pure ()+ StatementOpen{statementSchemaCache} -> releaseSchemas statementSchemaCache++-- | Keep the parameter types that the action reads until the action ends. The+-- first read of a parameter type runs the binder. Each later read of that type+-- in the action uses the kept type, also after a column type changes. See+-- Note [Statement schemas].+withKeptParameterTypes :: Statement -> IO a -> IO a+withKeptParameterTypes Statement{statementParameterTypes} =+ bracket_ (writeIORef statementParameterTypes (Just IntMap.empty)) (writeIORef statementParameterTypes Nothing)++-- | Parse SQL and borrow the resulting statements for one action.+withParsedStatements :: Connection -> Query -> ([DuckDBV2SqlStatementHandle] -> IO a) -> IO a+withParsedStatements conn query action = withQueryError query $ withConnectionHandle conn \raw ->+ bracket (parseStatements raw query) releaseParsed (action . reverse)++-- | Parse SQL into owned handles, most recent first for cleanup. The caller+-- must transfer or release them under a mask.+parseStatements :: DuckDBV2ConnectionHandle -> Query -> IO [DuckDBV2SqlStatementHandle]+parseStatements raw query = mask_ $ withQueryCString query \sql -> do+ iterator <- output (c_duckdb_v2_parse_sql raw sql)+ let destroyIterator = destroy c_duckdb_v2_statement_iterator_destroy iterator+ gather owned = do+ parsed <- flip onException (releaseParsed owned) do+ -- Accept cancellation between acquisitions, while every handle has+ -- an owner. Native output acquisition remains masked.+ allowInterrupt+ output (c_duckdb_v2_statement_iterator_next iterator)+ if parsed == nullHandle then pure owned else gather (parsed : owned)+ parsed <- gather [] `onException` destroyIterator+ destroyIterator `onException` releaseParsed parsed+ pure parsed++-- | Release the handles from 'parseStatements' in reverse acquisition order.+releaseParsed :: [DuckDBV2SqlStatementHandle] -> IO ()+releaseParsed = releaseItems (destroy c_duckdb_v2_sql_statement_destroy)++-- | Prepare a borrowed parsed statement. The native handle copies its SQL.+createStatementFromParsed :: Connection -> Query -> DuckDBV2SqlStatementHandle -> IO Statement+createStatementFromParsed parent query parsed = createStatementFromParsedWithCollections parent query parsed []++-- | Prepare a borrowed parsed statement with named input collections. A+-- collection is a DuckDB column data collection: rows in memory that the caller+-- owns. @duckdb_v2_statement_add_collection@ binds a collection to a statement+-- under a name, and the SQL of the statement reads it as a table with that name.+-- Each triple is the name, the collection handle, and the column names. An empty+-- list of column names selects the default names @col1@ to @colN@.+-- The function binds the collections to the parsed statement that it prepares,+-- and to the copy that it keeps for parameter names and schemas. The caller must+-- keep each collection alive and unchanged until the statement closes and the+-- caller destroys the parsed statement.+createStatementFromParsedWithCollections+ :: Connection -> Query -> DuckDBV2SqlStatementHandle -> [(Text, DuckDBV2ColumnDataCollectionHandle, [Text])] -> IO Statement+createStatementFromParsedWithCollections parent query parsed collections = withQueryError query $ withConnectionHandle parent \raw -> mask_ do+ addStatementCollections query collections parsed+ handle <- prepareStatementHandle parent raw parsed+ flip onException (destroy c_duckdb_v2_prepared_statement_destroy handle) do+ -- The caller retains its parsed handle, so keep a separate metadata copy.+ text <- output (c_duckdb_v2_sql_statement_get_text parsed) >>= readText+ copied <- withQueryCString (Query text) \sql -> bracket (output (c_duckdb_v2_parse_sql raw sql)) (destroy c_duckdb_v2_statement_iterator_destroy) \iterator ->+ output (c_duckdb_v2_statement_iterator_next iterator)+ flip onException (destroy c_duckdb_v2_sql_statement_destroy copied) do+ addStatementCollections query collections copied+ managedStatement parent query handle copied++-- | Parse one statement and retain its owned parsed handle for metadata.+-- Collections remain borrowed until the managed statement closes.+createStatementWithCollections+ :: Connection -> Query -> [(Text, DuckDBV2ColumnDataCollectionHandle, [Text])] -> IO Statement+createStatementWithCollections parent query collections = withQueryError query $ withConnectionHandle parent \raw -> mask_ do+ -- Parse all SQL before selecting one statement. A later syntax error has+ -- precedence over the empty or multiple-statement diagnostics.+ parsed <- parseStatements raw query+ flip onException (releaseParsed parsed) $+ case parsed of+ [statement] -> do+ addStatementCollections query collections statement+ handle <- prepareStatementHandle parent raw statement+ managedStatement parent query handle statement+ `onException` destroy c_duckdb_v2_prepared_statement_destroy handle+ [] -> throwInvalidArgument "SQL query contains no statement"+ _ -> throwInvalidArgument "prepared SQL contains more than one statement"++-- | Prepare a borrowed AST. Keep its connection locked until native work ends.+prepareStatementHandle :: Connection -> DuckDBV2ConnectionHandle -> DuckDBV2SqlStatementHandle -> IO DuckDBV2PreparedStatementHandle+prepareStatementHandle parent raw parsed = alloca \slot -> do+ poke slot nullHandle+ flip onException (peek slot >>= destroy c_duckdb_v2_prepared_statement_destroy) do+ -- DuckDB also rejects preparation while another result is open. See+ -- Note [Connection lock].+ withConnectionLock parent (runInterruptibleQuery parent (checked (c_duckdb_v2_prepared_statement_create raw parsed 0 slot)))+ peek slot++-- | Build a managed statement. Take both native handles only on success.+managedStatement :: Connection -> Query -> DuckDBV2PreparedStatementHandle -> DuckDBV2SqlStatementHandle -> IO Statement+managedStatement parent query handle parsed = do+ count <- fromIntegral <$> output (c_duckdb_v2_sql_statement_get_parameter_count parsed)+ names <- parameterNames parsed count+ schemas <- newIORef Nothing+ bindings <- newIORef IntMap.empty+ stream <- newIORef StatementStreamIdle+ arrow <- newIORef nullHandle+ kept <- newIORef Nothing+ ref <- newIORef (StatementOpen handle parsed schemas)+ let statement = Statement ref parent query stream arrow bindings (listArray (1, count) names) kept+ children <-+ readIORef (connectionState parent) >>= \case+ ConnectionClosed -> throwIO connectionClosedError+ ConnectionOpen{connectionStatements} -> pure connectionStatements+ -- The finalizer does not keep its weak key alive through this closure.+ -- See Note [Native handle ownership].+ registerChild id (\case StatementOpen{} -> True; _ -> False) children ref (keepAlive parent (closeStatementState statement))+ pure statement++-- | Attach borrowed collections to one parsed statement.+addStatementCollections :: Query -> [(Text, DuckDBV2ColumnDataCollectionHandle, [Text])] -> DuckDBV2SqlStatementHandle -> IO ()+addStatementCollections query collections statement = forM_ collections \(name, collection, columns) -> do+ when (any (Text.any (== '\0')) (name : columns)) $+ withQueryError query (throwInvalidArgument "collection name or column name contains NUL")+ withIdentifier name \nativeName -> withIdentifierArray columns \nativeColumns ->+ checked+ ( c_duckdb_v2_statement_add_collection+ statement+ nativeName+ collection+ (ConstPtr (if null columns then nullPtr else nativeColumns))+ (fromIntegral (length columns))+ )++-- | Prepare SQL with borrowed column collections as named input tables.+-- The caller must keep each collection alive and unchanged until the callback+-- returns. An empty column-name list selects DuckDB's default column names.+-- The statement and all its results close before this function returns.+withCollectionStatement :: Connection -> Query -> [(Text, DuckDBV2ColumnDataCollectionHandle, [Text])] -> (Statement -> IO a) -> IO a+withCollectionStatement conn queryText collections action =+ withQueryError queryText $+ bracket (createStatementWithCollections conn queryText collections) closeStatementState action++-- | Read the authoritative parameter names in binding order.+parameterNames :: DuckDBV2SqlStatementHandle -> Int -> IO [Text]+parameterNames statement count = reverse <$> foldM readName [] [0 .. count - 1]+ where+ -- V2 has no bulk accessor. A parser extension can supply names that+ -- differ from the statement text. Read the names in binding order.+ readName names index = do+ name <- output (c_duckdb_v2_sql_statement_get_parameter_name statement (fromIntegral index)) >>= readIdentifier+ pure (name : names)++-- | Error raised when an operation targets a closed connection.+connectionClosedError :: SQLError+connectionClosedError = SQLError "duckdb-simple: connection is closed" Nothing Nothing Nothing++-- | Error raised when an operation targets a closed statement.+statementClosedError :: Statement -> SQLError+statementClosedError Statement{statementQuery} = SQLError "duckdb-simple: statement is closed" Nothing (Just statementQuery) Nothing++-- | Keep a Haskell owner reachable until an action finishes. See+-- Note [Native handle ownership].+keepAlive :: a -> IO b -> IO b+keepAlive owner (IO action) = IO \state -> keepAlive# owner state action++-- | Borrow a live instance handle.+withInstanceHandle :: Instance -> (DuckDBV2InstanceHandle -> IO a) -> IO a+withInstanceHandle owner action = withOpenInstance owner (\raw _ -> action raw)++-- | Borrow the factory for DuckDB's built-in casts. All connections on the+-- instance use this factory.+instanceCastFactory :: Instance -> IO DuckDBV2FactoryHandle+instanceCastFactory owner = withOpenInstance owner (\_ casts -> pure casts)++-- | Borrow the instance and its factory while the managed owner is live.+withOpenInstance :: Instance -> (DuckDBV2InstanceHandle -> DuckDBV2FactoryHandle -> IO a) -> IO a+withOpenInstance owner@Instance{instanceState} action =+ keepAlive owner $+ readIORef instanceState >>= \case+ InstanceClosed -> throwIO (SQLError "duckdb-simple: instance is closed" Nothing Nothing Nothing)+ InstanceOpen raw casts _ -> action raw casts++-- | Borrow the instance that owns a live connection.+withDatabaseHandle :: Connection -> (DuckDBV2InstanceHandle -> IO a) -> IO a+withDatabaseHandle conn action = withConnectionHandle conn \_ -> withInstanceHandle (connectionInstance conn) action++-- | Borrow a live connection handle.+withConnectionHandle :: Connection -> (DuckDBV2ConnectionHandle -> IO a) -> IO a+withConnectionHandle owner@Connection{connectionState} action =+ keepAlive owner $+ readIORef connectionState >>= \case+ ConnectionClosed -> throwIO connectionClosedError+ ConnectionOpen{connectionHandle} -> action connectionHandle++{- Note [Connection lock]+~~~~~~~~~~~~~~~~~~~~~~~~~+DuckDB permits one open result for each connection, and it rejects a second+execution. DuckDB also rejects a preparation while another result is open.+Thus connectionLock serializes these native calls of a connection, and the+threads that share a connection wait for each other. withConnectionLock takes+the lock.++These calls take the lock:++- The preparation of a statement (prepareStatementHandle).++- The metadata bind of a statement (statementSchemas).++- The native calls of one result in the result window (withResultWindow). A+ materialized result holds the lock from its execution until it closes. See+ Note [Result window].++- Each fetch of an explicit stream after the result window (fetchChunk in+ "Database.DuckDB.Simple.Result").++- The execution of an explicit Arrow stream, and each fetch of a batch+ (foldArrowWith in "Database.DuckDB.Simple.Arrow.Internal").++An explicit stream keeps its result open between two fetches. It holds the+lock for its execution and for each fetch, but not between two fetches. A+statement of another thread on the connection can then take the lock, and+DuckDB rejects that statement while the result is open.++Each fetch of an explicit stream holds the lock for this reason. The header+says that a result has a single consumer, and that a bind runs at the same time+only on a second connection. Without the lock, another thread can start a+statement when the stream is finished in DuckDB but still has buffered chunks.+A cancellation of the fetch then interrupts the query that runs on the+connection, which is the query of the other thread.++The lock is not reentrant. A thread that holds the lock must not take it again.+Thus:++- A call in the result window does not take the lock. A fetch of an explicit+ stream after the result window takes the lock, because its thread does not+ hold the lock then.++- No step function runs under the lock. A step function is a function of the+ user that a fold calls for each row or for each Arrow batch. It can run a+ statement on the same connection, and that statement takes the lock. The+ folds call a step function after the fetch, or after the result closes.++DuckDB callbacks run while a native call holds the lock. withConnectionLock+gives a bound to the wait of a thread that is inside a callback. See+Note [Callback marks] in "Database.DuckDB.Simple.CallbackMark".+-}++-- | Run an action with the lock of the connection. The lock is not reentrant.+-- Do not run a function of the user, such as the step of a fold, in the+-- action. A thread with a callback mark waits for the lock at most the bound+-- of the connection ('connectionCallbackLockWait'), and then raises an t'SQLError' of+-- the type 'ErrorResourceInUse'. Each other thread waits without a bound. See+-- Note [Connection lock], and Note [Callback marks] in+-- <src/Database.DuckDB.Simple.CallbackMark.html Database.DuckDB.Simple.CallbackMark>.+--+-- The take of the lock is masked. The timeout of the bound can thus arrive only+-- while 'takeMVar' blocks, and then the take has not occurred. After the take,+-- the same handlers as in 'withMVar' put the lock back.+withConnectionLock :: Connection -> IO a -> IO a+withConnectionLock Connection{connectionLock, connectionOptions} action = do+ bound <- case connectionCallbackLockWait connectionOptions of+ Nothing -> pure Nothing+ Just micros -> do+ inside <- hasCallbackMark+ pure (if inside then Just micros else Nothing)+ case bound of+ Nothing -> withMVar connectionLock (const action)+ Just micros -> mask \restore -> do+ -- A free lock needs no timer. 'timeout' with a bound of 0 does not run+ -- its action, so the bound 0 also takes a free lock.+ taken <-+ tryTakeMVar connectionLock >>= \case+ Just () -> pure True+ Nothing -> isJust <$> timeout micros (takeMVar connectionLock)+ unless taken $ throwIO (callbackLockWaitError micros)+ result <- restore action `onException` putMVar connectionLock ()+ putMVar connectionLock ()+ pure result++-- | The error of a callback that waited for a connection lock for its bound.+callbackLockWaitError :: Int -> SQLError+callbackLockWaitError micros =+ SQLError+ ( Text.pack+ ( "duckdb-simple: a callback waited "+ <> seconds+ <> " s for a connection that runs a statement. A callback must not run a statement on the connection that runs the callback. Use another connection of the same instance, or change connectionCallbackLockWait."+ )+ )+ (Just ErrorResourceInUse)+ Nothing+ Nothing+ where+ (whole, fraction) = micros `quotRem` 1000000+ digits = dropWhileEnd (== '0') (let text = show fraction in replicate (6 - length text) '0' <> text)+ seconds = show whole <> (if null digits then "" else '.' : digits)++{- Note [Result window]+~~~~~~~~~~~~~~~~~~~~~~~+Each interruptible native call uses a helper thread. A bound caller, such as+the main thread, changes operating system threads to wait for one. A result+needs several native calls, for example its execution and its fetches. Thus+withResultWindow runs the native calls of one result as one unit, in one helper+thread. The caller changes threads once for the result, and not once for each+call. The result window is the time in which this helper thread runs the+action.++withResultWindow takes the connection lock, and it runs the action with+runInterruptible. The helper thread writes its ThreadId to connectionWindow+before the action, and it writes Nothing after the action. runInterruptible+compares connectionWindow with the current thread. If they are equal, it runs+the work directly. Thus each interruptible call in the action, for example in+executePreparedResult or fetchResultChunk, runs in the helper thread of the+window. The connection lock lets a connection have at most one window at a+time.++The helper thread is interruptible as one unit. A cancellation interrupts+DuckDB until the whole action returns. The caller can be cancelled after the+helper thread has made its result. The release action of withResultWindow then+releases that result. See Note [Interruptible native calls].++These functions use a result window:++- executeParsedStatement.++- executeStatement in "Database.DuckDB.Simple".++- startStatementStream and withParsedRows in "Database.DuckDB.Simple.Result".++- foldArrowWith in "Database.DuckDB.Simple.Arrow.Internal", for a materialized+ result.++A row stream runs its execution and its first fetch in the window. Each later+fetch is outside the window: it takes the connection lock and a helper thread+of its own. An Arrow stream does not use a window. See Note [Connection lock].++The action holds the connection lock. Thus no step function runs in the action.+-}++-- | Run the native calls of one result in one helper thread, and hold the+-- connection lock. The release action runs when the caller is cancelled after+-- the helper thread has made its result. Do not+-- run a function of the user in the action. See Note [Result window].+withResultWindow :: Connection -> (a -> IO ()) -> IO a -> IO a+withResultWindow conn@Connection{connectionWindow} release action =+ withConnectionLock conn . runInterruptible conn release $ do+ self <- myThreadId+ bracket_ (writeIORef connectionWindow (Just self)) (writeIORef connectionWindow Nothing) action++{- Note [Concurrent VARIANT decoding]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+DuckDB permits one active result per connection. The VARIANT decoder runs SQL+while a caller can still be streaming rows, so it needs its own connection.+Each caller opens one helper on its instance at the first VARIANT read and+reuses the helper's prepared statements. Two callers on the same instance can+therefore decode in parallel.++The decoder mutates its statements and cast types while it copies a vector.+The MVar gives one thread exclusive use until the copy finishes. Closing the+caller takes the same MVar before destroying the decoder and its helper.+A reader that finds the slot empty checks that the caller is still open under+the lock. This prevents it from opening a new helper after close.+-}++-- | Copy a VARIANT vector with this connection's decoder. See+-- Note [Concurrent VARIANT decoding].+readConnectionVariant :: Connection -> DuckDBV2VectorHandle -> IO (Int -> IO FieldValue)+readConnectionVariant owner@Connection{connectionState, connectionInstance} vector = keepAlive owner $ mask \restore -> do+ readIORef connectionState >>= \case+ ConnectionClosed -> throwIO connectionClosedError+ ConnectionOpen{connectionDecoder} -> do+ outcome <- modifyMVarMasked connectionDecoder \current -> do+ state <- case current of+ Just state -> pure state+ Nothing -> withConnectionHandle owner \_ -> withOpenInstance connectionInstance \raw casts ->+ bracketOnError (output (c_duckdb_v2_connection_create raw)) (destroy c_duckdb_v2_connection_destroy) \helper ->+ newVariantState helper casts+ result <- try (restore (prepareVariantReader state vector))+ pure (Just state, result)+ either (throwIO :: SomeException -> IO a) pure outcome++-- | Borrow a live prepared statement handle.+withStatementHandle :: Statement -> (DuckDBV2PreparedStatementHandle -> IO a) -> IO a+withStatementHandle owner@Statement{statementState, statementConnection, statementQuery} action = keepAlive owner $+ withQueryError statementQuery $ withConnectionHandle statementConnection \_ ->+ readIORef statementState >>= \case+ StatementClosed -> throwIO (statementClosedError owner)+ StatementOpen{statementHandle} -> action statementHandle++-- | Attach the SQL text to errors from native calls. An error that already has+-- SQL text keeps it, so an error from a nested query reports that query.+withQueryError :: Query -> IO a -> IO a+withQueryError query action = action `catch` \err -> throwIO err{sqlErrorQuery = Just (fromMaybe query (sqlErrorQuery err))}++-- | Borrow SQL as a NUL-terminated UTF-8 string.+withQueryCString :: Query -> (ConstPtr CChar -> IO a) -> IO a+withQueryCString query@(Query text) action = do+ when (Text.any (== '\0') text) $ withQueryError query (throwInvalidArgument "query contains NUL")+ TextForeign.withCString text (action . ConstPtr)++-- | Copy a UTF-8 C string.+peekUtf8CString :: (Coercible a CString) => a -> IO Text+peekUtf8CString ptr = Text.decodeUtf8 <$> BS.packCString (coerce ptr)++{- Note [Interruptible native calls]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+Native work that can run for a long time runs in a helper thread, so that an+asynchronous exception can interrupt it. An asynchronous exception does not+interrupt a thread that is in a foreign call: the documentation of throwTo says+that the exception is not raised until the call has completed. Thus the caller+does not make the call itself. runInterruptible starts the helper thread, and+the caller waits for the outcome of the work. An asynchronous exception that+the caller gets during this wait is a cancellation.++On a cancellation, the caller does these steps:++- It interrupts DuckDB with duckdb_v2_connection_interrupt until the work+ returns. The header says that the interrupt is "A no-op when no query is+ active". An interrupt that arrives before the query starts is thus lost, and+ the caller repeats the interrupt.++- It waits interruptRetryDelayMicros between two interrupts, so that the loop+ is not a busy loop.++- If the work gives a result, the caller releases it with the release action+ of runInterruptible, because the caller does not return that result. If the+ work raises an error, the caller ignores that error.++- It raises the cancellation again.++These steps are uninterruptible (uninterruptibleMask_). Thus the caller+releases its inputs, such as the arrays that withBoundValues lends, only after+the native work returns. Cancellation is cooperative: the native code, and each+DuckDB callback that it runs, must return before the cleanup can finish.++runInterruptibleQuery gives no release action. Its work must keep each native+result in storage that the caller releases. For example, executeParsedResult+gives the work a slot for the result handle. If an exception occurs, the caller+destroys the handle in the slot.++A call from the helper thread of the result window runs the work directly,+without a new helper thread. See Note [Result window].+-}++-- | Run native work with cancellation. Wait for the work before releasing inputs.+-- The work must keep its native results in storage that the caller releases.+-- See Note [Interruptible native calls].+runInterruptibleQuery :: Connection -> IO a -> IO a+runInterruptibleQuery conn = runInterruptible conn (const (pure ()))++-- | Run native work in a helper thread. On cancellation, interrupt DuckDB+-- until the work returns, and release a result that the caller does not receive.+-- A call from the helper thread of 'withResultWindow' runs the work directly.+-- See Note [Interruptible native calls] and Note [Result window].+runInterruptible :: forall a. Connection -> (a -> IO ()) -> IO a -> IO a+runInterruptible conn release action = do+ self <- myThreadId+ window <- readIORef (connectionWindow conn)+ if window == Just self then action else interruptible+ where+ interruptible = withConnectionHandle conn \handle -> mask \restore -> do+ done <- newEmptyMVar+ _ <- forkIO do+ outcome <- try action :: IO (Either SomeException a)+ putMVar done outcome+ let wait =+ tryReadMVar done >>= \case+ Just outcome -> pure outcome+ -- Repeat the interrupt until the work returns. See+ -- Note [Interruptible native calls].+ Nothing -> void (c_duckdb_v2_connection_interrupt handle nullPtr) >> threadDelay interruptRetryDelayMicros >> wait+ stop = uninterruptibleMask_ (wait >>= either (const (pure ())) release)+ outcome <- restore (readMVar done) `onException` stop+ either throwIO pure outcome++-- | Repeat the interrupt at most 100 times each second while the native work+-- finishes. The interval prevents a busy loop. It can add up to one interval to+-- the time at which the caller sees the end of the work. See+-- Note [Interruptible native calls].+interruptRetryDelayMicros :: Int+interruptRetryDelayMicros = 10 * 1000++-- | Execute a borrowed parsed statement. The caller owns the returned cursor.+-- See Note [Interruptible native calls].+executeParsedResult :: Connection -> Query -> DuckDBV2SqlStatementHandle -> IO DuckDBV2ResultHandle+executeParsedResult conn query parsed = withQueryError query $ withConnectionHandle conn \raw -> mask_ $ alloca \slot -> do+ poke slot nullHandle+ flip onException (peek slot >>= destroy c_duckdb_v2_result_destroy) do+ runInterruptibleQuery conn $+ checked (c_duckdb_v2_statement_execute raw parsed (ConstPtr nullPtr) (ConstPtr nullPtr) 0 slot)+ peek slot++-- | Execute a parsed statement directly and drain all its side effects. See+-- Note [Result window].+executeParsedStatement :: Connection -> Query -> DuckDBV2SqlStatementHandle -> IO Int+executeParsedStatement conn queryText parsed = withQueryError queryText . withResultWindow conn (const (pure ())) $+ bracket (executeParsedResult conn queryText parsed) (destroy c_duckdb_v2_result_destroy) \result ->+ fromIntegral <$> runInterruptibleQuery conn (output (c_duckdb_v2_result_drain result))++-- | Execute a prepared statement. The caller owns the returned result. See+-- Note [Interruptible native calls].+executePreparedResult :: Statement -> IO DuckDBV2ResultHandle+executePreparedResult stmt = withStatementHandle stmt \handle -> mask_ $ alloca \slot -> do+ poke slot nullHandle+ flip onException (peek slot >>= destroy c_duckdb_v2_result_destroy) $+ withBoundValues stmt \names values count -> do+ runInterruptibleQuery (statementConnection stmt) $+ checked (c_duckdb_v2_prepared_statement_execute handle names values count slot)+ peek slot++-- | Borrow parameter names and values until native execution returns.+withBoundValues :: Statement -> (ConstPtr DuckDBV2IdentifierT -> ConstPtr DuckDBV2ValueHandle -> DuckDBIdx -> IO a) -> IO a+withBoundValues stmt action = do+ bindings <- IntMap.toAscList <$> readIORef (statementBindings stmt)+ let names = map (\(idx, _) -> statementParameterNames stmt ! idx) bindings+ values = map snd bindings+ withIdentifierArray names \namesPtr -> withArray values \valuesPtr ->+ action (ConstPtr namesPtr) (ConstPtr valuesPtr) (fromIntegral (length bindings))++-- | Execute a prepared Arrow result and register it for statement cleanup.+-- Each array of the result has at most 'connectionArrowBatchSize' rows of the+-- connection. See Note [Interruptible native calls].+executePreparedArrowResult :: Statement -> IO DuckDBV2ArrowResultHandle+executePreparedArrowResult stmt = withStatementHandle stmt \handle -> mask_ $ alloca \slot -> do+ poke slot nullHandle+ flip onException (peek slot >>= destroy c_duckdb_v2_arrow_result_destroy) $+ withBoundValues stmt \names values count -> do+ let batch = fromIntegral (connectionArrowBatchSize (connectionOptions (statementConnection stmt)))+ runInterruptibleQuery (statementConnection stmt) $+ checked (c_duckdb_v2_prepared_statement_execute_arrow handle names values count batch slot)+ result <- peek slot+ writeIORef (statementArrowResult stmt) result+ pure result++-- | Execute an Arrow result and release it on every exit path.+withArrowResult :: Statement -> (DuckDBV2ArrowResultHandle -> IO a) -> IO a+withArrowResult stmt = bracket (executePreparedArrowResult stmt) (const (clearArrowResult stmt))++-- | Release an Arrow result before callbacks use materialized batches.+clearArrowResult :: Statement -> IO ()+clearArrowResult = clearArrowResultRef . statementArrowResult++-- | Remove the owned Arrow cursor before releasing it.+clearArrowResultRef :: IORef DuckDBV2ArrowResultHandle -> IO ()+clearArrowResultRef ref =+ mask_ $+ atomicModifyIORef' ref (nullHandle,) >>= destroy c_duckdb_v2_arrow_result_destroy++-- | Execute and release one result on every exit path.+withResult :: Statement -> (DuckDBV2ResultHandle -> IO a) -> IO a+withResult stmt = bracket (executePreparedResult stmt) (destroy c_duckdb_v2_result_destroy)++-- | Fetch one owned chunk. A null handle denotes completion. See+-- Note [Interruptible native calls].+fetchResultChunk :: Connection -> DuckDBV2ResultHandle -> IO DuckDBV2DataChunkHandle+fetchResultChunk conn result = mask_ $ alloca \slot -> do+ poke slot nullHandle+ flip onException (peek slot >>= destroyDataChunk) do+ runInterruptibleQuery conn (checked (c_duckdb_v2_result_fetch_chunk result slot))+ peek slot++-- | Release an owned native data chunk.+destroyDataChunk :: DuckDBV2DataChunkHandle -> IO ()+destroyDataChunk = destroy c_duckdb_v2_data_chunk_destroy++-- | Release the active cursor and return to the idle state. See+-- Note [Statement cursor].+cleanupStatementStreamRef :: IORef StatementStreamState -> IO ()+cleanupStatementStreamRef ref = mask_ do+ state <- atomicModifyIORef' ref (StatementStreamIdle,)+ case state of+ StatementStreamActive stream -> finalizeStream stream+ _ -> pure ()++-- | Close metadata access, then release the current chunk, pending chunks,+-- and result. Each release runs even if an earlier release raises an error.+finalizeStream :: StatementStream -> IO ()+finalizeStream StatementStream{statementStreamTypes, statementStreamResult, statementStreamChunk, statementStreamPendingChunks} =+ closeTypeScope statementStreamTypes+ `finally` ( maybe (pure ()) finalizeChunk statementStreamChunk+ `finally` (releaseItems destroyDataChunk statementStreamPendingChunks `finally` destroy c_duckdb_v2_result_destroy statementStreamResult)+ )++-- | Release the native chunk of a stream chunk.+finalizeChunk :: StatementStreamChunk -> IO ()+finalizeChunk StatementStreamChunk{statementStreamChunkPtr} = destroyDataChunk statementStreamChunkPtr++-- | Transfer ownership of one successfully constructed parameter value.+storeBoundValue :: Statement -> DuckDBIdx -> DuckDBV2ValueHandle -> IO ()+storeBoundValue stmt index value = withStatementHandle stmt \_ -> mask_ do+ let idx = fromIntegral index+ unless (inRange (bounds (statementParameterNames stmt)) idx) $+ withQueryError (statementQuery stmt) (throwInvalidArgument "parameter index is out of bounds")+ previous <- IntMap.lookup idx <$> readIORef (statementBindings stmt)+ mapM_ destroyValue previous+ atomicModifyIORef' (statementBindings stmt) \values -> (IntMap.insert idx value values, ())++-- | Release all owned parameter values.+clearBoundValues :: IORef (IntMap.IntMap DuckDBV2ValueHandle) -> IO ()+clearBoundValues ref = mask_ $ atomicModifyIORef' ref (IntMap.empty,) >>= releaseItems destroyValue . IntMap.elems++{- Note [Statement schemas]+~~~~~~~~~~~~~~~~~~~~~~~~~~~+An open statement keeps a cache of two schemas (statementSchemaCache): the+schema of its result and the schema of its parameters. statementSchemas fills+the cache on its first call. It runs the metadata bind+(duckdb_v2_statement_bind) on the retained parsed statement under the+connection lock, and it stores both handles. The next calls give the same+handles until the cache is empty again. columnCount and columnName in+"Database.DuckDB.Simple" read the result schema, and preparedParameterType+reads the type of one parameter.++The SQL constructors retain the owned parsed handle from the iterator. Native+preparation copies its AST. The constructors for borrowed parsed handles keep a+separate metadata copy, because the caller owns the input handle.++The statement owns both schemas. withStatementSchemas keeps the statement alive+while an action reads a schema and copies its borrowed names and types. Keeping+only a schema handle or another field of the statement does not keep the state+IORef alive. The finalizer can then close the statement and destroy the schemas.++clearStatementSchemas releases both handles and empties the cache.+resetStatementStream in "Database.DuckDB.Simple.Result" calls it. Thus each+call of bind, bindNamed, clearStatementBindings, and executeStatement clears+the cache first. Each fold of "Database.DuckDB.Simple" clears it before its+first row and after its last row. closeStatementState releases the handles+when the statement closes.++- Why each bind clears the cache. A statement that the caller keeps must see a+ change of the catalog between two binds, for example an ALTER TABLE that+ changes the type of a column. A binding that depends on the type of its+ parameter then builds its value for the new type. With a type from before the+ change, a UTCTime for a column that became TIMESTAMPTZ_NS loses its+ nanoseconds.++- The cost. The first read of a parameter type after a bind runs the binder of+ DuckDB again. Only a binding that reads the type of its parameter has this+ cost: a TimeOfDay, a LocalTime, or a UTCTime (or an Unbounded form) with+ digits below a microsecond (see Note [Nanosecond parameters] in+ "Database.DuckDB.Simple.ToField"), a SizedArray, a list, a Map, a manual STRUCT,+ TUPLE, or UNION value, a FieldList, a FieldArray, or a FieldMap value, and an+ Integer or a Natural outside the HUGEINT range+ (see Note [Integer parameters] in "Database.DuckDB.Simple.ToField").++- The exception is executeMany. One executeMany call binds many rows on one+ statement that no caller can reach. withKeptParameterTypes keeps the decoded+ type of each parameter (statementParameterTypes) from the first row that+ reads it to the end of the call. Thus the binder runs one time for the call,+ and not one time for each row. The schemas are still cleared for each row.+ If the type of a column changes during the call, the later rows of the call+ bind for the old type. Another connection can make such a change, and also+ another thread that uses the same connection.+-}++-- | Bind SQL metadata and borrow the cached output and parameter schemas.+-- The statement owns the handles. They remain valid until the statement resets,+-- rebinds, or closes. Use 'withStatementSchemas' to keep the statement alive+-- while reading them. See Note [Statement schemas].+statementSchemas :: Statement -> IO (DuckDBV2SchemaHandle, DuckDBV2SchemaHandle)+statementSchemas stmt = withStatementHandle stmt \_ -> mask_ do+ state <- readIORef (statementState stmt)+ case state of+ StatementClosed -> throwIO (statementClosedError stmt)+ StatementOpen{statementParsed, statementSchemaCache} ->+ readIORef statementSchemaCache >>= \case+ Just schemas -> pure schemas+ Nothing -> withConnectionHandle (statementConnection stmt) \conn -> alloca \result -> alloca \parameters -> do+ poke result nullHandle+ poke parameters nullHandle+ let release = (peek result >>= destroy c_duckdb_v2_schema_destroy) `finally` (peek parameters >>= destroy c_duckdb_v2_schema_destroy)+ flip onException release do+ withConnectionLock+ (statementConnection stmt)+ (runInterruptibleQuery (statementConnection stmt) (checked (c_duckdb_v2_statement_bind conn statementParsed result parameters)))+ schemas <- (,) <$> peek result <*> peek parameters+ writeIORef statementSchemaCache (Just schemas)+ pure schemas++-- | Borrow both schemas and keep their statement alive until the action ends.+-- Copy borrowed names and types in the action. See Note [Statement schemas].+withStatementSchemas :: Statement -> (DuckDBV2SchemaHandle -> DuckDBV2SchemaHandle -> IO a) -> IO a+withStatementSchemas stmt action = keepAlive stmt (statementSchemas stmt >>= uncurry action)++-- | Read a parameter type when the binder can determine it without values.+-- While 'withKeptParameterTypes' runs, read each type one time. See+-- Note [Statement schemas].+preparedParameterType :: Statement -> DuckDBIdx -> IO (Maybe LogicalTypeRep)+preparedParameterType stmt idx = do+ kept <- readIORef (statementParameterTypes stmt)+ case kept >>= IntMap.lookup (fromIntegral idx) of+ Just rep -> pure rep+ Nothing -> do+ rep <- readParameterType+ -- Outside 'withKeptParameterTypes' the value is Nothing, and the insert does nothing.+ modifyIORef' (statementParameterTypes stmt) (fmap (IntMap.insert (fromIntegral idx) rep))+ pure rep+ where+ readParameterType = withStatementSchemas stmt \_ schema ->+ alloca \name -> do+ dtype <- output (c_duckdb_v2_schema_get_field schema (idx - 1) name)+ ident <- output (c_duckdb_v2_logical_type_get_id dtype)+ -- The binding uses the base type. See Note [Type aliases] in+ -- "Database.DuckDB.Simple.LogicalRep".+ if ident == DuckDBV2LogicalTypeIdUnknown || ident == DuckDBV2LogicalTypeIdSQLNull+ then pure Nothing+ else Just <$> logicalTypeToBaseRep dtype++-- | Column type names and the logical types that they denote. When a type has+-- more than one name, 'duckDBTypeToName' gives the first name.+duckDBTypeNames :: [(Text, LogicalTypeId)]+duckDBTypeNames =+ [ ("BOOLEAN", LogicalTypeIdBoolean)+ , ("TINYINT", LogicalTypeIdTinyInt)+ , ("SMALLINT", LogicalTypeIdSmallInt)+ , ("INTEGER", LogicalTypeIdInteger)+ , ("BIGINT", LogicalTypeIdBigInt)+ , ("HUGEINT", LogicalTypeIdHugeInt)+ , ("UTINYINT", LogicalTypeIdUTinyInt)+ , ("USMALLINT", LogicalTypeIdUSmallInt)+ , ("UINTEGER", LogicalTypeIdUInteger)+ , ("UBIGINT", LogicalTypeIdUBigInt)+ , ("UHUGEINT", LogicalTypeIdUHugeInt)+ , ("FLOAT", LogicalTypeIdFloat)+ , ("DOUBLE", LogicalTypeIdDouble)+ , ("DATE", LogicalTypeIdDate)+ , ("TIME", LogicalTypeIdTime)+ , ("TIMETZ", LogicalTypeIdTimeTz)+ , ("TIMESTAMP", LogicalTypeIdTimestamp)+ , ("TIMESTAMPTZ", LogicalTypeIdTimestampTz)+ , ("TIMESTAMPTZ_NS", LogicalTypeIdTimestampTzNs)+ , ("TIMESTAMP_NS WITH TIME ZONE", LogicalTypeIdTimestampTzNs)+ , ("INTERVAL", LogicalTypeIdInterval)+ , ("VARCHAR", LogicalTypeIdVarchar)+ , ("TEXT", LogicalTypeIdVarchar)+ , ("BLOB", LogicalTypeIdBlob)+ , ("GEOMETRY", LogicalTypeIdGeometry)+ , ("VARIANT", LogicalTypeIdVariant)+ , ("UUID", LogicalTypeIdUUID)+ , ("BIT", LogicalTypeIdBit)+ , ("BIGNUM", LogicalTypeIdBigNum)+ , -- NULL gives an element type to Maybe values without data.+ ("NULL", LogicalTypeIdSQLNull)+ ]++-- | Find the type that a column type name denotes.+duckDBTypeFromName :: Text -> Maybe LogicalTypeId+duckDBTypeFromName name = lookup name duckDBTypeNames++-- | Find the column type name of a type. Other types use their 'Show' text.+duckDBTypeToName :: LogicalTypeId -> Text+duckDBTypeToName dtype =+ maybe (Text.pack (show dtype)) fst (find ((== dtype) . snd) duckDBTypeNames)
src/Database/DuckDB/Simple/Logging.hs view
@@ -1,95 +1,24 @@-{-# LANGUAGE BlockArguments #-}-{-# LANGUAGE NamedFieldPuns #-}--{- |-Module : Database.DuckDB.Simple.Logging-Description : High-level wrappers for DuckDB custom log storage.--}+-- |+-- Module : Database.DuckDB.Simple.Logging+-- Description : V2 log sinks with managed callbacks and copied log entries. module Database.DuckDB.Simple.Logging (- LogEntry (..),- registerLogStorage,+ LogEntry (..),+ LogLevel (..),+ LogScope (..),+ registerLogStorage,+ registerLogSink,+ registerInstanceLogSink,+ logConnectionMessage,+ logInstanceMessage, ) where -import Control.Exception (bracket, mask_)-import Control.Monad (when)-import Data.Ratio ((%))-import Data.Text (Text)-import qualified Data.Text as Text-import qualified Data.Text.Foreign as TextForeign-import Data.Time.Clock (UTCTime)-import Data.Time.Clock.POSIX (posixSecondsToUTCTime)-import Database.DuckDB.FFI-import Database.DuckDB.Simple.Callback (ignoreCallbackExceptions, withCallbackResources)-import Database.DuckDB.Simple.Internal (Connection, peekUtf8CString, throwRegistrationError, withDatabaseHandle)-import Foreign.C.String (CString)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (Ptr, nullFunPtr, nullPtr)-import Foreign.Storable (peek, poke)---- | A single log event delivered through DuckDB's log-storage callback.-data LogEntry = LogEntry- { logEntryTimestamp :: !(Maybe UTCTime)- , logEntryLevel :: !Text- , logEntryType :: !Text- , logEntryMessage :: !Text- }- deriving (Eq, Show)---- | Register a custom log storage callback on the database behind a connection.-registerLogStorage :: Connection -> Text -> (LogEntry -> IO ()) -> IO ()-registerLogStorage conn name callback = do- when (Text.null name || Text.any (== '\0') name) $- throwRegistrationError "invalid log storage name"- bracket c_duckdb_create_log_storage destroyLogStorage \storage -> do- when (storage == nullPtr) $ throwRegistrationError "allocate log storage"- withCallbackResources- (\allocate -> allocate (mkWriteLogEntryCallback (logStorageHandler callback)))- (c_duckdb_log_storage_set_extra_data storage)- \writeCb -> do- TextForeign.withCString name $ c_duckdb_log_storage_set_name storage- c_duckdb_log_storage_set_write_log_entry storage writeCb- withDatabaseHandle conn \db -> mask_ do- rc <- c_duckdb_register_log_storage db storage- -- DuckDB consumes extra data on duplicate-name failure.- -- Clear the wrapper to prevent a second destruction.- c_duckdb_log_storage_set_extra_data storage nullPtr nullFunPtr- when (rc /= DuckDBSuccess) $ throwRegistrationError "register log storage"--logStorageHandler ::- (LogEntry -> IO ()) ->- Ptr () ->- Ptr DuckDBTimestamp ->- CString ->- CString ->- CString ->- IO ()-logStorageHandler callback _ timestampPtr levelPtr logTypePtr messagePtr =- ignoreCallbackExceptions do- entry <- do- logEntryTimestamp <- readTimestamp timestampPtr- logEntryLevel <- readCStringText levelPtr- logEntryType <- readCStringText logTypePtr- logEntryMessage <- readCStringText messagePtr- pure LogEntry{logEntryTimestamp, logEntryLevel, logEntryType, logEntryMessage}- callback entry--readTimestamp :: Ptr DuckDBTimestamp -> IO (Maybe UTCTime)-readTimestamp ptr- | ptr == nullPtr = pure Nothing- | otherwise = do- DuckDBTimestamp micros <- peek ptr- pure (Just (posixSecondsToUTCTime (fromRational (toInteger micros % 1000000))))--readCStringText :: CString -> IO Text-readCStringText ptr- | ptr == nullPtr = pure Text.empty- | otherwise = peekUtf8CString ptr--destroyLogStorage :: DuckDBLogStorage -> IO ()-destroyLogStorage storage =- alloca \ptr -> poke ptr storage >> c_duckdb_destroy_log_storage ptr--foreign import ccall "wrapper"- mkWriteLogEntryCallback ::- (Ptr () -> Ptr DuckDBTimestamp -> CString -> CString -> CString -> IO ()) ->- IO DuckDBLoggerWriteLogEntryFun+import Database.DuckDB.Simple.Logging.Internal (+ LogEntry (..),+ LogLevel (..),+ LogScope (..),+ logConnectionMessage,+ logInstanceMessage,+ registerInstanceLogSink,+ registerLogSink,+ registerLogStorage,+ )
+ src/Database/DuckDB/Simple/Logging/Internal.hs view
@@ -0,0 +1,178 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE StrictData #-}++-- |+-- Module : Database.DuckDB.Simple.Logging.Internal+-- Description : Share the log sink registration for connections and extensions.+--+-- The cabal file lists this hidden module in @other-modules@.+-- 'registerLogSinkOn' takes a registration target.+-- "Database.DuckDB.Simple.Logging" exports the connection and instance forms.+-- "Database.DuckDB.Simple.Extension" exports the extension form.+module Database.DuckDB.Simple.Logging.Internal (+ LogEntry (..),+ LogLevel (..),+ LogScope (..),+ registerLogStorage,+ registerLogSink,+ registerInstanceLogSink,+ registerLogSinkOn,+ logConnectionMessage,+ logInstanceMessage,+) where++import Control.Exception (bracket)+import Control.Monad (when)+import Data.Ratio ((%))+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Time.Clock (UTCTime)+import Data.Time.Clock.POSIX (posixSecondsToUTCTime)+import Data.Word (Word64)+import Database.DuckDB.Simple.Callback (+ RegistrationTarget (..),+ attachOpaque,+ install,+ installAll,+ requireWorkerCallbackRuntime,+ runCallback,+ withCallbackResourcesFinally,+ )+import Database.DuckDB.Simple.Internal (Connection, Instance, withConnectionHandle, withDatabaseHandle, withInstanceHandle)+import Database.DuckDB.Simple.Native+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.C.Types (CBool)+import Foreign.Marshal.Alloc (alloca)+import Foreign.Ptr (Ptr)+import Foreign.Storable (peek)++-- | A copied log event. Native context identifiers are absent outside their scope.+data LogEntry = LogEntry+ { logEntryTimestamp :: Maybe UTCTime+ , logEntryLevel :: Text+ , logEntryType :: Text+ , logEntryMessage :: Text+ , logEntrySeverity :: LogLevel+ , logEntryScope :: LogScope+ , logEntryConnectionId :: Maybe Word64+ , logEntryTransactionId :: Maybe Word64+ , logEntryQueryId :: Maybe Word64+ , logEntryThreadId :: Maybe Word64+ }+ deriving (Eq, Show)++-- | Register a log sink without a flush action.+-- Select the sink through DuckDB's logging configuration to receive entries.+registerLogStorage :: Connection -> Text -> (LogEntry -> IO ()) -> IO ()+registerLogStorage conn name write = registerLogSink conn name write (pure ())++-- | Register an instance log sink through a live connection. Log sinks require+-- the threaded GHC runtime (@-threaded@), including with one query thread.+-- Callbacks can run concurrently on native threads. They must not log or change+-- logging settings. Native logging discards callback errors after containment.+-- The flush action also runs when DuckDB releases the sink. If the registration+-- fails after the name check, the flush action runs one time.+registerLogSink :: Connection -> Text -> (LogEntry -> IO ()) -> IO () -> IO ()+registerLogSink conn name write flush = withDatabaseHandle conn \instance_ ->+ withInstanceFactory instance_ \factory ->+ registerSink+ factory+ (checked . c_duckdb_v2_instance_register_log_sink instance_)+ name+ write+ flush++-- | Register a log sink on an explicitly managed instance. If the registration+-- fails after the name check, the flush action runs one time.+registerInstanceLogSink :: Instance -> Text -> (LogEntry -> IO ()) -> IO () -> IO ()+registerInstanceLogSink instance_ name write flush = withInstanceHandle instance_ \native ->+ withInstanceFactory native \factory ->+ registerSink+ factory+ (checked . c_duckdb_v2_instance_register_log_sink native)+ name+ write+ flush++-- | Register a log sink on a connection or a loading extension.+registerLogSinkOn :: RegistrationTarget -> Text -> (LogEntry -> IO ()) -> IO () -> IO ()+registerLogSinkOn target name write flush = case target of+ ConnectionTarget conn -> registerLogSink conn name write flush+ ExtensionTarget extension context -> withContextFactory context \factory ->+ registerSink+ factory+ (checked . c_duckdb_v2_extension_register_log_sink extension)+ name+ write+ flush++-- | Transfer callback ownership to the native sink and release its builder.+registerSink+ :: DuckDBV2FactoryHandle -> (DuckDBV2LogSinkHandle -> IO ()) -> Text -> (LogEntry -> IO ()) -> IO () -> IO ()+registerSink factory register name write flush = do+ when (Text.null name || Text.any (== '\0') name) $ throwInvalidArgument "log sink name must be nonempty and must not contain NUL"+ bracket (output (c_duckdb_v2_log_sink_create factory)) (destroy c_duckdb_v2_log_sink_destroy) \sink ->+ withCallbackResourcesFinally+ (requireWorkerCallbackRuntime "log sink")+ flush+ ( \allocate ->+ installAll+ [ install+ allocate+ DuckDBV2LogSinkWriteCallbackFn+ ( DuckDBV2LogSinkWriteCallbackFn_Aux \info _ timestamp level scope (ConstPtr kind) (ConstPtr message) err -> runCallback err do+ category <- peek kind >>= readText+ body <- peek message >>= readText+ connection <- optionalId (c_duckdb_v2_log_sink_write_info_get_connection_id info)+ transaction <- optionalId (c_duckdb_v2_log_sink_write_info_get_transaction_id info)+ query <- optionalId (c_duckdb_v2_log_sink_write_info_get_query_id info)+ thread <- optionalId (c_duckdb_v2_log_sink_write_info_get_thread_id info)+ severity <- logLevelFromNative level+ origin <- logScopeFromNative scope+ let time = posixSecondsToUTCTime (fromRational (toInteger timestamp % 1000000))+ write (LogEntry (Just time) (levelName severity) category body severity origin connection transaction query thread)+ )+ (c_duckdb_v2_log_sink_set_write_callback sink)+ , install+ allocate+ DuckDBV2LogSinkFlushCallbackFn+ (DuckDBV2LogSinkFlushCallbackFn_Aux \_ _ err -> runCallback err flush)+ (c_duckdb_v2_log_sink_set_flush_callback sink)+ ]+ )+ (attachOpaque (c_duckdb_v2_log_sink_set_user_data sink))+ \installCallbacks -> do+ withIdentifier name $ checked . c_duckdb_v2_log_sink_set_name sink+ installCallbacks+ register sink++-- | Copy an optional identifier from a borrowed log-entry context.+optionalId :: (Ptr DuckDBIdx -> Ptr CBool -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO (Maybe Word64)+optionalId getter = alloca \present -> do+ value <- output (\slot -> getter slot present)+ exists <- peek present+ pure (if exists == 0 then Nothing else Just (fromIntegral value))++-- | Retain the textual severity names of the log-storage interface.+levelName :: LogLevel -> Text+levelName = \case+ LogLevelTrace -> "TRACE"+ LogLevelDebug -> "DEBUG"+ LogLevelInfo -> "INFO"+ LogLevelWarning -> "WARNING"+ LogLevelError -> "ERROR"+ LogLevelFatal -> "FATAL"++-- | Emit a connection-scoped message through its current logging configuration.+logConnectionMessage :: Connection -> LogLevel -> Text -> Text -> IO ()+logConnectionMessage conn level category message = withConnectionHandle conn \native ->+ withText category \kind -> withText message \body ->+ checked (c_duckdb_v2_connection_log native (logLevelToNative level) kind body)++-- | Emit a database-scoped message on an explicit instance.+logInstanceMessage :: Instance -> LogLevel -> Text -> Text -> IO ()+logInstanceMessage instance_ level category message = withInstanceHandle instance_ \native ->+ withText category \kind -> withText message \body ->+ checked (c_duckdb_v2_instance_log native (logLevelToNative level) kind body)
src/Database/DuckDB/Simple/LogicalRep.hs view
@@ -1,252 +1,315 @@-{-# LANGUAGE BlockArguments #-} {-# LANGUAGE LambdaCase #-} {-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-} -{- |-Module : Database.DuckDB.Simple.LogicalRep-Description : Structured logical-type and value representations for DuckDB.--}+-- |+-- Module : Database.DuckDB.Simple.LogicalRep+-- Description : Structured logical-type and value representations for DuckDB.+--+-- "Database.DuckDB.Simple.V2.LogicalRep" converts these types to and from native+-- logical types. module Database.DuckDB.Simple.LogicalRep (- -- * Structured value helpers- StructField (..),- StructValue (..),- UnionMemberType (..),- UnionValue (..),- LogicalTypeRep (..),- structValueTypeRep,- unionValueTypeRep,- logicalTypeToRep,- logicalTypeFromRep,- logicalTypeFromRepWith,- destroyLogicalType,+ -- * Structured value helpers+ StructField (..),+ StructValue (..),+ TupleValue (..),+ UnionMemberType (..),+ UnionValue (..),+ LogicalTypeRep (..),+ LogicalTypeId (..),+ TypeDeclaration (..),+ withoutAliases,+ aliasBase,+ listElementType,+ mapEntryTypes,+ logicalTypeSQL,+ structValueTypeRep,+ tupleValueTypeRep,+ unionValueTypeRep, ) where -import Control.Exception (bracket, throwIO)-import Control.Monad (forM, when)-import Data.Array (Array, elems, listArray)-import qualified Data.ByteString as BS+import Data.Array (Array, elems)+import Data.Char (isAsciiLower, isAsciiUpper, isDigit) import Data.Map.Strict (Map)+import Data.Set (Set)+import qualified Data.Set as Set import Data.Text (Text) import qualified Data.Text as Text-import qualified Data.Text.Encoding as TextEncoding import Data.Word (Word16, Word64, Word8)-import Database.DuckDB.FFI-import Foreign.C.String (CString)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Marshal.Array (withArray)-import Foreign.Marshal.Utils (withMany)-import Foreign.Ptr (castPtr, nullPtr)-import Foreign.Storable (poke)+import Database.DuckDB.Simple.Native (LogicalTypeId (..), logicalTypeIdName) --- | A Haskell description of a DuckDB logical type tree.+-- | A Haskell description of a DuckDB logical type tree. See+-- Note [Type aliases] for 'LogicalTypeAlias'. data LogicalTypeRep- = LogicalTypeScalar DuckDBType- | LogicalTypeDecimal !Word8 !Word8- | LogicalTypeGeometry !(Maybe Text)- | LogicalTypeList LogicalTypeRep- | LogicalTypeArray LogicalTypeRep !Word64- | LogicalTypeMap LogicalTypeRep LogicalTypeRep- | LogicalTypeStruct !(Array Int (StructField LogicalTypeRep))- | LogicalTypeUnion !(Array Int UnionMemberType)- | LogicalTypeEnum !(Array Int Text)- deriving (Eq, Show)+ = LogicalTypeScalar LogicalTypeId+ | LogicalTypeDecimal !Word8 !Word8+ | LogicalTypeCollatedVarchar !Text+ | LogicalTypeGeometry !(Maybe Text)+ | LogicalTypeList LogicalTypeRep+ | LogicalTypeArray LogicalTypeRep !Word64+ | LogicalTypeMap LogicalTypeRep LogicalTypeRep+ | LogicalTypeStruct !(Array Int (StructField LogicalTypeRep))+ | LogicalTypeTuple !(Array Int LogicalTypeRep)+ | LogicalTypeUnion !(Array Int UnionMemberType)+ | LogicalTypeEnum !(Array Int Text)+ | LogicalTypeAlias !Text LogicalTypeRep+ deriving (Eq, Show) +{- Note [Type aliases]+~~~~~~~~~~~~~~~~~~~~~~+DuckDB can keep an alias on a type. The alias uses the value representation of+its base type, and the type text of DuckDB gives the alias. In DuckDB 2.0.0 these+types have an alias:++- JSON. It is VARCHAR with the alias JSON, also as a child type, for example+ the element type of JSON[].+- A type from 'Database.DuckDB.Simple.Registration.registerCustomType'.+- A type from duckdb_v2_factory_create_type_with_alias.++A column of a type from SQL @CREATE TYPE name AS base@ has the base type and no+alias. A column of an ENUM type from @CREATE TYPE name AS ENUM (...)@ also has+no name.++LogicalTypeAlias holds the alias and the base type. logicalTypeToRep in+"Database.DuckDB.Simple.V2.LogicalRep" makes it at each depth. The V2 API has+no function that gives only the alias. duckdb_v2_logical_type_get_name gives+the alias of a type, or the name of the type id for a type with no alias. A+name that is not the name of the type id (logicalTypeIdName in+"Database.DuckDB.Simple.Native") is an alias. When the names are equal,+logicalTypeToRep makes a temporary copy with that alias and compares the two+native types. DuckDB includes the alias in type equality, so equal types mean+that the original type has the alias. This also distinguishes INTEGER from+INTEGER with the alias INTEGER, whose type texts are equal.++logicalTypeToRep borrows a factory for this comparison. The factory can be+from any instance. The comparison does not parse a type, run SQL, or call user+code. The temporary type is destroyed before the function returns.+logicalTypeFromRep writes the alias with+duckdb_v2_factory_create_type_with_alias.++The base type decides value conversion and vector layout. These operations use+logicalTypeToBaseRep, which reads no type name. Collection binding also compares+native type identity to keep aliases during element type inference. A temporal+promotion reads full type trees to keep aliases on unchanged leaves. See+Note [Element types of a list] in "Database.DuckDB.Simple.ToField".+These places keep the alias:++- A 'Database.DuckDB.Simple.Catalog.ColumnDescription'.+- The type fields of a STRUCT, TUPLE, or UNION value that a reader gives. The+ reader of a vector takes its tree from logicalTypeToRep, and it decides with+ aliasBase. A value that the library reads and binds again keeps its type when+ the native default cast supports its aliases, for example STRUCT(k JSON).++Alias metadata does not imply native cast support. DuckDB 2.0.0 cannot cast a+TIME value to an alias of TIME through the default cast. The library writes+TIME alias payloads through public vector storage. See Note [TIME alias values]+in "Database.DuckDB.Simple.ToField". LogicalTypeRep stores the alias+name and public type parameters. It does not store other C++ extension metadata.++Use withoutAliases to compare base types.+-}++-- | Replace each alias with its base type, at each depth. See+-- Note [Type aliases].+withoutAliases :: LogicalTypeRep -> LogicalTypeRep+withoutAliases = \case+ LogicalTypeAlias _ base -> withoutAliases base+ LogicalTypeList child -> LogicalTypeList (withoutAliases child)+ LogicalTypeArray child size -> LogicalTypeArray (withoutAliases child) size+ LogicalTypeMap key value -> LogicalTypeMap (withoutAliases key) (withoutAliases value)+ LogicalTypeStruct fields -> LogicalTypeStruct (fmap (\field -> field{structFieldValue = withoutAliases (structFieldValue field)}) fields)+ LogicalTypeTuple fields -> LogicalTypeTuple (fmap withoutAliases fields)+ LogicalTypeUnion members -> LogicalTypeUnion (fmap (\member -> member{unionMemberType = withoutAliases (unionMemberType member)}) members)+ rep -> rep++-- | Remove the aliases at the top of a type. The child types keep their+-- aliases. See Note [Type aliases].+aliasBase :: LogicalTypeRep -> LogicalTypeRep+aliasBase (LogicalTypeAlias _ base) = aliasBase base+aliasBase rep = rep++-- | The type of an element of a LIST or an ARRAY, from the type of its parent.+-- The function looks through an alias. Another parent type gives VARIANT: the+-- type of each stored value decides. See Note [Column types] in+-- "Database.DuckDB.Simple.FromField".+listElementType :: LogicalTypeRep -> LogicalTypeRep+listElementType parent = case aliasBase parent of+ LogicalTypeList element -> element+ LogicalTypeArray element _ -> element+ _ -> LogicalTypeScalar LogicalTypeIdVariant++-- | The key type and the value type of a MAP, from the type of its parent, as+-- 'listElementType' gives an element type.+mapEntryTypes :: LogicalTypeRep -> (LogicalTypeRep, LogicalTypeRep)+mapEntryTypes parent = case aliasBase parent of+ LogicalTypeMap key value -> (key, value)+ _ -> (LogicalTypeScalar LogicalTypeIdVariant, LogicalTypeScalar LogicalTypeIdVariant)++{- Note [Type texts]+~~~~~~~~~~~~~~~~~~~~+logicalTypeSQL gives the text of a type that duckdb_v2_logical_type_to_text of+DuckDB 2.0.0 gives, for example DECIMAL(18,3), BIGINT[], INTEGER[3],+MAP(VARCHAR, BIGINT), ENUM('sad', 'happy'), and GEOMETRY('OGC:CRS84'). An+alias gives its name, for example JSON. The function is pure, so an error of a+FromField instance can name the type with no native handle, and the text costs+nothing until an error renders.++These rules follow LogicalType::ToString of DuckDB 2.0.0:++- A scalar type has the name of its type id (logicalTypeIdName in+ "Database.DuckDB.Simple.Native"), and SQL NULL is "NULL" in double quotes. A+ VARCHAR with a collation is VARCHAR: the text has no collation.+- A label of an ENUM and a CRS are in single quotes, with each single quote+ doubled.+- A field name of a STRUCT and a member name of a UNION are in double quotes,+ with each double quote doubled, when the name needs quotes. A name needs+ quotes when it has a character that is not an ASCII letter, a digit, or an+ underscore, when it starts with a digit or an underscore followed by a digit,+ or when it is a keyword of DuckDB in any case (SQLIdentifier::ToString).+ The keywords are the 502 names of duckdb_keywords() of DuckDB 2.0.0.+- An empty STRUCT or TUPLE has only its type name, with no parentheses.++A test case compares the text with duckdb_v2_logical_type_to_text for one type of+each kind and for a STRUCT with each keyword as a field name.+-}++-- | The SQL text of a type, as DuckDB gives it, for example DECIMAL(18,3) or+-- JSON. See Note [Type texts].+logicalTypeSQL :: LogicalTypeRep -> Text+logicalTypeSQL = \case+ LogicalTypeScalar LogicalTypeIdSQLNull -> "\"NULL\""+ LogicalTypeScalar ident -> logicalTypeIdName ident+ LogicalTypeDecimal width scale -> "DECIMAL(" <> number width <> "," <> number scale <> ")"+ LogicalTypeCollatedVarchar _ -> "VARCHAR"+ LogicalTypeGeometry Nothing -> "GEOMETRY"+ LogicalTypeGeometry (Just crs) -> "GEOMETRY(" <> quoted '\'' crs <> ")"+ LogicalTypeList element -> logicalTypeSQL element <> "[]"+ LogicalTypeArray element size -> logicalTypeSQL element <> "[" <> number size <> "]"+ LogicalTypeMap key value -> "MAP(" <> logicalTypeSQL key <> ", " <> logicalTypeSQL value <> ")"+ LogicalTypeStruct fields+ | null (elems fields) -> "STRUCT"+ | otherwise -> "STRUCT(" <> commas [identifier name <> " " <> logicalTypeSQL rep | StructField name rep <- elems fields] <> ")"+ LogicalTypeTuple fields+ | null (elems fields) -> "TUPLE"+ | otherwise -> "TUPLE(" <> commas (map logicalTypeSQL (elems fields)) <> ")"+ LogicalTypeUnion members -> "UNION(" <> commas [identifier name <> " " <> logicalTypeSQL rep | UnionMemberType name rep <- elems members] <> ")"+ LogicalTypeEnum labels -> "ENUM(" <> commas (map (quoted '\'') (elems labels)) <> ")"+ LogicalTypeAlias name _ -> name+ where+ number :: (Show n) => n -> Text+ number = Text.pack . show+ commas = Text.intercalate ", "+ quoted mark text = Text.singleton mark <> Text.replace (Text.singleton mark) (Text.pack [mark, mark]) text <> Text.singleton mark+ identifier name+ | needsQuotes name = quoted '"' name+ | otherwise = name+ needsQuotes name =+ Text.null name+ || (Text.isPrefixOf "_" name && maybe False (isDigit . fst) (Text.uncons (Text.drop 1 name)))+ || not (and (zipWith plain [0 :: Int ..] (Text.unpack name)))+ || Set.member (Text.toLower name) duckDBKeywords+ plain index char = isAsciiLower char || isAsciiUpper char || char == '_' || (index > 0 && isDigit char)++-- | The keywords of DuckDB 2.0.0, from duckdb_keywords(). A field name that is+-- a keyword needs quotes. See Note [Type texts].+duckDBKeywords :: Set Text+duckDBKeywords =+ Set.fromList . Text.words $+ "abort absolute access action add admin after aggregate all also alter always analyse analyze and anti any \+ \ approx array as asc asof assertion assignment asymmetric at attach attribute authorization backward before \+ \ begin between bigint binary bit boolean both by cache call called cascade cascaded case cast catalog centuries \+ \ century chain char character characteristics check checkpoint class close cluster coalesce collate collation \+ \ column columns comment comments commit committed compression concurrently configuration conflict connect \+ \ connection constraint constraints content continue conversion copy cost create cross csv cube current cursor \+ \ cycle data database day days deallocate dec decade decades decimal declare default defaults deferrable \+ \ deferred definer delete delimiter delimiters depends desc describe destroy detach dictionary disable discard \+ \ disconnect distance distinct do document domain double drop each else enable encoding encrypted end enum error \+ \ escape event exact except exclude excluding exclusive execute exists explain export export_state extension \+ \ extensions external extract false family fetch filter first float following for force foreign forward freeze \+ \ from full function functions generated glob global grant granted group grouping grouping_id groups handler \+ \ having header hold hour hours identity if ignore ilike immediate immutable implicit import in include \+ \ including increment index indexes inherit inherits initially inline inner inout input insensitive insert \+ \ install instead int integer intersect interval into invoker is isnull isolation join json key label lambda \+ \ language large last lateral leading leakproof left level like limit listen load local location lock locked \+ \ logged macro map mapping match matched materialized maxvalue merge method microsecond microseconds millennia \+ \ millennium millisecond milliseconds minute minutes minvalue mode month months move name names national natural \+ \ nchar nearest new next no none not nothing notify notnull nowait null nullif nulls numeric object of off \+ \ offset oids old on only operator option options or order ordinality others out outer over overlaps overlay \+ \ overriding owned owner parallel parser partial partition partitioned passing password percent persistent pivot \+ \ pivot_longer pivot_wider placing plans policy position positional pragma preceding precision prepare prepared \+ \ preserve primary prior privileges procedural procedure program publication qualify quarter quarters quote \+ \ range read real reassign recheck recursive ref references referencing refresh register reindex relative \+ \ release rename repeatable replace replica reset resource resources respect restart restrict returning returns \+ \ revoke right role rollback rollup row rows rule sample savepoint schema schemas scope scroll search second \+ \ seconds secret secure security select semi sequence sequences serializable server session set setof sets share \+ \ show similar similarity simple skip smallint snapshot some sorted source sql stable standalone start statement \+ \ statistics stdin stdout storage stored strict strip struct subscription substring summarize symmetric sysid \+ \ system table tables tablesample tablespace target temp template temporary text then ties time timestamp to \+ \ trailing transaction transform treat trigger trim true truncate trusted try_cast tuple type types unbounded \+ \ uncommitted unencrypted union unique unknown unlisten unlogged unpack unpivot until update use user using \+ \ vacuum valid validate validator value values varchar variable variadic varying verbose version view views \+ \ virtual volatile week weeks when where whitespace window with within without work wrapper write xml \+ \ xmlattributes xmlconcat xmlelement xmlexists xmlforest xmlnamespaces xmlparse xmlpi xmlroot xmlserialize \+ \ xmltable year years yes zone"++-- | A structured type or a SQL type expression. A 'TypeExpression' with the+-- name of a type from 'Database.DuckDB.Simple.Registration.registerCustomType'+-- keeps that name, so a cast to it is a cast to the custom type. A name from+-- SQL @CREATE TYPE name AS base@ is only another name for the base type, so a+-- cast to it is a cast to the base type. In a test with DuckDB 2.0.0, a+-- registered cast from BIGINT to a name for BIGINT did not run. The header of+-- DuckDB 2.0.0 says that a catalog type name is not kept, but DuckDB 2.0.0+-- keeps the name of a type from+-- 'Database.DuckDB.Simple.Registration.registerCustomType'.+data TypeDeclaration = StructuredType LogicalTypeRep | TypeExpression Text+ deriving (Eq, Show)+ -- | A named field within a STRUCT-like value or type. data StructField a = StructField- { structFieldName :: !Text- , structFieldValue :: !a- }- deriving (Eq, Show)+ { structFieldName :: !Text+ , structFieldValue :: !a+ }+ deriving (Eq, Show) -- | A fully materialized STRUCT value together with its type metadata. data StructValue a = StructValue- { structValueFields :: !(Array Int (StructField a))- , structValueTypes :: !(Array Int (StructField LogicalTypeRep))- , structValueIndex :: !(Map Text Int)- }- deriving (Eq, Show)+ { structValueFields :: !(Array Int (StructField a))+ , structValueTypes :: !(Array Int (StructField LogicalTypeRep))+ , structValueIndex :: !(Map Text Int)+ }+ deriving (Eq, Show) +-- | A positional TUPLE value with its field types. Array bounds are independent.+data TupleValue a = TupleValue+ { tupleValueFields :: !(Array Int a)+ , tupleValueTypes :: !(Array Int LogicalTypeRep)+ }+ deriving (Eq, Show)+ -- | A named member within a UNION type. data UnionMemberType = UnionMemberType- { unionMemberName :: !Text- , unionMemberType :: !LogicalTypeRep- }- deriving (Eq, Show)+ { unionMemberName :: !Text+ , unionMemberType :: !LogicalTypeRep+ }+ deriving (Eq, Show) -- | A fully materialized UNION value together with its member metadata. data UnionValue a = UnionValue- { unionValueIndex :: !Word16- , unionValueLabel :: !Text- , unionValuePayload :: !a- , unionValueMembers :: !(Array Int UnionMemberType)- }- deriving (Eq, Show)+ { unionValueIndex :: !Word16+ , unionValueLabel :: !Text+ , unionValuePayload :: !a+ , unionValueMembers :: !(Array Int UnionMemberType)+ }+ deriving (Eq, Show) -- | Recover the logical STRUCT type corresponding to a @StructValue@. structValueTypeRep :: StructValue a -> LogicalTypeRep structValueTypeRep StructValue{structValueTypes} = LogicalTypeStruct structValueTypes +-- | Recover the logical TUPLE type of a positional value.+tupleValueTypeRep :: TupleValue a -> LogicalTypeRep+tupleValueTypeRep TupleValue{tupleValueTypes} = LogicalTypeTuple tupleValueTypes+ -- | Recover the logical UNION type corresponding to a @UnionValue@. unionValueTypeRep :: UnionValue a -> LogicalTypeRep unionValueTypeRep UnionValue{unionValueMembers} = LogicalTypeUnion unionValueMembers---- | Destroy a logical type handle obtained from DuckDB.-destroyLogicalType :: DuckDBLogicalType -> IO ()-destroyLogicalType logical =- alloca \ptr -> do- poke ptr logical- c_duckdb_destroy_logical_type ptr---- | Convert a DuckDB logical type handle into the pure @LogicalTypeRep@ tree.-logicalTypeToRep :: DuckDBLogicalType -> IO LogicalTypeRep-logicalTypeToRep logical = do- dtype <- c_duckdb_get_type_id logical- case dtype of- DuckDBTypeGeometry ->- bracket (c_duckdb_geometry_type_get_crs logical) (c_duckdb_free . castPtr) \ptr ->- LogicalTypeGeometry <$> if ptr == nullPtr then pure Nothing else Just . TextEncoding.decodeUtf8 <$> BS.packCString ptr- DuckDBTypeStruct -> do- childCountRaw <- c_duckdb_struct_type_child_count logical- childCount <- word64ToInt (Text.pack "struct child count") childCountRaw- fields <-- forM [0 .. childCount - 1] \idx -> do- name <- bracket (c_duckdb_struct_type_child_name logical (fromIntegral idx)) (c_duckdb_free . castPtr) $ \ptr -> do- when (ptr == nullPtr) $- throwIO (userError "duckdb-simple: struct child name is null")- TextEncoding.decodeUtf8 <$> BS.packCString ptr- childRep <-- bracket (c_duckdb_struct_type_child_type logical (fromIntegral idx)) destroyLogicalType logicalTypeToRep- pure StructField{structFieldName = name, structFieldValue = childRep}- pure $- LogicalTypeStruct- (listArray (0, childCount - 1) fields)- DuckDBTypeUnion -> do- memberCountRaw <- c_duckdb_union_type_member_count logical- memberCount <- word64ToInt (Text.pack "union member count") memberCountRaw- members <-- forM [0 .. memberCount - 1] \idx -> do- name <- bracket (c_duckdb_union_type_member_name logical (fromIntegral idx)) (c_duckdb_free . castPtr) $ \ptr -> do- when (ptr == nullPtr) $- throwIO (userError "duckdb-simple: union member name is null")- TextEncoding.decodeUtf8 <$> BS.packCString ptr- memberRep <-- bracket (c_duckdb_union_type_member_type logical (fromIntegral idx)) destroyLogicalType logicalTypeToRep- pure UnionMemberType{unionMemberName = name, unionMemberType = memberRep}- pure $- LogicalTypeUnion- (listArray (0, memberCount - 1) members)- DuckDBTypeList -> do- childRep <- bracket (c_duckdb_list_type_child_type logical) destroyLogicalType logicalTypeToRep- pure (LogicalTypeList childRep)- DuckDBTypeArray -> do- childRep <- bracket (c_duckdb_array_type_child_type logical) destroyLogicalType logicalTypeToRep- size <- c_duckdb_array_type_array_size logical- pure (LogicalTypeArray childRep size)- DuckDBTypeMap -> do- keyRep <- bracket (c_duckdb_map_type_key_type logical) destroyLogicalType logicalTypeToRep- valueRep <- bracket (c_duckdb_map_type_value_type logical) destroyLogicalType logicalTypeToRep- pure (LogicalTypeMap keyRep valueRep)- DuckDBTypeDecimal -> do- width <- c_duckdb_decimal_width logical- scale <- c_duckdb_decimal_scale logical- pure (LogicalTypeDecimal width scale)- DuckDBTypeEnum -> do- dictSize <- c_duckdb_enum_dictionary_size logical- let count = fromIntegral dictSize :: Int- entries <-- forM [0 .. count - 1] \idx -> do- entry <- bracket (c_duckdb_enum_dictionary_value logical (fromIntegral idx)) (c_duckdb_free . castPtr) $ \ptr -> do- when (ptr == nullPtr) $- throwIO (userError "duckdb-simple: enum dictionary value is null")- TextEncoding.decodeUtf8 <$> BS.packCString ptr- pure entry- pure $- LogicalTypeEnum- (listArray (0, count - 1) entries)- _ ->- pure (LogicalTypeScalar dtype)--{- | Materialize a DuckDB logical type handle from a @LogicalTypeRep@ tree.-The C API cannot create a usable VARIANT type or a GEOMETRY type with a CRS.-This function raises an error for VARIANT. It creates GEOMETRY without a CRS-and ignores the CRS of 'LogicalTypeGeometry'. Parameter binding uses the-types that a connection reads the first time a parameter needs one.--}-logicalTypeFromRep :: LogicalTypeRep -> IO DuckDBLogicalType-logicalTypeFromRep = logicalTypeFromRepWith \case- -- TODO: improve this when this becomes available in the C API.- -- See https://github.com/Tritlo/duckdb-haskell/issues/26 and- -- https://github.com/Tritlo/duckdb-haskell/issues/27.- LogicalTypeScalar DuckDBTypeVariant ->- throwIO (userError "duckdb-simple: a VARIANT type needs the type cache of a connection; bind the value as a parameter or cast a plain value with ?::VARIANT")- _ -> c_duckdb_create_logical_type DuckDBTypeGeometry--{- | Materialize a type tree. The function argument creates the leaves that the-C API cannot create: VARIANT, and GEOMETRY with a CRS. The caller must-destroy the result.--}-logicalTypeFromRepWith :: (LogicalTypeRep -> IO DuckDBLogicalType) -> LogicalTypeRep -> IO DuckDBLogicalType-logicalTypeFromRepWith resolve rep = do- logical <- create rep- when (logical == nullPtr) $- throwIO (userError "duckdb-simple: DuckDB logical type construction failed")- pure logical- where- create = \case- leaf@(LogicalTypeScalar DuckDBTypeVariant) -> resolve leaf- LogicalTypeScalar dtype -> c_duckdb_create_logical_type dtype- leaf@(LogicalTypeGeometry (Just _)) -> resolve leaf- LogicalTypeGeometry Nothing -> c_duckdb_create_logical_type DuckDBTypeGeometry- LogicalTypeDecimal width scale -> do- when (width < 1 || width > 38 || scale > width) $- throwIO (userError "duckdb-simple: invalid DECIMAL width or scale")- c_duckdb_create_decimal_type width scale- LogicalTypeList elemRep ->- bracket (logicalTypeFromRepWith resolve elemRep) destroyLogicalType c_duckdb_create_list_type- LogicalTypeArray elemRep size ->- bracket (logicalTypeFromRepWith resolve elemRep) destroyLogicalType $- flip c_duckdb_create_array_type size- LogicalTypeMap keyRep valueRep ->- bracket (logicalTypeFromRepWith resolve keyRep) destroyLogicalType \keyType ->- bracket (logicalTypeFromRepWith resolve valueRep) destroyLogicalType $- c_duckdb_create_map_type keyType- LogicalTypeStruct fieldArray -> do- let fields = elems fieldArray- withMany (\field -> bracket (logicalTypeFromRepWith resolve (structFieldValue field)) destroyLogicalType) fields \childTypes ->- withMany withTypeName (map structFieldName fields) \names ->- withArray names \nameArray ->- withArray childTypes \typeArray ->- c_duckdb_create_struct_type typeArray nameArray (fromIntegral (length fields))- LogicalTypeUnion memberArray -> do- let members = elems memberArray- withMany (\member -> bracket (logicalTypeFromRepWith resolve (unionMemberType member)) destroyLogicalType) members \memberTypes ->- withMany withTypeName (map unionMemberName members) \names ->- withArray names \nameArray ->- withArray memberTypes \typeArray ->- c_duckdb_create_union_type typeArray nameArray (fromIntegral (length members))- LogicalTypeEnum dictArray ->- withMany withTypeName (elems dictArray) \names ->- withArray names \nameArray ->- c_duckdb_create_enum_type nameArray (fromIntegral (length names))---- | Encode native type names as UTF-8 and reject embedded NUL.-withTypeName :: Text -> (CString -> IO a) -> IO a-withTypeName name action = do- when (Text.any (== '\0') name) $- throwIO (userError "duckdb-simple: logical type name contains NUL")- BS.useAsCString (TextEncoding.encodeUtf8 name) action--word64ToInt :: Text -> Word64 -> IO Int-word64ToInt label value =- let actual = toInteger value- limit = toInteger (maxBound :: Int)- in if actual <= limit- then pure (fromIntegral value)- else- throwIO- ( userError- ( "duckdb-simple: "- <> Text.unpack label- <> " exceeds Int range"- )- )
src/Database/DuckDB/Simple/Materialize.hs view
@@ -1,308 +1,749 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-} {-# LANGUAGE LambdaCase #-} {-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE TypeApplications #-} +-- |+-- Module : Database.DuckDB.Simple.Materialize+-- Description : Readers that copy native vectors and values to field values. module Database.DuckDB.Simple.Materialize (- prepareValueReader,- prepareVectorReader,+ VectorReader,+ FallbackReaders (..),+ rejectRows,+ prepareVariantReader,+ readCallbackVariant,+ readInstanceVariant,+ createHelperInstance,+ ReadFactory (..),+ callbackReadFactory,+ readFactoryHandle,+ prepareGeometryReader,+ materializeValue,+ prepareVectorReader,+ TypeTree,+ EnumLabels,+ prepareVectorReaderWith, ) where -import Control.Exception (throwIO)-import Control.Monad (forM, when)-import Data.Array (Array, elems, listArray)+import Control.Exception (bracket)+import Control.Monad (foldM, forM, when)+import Data.Array (Array, bounds, elems, listArray, rangeSize, (!))+import Data.IORef (IORef, modifyIORef', newIORef, readIORef, writeIORef) import Data.Int (Int16, Int32, Int64)+import qualified Data.IntMap.Strict as IntMap import qualified Data.Map.Strict as Map import Data.Text (Text) import qualified Data.Text as Text+import Data.Void (Void) import Data.Word (Word16, Word32, Word64, Word8)-import Database.DuckDB.FFI import Database.DuckDB.Simple.Element (- chunkDecodeBlob,- chunkIsRowValid,- decodeElement,- duckDBHugeIntToInteger,- vectorElementType,- withVectorType,+ chunkIsRowValid,+ decodeElement,+ duckDBHugeIntToInteger,+ vectorElementType,+ withVectorType, ) import Database.DuckDB.Simple.FromField (- DecimalValue (..),- FieldValue (..),- RawGeometry (..),+ DecimalValue (..),+ FieldValue (..),+ RawGeometry (..), ) import Database.DuckDB.Simple.LogicalRep (- LogicalTypeRep (..),- StructField (..),- StructValue (..),- UnionMemberType (..),- UnionValue (..),- logicalTypeToRep,+ LogicalTypeRep (..),+ StructField (..),+ StructValue (..),+ TupleValue (..),+ UnionMemberType (..),+ UnionValue (..),+ aliasBase,+ logicalTypeSQL, )-import Database.DuckDB.Simple.VariantCodec (decodeVariant, prepareVariantDecoder)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeToBaseRep, logicalTypeToRep)+import qualified Database.DuckDB.Simple.VariantCodec as VariantCodec+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Marshal.Alloc (alloca) import Foreign.Ptr (Ptr, castPtr, nullPtr) import Foreign.Storable (Storable (..), peekElemOff)+import GHC.Records (getField) --- | Read the type, data, and validity of a vector, and prepare its reader.-prepareVectorReader :: DuckDBVector -> IO (Int -> IO FieldValue)-prepareVectorReader vector = do- dtype <- vectorElementType vector- dataPtr <- c_duckdb_vector_get_data vector- validity <- c_duckdb_vector_get_validity vector- prepareValueReader dtype vector dataPtr validity+-- | Prepare a reader for the rows of a vector. The reader must not outlive the vector.+type VectorReader = DuckDBV2VectorHandle -> IO (Int -> IO FieldValue) +-- | Readers that need state outside the vector. Type metadata needs a factory,+-- VARIANT needs a helper connection, and GEOMETRY needs a factory for casts.+-- See Note [Type trees]. For VARIANT, see Note [VARIANT decoding] in+-- "Database.DuckDB.Simple.VariantCodec".+data FallbackReaders = FallbackReaders+ { typeReader :: DuckDBV2LogicalTypeHandle -> IO LogicalTypeRep+ , variantReader :: VectorReader+ , geometryReader :: VectorReader+ }++-- | A reader for a type that the caller cannot read. It reads each row when the+-- row is read, not when the vector is prepared. A NULL row gives 'FieldNull',+-- and a row that holds a value raises the error. Thus a value of a nested type+-- with no value of that type, such as an empty VARIANT list or a NULL STRUCT,+-- gives its value. The header commits to no view layout for VARIANT and+-- GEOMETRY, so the reader reads each row with @duckdb_v2_vector_get_value@,+-- whose row is logical. 'prepareVectorReaderWith' flattens the vector, so the+-- row that the reader gets is the logical row. See Note [Vector flatten]. The+-- reader creates and destroys one native value for each row.+rejectRows :: IO FieldValue -> VectorReader+rejectRows raise vector = pure \row ->+ bracket (output (c_duckdb_v2_vector_get_value vector (fromIntegral row))) destroyValue \value -> do+ isNull <- output (c_duckdb_v2_value_is_null value)+ if isNull /= 0 then pure FieldNull else raise++-- | Copy VARIANT rows before releasing the borrowed helper connection. The+-- state belongs to that connection. The decoder reads logical rows. See+-- Note [Vector flatten] and Note [VARIANT decoding] in+-- "Database.DuckDB.Simple.VariantCodec".+prepareVariantReader :: VariantCodec.VariantState -> VectorReader+prepareVariantReader =+ VariantCodec.prepareVariantDecoder+ (materializeValue noVariant . InstanceFactory)+ ( \factory vector ->+ prepareVectorReader (FallbackReaders (logicalTypeToRep factory) noVariant (prepareGeometryReader (InstanceFactory factory))) vector+ >>= ($ 0)+ )+ where+ noVariant = rejectRows (throwInternalError (Text.pack "expected a value without VARIANT"))++{- Note [VARIANT reads inside callbacks]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+The VARIANT decoder runs SQL. A callback cannot run it on the executing+connection. Even a new connection to the host instance can need a lock that+the caller holds, for example during SET GLOBAL. An extension has no host+instance handle at all.++The decoder uses an instance with only built-in types, casts and system+functions. A scalar or aggregate registration opens its instance on demand.+Each vector uses a separate connection on that instance. Other typed+callbacks, including extension COPY TO, open a private instance for each+VARIANT vector. The decoder copies the rows before it closes the connection.+See Note [Function argument readers] in "Database.DuckDB.Simple.Typed"+and Note [VARIANT decoding] in "Database.DuckDB.Simple.VariantCodec".+-}++-- | Decode a callback's VARIANT vector independently of the invoking database.+readCallbackVariant :: VectorReader+readCallbackVariant vector =+ bracket (output c_duckdb_v2_environment_create) (destroy c_duckdb_v2_environment_destroy) \env ->+ bracket (createHelperInstance env) (destroy c_duckdb_v2_instance_destroy) (`readInstanceVariant` vector)++-- | Copy a VARIANT vector through a temporary connection on the supplied instance.+readInstanceVariant :: DuckDBV2InstanceHandle -> VectorReader+readInstanceVariant instance_ vector =+ bracket (output (c_duckdb_v2_connection_create instance_)) (destroy c_duckdb_v2_connection_destroy) \helper ->+ withInstanceFactory instance_ \factory ->+ bracket (VariantCodec.newVariantState helper factory) VariantCodec.destroyVariantState (`prepareVariantReader` vector)++-- | Create an instance for synchronous metadata work, without background workers.+createHelperInstance :: DuckDBV2EnvironmentHandle -> IO DuckDBV2InstanceHandle+createHelperInstance env =+ withTextArray (map Text.pack ["threads", "async_threads"]) \names ->+ withTextArray (map Text.pack ["1", "0"]) \values ->+ output (c_duckdb_v2_instance_create_with_options env (ConstPtr names) (ConstPtr values) 2)++-- | The factory of a read, and its kind. The kind decides how+-- 'prepareGeometryReader' casts GEOMETRY to BLOB. See+-- Note [User casts and the casts of the library].+data ReadFactory+ = -- | The factory of an instance. Its casts are the built-in casts.+ InstanceFactory DuckDBV2FactoryHandle+ | -- | The factory of a callback context. Its casts include the casts that a+ -- user registers.+ ContextFactory DuckDBV2FactoryHandle++-- | The factory of a read in a callback: the factory of the instance from+-- 'Database.DuckDB.Simple.Callback.targetCastFactory', or the factory of the+-- context for an extension target, which has no instance.+callbackReadFactory :: Maybe DuckDBV2FactoryHandle -> DuckDBV2FactoryHandle -> ReadFactory+callbackReadFactory casts context = maybe (ContextFactory context) InstanceFactory casts++-- | The handle of a read factory.+readFactoryHandle :: ReadFactory -> DuckDBV2FactoryHandle+readFactoryHandle (InstanceFactory factory) = factory+readFactoryHandle (ContextFactory factory) = factory++-- | Copy one value through an owned vector. The caller retains the factory owner.+-- The first argument reads VARIANT. 'prepareGeometryReader' reads GEOMETRY with+-- the factory.+materializeValue :: VectorReader -> ReadFactory -> DuckDBV2ValueHandle -> IO FieldValue+materializeValue readVariant factory value =+ bracket (output (c_duckdb_v2_value_get_logical_type value)) (destroy c_duckdb_v2_logical_type_destroy) \logical ->+ withOwnedVector (readFactoryHandle factory) logical \vector -> do+ checked (c_duckdb_v2_vector_set_size vector 1)+ checked (c_duckdb_v2_vector_set_value vector 0 value)+ prepareVectorReader (FallbackReaders (logicalTypeToRep (readFactoryHandle factory)) readVariant (prepareGeometryReader factory)) vector+ >>= ($ 0)++{- Note [Type trees]+~~~~~~~~~~~~~~~~~~~~+A reader of a vector uses two kinds of type information.++- A layout parameter gives the layout of the bytes, for example the width of a+ DECIMAL, the size of an ARRAY, or the size of an ENUM dictionary. A decoder+ takes each layout parameter from the type of its own vector, because that+ type gives the layout of the bytes. The header of duckdb_v2_vector_get_view+ says that the data has the layout that the logical type of the vector+ implies.++- A description is the field names and field types of a STRUCT or TUPLE, or+ the members of a UNION. A decoder takes a description from a type tree: the+ LogicalTypeRep that an action of type TypeTree gives. A decoder runs the+ action only when it needs a description.++Thus a wrong tree gives an error or a wrong description, but it does not change+how bytes are read. checkChildCount raises an error when a tree gives another+child count than the vector has.++These are the sources of a tree.++- prepareVectorReader converts the type of the vector.+- A caller of prepareVectorReaderWith gives the action. A result converts the+ tree of a column from the vector of the first chunk that needs it, and keeps+ the tree for one execution (prepareChunkReaders).+- A nested decoder gives each child an action that takes the tree of the child+ from its own tree (elementTree, prepareStructDecoder, prepareTupleDecoder, and+ prepareUnionDecoder). The action of a child runs the action of the parent+ only when the child needs its tree.+- The MAP decoder gives its key reader and its value reader no tree, and the+ UNION decoder gives its tag reader no tree. Each of these readers converts+ the type of its own vector when it needs a description.++The readers do not use the result schema. The header of+duckdb_v2_result_get_schema calls the schema the authoritative source of the+column types, and says that vectors do not carry their own type. But the header+of duckdb_v2_vector_get_logical_type says that a vector has the type that was+set on creation, and the layout follows the type of the vector. So the vector+type is the source of each type.+-}++-- | An action that gives the type tree of one vector. A decoder runs it only+-- when it needs a description. See Note [Type trees].+type TypeTree = IO LogicalTypeRep++-- | Read the type, data, and validity of a vector, and prepare its reader. A+-- decoder that needs the type tree converts the type of the vector. See+-- Note [Type trees].+prepareVectorReader :: FallbackReaders -> DuckDBV2VectorHandle -> IO (Int -> IO FieldValue)+prepareVectorReader readers vector = do+ -- The cache of ENUM labels lives as long as this reader. See Note [ENUM labels].+ labels <- newIORef IntMap.empty+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ prepareVectorReaderWith readers (withVectorType vector (typeReader readers)) (Just labels) vector++-- | The labels of an ENUM type that readers read, by index. 'Nothing' means+-- that the type tree holds the labels. See Note [ENUM labels].+type EnumLabels = Maybe (IORef (IntMap.IntMap Text))++{- Note [Vector flatten]+~~~~~~~~~~~~~~~~~~~~~~~~+vector_flatten changes a vector to the FLAT representation in place. A reader+flattens each vector before it reads the view. A writer of a callback output+vector flattens the vector before it writes rows.++Read (prepareVectorReaderWith).++- The header of duckdb_v2_vector_get_view says: "Rejects VECTOR_TYPE_OTHER;+ call vector_flatten first." The flatten also changes a CONSTANT or DICTIONARY+ vector to FLAT.++- In DuckDB 2.0.0 the view of a FLAT vector has no selection+ (duckdb_v2_vector_get_view in src/main/capi/v2/capi_v2_vector.cpp). The header+ does not say this. Then the physical row is the logical row. The readers of+ VARIANT and GEOMETRY need this, because they read logical rows (rejectRows,+ prepareVariantReader, and prepareGeometryReader). The reader still applies+ the selection of a view that has one.++- The flatten changes the buffers of the vector, so it runs before the view is+ read.++- In DuckDB 2.0.0 the flatten is recursive: it also flattens the child of a+ LIST or MAP vector and the fields of a STRUCT vector (VectorListBuffer::Flatten+ and VectorStructBuffer::Flatten). The header does not say this. The MAP+ decoder (prepareMapDecoder) depends on this, because it takes the key and+ value vectors from the entry STRUCT and does not flatten that STRUCT itself.++Write. A writer of a callback output vector sets the size of the vector, and+then flattens the vector (writeRows and writeResults). A writer of an+owned vector from withOwnedVector does not flatten (materializeValue, and the+writers of one-row vectors in "Database.DuckDB.Simple.ToField").++- The writer sets the size before the flatten. In DuckDB 2.0.0,+ duckdb_v2_vector_flatten flattens the rows of the current size of the vector:+ it calls Vector::Flatten without a row count+ (src/main/capi/v2/capi_v2_vector.cpp). duckdb_v2_vector_set_size reserves+ memory only for a vector that is already FLAT. Thus after this order the+ vector is FLAT and has the new number of rows. The header gives no order.++- A new vector of a data chunk is FLAT. The header of+ duckdb_v2_data_chunk_create says: "Allocates one FLAT vector per element of+ the types array". Thus the writer of an owned vector needs no flatten. In+ DuckDB 2.0.0 the children of such a vector are FLAT too: the vector cache+ resets each child of a LIST, ARRAY, or STRUCT vector to FLAT+ (VectorCacheEntry::ResetFromCache in src/common/types/vector_cache.cpp). The+ header does not say this for the children. The writer of a UNION value writes+ the child vectors of a new vector (unionValueDuckValue in+ "Database.DuckDB.Simple.ToField").++- The set operations check the row against the size. The header of+ duckdb_v2_vector_set_value and duckdb_v2_vector_set_null says that each call+ requires a FLAT vector and checks the row against vector_get_size.++- writeResults reads the data pointer and the validity pointer after the+ flatten. The header of duckdb_v2_vector_get_data_mutable says that the storage+ shape of the vector must not change while the pointer is in use.++- prepareResultWriter does not change the size and does not flatten. It raises+ an error for a vector that is not FLAT.+-}++-- | Read the type, data, and validity of a vector, and prepare its reader. The+-- action gives the type tree of the vector. A nested decoder gives each child+-- an action that takes the tree of the child from this tree. The function+-- flattens the vector first. See Note [Vector flatten] and Note [Type trees].+prepareVectorReaderWith :: FallbackReaders -> TypeTree -> EnumLabels -> DuckDBV2VectorHandle -> IO (Int -> IO FieldValue)+prepareVectorReaderWith readers tree labels vector = do+ -- Flatten before the view is read. See Note [Vector flatten].+ checked (c_duckdb_v2_vector_flatten vector)+ dtype <- vectorElementType vector+ view <- output (c_duckdb_v2_vector_get_view vector)+ let ConstPtr dataPtr = getField @"data'" view+ ConstPtr validity = getField @"validity" view+ ConstPtr selection = getField @"sel" view+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ readValue <- prepareValueReader readers (aliasBase <$> tree) labels dtype vector dataPtr validity+ pure \row -> do+ when (row < 0 || fromIntegral row >= getField @"count" view) $+ throwInternalError (Text.pack "vector row index out of range")+ physical <- if selection == nullPtr then pure row else fromIntegral <$> peekElemOff selection row+ readValue physical++{- Note [ENUM labels]+~~~~~~~~~~~~~~~~~~~~+A FieldEnum holds the index and the label of an ENUM value. The label is the+text of the cast of DuckDB to VARCHAR. The reader does not cast the vector, so+a user cast cannot change a label. The reader takes the label from one of two+sources:++- A cache (EnumLabels). The reader reads the label of an index through the+ dictionary of the vector type (duckdb_v2_logical_type_get_param) at the first+ row that has the index, and keeps it. Thus a query reads only the labels of+ the values that it reads.+- The type tree. A STRUCT, TUPLE, or UNION decoder converts the type of the+ value, which reads the whole dictionary of each ENUM type in it (see+ logicalTypeToRep). The reader of a child uses the labels of that tree, and it+ raises an error when the tree has another dictionary size than the vector.++A result column has one cache, which lives for one execution, as the tree of+the column (statementStreamColumnLabels in "Database.DuckDB.Simple.Internal").+The column reader gives the cache to the element reader of a LIST or an ARRAY,+because one column has at most one ENUM type outside a STRUCT, a TUPLE, a+UNION, and a MAP. prepareVectorReader makes a new cache for each reader. These+readers use it: the key and value readers of a MAP, the tag reader of a UNION,+the readers of typed arguments, function arguments, and COPY TO rows, and+materializeValue. There the cache lives for one vector.++Each index is checked against the parameter count of the vector type. An index+out of range raises an error, so the reader cannot give a wrong label.+-}+ -- | Prepare metadata once for a vector. The reader must not outlive its chunk.-prepareValueReader :: DuckDBType -> DuckDBVector -> Ptr () -> Ptr Word64 -> IO (Int -> IO FieldValue)-prepareValueReader dtype vector dataPtr validity = case dtype of- DuckDBTypeVariant -> prepareVariantDecoder vector- DuckDBTypeGeometry -> whenValid FieldGeometry <$> prepareGeometryDecoder vector dataPtr- DuckDBTypeStruct -> whenValid FieldStruct <$> prepareStructDecoder vector- DuckDBTypeUnion -> whenValid FieldUnion <$> prepareUnionDecoder vector- _ -> pure (materializeValue dtype vector dataPtr validity)+prepareValueReader+ :: FallbackReaders+ -> TypeTree+ -> EnumLabels+ -> DuckDBV2LogicalTypeId+ -> DuckDBV2VectorHandle+ -> Ptr Void+ -> Ptr Word64+ -> IO (Int -> IO FieldValue)+prepareValueReader readers tree labels dtype vector dataPtr validity = case dtype of+ DuckDBV2LogicalTypeIdVariant -> variantReader readers vector+ DuckDBV2LogicalTypeIdGeometry -> geometryReader readers vector+ DuckDBV2LogicalTypeIdArray -> whenValid FieldArray <$> prepareArrayDecoder readers tree labels vector+ DuckDBV2LogicalTypeIdList -> whenValid FieldList <$> prepareListDecoder readers tree labels vector dataPtr+ DuckDBV2LogicalTypeIdMap -> whenValid FieldMap <$> prepareMapDecoder readers vector dataPtr+ DuckDBV2LogicalTypeIdStruct -> whenValid FieldStruct <$> prepareStructDecoder readers tree vector+ DuckDBV2LogicalTypeIdTuple -> whenValid FieldTuple <$> prepareTupleDecoder readers tree vector+ DuckDBV2LogicalTypeIdUnion -> whenValid FieldUnion <$> prepareUnionDecoder readers tree vector+ DuckDBV2LogicalTypeIdDecimal -> do+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ (width, scale) <-+ withVectorType vector logicalTypeToBaseRep >>= \case+ LogicalTypeDecimal width scale -> pure (width, scale)+ _ -> throwInternalError (Text.pack "invalid DECIMAL type")+ pure (whenValid (FieldDecimal . DecimalValue width scale) (decimalStorage dataPtr width))+ DuckDBV2LogicalTypeIdEnum -> do+ -- See Note [ENUM labels].+ count <- withVectorType vector (output . c_duckdb_v2_logical_type_get_param_count)+ label <- case labels of+ Just cache -> pure (cachedLabel cache)+ Nothing ->+ tree >>= \case+ LogicalTypeEnum entries | toInteger (rangeSize (bounds entries)) == toInteger count -> pure (pure . (entries !) . fromIntegral)+ _ -> throwInternalError (Text.pack "invalid ENUM type")+ let readEnum row = do+ index <- enumStorage dataPtr count row+ when (toInteger index >= toInteger count) $ throwInternalError (Text.pack "ENUM index out of range")+ FieldEnum index <$> label index+ pure (whenValid id readEnum)+ DuckDBV2LogicalTypeIdSQLNull -> pure \_ -> pure FieldNull+ _ -> pure (whenValid id (decodeElement dtype dataPtr)) where+ -- Read a label through the dictionary of the vector type at the first row+ -- that needs it. See Note [ENUM labels].+ cachedLabel cache index =+ IntMap.lookup (fromIntegral index) <$> readIORef cache >>= \case+ Just text -> pure text+ Nothing -> do+ text <- withVectorType vector \logical -> alloca \nameSlot ->+ bracket (output (c_duckdb_v2_logical_type_get_param logical (fromIntegral index) nameSlot)) destroyValue \value ->+ output (c_duckdb_v2_value_get_varchar value) >>= readText+ modifyIORef' cache (IntMap.insert (fromIntegral index) text)+ pure text+ -- See Note [Vector memory layouts] in "Database.DuckDB.Simple.Element". whenValid wrap decode row = do- valid <- chunkIsRowValid validity (fromIntegral row)- if valid then wrap <$> decode row else pure FieldNull+ valid <- chunkIsRowValid validity (fromIntegral row)+ if valid then wrap <$> decode row else pure FieldNull -{- | Copy CRS metadata once for a vector. The decoder copies the WKB bytes of-a valid row. It must not outlive its chunk.+{- Note [User casts and the casts of the library]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+A cast that a user registers is a cast function of each connection of the+database, also of the helper connection of the VARIANT codec. DuckDB uses a+registered cast in place of the built-in cast for the same pair of types. The+header says: "A connection's or callback's factory casts with that client's+cast functions and settings". It also says that the factory of an instance+"supports only the built-in casts, with default settings".++The library runs casts of DuckDB for its own work. A user cast does not change+their result:++- Reads. prepareGeometryReader casts a GEOMETRY vector to BLOB. A result, the+ value of an option (getConfigValue), and the default value of an option+ (getConfigDefaultValue) read a GEOMETRY value with the factory of the+ instance, and so does queryMetrics. Callbacks registered on a connection+ use the same instance factory.+ A scalar or aggregate function registered by a loadable extension uses+ its private instance factory. See Note [Function argument readers] in+ "Database.DuckDB.Simple.Typed". Both paths cast the whole vector.+ Other extension callbacks have a context factory, which includes user+ casts. For those callbacks, withDefaultCastVector writes each row to a+ BLOB vector with the default cast ("Database.DuckDB.Simple.Native").+ This prevents a user cast from GEOMETRY to BLOB from calling itself when+ its callback reads a GEOMETRY argument.+ TODO: When V2 exposes WKB directly, copy it from the input vector and+ remove the intermediate BLOB cast.+- withObjectChildren and withCast cast VARIANT values to the types of their+ payloads ("Database.DuckDB.Simple.VariantCodec"). The codec also reads a+ GEOMETRY payload. It uses the factory of the instance of its helper+ connection (variantCasts).+- Values. rawGeometryDuckValue casts BLOB to GEOMETRY, the ToDuckValue instance+ of Geometry casts VARCHAR to GEOMETRY, enumDuckValue casts VARCHAR to ENUM,+ and variantDuckValue casts its payload to VARIANT+ ("Database.DuckDB.Simple.ToField"). defaultCast writes each value to a+ one-row vector of the target type and reads it back. vector_set_value uses+ "the non-strict default cast". This is the documented cell writer for+ GEOMETRY and VARIANT, which have no committed view layout. For ENUM, it avoids+ value_cast, which includes registered casts. A VARCHAR-to-ENUM user cast+ must not change the label of a FieldEnum. The factory resolves the target+ type. Only a connection or callback factory can construct a GEOMETRY+ type with a CRS. See Note [Raw GEOMETRY parameters] in+ "Database.DuckDB.Simple.ToField". UNION values use a vector to preserve the+ explicit member tag. See Note [UNION values] in that module.+- The bind-time CRS query passes a BLOB literal to ST_GeomFromWKB and binds+ the label as VARCHAR. It needs no user cast. See Note [Bind-time CRS types] in+ "Database.DuckDB.Simple.V2.LogicalRep".++These casts still use the cast functions of the session, so a user cast for+the same pair changes their result:++- setConfigValue and setConfigOption. DuckDB casts the value to the type of the+ option as SQL SET does.+- The casts of the SQL text. DuckDB runs them with the casts of the session.+ The library does not choose them. -}-prepareGeometryDecoder :: DuckDBVector -> Ptr () -> IO (Int -> IO RawGeometry)-prepareGeometryDecoder vector dataPtr = do- crs <- withVectorType vector \logical -> do- rep <- logicalTypeToRep logical- case rep of- LogicalTypeGeometry value -> pure value- _ -> throwIO (userError "duckdb-simple: invalid GEOMETRY type")- pure \row -> (`RawGeometry` crs) <$> chunkDecodeBlob dataPtr (fromIntegral row) -materializeValue :: DuckDBType -> DuckDBVector -> Ptr () -> Ptr Word64 -> Int -> IO FieldValue-materializeValue dtype vector dataPtr validity rowIdx = do- valid <- chunkIsRowValid validity (fromIntegral rowIdx)- if not valid- then pure FieldNull- else case dtype of- DuckDBTypeGeometry -> FieldGeometry <$> (prepareGeometryDecoder vector dataPtr >>= ($ rowIdx))- DuckDBTypeVariant -> decodeVariant vector rowIdx- DuckDBTypeDecimal ->- withVectorType vector \logical -> do- width <- c_duckdb_decimal_width logical- scale <- c_duckdb_decimal_scale logical- internalTy <- c_duckdb_decimal_internal_type logical- rawValue <-- case internalTy of- DuckDBTypeSmallInt ->- toInteger <$> peekElemOff (castPtr dataPtr :: Ptr Int16) rowIdx- DuckDBTypeInteger ->- toInteger <$> peekElemOff (castPtr dataPtr :: Ptr Int32) rowIdx- DuckDBTypeBigInt ->- toInteger <$> peekElemOff (castPtr dataPtr :: Ptr Int64) rowIdx- DuckDBTypeHugeInt ->- duckDBHugeIntToInteger <$> peekElemOff (castPtr dataPtr :: Ptr DuckDBHugeInt) rowIdx- _ ->- error "duckdb-simple: unsupported decimal internal storage type"- pure (FieldDecimal (DecimalValue width scale rawValue))- DuckDBTypeArray -> FieldArray <$> decodeArrayElements vector rowIdx- DuckDBTypeList -> FieldList <$> decodeListElements vector dataPtr rowIdx- DuckDBTypeMap -> FieldMap <$> decodeMapPairs vector dataPtr rowIdx- DuckDBTypeStruct -> FieldStruct <$> (prepareStructDecoder vector >>= ($ rowIdx))- DuckDBTypeUnion -> FieldUnion <$> (prepareUnionDecoder vector >>= ($ rowIdx))- DuckDBTypeEnum ->- withVectorType vector \logical -> do- enumInternal <- c_duckdb_enum_internal_type logical- case enumInternal of- DuckDBTypeUTinyInt ->- FieldEnum . fromIntegral <$> peekElemOff (castPtr dataPtr :: Ptr Word8) rowIdx- DuckDBTypeUSmallInt ->- FieldEnum . fromIntegral <$> peekElemOff (castPtr dataPtr :: Ptr Word16) rowIdx- DuckDBTypeUInteger ->- FieldEnum <$> peekElemOff (castPtr dataPtr :: Ptr Word32) rowIdx- _ ->- error "duckdb-simple: unsupported enum internal storage type"- DuckDBTypeSQLNull -> pure FieldNull- _ -> decodeElement dtype dataPtr rowIdx+-- | Copy the WKB bytes and CRS of every GEOMETRY row in a vector. GEOMETRY has+-- no committed vector layout, so a cast to BLOB reads the rows. The reader takes+-- logical rows. See Note [Vector flatten]. With the factory of an instance, one+-- native cast casts the vector. With the factory of a context, the default cast+-- of 'withDefaultCastVector' casts each row. See+-- Note [User casts and the casts of the library].+prepareGeometryReader :: ReadFactory -> DuckDBV2VectorHandle -> IO (Int -> IO FieldValue)+prepareGeometryReader readFactory vector = do+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ crs <-+ withVectorType vector logicalTypeToBaseRep >>= \case+ LogicalTypeGeometry value -> pure value+ _ -> throwInternalError (Text.pack "invalid GEOMETRY type")+ size <- output (c_duckdb_v2_vector_get_size vector)+ let count = fromIntegral size+ factory = readFactoryHandle readFactory+ readBlobs target = do+ let notBlob = rejectRows (throwInternalError (Text.pack "expected a BLOB vector"))+ readBlob <- prepareVectorReader (FallbackReaders (logicalTypeToRep factory) notBlob notBlob) target+ forM [0 .. count - 1] readBlob+ blobs <-+ bracket+ (output (c_duckdb_v2_factory_create_type_from_id factory DuckDBV2LogicalTypeIdBlob (ConstPtr nullPtr) (ConstPtr nullPtr) 0))+ destroyLogicalType+ \blob -> case readFactory of+ InstanceFactory _ -> withOwnedVector factory blob \target -> do+ checked (c_duckdb_v2_vector_cast factory vector target DuckDBV2CastModeNormal nullPtr nullPtr)+ readBlobs target+ ContextFactory _ ->+ withDefaultCastVector+ factory+ blob+ size+ (\row write -> bracket (output (c_duckdb_v2_vector_get_value vector row)) destroyValue write)+ readBlobs+ let rows = listArray (0, count - 1) blobs :: Array Int FieldValue+ pure \row -> case rows ! row of+ FieldBlob bytes -> pure (FieldGeometry (RawGeometry bytes crs))+ FieldNull -> pure FieldNull+ _ -> throwInternalError (Text.pack "GEOMETRY cast did not produce a BLOB") -decodeArrayElements :: DuckDBVector -> Int -> IO (Array Int FieldValue)-decodeArrayElements vector rowIdx = do- arraySize <-- withVectorType vector \logical -> do- sizeRaw <- c_duckdb_array_type_array_size logical- let sizeWord = fromIntegral sizeRaw :: Word64- ensureWithinIntRange (Text.pack "array size") sizeWord- childVec <- c_duckdb_array_vector_get_child vector- when (childVec == nullPtr) $- throwIO (userError "duckdb-simple: array child vector is null")- readChild <- prepareVectorReader childVec+-- | Read a DECIMAL. DuckDB stores it in the smallest integer type for its width.+-- See Note [Vector memory layouts] in "Database.DuckDB.Simple.Element".+decimalStorage :: Ptr Void -> Word8 -> Int -> IO Integer+decimalStorage dataPtr width rowIdx+ | width <= 4 = toInteger <$> peekElemOff (castPtr dataPtr :: Ptr Int16) rowIdx+ | width <= 9 = toInteger <$> peekElemOff (castPtr dataPtr :: Ptr Int32) rowIdx+ | width <= 18 = toInteger <$> peekElemOff (castPtr dataPtr :: Ptr Int64) rowIdx+ | otherwise = duckDBHugeIntToInteger <$> peekElemOff (castPtr dataPtr :: Ptr DuckDBV2HugeintT) rowIdx++-- | Read an ENUM index. DuckDB stores it in the smallest unsigned type for the dictionary size.+-- See Note [Vector memory layouts] in "Database.DuckDB.Simple.Element".+enumStorage :: Ptr Void -> DuckDBIdx -> Int -> IO Word32+enumStorage dataPtr count rowIdx+ | count <= 255 = fromIntegral <$> peekElemOff (castPtr dataPtr :: Ptr Word8) rowIdx+ | count <= 65535 = fromIntegral <$> peekElemOff (castPtr dataPtr :: Ptr Word16) rowIdx+ | otherwise = peekElemOff (castPtr dataPtr :: Ptr Word32) rowIdx++-- | Read the ARRAY size and prepare the element reader once for a vector. The+-- size gives the layout of the elements, so it comes from the type of this+-- vector. Parameter 1 of an ARRAY type is the size, as 'logicalTypeToRep' reads+-- it. The decoder does not convert the element type. The element reader takes+-- its tree from the tree of the ARRAY. See Note [Type trees].+-- See Note [Constant stack at safe calls] in "Database.DuckDB.Simple.Native".+prepareArrayDecoder :: FallbackReaders -> TypeTree -> EnumLabels -> DuckDBV2VectorHandle -> IO (Int -> IO (Array Int FieldValue))+prepareArrayDecoder readers tree labels vector = do+ size <- withVectorType vector \logical -> alloca \name ->+ bracket (output (c_duckdb_v2_logical_type_get_param logical 1 name)) destroyValue (output . c_duckdb_v2_value_get_bigint)+ arraySize <- ensureWithinIntRange (Text.pack "array size") (fromIntegral size)+ readChild <- vectorChild vector 0 >>= prepareVectorReaderWith readers (elementTree tree) labels+ pure \rowIdx -> do let baseIdx = rowIdx * arraySize- values <-- forM [0 .. arraySize - 1] \delta ->- readChild (baseIdx + delta)- pure $- listArray (0, arraySize - 1) values+ values <- reverse <$> foldM (\values delta -> (: values) <$> readChild (baseIdx + delta)) [] [0 .. arraySize - 1]+ pure (listArray (0, arraySize - 1) values) -decodeListElements :: DuckDBVector -> Ptr () -> Int -> IO [FieldValue]-decodeListElements vector dataPtr rowIdx = do- entry <- peekElemOff (castPtr dataPtr :: Ptr DuckDBListEntry) rowIdx+-- | Prepare the element reader once for a LIST vector. The element reader+-- takes its tree from the tree of the LIST. See Note [Type trees].+-- See Note [Constant stack at safe calls] in "Database.DuckDB.Simple.Native".+prepareListDecoder :: FallbackReaders -> TypeTree -> EnumLabels -> DuckDBV2VectorHandle -> Ptr Void -> IO (Int -> IO [FieldValue])+prepareListDecoder readers tree labels vector dataPtr = do+ readChild <- vectorChild vector 0 >>= prepareVectorReaderWith readers (elementTree tree) labels+ pure \rowIdx -> do+ entry <- peekElemOff (castPtr dataPtr :: Ptr DuckDBV2ListEntry) rowIdx (baseIdx, len) <- listEntryBounds (Text.pack "list") entry- childVec <- c_duckdb_list_vector_get_child vector- when (childVec == nullPtr) $- throwIO (userError "duckdb-simple: list child vector is null")- readChild <- prepareVectorReader childVec- forM [0 .. len - 1] \delta ->- readChild (baseIdx + delta)+ reverse <$> foldM (\values delta -> (: values) <$> readChild (baseIdx + delta)) [] [0 .. len - 1] -decodeMapPairs :: DuckDBVector -> Ptr () -> Int -> IO [(FieldValue, FieldValue)]-decodeMapPairs vector dataPtr rowIdx = do- entry <- peekElemOff (castPtr dataPtr :: Ptr DuckDBListEntry) rowIdx+-- | Prepare the key reader and the value reader once for a MAP vector. The+-- decoder gives them no type tree. Each converts the type of its own vector when+-- it needs a description. A tree of the MAP holds the key type and the value+-- type together. A value reader that took its tree from it would thus also+-- convert the key type, for example the dictionary of a large ENUM. The+-- decoder does not flatten the entry STRUCT. See Note [Vector flatten] and+-- Note [Type trees].+-- See Note [Constant stack at safe calls] in "Database.DuckDB.Simple.Native".+prepareMapDecoder :: FallbackReaders -> DuckDBV2VectorHandle -> Ptr Void -> IO (Int -> IO [(FieldValue, FieldValue)])+prepareMapDecoder readers vector dataPtr = do+ entries <- vectorChild vector 0+ readKey <- vectorChild entries 0 >>= prepareVectorReader readers+ readValue <- vectorChild entries 1 >>= prepareVectorReader readers+ pure \rowIdx -> do+ entry <- peekElemOff (castPtr dataPtr :: Ptr DuckDBV2ListEntry) rowIdx (baseIdx, len) <- listEntryBounds (Text.pack "map") entry- structVec <- c_duckdb_list_vector_get_child vector- when (structVec == nullPtr) $- throwIO (userError "duckdb-simple: map struct vector is null")- keyVec <- c_duckdb_struct_vector_get_child structVec 0- valueVec <- c_duckdb_struct_vector_get_child structVec 1- when (keyVec == nullPtr || valueVec == nullPtr) $- throwIO (userError "duckdb-simple: map child vectors are null")- readKey <- prepareVectorReader keyVec- readValue <- prepareVectorReader valueVec- forM [0 .. len - 1] \delta -> do- let childIdx = baseIdx + delta- keyValue <- readKey childIdx- valueValue <- readValue childIdx- pure (keyValue, valueValue)+ reverse+ <$> foldM+ ( \pairs delta -> do+ let childIdx = baseIdx + delta+ keyValue <- readKey childIdx+ valueValue <- readValue childIdx+ pure ((keyValue, valueValue) : pairs)+ )+ []+ [0 .. len - 1] -{- | Read the STRUCT type and prepare a reader for each child once for a vector.-The decoder reads a valid row. It must not outlive its chunk.--}-prepareStructDecoder :: DuckDBVector -> IO (Int -> IO (StructValue FieldValue))-prepareStructDecoder vector =- withVectorType vector \logical -> do- structTypeRep <- logicalTypeToRep logical- structFields <-- case structTypeRep of- LogicalTypeStruct typeArray -> pure typeArray- other ->- throwIO- ( userError- ( "duckdb-simple: expected STRUCT logical type, but saw "- <> show other- )- )- let typeList = elems structFields- count = length typeList- indexMap =- Map.fromList (zip (map structFieldName typeList) [0 ..])- childReaders <-- forM (zip [0 .. count - 1] typeList) \(childIdx, StructField{structFieldName}) -> do- childVec <- c_duckdb_struct_vector_get_child vector (fromIntegral childIdx)- when (childVec == nullPtr) $- throwIO (userError "duckdb-simple: struct child vector is null")- readChild <- prepareVectorReader childVec- pure (structFieldName, readChild)- pure \rowIdx -> do- valueFields <-- forM childReaders \(name, readChild) -> do- value <- readChild rowIdx- pure StructField{structFieldName = name, structFieldValue = value}- let fieldArray =- listArray (0, count - 1) valueFields- pure- StructValue- { structValueFields = fieldArray- , structValueTypes = structFields- , structValueIndex = indexMap- }+-- | Read the STRUCT type and prepare a reader for each child once for a vector.+-- The decoder reads a valid row. It must not outlive its chunk. See+-- Note [Type trees].+prepareStructDecoder :: FallbackReaders -> TypeTree -> DuckDBV2VectorHandle -> IO (Int -> IO (StructValue FieldValue))+prepareStructDecoder readers tree vector = do+ structFields <-+ tree >>= \case+ LogicalTypeStruct typeArray -> pure typeArray+ other ->+ throwInternalError (Text.pack "expected STRUCT logical type, but saw " <> logicalTypeSQL other)+ let typeList = elems structFields+ count = length typeList+ indexMap =+ Map.fromList (zip (map structFieldName typeList) [0 ..])+ checkChildCount "STRUCT" vector count+ childReaders <-+ forM (zip [0 ..] typeList) \(childIdx, StructField{structFieldName, structFieldValue}) -> do+ readChild <- vectorChild vector childIdx >>= prepareVectorReaderWith readers (pure structFieldValue) Nothing+ pure (structFieldName, readChild)+ pure \rowIdx -> do+ valueFields <-+ forM childReaders \(name, readChild) -> do+ value <- readChild rowIdx+ pure StructField{structFieldName = name, structFieldValue = value}+ let fieldArray =+ listArray (0, count - 1) valueFields+ pure+ StructValue+ { structValueFields = fieldArray+ , structValueTypes = structFields+ , structValueIndex = indexMap+ } -{- | Read the UNION type and prepare the tag reader and a reader for each member-once for a vector. The decoder reads a valid row. It must not outlive its chunk.--}-prepareUnionDecoder :: DuckDBVector -> IO (Int -> IO (UnionValue FieldValue))-prepareUnionDecoder vector =- withVectorType vector \logical -> do- unionTypeRep <- logicalTypeToRep logical- membersArray <-- case unionTypeRep of- LogicalTypeUnion members -> pure members- other ->- throwIO- ( userError- ( "duckdb-simple: expected UNION logical type, but saw "- <> show other- )- )- let membersList = elems membersArray- memberCount = length membersList- tagVec <- c_duckdb_struct_vector_get_child vector 0- when (tagVec == nullPtr) $- throwIO (userError "duckdb-simple: union tag vector is null")- readTag <- prepareVectorReader tagVec- memberReaders <-- forM [1 .. memberCount] \childIdx -> do- memberVec <- c_duckdb_struct_vector_get_child vector (fromIntegral childIdx)- when (memberVec == nullPtr) $- throwIO (userError "duckdb-simple: union member vector is null")- prepareVectorReader memberVec- pure \rowIdx -> do- memberIdx <- readTag rowIdx >>= unionTagIndex- when (memberIdx < 0 || memberIdx >= memberCount) $- throwIO (userError "duckdb-simple: union tag out of range")- payload <- (memberReaders !! memberIdx) rowIdx- pure- UnionValue- { unionValueIndex = fromIntegral memberIdx- , unionValueLabel = unionMemberName (membersList !! memberIdx)- , unionValuePayload = payload- , unionValueMembers = membersArray- }+-- | Read positional field metadata and prepare readers that borrow the tuple.+-- See Note [Type trees].+prepareTupleDecoder :: FallbackReaders -> TypeTree -> DuckDBV2VectorHandle -> IO (Int -> IO (TupleValue FieldValue))+prepareTupleDecoder readers tree vector = do+ types <-+ tree >>= \case+ LogicalTypeTuple fields -> pure fields+ _ -> throwInternalError (Text.pack "invalid TUPLE type")+ let typeList = elems types+ count = length typeList+ checkChildCount "TUPLE" vector count+ fieldReaders <- forM (zip [0 ..] typeList) \(index, fieldType) ->+ vectorChild vector index >>= prepareVectorReaderWith readers (pure fieldType) Nothing+ pure \row -> do+ fields <- mapM ($ row) fieldReaders+ pure (TupleValue (listArray (0, count - 1) fields) types) +-- | Read the UNION type and prepare the tag reader once for a vector. Prepare+-- the reader of a member at the first row that selects the member, so that a+-- member that no row selects makes no native call. The decoder reads a valid+-- row. It must not outlive its chunk. See Note [Type trees].+prepareUnionDecoder :: FallbackReaders -> TypeTree -> DuckDBV2VectorHandle -> IO (Int -> IO (UnionValue FieldValue))+prepareUnionDecoder readers tree vector = do+ membersArray <-+ tree >>= \case+ LogicalTypeUnion members -> pure members+ other ->+ throwInternalError (Text.pack "expected UNION logical type, but saw " <> logicalTypeSQL other)+ let memberCount = rangeSize (bounds membersArray)+ -- Child 0 of a UNION vector is the tag, and member j is child j + 1. The tag+ -- reader needs no type tree.+ checkChildCount "UNION" vector (memberCount + 1)+ readTag <- vectorChild vector 0 >>= prepareVectorReader readers+ readersByTag <- fmap (listArray (0, memberCount - 1)) $+ forM (zip [1 ..] (elems membersArray)) \(childIdx, UnionMemberType{unionMemberName, unionMemberType}) -> do+ memberVec <- vectorChild vector childIdx+ prepared <- newIORef Nothing+ pure \row -> do+ cached <- readIORef prepared+ readMember <- case cached of+ Just reader -> pure reader+ Nothing -> do+ reader <- prepareVectorReaderWith readers (pure unionMemberType) Nothing memberVec+ writeIORef prepared (Just reader)+ pure reader+ payload <- readMember row+ pure (unionMemberName, payload)+ pure \rowIdx -> do+ memberIdx <- readTag rowIdx >>= unionTagIndex+ when (memberIdx < 0 || memberIdx >= memberCount) $+ throwInternalError (Text.pack "union tag out of range")+ (name, payload) <- (readersByTag ! memberIdx) rowIdx+ pure+ UnionValue+ { unionValueIndex = fromIntegral memberIdx+ , unionValueLabel = name+ , unionValuePayload = payload+ , unionValueMembers = membersArray+ }+ -- | Convert a decoded UNION tag to a member index. unionTagIndex :: FieldValue -> IO Int unionTagIndex = \case- FieldWord8 tagWord -> pure (fromIntegral tagWord :: Int)- FieldWord16 tagWord -> pure (fromIntegral tagWord :: Int)- FieldWord32 tagWord ->- if tagWord <= fromIntegral (maxBound :: Word16)- then pure (fromIntegral tagWord)- else throwIO (userError "duckdb-simple: union tag exceeds Word16 range")- FieldWord64 tagWord ->- if tagWord <= fromIntegral (maxBound :: Word16)- then pure (fromIntegral tagWord)- else throwIO (userError "duckdb-simple: union tag exceeds Word16 range")- FieldInt8 tagInt- | tagInt >= 0 -> pure (fromIntegral tagInt)- | otherwise -> throwIO (userError "duckdb-simple: union tag negative")- FieldInt16 tagInt- | tagInt >= 0 -> pure (fromIntegral tagInt)- | otherwise -> throwIO (userError "duckdb-simple: union tag negative")- FieldInt32 tagInt- | tagInt >= 0 && tagInt <= fromIntegral (maxBound :: Word16) -> pure (fromIntegral tagInt)- | tagInt < 0 -> throwIO (userError "duckdb-simple: union tag negative")- | otherwise -> throwIO (userError "duckdb-simple: union tag exceeds Word16 range")- FieldInt64 tagInt- | tagInt >= 0 && tagInt <= fromIntegral (maxBound :: Word16) -> pure (fromIntegral tagInt)- | tagInt < 0 -> throwIO (userError "duckdb-simple: union tag negative")- | otherwise -> throwIO (userError "duckdb-simple: union tag exceeds Word16 range")- FieldNull ->- throwIO (userError "duckdb-simple: encountered NULL union tag")- other ->- throwIO- ( userError- ( "duckdb-simple: unexpected union tag value "- <> show other- )- )+ FieldWord8 tagWord -> pure (fromIntegral tagWord :: Int)+ FieldWord16 tagWord -> pure (fromIntegral tagWord :: Int)+ FieldWord32 tagWord ->+ if tagWord <= fromIntegral (maxBound :: Word16)+ then pure (fromIntegral tagWord)+ else throwInternalError (Text.pack "union tag exceeds Word16 range")+ FieldWord64 tagWord ->+ if tagWord <= fromIntegral (maxBound :: Word16)+ then pure (fromIntegral tagWord)+ else throwInternalError (Text.pack "union tag exceeds Word16 range")+ FieldInt8 tagInt+ | tagInt >= 0 -> pure (fromIntegral tagInt)+ | otherwise -> throwInternalError (Text.pack "union tag negative")+ FieldInt16 tagInt+ | tagInt >= 0 -> pure (fromIntegral tagInt)+ | otherwise -> throwInternalError (Text.pack "union tag negative")+ FieldInt32 tagInt+ | tagInt >= 0 && tagInt <= fromIntegral (maxBound :: Word16) -> pure (fromIntegral tagInt)+ | tagInt < 0 -> throwInternalError (Text.pack "union tag negative")+ | otherwise -> throwInternalError (Text.pack "union tag exceeds Word16 range")+ FieldInt64 tagInt+ | tagInt >= 0 && tagInt <= fromIntegral (maxBound :: Word16) -> pure (fromIntegral tagInt)+ | tagInt < 0 -> throwInternalError (Text.pack "union tag negative")+ | otherwise -> throwInternalError (Text.pack "union tag exceeds Word16 range")+ FieldNull ->+ throwInternalError (Text.pack "encountered NULL union tag")+ other ->+ throwInternalError (Text.pack ("unexpected union tag value " <> show other)) -listEntryBounds :: Text -> DuckDBListEntry -> IO (Int, Int)-listEntryBounds context DuckDBListEntry{duckDBListEntryOffset, duckDBListEntryLength} = do- base <- ensureWithinIntRange (context <> Text.pack " offset") duckDBListEntryOffset- len <- ensureWithinIntRange (context <> Text.pack " length") duckDBListEntryLength- let maxInt = toInteger (maxBound :: Int)- upperBound = toInteger base + toInteger len - 1- when (len > 0 && upperBound > maxInt) $- throwIO (userError ("duckdb-simple: " <> Text.unpack context <> " bounds exceed Int range"))- pure (base, len)+listEntryBounds :: Text -> DuckDBV2ListEntry -> IO (Int, Int)+listEntryBounds context (DuckDBV2ListEntry offset lenRaw) = do+ base <- ensureWithinIntRange (context <> Text.pack " offset") (fromIntegral offset)+ len <- ensureWithinIntRange (context <> Text.pack " length") (fromIntegral lenRaw)+ let maxInt = toInteger (maxBound :: Int)+ upperBound = toInteger base + toInteger len - 1+ when (len > 0 && upperBound > maxInt) $+ throwInternalError (context <> Text.pack " bounds exceed Int range")+ pure (base, len) ensureWithinIntRange :: Text -> Word64 -> IO Int ensureWithinIntRange context value =- let actual = toInteger value- limit = toInteger (maxBound :: Int)- in if actual <= limit- then pure (fromInteger actual)- else throwIO (userError ("duckdb-simple: " <> Text.unpack context <> " exceeds Int range"))+ let actual = toInteger value+ limit = toInteger (maxBound :: Int)+ in if actual <= limit+ then pure (fromInteger actual)+ else throwInternalError (context <> Text.pack " exceeds Int range")++-- | Take the element tree from the tree of a LIST or ARRAY. Read the parent+-- tree only when the element reader needs its tree. See+-- Note [Type trees].+elementTree :: TypeTree -> TypeTree+elementTree tree =+ tree >>= \case+ LogicalTypeList element -> pure element+ LogicalTypeArray element _ -> pure element+ _ -> throwInternalError (Text.pack "invalid LIST or ARRAY type")++-- | Raise an error when the type tree of a STRUCT, TUPLE, or UNION vector gives+-- another child count than the vector has. The tree can come from the type of a+-- parent vector, and the decoder must not drop a child. See Note [Type trees].+checkChildCount :: String -> DuckDBV2VectorHandle -> Int -> IO ()+checkChildCount kind vector expected = do+ actual <- output (c_duckdb_v2_vector_get_child_count vector)+ when (toInteger actual /= toInteger expected) $+ throwInternalError+ (Text.pack ("invalid " <> kind <> " type: the type has " <> show expected <> " children, but the vector has " <> show actual))++-- | Borrow a nested vector from its parent chunk.+vectorChild :: DuckDBV2VectorHandle -> DuckDBIdx -> IO DuckDBV2VectorHandle+vectorChild vector index = output (c_duckdb_v2_vector_get_child vector index)
+ src/Database/DuckDB/Simple/Native.hs view
@@ -0,0 +1,982 @@+{-# LANGUAGE DerivingStrategies #-}+{-# LANGUAGE FlexibleContexts #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : Database.DuckDB.Simple.Native+-- Description : Shared V2 error handling and UTF-8 views.+module Database.DuckDB.Simple.Native (+ module Database.DuckDB.FFI.V2,+ Query (..),+ SQLError (..),+ ErrorType (..),+ errorTypeCodes,+ errorTypeFromNative,+ SettingScope (..),+ settingScopeToNative,+ settingScopeFromNative,+ LogLevel (..),+ logLevelToNative,+ logLevelFromNative,+ LogScope (..),+ logScopeFromNative,+ TokenType (..),+ tokenTypeFromNative,+ MacroFunctionType (..),+ macroFunctionTypeToNative,+ LogicalTypeId (..),+ logicalTypeIdToNative,+ logicalTypeIdFromNative,+ logicalTypeIdName,+ arraySizeReason,+ maxArraySize,+ FileFlag (..),+ fileFlagToNative,+ checked,+ checkCode,+ output,+ destroy,+ destroyValue,+ destroyLogicalType,+ acquireItems,+ releaseItems,+ withCreatedValues,+ withOwnedVector,+ withDefaultCastVector,+ nullHandle,+ indices,+ withText,+ withTextArray,+ readText,+ valueText,+ withBytes,+ readBytes,+ withIdentifier,+ withIdentifierArray,+ readIdentifier,+ foldIdentifier,+ quoteIdentifier,+ throwInvalidArgument,+ throwConversionError,+ throwInternalError,+ exceptionText,+ withConnectionFactory,+ factoryOutput,+ withSQLStatement,+ withContextFactory,+ withInstanceFactory,+ ValueFactory,+ NativeValue,+) where++import Control.Applicative ((<|>))+import Control.Exception (Exception (..), SomeException (..), bracket, evaluate, finally, mask_, onException, throwIO, try)+import Control.Monad (foldM, forM_, mfilter, unless, when, (<$!>))+import qualified Data.ByteString as BS+import Data.Char (isAsciiUpper, toLower)+import Data.Coerce (Coercible, coerce)+import Data.Maybe (fromMaybe)+import Data.String (IsString (..))+import Data.Text (Text)+import qualified Data.Text as Text+import qualified Data.Text.Encoding as Text+import Data.Text.Encoding.Error (lenientDecode)+import qualified Data.Text.Foreign as TextForeign+import Database.DuckDB.FFI.V2+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.C.Types (CUInt)+import Foreign.Marshal.Alloc (alloca, allocaBytes)+import Foreign.Marshal.Array (withArray)+import Foreign.Marshal.Utils (fillBytes, with, withMany)+import Foreign.Ptr (Ptr, castPtr, nullPtr)+import Foreign.Storable (Storable, peek, poke, sizeOf)++-- | The largest size of a DuckDB ARRAY: @ArrayType::MAX_ARRAY_SIZE@ in+-- @src/include/duckdb/common/types.hpp@ of DuckDB 2.0.0. See+-- Note [Sized ARRAY bounds].+maxArraySize :: Int+maxArraySize = 100000++-- | Give the reason why a size has no DuckDB ARRAY type, or 'Nothing' for a+-- valid size. See Note [Sized ARRAY bounds].+arraySizeReason :: Integer -> Maybe Text+arraySizeReason size+ | size < 1 = Just "a DuckDB ARRAY has at least one element"+ | size > toInteger maxArraySize = Just ("a DuckDB ARRAY has at most " <> Text.pack (show maxArraySize) <> " elements")+ | otherwise = Nothing++{- Note [Sized ARRAY bounds]+~~~~~~~~~~~~~~~~~~~~~~~~~~~+A DuckDB ARRAY has 1 to maxArraySize elements. Its value API does not check+this limit, so both logical type construction and value binding must check+the size. Generic declarations check each ARRAY, also below an alias or a+container. A non-NULL SizedArray read also checks the size before it counts+or converts the elements.++SizedArray has a type-level Nat, but LogicalTypeArray stores a Word64. The+generic logical type saturates a Nat above maxBound Word64 to that bound.+This is an invalid-size sentinel. It cannot wrap into a valid size, and the+declaration and native construction checks reject it. No pure conversion+throws an exception.+-}++-- | SQL text encoded as UTF-8 when it is sent to DuckDB.+newtype Query = Query {fromQuery :: Text}+ deriving stock (Eq, Ord, Show)++instance Semigroup Query where+ Query a <> Query b = Query (a <> b)++instance IsString Query where+ fromString = Query . Text.pack++-- | A native V2 error or an error reported by duckdb-simple.+data SQLError = SQLError+ { sqlErrorMessage :: Text+ , sqlErrorType :: Maybe ErrorType+ , sqlErrorQuery :: Maybe Query+ , sqlErrorBody :: Maybe Text+ -- ^ The text of a native error without the prefix of its type. For example,+ -- the message @Conversion Error: Could not convert ...@ has the body+ -- @Could not convert ...@. An error of duckdb-simple has no body. A callback+ -- gives the type and the body of an error back to DuckDB, except for the+ -- types that the section /Errors/ of the README lists.+ }+ deriving stock (Eq, Show)++instance Exception SQLError++-- | The category of a native DuckDB error. An error that duckdb-simple reports+-- itself has an error type where one category fits the failure. An input check+-- for a value that its SQL type cannot hold raises 'ErrorTypeConversion', as+-- DuckDB does for the same value in SQL (see 'throwConversionError'). Each+-- other input check raises 'ErrorInputInvalid', also for a value of the wrong+-- shape or type (see 'throwInvalidArgument'). For a nanosecond target, DuckDB+-- can raise the error itself (see Note [Nanosecond parameters] in+-- "Database.DuckDB.Simple.ToField"). A callback that waits too long for a+-- connection lock gets 'ErrorResourceInUse'.+-- These errors of duckdb-simple have no error type: the use of a closed+-- environment, instance, connection, or statement, a loaded DuckDB+-- library of another version, a callback state read with the wrong type, and a+-- result value that does not convert to its Haskell type.+--+-- A failed check of the state of the library, or of a vector, a type, or a+-- value that DuckDB gave, raises an t'SQLError' of the type+-- 'ErrorRuntimeInternal'. Such an error is a defect of the library or of+-- DuckDB, not an error of the input. Report it. Inside a callback, the+-- statement that runs the callback fails with 'ErrorInputInvalid', because a+-- callback does not pass 'ErrorRuntimeInternal' (a cast gives+-- 'ErrorTypeMismatch'). See Note [Internal errors in callbacks].+data ErrorType+ = ErrorApi+ | ErrorIOFileNotFound+ | ErrorIOReadFailure+ | ErrorIOEof+ | ErrorIOGeneral+ | ErrorIONetwork+ | ErrorIOHTTP+ | ErrorInputInvalid+ | ErrorInputParameterInvalid+ | ErrorInputOutOfRange+ | ErrorInputObjectSize+ | ErrorInputMetadataDeferred+ | ErrorResourceInUse+ | ErrorResourceOutOfMemory+ | ErrorResourceConnection+ | ErrorResourceDependency+ | ErrorResourceMissingExtension+ | ErrorResourceAutoload+ | ErrorTypeConversion+ | ErrorTypeUnknown+ | ErrorTypeInvalid+ | ErrorTypeMismatch+ | ErrorTypeDecimal+ | ErrorTypeDivideByZero+ | ErrorQueryParser+ | ErrorQuerySyntax+ | ErrorQueryBinder+ | ErrorQueryPlanner+ | ErrorQueryOptimizer+ | ErrorQueryExpression+ | ErrorQueryExecutor+ | ErrorQueryScheduler+ | ErrorQueryNotImplemented+ | ErrorQueryParameterNotResolved+ | ErrorQueryParameterNotAllowed+ | ErrorDatabaseCatalog+ | ErrorDatabaseTransaction+ | ErrorDatabaseConstraint+ | ErrorDatabaseIndex+ | ErrorDatabaseSequence+ | ErrorDatabaseStatistics+ | ErrorDatabaseSerialization+ | ErrorConfigurationSettings+ | ErrorConfigurationInvalid+ | ErrorConfigurationPermission+ | ErrorRuntimeInternal+ | ErrorRuntimeFatal+ | ErrorRuntimeInterrupt+ | ErrorRuntimeNullPointer+ deriving (Eq, Ord, Show, Read, Enum, Bounded)++-- | Each error type and its native error code.+errorTypeCodes :: [(ErrorType, DuckDBV2Error)]+errorTypeCodes =+ [ (ErrorApi, DuckDBV2ErrorApi)+ , (ErrorIOFileNotFound, DuckDBV2ErrorIOFileNotFound)+ , (ErrorIOReadFailure, DuckDBV2ErrorIOReadFailure)+ , (ErrorIOEof, DuckDBV2ErrorIOEof)+ , (ErrorIOGeneral, DuckDBV2ErrorIOGeneral)+ , (ErrorIONetwork, DuckDBV2ErrorIONetwork)+ , (ErrorIOHTTP, DuckDBV2ErrorIOHTTP)+ , (ErrorInputInvalid, DuckDBV2ErrorInputInvalid)+ , (ErrorInputParameterInvalid, DuckDBV2ErrorInputParameterInvalid)+ , (ErrorInputOutOfRange, DuckDBV2ErrorInputOutOfRange)+ , (ErrorInputObjectSize, DuckDBV2ErrorInputObjectSize)+ , (ErrorInputMetadataDeferred, DuckDBV2ErrorInputMetadataDeferred)+ , (ErrorResourceInUse, DuckDBV2ErrorResourceInUse)+ , (ErrorResourceOutOfMemory, DuckDBV2ErrorResourceOutOfMemory)+ , (ErrorResourceConnection, DuckDBV2ErrorResourceConnection)+ , (ErrorResourceDependency, DuckDBV2ErrorResourceDependency)+ , (ErrorResourceMissingExtension, DuckDBV2ErrorResourceMissingExtension)+ , (ErrorResourceAutoload, DuckDBV2ErrorResourceAutoload)+ , (ErrorTypeConversion, DuckDBV2ErrorTypeConversion)+ , (ErrorTypeUnknown, DuckDBV2ErrorTypeUnknown)+ , (ErrorTypeInvalid, DuckDBV2ErrorTypeInvalid)+ , (ErrorTypeMismatch, DuckDBV2ErrorTypeMismatch)+ , (ErrorTypeDecimal, DuckDBV2ErrorTypeDecimal)+ , (ErrorTypeDivideByZero, DuckDBV2ErrorTypeDivideByZero)+ , (ErrorQueryParser, DuckDBV2ErrorQueryParser)+ , (ErrorQuerySyntax, DuckDBV2ErrorQuerySyntax)+ , (ErrorQueryBinder, DuckDBV2ErrorQueryBinder)+ , (ErrorQueryPlanner, DuckDBV2ErrorQueryPlanner)+ , (ErrorQueryOptimizer, DuckDBV2ErrorQueryOptimizer)+ , (ErrorQueryExpression, DuckDBV2ErrorQueryExpression)+ , (ErrorQueryExecutor, DuckDBV2ErrorQueryExecutor)+ , (ErrorQueryScheduler, DuckDBV2ErrorQueryScheduler)+ , (ErrorQueryNotImplemented, DuckDBV2ErrorQueryNotImplemented)+ , (ErrorQueryParameterNotResolved, DuckDBV2ErrorQueryParameterNotResolved)+ , (ErrorQueryParameterNotAllowed, DuckDBV2ErrorQueryParameterNotAllowed)+ , (ErrorDatabaseCatalog, DuckDBV2ErrorDatabaseCatalog)+ , (ErrorDatabaseTransaction, DuckDBV2ErrorDatabaseTransaction)+ , (ErrorDatabaseConstraint, DuckDBV2ErrorDatabaseConstraint)+ , (ErrorDatabaseIndex, DuckDBV2ErrorDatabaseIndex)+ , (ErrorDatabaseSequence, DuckDBV2ErrorDatabaseSequence)+ , (ErrorDatabaseStatistics, DuckDBV2ErrorDatabaseStatistics)+ , (ErrorDatabaseSerialization, DuckDBV2ErrorDatabaseSerialization)+ , (ErrorConfigurationSettings, DuckDBV2ErrorConfigurationSettings)+ , (ErrorConfigurationInvalid, DuckDBV2ErrorConfigurationInvalid)+ , (ErrorConfigurationPermission, DuckDBV2ErrorConfigurationPermission)+ , (ErrorRuntimeInternal, DuckDBV2ErrorRuntimeInternal)+ , (ErrorRuntimeFatal, DuckDBV2ErrorRuntimeFatal)+ , (ErrorRuntimeInterrupt, DuckDBV2ErrorRuntimeInterrupt)+ , (ErrorRuntimeNullPointer, DuckDBV2ErrorRuntimeNullPointer)+ ]++-- | Find the error type of a native error code. The success code and an unknown+-- code give 'Nothing'.+errorTypeFromNative :: DuckDBV2Error -> Maybe ErrorType+errorTypeFromNative code = lookup code [(native, errorType) | (errorType, native) <- errorTypeCodes]++-- | A null pointer represented by an opaque native handle.+nullHandle :: (Coercible (Ptr ()) a) => a+nullHandle = coerce (nullPtr @())++-- | Check a status from an operation without an error-info output.+checkCode :: DuckDBV2Error -> IO ()+checkCode code =+ unless (code == DuckDBV2ErrorNone) $+ throwIO (SQLError (Text.pack (show code)) (errorTypeFromNative code) Nothing Nothing)++-- | Copy a native diagnostic and release its error-info handle. The message is+-- the text of the diagnostic, and the body is its text without the prefix (see+-- 'sqlErrorBody'). Decode the error text with replacement characters, so that+-- a native text that is not valid UTF-8 cannot raise when the error is shown.+checked :: (Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO ()+checked action = mask_ $ alloca $ \slot -> do+ poke slot nullHandle+ -- A successful call leaves the slot null. Destroy only a handle in the slot.+ let release = do+ info <- peek slot+ unless (info == nullHandle) (c_duckdb_v2_error_info_destroy slot >>= checkCode)+ flip finally release $ do+ code <- action slot+ unless (code == DuckDBV2ErrorNone) $ do+ info <- peek slot+ let readInfo getText+ | info == nullHandle = pure Nothing+ | otherwise = alloca $ \view -> do+ rc <- getText info view+ if rc == DuckDBV2ErrorNone+ then Just . Text.decodeUtf8With lenientDecode <$> (peek view >>= readBytes)+ else pure Nothing+ message <- fromMaybe (Text.pack (show code)) <$> readInfo c_duckdb_v2_error_info_get_text+ body <- readInfo c_duckdb_v2_error_info_get_raw_text+ throwIO (SQLError message (errorTypeFromNative code) Nothing (mfilter (not . Text.null) body))++-- | Read one checked native output. Initialize its storage before the call.+output :: forall a. (Storable a) => (Ptr a -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a+output action = alloca $ \slot -> do+ fillBytes slot 0 (sizeOf (undefined :: a))+ checked (action slot)+ peek slot++-- | Destroy an owned handle held in a temporary native slot.+destroy :: (Storable a) => (Ptr a -> IO DuckDBV2Error) -> a -> IO ()+destroy release value = with value (\slot -> release slot >>= checkCode)++-- | Release an owned native value.+destroyValue :: DuckDBV2ValueHandle -> IO ()+destroyValue = destroy c_duckdb_v2_value_destroy++-- | Release an owned native logical type.+destroyLogicalType :: DuckDBV2LogicalTypeHandle -> IO ()+destroyLogicalType = destroy c_duckdb_v2_logical_type_destroy++{- Note [Constant stack at safe calls]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+Each c_duckdb_v2 binding is a safe foreign call. At a safe call, the GHC+runtime (threadPaused) reads the frames of the current stack chunk. A loop that+keeps a frame for each element or row adds those retained frames to this work.++Thus a loop whose steps make native calls keeps a constant stack. The rule+applies to a loop over the rows of a chunk, over the chunks of a result, and+over a list of native items. It applies also when only some steps make a native+call, as in the row loop of withStreamRows in+"Database.DuckDB.Simple.Result".++The code uses these forms:++- forM_ for a loop that gives no result (writeRows in+ "Database.DuckDB.Simple.Typed.Table").++- foldM with an accumulator. A fold can keep one value, as the first error in+ releaseItems. The LIST, ARRAY, and MAP readers in+ "Database.DuckDB.Simple.Materialize" prepend each element and then reverse+ the list. They read elements in order, so the first failing read stays first.++- A tail-recursive loop with an accumulator. acquireItems puts each item in+ front of the list, and then reverses the list. Its exception handler covers+ one step, including the evaluation of the next list constructor.++- A loop that reads the elements from the last to the first, and puts each+ element in front of the list. The list then needs no reverse (the row loop of+ withStreamRows).++These loops keep a frame for each element, and they stay as they are, because+the list is short in practice or the operation is rare:++- The row loop of materializeChunkRows in+ "Database.DuckDB.Simple.Copy.Internal". Only a BIGNUM element, the first+ row of a UNION member, and a rejectRows reader make a native call for each+ row there.++- The row loop of executeScalarFunction in+ "Database.DuckDB.Simple.Function.Internal", for the same readers.++- cleanupAll, for a connection with many open statements, in+ "Database.DuckDB.Simple.Internal".++- The option loop in "Database.DuckDB.Simple.Config".+-}++-- | Run the acquire actions in list order under 'mask_', and give the items in+-- list order. If evaluation of the input list or an acquire action fails,+-- release the items that the earlier actions gave, from the last to the first,+-- and raise that error. The loop keeps a constant stack. See+-- Note [Constant stack at safe calls].+acquireItems :: (a -> IO ()) -> [IO a] -> IO [a]+acquireItems release = mask_ . go []+ where+ go owned remaining = do+ next <- flip onException (releaseItems release owned) $ do+ actions <- evaluate remaining+ case actions of+ [] -> pure Nothing+ acquire : rest -> do+ item <- acquire+ pure (Just (item, rest))+ case next of+ Nothing -> pure (reverse owned)+ Just (item, rest) -> go (item : owned) rest++-- | Release each item in list order, also when a release raises an error.+-- Raise the first error after the last release. The fold keeps a constant stack.+-- See Note [Constant stack at safe calls].+releaseItems :: (a -> IO ()) -> [a] -> IO ()+releaseItems release items = foldM step (Nothing :: Maybe SomeException) items >>= mapM_ throwIO+ where+ step failure item = either (\err -> failure <|> Just err) (const failure) <$!> try (release item)++-- | Create the values in list order, and give them to the action. Destroy the+-- values in reverse order of creation after the action or after a failure. One+-- 'bracket' covers the whole list.+withCreatedValues :: [IO DuckDBV2ValueHandle] -> ([DuckDBV2ValueHandle] -> IO a) -> IO a+withCreatedValues creates = bracket (acquireItems destroyValue creates) (releaseItems destroyValue . reverse)++-- | Run an action with a new vector of the given type. The V2 API creates a+-- vector only as a column of a data chunk, so this function creates a chunk+-- with one column. The function destroys the chunk after the action, so the+-- vector must not be used after the action returns.+withOwnedVector :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> (DuckDBV2VectorHandle -> IO a) -> IO a+withOwnedVector factory logicalType action =+ with logicalType $ \types ->+ bracket (output (c_duckdb_v2_data_chunk_create factory (ConstPtr types) 1)) (destroy c_duckdb_v2_data_chunk_destroy) $ \chunk ->+ output (c_duckdb_v2_data_chunk_get_vector chunk 0) >>= action++-- | Cast values with the default cast of DuckDB. The function makes a new+-- vector of the type with the given number of rows (withOwnedVector). For each+-- row, the writer gives a borrowed value to its second argument, and+-- duckdb_v2_vector_set_value writes the value to the row. The header says that+-- this call casts "through the non-strict default cast". Then the action reads+-- the vector. See Note [User casts and the casts of the library] in+-- <src/Database.DuckDB.Simple.Materialize.html Database.DuckDB.Simple.Materialize>.+withDefaultCastVector+ :: DuckDBV2FactoryHandle+ -> DuckDBV2LogicalTypeHandle+ -> DuckDBIdx+ -> (DuckDBIdx -> (DuckDBV2ValueHandle -> IO ()) -> IO ())+ -> (DuckDBV2VectorHandle -> IO a)+ -> IO a+withDefaultCastVector factory logicalType count writeRow action =+ withOwnedVector factory logicalType $ \vector -> do+ checked (c_duckdb_v2_vector_set_size vector count)+ forM_ (indices count) $ \row -> writeRow row (checked . c_duckdb_v2_vector_set_value vector row)+ action vector++-- | List the indices below a native count without unsigned underflow.+indices :: DuckDBIdx -> [DuckDBIdx]+indices count = if count == 0 then [] else [0 .. count - 1]++-- | Lend UTF-8 text to DuckDB for the duration of the action.+withText :: Text -> (ConstPtr DuckDBV2Str -> IO a) -> IO a+withText value = withBytes (Text.encodeUtf8 value)++-- | Lend an array of UTF-8 views to DuckDB for the duration of the action.+withTextArray :: [Text] -> (Ptr DuckDBV2Str -> IO a) -> IO a+withTextArray values action =+ withMany withText values $ \views -> mapM (\(ConstPtr view) -> peek view) views >>= (`withArray` action)++-- | Copy the bytes of a UTF-8 view whose memory DuckDB owns. The result stays+-- valid after DuckDB frees that memory.+readText :: DuckDBV2Str -> IO Text+readText view = Text.decodeUtf8 <$> readBytes view++-- | Copy the native text of a value.+valueText :: DuckDBV2ValueHandle -> IO Text+valueText value = do+ count <- output (c_duckdb_v2_value_to_string value nullPtr 0)+ when (toInteger count >= toInteger (maxBound :: Int)) $ throwInternalError "value text exceeds Int range"+ allocaBytes (fromIntegral count + 1) $ \buffer -> do+ _ <- output (c_duckdb_v2_value_to_string value buffer (count + 1))+ Text.decodeUtf8 <$> BS.packCStringLen (buffer, fromIntegral count)++-- | Lend bytes to DuckDB through a length-delimited view for the duration of the action.+withBytes :: BS.ByteString -> (ConstPtr DuckDBV2Str -> IO a) -> IO a+withBytes value action = BS.useAsCStringLen value $ \(ptr, count) ->+ with (DuckDBV2Str (ConstPtr ptr) (fromIntegral count)) (action . ConstPtr)++-- | Copy the bytes of a view whose memory DuckDB owns, including interior NUL+-- bytes. The result stays valid after DuckDB frees that memory.+readBytes :: DuckDBV2Str -> IO BS.ByteString+readBytes (DuckDBV2Str (ConstPtr ptr) count)+ | count == 0 = pure BS.empty+ | otherwise = BS.packCStringLen (ptr, fromIntegral count)++-- | Lend a UTF-8 identifier to DuckDB for the duration of the action.+withIdentifier :: Text -> (ConstPtr DuckDBV2IdentifierT -> IO a) -> IO a+withIdentifier value action = withText value (action . coerce)++-- | Lend an array of UTF-8 identifiers to DuckDB for the duration of the action.+withIdentifierArray :: [Text] -> (Ptr DuckDBV2IdentifierT -> IO a) -> IO a+withIdentifierArray values action = withTextArray values (action . castPtr)++-- | Copy the bytes of an identifier whose memory DuckDB owns. The result stays+-- valid after DuckDB frees that memory.+readIdentifier :: DuckDBV2IdentifierT -> IO Text+readIdentifier = readText . coerce++-- | Fold ASCII letters to lower case, as DuckDB does for identifiers.+foldIdentifier :: Text -> Text+foldIdentifier = Text.map (\char -> if isAsciiUpper char then toLower char else char)++{- Note [SQL text of the library]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+The library runs some SQL statements of its own. It builds each statement from+constant text, and it gives each value of a user to DuckDB as a parameter.+Thus a value of a user cannot change the statement. These are the statements:++- withTransaction in "Database.DuckDB.Simple": BEGIN TRANSACTION, COMMIT, and+ ROLLBACK. The statements are constant.++- catalogTypeName in "Database.DuckDB.Simple.Catalog": a SELECT from+ duckdb_databases(). The database name is a parameter.++- lookupCatalogEntry in "Database.DuckDB.Simple.Catalog": a SELECT from a+ duckdb_*() table function. The module selects the table function and the+ column from constants. The catalog name, the schema name, and the entry name+ are parameters.++- queryGeometryType in "Database.DuckDB.Simple.V2.LogicalRep": a constant+ SELECT with a fixed WKB literal and one parameter for the CRS name.++- metadataStatement in "Database.DuckDB.Simple.VariantCodec": a constant+ SELECT. The list of VARIANT values is a parameter.++- extractionStatement in "Database.DuckDB.Simple.VariantCodec": a SELECT with+ one placeholder for each object key. The library makes the placeholders from+ the number of keys. The list of VARIANT values and each key are parameters.++- deleteFunction in "Database.DuckDB.Simple.Function": DROP FUNCTION IF EXISTS.+ This statement is the exception. DROP FUNCTION takes no parameter for the+ name of the function, so the name goes into the text. The library parses the+ name with duckdb_v2_qname_parse, the rules of DuckDB for a qualified name,+ and quoteIdentifier quotes each part of the result.++quoteIdentifier puts an identifier in double quotes, and it doubles each double+quote in the identifier. Thus DuckDB reads the identifier as one name, also+when it contains a quote, a dot, or a semicolon. Use quoteIdentifier for each+identifier that the library puts into SQL text, and add each new statement to+this list.+-}++-- | Put an SQL identifier in double quotes, and double each double quote in+-- it. See Note [SQL text of the library].+quoteIdentifier :: Text -> Text+quoteIdentifier identifier = "\"" <> Text.replace "\"" "\"\"" identifier <> "\""++-- | Raise the error of an input check: an argument of a library function, or a+-- parameter value, that the library rejects. The error is an t'SQLError' of the+-- type 'ErrorInputInvalid' with no query. The checks of arguments use this+-- function, and so do the checks of a parameter value of the wrong shape or+-- type, for example a UNION tag that is not a member, a leaf of another type+-- than its declaration, a STRUCT field that does not match its type, a DECIMAL+-- width outside 1 to 38, or an empty Array, also where DuckDB gives another+-- error type for the same input. A value that its SQL type cannot hold uses+-- 'throwConversionError'. An error that is not about an input, for example a+-- state error or a native failure, does not use this function.+throwInvalidArgument :: Text -> IO a+throwInvalidArgument message = throwIO (SQLError ("duckdb-simple: " <> message) (Just ErrorInputInvalid) Nothing Nothing)++-- | Raise the error of an input check for a value that its SQL type cannot+-- hold. The error is an t'SQLError' of the type 'ErrorTypeConversion' with no+-- query, as the "Conversion Error" of DuckDB for the same value in SQL, for+-- example for @300::TINYINT@, @'25:00:00'::TIME@, or @\'x\'::ENUM(\'a\', \'b\')@.+-- These checks use this function: an integer outside its parameter type or+-- its declared type, a HUGEINT or UHUGEINT value outside its type, a DECIMAL+-- value that its width cannot hold, a date or a timestamp outside the range of+-- DuckDB, a nanosecond timestamp outside its storage, a TIMETZ offset outside+-- -57599 to 57599 seconds, a time of day that is not a TIME value, and an ENUM+-- label that is not in the dictionary of its type.+throwConversionError :: Text -> IO a+throwConversionError message = throwIO (SQLError ("duckdb-simple: " <> message) (Just ErrorTypeConversion) Nothing Nothing)++-- | Raise the error of a check of the state of the library, or of a vector, a+-- type, or a value that DuckDB gave: an t'SQLError' of the type+-- 'ErrorRuntimeInternal' with no query. Such an error means a defect of the+-- library or of DuckDB, not a defect of an input. See+-- Note [Internal errors in callbacks].+throwInternalError :: Text -> IO a+throwInternalError message = throwIO (SQLError ("duckdb-simple: " <> message) (Just ErrorRuntimeInternal) Nothing Nothing)++-- | The 'displayException' text of the exception in a t'SomeException'. In+-- base 4.20 (GHC 9.10), the 'displayException' of t'SomeException' adds a+-- newline and the context of the exception, also when the context is empty.+-- The text of an error must be the same with each GHC version, so the+-- library shows the inner exception.+exceptionText :: SomeException -> String+exceptionText (SomeException failure) = displayException failure++{- Note [Internal errors in callbacks]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+throwInternalError can raise its SQLError inside a DuckDB callback, for+example in a decoder check of an argument vector or in a state check of an+aggregate. runCallback in "Database.DuckDB.Simple.Callback" writes the code+of the type of an SQLError into the error slot of a callback, but it holds+back ErrorRuntimeInternal and ErrorRuntimeFatal and writes the code+DuckDBV2ErrorApi and the message for them. See+Note [Errors that leave a callback] in "Database.DuckDB.Simple.Callback". The+type ErrorRuntimeInternal, and ErrorRuntimeFatal, must never go into the error+slot of a callback, because DuckDB 2.0.0 then invalidates the database, and+each later statement raises "FATAL Error: ... database has been invalidated".+This holds for a callback at execution time, for example a scalar function, a+table function, a cast, or COPY FROM; at bind time, for example in a+replacement scan or a query pragma, the database stays usable. The rule+applies at bind time too.+-}++-- | Borrow a factory while its connection remains alive. Do not destroy it.+withConnectionFactory :: DuckDBV2ConnectionHandle -> (DuckDBV2FactoryHandle -> IO a) -> IO a+withConnectionFactory connection action = output (c_duckdb_v2_connection_get_factory connection) >>= action++-- | Read one checked output from a call that borrows the connection's factory.+factoryOutput+ :: (Storable a) => DuckDBV2ConnectionHandle -> (DuckDBV2FactoryHandle -> Ptr a -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a+factoryOutput connection create = withConnectionFactory connection (output . create)++-- | Parse trusted SQL and borrow its first statement.+withSQLStatement :: DuckDBV2ConnectionHandle -> Text -> (DuckDBV2SqlStatementHandle -> IO a) -> IO a+withSQLStatement connection sql action =+ TextForeign.withCString sql $ \source ->+ bracket (output (c_duckdb_v2_parse_sql connection (ConstPtr source))) (destroy c_duckdb_v2_statement_iterator_destroy) $ \iterator ->+ bracket (output (c_duckdb_v2_statement_iterator_next iterator)) (destroy c_duckdb_v2_sql_statement_destroy) action++-- | Borrow a factory during its native callback. Do not retain or destroy it.+withContextFactory :: DuckDBV2ContextHandle -> (DuckDBV2FactoryHandle -> IO a) -> IO a+withContextFactory context action = output (c_duckdb_v2_context_get_factory context) >>= action++-- | Borrow an instance factory for built-in types, casts, and allocations.+withInstanceFactory :: DuckDBV2InstanceHandle -> (DuckDBV2FactoryHandle -> IO a) -> IO a+withInstanceFactory instance_ action = output (c_duckdb_v2_instance_get_factory instance_) >>= action++-- | The native factory that makes values and types. A connection, a callback+-- context, and an instance each lend one. The name has no C API version: a later+-- version changes only this synonym.+type ValueFactory = DuckDBV2FactoryHandle++-- | An owned native value. Destroy it after use. The name has no C API+-- version: a later version changes only this synonym.+type NativeValue = DuckDBV2ValueHandle++-- | The scope of a configuration value. The default scope selects the scope that the option declares.+data SettingScope+ = SettingScopeDefault+ | SettingScopeGlobal+ | SettingScopeSession+ deriving (Eq, Ord, Show, Read, Enum, Bounded)++-- | Convert a setting scope to its native value.+settingScopeToNative :: SettingScope -> DuckDBV2SettingScope+settingScopeToNative = \case+ SettingScopeDefault -> DuckDBV2SettingScopeDefault+ SettingScopeGlobal -> DuckDBV2SettingScopeGlobal+ SettingScopeSession -> DuckDBV2SettingScopeSession++-- | Convert a native setting scope. An unknown value raises t'SQLError'.+settingScopeFromNative :: DuckDBV2SettingScope -> IO SettingScope+settingScopeFromNative = \case+ DuckDBV2SettingScopeDefault -> pure SettingScopeDefault+ DuckDBV2SettingScopeGlobal -> pure SettingScopeGlobal+ DuckDBV2SettingScopeSession -> pure SettingScopeSession+ DuckDBV2SettingScope value -> throwUnknownNative "setting scope" value++-- | The severity of a log entry.+data LogLevel+ = LogLevelTrace+ | LogLevelDebug+ | LogLevelInfo+ | LogLevelWarning+ | LogLevelError+ | LogLevelFatal+ deriving (Eq, Ord, Show, Read, Enum, Bounded)++-- | Convert a log level to its native value.+logLevelToNative :: LogLevel -> DuckDBV2LogLevel+logLevelToNative = \case+ LogLevelTrace -> DuckDBV2LogLevelTrace+ LogLevelDebug -> DuckDBV2LogLevelDebug+ LogLevelInfo -> DuckDBV2LogLevelInfo+ LogLevelWarning -> DuckDBV2LogLevelWarning+ LogLevelError -> DuckDBV2LogLevelError+ LogLevelFatal -> DuckDBV2LogLevelFatal++-- | Convert a native log level. An unknown value raises t'SQLError'.+logLevelFromNative :: DuckDBV2LogLevel -> IO LogLevel+logLevelFromNative = \case+ DuckDBV2LogLevelTrace -> pure LogLevelTrace+ DuckDBV2LogLevelDebug -> pure LogLevelDebug+ DuckDBV2LogLevelInfo -> pure LogLevelInfo+ DuckDBV2LogLevelWarning -> pure LogLevelWarning+ DuckDBV2LogLevelError -> pure LogLevelError+ DuckDBV2LogLevelFatal -> pure LogLevelFatal+ DuckDBV2LogLevel value -> throwUnknownNative "log level" value++-- | The context of a log entry: the database, a connection, or a thread.+data LogScope+ = LogScopeDatabase+ | LogScopeConnection+ | LogScopeThread+ deriving (Eq, Ord, Show, Read, Enum, Bounded)++-- | Convert a native log scope. An unknown value raises t'SQLError'.+logScopeFromNative :: DuckDBV2LogScope -> IO LogScope+logScopeFromNative = \case+ DuckDBV2LogScopeDatabase -> pure LogScopeDatabase+ DuckDBV2LogScopeConnection -> pure LogScopeConnection+ DuckDBV2LogScopeThread -> pure LogScopeThread+ DuckDBV2LogScope value -> throwUnknownNative "log scope" value++-- | The kind of a SQL token.+data TokenType+ = TokenTypeInvalid+ | TokenTypeKeyword+ | TokenTypeIdentifier+ | TokenTypeStringLiteral+ | TokenTypeNumberLiteral+ | TokenTypeOperator+ | TokenTypeComment+ | TokenTypeTerminator+ | TokenTypeEndOfInput+ deriving (Eq, Ord, Show, Read, Enum, Bounded)++-- | Convert a native token type. An unknown value raises t'SQLError'.+tokenTypeFromNative :: DuckDBV2TokenType -> IO TokenType+tokenTypeFromNative = \case+ DuckDBV2TokenTypeInvalid -> pure TokenTypeInvalid+ DuckDBV2TokenTypeKeyword -> pure TokenTypeKeyword+ DuckDBV2TokenTypeIdentifier -> pure TokenTypeIdentifier+ DuckDBV2TokenTypeStringLiteral -> pure TokenTypeStringLiteral+ DuckDBV2TokenTypeNumberLiteral -> pure TokenTypeNumberLiteral+ DuckDBV2TokenTypeOperator -> pure TokenTypeOperator+ DuckDBV2TokenTypeComment -> pure TokenTypeComment+ DuckDBV2TokenTypeTerminator -> pure TokenTypeTerminator+ DuckDBV2TokenTypeEndOfInput -> pure TokenTypeEndOfInput+ DuckDBV2TokenType value -> throwUnknownNative "token type" value++-- | The kind of a macro: a scalar expression or a table SELECT.+data MacroFunctionType+ = MacroFunctionTypeScalar+ | MacroFunctionTypeTable+ deriving (Eq, Ord, Show, Read, Enum, Bounded)++-- | Convert a macro kind to its native value.+macroFunctionTypeToNative :: MacroFunctionType -> DuckDBV2MacroFunctionType+macroFunctionTypeToNative = \case+ MacroFunctionTypeScalar -> DuckDBV2MacroFunctionTypeScalar+ MacroFunctionTypeTable -> DuckDBV2MacroFunctionTypeTable++-- | The id of a logical type. The id gives the kind of the type, for example+-- BIGINT or STRUCT. It does not give the parameters of the type.+data LogicalTypeId+ = LogicalTypeIdInvalid+ | LogicalTypeIdSQLNull+ | LogicalTypeIdUnknown+ | LogicalTypeIdAny+ | LogicalTypeIdType+ | LogicalTypeIdBoolean+ | LogicalTypeIdTinyInt+ | LogicalTypeIdSmallInt+ | LogicalTypeIdInteger+ | LogicalTypeIdBigInt+ | LogicalTypeIdDate+ | LogicalTypeIdTime+ | LogicalTypeIdTimestampSec+ | LogicalTypeIdTimestampMs+ | LogicalTypeIdTimestamp+ | LogicalTypeIdTimestampNs+ | LogicalTypeIdDecimal+ | LogicalTypeIdFloat+ | LogicalTypeIdDouble+ | LogicalTypeIdVarchar+ | LogicalTypeIdBlob+ | LogicalTypeIdInterval+ | LogicalTypeIdUTinyInt+ | LogicalTypeIdUSmallInt+ | LogicalTypeIdUInteger+ | LogicalTypeIdUBigInt+ | LogicalTypeIdTimestampTz+ | LogicalTypeIdTimestampTzNs+ | LogicalTypeIdTimeTz+ | LogicalTypeIdTimeNs+ | LogicalTypeIdBit+ | LogicalTypeIdBigNum+ | LogicalTypeIdUHugeInt+ | LogicalTypeIdHugeInt+ | LogicalTypeIdUUID+ | LogicalTypeIdGeometry+ | LogicalTypeIdStruct+ | LogicalTypeIdList+ | LogicalTypeIdMap+ | LogicalTypeIdEnum+ | LogicalTypeIdUnion+ | LogicalTypeIdArray+ | LogicalTypeIdVariant+ | LogicalTypeIdTuple+ deriving (Eq, Ord, Show, Read, Enum, Bounded)++-- | Convert a logical type id to its native value.+logicalTypeIdToNative :: LogicalTypeId -> DuckDBV2LogicalTypeId+logicalTypeIdToNative = \case+ LogicalTypeIdInvalid -> DuckDBV2LogicalTypeIdInvalid+ LogicalTypeIdSQLNull -> DuckDBV2LogicalTypeIdSQLNull+ LogicalTypeIdUnknown -> DuckDBV2LogicalTypeIdUnknown+ LogicalTypeIdAny -> DuckDBV2LogicalTypeIdAny+ LogicalTypeIdType -> DuckDBV2LogicalTypeIdType+ LogicalTypeIdBoolean -> DuckDBV2LogicalTypeIdBoolean+ LogicalTypeIdTinyInt -> DuckDBV2LogicalTypeIdTinyInt+ LogicalTypeIdSmallInt -> DuckDBV2LogicalTypeIdSmallInt+ LogicalTypeIdInteger -> DuckDBV2LogicalTypeIdInteger+ LogicalTypeIdBigInt -> DuckDBV2LogicalTypeIdBigInt+ LogicalTypeIdDate -> DuckDBV2LogicalTypeIdDate+ LogicalTypeIdTime -> DuckDBV2LogicalTypeIdTime+ LogicalTypeIdTimestampSec -> DuckDBV2LogicalTypeIdTimestampSec+ LogicalTypeIdTimestampMs -> DuckDBV2LogicalTypeIdTimestampMs+ LogicalTypeIdTimestamp -> DuckDBV2LogicalTypeIdTimestamp+ LogicalTypeIdTimestampNs -> DuckDBV2LogicalTypeIdTimestampNs+ LogicalTypeIdDecimal -> DuckDBV2LogicalTypeIdDecimal+ LogicalTypeIdFloat -> DuckDBV2LogicalTypeIdFloat+ LogicalTypeIdDouble -> DuckDBV2LogicalTypeIdDouble+ LogicalTypeIdVarchar -> DuckDBV2LogicalTypeIdVarchar+ LogicalTypeIdBlob -> DuckDBV2LogicalTypeIdBlob+ LogicalTypeIdInterval -> DuckDBV2LogicalTypeIdInterval+ LogicalTypeIdUTinyInt -> DuckDBV2LogicalTypeIdUTinyInt+ LogicalTypeIdUSmallInt -> DuckDBV2LogicalTypeIdUSmallInt+ LogicalTypeIdUInteger -> DuckDBV2LogicalTypeIdUInteger+ LogicalTypeIdUBigInt -> DuckDBV2LogicalTypeIdUBigInt+ LogicalTypeIdTimestampTz -> DuckDBV2LogicalTypeIdTimestampTz+ LogicalTypeIdTimestampTzNs -> DuckDBV2LogicalTypeIdTimestampTzNs+ LogicalTypeIdTimeTz -> DuckDBV2LogicalTypeIdTimeTz+ LogicalTypeIdTimeNs -> DuckDBV2LogicalTypeIdTimeNs+ LogicalTypeIdBit -> DuckDBV2LogicalTypeIdBit+ LogicalTypeIdBigNum -> DuckDBV2LogicalTypeIdBigNum+ LogicalTypeIdUHugeInt -> DuckDBV2LogicalTypeIdUHugeInt+ LogicalTypeIdHugeInt -> DuckDBV2LogicalTypeIdHugeInt+ LogicalTypeIdUUID -> DuckDBV2LogicalTypeIdUUID+ LogicalTypeIdGeometry -> DuckDBV2LogicalTypeIdGeometry+ LogicalTypeIdStruct -> DuckDBV2LogicalTypeIdStruct+ LogicalTypeIdList -> DuckDBV2LogicalTypeIdList+ LogicalTypeIdMap -> DuckDBV2LogicalTypeIdMap+ LogicalTypeIdEnum -> DuckDBV2LogicalTypeIdEnum+ LogicalTypeIdUnion -> DuckDBV2LogicalTypeIdUnion+ LogicalTypeIdArray -> DuckDBV2LogicalTypeIdArray+ LogicalTypeIdVariant -> DuckDBV2LogicalTypeIdVariant+ LogicalTypeIdTuple -> DuckDBV2LogicalTypeIdTuple++-- | Convert a native logical type id. An unknown value raises t'SQLError'.+logicalTypeIdFromNative :: DuckDBV2LogicalTypeId -> IO LogicalTypeId+logicalTypeIdFromNative = \case+ DuckDBV2LogicalTypeIdInvalid -> pure LogicalTypeIdInvalid+ DuckDBV2LogicalTypeIdSQLNull -> pure LogicalTypeIdSQLNull+ DuckDBV2LogicalTypeIdUnknown -> pure LogicalTypeIdUnknown+ DuckDBV2LogicalTypeIdAny -> pure LogicalTypeIdAny+ DuckDBV2LogicalTypeIdType -> pure LogicalTypeIdType+ DuckDBV2LogicalTypeIdBoolean -> pure LogicalTypeIdBoolean+ DuckDBV2LogicalTypeIdTinyInt -> pure LogicalTypeIdTinyInt+ DuckDBV2LogicalTypeIdSmallInt -> pure LogicalTypeIdSmallInt+ DuckDBV2LogicalTypeIdInteger -> pure LogicalTypeIdInteger+ DuckDBV2LogicalTypeIdBigInt -> pure LogicalTypeIdBigInt+ DuckDBV2LogicalTypeIdDate -> pure LogicalTypeIdDate+ DuckDBV2LogicalTypeIdTime -> pure LogicalTypeIdTime+ DuckDBV2LogicalTypeIdTimestampSec -> pure LogicalTypeIdTimestampSec+ DuckDBV2LogicalTypeIdTimestampMs -> pure LogicalTypeIdTimestampMs+ DuckDBV2LogicalTypeIdTimestamp -> pure LogicalTypeIdTimestamp+ DuckDBV2LogicalTypeIdTimestampNs -> pure LogicalTypeIdTimestampNs+ DuckDBV2LogicalTypeIdDecimal -> pure LogicalTypeIdDecimal+ DuckDBV2LogicalTypeIdFloat -> pure LogicalTypeIdFloat+ DuckDBV2LogicalTypeIdDouble -> pure LogicalTypeIdDouble+ DuckDBV2LogicalTypeIdVarchar -> pure LogicalTypeIdVarchar+ DuckDBV2LogicalTypeIdBlob -> pure LogicalTypeIdBlob+ DuckDBV2LogicalTypeIdInterval -> pure LogicalTypeIdInterval+ DuckDBV2LogicalTypeIdUTinyInt -> pure LogicalTypeIdUTinyInt+ DuckDBV2LogicalTypeIdUSmallInt -> pure LogicalTypeIdUSmallInt+ DuckDBV2LogicalTypeIdUInteger -> pure LogicalTypeIdUInteger+ DuckDBV2LogicalTypeIdUBigInt -> pure LogicalTypeIdUBigInt+ DuckDBV2LogicalTypeIdTimestampTz -> pure LogicalTypeIdTimestampTz+ DuckDBV2LogicalTypeIdTimestampTzNs -> pure LogicalTypeIdTimestampTzNs+ DuckDBV2LogicalTypeIdTimeTz -> pure LogicalTypeIdTimeTz+ DuckDBV2LogicalTypeIdTimeNs -> pure LogicalTypeIdTimeNs+ DuckDBV2LogicalTypeIdBit -> pure LogicalTypeIdBit+ DuckDBV2LogicalTypeIdBigNum -> pure LogicalTypeIdBigNum+ DuckDBV2LogicalTypeIdUHugeInt -> pure LogicalTypeIdUHugeInt+ DuckDBV2LogicalTypeIdHugeInt -> pure LogicalTypeIdHugeInt+ DuckDBV2LogicalTypeIdUUID -> pure LogicalTypeIdUUID+ DuckDBV2LogicalTypeIdGeometry -> pure LogicalTypeIdGeometry+ DuckDBV2LogicalTypeIdStruct -> pure LogicalTypeIdStruct+ DuckDBV2LogicalTypeIdList -> pure LogicalTypeIdList+ DuckDBV2LogicalTypeIdMap -> pure LogicalTypeIdMap+ DuckDBV2LogicalTypeIdEnum -> pure LogicalTypeIdEnum+ DuckDBV2LogicalTypeIdUnion -> pure LogicalTypeIdUnion+ DuckDBV2LogicalTypeIdArray -> pure LogicalTypeIdArray+ DuckDBV2LogicalTypeIdVariant -> pure LogicalTypeIdVariant+ DuckDBV2LogicalTypeIdTuple -> pure LogicalTypeIdTuple+ DuckDBV2LogicalTypeId value -> throwUnknownNative "logical type id" value++-- | The name of a type id in DuckDB 2.0.0. duckdb_v2_logical_type_get_name+-- gives this name for a type with no alias. See Note [Type aliases] in+-- "Database.DuckDB.Simple.LogicalRep".+logicalTypeIdName :: LogicalTypeId -> Text+logicalTypeIdName = \case+ LogicalTypeIdInvalid -> "INVALID"+ LogicalTypeIdSQLNull -> "NULL"+ LogicalTypeIdUnknown -> "UNKNOWN"+ LogicalTypeIdAny -> "ANY"+ LogicalTypeIdType -> "TYPE"+ LogicalTypeIdBoolean -> "BOOLEAN"+ LogicalTypeIdTinyInt -> "TINYINT"+ LogicalTypeIdSmallInt -> "SMALLINT"+ LogicalTypeIdInteger -> "INTEGER"+ LogicalTypeIdBigInt -> "BIGINT"+ LogicalTypeIdDate -> "DATE"+ LogicalTypeIdTime -> "TIME"+ LogicalTypeIdTimestampSec -> "TIMESTAMP_S"+ LogicalTypeIdTimestampMs -> "TIMESTAMP_MS"+ LogicalTypeIdTimestamp -> "TIMESTAMP"+ LogicalTypeIdTimestampNs -> "TIMESTAMP_NS"+ LogicalTypeIdDecimal -> "DECIMAL"+ LogicalTypeIdFloat -> "FLOAT"+ LogicalTypeIdDouble -> "DOUBLE"+ LogicalTypeIdVarchar -> "VARCHAR"+ LogicalTypeIdBlob -> "BLOB"+ LogicalTypeIdInterval -> "INTERVAL"+ LogicalTypeIdUTinyInt -> "UTINYINT"+ LogicalTypeIdUSmallInt -> "USMALLINT"+ LogicalTypeIdUInteger -> "UINTEGER"+ LogicalTypeIdUBigInt -> "UBIGINT"+ LogicalTypeIdTimestampTz -> "TIMESTAMP WITH TIME ZONE"+ LogicalTypeIdTimestampTzNs -> "TIMESTAMPTZ_NS"+ LogicalTypeIdTimeTz -> "TIME WITH TIME ZONE"+ LogicalTypeIdTimeNs -> "TIME_NS"+ LogicalTypeIdBit -> "BIT"+ LogicalTypeIdBigNum -> "BIGNUM"+ LogicalTypeIdUHugeInt -> "UHUGEINT"+ LogicalTypeIdHugeInt -> "HUGEINT"+ LogicalTypeIdUUID -> "UUID"+ LogicalTypeIdGeometry -> "GEOMETRY"+ LogicalTypeIdStruct -> "STRUCT"+ LogicalTypeIdList -> "LIST"+ LogicalTypeIdMap -> "MAP"+ LogicalTypeIdEnum -> "ENUM"+ LogicalTypeIdUnion -> "UNION"+ LogicalTypeIdArray -> "ARRAY"+ LogicalTypeIdVariant -> "VARIANT"+ LogicalTypeIdTuple -> "TUPLE"++-- | A behaviour of a file open. The flags are not a bitmask. Each flag in the+-- list adds one behaviour.+data FileFlag+ = -- | Open the file for reading.+ FileFlagRead+ | -- | Open the file for writing.+ FileFlagWrite+ | -- | Create the file if it does not exist. Keep the contents of an existing file.+ FileFlagCreate+ | -- | Create the file if it does not exist. Truncate an existing file to empty.+ FileFlagCreateNew+ | -- | Open the file in append mode.+ FileFlagAppend+ | -- | Fail if the file exists. Use it only together with 'FileFlagCreate'.+ FileFlagExclusiveCreate+ | -- | Read and write at explicit offsets from more than one thread at the same time.+ FileFlagParallelAccess+ | -- | Read the file through the external file cache of DuckDB. Use it only for reading.+ FileFlagExternalFileCache+ deriving (Eq, Ord, Show, Read, Enum, Bounded)++-- | Convert a file flag to its native value.+fileFlagToNative :: FileFlag -> DuckDBV2FileFlag+fileFlagToNative = \case+ FileFlagRead -> DuckDBV2FileFlagRead+ FileFlagWrite -> DuckDBV2FileFlagWrite+ FileFlagCreate -> DuckDBV2FileFlagCreate+ FileFlagCreateNew -> DuckDBV2FileFlagCreateNew+ FileFlagAppend -> DuckDBV2FileFlagAppend+ FileFlagExclusiveCreate -> DuckDBV2FileFlagExclusiveCreate+ FileFlagParallelAccess -> DuckDBV2FileFlagParallelAccess+ FileFlagExternalFileCache -> DuckDBV2FileFlagExternalFileCache++-- | Raise t'SQLError' for a native enum value that has no Haskell constructor.+throwUnknownNative :: Text -> CUInt -> IO a+throwUnknownNative kind value = throwInternalError ("unknown native " <> kind <> " " <> Text.pack (show value))
src/Database/DuckDB/Simple/Ok.hs view
@@ -1,18 +1,17 @@ {-# LANGUAGE DeriveFunctor #-} {-# LANGUAGE DerivingStrategies #-} -{- |-Module : Database.DuckDB.Simple.Ok-Description : Lightweight error accumulation for row parsing.--The @Ok@ type mirrors the helper used by @sqlite-simple@: it behaves like an-@Either [SomeException]@ with sensible @Alternative@ semantics that accumulate-failure reasons instead of discarding them. This underpins the row parser-machinery in @Database.DuckDB.Simple.FromRow@.--}+-- |+-- Module : Database.DuckDB.Simple.Ok+-- Description : Accumulate errors from row parsing.+--+-- The @Ok@ type mirrors the helper used by @sqlite-simple@. It behaves like+-- @Either [SomeException]@. Its @Alternative@ instance keeps the errors when+-- both alternatives fail. @Database.DuckDB.Simple.Conversion.runConversion@+-- returns it after conversion effects and diagnostics have completed. module Database.DuckDB.Simple.Ok (- Ok (..),- ManyErrors (..),+ Ok (..),+ ManyErrors (..), ) where import Control.Applicative (Alternative (..))@@ -20,45 +19,45 @@ import Control.Monad (MonadPlus (..)) import qualified Control.Monad.Fail as Fail --- | Simple success-or-errors container with error accumulation.+-- | A value or a list of errors. data Ok a- = Errors [SomeException]- | Ok !a- deriving stock (Show)- deriving stock (Functor)+ = Errors [SomeException]+ | Ok !a+ deriving stock (Show)+ deriving stock (Functor) -- | Two failures are considered equal regardless of their payload. instance (Eq a) => Eq (Ok a) where- Errors _ == Errors _ = True- Ok a == Ok b = a == b- _ == _ = False+ Errors _ == Errors _ = True+ Ok a == Ok b = a == b+ _ == _ = False instance Applicative Ok where- pure = Ok- Ok f <*> Ok a = Ok (f a)- Errors es <*> Ok _ = Errors es- Ok _ <*> Errors es = Errors es- Errors es1 <*> Errors es2 = Errors (es1 <> es2)+ pure = Ok+ Ok f <*> Ok a = Ok (f a)+ Errors es <*> Ok _ = Errors es+ Ok _ <*> Errors es = Errors es+ Errors es1 <*> Errors es2 = Errors (es1 <> es2) instance Alternative Ok where- empty = Errors []- Ok x <|> _ = Ok x- Errors _ <|> Ok y = Ok y- Errors xs <|> Errors ys = Errors (xs <> ys)+ empty = Errors []+ Ok x <|> _ = Ok x+ Errors _ <|> Ok y = Ok y+ Errors xs <|> Errors ys = Errors (xs <> ys) instance Monad Ok where- Ok a >>= f = f a- Errors es >>= _ = Errors es+ Ok a >>= f = f a+ Errors es >>= _ = Errors es instance MonadPlus Ok where- mzero = empty- mplus = (<|>)+ mzero = empty+ mplus = (<|>) instance Fail.MonadFail Ok where- fail msg = Errors [toException (userError msg)]+ fail msg = Errors [toException (userError msg)] --- | Bundle multiple underlying exceptions into a single throwable value.+-- | Store multiple exceptions in one exception. newtype ManyErrors = ManyErrors [SomeException]- deriving stock (Show)+ deriving stock (Show) instance Exception ManyErrors
+ src/Database/DuckDB/Simple/Pragma.hs view
@@ -0,0 +1,13 @@+-- |+-- Module : Database.DuckDB.Simple.Pragma+-- Description : Register PRAGMA functions with typed Haskell callbacks.+--+-- A callback gets the arguments of the call as one row, decoded with+-- t'Database.DuckDB.Simple.FromRow.FromRow'.+-- "Database.DuckDB.Simple.V2.Pragma" has the callbacks that get native handles.+module Database.DuckDB.Simple.Pragma (+ PragmaCallback (..),+ registerPragmaFunction,+) where++import Database.DuckDB.Simple.Pragma.Internal (PragmaCallback (..), registerPragmaFunction)
+ src/Database/DuckDB/Simple/Pragma/Internal.hs view
@@ -0,0 +1,103 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}++-- |+-- Module : Database.DuckDB.Simple.Pragma.Internal+-- Description : Share the typed PRAGMA registration for connections and extensions.+--+-- The cabal file lists this hidden module in @other-modules@.+-- 'registerPragmaFunctionOn' takes a registration target.+-- "Database.DuckDB.Simple.Pragma" exports the connection form.+-- "Database.DuckDB.Simple.Extension" exports the extension form.+module Database.DuckDB.Simple.Pragma.Internal (+ PragmaCallback (..),+ registerPragmaFunction,+ registerPragmaFunctionOn,+) where++import Control.Exception (throwIO)+import Control.Monad (forM, when)+import Data.Maybe (fromMaybe)+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple.Callback (targetCastFactory)+import Database.DuckDB.Simple.Conversion (runConversion)+import Database.DuckDB.Simple.FromRow (FromRow (..), parseRow, rowErrorsToSqlError)+import Database.DuckDB.Simple.Internal (Connection)+import Database.DuckDB.Simple.Materialize (callbackReadFactory)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.Ok (Ok (..))+import Database.DuckDB.Simple.TypeScope (TypeScope, withTypeScope)+import Database.DuckDB.Simple.Typed (readBoundField)+import qualified Database.DuckDB.Simple.V2.Pragma as V2+import Database.DuckDB.Simple.V2.Registration (+ FunctionParameter,+ RegistrationTarget (..),+ )+import Foreign.Marshal.Array (advancePtr, allocaArray, peekArray)++-- | The callback of a pragma. The argument @args@ is the row of the+-- arguments of the call. A query callback gives SQL. DuckDB runs that SQL in+-- place of the PRAGMA statement. An execution callback does an action and gives+-- no rows. An exception becomes an SQL error.+data PragmaCallback args+ = QueryPragma (args -> IO Text)+ | ExecPragma (args -> IO ())++-- | Register a pragma on a connection. The pragma has no return type.+--+-- The row has one field for each argument, in this order: the positional-only+-- parameters, the standard parameters, the arguments of the positional variadic+-- parameter, the named-only parameters, and the arguments of the named variadic+-- parameter. Parameters of one kind keep the order of the declaration. The+-- arguments of a variadic parameter keep the order of the call. Thus, for+-- parameters from 'Database.DuckDB.Simple.Registration.parameter', the row has+-- one field for each parameter, in the order of the declaration.+--+-- A call gives a standard parameter by position or by name, for example+-- @PRAGMA p(1, label = \'x\')@. The field is at the position of the parameter in+-- both cases. When the call omits a parameter, the field has the default value of+-- the parameter. A NULL argument needs a 'Maybe' field. When the row does not+-- decode, the call fails with an SQL error that names the pragma.+--+-- > registerPragmaFunction conn "repeat_answer" [parameter "times" (TypeExpression "BIGINT")] $+-- > QueryPragma \(Only (times :: Int64)) -> pure ("SELECT 42 FROM range(" <> Text.pack (show times) <> ")")+registerPragmaFunction :: (FromRow args) => Connection -> Text -> [FunctionParameter] -> PragmaCallback args -> IO ()+registerPragmaFunction = registerPragmaFunctionOn . ConnectionTarget++-- | Register a pragma on a connection or a loading extension.+-- 'registerPragmaFunction' gives the rules for the argument row.+registerPragmaFunctionOn :: (FromRow args) => RegistrationTarget -> Text -> [FunctionParameter] -> PragmaCallback args -> IO ()+registerPragmaFunctionOn target name parameters callback = do+ -- Check the name before the read of the state of the target, as each+ -- registration does. 'V2.registerPragmaFunctionOn' checks it again.+ when (Text.null name || Text.any (== '\0') name) $+ throwInvalidArgument "pragma function name must be nonempty and must not contain NUL"+ casts <- targetCastFactory target+ V2.registerPragmaFunctionOn target name parameters case callback of+ QueryPragma action -> V2.QueryPragma \info context -> withTypeScope \scope -> pragmaArguments scope casts name info context >>= action+ ExecPragma action -> V2.ExecPragma \info context -> withTypeScope \scope -> pragmaArguments scope casts name info context >>= action++-- | Copy the arguments of a pragma call and decode them as one row. The first+-- argument is the name of the pragma for the error message.+pragmaArguments+ :: (FromRow args) => TypeScope -> Maybe DuckDBV2FactoryHandle -> Text -> DuckDBV2FunctionBindInfoHandle -> DuckDBV2ContextHandle -> IO args+pragmaArguments scope casts name info context = withContextFactory context \contextFactory -> do+ let factory = callbackReadFactory casts contextFactory+ count <- allocaArray 4 \counts -> do+ checked (c_duckdb_v2_function_bind_get_arg_count info counts (advancePtr counts 1) (advancePtr counts 2) (advancePtr counts 3))+ sum <$> peekArray 4 counts+ fields <- forM (indices count) \index -> do+ argument <- output (c_duckdb_v2_function_bind_get_arg_name info index) >>= readIdentifier+ readBoundField scope factory info argument index+ converted <- runConversion (parseRow fromRow fields)+ case converted of+ Ok args -> pure args+ Errors failures -> do+ let message = sqlErrorMessage (rowErrorsToSqlError (Query name) failures)+ throwIO+ ( userError+ ( Text.unpack+ ("duckdb-simple: cannot decode the arguments of pragma " <> name <> ": " <> fromMaybe message (Text.stripPrefix "duckdb-simple: " message))+ )+ )
+ src/Database/DuckDB/Simple/QueryMetrics.hs view
@@ -0,0 +1,68 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}++-- |+-- Module : Database.DuckDB.Simple.QueryMetrics+-- Description : Copied query profiles.+--+-- Returned values do not retain native snapshot handles.+module Database.DuckDB.Simple.QueryMetrics (+ QueryMetrics (..),+ enableProfiling,+ queryMetrics,+) where++import Control.Exception (bracket)+import Control.Monad (forM, when)+import Data.Text (Text)+import Database.DuckDB.Simple.FromField (FieldValue)+import Database.DuckDB.Simple.Internal (Connection, connectionInstance, instanceCastFactory, withConnectionHandle)+import Database.DuckDB.Simple.Materialize (ReadFactory (..), materializeValue, rejectRows)+import Database.DuckDB.Simple.Native++-- | One query or operator node. Metric order is unspecified. Any metric or+-- operator can be absent. Native V2 reports numeric metrics only.+-- Values and child nodes use Haskell-owned storage.+data QueryMetrics = QueryMetrics+ { metricsNodeName :: Text+ , metricsValues :: [(Text, FieldValue)]+ , metricsChildren :: [QueryMetrics]+ }+ deriving (Eq, Show)++-- | Enable or disable profiling for later queries on this connection.+-- Enabling sets the native @enable_profiling@ option to @no_output@. DuckDB then+-- collects the metrics and does not print a profile. The option has the+-- session scope, so another connection to the same database keeps its own+-- setting.+enableProfiling :: Connection -> Bool -> IO ()+enableProfiling connection enabled = withConnectionHandle connection \native -> do+ when enabled do+ config <- output (c_duckdb_v2_connection_get_config native)+ withIdentifier "enable_profiling" \name -> withText "no_output" \value ->+ checked (c_duckdb_v2_config_set_option_text config name value DuckDBV2SettingScopeDefault)+ checked (c_duckdb_v2_connection_enable_profiling native (if enabled then 1 else 0))++-- | Copy the profile of the last completed query. Drain or close any active+-- result first. The returned snapshot does not change when later queries run.+-- Without profiling, the root has no metrics and no children. The function+-- releases all native snapshot and value handles before it returns.+queryMetrics :: Connection -> IO QueryMetrics+queryMetrics connection = withConnectionHandle connection \native ->+ bracket (output (c_duckdb_v2_connection_get_query_metrics native)) (destroy c_duckdb_v2_query_metrics_destroy) \snapshot -> do+ casts <- instanceCastFactory (connectionInstance connection)+ root <- output (c_duckdb_v2_query_metrics_get_root snapshot)+ copyNode (materializeValue unexpectedVariant (InstanceFactory casts)) root+ where+ unexpectedVariant = rejectRows (throwInternalError "query metrics must be numeric")+ copyNode readValue node = do+ name <- output (c_duckdb_v2_query_metrics_node_get_name node) >>= readText+ metricCount <- output (c_duckdb_v2_query_metrics_node_get_metric_count node)+ metrics <- forM (indices metricCount) \index -> do+ key <- output (c_duckdb_v2_query_metrics_node_get_metric_name node index) >>= readText+ value <- bracket (output (c_duckdb_v2_query_metrics_node_get_metric_value node index)) (destroy c_duckdb_v2_value_destroy) readValue+ pure (key, value)+ childCount <- output (c_duckdb_v2_query_metrics_node_get_child_count node)+ children <- forM (indices childCount) \index ->+ output (c_duckdb_v2_query_metrics_node_get_child node index) >>= copyNode readValue+ pure (QueryMetrics name metrics children)
+ src/Database/DuckDB/Simple/Registration.hs view
@@ -0,0 +1,84 @@+-- |+-- Module : Database.DuckDB.Simple.Registration+-- Description : Typed registration of custom types, replacement scans, and shared declarations.+--+-- This module has the declarations that typed registrations share. It also+-- registers custom types and replacement scans on a connection or an instance.+-- Its signatures have no native handle. A t'FunctionParameter' from 'parameter' is+-- a required parameter. The V2 module has the field for the other parameter+-- kinds. "Database.DuckDB.Simple.V2.Registration" has the registrations with+-- native callbacks.+module Database.DuckDB.Simple.Registration (+ TypeDeclaration (..),+ DefaultValue,+ defaultValue,+ -- See Note [Abstract parameter selectors].+ FunctionParameter,+ parameterName,+ parameterType,+ parameterDefault,+ parameter,+ registerCustomType,+ Replacement (..),+ registerReplacementScan,+ registerInstanceReplacementScan,+) where++import Data.Text (Text)+import Database.DuckDB.Simple.Internal (Connection, Instance)+import Database.DuckDB.Simple.Typed (Replacement (..), replacementScanCallback)+import Database.DuckDB.Simple.V2.Registration (+ DefaultValue,+ FunctionParameter (..),+ RegistrationTarget (ConnectionTarget),+ TypeDeclaration (..),+ defaultValue,+ parameter,+ registerCustomType,+ )+import qualified Database.DuckDB.Simple.V2.Registration as V2++{- Note [Abstract parameter selectors]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+Haddock 2.33.0 omits selector declarations from bundled exports when the+constructor is hidden. Export the selectors separately so the index has+matching destinations. GHC retains the parent type of each selector.+Bundled imports and record updates still work.+-}++-- | Register a replacement scan on a connection. DuckDB calls the function+-- while it plans a query on this connection that reads a table that the catalog+-- does not have. Other connections of the instance do not use the scan.+--+-- The function gets the parts of the table name as the query writes them. A name+-- without a schema has one part. A qualified name has its catalog and schema+-- before the table name, for example @[\"memory\", \"main\", \"t\"]@. DuckDB+-- gives the parts also when the catalog or the schema does not exist. Each part+-- keeps the case of the query text. DuckDB removes the quotes of a quoted part. A+-- dot in a quoted part stays in that part. A file path in single quotes, for+-- example @\'data.csv\'@, is one part without the quotes.+--+-- 'Nothing' declines the name. DuckDB then asks the next scan. When no scan+-- replaces the name, the query fails with the usual error for an unknown table.+-- An exception in the function fails the query with an SQL error that contains+-- the message of the exception.+--+-- DuckDB asks the scans of the connection first, in the order of registration.+-- Then it asks the scans of the instance, which include the built-in scans for+-- file paths. The first scan that replaces the name wins. Thus a connection scan+-- can replace a file path. A scan stays registered until the connection closes.+-- No function removes a scan.+registerReplacementScan :: Connection -> ([Text] -> IO (Maybe Replacement)) -> IO ()+registerReplacementScan connection = V2.registerReplacementScanOn (ConnectionTarget connection) . replacementScanCallback++-- | Register a replacement scan on an instance. Each connection of the+-- instance uses the scan until the instance closes. No function removes a scan.+-- DuckDB asks the scans of a connection before the scans of its instance. It asks+-- the scans of the instance in the order of registration.+-- 'registerReplacementScan' gives the rules for the function.+--+-- Register the scan before a connection runs a query. Registration while another+-- connection plans a query is not thread safe. The function can run on several+-- connections at the same time.+registerInstanceReplacementScan :: Instance -> ([Text] -> IO (Maybe Replacement)) -> IO ()+registerInstanceReplacementScan instance' = V2.registerInstanceReplacementScan instance' . replacementScanCallback
src/Database/DuckDB/Simple/Result.hs view
@@ -3,225 +3,288 @@ {-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE TupleSections #-} --- | Shared result decoding and cursor ownership for both execution modes.+-- |+-- Module : Database.DuckDB.Simple.Result+-- Description : Shared result decoding and cursor ownership for both execution modes. module Database.DuckDB.Simple.Result (- collectRows,- foldStatementWith,- nextRowWith,- resetStatementStream,- cleanupStatementStreamRef,+ withRows,+ withParsedRows,+ foldStatementWith,+ nextRowWith,+ resetStatementStream, ) where -import Control.Exception (bracket, evaluate, finally, mask, mask_, onException, throwIO)-import Control.Monad (forM, when, zipWithM)-import Data.IORef (IORef, atomicModifyIORef', readIORef, writeIORef)+import Control.Exception (bracket, finally, mask, mask_, onException, throwIO)+import Control.Monad (foldM, forM, when, zipWithM)+import Data.IORef (IORef, atomicModifyIORef', newIORef, readIORef, writeIORef)+import qualified Data.IntMap.Strict as IntMap import qualified Data.Text as Text-import Database.DuckDB.FFI+import Database.DuckDB.Simple.Conversion (runConversion)+import Database.DuckDB.Simple.Element (withVectorType) import Database.DuckDB.Simple.FromField (Field (..), FieldValue) import Database.DuckDB.Simple.FromRow (RowParser, parseRow, rowErrorsToSqlError) import Database.DuckDB.Simple.Internal-import Database.DuckDB.Simple.Materialize (prepareVectorReader)+import Database.DuckDB.Simple.Materialize (+ FallbackReaders (..),+ ReadFactory (..),+ prepareGeometryReader,+ prepareVectorReaderWith,+ ) import Database.DuckDB.Simple.Ok (Ok (..))-import Foreign.Marshal.Alloc (free, malloc)-import Foreign.Marshal.Utils (fillBytes)-import Foreign.Ptr (Ptr, nullPtr)-import Foreign.Storable (sizeOf)+import Database.DuckDB.Simple.TypeScope (TypeScope, closeTypeScope, newTypeScope, scopedTypeRep)+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeToRep)+import Foreign.Marshal.Alloc (alloca)+import Foreign.Storable (peek) --- | Fold rows from one statement and release its result on every exit path.+-- | Fold rows from a statement owned by the enclosing @withStatement@. foldStatementWith :: ResultMode -> RowParser row -> Statement -> a -> (a -> row -> IO a) -> IO a-foldStatementWith mode parser stmt initial step =- let loop acc = do- nextVal <- nextRowWith mode parser stmt- case nextVal of- Nothing -> pure acc- Just row -> do- acc' <- step acc row- acc' `seq` loop acc'- in loop initial `finally` resetStatementStream stmt+foldStatementWith mode parser stmt initial step = loop initial+ where+ loop acc =+ nextRowWith mode parser stmt >>= \case+ Nothing -> pure acc+ Just row -> do+ acc' <- step acc row+ acc' `seq` loop acc' -- | Read one row. The first fetch selects execution until the cursor resets.+-- See Note [Statement cursor] in "Database.DuckDB.Simple.Internal". nextRowWith :: ResultMode -> RowParser r -> Statement -> IO (Maybe r) nextRowWith mode parser stmt@Statement{statementStream} =- withStatementHandle stmt \_ -> mask \restore -> do- state <- readIORef statementStream- case state of- StatementStreamExhausted -> pure Nothing- StatementStreamIdle -> do- newStream <- startStatementStream mode stmt- case newStream of- Nothing -> writeIORef statementStream StatementStreamExhausted >> pure Nothing- Just stream -> do- writeIORef statementStream (StatementStreamActive stream)- restore (consumeStream statementStream parser stmt stream)- `onException` exhaustStatementStream statementStream- StatementStreamActive stream ->- restore (consumeStream statementStream parser stmt stream)- `onException` exhaustStatementStream statementStream+ withStatementHandle stmt \_ -> mask \restore -> do+ let consume stream =+ restore (consumeStream statementStream parser stmt stream)+ `onException` exhaustStatementStream statementStream (statementStreamTypes stream)+ state <- readIORef statementStream+ case state of+ StatementStreamExhausted -> pure Nothing+ StatementStreamIdle -> do+ newStream <- startStatementStream mode stmt+ case newStream of+ Nothing -> writeIORef statementStream StatementStreamExhausted >> pure Nothing+ Just stream -> do+ writeIORef statementStream (StatementStreamActive stream)+ consume stream+ StatementStreamActive stream -> consume stream --- | Release an active cursor and allow the statement to execute again.+-- | Release an active cursor and allow the statement to execute again. Then+-- clear the schema cache of the statement. See Note [Statement schemas] in+-- "Database.DuckDB.Simple.Internal". resetStatementStream :: Statement -> IO ()-resetStatementStream Statement{statementStream} =- cleanupStatementStreamRef statementStream+resetStatementStream stmt@Statement{statementStream} =+ cleanupStatementStreamRef statementStream `finally` clearStatementSchemas stmt consumeStream :: IORef StatementStreamState -> RowParser r -> Statement -> StatementStream -> IO (Maybe r) consumeStream streamRef parser stmt stream = mask \restore -> do- loaded <- case statementStreamChunk stream of- Nothing -> fetchChunk (statementConnection stmt) (statementQuery stmt) stream- Just _ -> pure stream- writeIORef streamRef (StatementStreamActive loaded)- case statementStreamChunk loaded of- Nothing -> exhaustStatementStream streamRef >> pure Nothing- Just chunk -> do- fields <- restore $ buildMaterializedRow (statementStreamColumns loaded) (statementStreamChunkReaders chunk) (statementStreamChunkIndex chunk)- parsed <- restore (evaluate (parseRow parser fields))- case parsed of- Errors rowErr -> throwIO $ rowErrorsToSqlError (statementQuery stmt) rowErr- Ok value -> do- let nextIndex = statementStreamChunkIndex chunk + 1- if nextIndex < statementStreamChunkSize chunk- then writeIORef streamRef (StatementStreamActive loaded{statementStreamChunk = Just chunk{statementStreamChunkIndex = nextIndex}})- else do- writeIORef streamRef (StatementStreamActive loaded{statementStreamChunk = Nothing})- finalizeChunk chunk- pure (Just value)+ loaded <- case statementStreamChunk stream of+ Nothing -> restore (fetchChunk streamRef (statementConnection stmt) (statementQuery stmt) stream)+ Just _ -> pure stream+ writeIORef streamRef (StatementStreamActive loaded)+ case statementStreamChunk loaded of+ Nothing -> exhaustStatementStream streamRef (statementStreamTypes loaded) >> pure Nothing+ Just chunk -> do+ fields <-+ restore $ buildMaterializedRow (statementStreamColumns loaded) (statementStreamChunkReaders chunk) (statementStreamChunkIndex chunk)+ let nextIndex = statementStreamChunkIndex chunk + 1+ if nextIndex < statementStreamChunkSize chunk+ then writeIORef streamRef (StatementStreamActive loaded{statementStreamChunk = Just chunk{statementStreamChunkIndex = nextIndex}})+ else do+ writeIORef streamRef (StatementStreamActive loaded{statementStreamChunk = Nothing})+ finalizeChunk chunk+ restore (runConversion (parseRow parser fields)) >>= \case+ Errors rowErr -> throwIO $ rowErrorsToSqlError (statementQuery stmt) rowErr+ Ok value -> pure (Just value) --- | Release the cursor and retain its exhausted state until an explicit reset.-exhaustStatementStream :: IORef StatementStreamState -> IO ()-exhaustStatementStream ref = mask_ do- state <- atomicModifyIORef' ref (StatementStreamExhausted,)- finalizeStreamState state+-- | Release the cursor only if it still belongs to this execution.+-- See Note [Statement cursor] in "Database.DuckDB.Simple.Internal".+exhaustStatementStream :: IORef StatementStreamState -> TypeScope -> IO ()+exhaustStatementStream ref types = mask_ do+ owned <- atomicModifyIORef' ref \case+ StatementStreamActive stream | statementStreamTypes stream == types -> (StatementStreamExhausted, Just stream)+ state -> (state, Nothing)+ mapM_ finalizeStream owned startStatementStream :: ResultMode -> Statement -> IO (Maybe StatementStream)-startStatementStream mode stmt =- withStatementHandle stmt \handle -> do- resultPtr <- malloc- fillBytes resultPtr 0 (sizeOf (undefined :: DuckDBResult))- let release = c_duckdb_destroy_result resultPtr `finally` free resultPtr- flip onException release do- rc <- runInterruptibleQuery (statementConnection stmt) (executePreparedResult mode handle resultPtr)- when (rc /= DuckDBSuccess) do- (errMsg, errType) <- fetchResultError resultPtr- throwIO $ mkExecuteError (statementQuery stmt) errMsg errType- resultType <- c_duckdb_result_return_type resultPtr- if resultType /= DuckDBResultTypeQueryResult- then release >> pure Nothing- else do- columns <- collectResultColumns resultPtr- pure (Just (StatementStream resultPtr columns Nothing mode))+startStatementStream mode stmt = withQueryError (statementQuery stmt) . withResultWindow (statementConnection stmt) (mapM_ finalizeStream) $ mask_ do+ -- A materialized result closes here. A stream stays open after the lock is released.+ -- See Note [Result window] in "Database.DuckDB.Simple.Internal".+ result <- executePreparedResult stmt+ startResultStream mode (statementConnection stmt) result -fetchChunk :: Connection -> Query -> StatementStream -> IO StatementStream-fetchChunk conn queryText stream@StatementStream{statementStreamResult} = do- chunk <- fetchResultChunk (statementStreamMode stream) conn statementStreamResult- if chunk == nullPtr+-- | Take ownership of a cursor and retain its chunks in the selected mode.+-- A changed-row result gives its row with the count, as a query result gives its rows.+startResultStream :: ResultMode -> Connection -> DuckDBV2ResultHandle -> IO (Maybe StatementStream)+startResultStream mode conn result = mask_ do+ (owned, cursor, types) <-+ ((,,) <$> newIORef [] <*> newIORef result <*> newTypeScope)+ `onException` destroy c_duckdb_v2_result_destroy result+ -- Take each handle out before its release, so that a second call releases no handle again.+ let destroyResult = atomicModifyIORef' cursor (nullHandle,) >>= destroy c_duckdb_v2_result_destroy+ release = closeTypeScope types `finally` ((atomicModifyIORef' owned ([],) >>= releaseItems destroyDataChunk) `finally` destroyResult)+ fetch = fetchResultChunk conn result+ gather = do+ chunk <- fetch+ when (chunk /= nullHandle) do+ atomicModifyIORef' owned (\chunks -> (chunk : chunks, ()))+ gather+ flip onException release do+ -- Expanded statements expose metadata after the first native fetch.+ first <- fetch+ when (first /= nullHandle) $+ atomicModifyIORef' owned (\chunks -> (first : chunks, ()))+ resultType <- output (c_duckdb_v2_result_get_result_type result)+ if resultType /= DuckDBV2ResultTypeQueryResult && resultType /= DuckDBV2ResultTypeChangedRows+ then do+ _ <- runInterruptibleQuery conn (output (c_duckdb_v2_result_drain result))+ release+ pure Nothing+ else do+ columns <- collectResultColumns types (withConnectionHandle conn \_ -> instanceCastFactory (connectionInstance conn)) result+ case mode of+ MaterializedResult -> do+ gather+ chunks <- reverse <$> readIORef owned+ destroyResult+ pure (Just (StatementStream nullHandle types columns chunks Nothing))+ StreamingResult -> do+ chunks <- reverse <$> readIORef owned+ pure (Just (StatementStream result types columns chunks Nothing))++fetchChunk :: IORef StatementStreamState -> Connection -> Query -> StatementStream -> IO StatementStream+fetchChunk ref conn queryText stream = withQueryError queryText $ mask \restore -> do+ (chunk, remaining) <- case statementStreamPendingChunks stream of+ next : rest -> pure (next, rest)+ [] | statementStreamResult stream == nullHandle -> pure (nullHandle, [])+ -- The fetch of an explicit stream holds the connection lock. This call is+ -- not in the result window, so this thread does not hold the lock here.+ -- See Note [Connection lock] in "Database.DuckDB.Simple.Internal".+ [] -> (,[]) <$> withConnectionLock conn (fetchResultChunk conn (statementStreamResult stream))+ let nextStream = stream{statementStreamPendingChunks = remaining}+ if chunk == nullHandle+ then pure nextStream+ else do+ -- Transfer ownership before native metadata or reader setup can fail.+ -- See Note [Statement cursor] in "Database.DuckDB.Simple.Internal".+ let owned = nextStream{statementStreamChunk = Just (StatementStreamChunk chunk 0 0 [])}+ writeIORef ref (StatementStreamActive owned)+ rowCount <- fromIntegral <$> output (c_duckdb_v2_data_chunk_get_size chunk)+ if rowCount <= 0 then do- throwResultError queryText statementStreamResult- pure stream+ writeIORef ref (StatementStreamActive nextStream)+ destroyDataChunk chunk+ restore (fetchChunk ref conn queryText nextStream) else do- rawSize <- c_duckdb_data_chunk_get_size chunk- let rowCount = fromIntegral rawSize :: Int- if rowCount <= 0- then do- destroyDataChunk chunk- fetchChunk conn queryText stream- else do- readers <-- prepareChunkReaders chunk (statementStreamColumns stream)- `onException` destroyDataChunk chunk- let chunkState =- StatementStreamChunk- { statementStreamChunkPtr = chunk- , statementStreamChunkSize = rowCount- , statementStreamChunkIndex = 0- , statementStreamChunkReaders = readers- }- pure stream{statementStreamChunk = Just chunkState}---- | Prepare one reader for each column. The readers must not outlive the chunk.-prepareChunkReaders :: DuckDBDataChunk -> [StatementStreamColumn] -> IO [Int -> IO FieldValue]-prepareChunkReaders chunk columns =- forM columns \StatementStreamColumn{statementStreamColumnIndex} ->- c_duckdb_data_chunk_get_vector chunk (fromIntegral statementStreamColumnIndex) >>= prepareVectorReader---- | Clear cursor ownership before releasing native resources.-cleanupStatementStreamRef :: IORef StatementStreamState -> IO ()-cleanupStatementStreamRef ref = mask_ do- state <- atomicModifyIORef' ref (StatementStreamIdle,)- finalizeStreamState state--finalizeStreamState :: StatementStreamState -> IO ()-finalizeStreamState = \case- StatementStreamIdle -> pure ()- StatementStreamExhausted -> pure ()- StatementStreamActive stream -> finalizeStream stream+ readers <- restore (prepareChunkReaders conn chunk (statementStreamColumns stream))+ pure nextStream{statementStreamChunk = Just (StatementStreamChunk chunk rowCount 0 readers)} -finalizeStream :: StatementStream -> IO ()-finalizeStream StatementStream{statementStreamResult, statementStreamChunk} = do- maybe (pure ()) finalizeChunk statementStreamChunk- c_duckdb_destroy_result statementStreamResult- free statementStreamResult+-- | Prepare readers that borrow the vectors in an owned chunk. A reader takes+-- each layout parameter from its vector type and each description from the+-- column's type tree. The reader builds the tree on its first use, from the+-- current vector type. The tree lives for one execution. The readers do not+-- use the result schema. See Note [Type trees] in+-- "Database.DuckDB.Simple.Materialize".+prepareChunkReaders :: Connection -> DuckDBV2DataChunkHandle -> [StatementStreamColumn] -> IO [Int -> IO FieldValue]+prepareChunkReaders conn chunk columns = do+ casts <- instanceCastFactory (connectionInstance conn)+ -- See Note [VARIANT decoding] in "Database.DuckDB.Simple.VariantCodec" and+ -- Note [User casts and the casts of the library] in+ -- "Database.DuckDB.Simple.Materialize".+ let readType = logicalTypeToRep casts+ readers = FallbackReaders readType (readConnectionVariant conn) (prepareGeometryReader (InstanceFactory casts))+ forM columns \StatementStreamColumn{statementStreamColumnIndex, statementStreamColumnTree, statementStreamColumnLabels} -> do+ vector <- output (c_duckdb_v2_data_chunk_get_vector chunk (fromIntegral statementStreamColumnIndex))+ let tree =+ readIORef statementStreamColumnTree >>= \case+ Just rep -> pure rep+ Nothing -> do+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ rep <- withVectorType vector readType+ writeIORef statementStreamColumnTree (Just rep)+ pure rep+ prepareVectorReaderWith readers tree (Just statementStreamColumnLabels) vector -finalizeChunk :: StatementStreamChunk -> IO ()-finalizeChunk StatementStreamChunk{statementStreamChunkPtr} =- destroyDataChunk statementStreamChunkPtr+-- | Materialize all rows, then run the decoder while column metadata is live.+withRows :: Statement -> ([[Field]] -> IO a) -> IO a+withRows stmt = withStreamRows (statementConnection stmt) (statementQuery stmt) (startStatementStream MaterializedResult stmt) --- | Copy all rows from a materialized native result.-collectRows :: Query -> Ptr DuckDBResult -> IO [[Field]]-collectRows queryText resPtr = do- columns <- collectResultColumns resPtr- collectChunks columns []+-- | Execute a parsed statement directly and materialize its owned native chunks.+-- See Note [Result window] in "Database.DuckDB.Simple.Internal".+withParsedRows :: Connection -> Query -> DuckDBV2SqlStatementHandle -> ([[Field]] -> IO a) -> IO a+withParsedRows conn query parsed = withStreamRows conn query acquire where- collectChunks columns acc = do- fetched <- bracket (c_duckdb_fetch_chunk resPtr) destroyDataChunk \chunk ->- if chunk == nullPtr- then throwResultError queryText resPtr >> pure Nothing- else Just <$> decodeChunk columns chunk- case fetched of- Nothing -> pure (concat (reverse acc))- Just rows -> do- let acc' = maybe acc (: acc) rows- collectChunks columns acc'+ acquire = withResultWindow conn (mapM_ finalizeStream) $ mask_ do+ result <- executeParsedResult conn query parsed+ startResultStream MaterializedResult conn result - decodeChunk columns chunk = do- rawSize <- c_duckdb_data_chunk_get_size chunk- let rowCount = fromIntegral rawSize :: Int- if rowCount <= 0- then pure Nothing- else- if null columns- then pure (Just (replicate rowCount []))- else do- readers <- prepareChunkReaders chunk columns- rows <- mapM (buildMaterializedRow columns readers) [0 .. rowCount - 1]- pure (Just rows)+-- | Copy rows from owned chunks, then run the decoder before metadata closes.+-- Destroy each chunk after its values are copied.+withStreamRows :: Connection -> Query -> IO (Maybe StatementStream) -> ([[Field]] -> IO a) -> IO a+withStreamRows conn query acquire decode = withQueryError query $+ bracket (acquire >>= traverse adopt) (mapM_ release) \case+ Nothing -> decode []+ Just (owned, pending) -> do+ -- Both loops keep a constant stack. See+ -- Note [Constant stack at safe calls] in "Database.DuckDB.Simple.Native".+ --+ -- The chunk loop must keep a constant stack, because the code for+ -- each chunk makes native calls.+ --+ -- The reference holds the chunks that the fold has not reached, in+ -- the order of the fold. Each step takes its chunk out of the+ -- reference and destroys that chunk after its rows. The bracket of+ -- the step thus owns the chunk while its rows are decoded, and the+ -- release of this function destroys only the chunks in the reference.+ let readChunk done expected = bracket (takeChunk pending expected) destroyDataChunk \chunk -> do+ rowCount <- fromIntegral <$> output (c_duckdb_v2_data_chunk_get_size chunk)+ readers <- prepareChunkReaders conn chunk (statementStreamColumns owned)+ -- The row loop reads the rows from the last to the first, and+ -- puts each row in front of the list. The list then needs no+ -- reverse. The constant stack of this loop is necessary for a+ -- reader that makes a native call for each row: a BIGNUM+ -- column, and the first use of a UNION member.+ let readFrom row decoded+ | row < 0 = pure decoded+ | otherwise = buildMaterializedRow (statementStreamColumns owned) readers row >>= \fields -> readFrom (row - 1) (fields : decoded)+ (: done) <$> readFrom (rowCount - 1) []+ -- The fold prepends each decoded chunk. Reverse that list after the fold.+ rows <- foldM readChunk [] (statementStreamPendingChunks owned)+ decode (concat (reverse rows))+ where+ adopt stream =+ ((,) stream <$> newIORef (statementStreamPendingChunks stream))+ `onException` finalizeStream stream+ release (stream, pending) = do+ remaining <- atomicModifyIORef' pending ([],)+ finalizeStream stream{statementStreamPendingChunks = remaining}+ -- Take the first chunk out of the reference. It must be the chunk of the+ -- fold step. Otherwise leave the reference as it is, so that the release+ -- destroys each chunk in it.+ takeChunk pending expected =+ atomicModifyIORef' pending (\case chunk : rest | chunk == expected -> (rest, Just chunk); chunks -> (chunks, Nothing))+ >>= maybe (throwInternalError (Text.pack "internal error: the owned chunks do not match the chunk fold")) pure -collectResultColumns :: Ptr DuckDBResult -> IO [StatementStreamColumn]-collectResultColumns resPtr = do- rawCount <- c_duckdb_column_count resPtr- let cc = fromIntegral rawCount :: Int- forM [0 .. cc - 1] \columnIndex -> do- namePtr <- c_duckdb_column_name resPtr (fromIntegral columnIndex)- name <-- if namePtr == nullPtr- then pure (Text.pack ("column" <> show columnIndex))- else peekUtf8CString namePtr- dtype <- c_duckdb_column_type resPtr (fromIntegral columnIndex)- pure- StatementStreamColumn- { statementStreamColumnIndex = columnIndex- , statementStreamColumnName = name- , statementStreamColumnType = dtype- }+collectResultColumns :: TypeScope -> IO DuckDBV2FactoryHandle -> DuckDBV2ResultHandle -> IO [StatementStreamColumn]+collectResultColumns types acquireFactory result =+ bracket (output (c_duckdb_v2_result_get_schema result)) (destroy c_duckdb_v2_schema_destroy) \schema -> do+ count <- fromIntegral <$> output (c_duckdb_v2_schema_get_count schema)+ forM [0 .. count - 1] \idx -> alloca \name -> do+ borrowed <- output (c_duckdb_v2_schema_get_field schema (fromIntegral idx) name)+ columnName <- peek name >>= readIdentifier+ -- The result scope owns the copied type. The schema closes after this loop.+ columnType <- scopedTypeRep types acquireFactory (output (c_duckdb_v2_logical_type_copy borrowed))+ StatementStreamColumn idx columnName columnType <$> newIORef Nothing <*> newIORef IntMap.empty buildMaterializedRow :: [StatementStreamColumn] -> [Int -> IO FieldValue] -> Int -> IO [Field] buildMaterializedRow columns readers rowIdx =- zipWithM (buildMaterializedField rowIdx) columns readers+ zipWithM (buildMaterializedField rowIdx) columns readers buildMaterializedField :: Int -> StatementStreamColumn -> (Int -> IO FieldValue) -> IO Field buildMaterializedField rowIdx column readValue = do- value <- readValue rowIdx- pure- Field- { fieldName = statementStreamColumnName column- , fieldIndex = statementStreamColumnIndex column- , fieldValue = value- }+ value <- readValue rowIdx+ pure+ Field+ { fieldName = statementStreamColumnName column+ , fieldIndex = statementStreamColumnIndex column+ , fieldValue = value+ , fieldType = statementStreamColumnType column+ }
+ src/Database/DuckDB/Simple/Streaming.hs view
@@ -0,0 +1,91 @@+{-# LANGUAGE BlockArguments #-}++-- |+-- Module : Database.DuckDB.Simple.Streaming+-- Description : Optional streaming through the native V2 execution API.+--+-- Import this module qualified to request native streaming. Operators such as+-- sorting can buffer data. Streaming reduces result storage when the query+-- supports it. It does not bound query memory. The default+-- "Database.DuckDB.Simple" API materializes native results.+--+-- Rows and Arrow batches use the same decoders and cleanup as the default API.+-- Cancellation also interrupts native chunk fetching. Prompt interruption requires+-- @-threaded@ and waits for native code and Haskell callbacks to return. A+-- program that registers a callback needs @-threaded@ in each case: see+-- "Database.DuckDB.Simple.Function".+--+-- Serialize connection use for the entire cursor or fold. Another query on the+-- same connection fails while its stream is active. Call+-- 'Database.DuckDB.Simple.closeStatement' or+-- 'Database.DuckDB.Simple.clearStatementBindings' on an abandoned statement to+-- release its result.+module Database.DuckDB.Simple.Streaming (+ fold,+ fold_,+ foldNamed,+ nextRow,+ nextRowWith,+ foldArrow,+ foldArrow_,+) where++import Database.DuckDB.FFI.V2 (ArrowArray, ArrowSchema)+import Database.DuckDB.Simple (bind, bindNamed, withStatement)+import qualified Database.DuckDB.Simple.Arrow.Internal as Arrow+import Database.DuckDB.Simple.FromRow (FromRow (..), RowParser)+import Database.DuckDB.Simple.Internal (Connection, Query, ResultMode (StreamingResult), Statement)+import qualified Database.DuckDB.Simple.Result as Result+import Database.DuckDB.Simple.ToField (NamedParam)+import Database.DuckDB.Simple.ToRow (ToRow (..))+import Foreign.Ptr (Ptr)++-- | Fold a parameterized query, requesting native streaming.+fold :: (FromRow row, ToRow params) => Connection -> Query -> params -> a -> (a -> row -> IO a) -> IO a+fold conn sql params initial step =+ withStatement conn sql \stmt -> do+ bind stmt (toRow params)+ Result.foldStatementWith StreamingResult fromRow stmt initial step++-- | Fold a query without parameters, requesting native streaming.+fold_ :: (FromRow row) => Connection -> Query -> a -> (a -> row -> IO a) -> IO a+fold_ conn sql initial step =+ withStatement conn sql \stmt ->+ Result.foldStatementWith StreamingResult fromRow stmt initial step++-- | Fold a query with named parameters, requesting native streaming.+foldNamed :: (FromRow row) => Connection -> Query -> [NamedParam] -> a -> (a -> row -> IO a) -> IO a+foldNamed conn sql params initial step =+ withStatement conn sql \stmt -> do+ bindNamed stmt params+ Result.foldStatementWith StreamingResult fromRow stmt initial step++-- | Fetch the next row, requesting native streaming on the first fetch.+nextRow :: (FromRow r) => Statement -> IO (Maybe r)+nextRow = nextRowWith fromRow++-- | Fetch a row with a custom parser, requesting streaming on the first fetch.+-- The active result retains its execution mode until+-- 'Database.DuckDB.Simple.bind', 'Database.DuckDB.Simple.bindNamed',+-- 'Database.DuckDB.Simple.clearStatementBindings', or+-- 'Database.DuckDB.Simple.executeStatement' releases it. Switching between this+-- module and 'Database.DuckDB.Simple.nextRow' does not restart or change an+-- active result.+-- After EOF, this function returns 'Nothing' until one of those functions runs.+nextRowWith :: RowParser r -> Statement -> IO (Maybe r)+nextRowWith = Result.nextRowWith StreamingResult++-- | Fold Arrow batches, requesting native streaming.+-- Each batch has a separate schema and array. The callback may read, release,+-- or move them under the Arrow C Data Interface. The pointers themselves are+-- valid during the callback only. See "Database.DuckDB.Simple.Arrow" for the+-- ownership rules.+-- Query execution and Arrow export use the native V2 API.+-- The fold releases contents that the consumer has not+-- released or moved. The consumer owns any moved contents.+foldArrow :: (ToRow params) => Connection -> Query -> params -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a+foldArrow = Arrow.foldArrowWith StreamingResult++-- | Fold Arrow batches from a query without parameters.+foldArrow_ :: Connection -> Query -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a+foldArrow_ conn sql = foldArrow conn sql ()
src/Database/DuckDB/Simple/Time.hs view
@@ -1,21 +1,24 @@ {-# LANGUAGE DeriveFunctor #-}+{-# LANGUAGE StrictData #-} --- | Dates and timestamps that can represent DuckDB's infinities.+-- |+-- Module : Database.DuckDB.Simple.Time+-- Description : Dates and timestamps that can represent DuckDB's infinities. module Database.DuckDB.Simple.Time (- Unbounded (..),- Date,- LocalTimestamp,- UTCTimestamp,+ Unbounded (..),+ Date,+ LocalTimestamp,+ UTCTimestamp, ) where import Data.Time (Day, LocalTime, UTCTime) -- | A finite value or either infinity, as in @postgresql-simple@. data Unbounded a- = NegInfinity- | Finite !a- | PosInfinity- deriving (Eq, Ord, Show, Read, Functor)+ = NegInfinity+ | Finite a+ | PosInfinity+ deriving (Eq, Ord, Show, Read, Functor) -- | A DuckDB DATE, including infinity. type Date = Unbounded Day
src/Database/DuckDB/Simple/ToField.hs view
@@ -2,1009 +2,2308 @@ {-# LANGUAGE DefaultSignatures #-} {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE GADTs #-}-{-# LANGUAGE NamedFieldPuns #-}-{-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE ScopedTypeVariables #-}--{- |-Module : Database.DuckDB.Simple.ToField-Description : Convert Haskell parameters into DuckDB bindable values.--The @ToField@ class mirrors the interface provided by @sqlite-simple@ while-delegating to the DuckDB C API under the hood.--}-module Database.DuckDB.Simple.ToField (- FieldBinding,- ToDuckValue (..),- ToField (..),- DuckDBColumnType (..),- NamedParam (..),- duckdbColumnType,- bindFieldBinding,- renderFieldBinding,-) where--import Control.Exception (bracket, throwIO)-import Control.Monad (filterM, forM_, when)-import Data.Array (Array, elems)-import Data.Bits (complement, shiftL, shiftR, (.&.), (.|.))-import qualified Data.ByteString as BS-import Data.Fixed (Pico)-import qualified Data.Geometry as G-import qualified Data.Geometry.WKT as WKT-import Data.Int (Int16, Int32, Int64, Int8)-import qualified Data.List as List-import Data.Proxy (Proxy (..))-import Data.Text (Text)-import qualified Data.Text as Text-import qualified Data.Text.Encoding as TextEncoding-import Data.Time.Calendar (Day, diffDays, fromGregorian)-import Data.Time.Clock (UTCTime (..), diffTimeToPicoseconds)-import Data.Time.LocalTime (LocalTime (..), TimeOfDay (..), TimeZone (..), timeOfDayToTime, timeZoneMinutes, utc, utcToLocalTime)-import qualified Data.UUID as UUID-import Data.Word (Word16, Word32, Word64, Word8)-import Database.DuckDB.FFI-import Database.DuckDB.Simple.FromField (BigNum (..), BitString (..), DecimalValue (..), FieldValue (..), IntervalValue (..), TimeWithZone (..), toBigNumBytes)-import Database.DuckDB.Simple.Internal (- SQLError (..),- Statement (..),- destroyValue,- duckDBTypeFromName,- fetchPrepareError,- withStatementHandle,- withTypeCache,- )-import Database.DuckDB.Simple.LogicalRep (- LogicalTypeRep (..),- StructField (..),- StructValue (..),- UnionMemberType (..),- UnionValue (..),- destroyLogicalType,- logicalTypeFromRep,- logicalTypeFromRepWith,- logicalTypeToRep,- structValueTypeRep,- unionValueTypeRep,- )-import Database.DuckDB.Simple.Time (Date, LocalTimestamp, UTCTimestamp, Unbounded (..))-import Database.DuckDB.Simple.TypeCache (TypeCache, cachedLogicalType)-import Database.DuckDB.Simple.Types (Null (..))-import Database.DuckDB.Simple.Variant (Variant (..))-import Foreign.C.Types (CDouble (..), CFloat (..))-import Foreign.Marshal (fromBool)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Marshal.Array (withArray)-import Foreign.Marshal.Utils (withMany)-import Foreign.Ptr (Ptr, castPtr, nullPtr)-import Foreign.Storable (poke)-import Numeric.Natural (Natural)---- | Represents a named parameter binding using the @:=@ operator.-data NamedParam where- (:=) :: (ToField a) => Text -> a -> NamedParam--infixr 3 :=---- | Encapsulates the action required to bind a single positional parameter, together with a textual description used in diagnostics.-data FieldBinding = FieldBinding- { fieldBindingValue :: !(TypeCache -> IO DuckDBValue)- , fieldBindingDisplay :: !String- }---- | Low-level class for values that can be marshalled directly into `DuckDBValue`s.-class (DuckDBColumnType a) => ToDuckValue a where- -- | Convert a Haskell value into an owned DuckDB boxed value.- toDuckValue :: a -> IO DuckDBValue--valueBinding :: String -> IO DuckDBValue -> FieldBinding-valueBinding display = cacheValueBinding display . const---- | Construct a value with the type cache of the statement's connection.-cacheValueBinding :: String -> (TypeCache -> IO DuckDBValue) -> FieldBinding-cacheValueBinding display makeValue =- FieldBinding- { fieldBindingValue = makeValue- , fieldBindingDisplay = display- }---- | Build types with the cached types for VARIANT and GEOMETRY with a CRS.-cachedTypeFromRep :: TypeCache -> LogicalTypeRep -> IO DuckDBLogicalType-cachedTypeFromRep = logicalTypeFromRepWith . cachedLogicalType---- | Types that map to a concrete DuckDB column type when used with @ToField@.-class DuckDBColumnType a where- duckdbColumnTypeFor :: Proxy a -> Text---- | Report the DuckDB column type that best matches a given @ToField@ instance.-duckdbColumnType :: forall a. (DuckDBColumnType a) => Proxy a -> Text-duckdbColumnType = duckdbColumnTypeFor---- | Apply a @FieldBinding@ to the given statement/index.-bindFieldBinding :: Statement -> DuckDBIdx -> FieldBinding -> IO ()-bindFieldBinding stmt idx FieldBinding{fieldBindingValue} =- withTypeCache (statementConnection stmt) \cache ->- bindDuckValue stmt idx (fieldBindingValue cache)---- | Render a bound parameter for error reporting.-renderFieldBinding :: FieldBinding -> String-renderFieldBinding FieldBinding{fieldBindingDisplay} = fieldBindingDisplay---- | Types that can be used as positional parameters.-class ToField a where- toField :: a -> FieldBinding- default toField :: (Show a, ToDuckValue a) => a -> FieldBinding- toField value = valueBinding (show value) (toDuckValue value)--instance ToField Null where- toField Null = nullBinding "NULL"--instance ToField Bool-instance ToField Int-instance ToField Int8-instance ToField Int16-instance ToField Int32-instance ToField Int64-instance ToField Integer-instance ToField Natural-instance ToField UUID.UUID-instance ToField Word-instance ToField Word8-instance ToField Word16-instance ToField Word32-instance ToField Word64-instance ToField Double-instance ToField Float-instance ToField Text-instance ToField String-instance ToField BitString-instance ToField Day-instance ToField TimeOfDay-instance ToField LocalTime---- | Bind the shape as @GEOMETRY@ with no CRS.-instance ToField G.Geometry---- | Bind the payload as a VARIANT with the type cache of the connection.-instance ToField Variant where- toField value =- cacheValueBinding (show value) \cache ->- variantDuckValue (cachedTypeFromRep cache) (variantPayload value)--instance ToField UTCTime-instance ToField (Unbounded Day)-instance ToField (Unbounded LocalTime)-instance ToField (Unbounded UTCTime)--instance ToField BigNum where- toField big@(BigNum n) = valueBinding (show n) (bigNumDuckValue big)--instance ToField (StructValue FieldValue) where- toField structVal =- cacheValueBinding "<struct>" \cache ->- structValueDuckValue (cachedTypeFromRep cache) structVal--instance ToField (UnionValue FieldValue) where- toField unionVal =- let label = Text.unpack (unionValueLabel unionVal)- in cacheValueBinding ("<union " <> label <> ">") \cache ->- unionValueDuckValue (cachedTypeFromRep cache) unionVal--instance DuckDBColumnType BitString where- duckdbColumnTypeFor _ = "BIT"--instance ToField BS.ByteString where- toField bs =- valueBinding- ("<blob length=" <> show (BS.length bs) <> ">")- (toDuckValue bs)--instance (DuckDBColumnType a, ToField a) => ToField (Array Int a) where- toField arr =- cacheValueBinding- ("<array length=" <> show (length (elems arr)) <> ">")- \cache -> arrayDuckValue (cachedTypeFromRep cache) (\value -> fieldBindingValue (toField value) cache) arr--instance (ToField a) => ToField (Maybe a) where- toField Nothing = nullBinding "Nothing"- toField (Just value) =- let binding = toField value- in binding- { fieldBindingDisplay = "Just " <> renderFieldBinding binding- }--instance DuckDBColumnType G.Geometry where- duckdbColumnTypeFor _ = "GEOMETRY"--instance DuckDBColumnType Variant where- duckdbColumnTypeFor _ = "VARIANT"--instance DuckDBColumnType Null where- duckdbColumnTypeFor _ = "NULL"--instance DuckDBColumnType Bool where- duckdbColumnTypeFor _ = "BOOLEAN"--instance DuckDBColumnType Int where- duckdbColumnTypeFor _ = "BIGINT"--instance DuckDBColumnType Int8 where- duckdbColumnTypeFor _ = "TINYINT"--instance DuckDBColumnType Int16 where- duckdbColumnTypeFor _ = "SMALLINT"--instance DuckDBColumnType Int32 where- duckdbColumnTypeFor _ = "INTEGER"--instance DuckDBColumnType Int64 where- duckdbColumnTypeFor _ = "BIGINT"--instance DuckDBColumnType BigNum where- duckdbColumnTypeFor _ = "BIGNUM"--instance DuckDBColumnType UUID.UUID where- duckdbColumnTypeFor _ = "UUID"--instance DuckDBColumnType Integer where- duckdbColumnTypeFor _ = "BIGNUM"--instance DuckDBColumnType Natural where- duckdbColumnTypeFor _ = "BIGNUM"--instance DuckDBColumnType Word where- duckdbColumnTypeFor _ = "UBIGINT"--instance DuckDBColumnType Word8 where- duckdbColumnTypeFor _ = "UTINYINT"--instance DuckDBColumnType Word16 where- duckdbColumnTypeFor _ = "USMALLINT"--instance DuckDBColumnType Word32 where- duckdbColumnTypeFor _ = "UINTEGER"--instance DuckDBColumnType Word64 where- duckdbColumnTypeFor _ = "UBIGINT"--instance DuckDBColumnType Double where- duckdbColumnTypeFor _ = "DOUBLE"--instance DuckDBColumnType Float where- duckdbColumnTypeFor _ = "FLOAT"--instance DuckDBColumnType Text where- duckdbColumnTypeFor _ = "TEXT"--instance DuckDBColumnType String where- duckdbColumnTypeFor _ = "TEXT"--instance DuckDBColumnType BS.ByteString where- duckdbColumnTypeFor _ = "BLOB"--instance DuckDBColumnType Day where- duckdbColumnTypeFor _ = "DATE"--instance DuckDBColumnType TimeOfDay where- duckdbColumnTypeFor _ = "TIME"--instance DuckDBColumnType LocalTime where- duckdbColumnTypeFor _ = "TIMESTAMP"--instance DuckDBColumnType UTCTime where- duckdbColumnTypeFor _ = "TIMESTAMPTZ"--instance DuckDBColumnType (Unbounded Day) where- duckdbColumnTypeFor _ = "DATE"--instance DuckDBColumnType (Unbounded LocalTime) where- duckdbColumnTypeFor _ = "TIMESTAMP"--instance DuckDBColumnType (Unbounded UTCTime) where- duckdbColumnTypeFor _ = "TIMESTAMPTZ"--instance DuckDBColumnType (StructValue FieldValue) where- duckdbColumnTypeFor _ = "STRUCT"--instance DuckDBColumnType (UnionValue FieldValue) where- duckdbColumnTypeFor _ = "UNION"--instance (DuckDBColumnType a) => DuckDBColumnType (Maybe a) where- duckdbColumnTypeFor _ = duckdbColumnTypeFor (Proxy :: Proxy a)--instance (DuckDBColumnType a) => DuckDBColumnType (Array Int a) where- duckdbColumnTypeFor _ = duckdbColumnTypeFor (Proxy :: Proxy a) <> Text.pack "[]"--nullBinding :: String -> FieldBinding-nullBinding repr = valueBinding repr nullDuckValue--nullDuckValue :: IO DuckDBValue-nullDuckValue = c_duckdb_create_null_value--boolDuckValue :: Bool -> IO DuckDBValue-boolDuckValue value = c_duckdb_create_bool (if value then 1 else 0)--int8DuckValue :: Int8 -> IO DuckDBValue-int8DuckValue = c_duckdb_create_int8--int16DuckValue :: Int16 -> IO DuckDBValue-int16DuckValue = c_duckdb_create_int16--int32DuckValue :: Int32 -> IO DuckDBValue-int32DuckValue = c_duckdb_create_int32--int64DuckValue :: Int64 -> IO DuckDBValue-int64DuckValue = c_duckdb_create_int64--uint64DuckValue :: Word64 -> IO DuckDBValue-uint64DuckValue = c_duckdb_create_uint64--uint32DuckValue :: Word32 -> IO DuckDBValue-uint32DuckValue = c_duckdb_create_uint32--uint16DuckValue :: Word16 -> IO DuckDBValue-uint16DuckValue = c_duckdb_create_uint16--uint8DuckValue :: Word8 -> IO DuckDBValue-uint8DuckValue = c_duckdb_create_uint8--doubleDuckValue :: Double -> IO DuckDBValue-doubleDuckValue = c_duckdb_create_double . CDouble--floatDuckValue :: Float -> IO DuckDBValue-floatDuckValue = c_duckdb_create_float . CFloat--textDuckValue :: Text -> IO DuckDBValue-textDuckValue txt =- BS.useAsCStringLen (TextEncoding.encodeUtf8 txt) \(ptr, len) ->- c_duckdb_create_varchar_length ptr (fromIntegral len)--stringDuckValue :: String -> IO DuckDBValue-stringDuckValue = textDuckValue . Text.pack--blobDuckValue :: BS.ByteString -> IO DuckDBValue-blobDuckValue bs =- BS.useAsCStringLen bs \(ptr, len) ->- c_duckdb_create_blob (castPtr ptr :: Ptr Word8) (fromIntegral len)--uuidDuckValue :: UUID.UUID -> IO DuckDBValue-uuidDuckValue uuid =- alloca $ \ptr -> do- let (upper, lower) = UUID.toWords64 uuid- poke- ptr- DuckDBUHugeInt- { duckDBUHugeIntLower = lower- , duckDBUHugeIntUpper = upper- }- c_duckdb_create_uuid ptr--bitDuckValue :: BitString -> IO DuckDBValue-bitDuckValue (BitString padding bits) = do- when (BS.null bits || padding > 7) $- throwIO (userError "duckdb-simple: BIT requires nonempty data and padding from 0 to 7")- let nativePadding = complement ((1 `shiftL` (8 - fromIntegral padding)) - 1) :: Word8- payload = BS.cons padding (BS.cons (BS.head bits .|. nativePadding) (BS.tail bits))- BS.useAsCStringLen payload \(rawPtr, len) ->- alloca \ptr -> do- poke ptr DuckDBBit{duckDBBitData = castPtr rawPtr, duckDBBitSize = fromIntegral len}- c_duckdb_create_bit ptr--bigNumDuckValue :: BigNum -> IO DuckDBValue-bigNumDuckValue (BigNum big) =- let neg = fromBool (big < 0)- payload =- BS.pack $- if big < 0- then map complement (drop 3 $ toBigNumBytes big)- else drop 3 $ toBigNumBytes big- withPayload action =- if BS.null payload- then alloca \ptr -> do- poke- ptr- DuckDBBignum- { duckDBBignumData = nullPtr- , duckDBBignumSize = 0- , duckDBBignumIsNegative = neg- }- action ptr- else BS.useAsCStringLen payload \(rawPtr, len) ->- alloca \ptr -> do- poke- ptr- DuckDBBignum- { duckDBBignumData = castPtr rawPtr- , duckDBBignumSize = fromIntegral len- , duckDBBignumIsNegative = neg- }- action ptr- in withPayload c_duckdb_create_bignum--dayDuckValue :: Day -> IO DuckDBValue-dayDuckValue day = do- duckDate <- encodeDay day- c_duckdb_create_date duckDate--timeOfDayDuckValue :: TimeOfDay -> IO DuckDBValue-timeOfDayDuckValue tod = do- duckTime <- encodeTimeOfDay tod- c_duckdb_create_time duckTime--localTimeDuckValue :: LocalTime -> IO DuckDBValue-localTimeDuckValue ts = do- duckTimestamp <- encodeLocalTime ts- c_duckdb_create_timestamp duckTimestamp--utcTimeDuckValue :: UTCTime -> IO DuckDBValue-utcTimeDuckValue utcTime =- encodeLocalTime (utcToLocalTime utc utcTime) >>= c_duckdb_create_timestamp_tz---- | Bind a date, including either infinity sentinel.-dateDuckValue :: Date -> IO DuckDBValue-dateDuckValue value =- encodeUnbounded (fmap unDuckDBDate . encodeDay) value >>= c_duckdb_create_date . DuckDBDate---- | Bind a timestamp without a time zone, including infinity.-localTimestampDuckValue :: LocalTimestamp -> IO DuckDBValue-localTimestampDuckValue value =- encodeUnbounded (encodeTimestampUnits 1000000) value >>= c_duckdb_create_timestamp . DuckDBTimestamp---- | Bind a timestamp with a time zone, including infinity.-utcTimestampDuckValue :: UTCTimestamp -> IO DuckDBValue-utcTimestampDuckValue value =- encodeUnbounded (encodeTimestampUnits 1000000 . utcToLocalTime utc) value >>= c_duckdb_create_timestamp_tz . DuckDBTimestamp--{- | Build an array value with a function for each element. A scalar element-type comes from the column type name of the element, so an empty array keeps-it. Other element types, such as STRUCT, UNION, and ARRAY, come from the-first element that is not NULL. All present elements must have the same type.--}-arrayDuckValue ::- forall a.- (DuckDBColumnType a) =>- (LogicalTypeRep -> IO DuckDBLogicalType) ->- (a -> IO DuckDBValue) ->- Array Int a ->- IO DuckDBValue-arrayDuckValue typeFromRep elementValue arr =- withCreatedValues (map elementValue (elems arr)) \values ->- withElementType values \elementType ->- withDuckValues values \ptr ->- checkedValue (c_duckdb_create_array_value elementType ptr (fromIntegral (length values)))- where- typeName = duckdbColumnType (Proxy :: Proxy a)- withElementType values action =- case duckDBTypeFromName typeName of- Just dtype -> bracket (typeFromRep (LogicalTypeScalar dtype)) destroyLogicalType action- Nothing -> do- present <- filterM (fmap (== 0) . c_duckdb_is_null_value) values- case present of- -- The value owns this type.- value : rest -> do- logical <- c_duckdb_get_value_type value- expected <- logicalTypeToRep logical- forM_ rest \element -> do- actual <- c_duckdb_get_value_type element >>= logicalTypeToRep- when (actual /= expected) $- throwIO (userError "duckdb-simple: array elements have different logical types")- action logical- [] ->- throwIO- SQLError- { sqlErrorMessage = "duckdb-simple: an empty or all-NULL array of " <> typeName <> " elements has no element type"- , sqlErrorType = Nothing- , sqlErrorQuery = Nothing- }--structValueDuckValue :: (LogicalTypeRep -> IO DuckDBLogicalType) -> StructValue FieldValue -> IO DuckDBValue-structValueDuckValue typeFromRep StructValue{structValueFields, structValueTypes, structValueIndex = _} = do- let valueFields = elems structValueFields- typeFields = elems structValueTypes- typeNames = map structFieldName typeFields- valueNames = map structFieldName valueFields- when (length valueFields /= length typeFields) $- throwIO (userError "duckdb-simple: struct value/type arity mismatch")- when (typeNames /= valueNames) $- throwIO (userError "duckdb-simple: struct value/type field names mismatch")- let actions =- zipWith- ( \StructField{structFieldValue = typeRep} StructField{structFieldValue = fieldVal} ->- fieldValueWithTypeDuckValue typeFromRep typeRep fieldVal- )- typeFields- valueFields- bracket (typeFromRep (LogicalTypeStruct structValueTypes)) destroyLogicalType \structLogical ->- withCreatedValues actions \childValues ->- withDuckValues childValues $ \ptr ->- checkedValue (c_duckdb_create_struct_value structLogical ptr)--unionValueDuckValue :: (LogicalTypeRep -> IO DuckDBLogicalType) -> UnionValue FieldValue -> IO DuckDBValue-unionValueDuckValue typeFromRep UnionValue{unionValueIndex, unionValueLabel, unionValuePayload, unionValueMembers} = do- let membersList = elems unionValueMembers- idx = fromIntegral unionValueIndex :: Int- memberCount = length membersList- when (idx < 0 || idx >= memberCount) $- throwIO (userError "duckdb-simple: union value tag out of range")- let UnionMemberType{unionMemberName, unionMemberType = memberType} = membersList !! idx- when (unionValueLabel /= unionMemberName) $- throwIO (userError "duckdb-simple: union tag and member name mismatch")- bracket (typeFromRep (LogicalTypeUnion unionValueMembers)) destroyLogicalType \unionLogical ->- bracket (checkedValue (fieldValueWithTypeDuckValue typeFromRep memberType unionValuePayload)) destroyValue \payloadValue ->- checkedValue (c_duckdb_create_union_value unionLogical (fromIntegral unionValueIndex) payloadValue)--fieldValueWithTypeDuckValue :: (LogicalTypeRep -> IO DuckDBLogicalType) -> LogicalTypeRep -> FieldValue -> IO DuckDBValue-fieldValueWithTypeDuckValue typeFromRep typeRep FieldNull =- bracket (typeFromRep typeRep) destroyLogicalType \logical ->- withCreatedValues [nullDuckValue] \values ->- withDuckValues values \ptr ->- bracket (checkedValue (c_duckdb_create_list_value logical ptr 1)) destroyValue \list ->- checkedValue (c_duckdb_get_list_child list 0)-fieldValueWithTypeDuckValue typeFromRep rep value =- case rep of- LogicalTypeScalar DuckDBTypeVariant -> variantDuckValue typeFromRep value- LogicalTypeScalar dtype -> scalarFieldValueDuckValue dtype value- LogicalTypeGeometry _ ->- case value of- FieldGeometry{} -> unsupportedRawGeometryBinding- other -> typeMismatch "GEOMETRY" other- LogicalTypeDecimal width scale ->- case value of- FieldDecimal decVal@DecimalValue{decimalWidth, decimalScale}- | decimalWidth == width && decimalScale == scale -> decimalDuckValue decVal- | otherwise -> throwIO (userError "duckdb-simple: decimal value metadata mismatch")- other -> typeMismatch "DECIMAL" other- LogicalTypeList elemRep ->- case value of- FieldList elemsList ->- bracket (typeFromRep elemRep) destroyLogicalType \childLogical ->- withCreatedValues (map (fieldValueWithTypeDuckValue typeFromRep elemRep) elemsList) \values ->- withDuckValues values \ptr ->- checkedValue (c_duckdb_create_list_value childLogical ptr (fromIntegral (length values)))- other -> typeMismatch "LIST" other- LogicalTypeArray elemRep size ->- case value of- FieldArray arr -> do- let elemsList = elems arr- actualCount = length elemsList- when (fromIntegral actualCount /= size) $- throwIO (userError "duckdb-simple: array length mismatch")- bracket (typeFromRep elemRep) destroyLogicalType \childLogical ->- withCreatedValues (map (fieldValueWithTypeDuckValue typeFromRep elemRep) elemsList) \values ->- withDuckValues values \ptr ->- checkedValue (c_duckdb_create_array_value childLogical ptr (fromIntegral actualCount))- other -> typeMismatch "ARRAY" other- LogicalTypeMap keyRep valueRep ->- case value of- FieldMap pairs ->- bracket (typeFromRep (LogicalTypeMap keyRep valueRep)) destroyLogicalType \mapLogical ->- withCreatedValues (map (fieldValueWithTypeDuckValue typeFromRep keyRep . fst) pairs) \keyValues ->- withCreatedValues (map (fieldValueWithTypeDuckValue typeFromRep valueRep . snd) pairs) \valValues ->- withDuckValues keyValues \keyPtr ->- withDuckValues valValues \valPtr ->- checkedValue (c_duckdb_create_map_value mapLogical keyPtr valPtr (fromIntegral (length pairs)))- other -> typeMismatch "MAP" other- LogicalTypeStruct structRep ->- case value of- FieldStruct structVal- | structValueTypeRep structVal == LogicalTypeStruct structRep -> structValueDuckValue typeFromRep structVal- | otherwise -> throwIO (userError "duckdb-simple: struct value type mismatch")- other -> typeMismatch "STRUCT" other- LogicalTypeUnion unionRep ->- case value of- FieldUnion unionVal- | unionValueTypeRep unionVal == LogicalTypeUnion unionRep -> unionValueDuckValue typeFromRep unionVal- | otherwise -> throwIO (userError "duckdb-simple: union value type mismatch")- other -> typeMismatch "UNION" other- LogicalTypeEnum dict ->- case value of- FieldEnum enumIdx -> enumDuckValue dict enumIdx- other -> typeMismatch "ENUM" other--scalarFieldValueDuckValue :: DuckDBType -> FieldValue -> IO DuckDBValue-scalarFieldValueDuckValue dtype value =- case (dtype, value) of- (DuckDBTypeBoolean, FieldBool b) -> boolDuckValue b- (DuckDBTypeTinyInt, FieldInt8 i) -> int8DuckValue i- (DuckDBTypeSmallInt, FieldInt16 i) -> int16DuckValue i- (DuckDBTypeInteger, FieldInt32 i) -> int32DuckValue i- (DuckDBTypeBigInt, FieldInt64 i) -> int64DuckValue i- (DuckDBTypeUTinyInt, FieldWord8 w) -> uint8DuckValue w- (DuckDBTypeUSmallInt, FieldWord16 w) -> uint16DuckValue w- (DuckDBTypeUInteger, FieldWord32 w) -> uint32DuckValue w- (DuckDBTypeUBigInt, FieldWord64 w) -> uint64DuckValue w- (DuckDBTypeFloat, FieldFloat f) -> floatDuckValue f- (DuckDBTypeDouble, FieldDouble d) -> doubleDuckValue d- (DuckDBTypeVarchar, FieldText t) -> textDuckValue t- (DuckDBTypeBlob, FieldBlob b) -> blobDuckValue b- (DuckDBTypeGeometry, FieldGeometry{}) -> unsupportedRawGeometryBinding- (DuckDBTypeUUID, FieldUUID u) -> uuidDuckValue u- (DuckDBTypeBit, FieldBit bits) -> bitDuckValue bits- (DuckDBTypeDate, FieldDate d) -> dateDuckValue d- (DuckDBTypeTime, FieldTime t) -> timeOfDayDuckValue t- (DuckDBTypeTimeNs, FieldTime t) ->- timeOfDayUnits 1000000000 t >>= c_duckdb_create_time_ns . DuckDBTimeNs . fromInteger- (DuckDBTypeTimeTz, FieldTimeTZ tz) -> timeWithZoneDuckValue tz- (DuckDBTypeTimestamp, FieldTimestamp ts) -> localTimestampDuckValue ts- (DuckDBTypeTimestampS, FieldTimestamp ts) ->- encodeUnbounded (encodeTimestampUnits 1) ts >>= c_duckdb_create_timestamp_s . DuckDBTimestampS- (DuckDBTypeTimestampMs, FieldTimestamp ts) ->- encodeUnbounded (encodeTimestampUnits 1000) ts >>= c_duckdb_create_timestamp_ms . DuckDBTimestampMs- (DuckDBTypeTimestampNs, FieldTimestamp ts) ->- encodeUnbounded (encodeTimestampUnits 1000000000) ts >>= c_duckdb_create_timestamp_ns . DuckDBTimestampNs- (DuckDBTypeTimestampTz, FieldTimestampTZ ts) -> utcTimestampDuckValue ts- (DuckDBTypeInterval, FieldInterval iv) -> intervalDuckValue iv- (DuckDBTypeHugeInt, FieldHugeInt i) -> hugeIntDuckValue i- (DuckDBTypeUHugeInt, FieldUHugeInt i) -> uhugeIntDuckValue i- (DuckDBTypeBigNum, FieldBigNum big) -> bigNumDuckValue big- (DuckDBTypeSQLNull, _) -> nullDuckValue- _ ->- case value of- FieldNull -> nullDuckValue- other ->- throwIO- ( userError- ( "duckdb-simple: unsupported scalar conversion for "- <> show dtype- <> " from "- <> show other- )- )--enumDuckValue :: Array Int Text -> Word32 -> IO DuckDBValue-enumDuckValue dict idx = do- when (toInteger idx >= toInteger (length (elems dict))) $- throwIO (userError "duckdb-simple: ENUM index out of range")- bracket (logicalTypeFromRep (LogicalTypeEnum dict)) destroyLogicalType \enumLogical ->- checkedValue (c_duckdb_create_enum_value enumLogical (fromIntegral idx))--hugeIntDuckValue :: Integer -> IO DuckDBValue-hugeIntDuckValue value =- integerToHugeInt value >>= \huge ->- alloca \ptr -> do- poke ptr huge- c_duckdb_create_hugeint ptr--uhugeIntDuckValue :: Integer -> IO DuckDBValue-uhugeIntDuckValue value =- integerToUHugeInt value >>= \uhu ->- alloca \ptr -> do- poke ptr uhu- c_duckdb_create_uhugeint ptr--decimalDuckValue :: DecimalValue -> IO DuckDBValue-decimalDuckValue DecimalValue{decimalWidth, decimalScale, decimalInteger} = do- when (decimalWidth < 1 || decimalWidth > 38 || decimalScale > decimalWidth) $- throwIO (userError "duckdb-simple: invalid DECIMAL width or scale")- when (abs decimalInteger >= 10 ^ decimalWidth) $- throwIO (userError "duckdb-simple: DECIMAL value exceeds declared precision")- huge <- integerToHugeInt decimalInteger- alloca \ptr -> do- poke- ptr- DuckDBDecimal- { duckDBDecimalWidth = decimalWidth- , duckDBDecimalScale = decimalScale- , duckDBDecimalValue = huge- }- c_duckdb_create_decimal ptr--intervalDuckValue :: IntervalValue -> IO DuckDBValue-intervalDuckValue IntervalValue{intervalMonths, intervalDays, intervalMicros} =- alloca \ptr -> do- poke ptr (DuckDBInterval intervalMonths intervalDays intervalMicros)- c_duckdb_create_interval ptr--timeWithZoneDuckValue :: TimeWithZone -> IO DuckDBValue-timeWithZoneDuckValue TimeWithZone{timeWithZoneTime, timeWithZoneZone} = do- totalMicros <- timeOfDayUnits 1000000 timeWithZoneTime- let offsetSeconds = toInteger (timeZoneMinutes timeWithZoneZone) * 60- when (abs offsetSeconds > 57599) $- throwIO (userError "duckdb-simple: TIME WITH TIME ZONE offset out of range")- tzValue <- c_duckdb_create_time_tz (fromIntegral totalMicros) (fromIntegral offsetSeconds)- c_duckdb_create_time_tz_value tzValue--integerToHugeInt :: Integer -> IO DuckDBHugeInt-integerToHugeInt value = do- let minVal = negate (1 `shiftL` 127)- maxVal = (1 `shiftL` 127) - 1- when (value < minVal || value > maxVal) $- throwIO (userError "duckdb-simple: HUGEINT value out of range")- let lowerMask = (1 `shiftL` 64) - 1- lower = fromIntegral (value .&. lowerMask)- upper = fromIntegral (value `shiftR` 64)- pure DuckDBHugeInt{duckDBHugeIntLower = lower, duckDBHugeIntUpper = upper}--integerToUHugeInt :: Integer -> IO DuckDBUHugeInt-integerToUHugeInt value = do- let minVal = 0- maxVal = (1 `shiftL` 128) - 1- when (value < minVal || value > maxVal) $- throwIO (userError "duckdb-simple: UHUGEINT value out of range")- let lowerMask = (1 `shiftL` 64) - 1- lower = fromIntegral (value .&. lowerMask)- upper = fromIntegral (value `shiftR` 64)- pure DuckDBUHugeInt{duckDBUHugeIntLower = lower, duckDBUHugeIntUpper = upper}---- | Release every child handle when construction fails or completes.-withCreatedValues :: [IO DuckDBValue] -> ([DuckDBValue] -> IO a) -> IO a-withCreatedValues = withMany (\action -> bracket (checkedValue action) destroyValue)---- | Reject failed native constructors before a handle is used.-checkedValue :: IO DuckDBValue -> IO DuckDBValue-checkedValue action = do- value <- action- when (value == nullPtr) $- throwIO (userError "duckdb-simple: DuckDB value construction failed")- pure value--withDuckValues :: [DuckDBValue] -> (Ptr DuckDBValue -> IO a) -> IO a-withDuckValues xs action = withArray xs action--typeMismatch :: String -> FieldValue -> IO a-typeMismatch expected actual =- throwIO- ( userError- ( "duckdb-simple: cannot encode "- <> show actual- <> " as "- <> expected- )- )---- | Reject raw values that the C API cannot bind without format conversion.-unsupportedRawGeometryBinding :: IO a-unsupportedRawGeometryBinding =- throwIO (userError "duckdb-simple: raw GEOMETRY binding requires explicit ST_GeomFromWKB and ST_SetCRS parameters")--{- | Construct an owned VARIANT value. A scalar keeps its native type. Lists,-arrays, and STRUCT fields contain VARIANT values. The C API casts the payload-through a one-element VARIANT list.--}-variantDuckValue :: (LogicalTypeRep -> IO DuckDBLogicalType) -> FieldValue -> IO DuckDBValue-variantDuckValue typeFromRep value = do- (rep, payload) <- variantPayloadType value- bracket (typeFromRep (LogicalTypeScalar DuckDBTypeVariant)) destroyLogicalType \variantType ->- withCreatedValues [fieldValueWithTypeDuckValue typeFromRep rep payload] \values ->- withDuckValues values \ptr ->- bracket (checkedValue (c_duckdb_create_list_value variantType ptr 1)) destroyValue \list ->- checkedValue (c_duckdb_get_list_child list 0)--{- | Choose the native type of a VARIANT payload. Containers get VARIANT-elements and fields. Time values with sub-microsecond digits use nanosecond-types. Wide timestamps use milliseconds or seconds when these preserve the-value and fit the native range.--}-variantPayloadType :: FieldValue -> IO (LogicalTypeRep, FieldValue)-variantPayloadType value = case value of- FieldNull -> pure (variant, value)- FieldBool{} -> scalar DuckDBTypeBoolean- FieldInt8{} -> scalar DuckDBTypeTinyInt- FieldInt16{} -> scalar DuckDBTypeSmallInt- FieldInt32{} -> scalar DuckDBTypeInteger- FieldInt64{} -> scalar DuckDBTypeBigInt- FieldWord8{} -> scalar DuckDBTypeUTinyInt- FieldWord16{} -> scalar DuckDBTypeUSmallInt- FieldWord32{} -> scalar DuckDBTypeUInteger- FieldWord64{} -> scalar DuckDBTypeUBigInt- FieldHugeInt{} -> scalar DuckDBTypeHugeInt- FieldUHugeInt{} -> scalar DuckDBTypeUHugeInt- FieldFloat{} -> scalar DuckDBTypeFloat- FieldDouble{} -> scalar DuckDBTypeDouble- FieldDecimal DecimalValue{decimalWidth, decimalScale} -> pure (LogicalTypeDecimal decimalWidth decimalScale, value)- FieldText{} -> scalar DuckDBTypeVarchar- FieldBlob{} -> scalar DuckDBTypeBlob- FieldUUID{} -> scalar DuckDBTypeUUID- FieldDate{} -> scalar DuckDBTypeDate- FieldTime time- | hasNanos time -> scalar DuckDBTypeTimeNs- | otherwise -> scalar DuckDBTypeTime- FieldTimestamp (Finite LocalTime{localDay, localTimeOfDay})- | hasNanos localTimeOfDay -> scalar DuckDBTypeTimestampNs- | inFiniteRange (minBound :: Int64) maxBound micros -> scalar DuckDBTypeTimestamp- | micros `rem` 1000 == 0 && inFiniteRange (minBound :: Int64) maxBound (micros `div` 1000) -> scalar DuckDBTypeTimestampMs- | micros `rem` 1000000 == 0 && inFiniteRange (minBound :: Int64) maxBound (micros `div` 1000000) -> scalar DuckDBTypeTimestampS- where- micros = diffDays localDay (fromGregorian 1970 1 1) * 86400 * 1000000 + diffTimeToPicoseconds (timeOfDayToTime localTimeOfDay) `div` 1000000- FieldTimestamp{} -> scalar DuckDBTypeTimestamp- FieldTimestampTZ{} -> scalar DuckDBTypeTimestampTz- FieldTimeTZ{} -> scalar DuckDBTypeTimeTz- FieldInterval{} -> scalar DuckDBTypeInterval- FieldBigNum{} -> scalar DuckDBTypeBigNum- FieldBit{} -> scalar DuckDBTypeBit- FieldList{} -> pure (LogicalTypeList variant, value)- FieldArray items -> pure (LogicalTypeList variant, FieldList (elems items))- FieldStruct structValue@StructValue{structValueTypes} -> do- let names = map structFieldName (elems structValueTypes)- when (any Text.null names || length names /= length (List.nub names)) $- throwIO (userError "duckdb-simple: VARIANT objects need unique, nonempty keys")- let types = fmap (\field -> field{structFieldValue = variant}) structValueTypes- pure (LogicalTypeStruct types, FieldStruct structValue{structValueTypes = types})- FieldUnion unionValue -> pure (unionValueTypeRep unionValue, value)- FieldGeometry{} -> unsupportedRawGeometryBinding- FieldMap{} -> throwIO (userError "duckdb-simple: VARIANT payloads cannot contain MAP values")- FieldEnum{} -> throwIO (userError "duckdb-simple: VARIANT payloads cannot contain ENUM values")- where- variant = LogicalTypeScalar DuckDBTypeVariant- scalar dtype = pure (LogicalTypeScalar dtype, value)- hasNanos time = snd (properFraction (todSec time * 1000000) :: (Integer, Pico)) /= 0--instance ToDuckValue G.Geometry where- toDuckValue geometry = do- wkt <- either (throwIO . userError) pure (WKT.encodeWKT geometry)- bracket (logicalTypeFromRep (LogicalTypeGeometry Nothing)) destroyLogicalType \logical ->- withCreatedValues [textDuckValue wkt] \values ->- withDuckValues values \ptr ->- bracket (checkedValue (c_duckdb_create_list_value logical ptr 1)) destroyValue \list ->- checkedValue (c_duckdb_get_list_child list 0)--instance ToDuckValue Null where- toDuckValue _ = nullDuckValue--instance ToDuckValue Bool where- toDuckValue = boolDuckValue--instance ToDuckValue Int where- toDuckValue = int64DuckValue . fromIntegral--instance ToDuckValue Int8 where- toDuckValue = int8DuckValue--instance ToDuckValue Int16 where- toDuckValue = int16DuckValue--instance ToDuckValue Int32 where- toDuckValue = int32DuckValue--instance ToDuckValue Int64 where- toDuckValue = int64DuckValue--instance ToDuckValue BigNum where- toDuckValue = bigNumDuckValue--instance ToDuckValue UUID.UUID where- toDuckValue = uuidDuckValue--instance ToDuckValue Integer where- toDuckValue = bigNumDuckValue . BigNum--instance ToDuckValue Natural where- toDuckValue = bigNumDuckValue . BigNum . toInteger--instance ToDuckValue Word where- toDuckValue = uint64DuckValue . fromIntegral--instance ToDuckValue Word16 where- toDuckValue = uint16DuckValue--instance ToDuckValue Word32 where- toDuckValue = uint32DuckValue--instance ToDuckValue Word64 where- toDuckValue = uint64DuckValue--instance ToDuckValue Word8 where- toDuckValue = uint8DuckValue--instance ToDuckValue Double where- toDuckValue = doubleDuckValue--instance ToDuckValue Float where- toDuckValue = floatDuckValue--instance ToDuckValue Text where- toDuckValue = textDuckValue--instance ToDuckValue String where- toDuckValue = stringDuckValue--instance ToDuckValue BS.ByteString where- toDuckValue = blobDuckValue--instance ToDuckValue BitString where- toDuckValue = bitDuckValue--instance ToDuckValue Day where- toDuckValue = dayDuckValue--instance ToDuckValue TimeOfDay where- toDuckValue = timeOfDayDuckValue--instance ToDuckValue LocalTime where- toDuckValue = localTimeDuckValue--instance ToDuckValue UTCTime where- toDuckValue = utcTimeDuckValue--instance ToDuckValue (Unbounded Day) where- toDuckValue = dateDuckValue--instance ToDuckValue (Unbounded LocalTime) where- toDuckValue = localTimestampDuckValue--instance ToDuckValue (Unbounded UTCTime) where- toDuckValue = utcTimestampDuckValue--instance ToDuckValue (StructValue FieldValue) where- toDuckValue = structValueDuckValue logicalTypeFromRep--instance ToDuckValue (UnionValue FieldValue) where- toDuckValue = unionValueDuckValue logicalTypeFromRep--{- | Build an array without a connection. The elements need 'ToDuckValue', so-this instance does not accept t'Variant' elements. 'toField' binds those.--}-instance (DuckDBColumnType a, ToDuckValue a) => ToDuckValue (Array Int a) where- toDuckValue = arrayDuckValue logicalTypeFromRep toDuckValue--instance (ToDuckValue a) => ToDuckValue (Maybe a) where- toDuckValue Nothing = nullDuckValue- toDuckValue (Just value) = toDuckValue value---- | Preserve infinity sentinels and validate finite values before narrowing.-encodeUnbounded :: (Integral b, Bounded b) => (a -> IO b) -> Unbounded a -> IO b-encodeUnbounded _ NegInfinity = pure (negate maxBound)-encodeUnbounded _ PosInfinity = pure maxBound-encodeUnbounded encode (Finite value) = encode value---- | Encode finite dates as days from the Unix epoch.-encodeDay :: Day -> IO DuckDBDate-encodeDay day = do- let days = diffDays day (fromGregorian 1970 1 1)- checkFiniteRange "DATE" (minBound :: Int32) maxBound days- pure (DuckDBDate (fromInteger days))---- | Encode a time of day at DuckDB microsecond precision.-encodeTimeOfDay :: TimeOfDay -> IO DuckDBTime-encodeTimeOfDay tod = DuckDBTime . fromInteger <$> timeOfDayUnits 1000000 tod---- | Encode finite timestamps without native calendar conversions.-encodeLocalTime :: LocalTime -> IO DuckDBTimestamp-encodeLocalTime ts = DuckDBTimestamp <$> encodeTimestampUnits 1000000 ts---- | Encode finite timestamps in the requested number of units per second.-encodeTimestampUnits :: Integer -> LocalTime -> IO Int64-encodeTimestampUnits units LocalTime{localDay, localTimeOfDay} = do- time <- timeOfDayUnits units localTimeOfDay- let total = diffDays localDay (fromGregorian 1970 1 1) * 86400 * units + time- checkFiniteRange "TIMESTAMP" (minBound :: Int64) maxBound total- pure (fromInteger total)---- | Check storage limits and the two DuckDB infinity sentinels.-checkFiniteRange :: (Integral a) => String -> a -> a -> Integer -> IO ()-checkFiniteRange label lower upper value =- when (not (inFiniteRange lower upper value)) $- throwIO (userError ("duckdb-simple: " <> label <> " value out of finite range"))---- | Check storage limits without accepting the two infinity sentinels.-inFiniteRange :: (Integral a) => a -> a -> Integer -> Bool-inFiniteRange lower upper value =- value >= toInteger lower && value < toInteger upper && value /= negate (toInteger upper)---- | Validate time components and convert to the requested units per second.-timeOfDayUnits :: Integer -> TimeOfDay -> IO Integer-timeOfDayUnits units tod@(TimeOfDay hours minutes seconds) = do- when (hours < 0 || hours > 24 || minutes < 0 || minutes > 59 || seconds < 0 || seconds >= 61 || (hours == 24 && (minutes /= 0 || seconds /= 0))) $- throwIO (userError "duckdb-simple: invalid time of day")- let value = diffTimeToPicoseconds (timeOfDayToTime tod) `div` (1000000000000 `div` units)- when (value > 86400 * units) $- throwIO (userError "duckdb-simple: time of day out of range")- pure value--bindDuckValue :: Statement -> DuckDBIdx -> IO DuckDBValue -> IO ()-bindDuckValue stmt idx makeValue =- withStatementHandle stmt \handle ->- bracket (checkedValue makeValue) destroyValue \value -> do- rc <- c_duckdb_bind_value handle idx value- when (rc /= DuckDBSuccess) $ do- err <- fetchPrepareError (Text.pack "duckdb-simple: parameter binding failed") handle- throwBindError stmt err--throwBindError :: Statement -> Text -> IO a-throwBindError Statement{statementQuery} msg =- throwIO- SQLError- { sqlErrorMessage = msg- , sqlErrorType = Nothing- , sqlErrorQuery = Just statementQuery- }+{-# LANGUAGE MagicHash #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : Database.DuckDB.Simple.ToField+-- Description : Convert Haskell parameters into DuckDB bindable values.+--+-- The @ToField@ class mirrors the interface provided by @sqlite-simple@ while+-- delegating to the DuckDB C API under the hood.+--+-- A 'ToDuckValue' instance of a new type needs no call of the C API. A newtype+-- instance uses the instance of the wrapped type, for example+-- @toDuckValue factory (Age years) = toDuckValue factory years@, and the+-- newtype also needs a 'DuckDBColumnType' instance. A record or a sum type gets+-- both instances through 'Database.DuckDB.Simple.Generic.ViaDuckDB':+--+-- > data User = User {userId :: Int64, userName :: Text}+-- > deriving stock (Generic)+-- > deriving (DuckDBColumnType, ToDuckValue) via (ViaDuckDB User)+module Database.DuckDB.Simple.ToField (+ FieldBinding,+ ToDuckValue (..),+ ValueFactory,+ NativeValue,+ ToField (..),+ DuckDBColumnType (..),+ NamedParam (..),+ duckdbColumnType,+ bindFieldBinding,+ renderFieldBinding,+) where++import Control.Exception (bracket, mask_, onException)+import Control.Monad (foldM, forM_, unless, when)+import Data.Array (Array, bounds, elems, listArray)+import Data.Bits (complement, shiftL, shiftR, xor, (.|.))+import qualified Data.ByteString as BS+import Data.Fixed (Pico)+import qualified Data.Geometry as G+import qualified Data.Geometry.WKT as WKT+import Data.Int (Int16, Int32, Int64, Int8)+import qualified Data.Map.Strict as Map+import Data.Proxy (Proxy (..))+import qualified Data.Set as Set+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Time.Calendar (Day, diffDays, fromGregorian)+import Data.Time.Clock (UTCTime (..), diffTimeToPicoseconds)+import Data.Time.LocalTime (LocalTime (..), TimeOfDay (..), timeOfDayToTime, utc, utcToLocalTime)+import qualified Data.UUID as UUID+import Data.Word (Word16, Word32, Word64, Word8)+import Database.DuckDB.Simple.Element (containsType, timeTzMaxOffsetSeconds, timeTzOffsetBits, valueTypeName)+import Database.DuckDB.Simple.FromField (+ BigNum (..),+ BitString (..),+ DecimalValue (..),+ FieldValue (..),+ IntervalValue (..),+ SizedArray,+ TimeWithZone (..),+ sizedArrayToArray,+ )+import Database.DuckDB.Simple.Geometry (RawGeometry (..))+import Database.DuckDB.Simple.Internal (+ Statement (..),+ connectionInstance,+ duckDBTypeFromName,+ duckDBTypeToName,+ preparedParameterType,+ storeBoundValue,+ withConnectionHandle,+ withInstanceHandle,+ )+import Database.DuckDB.Simple.LogicalRep (+ LogicalTypeRep (..),+ StructField (..),+ StructValue (..),+ TupleValue (..),+ TypeDeclaration (..),+ UnionMemberType (..),+ UnionValue (..),+ logicalTypeSQL,+ structValueTypeRep,+ tupleValueTypeRep,+ unionValueTypeRep,+ withoutAliases,+ )+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.Time (Date, LocalTimestamp, UTCTimestamp, Unbounded (..))+import Database.DuckDB.Simple.Types (Null (..))+import Database.DuckDB.Simple.V2.LogicalRep (+ bindingLogicalType,+ logicalTypeFromRep,+ logicalTypeToBaseRep,+ logicalTypeToRep,+ valuesHaveType,+ )+import Database.DuckDB.Simple.Variant (Variant (..))+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.C.Types (CDouble (..), CFloat (..))+import Foreign.Marshal (fromBool)+import Foreign.Marshal.Alloc (alloca, allocaBytes)+import Foreign.Marshal.Array (withArray)+import Foreign.Marshal.Utils (with)+import Foreign.Ptr (Ptr, castPtr, nullPtr)+import Foreign.Storable (peek, poke)+import GHC.Exts (Ptr (..))+import GHC.Num.Integer (integerLog2, integerToAddr)+import GHC.TypeLits (KnownNat, natVal)+import Numeric.Natural (Natural)++-- | Represents a named parameter binding using the @:=@ operator.+data NamedParam where+ (:=) :: (ToField a) => Text -> a -> NamedParam++infixr 3 :=++-- | Encapsulates the action required to bind a single positional parameter, together with a textual description used in diagnostics.+data FieldBinding = FieldBinding+ { fieldBindingValue :: !(ValueBuilder -> IO (Maybe LogicalTypeRep) -> IO DuckDBV2ValueHandle)+ , fieldBindingDisplay :: !String+ }++-- | A factory and a type constructor of the same scope. The builder of a+-- statement resolves CRS types when it binds a parameter. See+-- Note [Bind-time CRS types] in "Database.DuckDB.Simple.V2.LogicalRep".+data ValueBuilder = ValueBuilder+ { builderFactory :: !DuckDBV2FactoryHandle+ , builderType :: !(LogicalTypeRep -> IO DuckDBV2LogicalTypeHandle)+ }++-- | A builder that constructs each type directly through the factory.+plainBuilder :: DuckDBV2FactoryHandle -> ValueBuilder+plainBuilder factory = ValueBuilder factory (logicalTypeFromRep factory)++-- | Low-level class for values that convert directly to a 'NativeValue'.+class (DuckDBColumnType a) => ToDuckValue a where+ -- | Construct an owned value through a borrowed native factory. Keep the+ -- owner of the factory (a connection, a callback context, or an instance)+ -- alive until you destroy the value. This is a rule of the library. The+ -- header says that a factory is valid for as long as its owner is. The+ -- header gives no rule for a value after the owner of its factory ends.+ -- A value with a GEOMETRY type that has a CRS needs the factory of a+ -- connection or of a callback context, because the factory of an+ -- instance cannot make that type. Use 'ToField' to resolve a custom CRS+ -- label when binding a parameter.+ toDuckValue :: ValueFactory -> a -> IO NativeValue++valueBinding :: String -> (DuckDBV2FactoryHandle -> IO DuckDBV2ValueHandle) -> FieldBinding+valueBinding display make = builderValueBinding display (make . builderFactory)++-- | Construct a value with the builder of the statement's connection.+builderValueBinding :: String -> (ValueBuilder -> IO DuckDBV2ValueHandle) -> FieldBinding+builderValueBinding display makeValue =+ targetValueBinding display (\builder _ -> makeValue builder)++-- | Construct a value with the prepared parameter type, when it is available.+targetValueBinding :: String -> (ValueBuilder -> IO (Maybe LogicalTypeRep) -> IO DuckDBV2ValueHandle) -> FieldBinding+targetValueBinding display makeValue =+ FieldBinding+ { fieldBindingValue = makeValue+ , fieldBindingDisplay = display+ }++{- Note [Nanosecond parameters]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+A TimeOfDay, a LocalTime, and a UTCTime hold picoseconds. A TIME, a TIMESTAMP,+and a TIMESTAMPTZ hold microseconds. TIME_NS, TIMESTAMP_NS, and TIMESTAMPTZ_NS+hold nanoseconds. timeValueBinding binds these Haskell values, and their+Unbounded forms, with this rule:++- A value with no digits below a microsecond binds as a microsecond value. The+ binding does not read the parameter type.+- A value with digits below a microsecond makes the binding read the parameter+ type. A nanosecond parameter (TIME_NS, TIMESTAMP_NS, or TIMESTAMPTZ_NS for+ the three types) gets a nanosecond value. A VARIANT parameter also gets a+ nanosecond value, when the value fits the range of the nanosecond type+ (fitsNanoseconds). A VARIANT holds TIME_NS, TIMESTAMP_NS, and TIMESTAMPTZ_NS+ values. A value outside that range, for example in the year 3000, gets the+ microsecond value, as for a parameter with no known type. The payload of a+ Variant follows the same rule (variantPayloadType). Each other parameter,+ and a parameter with no known type, gets the microsecond value. The binding+ discards the digits below one microsecond, as the SQL literal does.+- The binding discards the digits below one nanosecond in each case.++A value in a SizedArray, a list, a Map, a STRUCT, a TUPLE, or a UNION parameter+follows the type of its leaf in the target: timestampTypeForTarget gives a+nanosecond type to a TIME, TIMESTAMP, or TIMESTAMPTZ leaf of a nanosecond+target leaf. These parameters read the parameter type for each value. A+VARIANT leaf does not give a nanosecond type: bindElement gives an element of+a VARIANT list or map no target, and timestampTypeForTarget has no VARIANT+case. A list, a SizedArray, or a Map for a VARIANT parameter has no element+type in the target (collectionChildType and mapPartTypes give Nothing), so its+time elements also get the microsecond value. One LIST has one element type+for all its elements. For a Haskell list, a SizedArray, or a Map of TimeOfDay,+LocalTime, or UTCTime values, withElementType declares the microsecond type of+the element, for example TIMESTAMP for a LocalTime, so a nanosecond element+would be cast to that type, and the cast rounds. A FieldList or a FieldMap+takes the type that its elements give, as the SQL list literal does+(withValueType). TIME and TIME_NS elements, and+TIMESTAMPTZ and TIMESTAMPTZ_NS elements, have no common type, and DuckDB raises+an error. TIMESTAMP and TIMESTAMP_NS elements give TIMESTAMP_NS[], but an+element outside the range of TIMESTAMP_NS stays a TIMESTAMP (fitsNanoseconds),+and DuckDB cannot cast that element to TIMESTAMP_NS. Thus a time element gets+the microsecond value in each of these collections, because a nanosecond value+would round or raise an error in some of them.++- Why the binding reads the parameter type only for such digits. A read of the+ parameter type runs the binder of DuckDB. See Note [Statement schemas] in+ "Database.DuckDB.Simple.Internal". A microsecond value gives the same result+ for each parameter type, so the read has no use for it.++- Why a microsecond parameter gets the microsecond value. In DuckDB 2.0.0, the+ SQL literal '03:04:05.123456999'::TIME gives 03:04:05.123456, but a cast from+ TIME_NS to TIME rounds and gives 03:04:05.123457. The same holds for+ TIMESTAMP. A nanosecond value would thus change what a microsecond column+ stores.++- An explicit cast of the parameter to a nanosecond type, as in+ ?::TIMESTAMP_NS, makes it a nanosecond parameter. DuckDB then rounds when it+ casts the value to a microsecond type, for example for a TIMESTAMP column or+ in ?::TIMESTAMP_NS + INTERVAL 1 DAY. A plain parameter for a microsecond+ type cuts the digits.++- A LocalTime for a TIMESTAMP_NS parameter, or for the TIMESTAMP_NS elements+ of a SizedArray, a list, or a Map, is in one of three zones:+ * In the range of TIMESTAMP_NS (about the years 1677 to 2262): no error.+ * Outside the range, when DuckDB also cannot cast the microsecond value to+ TIMESTAMP_NS (outsideNanosecondRange): the value binds as a microsecond+ value. For a parameter, DuckDB raises its SQLError "Could not convert+ Timestamp to higher precision", of the type ErrorTypeConversion, when it+ casts the value. For an element, the native constructor of the LIST, the+ ARRAY, or the MAP casts each element when the value binds, and it would+ raise the ErrorInputInvalid "Failed to cast value". Before that cast,+ checkNanosecondElements raises the SQLError of the library "duckdb-simple:+ TIMESTAMP value out of finite range", of the type ErrorTypeConversion, for+ each element that the cast cannot hold. Thus the element gets the type of+ a parameter, and no element gets an error that the cast does not raise.+ * From 2262-04-11 23:47:16.854775807 to 23:47:16.854775999: DuckDB can cast+ the microsecond value, but the cast cuts the digits with no error. Thus+ the encoder raises the SQLError of the library "duckdb-simple: TIMESTAMP+ value out of finite range", of the type ErrorTypeConversion. The low end of+ the range has no such zone, because the microsecond value is cut toward+ the earlier time.+ A UTCTime for a TIMESTAMPTZ_NS parameter, or for the TIMESTAMPTZ_NS elements+ of a SizedArray, a list, or a Map, is in the same three zones, on its instant in+ UTC. Each leaf of a STRUCT, TUPLE, or UNION outside the range gets the error+ of the encoder, because in TRY_CAST DuckDB keeps the other leaves of a+ microsecond value with their digits cut.+-}++-- | Test whether a time of day has digits below one microsecond. See+-- Note [Nanosecond parameters].+hasSubMicroseconds :: TimeOfDay -> Bool+hasSubMicroseconds time = snd (properFraction (todSec time * 1000000) :: (Integer, Pico)) /= 0++-- | Bind a TIME, TIMESTAMP, or TIMESTAMPTZ value. The source is the microsecond+-- type of the value, and the last argument makes the microsecond value. See+-- Note [Nanosecond parameters].+timeValueBinding :: String -> LogicalTypeRep -> FieldValue -> (ValueBuilder -> IO DuckDBV2ValueHandle) -> FieldBinding+timeValueBinding display source value makeDefault+ | subMicroseconds value = temporalValueBinding display source value makeDefault+ | otherwise = builderValueBinding display makeDefault+ where+ subMicroseconds (FieldTime time) = hasSubMicroseconds time+ subMicroseconds (FieldTimestamp (Finite LocalTime{localTimeOfDay})) = hasSubMicroseconds localTimeOfDay+ subMicroseconds (FieldTimestampTZ (Finite instant)) = hasSubMicroseconds (localTimeOfDay (utcToLocalTime utc instant))+ subMicroseconds _ = False++-- | Bind a value with the nanosecond types of the prepared parameter. See+-- Note [Nanosecond parameters].+temporalValueBinding :: String -> LogicalTypeRep -> FieldValue -> (ValueBuilder -> IO DuckDBV2ValueHandle) -> FieldBinding+temporalValueBinding display source value makeDefault =+ targetValueBinding display \builder readTarget -> do+ target <- readTarget+ let precise = case (target, source) of+ -- A time value with digits below a microsecond for a VARIANT+ -- parameter. See Note [Nanosecond parameters].+ (Just (LogicalTypeScalar LogicalTypeIdVariant), LogicalTypeScalar _)+ | fitsNanoseconds value -> nanosecondType source+ _ -> timestampTypeForTarget target source+ if precise == source || outsideNanosecondRange precise value+ then makeDefault builder+ else fieldValueWithTypeDuckValue builder precise (valueWithTypeMetadata precise value)++-- | Give the nanosecond type of a TIME, TIMESTAMP, or TIMESTAMPTZ type.+nanosecondType :: LogicalTypeRep -> LogicalTypeRep+nanosecondType (LogicalTypeScalar LogicalTypeIdTime) = LogicalTypeScalar LogicalTypeIdTimeNs+nanosecondType (LogicalTypeScalar LogicalTypeIdTimestamp) = LogicalTypeScalar LogicalTypeIdTimestampNs+nanosecondType (LogicalTypeScalar LogicalTypeIdTimestampTz) = LogicalTypeScalar LogicalTypeIdTimestampTzNs+nanosecondType rep = rep++-- | Test whether the nanosecond value of a time value fits the range of its+-- nanosecond type, with the range test of the encoder. Each time of day fits.+fitsNanoseconds :: FieldValue -> Bool+fitsNanoseconds (FieldTimestamp (Finite time)) = inFiniteRange (minBound :: Int64) maxBound (timestampUnits 1000000000 time)+fitsNanoseconds (FieldTimestampTZ (Finite instant)) = fitsNanoseconds (FieldTimestamp (Finite (utcToLocalTime utc instant)))+fitsNanoseconds _ = True++-- | Bind an element of a LIST, an ARRAY, or a MAP for its element type in the+-- target. A VARIANT element type gives no target, so the element binds as for+-- a parameter with no known type. See Note [Nanosecond parameters].+bindElement :: (ToField a) => ValueBuilder -> Maybe LogicalTypeRep -> a -> IO DuckDBV2ValueHandle+bindElement builder target value = fieldBindingValue (toField value) builder (pure elementTarget)+ where+ elementTarget = case target of+ Just (LogicalTypeScalar LogicalTypeIdVariant) -> Nothing+ _ -> target++-- | Test whether a t'LocalTime' for a TIMESTAMP_NS parameter, or a t'UTCTime'+-- for a TIMESTAMPTZ_NS parameter on its instant in UTC, is outside the range of+-- the nanosecond type, and DuckDB also cannot cast its microsecond value to that+-- type. See Note [Nanosecond parameters].+outsideNanosecondRange :: LogicalTypeRep -> FieldValue -> Bool+outsideNanosecondRange (LogicalTypeScalar LogicalTypeIdTimestampNs) (FieldTimestamp (Finite time)) =+ not (fits (timestampUnits 1000000000 time)) && not (fits (1000 * timestampUnits 1000000 time))+ where+ fits = inFiniteRange (minBound :: Int64) maxBound+outsideNanosecondRange (LogicalTypeScalar LogicalTypeIdTimestampTzNs) (FieldTimestampTZ (Finite instant)) =+ outsideNanosecondRange (LogicalTypeScalar LogicalTypeIdTimestampNs) (FieldTimestamp (Finite (utcToLocalTime utc instant)))+outsideNanosecondRange _ _ = False++-- | Read the element type of a prepared LIST or ARRAY parameter.+collectionChildType :: Maybe LogicalTypeRep -> Maybe LogicalTypeRep+collectionChildType (Just (LogicalTypeList child)) = Just child+collectionChildType (Just (LogicalTypeArray child _)) = Just child+collectionChildType _ = Nothing++-- | Read the key type and the value type of a prepared MAP parameter.+mapPartTypes :: Maybe LogicalTypeRep -> (Maybe LogicalTypeRep, Maybe LogicalTypeRep)+mapPartTypes (Just (LogicalTypeMap key value)) = (Just key, Just value)+mapPartTypes _ = (Nothing, Nothing)++-- | Give a nanosecond type to each TIME, TIMESTAMP, and TIMESTAMPTZ leaf of the+-- source whose target leaf has that nanosecond type. See+-- Note [Nanosecond parameters].+timestampTypeForTarget :: Maybe LogicalTypeRep -> LogicalTypeRep -> LogicalTypeRep+timestampTypeForTarget target source = case (target, source) of+ -- Keep an alias only when its base type stays unchanged.+ (_, LogicalTypeAlias _ base) ->+ let promoted = timestampTypeForTarget target base+ in if promoted == base then source else promoted+ (Just (LogicalTypeScalar LogicalTypeIdTimestampTzNs), LogicalTypeScalar LogicalTypeIdTimestampTz) ->+ LogicalTypeScalar LogicalTypeIdTimestampTzNs+ (Just (LogicalTypeScalar LogicalTypeIdTimestampNs), LogicalTypeScalar LogicalTypeIdTimestamp) ->+ LogicalTypeScalar LogicalTypeIdTimestampNs+ (Just (LogicalTypeScalar LogicalTypeIdTimeNs), LogicalTypeScalar LogicalTypeIdTime) ->+ LogicalTypeScalar LogicalTypeIdTimeNs+ (_, LogicalTypeList child) ->+ maybe source (\expected -> LogicalTypeList (timestampTypeForTarget (Just expected) child)) (collectionChildType target)+ (_, LogicalTypeArray child size) ->+ maybe source (\expected -> LogicalTypeArray (timestampTypeForTarget (Just expected) child) size) (collectionChildType target)+ (Just (LogicalTypeMap targetKey targetValue), LogicalTypeMap key value) ->+ LogicalTypeMap (timestampTypeForTarget (Just targetKey) key) (timestampTypeForTarget (Just targetValue) value)+ (Just (LogicalTypeStruct targetFields), LogicalTypeStruct fields) ->+ let sourceFields = elems fields+ targets = [(identifier name, rep) | StructField name rep <- elems targetFields]+ matching = [lookup (identifier (structFieldName field)) targets | field <- sourceFields]+ updated = zipWith (\field expected -> field{structFieldValue = timestampTypeForTarget expected (structFieldValue field)}) sourceFields matching+ in LogicalTypeStruct (listArray (bounds fields) updated)+ (Just (LogicalTypeTuple targetFields), LogicalTypeStruct fields) ->+ let matching = map Just (elems targetFields) <> repeat Nothing+ updated = zipWith (\field expected -> field{structFieldValue = timestampTypeForTarget expected (structFieldValue field)}) (elems fields) matching+ in LogicalTypeStruct (listArray (bounds fields) updated)+ (Just (LogicalTypeTuple targetFields), LogicalTypeTuple fields) ->+ positionalTuple (elems targetFields) fields+ (Just (LogicalTypeStruct targetFields), LogicalTypeTuple fields) ->+ positionalTuple (map structFieldValue (elems targetFields)) fields+ (Just (LogicalTypeUnion targetMembers), LogicalTypeUnion members) ->+ let targets = [(identifier name, rep) | UnionMemberType name rep <- elems targetMembers]+ in LogicalTypeUnion+ ( fmap+ (\member -> member{unionMemberType = timestampTypeForTarget (lookup (identifier (unionMemberName member)) targets) (unionMemberType member)})+ members+ )+ _ -> source+ where+ -- DuckDB compares identifiers without ASCII case.+ identifier = foldIdentifier+ positionalTuple targets fields = LogicalTypeTuple (listArray (bounds fields) (zipWith timestampTypeForTarget (map Just targets <> repeat Nothing) (elems fields)))++-- | Copy adapted type metadata into nested values without changing their payloads.+valueWithTypeMetadata :: LogicalTypeRep -> FieldValue -> FieldValue+valueWithTypeMetadata logical value = case (logical, value) of+ (LogicalTypeList child, FieldList values) -> FieldList (map (valueWithTypeMetadata child) values)+ (LogicalTypeArray child _, FieldArray values) -> FieldArray (fmap (valueWithTypeMetadata child) values)+ (LogicalTypeMap keyType valueType, FieldMap pairs) ->+ FieldMap [(valueWithTypeMetadata keyType key, valueWithTypeMetadata valueType item) | (key, item) <- pairs]+ (LogicalTypeStruct types, FieldStruct struct) ->+ let sourceFields = elems (structValueFields struct)+ fields = [(structFieldName field, structFieldValue field) | field <- elems types]+ matching = [lookup (structFieldName field) fields | field <- sourceFields]+ update field expected = case expected of+ Nothing -> field+ Just rep -> field{structFieldValue = valueWithTypeMetadata rep (structFieldValue field)}+ in FieldStruct+ struct{structValueTypes = types, structValueFields = listArray (bounds (structValueFields struct)) (zipWith update sourceFields matching)}+ (LogicalTypeTuple types, FieldTuple tuple) ->+ let update item expected = maybe item (`valueWithTypeMetadata` item) expected+ fields = zipWith update (elems (tupleValueFields tuple)) (map Just (elems types) <> repeat Nothing)+ in FieldTuple tuple{tupleValueTypes = types, tupleValueFields = listArray (bounds (tupleValueFields tuple)) fields}+ (LogicalTypeUnion members, FieldUnion union) ->+ let payload = case drop (fromIntegral (unionValueIndex union)) (elems members) of+ member : _ -> valueWithTypeMetadata (unionMemberType member) (unionValuePayload union)+ [] -> unionValuePayload union+ in FieldUnion union{unionValueMembers = members, unionValuePayload = payload}+ _ -> value++{- Note [Type declarations]+~~~~~~~~~~~~~~~~~~~~~~~~~~~+duckdbTypeDeclaration gives the SQL type that a typed registration declares for+a Haskell type: an argument or the result of a scalar function, a parameter+or a column of a typed table function, and an argument or the result of a+typed aggregate function. Right gives the declaration. Left gives the reason+why the type has no declaration.++- The default is the text of duckdbColumnTypeFor. DuckDB parses the text when+ the registration declares the type. Thus a newtype with its own+ DuckDBColumnType instance gives its own type text, for example+ DECIMAL(18,3).+- Integer and Natural give HUGEINT and UHUGEINT, but their text is BIGNUM.+ See Note [Integer parameters].+- ViaDuckDB gives the structured type of the record or the sum type+ (genericLogicalType in "Database.DuckDB.Simple.Generic"). A SizedArray+ field, or a field of a newtype whose DuckValue instance gives the size,+ declares an ARRAY.+- Maybe a has the declaration of a.+- A list gives a LIST (listDeclaration): the text T[] for the element text T,+ or a structured LIST for a structured element type.+- A SizedArray gives the ARRAY of its size (arrayDeclaration, see+ Note [Sized arrays]). A newtype with the type text of one size, for example+ BIGINT[3], also declares that ARRAY. DuckDB then rejects an ARRAY of another+ size when it binds the call (a Binder Error). It casts a LIST of the same length to the ARRAY with no error,+ because the cast from LIST to ARRAY is implicit in DuckDB. A LIST of+ another length raises the conversion error of DuckDB when the statement+ runs.+- A Map gives a MAP (mapDeclaration). Two texts give the text MAP(K, V). In+ the other cases, a structured MAP needs a scalar type name as the text of a+ part (duckDBTypeFromName). With another text, a structured part that is a+ scalar type, or a LIST or a MAP of scalar types, gives its text, and the+ declaration is the text MAP(K, V), for example MAP(INTEGER[], HUGEINT[]).+ Each other case gives Left.+- StructValue, TupleValue, and UnionValue give Left. They have no static+ field types. Their text is the bare name STRUCT, TUPLE, or UNION. The bare+ text STRUCT is a valid type, a STRUCT with no fields. That type is also the+ type of a record with no fields. Thus a check of the declared type cannot+ find these three types.++A registration raises the Left text as an SQLError before it registers the+SQL object. For a parameter with no known type, a LIST or a MAP value with no+element that is not NULL takes its element type from a structured declaration+(withElementType). See Note [Element types of a list]. The ToDuckValue+instances of a list, a SizedArray, and a Map give the structured declaration of+the element as its target (declaredTarget).+-}++-- | Types that map to a concrete DuckDB column type when used with @ToField@.+-- A newtype with its own instance gives its own type text. A 'SizedArray' gives the+-- text of its ARRAY type, for example @BIGINT[3]@ for @SizedArray 3 Int64@.+-- See Note [Type declarations].+class DuckDBColumnType a where+ -- | The DuckDB type text of @a@.+ duckdbColumnTypeFor :: Proxy a -> Text++ -- | The SQL type that a typed registration declares for @a@, or the reason+ -- why @a@ has no declaration. The default is the text of+ -- 'duckdbColumnTypeFor'. A list declares a LIST, and a 'SizedArray' an+ -- ARRAY. See Note [Type declarations].+ duckdbTypeDeclaration :: Proxy a -> Either Text TypeDeclaration+ duckdbTypeDeclaration proxy = Right (TypeExpression (duckdbColumnTypeFor proxy))++-- | The structured declaration of a type, as the target type of an element.+-- 'Nothing' for a type text or a type with no declaration. See+-- Note [Type declarations].+declaredTarget :: (DuckDBColumnType a) => Proxy a -> Maybe LogicalTypeRep+declaredTarget proxy = case duckdbTypeDeclaration proxy of+ Right (StructuredType rep) -> Just rep+ _ -> Nothing++-- | Declare a LIST with the declaration of its element type. See+-- Note [Type declarations].+listDeclaration :: TypeDeclaration -> TypeDeclaration+listDeclaration (TypeExpression element) = TypeExpression (element <> "[]")+listDeclaration (StructuredType element) = StructuredType (LogicalTypeList element)++-- | Declare a MAP with the declarations of its key type and its value type.+-- See Note [Type declarations].+mapDeclaration :: TypeDeclaration -> TypeDeclaration -> Either Text TypeDeclaration+mapDeclaration key value = case (key, value) of+ (TypeExpression keyName, TypeExpression valueName) -> Right (TypeExpression (mapText keyName valueName))+ _ -> case LogicalTypeMap <$> structured key <*> structured value of+ Right rep -> Right (StructuredType rep)+ Left reason -> maybe (Left reason) (Right . TypeExpression) (mapText <$> typeText key <*> typeText value)+ where+ mapText keyName valueName = "MAP(" <> keyName <> ", " <> valueName <> ")"+ -- A scalar type, and a LIST or a MAP of such types, has a type text, for+ -- example HUGEINT[] for a list of Integer values.+ typeText (TypeExpression text) = Just text+ typeText (StructuredType rep) = repText rep+ repText (LogicalTypeScalar dtype) = Just (duckDBTypeToName dtype)+ repText (LogicalTypeList element) = (<> "[]") <$> repText element+ repText (LogicalTypeMap keyRep valueRep) = mapText <$> repText keyRep <*> repText valueRep+ repText _ = Nothing+ structured (StructuredType rep) = Right rep+ structured (TypeExpression text) = case duckDBTypeFromName text of+ Just dtype -> Right (LogicalTypeScalar dtype)+ Nothing ->+ Left+ ( "a MAP with the type text "+ <> text+ <> " and a structured type has no declaration. Use a newtype of the Map with its own DuckDBColumnType instance."+ )++-- | The reason why a type with no static field types has no declaration. See+-- Note [Type declarations].+noStaticFieldTypes :: Text -> Text+noStaticFieldTypes name =+ name <> " has no static field types. Use a record with ViaDuckDB, or a newtype with its own DuckDBColumnType instance."++-- | Report the DuckDB column type that best matches a given @ToField@ instance.+duckdbColumnType :: forall a. (DuckDBColumnType a) => Proxy a -> Text+duckdbColumnType = duckdbColumnTypeFor++-- | Apply a @FieldBinding@ to the given statement/index. The builder takes its+-- factory from the statement's connection. See Note [Bind-time CRS types]+-- in "Database.DuckDB.Simple.V2.LogicalRep".+bindFieldBinding :: Statement -> DuckDBIdx -> FieldBinding -> IO ()+bindFieldBinding stmt idx FieldBinding{fieldBindingValue} =+ withConnectionHandle (statementConnection stmt) \connection -> withConnectionFactory connection \factory ->+ let makeType rep = withInstanceHandle (connectionInstance (statementConnection stmt)) \inst -> bindingLogicalType factory inst rep+ in bindDuckValue stmt idx (fieldBindingValue (ValueBuilder factory makeType) (preparedParameterType stmt idx))++-- | Render a bound parameter for error reporting.+renderFieldBinding :: FieldBinding -> String+renderFieldBinding FieldBinding{fieldBindingDisplay} = fieldBindingDisplay++-- | Types that can be used as positional parameters.+class ToField a where+ toField :: a -> FieldBinding+ default toField :: (Show a, ToDuckValue a) => a -> FieldBinding+ toField value = valueBinding (show value) (\factory -> toDuckValue factory value)++instance ToField Null where+ toField Null = nullBinding "NULL"++instance ToField Bool++instance ToField Int++instance ToField Int8++instance ToField Int16++instance ToField Int32++instance ToField Int64++-- | A value from -2^127 to 2^127 - 1 binds as a HUGEINT. A value outside this+-- range binds as a UHUGEINT for a UHUGEINT parameter that can hold it, and as a+-- BIGNUM for a parameter that has no known type or is not an integer or DECIMAL+-- type. For another integer or DECIMAL parameter, such a value raises+-- t'SQLError'. Bind a t'BigNum' to bind a BIGNUM value in each case. See+-- Note [Integer parameters].+instance ToField Integer where+ toField = integerBinding++-- | A value from 0 to 2^127 - 1 binds as a HUGEINT. A larger value binds as a+-- UHUGEINT for a UHUGEINT parameter that can hold it, and as a BIGNUM for a+-- parameter that has no known type or is not an integer or DECIMAL type. For+-- another integer or DECIMAL parameter, such a value raises t'SQLError'. Bind a+-- t'BigNum' to bind a BIGNUM value in each case. See Note [Integer parameters].+instance ToField Natural where+ toField = integerBinding . toInteger++{- Note [Integer parameters]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~+integerBinding binds an Integer or a Natural parameter with this rule:++- A value from -2^127 to 2^127 - 1 binds as a HUGEINT. The binding does not+ read the parameter type.+- A value outside this range makes the binding read the parameter type. A+ UHUGEINT parameter that can hold the value gets a UHUGEINT value. Another+ integer parameter (TINYINT to UHUGEINT) or a DECIMAL parameter gets an+ SQLError with the type ErrorTypeConversion. Each other parameter, and a+ parameter with no known type, gets a BIGNUM value.++- Why a HUGEINT. In DuckDB 2.0.0, the casts from BIGNUM to the integer types+ have defects (Bignum::BignumToInt in+ src/include/duckdb/common/types/bignum.hpp):+ - A cast to HUGEINT or UHUGEINT fails for each value except 0 and -1.+ - A cast of a negative value to an unsigned type gives a wrapped value and+ no error. For example, -5 for a UBIGINT gives 18446744073709551611.+ - A cast reduces a value of 2^128 or more modulo 2^128 before it checks the+ range. For example, 2^128 + 5 for a BIGINT gives 5.+ DuckDB 2.0.0 has no cast from BIGNUM to DECIMAL. DuckDB checks the range of+ each cast from HUGEINT and UHUGEINT. For TINYINT to UBIGINT, DOUBLE,+ VARCHAR, BIGNUM, JSON, and VARIANT, the cast from HUGEINT gives the same+ result as the cast from BIGNUM where the cast from BIGNUM has no defect.++- Why the binding reads the parameter type only outside the HUGEINT range. A+ read of the parameter type runs the binder of DuckDB. See+ Note [Statement schemas] in "Database.DuckDB.Simple.Internal". A value in the+ HUGEINT range does not have this cost.++- What stays BIGNUM:+ - A value outside the HUGEINT range for a parameter with no known type, as+ in SELECT ?, binds as a BIGNUM. A cast that DuckDB adds later can still+ have the defects above.+ - A BigNum always binds as a BIGNUM.++ Collection values infer their common type as described below and in+ Note [Element types of a list].++- toDuckValue. ToDuckValue Integer and ToDuckValue Natural use the same range+ rule (hugeIntInRange): a value from -2^127 to 2^127 - 1 gives a HUGEINT+ value. A Natural from 2^127 to 2^128 - 1 gives a UHUGEINT value. Each other+ value raises an SQLError of the type ErrorTypeConversion (outsideRange).+ toDuckValue has no target type. A BIGNUM value would thus reach a HUGEINT+ column, and DuckDB 2.0.0 casts the BIGNUM 2^128 to the HUGEINT 0 with no+ error. The callers of toDuckValue are setConfigValue, defaultValue, the+ typed rows of table functions and COPY FROM, and the writer of a scalar+ function result. Use BigNum for a larger value.+ The default toField of the ToField class also calls toDuckValue. Thus a+ newtype over Integer with the default toField raises this error for each+ parameter value outside the HUGEINT range. An Integer parameter uses the+ rule of integerBinding above, for example a BIGNUM value for a parameter+ with no known type. Derive ToField with newtype deriving to keep that rule.++- Typed registrations. The DuckDBColumnType text of Integer and Natural stays+ BIGNUM, the type that holds each value: the element rule of a list+ parameter and the error texts use it. The declaration is HUGEINT for+ Integer and UHUGEINT for Natural, as for a field of a generic record (the+ DuckValue instances in "Database.DuckDB.Simple.Generic"). Thus a column, an+ argument, and a result of a typed registration have the type of the field.+ The ToDuckValue instances of a list, a SizedArray, and a Map give the+ declaration of the element as its target, so a list of Integer values in a+ typed row has HUGEINT elements. See Note [Type declarations].++- Arrays, lists, and maps. DuckDB casts each element of an ARRAY or a LIST+ value, and each key and value of a MAP value, to the declared type when it+ builds the value. The DuckDBColumnType of Integer, Natural, and BigNum is+ BIGNUM. For a target with an integer or DECIMAL element type, key type, or+ value type, withElementType declares that type in place of BIGNUM. Each+ Integer element is a HUGEINT or a UHUGEINT value, so DuckDB checks the range+ of the cast of each element. For example, [5, -6] for a UBIGINT[2] column+ raises an error. A SizedArray or a list of BigNum values also takes the element+ type of the target, but each element is a BIGNUM value. DuckDB casts each+ element with the casts from BIGNUM, so the result for BigNum elements does+ not improve. For each other target, and for a parameter with no known type,+ a list takes the type that its elements give (withValueType), as each scalar+ Integer and as a SQL list literal of the HUGEINT and BIGNUM values: [1, 2^64]+ gives HUGEINT[], and [5, 2^130] gives BIGNUM[]. An Integer has no fixed SQL+ type, so a list with no element, or with only NULL elements, takes the+ element type of a known target, and otherwise the type NULL, as [] and+ [NULL] in SQL. The test is on the column type name BIGNUM, so the rule also+ holds for BigNum elements and for each type whose column type name is+ BIGNUM, for example a newtype that derives its instances through Integer. A+ BigNum element is a BIGNUM value in each case.++Issue 41 of duckdb-haskell tracks the defects.+-}++-- | Bind an integer as a HUGEINT, a UHUGEINT, or a BIGNUM value. See+-- Note [Integer parameters].+integerBinding :: Integer -> FieldBinding+integerBinding value = case hugeIntInRange value of+ Just make -> valueBinding (show value) make+ Nothing -> targetValueBinding (show value) \ValueBuilder{builderFactory = factory} readTarget -> do+ target <- readTarget+ case target of+ Just (LogicalTypeScalar LogicalTypeIdUHugeInt) | value >= 0 && value < 1 `shiftL` 128 -> uhugeIntDuckValue factory value+ Just (LogicalTypeScalar dtype) | dtype `elem` integerTypes -> outOfRange (duckDBTypeToName dtype)+ Just (LogicalTypeDecimal width scale) -> outOfRange ("DECIMAL(" <> Text.pack (show width) <> "," <> Text.pack (show scale) <> ")")+ _ -> bigNumDuckValue factory (BigNum value)+ where+ outOfRange typeName = throwConversionError ("the value " <> Text.pack (show value) <> " is out of range for a parameter of the type " <> typeName)++-- | Construct a HUGEINT value for an integer from -2^127 to 2^127 - 1. Give+-- Nothing for each other integer. See Note [Integer parameters].+hugeIntInRange :: Integer -> Maybe (DuckDBV2FactoryHandle -> IO DuckDBV2ValueHandle)+hugeIntInRange value+ | value >= negate (1 `shiftL` 127) && value < 1 `shiftL` 127 = Just (`hugeIntDuckValue` value)+ | otherwise = Nothing++-- | The integer types, from TINYINT to UHUGEINT. See Note [Integer parameters].+integerTypes :: [LogicalTypeId]+integerTypes =+ [ LogicalTypeIdTinyInt+ , LogicalTypeIdSmallInt+ , LogicalTypeIdInteger+ , LogicalTypeIdBigInt+ , LogicalTypeIdHugeInt+ , LogicalTypeIdUTinyInt+ , LogicalTypeIdUSmallInt+ , LogicalTypeIdUInteger+ , LogicalTypeIdUBigInt+ , LogicalTypeIdUHugeInt+ ]++instance ToField UUID.UUID++instance ToField Word++instance ToField Word8++instance ToField Word16++instance ToField Word32++instance ToField Word64++instance ToField Double++instance ToField Float++instance ToField Text++instance ToField String++instance ToField BitString++-- | Bind a day as a DATE. A day outside the range of DuckDB, from -5877641-06-25+-- (5877642-06-25 BC) to 5881580-07-10 in the notation of "Data.Time", raises an+-- t'SQLError' of the type 'ErrorTypeConversion'. See Note [Time ranges].+instance ToField Day++-- | Bind a time of day as a TIME, or as a TIME_NS. See+-- Note [Nanosecond parameters]. A time of exactly 60 seconds (a leap second)+-- binds as the next second, as @make_time@ of DuckDB gives it: 23:59:60 binds as+-- 24:00:00. A later time, for example 23:59:60.5, raises an t'SQLError'. The+-- section \"Errors\" of the README gives its type. See Note [Time ranges].+instance ToField TimeOfDay where+ toField value =+ timeValueBinding (show value) (LogicalTypeScalar LogicalTypeIdTime) (FieldTime value) \builder -> timeOfDayDuckValue (builderFactory builder) value++-- | Bind a local time as a TIMESTAMP, or as a TIMESTAMP_NS. See+-- Note [Nanosecond parameters]. A time outside the range of DuckDB, from+-- -290308-12-22 00:00:00 (290309-12-22 BC) to 294247-01-10 04:00:54.775806 in+-- the notation of "Data.Time", raises an t'SQLError' of the type+-- 'ErrorTypeConversion'. For a TIMESTAMP_NS target, a time outside the range+-- of TIMESTAMP_NS can get another error, also a time outside the range of+-- DuckDB in a STRUCT, a TUPLE, or a UNION. See Note [Nanosecond parameters]. A+-- time of exactly 60 seconds (a leap second) binds as the next second. A later+-- time, for example 23:59:60.5, raises an t'SQLError'. The section+-- \"Errors\" of the README gives its type. See Note [Time ranges].+instance ToField LocalTime where+ toField value =+ timeValueBinding (show value) (LogicalTypeScalar LogicalTypeIdTimestamp) (FieldTimestamp (Finite value)) \builder -> localTimeDuckValue (builderFactory builder) value++-- | Bind the months, days, and microseconds as an INTERVAL. DuckDB keeps the+-- three fields.+instance ToField IntervalValue++-- | Bind the time and its offset in seconds as a TIMETZ. An offset outside+-- -57599 to 57599 seconds raises an t'SQLError' of the type+-- 'ErrorTypeConversion'. A time with digits below one+-- microsecond loses these digits, as in SQL text. A time of exactly 60 seconds+-- (a leap second) binds as the next second. A later time, for example+-- 23:59:60.5, raises an t'SQLError'. The section \"Errors\" of the README gives+-- its type. See Note [Time ranges].+instance ToField TimeWithZone++-- | Bind the value as a DECIMAL with its own width and scale. DuckDB casts it+-- to the type of the parameter. A width outside 1 to 38, or a scale larger than+-- the width, raises an t'SQLError' of the type 'ErrorInputInvalid'. An integer+-- that the width cannot hold raises an t'SQLError' of the type+-- 'ErrorTypeConversion'.+instance ToField DecimalValue where+ toField value = valueBinding (show value) (`decimalDuckValue` value)++-- | Bind the shape as @GEOMETRY@ with no CRS.+instance ToField G.Geometry++-- | Bind the WKB bytes as a GEOMETRY value with the CRS of the raw geometry.+-- DuckDB then casts the value to the type of the parameter. See+-- Note [Raw GEOMETRY parameters].+instance ToField RawGeometry where+ toField raw@RawGeometry{rawGeometryWKB, rawGeometryCRS} =+ builderValueBinding+ ("<geometry length=" <> show (BS.length rawGeometryWKB) <> ">")+ \builder -> rawGeometryDuckValue builder rawGeometryCRS raw++-- | Bind the payload as a VARIANT.+instance ToField Variant where+ toField value =+ builderValueBinding (show value) \builder -> variantDuckValue builder (variantPayload value)++-- | Bind an instant as a TIMESTAMPTZ, or as a TIMESTAMPTZ_NS. An instant+-- outside the range of the t'LocalTime' instance, in UTC, raises an+-- t'SQLError' of the type 'ErrorTypeConversion'. For a TIMESTAMPTZ_NS target,+-- an instant outside the range of TIMESTAMPTZ_NS can get another error, also+-- an instant outside the range of DuckDB in a STRUCT, a TUPLE, or a UNION, as+-- for a t'LocalTime'. See Note [Nanosecond parameters]. A leap second of+-- exactly 60 seconds binds as the next second. A later time in a leap second,+-- for example 60.5 seconds, raises an t'SQLError'. The section \"Errors\" of+-- the README gives its type. See Note [Time ranges].+instance ToField UTCTime where+ toField value =+ timeValueBinding+ (show value)+ (LogicalTypeScalar LogicalTypeIdTimestampTz)+ (FieldTimestampTZ (Finite value))+ (\builder -> utcTimeDuckValue (builderFactory builder) value)++-- | Bind a day or an infinity as a DATE. A finite day has the range of the+-- 'Day' instance.+instance ToField (Unbounded Day)++-- | Bind a local time or an infinity as a TIMESTAMP, or a local time as a+-- TIMESTAMP_NS. See Note [Nanosecond parameters]. A finite time has the range+-- of the t'LocalTime' instance.+instance ToField (Unbounded LocalTime) where+ toField value =+ timeValueBinding (show value) (LogicalTypeScalar LogicalTypeIdTimestamp) (FieldTimestamp value) \builder -> localTimestampDuckValue (builderFactory builder) value++-- | Bind an instant or an infinity as a TIMESTAMPTZ, or an instant as a+-- TIMESTAMPTZ_NS. See Note [Nanosecond parameters]. A finite instant has the+-- range of the t'UTCTime' instance.+instance ToField (Unbounded UTCTime) where+ toField value =+ timeValueBinding+ (show value)+ (LogicalTypeScalar LogicalTypeIdTimestampTz)+ (FieldTimestampTZ value)+ (\builder -> utcTimestampDuckValue (builderFactory builder) value)++instance ToField BigNum where+ toField big@(BigNum n) = valueBinding (show n) (\factory -> bigNumDuckValue factory big)++instance ToField (StructValue FieldValue) where+ toField structVal =+ temporalValueBinding "<struct>" (structValueTypeRep structVal) (FieldStruct structVal) (`structValueDuckValue` structVal)++instance ToField (TupleValue FieldValue) where+ toField tuple =+ temporalValueBinding "<tuple>" (tupleValueTypeRep tuple) (FieldTuple tuple) (`tupleValueDuckValue` tuple)++instance ToField (UnionValue FieldValue) where+ toField unionVal =+ let label = Text.unpack (unionValueLabel unionVal)+ in temporalValueBinding ("<union " <> label <> ">") (unionValueTypeRep unionVal) (FieldUnion unionVal) (`unionValueDuckValue` unionVal)++instance DuckDBColumnType BitString where+ duckdbColumnTypeFor _ = "BIT"++instance ToField BS.ByteString where+ toField bs =+ valueBinding+ ("<blob length=" <> show (BS.length bs) <> ">")+ (\factory -> toDuckValue factory bs)++-- | Bind a 'SizedArray' as an ARRAY of size @n@. Each element takes the+-- element type of the parameter as its target. See Note [Sized arrays].+instance (DuckDBColumnType a, ToField a) => ToField (SizedArray n a) where+ toField values =+ targetValueBinding+ ("<array length=" <> show (length values) <> ">")+ \builder readTarget -> do+ target <- readTarget+ let childTarget = collectionChildType target+ arrayDuckValue builder childTarget (bindElement builder childTarget) (elems (sizedArrayToArray values))++-- | Bind a list as a LIST value with the element rules of a 'SizedArray'. An empty+-- list binds as an empty LIST of the element type. When the element type has+-- no type name, an empty list takes the element type of the parameter, or the+-- type NULL for a parameter with no known type, as the SQL literal [] does.+-- See Note [Element types of a list]. 'String' keeps its own instance. A list+-- as a row is a row of parameters, so bind a list parameter as @Only [..]@.+instance {-# OVERLAPPABLE #-} (DuckDBColumnType a, ToField a) => ToField [a] where+ toField values =+ targetValueBinding+ ("<list length=" <> show (length values) <> ">")+ \builder readTarget -> do+ target <- readTarget+ let childTarget = collectionChildType target+ listDuckValue builder childTarget (bindElement builder childTarget) values++-- | Bind a map as a MAP value. The pairs go in ascending key order. The keys+-- and the values have the element rules of a list, so an empty map binds as an+-- empty MAP of the key type and the value type, or with the type NULL for a key+-- or value type with no type name and no type from the parameter. DuckDB+-- compares two MAP values pair by pair in their order.+instance (DuckDBColumnType k, DuckDBColumnType v, ToField k, ToField v) => ToField (Map.Map k v) where+ toField pairs =+ targetValueBinding+ ("<map size=" <> show (Map.size pairs) <> ">")+ \builder readTarget -> do+ (keyTarget, valueTarget) <- mapPartTypes <$> readTarget+ mapDuckValue+ (withElementType (Proxy @k) builder keyTarget)+ (withElementType (Proxy @v) builder valueTarget)+ builder+ (bindElement builder keyTarget)+ (bindElement builder valueTarget)+ (Map.toAscList pairs)++-- | Bind a dynamic value, for example one that a query read as a t'FieldValue'.+-- Each scalar constructor binds with the instance of its payload type, so it+-- has the DuckDB type of the constructor. 'FieldHugeInt' and 'FieldUHugeInt'+-- bind as HUGEINT and UHUGEINT. 'FieldNull' binds as NULL. 'FieldStruct',+-- 'FieldTuple', and 'FieldUnion' bind with their own types. 'FieldGeometry'+-- binds as a GEOMETRY with its CRS.+--+-- 'FieldList', 'FieldArray', and 'FieldMap' read the parameter type, and take+-- the element type that their elements give. See+-- Note [Element types of a list].+--+-- Some values lose their type:+--+-- * A 'FieldTimestamp' binds as a TIMESTAMP, or as a TIMESTAMP_NS when it has+-- digits below a microsecond and the parameter has the type TIMESTAMP_NS or+-- VARIANT, whatever the unit of the column it came from. The same holds for+-- 'FieldTime' and 'FieldTimestampTZ'. In a 'FieldList' or a 'FieldMap' for a+-- VARIANT parameter, such a value gets the microsecond value. See+-- Note [Nanosecond parameters].+-- * A 'FieldText' binds as a VARCHAR with no collation, also when it came from+-- a JSON value.+-- * A value that came from a VARIANT binds as its payload.+-- * A 'FieldEnum' binds its label as a VARCHAR. DuckDB casts the text to the+-- ENUM type of a column.+--+-- A 'FieldList', a 'FieldArray', or a 'FieldMap' that is empty or has only+-- 'FieldNull' elements takes the element type of the parameter, or the type+-- NULL for a parameter with no known type, as a SQL list literal does.+--+-- An empty 'FieldArray' raises an t'SQLError' of the type 'ErrorInputInvalid',+-- because a DuckDB ARRAY has at least one element. Inside a 'FieldStruct', a+-- 'FieldTuple', or a 'FieldUnion', a 'FieldEnum' binds its label to the+-- declared ENUM type.+--+-- t'FieldValue' has no 'DuckDBColumnType' and no 'ToDuckValue' instance,+-- because no one DuckDB type holds each of its values.+instance ToField FieldValue where+ toField value = case value of+ FieldNull -> nullBinding "NULL"+ FieldInt8 v -> toField v+ FieldInt16 v -> toField v+ FieldInt32 v -> toField v+ FieldInt64 v -> toField v+ FieldWord8 v -> toField v+ FieldWord16 v -> toField v+ FieldWord32 v -> toField v+ FieldWord64 v -> toField v+ FieldUUID v -> toField v+ FieldFloat v -> toField v+ FieldDouble v -> toField v+ FieldText v -> toField v+ FieldBool v -> toField v+ FieldBlob v -> toField v+ FieldGeometry v -> toField v+ FieldDate v -> toField v+ FieldTime v -> toField v+ FieldTimestamp v -> toField v+ FieldInterval v -> toField v+ FieldHugeInt v -> valueBinding (show v) (`hugeIntDuckValue` v)+ FieldUHugeInt v -> valueBinding (show v) (`uhugeIntDuckValue` v)+ FieldDecimal v -> toField v+ FieldTimestampTZ v -> toField v+ FieldTimeTZ v -> toField v+ FieldBit v -> toField v+ FieldBigNum v -> toField v+ FieldEnum _ label -> toField label+ FieldArray items+ | Just failure <- arraySizeError (length items) -> valueBinding (show value) (const failure)+ | otherwise -> fieldValueSequence "FieldArray" c_duckdb_v2_value_create_array (elems items)+ FieldList items -> fieldValueSequence "FieldList" c_duckdb_v2_value_create_list items+ FieldMap pairs -> fieldValueMap pairs+ FieldStruct v -> toField v+ FieldTuple v -> toField v+ FieldUnion v -> toField v++-- | Bind the elements of a 'FieldList' or a 'FieldArray' with the element type+-- that they give. See Note [Element types of a list].+fieldValueSequence+ :: Text+ -> ( DuckDBV2FactoryHandle+ -> DuckDBV2LogicalTypeHandle+ -> ConstPtr DuckDBV2ValueHandle+ -> DuckDBIdx+ -> Ptr DuckDBV2ValueHandle+ -> Ptr DuckDBV2ErrorInfoHandle+ -> IO DuckDBV2Error+ )+ -> [FieldValue]+ -> FieldBinding+fieldValueSequence name create items =+ targetValueBinding ("<" <> Text.unpack name <> " length=" <> show (length items) <> ">") \builder readTarget -> do+ childTarget <- collectionChildType <$> readTarget+ sequenceDuckValue (withValueType builder childTarget) create builder (bindElement builder childTarget) items++-- | Bind the pairs of a 'FieldMap' in their order, with the key type and the+-- value type that they give. See Note [Element types of a list].+fieldValueMap :: [(FieldValue, FieldValue)] -> FieldBinding+fieldValueMap pairs =+ targetValueBinding ("<FieldMap size=" <> show (length pairs) <> ">") \builder readTarget -> do+ (keyTarget, valueTarget) <- mapPartTypes <$> readTarget+ mapDuckValue+ (withValueType builder keyTarget)+ (withValueType builder valueTarget)+ builder+ (bindElement builder keyTarget)+ (bindElement builder valueTarget)+ pairs++instance (ToField a) => ToField (Maybe a) where+ toField Nothing = nullBinding "Nothing"+ toField (Just value) =+ let binding = toField value+ in binding+ { fieldBindingDisplay = "Just " <> renderFieldBinding binding+ }++instance DuckDBColumnType G.Geometry where+ duckdbColumnTypeFor _ = "GEOMETRY"++instance DuckDBColumnType Variant where+ duckdbColumnTypeFor _ = "VARIANT"++instance DuckDBColumnType Null where+ duckdbColumnTypeFor _ = "NULL"++instance DuckDBColumnType Bool where+ duckdbColumnTypeFor _ = "BOOLEAN"++instance DuckDBColumnType Int where+ duckdbColumnTypeFor _ = "BIGINT"++instance DuckDBColumnType Int8 where+ duckdbColumnTypeFor _ = "TINYINT"++instance DuckDBColumnType Int16 where+ duckdbColumnTypeFor _ = "SMALLINT"++instance DuckDBColumnType Int32 where+ duckdbColumnTypeFor _ = "INTEGER"++instance DuckDBColumnType Int64 where+ duckdbColumnTypeFor _ = "BIGINT"++instance DuckDBColumnType BigNum where+ duckdbColumnTypeFor _ = "BIGNUM"++instance DuckDBColumnType UUID.UUID where+ duckdbColumnTypeFor _ = "UUID"++-- | The text is BIGNUM, the type that holds each value. A typed registration+-- declares HUGEINT. See Note [Integer parameters].+instance DuckDBColumnType Integer where+ duckdbColumnTypeFor _ = "BIGNUM"+ duckdbTypeDeclaration _ = Right (StructuredType (LogicalTypeScalar LogicalTypeIdHugeInt))++-- | The text is BIGNUM, the type that holds each value. A typed registration+-- declares UHUGEINT. See Note [Integer parameters].+instance DuckDBColumnType Natural where+ duckdbColumnTypeFor _ = "BIGNUM"+ duckdbTypeDeclaration _ = Right (StructuredType (LogicalTypeScalar LogicalTypeIdUHugeInt))++instance DuckDBColumnType Word where+ duckdbColumnTypeFor _ = "UBIGINT"++instance DuckDBColumnType Word8 where+ duckdbColumnTypeFor _ = "UTINYINT"++instance DuckDBColumnType Word16 where+ duckdbColumnTypeFor _ = "USMALLINT"++instance DuckDBColumnType Word32 where+ duckdbColumnTypeFor _ = "UINTEGER"++instance DuckDBColumnType Word64 where+ duckdbColumnTypeFor _ = "UBIGINT"++instance DuckDBColumnType Double where+ duckdbColumnTypeFor _ = "DOUBLE"++instance DuckDBColumnType Float where+ duckdbColumnTypeFor _ = "FLOAT"++instance DuckDBColumnType Text where+ duckdbColumnTypeFor _ = "VARCHAR"++instance DuckDBColumnType String where+ duckdbColumnTypeFor _ = "VARCHAR"++instance DuckDBColumnType BS.ByteString where+ duckdbColumnTypeFor _ = "BLOB"++instance DuckDBColumnType Day where+ duckdbColumnTypeFor _ = "DATE"++instance DuckDBColumnType TimeOfDay where+ duckdbColumnTypeFor _ = "TIME"++instance DuckDBColumnType LocalTime where+ duckdbColumnTypeFor _ = "TIMESTAMP"++instance DuckDBColumnType IntervalValue where+ duckdbColumnTypeFor _ = "INTERVAL"++instance DuckDBColumnType TimeWithZone where+ duckdbColumnTypeFor _ = "TIMETZ"++instance DuckDBColumnType UTCTime where+ duckdbColumnTypeFor _ = "TIMESTAMPTZ"++instance DuckDBColumnType (Unbounded Day) where+ duckdbColumnTypeFor _ = "DATE"++instance DuckDBColumnType (Unbounded LocalTime) where+ duckdbColumnTypeFor _ = "TIMESTAMP"++instance DuckDBColumnType (Unbounded UTCTime) where+ duckdbColumnTypeFor _ = "TIMESTAMPTZ"++-- | No declaration. See Note [Type declarations].+instance DuckDBColumnType (StructValue FieldValue) where+ duckdbColumnTypeFor _ = "STRUCT"+ duckdbTypeDeclaration _ = Left (noStaticFieldTypes "StructValue")++-- | No declaration. See Note [Type declarations].+instance DuckDBColumnType (TupleValue FieldValue) where+ duckdbColumnTypeFor _ = "TUPLE"+ duckdbTypeDeclaration _ = Left (noStaticFieldTypes "TupleValue")++-- | No declaration. See Note [Type declarations].+instance DuckDBColumnType (UnionValue FieldValue) where+ duckdbColumnTypeFor _ = "UNION"+ duckdbTypeDeclaration _ = Left (noStaticFieldTypes "UnionValue")++instance (DuckDBColumnType a) => DuckDBColumnType (Maybe a) where+ duckdbColumnTypeFor _ = duckdbColumnTypeFor (Proxy @a)+ duckdbTypeDeclaration _ = duckdbTypeDeclaration (Proxy @a)++-- | The type text is @T[n]@, and the declaration is the ARRAY type of size @n@.+-- See Note [Sized arrays].+instance (KnownNat n, DuckDBColumnType a) => DuckDBColumnType (SizedArray n a) where+ duckdbColumnTypeFor _ = duckdbColumnTypeFor (Proxy @a) <> Text.pack ("[" <> show (natVal (Proxy @n)) <> "]")+ duckdbTypeDeclaration _ = case arraySizeReason size of+ Just reason -> Left ("a SizedArray of size " <> Text.pack (show size) <> " has no ARRAY type, because " <> reason)+ Nothing -> arrayDeclaration size <$> duckdbTypeDeclaration (Proxy @a)+ where+ size = natVal (Proxy @n)++{- Note [Sized arrays]+~~~~~~~~~~~~~~~~~~~~~~+A DuckDB ARRAY type has a size, for example BIGINT[3]. SizedArray n a is the+Haskell type of T[n]: its type text is T[n], and a typed registration declares+the ARRAY type of size n. A SizedArray holds an Array Int a with the bounds+(0, n - 1). sizedArray takes an Array with any bounds and stores its elements+with these bounds, so the derived Eq and Ord compare the elements only. It is+the one Haskell type of an ARRAY. For an ARRAY+of a size that varies, a list reads an ARRAY of each size, and DuckDB casts a+bound list to an ARRAY column of the same length. A FieldArray is an ARRAY+value that a program builds at run time.++- Binds. A SizedArray binds as an ARRAY of size n. Each element takes the+ element type of the parameter as its target, and the element type of the+ value follows Note [Element types of a list].+- Reads. A SizedArray reads an ARRAY and a LIST, as a list does, and the+ value must have n elements. The cast of DuckDB from a LIST or an ARRAY to+ T[n] gives the same value, or an error for another length, so this is a+ read by the rule of Note [Reads of another SQL type] in+ "Database.DuckDB.Simple.FromField". The check uses the length of the+ decoded value and not the column type, so it is the same for a LIST column,+ an ARRAY column of another size, and a VARIANT. A value of another length+ raises Incompatible, which names the column type and the SizedArray type.+- Sizes. A DuckDB ARRAY has 1 to 100000 elements (maxArraySize). DuckDB+ rejects the SQL type T[0] and a larger type, for example with "ARRAY type+ size must be at most 100000". The value API of DuckDB 2.0.0 does not check+ the largest size, so a parameter of a larger ARRAY would bind a value that+ the SQL of DuckDB cannot write. Thus the library checks both sizes, as an+ input check: the declaration of a SizedArray of another size gives Left,+ and the bind of a SizedArray or a FieldArray of another size raises an+ SQLError of the type ErrorInputInvalid (arraySizeError).+- Fields. A SizedArray field of a generic record declares the ARRAY of size+ n through its DuckValue instance, and reads as a SizedArray does.+-}++-- | Declare an ARRAY of a size with the declaration of its element type. See+-- Note [Sized arrays].+arrayDeclaration :: Integer -> TypeDeclaration -> TypeDeclaration+arrayDeclaration size (TypeExpression element) = TypeExpression (element <> Text.pack ("[" <> show size <> "]"))+arrayDeclaration size (StructuredType element) = StructuredType (LogicalTypeArray element (fromInteger size))++-- | See Note [Type declarations].+instance (DuckDBColumnType k, DuckDBColumnType v) => DuckDBColumnType (Map.Map k v) where+ duckdbColumnTypeFor _ = Text.pack "MAP(" <> duckdbColumnTypeFor (Proxy @k) <> Text.pack ", " <> duckdbColumnTypeFor (Proxy @v) <> Text.pack ")"+ duckdbTypeDeclaration _ = do+ key <- duckdbTypeDeclaration (Proxy @k)+ value <- duckdbTypeDeclaration (Proxy @v)+ mapDeclaration key value++-- | 'String' keeps its own instance. See Note [Type declarations].+instance {-# OVERLAPPABLE #-} (DuckDBColumnType a) => DuckDBColumnType [a] where+ duckdbColumnTypeFor _ = duckdbColumnTypeFor (Proxy @a) <> Text.pack "[]"+ duckdbTypeDeclaration _ = listDeclaration <$> duckdbTypeDeclaration (Proxy @a)++nullBinding :: String -> FieldBinding+nullBinding repr = valueBinding repr nullDuckValue++nullDuckValue :: DuckDBV2FactoryHandle -> IO DuckDBV2ValueHandle+nullDuckValue factory =+ bracket+ (logicalTypeFromRep factory (LogicalTypeScalar LogicalTypeIdSQLNull))+ destroyLogicalType+ (\logical -> output (c_duckdb_v2_value_create_null factory logical))++boolDuckValue :: DuckDBV2FactoryHandle -> Bool -> IO DuckDBV2ValueHandle+boolDuckValue factory value = output (c_duckdb_v2_value_create_bool factory (if value then 1 else 0))++int8DuckValue :: DuckDBV2FactoryHandle -> Int8 -> IO DuckDBV2ValueHandle+int8DuckValue factory value = output (c_duckdb_v2_value_create_tinyint factory value)++int16DuckValue :: DuckDBV2FactoryHandle -> Int16 -> IO DuckDBV2ValueHandle+int16DuckValue factory value = output (c_duckdb_v2_value_create_smallint factory value)++int32DuckValue :: DuckDBV2FactoryHandle -> Int32 -> IO DuckDBV2ValueHandle+int32DuckValue factory value = output (c_duckdb_v2_value_create_int factory value)++int64DuckValue :: DuckDBV2FactoryHandle -> Int64 -> IO DuckDBV2ValueHandle+int64DuckValue factory value = output (c_duckdb_v2_value_create_bigint factory value)++uint64DuckValue :: DuckDBV2FactoryHandle -> Word64 -> IO DuckDBV2ValueHandle+uint64DuckValue factory value = output (c_duckdb_v2_value_create_ubigint factory value)++uint32DuckValue :: DuckDBV2FactoryHandle -> Word32 -> IO DuckDBV2ValueHandle+uint32DuckValue factory value = output (c_duckdb_v2_value_create_uint factory value)++uint16DuckValue :: DuckDBV2FactoryHandle -> Word16 -> IO DuckDBV2ValueHandle+uint16DuckValue factory value = output (c_duckdb_v2_value_create_usmallint factory value)++uint8DuckValue :: DuckDBV2FactoryHandle -> Word8 -> IO DuckDBV2ValueHandle+uint8DuckValue factory value = output (c_duckdb_v2_value_create_utinyint factory value)++doubleDuckValue :: DuckDBV2FactoryHandle -> Double -> IO DuckDBV2ValueHandle+doubleDuckValue factory value = output (c_duckdb_v2_value_create_double factory (CDouble value))++floatDuckValue :: DuckDBV2FactoryHandle -> Float -> IO DuckDBV2ValueHandle+floatDuckValue factory value = output (c_duckdb_v2_value_create_float factory (CFloat value))++textDuckValue :: DuckDBV2FactoryHandle -> Text -> IO DuckDBV2ValueHandle+textDuckValue factory value = withText value (\text -> output (c_duckdb_v2_value_create_varchar factory text))++stringDuckValue :: DuckDBV2FactoryHandle -> String -> IO DuckDBV2ValueHandle+stringDuckValue factory = textDuckValue factory . Text.pack++blobDuckValue :: DuckDBV2FactoryHandle -> BS.ByteString -> IO DuckDBV2ValueHandle+blobDuckValue factory value = withBytes value (\bytes -> output (c_duckdb_v2_value_create_blob factory bytes))++uuidDuckValue :: DuckDBV2FactoryHandle -> UUID.UUID -> IO DuckDBV2ValueHandle+uuidDuckValue factory uuid =+ let (upper, lower) = UUID.toWords64 uuid+ value = DuckDBV2HugeintT lower (fromIntegral (upper `xor` 0x8000000000000000))+ in with value (\ptr -> output (c_duckdb_v2_value_create_uuid factory (ConstPtr ptr)))++bitDuckValue :: DuckDBV2FactoryHandle -> BitString -> IO DuckDBV2ValueHandle+bitDuckValue factory (BitString padding bits) = do+ when (BS.null bits || padding > 7) $+ throwInvalidArgument "BIT requires nonempty data and padding from 0 to 7"+ let nativePadding = complement ((1 `shiftL` (8 - fromIntegral padding)) - 1) :: Word8+ payload = BS.cons padding (BS.cons (BS.head bits .|. nativePadding) (BS.tail bits))+ withBytes payload (\bytes -> output (c_duckdb_v2_value_create_bit factory bytes))++bigNumDuckValue :: DuckDBV2FactoryHandle -> BigNum -> IO DuckDBV2ValueHandle+bigNumDuckValue factory (BigNum value) =+ withBigEndianMagnitude (abs value) \raw len -> alloca \lengthSlot -> do+ let encode ptr capacity = checked (c_duckdb_v2_bignum_encode (ConstPtr (castPtr raw)) (fromIntegral len) (fromBool (value < 0)) ptr capacity lengthSlot)+ encode nullPtr 0+ size <- peek lengthSlot+ allocaBytes (fromIntegral size) \storage -> do+ encode storage size+ payload <- BS.packCStringLen (castPtr storage, fromIntegral size)+ withBytes payload (\encoded -> output (c_duckdb_v2_value_create_bignum factory encoded))++-- | Write the unsigned big-endian magnitude of a non-negative number into a+-- buffer in linear time, and give the buffer and its length to the action. The+-- magnitude has no leading zero byte. Zero is one zero byte.+withBigEndianMagnitude :: Integer -> (Ptr Word8 -> Int -> IO a) -> IO a+withBigEndianMagnitude 0 action = with 0 \buffer -> action buffer 1+withBigEndianMagnitude magnitude action = allocaBytes size \buffer@(Ptr addr) -> integerToAddr magnitude addr 1# >> action buffer size+ where+ size = fromIntegral (integerLog2 magnitude `quot` 8) + 1++dayDuckValue :: DuckDBV2FactoryHandle -> Day -> IO DuckDBV2ValueHandle+dayDuckValue factory value = encodeDay value >>= (\encoded -> output (c_duckdb_v2_value_create_date factory encoded))++timeOfDayDuckValue :: DuckDBV2FactoryHandle -> TimeOfDay -> IO DuckDBV2ValueHandle+timeOfDayDuckValue factory value = encodeTimeOfDay value >>= (\encoded -> output (c_duckdb_v2_value_create_time factory encoded))++localTimeDuckValue :: DuckDBV2FactoryHandle -> LocalTime -> IO DuckDBV2ValueHandle+localTimeDuckValue factory value = encodeLocalTime value >>= (\encoded -> output (c_duckdb_v2_value_create_timestamp factory encoded))++utcTimeDuckValue :: DuckDBV2FactoryHandle -> UTCTime -> IO DuckDBV2ValueHandle+utcTimeDuckValue factory = utcTimestampDuckValue factory . Finite++dateDuckValue :: DuckDBV2FactoryHandle -> Date -> IO DuckDBV2ValueHandle+dateDuckValue factory value = encodeUnbounded encodeDay value >>= (\encoded -> output (c_duckdb_v2_value_create_date factory encoded))++localTimestampDuckValue :: DuckDBV2FactoryHandle -> LocalTimestamp -> IO DuckDBV2ValueHandle+localTimestampDuckValue factory value = encodeUnbounded (encodeTimestampUnits 1000000) value >>= (\encoded -> output (c_duckdb_v2_value_create_timestamp factory encoded))++utcTimestampDuckValue :: DuckDBV2FactoryHandle -> UTCTimestamp -> IO DuckDBV2ValueHandle+utcTimestampDuckValue factory value =+ encodeUnbounded (encodeTimestampUnits 1000000 . utcToLocalTime utc) value+ >>= (\encoded -> output (c_duckdb_v2_value_create_timestamp_tz factory encoded))++-- | Construct an ARRAY value. An array with no element or with more elements+-- than DuckDB permits raises t'SQLError'. See 'withElementType' for the element+-- type.+arrayDuckValue+ :: forall a+ . (DuckDBColumnType a)+ => ValueBuilder+ -> Maybe LogicalTypeRep+ -> (a -> IO DuckDBV2ValueHandle)+ -> [a]+ -> IO DuckDBV2ValueHandle+arrayDuckValue builder target elementValue values = do+ sequence_ (arraySizeError (length values))+ sequenceDuckValue (withElementType (Proxy @a) builder target) c_duckdb_v2_value_create_array builder elementValue values++-- | Give the input error of an ARRAY parameter of a size that DuckDB does not+-- have: no element, or more elements than DuckDB permits. A 'SizedArray' and a+-- 'FieldArray' raise this error. See Note [Sized arrays].+arraySizeError :: Int -> Maybe (IO a)+arraySizeError count = case arraySizeReason (toInteger count) of+ Nothing -> Nothing+ Just reason ->+ Just+ ( throwInvalidArgument case count of+ 0 -> "an empty ARRAY parameter is not possible, because " <> reason <> "; bind an empty list for an empty LIST"+ _ -> "an ARRAY parameter of " <> Text.pack (show count) <> " elements is not possible, because " <> reason <> "; bind a list for a LIST"+ )++-- | Construct a LIST value. An empty list is an empty LIST of the element type.+-- See 'withElementType' for the element type.+listDuckValue+ :: forall a+ . (DuckDBColumnType a)+ => ValueBuilder+ -> Maybe LogicalTypeRep+ -> (a -> IO DuckDBV2ValueHandle)+ -> [a]+ -> IO DuckDBV2ValueHandle+listDuckValue builder target = sequenceDuckValue (withElementType (Proxy @a) builder target) c_duckdb_v2_value_create_list builder++-- | Construct a LIST or an ARRAY value with the native constructor. The first+-- argument gives the element type of the values to its action:+-- 'withElementType' for a Haskell element type, or 'withValueType'.+sequenceDuckValue+ :: ([DuckDBV2ValueHandle] -> (DuckDBV2LogicalTypeHandle -> IO DuckDBV2ValueHandle) -> IO DuckDBV2ValueHandle)+ -> ( DuckDBV2FactoryHandle+ -> DuckDBV2LogicalTypeHandle+ -> ConstPtr DuckDBV2ValueHandle+ -> DuckDBIdx+ -> Ptr DuckDBV2ValueHandle+ -> Ptr DuckDBV2ErrorInfoHandle+ -> IO DuckDBV2Error+ )+ -> ValueBuilder+ -> (a -> IO DuckDBV2ValueHandle)+ -> [a]+ -> IO DuckDBV2ValueHandle+sequenceDuckValue withType create ValueBuilder{builderFactory = factory} elementValue elements =+ withCreatedValues (map elementValue elements) \values ->+ withType values \elementType ->+ withArray values \ptr ->+ output (create factory elementType (ConstPtr ptr) (fromIntegral (length values)))++-- | Construct a MAP value from pairs in their order. The first two arguments+-- give the key type and the value type, as for 'sequenceDuckValue'.+mapDuckValue+ :: ([DuckDBV2ValueHandle] -> (DuckDBV2LogicalTypeHandle -> IO DuckDBV2ValueHandle) -> IO DuckDBV2ValueHandle)+ -> ([DuckDBV2ValueHandle] -> (DuckDBV2LogicalTypeHandle -> IO DuckDBV2ValueHandle) -> IO DuckDBV2ValueHandle)+ -> ValueBuilder+ -> (k -> IO DuckDBV2ValueHandle)+ -> (v -> IO DuckDBV2ValueHandle)+ -> [(k, v)]+ -> IO DuckDBV2ValueHandle+mapDuckValue withKeyType withMapValueType ValueBuilder{builderFactory = factory} keyValue valueValue pairs =+ withCreatedValues (map (keyValue . fst) pairs) \keys ->+ withCreatedValues (map (valueValue . snd) pairs) \values ->+ withKeyType keys \keyType ->+ withMapValueType values \valueType ->+ withArray keys \keyPtr -> withArray values \valuePtr ->+ output (c_duckdb_v2_value_create_map factory keyType valueType (ConstPtr keyPtr) (ConstPtr valuePtr) (fromIntegral (length pairs)))++{- Note [Element types of a list]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+withElementType and withValueType give the element type of a LIST or an ARRAY+value, and the key type or the value type of a MAP value, to the native+constructor:++- A scalar element type comes from the column type name of the element. Some+ elements take the element type of the target in place of that type: an+ Integer, a Natural, or a BigNum element for an integer or DECIMAL target (see+ Note [Integer parameters]), and a TimeOfDay, a LocalTime, or a UTCTime+ element for a nanosecond target (see Note [Nanosecond parameters]). For each+ other target, a list of Integer, Natural, or BigNum elements that is not+ empty takes the type that its elements give, as the elements below do.+- An element type with no recognized scalar type name uses withValueType to+ infer its type from values. Examples are a STRUCT, a TUPLE, a UNION, a generic+ record, a list, an array, and a FieldValue. A complete structured declaration+ supplies the type only when no value contributes a type, as described below.+ A typed NULL contributes its type, including aliases and collations.+ Only a NULL of the unaliased SQLNULL type has no type to contribute. The+ filter compares each NULL type with one temporary unaliased SQLNULL type.+ Native equality distinguishes aliases, including the alias "NULL". When+ each remaining element has the type of the first one, that type is the+ element type. The check uses native equality to keep aliases and ENUM dictionaries.+ valuesHaveType checks the metadata that equality omits. In DuckDB 2.0.0,+ VARCHAR equality ignores collations. STRUCT and UNION names compare without+ ASCII case. GEOMETRY CRS definitions can compare equal through a shared+ identifier. The metadata check keeps exact names, collations, and definitions.+ It reads the first comparison key once, then checks each remaining value.+ It visits child types but reads no ENUM labels.+- When the types are different, but timestampTypeForTarget gives each of them+ the same nanosecond type for the target, that type is the element type. For+ example, a FieldTime with digits below a microsecond binds as a TIME_NS for a+ TIME_NS[] parameter, and one with no such digits binds as a TIME. DuckDB has+ no common type for TIME and TIME_NS, or for TIMESTAMPTZ and TIMESTAMPTZ_NS.+ This check reads complete type trees only after the equality check fails+ and the target contains a nanosecond temporal type. The promoted trees must+ agree, including aliases on unchanged leaves. At least one type must change.+ The promotion loop releases each native type before its next iteration.+- When these types are different, the library names no element type. The+ native constructor then gives the type of a SQL list literal of the elements+ (duckdb_v2.h: "The child type is the common type of the elements, resolved by+ the same rule a SQL list literal follows, and every element is cast to it; a+ set with no common type surfaces the cast error."). For example, a STRUCT(a+ INTEGER) and a STRUCT(b INTEGER) give STRUCT(a INTEGER, b INTEGER), with NULL+ for a field that an element does not have. A BIGINT field and a DOUBLE field+ give a DOUBLE field, which rounds a large BIGINT, as in SQL.+- Elements with no common type raise the SQLError of DuckDB. For a STRUCT(a+ INTEGER) and a STRUCT(a VARCHAR), it is "Cannot combine types" of the type+ ErrorQueryNotImplemented. The SQL list literal raises a Binder Error for the+ same elements, but array_value raises the same error as the constructor.+- The target does not change the type of these elements. In SQL, DuckDB also+ gives the list literal its type before it casts the list to the type of a+ column.+- A list with no element or only untyped NULL elements of this kind takes the+ element type of the target, when the parameter has a known type. DuckDB casts+ the list literal [] or [NULL, NULL] to the type of a column in the same way.+- For a parameter with no known type, the element type of a list with no+ element or only untyped NULL elements is the structured declaration of the+ element type, when it has one, for example a record through ViaDuckDB (see+ Note [Type declarations]). The two rules below apply to each other element+ type.+- For a parameter with no known type, a list with only untyped NULL elements+ names no element type. The native constructor then gives the type of the SQL+ list literal: "NULL"[] for a LIST, "NULL"[2] for an ARRAY of two NULL elements,+ and MAP(VARCHAR, "NULL") for a MAP of VARCHAR keys and NULL values.+- For a parameter with no known type, an empty list or map of this kind gets+ the element type NULL. The SQL literals [] and MAP {} have the types+ "NULL"[] and MAP("NULL", "NULL"). The native constructor gives no type to an+ empty set (it raises "cannot resolve the element type of an empty set"), so+ the library names the type NULL itself. A SizedArray or a FieldArray with no+ element raises an error before this point, because a DuckDB ARRAY has at+ least one element.+-}++-- | Give the element type of the values of a LIST, an ARRAY, or one part of a+-- MAP to the action. The target is the element type of the parameter. See+-- Note [Element types of a list].+withElementType+ :: forall a b+ . (DuckDBColumnType a)+ => Proxy a+ -> ValueBuilder+ -> Maybe LogicalTypeRep+ -> [DuckDBV2ValueHandle]+ -> (DuckDBV2LogicalTypeHandle -> IO b)+ -> IO b+withElementType proxy builder@ValueBuilder{builderType = makeType} target values action = case duckDBTypeFromName typeName of+ -- See Note [Integer parameters].+ Just LogicalTypeIdBigNum | not integerTarget -> withValueType builder target values action+ Just dtype -> do+ checkNanosecondElements (declaredType dtype) values+ bracket (makeType (declaredType dtype)) destroyLogicalType action+ Nothing -> withValueType builder elementTarget values action+ where+ typeName = duckdbColumnType proxy+ -- With no target, the structured declaration of a record is the element+ -- type of a list with no element that is not NULL. See+ -- Note [Element types of a list].+ elementTarget = maybe (declaredTarget proxy) Just target+ -- See Note [Integer parameters].+ integerTarget = case target of+ Just (LogicalTypeScalar dtype) -> dtype `elem` integerTypes+ Just LogicalTypeDecimal{} -> True+ _ -> False+ declaredType LogicalTypeIdBigNum | integerTarget, Just rep <- target = rep+ declaredType dtype = timestampTypeForTarget target (LogicalTypeScalar dtype)++-- | Raise the range error of the library for each TIMESTAMP or TIMESTAMPTZ+-- element that a TIMESTAMP_NS or TIMESTAMPTZ_NS element type cannot hold, before+-- the native constructor casts the element. An infinity value and a NULL value+-- pass, as in the cast. See Note [Nanosecond parameters].+checkNanosecondElements :: LogicalTypeRep -> [DuckDBV2ValueHandle] -> IO ()+checkNanosecondElements declared values = case declared of+ LogicalTypeScalar LogicalTypeIdTimestampNs -> forM_ values (check LogicalTypeIdTimestamp c_duckdb_v2_value_get_timestamp)+ LogicalTypeScalar LogicalTypeIdTimestampTzNs -> forM_ values (check LogicalTypeIdTimestampTz c_duckdb_v2_value_get_timestamp_tz)+ _ -> pure ()+ where+ check source get value = do+ isNull <- output (c_duckdb_v2_value_is_null value)+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ rep <- bracket (output (c_duckdb_v2_value_get_logical_type value)) destroyLogicalType logicalTypeToBaseRep+ when (isNull == 0 && rep == LogicalTypeScalar source) do+ micros <- output (get value)+ unless (micros == maxBound || micros == negate maxBound) $+ checkFiniteRange "TIMESTAMP" (minBound :: Int64) maxBound (1000 * toInteger micros)++-- | Give the element type that the values give to the action. The target is+-- the element type of the parameter. See Note [Element types of a list].+withValueType :: ValueBuilder -> Maybe LogicalTypeRep -> [DuckDBV2ValueHandle] -> (DuckDBV2LogicalTypeHandle -> IO b) -> IO b+withValueType ValueBuilder{builderFactory = factory, builderType = makeType} target values action = do+ -- Keep each value except an untyped NULL, in list order. The fold keeps a+ -- constant stack. See Note [Constant stack at safe calls] in+ -- "Database.DuckDB.Simple.Native".+ let keep plainNull kept value =+ output (c_duckdb_v2_value_is_null value) >>= \flag ->+ if flag == 0+ then pure (value : kept)+ else do+ untyped <-+ bracket (output (c_duckdb_v2_value_get_logical_type value)) destroyLogicalType \logical ->+ (/= 0) <$> output (c_duckdb_v2_logical_type_is_equal logical plainNull)+ pure (if untyped then kept else value : kept)+ present <-+ bracket (makeType (LogicalTypeScalar LogicalTypeIdSQLNull)) destroyLogicalType \plainNull ->+ reverse <$> foldM (keep plainNull) [] values+ case present of+ [value] -> bracket (output (c_duckdb_v2_value_get_logical_type value)) destroyLogicalType action+ value : rest -> bracket (output (c_duckdb_v2_value_get_logical_type value)) destroyLogicalType \logical -> do+ same <- valuesHaveType logical rest+ let nanoseconds (LogicalTypeScalar dtype) = dtype `elem` [LogicalTypeIdTimeNs, LogicalTypeIdTimestampNs, LogicalTypeIdTimestampTzNs]+ nanoseconds _ = False+ if same+ then action logical+ else+ if not (maybe False (containsType nanoseconds) target)+ then action nullHandle+ else do+ original <- logicalTypeToRep factory logical+ let promoted = timestampTypeForTarget target original+ agree changed [] = pure (if changed then Just promoted else Nothing)+ agree changed (element : remaining) = do+ -- Release this type before the next iteration.+ actual <- bracket (output (c_duckdb_v2_value_get_logical_type element)) destroyLogicalType (logicalTypeToRep factory)+ let updated = timestampTypeForTarget target actual+ if updated /= promoted+ then pure Nothing+ else do+ let changed' = changed || updated /= actual+ changed' `seq` agree changed' remaining+ common <- agree (promoted /= original) rest+ case common of+ Just rep -> bracket (makeType rep) destroyLogicalType action+ Nothing -> action nullHandle+ -- See Note [Element types of a list].+ [] -> case target of+ Just rep -> bracket (makeType rep) destroyLogicalType action+ Nothing+ | not (null values) -> action nullHandle+ | otherwise -> bracket (makeType (LogicalTypeScalar LogicalTypeIdSQLNull)) destroyLogicalType action++-- | Construct named fields. The native call copies the field values.+structValueDuckValue+ :: ValueBuilder -> StructValue FieldValue -> IO DuckDBV2ValueHandle+structValueDuckValue builder@ValueBuilder{builderFactory = factory} StructValue{structValueFields, structValueTypes} = do+ let fields = elems structValueFields+ types = elems structValueTypes+ names = map structFieldName fields+ when (names /= map structFieldName types) $+ throwInvalidArgument "struct value/type fields mismatch"+ withCreatedValues+ (zipWith (\field logical -> fieldValueWithTypeDuckValue builder (structFieldValue logical) (structFieldValue field)) fields types)+ \values ->+ withIdentifiers names \namePtr -> withArray values \valuePtr ->+ declaredCompositeValue builder (LogicalTypeStruct structValueTypes) $+ output (c_duckdb_v2_value_create_struct factory (ConstPtr namePtr) (ConstPtr valuePtr) (fromIntegral (length values)))++-- | Construct positional fields without inferring a TUPLE from field names.+tupleValueDuckValue+ :: ValueBuilder -> TupleValue FieldValue -> IO DuckDBV2ValueHandle+tupleValueDuckValue builder@ValueBuilder{builderFactory = factory} TupleValue{tupleValueFields, tupleValueTypes} = do+ let fields = elems tupleValueFields+ types = elems tupleValueTypes+ when (length fields /= length types) $+ throwInvalidArgument "tuple value/type fields mismatch"+ withCreatedValues (zipWith (fieldValueWithTypeDuckValue builder) types fields) \values ->+ withArray values \valuePtr ->+ declaredCompositeValue builder (LogicalTypeTuple tupleValueTypes) $+ output (c_duckdb_v2_value_create_tuple factory (ConstPtr valuePtr) (fromIntegral (length values)))++-- | Preserve declared field types through an owned vector. Scalar VARCHAR+-- values do not retain collations. Most alias values use their base type,+-- but their enclosing composite type keeps both. See+-- Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+declaredCompositeValue+ :: ValueBuilder+ -> LogicalTypeRep+ -> IO DuckDBV2ValueHandle+ -> IO DuckDBV2ValueHandle+declaredCompositeValue _ rep create | not (needsDeclaredType rep) = create+declaredCompositeValue ValueBuilder{builderFactory = factory, builderType = makeType} rep create = bracket create destroyValue \value ->+ bracket (makeType rep) destroyLogicalType (defaultCast factory value)++-- | Test whether a type tree contains a VARCHAR collation or an alias.+needsDeclaredType :: LogicalTypeRep -> Bool+needsDeclaredType = containsType declared+ where+ declared (LogicalTypeCollatedVarchar _) = True+ declared LogicalTypeAlias{} = True+ declared _ = False++{- Note [UNION values]+~~~~~~~~~~~~~~~~~~~~~+V2 has no constructor for a UNION value. unionValueDuckValue writes the tag+and active member into a one-row UNION vector, sets the inactive members to+NULL, and reads the row as a value. The header specifies child 0 as the tag+and children 1 to N as the members. vector_set_value uses the non-strict+default cast, so a user cast cannot change the payload. See Note [User casts+and the casts of the library] in "Database.DuckDB.Simple.Materialize".++The header recommends casting a member value to the UNION type, but that cast+selects the member by type. It is ambiguous for two BIGINT members or a nested+UNION with a member name of the outer UNION. An intermediate UNION with only+the active member still fails for the nested UNION and for a STRUCT that+DuckDB interprets as the STRUCT form of a UNION. Casting a STRUCT with a+UTINYINT tag and one field per member is another alternative, but the header+does not document that cast.++The vector construction uses the documented layout and writes. It requires+FLAT child vectors. DuckDB 2.0.0 supplies them in a new vector, but the header+does not guarantee this for the children. A change can make the writes fail+or give a wrong value. See Note [Vector flatten] in+"Database.DuckDB.Simple.Materialize". Regression cases check the selected+member at the 255-member limit and a NULL payload in ValueRegressionTests.hs+("a UNION value binds the member of its tag up to the member limit of DuckDB"+and "UNION NULL payload retains its member type"). Spec.hs also checks+"round-trips struct and union parameters".+-}++-- | Build one UNION row through its tag and member vectors. See+-- Note [UNION values].+unionValueDuckValue+ :: ValueBuilder -> UnionValue FieldValue -> IO DuckDBV2ValueHandle+unionValueDuckValue builder@ValueBuilder{builderFactory = factory, builderType = makeType} UnionValue{unionValueIndex, unionValueLabel, unionValuePayload, unionValueMembers} = do+ let members = elems unionValueMembers+ index = fromIntegral unionValueIndex+ when (index >= length members) $ throwInvalidArgument "union value tag out of range"+ let UnionMemberType name rep = members !! index+ when (name /= unionValueLabel) $ throwInvalidArgument "union tag and member name mismatch"+ bracket (makeType (LogicalTypeUnion unionValueMembers)) destroyLogicalType \logical ->+ withOwnedVector factory logical \vector -> do+ checked (c_duckdb_v2_vector_set_size vector 1)+ tag <- output (c_duckdb_v2_vector_get_child vector 0)+ checked (c_duckdb_v2_vector_set_size tag 1)+ bracket (uint8DuckValue factory (fromIntegral index)) destroyValue \value ->+ checked (c_duckdb_v2_vector_set_value tag 0 value)+ forM_ [0 .. length members - 1] \memberIndex -> do+ member <- output (c_duckdb_v2_vector_get_child vector (fromIntegral memberIndex + 1))+ checked (c_duckdb_v2_vector_set_size member 1)+ if memberIndex == index+ then bracket (fieldValueWithTypeDuckValue builder rep unionValuePayload) destroyValue \value ->+ checked (c_duckdb_v2_vector_set_value member 0 value)+ else checked (c_duckdb_v2_vector_set_null member 0)+ output (c_duckdb_v2_vector_get_value vector 0)++{- Note [TIME alias values]+~~~~~~~~~~~~~~~~~~~~~~~~~~+DuckDB 2.0.0 rejects the default cast from TIME to a TIME alias. The public+API specifies that an alias keeps its base representation. TIME vector data+uses int64 microseconds since midnight, as encodeTimeOfDay does. Thus a checked+TIME payload can be written through vector_get_data_mutable without a cast.+vector_get_value preserves the declared alias. The owned vector has one FLAT+row, and its validity is set explicitly. NULL uses value_create_null instead.+-}++-- | Construct a value with explicit logical metadata, including typed NULL.+fieldValueWithTypeDuckValue+ :: ValueBuilder -> LogicalTypeRep -> FieldValue -> IO DuckDBV2ValueHandle+fieldValueWithTypeDuckValue ValueBuilder{builderFactory = factory, builderType = makeType} rep FieldNull = bracket (makeType rep) destroyLogicalType \logical ->+ output (c_duckdb_v2_value_create_null factory logical)+fieldValueWithTypeDuckValue builder@ValueBuilder{builderFactory = factory, builderType = makeType} rep value = case (rep, value) of+ (LogicalTypeAlias _ base, FieldTime time)+ | withoutAliases base == LogicalTypeScalar LogicalTypeIdTime -> do+ -- See Note [TIME alias values].+ micros <- encodeTimeOfDay time+ bracket (makeType rep) destroyLogicalType \logical -> withOwnedVector factory logical \vector -> do+ checked (c_duckdb_v2_vector_flatten vector)+ checked (c_duckdb_v2_vector_set_size vector 1)+ raw <- output (c_duckdb_v2_vector_get_data_mutable vector)+ poke (castPtr raw) micros+ validity <- output (c_duckdb_v2_vector_flat_get_validity_mutable vector)+ peek validity >>= poke validity . (.|. 1)+ output (c_duckdb_v2_vector_get_value vector 0)+ -- A value of the base type. DuckDB casts it to the alias type. See+ -- Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ (LogicalTypeAlias _ base, _) -> fieldValueWithTypeDuckValue builder base value+ (LogicalTypeScalar LogicalTypeIdVariant, _) -> variantDuckValue builder value+ (LogicalTypeScalar dtype, _) -> scalarFieldValueDuckValue factory dtype value+ (LogicalTypeCollatedVarchar _, FieldText text) -> textDuckValue factory text+ (LogicalTypeGeometry crs, FieldGeometry raw) -> rawGeometryDuckValue builder crs raw+ (LogicalTypeDecimal width scale, FieldDecimal decimal@DecimalValue{decimalWidth, decimalScale})+ | width == decimalWidth && scale == decimalScale -> decimalDuckValue factory decimal+ (LogicalTypeList child, FieldList values) -> sequenceValue c_duckdb_v2_value_create_list child values+ (LogicalTypeArray child size, FieldArray array)+ | fromIntegral (length (elems array)) == size -> do+ sequence_ (arraySizeError (length (elems array)))+ sequenceValue c_duckdb_v2_value_create_array child (elems array)+ (LogicalTypeMap keyRep valueRep, FieldMap pairs) ->+ bracket (makeType keyRep) destroyLogicalType \keyType ->+ bracket (makeType valueRep) destroyLogicalType \valueType ->+ withCreatedValues (map (fieldValueWithTypeDuckValue builder keyRep . fst) pairs) \keys ->+ withCreatedValues (map (fieldValueWithTypeDuckValue builder valueRep . snd) pairs) \values ->+ withArray keys \keyPtr -> withArray values \valuePtr ->+ output (c_duckdb_v2_value_create_map factory keyType valueType (ConstPtr keyPtr) (ConstPtr valuePtr) (fromIntegral (length pairs)))+ (LogicalTypeStruct _, FieldStruct struct)+ | sameBase (structValueTypeRep struct) -> structValueDuckValue builder struct+ (LogicalTypeTuple _, FieldTuple tuple)+ | sameBase (tupleValueTypeRep tuple) -> tupleValueDuckValue builder tuple+ (LogicalTypeUnion _, FieldUnion union)+ | sameBase (unionValueTypeRep union) -> unionValueDuckValue builder union+ (LogicalTypeEnum entries, FieldEnum _ label) -> enumDuckValue factory entries label+ _ -> typeMismatch rep value+ where+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ sameBase other = withoutAliases rep == withoutAliases other+ sequenceValue create child values = bracket (makeType child) destroyLogicalType \logical ->+ withCreatedValues (map (fieldValueWithTypeDuckValue builder child) values) \native -> withArray native \ptr ->+ output (create factory logical (ConstPtr ptr) (fromIntegral (length native)))++scalarFieldValueDuckValue :: DuckDBV2FactoryHandle -> LogicalTypeId -> FieldValue -> IO DuckDBV2ValueHandle+scalarFieldValueDuckValue factory dtype value = case (dtype, value) of+ (LogicalTypeIdBoolean, FieldBool b) -> boolDuckValue factory b+ (LogicalTypeIdTinyInt, FieldInt8 i) -> int8DuckValue factory i+ (LogicalTypeIdSmallInt, FieldInt16 i) -> int16DuckValue factory i+ (LogicalTypeIdInteger, FieldInt32 i) -> int32DuckValue factory i+ (LogicalTypeIdBigInt, FieldInt64 i) -> int64DuckValue factory i+ (LogicalTypeIdUTinyInt, FieldWord8 w) -> uint8DuckValue factory w+ (LogicalTypeIdUSmallInt, FieldWord16 w) -> uint16DuckValue factory w+ (LogicalTypeIdUInteger, FieldWord32 w) -> uint32DuckValue factory w+ (LogicalTypeIdUBigInt, FieldWord64 w) -> uint64DuckValue factory w+ (LogicalTypeIdFloat, FieldFloat f) -> floatDuckValue factory f+ (LogicalTypeIdDouble, FieldDouble d) -> doubleDuckValue factory d+ (LogicalTypeIdVarchar, FieldText t) -> textDuckValue factory t+ (LogicalTypeIdBlob, FieldBlob b) -> blobDuckValue factory b+ -- The type id has no CRS. See Note [Raw GEOMETRY parameters].+ (LogicalTypeIdGeometry, FieldGeometry raw) -> rawGeometryDuckValue (plainBuilder factory) Nothing raw+ (LogicalTypeIdUUID, FieldUUID u) -> uuidDuckValue factory u+ (LogicalTypeIdBit, FieldBit bits) -> bitDuckValue factory bits+ (LogicalTypeIdDate, FieldDate d) -> dateDuckValue factory d+ (LogicalTypeIdTime, FieldTime t) -> timeOfDayDuckValue factory t+ (LogicalTypeIdTimeNs, FieldTime t) ->+ timeOfDayUnits 1000000000 t >>= (\encoded -> output (c_duckdb_v2_value_create_time_ns factory encoded)) . fromInteger+ (LogicalTypeIdTimeTz, FieldTimeTZ tz) -> timeWithZoneDuckValue factory tz+ (LogicalTypeIdTimestamp, FieldTimestamp ts) -> localTimestampDuckValue factory ts+ (LogicalTypeIdTimestampSec, FieldTimestamp ts) ->+ encodeUnbounded (encodeTimestampUnits 1) ts >>= (\encoded -> output (c_duckdb_v2_value_create_timestamp_sec factory encoded))+ (LogicalTypeIdTimestampMs, FieldTimestamp ts) ->+ encodeUnbounded (encodeTimestampUnits 1000) ts >>= (\encoded -> output (c_duckdb_v2_value_create_timestamp_ms factory encoded))+ (LogicalTypeIdTimestampNs, FieldTimestamp ts) ->+ encodeUnbounded (encodeTimestampUnits 1000000000) ts >>= (\encoded -> output (c_duckdb_v2_value_create_timestamp_ns factory encoded))+ (LogicalTypeIdTimestampTz, FieldTimestampTZ ts) -> utcTimestampDuckValue factory ts+ (LogicalTypeIdTimestampTzNs, FieldTimestampTZ ts) -> utcTimestampNsDuckValue factory ts+ (LogicalTypeIdInterval, FieldInterval iv) -> intervalDuckValue factory iv+ (LogicalTypeIdHugeInt, FieldHugeInt i) -> hugeIntDuckValue factory i+ (LogicalTypeIdUHugeInt, FieldUHugeInt i) -> uhugeIntDuckValue factory i+ (LogicalTypeIdBigNum, FieldBigNum big) -> bigNumDuckValue factory big+ _ ->+ throwInvalidArgument+ ("unsupported scalar conversion for " <> logicalTypeSQL (LogicalTypeScalar dtype) <> " from a value of type " <> valueTypeName value)++-- | Construct an ENUM value by the default cast of the label to the ENUM type.+-- The label of a 'FieldEnum' gives the value, not its index: an index of+-- another dictionary would give another label with no error. A label that is+-- not in the dictionary raises an t'SQLError' of the type 'ErrorTypeConversion',+-- as the cast of DuckDB from VARCHAR does. See+-- Note [User casts and the casts of the library] in+-- "Database.DuckDB.Simple.Materialize".+enumDuckValue :: DuckDBV2FactoryHandle -> Array Int Text -> Text -> IO DuckDBV2ValueHandle+enumDuckValue factory entries label = do+ unless (label `elem` elems entries) $+ throwConversionError ("the ENUM label '" <> label <> "' is not in the dictionary of the type")+ bracket (logicalTypeFromRep factory (LogicalTypeEnum entries)) destroyLogicalType \logical ->+ bracket (textDuckValue factory label) destroyValue \value ->+ defaultCast factory value logical++-- | Cast a value to a type with the default cast of DuckDB+-- ('withDefaultCastVector' with one row). The caller destroys the result. See+-- Note [User casts and the casts of the library] in+-- "Database.DuckDB.Simple.Materialize".+defaultCast :: DuckDBV2FactoryHandle -> DuckDBV2ValueHandle -> DuckDBV2LogicalTypeHandle -> IO DuckDBV2ValueHandle+defaultCast factory value logical =+ withDefaultCastVector factory logical 1 (\_ write -> write value) \vector -> output (c_duckdb_v2_vector_get_value vector 0)++hugeIntDuckValue :: DuckDBV2FactoryHandle -> Integer -> IO DuckDBV2ValueHandle+hugeIntDuckValue factory value = integerToHugeInt value >>= \huge -> with huge (\ptr -> output (c_duckdb_v2_value_create_hugeint factory (ConstPtr ptr)))++uhugeIntDuckValue :: DuckDBV2FactoryHandle -> Integer -> IO DuckDBV2ValueHandle+uhugeIntDuckValue factory value = integerToUHugeInt value >>= \huge -> with huge (\ptr -> output (c_duckdb_v2_value_create_uhugeint factory (ConstPtr ptr)))++decimalDuckValue :: DuckDBV2FactoryHandle -> DecimalValue -> IO DuckDBV2ValueHandle+decimalDuckValue factory DecimalValue{decimalWidth, decimalScale, decimalInteger} = do+ when (decimalWidth < 1 || decimalWidth > 38 || decimalScale > decimalWidth) $+ throwInvalidArgument "invalid DECIMAL width or scale"+ when (abs decimalInteger >= 10 ^ decimalWidth) $ throwConversionError "DECIMAL value exceeds declared precision"+ huge <- integerToHugeInt decimalInteger+ with huge (\ptr -> output (c_duckdb_v2_value_create_decimal factory (ConstPtr ptr) decimalWidth decimalScale))++intervalDuckValue :: DuckDBV2FactoryHandle -> IntervalValue -> IO DuckDBV2ValueHandle+intervalDuckValue factory IntervalValue{intervalMonths, intervalDays, intervalMicros} =+ with+ (DuckDBV2IntervalT intervalMonths intervalDays intervalMicros)+ (\ptr -> output (c_duckdb_v2_value_create_interval factory (ConstPtr ptr)))++timeWithZoneDuckValue :: DuckDBV2FactoryHandle -> TimeWithZone -> IO DuckDBV2ValueHandle+timeWithZoneDuckValue factory TimeWithZone{timeWithZoneTime, timeWithZoneOffsetSeconds} = do+ micros <- timeOfDayUnits 1000000 timeWithZoneTime+ let offset = toInteger timeWithZoneOffsetSeconds+ when (abs offset > toInteger timeTzMaxOffsetSeconds) $+ throwConversionError+ ("the TIMETZ offset " <> Text.pack (show offset) <> " seconds is out of range; DuckDB stores an offset from -57599 to 57599 seconds")+ output+ ( c_duckdb_v2_value_create_time_tz+ factory+ ((fromIntegral micros `shiftL` timeTzOffsetBits) .|. fromIntegral (toInteger timeTzMaxOffsetSeconds - offset))+ )++integerToHugeInt :: Integer -> IO DuckDBV2HugeintT+integerToHugeInt value = do+ when (value < negate (1 `shiftL` 127) || value >= 1 `shiftL` 127) $ throwConversionError "HUGEINT value out of range"+ pure (DuckDBV2HugeintT (fromIntegral value) (fromIntegral (value `shiftR` 64)))++integerToUHugeInt :: Integer -> IO DuckDBV2UhugeintT+integerToUHugeInt value = do+ when (value < 0 || value >= 1 `shiftL` 128) $ throwConversionError "UHUGEINT value out of range"+ pure (DuckDBV2UhugeintT (fromIntegral value) (fromIntegral (value `shiftR` 64)))++-- | Borrow generated identifier views and reject embedded NUL bytes.+withIdentifiers :: [Text] -> (Ptr DuckDBV2IdentifierT -> IO a) -> IO a+withIdentifiers names action = do+ when (any (Text.any (== '\0')) names) $+ throwInvalidArgument "logical type name contains NUL"+ withIdentifierArray names action++-- | Raise the error of a value that does not fit its declared type. The text+-- names the type as DuckDB does. See Note [Type texts] in+-- "Database.DuckDB.Simple.LogicalRep".+typeMismatch :: LogicalTypeRep -> FieldValue -> IO a+typeMismatch expected actual = throwInvalidArgument ("cannot encode a value of type " <> valueTypeName actual <> " as " <> logicalTypeSQL expected)++-- | Construct a GEOMETRY value of the declared CRS from a raw geometry. See+-- Note [Raw GEOMETRY parameters], and Note [User casts and the casts of the library] in+-- "Database.DuckDB.Simple.Materialize".+rawGeometryDuckValue :: ValueBuilder -> Maybe Text -> RawGeometry -> IO DuckDBV2ValueHandle+rawGeometryDuckValue ValueBuilder{builderFactory = factory, builderType = makeType} crs RawGeometry{rawGeometryWKB, rawGeometryCRS} = do+ when (rawGeometryCRS /= crs) $+ throwInvalidArgument ("a GEOMETRY value with " <> describeCRS rawGeometryCRS <> " cannot bind as a GEOMETRY with " <> describeCRS crs)+ bracket (makeType (LogicalTypeGeometry crs)) destroyLogicalType \logical ->+ bracket (blobDuckValue factory rawGeometryWKB) destroyValue \blob ->+ defaultCast factory blob logical+ where+ describeCRS = maybe "no CRS" (("the CRS " <>) . quotedCRS)++-- | Put a CRS between single quotes for an error message.+quotedCRS :: Text -> Text+quotedCRS crs = "'" <> crs <> "'"++{- Note [Raw GEOMETRY parameters]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+RawGeometry holds WKB bytes and an optional CRS. GEOMETRY results use this+representation, including FieldGeometry values inside composite values.+rawGeometryDuckValue casts a BLOB of the bytes to the declared GEOMETRY type.+It uses the default cast, so a user cast cannot change the value. See+Note [User casts and the casts of the library] in+"Database.DuckDB.Simple.Materialize".++A top-level RawGeometry parameter declares its own CRS. DuckDB then applies+the casts of SQL text to the parameter's target type:++- A value without a CRS takes the CRS of a target that has one.+- A target without a CRS removes the CRS without an error.+- A target with another CRS raises "Binder Error: Cannot cast GEOMETRY with+ CRS ... to GEOMETRY with different CRS ...". A comparison with = has the+ same rule.++In a STRUCT, TUPLE, or UNION, the CRS of each raw geometry must equal its+declared CRS, including when both have no CRS. A mismatch raises SQLError+with ErrorInputInvalid. The BLOB cast assigns the declared CRS without+checking the original label, so omitting this check could change or remove+the label silently. The statement builder resolves CRS types at bind time.+See Note [Bind-time CRS types] in "Database.DuckDB.Simple.V2.LogicalRep".+A plain GEOMETRY declaration has no CRS and uses the factory's plain builder.++A VARIANT accepts raw geometry only without a CRS. DuckDB silently removes+CRS metadata when it casts GEOMETRY to VARIANT, so the binding rejects a+CRS-tagged payload with ErrorInputInvalid. RawGeometry has no+DuckDBColumnType or ToDuckValue instance. A list or SizedArray of RawGeometry,+or a typed row containing it, would declare GEOMETRY without a CRS and lose+the labels.++An instance factory cannot construct a GEOMETRY type with a CRS. It raises+"Binder Error: Cannot create GEOMETRY type with coordinate system without a+connection". This limits toDuckValue for composite values with CRS-tagged+geometry leaves, including typed NULLs. A connection factory has no such+limit.++DuckDB 2.0.0 stores GEOMETRY as WKB. The BLOB cast and ST_GeomFromWKB both use+Geometry::FromBinary (src/function/cast/geo_casts.cpp and+src/function/scalar/geometry/geometry_functions.cpp). The result reader casts+GEOMETRY to BLOB with Geometry::ToBinary, which copies the stored bytes+(src/common/types/geometry.cpp). Thus a result binds again with the same+bytes. Import can normalize other input: big-endian WKB and EWKB become+little-endian ISO WKB. It rejects collections with different Z/M member+layouts. GeometryRegressionTests.hs covers result rebinding, top-level CRS+casts, composite CRS mismatches, and nested typed NULLs.+-}++-- | Cast the payload to VARIANT with the default cast of DuckDB. See+-- Note [User casts and the casts of the library] in+-- "Database.DuckDB.Simple.Materialize".+variantDuckValue :: ValueBuilder -> FieldValue -> IO DuckDBV2ValueHandle+variantDuckValue builder@ValueBuilder{builderFactory = factory, builderType = makeType} value = do+ (rep, payload) <- variantPayloadType value+ bracket (makeType (LogicalTypeScalar LogicalTypeIdVariant)) destroyLogicalType \logical ->+ bracket (fieldValueWithTypeDuckValue builder rep payload) destroyValue \native ->+ defaultCast factory native logical++-- | Construct a UTC nanosecond value without losing sub-microsecond digits.+utcTimestampNsDuckValue :: DuckDBV2FactoryHandle -> UTCTimestamp -> IO DuckDBV2ValueHandle+utcTimestampNsDuckValue factory value =+ encodeUnbounded (encodeTimestampUnits 1000000000 . utcToLocalTime utc) value+ >>= (\encoded -> output (c_duckdb_v2_value_create_timestamp_tz_ns factory encoded))++-- | Cast the WKT text of the shape to GEOMETRY with the default cast of DuckDB.+-- See Note [User casts and the casts of the library] in+-- <src/Database.DuckDB.Simple.Materialize.html Database.DuckDB.Simple.Materialize>.+instance ToDuckValue G.Geometry where+ toDuckValue factory geometry = do+ wkt <- either (throwInvalidArgument . Text.pack) pure (WKT.encodeWKT geometry)+ bracket (logicalTypeFromRep factory (LogicalTypeGeometry Nothing)) destroyLogicalType \logical ->+ bracket (textDuckValue factory wkt) destroyValue \value ->+ defaultCast factory value logical++instance ToDuckValue Variant where+ toDuckValue factory = variantDuckValue (plainBuilder factory) . variantPayload++-- | Choose the native type of a VARIANT payload. Containers get VARIANT+-- elements and fields. A time value with digits below a microsecond uses the+-- nanosecond type when it fits the range of that type (fitsNanoseconds). Each+-- other time value uses the microsecond type, as a VARIANT parameter does. See+-- Note [Nanosecond parameters].+-- The C API cannot construct a nonempty object whose keys are all empty.+-- DuckDB 2.0 treats that STRUCT type as a tuple and casts it to a VARIANT array.+-- A raw geometry without a CRS uses GEOMETRY. See Note [Raw GEOMETRY parameters].+-- A VARIANT does not keep a UNION, a MAP, an ENUM, or the CRS of a geometry, so+-- each of these payloads raises an error, at each depth.+variantPayloadType :: FieldValue -> IO (LogicalTypeRep, FieldValue)+variantPayloadType value = case value of+ FieldNull -> pure (variant, value)+ FieldBool{} -> scalar LogicalTypeIdBoolean+ FieldInt8{} -> scalar LogicalTypeIdTinyInt+ FieldInt16{} -> scalar LogicalTypeIdSmallInt+ FieldInt32{} -> scalar LogicalTypeIdInteger+ FieldInt64{} -> scalar LogicalTypeIdBigInt+ FieldWord8{} -> scalar LogicalTypeIdUTinyInt+ FieldWord16{} -> scalar LogicalTypeIdUSmallInt+ FieldWord32{} -> scalar LogicalTypeIdUInteger+ FieldWord64{} -> scalar LogicalTypeIdUBigInt+ FieldHugeInt{} -> scalar LogicalTypeIdHugeInt+ FieldUHugeInt{} -> scalar LogicalTypeIdUHugeInt+ FieldFloat{} -> scalar LogicalTypeIdFloat+ FieldDouble{} -> scalar LogicalTypeIdDouble+ FieldDecimal DecimalValue{decimalWidth, decimalScale} -> pure (LogicalTypeDecimal decimalWidth decimalScale, value)+ FieldText{} -> scalar LogicalTypeIdVarchar+ FieldBlob{} -> scalar LogicalTypeIdBlob+ FieldUUID{} -> scalar LogicalTypeIdUUID+ FieldDate{} -> scalar LogicalTypeIdDate+ FieldTime time+ | hasSubMicroseconds time -> scalar LogicalTypeIdTimeNs+ | otherwise -> scalar LogicalTypeIdTime+ FieldTimestamp (Finite LocalTime{localTimeOfDay})+ | hasSubMicroseconds localTimeOfDay && fitsNanoseconds value -> scalar LogicalTypeIdTimestampNs+ FieldTimestamp{} -> scalar LogicalTypeIdTimestamp+ FieldTimestampTZ (Finite instant)+ | hasSubMicroseconds (localTimeOfDay (utcToLocalTime utc instant)) && fitsNanoseconds value -> scalar LogicalTypeIdTimestampTzNs+ FieldTimestampTZ{} -> scalar LogicalTypeIdTimestampTz+ FieldTimeTZ{} -> scalar LogicalTypeIdTimeTz+ FieldInterval{} -> scalar LogicalTypeIdInterval+ FieldBigNum{} -> scalar LogicalTypeIdBigNum+ FieldBit{} -> scalar LogicalTypeIdBit+ FieldList{} -> pure (LogicalTypeList variant, value)+ FieldArray items -> pure (LogicalTypeList variant, FieldList (elems items))+ FieldTuple TupleValue{tupleValueFields} -> pure (LogicalTypeList variant, FieldList (elems tupleValueFields))+ FieldStruct structValue@StructValue{structValueTypes} -> do+ let names = map structFieldName (elems structValueTypes)+ when (Set.size (Set.fromList names) /= length names) $+ throwInvalidArgument "VARIANT objects need unique keys"+ when (names == [Text.empty]) $+ throwInvalidArgument "DuckDB's C API cannot construct a VARIANT object with only an empty key"+ let types = fmap (\field -> field{structFieldValue = variant}) structValueTypes+ pure (LogicalTypeStruct types, FieldStruct structValue{structValueTypes = types})+ FieldUnion{} -> throwInvalidArgument "a VARIANT keeps only the member value of a UNION, so a UNION value cannot bind as a VARIANT"+ FieldGeometry RawGeometry{rawGeometryCRS = Just crs} ->+ throwInvalidArgument+ ("a VARIANT does not keep the CRS of a GEOMETRY, so the GEOMETRY value with the CRS " <> quotedCRS crs <> " cannot bind as a VARIANT")+ FieldGeometry{} -> pure (LogicalTypeGeometry Nothing, value)+ FieldMap{} -> throwInvalidArgument "a VARIANT stores a MAP as a list of key and value objects, so a MAP value cannot bind as a VARIANT"+ FieldEnum{} -> throwInvalidArgument "a VARIANT stores an ENUM as its text, so an ENUM value cannot bind as a VARIANT"+ where+ variant = LogicalTypeScalar LogicalTypeIdVariant+ scalar dtype = pure (LogicalTypeScalar dtype, value)++instance ToDuckValue Null where+ toDuckValue factory _ = nullDuckValue factory++instance ToDuckValue Bool where+ toDuckValue = boolDuckValue++instance ToDuckValue Int where+ toDuckValue factory = int64DuckValue factory . fromIntegral++instance ToDuckValue Int8 where+ toDuckValue = int8DuckValue++instance ToDuckValue Int16 where+ toDuckValue = int16DuckValue++instance ToDuckValue Int32 where+ toDuckValue = int32DuckValue++instance ToDuckValue Int64 where+ toDuckValue = int64DuckValue++instance ToDuckValue BigNum where+ toDuckValue = bigNumDuckValue++instance ToDuckValue UUID.UUID where+ toDuckValue = uuidDuckValue++-- | A value from -2^127 to 2^127 - 1 gives a HUGEINT value. Each other value+-- raises an error. See Note [Integer parameters].+instance ToDuckValue Integer where+ toDuckValue factory value = case hugeIntInRange value of+ Just make -> make factory+ Nothing -> outsideRange "HUGEINT" value++-- | A value from 0 to 2^127 - 1 gives a HUGEINT value, and a value up to+-- 2^128 - 1 a UHUGEINT value. Each larger value raises an error. See+-- Note [Integer parameters].+instance ToDuckValue Natural where+ toDuckValue factory value = case hugeIntInRange (toInteger value) of+ Just make -> make factory+ Nothing+ | toInteger value < 1 `shiftL` 128 -> uhugeIntDuckValue factory (toInteger value)+ | otherwise -> outsideRange "UHUGEINT" (toInteger value)++-- | Raise the range error of 'toDuckValue' for an 'Integer' or a 'Natural'+-- outside the declared type. See Note [Integer parameters].+outsideRange :: Text -> Integer -> IO a+outsideRange typeName value =+ throwConversionError+ ("the value " <> Text.pack (show value) <> " is outside the range of " <> typeName <> ". Use BigNum for a larger value.")++instance ToDuckValue Word where+ toDuckValue factory = uint64DuckValue factory . fromIntegral++instance ToDuckValue Word16 where+ toDuckValue = uint16DuckValue++instance ToDuckValue Word32 where+ toDuckValue = uint32DuckValue++instance ToDuckValue Word64 where+ toDuckValue = uint64DuckValue++instance ToDuckValue Word8 where+ toDuckValue = uint8DuckValue++instance ToDuckValue Double where+ toDuckValue = doubleDuckValue++instance ToDuckValue Float where+ toDuckValue = floatDuckValue++instance ToDuckValue Text where+ toDuckValue = textDuckValue++instance ToDuckValue String where+ toDuckValue = stringDuckValue++instance ToDuckValue BS.ByteString where+ toDuckValue = blobDuckValue++instance ToDuckValue BitString where+ toDuckValue = bitDuckValue++instance ToDuckValue Day where+ toDuckValue = dayDuckValue++instance ToDuckValue TimeOfDay where+ toDuckValue = timeOfDayDuckValue++instance ToDuckValue LocalTime where+ toDuckValue = localTimeDuckValue++instance ToDuckValue IntervalValue where+ toDuckValue = intervalDuckValue++instance ToDuckValue TimeWithZone where+ toDuckValue = timeWithZoneDuckValue++instance ToDuckValue UTCTime where+ toDuckValue = utcTimeDuckValue++instance ToDuckValue (Unbounded Day) where+ toDuckValue = dateDuckValue++instance ToDuckValue (Unbounded LocalTime) where+ toDuckValue = localTimestampDuckValue++instance ToDuckValue (Unbounded UTCTime) where+ toDuckValue = utcTimestampDuckValue++instance ToDuckValue (StructValue FieldValue) where+ toDuckValue = structValueDuckValue . plainBuilder++instance ToDuckValue (TupleValue FieldValue) where+ toDuckValue = tupleValueDuckValue . plainBuilder++instance ToDuckValue (UnionValue FieldValue) where+ toDuckValue = unionValueDuckValue . plainBuilder++-- | Construct a MAP through the supplied factory with the key type of @k@ and+-- the value type of @v@. The pairs go in ascending key order. A structured+-- declaration of a part is its target. See Note [Type declarations].+instance (DuckDBColumnType k, DuckDBColumnType v, ToDuckValue k, ToDuckValue v) => ToDuckValue (Map.Map k v) where+ toDuckValue factory =+ mapDuckValue+ (withElementType (Proxy @k) builder (declaredTarget (Proxy @k)))+ (withElementType (Proxy @v) builder (declaredTarget (Proxy @v)))+ builder+ (toDuckValue factory)+ (toDuckValue factory)+ . Map.toAscList+ where+ builder = plainBuilder factory++-- | Construct an ARRAY of size @n@ through the supplied factory with the+-- element type of @a@. A structured declaration of @a@ is the target of the+-- elements. See Note [Sized arrays].+instance (KnownNat n, DuckDBColumnType a, ToDuckValue a) => ToDuckValue (SizedArray n a) where+ toDuckValue factory = arrayDuckValue (plainBuilder factory) (declaredTarget (Proxy @a)) (toDuckValue factory) . elems . sizedArrayToArray++-- | Construct a LIST through the supplied factory with the element type of @a@.+-- 'String' keeps its own instance. A structured declaration of @a@ is the+-- target of the elements. See Note [Type declarations].+instance {-# OVERLAPPABLE #-} (DuckDBColumnType a, ToDuckValue a) => ToDuckValue [a] where+ toDuckValue factory = listDuckValue (plainBuilder factory) (declaredTarget (Proxy @a)) (toDuckValue factory)++instance (ToDuckValue a) => ToDuckValue (Maybe a) where+ toDuckValue factory Nothing = nullDuckValue factory+ toDuckValue factory (Just value) = toDuckValue factory value++-- | Preserve infinity sentinels and validate finite values before narrowing.+encodeUnbounded :: (Integral b, Bounded b) => (a -> IO b) -> Unbounded a -> IO b+encodeUnbounded _ NegInfinity = pure (negate maxBound)+encodeUnbounded _ PosInfinity = pure maxBound+encodeUnbounded encode (Finite value) = encode value++-- | Encode finite dates as days from the Unix epoch. A day outside the range+-- of DuckDB raises an error. See Note [Time ranges].+encodeDay :: Day -> IO Int32+encodeDay day = do+ let days = diffDays day (fromGregorian 1970 1 1)+ when (days < -2147483646) $+ throwOutsideTimeRange (show day) "below the smallest DATE of DuckDB, -5877641-06-25"+ when (days > 2147483646) $+ throwOutsideTimeRange (show day) "above the largest DATE of DuckDB, 5881580-07-10"+ pure (fromInteger days)++-- | Encode a time of day at DuckDB microsecond precision.+encodeTimeOfDay :: TimeOfDay -> IO Int64+encodeTimeOfDay tod = fromInteger <$> timeOfDayUnits 1000000 tod++-- | Encode finite timestamps without native calendar conversions.+encodeLocalTime :: LocalTime -> IO Int64+encodeLocalTime = encodeTimestampUnits 1000000++-- | Encode finite timestamps in the requested number of units per second.+-- See Note [Time ranges].+encodeTimestampUnits :: Integer -> LocalTime -> IO Int64+encodeTimestampUnits units value@LocalTime{localTimeOfDay} = do+ -- timeOfDayUnits checks the time of day.+ _ <- timeOfDayUnits units localTimeOfDay+ checkTimestampRange units value+ let total = timestampUnits units value+ -- Only a nanosecond value reaches this check with a value outside the+ -- storage or at an infinity value. See Note [Time ranges].+ checkFiniteRange "TIMESTAMP" (minBound :: Int64) maxBound total+ pure (fromInteger total)++{- Note [Time ranges]+~~~~~~~~~~~~~~~~~~~~~+The storage of a DATE and of a TIMESTAMP can hold values that DuckDB 2.0.0+does not accept. The SQL literal of each type, and the casts to it, reject+these values:++- DATE: from -5877641-06-25 (5877642-06-25 BC) to 5881580-07-10, days+ -2147483646 to 2147483646 from the Unix epoch. The INT32 storage also holds+ day -2147483648.++- TIMESTAMP, TIMESTAMPTZ, TIMESTAMP_S, and TIMESTAMP_MS: from -290308-12-22+ 00:00:00 (290309-12-22 BC) to 294247-01-10 04:00:54.775806, microseconds+ -9223372022400000000 to 2^63 - 2. The INT64 storage of a microsecond value+ also holds about four hours before that start, and the storage of a second+ or a millisecond value holds much more at each end.++A value outside these ranges breaks ordinary SQL in DuckDB 2.0.0. A cast of+such a TIMESTAMP to VARCHAR raises an INTERNAL error, and DuckDB then+invalidates the database. A cast of such a TIMESTAMP to TIMESTAMP_S stores the+first valid value with no error. A DATE before the range sorts before+-infinity, and a cast of it to TIMESTAMP fails. Thus encodeDay and+encodeTimestampUnits raise an SQLError of the type ErrorTypeConversion for+each finite value outside these ranges (throwOutsideTimeRange), before any+check of the storage, so a value outside the storage and a value at an+infinity value get this error too. The text names the value and the end of+the range in the notation of Data.Time, where the year -5877641 is 5877642 BC.+A nanosecond value has the range of its INT64 storage, and the literal of+TIMESTAMP_NS accepts that range, so it gets no check here. checkFiniteRange+then catches only a nanosecond value outside its storage or at an infinity+value.++A leap second gets no check. A TimeOfDay with 60 seconds binds as the next+second, as make_time(23, 59, 60) and make_timestamp(2016, 12, 31, 23, 59, 60)+of DuckDB give it. The SQL literal '23:59:60' raises an error.+-}++-- | Reject a second, millisecond, or microsecond timestamp outside the range+-- of DuckDB. See Note [Time ranges].+checkTimestampRange :: Integer -> LocalTime -> IO ()+checkTimestampRange units value+ | units > 1000000 = pure ()+ | micros < -9223372022400000000 =+ throwOutsideTimeRange (show value) ("below the smallest " <> name <> " of DuckDB, -290308-12-22 00:00:00")+ | micros > 9223372036854775806 = throwOutsideTimeRange (show value) ("above the largest " <> name <> " of DuckDB, " <> largest)+ | otherwise = pure ()+ where+ micros = timestampUnits 1000000 value+ (name, largest) = case units of+ 1 -> ("TIMESTAMP_S", "294247-01-10 04:00:54")+ 1000 -> ("TIMESTAMP_MS", "294247-01-10 04:00:54.775")+ _ -> ("TIMESTAMP", "294247-01-10 04:00:54.775806")++-- | Raise the error for a date or a timestamp outside the range of DuckDB. The+-- text names the Haskell value and the end of the range. See+-- Note [Time ranges].+throwOutsideTimeRange :: String -> Text -> IO a+throwOutsideTimeRange value bound = throwConversionError ("the value " <> Text.pack value <> " is " <> bound)++-- | Count the units from the Unix epoch to a timestamp, with the digits below+-- one unit cut toward the earlier time. The count does not check the time of+-- day.+timestampUnits :: Integer -> LocalTime -> Integer+timestampUnits units LocalTime{localDay, localTimeOfDay} =+ diffDays localDay (fromGregorian 1970 1 1) * 86400 * units+ + diffTimeToPicoseconds (timeOfDayToTime localTimeOfDay) `div` (1000000000000 `div` units)++-- | Check storage limits and the two DuckDB infinity sentinels.+checkFiniteRange :: (Integral a) => String -> a -> a -> Integer -> IO ()+checkFiniteRange label lower upper value =+ unless (inFiniteRange lower upper value) $+ throwConversionError (Text.pack label <> " value out of finite range")++-- | Check storage limits without accepting the two infinity sentinels.+inFiniteRange :: (Integral a) => a -> a -> Integer -> Bool+inFiniteRange lower upper value =+ value >= toInteger lower && value < toInteger upper && value /= negate (toInteger upper)++-- | Validate time components and convert to the requested units per second.+timeOfDayUnits :: Integer -> TimeOfDay -> IO Integer+timeOfDayUnits units tod@(TimeOfDay hours minutes seconds) = do+ -- A time of day that is not a TIME value is outside the range of the type: a+ -- field outside its own range, for example 23:61:00, and a time after+ -- 24:00:00 with valid fields, for example 24:00:01, 25:00:00, or 23:59:60.5.+ -- The two texts name the two cases.+ when (hours < 0 || minutes < 0 || minutes > 59 || seconds < 0 || seconds >= 61) $+ throwConversionError "invalid time of day"+ let picos = diffTimeToPicoseconds (timeOfDayToTime tod)+ when (hours > 24 || picos > 86400 * 1000000000000) $+ throwConversionError "time of day out of range"+ pure (picos `div` (1000000000000 `div` units))++-- | Transfer a newly constructed value to the statement on success.+bindDuckValue :: Statement -> DuckDBIdx -> IO DuckDBV2ValueHandle -> IO ()+bindDuckValue statement index make = mask_ do+ value <- make+ storeBoundValue statement index value `onException` destroyValue value
src/Database/DuckDB/Simple/ToRow.hs view
@@ -2,64 +2,63 @@ {-# LANGUAGE FlexibleContexts #-} {-# LANGUAGE TypeOperators #-} -{- |-Module : Database.DuckDB.Simple.ToRow-Description : Rendering Haskell collections into parameter rows.--The `ToRow` class converts Haskell values into lists of `FieldBinding`s that-can be consumed by statement binding functions.--}+-- |+-- Module : Database.DuckDB.Simple.ToRow+-- Description : Rendering Haskell collections into parameter rows.+--+-- The 'ToRow' class converts Haskell values to lists of 'FieldBinding's for+-- statement binding. module Database.DuckDB.Simple.ToRow (- ToRow (..),- GToRow (..),+ ToRow (..),+ GToRow (..), ) where import Database.DuckDB.Simple.ToField (FieldBinding, ToField (..)) import Database.DuckDB.Simple.Types (Only (..), (:.) (..)) import GHC.Generics --- | Types that can be transformed into parameter bindings.+-- | Types that can convert to parameter bindings. class ToRow a where- toRow :: a -> [FieldBinding]- default toRow :: (Generic a, GToRow (Rep a)) => a -> [FieldBinding]- toRow = gtoRow . from+ toRow :: a -> [FieldBinding]+ default toRow :: (Generic a, GToRow (Rep a)) => a -> [FieldBinding]+ toRow = gtoRow . from --- | Generic helper for deriving `ToRow`.+-- | Generic helper for deriving 'ToRow'. class GToRow f where- gtoRow :: f p -> [FieldBinding]+ gtoRow :: f p -> [FieldBinding] instance GToRow U1 where- gtoRow _ = []+ gtoRow _ = [] instance (ToField a) => GToRow (K1 i a) where- gtoRow (K1 a) = [toField a]+ gtoRow (K1 a) = [toField a] instance (GToRow a, GToRow b) => GToRow (a :*: b) where- gtoRow (a :*: b) = gtoRow a ++ gtoRow b+ gtoRow (a :*: b) = gtoRow a ++ gtoRow b instance (GToRow a) => GToRow (M1 i c a) where- gtoRow (M1 a) = gtoRow a+ gtoRow (M1 a) = gtoRow a instance ToRow () where- toRow () = []+ toRow () = [] instance (ToField a) => ToRow (Only a) where- toRow (Only a) = [toField a]+ toRow (Only a) = [toField a] instance (ToField a, ToField b) => ToRow (a, b) where- toRow (a, b) = [toField a, toField b]+ toRow (a, b) = [toField a, toField b] instance (ToField a, ToField b, ToField c) => ToRow (a, b, c) where- toRow (a, b, c) = [toField a, toField b, toField c]+ toRow (a, b, c) = [toField a, toField b, toField c] instance (ToField a, ToField b, ToField c, ToField d) => ToRow (a, b, c, d) where- toRow (a, b, c, d) = [toField a, toField b, toField c, toField d]+ toRow (a, b, c, d) = [toField a, toField b, toField c, toField d] instance (ToField a, ToField b, ToField c, ToField d, ToField e) => ToRow (a, b, c, d, e) where- toRow (a, b, c, d, e) = [toField a, toField b, toField c, toField d, toField e]+ toRow (a, b, c, d, e) = [toField a, toField b, toField c, toField d, toField e] instance (ToField a) => ToRow [a] where- toRow = fmap toField+ toRow = fmap toField instance (ToRow a, ToRow b) => ToRow (a :. b) where- toRow (a :. b) = toRow a ++ toRow b+ toRow (a :. b) = toRow a ++ toRow b
+ src/Database/DuckDB/Simple/Tokenizer.hs view
@@ -0,0 +1,57 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE StrictData #-}++-- |+-- Module : Database.DuckDB.Simple.Tokenizer+-- Description : SQL tokenization using the grammar of a V2 connection.+module Database.DuckDB.Simple.Tokenizer (+ Token (..),+ TokenType (..),+ Tokenization (..),+ tokenize,+) where++import Control.Exception (bracket)+import qualified Data.ByteString as BS+import Data.Text (Text)+import qualified Data.Text.Encoding as Text+import Database.DuckDB.Simple.Internal (Connection, withConnectionHandle)+import Database.DuckDB.Simple.Native+import Foreign.Marshal.Alloc (alloca)+import Foreign.Storable (peek)++-- | A token and its location in the UTF-8 source bytes.+data Token = Token+ { tokenType :: TokenType+ , tokenOffset :: Int+ , tokenBytes :: BS.ByteString+ }+ deriving (Eq, Show)++-- | Tokens and the native indication of an unclosed final token.+data Tokenization = Tokenization+ { tokenizationTokens :: [Token]+ , tokenizationUnterminated :: Bool+ }+ deriving (Eq, Show)++-- | Tokenize text, including comments and interior NUL bytes.+tokenize :: Connection -> Text -> IO Tokenization+tokenize conn sql = withConnectionHandle conn \handle ->+ let bytes = Text.encodeUtf8 sql+ in withBytes bytes \source ->+ bracket (output (c_duckdb_v2_tokenize_sql handle source)) (destroy c_duckdb_v2_token_iterator_destroy) \iterator -> do+ tokens <- alloca \kind -> alloca \start -> alloca \count ->+ let collect acc = do+ checked (c_duckdb_v2_token_iterator_next iterator kind start count)+ typ <- peek kind+ if typ == DuckDBV2TokenTypeEndOfInput+ then pure (reverse acc)+ else do+ tokenKind <- tokenTypeFromNative typ+ offset <- fromIntegral <$> peek start+ size <- fromIntegral <$> peek count+ collect (Token tokenKind offset (BS.take size (BS.drop offset bytes)) : acc)+ in collect []+ unfinished <- output (c_duckdb_v2_token_iterator_ends_unterminated iterator)+ pure (Tokenization tokens (unfinished /= 0))
− src/Database/DuckDB/Simple/TypeCache.hs
@@ -1,174 +0,0 @@-{-# LANGUAGE BlockArguments #-}-{-# LANGUAGE LambdaCase #-}-{-# LANGUAGE NamedFieldPuns #-}-{-# LANGUAGE OverloadedStrings #-}--{- | Native types that the C API cannot create. A connection reads them the-first time a parameter needs one, and destroys them when it closes.--}-module Database.DuckDB.Simple.TypeCache (- TypeCache,- defaultGeometryCRS,- createTypeCache,- destroyTypeCache,- cachedLogicalType,-) where--import Control.Concurrent.MVar (MVar, modifyMVarMasked, modifyMVar_, newMVar)-import Control.Exception (bracket, finally, mask_, onException, throwIO)-import Control.Monad (forM_, when)-import qualified Data.ByteString as BS-import Data.List (nub)-import Data.Map.Strict (Map)-import qualified Data.Map.Strict as Map-import Data.Text (Text)-import qualified Data.Text as Text-import qualified Data.Text.Encoding as TextEncoding-import Database.DuckDB.FFI-import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), destroyLogicalType, logicalTypeToRep)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Marshal.Utils (fillBytes)-import Foreign.Ptr (Ptr, nullPtr)-import Foreign.Storable (peek, poke, sizeOf)--{- | The configured CRSs and the native types of a connection. The types are-absent until a parameter needs one. The connection owns the types.--}-data TypeCache = TypeCache- { typeCacheDatabase :: !DuckDBDatabase- , typeCacheCRS :: ![Text]- , typeCacheTypes :: !(MVar (Maybe NativeTypes))- }--{- | The VARIANT type and the GEOMETRY types for the configured CRSs. Each-GEOMETRY type has two keys: the configured CRS text and the CRS text that-DuckDB reports for the type.--}-data NativeTypes = NativeTypes- { nativeVariant :: !DuckDBLogicalType- , nativeGeometry :: !(Map Text DuckDBLogicalType)- , nativeOwned :: ![DuckDBLogicalType]- }---- | The CRS that a connection reads when the options do not give a list.-defaultGeometryCRS :: [Text]-defaultGeometryCRS = ["OGC:CRS84"]--{- | Make an empty cache for a database. The caller must destroy the cache. A-CRS must be nonempty and must not contain NUL.--}-createTypeCache :: DuckDBDatabase -> [Text] -> IO TypeCache-createTypeCache database crsList = do- let crss = nub crsList- when (any Text.null crss || any (Text.any (== '\0')) crss) $- throwIO (userError "duckdb-simple: a GEOMETRY CRS must be nonempty and must not contain NUL")- TypeCache database crss <$> newMVar Nothing---- | Destroy the types in the cache, if the connection read them.-destroyTypeCache :: TypeCache -> IO ()-destroyTypeCache TypeCache{typeCacheTypes} =- modifyMVar_ typeCacheTypes \types -> Nothing <$ mapM_ (mapM_ destroyLogicalType . nativeOwned) types--{- | Copy a cached type for a leaf that the C API cannot create. The caller-must destroy the copy. The first VARIANT leaf or GEOMETRY leaf with a CRS-reads the types. A GEOMETRY CRS that the cache does not hold gives GEOMETRY-without a CRS.--}-cachedLogicalType :: TypeCache -> LogicalTypeRep -> IO DuckDBLogicalType-cachedLogicalType cache = \case- LogicalTypeScalar DuckDBTypeVariant -> nativeTypes cache >>= copyLogicalType . nativeVariant- LogicalTypeGeometry (Just crs) -> do- NativeTypes{nativeGeometry} <- nativeTypes cache- maybe (c_duckdb_create_logical_type DuckDBTypeGeometry) copyLogicalType (Map.lookup crs nativeGeometry)- LogicalTypeGeometry Nothing -> c_duckdb_create_logical_type DuckDBTypeGeometry- other -> throwIO (userError ("duckdb-simple: the type cache cannot create " <> show other))--{- | Get the native types. Read them on first use with a separate connection,-so the query does not run in the transaction of the caller. A failed read-leaves the cache empty.--}-nativeTypes :: TypeCache -> IO NativeTypes-nativeTypes TypeCache{typeCacheDatabase, typeCacheCRS, typeCacheTypes} =- modifyMVarMasked typeCacheTypes \cached -> do- types <- maybe (withTypeConnection typeCacheDatabase (`readNativeTypes` typeCacheCRS)) pure cached- pure (Just types, types)---- | Read the VARIANT type and a GEOMETRY type for each CRS with one query.-readNativeTypes :: DuckDBConnection -> [Text] -> IO NativeTypes-readNativeTypes connection crss = do- let sql = Text.concat ("SELECT NULL::VARIANT" : [", system.main.ST_SetCRS('POINT EMPTY'::GEOMETRY, ?)" | _ <- crss])- withTypeQuery connection sql crss \result -> mask_ do- owned <- columnTypes result (length crss + 1)- case owned of- [] -> throwIO (userError "duckdb-simple: the type query returned no columns")- variant : geometry -> do- reported <- mapM logicalTypeToRep geometry `onException` mapM_ destroyLogicalType owned- pure- NativeTypes- { nativeVariant = variant- , nativeGeometry =- Map.fromList- ( concat- [ (crs, logical) : [(reportedCRS, logical) | LogicalTypeGeometry (Just reportedCRS) <- [rep]]- | (crs, rep, logical) <- zip3 crss reported geometry- ]- )- , nativeOwned = owned- }---- | Take the types of the first columns. Destroy the taken types on failure.-columnTypes :: Ptr DuckDBResult -> Int -> IO [DuckDBLogicalType]-columnTypes result count = go [] 0- where- go taken column- | column == count = pure (reverse taken)- | otherwise = do- logical <- c_duckdb_column_logical_type result (fromIntegral column) `onException` mapM_ destroyLogicalType taken- when (logical == nullPtr) do- mapM_ destroyLogicalType taken- throwIO (userError "duckdb-simple: the type query returned no type")- go (logical : taken) (column + 1)---- | Copy a type through a LIST type, because the C API has no copy function.-copyLogicalType :: DuckDBLogicalType -> IO DuckDBLogicalType-copyLogicalType logical =- -- TODO: use a copy function when the C API has one.- -- See https://github.com/Tritlo/duckdb-haskell/issues/30 and- -- https://github.com/duckdb/duckdb/issues/26664.- bracket (c_duckdb_create_list_type logical) destroyLogicalType c_duckdb_list_type_child_type---- | Run an action with a new connection to the database. Disconnect after it.-withTypeConnection :: DuckDBDatabase -> (DuckDBConnection -> IO a) -> IO a-withTypeConnection database action =- alloca \connectionPtr -> do- connected <- c_duckdb_connect database connectionPtr- when (connected /= DuckDBSuccess) $- throwIO (userError "duckdb-simple: cannot connect to read the VARIANT and GEOMETRY types")- (peek connectionPtr >>= action) `finally` c_duckdb_disconnect connectionPtr---- | Prepare and run a constant query with text parameters, and borrow its result.-withTypeQuery :: DuckDBConnection -> Text -> [Text] -> (Ptr DuckDBResult -> IO a) -> IO a-withTypeQuery connection sql parameters action =- BS.useAsCString (TextEncoding.encodeUtf8 sql) \sqlPtr ->- alloca \statementPtr -> do- poke statementPtr nullPtr- bracket (c_duckdb_prepare connection sqlPtr statementPtr) (const (c_duckdb_destroy_prepare statementPtr)) \prepared -> do- statement <- peek statementPtr- when (prepared /= DuckDBSuccess) do- errorPtr <- c_duckdb_prepare_error statement- message <- if errorPtr == nullPtr then pure "prepare failed" else TextEncoding.decodeUtf8 <$> BS.packCString errorPtr- queryFailed message- forM_ (zip [1 ..] parameters) \(index, parameter) ->- BS.useAsCStringLen (TextEncoding.encodeUtf8 parameter) \(ptr, len) -> do- bound <- c_duckdb_bind_varchar_length statement index ptr (fromIntegral len)- when (bound /= DuckDBSuccess) (queryFailed ("cannot bind CRS " <> parameter))- alloca \result -> do- fillBytes result 0 (sizeOf (undefined :: DuckDBResult))- bracket (c_duckdb_execute_prepared statement result) (const (c_duckdb_destroy_result result)) \executed -> do- when (executed /= DuckDBSuccess) do- errorPtr <- c_duckdb_result_error result- message <- if errorPtr == nullPtr then pure "execution failed" else TextEncoding.decodeUtf8 <$> BS.packCString errorPtr- queryFailed message- action result- where- queryFailed message = throwIO (userError ("duckdb-simple: cannot read the VARIANT and GEOMETRY types: " <> Text.unpack message))
+ src/Database/DuckDB/Simple/TypeScope.hs view
@@ -0,0 +1,99 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE OverloadedStrings #-}++-- |+-- Module : Database.DuckDB.Simple.TypeScope+-- Description : Scoped ownership of native field types.+module Database.DuckDB.Simple.TypeScope (+ TypeScope,+ newTypeScope,+ closeTypeScope,+ withTypeScope,+ scopedTypeRep,+) where++import Control.Concurrent.MVar (MVar, modifyMVarMasked, newMVar, putMVar, takeMVar, withMVar)+import Control.Exception (bracket, bracketOnError, evaluate, finally, throwIO, uninterruptibleMask_)+import Data.IORef (IORef, newIORef, readIORef, writeIORef)+import Database.DuckDB.Simple.Conversion (Conversion, liftConversion)+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep)+import Database.DuckDB.Simple.Native (DuckDBV2FactoryHandle, DuckDBV2LogicalTypeHandle, SQLError (..), destroyLogicalType, releaseItems)+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeToRep)++-- | The owner of native field types for one result or callback.+-- See Note [Scoped field types].+newtype TypeScope = TypeScope (MVar (Maybe [(DuckDBV2LogicalTypeHandle, IORef (Either (IO DuckDBV2FactoryHandle) LogicalTypeRep))]))+ deriving (Eq)++{- Note [Scoped field types]+~~~~~~~~~~~~~~~~~~~~~~~~~~~+A scope owns each native type and its optional Haskell snapshot. The MVar+serializes acquisition, type reads, and closure. A closed scope rejects each+read, including a cache hit. A snapshot already returned to the caller is pure+and remains valid after closure.++Only internal native operations run while the scope is locked. The type and+factory acquisition actions must not run user code. An uncached read needs a+live factory owner. A result reads the factory from the connection at that+point, so a decoder that closed the connection gets an error. An extension+aggregate can initialize its registration-owned helper at that point.+logicalTypeToRep reads type ids, names, and parameters, and compares temporary+alias copies. These operations do not run SQL or callbacks. A failed or+interrupted read leaves the native type available for another attempt. Only+a successful read fills the snapshot cache. Until then, the cache holds the+factory acquisition action. Closure drops that action too, so an escaped field+cannot retain the connection that it used for the read.++Closure waits for an active metadata operation, then destroys every native+type and clears every snapshot cache. Cleanup uses only internal metadata+operations. It is uninterruptible so cancellation cannot leave an owned type+behind. No user decoder or action runs under this mask. The scope stays locked+until cleanup ends, and becomes closed even if a destructor raises an+exception. Thus a concurrent reader or closer cannot use or destroy a released+handle.+-}++-- | Start a scope with no native types.+newTypeScope :: IO TypeScope+newTypeScope = TypeScope <$> newMVar (Just [])++-- | Release each native type and reject future type reads. Repeated calls are safe.+-- See Note [Scoped field types].+closeTypeScope :: TypeScope -> IO ()+closeTypeScope (TypeScope scope) = uninterruptibleMask_ do+ owned <- takeMVar scope+ flip finally (putMVar scope Nothing) $+ case owned of+ Nothing -> pure ()+ Just entries -> releaseItems release entries+ where+ release (native, snapshot) = destroyLogicalType native `finally` writeIORef snapshot (Left closedTypeScope)++-- | Run an action in a type scope and close it on every exit path.+withTypeScope :: (TypeScope -> IO a) -> IO a+withTypeScope = bracket newTypeScope closeTypeScope++-- | Acquire an owned native type and read its Haskell snapshot on demand.+-- The acquisition action must use only native metadata operations.+-- See Note [Scoped field types].+scopedTypeRep :: TypeScope -> IO DuckDBV2FactoryHandle -> IO DuckDBV2LogicalTypeHandle -> IO (Conversion LogicalTypeRep)+scopedTypeRep (TypeScope scope) acquireFactory create = modifyMVarMasked scope \case+ Nothing -> closedTypeScope+ Just entries -> bracketOnError create destroyLogicalType \native -> do+ snapshot <- newIORef (Left acquireFactory)+ let readType = withMVar scope \case+ Nothing -> closedTypeScope+ Just _ ->+ readIORef snapshot >>= \case+ Right logical -> pure logical+ Left acquire -> do+ factory <- acquire+ logical <- logicalTypeToRep factory native >>= evaluate+ writeIORef snapshot (Right logical)+ pure logical+ pure (Just ((native, snapshot) : entries), liftConversion readType)++-- | Reject an operation after its metadata owner has closed.+closedTypeScope :: IO a+closedTypeScope = throwIO (SQLError "duckdb-simple: field type scope is closed" Nothing Nothing Nothing)
+ src/Database/DuckDB/Simple/Typed.hs view
@@ -0,0 +1,249 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DefaultSignatures #-}+{-# LANGUAGE FlexibleContexts #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}+{-# LANGUAGE TypeOperators #-}++-- |+-- Module : Database.DuckDB.Simple.Typed+-- Description : Read and write callback vectors as typed Haskell values.+--+-- Typed registrations use these helpers. An argument reader gives the rows of a+-- borrowed vector as fields with the names from 'argumentName'.+-- 'conversionMessage' describes the arguments that do not convert. 'TypedRow'+-- gives the column types and the column values of a row. The module also+-- converts a typed replacement function to a native replacement scan callback.+module Database.DuckDB.Simple.Typed (+ TypedRow (..),+ ColumnValue (..),+ argumentName,+ rowDeclarations,+ positionalSignature,+ readBoundField,+ ArgumentReader,+ newArgumentReader,+ conversionMessage,+ Replacement (..),+ replacementScanCallback,+) where++import Control.Concurrent.MVar (modifyMVarMasked, newMVar)+import Control.Exception (SomeException, bracket, bracketOnError, finally, fromException, mask_)+import Control.Monad (forM, forM_)+import Data.List (intercalate)+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple.Callback (RegistrationTarget, targetCastFactory)+import Database.DuckDB.Simple.FromField (Field (..), ResultError (..))+import Database.DuckDB.Simple.Materialize (+ FallbackReaders (..),+ ReadFactory (InstanceFactory),+ createHelperInstance,+ materializeValue,+ prepareGeometryReader,+ prepareVectorReader,+ readCallbackVariant,+ readFactoryHandle,+ readInstanceVariant,+ )+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.ToField (DuckDBColumnType (..), ToDuckValue (..))+import Database.DuckDB.Simple.TypeScope (TypeScope, scopedTypeRep)+import Database.DuckDB.Simple.Types (Only (..))+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeToRep)+import Database.DuckDB.Simple.V2.Registration (+ ContextCallback,+ FunctionParameter (parameterKind),+ FunctionSignature (..),+ TypeDeclaration,+ parameter,+ )+import GHC.Generics++-- | A row with a fixed list of typed columns. A typed registration uses the+-- column types in the signature of the SQL object, and writes the column values+-- to DuckDB. t'Only', tuples of 2 to 7 columns, and @()@ are rows. A record with a+-- 'Generic' instance is a row when each field has a 'ToDuckValue' instance:+--+-- > data Point = Point {x :: Double, y :: Double} deriving (Generic)+-- > instance TypedRow Point+--+-- A @Maybe@ column can hold NULL. A row of arguments has the same rule as a row+-- of results: each column type needs a 'ToDuckValue' instance, because that+-- instance gives the SQL type of the column.+class TypedRow r where+ -- | The declaration of each column, from 'duckdbTypeDeclaration'. See+ -- Note [Type declarations] in "Database.DuckDB.Simple.ToField".+ rowColumnDeclarations :: Proxy r -> [Either Text TypeDeclaration]+ default rowColumnDeclarations :: (GTypedRow (Rep r)) => Proxy r -> [Either Text TypeDeclaration]+ rowColumnDeclarations _ = gRowColumnDeclarations (Proxy @(Rep r))++ -- | The value of each column.+ rowColumnValues :: r -> [ColumnValue]+ default rowColumnValues :: (Generic r, GTypedRow (Rep r)) => r -> [ColumnValue]+ rowColumnValues = gRowColumnValues . from++-- | The value of one column of a row. The action makes an owned native value+-- through a borrowed factory. The caller destroys the value.+newtype ColumnValue = ColumnValue (DuckDBV2FactoryHandle -> IO DuckDBV2ValueHandle)++-- | The generic implementation of 'TypedRow' for a type with one constructor.+class GTypedRow f where+ gRowColumnDeclarations :: Proxy f -> [Either Text TypeDeclaration]+ gRowColumnValues :: f p -> [ColumnValue]++instance GTypedRow U1 where+ gRowColumnDeclarations _ = []+ gRowColumnValues U1 = []++instance (GTypedRow a) => GTypedRow (M1 i c a) where+ gRowColumnDeclarations _ = gRowColumnDeclarations (Proxy @a)+ gRowColumnValues (M1 value) = gRowColumnValues value++instance (ToDuckValue a) => GTypedRow (K1 i a) where+ gRowColumnDeclarations _ = [duckdbTypeDeclaration (Proxy @a)]+ gRowColumnValues (K1 value) = [ColumnValue (`toDuckValue` value)]++instance (GTypedRow a, GTypedRow b) => GTypedRow (a :*: b) where+ gRowColumnDeclarations _ = gRowColumnDeclarations (Proxy @a) <> gRowColumnDeclarations (Proxy @b)+ gRowColumnValues (left :*: right) = gRowColumnValues left <> gRowColumnValues right++instance TypedRow ()++instance (ToDuckValue a) => TypedRow (Only a) where+ rowColumnDeclarations _ = [duckdbTypeDeclaration (Proxy @a)]+ rowColumnValues (Only value) = [ColumnValue (`toDuckValue` value)]++instance (ToDuckValue a, ToDuckValue b) => TypedRow (a, b)++instance (ToDuckValue a, ToDuckValue b, ToDuckValue c) => TypedRow (a, b, c)++instance (ToDuckValue a, ToDuckValue b, ToDuckValue c, ToDuckValue d) => TypedRow (a, b, c, d)++instance (ToDuckValue a, ToDuckValue b, ToDuckValue c, ToDuckValue d, ToDuckValue e) => TypedRow (a, b, c, d, e)++instance (ToDuckValue a, ToDuckValue b, ToDuckValue c, ToDuckValue d, ToDuckValue e, ToDuckValue f) => TypedRow (a, b, c, d, e, f)++instance (ToDuckValue a, ToDuckValue b, ToDuckValue c, ToDuckValue d, ToDuckValue e, ToDuckValue f, ToDuckValue g) => TypedRow (a, b, c, d, e, f, g)++-- | The name of the argument at an index, from 0: @arg0@, @arg1@, and so on.+argumentName :: Int -> Text+argumentName index = Text.pack ("arg" <> show index)++-- | Give the declaration of each column of a row, or raise the first 'Left' as+-- an @SQLError@ before a registration. The error names the registration and+-- the place of the column. The list of places has a name for each column. See+-- Note [Type declarations] in "Database.DuckDB.Simple.ToField".+rowDeclarations :: Text -> [Text] -> [Either Text TypeDeclaration] -> IO [TypeDeclaration]+rowDeclarations name places declarations =+ forM (zip places declarations) \(place, declaration) ->+ either (\reason -> throwInvalidArgument (Text.intercalate (Text.pack ": ") [name, place, reason])) pure declaration++-- | Declare the positional arguments of a typed function. A table function+-- declares its result columns during bind and supplies no return type here.+positionalSignature :: [TypeDeclaration] -> Maybe TypeDeclaration -> FunctionSignature+positionalSignature types =+ FunctionSignature+ [ (parameter (argumentName index) dtype){parameterKind = DuckDBV2FunctionParameterKindPositionalOnly}+ | (index, dtype) <- zip [0 ..] types+ ]++-- | Copy one argument of a table function or pragma. Acquire its logical+-- type within the value's lifetime. The type scope owns it during conversion.+readBoundField :: TypeScope -> ReadFactory -> DuckDBV2FunctionBindInfoHandle -> Text -> DuckDBIdx -> IO Field+readBoundField scope factory info name index =+ bracket (output (c_duckdb_v2_function_bind_get_arg_value info index)) destroyValue \value ->+ Field name (fromIntegral index)+ <$> materializeValue readCallbackVariant factory value+ <*> scopedTypeRep scope (pure (readFactoryHandle factory)) (output (c_duckdb_v2_value_get_logical_type value))++{- Note [Function argument readers]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+A scalar or aggregate function belongs to the catalog. Its callbacks can+run after the registering connection closes. An extension has no managed+connection, and an aggregate update has no callback context.++A registration takes the host instance factory when available. An extension+opens a private instance on the first argument read or aggregate result+write that needs a factory. GEOMETRY inputs use its built-in cast to BLOB.+A VARIANT read always needs the private instance because its metadata query+can run while the caller holds a host instance lock. See Note [VARIANT reads inside callbacks]+in "Database.DuckDB.Simple.Materialize".++An MVar protects instance creation. Each VARIANT vector gets its own+connection and decoder state, so concurrent calls can decode independently.+The native callback owner releases the instance and its environment after+the last bound or executing function releases the callbacks.+-}++-- | Prepare a vector's fields within the current callback's type scope.+type ArgumentReader = TypeScope -> Int -> DuckDBV2VectorHandle -> IO (Int -> IO Field)++-- | Create the argument reader, the action to borrow its instance factory,+-- and the cleanup for one registration.+-- Native resources are acquired on demand. See Note [Function argument readers].+newArgumentReader :: RegistrationTarget -> IO (ArgumentReader, IO DuckDBV2FactoryHandle, IO ())+newArgumentReader target = do+ hostFactory <- targetCastFactory target+ cache <- newMVar Nothing+ let helper = modifyMVarMasked cache \case+ Just owned@(_, instance_) -> pure (Just owned, instance_)+ Nothing -> bracketOnError (output c_duckdb_v2_environment_create) (destroy c_duckdb_v2_environment_destroy) \env -> do+ instance_ <- createHelperInstance env+ pure (Just (env, instance_), instance_)+ close = mask_ do+ owned <- modifyMVarMasked cache (\state -> pure (Nothing, state))+ forM_ owned \(env, instance_) ->+ destroy c_duckdb_v2_instance_destroy instance_ `finally` destroy c_duckdb_v2_environment_destroy env+ factory = maybe (helper >>= output . c_duckdb_v2_instance_get_factory) pure hostFactory+ readers =+ FallbackReaders+ (\logical -> factory >>= \f -> logicalTypeToRep f logical)+ (\vector -> helper >>= \instance_ -> readInstanceVariant instance_ vector)+ (\vector -> factory >>= \f -> prepareGeometryReader (InstanceFactory f) vector)+ readArgument scope index vector = do+ readValue <- prepareVectorReader readers vector+ -- See Note [Scoped field types] in "Database.DuckDB.Simple.TypeScope".+ argumentType <- scopedTypeRep scope factory (output (c_duckdb_v2_vector_get_logical_type vector))+ pure \row -> (\value -> Field (argumentName index) index value argumentType) <$> readValue row+ pure (readArgument, factory, close)++-- | Describe the arguments that 'Database.DuckDB.Simple.FromField.FromField' or+-- @FromRow@ cannot convert. The text names each argument with the name of its+-- field.+conversionMessage :: [SomeException] -> String+conversionMessage failures = intercalate "; " (map describe failures)+ where+ describe failure = case fromException failure of+ Just result@UnexpectedNull{} -> "argument " <> Text.unpack (errSQLField result) <> " is NULL, but " <> Text.unpack (errHaskellType result) <> " cannot hold NULL"+ Just result ->+ "cannot convert argument "+ <> Text.unpack (errSQLField result)+ <> " from "+ <> Text.unpack (errSQLType result)+ <> " to "+ <> Text.unpack (errHaskellType result)+ <> if Text.null (errMessage result) then "" else ": " <> Text.unpack (errMessage result)+ Nothing -> exceptionText failure++-- | What DuckDB reads in place of an unknown table. DuckDB reads the result+-- under the name of the table, unless the query gives an alias.+data Replacement+ = -- | Read the rows of a query. The text must contain exactly one SELECT+ -- statement. Other text fails the query with an SQL error.+ ReplaceWithQuery Text+ deriving (Eq, Show, Read)++-- | Convert a typed replacement function to a native replacement scan+-- callback. The callback reads the parts of the unknown name and gives them to the+-- function. 'Nothing' declines the name.+replacementScanCallback :: ([Text] -> IO (Maybe Replacement)) -> ContextCallback DuckDBV2ReplacementScanInfoHandle+replacementScanCallback replace info _ = do+ parts <- bracket (output (c_duckdb_v2_replacement_scan_get_name info)) (destroy c_duckdb_v2_qname_destroy) \name -> do+ count <- output (c_duckdb_v2_qname_get_part_count name)+ forM [0 .. fromIntegral count - 1 :: Int] \index -> output (c_duckdb_v2_qname_get_part name (fromIntegral index)) >>= readIdentifier+ replace parts >>= mapM_ \case+ ReplaceWithQuery sql -> withText sql (checked . c_duckdb_v2_replacement_scan_set_subquery info)
+ src/Database/DuckDB/Simple/Typed/Aggregate.hs view
@@ -0,0 +1,402 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE ForeignFunctionInterface #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : Database.DuckDB.Simple.Typed.Aggregate+-- Description : Register an aggregate function as a fold over Haskell values.+--+-- The Haskell state of a group lives outside DuckDB. The native state holds a+-- 'StablePtr' to an 'IORef' with the Haskell state. The bind data of each call+-- site holds a state table with the live stable pointers. See+-- Note [Aggregate state life cycle].+module Database.DuckDB.Simple.Typed.Aggregate (+ Aggregate (..),+ createAggregateOn,+) where++import Control.Exception (evaluate, finally, mask_, onException, throwIO)+import Control.Monad (forM_, unless, when)+import Data.Array (Array, elems, listArray, (!))+import Data.IORef (IORef, atomicModifyIORef', newIORef, readIORef, writeIORef)+import qualified Data.IntSet as IntSet+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Void (Void)+import Database.DuckDB.Simple.Callback (attachOpaque, ignoreCallbackExceptions)+import Database.DuckDB.Simple.Conversion (runConversion)+import Database.DuckDB.Simple.FromField (Field (..), FieldValue (FieldNull))+import Database.DuckDB.Simple.FromRow (FromRow (..), parseRow)+import Database.DuckDB.Simple.Function.Internal (FunctionResult, prepareAggregateResultWriter)+import Database.DuckDB.Simple.Internal (Connection (..), ConnectionOptions (..), defaultAggregateStripes)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.Ok (Ok (..))+import Database.DuckDB.Simple.ToField (DuckDBColumnType (..))+import Database.DuckDB.Simple.TypeScope (withTypeScope)+import Database.DuckDB.Simple.Typed (+ ArgumentReader,+ TypedRow (..),+ argumentName,+ conversionMessage,+ newArgumentReader,+ positionalSignature,+ rowDeclarations,+ )+import Database.DuckDB.Simple.V2.Registration (+ AggregateCallbacks (..),+ BindCallback,+ RegistrationTarget (..),+ aggregateCallbacks,+ configureFunctionSignature,+ registerAggregateFunctionWithCleanup,+ withRegistrationFactory,+ )+import Foreign.Ptr (FunPtr, Ptr, castFunPtr, castPtr, intPtrToPtr, nullPtr, ptrToIntPtr)+import Foreign.StablePtr (StablePtr, castPtrToStablePtr, castStablePtrToPtr, deRefStablePtr, freeStablePtr, newStablePtr)+import Foreign.Storable (peek, peekElemOff, poke)+import HsBindgen.Runtime.Marshal (staticSizeOf)++-- | An aggregate function as a fold. DuckDB keeps one state for each group, or+-- more than one partial state.+data Aggregate state args result = Aggregate+ { aggregateInitial :: state+ -- ^ The state of a group with no rows.+ , aggregateStep :: state -> args -> IO state+ -- ^ Add one row to a state.+ , aggregateCombine :: state -> state -> IO state+ -- ^ Merge the second partial state of a group into the first.+ , aggregateResult :: state -> IO result+ -- ^ Give the result of a group.+ }++{- Note [Aggregate state life cycle]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+The Haskell state of a group lives outside DuckDB. The native state of a group+has the size of one pointer (setStateSize). It holds a StablePtr to an IORef,+and the IORef holds the Haskell state.++These are the steps in the life of a state.++- Init (initializeStates). The callback first writes null to all native states+ of the call, so that an error leaves no native state uninitialized. Then for+ each state it makes the IORef and the StablePtr, adds the key of the+ StablePtr to the state table, and writes the StablePtr to the native state.++- Update and combine (updateStates and combineStates). DuckDB can move a native+ state with a byte copy. Thus these callbacks change only the IORef.++- Release (releaseState). After init, only a release changes the native state:+ it writes null to the native state. A release does not change a null native+ state. Thus the null write stops a second release of one native state. A read+ of a released native state raises an error (stateCell).++The bind data of each call site holds a state table with the keys of the live+stable pointers (StateTable and bindStateTable). A release removes the key from+the table, and only the caller that removes it frees the stable pointer. This+check stops a second free of a stable pointer that a byte copy of the native+state also holds. It does not stop the release of a byte copy after the runtime+has given its stable pointer to a new state. DuckDB 2.0.0 destroys each state at+one location. The destructor of the state table frees the stable pointers that+remain (releaseStateTable).++The registration always sets a bind callback, so each call site has bind data.+The code still reads the table as optional (readStateTable). Without bind data,+only the destroy callback and a failed finalize release the states, and a+release frees each stable pointer that it reads.++These are the states that DuckDB 2.0.0 destroys and does not destroy.++- The header says that the destroy callback runs on states that are discarded+ without being finalized. DuckDB 2.0.0 runs the destroy callback on finalized+ states too: a hash aggregate destroys each chunk of states after it finalizes+ them, and an ungrouped aggregate and a window destroy their states after+ finalize (radix_partitioned_hashtable.cpp, physical_ungrouped_aggregate.cpp,+ and window_segment_tree.cpp).++- DuckDB 2.0.0 does not run the destroy callback on these states:++ - The states of a finalize call that fails, in a hash aggregate and in a+ window.+ - The states that no finalize call gets when LIMIT stops a parallel hash+ aggregate.+ - The states after a combine call fails in a parallel hash aggregate.+ - The states after a finalize call fails in a parallel hash aggregate.++ The exception handler of finalizeStates releases the states of the finalize+ call that fails. The destructor of the state table frees the stable pointers+ of the other states.++DuckDB runs the destructor of the state table when it drops the bind data of a+call site. In DuckDB 2.0.0 a hash table and an ungrouped aggregate each hold a+copy of the bind data, and each copy shares the holder of the state table+through a shared pointer (CV2AggregateFunctionData::Copy and CV2UserData). Each+of them runs the destroy callback on its states before it drops its copy+(~GroupedAggregateHashTable and ~UngroupedAggregateState). In a window, the+same is true for WindowAggregateStates and WindowCustomAggregatorLocalState.+The naive aggregator and the segment tree use the copy of the window global+state, and run the destroy callback during the evaluation. Thus the destructor+runs after the last destroy callback of these owners that gets this bind data.++See Note [State table stripes] for the form of the table.+-}++{- Note [State table stripes]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+A state table holds the keys of the live states of one call site. Each new+state adds its key, and each release removes it. DuckDB makes and releases+states on several threads at the same time.++A stripe is one part of the table: an IORef with a set of keys. The table is an+array of stripes, and the key of a state selects its stripe (stateStripe). Thus+two threads change the same IORef only when their keys select the same stripe.++- Why the table has stripes. A change uses atomicModifyIORef', which stores a+ value that is not evaluated. If the runtime stops a thread during that+ evaluation, each thread that needs the value waits. With one IORef for all+ states, the DuckDB threads of a query with many groups waited for each other.++- The number of stripes. The option connectionAggregateStripes of+ ConnectionOptions sets the number for each aggregate that createAggregate+ registers through the connection. The registration keeps the number, also+ when another connection calls the aggregate. The default is+ defaultAggregateStripes, 64 stripes. A registration on an extension has no+ connection options, so it uses the default. The bind callback makes the+ table with this number (bindStateTable).++- No result depends on the number. It changes only how often two threads use+ the same stripe. A stripe costs one small IORef for each call site.++The destructor of the table reads all stripes (releaseStateTable).+-}++-- | The live states of one call site, as the keys of their stable pointers.+-- 'stateStripe' gives the stripe of a key. See Note [State table stripes].+--+-- With the table, the library frees the states that DuckDB 2.0.0 does not+-- destroy. See Note [Aggregate state life cycle].+type StateTable = Array Int (IORef IntSet.IntSet)++-- | The stripe that holds a key. In GHC 9.14 the key of a stable pointer is+-- one more than its index in the stable pointer table of the runtime+-- (@rts\/include\/rts\/StablePtr.h@). Thus consecutive keys go to different+-- stripes when the table has more than one stripe. The table is correct for+-- each distribution of the keys. The distribution changes only how often two+-- threads use the same stripe. See Note [State table stripes].+stateStripe :: StateTable -> Int -> IORef IntSet.IntSet+stateStripe table key = table ! (key `mod` length table)++-- | Register a typed aggregate function on a connection or a loading extension.+-- Check the name and declarations before accessing the registration target.+-- See Note [Type declarations] in "Database.DuckDB.Simple.ToField" and+-- Note [Result writers] in "Database.DuckDB.Simple.Function.Internal".+createAggregateOn+ :: forall state args result+ . (FromRow args, TypedRow args, FunctionResult result) => RegistrationTarget -> Text -> Aggregate state args result -> IO ()+createAggregateOn target name aggregate = do+ when (Text.null name || Text.any (== '\0') name) $+ throwInvalidArgument "aggregate function name must be nonempty and must not contain NUL"+ argumentTypes <- rowDeclarations name ["argument " <> argumentName index | index <- [0 ..]] columnTypes+ [resultType] <- rowDeclarations name ["result"] [duckdbTypeDeclaration (Proxy @result)]+ (readArgument, acquireFactory, closeReader) <- newArgumentReader target+ -- See Note [State table stripes]. Evaluate the count before the registration.+ -- An unevaluated count keeps the target, and thus the connection, reachable+ -- from the callbacks.+ stripes <- evaluate case target of+ ConnectionTarget connection -> connectionAggregateStripes (connectionOptions connection)+ ExtensionTarget{} -> defaultAggregateStripes+ let signature = positionalSignature argumentTypes (Just resultType)+ callbacks =+ ( aggregateCallbacks+ setStateSize+ (initializeStates aggregate)+ (updateStates readArgument name aggregate (length columnTypes))+ (combineStates aggregate)+ (finalizeStates aggregate acquireFactory)+ )+ { aggregateBindCallback = Just (bindStateTable stripes)+ , aggregateDestroyCallback = Just destroyStates+ }+ withRegistrationFactory target \factory ->+ registerAggregateFunctionWithCleanup target closeReader (configure signature factory) callbacks+ where+ columnTypes = rowColumnDeclarations (Proxy @args)+ configure signature factory function = do+ withIdentifier name (checked . c_duckdb_v2_aggregate_function_set_name function)+ output (c_duckdb_v2_aggregate_function_get_signature function) >>= configureFunctionSignature factory signature+ checked+ (c_duckdb_v2_aggregate_function_set_property function DuckDBV2FunctionPropertyNullHandling DuckDBV2FunctionPropertyNullHandlingSpecial)++-- | Make a native state the size of one pointer. See+-- Note [Aggregate state life cycle].+setStateSize :: DuckDBV2AggregateFunctionSizeInfoHandle -> IO ()+setStateSize info = checked (c_duckdb_v2_aggregate_function_size_set_state_size info (fromIntegral (staticSizeOf (Proxy @(Ptr ())))))++-- | Make an empty state table with the given number of stripes and transfer it+-- to the bind data of the call site. See Note [Aggregate state life cycle] and+-- Note [State table stripes].+bindStateTable :: Int -> BindCallback DuckDBV2AggregateFunctionBindInfoHandle+bindStateTable count input _ _ = mask_ do+ stripes <- mapM (const (newIORef IntSet.empty)) [1 .. count]+ -- Build the array here, so that no callback thread evaluates it.+ table <- evaluate (listArray (0, count - 1) stripes :: StateTable)+ stable <- newStablePtr table+ attachOpaque (c_duckdb_v2_function_bind_set_bind_data input) (castPtr (castStablePtrToPtr stable)) stateTableDestructor+ `onException` freeStablePtr stable++-- | Read the state table from the bind data. Without bind data, give 'Nothing'.+-- See Note [Aggregate state life cycle].+readStateTable :: (Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO (Maybe StateTable)+readStateTable getter = do+ raw <- output getter+ if raw == nullPtr then pure Nothing else Just <$> deRefStablePtr (castPtrToStablePtr (castPtr raw))++-- | Give each new native state the initial Haskell state. First set all native+-- states to null, so that an error leaves no native state uninitialized. See+-- Note [Aggregate state life cycle].+initializeStates :: Aggregate state args result -> DuckDBV2AggregateFunctionInitInfoHandle -> IO ()+initializeStates Aggregate{aggregateInitial} info = do+ table <- readStateTable (c_duckdb_v2_aggregate_function_init_get_bind_data info)+ count <- fromIntegral <$> output (c_duckdb_v2_aggregate_function_init_get_state_count info)+ states <- output (c_duckdb_v2_aggregate_function_init_get_states info)+ forM_ [0 .. count - 1] \index -> peekElemOff states index >>= \slot -> poke (castPtr slot) (nullPtr @())+ forM_ [0 .. count - 1] \index -> do+ slot <- peekElemOff states index+ mask_ do+ stable <- newIORef aggregateInitial >>= newStablePtr+ let key = stableKey stable+ forM_ table \keys -> atomicModifyIORef' (stateStripe keys key) \live -> (IntSet.insert key live, ())+ poke (castPtr slot) stable++-- | Add each row to its state. A row with NULL in an argument does not change+-- its state. A row that does not convert raises an error that names the+-- aggregate function. A GEOMETRY argument reads with the instance factory.+-- See Note [Aggregate state life cycle].+updateStates+ :: (FromRow args)+ => ArgumentReader+ -> Text+ -> Aggregate state args result+ -> Int+ -> DuckDBV2AggregateFunctionUpdateInfoHandle+ -> IO ()+updateStates readArgument name Aggregate{aggregateStep} columns info = withTypeScope \scope -> do+ count <- fromIntegral <$> output (c_duckdb_v2_aggregate_function_update_get_row_count info)+ states <- output (c_duckdb_v2_aggregate_function_update_get_states info)+ readers <-+ mapM+ ( \index ->+ output (c_duckdb_v2_aggregate_function_update_get_arg info (fromIntegral index))+ >>= readArgument scope index+ )+ [0 .. columns - 1]+ forM_ [0 .. count - 1] \row -> do+ fields <- mapM ($ row) readers+ unless (any ((== FieldNull) . fieldValue) fields) do+ arguments <-+ runConversion (parseRow fromRow fields) >>= \case+ Ok arguments -> pure arguments+ Errors failures -> throwIO (userError ("duckdb-simple: " <> Text.unpack name <> ": " <> conversionMessage failures))+ cell <- peekElemOff states row >>= stateCell+ readIORef cell >>= (`aggregateStep` arguments) >>= evaluate >>= writeIORef cell++-- | Merge each source state into its target state. Do not change the source.+-- Change only the cell of the target. See Note [Aggregate state life cycle].+combineStates :: Aggregate state args result -> DuckDBV2AggregateFunctionCombineInfoHandle -> IO ()+combineStates Aggregate{aggregateCombine} info = do+ count <- fromIntegral <$> output (c_duckdb_v2_aggregate_function_combine_get_state_count info)+ sources <- output (c_duckdb_v2_aggregate_function_combine_get_sources info)+ targets <- output (c_duckdb_v2_aggregate_function_combine_get_targets info)+ forM_ [0 .. count - 1] \index -> do+ source <- peekElemOff sources index >>= stateCell >>= readIORef+ target <- peekElemOff targets index >>= stateCell+ current <- readIORef target+ aggregateCombine current source >>= evaluate >>= writeIORef target++-- | Write each state at its offset in the result vector. DuckDB does not+-- destroy the states when finalize fails. See Note [Aggregate state life cycle].+finalizeStates+ :: (FunctionResult result) => Aggregate state args result -> IO DuckDBV2FactoryHandle -> DuckDBV2AggregateFunctionFinalizeInfoHandle -> IO ()+finalizeStates Aggregate{aggregateResult} acquireFactory info = do+ table <- readStateTable (c_duckdb_v2_aggregate_function_finalize_get_bind_data info)+ count <- output (c_duckdb_v2_aggregate_function_finalize_get_state_count info)+ states <- output (c_duckdb_v2_aggregate_function_finalize_get_states info)+ let rows = [0 .. fromIntegral count - 1]+ finalize = when (count /= 0) do+ offset <- output (c_duckdb_v2_aggregate_function_finalize_get_result_offset info)+ vector <- output (c_duckdb_v2_aggregate_function_finalize_get_result info)+ write <- prepareAggregateResultWriter acquireFactory vector (offset + count)+ forM_ rows \index ->+ peekElemOff states index >>= stateCell >>= readIORef >>= aggregateResult >>= write (fromIntegral offset + index)+ finalize `onException` forM_ rows \index -> peekElemOff states index >>= releaseState table++-- | Release each state. See Note [Aggregate state life cycle].+destroyStates :: DuckDBV2AggregateFunctionDestroyInfoHandle -> IO ()+destroyStates info = do+ table <- readStateTable (c_duckdb_v2_aggregate_function_destroy_get_bind_data info)+ count <- fromIntegral <$> output (c_duckdb_v2_aggregate_function_destroy_get_state_count info)+ states <- output (c_duckdb_v2_aggregate_function_destroy_get_states info)+ forM_ [0 .. count - 1 :: Int] \index -> peekElemOff states index >>= releaseState table++-- | Read the cell of a native state. A released native state raises an error.+-- See Note [Aggregate state life cycle].+stateCell :: Ptr Void -> IO (IORef state)+stateCell slot = do+ stable <- peek (castPtr slot)+ when (castStablePtrToPtr stable == nullPtr) $+ throwInternalError "the aggregate state was released"+ deRefStablePtr stable++-- | Release the Haskell state of a native state, and set the native state to+-- null. A null native state does not change. With a table, only the caller that+-- removes the key from the table frees the stable pointer. See+-- Note [Aggregate state life cycle].+releaseState :: Maybe StateTable -> Ptr Void -> IO ()+releaseState table slot = mask_ do+ stable <- peek (castPtr slot) :: IO (StablePtr ())+ unless (castStablePtrToPtr stable == nullPtr) do+ poke (castPtr slot) (nullPtr @())+ let key = stableKey stable+ owner <- case table of+ Nothing -> pure True+ Just keys -> atomicModifyIORef' (stateStripe keys key) \live -> (IntSet.delete key live, IntSet.member key live)+ when owner (freeStablePtr stable)++-- | The key of a stable pointer in a state table, from 'castStablePtrToPtr'+-- and 'ptrToIntPtr' of base. Base has no single function for the two steps.+-- 'stableFromKey' is the inverse.+stableKey :: StablePtr a -> Int+stableKey = fromIntegral . ptrToIntPtr . castStablePtrToPtr++-- | The stable pointer of a key in a state table, from 'intPtrToPtr' and+-- 'castPtrToStablePtr' of base. This is the inverse of 'stableKey'.+stableFromKey :: Int -> StablePtr a+stableFromKey = castPtrToStablePtr . intPtrToPtr . fromIntegral++-- | Free the remaining group states and the state table owner.+-- DuckDB calls this when it drops the bind data of a call site. See+-- Note [Aggregate state life cycle], and Note [Callback ownership] in+-- "Database.DuckDB.Simple.Callback" for the rules of a destructor.+releaseStateTable :: Ptr Void -> IO ()+releaseStateTable raw =+ mask_ $ ignoreCallbackExceptions $ unless (raw == nullPtr) do+ let stable = castPtrToStablePtr (castPtr raw) :: StablePtr StateTable+ ( do+ remaining <- deRefStablePtr stable >>= fmap IntSet.unions . mapM readIORef . elems+ forM_ (IntSet.toList remaining) (freeStablePtr . stableFromKey)+ )+ `finally` freeStablePtr stable++foreign export ccall "duckdb_simple_release_aggregate_states"+ releaseStateTable :: Ptr Void -> IO ()++foreign import ccall "&duckdb_simple_release_aggregate_states"+ stateTableDestructorAddress :: FunPtr (Ptr Void -> IO ())++-- | Use the generated callback type for the static state table destructor.+stateTableDestructor :: DuckDBV2OpaqueDestroyFn+stateTableDestructor = DuckDBV2OpaqueDestroyFn (castFunPtr stateTableDestructorAddress)
+ src/Database/DuckDB/Simple/Typed/Table.hs view
@@ -0,0 +1,220 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : Database.DuckDB.Simple.Typed.Table+-- Description : Register typed table functions and write typed rows to output chunks.+--+-- A typed table function converts its arguments with 'FromRow'. It gives the rows+-- of a scan as a Haskell list of 'TypedRow' values. The exec callback writes the+-- list to DuckDB one output chunk at a time. See Note [Row lists of typed scans].+module Database.DuckDB.Simple.Typed.Table (+ createTableFunctionOn,+ writeRows,+) where++import Control.Concurrent.MVar (MVar, modifyMVar_, newMVar)+import Control.Exception (bracket, throwIO)+import Control.Monad (forM, forM_, when, zipWithM_)+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple.Callback (attachOpaque, readCallbackState, targetCastFactory, transferCallbackState)+import Database.DuckDB.Simple.Conversion (runConversion)+import Database.DuckDB.Simple.FromRow (FromRow (..), parseRow)+import Database.DuckDB.Simple.Materialize (ReadFactory, callbackReadFactory)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.Ok (Ok (..))+import Database.DuckDB.Simple.TypeScope (TypeScope, withTypeScope)+import Database.DuckDB.Simple.Typed (+ ColumnValue (..),+ TypedRow (..),+ argumentName,+ conversionMessage,+ positionalSignature,+ readBoundField,+ rowDeclarations,+ )+import Database.DuckDB.Simple.V2.Registration (+ RegistrationTarget,+ TableCallbacks (..),+ configureFunctionSignature,+ registerTableFunctionWith,+ tableCallbacks,+ withRegistrationFactory,+ withTypeDeclaration,+ )++{- Note [Row lists of typed scans]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+A typed table function (createTableFunctionOn) and the COPY FROM side of a+typed format (copyFromSide) give the rows of a scan as a Haskell list of+TypedRow values. The two use the same callbacks.++- The bind callback reads the inputs of the call. Its bind data is the action+ that gives the list.++- The global init callback runs the action, and keeps the list in an MVar in+ the global state. The action runs one time for each scan.++- The exec callback takes the list from the MVar, writes the first rows to the+ output chunk (writeRows), and puts the other rows back. In the exec callback+ of a table function, a call that writes no row ends the scan.++DuckDB can run exec callbacks of one scan at the same time. For a plain table+function, the maximum number of threads of a scan "Defaults to 1, a+single-threaded scan" (duckdb_v2_table_function_init_global_set_max_threads in+the header), and the typed scans do not change it. But a multi-file function+(registerMultiFileFunction) over the table function "claims the batches of the+function itself, so that the rows of a file keep their order also when several+threads scan it" (duckdb_v2_table_function_set_claim_batch_callback in the+header). Then DuckDB 2.0.0 runs several exec callbacks of the scan of one file+at the same time. The lock of the MVar gives each row to one call only.++A typed table function declares projection pushdown. DuckDB then asks the exec+callback for only the columns that the query uses. When the query uses no+column, for example in SELECT count(*), DuckDB 2.0.0 asks for the first column:+the function declares no virtual column, so LogicalGet::GetAnyColumn gives+column 0 (logical_get.cpp), and CV2TableScanColumns gives column 0 for an empty+projection (capi_v2_func_table.cpp). writeRows makes the values of the+requested columns only, so it does not evaluate the other columns of a row. But+it evaluates each row that it writes to its constructor, and this evaluates+each strict field of the row. The COPY FROM side asks writeRows for all+columns.++writeRows evaluates only the spine of the rows that fit in the chunk+(splitRows). It does not evaluate the other rows. The list that it gives back+is the list after the written rows, not a reference to the whole list. DuckDB+takes the rows one chunk at a time, so the list can be lazy. DuckDB stops when+the query needs no more rows, so the list can be long or infinite.+-}++-- | Register a typed table function on a connection or a loading extension.+-- 'Database.DuckDB.Simple.Function.createTableFunction' gives the rules.+--+-- The bind callback converts the arguments and declares the result columns.+-- See Note [Row lists of typed scans]. A parameter type or a column type with+-- no declaration raises an error at registration. See+-- Note [Type declarations] in "Database.DuckDB.Simple.ToField".+createTableFunctionOn+ :: forall params row. (FromRow params, TypedRow params, TypedRow row) => RegistrationTarget -> Text -> [Text] -> (params -> IO [row]) -> IO ()+createTableFunctionOn target name columns function = do+ when (Text.null name || Text.any (== '\0') name) $+ throwInvalidArgument "table function name must be nonempty and must not contain NUL"+ when (null columnDeclarations) $+ throwInvalidArgument (name <> ": a table function must have at least one result column")+ when (length columns /= length columnDeclarations) $+ throwInvalidArgument (name <> ": " <> count columns <> " column names for " <> count columnDeclarations <> " result columns")+ parameterTypes <-+ rowDeclarations name ["parameter " <> argumentName index | index <- [0 ..]] (rowColumnDeclarations (Proxy @params))+ columnTypes <- rowDeclarations name (map ("column " <>) columns) columnDeclarations+ casts <- targetCastFactory target+ let signature = positionalSignature parameterTypes Nothing+ configure factory tableFunction = do+ withIdentifier name (checked . c_duckdb_v2_table_function_set_name tableFunction)+ output (c_duckdb_v2_table_function_get_signature tableFunction) >>= configureFunctionSignature factory signature+ bind input result context = withTypeScope \scope -> withContextFactory context \factory -> do+ params <- readArguments scope name (length parameterTypes) (callbackReadFactory casts factory) input+ forM_ (zip columns columnTypes) \(column, dtype) ->+ withTypeDeclaration factory dtype \logical ->+ withIdentifier column \identifier -> checked (c_duckdb_v2_table_function_bind_add_result_column result identifier logical)+ transferCallbackState (attachOpaque (c_duckdb_v2_function_bind_set_bind_data input)) (function params)+ withRegistrationFactory target \factory ->+ registerTableFunctionWith+ target+ (configure factory)+ (tableCallbacks bind execute){tableInitGlobalCallback = Just initialize, tableProjectionPushdown = True}+ where+ count = Text.pack . show . length+ columnDeclarations = rowColumnDeclarations (Proxy @row)+ initialize info _ = do+ scan <- readCallbackState (c_duckdb_v2_table_function_init_global_get_bind_data info)+ remaining <- scan >>= newMVar :: IO (MVar [row])+ transferCallbackState (attachOpaque (c_duckdb_v2_table_function_init_global_set_global_state info)) remaining+ execute info context = withContextFactory context \factory -> do+ remaining <- readCallbackState (c_duckdb_v2_table_function_exec_get_global_state info) :: IO (MVar [row])+ chunk <- output (c_duckdb_v2_table_function_exec_get_output_chunk info)+ -- The declared column of each vector. See Note [Row lists of typed scans].+ requested <- output (c_duckdb_v2_table_function_exec_get_column_count info)+ selected <- forM (takeWhile (< requested) [0 ..]) \index ->+ fromIntegral <$> output (c_duckdb_v2_table_function_exec_get_column_index info index)+ -- The lock gives each row to one call only. See+ -- Note [Row lists of typed scans].+ modifyMVar_ remaining (writeRows factory chunk selected)++-- | Write the first rows of a list to the vectors of an output chunk. Return+-- the rows that do not fit in the chunk. The function does not evaluate those+-- rows. The list of columns gives the column of the row for each vector of the+-- chunk, in the order of the vectors. The function makes the values of these+-- columns only, so it does not evaluate the other columns. It evaluates each+-- written row to its constructor, so it evaluates each strict field of the row.+-- The values come from the borrowed factory. The function destroys each value+-- after it writes it. The size of each vector becomes the number of written+-- rows. An empty list writes no row. In the exec callback of a table function,+-- no row ends the scan. See Note [Row lists of typed scans].+writeRows :: forall row. (TypedRow row) => DuckDBV2FactoryHandle -> DuckDBV2DataChunkHandle -> [Int] -> [row] -> IO [row]+writeRows factory chunk columns rows = do+ capacity <- fromIntegral <$> output (c_duckdb_v2_data_chunk_get_capacity chunk)+ vectorCount <- output (c_duckdb_v2_data_chunk_get_vector_count chunk)+ let width = length (rowColumnDeclarations (Proxy @row))+ when (fromIntegral vectorCount /= length columns || any (\column -> column < 0 || column >= width) columns) $+ throwInternalError+ ( "the output chunk has "+ <> Text.pack (show vectorCount)+ <> " vectors for the columns "+ <> Text.pack (show columns)+ <> " of a row of "+ <> Text.pack (show width)+ <> " columns"+ )+ case splitRows capacity rows of+ (now, later) -> do+ let size = fromIntegral (length now)+ vectors <- forM (take (length columns) [0 ..]) \index -> do+ vector <- output (c_duckdb_v2_data_chunk_get_vector chunk index)+ -- Flatten after the size changes. See Note [Vector flatten] in+ -- "Database.DuckDB.Simple.Materialize".+ checked (c_duckdb_v2_vector_set_size vector size)+ checked (c_duckdb_v2_vector_flatten vector)+ pure vector+ -- Use loops with a constant stack. See Note [Constant stack at safe calls]+ -- in "Database.DuckDB.Simple.Native".+ let select+ | columns == [0 .. width - 1] = id+ | otherwise = \values -> map (values !!) columns+ forM_ (zip [0 ..] now) \(index, row) -> do+ zipWithM_+ (\vector (ColumnValue make) -> bracket (make factory) destroyValue (checked . c_duckdb_v2_vector_set_value vector index))+ vectors+ (select (rowColumnValues row))+ pure later++-- | Split a list after at most the given number of elements. Evaluate the+-- spine of the first part only. The second part is the list after those+-- elements, not a reference to the whole list. See+-- Note [Row lists of typed scans].+--+-- In GHC 9.14.1, 'splitAt' of base gives its second part through a lazy+-- pattern binding on the pair of each recursive call, so that part is a thunk+-- on the inner pairs. This function gives the second part as the plain rest of+-- the list.+splitRows :: Int -> [a] -> ([a], [a])+splitRows = go []+ where+ go taken 0 rest = (reverse taken, rest)+ go taken _ [] = (reverse taken, [])+ go taken limit (row : rest) = go (row : taken) (limit - 1) rest++-- | Read the constant arguments of one call and convert them with 'FromRow'.+-- A NULL argument is @FieldNull@. An argument that does not convert raises an+-- error that names the table function.+readArguments :: (FromRow params) => TypeScope -> Text -> Int -> ReadFactory -> DuckDBV2FunctionBindInfoHandle -> IO params+readArguments scope name count factory input = do+ fields <- forM (take count [0 ..]) \index ->+ readBoundField scope factory input (argumentName index) (fromIntegral index)+ converted <- runConversion (parseRow fromRow fields)+ case converted of+ Ok params -> pure params+ Errors failures -> throwIO (userError ("duckdb-simple: " <> Text.unpack name <> ": " <> conversionMessage failures))
src/Database/DuckDB/Simple/Types.hs view
@@ -1,69 +1,69 @@ {-# LANGUAGE TypeOperators #-} -{- |-Module : Database.DuckDB.Simple.Types-Description : Shared data types for the duckdb-simple public surface.--The datatypes in this module are intentionally kept lightweight. The main-`Database.DuckDB.Simple` module exposes the types without their constructors so-that callers interact with them through the high-level API. The actual-definitions live in @Database.DuckDB.Simple.Internal@.--}+-- |+-- Module : Database.DuckDB.Simple.Types+-- Description : Shared data types for the duckdb-simple public surface.+--+-- This module defines t'Null', t'Only', @(:.)@, and t'FormatError'. It re-exports+-- 'Connection', 'Statement', t'Query', t'SQLError', and 'ErrorType' from+-- @Database.DuckDB.Simple.Internal@. This module and @Database.DuckDB.Simple@+-- export 'Connection' and 'Statement' without their constructors. They export+-- the other types with their constructors. module Database.DuckDB.Simple.Types (- Connection,- Statement,- Query (..),- SQLError (..),- FormatError (..),- Null (..),- Only (..),- (:.) (..),+ Connection,+ Statement,+ Query (..),+ SQLError (..),+ ErrorType (..),+ FormatError (..),+ Null (..),+ Only (..),+ (:.) (..), ) where import Control.Exception (Exception) import Data.Text (Text) import qualified Data.Text as Text- import Database.DuckDB.Simple.FromField (Field (..), FieldValue (..), FromField (..), ResultError (..), returnError) import Database.DuckDB.Simple.Internal (- Connection,- Query (..),- SQLError (..),- Statement,+ Connection,+ ErrorType (..),+ Query (..),+ SQLError (..),+ Statement, )-import Database.DuckDB.Simple.Ok (Ok (..)) -- | Placeholder representing SQL @NULL@. data Null = Null- deriving (Eq, Ord, Show, Read)+ deriving (Eq, Ord, Show, Read) -{- | This instance is in this module because @Internal@ imports @FromField@.-An import of this module from @FromField@ would cause an import cycle.--}+-- | Reads only a NULL value, of each SQL type. This instance is in this module+-- because @Internal@ imports @FromField@. An import of this module from+-- @FromField@ would cause an import cycle. instance FromField Null where- fromField f = case fieldValue f of- FieldNull -> Ok Null- _ -> returnError Incompatible f (Text.pack "expected NULL")+ fromField f = case fieldValue f of+ FieldNull -> pure Null+ _ -> returnError Incompatible f (Text.pack "expected NULL") -- | Wrapper used for single-column rows. newtype Only a = Only {fromOnly :: a}- deriving (Eq, Ord, Show, Read)+ deriving (Eq, Ord, Show, Read) -- | Convenience product type for combining @FromRow@ and @ToRow@ instances. data h :. t = h :. t- deriving (Eq, Ord, Show, Read)+ deriving (Eq, Ord, Show, Read) infixr 3 :. -- | Raised when parameter formatting fails before a statement is executed. data FormatError = FormatError- { formatErrorMessage :: !Text- -- ^ Human-readable description of the mismatch (e.g. parameter counts).- , formatErrorQuery :: !Query- -- ^ Query that triggered the formatting failure.- , formatErrorParams :: ![String]- -- ^ Rendered parameter values supplied by the caller (used for diagnostics).- }- deriving (Eq, Show)+ { formatErrorMessage :: !Text+ -- ^ Human-readable description of the mismatch (e.g. parameter counts).+ , formatErrorQuery :: !Query+ -- ^ Query that triggered the formatting failure.+ , formatErrorParams :: ![String]+ -- ^ Rendered parameter values supplied by the caller (used for diagnostics).+ }+ deriving (Eq, Show) instance Exception FormatError
+ src/Database/DuckDB/Simple/V2/Config.hs view
@@ -0,0 +1,56 @@+{-# LANGUAGE BlockArguments #-}++-- |+-- Module : Database.DuckDB.Simple.V2.Config+-- Description : Register custom options with native builders and update callbacks.+--+-- The functions of this module give native handles to their actions.+-- "Database.DuckDB.Simple.Config" has the typed custom options.+--+-- Use 'registerCustomOptionWith' when the update callback needs the native update info. With that info, the+-- callback can replace the new value with @c_duckdb_v2_custom_option_update_set_value@, read the scope of+-- the write, or find if the update is a RESET. The configuration action sets the name, the type, and the+-- other properties of the option with the @c_duckdb_v2_custom_option_set_@ functions of @Database.DuckDB.FFI.V2@.+module Database.DuckDB.Simple.V2.Config (+ registerCustomOptionWith,+) where++import Database.DuckDB.Simple.Callback (+ attachOpaque,+ checkTargetThreads,+ guardContext,+ install,+ installOptional,+ withCallbackResources,+ withTarget,+ )+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.V2.Registration (ContextCallback, RegistrationTarget (..))++-- | Configure and register a custom option with managed callback cleanup.+-- The configuration action borrows the builder. Do not set user data or callbacks.+-- An extension target includes the borrowed initialization context.+registerCustomOptionWith+ :: RegistrationTarget -> (DuckDBV2CustomOptionHandle -> IO ()) -> Maybe (ContextCallback DuckDBV2CustomOptionUpdateInfoHandle) -> IO ()+registerCustomOptionWith target configure callback =+ withTarget+ target+ c_duckdb_v2_custom_option_create+ c_duckdb_v2_custom_option_destroy+ c_duckdb_v2_connection_register_custom_option+ c_duckdb_v2_extension_register_custom_option+ \option register ->+ withCallbackResources+ (checkTargetThreads target)+ ( \allocate -> installOptional callback \update ->+ install+ allocate+ DuckDBV2CustomOptionUpdateCallbackFn+ (DuckDBV2CustomOptionUpdateCallbackFn_Aux (guardContext update))+ (c_duckdb_v2_custom_option_set_update_callback option)+ )+ (attachOpaque (c_duckdb_v2_custom_option_set_user_data option))+ \installCallback -> do+ configure option+ installCallback+ register
+ src/Database/DuckDB/Simple/V2/Copy.hs view
@@ -0,0 +1,295 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE RankNTypes #-}++-- |+-- Module : Database.DuckDB.Simple.V2.Copy+-- Description : Managed V2 COPY callbacks and multi-file table functions.+--+-- Callback info and context handles are borrowed and must not escape a callback.+-- The registration user-data slot is reserved for callback cleanup. Use closures+-- for registration data and 'setCallbackState' for bind, init, and batch state.+-- Transfer state to each slot once. Use an @IORef@ to change existing state.+-- COPY format callbacks require the threaded GHC runtime (@-threaded@).+-- Batch preparation may run concurrently. Flushes for one output file are serial.+-- Destroy a collection acquired with @copy_to_batch_take_input@ after use.+-- Multi-file functions use the callbacks of a registered single-file table+-- function.+--+-- Use this module when the row-oriented functions of "Database.DuckDB.Simple.Copy" are not sufficient, for+-- example to get a batch as a native column data collection, to set the batch size, or to give the+-- statistics for @RETURN_STATS@. Make a record of callbacks with 'copyToCallbacks' or 'copyFromCallbacks'.+-- Then register it with 'registerCopyToCallbacks' or 'registerCopyFromCallbacks', or register both sides+-- with 'registerCopyFunction'. 'registerMultiFileFunctionWith' adds file lists, globs, and hive partitions to+-- a registered single-file table function.+--+-- The examples use the imports and the definitions of the module header of+-- "Database.DuckDB.Simple.V2.Registration".+module Database.DuckDB.Simple.V2.Copy (+ CopyToCallbacks (..),+ copyToCallbacks,+ CopyFromCallbacks (..),+ copyFromCallbacks,+ registerCopyFunction,+ registerCopyFunctionWith,+ registerCopyToCallbacks,+ registerCopyFromCallbacks,+ registerMultiFileFunctionWith,+ RegistrationTarget (..),+ ContextCallback,+ setCallbackState,+ getCallbackState,+) where++import Control.Monad (when)+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple.Callback (+ Allocator,+ attachOpaque,+ guardContext,+ install,+ installAll,+ installOptional,+ requireWorkerCallbackRuntime,+ withCallbackResources,+ withTarget,+ )+import Database.DuckDB.Simple.Internal (Connection)+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.V2.Registration (ContextCallback, RegistrationTarget (..), getCallbackState, setCallbackState)++-- | Every COPY TO callback in the pinned V2 API. Batch and flush are required.+data CopyToCallbacks = CopyToCallbacks+ { copyToBindCallback :: Maybe (ContextCallback DuckDBV2CopyToBindInfoHandle)+ , copyToBatchSizeCallback :: Maybe (ContextCallback DuckDBV2CopyToBatchSizeInfoHandle)+ , copyToInitCallback :: Maybe (ContextCallback DuckDBV2CopyToInitInfoHandle)+ , copyToBatchCallback :: ContextCallback DuckDBV2CopyToBatchInfoHandle+ , copyToFlushCallback :: ContextCallback DuckDBV2CopyToFlushInfoHandle+ , copyToFinalizeCallback :: Maybe (ContextCallback DuckDBV2CopyToFinalizeInfoHandle)+ , copyToStatisticsCallback :: Maybe (ContextCallback DuckDBV2CopyToStatisticsInfoHandle)+ }++-- | Use batch and flush callbacks with no optional COPY TO callbacks.+copyToCallbacks :: ContextCallback DuckDBV2CopyToBatchInfoHandle -> ContextCallback DuckDBV2CopyToFlushInfoHandle -> CopyToCallbacks+copyToCallbacks batch flush = CopyToCallbacks Nothing Nothing Nothing batch flush Nothing Nothing++-- | Every COPY FROM callback in the pinned V2 API. Bind and exec are required.+data CopyFromCallbacks = CopyFromCallbacks+ { copyFromBindCallback :: ContextCallback DuckDBV2CopyFromBindInfoHandle+ , copyFromInitGlobalCallback :: Maybe (ContextCallback DuckDBV2CopyFromInitGlobalInfoHandle)+ , copyFromInitLocalCallback :: Maybe (ContextCallback DuckDBV2CopyFromInitLocalInfoHandle)+ , copyFromExecCallback :: ContextCallback DuckDBV2CopyFromExecInfoHandle+ , copyFromProgressCallback :: Maybe (ContextCallback DuckDBV2CopyFromProgressInfoHandle)+ }++-- | Use bind and exec callbacks with no optional COPY FROM callbacks.+copyFromCallbacks :: ContextCallback DuckDBV2CopyFromBindInfoHandle -> ContextCallback DuckDBV2CopyFromExecInfoHandle -> CopyFromCallbacks+copyFromCallbacks bind execute = CopyFromCallbacks bind Nothing Nothing execute Nothing++-- | Register either direction or both. The library allocates the callback pointers. It gives their cleanup+-- to DuckDB in the user data of the native object. DuckDB runs the cleanup when it releases the object. See+-- Note [Callback ownership] in+-- <src/Database.DuckDB.Simple.Callback.html Database.DuckDB.Simple.Callback>.+registerCopyFunction :: Connection -> Text -> Maybe CopyToCallbacks -> Maybe CopyFromCallbacks -> IO ()+registerCopyFunction connection name to from = do+ when (Text.null name || Text.any (== '\0') name) $ throwInvalidArgument "invalid copy function name"+ registerCopyFunctionWith+ (ConnectionTarget connection)+ (\function -> withIdentifier name (checked . c_duckdb_v2_copy_function_set_name function))+ to+ from++-- | Register COPY callbacks on a connection or a borrowed extension target.+-- The configuration action sets the name of the format:+--+-- > V2Copy.registerCopyFunctionWith+-- > target+-- > (\function -> R.withIdentifier "managed_output" (R.checked . c_duckdb_v2_copy_function_set_name function))+-- > (Just callbacks)+-- > Nothing+registerCopyFunctionWith+ :: RegistrationTarget -> (DuckDBV2CopyFunctionHandle -> IO ()) -> Maybe CopyToCallbacks -> Maybe CopyFromCallbacks -> IO ()+registerCopyFunctionWith target configure to from =+ withTarget+ target+ c_duckdb_v2_copy_function_create+ c_duckdb_v2_copy_function_destroy+ c_duckdb_v2_connection_register_copy_function+ c_duckdb_v2_extension_register_copy_function+ \function register ->+ withCallbackResources+ (requireWorkerCallbackRuntime "COPY")+ (\allocate -> installAll (toInstallers allocate function to <> fromInstallers allocate function from))+ (attachOpaque (c_duckdb_v2_copy_function_set_user_data function))+ \installCallbacks -> do+ configure function+ installCallbacks+ register++-- | Register the advanced COPY TO callback interface.+--+-- The example registers the format @managed_output@, which counts the rows. The init callback makes the+-- output file and stores a counter as init data. The batch callback takes the rows of a batch and stores+-- their count as batch data. The flush callback adds that count to the counter. The example does not write+-- the rows. A format that writes rows converts them in the batch callback and writes them in the flush+-- callback.+--+-- > withConnection ":memory:" \connection -> do+-- > total <- newIORef (0 :: DuckDBIdx)+-- > let initialize info _ = do+-- > path <- R.output (c_duckdb_v2_copy_to_init_get_file_path info) >>= R.readText+-- > writeFile (Text.unpack path) ""+-- > R.setCallbackState (c_duckdb_v2_copy_to_init_set_init_data info) total+-- > batch info _ = do+-- > count <-+-- > bracket+-- > (R.output (c_duckdb_v2_copy_to_batch_take_input info))+-- > (R.destroy c_duckdb_v2_column_data_collection_destroy)+-- > (R.output . c_duckdb_v2_column_data_collection_row_count)+-- > R.setCallbackState (c_duckdb_v2_copy_to_batch_set_batch_data info) count+-- > flush info _ = do+-- > count <- R.getCallbackState (c_duckdb_v2_copy_to_flush_get_batch_data info) :: IO DuckDBIdx+-- > state <- R.getCallbackState (c_duckdb_v2_copy_to_flush_get_init_data info) :: IO (IORef DuckDBIdx)+-- > atomicModifyIORef' state (\n -> (n + count, ()))+-- > callbacks = (V2Copy.copyToCallbacks batch flush){V2Copy.copyToInitCallback = Just initialize}+-- > V2Copy.registerCopyToCallbacks connection "managed_output" callbacks+-- > _ <- execute_ connection "COPY (SELECT i FROM range(5000) t(i)) TO 'numbers.out' (FORMAT managed_output)"+-- > readIORef total >>= print -- DuckDBIdx {unwrap = 5000}+registerCopyToCallbacks :: Connection -> Text -> CopyToCallbacks -> IO ()+registerCopyToCallbacks connection name callbacks = registerCopyFunction connection name (Just callbacks) Nothing++-- | Register the advanced COPY FROM callback interface.+registerCopyFromCallbacks :: Connection -> Text -> CopyFromCallbacks -> IO ()+registerCopyFromCallbacks connection name callbacks = registerCopyFunction connection name Nothing (Just callbacks)++-- | Allocate the configured COPY TO callbacks before installing them.+toInstallers :: Allocator -> DuckDBV2CopyFunctionHandle -> Maybe CopyToCallbacks -> [IO (IO ())]+toInstallers _ _ Nothing = []+toInstallers allocate function (Just callbacks) =+ [ installOptional (copyToBindCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2CopyToBindCallbackFn+ (DuckDBV2CopyToBindCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_copy_to_set_bind_callback function)+ , installOptional (copyToBatchSizeCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2CopyToBatchSizeCallbackFn+ (DuckDBV2CopyToBatchSizeCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_copy_to_set_batch_size_callback function)+ , installOptional (copyToInitCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2CopyToInitCallbackFn+ (DuckDBV2CopyToInitCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_copy_to_set_init_callback function)+ , install+ allocate+ DuckDBV2CopyToBatchCallbackFn+ (DuckDBV2CopyToBatchCallbackFn_Aux (guardContext (copyToBatchCallback callbacks)))+ (c_duckdb_v2_copy_to_set_batch_callback function)+ , install+ allocate+ DuckDBV2CopyToFlushCallbackFn+ (DuckDBV2CopyToFlushCallbackFn_Aux (guardContext (copyToFlushCallback callbacks)))+ (c_duckdb_v2_copy_to_set_flush_callback function)+ , installOptional (copyToFinalizeCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2CopyToFinalizeCallbackFn+ (DuckDBV2CopyToFinalizeCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_copy_to_set_finalize_callback function)+ , installOptional (copyToStatisticsCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2CopyToStatisticsCallbackFn+ (DuckDBV2CopyToStatisticsCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_copy_to_set_statistics_callback function)+ ]++-- | Allocate the configured COPY FROM callbacks before installing them.+fromInstallers :: Allocator -> DuckDBV2CopyFunctionHandle -> Maybe CopyFromCallbacks -> [IO (IO ())]+fromInstallers _ _ Nothing = []+fromInstallers allocate function (Just callbacks) =+ [ install+ allocate+ DuckDBV2CopyFromBindCallbackFn+ (DuckDBV2CopyFromBindCallbackFn_Aux (guardContext (copyFromBindCallback callbacks)))+ (c_duckdb_v2_copy_from_set_bind_callback function)+ , installOptional (copyFromInitGlobalCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2CopyFromInitGlobalCallbackFn+ (DuckDBV2CopyFromInitGlobalCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_copy_from_set_init_global_callback function)+ , installOptional (copyFromInitLocalCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2CopyFromInitLocalCallbackFn+ (DuckDBV2CopyFromInitLocalCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_copy_from_set_init_local_callback function)+ , install+ allocate+ DuckDBV2CopyFromExecCallbackFn+ (DuckDBV2CopyFromExecCallbackFn_Aux (guardContext (copyFromExecCallback callbacks)))+ (c_duckdb_v2_copy_from_set_exec_callback function)+ , installOptional (copyFromProgressCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2CopyFromProgressCallbackFn+ (DuckDBV2CopyFromProgressCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_copy_from_set_progress_callback function)+ ]++-- | Configure a multi-file wrapper on a connection or borrowed extension target.+-- 'Database.DuckDB.Simple.Copy.registerMultiFileFunction' sets the same fields without a native handle.+--+-- The example registers the table function @managed_single@, which reads the integer in one file. Then it+-- registers the multi-file function @managed_files@, which calls @managed_single@ for each file. The+-- configuration action sets the two names. DuckDB can run the exec callback of the single-file function on+-- more than one thread for one scan. Thus the exec callback takes its state with @atomicModifyIORef\'@.+--+-- > withConnection ":memory:" \connection -> do+-- > let bind input result context = do+-- > path <- bracket (R.output (c_duckdb_v2_function_bind_get_arg_value input 0)) (R.destroy c_duckdb_v2_value_destroy) \value ->+-- > R.output (c_duckdb_v2_value_get_varchar value) >>= R.readText+-- > value <- read <$> readFile (Text.unpack path) :: IO Int64+-- > R.setCallbackState (c_duckdb_v2_function_bind_set_bind_data input) value+-- > factory <- R.output (c_duckdb_v2_context_get_factory context)+-- > R.withTypeDeclaration factory bigint \dtype ->+-- > R.withIdentifier "value" \name -> R.checked (c_duckdb_v2_table_function_bind_add_result_column result name dtype)+-- > initialize info _ = do+-- > value <- R.getCallbackState (c_duckdb_v2_table_function_init_global_get_bind_data info)+-- > state <- newIORef (Just (value :: Int64))+-- > R.setCallbackState (c_duckdb_v2_table_function_init_global_set_global_state info) state+-- > executeRows info _ = do+-- > state <- R.getCallbackState (c_duckdb_v2_table_function_exec_get_global_state info) :: IO (IORef (Maybe Int64))+-- > value <- atomicModifyIORef' state (\previous -> (Nothing, previous))+-- > chunk <- R.output (c_duckdb_v2_table_function_exec_get_output_chunk info)+-- > vector <- R.output (c_duckdb_v2_data_chunk_get_vector chunk 0)+-- > writeIntegers vector (maybe [] pure value)+-- > R.registerTableFunction+-- > connection+-- > "managed_single"+-- > [R.parameter "path" (R.TypeExpression "VARCHAR")]+-- > (R.tableCallbacks bind executeRows){R.tableInitGlobalCallback = Just initialize}+-- > V2Copy.registerMultiFileFunctionWith (R.ConnectionTarget connection) \function -> do+-- > R.withText "managed_files" \(ConstPtr view) -> R.checked (c_duckdb_v2_multi_file_function_set_name function view)+-- > R.withText "managed_single" \(ConstPtr view) -> R.checked (c_duckdb_v2_multi_file_function_set_single_file_function function view)+-- > writeFile "first.mock" "7"+-- > writeFile "second.mock" "11"+-- > rows <- query_ connection "SELECT value FROM managed_files('*.mock') ORDER BY value"+-- > print (map fromOnly rows :: [Int64]) -- [7,11]+registerMultiFileFunctionWith :: RegistrationTarget -> (DuckDBV2MultiFileFunctionHandle -> IO ()) -> IO ()+registerMultiFileFunctionWith target configure =+ withTarget+ target+ c_duckdb_v2_multi_file_function_create+ c_duckdb_v2_multi_file_function_destroy+ c_duckdb_v2_connection_register_multi_file_function+ c_duckdb_v2_extension_register_multi_file_function+ \function register -> do+ configure function+ register
+ src/Database/DuckDB/Simple/V2/FunctionDocs.hs view
@@ -0,0 +1,51 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE NamedFieldPuns #-}++-- |+-- Module : Database.DuckDB.Simple.V2.FunctionDocs+-- Description : Documentation attached to scalar, aggregate, table, macro, and pragma functions.+--+-- The examples use the imports of the module header of "Database.DuckDB.Simple.V2.Registration".+module Database.DuckDB.Simple.V2.FunctionDocs (+ FunctionDocs (..),+ defaultFunctionDocs,+ setFunctionDocs,+) where++import Control.Monad (forM_)+import Data.Text (Text)+import Database.DuckDB.Simple.Native++-- | A description, titled SQL examples, and category names.+-- Each example is a pair of a title and the SQL text:+--+-- > Docs.FunctionDocs (Just "Preserve the input type.") [("Example", "SELECT managed_resolved(42)")] ["test"]+data FunctionDocs = FunctionDocs+ { functionDescription :: Maybe Text+ , functionExamples :: [(Text, Text)]+ , functionCategories :: [Text]+ }+ deriving (Eq, Show, Read)++-- | Leave the description unset and add no examples or categories.+defaultFunctionDocs :: FunctionDocs+defaultFunctionDocs = FunctionDocs Nothing [] []++-- | Apply documentation to a borrowed native documentation handle.+-- Get the handle from the function builder in its configuration action.+-- The action replaces the description when present and appends examples and categories.+-- DuckDB copies all text before the action returns.+-- For example, in the configuration action of+-- 'Database.DuckDB.Simple.V2.Registration.registerScalarFunctionWith':+--+-- > docs <- R.output (c_duckdb_v2_scalar_function_get_docs function)+-- > Docs.setFunctionDocs docs Docs.defaultFunctionDocs{Docs.functionDescription = Just "Add two BIGINT values."}+setFunctionDocs :: DuckDBV2FunctionDocsHandle -> FunctionDocs -> IO ()+setFunctionDocs docs FunctionDocs{functionDescription, functionExamples, functionCategories} = do+ forM_ functionDescription \description ->+ withText description (checked . c_duckdb_v2_function_docs_set_description docs)+ forM_ functionExamples \(title, sql) ->+ withText title \nativeTitle ->+ withText sql \nativeSQL -> checked (c_duckdb_v2_function_docs_add_example docs nativeTitle nativeSQL)+ forM_ functionCategories \category ->+ withText category (checked . c_duckdb_v2_function_docs_add_category docs)
+ src/Database/DuckDB/Simple/V2/Logging.hs view
@@ -0,0 +1,27 @@+{-# LANGUAGE BlockArguments #-}++-- |+-- Module : Database.DuckDB.Simple.V2.Logging+-- Description : Write log messages through the context of a native callback.+--+-- The function of this module gets the borrowed context of a native callback.+-- "Database.DuckDB.Simple.Logging" has the log sinks and the functions that write+-- messages for a connection or an instance.+--+-- The example uses the imports of the module header of "Database.DuckDB.Simple.V2.Registration".+module Database.DuckDB.Simple.V2.Logging (+ logMessage,+) where++import Data.Text (Text)+import Database.DuckDB.Simple.Native++-- | Emit a message from a native callback's borrowed context.+-- A log sink gets the first text as 'Database.DuckDB.Simple.Logging.logEntryType' and the second text as+-- 'Database.DuckDB.Simple.Logging.logEntryMessage'. For example, in the exec callback of a scalar function:+--+-- > V2Logging.logMessage context Logging.LogLevelWarning "HaskellContext" "from a callback"+logMessage :: DuckDBV2ContextHandle -> LogLevel -> Text -> Text -> IO ()+logMessage context level category message =+ withText category \kind -> withText message \body ->+ checked (c_duckdb_v2_context_log context (logLevelToNative level) kind body)
+ src/Database/DuckDB/Simple/V2/LogicalRep.hs view
@@ -0,0 +1,304 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE LambdaCase #-}++-- |+-- Module : Database.DuckDB.Simple.V2.LogicalRep+-- Description : Convert logical type trees to and from native logical types.+--+-- The functions of this module read and create native logical type handles.+-- "Database.DuckDB.Simple.LogicalRep" has the Haskell type trees.+--+-- The examples use the imports of the module header of "Database.DuckDB.Simple.V2.Registration".+module Database.DuckDB.Simple.V2.LogicalRep (+ logicalTypeToRep,+ logicalTypeToBaseRep,+ valuesHaveType,+ logicalTypeFromRep,+ logicalTypeFromRepWith,+ bindingLogicalType,+ destroyLogicalType,+) where++import Control.Exception (bracket, mask_, onException, tryJust)+import Control.Monad (foldM, forM, when)+import Data.Array (elems, listArray)+import Data.Char (isSpace)+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructField (..), UnionMemberType (..))+import Database.DuckDB.Simple.Native+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Marshal.Alloc (alloca)+import Foreign.Marshal.Array (withArray)+import Foreign.Marshal.Utils (with)+import Foreign.Ptr (nullPtr)+import Foreign.Storable (peek)++-- | Copy a native logical type into a Haskell type tree. A type with an alias+-- gives 'LogicalTypeAlias', at each depth. See Note [Type aliases] in+-- "Database.DuckDB.Simple.LogicalRep". The function does not destroy the+-- handle. Keep the factory's owner alive until the function returns:+--+-- > bracket (R.output (c_duckdb_v2_vector_get_logical_type vector)) destroyLogicalType (logicalTypeToRep factory)+logicalTypeToRep :: DuckDBV2FactoryHandle -> DuckDBV2LogicalTypeHandle -> IO LogicalTypeRep+logicalTypeToRep factory = convertLogicalType (Just factory) ReadEnumDictionary++-- | Copy a native logical type into a Haskell type tree with no alias, at each+-- depth. Use it when decoding or binding needs only base types. The function+-- reads no type name. See Note [Type aliases] in+-- "Database.DuckDB.Simple.LogicalRep". The function does not destroy the+-- handle.+logicalTypeToBaseRep :: DuckDBV2LogicalTypeHandle -> IO LogicalTypeRep+logicalTypeToBaseRep = convertLogicalType Nothing ReadEnumDictionary++-- | Check the types of borrowed values against a borrowed logical type,+-- including exact field names, collations, and CRS definitions. Each temporary+-- value type is released before the next value is checked. See+-- Note [Element types of a list] in "Database.DuckDB.Simple.ToField".+valuesHaveType :: DuckDBV2LogicalTypeHandle -> [DuckDBV2ValueHandle] -> IO Bool+valuesHaveType logical values = do+ expected <- key logical+ let check [] = pure True+ check (value : remaining) = do+ matches <- bracket (output (c_duckdb_v2_value_get_logical_type value)) destroyLogicalType \actual -> do+ equal <- output (c_duckdb_v2_logical_type_is_equal logical actual)+ if equal == 0 then pure False else (== expected) <$> key actual+ if matches then check remaining else pure False+ check values+ where+ -- Native equality checks aliases and ENUM dictionaries. The partial trees+ -- are comparison keys only. They do not leave this function.+ key = convertLogicalType Nothing OmitEnumDictionary++-- | Select complete dictionaries or private comparison keys.+data EnumDictionary = ReadEnumDictionary | OmitEnumDictionary++-- | Copy a native logical type into a Haskell type tree. With a factory, keep+-- each alias. Without a factory, read no type name.+convertLogicalType :: Maybe DuckDBV2FactoryHandle -> EnumDictionary -> DuckDBV2LogicalTypeHandle -> IO LogicalTypeRep+convertLogicalType factory dictionary logical = do+ dtype <- output (c_duckdb_v2_logical_type_get_id logical)+ let parameter index action = alloca \nameSlot ->+ bracket (output (c_duckdb_v2_logical_type_get_param logical index nameSlot)) (destroy c_duckdb_v2_value_destroy) \value -> do+ name <- peek nameSlot >>= readIdentifier+ action name value+ child index = parameter index \_ value ->+ bracket (output (c_duckdb_v2_value_get_type value)) destroyLogicalType (convertLogicalType factory dictionary)+ string index = parameter index \_ value -> output (c_duckdb_v2_value_get_varchar value) >>= readText+ count = output (c_duckdb_v2_logical_type_get_param_count logical) >>= word64ToInt+ fields make = do+ size <- count+ listArray (0, size - 1)+ <$> forM+ [0 .. size - 1]+ ( \index -> parameter (fromIntegral index) \name value -> do+ rep <- bracket (output (c_duckdb_v2_value_get_type value)) destroyLogicalType (convertLogicalType factory dictionary)+ pure (make name rep)+ )+ base <- case dtype of+ DuckDBV2LogicalTypeIdGeometry -> do+ size <- count+ LogicalTypeGeometry <$> if size == 0 then pure Nothing else Just <$> string 0+ DuckDBV2LogicalTypeIdStruct -> LogicalTypeStruct <$> fields StructField+ DuckDBV2LogicalTypeIdTuple -> LogicalTypeTuple <$> fields (\_ rep -> rep)+ DuckDBV2LogicalTypeIdVarchar -> do+ size <- count+ if size == 0 then pure (LogicalTypeScalar LogicalTypeIdVarchar) else LogicalTypeCollatedVarchar <$> string 0+ DuckDBV2LogicalTypeIdUnion -> LogicalTypeUnion <$> fields UnionMemberType+ DuckDBV2LogicalTypeIdList -> LogicalTypeList <$> child 0+ DuckDBV2LogicalTypeIdArray -> do+ element <- child 0+ size <- parameter 1 \_ value -> output (c_duckdb_v2_value_get_bigint value)+ pure (LogicalTypeArray element (fromIntegral size))+ DuckDBV2LogicalTypeIdMap -> LogicalTypeMap <$> child 0 <*> child 1+ DuckDBV2LogicalTypeIdDecimal ->+ LogicalTypeDecimal+ <$> parameter 0 (\_ value -> output (c_duckdb_v2_value_get_utinyint value))+ <*> parameter 1 (\_ value -> output (c_duckdb_v2_value_get_utinyint value))+ DuckDBV2LogicalTypeIdEnum -> case dictionary of+ OmitEnumDictionary -> pure (LogicalTypeEnum (listArray (0, -1) []))+ ReadEnumDictionary -> do+ size <- count+ -- A dictionary can have many entries, and each entry makes native+ -- calls. The fold keeps a constant stack. See+ -- Note [Constant stack at safe calls] in "Database.DuckDB.Simple.Native".+ LogicalTypeEnum . listArray (0, size - 1) . reverse+ <$> foldM (\entries index -> (: entries) <$> string (fromIntegral index)) [] [0 .. size - 1]+ _ -> LogicalTypeScalar <$> logicalTypeIdFromNative dtype+ case factory of+ Nothing -> pure base+ Just nativeFactory -> do+ ident <- logicalTypeIdFromNative dtype+ typeName <- output (c_duckdb_v2_logical_type_get_name logical) >>= readIdentifier+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ hasAlias <-+ if typeName /= logicalTypeIdName ident+ then pure True+ else withIdentifier typeName \name ->+ bracket+ (output (c_duckdb_v2_factory_create_type_with_alias nativeFactory logical name))+ destroyLogicalType+ (\aliased -> (/= 0) <$> output (c_duckdb_v2_logical_type_is_equal logical aliased))+ pure (if hasAlias then LogicalTypeAlias typeName base else base)++-- | Construct an owned logical type through a borrowed native factory.+-- Keep the owner of the factory alive until you destroy the type.+-- CRS labels must be recognized by the native type constructor. Use the resolver+-- in 'logicalTypeFromRepWith' for SQL-derived types with arbitrary CRS labels.+-- 'Database.DuckDB.Simple.V2.Registration.withTypeDeclaration' constructs the type for an action and then destroys it.+--+-- > bracket (logicalTypeFromRep factory (LogicalTypeList (LogicalTypeScalar LogicalTypeIdBigInt))) destroyLogicalType (logicalTypeToRep factory)+logicalTypeFromRep :: DuckDBV2FactoryHandle -> LogicalTypeRep -> IO DuckDBV2LogicalTypeHandle+logicalTypeFromRep factory = logicalTypeFromRepWith factory (const (pure Nothing))++-- | Construct a type tree through a borrowed native factory. Use the resolver+-- for owned native leaf types. It runs before the parent types are constructed.+logicalTypeFromRepWith+ :: DuckDBV2FactoryHandle -> (LogicalTypeRep -> IO (Maybe DuckDBV2LogicalTypeHandle)) -> LogicalTypeRep -> IO DuckDBV2LogicalTypeHandle+logicalTypeFromRepWith factory resolve = create+ where+ boxed rep = bracket (create rep) destroyLogicalType (output . c_duckdb_v2_value_create_type factory)+ text value = withText value (output . c_duckdb_v2_value_create_varchar factory)+ fromText value = withText value (output . c_duckdb_v2_factory_create_type_from_text factory)+ fromName value = withIdentifier value \part ->+ bracket (output (c_duckdb_v2_qname_create part 1)) (destroy c_duckdb_v2_qname_destroy) \name ->+ output (c_duckdb_v2_factory_create_type_from_name factory name (ConstPtr nullPtr) (ConstPtr nullPtr) 0)+ construct dtype names actions = do+ when (any (Text.any (== '\0')) names) $+ throwInvalidArgument (Text.pack "logical type name contains NUL")+ withCreatedValues actions \values ->+ withArray values \valuePtr ->+ if null names+ then output (c_duckdb_v2_factory_create_type_from_id factory dtype (ConstPtr nullPtr) (ConstPtr valuePtr) (fromIntegral (length values)))+ else withIdentifierArray names \namePtr ->+ output (c_duckdb_v2_factory_create_type_from_id factory dtype (ConstPtr namePtr) (ConstPtr valuePtr) (fromIntegral (length values)))+ -- The binding resolver supplies CRS leaves. See Note [Bind-time CRS types].+ create rep = resolve rep >>= maybe (constructRep rep) pure+ constructRep = \case+ LogicalTypeScalar LogicalTypeIdVariant -> fromName (Text.pack "VARIANT")+ LogicalTypeScalar LogicalTypeIdGeometry -> fromText (Text.pack "GEOMETRY")+ LogicalTypeScalar LogicalTypeIdSQLNull -> fromName (Text.pack "NULL")+ LogicalTypeScalar dtype -> construct (logicalTypeIdToNative dtype) [] []+ LogicalTypeCollatedVarchar collation -> construct DuckDBV2LogicalTypeIdVarchar [Text.pack "collation"] [text collation]+ LogicalTypeGeometry Nothing -> fromText (Text.pack "GEOMETRY")+ LogicalTypeGeometry (Just crs) ->+ directGeometryType factory crs+ >>= maybe+ ( throwInvalidArgument+ (Text.pack "DuckDB did not preserve the requested GEOMETRY CRS; use a recognized identifier or a complete CRS definition")+ )+ pure+ LogicalTypeDecimal width scale -> do+ when (width < 1 || width > 38 || scale > width) $+ throwInvalidArgument (Text.pack "invalid DECIMAL width or scale")+ construct+ DuckDBV2LogicalTypeIdDecimal+ []+ [output (c_duckdb_v2_value_create_utinyint factory width), output (c_duckdb_v2_value_create_utinyint factory scale)]+ LogicalTypeList element -> construct DuckDBV2LogicalTypeIdList [] [boxed element]+ -- See Note [Sized ARRAY bounds] in "Database.DuckDB.Simple.Native".+ LogicalTypeArray element size -> do+ maybe (pure ()) throwInvalidArgument (arraySizeReason (toInteger size))+ construct DuckDBV2LogicalTypeIdArray [] [boxed element, output (c_duckdb_v2_value_create_bigint factory (fromIntegral size))]+ LogicalTypeMap key value -> construct DuckDBV2LogicalTypeIdMap [] [boxed key, boxed value]+ LogicalTypeStruct fields+ | null (elems fields) ->+ bracket+ (output (c_duckdb_v2_value_create_struct factory (ConstPtr nullPtr) (ConstPtr nullPtr) 0))+ destroyValue+ (output . c_duckdb_v2_value_get_logical_type)+ LogicalTypeStruct fields -> construct DuckDBV2LogicalTypeIdStruct (map structFieldName (elems fields)) (map (boxed . structFieldValue) (elems fields))+ LogicalTypeTuple fields+ | null (elems fields) ->+ bracket (output (c_duckdb_v2_value_create_tuple factory (ConstPtr nullPtr) 0)) destroyValue (output . c_duckdb_v2_value_get_logical_type)+ | otherwise -> construct DuckDBV2LogicalTypeIdTuple [] (map boxed (elems fields))+ LogicalTypeUnion members -> construct DuckDBV2LogicalTypeIdUnion (map unionMemberName (elems members)) (map (boxed . unionMemberType) (elems members))+ LogicalTypeEnum entries -> construct DuckDBV2LogicalTypeIdEnum [] (map text (elems entries))+ -- See Note [Type aliases] in "Database.DuckDB.Simple.LogicalRep".+ LogicalTypeAlias name base -> do+ when (Text.any (== '\0') name) $+ throwInvalidArgument (Text.pack "logical type name contains NUL")+ bracket (create base) destroyLogicalType \logical ->+ withIdentifier name (output . c_duckdb_v2_factory_create_type_with_alias factory logical)++-- | Reject a native count that does not fit a Haskell array index.+word64ToInt :: (Integral a) => a -> IO Int+word64ToInt value+ | toInteger value <= toInteger (maxBound :: Int) = pure (fromIntegral value)+ | otherwise = throwInternalError (Text.pack "logical type parameter count exceeds Int range")++-- | Construct an owned binding type. Resolve CRS leaves at every depth.+-- The factory must belong to a connection on the supplied instance.+bindingLogicalType+ :: DuckDBV2FactoryHandle+ -> DuckDBV2InstanceHandle+ -> LogicalTypeRep+ -> IO DuckDBV2LogicalTypeHandle+bindingLogicalType factory instance_ = logicalTypeFromRepWith factory resolve+ where+ resolve (LogicalTypeGeometry (Just crs)) = do+ direct <- tryJust unrecognized (directGeometryType factory crs)+ Just <$> case direct of+ Right (Just logical) -> pure logical+ _ -> queryGeometryType crs instance_+ resolve _ = pure Nothing+ unrecognized err+ | sqlErrorType err == Just ErrorQueryBinder+ , Just body <- sqlErrorBody err+ , Text.pack "Encountered unrecognized coordinate system '" `Text.isPrefixOf` body =+ Just ()+ | otherwise = Nothing++{- Note [Bind-time CRS types]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~+Bindings construct native types from their declarations at every depth. The+connection factory constructs recognized CRS types directly. For an unknown+label, DuckDB either reports its specific binder error or drops the CRS when+ignore_unknown_crs is set. Only these two outcomes use the SQL fallback.+Any other error from the native constructor propagates to the caller.++The fallback opens a separate connection on the same instance. It cannot run+SQL on the caller connection because the caller can have an active result.+The query applies fully qualified built-ins to a WKB literal and the bound label.+The SELECT has a schema before its first execution step. Copy its field type+while the schema is live. The helper can close without fetching a row.+Typed NULLs and empty containers use the same resolver as nonempty values.+-}++-- | Construct a CRS type directly. Return 'Nothing' if DuckDB drops its CRS.+-- Keep native errors so the binding resolver can select the unknown CRS error.+directGeometryType :: DuckDBV2FactoryHandle -> Text -> IO (Maybe DuckDBV2LogicalTypeHandle)+directGeometryType factory crs = mask_ do+ validateGeometryCRS crs+ logical <- withText crs \text ->+ bracket (output (c_duckdb_v2_value_create_varchar factory text)) destroyValue \value ->+ with value \values ->+ output (c_duckdb_v2_factory_create_type_from_id factory DuckDBV2LogicalTypeIdGeometry (ConstPtr nullPtr) (ConstPtr values) 1)+ count <- output (c_duckdb_v2_logical_type_get_param_count logical) `onException` destroyLogicalType logical+ if count == 0 then destroyLogicalType logical >> pure Nothing else pure (Just logical)++-- | Reject CRS text that cannot identify a coordinate system.+validateGeometryCRS :: Text -> IO ()+validateGeometryCRS crs =+ when (Text.all isSpace crs || Text.any (== '\0') crs) $+ throwInvalidArgument (Text.pack "a GEOMETRY CRS must be nonblank and must not contain NUL")++-- | Read one owned GEOMETRY type from a short-lived helper connection.+queryGeometryType :: Text -> DuckDBV2InstanceHandle -> IO DuckDBV2LogicalTypeHandle+queryGeometryType crs instance_ = mask_ $+ bracket (output (c_duckdb_v2_connection_create instance_)) (destroy c_duckdb_v2_connection_destroy) \connection ->+ withSQLStatement connection sql \parsed -> withConnectionFactory connection \factory ->+ withText crs \name -> bracket (output (c_duckdb_v2_value_create_varchar factory name)) destroyValue \label ->+ with label \param ->+ bracket+ (output (c_duckdb_v2_statement_execute connection parsed (ConstPtr nullPtr) (ConstPtr param) 1))+ (destroy c_duckdb_v2_result_destroy)+ \result ->+ bracket (output (c_duckdb_v2_result_get_schema result)) (destroy c_duckdb_v2_schema_destroy) \schema -> alloca \nameSlot -> do+ logical <- output (c_duckdb_v2_schema_get_field schema 0 nameSlot)+ count <- output (c_duckdb_v2_logical_type_get_param_count logical)+ when (count == 0) $ throwInvalidArgument (Text.pack "the type query did not preserve the GEOMETRY CRS")+ output (c_duckdb_v2_logical_type_copy logical)+ where+ -- Little-endian WKB for GEOMETRYCOLLECTION EMPTY. The hex literal has type BLOB.+ sql = Text.pack "SELECT system.main.ST_SetCRS(system.main.ST_GeomFromWKB(X'010700000000000000'), ?)"
+ src/Database/DuckDB/Simple/V2/Pragma.hs view
@@ -0,0 +1,121 @@+{-# LANGUAGE BlockArguments #-}++-- |+-- Module : Database.DuckDB.Simple.V2.Pragma+-- Description : Register PRAGMA functions with callbacks that get native handles.+--+-- The callbacks of this module get the borrowed bind info and context of the+-- call. "Database.DuckDB.Simple.Pragma" has the typed callbacks.+--+-- Use this module to read the arguments of a call as native values, to set the documentation of a pragma,+-- or to register a pragma on an extension target. 'registerPragmaFunctionOn' declares the name and the+-- parameters. 'registerPragmaFunctionWith' gives the native builder to a configuration action. A pragma has+-- one callback: a 'QueryPragma' or an 'ExecPragma'.+--+-- The example uses the imports and the definitions of the module header of+-- "Database.DuckDB.Simple.V2.Registration".+module Database.DuckDB.Simple.V2.Pragma (+ PragmaCallback (..),+ registerPragmaFunctionOn,+ registerPragmaFunctionWith,+) where++import Control.Monad (when)+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple.Callback (+ attachOpaque,+ checkTargetThreads,+ guardContext,+ install,+ runCallback,+ withCallbackResources,+ withTarget,+ )+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.V2.Registration (+ ContextCallback,+ FunctionParameter,+ FunctionSignature (..),+ RegistrationTarget (..),+ configureFunctionSignature,+ withRegistrationFactory,+ )++-- | A query callback returns SQL to run in place of the PRAGMA statement.+-- An execution callback performs an action and returns no rows. Callback handles+-- are borrowed. Exceptions become SQL errors. DuckDB permits exactly one callback.+data PragmaCallback+ = QueryPragma (DuckDBV2FunctionBindInfoHandle -> DuckDBV2ContextHandle -> IO Text)+ | ExecPragma (ContextCallback DuckDBV2FunctionBindInfoHandle)++-- | Register a pragma on a connection or a loading extension. The name must not+-- be empty and must not contain NUL.+registerPragmaFunctionOn :: RegistrationTarget -> Text -> [FunctionParameter] -> PragmaCallback -> IO ()+registerPragmaFunctionOn target name parameters callback = do+ when (Text.null name || Text.any (== '\0') name) $+ throwInvalidArgument (Text.pack "pragma function name must be nonempty and must not contain NUL")+ withRegistrationFactory target \factory ->+ registerPragmaFunctionWith+ target+ ( \function -> do+ withIdentifier name (checked . c_duckdb_v2_pragma_function_set_name function)+ signature <- output (c_duckdb_v2_pragma_function_get_signature function)+ configureFunctionSignature factory (FunctionSignature parameters Nothing) signature+ )+ callback++-- | Configure a pragma on a connection or a loading extension.+-- The action borrows the builder and can set its name, signature, and documentation.+-- Do not set native callbacks or user data. The library owns the callback pointer.+-- Use Haskell closures to retain registration data.+--+-- The example registers an execution pragma that counts its calls. The configuration action sets the name+-- and the description of the pragma.+--+-- > withConnection ":memory:" \connection -> do+-- > calls <- newIORef (0 :: Int)+-- > Pragma.registerPragmaFunctionWith+-- > (R.ConnectionTarget connection)+-- > ( \function -> do+-- > R.withIdentifier "managed_exec" (R.checked . c_duckdb_v2_pragma_function_set_name function)+-- > docs <- R.output (c_duckdb_v2_pragma_function_get_docs function)+-- > Docs.setFunctionDocs docs Docs.defaultFunctionDocs{Docs.functionDescription = Just "Count calls."}+-- > )+-- > (Pragma.ExecPragma (\_ _ -> modifyIORef' calls (+ 1)))+-- > _ <- execute_ connection "PRAGMA managed_exec"+-- > readIORef calls >>= print -- 1+registerPragmaFunctionWith :: RegistrationTarget -> (DuckDBV2PragmaFunctionHandle -> IO ()) -> PragmaCallback -> IO ()+registerPragmaFunctionWith target configure callback =+ withTarget+ target+ c_duckdb_v2_pragma_function_create+ c_duckdb_v2_pragma_function_destroy+ c_duckdb_v2_connection_register_pragma_function+ c_duckdb_v2_extension_register_pragma_function+ \function register ->+ withCallbackResources+ (checkTargetThreads target)+ ( \allocate -> case callback of+ QueryPragma action ->+ install+ allocate+ DuckDBV2PragmaFunctionQueryCallbackFn+ ( DuckDBV2PragmaFunctionQueryCallbackFn_Aux \info result context err ->+ runCallback err do+ sql <- action info context+ withText sql (checked . c_duckdb_v2_pragma_function_query_set_query result)+ )+ (c_duckdb_v2_pragma_function_set_query_callback function)+ ExecPragma action ->+ install+ allocate+ DuckDBV2PragmaFunctionExecCallbackFn+ (DuckDBV2PragmaFunctionExecCallbackFn_Aux (guardContext action))+ (c_duckdb_v2_pragma_function_set_exec_callback function)+ )+ (attachOpaque (c_duckdb_v2_pragma_function_set_user_data function))+ \installCallback -> do+ configure function+ installCallback+ register
+ src/Database/DuckDB/Simple/V2/Registration.hs view
@@ -0,0 +1,962 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : Database.DuckDB.Simple.V2.Registration+-- Description : Advanced V2 registration with Haskell callbacks.+--+-- The library allocates the callback pointers. It gives their cleanup to DuckDB in the user data of the+-- native object. DuckDB runs the cleanup when it releases the object. See Note [Callback ownership] in+-- <src/Database.DuckDB.Simple.Callback.html Database.DuckDB.Simple.Callback>.+-- Callback info and context handles are borrowed and must+-- not escape the callback. Use @Database.DuckDB.FFI.V2@ operations with 'checked'+-- and 'output' in callbacks.+--+-- The registration user-data slot is reserved for callback cleanup. Use Haskell+-- closures for registration data and 'setCallbackState' for bind or execution+-- state. A state getter with a different type than its setter raises an error.+-- Synchronize access to state that DuckDB shares between workers.+--+-- The pinned V2 API has no custom file-system registration API. Use+-- @Database.DuckDB.Simple.Logging@ for managed log-sink registration.+--+-- Use this module when the typed functions of "Database.DuckDB.Simple.Function" cannot express a+-- registration. This module adds named-only and variadic parameters, parameter defaults, a return type that+-- a callback resolves, function properties, function documentation, and callbacks that read and write+-- native vectors. A registration has three parts: the parameters from 'parameter', a record of callbacks+-- from 'scalarCallbacks', 'aggregateCallbacks', or 'tableCallbacks', and a call of+-- 'registerScalarFunction', 'registerAggregateFunction', or 'registerTableFunction'. The functions with the+-- suffix @With@ give the native builder to a configuration action and accept an extension target.+--+-- The examples of the V2 modules use the extensions @BlockArguments@ and @OverloadedStrings@, and these+-- imports:+--+-- > import Control.Exception (bracket)+-- > import Control.Monad (forM, forM_)+-- > import Data.IORef (IORef, atomicModifyIORef', modifyIORef', newIORef, readIORef)+-- > import Data.Int (Int64)+-- > import qualified Data.Text as Text+-- > import Database.DuckDB.FFI.V2+-- > import Database.DuckDB.Simple hiding (bind, execute)+-- > import qualified Database.DuckDB.Simple.Logging as Logging+-- > import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..))+-- > import qualified Database.DuckDB.Simple.V2.Copy as V2Copy+-- > import qualified Database.DuckDB.Simple.V2.FunctionDocs as Docs+-- > import qualified Database.DuckDB.Simple.V2.Logging as V2Logging+-- > import Database.DuckDB.Simple.V2.LogicalRep (destroyLogicalType, logicalTypeFromRep, logicalTypeToRep)+-- > import qualified Database.DuckDB.Simple.V2.Pragma as Pragma+-- > import qualified Database.DuckDB.Simple.V2.Registration as R+-- > import Foreign.C.ConstPtr (ConstPtr (..))+-- > import Foreign.Ptr (castPtr)+-- > import Foreign.Storable (peek, peekElemOff, poke, pokeElemOff, sizeOf)+--+-- The examples also use these definitions. The helpers read and write the rows of a BIGINT vector. Each+-- helper flattens the vector before it reads or writes the data. Then each row has its own slot in the data.+--+-- > bigint :: R.TypeDeclaration+-- > bigint = R.StructuredType (LogicalTypeScalar LogicalTypeIdBigInt)+-- >+-- > readIntegers :: Int -> DuckDBV2VectorHandle -> IO [Int64]+-- > readIntegers count vector = do+-- > R.checked (c_duckdb_v2_vector_flatten vector)+-- > raw <- R.output (c_duckdb_v2_vector_get_data_mutable vector)+-- > mapM (peekElemOff (castPtr raw)) [0 .. count - 1]+-- >+-- > writeIntegers :: DuckDBV2VectorHandle -> [Int64] -> IO ()+-- > writeIntegers vector values = do+-- > R.checked (c_duckdb_v2_vector_set_size vector (fromIntegral (length values)))+-- > R.checked (c_duckdb_v2_vector_flatten vector)+-- > raw <- R.output (c_duckdb_v2_vector_get_data_mutable vector)+-- > forM_ (zip [0 ..] values) \(index, value) -> pokeElemOff (castPtr raw) index value+module Database.DuckDB.Simple.V2.Registration (+ RegistrationTarget (..),+ withRegistrationFactory,+ TypeDeclaration (..),+ DefaultValue (..),+ defaultValue,+ FunctionParameter (..),+ parameter,+ FunctionSignature (..),+ FunctionProperties,+ ContextCallback,+ BindCallback,+ ScalarCallbacks (..),+ scalarCallbacks,+ AggregateCallbacks (..),+ aggregateCallbacks,+ TableCallbacks (..),+ tableCallbacks,+ registerScalarFunction,+ registerScalarFunctionWith,+ registerAggregateFunction,+ registerAggregateFunctionWith,+ registerAggregateFunctionWithCleanup,+ registerTableFunction,+ registerTableFunctionWith,+ registerCast,+ registerCastWith,+ registerCustomType,+ registerCustomTypeOn,+ registerCustomTypeWith,+ registerMacroFunctionWith,+ registerReplacementScan,+ registerReplacementScanOn,+ registerInstanceReplacementScan,+ setCallbackState,+ getCallbackState,+ withTypeDeclaration,+ configureFunctionSignature,+ checked,+ output,+ destroy,+ nullHandle,+ withText,+ readText,+ withIdentifier,+ readIdentifier,+) where++import Control.Exception (bracket, throwIO)+import Control.Monad (forM_, when)+import Data.Dynamic (dynTypeRep, fromDynamic, toDyn)+import Data.Int (Int64)+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Typeable (Typeable, typeRep)+import Data.Void (Void)+import Database.DuckDB.Simple.Callback (+ RegistrationTarget (..),+ attachOpaque,+ checkQueryThreads,+ checkTargetThreads,+ guardContext,+ install,+ installAll,+ installOptional,+ readCallbackState,+ runCallback,+ runCallbackWith,+ transferCallbackState,+ withCallbackResources,+ withCallbackResourcesFinally,+ withRegistrationFactory,+ withTarget,+ )+import Database.DuckDB.Simple.Internal (Connection, Instance, withConnectionHandle, withInstanceHandle)+import Database.DuckDB.Simple.LogicalRep (TypeDeclaration (..))+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.ToField (ToDuckValue (..))+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeFromRep)+import Foreign.Ptr (Ptr)++-- | Construct an owned native default value through the borrowed factory of+-- the registration target. Registration releases the value.+--+-- > R.DefaultValue (\factory -> R.output (c_duckdb_v2_value_create_bigint factory 35))+newtype DefaultValue = DefaultValue (DuckDBV2FactoryHandle -> IO DuckDBV2ValueHandle)++-- | Use a typed Haskell value as a default. It works on a connection target and+-- on an extension target.+defaultValue :: (ToDuckValue a) => a -> DefaultValue+defaultValue value = DefaultValue (\factory -> toDuckValue factory value)++-- | One parameter. Native V2 validates parameter order, names, and defaults.+data FunctionParameter = FunctionParameter+ { parameterName :: Text+ -- ^ The parameter name.+ , parameterType :: TypeDeclaration+ -- ^ The declared parameter type.+ , parameterDefault :: Maybe DefaultValue+ -- ^ The optional default value.+ , parameterKind :: DuckDBV2FunctionParameterKind+ }++-- | Declare a required parameter accepted by position or by name.+-- Change a field to add a default or to use another parameter kind:+--+-- > (R.parameter "count" bigint){R.parameterDefault = Just (R.defaultValue (3 :: Int64))}+-- > (R.parameter "right" bigint){R.parameterKind = DuckDBV2FunctionParameterKindNamedOnly}+parameter :: Text -> TypeDeclaration -> FunctionParameter+parameter name dtype = FunctionParameter name dtype Nothing DuckDBV2FunctionParameterKindStandard++-- | Function parameters and an optional return type. Table functions use Nothing.+data FunctionSignature = FunctionSignature+ { signatureParameters :: [FunctionParameter]+ , signatureReturnType :: Maybe TypeDeclaration+ }++-- | Native function properties. Scalar and aggregate functions default to fallible.+type FunctionProperties = [(DuckDBV2FunctionPropertyKey, DuckDBV2FunctionPropertyValue)]++-- | A callback with a borrowed execution context. Exceptions become SQL errors.+type ContextCallback info = info -> DuckDBV2ContextHandle -> IO ()++-- | A bind callback with borrowed input metadata, output metadata, and context.+type BindCallback result = DuckDBV2FunctionBindInfoHandle -> result -> DuckDBV2ContextHandle -> IO ()++-- | All scalar function callbacks. The exec callback is required.+data ScalarCallbacks = ScalarCallbacks+ { scalarResolveTypesCallback :: Maybe (BindCallback DuckDBV2ScalarFunctionResolveTypesInfoHandle)+ , scalarBindCallback :: Maybe (BindCallback DuckDBV2ScalarFunctionBindInfoHandle)+ , scalarInitCallback :: Maybe (ContextCallback DuckDBV2ScalarFunctionInitInfoHandle)+ , scalarExecCallback :: ContextCallback DuckDBV2ScalarFunctionExecInfoHandle+ }++-- | Use one exec callback and no optional scalar callbacks.+scalarCallbacks :: ContextCallback DuckDBV2ScalarFunctionExecInfoHandle -> ScalarCallbacks+scalarCallbacks execute = ScalarCallbacks Nothing Nothing Nothing execute++-- | All aggregate callbacks. Bind and state destruction are optional.+-- Aggregate state callbacks have no context in native V2. Native cleanup ignores+-- destroy-callback errors after exception containment.+data AggregateCallbacks = AggregateCallbacks+ { aggregateBindCallback :: Maybe (BindCallback DuckDBV2AggregateFunctionBindInfoHandle)+ , aggregateSizeCallback :: DuckDBV2AggregateFunctionSizeInfoHandle -> IO ()+ , aggregateInitCallback :: DuckDBV2AggregateFunctionInitInfoHandle -> IO ()+ , aggregateUpdateCallback :: DuckDBV2AggregateFunctionUpdateInfoHandle -> IO ()+ , aggregateCombineCallback :: DuckDBV2AggregateFunctionCombineInfoHandle -> IO ()+ , aggregateFinalizeCallback :: DuckDBV2AggregateFunctionFinalizeInfoHandle -> IO ()+ , aggregateDestroyCallback :: Maybe (DuckDBV2AggregateFunctionDestroyInfoHandle -> IO ())+ }++-- | Use the required aggregate callbacks and no optional callbacks.+aggregateCallbacks+ :: (DuckDBV2AggregateFunctionSizeInfoHandle -> IO ())+ -> (DuckDBV2AggregateFunctionInitInfoHandle -> IO ())+ -> (DuckDBV2AggregateFunctionUpdateInfoHandle -> IO ())+ -> (DuckDBV2AggregateFunctionCombineInfoHandle -> IO ())+ -> (DuckDBV2AggregateFunctionFinalizeInfoHandle -> IO ())+ -> AggregateCallbacks+aggregateCallbacks size initialize update combine finalize =+ AggregateCallbacks Nothing size initialize update combine finalize Nothing++-- | All table callbacks and the projection-pushdown setting.+-- The claim-batch callback is part of the pinned unstable V2 API.+--+-- A filter pushdown callback must not accept a BETWEEN predicate. DuckDB 2.0+-- merges two range comparisons on one column into one BETWEEN node, and the C API+-- has no call that reads whether a bound is inclusive+-- (<https://github.com/duckdb/duckdb/issues/25391 DuckDB issue 25391>). DuckDB+-- removes an accepted predicate from the query, so the scan returns wrong rows.+data TableCallbacks = TableCallbacks+ { tableBindCallback :: BindCallback DuckDBV2TableFunctionBindInfoHandle+ , tableInitGlobalCallback :: Maybe (ContextCallback DuckDBV2TableFunctionInitGlobalInfoHandle)+ , tableInitLocalCallback :: Maybe (ContextCallback DuckDBV2TableFunctionInitLocalInfoHandle)+ , tableExecCallback :: ContextCallback DuckDBV2TableFunctionExecInfoHandle+ , tableProgressCallback :: Maybe (ContextCallback DuckDBV2TableFunctionProgressInfoHandle)+ , tableFilterPushdownCallback :: Maybe (ContextCallback DuckDBV2TableFunctionFilterPushdownInfoHandle)+ , tablePartitionDataCallback :: Maybe (ContextCallback DuckDBV2TableFunctionPartitionDataInfoHandle)+ , tablePartitioningCallback :: Maybe (ContextCallback DuckDBV2TableFunctionPartitioningInfoHandle)+ , tableClaimBatchCallback :: Maybe (ContextCallback DuckDBV2TableFunctionClaimBatchInfoHandle)+ , tableGetBindInfoCallback :: Maybe (ContextCallback DuckDBV2TableFunctionGetBindInfoInfoHandle)+ , tableProjectionPushdown :: Bool+ }++-- | Use bind and exec callbacks. Disable optional callbacks and projection pushdown.+tableCallbacks :: BindCallback DuckDBV2TableFunctionBindInfoHandle -> ContextCallback DuckDBV2TableFunctionExecInfoHandle -> TableCallbacks+tableCallbacks bind execute = TableCallbacks bind Nothing Nothing execute Nothing Nothing Nothing Nothing Nothing Nothing False++-- | Register an advanced scalar function on the connection's database.+-- Set an ANY return type only when the resolve-types callback supplies a concrete type.+--+-- The example registers a function with two BIGINT parameters. Each parameter has a default, and the second+-- parameter is named-only. The exec callback adds the two arguments of each row.+--+-- > scalarSum :: R.ContextCallback DuckDBV2ScalarFunctionExecInfoHandle+-- > scalarSum info _ = do+-- > count <- R.output (c_duckdb_v2_scalar_function_exec_get_row_count info)+-- > columns <- forM [0, 1 :: Int] \index ->+-- > R.output (c_duckdb_v2_scalar_function_exec_get_arg info (fromIntegral index)) >>= readIntegers (fromIntegral count)+-- > vector <- R.output (c_duckdb_v2_scalar_function_exec_get_result info)+-- > writeIntegers vector [sum [column !! row | column <- columns] | row <- [0 .. fromIntegral count - 1]]+-- >+-- > withConnection ":memory:" \connection -> do+-- > let left = (R.parameter "left" bigint){R.parameterDefault = Just (R.defaultValue (7 :: Int64))}+-- > right =+-- > (R.parameter "right" bigint)+-- > { R.parameterDefault = Just (R.defaultValue (35 :: Int64))+-- > , R.parameterKind = DuckDBV2FunctionParameterKindNamedOnly+-- > }+-- > R.registerScalarFunction+-- > connection+-- > "managed_default"+-- > (R.FunctionSignature [left, right] (Just bigint))+-- > []+-- > (R.scalarCallbacks scalarSum)+-- > rows <- query_ connection "SELECT managed_default(), managed_default(right := 5, left := 3)"+-- > print (rows :: [(Int64, Int64)]) -- [(42,8)]+registerScalarFunction :: Connection -> Text -> FunctionSignature -> FunctionProperties -> ScalarCallbacks -> IO ()+registerScalarFunction connection name signature properties callbacks = do+ validateName name+ withConnectionHandle connection \native -> withConnectionFactory native \factory ->+ registerScalarFunctionWith+ (ConnectionTarget connection)+ ( \function -> do+ withIdentifier name $ checked . c_duckdb_v2_scalar_function_set_name function+ output (c_duckdb_v2_scalar_function_get_signature function) >>= configureFunctionSignature factory signature+ forM_ properties $ \(key, value) -> checked (c_duckdb_v2_scalar_function_set_property function key value)+ )+ callbacks++-- | Configure and register a managed scalar function on a connection or extension.+-- The configuration action borrows the builder. Do not set callbacks or user data.+--+-- The example sets the name, the signature, and the description of the function. It uses @scalarSum@ from+-- the example of 'registerScalarFunction'. 'withRegistrationFactory' gives the factory for the types of the+-- signature.+--+-- > withConnection ":memory:" \connection -> do+-- > let target = R.ConnectionTarget connection+-- > R.withRegistrationFactory target \factory ->+-- > R.registerScalarFunctionWith+-- > target+-- > ( \function -> do+-- > R.withIdentifier "managed_target_default" (R.checked . c_duckdb_v2_scalar_function_set_name function)+-- > signature <- R.output (c_duckdb_v2_scalar_function_get_signature function)+-- > R.configureFunctionSignature+-- > factory+-- > (R.FunctionSignature [R.parameter "left" bigint, R.parameter "right" bigint] (Just bigint))+-- > signature+-- > docs <- R.output (c_duckdb_v2_scalar_function_get_docs function)+-- > Docs.setFunctionDocs docs Docs.defaultFunctionDocs{Docs.functionDescription = Just "Add two BIGINT values."}+-- > )+-- > (R.scalarCallbacks scalarSum)+-- > rows <- query_ connection "SELECT description FROM duckdb_functions() WHERE function_name = 'managed_target_default'"+-- > print (map fromOnly rows :: [Text.Text]) -- ["Add two BIGINT values."]+registerScalarFunctionWith :: RegistrationTarget -> (DuckDBV2ScalarFunctionHandle -> IO ()) -> ScalarCallbacks -> IO ()+registerScalarFunctionWith target configure callbacks =+ withTarget+ target+ c_duckdb_v2_scalar_function_create+ c_duckdb_v2_scalar_function_destroy+ c_duckdb_v2_connection_register_scalar_function+ c_duckdb_v2_extension_register_scalar_function+ \function register ->+ withCallbackResources+ (checkTargetThreads target)+ ( \allocate ->+ installAll+ [ installOptional (scalarResolveTypesCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2ScalarFunctionResolveTypesCallbackFn+ (DuckDBV2ScalarFunctionResolveTypesCallbackFn_Aux (guardBind callback))+ (c_duckdb_v2_scalar_function_set_resolve_types_callback function)+ , installOptional (scalarBindCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2ScalarFunctionBindCallbackFn+ (DuckDBV2ScalarFunctionBindCallbackFn_Aux (guardBind callback))+ (c_duckdb_v2_scalar_function_set_bind_callback function)+ , installOptional (scalarInitCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2ScalarFunctionInitCallbackFn+ (DuckDBV2ScalarFunctionInitCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_scalar_function_set_init_callback function)+ , install+ allocate+ DuckDBV2ScalarFunctionExecCallbackFn+ (DuckDBV2ScalarFunctionExecCallbackFn_Aux (guardContext (scalarExecCallback callbacks)))+ (c_duckdb_v2_scalar_function_set_exec_callback function)+ ]+ )+ (attachOpaque (c_duckdb_v2_scalar_function_set_user_data function))+ \installCallbacks -> do+ checked (c_duckdb_v2_scalar_function_set_property function DuckDBV2FunctionPropertyFallibility DuckDBV2FunctionPropertyFallibilityFallible)+ configure function+ installCallbacks+ register++-- | Register an aggregate function on the connection's database.+-- State memory and its size follow the native V2 aggregate contract.+--+-- The example registers a sum of BIGINT values. The state of a group is one 'Int64' in native memory. The+-- size callback gives the size of the state. The other callbacks read and write the state with @peek@ and+-- @poke@. The finalize callback writes each total at the result offset of the result vector.+--+-- > sumCallbacks :: R.AggregateCallbacks+-- > sumCallbacks = R.aggregateCallbacks stateSize initialize update combine finalize+-- > where+-- > stateSize info = R.checked (c_duckdb_v2_aggregate_function_size_set_state_size info (fromIntegral (sizeOf (0 :: Int64))))+-- > initialize info = do+-- > count <- R.output (c_duckdb_v2_aggregate_function_init_get_state_count info)+-- > states <- R.output (c_duckdb_v2_aggregate_function_init_get_states info)+-- > forM_ [0 .. fromIntegral count - 1] \index -> peekElemOff states index >>= \state -> poke (castPtr state) (0 :: Int64)+-- > update info = do+-- > count <- R.output (c_duckdb_v2_aggregate_function_update_get_row_count info)+-- > states <- R.output (c_duckdb_v2_aggregate_function_update_get_states info)+-- > values <- R.output (c_duckdb_v2_aggregate_function_update_get_arg info 0) >>= readIntegers (fromIntegral count)+-- > forM_ (zip [0 ..] values) \(index, value) -> do+-- > state <- castPtr <$> peekElemOff states index+-- > total <- peek state+-- > poke state (total + value :: Int64)+-- > combine info = do+-- > count <- R.output (c_duckdb_v2_aggregate_function_combine_get_state_count info)+-- > sources <- R.output (c_duckdb_v2_aggregate_function_combine_get_sources info)+-- > targets <- R.output (c_duckdb_v2_aggregate_function_combine_get_targets info)+-- > forM_ [0 .. fromIntegral count - 1] \index -> do+-- > source <- peekElemOff sources index >>= peek . castPtr+-- > target <- castPtr <$> peekElemOff targets index+-- > total <- peek target+-- > poke target (total + source :: Int64)+-- > finalize info = do+-- > count <- R.output (c_duckdb_v2_aggregate_function_finalize_get_state_count info)+-- > states <- R.output (c_duckdb_v2_aggregate_function_finalize_get_states info)+-- > offset <- R.output (c_duckdb_v2_aggregate_function_finalize_get_result_offset info)+-- > result <- R.output (c_duckdb_v2_aggregate_function_finalize_get_result info)+-- > raw <- R.output (c_duckdb_v2_vector_get_data_mutable result)+-- > forM_ [0 .. fromIntegral count - 1] \index -> do+-- > total <- peekElemOff states index >>= peek . castPtr+-- > pokeElemOff (castPtr raw) (fromIntegral offset + index) (total :: Int64)+-- >+-- > withConnection ":memory:" \connection -> do+-- > R.registerAggregateFunction connection "managed_sum" (R.FunctionSignature [R.parameter "value" bigint] (Just bigint)) [] sumCallbacks+-- > rows <- query_ connection "SELECT managed_sum(i::BIGINT) FROM range(12) t(i) GROUP BY i % 3 ORDER BY i % 3"+-- > print (map fromOnly rows :: [Int64]) -- [18,22,26]+registerAggregateFunction :: Connection -> Text -> FunctionSignature -> FunctionProperties -> AggregateCallbacks -> IO ()+registerAggregateFunction connection name signature properties callbacks = do+ validateName name+ withConnectionHandle connection \native -> withConnectionFactory native \factory ->+ registerAggregateFunctionWith+ (ConnectionTarget connection)+ ( \function -> do+ withIdentifier name $ checked . c_duckdb_v2_aggregate_function_set_name function+ output (c_duckdb_v2_aggregate_function_get_signature function) >>= configureFunctionSignature factory signature+ forM_ properties $ \(key, value) -> checked (c_duckdb_v2_aggregate_function_set_property function key value)+ )+ callbacks++-- | Configure and register a managed aggregate function on a connection or extension.+-- The configuration action borrows the builder. Do not set callbacks or user data.+registerAggregateFunctionWith :: RegistrationTarget -> (DuckDBV2AggregateFunctionHandle -> IO ()) -> AggregateCallbacks -> IO ()+registerAggregateFunctionWith target = registerAggregateFunctionWithCleanup target (pure ())++-- | Register an aggregate with cleanup owned by its native callback resources.+-- Acquire resources for this cleanup only from the callbacks. Native builder+-- acquisition can fail before setup of callback resources starts. Cleanup runs+-- if that setup fails. After attachment, it runs when the last native+-- callback owner is released, including after a failed configuration or+-- registration. Cleanup exceptions are contained. Cleanup must not use a+-- borrowed registration context. The configuration action must not set callbacks+-- or user data.+registerAggregateFunctionWithCleanup+ :: RegistrationTarget -> IO () -> (DuckDBV2AggregateFunctionHandle -> IO ()) -> AggregateCallbacks -> IO ()+registerAggregateFunctionWithCleanup target cleanup configure callbacks =+ withTarget+ target+ c_duckdb_v2_aggregate_function_create+ c_duckdb_v2_aggregate_function_destroy+ c_duckdb_v2_connection_register_aggregate_function+ c_duckdb_v2_extension_register_aggregate_function+ \function register ->+ withCallbackResourcesFinally+ (checkTargetThreads target)+ cleanup+ ( \allocate ->+ installAll+ [ installOptional (aggregateBindCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2AggregateFunctionBindCallbackFn+ (DuckDBV2AggregateFunctionBindCallbackFn_Aux (guardBind callback))+ (c_duckdb_v2_aggregate_function_set_bind_callback function)+ , install+ allocate+ DuckDBV2AggregateFunctionSizeCallbackFn+ (DuckDBV2AggregateFunctionSizeCallbackFn_Aux (guardCallback (aggregateSizeCallback callbacks)))+ (c_duckdb_v2_aggregate_function_set_size_callback function)+ , install+ allocate+ DuckDBV2AggregateFunctionInitCallbackFn+ (DuckDBV2AggregateFunctionInitCallbackFn_Aux (guardCallback (aggregateInitCallback callbacks)))+ (c_duckdb_v2_aggregate_function_set_init_callback function)+ , install+ allocate+ DuckDBV2AggregateFunctionUpdateCallbackFn+ (DuckDBV2AggregateFunctionUpdateCallbackFn_Aux (guardCallback (aggregateUpdateCallback callbacks)))+ (c_duckdb_v2_aggregate_function_set_update_callback function)+ , install+ allocate+ DuckDBV2AggregateFunctionCombineCallbackFn+ (DuckDBV2AggregateFunctionCombineCallbackFn_Aux (guardCallback (aggregateCombineCallback callbacks)))+ (c_duckdb_v2_aggregate_function_set_combine_callback function)+ , install+ allocate+ DuckDBV2AggregateFunctionFinalizeCallbackFn+ (DuckDBV2AggregateFunctionFinalizeCallbackFn_Aux (guardCallback (aggregateFinalizeCallback callbacks)))+ (c_duckdb_v2_aggregate_function_set_finalize_callback function)+ , installOptional (aggregateDestroyCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2AggregateFunctionDestroyCallbackFn+ (DuckDBV2AggregateFunctionDestroyCallbackFn_Aux (guardCallback callback))+ (c_duckdb_v2_aggregate_function_set_destroy_callback function)+ ]+ )+ (attachOpaque (c_duckdb_v2_aggregate_function_set_user_data function))+ \installCallbacks -> do+ checked+ (c_duckdb_v2_aggregate_function_set_property function DuckDBV2FunctionPropertyFallibility DuckDBV2FunctionPropertyFallibilityFallible)+ configure function+ installCallbacks+ register++-- | Register a table function. Its bind callback declares the result columns.+--+-- The example registers a table function with one BIGINT parameter. The bind callback stores the argument as+-- bind data and declares the result column. The 'tableInitGlobalCallback' callback copies the bind data into+-- a global state. The exec callback takes the state one time. Thus the first chunk has the rows and the+-- next chunk is empty. An empty chunk ends the scan.+--+-- > tableDefinition :: R.TableCallbacks+-- > tableDefinition = (R.tableCallbacks bind execute){R.tableInitGlobalCallback = Just initialize}+-- > where+-- > bind input result context = do+-- > amount <- bracket (R.output (c_duckdb_v2_function_bind_get_arg_value input 0)) (R.destroy c_duckdb_v2_value_destroy) \value ->+-- > R.output (c_duckdb_v2_value_get_bigint value)+-- > R.setCallbackState (c_duckdb_v2_function_bind_set_bind_data input) (amount :: Int64)+-- > factory <- R.output (c_duckdb_v2_context_get_factory context)+-- > R.withTypeDeclaration factory bigint \dtype ->+-- > R.withIdentifier "amount" \name -> R.checked (c_duckdb_v2_table_function_bind_add_result_column result name dtype)+-- > initialize info _ = do+-- > amount <- R.getCallbackState (c_duckdb_v2_table_function_init_global_get_bind_data info)+-- > state <- newIORef (Just (amount :: Int64))+-- > R.setCallbackState (c_duckdb_v2_table_function_init_global_set_global_state info) state+-- > execute info _ = do+-- > state <- R.getCallbackState (c_duckdb_v2_table_function_exec_get_global_state info) :: IO (IORef (Maybe Int64))+-- > amount <- atomicModifyIORef' state (\value -> (Nothing, value))+-- > chunk <- R.output (c_duckdb_v2_table_function_exec_get_output_chunk info)+-- > vector <- R.output (c_duckdb_v2_data_chunk_get_vector chunk 0)+-- > writeIntegers vector (maybe [] (\count -> [42 .. 41 + count]) amount)+-- >+-- > withConnection ":memory:" \connection -> do+-- > let count = (R.parameter "count" bigint){R.parameterDefault = Just (R.defaultValue (3 :: Int64))}+-- > R.registerTableFunction connection "managed_scan" [count] tableDefinition+-- > rows <- query_ connection "SELECT amount FROM managed_scan()"+-- > print (map fromOnly rows :: [Int64]) -- [42,43,44]+registerTableFunction :: Connection -> Text -> [FunctionParameter] -> TableCallbacks -> IO ()+registerTableFunction connection name parameters callbacks = do+ validateName name+ withConnectionHandle connection \native -> withConnectionFactory native \factory ->+ registerTableFunctionWith+ (ConnectionTarget connection)+ ( \function -> do+ withIdentifier name $ checked . c_duckdb_v2_table_function_set_name function+ output (c_duckdb_v2_table_function_get_signature function) >>= configureFunctionSignature factory (FunctionSignature parameters Nothing)+ )+ callbacks++-- | Configure and register a managed table function on a connection or extension.+-- The configuration action borrows the builder. Do not set callbacks or user data.+registerTableFunctionWith :: RegistrationTarget -> (DuckDBV2TableFunctionHandle -> IO ()) -> TableCallbacks -> IO ()+registerTableFunctionWith target configure callbacks =+ withTarget+ target+ c_duckdb_v2_table_function_create+ c_duckdb_v2_table_function_destroy+ c_duckdb_v2_connection_register_table_function+ c_duckdb_v2_extension_register_table_function+ \function register ->+ withCallbackResources+ (checkTargetThreads target)+ ( \allocate ->+ installAll+ [ install+ allocate+ DuckDBV2TableFunctionBindCallbackFn+ (DuckDBV2TableFunctionBindCallbackFn_Aux (guardBind (tableBindCallback callbacks)))+ (c_duckdb_v2_table_function_set_bind_callback function)+ , installOptional (tableInitGlobalCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2TableFunctionInitGlobalCallbackFn+ (DuckDBV2TableFunctionInitGlobalCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_table_function_set_init_global_callback function)+ , installOptional (tableInitLocalCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2TableFunctionInitLocalCallbackFn+ (DuckDBV2TableFunctionInitLocalCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_table_function_set_init_local_callback function)+ , install+ allocate+ DuckDBV2TableFunctionExecCallbackFn+ (DuckDBV2TableFunctionExecCallbackFn_Aux (guardContext (tableExecCallback callbacks)))+ (c_duckdb_v2_table_function_set_exec_callback function)+ , installOptional (tableProgressCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2TableFunctionProgressCallbackFn+ (DuckDBV2TableFunctionProgressCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_table_function_set_progress_callback function)+ , installOptional (tableFilterPushdownCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2TableFunctionFilterPushdownCallbackFn+ (DuckDBV2TableFunctionFilterPushdownCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_table_function_set_filter_pushdown_callback function)+ , installOptional (tablePartitionDataCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2TableFunctionPartitionDataCallbackFn+ (DuckDBV2TableFunctionPartitionDataCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_table_function_set_partition_data_callback function)+ , installOptional (tablePartitioningCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2TableFunctionPartitioningCallbackFn+ (DuckDBV2TableFunctionPartitioningCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_table_function_set_partitioning_callback function)+ , installOptional (tableClaimBatchCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2TableFunctionClaimBatchCallbackFn+ (DuckDBV2TableFunctionClaimBatchCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_table_function_set_claim_batch_callback function)+ , installOptional (tableGetBindInfoCallback callbacks) $ \callback ->+ install+ allocate+ DuckDBV2TableFunctionGetBindInfoCallbackFn+ (DuckDBV2TableFunctionGetBindInfoCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_table_function_set_get_bind_info_callback function)+ ]+ )+ (attachOpaque (c_duckdb_v2_table_function_set_user_data function))+ \installCallbacks -> do+ checked (c_duckdb_v2_table_function_set_projection_pushdown function (if tableProjectionPushdown callbacks then 1 else 0))+ configure function+ installCallbacks+ register++-- | Register a cast. Nothing keeps the native explicit-cast cost.+-- The callback must inspect the native cast mode to implement TRY_CAST behavior.+-- An exception in the callback fails the cast with the error type+-- 'ErrorTypeMismatch', whatever the type of the exception. With another error+-- code, DuckDB can invalidate the database. When DuckDB casts a constant+-- through the cast while it binds the query, the error has the type+-- 'ErrorInputInvalid' and the text @Failed to cast value: @ before the text of+-- the exception. See the section /Errors/ of the README and+-- Note [The error code of a failed cast].+-- A target type from SQL @CREATE TYPE name AS base@ is only another name for+-- the base type, so the cast is a cast to the base type. A type from+-- 'registerCustomType' keeps its name. See 'TypeDeclaration'.+--+-- The cast replaces the cast of DuckDB for the pair of types on each+-- connection of the instance. A second registration of the same pair returns+-- with no error and has no effect, because DuckDB keeps the first cast.+--+-- Warning: DuckDB does not always run the callback. Three rules of the+-- optimizer of DuckDB 2.0.0 use the two types of a cast and not the cast+-- function:+--+-- * expression_rewriter (ComparisonSimplificationRule, with+-- BoundCastExpression::CastIsInvertible) casts the constant of a comparison+-- to the type of the column, and compares the column without the cast.+-- * join_filter_pushdown makes a filter of a join from the values of the other+-- side through an integral cast.+-- * statistics_propagation gives min and max from the statistics of the+-- column through the cast.+--+-- In runs with DuckDB 2.0.0, a cast from INTEGER to BIGINT, from BIGINT to a+-- custom type with the base type BIGINT, and from that type to BIGINT, with a+-- callback that changes the value, gave wrong rows and no error. This occurred+-- in a filter, a filter with a parameter, IN, BETWEEN, a comparison of two+-- columns, a join, and min and max. The callback did not run on the column,+-- or, in a comparison of two columns, it ran on one side only. A cast with+-- VARCHAR or BLOB on one side gave the same rows with the optimizer on and off+-- in each of these runs: with 10 rows, and with 300000 rows except min and max+-- of a cast from BLOB to a custom type with the base type BLOB, which did not+-- run. For a custom target type, DuckDB also casts the constant of a+-- comparison through the callback. With a cast from INTEGER to BIGINT, DuckDB+-- also converts each INTEGER literal that must be a BIGINT through the cast:+-- @range(10)@ gave no row, and @LIMIT 5@ gave 10 rows. Use a scalar function+-- for a conversion that changes the value.+--+-- The source type is the first type, and the target type is the second type. The example registers a cast+-- from BIGINT to a custom type. The callback copies each value.+--+-- > withConnection ":memory:" \connection -> do+-- > R.registerCustomType connection "managed_copy" bigint+-- > R.registerCast connection bigint (R.TypeExpression "managed_copy") Nothing \info _ -> do+-- > count <- R.output (c_duckdb_v2_cast_function_exec_get_row_count info)+-- > values <- R.output (c_duckdb_v2_cast_function_exec_get_input info) >>= readIntegers (fromIntegral count)+-- > result <- R.output (c_duckdb_v2_cast_function_exec_get_output info)+-- > writeIntegers result values+-- > rows <- query_ connection "SELECT CAST(i::BIGINT AS managed_copy)::BIGINT FROM range(3) t(i)"+-- > print (map fromOnly rows :: [Int64]) -- [0,1,2]+registerCast+ :: Connection -> TypeDeclaration -> TypeDeclaration -> Maybe Int64 -> ContextCallback DuckDBV2CastFunctionExecInfoHandle -> IO ()+registerCast connection source target implicitCost execute =+ withConnectionHandle connection \native -> withConnectionFactory native \factory ->+ registerCastWith+ (ConnectionTarget connection)+ ( \function -> do+ withTypeDeclaration factory source $ checked . c_duckdb_v2_cast_function_set_source_type function+ withTypeDeclaration factory target $ checked . c_duckdb_v2_cast_function_set_target_type function+ forM_ implicitCost $ checked . c_duckdb_v2_cast_function_set_implicit_cast_cost function+ )+ execute++-- | Configure and register a managed cast on a connection or extension.+-- The configuration action borrows the builder. Do not set callbacks or user data.+-- An exception in the callback fails the cast with the error type+-- 'ErrorTypeMismatch', whatever the type of the exception. With another error+-- code, DuckDB can invalidate the database. See the section /Errors/ of the+-- README and Note [The error code of a failed cast].+registerCastWith+ :: RegistrationTarget -> (DuckDBV2CastFunctionHandle -> IO ()) -> ContextCallback DuckDBV2CastFunctionExecInfoHandle -> IO ()+registerCastWith target configure execute =+ withTarget+ target+ c_duckdb_v2_cast_function_create+ c_duckdb_v2_cast_function_destroy+ c_duckdb_v2_connection_register_cast_function+ c_duckdb_v2_extension_register_cast_function+ \function register ->+ withCallbackResources+ (checkTargetThreads target)+ ( \allocate ->+ install+ allocate+ DuckDBV2CastFunctionExecCallbackFn+ (DuckDBV2CastFunctionExecCallbackFn_Aux \info context err -> runCallbackWith DuckDBV2ErrorTypeMismatch err (execute info context))+ (c_duckdb_v2_cast_function_set_exec_callback function)+ )+ (attachOpaque (c_duckdb_v2_cast_function_set_user_data function))+ \installCallback -> do+ configure function+ installCallback+ register++{- Note [The error code of a failed cast]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+DuckDB 2.0.0 runs each cast as its scalar function "__cast". It marks this+function as fallible only from the source type and the target type+(CreateCastExpression in src/function/scalar/operator/cast.cpp, lines 174 to+190, and BoundCastExpression::CastCanThrow in+src/planner/expression/bound_cast_expression.cpp, lines 337 to 396). It does+not examine a registered cast. Thus DuckDB marks a cast to VARCHAR, a widening+cast such as INTEGER to BIGINT or FLOAT to DOUBLE, and a cast from an ENUM, a+LIST, or a STRUCT to VARCHAR as a function that cannot fail. The cast builder+of the C API has no property that changes this.++A function that cannot fail must not raise an Invalid Input, Out of Range, or+Conversion error. DuckDB changes such an error to an INTERNAL error+(ThrowNonFallibleFunctionError in src/function/scalar_function.cpp, lines 17+to 25). A code with no exception type, for example Api, becomes an Invalid+Input error first. An INTERNAL error during the execution of a query+invalidates the database (FailQueryInternal in+src/main/client_context_execution.cpp, lines 439 to 443). Then each statement+on each connection of the database fails until the database opens again. The+header says for DUCKDB_V2_CAST_MODE_NORMAL: "A conversion failure reported+through the error slot aborts the query." DuckDB does not keep this sentence.++Thus each cast callback of the library writes the code+DUCKDB_V2_ERROR_TYPE_MISMATCH for each failure (runCallbackWith in+"Database.DuckDB.Simple.Callback"). The code of a cast does not depend on the+exception. It replaces the type of each t'SQLError' from the callback, also the+type Conversion of a failed query in the callback. Thus a change of the code+of runCallback does not change the code of a cast. DuckDB rethrows an error of+this type with no change. The statement fails with "Mismatch Type Error: " and+the text of the failure. An explicit transaction is aborted under each+transaction invalidation policy. The code is the same for a cast that DuckDB+marks as fallible.++The 49 codes of 'ErrorType' give these results in the column form and in an+explicit transaction:++- 30 codes, ErrorTypeMismatch included, keep their type and the database, and+ abort an explicit transaction under each policy.+- 7 codes keep an explicit transaction open under the policy+ SYNTACTIC_ERRORS_DO_NOT_INVALIDATE: ErrorQueryBinder, ErrorDatabaseCatalog,+ ErrorResourceConnection, ErrorQueryParameterNotAllowed, ErrorQueryParser,+ ErrorConfigurationPermission, and ErrorResourceInUse.+- 11 codes invalidate the database. ErrorApi, ErrorIOFileNotFound,+ ErrorIOReadFailure, ErrorIOEof, ErrorInputInvalid,+ ErrorInputParameterInvalid, ErrorInputOutOfRange,+ ErrorInputMetadataDeferred, ErrorTypeConversion, and ErrorRuntimeInternal+ give an INTERNAL error. ErrorRuntimeFatal gives a FATAL error.+- ErrorIONetwork keeps the database, but the error has the type ErrorApi.++Do not replace the code without a run of each code in the column form and in+an explicit transaction under each policy.++In DUCKDB_V2_CAST_MODE_TRY, DuckDB discards the error and keeps what the+callback wrote, so TRY_CAST gives NULL for a failed row only when the callback+wrote NULL there. TRY(...) catches only the Invalid Input, Out of Range, and+Conversion errors, so TRY(CAST(...)) raises the error of a failed cast.+-}++-- | Register a custom type with a native base type. Casts are registered separately.+registerCustomType :: Connection -> Text -> TypeDeclaration -> IO ()+registerCustomType connection = registerCustomTypeOn (ConnectionTarget connection)++-- | Register a custom type with a native base type on a connection or a+-- loading extension. The name must not be empty and must not contain NUL.+registerCustomTypeOn :: RegistrationTarget -> Text -> TypeDeclaration -> IO ()+registerCustomTypeOn target name baseType = do+ validateName name+ withRegistrationFactory target \factory ->+ registerCustomTypeWith target \custom -> do+ withIdentifier name $ checked . c_duckdb_v2_custom_type_set_name custom+ withTypeDeclaration factory baseType $ checked . c_duckdb_v2_custom_type_set_base_type custom++-- | Configure and register a custom type on a connection or extension.+-- The configuration action borrows the builder.+registerCustomTypeWith :: RegistrationTarget -> (DuckDBV2CustomTypeHandle -> IO ()) -> IO ()+registerCustomTypeWith target configure =+ withTarget+ target+ c_duckdb_v2_custom_type_create+ c_duckdb_v2_custom_type_destroy+ c_duckdb_v2_connection_register_custom_type+ c_duckdb_v2_extension_register_custom_type+ \custom register ->+ configure custom >> register++-- | Configure and register a macro of a loading extension. The configuration+-- action borrows the builder. It can set the name, the body, the signature, and+-- the documentation. DuckDB keeps copies after this action returns. The V2 API+-- registers a macro only for an extension. A connection target raises an+-- @SQLError@ before a native call. Use SQL @CREATE MACRO@ on a connection.+registerMacroFunctionWith :: RegistrationTarget -> MacroFunctionType -> (DuckDBV2MacroFunctionHandle -> IO ()) -> IO ()+registerMacroFunctionWith target kind configure = case target of+ ConnectionTarget _ -> throwInvalidArgument "the V2 API registers a macro only for an extension; use CREATE MACRO on a connection"+ ExtensionTarget extension context ->+ withContextFactory context \factory ->+ bracket (output (c_duckdb_v2_macro_function_create factory (macroFunctionTypeToNative kind))) (destroy c_duckdb_v2_macro_function_destroy) \function -> do+ configure function+ checked (c_duckdb_v2_extension_register_macro_function extension function)++-- | Register a connection-local replacement scan. It precedes instance scans.+registerReplacementScan :: Connection -> ContextCallback DuckDBV2ReplacementScanInfoHandle -> IO ()+registerReplacementScan connection = registerReplacementScanOn (ConnectionTarget connection)++-- | Register a managed replacement scan on a connection or loading extension.+-- Extension scans belong to the instance.+registerReplacementScanOn :: RegistrationTarget -> ContextCallback DuckDBV2ReplacementScanInfoHandle -> IO ()+registerReplacementScanOn target callback =+ withTarget+ target+ c_duckdb_v2_replacement_scan_create+ c_duckdb_v2_replacement_scan_destroy+ c_duckdb_v2_connection_register_replacement_scan+ c_duckdb_v2_extension_register_replacement_scan+ (configureScan (checkTargetThreads target) callback)++-- | Register an instance-wide replacement scan. It is shared by its connections.+-- Register the scan before a connection runs a query. Registration while another+-- connection binds a query is not thread safe.+registerInstanceReplacementScan :: Instance -> ContextCallback DuckDBV2ReplacementScanInfoHandle -> IO ()+registerInstanceReplacementScan instance' callback = withInstanceHandle instance' $ \native ->+ withInstanceFactory native $ \factory ->+ bracket (output (c_duckdb_v2_replacement_scan_create factory)) (destroy c_duckdb_v2_replacement_scan_destroy) $ \scan ->+ configureScan+ (checkQueryThreads (output (c_duckdb_v2_instance_get_config native)))+ callback+ scan+ (checked (c_duckdb_v2_instance_register_replacement_scan native scan))++-- | Store Haskell state in a native-owned bind, global, local, or init data slot.+-- Pass the corresponding native opaque setter. Do not use registration user data.+-- This helper uses native default equality. Use raw opaque callbacks for custom equality.+-- Transfer a value at most once to each native state slot. In DuckDB 2.0.0, a+-- setter for init data, batch data, global state, or local state does not release+-- a pointer that it replaces. See Note [Callback ownership] in+-- <src/Database.DuckDB.Simple.Callback.html Database.DuckDB.Simple.Callback>. Store an @IORef@ or another mutable state+-- cell when the callback must change its state.+--+-- > R.setCallbackState (c_duckdb_v2_function_bind_set_bind_data input) (amount :: Int64)+setCallbackState :: (Typeable a) => (Ptr DuckDBV2Opaque -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> a -> IO ()+setCallbackState setter = transferCallbackState (attachOpaque setter) . toDyn++-- | Read state previously stored with 'setCallbackState'.+-- Pass the corresponding native pointer getter. A missing state or a state of+-- another type raises an error.+--+-- > amount <- R.getCallbackState (c_duckdb_v2_table_function_init_global_get_bind_data info) :: IO Int64+getCallbackState :: forall a. (Typeable a) => (Ptr (Ptr Void) -> Ptr DuckDBV2ErrorInfoHandle -> IO DuckDBV2Error) -> IO a+getCallbackState getter = do+ stored <- readCallbackState getter+ case fromDynamic stored of+ Just value -> pure value+ Nothing ->+ throwIO+ ( SQLError+ ( "duckdb-simple: callback state has type "+ <> Text.pack (show (dynTypeRep stored))+ <> ", not "+ <> Text.pack (show (typeRep (Proxy @a)))+ )+ Nothing+ Nothing+ Nothing+ )++-- | Construct a type through a borrowed factory, borrow it for an action, and+-- release it on every exit path. The factory can come from a connection or from+-- an extension context.+--+-- > R.withTypeDeclaration factory bigint $ R.checked . c_duckdb_v2_cast_function_set_source_type function+withTypeDeclaration :: DuckDBV2FactoryHandle -> TypeDeclaration -> (DuckDBV2LogicalTypeHandle -> IO a) -> IO a+withTypeDeclaration factory declaration = bracket acquire destroyLogicalType+ where+ acquire = case declaration of+ StructuredType rep -> logicalTypeFromRep factory rep+ TypeExpression sql -> withText sql (output . c_duckdb_v2_factory_create_type_from_text factory)++-- | Add parameters and an optional return type to a borrowed native signature.+-- Types and defaults use the borrowed factory. The factory can come from a+-- connection or from an extension context.+-- Get the signature from the builder:+--+-- > R.output (c_duckdb_v2_scalar_function_get_signature function)+-- > >>= R.configureFunctionSignature factory (R.FunctionSignature [R.parameter "value" bigint] (Just bigint))+configureFunctionSignature :: DuckDBV2FactoryHandle -> FunctionSignature -> DuckDBV2FunctionSignatureHandle -> IO ()+configureFunctionSignature factory FunctionSignature{signatureParameters, signatureReturnType} signature = do+ forM_ signatureParameters $ \FunctionParameter{parameterName, parameterType, parameterDefault, parameterKind} ->+ withIdentifier parameterName $ \name ->+ withTypeDeclaration factory parameterType $ \dtype ->+ case parameterDefault of+ Nothing -> checked (c_duckdb_v2_function_signature_add_parameter signature name dtype nullHandle parameterKind)+ Just (DefaultValue make) -> bracket (make factory) destroyValue $ \native ->+ checked (c_duckdb_v2_function_signature_add_parameter signature name dtype native parameterKind)+ forM_ signatureReturnType $ \dtype ->+ withTypeDeclaration factory dtype $ checked . c_duckdb_v2_function_signature_set_return_type signature++-- | Set the callback of a replacement scan builder, and then run the+-- registration action. The user data of the scan releases the callback pointer.+configureScan+ :: IO () -> ContextCallback DuckDBV2ReplacementScanInfoHandle -> DuckDBV2ReplacementScanHandle -> IO () -> IO ()+configureScan checkRuntime callback scan register =+ withCallbackResources+ checkRuntime+ ( \allocate ->+ install+ allocate+ DuckDBV2ReplacementScanCallbackFn+ (DuckDBV2ReplacementScanCallbackFn_Aux (guardContext callback))+ (c_duckdb_v2_replacement_scan_set_callback scan)+ )+ (attachOpaque (c_duckdb_v2_replacement_scan_set_user_data scan))+ \installCallback -> installCallback >> register++-- | Run a callback that gets no context. An exception becomes a native error.+guardCallback :: (info -> IO ()) -> info -> Ptr DuckDBV2ErrorInfoHandle -> IO ()+guardCallback action info err = runCallback err (action info)++-- | Run a bind callback. An exception becomes a native error.+guardBind+ :: BindCallback result -> DuckDBV2FunctionBindInfoHandle -> result -> DuckDBV2ContextHandle -> Ptr DuckDBV2ErrorInfoHandle -> IO ()+guardBind action info result context err = runCallback err (action info result context)++-- | Reject an empty registration name and a name that contains NUL.+validateName :: Text -> IO ()+validateName name =+ when (Text.null name || Text.any (== '\0') name) $+ throwInvalidArgument "registration name must be nonempty and must not contain NUL"
src/Database/DuckDB/Simple/Variant.hs view
@@ -1,40 +1,97 @@+{-# LANGUAGE LambdaCase #-} {-# LANGUAGE NamedFieldPuns #-}-{-# LANGUAGE PatternSynonyms #-} --- | DuckDB VARIANT values.+-- |+-- Module : Database.DuckDB.Simple.Variant+-- Description : DuckDB VARIANT values. module Database.DuckDB.Simple.Variant (- Variant (..),- variantObject,+ Variant (..),+ variantObject, ) where -import Data.Array (listArray)+import Data.Array (elems, listArray) import qualified Data.Map.Strict as Map import Data.Text (Text)-import Database.DuckDB.FFI (pattern DuckDBTypeVariant)-import Database.DuckDB.Simple.FromField (Field (..), FieldValue (..), FromField (..))-import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructField (..), StructValue (..))-import Database.DuckDB.Simple.Ok (Ok (..))+import qualified Data.Text as Text+import Database.DuckDB.Simple.Element (containsType)+import Database.DuckDB.Simple.FromField (Field (..), FieldValue (..), FromField (..), ResultError (..), returnError)+import Database.DuckDB.Simple.Geometry (RawGeometry (..))+import Database.DuckDB.Simple.LogicalRep (LogicalTypeId (..), LogicalTypeRep (..), StructField (..), StructValue (..), withoutAliases) -{- | A VARIANT value. The payload is a 'FieldValue' with the native type of the-stored value. Arrays are 'FieldList' values. Objects are 'FieldStruct' values-whose fields have the VARIANT type. SQL NULL is 'FieldNull'.-A t'Variant' parameter binds its payload as a VARIANT.--}+-- | A VARIANT value. The payload is a 'FieldValue' with the native type of the+-- stored value. Arrays are 'FieldList' values. Objects are 'FieldStruct' values+-- whose fields have the VARIANT type. Decoded object keys use DuckDB's native+-- lexicographic order. Array elements retain their order. SQL NULL is 'FieldNull'.+-- A t'Variant' parameter binds its payload as a VARIANT. A VARIANT does not keep+-- a UNION, a MAP, an ENUM, or the CRS of a GEOMETRY, so a parameter with such a+-- payload, at any depth, raises an error. Results preserve their native scalar+-- types and nested values through public V2 operations. newtype Variant = Variant {variantPayload :: FieldValue}- deriving (Eq, Show)+ deriving (Eq, Show) -{- | Build the payload of a VARIANT object. Each field has the VARIANT type,-in the order of the entries.--}+-- | Build the payload of a VARIANT object. Each field has the VARIANT type,+-- in the order of the entries. variantObject :: [(Text, FieldValue)] -> FieldValue variantObject entries =- FieldStruct- StructValue- { structValueFields = listArray (0, length entries - 1) [StructField name value | (name, value) <- entries]- , structValueTypes = listArray (0, length entries - 1) [StructField name (LogicalTypeScalar DuckDBTypeVariant) | (name, _) <- entries]- , structValueIndex = Map.fromList (zip (map fst entries) [0 ..])- }+ FieldStruct+ StructValue+ { structValueFields = listArray (0, length entries - 1) [StructField name value | (name, value) <- entries]+ , structValueTypes = listArray (0, length entries - 1) [StructField name (LogicalTypeScalar LogicalTypeIdVariant) | (name, _) <- entries]+ , structValueIndex = Map.fromList (zip (map fst entries) [0 ..])+ } --- | Read any column. A VARIANT column gives its decoded payload.+-- | Reads a value that a VARIANT can hold: the value of a VARIANT column, or a+-- value of another SQL type with that shape. The shape is a scalar that is not+-- an ENUM, a GEOMETRY with no CRS, a LIST of such values, or a STRUCT in the+-- form of a VARIANT object: its fields have the type VARIANT and hold such+-- values, and its keys are in ascending order. A value of another shape fails+-- with 'Incompatible', because the cast of DuckDB to VARIANT gives another+-- value. Write @c::VARIANT@ in SQL for such a column. A column whose type is or+-- contains JSON, at each depth, also fails with 'Incompatible': DuckDB casts+-- JSON to VARIANT by parsing the text, so @c::VARIANT@ gives the parsed value,+-- not the text. See Note [Reads of another SQL type] in+-- "Database.DuckDB.Simple.FromField". instance FromField Variant where- fromField Field{fieldValue} = Ok (Variant fieldValue)+ fromField field@Field{fieldName, fieldValue, fieldType} = do+ logical <- fieldType+ if containsType isJson logical+ then+ returnError+ Incompatible+ field+ (Text.concat [Text.pack "DuckDB casts JSON to VARIANT by parsing the text; write ", column, Text.pack "::VARIANT"])+ else+ if variantShape fieldValue+ then pure (Variant fieldValue)+ else returnError Incompatible field (Text.pack "a VARIANT does not hold a value of this shape; write c::VARIANT")+ where+ isJson (LogicalTypeAlias name (LogicalTypeScalar LogicalTypeIdVarchar)) = name == Text.pack "JSON"+ isJson (LogicalTypeAlias name LogicalTypeCollatedVarchar{}) = name == Text.pack "JSON"+ isJson _ = False+ column = if Text.null fieldName then Text.pack "c" else fieldName++-- | Whether a VARIANT can hold a decoded value. A VARIANT column decodes only to+-- these values: a scalar that is not an ENUM, a GEOMETRY with no CRS, a+-- 'FieldList' of such values, and a 'FieldStruct' in the form of a VARIANT+-- object. Such an object has fields of the type VARIANT that hold such values,+-- and its keys are in ascending order: DuckDB sorts the keys of an object by+-- their UTF-8 bytes. The order of 'Text' is the order of code points, and UTF-8+-- keeps that order in its bytes, so the two orders are the same. Each other+-- value has no VARIANT form: a 'FieldArray', a 'FieldTuple', another+-- 'FieldStruct', a 'FieldEnum', a 'FieldMap', a 'FieldUnion', and a GEOMETRY+-- with a CRS.+variantShape :: FieldValue -> Bool+variantShape = \case+ FieldEnum{} -> False+ FieldGeometry RawGeometry{rawGeometryCRS = Just _} -> False+ FieldMap{} -> False+ FieldUnion{} -> False+ FieldArray{} -> False+ FieldTuple{} -> False+ FieldList values -> all variantShape values+ FieldStruct StructValue{structValueFields, structValueTypes} ->+ let names = map structFieldName (elems structValueTypes)+ in all ((== LogicalTypeScalar LogicalTypeIdVariant) . withoutAliases . structFieldValue) (elems structValueTypes)+ && and (zipWith (<) names (drop 1 names))+ && all (variantShape . structFieldValue) (elems structValueFields)+ _ -> True
src/Database/DuckDB/Simple/VariantCodec.hs view
@@ -1,452 +1,499 @@ {-# LANGUAGE BlockArguments #-} {-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE StrictData #-} -{- | Checked access to DuckDB 1.5's private VARIANT payload.-Only public C handles and vector accessors cross the native boundary.--}+-- |+-- Module : Database.DuckDB.Simple.VariantCodec+-- Description : Lossless VARIANT decoding through public V2 values and SQL type metadata. module Database.DuckDB.Simple.VariantCodec (- decodeVariant,- prepareVariantDecoder,- decodeVariantPayload,+ VariantState,+ variantConnection,+ newVariantState,+ destroyVariantState,+ prepareVariantDecoder, ) where -import Control.Exception (bracket, throwIO)-import Control.Monad (forM, forM_, unless, when)-import Control.Monad.Trans.Class (lift)-import Control.Monad.Trans.State.Strict (StateT, evalStateT, gets, modify')-import Data.Array (Array, bounds, listArray, (!))-import Data.Bits (complement, finiteBitSize, shiftL, (.&.), (.|.))-import Data.ByteString (ByteString)-import qualified Data.ByteString as BS+import Control.Exception (bracket, evaluate, mask_)+import Control.Monad (forM, unless)+import Data.Array (elems, listArray, (!))+import Data.IORef (IORef, modifyIORef', newIORef, readIORef, writeIORef) import qualified Data.IntMap.Strict as IntMap-import qualified Data.IntSet as IntSet+import qualified Data.Map.Strict as Map import qualified Data.Set as Set import Data.Text (Text) import qualified Data.Text as Text-import qualified Data.Text.Encoding as Text-import Data.Word (Word32, Word8)-import Database.DuckDB.FFI-import Database.DuckDB.Simple.Element (bitStringFromBytes, chunkDecodeBlob, chunkIsRowValid, decodeElement)-import Database.DuckDB.Simple.FromField (- BigNum (..),- DecimalValue (..),- FieldValue (..),- RawGeometry (..),- fromBigNumBytes,- )-import Database.DuckDB.Simple.Internal (destroyLogicalType)+import qualified Data.Text.Foreign as TextForeign+import Database.DuckDB.Simple.FromField (FieldValue (..))+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructField (..), StructValue (..))+import Database.DuckDB.Simple.Native+import Database.DuckDB.Simple.V2.LogicalRep (logicalTypeFromRep) import Database.DuckDB.Simple.Variant (variantObject)-import Foreign.C.String (peekCString)-import Foreign.Marshal.Alloc (alloca, allocaBytesAligned)-import Foreign.Marshal.Utils (copyBytes)-import Foreign.Ptr (Ptr, castPtr, nullPtr)-import Foreign.Storable (Storable, peek, peekElemOff, poke, sizeOf)-import Text.Read (readMaybe)+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Marshal.Array (withArray)+import Foreign.Marshal.Utils (with)+import Foreign.Ptr (nullPtr) --- | Raise a codec error before an invalid native operation.-codecError :: String -> IO a-codecError message = throwIO (userError ("duckdb-simple: VARIANT: " <> message))+{- Note [VARIANT decoding]+~~~~~~~~~~~~~~~~~~~~~~~~~~+The decoder copies each VARIANT row of a vector to a FieldValue with the native+type of the stored value. It uses SQL metadata and native casts on a helper+connection. It does not read the private payload format of DuckDB. --- | Convert a checked pure result to an IO result.-checked :: Either String a -> IO a-checked = either codecError pure+- Why SQL. The header commits to no view layout for VARIANT, so the decoder+ reads each row with duckdb_v2_vector_get_value. V2 has no accessor for the+ children of a VARIANT value. SQL functions and native casts give the tag, the+ keys, and the children of a value. --- | Reject native versions outside the supported private-format range.-checkVersion :: IO ()-checkVersion = do- version <- c_duckdb_library_version >>= peekCString- let parts = Text.splitOn "." (Text.pack version)- patch = case parts of- ["v1", "5", p] -> readMaybe (Text.unpack p) :: Maybe Int- _ -> Nothing- unless (maybe False (>= 3) patch) (codecError ("unsupported native version " <> version))+- The metadata statement (metadataStatement). One statement gives the tag+ (variant_typeof) and the object keys (variant_keys) of each value in a+ VARIANT list. The state keeps one prepared statement, so executions reuse+ its bound plan. Its parameter has the fixed type VARIANT[]. The decoder+ reads the borrowed result vector before its chunk closes. This avoids copying the+ complete metadata list into an owned value and then into another vector. --- | Check the byte order and pointer width required by native scalar loads.-checkPlatform :: IO ()-checkPlatform = alloca \ptr -> do- poke ptr (1 :: Word32)- first <- peek (castPtr ptr :: Ptr Word8)- unless (first == 1 && sizeOf (nullPtr :: Ptr ()) == 8) $- codecError "the private codec requires a 64-bit little-endian host"+- The casts. The tag of a scalar is the text of variant_typeof. scalarType+ gives the type text of the exact cast target of each tag. The scalars with+ the same tag share one cast to a list of the target type. A tag with one+ scalar gets a direct cast of that scalar. Each array gets its own cast to a+ VARIANT list. The state keeps each parsed type (castType). --- | Acquire a non-NULL native handle.-nonNull :: String -> IO (Ptr a) -> IO (Ptr a)-nonNull label action = do- ptr <- action- when (ptr == nullPtr) (codecError (label <> " returned NULL"))- pure ptr+- The objects (withObjectChildren). The objects with the same keys share one+ extraction. If the keys can be STRUCT field names (validStructFields), one+ cast to a STRUCT list extracts the children. Otherwise one extraction query+ binds the keys as values, so empty and NUL keys stay exact. The SQL text of+ the query depends only on the key count. The state keeps the parsed+ extraction statement of the last key count (extractionStatement). A group+ with another key count replaces that statement. The state keeps one+ extraction statement only. Thus the memory of the state does not grow with+ the number of key counts in the data. An object without keys needs no+ extraction (queryObjectGroup). --- | Check a logical type's tag.-expectType :: DuckDBType -> DuckDBLogicalType -> IO ()-expectType expected logical = do- actual <- c_duckdb_get_type_id logical- unless (actual == expected) (codecError "unexpected physical child type")+- The levels (decodeValues). The decoder reads the metadata of all values of+ one nesting level together. Then it decodes the children of all nodes of the+ level together, and then it builds the nodes of the level (assemble). --- | Check a STRUCT's names and owned child descriptors.-checkStruct :: DuckDBLogicalType -> [(Text, DuckDBLogicalType -> IO ())] -> IO ()-checkStruct logical fields = do- count <- c_duckdb_struct_type_child_count logical- unless (count == fromIntegral (length fields)) (codecError "unexpected physical child count")- sequence_- [ do- bracket (nonNull "child name" (c_duckdb_struct_type_child_name logical index)) (c_duckdb_free . castPtr) \name -> do- bytes <- BS.packCString name- unless (bytes == Text.encodeUtf8 expectedName) (codecError "unexpected physical child name")- bracket (nonNull "child type" (c_duckdb_struct_type_child_type logical index)) destroyLogicalType checkChild- | (index, (expectedName, checkChild)) <- zip [0 ..] fields- ]+The helper connection. The queries of the decoder do not run on the caller+connection. An explicit stream keeps its result open on the caller connection+between two fetches, and the decoder reads the rows of each chunk while that+result is open (fetchChunk in "Database.DuckDB.Simple.Result"). DuckDB permits+one open result for each connection, and it rejects a second execution. Thus+the queries of the decoder run on a helper connection, so that the decode does+not interrupt the stream. See Note [Connection lock] in+"Database.DuckDB.Simple.Internal". A materialized result closes before its rows+are decoded, but it uses the same reader. A helper connection on the host+instance has its GLOBAL settings, not the caller's SESSION settings. The+metadata query needs an expression depth of 11 when it is prepared. A GLOBAL+max_expression_depth below 11 fails its preparation with a Binder Error. --- | Check a LIST and its owned element descriptor.-checkList :: (DuckDBLogicalType -> IO ()) -> DuckDBLogicalType -> IO ()-checkList checkChild logical = do- expectType DuckDBTypeList logical- bracket (nonNull "list child type" (c_duckdb_list_type_child_type logical)) destroyLogicalType checkChild+VariantState holds the helper connection, the prepared metadata statement, the+parsed extraction statement of the last key count, and the cast types. The+owner of the helper connection destroys the state before the connection. --- | Check the complete, unshredded four-child VARIANT schema.-checkSchema :: DuckDBLogicalType -> IO ()-checkSchema logical = do- expectType DuckDBTypeVariant logical- checkStruct- logical- [ ("keys", checkList (expectType DuckDBTypeVarchar))- ,- ( "children"- , checkList \child -> do- expectType DuckDBTypeStruct child- checkStruct child [("keys_index", expectType DuckDBTypeUInteger), ("values_index", expectType DuckDBTypeUInteger)]- )- ,- ( "values"- , checkList \child -> do- expectType DuckDBTypeStruct child- checkStruct child [("type_id", expectType DuckDBTypeUTinyInt), ("byte_offset", expectType DuckDBTypeUInteger)]- )- , ("data", expectType DuckDBTypeBlob)- ]+- Each caller connection reuses its own decoder. See Note [Concurrent VARIANT+ decoding] in "Database.DuckDB.Simple.Internal". --- | Check an element index before native pointer arithmetic.-checkIndex :: Int -> Int -> IO ()-checkIndex width index =- when (index < 0 || index > maxBound `div` width) (codecError "native element index exceeds Int range")+- A connection's COPY TO reader uses a helper connection of that instance+ for each vector. An extension's COPY TO reader and typed argument readers+ use a private instance. Scalar and aggregate functions reuse that instance+ across vectors. See Note [VARIANT reads inside callbacks] in+ "Database.DuckDB.Simple.Materialize". --- | Prepare validity access for a chunk. The caller checks row bounds.-prepareValidity :: DuckDBVector -> IO (Int -> IO Bool)-prepareValidity vector = do- validity <- c_duckdb_vector_get_validity vector- pure (chunkIsRowValid validity . fromIntegral)+TODO: When V2 exposes VARIANT tags and children directly, use those accessors+to replace the metadata and object extraction queries. Preserve exact scalar+types and object keys, including empty keys and keys containing NUL. --- | Borrow a fixed-width buffer until its chunk is destroyed.-prepareElementReader :: (Storable a) => DuckDBVector -> IO (Int -> IO a)-prepareElementReader vector = do- ptr <- c_duckdb_vector_get_data vector- valid <- prepareValidity vector- pure (readAt valid (castPtr ptr))- where- readAt :: (Storable a) => (Int -> IO Bool) -> Ptr a -> Int -> IO a- readAt valid ptr index = do- checkIndex (sizeOfElement ptr) index- when (ptr == nullPtr) (codecError "NULL vector data")- present <- valid index- unless present (codecError "NULL physical payload element")- peekElemOff ptr index- sizeOfElement :: (Storable a) => Ptr a -> Int- sizeOfElement ptr = sizeOf (undefined `asTypeOf` element ptr)- element :: Ptr a -> a- element _ = undefined+The decoder casts with the factory of the instance (variantCasts). It decodes+each value that contains no VARIANT, such as a cast result, with the same+factory. See Note [User casts and the casts of the library] in+"Database.DuckDB.Simple.Materialize".+-} --- | Borrow a string buffer and copy each requested value into Haskell memory.-prepareBytesReader :: DuckDBVector -> IO (Int -> IO ByteString)-prepareBytesReader vector = do- base <- c_duckdb_vector_get_data vector- valid <- prepareValidity vector- pure \index -> do- checkIndex 16 index- present <- valid index- unless present (codecError "NULL physical string element")- when (base == nullPtr) (codecError "NULL string vector data")- chunkDecodeBlob base (fromIntegral index)+-- | The helper connection of the VARIANT decoder, and the native objects that+-- the decoder keeps for it: the prepared metadata statement, the parsed+-- extraction statement of the last key count, and the types for casts and+-- VARIANT lists. The state borrows the connection. The owner destroys+-- the state with 'destroyVariantState' before it destroys the connection. Use+-- the state from one thread at a time. See Note [VARIANT decoding].+data VariantState = VariantState+ { variantConnection :: DuckDBV2ConnectionHandle+ -- ^ The helper on which the decoder runs its queries.+ , variantCasts :: DuckDBV2FactoryHandle+ -- ^ The factory of the instance of the helper connection. The decoder casts+ -- with it.+ , variantStatement :: IORef (Maybe DuckDBV2PreparedStatementHandle)+ , variantExtraction :: IORef (Maybe (Int, DuckDBV2SqlStatementHandle))+ , variantTypes :: CastTypes+ } --- | Convert a nonnegative bounded integer to Int.-checkedInt :: Integer -> IO Int-checkedInt n- | n < 0 || n > toInteger (maxBound :: Int) = codecError "payload size exceeds Int range"- | otherwise = pure (fromInteger n)+-- | Make an empty state for a helper connection and the factory of its+-- instance. This allocates no native object.+newVariantState :: DuckDBV2ConnectionHandle -> DuckDBV2FactoryHandle -> IO VariantState+newVariantState connection casts = VariantState connection casts <$> newIORef Nothing <*> newIORef Nothing <*> newIORef Map.empty --- | Prepare LIST bounds checks against the chunk's fixed child size.-prepareListBounds :: DuckDBVector -> IO (Int -> IO (Int, Int))-prepareListBounds vector = do- readEntry <- prepareElementReader vector- size <- c_duckdb_list_vector_get_size vector- pure \row -> do- DuckDBListEntry offset count <- readEntry row- unless (offset <= size && count <= size - offset) (codecError "LIST bounds exceed child size")- start <- checkedInt (toInteger offset)- len <- checkedInt (toInteger count)- _ <- checkedInt (toInteger offset + toInteger count)- pure (start, len)+-- | Destroy the prepared metadata statement, the extraction statement, and+-- every cached type. Release every item even if one destructor fails, and raise the first+-- error after the last release. The state is empty afterwards, so a second call+-- does nothing.+destroyVariantState :: VariantState -> IO ()+destroyVariantState VariantState{variantStatement, variantExtraction, variantTypes} = mask_ do+ statement <- readIORef variantStatement+ extraction <- readIORef variantExtraction+ types <- readIORef variantTypes+ writeIORef variantStatement Nothing+ writeIORef variantExtraction Nothing+ writeIORef variantTypes Map.empty+ releaseItems+ id+ ( map (destroy c_duckdb_v2_prepared_statement_destroy) (maybe [] pure statement)+ <> map (destroy c_duckdb_v2_sql_statement_destroy . snd) (maybe [] pure extraction)+ <> map destroyLogicalType (Map.elems types)+ ) --- | Decode one row while its flattened result chunk remains alive.-decodeVariant :: DuckDBVector -> Int -> IO FieldValue-decodeVariant vector row = prepareVariantDecoder vector >>= ($ row)+-- | Copy a vector before returning its row reader. The connection of the state+-- is a dedicated helper connection. It must remain exclusively borrowed until+-- this function returns. The callbacks decode a value, or the first row of a+-- borrowed vector, that contains no VARIANT. They use the factory they get and+-- complete all native reads before returning. Each level batches type and key+-- metadata. See Note [VARIANT decoding].+prepareVariantDecoder+ :: (DuckDBV2FactoryHandle -> DuckDBV2ValueHandle -> IO FieldValue)+ -> (DuckDBV2FactoryHandle -> DuckDBV2VectorHandle -> IO FieldValue)+ -> VariantState+ -> DuckDBV2VectorHandle+ -> IO (Int -> IO FieldValue)+prepareVariantDecoder plain plainVector state vector = do+ count <- output (c_duckdb_v2_vector_get_size vector)+ bracket (acquireItems destroyValue [output (c_duckdb_v2_vector_get_value vector i) | i <- indices count]) (releaseItems destroyValue) \values -> do+ decoded <- decodeValues plain plainVector state values+ let rows = listArray (0, length decoded - 1) decoded+ -- Give the element, not a thunk that keeps the array alive.+ pure (evaluate . (rows !)) -{- | Check the format once and borrow buffers for a flattened result chunk.-The returned reader must not outlive the chunk. The caller supplies row indices-within that chunk. DuckDB result Fetch flattens nested vectors in 1.5.-Each read copies its payload into Haskell memory, including referenced keys.--}-prepareVariantDecoder :: DuckDBVector -> IO (Int -> IO FieldValue)-prepareVariantDecoder vector = do- checkVersion- checkPlatform- when (vector == nullPtr) (codecError "NULL vector")- bracket (nonNull "vector type" (c_duckdb_vector_get_column_type vector)) destroyLogicalType checkSchema- valid <- prepareValidity vector- keys <- child vector 0- children <- child vector 1- values <- child vector 2- blob <- child vector 3- keyBounds <- prepareListBounds keys- childBounds <- prepareListBounds children- valueBounds <- prepareListBounds values- keyVector <- nonNull "keys vector" (c_duckdb_list_vector_get_child keys)- childVector <- nonNull "children vector" (c_duckdb_list_vector_get_child children)- valueVector <- nonNull "values vector" (c_duckdb_list_vector_get_child values)- keyIndices <- child childVector 0- valueIndices <- child childVector 1- tags <- child valueVector 0- offsets <- child valueVector 1- readTag <- prepareElementReader tags- readOffset <- prepareElementReader offsets- hasKey <- prepareValidity keyIndices- readKey <- prepareElementReader keyIndices- readValue <- prepareElementReader valueIndices- readKeyBytes <- prepareBytesReader keyVector- readBlob <- prepareBytesReader blob- pure \row -> do- checkIndex 16 row- present <- valid row- if not present- then pure FieldNull- else do- (keyStart, keyCount) <- keyBounds row- (childStart, childCount) <- childBounds row- (valueStart, valueCount) <- valueBounds row- valueRows <- forM [valueStart .. valueStart + valueCount - 1] \index ->- (,) <$> readTag index <*> readOffset index- childRows <- forM [childStart .. childStart + childCount - 1] \index -> do- keyed <- hasKey index- key <- if keyed then Just <$> readKey index else pure Nothing- value <- readValue index- when (toInteger value >= toInteger valueCount) (codecError "child value index out of bounds")- case key of- Just k | toInteger k >= toInteger keyCount -> codecError "child key index out of bounds"- _ -> pure ()- pure (key, value)- let usedKeys = Set.toList (Set.fromList [k | (Just k, _) <- childRows])- keyRows <- forM usedKeys \index -> do- bytes <- readKeyBytes (keyStart + fromIntegral index)- text <- checked (either (Left . show) Right (Text.decodeUtf8' bytes))- pure (index, text)- bytes <- readBlob row- decodeVariantPayload valueRows childRows keyRows bytes+-- | A decoded node owns the child values needed at the next nesting level.+data VariantNode+ = ScalarNode FieldValue+ | ListNode [DuckDBV2ValueHandle]+ | ObjectNode [(Text, DuckDBV2ValueHandle)]++-- | Decode all siblings together before reconstructing their parent nodes.+-- See Note [VARIANT decoding].+decodeValues+ :: (DuckDBV2FactoryHandle -> DuckDBV2ValueHandle -> IO FieldValue)+ -> (DuckDBV2FactoryHandle -> DuckDBV2VectorHandle -> IO FieldValue)+ -> VariantState+ -> [DuckDBV2ValueHandle]+ -> IO [FieldValue]+decodeValues _ _ _ [] = pure []+decodeValues plain plainVector state values = do+ metadata <- queryMetadata plainVector state values+ unless (length metadata == length values) (codecError "metadata query returned an unexpected number of nodes")+ let indexed = zip3 [0 ..] metadata values+ grouped key select = Map.toList (Map.fromListWith (++) [(key node, [(index, value)]) | (index, node@(tag, _), value) <- indexed, select tag])+ nulls = [(index, ScalarNode FieldNull) | (index, (tag, _), _) <- indexed, isNull tag]+ releaseIndexed = releaseNode . snd+ scalars <- mapM (uncurry decodeScalars) (grouped fst (\tag -> not (isNull tag || isArray tag || isObject tag)))+ bracket+ (acquireItems releaseIndexed [(,) index <$> decodeArray value | (index, (tag, _), value) <- indexed, isArray tag])+ (releaseItems releaseIndexed)+ \arrays ->+ bracket+ (acquireItems (releaseItems releaseIndexed) [queryObjectGroup state keys rows | (keys, rows) <- grouped snd isObject])+ (releaseItems (releaseItems releaseIndexed))+ \objectGroups -> do+ let indexedNodes = IntMap.fromList (nulls <> concat scalars <> arrays <> concat objectGroups)+ nodes <- forM [0 .. length values - 1] \index ->+ maybe (codecError "object extraction omitted a node") pure (IntMap.lookup index indexedNodes)+ children <- decodeValues plain plainVector state (concatMap nodeChildren nodes)+ assemble nodes children where- child parent index = nonNull "STRUCT vector child" (c_duckdb_struct_vector_get_child parent index)+ casts = variantCasts state+ isNull = (== "VARIANT_NULL")+ isArray = Text.isPrefixOf "ARRAY("+ isObject = Text.isPrefixOf "OBJECT("+ decodeArray value =+ withCast state value "VARIANT[]" \list -> do+ count <- output (c_duckdb_v2_value_get_child_count list)+ ListNode <$> acquireItems destroyValue [output (c_duckdb_v2_value_get_child list i) | i <- indices count]+ -- Cast the scalars of one tag as one list. The cast keeps the exact type.+ -- Cast a single scalar directly, because a list needs more native calls.+ decodeScalars tag rows = do+ target <- scalarType tag+ case rows of+ [(index, value)] -> withCast state value target \cast -> do+ -- The cast fixes the type. Read these payloads without an owned vector.+ decoded <- case tag of+ "INT64" -> FieldInt64 <$> output (c_duckdb_v2_value_get_bigint cast)+ "UINT64" -> FieldWord64 <$> output (c_duckdb_v2_value_get_ubigint cast)+ _ -> plain casts cast+ pure [(index, ScalarNode decoded)]+ _ ->+ withVariantList state (map snd rows) \list ->+ withCast state list (target <> "[]") \cast ->+ plain casts cast >>= \case+ FieldList decoded | length decoded == length rows -> pure (zip (map fst rows) (map ScalarNode decoded))+ _ -> codecError "scalar cast returned an unexpected value" -{- | Decode copied 1.5 payload data for one non-NULL row.-Values are (tag, byte offset). Children are (optional key index, value index).-Keys pair an index with its text. Indices are relative to this row's LISTs.-The root value has index zero. This helper checks all metadata before decoding.-It raises an error for cycles. Shared values use a memo table.--}-decodeVariantPayload :: [(Word8, Word32)] -> [(Maybe Word32, Word32)] -> [(Word32, Text)] -> ByteString -> IO FieldValue-decodeVariantPayload valueRows childRows keyRows bytes = do- when (finiteBitSize (0 :: Int) < 64) (codecError "the private codec requires a 64-bit host")- let values = listArray (0, length valueRows - 1) valueRows- children = listArray (0, length childRows - 1) childRows- keys = IntMap.fromList [(fromIntegral k, t) | (k, t) <- keyRows]- valueCount = length valueRows- childCount = length childRows- when (null valueRows) (codecError "missing root value")- unless (IntMap.size keys == length keyRows) (codecError "duplicate key dictionary index")- forM_ valueRows \(tag, offset) -> do- when (tag > 33) (codecError "unknown payload tag")- when (toInteger offset > toInteger (BS.length bytes)) (codecError "byte offset exceeds data size")- forM_ childRows \(key, value) -> do- when (toInteger value >= toInteger valueCount) (codecError "child value index out of bounds")- case key of- Just k -> unless (IntMap.member (fromIntegral k) keys) (codecError "missing child key")- Nothing -> pure ()- evalStateT (visit values children keys childCount IntSet.empty 0) IntMap.empty+-- | Extract the children of objects with one key layout. One native cast or+-- one extraction query extracts the children of all objects. Each object gets one+-- child for each key, in key order. Check the object count and the child count of+-- each object.+queryObjectGroup :: VariantState -> [Text] -> [(Int, DuckDBV2ValueHandle)] -> IO [(Int, VariantNode)]+queryObjectGroup _ [] rows = pure [(index, ObjectNode []) | (index, _) <- rows]+queryObjectGroup state keys rows =+ withVariantList state (map snd rows) \list ->+ withObjectChildren state keys list \objects -> do+ count <- output (c_duckdb_v2_value_get_child_count objects)+ unless (toInteger count == toInteger (length rows)) (codecError "object extraction returned an unexpected row count")+ acquireItems+ (releaseNode . snd)+ [ bracket (output (c_duckdb_v2_value_get_child objects i)) destroyValue \children -> do+ childCount <- output (c_duckdb_v2_value_get_child_count children)+ unless (toInteger childCount == toInteger (length keys)) (codecError "object extraction returned an unexpected child count")+ owned <- acquireItems destroyValue [output (c_duckdb_v2_value_get_child children j) | j <- indices childCount]+ pure (index, ObjectNode (zip keys owned))+ | (i, (index, _)) <- zip [0 ..] rows+ ]++-- | Borrow one value for each object in a VARIANT list. Each value has one+-- child for each key, in key order. If the keys are valid STRUCT field names, a+-- native cast to a STRUCT list makes the values. Otherwise an extraction query+-- makes them. The binder needs constant extraction keys. The query binds each+-- key once and applies the extractions to all objects. The bound keys can be+-- empty or contain NUL. The state keeps the parsed extraction statement of the+-- last key count ('extractionStatement'). A group with another key count+-- replaces it. V2 has no accessor for the children of a VARIANT value. The+-- STRUCT cast is the fast path. The query is the only exact path for the other+-- key lists. See Note [VARIANT decoding].+withObjectChildren :: VariantState -> [Text] -> DuckDBV2ValueHandle -> (DuckDBV2ValueHandle -> IO a) -> IO a+withObjectChildren state@VariantState{variantConnection = connection} keys list action+ | validStructFields keys =+ -- The cast is lenient. It matches a key and a field name without+ -- regard to ASCII case. It gives NULL for a missing key. It ignores an+ -- extra key. Thus a wrong field list does not cause an error. The+ -- result is exact only because the fields are the keys of each object+ -- and 'validStructFields' accepts them. Do not relax that predicate.+ withConnectionFactory connection \factory ->+ bracket (logicalTypeFromRep factory structList) destroyLogicalType \logical ->+ bracket (output (c_duckdb_v2_value_cast (variantCasts state) list logical)) destroyValue action+ | otherwise =+ withConnectionFactory connection \factory ->+ bracket+ (acquireItems destroyValue [withText key (output . c_duckdb_v2_value_create_varchar factory) | key <- keys])+ (releaseItems destroyValue)+ \keyValues -> do+ statement <- extractionStatement state (length keys)+ withStatementValue connection statement (list : keyValues) action where- visit :: Array Int (Word8, Word32) -> Array Int (Maybe Word32, Word32) -> IntMap.IntMap Text -> Int -> IntSet.IntSet -> Int -> StateT (IntMap.IntMap FieldValue) IO FieldValue- visit values children keys childCount ancestors index = do- when (IntSet.member index ancestors) (lift (codecError "cyclic child reference"))- cached <- gets (IntMap.lookup index)- case cached of- Just result -> pure result- Nothing -> do- (tag, offset) <- lift (checked (arrayElement values index))- let payload = BS.drop (fromIntegral offset) bytes- result <- case tag of- 29 -> nested True payload- 30 -> nested False payload- _ -> lift (decodeScalar tag payload)- modify' (IntMap.insert index result)- pure result- where- nested object payload = do- (count, rest) <- lift (checked (readVarint payload))- start <- if count == 0 then pure 0 else fst <$> lift (checked (readVarint rest))- unless (toInteger start + toInteger count <= toInteger childCount) $- lift (codecError "container child range out of bounds")- entries <- forM [fromIntegral start .. fromIntegral start + fromIntegral count - 1] \childIndex -> do- (key, childIndexValue) <- lift (checked (arrayElement children childIndex))- name <- case (object, key) of- (True, Just k) -> case IntMap.lookup (fromIntegral k) keys of- Just text -> pure text- Nothing -> lift (codecError "missing object key")- (False, Nothing) -> pure Text.empty- _ -> lift (codecError "container key validity does not match its tag")- item <- visit values children keys childCount (IntSet.insert index ancestors) (fromIntegral childIndexValue)- pure (name, item)- if object- then do- unless (Set.size (Set.fromList (map fst entries)) == length entries) $- lift (codecError "duplicate object key")- pure (variantObject entries)- else pure (FieldList (map snd entries))+ structList =+ LogicalTypeList+ (LogicalTypeStruct (listArray (0, length keys - 1) [StructField key (LogicalTypeScalar LogicalTypeIdVariant) | key <- keys])) --- | Read an array element after checking both bounds.-arrayElement :: Array Int a -> Int -> Either String a-arrayElement array index- | index < lower || index > upper = Left "value index out of bounds"- | otherwise = Right (array ! index)+-- | Tell if the keys can be the field names of one STRUCT type. A name must+-- not be empty and must not contain NUL. Two names must not be equal after ASCII+-- case folding.+validStructFields :: [Text] -> Bool+validStructFields keys =+ all (\key -> not (Text.null key || Text.any (== '\0') key)) keys+ && Set.size (Set.fromList (map foldIdentifier keys)) == length keys++-- | Rebuild a level from its copied children in their native order.+assemble :: [VariantNode] -> [FieldValue] -> IO [FieldValue]+assemble [] [] = pure []+assemble [] _ = codecError "decoded child count exceeds its parent count"+assemble (node : rest) children = do+ let (count, build) = case node of+ ScalarNode value -> (0, const value)+ ListNode values -> (length values, FieldList)+ ObjectNode entries -> (length entries, variantObject . zip (map fst entries))+ (ownChildren, remaining) = splitAt count children+ unless (length ownChildren == count) (codecError "decoded child count is smaller than its parent count")+ -- Build the constructor of this node now. This reads no native value and+ -- cannot fail.+ value <- evaluate (build ownChildren)+ (value :) <$> assemble rest remaining++-- | Native types by type text, for casts and VARIANT lists. The map is small+-- and bounded. It holds the types of the fixed scalar tags and their list forms,+-- "VARIANT" and "VARIANT[]", and one DECIMAL type and its list form for each+-- width and scale. It never holds a STRUCT type.+type CastTypes = IORef (Map.Map Text DuckDBV2LogicalTypeHandle)++-- | Borrow the native cast result while copying its payload. The cast uses the+-- factory of the instance. See Note [User casts and the casts of the library]+-- in "Database.DuckDB.Simple.Materialize".+withCast :: VariantState -> DuckDBV2ValueHandle -> Text -> (DuckDBV2ValueHandle -> IO a) -> IO a+withCast state value target action = do+ logical <- castType state target+ bracket (output (c_duckdb_v2_value_cast (variantCasts state) value logical)) destroyValue action++-- | Borrow a type from the cache. Parse the type text on its first use.+castType :: VariantState -> Text -> IO DuckDBV2LogicalTypeHandle+castType VariantState{variantConnection = connection, variantTypes = castTypes} target =+ Map.lookup target <$> readIORef castTypes >>= \case+ Just cached -> pure cached+ Nothing -> mask_ do+ created <- withText target (\text -> factoryOutput connection (\factory -> c_duckdb_v2_factory_create_type_from_text factory text))+ modifyIORef' castTypes (Map.insert target created)+ pure created++-- | Borrow the prepared metadata statement. Prepare it on first use and keep+-- it in the state. An execution does not consume it. See Note [VARIANT decoding].+metadataStatement :: VariantState -> IO DuckDBV2PreparedStatementHandle+metadataStatement VariantState{variantConnection = connection, variantStatement} =+ readIORef variantStatement >>= \case+ Just statement -> pure statement+ Nothing -> mask_ $ parseKept connection sql \parsed ->+ bracket (pure parsed) (destroy c_duckdb_v2_sql_statement_destroy) \statement -> do+ prepared <- output (c_duckdb_v2_prepared_statement_create connection statement 0)+ writeIORef variantStatement (Just prepared)+ pure prepared where- (lower, upper) = bounds array+ -- See Note [SQL text of the library] in "Database.DuckDB.Simple.Native".+ sql =+ "SELECT system.main.list_transform(?::VARIANT[], lambda value: system.main.struct_pack(tag := system.main.variant_typeof(value), keys := system.main.variant_keys(value))) AS metadata" --- | Read an unsigned base-128 uint32 without overflow or truncation.-readVarint :: ByteString -> Either String (Word32, ByteString)-readVarint = go 0 0+-- | Borrow the extraction statement for a key count. The state keeps the+-- statement of the last key count. A call with another key count parses a new+-- statement and destroys the old statement. The statement is valid until the+-- next call of this function or 'destroyVariantState'. See+-- Note [VARIANT decoding].+extractionStatement :: VariantState -> Int -> IO DuckDBV2SqlStatementHandle+extractionStatement VariantState{variantConnection = connection, variantExtraction} count =+ readIORef variantExtraction >>= \case+ Just (kept, statement) | kept == count -> pure statement+ previous ->+ -- Store the new statement before the old statement is destroyed. If the+ -- parse fails, the slot keeps the old statement. If the destroy of the+ -- old statement fails, the error propagates. Then the slot holds the new+ -- statement. Nothing destroys the old statement again, as in+ -- 'destroyVariantState'.+ mask_ $ parseKept connection sql \statement -> do+ writeIORef variantExtraction (Just (count, statement))+ mapM_ (destroy c_duckdb_v2_sql_statement_destroy . snd) previous+ pure statement where- go shift value input = case BS.uncons input of- Nothing -> Left "truncated varint"- Just (byte, rest)- | shift == 28 && byte > 15 -> Left "varint exceeds uint32"- | otherwise ->- let result = value .|. (fromIntegral (byte .&. 127) `shiftL` shift)- in if byte .&. 128 == 0 then Right (result, rest) else go (shift + 7) result rest+ -- See Note [SQL text of the library] in "Database.DuckDB.Simple.Native".+ sql =+ "SELECT system.main.list_transform($1::VARIANT[], lambda value: system.main.list_value(" <> Text.intercalate "," extracts <> ")) AS payload"+ extracts = ["system.main.variant_extract(value,$" <> Text.pack (show index) <> "::VARCHAR)" | index <- [2 .. count + 1]] --- | Read a bounded fixed-size payload in little-endian order.-readUnsigned :: Int -> ByteString -> Either String (Integer, ByteString)-readUnsigned count bytes- | BS.length bytes < count = Left "truncated scalar payload"- | otherwise =- let (part, rest) = BS.splitAt count bytes- in Right (BS.foldr (\byte value -> value `shiftL` 8 .|. toInteger byte) 0 part, rest)+-- | Parse trusted SQL and give its first statement to the action before the+-- iterator closes. The action must keep or destroy the statement, also when+-- the iterator destroy fails. Call this function under 'mask_'.+parseKept :: DuckDBV2ConnectionHandle -> Text -> (DuckDBV2SqlStatementHandle -> IO a) -> IO a+parseKept connection sql keep =+ TextForeign.withCString sql \source ->+ bracket (output (c_duckdb_v2_parse_sql connection (ConstPtr source))) (destroy c_duckdb_v2_statement_iterator_destroy) \iterator -> do+ statement <- output (c_duckdb_v2_statement_iterator_next iterator)+ keep statement --- | Read a two's-complement integer with a checked payload length.-readSigned :: Int -> ByteString -> Either String Integer-readSigned count bytes = do- (value, _) <- readUnsigned count bytes- pure (if value >= 2 ^ (count * 8 - 1) then value - 2 ^ (count * 8) else value)+-- | Read exact scalar tags and object keys without serializing payloads.+queryMetadata+ :: (DuckDBV2FactoryHandle -> DuckDBV2VectorHandle -> IO FieldValue)+ -> VariantState+ -> [DuckDBV2ValueHandle]+ -> IO [(Text, [Text])]+queryMetadata plainVector state values = withVariantList state values \list -> do+ statement <- metadataStatement state+ with list \items ->+ withResultVector+ (output (c_duckdb_v2_prepared_statement_execute statement (ConstPtr nullPtr) (ConstPtr items) 1))+ \vector ->+ plainVector (variantCasts state) vector >>= maybe (codecError "metadata query returned an unexpected value") pure . metadataNodes --- | Read a length-prefixed string or blob.-readString :: ByteString -> Either String ByteString-readString bytes = do- (count, rest) <- readVarint bytes- if toInteger count > toInteger (BS.length rest)- then Left "string length exceeds payload size"- else Right (BS.take (fromIntegral count) rest)+-- | Read the tag and the object keys of each metadata node. A NULL key list+-- gives no keys. Return 'Nothing' for an unexpected shape.+metadataNodes :: FieldValue -> Maybe [(Text, [Text])]+metadataNodes = \case+ FieldList nodes -> traverse node nodes+ _ -> Nothing+ where+ node = \case+ FieldStruct struct -> case elems (structValueFields struct) of+ [StructField "tag" (FieldText tag), StructField "keys" keys] -> (,) tag <$> keyList keys+ _ -> Nothing+ _ -> Nothing+ keyList = \case+ FieldNull -> Just []+ FieldList keys -> traverse keyText keys+ _ -> Nothing+ keyText = \case+ FieldText key -> Just key+ _ -> Nothing --- | Decode the scalar tags from VariantLogicalType in DuckDB 1.5.-decodeScalar :: Word8 -> ByteString -> IO FieldValue-decodeScalar tag bytes = case tag of- 0 -> pure FieldNull- 1 -> pure (FieldBool True)- 2 -> pure (FieldBool False)- 15 -> checked do- (precision, rest) <- readVarint bytes- (scale, digits) <- readVarint rest- when (precision < 1 || precision > 38 || scale > precision) (Left "invalid decimal metadata")- let count = if precision <= 4 then 2 else if precision <= 9 then 4 else if precision <= 18 then 8 else 16- value <- readSigned count digits- when (abs value >= 10 ^ precision) (Left "unscaled decimal exceeds its precision")- pure (FieldDecimal (DecimalValue (fromIntegral precision) (fromIntegral scale) value))- 16 -> checked do- string <- readString bytes- FieldText <$> either (Left . show) Right (Text.decodeUtf8' string)- 17 -> FieldBlob <$> checked (readString bytes)- 31 -> FieldBigNum . BigNum <$> checked (readString bytes >>= decodeBigNum)- 32 -> checked do- bitBytes <- readString bytes- case BS.unpack (BS.take 2 bitBytes) of- [padding, first] -> do- when (padding > 7) (Left "BIT padding exceeds seven")- let maskBits = paddingMask padding- unless (first .&. maskBits == maskBits) (Left "invalid native BIT padding")- pure (FieldBit (bitStringFromBytes bitBytes))- _ -> Left "BIT requires a padding byte and nonempty data"- 33 -> FieldGeometry . (`RawGeometry` Nothing) <$> checked (readString bytes)- _ -> case fixedWidthTag tag of- Just (dtype, size) -> decodeFixedWidth dtype size bytes- Nothing -> codecError "unknown scalar tag"+-- | Borrow an owned VARIANT list that copies all supplied values.+withVariantList :: VariantState -> [DuckDBV2ValueHandle] -> (DuckDBV2ValueHandle -> IO a) -> IO a+withVariantList state values action = do+ logical <- castType state "VARIANT"+ withConnectionFactory (variantConnection state) \factory ->+ withArray values \items ->+ bracket (output (c_duckdb_v2_value_create_list factory logical (ConstPtr items) (fromIntegral (length values)))) destroyValue action -{- | The type and the payload size of each fixed-width scalar tag. These-payloads have the memory layout of one vector element of the type.--}-fixedWidthTag :: Word8 -> Maybe (DuckDBType, Int)-fixedWidthTag = \case- 3 -> Just (DuckDBTypeTinyInt, 1)- 4 -> Just (DuckDBTypeSmallInt, 2)- 5 -> Just (DuckDBTypeInteger, 4)- 6 -> Just (DuckDBTypeBigInt, 8)- 7 -> Just (DuckDBTypeHugeInt, 16)- 8 -> Just (DuckDBTypeUTinyInt, 1)- 9 -> Just (DuckDBTypeUSmallInt, 2)- 10 -> Just (DuckDBTypeUInteger, 4)- 11 -> Just (DuckDBTypeUBigInt, 8)- 12 -> Just (DuckDBTypeUHugeInt, 16)- 13 -> Just (DuckDBTypeFloat, 4)- 14 -> Just (DuckDBTypeDouble, 8)- 18 -> Just (DuckDBTypeUUID, 16)- 19 -> Just (DuckDBTypeDate, 4)- 20 -> Just (DuckDBTypeTime, 8)- 21 -> Just (DuckDBTypeTimeNs, 8)- 22 -> Just (DuckDBTypeTimestampS, 8)- 23 -> Just (DuckDBTypeTimestampMs, 8)- 24 -> Just (DuckDBTypeTimestamp, 8)- 25 -> Just (DuckDBTypeTimestampNs, 8)- 26 -> Just (DuckDBTypeTimeTz, 8)- 27 -> Just (DuckDBTypeTimestampTz, 8)- 28 -> Just (DuckDBTypeInterval, 16)- _ -> Nothing+-- | Borrow the first result value while its statement result stays alive.+withStatementValue+ :: DuckDBV2ConnectionHandle -> DuckDBV2SqlStatementHandle -> [DuckDBV2ValueHandle] -> (DuckDBV2ValueHandle -> IO a) -> IO a+withStatementValue connection statement parameters action =+ withArray parameters \items ->+ withResultVector+ (output (c_duckdb_v2_statement_execute connection statement (ConstPtr nullPtr) (ConstPtr items) (fromIntegral (length parameters))))+ \vector -> bracket (output (c_duckdb_v2_vector_get_value vector 0)) destroyValue action --- | Copy a fixed-width payload to aligned memory and decode it as one element.-decodeFixedWidth :: DuckDBType -> Int -> ByteString -> IO FieldValue-decodeFixedWidth dtype size bytes = do- when (BS.length bytes < size) (codecError "truncated scalar payload")- allocaBytesAligned size 16 \buffer -> do- BS.useAsCStringLen bytes \(source, _) -> copyBytes buffer (castPtr source) size- decodeElement dtype (castPtr buffer) 0+-- | Borrow the first result vector while its result and chunk stay alive.+-- Complete native reads before the action returns.+withResultVector :: IO DuckDBV2ResultHandle -> (DuckDBV2VectorHandle -> IO a) -> IO a+withResultVector execute action =+ bracket execute (destroy c_duckdb_v2_result_destroy) \result ->+ bracket (output (c_duckdb_v2_result_fetch_chunk result)) (destroy c_duckdb_v2_data_chunk_destroy) \chunk -> do+ unless (chunk /= nullHandle) (codecError "helper query returned no rows")+ vector <- output (c_duckdb_v2_data_chunk_get_vector chunk 0)+ action vector --- | Get the high-bit mask used for native BIT padding.-paddingMask :: Word8 -> Word8-paddingMask padding = complement ((1 `shiftL` (8 - fromIntegral padding)) - 1)+-- | Map every native VARIANT scalar tag to its exact cast target. The tags are+-- the text of variant_typeof. The result is type text, because a DECIMAL tag+-- holds its width and scale. 'castType' parses each text once for each state.+-- See Note [VARIANT decoding].+scalarType :: Text -> IO Text+scalarType tag = case tag of+ "BOOL_TRUE" -> pure "BOOLEAN"+ "BOOL_FALSE" -> pure "BOOLEAN"+ "INT8" -> pure "TINYINT"+ "INT16" -> pure "SMALLINT"+ "INT32" -> pure "INTEGER"+ "INT64" -> pure "BIGINT"+ "INT128" -> pure "HUGEINT"+ "UINT8" -> pure "UTINYINT"+ "UINT16" -> pure "USMALLINT"+ "UINT32" -> pure "UINTEGER"+ "UINT64" -> pure "UBIGINT"+ "UINT128" -> pure "UHUGEINT"+ "TIME_MICROS" -> pure "TIME"+ "TIME_NANOS" -> pure "TIME_NS"+ "TIME_MICROS_TZ" -> pure "TIMETZ"+ "TIMESTAMP_SEC" -> pure "TIMESTAMP_S"+ "TIMESTAMP_MILIS" -> pure "TIMESTAMP_MS"+ "TIMESTAMP_MICROS" -> pure "TIMESTAMP"+ "TIMESTAMP_NANOS" -> pure "TIMESTAMP_NS"+ "TIMESTAMP_MICROS_TZ" -> pure "TIMESTAMPTZ"+ "TIMESTAMP_NANOS_TZ" -> pure "TIMESTAMPTZ_NS"+ "BITSTRING" -> pure "BIT"+ _+ | tag `elem` ["FLOAT", "DOUBLE", "VARCHAR", "BLOB", "UUID", "DATE", "INTERVAL", "BIGNUM", "GEOMETRY"] -> pure tag+ | "DECIMAL(" `Text.isPrefixOf` tag && ")" `Text.isSuffixOf` tag -> pure tag+ | otherwise -> codecError ("unsupported scalar tag " <> tag) --- | Check BIGNUM's sign, three-byte length header, and magnitude, then decode it.-decodeBigNum :: ByteString -> Either String Integer-decodeBigNum bytes = do- when (BS.length bytes < 4) (Left "truncated BIGNUM header or magnitude")- let (header, raw) = BS.splitAt 3 bytes- encoded = BS.foldl' (\n byte -> n `shiftL` 8 .|. fromIntegral byte) (0 :: Word32) header- negative = encoded .&. 0x800000 == 0- decoded = if negative then complement encoded .&. 0xffffff else encoded- count = decoded .&. 0x7fffff- magnitude = if negative then BS.map complement raw else raw- unless (toInteger count == toInteger (BS.length magnitude)) (Left "BIGNUM header length mismatch")- case BS.uncons magnitude of- Just (0, rest) | not (BS.null rest) || negative -> Left "noncanonical BIGNUM magnitude"- _ -> Right ()- pure (fromBigNumBytes (BS.unpack bytes))+-- | List a node's owned children in their native order.+nodeChildren :: VariantNode -> [DuckDBV2ValueHandle]+nodeChildren (ScalarNode _) = []+nodeChildren (ListNode values) = values+nodeChildren (ObjectNode entries) = map snd entries++-- | Release the children owned by a decoded node.+releaseNode :: VariantNode -> IO ()+releaseNode = releaseItems destroyValue . nodeChildren++-- | Report a metadata or payload mismatch before decoding a native value.+codecError :: Text -> IO a+codecError message = throwInternalError ("VARIANT: " <> message)
− test-codec/Main.hs
@@ -1,88 +0,0 @@-{-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE ScopedTypeVariables #-}---- | Check the private codec's validation without constructing invalid native vectors.-module Main (main) where--import Control.Exception (IOException, try)-import Control.Monad (forM_)-import Data.Bits (shiftR, (.&.), (.|.))-import qualified Data.ByteString as BS-import Data.Text (Text)-import Data.Word (Word32, Word8)-import Database.DuckDB.Simple.FromField (FieldValue (..))-import Database.DuckDB.Simple.Variant (variantObject)-import Database.DuckDB.Simple.VariantCodec (decodeVariantPayload)-import Test.Tasty (TestTree, defaultMain, testGroup)-import Test.Tasty.HUnit---- | Run malformed payload checks and boundary cases for the native format.-main :: IO ()-main =- defaultMain $- testGroup- "VARIANT payload validation"- [ rejected "missing root" [] [] [] []- , rejected "unknown tag" [(34, 0)] [] [] []- , rejected "offset outside blob" [(1, 1)] [] [] []- , rejected "truncated BIGINT" [(6, 0)] [] [] (replicate 7 0)- , rejected "truncated varint" [(16, 0)] [] [] [128]- , rejected "varint overflows uint32" [(16, 0)] [] [] [255, 255, 255, 255, 16]- , rejected "unterminated varint" [(16, 0)] [] [] [255, 255, 255, 255, 255, 0]- , rejected "string outside blob" [(16, 0)] [] [] [3, 97, 98]- , rejected "invalid UTF-8" [(16, 0)] [] [] [1, 255]- , rejected "decimal scale exceeds precision" [(15, 0)] [] [] [2, 3, 1, 0]- , rejected "decimal value exceeds precision" [(15, 0)] [] [] [1, 0, 100, 0]- , rejected "BIGNUM header is truncated" [(31, 0)] [] [] [2, 128, 0]- , rejected "BIGNUM length mismatch" [(31, 0)] [] [] [4, 128, 0, 2, 1]- , rejected "BIT padding exceeds seven" [(32, 0)] [] [] [2, 8, 255]- , rejected "BIT padding bits are absent" [(32, 0)] [] [] [2, 3, 21]- , rejected "child value out of range" [(30, 0)] [(Nothing, 1)] [] [1, 0]- , rejected "container range out of bounds" [(30, 0)] [] [] [1, 0]- , rejected "object key is missing" [(29, 0), (1, 2)] [(Just 0, 1)] [] [1, 0]- , rejected "array child has a key" [(30, 0), (1, 2)] [(Just 0, 1)] [(0, "x")] [1, 0]- , rejected "object child has no key" [(29, 0), (1, 2)] [(Nothing, 1)] [] [1, 0]- , rejected "duplicate dictionary index" [(1, 0)] [] [(0, "x"), (0, "y")] []- , rejected "duplicate object key" [(29, 0), (1, 2)] [(Just 0, 1), (Just 1, 1)] [(0, "x"), (1, "x")] [2, 0]- , rejected "cyclic child reference" [(30, 0)] [(Nothing, 0)] [] [1, 0]- , testCase "empty values need no payload bytes" $- decodeVariantPayload [(30, 0)] [] [] (BS.singleton 0) >>= (@?= FieldList [])- , testCase "shared children decode without duplicate traversal" $- decodeVariantPayload [(30, 0), (3, 2)] [(Nothing, 1), (Nothing, 1)] [] (BS.pack [2, 0, 7])- >>= (@?= FieldList [FieldInt8 7, FieldInt8 7])- , testCase "length-aware keys preserve embedded NUL" $- decodeVariantPayload [(29, 0), (1, 2)] [(Just 0, 1)] [(0, "a\0b")] (BS.pack [1, 0])- >>= (@?= variantObject [("a\0b", FieldBool True)])- , testCase "deep acyclic containers are accepted" $- forM_ [128, 129, 257, 1025] $ \count ->- let (values, children, bytes) = nested count- in decodeVariantPayload values children [] bytes >>= (@?= foldr (const (FieldList . pure)) (FieldInt8 7) [1 .. count - 1])- ]---- | Require a validation error for copied native payload data.-rejected :: String -> [(Word8, Word32)] -> [(Maybe Word32, Word32)] -> [(Word32, Text)] -> [Word8] -> TestTree-rejected label values children keys bytes =- testCase label $ assertRejected (decodeVariantPayload values children keys (BS.pack bytes))---- | Require a codec error.-assertRejected :: IO FieldValue -> Assertion-assertRejected action = do- result <- try action- case result of- Left (_ :: IOException) -> pure ()- Right value -> assertFailure ("expected payload rejection, got " <> show value)---- | Construct a chain of arrays with one scalar leaf.-nested :: Int -> ([(Word8, Word32)], [(Maybe Word32, Word32)], BS.ByteString)-nested count =- ( zip (replicate (count - 1) 30 <> [3]) offsets- , [(Nothing, fromIntegral (n + 1)) | n <- [0 .. count - 2]]- , BS.pack (concat payloads)- )- where- payloads = [1 : varint (fromIntegral n) | n <- [0 .. count - 2]] <> [[7]]- offsets = scanl (+) 0 (map (fromIntegral . length) payloads)- varint :: Word32 -> [Word8]- varint value- | value < 128 = [fromIntegral value]- | otherwise = fromIntegral (value .&. 127 .|. 128) : varint (value `shiftR` 7)
+ test-quack/Main.hs view
@@ -0,0 +1,98 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}++-- |+-- Module : Main+-- Description : Check authenticated Quack connections with local extension artifacts.+module Main (main) where++import Control.Exception (finally, try)+import Control.Monad (unless, void)+import Data.Int (Int64)+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Time (UTCTime)+import Data.Time.Clock.POSIX (posixSecondsToUTCTime)+import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Streaming as Streaming+import System.Directory (canonicalizePath, doesFileExist)+import System.Environment (lookupEnv)+import Test.Tasty (defaultMain)+import Test.Tasty.HUnit++-- | Require the matching artifacts before starting the integration test.+main :: IO ()+main = do+ httpfs <- extensionPath "DUCKDB_HTTPFS_EXTENSION"+ quack <- extensionPath "DUCKDB_QUACK_EXTENSION"+ defaultMain $+ testCase "authenticated Quack CONNECT, ATTACH, parameters and streaming" $+ withQuackConnection [httpfs, quack] \server -> do+ void $ execute_ server "CREATE TABLE items(id BIGINT, label VARCHAR)"+ void $ executeMany server "INSERT INTO items VALUES (?, ?)" [(1 :: Int64, "one" :: Text), (2, "two")]+ void $ execute_ server "CREATE TABLE large AS SELECT i FROM range(5000) t(i)"+ void $ execute_ server "CREATE TABLE instants(value TIMESTAMPTZ_NS)"+ let precise = posixSecondsToUTCTime 0.123456789+ void $ execute server "INSERT INTO instants VALUES (?)" (Only precise)+ [(uri, _, _)] <-+ query_ server "CALL quack_serve('quack:localhost:0', token='duckdb-haskell-loopback', disable_ssl=true)" :: IO [(Text, Text, Text)]+ finally+ ( do+ withQuackConnection [httpfs, quack] \client -> do+ void $ execute_ client (remoteConnection "CONNECT" uri "duckdb-haskell-loopback")+ (query_ client "SELECT id, label FROM items ORDER BY id" :: IO [(Int64, Text)])+ >>= (@?= [(1, "one"), (2, "two")])+ void $ execute_ client "DISCONNECT"+ void $ execute_ client (remoteConnection "ATTACH" uri "duckdb-haskell-loopback")+ void $ execute_ client "SET disabled_optimizers='remote_pushdown'"+ (query client "SELECT id, label FROM remote.items WHERE id >= ? ORDER BY id" (Only (2 :: Int64)) :: IO [(Int64, Text)])+ >>= (@?= [(2, "two")])+ (queryNamed client "SELECT label FROM remote.items WHERE id=$id" ["id" := (1 :: Int64)] :: IO [Only Text])+ >>= (@?= [Only "one"])+ void $ execute client "INSERT INTO remote.items VALUES (?, ?)" (3 :: Int64, "bound 'λ'" :: Text)+ (query_ server "SELECT label FROM items WHERE id=3" :: IO [Only Text])+ >>= (@?= [Only "bound 'λ'"])+ streamed <- Streaming.fold client "SELECT i FROM remote.large WHERE i < ?" (Only (5000 :: Int64)) (0 :: Int, 0 :: Int64) \(count, total) (Only value) ->+ pure (count + 1, total + value)+ streamed @?= (5000, 12497500)+ (query_ client "SELECT value FROM remote.instants" :: IO [Only UTCTime])+ >>= (@?= [Only precise])+ withQuackConnection [httpfs, quack] \rejected -> do+ result <- try (execute_ rejected (remoteConnection "CONNECT" uri "wrong-token"))+ case result of+ Left SQLError{sqlErrorMessage} ->+ assertBool "wrong credentials must fail authentication" ("Authentication failed" `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure "expected Quack to reject the wrong token"+ (query_ server "SELECT count(*) FROM items" :: IO [Only Int64]) >>= (@?= [Only 3])+ )+ (void (query server "CALL quack_stop(?)" (Only uri) :: IO [Only Text]))++-- | Fail when an extension path is absent or does not name a local file.+extensionPath :: String -> IO FilePath+extensionPath name = do+ configured <- lookupEnv name+ path <- case configured of+ Just value | not (null value) -> pure value+ _ -> fail (name <> " must name a local DuckDB extension artifact")+ exists <- doesFileExist path+ unless exists $ fail (name <> " does not name an existing file: " <> path)+ canonicalizePath path++-- | Load the local extensions with the configuration required by this fixture.+withQuackConnection :: [FilePath] -> (Connection -> IO a) -> IO a+withQuackConnection extensions action =+ withConnectionWithConfig ":memory:" [("allow_unsigned_extensions", "true"), ("threads", "2")] \connection -> do+ mapM_ (\path -> void (execute_ connection (Query ("LOAD " <> sqlLiteral (Text.pack path))))) extensions+ action connection++-- | Build CONNECT or ATTACH SQL with quoted URI and token literals.+remoteConnection :: Text -> Text -> Text -> Query+remoteConnection command uri token =+ Query (command <> " " <> sqlLiteral uri <> alias <> " (TOKEN " <> sqlLiteral token <> ")")+ where+ alias = if command == "ATTACH" then " AS remote" else ""++-- | Quote a SQL string literal, including apostrophes in local paths.+sqlLiteral :: Text -> Text+sqlLiteral value = "'" <> Text.replace "'" "''" value <> "'"
test/ArrowTests.hs view
@@ -1,183 +1,282 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE DuplicateRecordFields #-}+{-# LANGUAGE LambdaCase #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-}-{-# OPTIONS_GHC -Wno-deprecations #-}+{-# LANGUAGE TypeApplications #-} --- | Integration tests for scoped Arrow batches and their native ownership.+-- |+-- Module : ArrowTests+-- Description : Integration tests for scoped Arrow batches and their native ownership. module ArrowTests (arrowTests) where import Control.Concurrent (forkIO, killThread, newEmptyMVar, putMVar, takeMVar) import Control.Exception (AsyncException (ThreadKilled), IOException, SomeException, bracket, bracket_, fromException, mask_, throwIO, try)-import Control.Monad (forM, forM_, when)+import Control.Monad (forM, forM_, void, when) import Data.Bits (testBit)+import Data.Coerce (coerce) import Data.IORef (IORef, modifyIORef', newIORef, readIORef) import Data.Int (Int64)+import Data.List (sort)+import Data.Proxy (Proxy (..)) import qualified Data.Text as Text+import Data.Void (Void) import Data.Word (Word8)-import Database.DuckDB.FFI+import Database.DuckDB.FFI.V2 import Database.DuckDB.Simple+import Database.DuckDB.Simple.Arrow (releaseArrowArray, releaseArrowSchema) import qualified Database.DuckDB.Simple.Arrow as Arrow-import qualified Database.DuckDB.Simple.Deprecated.Streaming as Streaming+import Database.DuckDB.Simple.FromField (DecimalValue (..)) import Database.DuckDB.Simple.Internal (peekUtf8CString)+import qualified Database.DuckDB.Simple.Streaming as Streaming+import Foreign.C.ConstPtr (ConstPtr (..)) import Foreign.C.String (peekCString) import Foreign.Marshal.Alloc (alloca) import Foreign.Marshal.Utils (fillBytes)-import Foreign.Ptr (Ptr, castPtr, freeHaskellFunPtr, nullFunPtr, nullPtr)-import Foreign.Storable (peek, peekElemOff, poke, sizeOf)+import Foreign.Ptr (Ptr, freeHaskellFunPtr, nullFunPtr, nullPtr)+import Foreign.Storable (peek, peekElemOff, poke)+import GHC.Records (getField)+import HsBindgen.Runtime.HasCField (fromPtr)+import HsBindgen.Runtime.Marshal (staticSizeOf)+import HsBindgen.Runtime.Support.FunPtr (fromFunPtr, toFunPtr) import Test.Tasty (TestTree, testGroup) import Test.Tasty.HUnit+import TestUtils (filteredQueries) -- | Exercise native results and count their Arrow release callbacks. arrowTests :: TestTree arrowTests =- testGroup "Arrow batches" [arrowModeTests False, arrowModeTests True]+ testGroup "Arrow batches" [arrowModeTests False, arrowModeTests True] -- | Run the same ownership checks for both public Arrow interfaces. arrowModeTests :: Bool -> TestTree arrowModeTests streaming =- testGroup- (if streaming then "deprecated streaming" else "materialized")- [ testCase "parameters, Unicode names, NULLs and multiple batches" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- (batches, chunks) <-- foldArrow conn "SELECT CASE WHEN i % 3 = 0 THEN NULL ELSE i END AS \"íslenska_λ\" FROM range(?::BIGINT) t(i)" (Only (5000 :: Int64)) (0 :: Int, []) \(count, acc) schemaPtr arrayPtr -> do- schema <- peek schemaPtr- arrowSchemaChildCount schema @?= 1- child <- peekElemOff (arrowSchemaChildren schema) 0 >>= peek- peekUtf8CString (arrowSchemaName child) >>= (@?= "íslenska_λ")- peekCString (arrowSchemaFormat child) >>= (@?= "l")- values <- readInt64Batch arrayPtr- pure (count + 1, values : acc)- assertBool "more than one native batch" (batches > 1)- concat (reverse chunks) @?= [if i `rem` 3 == 0 then Nothing else Just i | i <- [0 .. 4999]]- , testCase "empty results retain the initial accumulator" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- result <- foldArrow_ conn "SELECT 1::BIGINT WHERE false" (17 :: Int) \_ _ _ -> assertFailure "unexpected batch" >> pure 0- result @?= 17- , testCase "result metadata follows bound parameter types" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- values <- foldArrow conn "SELECT ? AS value" (Only (5000000000 :: Int64)) [] \acc _ array -> do- batch <- readInt64Batch array- pure (acc <> batch)- values @?= [Just 5000000000]- , testCase "success releases the schema and every batch" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- withReleaseCounters \schemaReleases arrayReleases observe -> do- batches <- foldArrow_ conn "SELECT i FROM range(5000) t(i)" (0 :: Int) \count schema array -> do- observe schema array- pure (count + 1)- readIORef schemaReleases >>= (@?= batches)- readIORef arrayReleases >>= (@?= batches)- assertConnectionUsable conn- , testCase "consumers can release each schema and batch" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- withReleaseCounters \schemaReleases arrayReleases observe -> do- batches <- foldArrow_ conn "SELECT i FROM range(5000) t(i)" (0 :: Int) \count schema array -> do- observe schema array- releaseArrowArray array- releaseArrowSchema schema- pure (count + 1)- assertBool "more than one native batch" (batches > 1)- readIORef schemaReleases >>= (@?= batches)- readIORef arrayReleases >>= (@?= batches)- assertConnectionUsable conn- , testCase "moved objects survive query and connection close" $- withReleaseCounters \schemaReleases arrayReleases observe -> do- alloca \savedSchema -> alloca \savedArray ->- bracket_- ( do- fillBytes savedSchema 0 (sizeOf (undefined :: ArrowSchema))- fillBytes savedArray 0 (sizeOf (undefined :: ArrowArray))- )- (releaseArrowArray savedArray >> releaseArrowSchema savedSchema)- do- withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- foldArrow_ conn "SELECT 42::BIGINT AS id" () \() schemaPtr arrayPtr -> mask_ do- observe schemaPtr arrayPtr- schema <- peek schemaPtr- array <- peek arrayPtr- poke savedSchema schema- poke schemaPtr schema{arrowSchemaRelease = nullFunPtr}- poke savedArray array- poke arrayPtr array{arrowArrayRelease = nullFunPtr}- readIORef schemaReleases >>= (@?= 0)- readIORef arrayReleases >>= (@?= 0)- schema <- peek savedSchema- child <- peekElemOff (arrowSchemaChildren schema) 0 >>= peek- peekCString (arrowSchemaName child) >>= (@?= "id")- readInt64Batch savedArray >>= (@?= [Just 42])- readIORef schemaReleases >>= (@?= 1)- readIORef arrayReleases >>= (@?= 1)- , testCase "exceptions during consumption release each object once" $- forM_ [False, True] \consumeSchema ->+ testGroup+ (if streaming then "native streaming" else "materialized")+ [ testCase "parameters, Unicode names, NULLs and multiple batches" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ (batches, chunks) <-+ foldArrow+ conn+ "SELECT CASE WHEN i % 3 = 0 THEN NULL ELSE i END AS \"íslenska_λ\" FROM range(?::BIGINT) t(i)"+ (Only (5000 :: Int64))+ (0 :: Int, [])+ \(count, acc) schemaPtr arrayPtr -> do+ schema <- peek schemaPtr+ (getField @"n_children") schema @?= 1+ child <- peekElemOff ((getField @"children") schema) 0 >>= peek+ peekUtf8CString ((getField @"name") child) >>= (@?= "íslenska_λ")+ (peekCString . coerce) ((getField @"format") child) >>= (@?= "l")+ values <- readInt64Batch arrayPtr+ pure (count + 1, values : acc)+ assertBool "more than one native batch" (batches > 1)+ concat (reverse chunks) @?= [if i `rem` 3 == 0 then Nothing else Just i | i <- [0 .. 4999]]+ , testCase "the batch size option sets the rows of each batch" $ do+ let run options count = withConnectionWithOptions ":memory:" options{connectionConfig = [("threads", "1")]} \conn ->+ reverse <$> foldArrow conn "SELECT i::BIGINT AS i FROM range(?::BIGINT) t(i)" (Only (count :: Int64)) [] \acc _ arrayPtr -> do+ values <- readInt64Batch arrayPtr+ pure (values : acc)+ -- Each array has at most the given number of rows. The arrays give+ -- each row in order.+ checkBatches size limit count = do+ batches <- run defaultConnectionOptions{connectionArrowBatchSize = size} count+ assertBool ("array lengths " <> show (map length batches)) (all ((<= limit) . length) batches)+ concat batches @?= map Just [0 .. count - 1]+ defaultRows <- run defaultConnectionOptions 5+ singleRows <- run defaultConnectionOptions{connectionArrowBatchSize = 1} 5+ concat defaultRows @?= map Just [0 .. 4]+ concat singleRows @?= concat defaultRows+ (length defaultRows, length singleRows) @?= (1, 5)+ -- 131072 is the batch size of DuckDB.+ checkBatches 131072 131072 300000+ checkBatches 20000 20000 10000+ checkBatches 3000 3000 10000+ forM_ [-1, 0] \size ->+ try (withConnectionWithOptions ":memory:" defaultConnectionOptions{connectionArrowBatchSize = size} (const (pure ()))) >>= \case+ Left (err :: SQLError) -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ assertBool ("unexpected message: " <> show err) ("connectionArrowBatchSize is not positive" `Text.isInfixOf` sqlErrorMessage err)+ Right () -> assertFailure ("the batch size " <> show size <> " opened a connection")+ , testCase "an error of the step keeps its own query, and a native error of the fold has the fold query" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> withConnection ":memory:" \other -> do+ let failingStep :: IO ()+ failingStep = void (query other "SELECT ?" (Only (DecimalValue 2 1 100)) :: IO [Only Int64])+ stepError :: String -> IO a -> IO ()+ stepError label action =+ try action >>= \case+ Left (err :: SQLError) -> do+ assertEqual label (Just ErrorTypeConversion) (sqlErrorType err)+ assertEqual label Nothing (sqlErrorQuery err)+ Right _ -> assertFailure (label <> ": expected the DECIMAL error")+ stepError "Arrow.foldArrow" (Arrow.foldArrow conn "SELECT 1" () () \acc _ _ -> failingStep >> pure acc)+ stepError "Streaming.foldArrow" (Streaming.foldArrow conn "SELECT 1" () () \acc _ _ -> failingStep >> pure acc)+ stepError "fold" (fold conn "SELECT 1" () () \acc (Only (_ :: Int64)) -> failingStep >> pure acc)+ stepError "Streaming.fold" (Streaming.fold conn "SELECT 1" () () \acc (Only (_ :: Int64)) -> failingStep >> pure acc)+ forM_ [("Arrow.foldArrow", Arrow.foldArrow), ("Streaming.foldArrow", Streaming.foldArrow)] \(label, foldBatches) ->+ try (foldBatches conn "SELECT error('boom')" () () \acc _ _ -> pure acc) >>= \case+ Left (err :: SQLError) -> assertEqual label (Just "SELECT error('boom')") (sqlErrorQuery err)+ Right () -> assertFailure (label <> ": expected the error of DuckDB")+ , testCase "empty results retain the initial accumulator" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ result <- foldArrow_ conn "SELECT 1::BIGINT WHERE false" (17 :: Int) \_ _ _ -> assertFailure "unexpected batch" >> pure 0+ result @?= 17+ , testCase "result metadata follows bound parameter types" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ values <- foldArrow conn "SELECT ? AS value" (Only (5000000000 :: Int64)) [] \acc _ array -> do+ batch <- readInt64Batch array+ pure (acc <> batch)+ values @?= [Just 5000000000]+ , testCase "success releases the schema and every batch" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ withReleaseCounters \schemaReleases arrayReleases observe -> do+ batches <- foldArrow_ conn "SELECT i FROM range(5000) t(i)" (0 :: Int) \count schema array -> do+ observe schema array+ pure (count + 1)+ readIORef schemaReleases >>= (@?= batches)+ readIORef arrayReleases >>= (@?= batches)+ assertConnectionUsable conn+ , testCase "consumers can release each schema and batch" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ withReleaseCounters \schemaReleases arrayReleases observe -> do+ batches <- foldArrow_ conn "SELECT i FROM range(5000) t(i)" (0 :: Int) \count schema array -> do+ observe schema array+ releaseArrowArray array+ releaseArrowSchema schema+ pure (count + 1)+ assertBool "more than one native batch" (batches > 1)+ readIORef schemaReleases >>= (@?= batches)+ readIORef arrayReleases >>= (@?= batches)+ assertConnectionUsable conn+ , testCase "moved objects survive query and connection close" $+ withReleaseCounters \schemaReleases arrayReleases observe -> do+ alloca \savedSchema -> alloca \savedArray ->+ bracket_+ ( do+ fillBytes savedSchema 0 (staticSizeOf (Proxy @ArrowSchema))+ fillBytes savedArray 0 (staticSizeOf (Proxy @ArrowArray))+ )+ (releaseArrowArray savedArray >> releaseArrowSchema savedSchema)+ do withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- withReleaseCounters \schemaReleases arrayReleases observe -> do- result <- try $ foldArrow_ conn "SELECT 42::BIGINT" () \() schema array -> do- observe schema array- releaseArrowArray array- when consumeSchema (releaseArrowSchema schema)- throwIO (userError "consumer failed after release")- case result of- Left (err :: IOException) -> assertBool "consumer exception" ("consumer failed" `Text.isInfixOf` Text.pack (show err))- Right () -> assertFailure "expected consumer exception"- readIORef schemaReleases >>= (@?= 1)- readIORef arrayReleases >>= (@?= 1)- assertConnectionUsable conn- , testCase "callback exceptions release the active batch and schema" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- withReleaseCounters \schemaReleases arrayReleases observe -> do- result <- try $ foldArrow_ conn "SELECT i FROM range(5000) t(i)" () \_ schema array -> do- observe schema array- throwIO (userError "Arrow callback failed")- case result of- Left (err :: IOException) -> assertBool "original error" ("Arrow callback failed" `Text.isInfixOf` Text.pack (show err))- Right () -> assertFailure "expected callback exception"- readIORef schemaReleases >>= (@?= 1)- readIORef arrayReleases >>= (@?= 1)- assertConnectionUsable conn- , testCase "cancellation releases the active batch and schema" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- withReleaseCounters \schemaReleases arrayReleases observe -> do- entered <- newEmptyMVar- blocked <- newEmptyMVar- finished <- newEmptyMVar- worker <- forkIO do- result <- try $ foldArrow_ conn "SELECT i FROM range(5000) t(i)" () \_ schema array -> do- observe schema array- putMVar entered ()- takeMVar blocked- putMVar finished (result :: Either SomeException ())- takeMVar entered- killThread worker- result <- takeMVar finished- case result of- Left err -> fromException err @?= Just ThreadKilled- Right () -> assertFailure "expected cancellation"- readIORef schemaReleases >>= (@?= 1)- readIORef arrayReleases >>= (@?= 1)- assertConnectionUsable conn- , testCase "SQL errors leave the connection usable" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- result <- try $ foldArrow_ conn "SELECT error('Arrow query failed')" () \_ _ _ -> assertFailure "unexpected batch"- case result of- Left (err :: SQLError) -> assertBool "native error" ("Arrow query failed" `Text.isInfixOf` sqlErrorMessage err)- Right () -> assertFailure "expected query error"- assertConnectionUsable conn- , testCase "closing the connection stops further batch callbacks" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- withReleaseCounters \schemaReleases arrayReleases observe -> do- callbacks <- newIORef (0 :: Int)- result <- try $ foldArrow_ conn "SELECT i FROM range(5000) t(i)" () \_ schema array -> do- observe schema array- modifyIORef' callbacks (+ 1)- close conn- case result of- Left (err :: SQLError) -> sqlErrorMessage err @?= "duckdb-simple: connection is closed"- Right () -> assertFailure "expected closed connection error"- readIORef callbacks >>= (@?= 1)- readIORef schemaReleases >>= (@?= 1)- readIORef arrayReleases >>= (@?= 1)- ]+ foldArrow_ conn "SELECT 42::BIGINT AS id" () \() schemaPtr arrayPtr -> mask_ do+ observe schemaPtr arrayPtr+ schema <- peek schemaPtr+ array <- peek arrayPtr+ poke savedSchema schema+ poke (fromPtr (Proxy @"release") (schemaPtr :: Ptr ArrowSchema)) (coerce (nullFunPtr @Void))+ poke savedArray array+ poke (fromPtr (Proxy @"release") (arrayPtr :: Ptr ArrowArray)) (coerce (nullFunPtr @Void))+ readIORef schemaReleases >>= (@?= 0)+ readIORef arrayReleases >>= (@?= 0)+ schema <- peek savedSchema+ child <- peekElemOff ((getField @"children") schema) 0 >>= peek+ (peekCString . coerce) ((getField @"name") child) >>= (@?= "id")+ readInt64Batch savedArray >>= (@?= [Just 42])+ readIORef schemaReleases >>= (@?= 1)+ readIORef arrayReleases >>= (@?= 1)+ , testCase "exceptions during consumption release each object once" $+ forM_ [False, True] \consumeSchema ->+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ withReleaseCounters \schemaReleases arrayReleases observe -> do+ result <- try $ foldArrow_ conn "SELECT 42::BIGINT" () \() schema array -> do+ observe schema array+ releaseArrowArray array+ when consumeSchema (releaseArrowSchema schema)+ throwIO (userError "consumer failed after release")+ case result of+ Left (err :: IOException) -> assertBool "consumer exception" ("consumer failed" `Text.isInfixOf` Text.pack (show err))+ Right () -> assertFailure "expected consumer exception"+ readIORef schemaReleases >>= (@?= 1)+ readIORef arrayReleases >>= (@?= 1)+ assertConnectionUsable conn+ , testCase "callback exceptions release the active batch and schema" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ withReleaseCounters \schemaReleases arrayReleases observe -> do+ result <- try $ foldArrow_ conn "SELECT i FROM range(5000) t(i)" () \_ schema array -> do+ observe schema array+ throwIO (userError "Arrow callback failed")+ case result of+ Left (err :: IOException) -> assertBool "original error" ("Arrow callback failed" `Text.isInfixOf` Text.pack (show err))+ Right () -> assertFailure "expected callback exception"+ readIORef schemaReleases >>= (@?= 1)+ readIORef arrayReleases >>= (@?= 1)+ assertConnectionUsable conn+ , testCase "cancellation releases the active batch and schema" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ withReleaseCounters \schemaReleases arrayReleases observe -> do+ entered <- newEmptyMVar+ blocked <- newEmptyMVar+ finished <- newEmptyMVar+ worker <- forkIO do+ result <- try $ foldArrow_ conn "SELECT i FROM range(5000) t(i)" () \_ schema array -> do+ observe schema array+ putMVar entered ()+ takeMVar blocked+ putMVar finished (result :: Either SomeException ())+ takeMVar entered+ killThread worker+ result <- takeMVar finished+ case result of+ Left err -> fromException err @?= Just ThreadKilled+ Right () -> assertFailure "expected cancellation"+ readIORef schemaReleases >>= (@?= 1)+ readIORef arrayReleases >>= (@?= 1)+ assertConnectionUsable conn+ , testCase "SQL errors leave the connection usable" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ result <- try $ foldArrow_ conn "SELECT error('Arrow query failed')" () \_ _ _ -> assertFailure "unexpected batch"+ case result of+ Left (err :: SQLError) -> assertBool "native error" ("Arrow query failed" `Text.isInfixOf` sqlErrorMessage err)+ Right () -> assertFailure "expected query error"+ assertConnectionUsable conn+ , testCase "callback connection use follows the execution mode" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ count <- foldArrow_ conn "SELECT i FROM range(5000) t(i)" (0 :: Int) \acc _ _ -> do+ if streaming+ then do+ outcome <- try (query_ conn "SELECT 42" :: IO [Only Int64])+ case outcome of+ Left (err :: SQLError) -> sqlErrorType err @?= Just ErrorResourceInUse+ Right _ -> assertFailure "expected a busy streaming connection"+ else query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ pure (acc + 1)+ assertBool "expected multiple callbacks" (count > 1)+ , testCase "SQL errors report the query that failed" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ let outer = "SELECT error('Arrow query failed')"+ inner = "SELECT error('callback query failed')"+ failed <- try $ foldArrow_ conn outer () \_ _ _ -> assertFailure "unexpected batch"+ case failed of+ Left (err :: SQLError) -> sqlErrorQuery err @?= Just outer+ Right () -> assertFailure "expected query error"+ nested <- try $ foldArrow_ conn "SELECT 42" () \_ _ _ -> void (query_ conn inner :: IO [Only Int64])+ case nested of+ Left (err :: SQLError) -> sqlErrorQuery err @?= Just inner+ Right () -> assertFailure "expected callback query error"+ assertConnectionUsable conn+ , testCase "closing the connection stops further batch callbacks" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ withReleaseCounters \schemaReleases arrayReleases observe -> do+ callbacks <- newIORef (0 :: Int)+ result <- try $ foldArrow_ conn "SELECT i FROM range(5000) t(i)" () \_ schema array -> do+ observe schema array+ modifyIORef' callbacks (+ 1)+ close conn+ case result of+ Left (err :: SQLError) -> sqlErrorMessage err @?= "duckdb-simple: connection is closed"+ Right () -> assertFailure "expected closed connection error"+ readIORef callbacks >>= (@?= 1)+ readIORef schemaReleases >>= (@?= 1)+ readIORef arrayReleases >>= (@?= 1)+ , testCase "filtered results give each row in the batches" $+ forM_ ["1", "4"] \threads -> withConnectionWithConfig ":memory:" [("threads", threads)] \conn ->+ forM_ filteredQueries \(sql, expected) -> do+ batches <- foldArrow_ conn sql [] \acc _ array -> (: acc) <$> readInt64Batch array+ let rows = concat (reverse batches)+ -- With more than one thread, DuckDB can give the rows in another order.+ assertEqual (Text.unpack threads <> " threads: " <> show sql) (map Just expected) (if threads == "1" then rows else sort rows)+ ] where foldArrow :: (ToRow q) => Connection -> Query -> q -> a -> (a -> Ptr ArrowSchema -> Ptr ArrowArray -> IO a) -> IO a foldArrow = if streaming then Streaming.foldArrow else Arrow.foldArrow@@ -188,53 +287,53 @@ -- | Copy a BIGINT Arrow column, including its validity bitmap. readInt64Batch :: Ptr ArrowArray -> IO [Maybe Int64] readInt64Batch arrayPtr = do- array <- peek arrayPtr- child <- peekElemOff (arrowArrayChildren array) 0 >>= peek- validity <- peekElemOff (arrowArrayBuffers child) 0- values <- peekElemOff (arrowArrayBuffers child) 1- forM [0 .. fromIntegral (arrowArrayLength array) - 1] \row -> do- let index = fromIntegral (arrowArrayOffset child) + row- valid <-- if validity == nullPtr- then pure True- else do- byte <- peekElemOff (castPtr validity :: Ptr Word8) (index `div` 8)- pure (testBit byte (index `rem` 8))- if valid then Just <$> peekElemOff (castPtr values) index else pure Nothing+ array <- peek arrayPtr+ child <- peekElemOff ((getField @"children") array) 0 >>= peek+ validity <- peekElemOff ((getField @"buffers") child) 0+ values <- peekElemOff ((getField @"buffers") child) 1+ forM [0 .. fromIntegral ((getField @"length") array) - 1] \row -> do+ let index = fromIntegral ((getField @"offset") child) + row+ valid <-+ if validity == (coerce (nullPtr @Void))+ then pure True+ else do+ byte <- peekElemOff (coerce validity :: Ptr Word8) (index `div` 8)+ pure (testBit byte (index `rem` 8))+ if valid then Just <$> peekElemOff (coerce values) index else pure Nothing -- | Wrap real release callbacks to observe cleanup without changing ownership. withReleaseCounters :: (IORef Int -> IORef Int -> (Ptr ArrowSchema -> Ptr ArrowArray -> IO ()) -> IO a) -> IO a withReleaseCounters action = do- schemaReleases <- newIORef 0- arrayReleases <- newIORef 0- originalSchema <- newIORef Nothing- originalArray <- newIORef Nothing- bracket- ( wrapArrowSchemaRelease \ptr -> do- modifyIORef' schemaReleases (+ 1)- callback <- readIORef originalSchema- maybe (assertFailure "missing schema release") (\release -> mkArrowSchemaRelease release ptr) callback+ schemaReleases <- newIORef 0+ arrayReleases <- newIORef 0+ originalSchema <- newIORef Nothing+ originalArray <- newIORef Nothing+ bracket+ ( toFunPtr \ptr -> do+ modifyIORef' schemaReleases (+ 1)+ callback <- readIORef originalSchema+ maybe (assertFailure "missing schema release") (\release -> fromFunPtr release ptr) callback+ )+ (freeHaskellFunPtr . coerce)+ \schemaCallback ->+ bracket+ ( toFunPtr \ptr -> do+ modifyIORef' arrayReleases (+ 1)+ callback <- readIORef originalArray+ maybe (assertFailure "missing array release") (\release -> fromFunPtr release ptr) callback )- freeHaskellFunPtr- \schemaCallback ->- bracket- ( wrapArrowArrayRelease \ptr -> do- modifyIORef' arrayReleases (+ 1)- callback <- readIORef originalArray- maybe (assertFailure "missing array release") (\release -> mkArrowArrayRelease release ptr) callback- )- freeHaskellFunPtr- \arrayCallback -> do- let observe schemaPtr arrayPtr = do- schema <- peek schemaPtr- assertBool "schema must not have been released" (arrowSchemaRelease schema /= nullFunPtr)- when (arrowSchemaRelease schema /= schemaCallback) do- modifyIORef' originalSchema (const (Just (arrowSchemaRelease schema)))- poke schemaPtr schema{arrowSchemaRelease = schemaCallback}- array <- peek arrayPtr- modifyIORef' originalArray (const (Just (arrowArrayRelease array)))- poke arrayPtr array{arrowArrayRelease = arrayCallback}- action schemaReleases arrayReleases observe+ (freeHaskellFunPtr . coerce)+ \arrayCallback -> do+ let observe schemaPtr arrayPtr = do+ schema <- peek schemaPtr+ assertBool "schema must not have been released" ((getField @"release") schema /= (coerce (nullFunPtr @Void)))+ when ((getField @"release") schema /= schemaCallback) do+ modifyIORef' originalSchema (const (Just ((getField @"release") schema)))+ poke (fromPtr (Proxy @"release") (schemaPtr :: Ptr ArrowSchema)) schemaCallback+ array <- peek arrayPtr+ modifyIORef' originalArray (const (Just ((getField @"release") array)))+ poke (fromPtr (Proxy @"release") (arrayPtr :: Ptr ArrowArray)) arrayCallback+ action schemaReleases arrayReleases observe -- | Check that no failed Arrow operation leaves the connection busy. assertConnectionUsable :: Connection -> Assertion
test/CancellationTests.hs view
@@ -2,166 +2,436 @@ {-# LANGUAGE LambdaCase #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-}-{-# OPTIONS_GHC -Wno-deprecations #-} --- | Cancellation during native execution must finish before handles are released.-module CancellationTests (cancellationTests) where+-- |+-- Module : CancellationTests+-- Description : Cancellation during native execution must finish before handles are released, and a callback must not wait without a bound for its own connection.+module CancellationTests (cancellationTests, callbackLockTests) where -import Control.Concurrent (MVar, ThreadId, forkIOWithUnmask, newEmptyMVar, putMVar, readMVar, rtsSupportsBoundThreads, threadDelay, throwTo, tryPutMVar, tryReadMVar)-import Control.Exception (AsyncException (..), SomeException, bracket, fromException, mask_, try)-import Control.Monad (unless, void)-import Data.IORef (atomicWriteIORef, newIORef, readIORef)+import Control.Concurrent (+ MVar,+ ThreadId,+ forkIOWithUnmask,+ myThreadId,+ newEmptyMVar,+ putMVar,+ readMVar,+ rtsSupportsBoundThreads,+ threadDelay,+ throwTo,+ tryPutMVar,+ tryReadMVar,+ )+import Control.Exception (AsyncException (..), ErrorCall (..), SomeException, bracket, fromException, mask_, throwIO, try)+import Control.Monad (unless, void, when)+import Data.IORef (IORef, atomicModifyIORef', atomicWriteIORef, modifyIORef', newIORef, readIORef) import Data.Int (Int64)-import Database.DuckDB.FFI (c_duckdb_interrupt, c_duckdb_query)+import Data.Maybe (isNothing)+import Data.Text (Text)+import qualified Data.Text as Text import Database.DuckDB.Simple import Database.DuckDB.Simple.Arrow (foldArrow_)-import qualified Database.DuckDB.Simple.Deprecated.Streaming as Streaming-import Database.DuckDB.Simple.Internal (withConnectionHandle, withQueryCString, withResult)+import Database.DuckDB.Simple.Internal (+ c_duckdb_v2_connection_interrupt,+ c_duckdb_v2_result_drain,+ checked,+ output,+ runInterruptibleQuery,+ withConnectionHandle,+ withResult,+ withResultWindow,+ )+import qualified Database.DuckDB.Simple.Streaming as Streaming import GHC.Conc (BlockReason (..), ThreadStatus (..), threadStatus) import System.Timeout (timeout) import Test.Tasty (TestTree, testGroup) import Test.Tasty.HUnit+import TestUtils (assertReusable, raiseCapabilities) -- | Exercise each native execution entry point and two cancellation races. cancellationTests :: TestTree cancellationTests =- testGroup "native cancellation" $- [ testCase name $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- assertBool "cancellation requires the threaded runtime" rtsSupportsBoundThreads- entered <- newEmptyMVar- createFunction conn "query_started" (signal entered >> pure (1 :: Int64))- withCaller conn (pure ()) (run conn longQuery) \caller done -> do- await "native query did not start" (readMVar entered)- (_, sent) <- startCaller (throwTo caller UserInterrupt)- outcome <- await "native query did not cancel" (readMVar done)- assertAsync [UserInterrupt] outcome- await "interrupt sender did not finish" (readMVar sent) >>= assertSucceeded- assertReusable conn- | (name, run) <-- [ ("query_", \conn sql -> void (query_ conn sql :: IO [Only Double]))- , ("query with a prepared statement", \conn sql -> void (query conn sql () :: IO [Only Double]))- , ("execute_", \conn sql -> void (execute_ conn sql))- , ("execute with a prepared statement", \conn sql -> void (execute conn sql ()))- , ("statement cursor", \conn sql -> withStatement conn sql \stmt -> void (nextRow stmt :: IO (Maybe (Only Double))))- , ("fold_", \conn sql -> void (fold_ conn sql (0 :: Double) (\acc (Only value) -> pure (acc + value))))- , ("Arrow fold", \conn sql -> foldArrow_ conn sql () (\() _ _ -> pure ()))- ]+ testGroup "native cancellation" $+ [ testCase name $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ assertBool "cancellation requires the threaded runtime" rtsSupportsBoundThreads+ entered <- newEmptyMVar+ createFunction conn "query_started" (signal entered >> pure (1 :: Int64))+ withCaller conn (pure ()) (run conn longQuery) \caller done -> do+ await "native query did not start" (readMVar entered)+ (_, sent) <- startCaller (throwTo caller UserInterrupt)+ outcome <- await "native query did not cancel" (readMVar done)+ assertAsync [UserInterrupt] outcome+ await "interrupt sender did not finish" (readMVar sent) >>= assertSucceeded+ assertReusable conn+ | (name, run) <-+ [ ("query_", \conn sql -> void (query_ conn sql :: IO [Only Double]))+ , ("query with a prepared statement", \conn sql -> void (query conn sql () :: IO [Only Double]))+ , ("execute_", \conn sql -> void (execute_ conn sql))+ , ("execute with a prepared statement", \conn sql -> void (execute conn sql ()))+ , ("statement cursor", \conn sql -> withStatement conn sql \stmt -> void (nextRow stmt :: IO (Maybe (Only Double))))+ , ("fold_", \conn sql -> void (fold_ conn sql (0 :: Double) (\acc (Only value) -> pure (acc + value))))+ , ("Arrow fold", \conn sql -> foldArrow_ conn sql () (\() _ _ -> pure ())) ]- <> [ testCase name $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- void (execute_ conn "SET streaming_buffer_size = '64KB'")- filtering <- newIORef False- entered <- newEmptyMVar- createFunction conn "discard_remaining" \(_ :: Int64) -> do- active <- readIORef filtering- if active then signal entered >> pure True else pure False- let sql = "SELECT i FROM range(1000000000000) t(i) WHERE NOT discard_remaining(i)"- withCaller conn (pure ()) (run conn sql (atomicWriteIORef filtering True)) \caller done -> do- await "fetch did not start after the first delivered batch" (readMVar entered)- (_, sent) <- startCaller (throwTo caller UserInterrupt)- await "streaming fetch did not cancel" (readMVar done) >>= assertAsync [UserInterrupt]- await "interrupt sender did not finish" (readMVar sent) >>= assertSucceeded- assertReusable conn- | (name, run) <-- [ ("cancel a native row fetch", \conn sql delivered -> Streaming.fold_ conn sql () (\() (Only (_ :: Int64)) -> delivered))- , ("cancel a native Arrow fetch", \conn sql delivered -> Streaming.foldArrow_ conn sql () (\() _ _ -> delivered))- ,- ( "mixed cursor entry points retain interruptible fetching"- , \conn sql delivered -> withStatement conn sql \stmt -> do- Streaming.nextRow stmt >>= (@?= Just (Only (0 :: Int64)))- delivered- let consume =- nextRow stmt >>= \case- Nothing -> pure ()- Just (Only (_ :: Int64)) -> consume- consume- )- ]- ]- <> [ testCase "cancellation before native entry is not cleared by startup" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- queued <- newEmptyMVar- begin <- newEmptyMVar- createFunction conn "query_started" (pure 1 :: IO Int64)- let action = withConnectionHandle conn \handle ->- withQueryCString longQuery \sql ->- withResult- conn- longQuery- (\result -> signal queued >> readMVar begin >> c_duckdb_query handle sql result)- (const (pure ()))- withCaller conn (signal begin) action \caller done -> do- await "query worker did not reach its entry gate" (readMVar queued)- (_, sent) <- startCaller (throwTo caller UserInterrupt)- await "first interrupt was not delivered" (readMVar sent) >>= assertSucceeded- awaitCleanup caller- signal begin- await "startup cleared the cancellation" (readMVar done) >>= assertAsync [UserInterrupt]- assertReusable conn- , testCase "a second cancellation waits for a blocked callback to return" $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- entered <- newEmptyMVar- release <- newEmptyMVar- createFunction conn "blocked_callback" (signal entered >> readMVar release >> pure (1 :: Int64))- withCaller conn (signal release) (void (query_ conn "SELECT blocked_callback()" :: IO [Only Int64])) \caller done -> do- await "callback did not start" (readMVar entered)- (_, firstSent) <- startCaller (throwTo caller UserInterrupt)- await "first interrupt was not delivered" (readMVar firstSent) >>= assertSucceeded- secondAttempted <- newEmptyMVar- (_, secondSent) <- startCaller (signal secondAttempted >> throwTo caller ThreadKilled)- await "second interrupt sender did not start" (readMVar secondAttempted)- premature <- timeout 50000 (readMVar secondSent)- case premature of- Nothing -> pure ()- Just _ -> assertFailure "second interrupt escaped cleanup"- tryReadMVar done >>= \case- Nothing -> pure ()- Just _ -> assertFailure "query returned while its callback was still running"- signal release- await "callback cancellation did not finish" (readMVar done) >>= assertAsync [UserInterrupt, ThreadKilled]- await "second interrupt sender did not finish" (readMVar secondSent) >>= assertSucceeded- assertReusable conn- ]+ ]+ <> [ testCase name $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ void (execute_ conn "SET streaming_buffer_size = '64KB'")+ filtering <- newIORef False+ entered <- newEmptyMVar+ createFunction conn "discard_remaining" \(_ :: Int64) -> do+ active <- readIORef filtering+ if active then signal entered >> pure True else pure False+ let sql = "SELECT i FROM range(1000000000000) t(i) WHERE NOT discard_remaining(i)"+ withCaller conn (pure ()) (run conn sql (atomicWriteIORef filtering True)) \caller done -> do+ await "fetch did not start after the first delivered batch" (readMVar entered)+ (_, sent) <- startCaller (throwTo caller UserInterrupt)+ await "streaming fetch did not cancel" (readMVar done) >>= assertAsync [UserInterrupt]+ await "interrupt sender did not finish" (readMVar sent) >>= assertSucceeded+ assertReusable conn+ | (name, run) <-+ [ ("cancel a native row fetch", \conn sql delivered -> Streaming.fold_ conn sql () (\() (Only (_ :: Int64)) -> delivered))+ , ("cancel a native Arrow fetch", \conn sql delivered -> Streaming.foldArrow_ conn sql () (\() _ _ -> delivered))+ ,+ ( "mixed cursor entry points retain interruptible fetching"+ , \conn sql delivered -> withStatement conn sql \stmt -> do+ Streaming.nextRow stmt >>= (@?= Just (Only (0 :: Int64)))+ delivered+ let consume =+ nextRow stmt >>= \case+ Nothing -> pure ()+ Just (Only (_ :: Int64)) -> consume+ consume+ )+ ]+ ]+ <> [ testCase "cancellation before native entry is not cleared by startup" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ queued <- newEmptyMVar+ begin <- newEmptyMVar+ createFunction conn "query_started" (pure 1 :: IO Int64)+ let action = withStatement conn longQuery \stmt ->+ withResult stmt \result -> runInterruptibleQuery conn do+ signal queued+ readMVar begin+ void (output (c_duckdb_v2_result_drain result))+ withCaller conn (signal begin) action \caller done -> do+ await "query worker did not reach its entry gate" (readMVar queued)+ (_, sent) <- startCaller (throwTo caller UserInterrupt)+ await "first interrupt was not delivered" (readMVar sent) >>= assertSucceeded+ awaitCleanup caller+ signal begin+ await "startup cleared the cancellation" (readMVar done) >>= assertAsync [UserInterrupt]+ assertReusable conn+ , testCase "a second cancellation waits for a blocked callback to return" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ entered <- newEmptyMVar+ release <- newEmptyMVar+ createFunction conn "blocked_callback" (signal entered >> readMVar release >> pure (1 :: Int64))+ withCaller conn (signal release) (void (query_ conn "SELECT blocked_callback()" :: IO [Only Int64])) \caller done -> do+ await "callback did not start" (readMVar entered)+ (_, firstSent) <- startCaller (throwTo caller UserInterrupt)+ await "first interrupt was not delivered" (readMVar firstSent) >>= assertSucceeded+ secondAttempted <- newEmptyMVar+ (_, secondSent) <- startCaller (signal secondAttempted >> throwTo caller ThreadKilled)+ await "second interrupt sender did not start" (readMVar secondAttempted)+ premature <- timeout 50000 (readMVar secondSent)+ case premature of+ Nothing -> pure ()+ Just _ -> assertFailure "second interrupt escaped cleanup"+ tryReadMVar done >>= \case+ Nothing -> pure ()+ Just _ -> assertFailure "query returned while its callback was still running"+ signal release+ await "callback cancellation did not finish" (readMVar done) >>= assertAsync [UserInterrupt, ThreadKilled]+ await "second interrupt sender did not finish" (readMVar secondSent) >>= assertSucceeded+ assertReusable conn+ , testCase "a cancellation in the result window releases the result that the window makes" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ queued <- newEmptyMVar+ gate <- newEmptyMVar+ released <- newIORef []+ -- The helper thread of the window makes its result only after the+ -- caller is cancelled. The caller does not return that result, so+ -- the release action of the window must get it.+ let action = withResultWindow conn (\result -> modifyIORef' released (result :)) do+ signal queued+ readMVar gate+ pure (7 :: Int)+ withCaller conn (signal gate) (void action) \caller done -> do+ await "result window did not start" (readMVar queued)+ (_, sent) <- startCaller (throwTo caller UserInterrupt)+ await "interrupt was not delivered" (readMVar sent) >>= assertSucceeded+ awaitCleanup caller+ signal gate+ await "cancelled result window did not finish" (readMVar done) >>= assertAsync [UserInterrupt]+ readIORef released >>= (@?= [7])+ assertReusable conn+ ] +-- | A callback that runs a statement on the connection that runs the callback+-- waits for the lock of that connection. Each case sets a bound of 0.2 s.+callbackLockTests :: TestTree+callbackLockTests =+ testGroup+ "a callback that waits for a connection lock"+ [ testCase "a callback that runs a statement on its own connection in query_ fails at the bound" $+ withBoundedConnections "1" testLockWait \conn _ -> do+ (threads, errors) <- registerStatementCallback conn conn+ outcome <- boundedOuterCall 30000000 threads (void (query_ conn "SELECT own_statement(i) FROM range(3) t(i)" :: IO [Only Int64]))+ assertBoundError outcome+ assertInnerErrors errors+ assertReusable conn+ , testCase "a callback that runs a statement on its own connection in a later stream fetch fails at the bound" $+ withBoundedConnections "1" testLockWait \conn _ -> do+ (threads, errors) <- registerStatementCallback conn conn+ rows <- newIORef (0 :: Int)+ void (execute_ conn "SET streaming_buffer_size = '16KB'")+ outcome <- boundedOuterCall 30000000 threads $ withStatement conn laterFetchQuery \stmt -> do+ let consume =+ Streaming.nextRow stmt >>= \case+ Nothing -> pure ()+ Just (Only (_ :: Int64)) -> modifyIORef' rows (+ 1) >> consume+ consume+ assertBoundError outcome+ assertInnerErrors errors+ readIORef rows >>= \count -> assertBool "the callback ran before the stream delivered a row" (count > 0)+ assertReusable conn+ , testCase "a callback that runs a statement on its own connection in a streaming fold fails at the bound" $+ withBoundedConnections "1" testLockWait \conn _ -> do+ (threads, errors) <- registerStatementCallback conn conn+ rows <- newIORef (0 :: Int)+ void (execute_ conn "SET streaming_buffer_size = '16KB'")+ outcome <- boundedOuterCall 30000000 threads $ Streaming.fold_ conn laterFetchQuery () \() (Only (_ :: Int64)) -> modifyIORef' rows (+ 1)+ assertBoundError outcome+ assertInnerErrors errors+ readIORef rows >>= \count -> assertBool "the callback ran before the fold received a row" (count > 0)+ assertReusable conn+ , testCase "a callback that runs a statement on its own connection in a later Arrow fetch fails at the bound" $+ withBoundedConnections "1" testLockWait \conn _ -> do+ (threads, errors) <- registerStatementCallback conn conn+ batches <- newIORef (0 :: Int)+ void (execute_ conn "SET streaming_buffer_size = '16KB'")+ outcome <- boundedOuterCall 30000000 threads $ Streaming.foldArrow_ conn laterFetchQuery () \() _ _ -> modifyIORef' batches (+ 1)+ assertBoundError outcome+ assertInnerErrors errors+ readIORef batches >>= \count -> assertBool "the callback ran before the fold received a batch" (count > 0)+ assertReusable conn+ , testCase "a callback that runs a statement on its own connection before the first Arrow batch fails at the bound" $+ withBoundedConnections "1" testLockWait \conn _ -> do+ (threads, errors) <- registerStatementCallback conn conn+ batches <- newIORef (0 :: Int)+ outcome <-+ boundedOuterCall 30000000 threads $+ Streaming.foldArrow_ conn "SELECT own_statement(i) FROM range(3) t(i)" () \() _ _ -> modifyIORef' batches (+ 1)+ assertBoundError outcome+ assertInnerErrors errors+ readIORef batches >>= (@?= 0)+ assertReusable conn+ , testCase "a callback that prepares a statement on its own connection fails at the bound" $+ withBoundedConnections "1" testLockWait \conn _ -> do+ -- The preparation is the only step of the callback that takes the connection lock.+ (threads, errors) <- registerOwnStatement conn (withStatement conn "SELECT ?::BIGINT" (\_ -> pure ()))+ outcome <- boundedOuterCall 30000000 threads (void (query_ conn "SELECT own_statement(i) FROM range(3) t(i)" :: IO [Only Int64]))+ assertBoundError outcome+ assertInnerErrors errors+ assertReusable conn+ , testCase "a callback that reads the column count of a statement on its own connection fails at the bound" $+ withBoundedConnections "1" testLockWait \conn _ ->+ -- The statement has no schemas yet. Thus the column count binds the+ -- metadata of the statement, and that bind takes the connection lock.+ withStatement conn "SELECT 1 AS a, 2 AS b" \pending -> do+ (threads, errors) <- registerOwnStatement conn (void (columnCount pending))+ outcome <- boundedOuterCall 30000000 threads (void (query_ conn "SELECT own_statement(i) FROM range(3) t(i)" :: IO [Only Int64]))+ assertBoundError outcome+ assertInnerErrors errors+ columnCount pending >>= (@?= 2)+ assertReusable conn+ , testCase "a callback statement that fails on a second connection leaves the lock of that connection free" $+ withBoundedConnections "1" testLockWait \conn other -> do+ createFunction conn "failing_other" \(i :: Int64) -> do+ inner <- try (query_ other "SELECT error('inner')" :: IO [Only Int64])+ case inner of+ Left (_ :: SQLError) -> pure ()+ Right _ -> assertFailure "expected the inner error"+ -- The failed statement held the lock of the second connection.+ -- The next statement of the callback must get that lock at once.+ query_ other "SELECT 1" >>= (@?= [Only (1 :: Int64)])+ pure i+ query_ conn "SELECT failing_other(i) FROM range(3) t(i)" >>= (@?= map Only [0, 1, 2 :: Int64])+ timeout 5000000 (query_ other "SELECT 42") >>= (@?= Just [Only (42 :: Int64)])+ , testCase "callbacks on 4 DuckDB threads that run statements on their own connection each fail at the bound" $+ withBoundedConnections "4" testLockWait \conn _ -> do+ void (execute_ conn "CREATE TABLE callback_rows AS SELECT i FROM range(1000000) t(i)")+ (threads, errors) <- registerStatementCallback conn conn+ outcome <- boundedOuterCall 30000000 threads (void (query_ conn "SELECT sum(own_statement(i)) FROM callback_rows" :: IO [Only Integer]))+ assertBoundError outcome+ assertInnerErrors errors+ assertReusable conn+ , testCase "callbacks on 4 capabilities that are inside at the same time each fail at the bound" $+ do+ -- See Note [Capabilities of the tests] in "TestUtils".+ raiseCapabilities 4+ withBoundedConnections "4" testLockWait \conn _ -> do+ void (execute_ conn "CREATE TABLE callback_rows AS SELECT i FROM range(1000000) t(i)")+ inside <- newIORef (0 :: Int)+ together <- newEmptyMVar+ successes <- newIORef (0 :: Int)+ -- The first two calls wait for each other before their statements.+ -- Thus two callback threads have a mark at the same time, and the+ -- mark of the second thread must keep the mark of the first thread.+ let meet = do+ count <- atomicModifyIORef' inside (\known -> (known + 1, known + 1))+ when (count >= 2) (signal together)+ met <- timeout 10000000 (readMVar together)+ when (isNothing met) (throwIO (userError "no second callback was inside at the same time"))+ (threads, errors) <- registerOwnStatement conn do+ meet+ void (query_ conn "SELECT 1" :: IO [Only Int64])+ atomicModifyIORef' successes (\known -> (known + 1, ()))+ outcome <- boundedOuterCall 30000000 threads (void (query_ conn "SELECT sum(own_statement(i)) FROM callback_rows" :: IO [Only Integer]))+ assertBoundError outcome+ assertInnerErrors errors+ readIORef successes >>= (@?= 0)+ readIORef inside >>= \count -> assertBool ("callbacks before the error: " <> show count) (count >= 2)+ assertReusable conn+ , testCase "callbacks on 4 DuckDB threads can share a second connection" $+ withBoundedConnections "4" (Just defaultCallbackLockWait) \conn other -> do+ void (execute_ conn "CREATE TABLE callback_rows AS SELECT i FROM range(1000000) t(i)")+ threads <- newIORef []+ createFunction conn "other_statement" \(i :: Int64) -> do+ when (i `mod` 200 == 0) do+ myThreadId >>= \self -> atomicModifyIORef' threads (\known -> (self : known, ()))+ query_ other "SELECT 1" >>= (@?= [Only (1 :: Int64)])+ pure (i `mod` 7)+ result <- newIORef []+ outcome <- boundedOuterCall 120000000 threads (query_ conn "SELECT sum(other_statement(i)) FROM callback_rows" >>= atomicWriteIORef result)+ either (assertFailure . show) pure outcome+ readIORef result >>= (@?= [Only (sum [i `mod` 7 | i <- [0 .. 999999 :: Integer]])])+ assertReusable other+ , testCase "the callback lock wait rejects a negative bound and accepts no bound" $ do+ negative <- try (openWithOptions ":memory:" defaultConnectionOptions{connectionCallbackLockWait = Just (-1)})+ case negative of+ Left (err :: SQLError) -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ assertBool ("unexpected message: " <> show err) ("connectionCallbackLockWait" `Text.isInfixOf` sqlErrorMessage err)+ Right conn -> close conn >> assertFailure "a negative bound opened a connection"+ withConnectionWithOptions ":memory:" defaultConnectionOptions{connectionCallbackLockWait = Nothing} assertReusable+ ]++-- | The bound of the cases in microseconds: 0.2 s.+testLockWait :: Maybe Int+testLockWait = Just 200000++-- | Run an action with two connections of one instance, each with the bound.+withBoundedConnections :: Text -> Maybe Int -> (Connection -> Connection -> IO a) -> IO a+withBoundedConnections threads bound action =+ withEnvironment \environment -> withInstance environment [("threads", threads)] \inst -> do+ attachDatabase inst ":memory:" "callback_lock" []+ setDefaultDatabase inst "callback_lock"+ let options = defaultConnectionOptions{connectionCallbackLockWait = bound}+ bracket (connectWithOptions inst options) close \conn ->+ bracket (connectWithOptions inst options) close (action conn)++-- | Register @own_statement@ on a connection. It runs a statement on the target+-- connection, records its thread and the error of the statement, and raises the error.+registerStatementCallback :: Connection -> Connection -> IO (IORef [ThreadId], IORef [SQLError])+registerStatementCallback conn target = registerOwnStatement conn (void (query_ target "SELECT 1" :: IO [Only Int64]))++-- | Register @own_statement@ on a connection. It runs an action, records its+-- thread and each 'SQLError' of the action, and raises the error.+registerOwnStatement :: Connection -> IO () -> IO (IORef [ThreadId], IORef [SQLError])+registerOwnStatement conn statement = do+ threads <- newIORef []+ errors <- newIORef []+ createFunction conn "own_statement" \(i :: Int64) -> do+ myThreadId >>= \self -> atomicModifyIORef' threads (\known -> (self : known, ()))+ try statement >>= \case+ Left err -> atomicModifyIORef' errors (\known -> (err : known, ())) >> throwIO (err :: SQLError)+ Right () -> pure i+ pure (threads, errors)++-- | The stream delivers 500000 rows before DuckDB computes a row that calls the+-- callback. With a small stream buffer, a later fetch computes that row.+laterFetchQuery :: Query+laterFetchQuery = "SELECT CASE WHEN i >= 500000 AND i % 100000 = 0 THEN own_statement(i) ELSE i END FROM range(1000000) t(i)"++-- | Run an outer call in its own thread, and wait for it at most the test+-- bound. If the bound expires, send an exception to each callback thread.+-- Cleanup can still wait for exception delivery or a native call. Use an+-- external process timeout to limit the complete test run.+boundedOuterCall :: Int -> IORef [ThreadId] -> IO () -> IO (Either SomeException ())+boundedOuterCall micros threads action = do+ (_, done) <- startCaller action+ timeout micros (readMVar done) >>= \case+ Just outcome -> pure outcome+ Nothing -> do+ readIORef threads >>= mapM_ (`throwTo` ErrorCall "test bound: release the callback")+ _ <- await "the outer call did not return after the release of its callbacks" (readMVar done)+ assertFailure "the outer call did not return within the test bound"++-- | Require the error of a callback that waited for the bound. The outer+-- statement has the text of the error of the callback and the type+-- ErrorInputInvalid, because ErrorResourceInUse does not pass. See+-- Note [Errors that leave a callback] in "Database.DuckDB.Simple.Callback".+assertBoundError :: Either SomeException () -> Assertion+assertBoundError = \case+ Left err -> case fromException err of+ Just (failure :: SQLError) -> do+ let message = sqlErrorMessage failure+ sqlErrorType failure @?= Just ErrorInputInvalid+ assertBool+ ("the rule is missing: " <> show failure)+ ("A callback must not run a statement on the connection that runs the callback" `Text.isInfixOf` message)+ assertBool ("the bound is missing: " <> show failure) ("waited 0.2 s" `Text.isInfixOf` message)+ Nothing -> assertFailure ("expected an SQLError, got " <> show err)+ Right () -> assertFailure "the outer call returned without the error of the bound"++-- | Require that each statement of a callback failed with the busy error type.+assertInnerErrors :: IORef [SQLError] -> Assertion+assertInnerErrors errors = do+ known <- readIORef errors+ assertBool "no statement of a callback failed" (not (null known))+ map sqlErrorType known @?= map (const (Just ErrorResourceInUse)) known+ -- | Run one volatile callback before a CPU query that cannot finish in the test window. longQuery :: Query longQuery =- "WITH started AS MATERIALIZED (SELECT query_started() AS seed) \- \SELECT sum(sin((a.i + b.j + started.seed)::DOUBLE)) \- \FROM started, range(1000000) a(i), range(1000000) b(j)"+ "WITH started AS MATERIALIZED (SELECT query_started() AS seed) \+ \SELECT sum(sin((a.i + b.j + started.seed)::DOUBLE)) \+ \FROM started, range(1000000) a(i), range(1000000) b(j)" -- | Publish a thread's outcome while asynchronous exceptions are masked. startCaller :: IO () -> IO (ThreadId, MVar (Either SomeException ())) startCaller action = mask_ do- done <- newEmptyMVar- caller <- forkIOWithUnmask \unmask -> try (unmask action) >>= putMVar done- pure (caller, done)+ done <- newEmptyMVar+ caller <- forkIOWithUnmask \unmask -> try (unmask action) >>= putMVar done+ pure (caller, done) -- | Release gates and interrupt unfinished SQL before the connection can close. withCaller :: Connection -> IO () -> IO () -> (ThreadId -> MVar (Either SomeException ()) -> IO a) -> IO a withCaller conn release action use =- bracket (startCaller action) cleanup (uncurry use)+ bracket (startCaller action) cleanup (uncurry use) where cleanup (_, done) = do- release- await "emergency query cleanup did not finish" (stop done)+ release+ await "emergency query cleanup did not finish" (stop done) stop done = do- finished <- tryReadMVar done- case finished of- Just _ -> pure ()- Nothing -> do- withConnectionHandle conn c_duckdb_interrupt- threadDelay 10000- stop done+ finished <- tryReadMVar done+ case finished of+ Just _ -> pure ()+ Nothing -> do+ withConnectionHandle conn (\raw -> checked (c_duckdb_v2_connection_interrupt raw))+ threadDelay 10000+ stop done -- | Wait until the cancelled caller has reached its interrupt-and-join loop. awaitCleanup :: ThreadId -> IO () awaitCleanup caller = awaitCondition "caller did not enter cancellation cleanup" do- threadStatus caller >>= \case- ThreadBlocked BlockedOnForeignCall -> pure False- ThreadBlocked _ -> pure True- ThreadFinished -> assertFailure "caller finished before the native worker" >> pure False- ThreadDied -> assertFailure "caller died before publishing its result" >> pure False- ThreadRunning -> pure False+ threadStatus caller >>= \case+ ThreadBlocked BlockedOnForeignCall -> pure False+ ThreadBlocked _ -> pure True+ ThreadFinished -> assertFailure "caller finished before the native worker" >> pure False+ ThreadDied -> assertFailure "caller died before publishing its result" >> pure False+ ThreadRunning -> pure False -- | Bound waits in the test thread, independently of the cancelled caller. await :: String -> IO a -> IO a@@ -172,8 +442,8 @@ awaitCondition message condition = await message loop where loop = do- ready <- condition- unless ready (threadDelay 1000 >> loop)+ ready <- condition+ unless ready (threadDelay 1000 >> loop) -- | Open a test gate once; cleanup can safely call this again. signal :: MVar () -> IO ()@@ -182,17 +452,11 @@ -- | Require the requested asynchronous exception to reach the caller. assertAsync :: [AsyncException] -> Either SomeException () -> Assertion assertAsync expected = \case- Left err -> case fromException err of- Just actual -> assertBool ("unexpected asynchronous exception: " <> show actual) (actual `elem` expected)- Nothing -> assertFailure ("expected asynchronous cancellation, got " <> show err)- Right () -> assertFailure "query completed instead of reporting cancellation"+ Left err -> case fromException err of+ Just actual -> assertBool ("unexpected asynchronous exception: " <> show actual) (actual `elem` expected)+ Nothing -> assertFailure ("expected asynchronous cancellation, got " <> show err)+ Right () -> assertFailure "query completed instead of reporting cancellation" -- | Check that an interrupt sender completed normally. assertSucceeded :: Either SomeException () -> Assertion assertSucceeded = either (assertFailure . show) pure---- | The interrupted native query must have released its connection state.-assertReusable :: Connection -> Assertion-assertReusable conn = do- rows <- query_ conn "SELECT 42" :: IO [Only Int64]- rows @?= [Only 42]
+ test/CopyTests.hs view
@@ -0,0 +1,426 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}++-- |+-- Module : CopyTests+-- Description : Native integration checks for managed COPY and multi-file callbacks.+module CopyTests (copyTests) where++import Control.Exception (bracket, try)+import Control.Monad (forM_, void)+import Data.IORef+import Data.Int (Int64)+import Data.List (sort)+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.FFI.V2+import Database.DuckDB.Simple hiding (bind)+import qualified Database.DuckDB.Simple.Catalog as Catalog+import qualified Database.DuckDB.Simple.Copy as Copy+import Database.DuckDB.Simple.FromField (Field (..), FieldValue (..), LogicalTypeRep (..))+import Database.DuckDB.Simple.LogicalRep (withoutAliases)+import Database.DuckDB.Simple.Ok (Ok (..))+import qualified Database.DuckDB.Simple.V2.Copy as V2Copy+import qualified Database.DuckDB.Simple.V2.Registration as R+import Database.DuckDB.Simple.Variant (variantObject)+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Ptr (nullPtr)+import System.Directory (createDirectory, getTemporaryDirectory, removeFile, removePathForcibly)+import System.FilePath ((</>))+import System.IO (hClose, openTempFile)+import System.Mem (performMajorGC)+import Test.Tasty (TestTree, testGroup)+import Test.Tasty.HUnit+import TestUtils (writeIntegers)++-- | Exercise the available COPY directions and multi-file wrapper.+copyTests :: TestTree+copyTests =+ testGroup+ "managed COPY"+ [ testCase "COPY TO scans batches and reports statistics" testCopyTo+ , testCase "row callbacks receive per-partition file paths" testPartitionedCopy+ , testCase "row callbacks decode VARIANT after the registering connection closes" testVariantCopy+ , testCase "COPY FROM runs bind global local and exec callbacks" testCopyFrom+ , testCase "callback exceptions become SQL errors and permit reuse" testCopyFailure+ , testCase "multi-file wrappers support lists globs filename and hive columns" testMultiFile+ , testCase "COPY TO and table descriptions report the logical type id of each column type" testColumnTypeIds+ , testCase "COPY TO snapshots survive but native metadata expires after the callback" testCopyFieldTypes+ ]++-- | Run every COPY TO callback and check the reported file statistics.+testCopyTo :: Assertion+testCopyTo = withDirectory \directory -> withConnection ":memory:" \connection -> do+ seen <- newIORef ([] :: [Text])+ total <- newIORef (0 :: DuckDBIdx)+ let mark phase = atomicModifyIORef' seen (\phases -> (phase : phases, ()))+ bind info _ = do+ mark "bind"+ R.output (c_duckdb_v2_copy_to_bind_get_column_count info) >>= (@?= 1)+ R.setCallbackState (c_duckdb_v2_copy_to_bind_set_bind_data info) ()+ batchSize info _ = mark "batch size" >> R.checked (c_duckdb_v2_copy_to_batch_size_set_target info 1024)+ initialize info _ = do+ mark "init"+ path <- R.output (c_duckdb_v2_copy_to_init_get_file_path info) >>= R.readText+ writeFile (Text.unpack path) ""+ R.setCallbackState (c_duckdb_v2_copy_to_init_set_init_data info) total+ batch info _ = do+ mark "batch"+ count <-+ bracket+ (R.output (c_duckdb_v2_copy_to_batch_take_input info))+ (R.destroy c_duckdb_v2_column_data_collection_destroy)+ (R.output . c_duckdb_v2_column_data_collection_row_count)+ R.setCallbackState (c_duckdb_v2_copy_to_batch_set_batch_data info) count+ flush info _ = do+ mark "flush"+ count <- R.getCallbackState (c_duckdb_v2_copy_to_flush_get_batch_data info) :: IO DuckDBIdx+ state <- R.getCallbackState (c_duckdb_v2_copy_to_flush_get_init_data info) :: IO (IORef DuckDBIdx)+ atomicModifyIORef' state (\n -> (n + count, ()))+ finalize _ _ = mark "finalize"+ statistics info _ = do+ mark "statistics"+ state <- R.getCallbackState (c_duckdb_v2_copy_to_statistics_get_init_data info) :: IO (IORef DuckDBIdx)+ count <- readIORef state+ R.checked (c_duckdb_v2_copy_to_statistics_set_row_count info count)+ R.checked (c_duckdb_v2_copy_to_statistics_set_file_size info 17)+ callbacks =+ (V2Copy.copyToCallbacks batch flush)+ { V2Copy.copyToBindCallback = Just bind+ , V2Copy.copyToBatchSizeCallback = Just batchSize+ , V2Copy.copyToInitCallback = Just initialize+ , V2Copy.copyToFinalizeCallback = Just finalize+ , V2Copy.copyToStatisticsCallback = Just statistics+ }+ V2Copy.registerCopyToCallbacks connection "managed_output" callbacks+ rows <-+ query+ connection+ "COPY (SELECT i FROM range(5000) t(i)) TO ? (FORMAT managed_output, RETURN_STATS)"+ (Only (Text.pack (directory </> "stats")))+ :: IO [[FieldValue]]+ readIORef total >>= (@?= 5000)+ phases <- readIORef seen+ forM_ ["bind", "batch size", "init", "batch", "flush", "finalize", "statistics"] \phase -> assertBool (Text.unpack phase) (phase `elem` phases)+ case rows of+ [FieldText _ : count : size : _] -> do+ number count @?= 5000+ number size @?= 17+ _ -> assertFailure (show rows)++-- | Preserve the row-oriented API while native COPY creates partition files.+testPartitionedCopy :: Assertion+testPartitionedCopy = withDirectory \directory -> withConnection ":memory:" \connection -> do+ files <- newIORef ([] :: [FilePath])+ values <- newIORef ([] :: [Int64])+ Copy.registerCopyToFunction+ connection+ "managed_partitions"+ (\info -> Copy.copyBindColumnTypes info @?= [Copy.LogicalTypeIdBigInt])+ ( \info -> do+ let path = Copy.copyInitFilePath info+ writeFile path ""+ atomicModifyIORef' files (\paths -> (path : paths, ()))+ pure path+ )+ ( \info rows -> do+ let decoded = [value | [Field{fieldValue = FieldInt64 value}] <- rows]+ atomicModifyIORef' values (\previous -> (decoded <> previous, ()))+ appendFile (Copy.copySinkGlobalState info) (unlines (map show decoded))+ )+ (const (pure ()))+ void+ ( execute+ connection+ "COPY (SELECT i, i%2 AS part FROM range(12) t(i)) TO ? (FORMAT managed_partitions, PARTITION_BY(part))"+ (Only (Text.pack (directory </> "partitioned")))+ )+ readIORef values >>= (@?= [0 .. 11]) . sort+ paths <- readIORef files+ length paths @?= 2+ assertBool "partition paths" (all (Text.isInfixOf "part=" . Text.pack) paths)++-- | COPY TO keeps pure type snapshots and expires escaped native actions.+testCopyFieldTypes :: Assertion+testCopyFieldTypes = withDirectory \directory -> withConnection ":memory:" \connection -> do+ received <- newIORef ([] :: [Field])+ snapshots <- newIORef ([] :: [LogicalTypeRep])+ Copy.registerCopyToFunction+ connection+ "field_types"+ (const (pure ()))+ (\info -> writeFile (Copy.copyInitFilePath info) "")+ ( \_ rows -> do+ forM_ rows \fields -> case fields of+ [decimal, json, _] -> do+ runConversion (traverse fieldType [decimal, json]) >>= \case+ Ok types -> atomicModifyIORef' snapshots (\previous -> (previous <> types, ()))+ Errors errors -> assertFailure (show errors)+ _ -> assertFailure "expected three COPY TO columns"+ atomicModifyIORef' received (\previous -> (previous <> concat rows, ()))+ )+ (const (pure ()))+ void+ ( execute+ connection+ "COPY (SELECT 1.5::DECIMAL(4,1) AS d, '{}'::JSON AS j, 7::BIGINT AS unused) TO ? (FORMAT field_types)"+ (Only (Text.pack (directory </> "types")))+ )+ performMajorGC+ readIORef snapshots >>= (@?= [LogicalTypeDecimal 4 1, LogicalTypeAlias "JSON" (LogicalTypeScalar LogicalTypeIdVarchar)])+ readIORef received >>= mapM_ \entry ->+ try (runConversion (fieldType entry)) >>= \case+ Left SQLError{sqlErrorMessage} -> sqlErrorMessage @?= "duckdb-simple: field type scope is closed"+ Right _ -> assertFailure "COPY TO metadata survived its callback"++-- | COPY registrations share the instance and copy nested VARIANT payloads.+testVariantCopy :: Assertion+testVariantCopy = withDirectory \directory -> withEnvironment \environment -> withInstance environment [] \instance' -> do+ received <- newIORef ([] :: [FieldValue])+ withConnectionOn instance' \connection ->+ Copy.registerCopyToFunction+ connection+ "managed_variant"+ (\info -> Copy.copyBindColumnTypes info @?= [Copy.LogicalTypeIdVariant])+ (\info -> writeFile (Copy.copyInitFilePath info) "")+ (\_ rows -> atomicModifyIORef' received (\previous -> (previous <> [fieldValue entry | [entry] <- rows], ())))+ (const (pure ()))+ withConnectionOn instance' \connection -> do+ void+ ( execute+ connection+ "COPY (SELECT {'n': i, 'values': [i, NULL]}::VARIANT FROM range(3) t(i)) TO ? (FORMAT managed_variant)"+ (Only (Text.pack (directory </> "variant")))+ )+ readIORef received >>= (@?= [variantObject [("n", FieldInt64 i), ("values", FieldList [FieldInt64 i, FieldNull])] | i <- [0 .. 2]])++-- | Create a column of each type that DuckDB permits. Check the id that the+-- COPY TO bind info reports for each column. Check the id in the table+-- description of each column that has a scalar type. Check that no column type+-- in the table description has an alias, so the name of each id in+-- logicalTypeIdName is the name that DuckDB gives.+testColumnTypeIds :: Assertion+testColumnTypeIds = withDirectory \directory -> withConnection ":memory:" \connection -> do+ reported <- newIORef []+ Copy.registerCopyToFunction+ connection+ "column_types"+ (writeIORef reported . Copy.copyBindColumnTypes)+ (\info -> writeFile (Copy.copyInitFilePath info) "")+ (\_ _ -> pure ())+ (const (pure ()))+ let named = zip [Text.pack ("c" <> show index) | index <- [0 :: Int ..]] tableColumns+ void (execute_ connection (Query ("CREATE TABLE typed (" <> Text.intercalate ", " [name <> " " <> sql | (name, (sql, _)) <- named] <> ")")))+ -- A table cannot have a TYPE column. A query can.+ void (execute connection "COPY (SELECT NULL::TYPE, * FROM typed) TO ? (FORMAT column_types)" (Only (Text.pack (directory </> "types"))))+ readIORef reported >>= (@?= LogicalTypeIdType : map snd tableColumns)+ sort (LogicalTypeIdType : map snd tableColumns) @?= filter (`notElem` unreachable) [minBound .. maxBound]+ description <- Catalog.describeTable connection "typed"+ [ (name, ident)+ | Catalog.ColumnDescription{Catalog.columnDescriptionName = name, Catalog.columnDescriptionType = LogicalTypeScalar ident} <-+ Catalog.tableDescriptionColumns description+ ]+ @?= [(name, ident) | (name, (_, ident)) <- named, ident `notElem` nested]+ let types = map Catalog.columnDescriptionType (Catalog.tableDescriptionColumns description)+ map withoutAliases types @?= types+ where+ -- DuckDB uses these ids for an unset type, an unresolved parameter, and a+ -- function parameter of any type. SQL cannot declare a column of these types.+ unreachable = [LogicalTypeIdInvalid, LogicalTypeIdUnknown, LogicalTypeIdAny]+ -- LogicalTypeRep has a constructor with the type parameters for these ids.+ nested =+ [ LogicalTypeIdDecimal+ , LogicalTypeIdGeometry+ , LogicalTypeIdStruct+ , LogicalTypeIdList+ , LogicalTypeIdMap+ , LogicalTypeIdEnum+ , LogicalTypeIdUnion+ , LogicalTypeIdArray+ , LogicalTypeIdTuple+ ]+ tableColumns =+ [ ("\"NULL\"", LogicalTypeIdSQLNull)+ , ("BOOLEAN", LogicalTypeIdBoolean)+ , ("TINYINT", LogicalTypeIdTinyInt)+ , ("SMALLINT", LogicalTypeIdSmallInt)+ , ("INTEGER", LogicalTypeIdInteger)+ , ("BIGINT", LogicalTypeIdBigInt)+ , ("DATE", LogicalTypeIdDate)+ , ("TIME", LogicalTypeIdTime)+ , ("TIMESTAMP_S", LogicalTypeIdTimestampSec)+ , ("TIMESTAMP_MS", LogicalTypeIdTimestampMs)+ , ("TIMESTAMP", LogicalTypeIdTimestamp)+ , ("TIMESTAMP_NS", LogicalTypeIdTimestampNs)+ , ("DECIMAL(10, 2)", LogicalTypeIdDecimal)+ , ("FLOAT", LogicalTypeIdFloat)+ , ("DOUBLE", LogicalTypeIdDouble)+ , ("VARCHAR", LogicalTypeIdVarchar)+ , ("BLOB", LogicalTypeIdBlob)+ , ("INTERVAL", LogicalTypeIdInterval)+ , ("UTINYINT", LogicalTypeIdUTinyInt)+ , ("USMALLINT", LogicalTypeIdUSmallInt)+ , ("UINTEGER", LogicalTypeIdUInteger)+ , ("UBIGINT", LogicalTypeIdUBigInt)+ , ("TIMESTAMPTZ", LogicalTypeIdTimestampTz)+ , ("TIMESTAMPTZ_NS", LogicalTypeIdTimestampTzNs)+ , ("TIMETZ", LogicalTypeIdTimeTz)+ , ("TIME_NS", LogicalTypeIdTimeNs)+ , ("BIT", LogicalTypeIdBit)+ , ("BIGNUM", LogicalTypeIdBigNum)+ , ("UHUGEINT", LogicalTypeIdUHugeInt)+ , ("HUGEINT", LogicalTypeIdHugeInt)+ , ("UUID", LogicalTypeIdUUID)+ , ("GEOMETRY", LogicalTypeIdGeometry)+ , ("STRUCT(a INTEGER)", LogicalTypeIdStruct)+ , ("INTEGER[]", LogicalTypeIdList)+ , ("MAP(INTEGER, VARCHAR)", LogicalTypeIdMap)+ , ("ENUM('a', 'b')", LogicalTypeIdEnum)+ , ("UNION(a INTEGER, b VARCHAR)", LogicalTypeIdUnion)+ , ("INTEGER[2]", LogicalTypeIdArray)+ , ("VARIANT", LogicalTypeIdVariant)+ , ("TUPLE(INTEGER, VARCHAR)", LogicalTypeIdTuple)+ ]++-- | Fill one output vector from state created in each native COPY FROM phase.+testCopyFrom :: Assertion+testCopyFrom = withConnection ":memory:" \connection -> do+ phases <- newIORef ([] :: [Text])+ let mark phase = atomicModifyIORef' phases (\previous -> (phase : previous, ()))+ bind info _ = do+ mark "bind"+ R.output (c_duckdb_v2_copy_from_bind_get_column_count info) >>= (@?= 1)+ R.setCallbackState (c_duckdb_v2_copy_from_bind_set_bind_data info) (7 :: Int64)+ R.checked (c_duckdb_v2_copy_from_bind_set_cardinality info 3 1)+ initializeGlobal info _ = do+ mark "global"+ start <- R.getCallbackState (c_duckdb_v2_copy_from_init_global_get_bind_data info)+ state <- newIORef (Just (start :: Int64))+ R.setCallbackState (c_duckdb_v2_copy_from_init_global_set_global_state info) state+ R.checked (c_duckdb_v2_copy_from_init_global_set_max_threads info 1)+ initializeLocal info _ = do+ mark "local"+ R.setCallbackState (c_duckdb_v2_copy_from_init_local_set_local_state info) ()+ executeRows info _ = do+ mark "exec"+ () <- R.getCallbackState (c_duckdb_v2_copy_from_exec_get_local_state info)+ state <- R.getCallbackState (c_duckdb_v2_copy_from_exec_get_global_state info) :: IO (IORef (Maybe Int64))+ start <- atomicModifyIORef' state (\value -> (Nothing, value))+ chunk <- R.output (c_duckdb_v2_copy_from_exec_get_output_chunk info)+ vector <- R.output (c_duckdb_v2_data_chunk_get_vector chunk 0)+ writeIntegers vector (maybe [] (\first -> [first .. first + 2]) start)+ -- DuckDB calls this only to display progress, so the test checks+ -- registration and does not require a call.+ progress info _ = R.checked (c_duckdb_v2_copy_from_progress_set_progress info 0.5)+ callbacks =+ (V2Copy.copyFromCallbacks bind executeRows)+ { V2Copy.copyFromInitGlobalCallback = Just initializeGlobal+ , V2Copy.copyFromInitLocalCallback = Just initializeLocal+ , V2Copy.copyFromProgressCallback = Just progress+ }+ V2Copy.registerCopyFromCallbacks connection "managed_input" callbacks+ void (execute_ connection "CREATE TABLE imported (value BIGINT)")+ void (execute_ connection "COPY imported FROM 'unused' (FORMAT managed_input)")+ query_ connection "SELECT value FROM imported ORDER BY value" >>= (@?= map Only ([7, 8, 9] :: [Int64]))+ called <- readIORef phases+ forM_ ["bind", "global", "local", "exec"] \phase -> assertBool (Text.unpack phase) (phase `elem` called)++-- | Contain exceptions from native COPY callbacks.+testCopyFailure :: Assertion+testCopyFailure = withConnection ":memory:" \connection -> do+ V2Copy.registerCopyFromCallbacks+ connection+ "managed_failure"+ (V2Copy.copyFromCallbacks (\_ _ -> pure ()) (\_ _ -> ioError (userError "managed COPY failure")))+ void (execute_ connection "CREATE TABLE failed_import (value BIGINT)")+ result <- try (execute_ connection "COPY failed_import FROM 'unused' (FORMAT managed_failure)")+ case result of+ Left SQLError{sqlErrorMessage} -> assertBool "callback diagnostic" ("managed COPY failure" `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure "expected callback failure"+ query_ connection "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | Wrap a managed one-file table reader with native file discovery.+testMultiFile :: Assertion+testMultiFile = withDirectory \directory -> withConnection ":memory:" \connection -> do+ let first = directory </> "first.mock"+ second = directory </> "second.mock"+ writeFile first "7"+ writeFile second "11"+ bindInfoCalls <- registerSingleFileReader connection+ Copy.registerMultiFileFunction connection "managed_files" "managed_single" (Copy.MultiFileOptions (Just "managed fixture") (Just "mock"))+ query connection "SELECT value FROM managed_files(?) ORDER BY value" (Only (Text.pack (directory </> "*.mock")))+ >>= (@?= map Only ([7, 11] :: [Int64]))+ let paths = [Text.pack first, Text.pack second]+ query connection "SELECT value, filename FROM managed_files(?::VARCHAR[], filename=true) ORDER BY value" (Only paths)+ >>= (@?= [(7 :: Int64, Text.pack first), (11, Text.pack second)])+ forM_ [("left", 3), ("right", 5 :: Int)] \(part, value) -> do+ let path = directory </> ("part=" <> part)+ createDirectory path+ writeFile (path </> "data.mock") (show value)+ query+ connection+ "SELECT value,part FROM managed_files(?,hive_partitioning=true,union_by_name=true) ORDER BY value"+ (Only (Text.pack (directory </> "part=*" </> "*.mock")))+ >>= (@?= [(3 :: Int64, "left" :: Text), (5, "right")])+ readIORef bindInfoCalls >>= assertBool "table get-bind-info callback" . (> 0)++-- | Register a table function that returns the integer stored in one text file.+registerSingleFileReader :: Connection -> IO (IORef Int)+registerSingleFileReader connection = do+ calls <- newIORef 0+ R.registerTableFunction+ connection+ "managed_single"+ [R.parameter "path" (R.TypeExpression "VARCHAR")]+ (R.tableCallbacks bind executeRows){R.tableInitGlobalCallback = Just initialize, R.tableGetBindInfoCallback = Just (getBindInfo calls)}+ pure calls+ where+ bind input result context = do+ path <- bracket (R.output (c_duckdb_v2_function_bind_get_arg_value input 0)) (R.destroy c_duckdb_v2_value_destroy) \value -> R.output (c_duckdb_v2_value_get_varchar value) >>= R.readText+ value <- read <$> readFile (Text.unpack path) :: IO Int64+ R.setCallbackState (c_duckdb_v2_function_bind_set_bind_data input) value+ bracket+ ( R.output (c_duckdb_v2_context_get_factory context) >>= \factory -> R.withText "BIGINT" (R.output . c_duckdb_v2_factory_create_type_from_text factory)+ )+ (R.destroy c_duckdb_v2_logical_type_destroy)+ \dtype ->+ R.withIdentifier "value" \name -> R.checked (c_duckdb_v2_table_function_bind_add_result_column result name dtype)+ getBindInfo calls info context = do+ modifyIORef' calls (+ 1)+ value <- R.getCallbackState (c_duckdb_v2_table_function_get_bind_info_get_bind_data info) :: IO Int64+ factory <- R.output (c_duckdb_v2_context_get_factory context)+ bracket (R.withText "value" (R.output . c_duckdb_v2_value_create_varchar factory)) (R.destroy c_duckdb_v2_value_destroy) $ \identifier ->+ R.checked (c_duckdb_v2_table_function_get_bind_info_set_column_identifier info 0 (ConstPtr nullPtr) 0 identifier)+ bracket (R.output (c_duckdb_v2_value_create_bigint factory value)) (R.destroy c_duckdb_v2_value_destroy) $ \native ->+ R.withText "fixture_value" $ \(ConstPtr key) -> R.checked (c_duckdb_v2_table_function_get_bind_info_set_option info key native)+ initialize info _ = do+ value <- R.getCallbackState (c_duckdb_v2_table_function_init_global_get_bind_data info)+ state <- newIORef (Just (value :: Int64))+ R.setCallbackState (c_duckdb_v2_table_function_init_global_set_global_state info) state+ executeRows info _ = do+ state <- R.getCallbackState (c_duckdb_v2_table_function_exec_get_global_state info) :: IO (IORef (Maybe Int64))+ value <- atomicModifyIORef' state (\previous -> (Nothing, previous))+ chunk <- R.output (c_duckdb_v2_table_function_exec_get_output_chunk info)+ vector <- R.output (c_duckdb_v2_data_chunk_get_vector chunk 0)+ writeIntegers vector (maybe [] pure value)++-- | Read a nonnegative integer from native statistics.+number :: FieldValue -> Integer+number (FieldWord64 value) = toInteger value+number (FieldInt64 value) = toInteger value+number value = error ("unexpected statistics value " <> show value)++-- | Create and remove a unique directory for native file operations.+withDirectory :: (FilePath -> IO a) -> IO a+withDirectory action = bracket create removePathForcibly action+ where+ create = do+ root <- getTemporaryDirectory+ (path, handle) <- openTempFile root "duckdb-managed-copy"+ hClose handle+ removeFile path+ createDirectory path+ pure path
test/CoreRegressionTests.hs view
@@ -1,153 +1,1027 @@-{-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE ScopedTypeVariables #-}-{-# OPTIONS_GHC -Wno-deprecations #-}---- | Regression tests for result metadata, cursor state, and handle lifetime.-module CoreRegressionTests (coreRegressionTests) where--import Control.Exception (SomeException, throwIO, try)-import Control.Monad (forM_, replicateM_, void)-import Data.IORef (readIORef)-import Data.Int (Int64)-import Data.List (isInfixOf)-import Data.Text (Text)-import Database.DuckDB.FFI.Deprecated (c_duckdb_result_is_streaming)-import Database.DuckDB.Simple-import Database.DuckDB.Simple.FromField (FieldValue)-import Database.DuckDB.Simple.Internal (Statement (statementStream), StatementStream (statementStreamResult), StatementStreamState (..))-import System.Mem (performMajorGC)-import Test.Tasty (TestTree, testGroup)-import Test.Tasty.HUnit---- | Exercise public operations against an in-memory database.-coreRegressionTests :: TestTree-coreRegressionTests =- testGroup- "core regressions"- [ testCase "stream metadata follows parameter types" $- withConnection ":memory:" $ \conn -> do- rows <- fold conn "SELECT ?" (Only (42 :: Int64)) [] collect- rows @?= [Only (42 :: Int64)]- , testCase "cursor uses the supported materialized execution API" $- withConnection ":memory:" $ \conn ->- withStatement conn "SELECT i FROM range(10000) t(i)" $ \stmt -> do- void (nextRow stmt :: IO (Maybe (Only Int64)))- state <- readIORef (statementStream stmt)- case state of- StatementStreamActive stream -> do- streaming <- c_duckdb_result_is_streaming (statementStreamResult stream)- assertBool "expected a materialized native result" (streaming == 0)- _ -> assertFailure "expected an active cursor"- , testCase "stream metadata preserves 64-bit values" $- withConnection ":memory:" $ \conn -> do- rows <- fold conn "SELECT coalesce(?, 1) FROM range(3)" (Only (5000000000 :: Int64)) [] collect- rows @?= replicate 3 (Only (5000000000 :: Int64))- , testCase "stream numeric to Text conversion uses result metadata" $- withConnection ":memory:" $ \conn -> do- rows <- fold conn "SELECT coalesce(?, 'x')" (Only (100 :: Int64)) [] collect- rows @?= [Only ("100" :: Text)]- , testCase "stream metadata follows schema rebinding" $- withConnection ":memory:" $ \conn -> do- void $ execute_ conn "CREATE TABLE rebind_probe(x VARCHAR)"- withStatement conn "SELECT x FROM rebind_probe" $ \stmt -> do- void $ execute_ conn "DROP TABLE rebind_probe"- void $ execute_ conn "CREATE TABLE rebind_probe(x BIGINT)"- void $ execute_ conn "INSERT INTO rebind_probe VALUES (5000000000)"- nextRow stmt >>= (@?= Just (Only (5000000000 :: Int64)))- , testCase "EOF remains exhausted" $- withConnection ":memory:" $ \conn ->- withStatement conn "SELECT 42" $ \stmt -> do- nextRow stmt >>= (@?= Just (Only (42 :: Int64)))- replicateM_ 3 $ nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))- , testCase "DML cursor executes only once" $- withConnection ":memory:" $ \conn -> do- void $ execute_ conn "CREATE TABLE once_probe(x INTEGER)"- withStatement conn "INSERT INTO once_probe VALUES (1)" $ \stmt ->- replicateM_ 3 $ void (nextRow stmt :: IO (Maybe (Only Int64)))- query_ conn "SELECT count(*) FROM once_probe" >>= (@?= [Only (1 :: Int64)])- , testCase "rebinding resets an exhausted cursor" $- withConnection ":memory:" $ \conn ->- withStatement conn "SELECT ?::BIGINT" $ \stmt -> do- bind stmt [toField (1 :: Int64)]- nextRow stmt >>= (@?= Just (Only (1 :: Int64)))- nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))- bind stmt [toField (2 :: Int64)]- nextRow stmt >>= (@?= Just (Only (2 :: Int64)))- , testCase "clear bindings invalidates an active cursor" $- withConnection ":memory:" $ \conn ->- withStatement conn "SELECT ?::BIGINT FROM range(3)" $ \stmt -> do- bind stmt [toField (7 :: Int64)]- void (nextRow stmt :: IO (Maybe (Only Int64)))- clearStatementBindings stmt- assertThrows (nextRow stmt :: IO (Maybe (Only Int64)))- , testCase "fetch errors are not EOF" $- withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn ->- assertThrows (fold_ conn "SELECT CASE WHEN i = 5000 THEN error('late failure') ELSE i END FROM range(10000) t(i)" (0 :: Int64) (\acc (Only n) -> pure (acc + n)))- , testCase "composite cursors preserve values after chunk cleanup" $- withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn ->- forM_- [ "SELECT {'n': i, 'text': 'λ' || i::VARCHAR} FROM range(5000) t(i)"- , "SELECT CASE WHEN i % 2 = 0 THEN union_value(n := i)::UNION(n BIGINT, text VARCHAR) ELSE union_value(text := i::VARCHAR)::UNION(n BIGINT, text VARCHAR) END FROM range(5000) t(i)"- , "SELECT CASE WHEN i % 3 = 0 THEN NULL ELSE {'n': i, 'list': [i, NULL], 'union': union_value(v := i)} END FROM range(5000) t(i)"- ]- $ \sql -> do- eager <- query_ conn sql :: IO [Only FieldValue]- streamed <- reverse <$> fold_ conn sql [] (\acc row -> pure (row : acc))- streamed @?= eager- , testCase "composite cursor failures release the result" $- withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn -> do- let sql = "SELECT {'n': i, 'values': [i, NULL]} FROM range(5000) t(i)"- assertThrows (fold_ conn sql () (\() (Only (_ :: Int64)) -> pure ()))- assertThrows $ fold_ conn sql (0 :: Int) $ \n (Only (_ :: FieldValue)) ->- if n == 2050 then throwIO (userError "composite fold failure") else pure (n + 1)- query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])- , testCase "closed connection rejects active cursor reads" $ do- conn <- open ":memory:"- stmt <- openStatement conn "SELECT i FROM range(3) t(i)"- void (nextRow stmt :: IO (Maybe (Only Int64)))- close conn- assertThrows (nextRow stmt :: IO (Maybe (Only Int64)))- closeStatement stmt- , testCase "NUL in SQL is rejected" $- withConnection ":memory:" $ \conn ->- assertThrows (query_ conn "SELECT 1\0 + 2" :: IO [Only Int64])- , testCase "NUL in database path is rejected" $- assertThrows (withConnection ":memory:\0suffix" (const (pure ())))- , testCase "NUL in configuration value is rejected" $- assertThrows (withConnectionWithConfig ":memory:" [("threads", "1\0suffix")] (const (pure ())))- , testCase "NUL in parameter name is rejected" $- withConnection ":memory:" $ \conn ->- withStatement conn "SELECT $value" $ \stmt ->- assertThrows (namedParameterIndex stmt "value\0suffix")- , testCase "duplicate named bindings are rejected" $- withConnection ":memory:" $ \conn ->- withStatement conn "SELECT $a + $b" $ \stmt -> do- result <- try (bindNamed stmt ["a" := (1 :: Int64), "$a" := (2 :: Int64)])- case result of- Left (_ :: FormatError) -> pure ()- Right () -> assertFailure "expected duplicate binding error before execution"- , testCase "transaction preserves the original exception if rollback fails" $- withConnection ":memory:" $ \conn -> do- result <- try $ withTransaction conn $ do- void $ execute_ conn "ROLLBACK"- throwIO (userError "original transaction failure")- case result of- Left (err :: IOError) -> assertBool "original exception" ("original transaction failure" `isInfixOf` show err)- Right () -> assertFailure "expected transaction failure"- , testCase "GC during the last connection use keeps native owners alive" $ do- -- A bracket cleanup would retain conn and hide early finalization.- conn <- open ":memory:"- createFunction conn "gc_identity" (\(n :: Int64) -> performMajorGC >> pure n)- rows <- query_ conn "SELECT gc_identity(i) FROM range(10) t(i)"- rows @?= map Only ([0 .. 9] :: [Int64])- ]- where- collect acc row = pure (acc <> [row])---- | Require an exception without constraining its concrete type.-assertThrows :: IO a -> Assertion-assertThrows action = do- result <- try (void action)- case result of- Left (_ :: SomeException) -> pure ()- Right () -> assertFailure "expected an exception"+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}++-- |+-- Module : CoreRegressionTests+-- Description : Regression tests for result metadata, cursor state, and handle lifetime.+module CoreRegressionTests (coreRegressionTests) where++import Control.Concurrent (forkFinally, killThread, newEmptyMVar, putMVar, readMVar, takeMVar, threadDelay)+import Control.Exception (AsyncException (ThreadKilled), IOException, SomeException, bracket, finally, throw, throwIO, try)+import Control.Monad (forM_, forever, replicateM, replicateM_, void, when)+import Data.Either (partitionEithers)+import Data.IORef (IORef, atomicModifyIORef', modifyIORef', newIORef, readIORef)+import Data.Int (Int64)+import Data.List (isInfixOf)+import Data.Text (Text)+import qualified Data.Text as Text+import qualified Database.DuckDB.FFI.V2 as Raw+import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Catalog as Catalog+import qualified Database.DuckDB.Simple.Config as Config+import qualified Database.DuckDB.Simple.Copy as Copy+import Database.DuckDB.Simple.FromField (BitString (..), FieldValue)+import Database.DuckDB.Simple.Internal (+ DuckDBV2ResultHandle (..),+ Statement (statementStream),+ StatementStream (statementStreamResult),+ StatementStreamState (..),+ checked,+ executePreparedResult,+ nullHandle,+ output,+ withCollectionStatement,+ withConnectionFactory,+ withConnectionHandle,+ )+import qualified Database.DuckDB.Simple.Internal as Internal+import qualified Database.DuckDB.Simple.Logging as Logging+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..))+import Database.DuckDB.Simple.Pragma (PragmaCallback (..), registerPragmaFunction)+import Database.DuckDB.Simple.Registration (+ Replacement (..),+ TypeDeclaration (..),+ parameter,+ registerInstanceReplacementScan,+ registerReplacementScan,+ )+import qualified Database.DuckDB.Simple.Streaming as Streaming+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Marshal.Array (withArray)+import System.Directory (canonicalizePath, createDirectory, getTemporaryDirectory, listDirectory, removeFile, removePathForcibly)+import System.FilePath ((</>))+import System.IO (hClose, openBinaryTempFile)+import System.Mem (performMajorGC)+import System.Timeout (timeout)+import Test.Tasty (TestTree, testGroup)+import Test.Tasty.HUnit+import TestUtils (assertInputError, assertReusable, assertSqlErrorType, raiseCapabilities)++-- | Exercise public operations against an in-memory database.+coreRegressionTests :: TestTree+coreRegressionTests =+ testGroup+ "core regressions"+ [ testCase "stream metadata follows parameter types" $+ withConnection ":memory:" $ \conn -> do+ rows <- fold conn "SELECT ?" (Only (42 :: Int64)) [] collect+ rows @?= [Only (42 :: Int64)]+ , testCase "cursor uses the supported materialized execution API" $+ withConnection ":memory:" $ \conn ->+ withStatement conn "SELECT i FROM range(10000) t(i)" $ \stmt -> do+ void (nextRow stmt :: IO (Maybe (Only Int64)))+ state <- readIORef (statementStream stmt)+ case state of+ StatementStreamActive stream -> do+ assertBool "expected a released execution cursor" (statementStreamResult stream == nullHandle)+ _ -> assertFailure "expected an active cursor"+ , testCase "stream metadata preserves 64-bit values" $+ withConnection ":memory:" $ \conn -> do+ rows <- fold conn "SELECT coalesce(?, 1) FROM range(3)" (Only (5000000000 :: Int64)) [] collect+ rows @?= replicate 3 (Only (5000000000 :: Int64))+ , testCase "stream numeric to Text conversion uses result metadata" $+ withConnection ":memory:" $ \conn -> do+ rows <- fold conn "SELECT coalesce(?, 'x')" (Only (100 :: Int64)) [] collect+ rows @?= [Only ("100" :: Text)]+ , testCase "stream metadata follows schema rebinding" $+ withConnection ":memory:" $ \conn -> do+ void $ execute_ conn "CREATE TABLE rebind_probe(x VARCHAR)"+ withStatement conn "SELECT x FROM rebind_probe" $ \stmt -> do+ void $ execute_ conn "DROP TABLE rebind_probe"+ void $ execute_ conn "CREATE TABLE rebind_probe(x BIGINT)"+ void $ execute_ conn "INSERT INTO rebind_probe VALUES (5000000000)"+ nextRow stmt >>= (@?= Just (Only (5000000000 :: Int64)))+ , testCase "reset metadata follows an altered result schema" $+ withConnection ":memory:" $ \conn -> do+ void $ execute_ conn "CREATE TABLE metadata_rebind(x BIGINT)"+ withStatement conn "SELECT * FROM metadata_rebind" $ \stmt -> do+ columnCount stmt >>= (@?= 1)+ void $ execute_ conn "ALTER TABLE metadata_rebind ADD COLUMN y VARCHAR"+ clearStatementBindings stmt+ columnCount stmt >>= (@?= 2)+ columnName stmt 1 >>= (@?= "y")+ , testCase "EOF remains exhausted" $+ withConnection ":memory:" $ \conn ->+ withStatement conn "SELECT 42" $ \stmt -> do+ nextRow stmt >>= (@?= Just (Only (42 :: Int64)))+ replicateM_ 3 $ nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ , testCase "DML cursor executes only once" $+ withConnection ":memory:" $ \conn -> do+ void $ execute_ conn "CREATE TABLE once_probe(x INTEGER)"+ withStatement conn "INSERT INTO once_probe VALUES (1)" $ \stmt -> do+ nextRow stmt >>= (@?= Just (Only (1 :: Int64)))+ replicateM_ 3 $ nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ query_ conn "SELECT count(*) FROM once_probe" >>= (@?= [Only (1 :: Int64)])+ , testCase "queries and folds give the changed-row count of a DML statement" $+ withConnection ":memory:" $ \conn -> do+ void (execute_ conn "CREATE TABLE changed_rows(value BIGINT)")+ (query_ conn "INSERT INTO changed_rows VALUES (1),(2)" :: IO [Only Int64]) >>= (@?= [Only 2])+ (query conn "INSERT INTO changed_rows VALUES (?)" (Only (3 :: Int64)) :: IO [Only Int64]) >>= (@?= [Only 1])+ (queryNamed conn "UPDATE changed_rows SET value=value+10 WHERE value<=$limit" ["limit" := (2 :: Int64)] :: IO [Only Int64])+ >>= (@?= [Only 2])+ (query_ conn "DELETE FROM changed_rows WHERE false" :: IO [Only Int64]) >>= (@?= [Only 0])+ fold_ conn "INSERT INTO changed_rows VALUES (4)" [] collect >>= (@?= [Only (1 :: Int64)])+ fold conn "INSERT INTO changed_rows VALUES (?)" (Only (5 :: Int64)) [] collect >>= (@?= [Only (1 :: Int64)])+ foldNamed conn "UPDATE changed_rows SET value=value+10 WHERE value<=$limit" ["limit" := (4 :: Int64)] [] collect+ >>= (@?= [Only (2 :: Int64)])+ query_ conn "SELECT value FROM changed_rows ORDER BY value" >>= (@?= map Only [5, 11, 12, 13, 14 :: Int64])+ , testCase "rebinding resets an exhausted cursor" $+ withConnection ":memory:" $ \conn ->+ withStatement conn "SELECT ?::BIGINT" $ \stmt -> do+ bind stmt [toField (1 :: Int64)]+ nextRow stmt >>= (@?= Just (Only (1 :: Int64)))+ nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ bind stmt [toField (2 :: Int64)]+ nextRow stmt >>= (@?= Just (Only (2 :: Int64)))+ , testCase "failed replacement retains complete bindings and resets the cursor" $+ withConnection ":memory:" $ \conn ->+ withStatement conn "SELECT ?::BIGINT, ?::BIGINT, ?::BIGINT FROM range(2)" $ \stmt -> do+ let original = (1, 2, 3) :: (Int64, Int64, Int64)+ bind stmt [toField (1 :: Int64), toField (2 :: Int64), toField (3 :: Int64)]+ nextRow stmt >>= (@?= Just original)+ assertInputError (bind stmt [toField (4 :: Int64), toField (BitString 8 "x"), toField (6 :: Int64)])+ nextRow stmt >>= (@?= Just original)+ interrupted <- try (bind stmt [toField (7 :: Int64), throw ThreadKilled, toField (9 :: Int64)]) :: IO (Either AsyncException ())+ interrupted @?= Left ThreadKilled+ nextRow stmt >>= (@?= Just original)+ nextRow stmt >>= (@?= Just original)+ nextRow stmt >>= (@?= (Nothing :: Maybe (Int64, Int64, Int64)))+ bind stmt [toField (10 :: Int64), toField (11 :: Int64), toField (12 :: Int64)]+ nextRow stmt >>= (@?= Just (10 :: Int64, 11 :: Int64, 12 :: Int64))+ , testCase "clear bindings invalidates an active cursor" $+ withConnection ":memory:" $ \conn ->+ withStatement conn "SELECT ?::BIGINT FROM range(3)" $ \stmt -> do+ bind stmt [toField (7 :: Int64)]+ void (nextRow stmt :: IO (Maybe (Only Int64)))+ clearStatementBindings stmt+ assertSqlErrorType ErrorInputInvalid (nextRow stmt :: IO (Maybe (Only Int64)))+ , testCase "fetch errors are not EOF" $+ withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn ->+ assertSqlErrorType+ ErrorInputInvalid+ ( fold_+ conn+ "SELECT CASE WHEN i = 5000 THEN error('late failure') ELSE i END FROM range(10000) t(i)"+ (0 :: Int64)+ (\acc (Only n) -> pure (acc + n))+ )+ , testCase "composite cursors preserve values after chunk cleanup" $+ withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn ->+ forM_+ [ "SELECT {'n': i, 'text': 'λ' || i::VARCHAR} FROM range(5000) t(i)"+ , "SELECT CASE WHEN i % 2 = 0 THEN union_value(n := i)::UNION(n BIGINT, text VARCHAR) ELSE union_value(text := i::VARCHAR)::UNION(n BIGINT, text VARCHAR) END FROM range(5000) t(i)"+ , "SELECT CASE WHEN i % 3 = 0 THEN NULL ELSE {'n': i, 'list': [i, NULL], 'union': union_value(v := i)} END FROM range(5000) t(i)"+ ]+ $ \sql -> do+ eager <- query_ conn sql :: IO [Only FieldValue]+ streamed <- reverse <$> fold_ conn sql [] (\acc row -> pure (row : acc))+ streamed @?= eager+ , -- The case of the leak test for a statement that closes with a live cursor+ -- checks the release of the chunks.+ testCase "composite fold failures raise their error and leave the connection usable" $+ withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn -> do+ let sql = "SELECT {'n': i, 'values': [i, NULL]} FROM range(5000) t(i)"+ -- A STRUCT does not convert to Int64. The conversion error names the+ -- Haskell type and the query.+ conversion <- try (fold_ conn sql () (\() (Only (_ :: Int64)) -> pure ()))+ case conversion of+ Left err -> do+ assertBool (Text.unpack (sqlErrorMessage err)) ("Int64" `Text.isInfixOf` sqlErrorMessage err)+ sqlErrorQuery err @?= Just sql+ Right () -> assertFailure "expected the conversion error"+ stepFailure <- try $ fold_ conn sql (0 :: Int) $ \n (Only (_ :: FieldValue)) ->+ if n == 2050 then throwIO (userError "composite fold failure") else pure (n + 1)+ case stepFailure of+ Left (err :: IOException) -> show err @?= "user error (composite fold failure)"+ Right count -> assertFailure ("expected the step failure, got " <> show count)+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "closed connection rejects active cursor reads" $ do+ conn <- open ":memory:"+ stmt <- openStatement conn "SELECT i FROM range(3) t(i)"+ void (nextRow stmt :: IO (Maybe (Only Int64)))+ close conn+ assertThrows (nextRow stmt :: IO (Maybe (Only Int64)))+ closeStatement stmt+ , testCase "NUL in SQL is rejected" $+ withConnection ":memory:" $ \conn ->+ assertInputError (query_ conn "SELECT 1\0 + 2" :: IO [Only Int64])+ , testCase "NUL in database path is rejected" $+ assertInputError (withConnection ":memory:\0suffix" (const (pure ())))+ , testCase "NUL in configuration value is rejected" $+ assertInputError (withConnectionWithConfig ":memory:" [("threads", "1\0suffix")] (const (pure ())))+ , testCase "a column index out of bounds raises the input error with the query" $+ withConnection ":memory:" $ \conn ->+ withStatement conn "SELECT 1 AS a" $ \stmt ->+ forM_ [-1, 1] $ \index -> do+ assertInputError (columnName stmt index)+ result <- try (columnName stmt index)+ case result of+ Left (err :: SQLError) -> sqlErrorQuery err @?= Just "SELECT 1 AS a"+ Right name -> assertFailure ("expected the column index error, got " <> show name)+ , testCase "NUL in parameter name is rejected" $+ withConnection ":memory:" $ \conn ->+ withStatement conn "SELECT $value" $ \stmt ->+ assertInputError (namedParameterIndex stmt "value\0suffix")+ , testCase "duplicate named bindings are rejected" $+ withConnection ":memory:" $ \conn ->+ withStatement conn "SELECT $foo + $bar" $ \stmt ->+ forM_ ["$foo", "FOO"] $ \duplicate -> do+ result <- try (bindNamed stmt ["foo" := (1 :: Int64), duplicate := (2 :: Int64)])+ case result of+ Left (_ :: FormatError) -> pure ()+ Right () -> assertFailure "expected duplicate binding error before execution"+ , testCase "named bindings retain distinct Unicode names" $+ withConnection ":memory:" $ \conn ->+ (queryNamed conn "SELECT $Straße::BIGINT, $STRASSE::BIGINT" ["Straße" := (1 :: Int64), "STRASSE" := (2 :: Int64)] :: IO [(Int64, Int64)])+ >>= (@?= [(1, 2)])+ , testCase "numeric named bindings retain sparse and reordered parameter names" $+ withConnection ":memory:" $ \conn -> do+ (queryNamed conn "SELECT $3::BIGINT" ["3" := (33 :: Int64)] :: IO [Only Int64]) >>= (@?= [Only 33])+ (queryNamed conn "SELECT $3::BIGINT, $1::BIGINT, $3::BIGINT" ["1" := (11 :: Int64), "3" := (33 :: Int64)] :: IO [(Int64, Int64, Int64)])+ >>= (@?= [(33, 11, 33)])+ (query conn "SELECT ?::BIGINT, ?::BIGINT" ((1 :: Int64), (2 :: Int64)) :: IO [(Int64, Int64)]) >>= (@?= [(1, 2)])+ (query conn "SELECT $02::BIGINT, $01::BIGINT" ((11 :: Int64), (22 :: Int64)) :: IO [(Int64, Int64)]) >>= (@?= [(22, 11)])+ , testCase "a quoted parameter can start with a binding prefix" $+ withConnection ":memory:" $ \conn -> do+ let sql = "SELECT $foo::BIGINT, $\"$foo\"::BIGINT, $\":foo\"::BIGINT, $\"@foo\"::BIGINT"+ names = ["$FOO", "$$foo", "$:foo", "$@foo"]+ withStatement conn sql $ \stmt -> do+ mapM (namedParameterIndex stmt) names >>= (@?= map Just [1 .. 4])+ bindNamed stmt [name := value | (name, value) <- zip names [1 .. 4 :: Int64]]+ nextRow stmt >>= (@?= Just (1 :: Int64, 2 :: Int64, 3 :: Int64, 4 :: Int64))+ withStatement conn "SELECT $\"$1\"::BIGINT" $ \stmt -> do+ outcome <- try (bind stmt [toField (5 :: Int64)])+ case outcome of+ Left (_ :: FormatError) -> pure ()+ Right () -> assertFailure "a literal dollar sign is part of the name"+ bindNamed stmt ["$$1" := (5 :: Int64)]+ nextRow stmt >>= (@?= Just (Only (5 :: Int64)))+ , testCase "quoted numeric names remain named parameters" $+ withConnection ":memory:" $ \conn ->+ forM_+ [ ("SELECT $foo::BIGINT, $\"2\"::BIGINT", "foo")+ , ("SELECT $\"01\"::BIGINT, $\"2\"::BIGINT", "01")+ , ("SELECT $`0`::BIGINT, $`2`::BIGINT", "0")+ ]+ $ \(sql, first) -> do+ (queryNamed conn sql [first := (11 :: Int64), "2" := (22 :: Int64)] :: IO [(Int64, Int64)]) >>= (@?= [(11, 22)])+ Internal.withParsedStatements conn sql $ \parsed -> case parsed of+ [raw] -> bracket (Internal.createStatementFromParsed conn "SELECT ?, ?" raw) closeStatement $ \stmt -> do+ namedParameterIndex stmt first >>= (@?= Just 1)+ outcome <- try (bind stmt [toField (11 :: Int64), toField (22 :: Int64)])+ case outcome of+ Left (_ :: FormatError) -> pure ()+ Right () -> assertFailure "expected positional binding rejection for quoted names"+ _ -> assertFailure "expected one parsed statement"+ , testCase "statements bind several thousand parameters by position and by name" $+ withConnection ":memory:" $ \conn -> do+ let count = 3000 :: Int+ values = [1 .. fromIntegral count] :: [Int64]+ listSum items = Query ("SELECT list_sum([" <> Text.intercalate "," items <> "])")+ name index = "p" <> Text.pack (show index)+ (query conn (listSum (replicate count "?::BIGINT")) values :: IO [Only Int64]) >>= (@?= [Only (sum values)])+ ( queryNamed+ conn+ (listSum ["$" <> name index <> "::BIGINT" | index <- [1 .. count]])+ [name index := value | (index, value) <- zip [1 :: Int ..] values]+ :: IO [Only Int64]+ )+ >>= (@?= [Only (sum values)])+ -- Dense numbered parameters bind by position despite their SQL order.+ (query conn "SELECT $2::BIGINT, $1::BIGINT, $3::BIGINT" (1 :: Int64, 2 :: Int64, 3 :: Int64) :: IO [(Int64, Int64, Int64)])+ >>= (@?= [(2, 1, 3)])+ (queryNamed conn "SELECT $1::BIGINT, $3::BIGINT" ["1" := (11 :: Int64), "3" := (33 :: Int64)] :: IO [(Int64, Int64)]) >>= (@?= [(11, 33)])+ assertThrows (query conn "SELECT $1::BIGINT, $3::BIGINT" (11 :: Int64, 33 :: Int64) :: IO [(Int64, Int64)])+ , testCase "transaction preserves the original exception if rollback fails" $+ withConnection ":memory:" $ \conn -> do+ result <- try $ withTransaction conn $ do+ void $ execute_ conn "ROLLBACK"+ throwIO (userError "original transaction failure")+ case result of+ Left (err :: IOError) -> assertBool "original exception" ("original transaction failure" `isInfixOf` show err)+ Right () -> assertFailure "expected transaction failure"+ , testCase "a callback exception of each kind fails its statement in a transaction and leaves no rows" $+ mapM_ runFailingCallback failingCallbacks+ , testCase "materialization reports a late error before the first callback" $+ withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn -> do+ calls <- newIORef (0 :: Int)+ assertSqlErrorType ErrorInputInvalid $ fold_ conn "SELECT CASE WHEN i = 5000 THEN error('late failure') ELSE i END FROM range(10000) t(i)" () $ \() (Only (_ :: Int64)) ->+ modifyIORef' calls (+ 1)+ readIORef calls >>= (@?= 0)+ , testCase "a materialized callback can query the same connection" $+ withConnection ":memory:" $ \conn -> do+ count <- fold_ conn "SELECT i FROM range(3) t(i)" (0 :: Int) $ \acc (Only (_ :: Int64)) -> do+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ pure (acc + 1)+ count @?= 3+ , testCase "threads share a connection for default queries" $+ do+ -- See Note [Capabilities of the tests] in "TestUtils".+ raiseCapabilities 4+ withConnection ":memory:" $ \conn -> do+ void (execute_ conn "CREATE TABLE shared_rows(worker BIGINT, i BIGINT)")+ start <- newEmptyMVar+ completed <- replicateM 8 newEmptyMVar+ forM_ (zip [1 :: Int64 ..] completed) $ \(worker, result) ->+ void $ flip forkFinally (putMVar result) $ do+ readMVar start+ forM_ [1 .. 50 :: Int64] $ \i -> do+ query conn "SELECT ? + ?" (worker, i) >>= (@?= [Only (worker + i)])+ execute conn "INSERT INTO shared_rows VALUES (?, ?)" (worker, i) >>= (@?= 1)+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ -- The threads share the VARIANT decoder state of the connection.+ query conn "SELECT ?::VARIANT, [?, ?]::VARIANT" (worker + i, worker, i) >>= (@?= [(worker + i, [worker, i])])+ fold_ conn "SELECT i FROM range(3) t(i)" (0 :: Int64) (\acc (Only n) -> pure (acc + n)) >>= (@?= 3)+ execute_ conn "UPDATE shared_rows SET i = i WHERE worker = 0" >>= (@?= 0)+ putMVar start ()+ outcomes <- mapM takeMVar completed+ case partitionEithers outcomes of+ ([], _) -> pure ()+ (failure : _, _) -> throwIO failure+ query_ conn "SELECT count(*) FROM shared_rows" >>= (@?= [Only (400 :: Int64)])+ , testCase "ad hoc SQL executes statement-expanded groups" $+ withConnection ":memory:" $ \conn -> do+ void+ ( execute_+ conn+ "CREATE TABLE pivot_probe(category VARCHAR, yr INTEGER, amount BIGINT); INSERT INTO pivot_probe VALUES ('a',2020,1),('a',2021,2)"+ )+ query_ conn "PIVOT pivot_probe ON yr USING sum(amount)" >>= (@?= [("a" :: Text, 1 :: Int64, 2 :: Int64)])+ void (execute_ conn "ALTER TABLE pivot_probe ADD COLUMN extra DOUBLE DEFAULT random()")+ query_ conn "SELECT count(*) FROM pivot_probe WHERE extra IS NOT NULL" >>= (@?= [Only (2 :: Int64)])+ , testCase "a prepared PIVOT refreshes its result schema on each execution" $+ withConnection ":memory:" $ \conn -> do+ void+ ( execute_+ conn+ "CREATE TABLE prepared_pivot(category VARCHAR, yr INTEGER, amount BIGINT); INSERT INTO prepared_pivot VALUES ('a',2020,1),('a',2021,2)"+ )+ withStatement conn "PIVOT prepared_pivot ON yr USING sum(amount)" $ \stmt -> do+ nextRow stmt >>= (@?= Just ("a" :: Text, 1 :: Int64, 2 :: Int64))+ nextRow stmt >>= (@?= (Nothing :: Maybe (Text, Int64, Int64)))+ void (execute_ conn "INSERT INTO prepared_pivot VALUES ('a',2022,3)")+ clearStatementBindings stmt+ nextRow stmt >>= (@?= Just ("a" :: Text, 1 :: Int64, 2 :: Int64, 3 :: Int64))+ nextRow stmt >>= (@?= (Nothing :: Maybe (Text, Int64, Int64, Int64)))+ , testCase "a scoped collection statement retains named input and parameter metadata" $+ withConnection ":memory:" $ \conn ->+ withConnectionHandle conn $ \native -> withConnectionFactory native $ \factory ->+ withStatement conn "SELECT 42::BIGINT" $ \source ->+ bracket (executePreparedResult source) (Internal.destroy Raw.c_duckdb_v2_result_destroy) $ \result ->+ bracket (output (Raw.c_duckdb_v2_result_fetch_chunk result)) (Internal.destroy Raw.c_duckdb_v2_data_chunk_destroy) $ \chunk -> do+ bracket (output (Raw.c_duckdb_v2_result_fetch_chunk result)) (Internal.destroy Raw.c_duckdb_v2_data_chunk_destroy) $ \end ->+ end @?= nullHandle+ vector <- output (Raw.c_duckdb_v2_data_chunk_get_vector chunk 0)+ bracket (output (Raw.c_duckdb_v2_vector_get_logical_type vector)) (Internal.destroy Raw.c_duckdb_v2_logical_type_destroy) $ \logical ->+ withArray [logical] $ \types ->+ bracket+ (output (Raw.c_duckdb_v2_column_data_collection_create factory (ConstPtr types) 1))+ (Internal.destroy Raw.c_duckdb_v2_column_data_collection_destroy)+ $ \collection -> do+ let checkRows :: Maybe (Only Int64) -> IO ()+ checkRows expected = withCollectionStatement conn "SELECT payload + ?::BIGINT FROM input_rows" [("input_rows", collection, ["payload"])] $ \stmt -> do+ bind stmt [toField (1 :: Int64)]+ nextRow stmt >>= (@?= expected)+ nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ checkRows Nothing+ bracket+ (output (Raw.c_duckdb_v2_column_data_collection_append_state_create collection))+ (Internal.destroy Raw.c_duckdb_v2_column_data_collection_append_state_destroy)+ $ \appendState ->+ checked (Raw.c_duckdb_v2_column_data_collection_append collection appendState chunk)+ checkRows (Just (Only (43 :: Int64)))+ , testCase "connections share an explicit instance" $+ withEnvironment $ \environment -> withInstance environment [] $ \instance_ -> do+ attachDatabase instance_ ":memory:" "shared" []+ setDefaultDatabase instance_ "shared"+ withConnectionOn instance_ $ \first -> withConnectionOn instance_ $ \second -> do+ void (execute_ first "CREATE TABLE shared_probe(x BIGINT); INSERT INTO shared_probe VALUES (42)")+ query_ second "SELECT x FROM shared_probe" >>= (@?= [Only (42 :: Int64)])+ , testCase "a detached database is not reachable" $+ withEnvironment $ \environment -> withInstance environment [] $ \instance_ -> do+ attachDatabase instance_ ":memory:" "shared" []+ setDefaultDatabase instance_ "shared"+ attachDatabase instance_ ":memory:" "side" []+ withConnectionOn instance_ $ \conn -> do+ void (execute_ conn "CREATE TABLE side.probe(x BIGINT); INSERT INTO side.probe VALUES (7)")+ query_ conn "SELECT x FROM side.probe" >>= (@?= [Only (7 :: Int64)])+ detachDatabase instance_ "side"+ withConnectionOn instance_ $ \conn -> do+ query_ conn "SELECT count(*) FROM duckdb_databases() WHERE database_name = 'side'" >>= (@?= [Only (0 :: Int64)])+ assertSqlErrorType ErrorDatabaseCatalog (query_ conn "SELECT x FROM side.probe" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ -- The native detach call gives this type. The SQL statement DETACH gives+ -- a Binder Error.+ assertSqlErrorType ErrorInputInvalid (detachDatabase instance_ "side")+ , testCase "NUL in attach options and aliases is rejected" $+ withEnvironment $ \environment -> withInstance environment [] $ \instance_ -> do+ attachDatabase instance_ ":memory:" "shared" []+ setDefaultDatabase instance_ "shared"+ forM_+ [ attachDatabase instance_ ":memory:" "side" [("read_only\0suffix", "false")]+ , attachDatabase instance_ ":memory:" "side" [("read_only", "false\0suffix")]+ , detachDatabase instance_ "shared\0suffix"+ , setDefaultDatabase instance_ "shared\0suffix"+ ]+ $ \action -> do+ result <- try action+ case result of+ Left (err :: SQLError) -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ assertBool "NUL diagnostic" ("contains NUL" `Text.isInfixOf` sqlErrorMessage err)+ Right () -> assertFailure "expected a NUL error"+ withConnectionOn instance_ $ \conn ->+ query_ conn "SELECT current_database(), count(*) FILTER (WHERE database_name = 'side') FROM duckdb_databases()"+ >>= (@?= [("shared" :: Text, 0 :: Int64)])+ , testCase "parent close releases concurrently created children" $+ do+ -- See Note [Capabilities of the tests] in "TestUtils".+ raiseCapabilities 2+ withEnvironment $ \environment -> withInstance environment [("threads", "1"), ("async_threads", "1")] $ \instance_ ->+ withConcurrentChildren 64 (connect instance_) close $ \connections -> do+ closeInstance instance_+ forM_ connections $ \connection -> do+ state <- readIORef (Internal.connectionState connection)+ case state of+ Internal.ConnectionClosed -> pure ()+ Internal.ConnectionOpen{} -> assertFailure "parent close missed a connection"+ assertThrows (query_ connection "SELECT 42" :: IO [Only Int64])+ withEnvironment $ \environment ->+ withConcurrentChildren 16 (createInstance environment [("threads", "1"), ("async_threads", "1")]) closeInstance $ \instances -> do+ closeEnvironment environment+ forM_ instances $ \instance_ -> do+ state <- readIORef (Internal.instanceState instance_)+ case state of+ Internal.InstanceClosed -> pure ()+ Internal.InstanceOpen{} -> assertFailure "parent close missed an instance"+ assertThrows (connect instance_)+ , testCase "startup configuration sets the whitelist before disabling external access" $ do+ path <- canonicalizePath "/tmp/duckdb-whitelist-probe"+ withConnectionWithConfig ":memory:" [("enable_external_access", "false"), ("allowed_paths", "['/tmp/duckdb-whitelist-probe']")] $ \conn ->+ query conn "SELECT current_setting('enable_external_access'), list_contains(current_setting('allowed_paths'), ?)" (Only path)+ >>= (@?= [(False, True)])+ , testCase "environment close ends the live cursors of its connections" $ do+ environment <- createEnvironment+ instance_ <- createInstance environment []+ attachDatabase instance_ ":memory:" "shared" []+ setDefaultDatabase instance_ "shared"+ conn <- connect instance_+ stmt <- openStatement conn "SELECT i FROM range(5000) t(i)"+ nextRow stmt >>= (@?= Just (Only (0 :: Int64)))+ replicateM_ 20 (void (openStatement conn "SELECT 42"))+ performMajorGC+ closeEnvironment environment+ cursor <- readIORef (statementStream stmt)+ case cursor of+ StatementStreamActive _ -> assertFailure "the environment close left the cursor of the statement active"+ _ -> pure ()+ assertThrows (nextRow stmt :: IO (Maybe (Only Int64)))+ assertThrows (query_ conn "SELECT 42" :: IO [Only Int64])+ closeStatement stmt+ close conn+ closeInstance instance_+ closeEnvironment environment+ , testCase "a refused environment close keeps the environment open" $ do+ environment <- createEnvironment+ state <- readIORef (Internal.environmentState environment)+ raw <- case state of+ Internal.EnvironmentOpen raw _ -> pure raw+ Internal.EnvironmentClosed -> assertFailure "a new environment is closed"+ -- The library does not track this instance, so it does not close it.+ untracked <- output (Raw.c_duckdb_v2_instance_create raw)+ refused <- try (closeEnvironment environment)+ case refused of+ Left (err :: SQLError) -> sqlErrorType err @?= Just ErrorResourceInUse+ Right () -> assertFailure "expected a refused close"+ -- Keep this instance open. The second close must find it in the child registry.+ instance_ <- createInstance environment []+ withConnectionOn instance_ $ \conn -> query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ Internal.destroy Raw.c_duckdb_v2_instance_destroy untracked+ closeEnvironment environment+ assertThrows (connect instance_)+ closed <- try (createInstance environment [])+ case closed of+ Left (err :: SQLError) -> sqlErrorMessage err @?= "duckdb-simple: environment is closed"+ Right _ -> assertFailure "expected a closed environment"+ , testCase "GC releases an unclosed convenience connection" $ do+ directory <- getTemporaryDirectory+ (path, file) <- openBinaryTempFile directory "duckdb-v2-gc.duckdb"+ hClose file+ removeFile path+ let allocate = do+ conn <- openWithConfig path [("threads", "1")]+ void (execute_ conn "CREATE TABLE gc_probe AS SELECT 42::BIGINT AS x")+ wait = do+ performMajorGC+ reopened <- try $ withConnectionWithConfig path [("threads", "1")] $ \conn ->+ query_ conn "SELECT x FROM gc_probe" >>= (@?= [Only (42 :: Int64)])+ case reopened of+ Right () -> pure ()+ Left (_ :: SQLError) -> threadDelay 10000 >> wait+ flip finally (removeFile path) $ do+ allocate+ released <- timeout 60000000 wait+ released @?= Just ()+ , testCase "GC during the last connection use keeps native owners alive" $ do+ -- A bracket cleanup would retain conn and hide early finalization.+ conn <- open ":memory:"+ createFunction conn "gc_identity" (\(n :: Int64) -> performMajorGC >> pure n)+ rows <- query_ conn "SELECT gc_identity(i) FROM range(10) t(i)"+ rows @?= map Only ([0 .. 9] :: [Int64])+ , testCase "GC during the last statement use keeps cached schemas alive" $+ withConnectionWithConfig ":memory:" [("threads", "1")] $ \conn -> do+ let name = Text.replicate 4096 "x"+ sql = Query ("SELECT 1 AS \"" <> name <> "\"")+ collectGarbage = forever (performMajorGC >> threadDelay 1000)+ done <- newEmptyMVar+ -- A bracket around each statement would keep it alive after the read.+ bracket (forkFinally collectGarbage (putMVar done)) (\worker -> killThread worker >> void (readMVar done)) $ \_ ->+ replicateM_ 100 $ do+ (openStatement conn sql >>= columnCount) >>= (@?= 1)+ (openStatement conn sql >>= \stmt -> columnName stmt 0) >>= (@?= name)+ (openStatement conn "SELECT ?::BIGINT" >>= \stmt -> Internal.preparedParameterType stmt 1)+ >>= (@?= Just (LogicalTypeScalar Internal.LogicalTypeIdBigInt))+ , testCase "a prepare of no statement or of two statements raises its error" $ withConnection ":memory:" $ \conn ->+ forM_+ [ ("", "duckdb-simple: SQL query contains no statement")+ , ("-- a comment only", "duckdb-simple: SQL query contains no statement")+ , ("SELECT 1; SELECT 2", "duckdb-simple: prepared SQL contains more than one statement")+ ]+ $ \(sql, message) -> do+ outcome <- try (openStatement conn (Query sql))+ case outcome of+ Left err -> do+ sqlErrorMessage err @?= message+ sqlErrorQuery err @?= Just (Query sql)+ Right stmt -> closeStatement stmt >> assertFailure ("expected an error for " <> show sql)+ , testCase "named and indexed statement access raises the errors of the library" $+ withConnection ":memory:" $ \conn -> withStatement conn "SELECT $value AS v" $ \stmt -> do+ nulName <- try (bindNamed stmt ["value\0x" := (1 :: Int64)])+ case nulName of+ Left err -> do+ sqlErrorMessage err @?= "duckdb-simple: parameter name contains NUL"+ sqlErrorQuery err @?= Just "SELECT $value AS v"+ Right () -> assertFailure "expected the NUL error"+ tooMany <- try (bindNamed stmt ["value" := (1 :: Int64), "other" := (2 :: Int64)])+ case tooMany of+ Left err -> formatErrorMessage err @?= "duckdb-simple: SQL query contains 1 named parameter(s), but 2 argument(s) were supplied"+ Right () -> assertFailure "expected the count error"+ negativeIndex <- try (columnName stmt (-1))+ case negativeIndex of+ Left err -> sqlErrorMessage err @?= "duckdb-simple: column index -1 out of bounds"+ Right name -> assertFailure ("expected the index error, got " <> show name)+ -- The statement still works after the errors.+ bindNamed stmt ["value" := (7 :: Int64)]+ nextRow stmt >>= (@?= Just (Only (7 :: Int64)))+ , testCase "connection options with NUL raise an error before the connection opens" $+ withEnvironment $ \environment -> withInstance environment [] $ \instance_ -> do+ attachDatabase instance_ ":memory:" "shared" []+ setDefaultDatabase instance_ "shared"+ outcome <- try (connectWithOptions instance_ defaultConnectionOptions{connectionConfig = [("threads\0x", "1")]})+ case outcome of+ Left err -> sqlErrorMessage err @?= "duckdb-simple: configuration contains NUL"+ Right conn -> close conn >> assertFailure "expected the NUL error"+ , testCase "statements without a query result give no row through each read function" $ withConnection ":memory:" $ \conn -> do+ let prepend rows (Only (value :: Int64)) = pure (value : rows)+ (query_ conn "CREATE TABLE no_rows_a (x BIGINT)" :: IO [Only Int64]) >>= (@?= [])+ fold_ conn "CREATE TABLE no_rows_b (x BIGINT)" [] prepend >>= (@?= [])+ withStatement conn "CREATE TABLE no_rows_c (x BIGINT)" nextRow >>= (@?= (Nothing :: Maybe (Only Int64)))+ Streaming.fold_ conn "CREATE TABLE no_rows_d (x BIGINT)" [] prepend >>= (@?= [])+ withStatement conn "CREATE TABLE no_rows_e (x BIGINT)" Streaming.nextRow >>= (@?= (Nothing :: Maybe (Only Int64)))+ -- DuckDB ran each statement one time.+ query_ conn "SELECT table_name FROM duckdb_tables() WHERE table_name LIKE 'no_rows_%' ORDER BY table_name"+ >>= (@?= map Only ["no_rows_a", "no_rows_b", "no_rows_c", "no_rows_d", "no_rows_e" :: Text])+ , testCase "a rejected runtime option raises the error of DuckDB and leaves the environment closable" $ do+ environment <- createEnvironment+ outcome <- try (createInstance environment [("logging_level", "no_such_level")])+ case outcome of+ Left err ->+ assertBool+ (Text.unpack (sqlErrorMessage err))+ ("unrecognized value \"no_such_level\" for enum \"LogLevel\"" `Text.isInfixOf` sqlErrorMessage err)+ Right instance_ -> closeInstance instance_ >> assertFailure "expected the error of DuckDB for logging_level"+ -- No instance remains, so DuckDB destroys the environment.+ closeEnvironment environment+ , testCase "a description of a missing table raises the catalog error of DuckDB" $ withConnection ":memory:" $ \conn -> do+ outcome <- try (Catalog.describeTable conn "no_such_table")+ case outcome of+ Left err -> assertBool (Text.unpack (sqlErrorMessage err)) ("Table with name no_such_table does not exist" `Text.isInfixOf` sqlErrorMessage err)+ Right description -> assertFailure ("expected the catalog error, got " <> show (Catalog.tableDescriptionName description))+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "an attach in a missing directory raises the IO error of DuckDB" $+ withEnvironment $ \environment -> withInstance environment [] $ \instance_ -> do+ outcome <- try (attachDatabase instance_ "/nonexistent-duckdb-simple/attached.db" "missing" [])+ case outcome of+ Left err ->+ assertBool+ (Text.unpack (sqlErrorMessage err))+ ("Cannot open file \"/nonexistent-duckdb-simple/attached.db\": No such file or directory" `Text.isInfixOf` sqlErrorMessage err)+ Right () -> assertFailure "expected the attach error"+ -- The instance attaches another database after the failure.+ attachDatabase instance_ ":memory:" "present" []+ setDefaultDatabase instance_ "present"+ withConnectionOn instance_ $ \conn -> query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ ]+ where+ collect acc row = pure (acc <> [row])++-- | Join all creation workers and release every acquired child after the action.+withConcurrentChildren :: Int -> IO a -> (a -> IO ()) -> ([a] -> IO ()) -> IO ()+withConcurrentChildren count acquire release action =+ bracket createChildren (mapM_ release) action+ where+ createChildren = do+ start <- newEmptyMVar+ completed <- replicateM count newEmptyMVar+ forM_ completed $ \result ->+ void $ forkFinally (readMVar start >> acquire) (putMVar result)+ putMVar start ()+ outcomes <- mapM takeMVar completed+ case partitionEithers outcomes of+ ([], children) -> pure children+ (failure : _, children) -> mapM_ release children >> throwIO failure++-- | Require an exception without constraining its concrete type.+assertThrows :: IO a -> Assertion+assertThrows action = do+ result <- try (void action)+ case result of+ Left (_ :: SomeException) -> pure ()+ Right () -> assertFailure "expected an exception"++-- | A callback of one kind and phase that raises an exception while DuckDB runs+-- a statement over more than one chunk, and that statement.+data FailingCallback = FailingCallback+ { failingKind :: String+ , failingLabel :: String+ , failingThreads :: Int+ , failingSetup :: FailingContext -> IO ()+ , failingStatement :: FailingContext -> IO ()+ , failingError :: Maybe (ErrorType, Text.Text)+ -- ^ The type and a text of the error of the statement, or 'Nothing' for a+ -- statement that does not fail, because DuckDB ignores the error of a log+ -- sink. The callback of such a row must count its calls in+ -- 'contextCalls'.+ }++-- | The connections and the directory of one row of 'failingCallbacks'.+data FailingContext = FailingContext+ { contextConnection :: Connection+ , contextOther :: Connection+ , contextInstance :: Instance+ , contextDirectory :: FilePath+ , contextCalls :: IORef Int+ }++-- | Run a statement on the connection of the row.+runIn :: FailingContext -> Query -> IO ()+runIn context sql = void (execute_ (contextConnection context) sql)++-- | Raise the exception of a row.+failWith :: String -> IO a+failWith text = throwIO (userError text)++-- | A sum whose step, combine, or result raises an exception.+failingSum :: (Int64 -> Bool) -> Bool -> Bool -> String -> Aggregate Int64 (Only Int64) Int64+failingSum failStep failCombine failResult text =+ Aggregate+ { aggregateInitial = 0+ , aggregateStep = \total (Only value) -> if failStep value then failWith text else pure (total + value)+ , aggregateCombine = \a b -> if failCombine then failWith text else pure (a + b)+ , aggregateResult = \total -> if failResult then failWith text else pure total+ }++-- | A table function of the pairs (n, f) for n from 0 to the count, whose column+-- f raises an exception at one row.+failingPairs :: Connection -> Text.Text -> Int64 -> String -> IO ()+failingPairs conn name row text =+ createTableFunction conn name ["n", "f"] \(Only count) ->+ pure [(n, if n == row then error text else n) | n <- [0 .. count - 1 :: Int64]]++-- | Each callback kind and phase in which a function of the user can fail. The+-- statements run over 5000 rows: two full chunks of 2048 rows come before row+-- 3000. Each kind runs one time more at 4 threads.+failingCallbacks :: [FailingCallback]+failingCallbacks = rows <> [row{failingThreads = 4} | row <- firstOfEachKind rows, failingThreads row == 1]+ where+ firstOfEachKind = foldr (\row kept -> row : filter ((/= failingKind row) . failingKind) kept) []+ insertRows sql context = runIn context ("INSERT INTO failing_rows " <> sql)+ countCall context = atomicModifyIORef' (contextCalls context) \count -> (count + 1, ())+ rows =+ [ FailingCallback+ "scalar"+ "scalar function with an IO result"+ 1+ ( \context ->+ createFunction (contextConnection context) "failing_io" \(x :: Int64) ->+ (when (x == 3000) (failWith "failing scalar io") >> pure x :: IO Int64)+ )+ (insertRows "SELECT failing_io(i) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing scalar io"))+ , FailingCallback+ "scalar"+ "scalar function with a pure result"+ 1+ (\context -> createFunction (contextConnection context) "failing_pure" \(x :: Int64) -> if x == 3000 then error "failing scalar pure" else x)+ (insertRows "SELECT failing_pure(i) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing scalar pure"))+ , FailingCallback+ "stateful scalar"+ "stateful scalar function call"+ 1+ ( \context ->+ createFunctionWithState (contextConnection context) "failing_call" (pure ()) \() (x :: Int64) ->+ (when (x == 3000) (failWith "failing stateful call") >> pure x :: IO Int64)+ )+ (insertRows "SELECT failing_call(i) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing stateful call"))+ , FailingCallback+ "stateful scalar"+ "stateful scalar function init"+ 1+ (\context -> createFunctionWithState (contextConnection context) "failing_init" (failWith "failing stateful init") \() (x :: Int64) -> x)+ (insertRows "SELECT failing_init(i) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing stateful init"))+ , FailingCallback+ "aggregate"+ "aggregate update with DISTINCT"+ 1+ (\context -> createAggregate (contextConnection context) "failing_distinct" (failingSum (== 3000) False False "failing distinct update"))+ (insertRows "SELECT failing_distinct(DISTINCT i) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing distinct update"))+ , FailingCallback+ "aggregate"+ "aggregate update with FILTER"+ 1+ (\context -> createAggregate (contextConnection context) "failing_filter" (failingSum (== 3000) False False "failing filter update"))+ (insertRows "SELECT failing_filter(i) FILTER (WHERE i >= 0) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing filter update"))+ , FailingCallback+ "aggregate"+ "aggregate update as a WINDOW aggregate"+ 1+ (\context -> createAggregate (contextConnection context) "failing_window" (failingSum (== 3000) False False "failing window update"))+ (insertRows "SELECT failing_window(i) OVER (ORDER BY i ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing window update"))+ , FailingCallback+ "aggregate"+ "aggregate combine of a plain aggregate"+ 1+ (\context -> createAggregate (contextConnection context) "failing_combine" (failingSum (const False) True False "failing plain combine"))+ (insertRows "SELECT failing_combine(i) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing plain combine"))+ , -- A grouped aggregate combines only the partial states of several+ -- threads, and DuckDB scans one row group of 122880 rows with one+ -- thread. Thus this row needs two row groups and 4 threads.+ FailingCallback+ "aggregate"+ "aggregate combine of a grouped aggregate"+ 4+ ( \context -> do+ runIn context "CREATE TABLE failing_groups AS SELECT i FROM range(130000) t(i)"+ createAggregate (contextConnection context) "failing_grouped" (failingSum (const False) True False "failing grouped combine")+ )+ (insertRows "SELECT failing_grouped(i) FROM failing_groups GROUP BY i % 1000")+ (Just (ErrorInputInvalid, "failing grouped combine"))+ , FailingCallback+ "aggregate"+ "aggregate finalize as a WINDOW aggregate"+ 1+ (\context -> createAggregate (contextConnection context) "failing_result" (failingSum (const False) False True "failing window finalize"))+ (insertRows "SELECT failing_result(i) OVER (ORDER BY i ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) FROM range(5000) t(i)")+ (Just (ErrorInputInvalid, "failing window finalize"))+ , FailingCallback+ "table function"+ "table function scan after rows, failing column only"+ 1+ (\context -> failingPairs (contextConnection context) "failing_later" 3000 "failing scan at row 3000")+ (insertRows "SELECT f FROM failing_later(5000)")+ (Just (ErrorInputInvalid, "failing scan at row 3000"))+ , FailingCallback+ "table function"+ "table function scan after rows, all columns"+ 1+ (\context -> failingPairs (contextConnection context) "failing_later" 3000 "failing scan at row 3000")+ (\context -> runIn context "INSERT INTO failing_pairs SELECT * FROM failing_later(5000)")+ (Just (ErrorInputInvalid, "failing scan at row 3000"))+ , FailingCallback+ "table function"+ "table function scan at row 0, failing column only"+ 1+ (\context -> failingPairs (contextConnection context) "failing_first" 0 "failing scan at row 0")+ (insertRows "SELECT f FROM failing_first(5000)")+ (Just (ErrorInputInvalid, "failing scan at row 0"))+ , FailingCallback+ "table function"+ "table function scan at row 0, all columns"+ 1+ (\context -> failingPairs (contextConnection context) "failing_first" 0 "failing scan at row 0")+ (\context -> runIn context "INSERT INTO failing_pairs SELECT * FROM failing_first(5000)")+ (Just (ErrorInputInvalid, "failing scan at row 0"))+ , FailingCallback+ "table function"+ "table function init"+ 1+ ( \context ->+ createTableFunction (contextConnection context) "failing_start" ["v"] \(Only (_ :: Int64)) ->+ failWith "failing table init" :: IO [Only Int64]+ )+ (insertRows "SELECT v FROM failing_start(5000)")+ (Just (ErrorInputInvalid, "failing table init"))+ , -- The bind callback runs no function of the user. A NULL for a+ -- parameter that cannot hold NULL fails there.+ FailingCallback+ "table function"+ "table function bind"+ 1+ (\context -> createTableFunction (contextConnection context) "failing_bind" ["v"] \(Only (count :: Int64)) -> pure [Only count])+ (insertRows "SELECT v FROM failing_bind(NULL::BIGINT)")+ (Just (ErrorInputInvalid, "failing_bind: argument arg0 is NULL, but Int64 cannot hold NULL"))+ , FailingCallback+ "table function"+ "table function init with a failing query on another connection"+ 1+ ( \context ->+ createTableFunction (contextConnection context) "failing_inner" ["v"] \(Only (_ :: Int64)) ->+ query_ (contextOther context) "SELECT 'abc'::INTEGER" :: IO [Only Int64]+ )+ (insertRows "SELECT v FROM failing_inner(5000)")+ (Just (ErrorTypeConversion, "Conversion Error: Could not convert string 'abc' to INT32"))+ , FailingCallback+ "pragma"+ "pragma in the call form"+ 1+ ( \context ->+ registerPragmaFunction+ (contextConnection context)+ "failing_pragma"+ [parameter "amount" (TypeExpression "BIGINT")]+ (ExecPragma \(Only (_ :: Int64)) -> failWith "failing pragma call")+ )+ (\context -> runIn context "PRAGMA failing_pragma(1)")+ (Just (ErrorInputInvalid, "failing pragma call"))+ , FailingCallback+ "custom option"+ "custom option update"+ 1+ ( \context ->+ Config.registerCustomOption+ (contextConnection context)+ (Config.customOption "failing_option" (TypeExpression "BIGINT")){Config.customOptionUpdate = Just \_ -> failWith "failing option update"}+ )+ (\context -> runIn context "SET failing_option = 1")+ (Just (ErrorInputInvalid, "failing option update"))+ , FailingCallback+ "replacement scan"+ "replacement scan on a connection"+ 1+ ( \context -> registerReplacementScan (contextConnection context) \parts ->+ if parts == ["failing_missing"] then failWith "failing replacement scan" else pure Nothing+ )+ (insertRows "SELECT * FROM failing_missing")+ (Just (ErrorInputInvalid, "failing replacement scan"))+ , FailingCallback+ "replacement scan"+ "replacement scan on an instance"+ 1+ ( \context -> registerInstanceReplacementScan (contextInstance context) \parts ->+ if parts == ["failing_missing"] then failWith "failing instance replacement scan" else pure (Nothing :: Maybe Replacement)+ )+ (insertRows "SELECT * FROM failing_missing")+ (Just (ErrorInputInvalid, "failing instance replacement scan"))+ , FailingCallback+ "COPY TO"+ "COPY TO sink after rows"+ 1+ ( \context -> do+ written <- newIORef (0 :: Int)+ Copy.registerCopyToFunction+ (contextConnection context)+ "failing_copy"+ (\_ -> pure ())+ (\info -> writeFile (Copy.copyInitFilePath info) "" >> pure (Copy.copyInitFilePath info))+ ( \info chunk -> do+ appendFile (Copy.copySinkGlobalState info) (concatMap (const "row\n") chunk)+ total <- atomicModifyIORef' written \count -> (count + length chunk, count + length chunk)+ when (total > 3000) (failWith "failing COPY TO sink")+ )+ (\_ -> pure ())+ )+ ( \context ->+ void+ ( execute+ (contextConnection context)+ "COPY (SELECT i FROM range(5000) t(i)) TO ? (FORMAT failing_copy)"+ (Only (Text.pack (contextDirectory context </> "out.txt")))+ )+ )+ (Just (ErrorInputInvalid, "failing COPY TO sink"))+ , FailingCallback+ "COPY FROM"+ "COPY FROM after rows"+ 1+ ( \context -> do+ writeFile (contextDirectory context </> "in.txt") ""+ Copy.registerCopyFromFunction (contextConnection context) "failing_from" \_ ->+ pure [Only (if i == 3000 then error "failing COPY FROM row" else i) | i <- [0 .. 4999 :: Int64]]+ )+ ( \context ->+ void+ ( execute+ (contextConnection context)+ "COPY failing_rows FROM ? (FORMAT failing_from)"+ (Only (Text.pack (contextDirectory context </> "in.txt")))+ )+ )+ (Just (ErrorInputInvalid, "failing COPY FROM row"))+ , FailingCallback+ "multi-file function"+ "multi-file function at its second file"+ 1+ ( \context -> do+ forM_ [1 :: Int .. 3] \file -> writeFile (contextDirectory context </> ("part" <> show file <> ".txt")) ""+ createTableFunction (contextConnection context) "failing_single" ["v"] \(Only (path :: Text.Text)) -> do+ when ("part2.txt" `Text.isSuffixOf` path) (failWith "failing second file")+ pure [Only i | i <- [0 .. 2999 :: Int64]]+ Copy.registerMultiFileFunction (contextConnection context) "failing_files" "failing_single" Copy.defaultMultiFileOptions+ )+ ( \context ->+ void+ ( execute+ (contextConnection context)+ "INSERT INTO failing_rows SELECT v FROM failing_files(?)"+ (Only (Text.pack (contextDirectory context </> "part*.txt")))+ )+ )+ (Just (ErrorInputInvalid, "failing second file"))+ , FailingCallback+ "log sink"+ "log sink write"+ 1+ ( \context -> Logging.registerLogSink (contextConnection context) "failing_log" (\_ -> countCall context >> failWith "failing log write") (pure ())+ )+ ( \context -> runIn context "CALL enable_logging(); SET logging_storage = 'failing_log'; SELECT write_log('failing'); SET logging_storage = 'memory'"+ )+ Nothing+ , FailingCallback+ "log sink"+ "log sink flush"+ 1+ ( \context -> Logging.registerLogSink (contextConnection context) "failing_log" (\_ -> pure ()) (countCall context >> failWith "failing log flush")+ )+ ( \context -> runIn context "CALL enable_logging(); SET logging_storage = 'failing_log'; SELECT write_log('failing'); SET logging_storage = 'memory'"+ )+ Nothing+ ]++-- | Run one row of 'failingCallbacks' in its own instance and directory, inside+-- an explicit transaction that has written a marker row. Require the error type+-- and the text of the callback, then COMMIT, then a usable connection and a+-- usable new connection of the instance, no row of the statement, no marker+-- row, and no new file. A log sink row requires no error and a call of the+-- sink, and keeps the marker. A new connection runs a query before the COMMIT+-- too. The same connection runs its query only after the COMMIT, because+-- DuckDB refuses each other statement in a transaction after a failed+-- statement.+runFailingCallback :: FailingCallback -> Assertion+runFailingCallback row = bracket createRowDirectory removePathForcibly \directory ->+ withEnvironment \environment -> withInstance environment [("threads", Text.pack (show (failingThreads row)))] \instance' ->+ withConnection ":memory:" \other -> do+ attachDatabase instance' ":memory:" "failing" []+ setDefaultDatabase instance' "failing"+ withConnectionOn instance' \conn -> do+ calls <- newIORef 0+ let context = FailingContext conn other instance' directory calls+ label = failingLabel row <> " at " <> show (failingThreads row) <> " threads"+ mapM_+ (void . execute_ conn)+ ["CREATE TABLE failing_rows (v BIGINT)", "CREATE TABLE failing_pairs (n BIGINT, f BIGINT)", "CREATE TABLE failing_marker (m INTEGER)"]+ failingSetup row context+ before <- listDirectory directory+ void (execute_ conn "BEGIN")+ void (execute_ conn "INSERT INTO failing_marker VALUES (1)")+ outcome <- try (failingStatement row context)+ case (failingError row, outcome) of+ (Just (errorType, text), Left err) -> do+ assertEqual label (Just errorType) (sqlErrorType err)+ assertBool (label <> ": " <> show (sqlErrorMessage err)) (text `Text.isInfixOf` sqlErrorMessage err)+ (Just _, Right ()) -> assertFailure (label <> ": the statement did not fail")+ (Nothing, Left err) -> assertFailure (label <> ": " <> show err)+ (Nothing, Right ()) -> readIORef calls >>= assertBool (label <> ": the callback ran") . (> 0)+ withConnectionOn instance' assertReusable+ void (execute_ conn "COMMIT")+ assertReusable conn+ withConnectionOn instance' assertReusable+ let marker = maybe 1 (const 0) (failingError row) :: Int64+ query_ conn "SELECT (SELECT count(*) FROM failing_rows), (SELECT count(*) FROM failing_pairs), (SELECT count(*) FROM failing_marker)"+ >>= assertEqual (label <> ": rows after COMMIT") [(0 :: Int64, 0 :: Int64, marker)]+ listDirectory directory >>= assertEqual (label <> ": files after the statement") before+ where+ createRowDirectory = do+ root <- getTemporaryDirectory+ (path, handle) <- openBinaryTempFile root "duckdb-failing-callback"+ hClose handle+ removeFile path+ createDirectory path+ pure path
test/ExtensionRegressionTests.hs view
@@ -2,303 +2,1018 @@ {-# LANGUAGE FlexibleInstances #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-}---- | Regression tests for extension callbacks and native handle ownership.-module ExtensionRegressionTests (main, extensionRegressionTests) where--import Control.Exception (AsyncException (ThreadKilled), ErrorCall, Exception (..), SomeException, bracket, throwIO, try)-import Control.Monad (forM_, replicateM_, void)-import qualified Data.ByteString as BS-import Data.IORef (modifyIORef', newIORef, readIORef)-import Data.Int (Int64)-import Data.Proxy (Proxy (..))-import qualified Data.Text as Text-import Data.Word (Word64)-import Database.DuckDB.FFI-import Database.DuckDB.Simple-import qualified Database.DuckDB.Simple.Catalog as Catalog-import qualified Database.DuckDB.Simple.Config as Config-import qualified Database.DuckDB.Simple.Copy as Copy-import qualified Database.DuckDB.Simple.FileSystem as FileSystem-import qualified Database.DuckDB.Simple.Logging as Logging-import System.Directory (removeFile)-import System.IO (hClose, openBinaryTempFile)-import Test.Tasty (TestTree, defaultMain, testGroup)-import Test.Tasty.HUnit (Assertion, assertBool, assertFailure, testCase, (@?=))---- | Run the extension regression tests as a standalone executable.-main :: IO ()-main = defaultMain extensionRegressionTests---- | Check callback failures, value conversion, and resource cleanup.-extensionRegressionTests :: TestTree-extensionRegressionTests =- testGroup- "extension regressions"- [ testCase "scalar closed connection cleanup" do- conn <- open ":memory:"- close conn- replicateM_ 32 $ assertSqlError $ createFunction conn "closed_scalar" (1 :: Int64)- , testCase "stateful closed connection cleanup" do- conn <- open ":memory:"- close conn- replicateM_ 32 $ assertSqlError $ createFunctionWithState conn "closed_state" (pure ()) (\() -> (1 :: Int64))- , testCase "COPY closed connection cleanup" do- conn <- open ":memory:"- close conn- replicateM_ 32 $ assertSqlError $ registerCopy conn "closed_copy" Nothing- , testCase "logging closed connection cleanup" do- conn <- open ":memory:"- close conn- replicateM_ 32 $ assertSqlError $ Logging.registerLogStorage conn "closed_log" (\_ -> pure ())- , testCase "scalar replacement registration cleanup" $- withConnection ":memory:" \conn -> do- createFunction conn "duplicate_scalar" (42 :: Int64)- replicateM_ 32 $ createFunction conn "duplicate_scalar" (7 :: Int64)- query_ conn "SELECT duplicate_scalar()" >>= (@?= [Only (7 :: Int64)])- , testCase "stateful replacement registration cleanup" $- withConnection ":memory:" \conn -> do- createFunctionWithState conn "duplicate_state" (pure ()) (\() -> (42 :: Int64))- replicateM_ 32 $ createFunctionWithState conn "duplicate_state" (pure ()) (\() -> (7 :: Int64))- query_ conn "SELECT duplicate_state()" >>= (@?= [Only (7 :: Int64)])- , testCase "scalar invalid signature registration cleanup" $- withConnection ":memory:" \conn ->- replicateM_ 32 $ assertSqlError $ createFunction conn "invalid_scalar" InvalidSignature- , testCase "stateful invalid signature registration cleanup" $- withConnection ":memory:" \conn ->- replicateM_ 32 $ assertSqlError $ createFunctionWithState conn "invalid_state" (pure ()) (\() -> InvalidSignature)- , testCase "logging duplicate registration cleanup" $- withConnection ":memory:" \conn -> do- Logging.registerLogStorage conn "duplicate_log" (\_ -> pure ())- replicateM_ 32 $ assertSqlError $ Logging.registerLogStorage conn "duplicate_log" (\_ -> pure ())- query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])- , testCase "empty registration names fail safely" $- withConnection ":memory:" \conn -> replicateM_ 32 do- assertSqlError $ createFunction conn "" (1 :: Int64)- assertSqlError $ createFunctionWithState conn "" (pure ()) (\() -> (1 :: Int64))- assertSqlError $ registerCopy conn "" Nothing- assertSqlError $ Logging.registerLogStorage conn "" (\_ -> pure ())- , testCase "NUL registration names fail safely" $- withConnection ":memory:" \conn -> do- assertSqlError $ createFunction conn "nul\0scalar" (1 :: Int64)- assertSqlError $ createFunctionWithState conn "nul\0state" (pure ()) (\() -> (1 :: Int64))- assertSqlError $ registerCopy conn "nul\0copy" Nothing- assertSqlError $ Logging.registerLogStorage conn "nul\0log" (\_ -> pure ())- , testCase "partial scalar setup cleanup" $- withConnection ":memory:" \conn -> replicateM_ 32 do- outcome <- try (createFunction conn "broken_signature" BrokenSignature) :: IO (Either ErrorCall ())- case outcome of- Left _ -> pure ()- Right () -> assertFailure "expected signature exception"- , testCase "scalar exceptions become SQL errors" $- withConnection ":memory:" \conn -> do- createFunction conn "throw_scalar" (throwIO (userError "scalar callback failure") :: IO Int64)- assertSqlErrorContaining "scalar callback failure" (query_ conn "SELECT throw_scalar()" :: IO [Only Int64])- query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])- , testCase "exception rendering cannot escape scalar callback" $- withConnection ":memory:" \conn -> do- createFunction conn "throw_bad_display" (throwIO BrokenDisplay :: IO Int64)- assertSqlErrorContaining "Haskell callback failed" (query_ conn "SELECT throw_bad_display()" :: IO [Only Int64])- , testCase "scalar async exceptions become SQL errors" $- withConnection ":memory:" \conn -> do- createFunction conn "throw_async_scalar" (throwIO ThreadKilled :: IO Int64)- assertSqlErrorContaining "thread killed" (query_ conn "SELECT throw_async_scalar()" :: IO [Only Int64])- , testCase "scalar initializer exceptions become SQL errors" $- withConnection ":memory:" \conn -> do- createFunctionWithState conn "throw_init" (throwIO (userError "state init failure") :: IO ()) (\() -> (1 :: Int64))- assertSqlErrorContaining "state init failure" (query_ conn "SELECT throw_init()" :: IO [Only Int64])- , testCase "scalar initializer async exceptions become SQL errors" $- withConnection ":memory:" \conn -> do- createFunctionWithState conn "throw_async_init" (throwIO ThreadKilled :: IO ()) (\() -> (1 :: Int64))- assertSqlErrorContaining "thread killed" (query_ conn "SELECT throw_async_init()" :: IO [Only Int64])- , testCase "stateful scalar exceptions become SQL errors" $- withConnection ":memory:" \conn -> do- createFunctionWithState conn "throw_state" (pure ()) (\() -> throwIO (userError "state execution failure") :: IO Int64)- assertSqlErrorContaining "state execution failure" (query_ conn "SELECT throw_state()" :: IO [Only Int64])- , testCase "Float scalar results preserve NaN infinity and negative zero" $- withConnection ":memory:" \conn ->- forM_- [ (0 / 0, isNaN)- , (1 / 0, \value -> isInfinite value && value > 0)- , (-1 / 0, \value -> isInfinite value && value < 0)- , (negate 0, isNegativeZero)- ]- \(value, matches) -> do- createFunction conn "float_value" (value :: Float)- createFunctionWithState conn "float_state" (pure ()) (\() -> value)- [(plain, stateful)] <- query_ conn "SELECT float_value(), float_state()" :: IO [(Double, Double)]- assertBool "plain callback preserves the floating-point value" (matches plain)- assertBool "stateful callback preserves the floating-point value" (matches stateful)- , testCase "Word64 scalar preserves unsigned maximum" $- withConnection ":memory:" \conn -> do- createFunction conn "word64_identity" (id :: Word64 -> Word64)- query_ conn "SELECT word64_identity(18446744073709551615::UBIGINT)" >>= (@?= [Only (maxBound :: Word64)])- query_ conn "SELECT typeof(word64_identity(0::UBIGINT))" >>= (@?= [Only ("UBIGINT" :: Text.Text)])- , testCase "Word scalar preserves unsigned maximum" $- withConnection ":memory:" \conn -> do- createFunction conn "word_maximum" (maxBound :: Word)- query_ conn "SELECT word_maximum()" >>= (@?= [Only (maxBound :: Word)])- , testCase "signed scalar preserves limits" $- withConnection ":memory:" \conn -> do- createFunction conn "int64_identity" (id :: Int64 -> Int64)- query_ conn "SELECT int64_identity(x) FROM (VALUES ((-9223372036854775808)::BIGINT), (9223372036854775807::BIGINT)) t(x)"- >>= (@?= [Only (minBound :: Int64), Only maxBound])- , testCase "scalar strings preserve UTF8 and embedded NUL" $- withConnection ":memory:" \conn -> do- let payload = "Matti \955\0 suffix" :: Text.Text- createFunction conn "text_payload" payload- createFunction conn "text_identity" (id :: Text.Text -> Text.Text)- query_ conn "SELECT text_identity(text_payload())" >>= (@?= [Only payload])- , testCase "scalar NULL arguments reach Maybe functions" $- withConnection ":memory:" \conn -> do- createFunction conn "null_default" (\(x :: Maybe Int64) -> maybe 42 id x)- query_ conn "SELECT null_default(NULL::BIGINT)" >>= (@?= [Only (42 :: Int64)])- , testCase "scalar nullable strings across chunks" $- withConnection ":memory:" \conn -> do- createFunction conn "optional_text" (\(x :: Int64) -> if even x then Just ("\955\0" <> Text.replicate 32 "x") else Nothing)- query_ conn "SELECT count(*), count(optional_text(i)), min(length(optional_text(i))) FROM range(5000) t(i)"- >>= (@?= [(5000 :: Int64, 2500 :: Int64, Just (34 :: Int64))])- , testCase "scalar DECIMAL inputs use supported numeric casts" $- withConnection ":memory:" \conn -> do- createFunction conn "double_identity" (id :: Double -> Double)- query_ conn "SELECT double_identity(123.25::DECIMAL(10,2))" >>= (@?= [Only (123.25 :: Double)])- , testCase "COPY exceptions become SQL errors in every phase" $- withConnection ":memory:" \conn ->- mapM_- ( \phase -> do- let name = "copy_fail_" <> Text.pack (show phase)- registerCopy conn name (Just phase)- assertSqlErrorContaining (Text.pack ("copy phase " <> show phase)) $- execute_ conn (Query ("COPY (SELECT 1) TO '/tmp/duckdb-simple-regression-copy' (FORMAT " <> name <> ")"))- )- [0 .. 3]- , testCase "COPY state is valid across repeated queries" $- withConnection ":memory:" \conn -> do- registerCopy conn "copy_repeat" Nothing- replicateM_ 32 $ void $ execute_ conn "COPY (SELECT * FROM range(5000)) TO '/tmp/duckdb-simple-regression-copy' (FORMAT copy_repeat)"- , testCase "logging exceptions stay inside callback" $- withConnection ":memory:" \conn -> do- calls <- newIORef (0 :: Int)- Logging.registerLogStorage conn "throwing_log" \_ -> do- modifyIORef' calls (+ 1)- throwIO (userError "log callback failure")- void $ execute_ conn "SET logging_storage = 'throwing_log'"- void $ execute_ conn "SET enable_logging = true"- void $ execute_ conn "SELECT write_log('callback regression', level := 'INFO', scope := 'connection')"- readIORef calls >>= \n -> assertBool "log callback ran" (n > 0)- query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])- , testCase "logging messages preserve UTF8" $- withConnection ":memory:" \conn -> do- messages <- newIORef []- Logging.registerLogStorage conn "unicode_log" \entry ->- modifyIORef' messages (Logging.logEntryMessage entry :)- void $ execute_ conn "SET logging_storage = 'unicode_log'"- void $ execute_ conn "SET enable_logging = true"- void $ execute_ conn "SELECT write_log('\955', level := 'INFO', scope := 'connection')"- readIORef messages >>= \entries -> assertBool "Unicode log message" ("\955" `elem` entries)- , testCase "filesystem action exceptions release handles" $- withTemporaryFile \path ->- withConnection ":memory:" \conn -> do- outcome <- try $ FileSystem.withFileHandle conn path [DuckDBFileFlagWrite] \_ -> throwIO (userError "file action failure")- case outcome :: Either SomeException () of- Left _ -> pure ()- Right () -> assertFailure "expected file action exception"- FileSystem.withFileHandle conn path [DuckDBFileFlagWrite] \handle -> do- FileSystem.writeFileHandleBytes handle (BS.pack [0, 255]) >>= (@?= 2)- FileSystem.fileHandleSync handle- FileSystem.withFileHandle conn path [DuckDBFileFlagRead] \handle -> do- FileSystem.readFileHandleChunk handle 0 >>= (@?= BS.empty)- FileSystem.readFileHandleChunk handle 8 >>= (@?= BS.pack [0, 255])- , testCase "filesystem errors remain controlled" $- withConnection ":memory:" \conn -> do- assertSqlError $ FileSystem.withFileHandle conn "/tmp/duckdb-review-missing-dir/no-file" [DuckDBFileFlagRead] (\_ -> pure ())- withTemporaryFile \path ->- assertSqlError $ FileSystem.withFileHandle conn (path <> "\0suffix") [DuckDBFileFlagRead] (\_ -> pure ())- , testCase "catalog unsupported kinds fail safely" $- withConnection ":memory:" \conn ->- withTransaction conn $- mapM_- (\kind -> assertSqlError $ Catalog.lookupCatalogEntry conn "memory" "main" "probe" kind)- [DuckDBCatalogEntryTypeInvalid, DuckDBCatalogEntryTypeSchema, DuckDBCatalogEntryTypePreparedStatement, DuckDBCatalogEntryTypeDatabase, DuckDBCatalogEntryType 99]- , testCase "catalog and config NUL names fail safely" $- withConnection ":memory:" \conn -> do- assertSqlError $ Config.getConfigOption conn "threads\0suffix"- withTransaction conn do- assertSqlError $ Catalog.catalogTypeName conn "memory\0suffix"- assertSqlError $ Catalog.lookupCatalogEntry conn "memory" "main\0suffix" "probe" DuckDBCatalogEntryTypeTable- , testCase "catalog names preserve UTF8" $- withConnection ":memory:" \conn -> do- void $ execute_ conn "CREATE TABLE \"\955\" (i INT)"- withTransaction conn do- entry <- Catalog.lookupCatalogEntry conn "memory" "main" "\955" DuckDBCatalogEntryTypeTable- fmap Catalog.catalogEntryName entry @?= Just "\955"- , testCase "catalog and config missing values remain safe" $- withConnection ":memory:" \conn -> do- Config.getConfigOption conn "no_such_config_option" >>= (@?= Nothing)- withTransaction conn do- Catalog.catalogTypeName conn "no_such_catalog" >>= (@?= Nothing)- Catalog.lookupCatalogEntry conn "memory" "main" "no_such_table" DuckDBCatalogEntryTypeTable >>= (@?= Nothing)- ]---- | Raise a second exception when the callback error is rendered.-data BrokenDisplay = BrokenDisplay- deriving (Show)--instance Exception BrokenDisplay where- displayException _ = error "broken error rendering"---- | Force a setup exception after scalar callbacks have been acquired.-data BrokenSignature = BrokenSignature--instance Function BrokenSignature where- argumentTypes _ = error "broken scalar signature"- returnType _ = error "unused return type"- isVolatile _ = False- applyFunction _ _ = error "unused function"---- | Supply an argument type which DuckDB rejects at registration.-data InvalidSignature = InvalidSignature--instance Function InvalidSignature where- argumentTypes _ = [DuckDBTypeInvalid]- returnType _ = returnType (Proxy :: Proxy Int64)- isVolatile _ = False- applyFunction _ _ = error "unregistered function"---- | Register a COPY callback and optionally fail one phase.-registerCopy :: Connection -> Text.Text -> Maybe Int -> IO ()-registerCopy conn name failingPhase =- Copy.registerCopyToFunction- conn- name- (\_ -> failPhase 0)- (\info -> (Copy.copyInitBindState info @?= ()) >> failPhase 1)- (\info _ -> (Copy.copySinkGlobalState info @?= ()) >> failPhase 2)- (\info -> (Copy.copyFinalizeGlobalState info @?= ()) >> failPhase 3)- where- failPhase phase =- if failingPhase == Just phase- then throwIO (userError ("copy phase " <> show phase))- else pure ()---- | Require a controlled SQL exception.-assertSqlError :: IO a -> Assertion-assertSqlError = assertSqlErrorContaining Text.empty---- | Require a SQL exception which includes the expected message.-assertSqlErrorContaining :: Text.Text -> IO a -> Assertion-assertSqlErrorContaining expected action = do- outcome <- try (void action)- case outcome of- Left (err :: SQLError) ->- assertBool ("expected SQL error containing " <> show expected <> ", got " <> show err) $- expected `Text.isInfixOf` sqlErrorMessage err- Right () -> assertFailure "expected SQL error"---- | Remove a temporary file after the action.-withTemporaryFile :: (FilePath -> IO a) -> IO a-withTemporaryFile action =- bracket- (do (path, handle) <- openBinaryTempFile "/tmp" "duckdb-extension-regression"; hClose handle; pure path)- removeFile- action+{-# LANGUAGE TypeApplications #-}+-- The tests call deleteFunction on purpose, so its warning does not apply.+{-# OPTIONS_GHC -Wno-deprecations #-}++-- |+-- Module : ExtensionRegressionTests+-- Description : Regression tests for extension callbacks and native handle ownership.+module ExtensionRegressionTests (main, extensionRegressionTests) where++import Control.Concurrent (forkIOWithUnmask, killThread, newEmptyMVar, putMVar, readMVar, takeMVar, threadDelay)+import Control.Exception (+ AsyncException (ThreadKilled),+ ErrorCall,+ Exception (..),+ IOException,+ SomeException,+ bracket,+ bracket_,+ finally,+ throwIO,+ try,+ )+import Control.Monad (forM, forM_, forever, replicateM_, void, when)+import qualified Data.ByteString as BS+import Data.IORef (IORef, atomicModifyIORef', mkWeakIORef, newIORef, readIORef, writeIORef)+import Data.Int (Int64)+import Data.List (find, isInfixOf, nub, sort)+import Data.Maybe (isJust)+import Data.Proxy (Proxy (..))+import qualified Data.Text as Text+import qualified Data.Text.Encoding as Text+import Data.Time.Clock (addUTCTime, getCurrentTime)+import Data.Word (Word64)+import Database.DuckDB.FFI.V2+import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Catalog as Catalog+import qualified Database.DuckDB.Simple.Config as Config+import qualified Database.DuckDB.Simple.Copy as Copy+import qualified Database.DuckDB.Simple.FileSystem as FileSystem+import Database.DuckDB.Simple.FromField (FieldValue (..))+import Database.DuckDB.Simple.Internal (+ MacroFunctionType,+ acquireItems,+ errorTypeCodes,+ errorTypeFromNative,+ logLevelFromNative,+ logLevelToNative,+ logScopeFromNative,+ macroFunctionTypeToNative,+ settingScopeFromNative,+ settingScopeToNative,+ throwInternalError,+ tokenTypeFromNative,+ )+import qualified Database.DuckDB.Simple.Logging as Logging+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..))+import qualified Database.DuckDB.Simple.Tokenizer as Tokenizer+import qualified Database.DuckDB.Simple.V2.Registration as Registration+import Numeric.Natural (Natural)+import System.Directory (doesFileExist, getHomeDirectory, getTemporaryDirectory, removeFile)+import System.FilePath (takeDirectory)+import System.IO (hClose, openBinaryTempFile)+import System.Mem (performMajorGC)+import System.Mem.Weak (deRefWeak)+import System.Timeout (timeout)+import Test.Tasty (TestTree, defaultMain, testGroup)+import Test.Tasty.HUnit (Assertion, assertBool, assertEqual, assertFailure, testCase, (@?=))+import TestUtils (assertInputErrorContaining, assertRangeError, assertReleasedAfter, assertSqlErrorContaining, assertSqlErrorType)++-- | Run the extension regression tests as a standalone executable.+main :: IO ()+main = defaultMain extensionRegressionTests++-- | Check callback failures, value conversion, and resource cleanup.+extensionRegressionTests :: TestTree+extensionRegressionTests =+ testGroup+ "extension regressions"+ [ -- A registration on a closed connection raises the error before it+ -- allocates a callback.+ testCase "scalar registration on a closed connection raises the closed error" do+ conn <- open ":memory:"+ close conn+ replicateM_ 32 $ assertSqlErrorContaining "connection is closed" $ createFunction conn "closed_scalar" (1 :: Int64)+ , testCase "stateful registration on a closed connection raises the closed error" do+ conn <- open ":memory:"+ close conn+ replicateM_ 32+ $ assertSqlErrorContaining "connection is closed"+ $ createFunctionWithState conn "closed_state" (pure ()) (\() -> (1 :: Int64))+ , testCase "COPY registration on a closed connection raises the closed error" do+ conn <- open ":memory:"+ close conn+ replicateM_ 32 $ assertSqlErrorContaining "connection is closed" $ registerCopy conn "closed_copy" Nothing+ , testCase "logging registration on a closed connection raises the closed error" do+ conn <- open ":memory:"+ close conn+ replicateM_ 32 $ assertSqlErrorContaining "connection is closed" $ Logging.registerLogStorage conn "closed_log" (\_ -> pure ())+ , testCase "scalar replacement releases each replaced callback before the connection closes" $+ withConnection ":memory:" \conn -> do+ createFunction conn "duplicate_scalar" (42 :: Int64)+ assertReleasedAfter "a replaced scalar function" \token -> do+ replicateM_ 32 $ createFunction conn "duplicate_scalar" (readIORef token)+ createFunction conn "duplicate_scalar" (7 :: Int64)+ query_ conn "SELECT duplicate_scalar()" >>= (@?= [Only (7 :: Int64)])+ , testCase "a scalar function of a rolled-back transaction is gone and releases its callback" $+ withConnection ":memory:" \conn -> do+ assertReleasedAfter "a rolled-back scalar function" \token -> do+ _ <- execute_ conn "BEGIN TRANSACTION"+ createFunction conn "rolled_scalar" (\(x :: Int64) -> (+ x) <$> readIORef token)+ query_ conn "SELECT rolled_scalar(1)" >>= (@?= [Only (2 :: Int64)])+ void (execute_ conn "ROLLBACK")+ assertSqlErrorType ErrorDatabaseCatalog (query_ conn "SELECT rolled_scalar(1)" :: IO [Only Int64])+ createFunction conn "rolled_scalar" (\(x :: Int64) -> x + 100)+ query_ conn "SELECT rolled_scalar(1)" >>= (@?= [Only (101 :: Int64)])+ -- A registration that stays keeps its callback.+ kept <- do+ token <- newIORef (1 :: Int64)+ weak <- mkWeakIORef token (pure ())+ createFunction conn "kept_scalar" (\(x :: Int64) -> (+ x) <$> readIORef token)+ pure weak+ performMajorGC+ performMajorGC+ deRefWeak kept >>= assertBool "a registration that stays keeps its callback" . isJust+ query_ conn "SELECT kept_scalar(1)" >>= (@?= [Only (2 :: Int64)])+ , testCase "scalar deletion refusal keeps function and connection usable" $+ withConnection ":memory:" \conn -> do+ createFunction conn "retained_scalar" (42 :: Int64)+ assertSqlErrorContaining "Cannot drop internal catalog entry" $ deleteFunction conn "retained_scalar"+ query_ conn "SELECT retained_scalar()" >>= (@?= [Only (42 :: Int64)])+ query_ conn "SELECT 7" >>= (@?= [Only (7 :: Int64)])+ , testCase "absent scalar deletion succeeds" $+ withConnection ":memory:" \conn -> do+ deleteFunction conn "absent_scalar"+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "clearing bindings permits statement reuse after cancellation" $+ withConnection ":memory:" \conn -> do+ let parameters = replicate 32 (toField (Text.replicate 4096 "x"))+ sql = Query ("SELECT length(concat(" <> Text.intercalate ", " (replicate 32 "?::VARCHAR") <> "))")+ withStatement conn sql \stmt -> replicateM_ 16 do+ started <- newEmptyMVar+ done <- newEmptyMVar+ -- Each worker alternates binding and clearing. The cancellation+ -- can occur in either operation. The last query checks reuse.+ -- Unmask the loop because bracket masks the creation of the worker.+ bracket+ ( forkIOWithUnmask \unmask ->+ try @SomeException (unmask (putMVar started () >> forever (bind stmt parameters >> clearStatementBindings stmt)))+ >>= putMVar done+ )+ ( \worker -> do+ stopped <- timeout (5 * 1000000) (killThread worker >> readMVar done)+ case stopped of+ Just _ -> pure ()+ Nothing -> assertFailure "the binding worker did not stop during cleanup"+ )+ \worker -> do+ outcome <- timeout (5 * 1000000) do+ takeMVar started+ threadDelay 1000+ killThread worker+ readMVar done+ case outcome of+ Just (Left failure) | fromException failure == Just ThreadKilled -> pure ()+ Just (Left failure) -> throwIO failure+ Just (Right ()) -> assertFailure "the binding worker returned before cancellation"+ Nothing -> assertFailure "the binding worker did not finish after cancellation"+ clearStatementBindings stmt+ bind stmt parameters+ nextRow stmt >>= (@?= Just (Only (131072 :: Int64)))+ , testCase "deleteFunction reads the name with the qualified-name rules of DuckDB" $+ withConnection ":memory:" \conn -> do+ let macroCount :: Text.Text -> Text.Text -> IO [Only Int64]+ macroCount schema name = query conn "SELECT count(*) FROM duckdb_functions() WHERE schema_name = ? AND function_name = ?" (schema, name)+ void $ execute_ conn "CREATE MACRO \"a.b\"(x) AS x + 1; CREATE SCHEMA s; CREATE MACRO s.f(x) AS x; CREATE MACRO s.g(x) AS x"+ mapM (uncurry macroCount) [("main", "a.b"), ("s", "f"), ("s", "g")] >>= (@?= replicate 3 [Only 1])+ deleteFunction conn "\"a.b\""+ macroCount "main" "a.b" >>= (@?= [Only 0])+ deleteFunction conn "s.f"+ macroCount "s" "f" >>= (@?= [Only 0])+ deleteFunction conn "memory.s.g"+ macroCount "s" "g" >>= (@?= [Only 0])+ -- The parser rejects a quote inside a part that has no quotes, so no+ -- statement runs.+ void $ execute_ conn "CREATE TABLE kept (i INTEGER)"+ assertSqlErrorContaining "Unexpected quote" $ deleteFunction conn "x\"; DROP TABLE kept; --"+ query_ conn "SELECT count(*) FROM duckdb_tables() WHERE table_name = 'kept'" >>= (@?= [Only (1 :: Int64)])+ assertSqlErrorContaining "Unexpected quote" $ deleteFunction conn "a\"b"+ assertSqlErrorContaining "between one and three parts" $ deleteFunction conn "a.b.c.d"+ , testCase "stateful replacement releases each replaced callback before the connection closes" $+ withConnection ":memory:" \conn -> do+ createFunctionWithState conn "duplicate_state" (pure ()) (\() -> (42 :: Int64))+ assertReleasedAfter "a replaced stateful function" \token -> do+ replicateM_ 32 $ createFunctionWithState conn "duplicate_state" (readIORef token) (id :: Int64 -> Int64)+ createFunctionWithState conn "duplicate_state" (pure ()) (\() -> (7 :: Int64))+ query_ conn "SELECT duplicate_state()" >>= (@?= [Only (7 :: Int64)])+ , -- DuckDB refuses the type of the argument after the registration+ -- allocates its callbacks.+ testCase "scalar registration with an invalid signature releases its callbacks at the failure" $+ withConnection ":memory:" \conn ->+ assertReleasedAfter "a scalar function with an invalid signature" \token ->+ replicateM_ 32 $ assertSqlError $ createFunction conn "invalid_scalar" (InvalidSignature token)+ , testCase "stateful registration with an invalid signature releases its callbacks at the failure" $+ withConnection ":memory:" \conn ->+ assertReleasedAfter "a stateful function with an invalid signature" \token ->+ replicateM_ 32 $ assertSqlError $ createFunctionWithState conn "invalid_state" (readIORef token) (\_ -> InvalidSignature token)+ , testCase "logging duplicate registration releases each rejected callback at the failure" $+ withConnection ":memory:" \conn -> do+ Logging.registerLogStorage conn "duplicate_log" (\_ -> pure ())+ assertReleasedAfter "a rejected log storage" \token ->+ replicateM_ 32 $ assertSqlError $ Logging.registerLogStorage conn "duplicate_log" (\_ -> void (readIORef token))+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "empty registration names fail safely" $+ withConnection ":memory:" \conn -> replicateM_ 32 do+ assertSqlError $ createFunction conn "" (1 :: Int64)+ assertSqlError $ createFunctionWithState conn "" (pure ()) (\() -> (1 :: Int64))+ assertSqlError $ registerCopy conn "" Nothing+ assertSqlError $ Logging.registerLogStorage conn "" (\_ -> pure ())+ , testCase "NUL registration names fail safely" $+ withConnection ":memory:" \conn -> do+ assertSqlError $ createFunction conn "nul\0scalar" (1 :: Int64)+ assertSqlError $ createFunctionWithState conn "nul\0state" (pure ()) (\() -> (1 :: Int64))+ assertSqlError $ registerCopy conn "nul\0copy" Nothing+ assertSqlError $ Logging.registerLogStorage conn "nul\0log" (\_ -> pure ())+ , testCase "a scalar registration whose setup fails releases its callbacks at the failure" $+ -- The registration forces the argument declaration after it acquires the callbacks.+ -- If a change forces the declaration before the acquisition, this check cannot fail.+ withConnection ":memory:" \conn ->+ assertReleasedAfter "a scalar function whose setup fails" \token -> replicateM_ 32 do+ outcome <- try (createFunction conn "broken_signature" (BrokenSignature token)) :: IO (Either ErrorCall ())+ case outcome of+ Left err -> assertBool (displayException err) ("broken scalar signature" `isInfixOf` displayException err)+ Right () -> assertFailure "expected signature exception"+ , testCase "a raw result of another type than the result vector is an SQL error" $+ withConnection ":memory:" \conn -> do+ createFunction conn "mismatched_result" MismatchedResult+ assertSqlErrorContaining+ "but the result value has the type BIGINT"+ (query_ conn "SELECT mismatched_result()" :: IO [Only Text.Text])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "scalar exceptions become SQL errors" $+ withConnection ":memory:" \conn -> do+ createFunction conn "throw_scalar" (throwIO (userError "scalar callback failure") :: IO Int64)+ assertSqlErrorContaining "scalar callback failure" (query_ conn "SELECT throw_scalar()" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "a scalar callback failure gives an SQL error of the type ErrorInputInvalid" $+ withConnection ":memory:" \conn -> do+ -- An untyped callback failure gets the code DuckDBV2ErrorApi.+ -- DuckDB 2.0.0 reports that code as an invalid input.+ createFunction conn "throw_typed_scalar" (throwIO (userError "typed callback failure") :: IO Int64)+ outcome <- try (query_ conn "SELECT throw_typed_scalar()" :: IO [Only Int64])+ case outcome of+ Left (err :: SQLError) -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ assertBool (show err) ("typed callback failure" `Text.isInfixOf` sqlErrorMessage err)+ Right _ -> assertFailure "expected SQL error"+ , testCase "an internal error of the library in a callback leaves the database usable" $+ withConnection ":memory:" \conn -> do+ -- See Note [Internal errors in callbacks] in "Database.DuckDB.Simple.Native".+ createFunction conn "throw_internal" (throwInternalError "internal callback failure" :: IO Int64)+ createAggregate+ conn+ "internal_step"+ Aggregate+ { aggregateInitial = 0 :: Int64+ , aggregateStep = \_ (Only (_ :: Int64)) -> throwInternalError "internal step failure"+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }+ forM_+ [ ("SELECT throw_internal()", "internal callback failure")+ , ("SELECT internal_step(i) FROM range(3) t(i)", "internal step failure") :: (Text.Text, Text.Text)+ ]+ \(sql, text) -> do+ result <- try (query_ conn (Query sql) :: IO [Only Int64])+ case result of+ Left err -> do+ assertBool (show err) (text `Text.isInfixOf` sqlErrorMessage err)+ -- runCallback holds back ErrorRuntimeInternal.+ sqlErrorType err @?= Just ErrorInputInvalid+ Right rows -> assertFailure (Text.unpack sql <> ": expected the error of the callback, got " <> show rows)+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "exception rendering cannot escape scalar callback" $+ withConnection ":memory:" \conn -> do+ createFunction conn "throw_bad_display" (throwIO BrokenDisplay :: IO Int64)+ assertSqlErrorContaining "Haskell callback failed" (query_ conn "SELECT throw_bad_display()" :: IO [Only Int64])+ , testCase "a failing exception category cannot turn a callback failure into success" $+ withConnection ":memory:" \conn -> do+ let failure = SQLError "bad category" (error "category failed") Nothing Nothing+ createFunction conn "bad_category" (throwIO failure :: IO Int64)+ result <- try (query_ conn "SELECT bad_category()" :: IO [Only Int64])+ case result of+ Left err -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ assertBool (show err) ("Haskell callback failed" `Text.isInfixOf` sqlErrorMessage err)+ Right rows -> assertFailure ("expected a callback failure, got " <> show rows)+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "scalar async exceptions become SQL errors" $+ withConnection ":memory:" \conn -> do+ createFunction conn "throw_async_scalar" (throwIO ThreadKilled :: IO Int64)+ assertSqlErrorContaining "thread killed" (query_ conn "SELECT throw_async_scalar()" :: IO [Only Int64])+ , testCase "scalar initializer exceptions become SQL errors" $+ withConnection ":memory:" \conn -> do+ createFunctionWithState conn "throw_init" (throwIO (userError "state init failure") :: IO ()) (\() -> (1 :: Int64))+ assertSqlErrorContaining "state init failure" (query_ conn "SELECT throw_init()" :: IO [Only Int64])+ , testCase "scalar initializer async exceptions become SQL errors" $+ withConnection ":memory:" \conn -> do+ createFunctionWithState conn "throw_async_init" (throwIO ThreadKilled :: IO ()) (\() -> (1 :: Int64))+ assertSqlErrorContaining "thread killed" (query_ conn "SELECT throw_async_init()" :: IO [Only Int64])+ , testCase "stateful scalar exceptions become SQL errors" $+ withConnection ":memory:" \conn -> do+ createFunctionWithState conn "throw_state" (pure ()) (\() -> throwIO (userError "state execution failure") :: IO Int64)+ assertSqlErrorContaining "state execution failure" (query_ conn "SELECT throw_state()" :: IO [Only Int64])+ , testCase "Float scalar results preserve NaN infinity and negative zero" $+ withConnection ":memory:" \conn ->+ forM_+ [ (0 / 0, isNaN)+ , (1 / 0, \value -> isInfinite value && value > 0)+ , (-1 / 0, \value -> isInfinite value && value < 0)+ , (negate 0, isNegativeZero)+ ]+ \(value, matches) -> do+ createFunction conn "float_value" (value :: Float)+ createFunctionWithState conn "float_state" (pure ()) (\() -> value)+ [(plain, stateful)] <- query_ conn "SELECT float_value(), float_state()" :: IO [(Double, Double)]+ assertBool "plain callback preserves the floating-point value" (matches plain)+ assertBool "stateful callback preserves the floating-point value" (matches stateful)+ , testCase "Word64 scalar preserves unsigned maximum" $+ withConnection ":memory:" \conn -> do+ createFunction conn "word64_identity" (id :: Word64 -> Word64)+ query_ conn "SELECT word64_identity(18446744073709551615::UBIGINT)" >>= (@?= [Only (maxBound :: Word64)])+ query_ conn "SELECT typeof(word64_identity(0::UBIGINT))" >>= (@?= [Only ("UBIGINT" :: Text.Text)])+ , testCase "Word scalar preserves unsigned maximum" $+ withConnection ":memory:" \conn -> do+ createFunction conn "word_maximum" (maxBound :: Word)+ query_ conn "SELECT word_maximum()" >>= (@?= [Only (maxBound :: Word)])+ , testCase "signed scalar preserves limits" $+ withConnection ":memory:" \conn -> do+ createFunction conn "int64_identity" (id :: Int64 -> Int64)+ query_ conn "SELECT int64_identity(x) FROM (VALUES ((-9223372036854775808)::BIGINT), (9223372036854775807::BIGINT)) t(x)"+ >>= (@?= [Only (minBound :: Int64), Only maxBound])+ , testCase "scalar strings preserve UTF8 and embedded NUL" $+ withConnection ":memory:" \conn -> do+ let payload = "Matti \955\0 suffix" :: Text.Text+ createFunction conn "text_payload" payload+ createFunction conn "text_identity" (id :: Text.Text -> Text.Text)+ query_ conn "SELECT text_identity(text_payload())" >>= (@?= [Only payload])+ , testCase "a throwing String result tail fails the query and permits another call" $+ withConnection ":memory:" \conn -> do+ createFunction conn "broken_text" \(index :: Int64) ->+ if index == 1+ then replicate 64 'x' <> error "text result tail failed"+ else replicate 64 'x'+ assertSqlErrorContaining "text result tail failed" $+ (query_ conn "SELECT broken_text(i) FROM range(3) t(i)" :: IO [Only String])+ query_ conn "SELECT broken_text(0)" >>= (@?= [Only (replicate 64 'x')])+ , testGroup+ "native acquisition cleanup"+ [ testCase label do+ released <- newIORef []+ outcome <- try (acquireItems (\item -> atomicModifyIORef' released (\items -> (items <> [item], ()))) actions)+ case outcome of+ Left (failure :: SomeException) ->+ assertBool "the acquisition exception reaches the caller" ("acquisition failed" `isInfixOf` displayException failure)+ Right _ -> assertFailure "the acquisition succeeded"+ readIORef released >>= (@?= [2, 1 :: Int])+ | (label, actions) <-+ [ ("a failing list spine releases the acquired prefix in reverse order", pure 1 : pure 2 : error "acquisition failed")+ , ("a failing acquisition action releases the acquired prefix in reverse order", [pure 1, pure 2, throwIO (userError "acquisition failed")])+ ]+ ]+ , testCase "a failing lazy list parameter permits another query" $+ withConnection ":memory:" \conn -> do+ outcome <- try (query conn "SELECT ?::BIGINT[]" (Only (1 : error "parameter tail failed" :: [Int64])) :: IO [Only [Int64]])+ case outcome of+ Left (failure :: ErrorCall) ->+ assertBool "the list exception reaches the caller" ("parameter tail failed" `isInfixOf` displayException failure)+ Right _ -> assertFailure "the lazy list parameter succeeded"+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "scalar NULL arguments reach Maybe functions" $+ withConnection ":memory:" \conn -> do+ createFunction conn "null_default" (\(x :: Maybe Int64) -> maybe 42 id x)+ query_ conn "SELECT null_default(NULL::BIGINT)" >>= (@?= [Only (42 :: Int64)])+ , testCase "scalar nullable strings across chunks" $+ withConnection ":memory:" \conn -> do+ createFunction conn "optional_text" (\(x :: Int64) -> if even x then Just ("\955\0" <> Text.replicate 32 "x") else Nothing)+ query_ conn "SELECT count(*), count(optional_text(i)), min(length(optional_text(i))) FROM range(5000) t(i)"+ >>= (@?= [(5000 :: Int64, 2500 :: Int64, Just (34 :: Int64))])+ , testCase "scalar DECIMAL inputs use supported numeric casts" $+ withConnection ":memory:" \conn -> do+ createFunction conn "double_identity" (id :: Double -> Double)+ query_ conn "SELECT double_identity(123.25::DECIMAL(10,2))" >>= (@?= [Only (123.25 :: Double)])+ , testCase "COPY exceptions become SQL errors in every phase" $+ withMissingFile \path -> withConnection ":memory:" \conn ->+ mapM_+ ( \phase -> do+ let name = "copy_fail_" <> Text.pack (show phase)+ registerCopy conn name (Just phase)+ assertSqlErrorContaining (Text.pack ("copy phase " <> show phase)) $+ execute conn (Query ("COPY (SELECT 1) TO ? (FORMAT " <> name <> ")")) (Only (Text.pack path))+ )+ [0 .. 3]+ , testCase "COPY state is valid across repeated queries" $+ withMissingFile \path -> withConnection ":memory:" \conn -> do+ registerCopy conn "copy_repeat" Nothing+ replicateM_ 32 $ void $ execute conn "COPY (SELECT * FROM range(5000)) TO ? (FORMAT copy_repeat)" (Only (Text.pack path))+ , testCase "log sink callback exceptions cannot fail a query" do+ calls <- newIORef (0 :: Int)+ withConnection ":memory:" \conn -> do+ Logging.registerLogStorage conn "throwing_log" (\_ -> atomicModifyIORef' calls (\n -> (n + 1, ())) >> throwIO ThreadKilled)+ void $ execute_ conn "CALL enable_logging(); SET logging_storage = 'throwing_log'; SELECT write_log('contained')"+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ readIORef calls >>= assertBool "the sink ran" . (> 0)+ , testCase "log sinks copy entries and flush when released" do+ entries <- newIORef []+ flushes <- newIORef (0 :: Int)+ start <- getCurrentTime+ withConnection ":memory:" \conn -> do+ Logging.registerLogSink+ conn+ "captured_log"+ (\entry -> atomicModifyIORef' entries (\values -> (entry : values, ())))+ (atomicModifyIORef' flushes (\count -> (count + 1, ())))+ void $+ execute_+ conn+ "CALL enable_logging(); SET logging_storage = 'captured_log'; SELECT write_log('Matti λ', level := 'warning', log_type := 'HaskellTest'); SET logging_storage = 'memory'"+ -- DuckDB 2.0.0 flushes the sink five times for these statements.+ readIORef flushes >>= (@?= 5)+ end <- getCurrentTime+ received <- filter ((== "HaskellTest") . Logging.logEntryType) <$> readIORef entries+ map Logging.logEntryMessage received @?= ["Matti λ"]+ map Logging.logEntrySeverity received @?= [Logging.LogLevelWarning]+ map Logging.logEntryScope received @?= [Logging.LogScopeConnection]+ assertBool "entry has a timestamp" (all ((/= Nothing) . Logging.logEntryTimestamp) received)+ -- A timestamp in the wrong unit is far from the clock of the test.+ let nearClock time = addUTCTime (-300) start <= time && time <= addUTCTime 300 end+ assertBool+ ("timestamps " <> show (map Logging.logEntryTimestamp received) <> " are within five minutes of " <> show (start, end))+ (all (maybe False nearClock . Logging.logEntryTimestamp) received)+ assertBool "entry has a connection id" (all ((/= Nothing) . Logging.logEntryConnectionId) received)+ -- The release flushes the sink one more time.+ readIORef flushes >>= (@?= 6)+ , testCase "instance log sinks receive connection and instance messages" do+ entries <- newIORef []+ withEnvironment \environment -> withInstance environment [] \instance_ -> do+ attachDatabase instance_ ":memory:" "shared" []+ setDefaultDatabase instance_ "shared"+ Logging.registerInstanceLogSink instance_ "instance_log" (\entry -> atomicModifyIORef' entries (\values -> (entry : values, ()))) (pure ())+ withConnectionOn instance_ \conn -> do+ void $ execute_ conn "CALL enable_logging(); SET logging_storage = 'instance_log'"+ Logging.logConnectionMessage conn Logging.LogLevelWarning "HaskellConnection" "from a connection"+ Logging.logInstanceMessage instance_ Logging.LogLevelWarning "HaskellInstance" "from an instance"+ void $ execute_ conn "SET logging_storage = 'memory'"+ received <- readIORef entries+ let ofType kind = filter ((== kind) . Logging.logEntryType) received+ map Logging.logEntryMessage (ofType "HaskellConnection") @?= ["from a connection"]+ map Logging.logEntryScope (ofType "HaskellConnection") @?= [Logging.LogScopeConnection]+ assertBool "connection entry has a connection id" (all ((/= Nothing) . Logging.logEntryConnectionId) (ofType "HaskellConnection"))+ map Logging.logEntryMessage (ofType "HaskellInstance") @?= ["from an instance"]+ map Logging.logEntryScope (ofType "HaskellInstance") @?= [Logging.LogScopeDatabase]+ map Logging.logEntryConnectionId (ofType "HaskellInstance") @?= [Nothing]+ , testCase "log entries carry the connection, transaction, and query ids of DuckDB" do+ entries <- newIORef []+ (connectionId, transactionId, queryId) <- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ Logging.registerLogSink conn "id_log" (\entry -> atomicModifyIORef' entries (\values -> (entry : values, ()))) (pure ())+ void $ execute_ conn "CALL enable_logging(); SET logging_storage = 'id_log'"+ -- In a transaction, the transaction id and the query id differ.+ void $ execute_ conn "BEGIN"+ _ <- query_ conn "SELECT 1" :: IO [Only Int64]+ rows <-+ query_+ conn+ "SELECT write_log('ids', level := 'warning', log_type := 'HaskellIds'), current_connection_id(), current_transaction_id(), current_query_id()"+ :: IO [(FieldValue, Word64, Word64, Word64)]+ void $ execute_ conn "COMMIT; SET logging_storage = 'memory'"+ case rows of+ [(_, connection, transaction, queryNumber)] -> pure (connection, transaction, queryNumber)+ _ -> assertFailure ("expected one row, got " <> show rows)+ assertEqual "the ids differ" 3 (length (nub [connectionId, transactionId, queryId]))+ received <- filter ((== "HaskellIds") . Logging.logEntryType) <$> readIORef entries+ map Logging.logEntryConnectionId received @?= [Just connectionId]+ map Logging.logEntryTransactionId received @?= [Just transactionId]+ map Logging.logEntryQueryId received @?= [Just queryId]+ map Logging.logEntryThreadId received @?= [Nothing]+ , testCase "filesystem action exceptions propagate and the file opens again" $+ withTemporaryFile \path ->+ withConnection ":memory:" \conn -> do+ outcome <- try $ FileSystem.withFileHandle conn path [FileSystem.FileFlagWrite] \_ -> throwIO (userError "file action failure")+ case outcome of+ Left (err :: IOException) -> displayException err @?= "user error (file action failure)"+ Right () -> assertFailure "expected file action exception"+ FileSystem.withFileHandle conn path [FileSystem.FileFlagWrite] \handle -> do+ FileSystem.writeFileHandleBytes handle (BS.pack [0, 255]) >>= (@?= 2)+ FileSystem.fileHandleSync handle+ FileSystem.withFileHandle conn path [FileSystem.FileFlagRead] \handle -> do+ FileSystem.readFileHandleChunk handle 0 >>= (@?= BS.empty)+ FileSystem.readFileHandleChunk handle 8 >>= (@?= BS.pack [0, 255])+ , testCase "filesystem errors remain controlled" $+ withConnection ":memory:" \conn -> do+ assertSqlError $ FileSystem.withFileHandle conn "/tmp/duckdb-review-missing-dir/no-file" [FileSystem.FileFlagRead] (\_ -> pure ())+ withTemporaryFile \path ->+ assertInputErrorContaining "file path contains NUL" $+ FileSystem.withFileHandle conn (path <> "\0suffix") [FileSystem.FileFlagRead] (\_ -> pure ())+ , testCase "catalog and config NUL names fail safely" $+ withConnection ":memory:" \conn -> do+ -- The NUL check runs before the handler that gives Nothing for an unknown option.+ nulName <- try (Config.getConfigOption conn "threads\0suffix")+ case nulName of+ Left (err :: SQLError) -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ sqlErrorMessage err @?= "duckdb-simple: config option name contains NUL"+ Right value -> assertFailure ("expected the NUL error, got " <> show (fmap Config.configValueText value))+ -- DuckDB reads the name with its length, so it gets the NUL byte and+ -- reports an unknown option. The library rejects the name before+ -- the native call.+ let rejectsNul = assertInputErrorContaining "config option name contains NUL"+ rejectsNul $ Config.setConfigOption conn Config.SettingScopeSession "threads\0suffix" "1"+ rejectsNul $ Config.setConfigValue conn Config.SettingScopeSession "threads\0suffix" (1 :: Int64)+ rejectsNul $ Config.getConfigValue conn "threads\0suffix"+ rejectsNul $ Config.getConfigDefaultValue conn "threads\0suffix"+ withTransaction conn do+ assertSqlError $ Catalog.catalogTypeName conn "memory\0suffix"+ assertSqlError $ Catalog.lookupCatalogEntry conn "memory" "main\0suffix" "probe" Catalog.CatalogTable+ , testCase "catalog names preserve UTF8" $+ withConnection ":memory:" \conn -> do+ void $ execute_ conn "CREATE TABLE \"\955\" (i INT)"+ withTransaction conn do+ entry <- Catalog.lookupCatalogEntry conn "memory" "main" "\955" Catalog.CatalogTable+ fmap Catalog.catalogEntryName entry @?= Just "\955"+ , testCase "catalog names use ASCII identifier comparison independently of collation" $+ withConnection ":memory:" \conn -> do+ void $ execute_ conn "CREATE TABLE MiXeD (i INT); CREATE TABLE \"Ä\" (i INT); CREATE TABLE \"ä\" (i INT)"+ void $ execute_ conn "SET default_collation = 'nocase'"+ Catalog.catalogTypeName conn "MEMORY" >>= (@?= Just "duckdb")+ entry <- Catalog.lookupCatalogEntry conn "MEMORY" "MAIN" "mixed" Catalog.CatalogTable+ fmap Catalog.catalogEntryName entry @?= Just "MiXeD"+ forM_ ["Ä", "ä"] \name -> do+ found <- Catalog.lookupCatalogEntry conn "memory" "main" name Catalog.CatalogTable+ fmap Catalog.catalogEntryName found @?= Just name+ , testCase "catalog lookup keeps nested and literal-dot schema names distinct" $+ withConnection ":memory:" \conn -> do+ void $ execute_ conn "CREATE SCHEMA parent; CREATE SCHEMA parent.child; CREATE TABLE parent.child.inside (i INT)"+ Catalog.lookupCatalogEntry conn "memory" "child" "inside" Catalog.CatalogTable >>= (@?= Nothing)+ void $ execute_ conn "CREATE SCHEMA \"parent.child\"; CREATE TABLE \"parent.child\".outside (i INT)"+ found <- Catalog.lookupCatalogEntry conn "memory" "parent.child" "outside" Catalog.CatalogTable+ fmap Catalog.catalogEntryName found @?= Just "outside"+ , testCase "catalog and config missing values remain safe" $+ withConnection ":memory:" \conn -> do+ Config.getConfigOption conn "no_such_config_option" >>= (@?= Nothing)+ withTransaction conn do+ Catalog.catalogTypeName conn "no_such_catalog" >>= (@?= Nothing)+ Catalog.lookupCatalogEntry conn "memory" "main" "no_such_table" Catalog.CatalogTable >>= (@?= Nothing)+ , testCase "configuration descriptors remain usable after connection close" $+ withConnection ":memory:" \conn -> do+ Config.setConfigOption conn Config.SettingScopeGlobal "threads" "1"+ options <- Config.listConfigOptions conn+ fmap Config.configValueText <$> Config.getConfigOption conn "threads" >>= (@?= Just (Just "1"))+ close conn+ case find ((== "threads") . Config.configOptionName) options of+ Nothing -> assertFailure "missing threads descriptor"+ Just option -> do+ Config.configOptionSetting option @?= Just "1"+ Config.configOptionName option @?= "threads"+ assertBool "description is retained" (not (Text.null (Config.configOptionDescription option)))+ , testCase "listed options give the aliases of DuckDB" $+ withConnection ":memory:" \conn -> do+ void $ execute_ conn "SET progress_bar_time = 1234"+ options <- Config.listConfigOptions conn+ let option name = find ((== name) . Config.configOptionName) options+ aliases <- query_ conn "SELECT name, aliases FROM duckdb_settings() WHERE len(aliases) > 0 ORDER BY name" :: IO [(Text.Text, [Text.Text])]+ assertBool "DuckDB lists options with aliases" (not (null aliases))+ forM_ aliases \(name, expected) ->+ assertEqual (Text.unpack name) (Just (sort expected)) (sort . Config.configOptionAliases <$> option name)+ Config.configOptionSetting <$> option "progress_bar_time" @?= Just (Just "1234")+ , testCase "option descriptors give Nothing for a NULL value and a missing default" $+ withConnection ":memory:" \conn -> do+ Config.registerCustomOption+ conn+ (Config.customOption "haskell_described" (Registration.TypeExpression "BIGINT"))+ { Config.customOptionDefault = Just (Registration.defaultValue (3 :: Int64))+ }+ void $ execute_ conn "RESET progress_bar_time"+ options <- Config.listConfigOptions conn+ let described get name = get <$> find ((== name) . Config.configOptionName) options+ described Config.configOptionDefault "haskell_described" @?= Just (Just "3")+ -- progress_bar_time has no static default. After RESET its value is 2000.+ described Config.configOptionDefault "progress_bar_time" @?= Just Nothing+ described Config.configOptionSetting "progress_bar_time" @?= Just (Just "2000")+ -- enable_profiling has a NULL value until a write.+ described Config.configOptionSetting "enable_profiling" @?= Just Nothing+ -- getConfigOption shows the NULL value in the same way.+ fmap Config.configValueText <$> Config.getConfigOption conn "enable_profiling" >>= (@?= Just Nothing)+ , testCase "instance options list global values" $+ withEnvironment \environment -> withInstance environment [("threads", "1")] \instance_ -> do+ options <- Config.listInstanceOptions instance_+ fmap Config.configOptionSetting (find ((== "threads") . Config.configOptionName) options) @?= Just (Just "1")+ , testCase "connection options set a session option on one connection only" $+ withEnvironment \environment -> withInstance environment [] \instance_ -> do+ attachDatabase instance_ ":memory:" "shared" []+ setDefaultDatabase instance_ "shared"+ -- DuckDB gives progress_bar_time the session scope by default.+ let options = defaultConnectionOptions{connectionConfig = [("progress_bar_time", "1234")]}+ bracket (connectWithOptions instance_ options) close \configured ->+ withConnectionOn instance_ \other -> do+ query_ configured "SELECT current_setting('progress_bar_time')" >>= (@?= [Only (1234 :: Int64)])+ query_ other "SELECT current_setting('progress_bar_time')" >>= (@?= [Only (2000 :: Int64)])+ , testCase "configuration values resolve case-insensitive names and aliases" $+ withConnection ":memory:" \conn -> do+ Config.setConfigOption conn Config.SettingScopeGlobal "threads" "1"+ forM_ ["threads", "THREADS", "worker_threads", "WORKER_THREADS"] \name ->+ fmap Config.configValueText <$> Config.getConfigOption conn name >>= (@?= Just (Just "1"))+ Config.getConfigOption conn "NO_SUCH_CONFIG_OPTION" >>= (@?= Nothing)+ close conn+ assertSqlError $ Config.getConfigOption conn "threads"+ , testCase "typed reads copy the value of every listed option" $+ withConnection ":memory:" \conn -> do+ options <- Config.listConfigOptions conn+ assertBool "options are listed" (not (null options))+ failures <- forM options \option -> do+ let name = Config.configOptionName option+ outcome <- try (Config.getConfigValue conn name)+ pure case outcome of+ Left (err :: SomeException) -> [(name, displayException err)]+ Right _ -> []+ concat failures @?= []+ assertSqlError $ Config.getConfigValue conn "NO_SUCH_CONFIG_OPTION"+ assertSqlError $ Config.getConfigDefaultValue conn "NO_SUCH_CONFIG_OPTION"+ , testCase "custom options retain typed defaults scopes and callback errors" $+ withConnection ":memory:" \conn -> do+ let option =+ (Config.customOption "haskell_limit" (Registration.StructuredType (LogicalTypeScalar LogicalTypeIdBigInt)))+ { Config.customOptionDefault = Just (Registration.defaultValue (3 :: Int64))+ , Config.customOptionDescription = "A Haskell option"+ , Config.customOptionDefaultScope = Config.SettingScopeSession+ , Config.customOptionScopes = [Config.SettingScopeGlobal, Config.SettingScopeSession]+ , Config.customOptionUpdate = Just \value -> case value of+ FieldInt64 number | number < 0 -> throwIO (userError "limit must be nonnegative")+ _ -> pure ()+ }+ Config.registerCustomOption conn option+ Config.getConfigDefaultValue conn "haskell_limit" >>= (@?= FieldInt64 3)+ Config.setConfigValue conn Config.SettingScopeSession "haskell_limit" (7 :: Int64)+ Config.getConfigValue conn "haskell_limit" >>= (@?= (FieldInt64 7, Config.SettingScopeSession))+ assertSqlErrorContaining "limit must be nonnegative" $ Config.setConfigValue conn Config.SettingScopeSession "haskell_limit" (-1 :: Int64)+ void $ execute_ conn "RESET haskell_limit"+ fmap Config.configValueText <$> Config.getConfigOption conn "haskell_limit" >>= (@?= Just (Just "3"))+ assertSqlError $ Config.registerCustomOption conn option+ , testCase "listConfigFlags lists a custom option of the connection" $+ withConnection ":memory:" \conn -> do+ Config.registerCustomOption conn (Config.customOption "haskell_listed" (Registration.TypeExpression "BIGINT"))+ flags <- Config.listConfigFlags conn+ options <- Config.listConfigOptions conn+ assertBool "the custom option is listed" ("haskell_listed" `elem` map Config.configFlagName flags)+ map Config.configFlagName flags @?= map Config.configOptionName options+ , testCase "typed reads of a VARIANT custom option give the value that it holds" $+ withConnection ":memory:" \conn -> do+ let variantType = Registration.TypeExpression "VARIANT"+ Config.registerCustomOption+ conn+ (Config.customOption "haskell_variant" variantType){Config.customOptionDefault = Just (Registration.defaultValue (5 :: Int64))}+ Config.getConfigDefaultValue conn "haskell_variant" >>= (@?= FieldInt64 5)+ Config.setConfigValue conn Config.SettingScopeGlobal "haskell_variant" (7 :: Int64)+ Config.getConfigValue conn "haskell_variant" >>= (@?= (FieldInt64 7, Config.SettingScopeGlobal))+ -- A nested value gives the value of the same VARIANT in a result.+ void $ execute_ conn "SET haskell_variant = {'a': 1, 'b': [1, 2]}"+ [Only expected] <- query_ conn "SELECT {'a': 1, 'b': [1, 2]}::VARIANT" :: IO [Only FieldValue]+ fst <$> Config.getConfigValue conn "haskell_variant" >>= (@?= expected)+ , testCase "VARIANT option updates decode GLOBAL and SESSION values and can reject a write" $+ withConnection ":memory:" \conn -> do+ seen <- newIORef Nothing+ let update value = case value of+ FieldText "reject" -> throwIO (userError "rejected VARIANT value")+ _ -> writeIORef seen (Just value)+ Config.registerCustomOption conn $+ (Config.customOption "haskell_variant_update" (Registration.TypeExpression "VARIANT"))+ { Config.customOptionScopes = [Config.SettingScopeGlobal, Config.SettingScopeSession]+ , Config.customOptionUpdate = Just update+ }+ [Only expected] <- query_ conn "SELECT {'a': [1,2]}::VARIANT" :: IO [Only FieldValue]+ void $ execute_ conn "SET GLOBAL haskell_variant_update = {'a': [1,2]}"+ readIORef seen >>= (@?= Just expected)+ Config.getConfigValue conn "haskell_variant_update" >>= (@?= (expected, Config.SettingScopeGlobal))+ void $ execute_ conn "SET SESSION haskell_variant_update = 7::BIGINT"+ readIORef seen >>= (@?= Just (FieldInt64 7))+ assertSqlErrorContaining "rejected VARIANT value" $ execute_ conn "SET SESSION haskell_variant_update = 'reject'"+ Config.getConfigValue conn "haskell_variant_update" >>= (@?= (FieldInt64 7, Config.SettingScopeSession))+ , testCase "setConfigValue with an Integer or a Natural checks the range of an unsigned option" $+ withConnection ":memory:" \conn -> do+ -- DuckDB 2.0.0 gives asof_loop_join_threshold the type UBIGINT.+ let option = "asof_loop_join_threshold"+ current = fst <$> Config.getConfigValue conn option+ Config.setConfigValue conn Config.SettingScopeSession option (9 :: Integer)+ current >>= (@?= FieldWord64 9)+ assertSqlErrorContaining "out of range for the destination type UINT64" $+ Config.setConfigValue conn Config.SettingScopeSession option (-5 :: Integer)+ current >>= (@?= FieldWord64 9)+ Config.setConfigValue conn Config.SettingScopeSession option (11 :: Natural)+ current >>= (@?= FieldWord64 11)+ , testCase "setConfigValue with an Integer or a Natural outside the HUGEINT range raises an error" $+ withConnection ":memory:" \conn -> do+ -- toDuckValue gives no BIGNUM value for an Integer or a Natural. See+ -- Note [Integer parameters] in "Database.DuckDB.Simple.ToField".+ let option = "asof_loop_join_threshold"+ assertRangeError $+ Config.setConfigValue conn Config.SettingScopeSession option (2 ^ (128 :: Int) + 5 :: Integer)+ -- A Natural from 2^127 is a UHUGEINT value. DuckDB checks its cast.+ assertSqlErrorContaining "out of range for the destination type UINT64" $+ Config.setConfigValue conn Config.SettingScopeSession option (2 ^ (127 :: Int) :: Natural)+ assertRangeError $+ Config.setConfigValue conn Config.SettingScopeSession option (2 ^ (128 :: Int) + 5 :: Natural)+ , testCase "table descriptions resolve names and copy borrowed column types" $+ withConnection ":memory:" \conn -> do+ void $ execute_ conn "CREATE TABLE \"\955\" (id BIGINT DEFAULT 7, twice BIGINT GENERATED ALWAYS AS (id * 2))"+ description <- Catalog.describeTable conn "\"\955\""+ Catalog.tableDescriptionName description @?= ["memory", "main", "\955"]+ Catalog.tableDescriptionReadonly description @?= False+ map Catalog.columnDescriptionName (Catalog.tableDescriptionColumns description) @?= ["id", "twice"]+ map Catalog.columnDescriptionHasDefault (Catalog.tableDescriptionColumns description) @?= [True, False]+ map Catalog.columnDescriptionGenerated (Catalog.tableDescriptionColumns description) @?= [False, True]+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ , testCase "tokenization copies Unicode byte offsets and unterminated state" $+ withConnection ":memory:" \conn -> do+ let sql = "SELECT \"\955\", 'x'; -- comment\n"+ bytes = Text.encodeUtf8 sql+ parsed <- Tokenizer.tokenize conn sql+ Tokenizer.tokenizationUnterminated parsed @?= False+ forM_ (Tokenizer.tokenizationTokens parsed) \token ->+ BS.take (BS.length (Tokenizer.tokenBytes token)) (BS.drop (Tokenizer.tokenOffset token) bytes) @?= Tokenizer.tokenBytes token+ assertBool "comment token" (any ((== Tokenizer.TokenTypeComment) . Tokenizer.tokenType) (Tokenizer.tokenizationTokens parsed))+ unfinished <- Tokenizer.tokenize conn "SELECT 'unclosed"+ Tokenizer.tokenizationUnterminated unfinished @?= True+ , testCase "tokenization reports the type of each token" $+ withConnection ":memory:" \conn -> do+ parsed <- Tokenizer.tokenize conn "SELECT amount, 'text', 42 + 1; -- note\n"+ [(Tokenizer.tokenType token, Tokenizer.tokenBytes token) | token <- Tokenizer.tokenizationTokens parsed]+ @?= [ (Tokenizer.TokenTypeKeyword, "SELECT")+ , (Tokenizer.TokenTypeIdentifier, "amount")+ , (Tokenizer.TokenTypeOperator, ",")+ , (Tokenizer.TokenTypeStringLiteral, "'text'")+ , (Tokenizer.TokenTypeOperator, ",")+ , (Tokenizer.TokenTypeNumberLiteral, "42")+ , (Tokenizer.TokenTypeOperator, "+")+ , (Tokenizer.TokenTypeNumberLiteral, "1")+ , (Tokenizer.TokenTypeTerminator, ";")+ , (Tokenizer.TokenTypeComment, "-- note\n")+ ]+ , testCase "configuration writes and reads each setting scope" $+ withConnection ":memory:" \conn -> do+ let option name scope =+ (Config.customOption name (Registration.StructuredType (LogicalTypeScalar LogicalTypeIdBigInt)))+ { Config.customOptionDefault = Just (Registration.defaultValue (0 :: Int64))+ , Config.customOptionDefaultScope = scope+ , Config.customOptionScopes = [Config.SettingScopeGlobal, Config.SettingScopeSession]+ }+ Config.registerCustomOption conn (option "haskell_global_scope" Config.SettingScopeGlobal)+ Config.registerCustomOption conn (option "haskell_session_scope" Config.SettingScopeSession)+ Config.setConfigValue conn Config.SettingScopeDefault "haskell_global_scope" (1 :: Int64)+ Config.getConfigValue conn "haskell_global_scope" >>= (@?= (FieldInt64 1, Config.SettingScopeGlobal))+ Config.setConfigOption conn Config.SettingScopeDefault "haskell_session_scope" "2"+ Config.getConfigValue conn "haskell_session_scope" >>= (@?= (FieldInt64 2, Config.SettingScopeSession))+ Config.setConfigValue conn Config.SettingScopeGlobal "haskell_global_scope" (3 :: Int64)+ Config.getConfigValue conn "haskell_global_scope" >>= (@?= (FieldInt64 3, Config.SettingScopeGlobal))+ Config.setConfigOption conn Config.SettingScopeSession "haskell_global_scope" "4"+ Config.getConfigValue conn "haskell_global_scope" >>= (@?= (FieldInt64 4, Config.SettingScopeSession))+ fmap Config.configValueScope <$> Config.getConfigOption conn "haskell_global_scope" >>= (@?= Just Config.SettingScopeSession)+ options <- Config.listConfigOptions conn+ sort+ [ (Config.configOptionName descriptor, Config.configOptionScope descriptor, Config.configOptionScopes descriptor)+ | descriptor <- options+ , "haskell_" `Text.isPrefixOf` Config.configOptionName descriptor+ ]+ @?= [ ("haskell_global_scope", Config.SettingScopeGlobal, [Config.SettingScopeGlobal, Config.SettingScopeSession])+ , ("haskell_session_scope", Config.SettingScopeSession, [Config.SettingScopeGlobal, Config.SettingScopeSession])+ ]+ , testCase "log sinks receive each log level and its scope" do+ entries <- newIORef []+ withConnection ":memory:" \conn -> do+ Logging.registerLogSink conn "level_log" (\entry -> atomicModifyIORef' entries (\values -> (entry : values, ()))) (pure ())+ void $ execute_ conn "CALL enable_logging(level := 'trace'); SET logging_storage = 'level_log'"+ forM_ [minBound .. maxBound] \level -> Logging.logConnectionMessage conn level "HaskellLevels" (Text.pack (show level))+ void $ execute_ conn "SET logging_storage = 'memory'"+ received <- filter ((== "HaskellLevels") . Logging.logEntryType) <$> readIORef entries+ sort [(Logging.logEntrySeverity entry, Logging.logEntryLevel entry, Logging.logEntryMessage entry) | entry <- received]+ @?= [ (level, name, Text.pack (show level))+ | (level, name) <- zip [minBound .. maxBound] ["TRACE", "DEBUG", "INFO", "WARNING", "ERROR", "FATAL"]+ ]+ map Logging.logEntryScope received @?= replicate 6 Logging.LogScopeConnection+ , testCase "native errors report their error type" $+ withConnection ":memory:" \conn ->+ forM_+ [ ("SELEC 1", ErrorQueryParser)+ , ("SELECT * FROM no_such_table", ErrorDatabaseCatalog)+ , ("SELECT no_such_column", ErrorQueryBinder)+ , ("SELECT 'abc'::INTEGER", ErrorTypeConversion)+ ]+ \(sql, expected) -> do+ outcome <- try (execute_ conn sql)+ case outcome of+ Left (err :: SQLError) -> sqlErrorType err @?= Just expected+ Right _ -> assertFailure ("expected an error from " <> show sql)+ , testCase "each enum value has the name of its native value" do+ assertNativeRoundTrip settingScopeToNative settingScopeFromNative DuckDBV2SettingScopeMaxEnum+ assertNativeRoundTrip logLevelToNative logLevelFromNative DuckDBV2LogLevelMaxEnum+ -- The package converts these three types in one direction only.+ assertFromNative logScopeFromNative [DuckDBV2LogScopeDatabase, DuckDBV2LogScopeConnection, DuckDBV2LogScopeThread] DuckDBV2LogScopeMaxEnum+ assertFromNative+ tokenTypeFromNative+ [ DuckDBV2TokenTypeInvalid+ , DuckDBV2TokenTypeKeyword+ , DuckDBV2TokenTypeIdentifier+ , DuckDBV2TokenTypeStringLiteral+ , DuckDBV2TokenTypeNumberLiteral+ , DuckDBV2TokenTypeOperator+ , DuckDBV2TokenTypeComment+ , DuckDBV2TokenTypeTerminator+ , DuckDBV2TokenTypeEndOfInput+ ]+ DuckDBV2TokenTypeMaxEnum+ let macroTypes = [minBound .. maxBound] :: [MacroFunctionType]+ map (show . macroFunctionTypeToNative) macroTypes @?= map (("DuckDBV2" <>) . show) macroTypes+ let errorTypes = [minBound .. maxBound] :: [ErrorType]+ map (\errorType -> lookup errorType errorTypeCodes >>= errorTypeFromNative) errorTypes @?= map Just errorTypes+ map (fmap show . (`lookup` errorTypeCodes)) errorTypes @?= map (Just . ("DuckDBV2" <>) . show) errorTypes+ map errorTypeFromNative [DuckDBV2ErrorNone, DuckDBV2ErrorMaxEnum] @?= [Nothing, Nothing]+ , testCase "positioned file operations preserve the sequential position" $+ withTemporaryFile \path -> withConnection ":memory:" \conn ->+ FileSystem.withFileHandle conn path [FileSystem.FileFlagRead, FileSystem.FileFlagWrite] \handle -> do+ FileSystem.writeFileHandleBytes handle (BS.pack [1, 2, 3, 4]) >>= (@?= 4)+ FileSystem.fileHandleSeek handle 1+ FileSystem.readFileHandleAt handle 2 2 >>= (@?= BS.pack [3, 4])+ FileSystem.fileHandleTell handle >>= (@?= 1)+ FileSystem.writeFileHandleAt handle 0 (BS.pack [8, 9])+ FileSystem.fileHandleTell handle >>= (@?= 1)+ FileSystem.fileHandleSeek handle 0+ FileSystem.readFileHandleChunk handle 8 >>= (@?= BS.pack [8, 9, 3, 4])+ , testCase "file reads shorter than the file give the requested bytes" $+ withTemporaryFile \path -> withConnection ":memory:" \conn -> do+ let bytes = BS.pack [0 .. 15]+ BS.writeFile path bytes+ FileSystem.withFileHandle conn path [FileSystem.FileFlagRead] \handle -> do+ FileSystem.readFileHandleChunk handle 4 >>= (@?= BS.take 4 bytes)+ FileSystem.fileHandleTell handle >>= (@?= 4)+ FileSystem.readFileHandleAt handle 8 4 >>= (@?= BS.take 4 (BS.drop 8 bytes))+ -- A chunk read gives the bytes that remain after the position.+ FileSystem.readFileHandleChunk handle 100 >>= (@?= BS.drop 4 bytes)+ , testCase "a positioned read past the end gives the error of DuckDB" $+ withTemporaryFile \path -> withConnection ":memory:" \conn -> do+ BS.writeFile path (BS.pack [0 .. 15])+ FileSystem.withFileHandle conn path [FileSystem.FileFlagRead] \handle -> do+ outcome <- try (FileSystem.readFileHandleAt handle 14 4)+ case outcome of+ Left (err :: SQLError) -> do+ sqlErrorType err @?= Just ErrorIOGeneral+ assertBool "DuckDB read error" ("Could not read enough bytes" `Text.isInfixOf` sqlErrorMessage err)+ Right bytes -> assertFailure ("expected an IO error, got " <> show bytes)+ -- The library checks only that the end of the range fits in Int64.+ assertInputErrorContaining "file read range exceeds Int64 range" $ FileSystem.readFileHandleAt handle maxBound 1+ assertInputErrorContaining "file offset is negative" $ FileSystem.readFileHandleAt handle (-1) 1+ assertInputErrorContaining "file read size is negative" $ FileSystem.readFileHandleAt handle 0 (-1)+ assertInputErrorContaining "file offset is negative" $ FileSystem.fileHandleSeek handle (-1)+ -- A size of 0 gives no bytes, also past the end, as in DuckDB.+ FileSystem.readFileHandleAt handle 20 0 >>= (@?= BS.empty)+ FileSystem.readFileHandleAt handle 12 4 >>= (@?= BS.pack [12 .. 15])+ , testCase "a positioned write checks that its end fits in Int64" $+ withTemporaryFile \path -> withConnection ":memory:" \conn ->+ FileSystem.withFileHandle conn path [FileSystem.FileFlagWrite] \handle -> do+ assertInputErrorContaining "file write range exceeds Int64 range" $ FileSystem.writeFileHandleAt handle (maxBound - 1) (BS.pack [1, 2])+ assertInputErrorContaining "file offset is negative" $ FileSystem.writeFileHandleAt handle (-1) (BS.pack [1])+ FileSystem.fileHandleSize handle >>= (@?= 0)+ , testCase "file flags of the other file tests create read and write a file" $+ withTemporaryFile \path -> withConnection ":memory:" \conn -> do+ removeFile path+ FileSystem.withFileHandle conn path [FileSystem.FileFlagCreate, FileSystem.FileFlagWrite] \handle ->+ FileSystem.writeFileHandleBytes handle (BS.pack [1, 2, 3]) >>= (@?= 3)+ FileSystem.withFileHandle conn path [FileSystem.FileFlagWrite] \handle ->+ FileSystem.writeFileHandleBytes handle (BS.pack [4]) >>= (@?= 1)+ FileSystem.withFileHandle conn path [FileSystem.FileFlagRead] \handle ->+ FileSystem.readFileHandleChunk handle 8 >>= (@?= BS.pack [4, 2, 3])+ FileSystem.withFileHandle conn path [FileSystem.FileFlagRead, FileSystem.FileFlagWrite] \handle -> do+ FileSystem.readFileHandleChunk handle 8 >>= (@?= BS.pack [4, 2, 3])+ FileSystem.writeFileHandleBytes handle (BS.pack [5]) >>= (@?= 1)+ FileSystem.fileHandleSize handle >>= (@?= 4)+ , testCase "file flags truncate a file and refuse to create an existing file" $+ withTemporaryFile \path -> withConnection ":memory:" \conn -> do+ BS.writeFile path (BS.pack [1, 2, 3])+ FileSystem.withFileHandle conn path [FileSystem.FileFlagWrite, FileSystem.FileFlagCreateNew] FileSystem.fileHandleSize >>= (@?= 0)+ let exclusive = [FileSystem.FileFlagWrite, FileSystem.FileFlagCreate, FileSystem.FileFlagExclusiveCreate]+ assertSqlError $ FileSystem.withFileHandle conn path exclusive (\_ -> pure ())+ removeFile path+ FileSystem.withFileHandle conn path exclusive \handle ->+ FileSystem.writeFileHandleBytes handle (BS.pack [7]) >>= (@?= 1)+ FileSystem.withFileHandle conn path [FileSystem.FileFlagRead] \handle ->+ FileSystem.readFileHandleChunk handle 8 >>= (@?= BS.pack [7])+ , testCase "file paths support expansion glob movement and removal" $+ withTemporaryFile \path -> withConnection ":memory:" \conn -> do+ FileSystem.pathExists conn path >>= (@?= True)+ -- A directory is not a file.+ FileSystem.pathExists conn (takeDirectory path) >>= (@?= False)+ home <- getHomeDirectory+ FileSystem.expandPath conn "~/duckdb-simple-path" >>= (@?= home <> "/duckdb-simple-path")+ FileSystem.globPaths conn (path <> "*") >>= (@?= [path])+ FileSystem.pathIsRemote conn "https://example.com/data.parquet" >>= (@?= True)+ FileSystem.pathIsRemote conn path >>= (@?= False)+ let moved = path <> ".moved"+ bracket_ (FileSystem.movePath conn path moved) (FileSystem.movePath conn moved path) do+ FileSystem.pathExists conn path >>= (@?= False)+ FileSystem.pathExists conn moved >>= (@?= True)+ ( do+ FileSystem.removePath conn path >>= (@?= True)+ FileSystem.removePath conn path >>= (@?= False)+ )+ `finally` BS.writeFile path BS.empty+ assertSqlError $ FileSystem.globPaths conn (path <> "\0suffix")+ ]++-- | Raise a second exception when the callback error is rendered.+data BrokenDisplay = BrokenDisplay+ deriving (Show)++instance Exception BrokenDisplay where+ displayException _ = error "broken error rendering"++-- | Force a setup exception after scalar callbacks have been acquired. The+-- registration reads the declaration of the argument in the setup. The token+-- lets a case check the release of the value.+newtype BrokenSignature = BrokenSignature (IORef Int64)++instance Function BrokenSignature where+ argumentTypes _ = [Right (error "broken scalar signature")]+ returnType _ = returnType (Proxy @Int64)+ isVolatile _ = False+ applyFunction _ _ = error "unused function"++-- | Declare a VARCHAR result, but give a BIGINT value.+data MismatchedResult = MismatchedResult++instance Function MismatchedResult where+ argumentTypes _ = []+ returnType _ = Right (Registration.TypeExpression "VARCHAR")+ isVolatile _ = True+ applyFunction fields _ = applyFunction fields (1 :: Int64)++-- | Supply an argument type which DuckDB rejects at registration. The token+-- lets a case check the release of the value.+newtype InvalidSignature = InvalidSignature (IORef Int64)++instance Function InvalidSignature where+ argumentTypes _ = [Right (Registration.TypeExpression "no_such_type")]+ returnType _ = returnType (Proxy @Int64)+ isVolatile _ = False+ applyFunction _ _ = error "unregistered function"++-- | Register a COPY callback and optionally fail one phase. The bind state and+-- the global state are the column count that DuckDB gives. Each query of this+-- module has one column.+registerCopy :: Connection -> Text.Text -> Maybe Int -> IO ()+registerCopy conn name failingPhase =+ Copy.registerCopyToFunction+ conn+ name+ (\info -> failPhase 0 >> pure (length (Copy.copyBindColumnTypes info)))+ (\info -> (Copy.copyInitBindState info @?= 1) >> failPhase 1 >> pure (Copy.copyInitBindState info))+ (\info _ -> (Copy.copySinkGlobalState info @?= 1) >> failPhase 2)+ (\info -> (Copy.copyFinalizeGlobalState info @?= 1) >> failPhase 3)+ where+ failPhase phase =+ if failingPhase == Just phase+ then throwIO (userError ("copy phase " <> show phase))+ else pure ()++-- | Require a controlled SQL exception.+assertSqlError :: IO a -> Assertion+assertSqlError = assertSqlErrorContaining Text.empty++-- | Convert each value to its native value and back. The native value must have+-- the pattern name of the value. The native sentinel must raise 'SQLError'.+assertNativeRoundTrip :: (Bounded a, Enum a, Eq a, Show a, Show native) => (a -> native) -> (native -> IO a) -> native -> Assertion+assertNativeRoundTrip toNative fromNative sentinel = do+ let values = [minBound .. maxBound]+ mapM (fromNative . toNative) values >>= (@?= values)+ map (show . toNative) values @?= map (("DuckDBV2" <>) . show) values+ assertSqlErrorContaining "unknown native" (fromNative sentinel)++-- | Check that the native values convert to all constructors in order, and that+-- each constructor has the pattern name of its native value. The native sentinel+-- must raise 'SQLError'.+assertFromNative :: (Bounded a, Enum a, Eq a, Show a, Show native) => (native -> IO a) -> [native] -> native -> Assertion+assertFromNative fromNative natives sentinel = do+ values <- mapM fromNative natives+ values @?= [minBound .. maxBound]+ map show natives @?= map (("DuckDBV2" <>) . show) values+ assertSqlErrorContaining "unknown native" (fromNative sentinel)++-- | Give the action a unique temporary path where no file exists. Remove the+-- file after the action if the action created it. DuckDB writes a COPY TO+-- target that exists through a second file, so the path has no file.+withMissingFile :: (FilePath -> IO a) -> IO a+withMissingFile = bracket acquire release+ where+ acquire = do+ directory <- getTemporaryDirectory+ (path, handle) <- openBinaryTempFile directory "duckdb-extension-regression"+ hClose handle+ removeFile path+ pure path+ release path = doesFileExist path >>= \exists -> when exists (removeFile path)++-- | Remove a temporary file after the action.+withTemporaryFile :: (FilePath -> IO a) -> IO a+withTemporaryFile action =+ bracket+ (do (path, handle) <- openBinaryTempFile "/tmp" "duckdb-extension-regression"; hClose handle; pure path)+ removeFile+ action
test/GeometryRegressionTests.hs view
@@ -1,404 +1,671 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-} {-# LANGUAGE OverloadedStrings #-}-{-# OPTIONS_GHC -Wno-deprecations #-} +-- |+-- Module : GeometryRegressionTests+-- Description : Regression tests for GEOMETRY construction, metadata, nesting, and streaming. module GeometryRegressionTests (tests) where -import Control.Exception (SomeException, bracket, displayException, try)-import Control.Monad (forM_)-import Data.Array (Array, listArray)+import Control.Exception (bracket, try)+import Control.Monad (forM, forM_)+import Data.Array (elems, listArray) import qualified Data.ByteString as BS import qualified Data.Geometry as G import qualified Data.Geometry.WKB as WKB import qualified Data.Geometry.WKT as WKT import Data.Int (Int64)-import Data.List (isInfixOf) import qualified Data.Map.Strict as Map import Data.Text (Text) import qualified Data.Text as Text import qualified Data.Vector as V import qualified Data.Vector.Unboxed as U import Database.DuckDB.Simple-import qualified Database.DuckDB.Simple.Deprecated.Streaming as Streaming-import Database.DuckDB.Simple.FromField (FieldValue (..), StructValue (..), UnionValue)+import Database.DuckDB.Simple.FromField (FieldValue (..), StructValue (..), UnionValue (..)) import Database.DuckDB.Simple.Geometry (RawGeometry (..), fromRawGeometry, toRawGeometry)-import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructField (..), destroyLogicalType, logicalTypeFromRep, logicalTypeToRep)+import Database.DuckDB.Simple.Internal (withConnectionFactory, withConnectionHandle)+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructField (..), UnionMemberType (..))+import qualified Database.DuckDB.Simple.Streaming as Streaming+import Database.DuckDB.Simple.V2.LogicalRep (destroyLogicalType, logicalTypeFromRep, logicalTypeToRep)+import Database.DuckDB.Simple.Variant (Variant (..)) import GHC.Float (castWord64ToDouble) import System.Mem (performMajorGC) import Test.Tasty (TestTree, testGroup) import Test.Tasty.HUnit-import TestUtils (assertFailureIO)+import TestUtils (assertFailureIO, assertSqlErrorType, sized) -- | Check native construction, metadata, nesting, and streaming. tests :: TestTree tests =- testGroup- "geometry integration"- [ testCase "Haskell shapes bind directly in every coordinate layout" $- withConnection ":memory:" \conn -> do- shapeRoundTrips conn G.DimXY G.PointXY G.CoordinatesXY (G.XY 1 2) (G.XY 3 4)- shapeRoundTrips conn G.DimXYZ G.PointXYZ G.CoordinatesXYZ (G.XYZ 1 2 3) (G.XYZ 4 5 6)- shapeRoundTrips conn G.DimXYM G.PointXYM G.CoordinatesXYM (G.XYM 1 2 3) (G.XYM 4 5 6)- shapeRoundTrips conn G.DimXYZM G.PointXYZM G.CoordinatesXYZM (G.XYZM 1 2 3 4) (G.XYZM 5 6 7 8)- , testCase "large coordinate buffers survive native round trips" $- withConnection ":memory:" \conn -> do- let coordinates = U.generate 10000 (\i -> G.XY (fromIntegral i) (fromIntegral (i * 2)))- points = U.imap (\i coordinate -> if i `mod` 17 == 0 then G.EmptyPoint G.DimXY else G.PointXY coordinate) coordinates- forM_ [G.LineString (G.CoordinatesXY coordinates), G.MultiPoint points] \shape -> do- (query conn "SELECT ?" (Only shape)) >>= (@?= [Only shape])- raw <- either assertFailure pure (toRawGeometry shape)- (importRaw conn raw) >>= (@?= [Only raw])- , testCase "WKT binding preserves exact finite coordinate bits" $- withConnection ":memory:" \conn ->- forM_ [0, 0x8000000000000000, 1, 0x8000000000000001, 0x000fffffffffffff, 0x0010000000000000, 0x3fb999999999999b, 0x3ff0000000000001, 0x44b52d02c7e14af6, 0x7fefffffffffffff, 0xffefffffffffffff] \bits -> do- let shape = G.PointGeometry (G.PointXY (G.XY (castWord64ToDouble bits) (-0.0)))- expected <- either assertFailure pure (WKB.encodeWKB shape)- [(actual, wkt)] <- query conn "SELECT ST_AsWKB(?), ST_AsText(?)" (shape, shape) :: IO [(BS.ByteString, Text)]- actual @?= expected- (WKT.decodeWKT wkt >>= WKB.encodeWKB) @?= Right expected- (query conn "SELECT system.main.ST_AsWKB(system.main.ST_GeomFromWKB(?))" (Only expected) :: IO [Only BS.ByteString])- >>= (@?= [Only expected])- , testCase "pure WKT decoding agrees with native geometry results" $- withConnection ":memory:" \conn ->- forM_ shapes \wkt -> do- [(native, rendered)] <- query conn "SELECT ?::GEOMETRY, ST_AsText(?::GEOMETRY)" (wkt, wkt) :: IO [(G.Geometry, Text)]- WKT.decodeWKT wkt @?= Right native- WKT.decodeWKT rendered @?= Right native- , testCase "native bare multipoint EMPTY syntax parses as WKT" $- withConnection ":memory:" \conn -> do- [(native, bytes, rendered)] <-- query_ conn "SELECT g, ST_AsWKB(g), ST_AsText(g) FROM (SELECT 'MULTIPOINT ((1 2), EMPTY, (3 4))'::GEOMETRY AS g)" :: IO [(G.Geometry, BS.ByteString, Text)]- native @?= G.MultiPoint (U.fromList [G.PointXY (G.XY 1 2), G.EmptyPoint G.DimXY, G.PointXY (G.XY 3 4)])- WKB.decodeWKB bytes @?= Right native- WKT.decodeWKT rendered @?= Right native- (WKT.encodeWKT native >>= WKT.decodeWKT) @?= Right native- , testCase "standalone WKT syntax binds through native geometry parameters" $- withConnection ":memory:" \conn ->- forM_ ["point(+.5 -1.e+2)", "MULTIPOINT (EMPTY, (1 2), (3 4))", "MULTIPOINT (EMPTY, 1 2, EMPTY, 3 4, EMPTY)"] \wkt -> do- shape <- either assertFailure pure (WKT.decodeWKT wkt)- (query conn "SELECT ?" (Only shape)) >>= (@?= [Only shape])- , testCase "native geometry depth limit produces a recoverable error" $- withConnection ":memory:" \conn -> do- let nested = iterate (G.GeometryCollection . V.singleton) (G.PointGeometry (G.PointXY (G.XY 1 2)))- accepted = nested !! 15- rejected = nested !! 16- (query conn "SELECT ?" (Only accepted)) >>= (@?= [Only accepted])- assertFailureIO (query conn "SELECT ?" (Only rejected) :: IO [Only G.Geometry])- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])- , testCase "decoded native shapes round trip through parameters and WKB" $- withConnection ":memory:" \conn ->- forM_ shapes \wkt -> do- [Only geometry] <- query conn "SELECT ?::GEOMETRY" (Only wkt) :: IO [Only G.Geometry]- (query conn "SELECT ?" (Only geometry) :: IO [Only G.Geometry]) >>= (@?= [Only geometry])- raw <- either assertFailure pure (toRawGeometry geometry)- fromRawGeometry raw @?= Right geometry- , testCase "decoded rows retain point and sequence layouts, including empties" $- withConnection ":memory:" \conn ->- forM_- [ ("POINT Z EMPTY", G.PointGeometry (G.EmptyPoint G.DimXYZ))- , ("LINESTRING M EMPTY", G.LineString (G.CoordinatesXYM U.empty))- , ("POINT Z (1 2 3)", G.PointGeometry (G.PointXYZ (G.XYZ 1 2 3)))- , ("POINT EMPTY", G.PointGeometry (G.EmptyPoint G.DimXY))- ]- \(wkt, expected) ->- (query conn "SELECT ?::GEOMETRY" (Only (wkt :: Text)) :: IO [Only G.Geometry])- >>= (@?= [Only expected])- , testCase "decoded arrays preserve shapes and NULL elements without CRS" $- withConnection ":memory:" \conn -> do- let geometry = G.PointGeometry (G.PointXYM (G.XYM 1 2 3))- values = listArray (0, 2) [Nothing, Just geometry, Just (G.PointGeometry (G.EmptyPoint G.DimXYM))]- (query conn "SELECT ?, ST_CRS((?)[2])" (values, values) :: IO [(Array Int (Maybe G.Geometry), Maybe Text)])- >>= (@?= [(values, Nothing)])- , testCase "only raw geometry retains CRS metadata" $- withConnection ":memory:" \conn -> do- [(raw, shape)] <- query_ conn "SELECT g, g FROM (SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84') AS g)" :: IO [(RawGeometry, G.Geometry)]- rawGeometryCRS raw @?= Just "OGC:CRS84"- shape @?= G.PointGeometry (G.PointXYZM (G.XYZM 1 2 3 4))- fromRawGeometry raw @?= Right shape- unannotated <- either assertFailure pure (toRawGeometry shape)- rawGeometryCRS unannotated @?= Nothing- rawGeometryWKB unannotated @?= rawGeometryWKB raw- (query conn "SELECT system.main.ST_CRS(system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?))" (rawParameters raw) :: IO [Only (Maybe Text)])- >>= (@?= [Only (Just "OGC:CRS84")])- (query conn "SELECT ST_CRS(?), ?" (shape, shape) :: IO [(Maybe Text, G.Geometry)])- >>= (@?= [(Nothing, shape)])- , testCase "decoded coordinates remain usable after closing their connection" $ do- geometry <- withConnection ":memory:" \conn -> do- [Only value] <- query_ conn "SELECT 'LINESTRING (1 2, 3 4, 5 6)'::GEOMETRY('OGC:CRS84')" :: IO [Only G.Geometry]- pure value- performMajorGC- case geometry of- G.LineString (G.CoordinatesXY coords) -> U.foldl' (\acc (G.XY x y) -> acc + x + y) 0 coords @?= 21- other -> assertFailure ("unexpected geometry: " <> show other)- , testCase "native WKB round trips through explicit binary import" $- withConnection ":memory:" \conn ->- forM_ shapes \wkt -> do- [Only geometry] <- query conn "SELECT ?::GEOMETRY" (Only wkt)- (query conn "SELECT typeof(system.main.ST_GeomFromWKB(?)), system.main.ST_GeomFromWKB(?)" (rawGeometryWKB geometry, rawGeometryWKB geometry) :: IO [(Text, RawGeometry)])- >>= (@?= [("GEOMETRY", geometry)])- , testCase "raw empty containers retain layouts that decoded values do not store" $- withConnection ":memory:" \conn ->- forM_- ( [ (family <> " " <> dimensions <> " EMPTY", family <> " EMPTY")- | family <- ["MULTIPOINT", "MULTILINESTRING", "MULTIPOLYGON", "GEOMETRYCOLLECTION"]- , dimensions <- ["Z", "M", "ZM"]- ]- ++ [("GEOMETRYCOLLECTION ZM (MULTIPOINT ZM EMPTY, GEOMETRYCOLLECTION ZM EMPTY)", "GEOMETRYCOLLECTION (MULTIPOINT EMPTY, GEOMETRYCOLLECTION EMPTY)")]- )- \(wkt, normalizedWKT) -> do- [Only raw] <- query conn "SELECT ?::GEOMETRY('OGC:CRS84')" (Only (wkt :: Text)) :: IO [Only RawGeometry]- (importRaw conn raw) >>= (@?= [Only raw])- decoded <- either assertFailure pure (fromRawGeometry raw)- WKT.encodeWKT decoded @?= Right normalizedWKT- normalized <- either assertFailure pure (toRawGeometry decoded)- rawGeometryCRS normalized @?= Nothing- assertBool "decoded empty containers must not retain an unstored dimension tag" (rawGeometryWKB normalized /= rawGeometryWKB raw)- , testCase "mixed-layout collections use binary import without WKT normalization" $- withConnection ":memory:" \conn -> do- let geometry = G.GeometryCollection (V.fromList [G.PointGeometry (G.PointXY (G.XY 1 2)), G.PointGeometry (G.PointXYZ (G.XYZ 3 4 5))])- raw <- either assertFailure pure (toRawGeometry geometry)- assertFailureIO (query conn "SELECT ?" (Only geometry) :: IO [Only G.Geometry])- (importRaw conn raw) >>= (@?= [Only raw])- (query conn "SELECT ST_GeomFromWKB(?)" (Only (rawGeometryWKB raw)) :: IO [Only G.Geometry])- >>= (@?= [Only geometry])- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])- , testCase "mixed multipoint layouts promote missing ordinates to NaN" $- withConnection ":memory:" \conn -> do- let geometry = G.MultiPoint (U.fromList [G.PointXY (G.XY 1 2), G.PointXYZ (G.XYZ 3 4 5)])- [Only actual] <- query conn "SELECT ?" (Only geometry) :: IO [Only G.Geometry]- case actual of- G.MultiPoint points -> do- U.length points @?= 2- case points U.! 0 of- G.PointXYZ (G.XYZ x y z) -> do- (x, y) @?= (1, 2)- assertBool "promoted Z must be NaN" (isNaN z)- other -> assertFailure ("unexpected promoted point: " <> show other)- points U.! 1 @?= G.PointXYZ (G.XYZ 3 4 5)- other -> assertFailure ("unexpected geometry: " <> show other)- , testCase "raw NaN points retain ordinates that decoded emptiness normalizes" $- withConnection ":memory:" \conn -> do- [Only raw] <- query_ conn "SELECT 'POINT Z (NaN NaN 7)'::GEOMETRY" :: IO [Only RawGeometry]- (importRaw conn raw) >>= (@?= [Only raw])- decoded <- either assertFailure pure (fromRawGeometry raw)- decoded @?= G.PointGeometry (G.EmptyPoint G.DimXYZ)- normalized <- either assertFailure pure (toRawGeometry decoded)- assertBool "decoding an XY-NaN point normalizes its extra ordinates" (rawGeometryWKB normalized /= rawGeometryWKB raw)- (query conn "SELECT ?" (Only decoded) :: IO [Only G.Geometry]) >>= (@?= [Only decoded])- , testCase "binary import preserves distinct NaN payload bits" $- withConnection ":memory:" \conn -> do- let shape = G.PointGeometry (G.PointXYZ (G.XYZ (castWord64ToDouble 0x7ff8000000000001) (castWord64ToDouble 0x7ff8000000000002) 7))- raw <- either assertFailure pure (toRawGeometry shape)- (importRaw conn raw) >>= (@?= [Only raw])- fromRawGeometry raw @?= Right (G.PointGeometry (G.EmptyPoint G.DimXYZ))- , testCase "binary import preserves mixed empty and nonempty member layouts" $- withConnection ":memory:" \conn -> do- let shape = G.GeometryCollection (V.fromList [G.PointGeometry (G.EmptyPoint G.DimXY), G.PointGeometry (G.PointXYZ (G.XYZ 1 2 3))])- raw <- either assertFailure pure (toRawGeometry shape)- (query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (rawGeometryWKB raw)) :: IO [Only RawGeometry]) >>= (@?= [Only raw])- (query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (rawGeometryWKB raw)) :: IO [Only G.Geometry]) >>= (@?= [Only shape])- , testCase "decoded polygon writing normalizes all-empty rings" $- withConnection ":memory:" \conn -> do- let emptyRing = G.CoordinatesXYZ U.empty- geometry = G.Polygon (G.PolygonRings emptyRing (V.singleton emptyRing))- expected = G.Polygon (G.PolygonRings emptyRing V.empty)- (query conn "SELECT ?" (Only geometry) :: IO [Only G.Geometry]) >>= (@?= [Only expected])- , testCase "invalid decoded construction fails without poisoning the connection" $- withConnection ":memory:" \conn -> do- let closedRing = G.CoordinatesXY (U.fromList [G.XY 0 0, G.XY 1 0, G.XY 1 1, G.XY 0 0])- openRing = G.CoordinatesXY (U.fromList [G.XY 0 0, G.XY 1 0, G.XY 1 1])- emptyRing = G.CoordinatesXY U.empty- forM_- [ G.LineString (G.CoordinatesXY (U.singleton (G.XY 1 2)))- , G.Polygon (G.PolygonRings openRing V.empty)- , G.Polygon (G.PolygonRings emptyRing (V.singleton closedRing))- ]- \shape ->- assertFailureIO (query conn "SELECT ?" (Only shape) :: IO [Only G.Geometry])- [Only raw] <- query_ conn "SELECT 'LINESTRING (1 2)'::GEOMETRY" :: IO [Only RawGeometry]- assertBool "raw result bytes remain available" (not (BS.null (rawGeometryWKB raw)))- assertFailureIO (query_ conn "SELECT 'LINESTRING (1 2)'::GEOMETRY" :: IO [Only G.Geometry])- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])- , testCase "CRS survives explicit binary import and table insertion" $- withConnection ":memory:" \conn -> do- [Only geometry] <- query_ conn "SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84')"- rawGeometryCRS geometry @?= Just "OGC:CRS84"- (importRaw conn geometry)- >>= (@?= [Only geometry])- _ <- execute_ conn "CREATE TABLE shapes (shape GEOMETRY('OGC:CRS84'))"- _ <- execute conn "INSERT INTO shapes VALUES (system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?))" (rawParameters geometry)- (query_ conn "SELECT shape FROM shapes" :: IO [Only RawGeometry]) >>= (@?= [Only geometry])- , testCase "CRS parameters preserve custom names and full definitions" $- withConnection ":memory:" \conn -> do- [Only geometry] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY"- forM_- [ "local' íslenska λ"- , "ENGCRS[\"Local grid\",EDATUM[\"Local datum\"],CS[Cartesian,2],AXIS[\"x\",east,ORDER[1],LENGTHUNIT[\"metre\",1]],AXIS[\"y\",north,ORDER[2],LENGTHUNIT[\"metre\",1]]]"- , "{\"type\":\"EngineeringCRS\",\"name\":\"Local grid\",\"datum\":{\"type\":\"EngineeringDatum\",\"name\":\"Local datum\"},\"coordinate_system\":{\"subtype\":\"Cartesian\",\"axis\":[{\"name\":\"x\",\"abbreviation\":\"x\",\"direction\":\"east\",\"unit\":\"metre\"},{\"name\":\"y\",\"abbreviation\":\"y\",\"direction\":\"north\",\"unit\":\"metre\"}]}}"- ]- \crs -> do- let annotated = geometry{rawGeometryCRS = Just crs}- (importRaw conn annotated) >>= (@?= [Only annotated])- , testCase "geometry construction ignores user macros with built-in names" $- withConnection ":memory:" \conn -> do- [Only raw] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]- decoded <- either assertFailure pure (fromRawGeometry raw)- _ <- execute_ conn "CREATE MACRO ST_AsText(x) AS 42"- _ <- execute_ conn "CREATE MACRO ST_GeomFromWKB(x) AS 'POINT (9 9)'::GEOMETRY"- _ <- execute_ conn "CREATE MACRO ST_SetCRS(x, crs) AS 42"- (query conn "SELECT system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?), ?" (rawGeometryWKB raw, rawGeometryCRS raw, decoded) :: IO [(RawGeometry, G.Geometry)])- >>= (@?= [(raw, decoded)])- , testCase "NULL and empty geometry stay distinct" $- withConnection ":memory:" \conn -> do- [Only empty] <- query_ conn "SELECT 'POINT EMPTY'::GEOMETRY" :: IO [Only RawGeometry]- (query conn "SELECT system.main.ST_GeomFromWKB(?), system.main.ST_GeomFromWKB(?)" (Nothing :: Maybe BS.ByteString, Just (rawGeometryWKB empty)) :: IO [(Maybe RawGeometry, Maybe RawGeometry)])- >>= (@?= [(Nothing, Just empty)])- , testCase "binary import constructs arrays with a common CRS and NULL elements" $- withConnection ":memory:" \conn -> do- [Only geometry] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]- let values = listArray (0, 2) [Nothing, Just geometry, Nothing]- (query conn "SELECT [NULL, system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?), NULL]::GEOMETRY('OGC:CRS84')[3]" (rawParameters geometry) :: IO [Only (Array Int (Maybe RawGeometry))]) >>= (@?= [Only values])- , testCase "nested LIST, ARRAY and MAP results preserve CRS and reject implicit raw binding" $- withConnection ":memory:" \conn -> do- [Only value] <-- query_- conn- "SELECT {'xs': ['POINT (1 2)'::GEOMETRY('OGC:CRS84'), NULL], 'fixed': ['POINT EMPTY'::GEOMETRY('OGC:CRS84')]::GEOMETRY('OGC:CRS84')[1], 'map': MAP {'one': 'POINT (3 4)'::GEOMETRY('OGC:CRS84')}}" ::- IO [Only (StructValue FieldValue)]- assertRawBindingFailure (query conn "SELECT ?" (Only value) :: IO [Only (StructValue FieldValue)])- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])- , testCase "UNION geometry results retain CRS and reject implicit raw binding" $- withConnection ":memory:" \conn -> do- [Only value] <- query_ conn "SELECT union_value(shape := 'POINT (1 2)'::GEOMETRY('OGC:CRS84'))" :: IO [Only (UnionValue FieldValue)]- assertRawBindingFailure (query conn "SELECT ?" (Only value) :: IO [Only (UnionValue FieldValue)])- [Only nullValue] <- query_ conn "SELECT union_value(shape := NULL::GEOMETRY('OGC:CRS84'))" :: IO [Only (UnionValue FieldValue)]- (query conn "SELECT ?" (Only nullValue) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only nullValue])- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])- , testCase "inactive UNION geometry members keep the CRS of the type cache" $- withConnection ":memory:" \conn -> do- [Only value] <- query_ conn "SELECT union_value(number := 42::BIGINT)::UNION(number BIGINT, shape GEOMETRY('OGC:CRS84'))" :: IO [Only (UnionValue FieldValue)]- (query conn "SELECT typeof(?)" (Only value) :: IO [Only Text]) >>= (@?= [Only "UNION(number BIGINT, shape GEOMETRY('OGC:CRS84'))"])- (query conn "SELECT ?" (Only value) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only value])- , testCase "NULL and empty nested geometry collections keep the CRS of the type cache" $- withConnection ":memory:" \conn -> do- _ <-- execute_- conn- "CREATE TABLE nested AS SELECT {'null_list': NULL::GEOMETRY('OGC:CRS84')[], 'empty_list': []::GEOMETRY('OGC:CRS84')[], 'null_array': NULL::GEOMETRY('OGC:CRS84')[2], 'array': [NULL, NULL]::GEOMETRY('OGC:CRS84')[2], 'null_map': NULL::MAP(VARCHAR, GEOMETRY('OGC:CRS84')), 'empty_map': map([], [])::MAP(VARCHAR, GEOMETRY('OGC:CRS84')), 'map': map(['one'], [NULL::GEOMETRY('OGC:CRS84')])} AS v"- [Only value] <- query_ conn "SELECT v FROM nested" :: IO [Only (StructValue FieldValue)]- (query conn "SELECT ?" (Only value) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only value])- _ <- execute conn "INSERT INTO nested VALUES (?)" (Only value)- (query_ conn "SELECT v FROM nested" :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only value, Only value])- , testCase "configured CRS definitions bind; other CRS definitions bind without a CRS" $ do- let crs = "local' íslenska λ"- withConnectionWithOptions ":memory:" defaultConnectionOptions{connectionGeometryCRS = [crs]} \conn -> do- [Only typed] <- query conn "SELECT typeof(?)" (Only (nullShape crs)) :: IO [Only Text]- assertBool (Text.unpack typed) ("íslenska λ" `Text.isInfixOf` typed)- (query conn "SELECT ?" (Only (nullShape crs)) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only (nullShape crs)])- (query conn "SELECT typeof(?)" (Only (nullShape "OGC:CRS84")) :: IO [Only Text]) >>= (@?= [Only "STRUCT(shape GEOMETRY)"])- , testCase "connection options validate CRS definitions and pass configuration flags" $ do- forM_ ["", "OGC:CRS84\0bad"] \crs ->- assertFailureIO (withConnectionWithOptions ":memory:" defaultConnectionOptions{connectionGeometryCRS = [crs]} (const (pure ())))- let options = defaultConnectionOptions{connectionConfig = [("threads", "1")], connectionGeometryCRS = ["local grid", "OGC:CRS84", "OGC:CRS84"]}- withConnectionWithOptions ":memory:" options \conn ->- (query_ conn "SELECT current_setting('threads')" :: IO [Only Int64]) >>= (@?= [Only 1])- , testCase "logical type construction drops CRS" $ do- let logical = LogicalTypeList (LogicalTypeGeometry (Just "OGC:CRS84"))- bracket (logicalTypeFromRep logical) destroyLogicalType logicalTypeToRep- >>= (@?= LogicalTypeList (LogicalTypeGeometry Nothing))- , testGroup- "folds cross chunk boundaries and preserve CRS"- [ testCase mode $- withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- [Only expected] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]- count <- foldRows conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84') FROM range(5000)" (0 :: Int64) \n (Only actual) -> do- actual @?= expected- pure (n + 1)- count @?= 5000- | (mode, foldRows) <- [("materialized", fold_), ("deprecated streaming", Streaming.fold_)]+ testGroup+ "geometry integration"+ [ testCase "Haskell shapes bind directly in every coordinate layout" $+ withConnection ":memory:" \conn -> do+ shapeRoundTrips conn G.DimXY G.PointXY G.CoordinatesXY (G.XY 1 2) (G.XY 3 4)+ shapeRoundTrips conn G.DimXYZ G.PointXYZ G.CoordinatesXYZ (G.XYZ 1 2 3) (G.XYZ 4 5 6)+ shapeRoundTrips conn G.DimXYM G.PointXYM G.CoordinatesXYM (G.XYM 1 2 3) (G.XYM 4 5 6)+ shapeRoundTrips conn G.DimXYZM G.PointXYZM G.CoordinatesXYZM (G.XYZM 1 2 3 4) (G.XYZM 5 6 7 8)+ , testCase "large coordinate buffers survive native round trips" $+ withConnection ":memory:" \conn -> do+ let coordinates = U.generate 10000 (\i -> G.XY (fromIntegral i) (fromIntegral (i * 2)))+ points = U.imap (\i coordinate -> if i `mod` 17 == 0 then G.EmptyPoint G.DimXY else G.PointXY coordinate) coordinates+ forM_ [G.LineString (G.CoordinatesXY coordinates), G.MultiPoint points] \shape -> do+ (query conn "SELECT ?" (Only shape)) >>= (@?= [Only shape])+ raw <- either assertFailure pure (toRawGeometry shape)+ (importRaw conn raw) >>= (@?= [Only raw])+ , testCase "WKT binding preserves exact finite coordinate bits" $+ withConnection ":memory:" \conn ->+ forM_+ [ 0+ , 0x8000000000000000+ , 1+ , 0x8000000000000001+ , 0x000fffffffffffff+ , 0x0010000000000000+ , 0x3fb999999999999b+ , 0x3ff0000000000001+ , 0x44b52d02c7e14af6+ , 0x7fefffffffffffff+ , 0xffefffffffffffff ]- , testCase "invalid binary import leaves the connection usable" $- withConnection ":memory:" \conn -> do- assertFailureIO (query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (BS.pack [1, 1, 0, 0, 0])) :: IO [Only RawGeometry])- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ \bits -> do+ let shape = G.PointGeometry (G.PointXY (G.XY (castWord64ToDouble bits) (-0.0)))+ expected <- either assertFailure pure (WKB.encodeWKB shape)+ [(actual, wkt)] <- query conn "SELECT ST_AsWKB(?), ST_AsText(?)" (shape, shape) :: IO [(BS.ByteString, Text)]+ actual @?= expected+ (WKT.decodeWKT wkt >>= WKB.encodeWKB) @?= Right expected+ (query conn "SELECT system.main.ST_AsWKB(system.main.ST_GeomFromWKB(?))" (Only expected) :: IO [Only BS.ByteString])+ >>= (@?= [Only expected])+ , testCase "pure WKT decoding agrees with native geometry results" $+ withConnection ":memory:" \conn ->+ forM_ shapes \wkt -> do+ [(native, rendered)] <- query conn "SELECT ?::GEOMETRY, ST_AsText(?::GEOMETRY)" (wkt, wkt) :: IO [(G.Geometry, Text)]+ WKT.decodeWKT wkt @?= Right native+ WKT.decodeWKT rendered @?= Right native+ , testCase "native bare multipoint EMPTY syntax parses as WKT" $+ withConnection ":memory:" \conn -> do+ [(native, bytes, rendered)] <-+ query_ conn "SELECT g, ST_AsWKB(g), ST_AsText(g) FROM (SELECT 'MULTIPOINT ((1 2), EMPTY, (3 4))'::GEOMETRY AS g)"+ :: IO [(G.Geometry, BS.ByteString, Text)]+ native @?= G.MultiPoint (U.fromList [G.PointXY (G.XY 1 2), G.EmptyPoint G.DimXY, G.PointXY (G.XY 3 4)])+ WKB.decodeWKB bytes @?= Right native+ WKT.decodeWKT rendered @?= Right native+ (WKT.encodeWKT native >>= WKT.decodeWKT) @?= Right native+ , testCase "standalone WKT syntax binds through native geometry parameters" $+ withConnection ":memory:" \conn ->+ forM_ ["point(+.5 -1.e+2)", "MULTIPOINT (EMPTY, (1 2), (3 4))", "MULTIPOINT (EMPTY, 1 2, EMPTY, 3 4, EMPTY)"] \wkt -> do+ shape <- either assertFailure pure (WKT.decodeWKT wkt)+ (query conn "SELECT ?" (Only shape)) >>= (@?= [Only shape])+ , testCase "native geometry depth limit produces a recoverable error" $+ withConnection ":memory:" \conn -> do+ let nested = iterate (G.GeometryCollection . V.singleton) (G.PointGeometry (G.PointXY (G.XY 1 2)))+ accepted = nested !! 15+ rejected = nested !! 16+ (query conn "SELECT ?" (Only accepted)) >>= (@?= [Only accepted])+ assertSqlErrorType ErrorInputInvalid (query conn "SELECT ?" (Only rejected) :: IO [Only G.Geometry])+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "decoded native shapes round trip through parameters and WKB" $+ withConnection ":memory:" \conn ->+ forM_ shapes \wkt -> do+ [Only geometry] <- query conn "SELECT ?::GEOMETRY" (Only wkt) :: IO [Only G.Geometry]+ (query conn "SELECT ?" (Only geometry) :: IO [Only G.Geometry]) >>= (@?= [Only geometry])+ raw <- either assertFailure pure (toRawGeometry geometry)+ fromRawGeometry raw @?= Right geometry+ , testCase "decoded rows retain point and sequence layouts, including empties" $+ withConnection ":memory:" \conn ->+ forM_+ [ ("POINT Z EMPTY", G.PointGeometry (G.EmptyPoint G.DimXYZ))+ , ("LINESTRING M EMPTY", G.LineString (G.CoordinatesXYM U.empty))+ , ("POINT Z (1 2 3)", G.PointGeometry (G.PointXYZ (G.XYZ 1 2 3)))+ , ("POINT EMPTY", G.PointGeometry (G.EmptyPoint G.DimXY))+ ]+ \(wkt, expected) ->+ (query conn "SELECT ?::GEOMETRY" (Only (wkt :: Text)) :: IO [Only G.Geometry])+ >>= (@?= [Only expected])+ , testCase "decoded arrays preserve shapes and NULL elements without CRS" $+ withConnection ":memory:" \conn -> do+ let geometry = G.PointGeometry (G.PointXYM (G.XYM 1 2 3))+ values = sized [Nothing, Just geometry, Just (G.PointGeometry (G.EmptyPoint G.DimXYM))] :: SizedArray 3 (Maybe G.Geometry)+ (query conn "SELECT ?, ST_CRS((?)[2])" (values, values) :: IO [(SizedArray 3 (Maybe G.Geometry), Maybe Text)])+ >>= (@?= [(values, Nothing)])+ , testCase "only raw geometry retains CRS metadata" $+ withConnection ":memory:" \conn -> do+ [(raw, shape)] <- query_ conn "SELECT g, g FROM (SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84') AS g)" :: IO [(RawGeometry, G.Geometry)]+ rawGeometryCRS raw @?= Just "OGC:CRS84"+ shape @?= G.PointGeometry (G.PointXYZM (G.XYZM 1 2 3 4))+ fromRawGeometry raw @?= Right shape+ unannotated <- either assertFailure pure (toRawGeometry shape)+ rawGeometryCRS unannotated @?= Nothing+ rawGeometryWKB unannotated @?= rawGeometryWKB raw+ ( query conn "SELECT system.main.ST_CRS(system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?))" (rawParameters raw)+ :: IO [Only (Maybe Text)]+ )+ >>= (@?= [Only (Just "OGC:CRS84")])+ (query conn "SELECT ST_CRS(?), ?" (shape, shape) :: IO [(Maybe Text, G.Geometry)])+ >>= (@?= [(Nothing, shape)])+ , testCase "decoded coordinates remain usable after closing their connection" $ do+ geometry <- withConnection ":memory:" \conn -> do+ [Only value] <- query_ conn "SELECT 'LINESTRING (1 2, 3 4, 5 6)'::GEOMETRY('OGC:CRS84')" :: IO [Only G.Geometry]+ pure value+ performMajorGC+ case geometry of+ G.LineString (G.CoordinatesXY coords) -> U.foldl' (\acc (G.XY x y) -> acc + x + y) 0 coords @?= 21+ other -> assertFailure ("unexpected geometry: " <> show other)+ , testCase "native WKB round trips through explicit binary import" $+ withConnection ":memory:" \conn ->+ forM_ shapes \wkt -> do+ [Only geometry] <- query conn "SELECT ?::GEOMETRY" (Only wkt)+ ( query conn "SELECT typeof(system.main.ST_GeomFromWKB(?)), system.main.ST_GeomFromWKB(?)" (rawGeometryWKB geometry, rawGeometryWKB geometry)+ :: IO [(Text, RawGeometry)]+ )+ >>= (@?= [("GEOMETRY", geometry)])+ , testCase "raw empty containers retain layouts that decoded values do not store" $+ withConnection ":memory:" \conn ->+ forM_+ ( [ (family <> " " <> dimensions <> " EMPTY", family <> " EMPTY")+ | family <- ["MULTIPOINT", "MULTILINESTRING", "MULTIPOLYGON", "GEOMETRYCOLLECTION"]+ , dimensions <- ["Z", "M", "ZM"]+ ]+ ++ [+ ( "GEOMETRYCOLLECTION ZM (MULTIPOINT ZM EMPTY, GEOMETRYCOLLECTION ZM EMPTY)"+ , "GEOMETRYCOLLECTION (MULTIPOINT EMPTY, GEOMETRYCOLLECTION EMPTY)"+ )+ ]+ )+ \(wkt, normalizedWKT) -> do+ [Only raw] <- query conn "SELECT ?::GEOMETRY('OGC:CRS84')" (Only (wkt :: Text)) :: IO [Only RawGeometry]+ (importRaw conn raw) >>= (@?= [Only raw])+ decoded <- either assertFailure pure (fromRawGeometry raw)+ WKT.encodeWKT decoded @?= Right normalizedWKT+ normalized <- either assertFailure pure (toRawGeometry decoded)+ rawGeometryCRS normalized @?= Nothing+ assertBool "decoded empty containers must not retain an unstored dimension tag" (rawGeometryWKB normalized /= rawGeometryWKB raw)+ , testCase "mixed-layout collections retain WKB but native import rejects them" $+ withConnection ":memory:" \conn -> do+ let geometry = G.GeometryCollection (V.fromList [G.PointGeometry (G.PointXY (G.XY 1 2)), G.PointGeometry (G.PointXYZ (G.XYZ 3 4 5))])+ raw <- either assertFailure pure (toRawGeometry geometry)+ fromRawGeometry raw @?= Right geometry+ (fromRawGeometry raw >>= toRawGeometry) @?= Right raw+ assertSqlErrorType ErrorInputInvalid (query conn "SELECT ?" (Only geometry) :: IO [Only G.Geometry])+ assertMixedLayoutFailure ErrorInputInvalid (importRaw conn raw)+ assertMixedLayoutFailure+ ErrorInputInvalid+ (query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (rawGeometryWKB raw)) :: IO [Only G.Geometry])+ assertMixedLayoutFailure ErrorTypeConversion (query conn "SELECT ?::GEOMETRY" (Only (rawGeometryWKB raw)) :: IO [Only RawGeometry])+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "mixed multipoint layouts promote missing ordinates to NaN" $+ withConnection ":memory:" \conn -> do+ let geometry = G.MultiPoint (U.fromList [G.PointXY (G.XY 1 2), G.PointXYZ (G.XYZ 3 4 5)])+ [Only actual] <- query conn "SELECT ?" (Only geometry) :: IO [Only G.Geometry]+ case actual of+ G.MultiPoint points -> do+ U.length points @?= 2+ case points U.! 0 of+ G.PointXYZ (G.XYZ x y z) -> do+ (x, y) @?= (1, 2)+ assertBool "promoted Z must be NaN" (isNaN z)+ other -> assertFailure ("unexpected promoted point: " <> show other)+ points U.! 1 @?= G.PointXYZ (G.XYZ 3 4 5)+ other -> assertFailure ("unexpected geometry: " <> show other)+ , testCase "raw NaN points retain ordinates that decoded emptiness normalizes" $+ withConnection ":memory:" \conn -> do+ [Only raw] <- query_ conn "SELECT 'POINT Z (NaN NaN 7)'::GEOMETRY" :: IO [Only RawGeometry]+ (importRaw conn raw) >>= (@?= [Only raw])+ decoded <- either assertFailure pure (fromRawGeometry raw)+ decoded @?= G.PointGeometry (G.EmptyPoint G.DimXYZ)+ normalized <- either assertFailure pure (toRawGeometry decoded)+ assertBool "decoding an XY-NaN point normalizes its extra ordinates" (rawGeometryWKB normalized /= rawGeometryWKB raw)+ (query conn "SELECT ?" (Only decoded) :: IO [Only G.Geometry]) >>= (@?= [Only decoded])+ , testCase "binary import preserves distinct NaN payload bits" $+ withConnection ":memory:" \conn -> do+ let shape = G.PointGeometry (G.PointXYZ (G.XYZ (castWord64ToDouble 0x7ff8000000000001) (castWord64ToDouble 0x7ff8000000000002) 7))+ raw <- either assertFailure pure (toRawGeometry shape)+ (importRaw conn raw) >>= (@?= [Only raw])+ fromRawGeometry raw @?= Right (G.PointGeometry (G.EmptyPoint G.DimXYZ))+ , testCase "mixed empty and nonempty layouts retain WKB but native import rejects them" $+ withConnection ":memory:" \conn -> do+ let shape = G.GeometryCollection (V.fromList [G.PointGeometry (G.EmptyPoint G.DimXY), G.PointGeometry (G.PointXYZ (G.XYZ 1 2 3))])+ raw <- either assertFailure pure (toRawGeometry shape)+ fromRawGeometry raw @?= Right shape+ (fromRawGeometry raw >>= toRawGeometry) @?= Right raw+ assertMixedLayoutFailure ErrorInputInvalid (importRaw conn raw)+ assertMixedLayoutFailure+ ErrorInputInvalid+ (query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (rawGeometryWKB raw)) :: IO [Only G.Geometry])+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "decoded polygon writing normalizes all-empty rings" $+ withConnection ":memory:" \conn -> do+ let emptyRing = G.CoordinatesXYZ U.empty+ geometry = G.Polygon (G.PolygonRings emptyRing (V.singleton emptyRing))+ expected = G.Polygon (G.PolygonRings emptyRing V.empty)+ (query conn "SELECT ?" (Only geometry) :: IO [Only G.Geometry]) >>= (@?= [Only expected])+ , testCase "invalid decoded construction fails without poisoning the connection" $+ withConnection ":memory:" \conn -> do+ let closedRing = G.CoordinatesXY (U.fromList [G.XY 0 0, G.XY 1 0, G.XY 1 1, G.XY 0 0])+ openRing = G.CoordinatesXY (U.fromList [G.XY 0 0, G.XY 1 0, G.XY 1 1])+ emptyRing = G.CoordinatesXY U.empty+ forM_+ [ G.LineString (G.CoordinatesXY (U.singleton (G.XY 1 2)))+ , G.Polygon (G.PolygonRings openRing V.empty)+ , G.Polygon (G.PolygonRings emptyRing (V.singleton closedRing))+ ]+ \shape ->+ assertFailureIO (query conn "SELECT ?" (Only shape) :: IO [Only G.Geometry])+ [Only raw] <- query_ conn "SELECT 'LINESTRING (1 2)'::GEOMETRY" :: IO [Only RawGeometry]+ assertBool "raw result bytes remain available" (not (BS.null (rawGeometryWKB raw)))+ assertFailureIO (query_ conn "SELECT 'LINESTRING (1 2)'::GEOMETRY" :: IO [Only G.Geometry])+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "CRS survives explicit binary import and table insertion" $+ withConnection ":memory:" \conn -> do+ [Only geometry] <- query_ conn "SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84')"+ rawGeometryCRS geometry @?= Just "OGC:CRS84"+ (importRaw conn geometry)+ >>= (@?= [Only geometry])+ _ <- execute_ conn "CREATE TABLE shapes (shape GEOMETRY('OGC:CRS84'))"+ _ <- execute conn "INSERT INTO shapes VALUES (system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?))" (rawParameters geometry)+ (query_ conn "SELECT shape FROM shapes" :: IO [Only RawGeometry]) >>= (@?= [Only geometry])+ , testCase "CRS parameters preserve custom names and full definitions" $+ withConnection ":memory:" \conn -> do+ [Only geometry] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY"+ forM_+ [ "local' íslenska λ"+ , "ENGCRS[\"Local grid\",EDATUM[\"Local datum\"],CS[Cartesian,2],AXIS[\"x\",east,ORDER[1],LENGTHUNIT[\"metre\",1]],AXIS[\"y\",north,ORDER[2],LENGTHUNIT[\"metre\",1]]]"+ , "{\"type\":\"EngineeringCRS\",\"name\":\"Local grid\",\"datum\":{\"type\":\"EngineeringDatum\",\"name\":\"Local datum\"},\"coordinate_system\":{\"subtype\":\"Cartesian\",\"axis\":[{\"name\":\"x\",\"abbreviation\":\"x\",\"direction\":\"east\",\"unit\":\"metre\"},{\"name\":\"y\",\"abbreviation\":\"y\",\"direction\":\"north\",\"unit\":\"metre\"}]}}"+ ]+ \crs -> do+ let annotated = geometry{rawGeometryCRS = Just crs}+ (importRaw conn annotated) >>= (@?= [Only annotated])+ , testCase "geometry construction ignores user macros with built-in names" $+ withConnection ":memory:" \conn -> do+ [Only raw] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ decoded <- either assertFailure pure (fromRawGeometry raw)+ _ <- execute_ conn "CREATE MACRO ST_AsText(x) AS 42"+ _ <- execute_ conn "CREATE MACRO ST_GeomFromWKB(x) AS 'POINT (9 9)'::GEOMETRY"+ _ <- execute_ conn "CREATE MACRO ST_SetCRS(x, crs) AS 42"+ ( query conn "SELECT system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?), ?" (rawGeometryWKB raw, rawGeometryCRS raw, decoded)+ :: IO [(RawGeometry, G.Geometry)]+ )+ >>= (@?= [(raw, decoded)])+ , testCase "NULL and empty geometry stay distinct" $+ withConnection ":memory:" \conn -> do+ [Only empty] <- query_ conn "SELECT 'POINT EMPTY'::GEOMETRY" :: IO [Only RawGeometry]+ ( query+ conn+ "SELECT system.main.ST_GeomFromWKB(?), system.main.ST_GeomFromWKB(?)"+ (Nothing :: Maybe BS.ByteString, Just (rawGeometryWKB empty))+ :: IO [(Maybe RawGeometry, Maybe RawGeometry)]+ )+ >>= (@?= [(Nothing, Just empty)])+ , testCase "binary import constructs arrays with a common CRS and NULL elements" $+ withConnection ":memory:" \conn -> do+ [Only geometry] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ let values = sized [Nothing, Just geometry, Nothing] :: SizedArray 3 (Maybe RawGeometry)+ ( query conn "SELECT [NULL, system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?), NULL]::GEOMETRY('OGC:CRS84')[3]" (rawParameters geometry)+ :: IO [Only (SizedArray 3 (Maybe RawGeometry))]+ )+ >>= (@?= [Only values])+ , testCase "nested STRUCT, LIST, ARRAY and MAP results bind again with their bytes and CRS" $+ withConnection ":memory:" \conn -> do+ wkbs <- shapeBytes conn+ forM_ [("GEOMETRY", Nothing), ("GEOMETRY('OGC:CRS84')", Just "OGC:CRS84")] \(geometryType, crs) ->+ forM_ wkbs \wkb -> do+ [(value, valueType)] <-+ query+ conn+ ( Query+ ( "SELECT v, typeof(v) FROM (SELECT {'shape': g, 'xs': [g, NULL], 'fixed': [g]::"+ <> geometryType+ <> "[1], 'map': MAP {'one': g}} AS v FROM (SELECT system.main.ST_GeomFromWKB(?)::"+ <> geometryType+ <> " AS g))"+ )+ )+ (Only wkb)+ :: IO [(StructValue FieldValue, Text)]+ take 1 (elems (structValueFields value)) @?= [StructField "shape" (FieldGeometry (RawGeometry wkb crs))]+ (query conn "SELECT ?, typeof(?)" (value, value)) >>= (@?= [(value, valueType)])+ , testCase "UNION geometry results retain CRS and bind again" $+ withConnection ":memory:" \conn -> do+ [(value, valueType)] <-+ query_ conn "SELECT u, typeof(u) FROM (SELECT union_value(shape := 'POINT (1 2)'::GEOMETRY('OGC:CRS84')) AS u)"+ :: IO [(UnionValue FieldValue, Text)]+ (query conn "SELECT ?, typeof(?)" (value, value)) >>= (@?= [(value, valueType)])+ [Only nullValue] <- query_ conn "SELECT union_value(shape := NULL::GEOMETRY('OGC:CRS84'))" :: IO [Only (UnionValue FieldValue)]+ (query conn "SELECT ?" (Only nullValue) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only nullValue])+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "a raw geometry binds only as a GEOMETRY with the same CRS" $+ withConnection ":memory:" \conn -> do+ [Only raw] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ let plain = raw{rawGeometryCRS = Nothing}+ fromCRS = "duckdb-simple: a GEOMETRY value with the CRS 'OGC:CRS84' cannot bind as a GEOMETRY with no CRS"+ toCRS = "duckdb-simple: a GEOMETRY value with no CRS cannot bind as a GEOMETRY with the CRS 'OGC:CRS84'"+ forM_+ [ ("GEOMETRY", LogicalTypeGeometry Nothing, raw, fromCRS)+ , ("GEOMETRY", LogicalTypeScalar LogicalTypeIdGeometry, raw, fromCRS)+ , ("GEOMETRY('OGC:CRS84')", LogicalTypeGeometry (Just "OGC:CRS84"), plain, toCRS)+ ]+ \(columnType, declared, value, message) -> do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v STRUCT(shape " <> columnType <> "))"))+ result <- try (execute conn "INSERT INTO stored VALUES (?)" (Only (geometryStruct declared (FieldGeometry value))))+ case result of+ Left err -> do+ sqlErrorMessage err @?= message+ sqlErrorType err @?= Just ErrorInputInvalid+ Right _ -> assertFailure "expected a CRS mismatch error"+ (query_ conn "SELECT count(*) FROM stored" :: IO [Only Int64]) >>= (@?= [Only 0])+ (query conn "SELECT ?" (Only (geometryStruct (LogicalTypeScalar LogicalTypeIdGeometry) (FieldGeometry plain))))+ >>= (@?= [Only (geometryStruct (LogicalTypeGeometry Nothing) (FieldGeometry plain))])+ , testCase "a raw geometry binds as a top-level parameter with its bytes and CRS" $+ withConnection ":memory:" \conn -> do+ wkbs <- shapeBytes conn+ forM_ [("GEOMETRY", Nothing), ("GEOMETRY('OGC:CRS84')", Just "OGC:CRS84")] \(geometryType, crs) -> do+ let values = [RawGeometry wkb crs | wkb <- wkbs]+ forM_ values \value ->+ (query conn "SELECT ?, typeof(?)" (value, value) :: IO [(RawGeometry, Text)]) >>= (@?= [(value, geometryType)])+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (i BIGINT, v " <> geometryType <> ")"))+ _ <- executeMany conn "INSERT INTO stored VALUES (?, ?)" (zip [0 :: Int64 ..] values)+ (query_ conn "SELECT v FROM stored ORDER BY i" :: IO [Only RawGeometry]) >>= (@?= map Only values)+ , testCase "a top-level raw geometry follows the CRS casts of DuckDB" $+ withConnection ":memory:" \conn -> do+ [Only raw] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ let plain = raw{rawGeometryCRS = Nothing}+ other = raw{rawGeometryCRS = Just "OGC:CRS83"}+ mismatch = "Binder Error: Cannot cast GEOMETRY with CRS 'OGC:CRS83' to GEOMETRY with different CRS 'OGC:CRS84'"+ store column value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ _ <- execute conn "INSERT INTO stored VALUES (?)" (Only value)+ query_ conn "SELECT v FROM stored" :: IO [Only RawGeometry]+ -- A value without a CRS takes the CRS of its column.+ store "GEOMETRY('OGC:CRS84')" plain >>= (@?= [Only raw])+ -- A column without a CRS removes the CRS of the value.+ store "GEOMETRY" raw >>= (@?= [Only plain])+ store "GEOMETRY" other >>= (@?= [Only plain])+ -- DuckDB does not cast a GEOMETRY to a GEOMETRY with another CRS.+ _ <- execute_ conn "CREATE OR REPLACE TABLE stored (v GEOMETRY('OGC:CRS84'))"+ inserted <- try (execute conn "INSERT INTO stored VALUES (?)" (Only other))+ case inserted of+ Left err -> do+ sqlErrorMessage err @?= mismatch+ sqlErrorType err @?= Just ErrorQueryBinder+ Right _ -> assertFailure "expected the CRS cast error of DuckDB"+ (query_ conn "SELECT count(*) FROM stored" :: IO [Only Int64]) >>= (@?= [Only 0])+ -- A comparison follows the same casts.+ _ <- execute conn "INSERT INTO stored VALUES (?)" (Only raw)+ (query conn "SELECT count(*) FROM stored WHERE v = ?" (Only plain) :: IO [Only Int64]) >>= (@?= [Only 1])+ compared <- try (query conn "SELECT count(*) FROM stored WHERE v = ?" (Only other) :: IO [Only Int64])+ case compared of+ Left err -> do+ sqlErrorMessage err @?= mismatch+ sqlErrorType err @?= Just ErrorQueryBinder+ Right rows -> assertFailure ("expected the CRS cast error of DuckDB, got " <> show rows)+ , testCase "a Maybe RawGeometry parameter binds NULL and a value" $+ withConnection ":memory:" \conn -> do+ [Only raw] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ _ <- execute_ conn "CREATE TABLE stored (i BIGINT, v GEOMETRY('OGC:CRS84'))"+ _ <- executeMany conn "INSERT INTO stored VALUES (?, ?)" [(0 :: Int64, Just raw), (1, Nothing)]+ (query_ conn "SELECT v FROM stored ORDER BY i" :: IO [Only (Maybe RawGeometry)]) >>= (@?= [Only (Just raw), Only Nothing])+ (query conn "SELECT ? IS NULL" (Only (Nothing :: Maybe RawGeometry)) :: IO [Only Bool]) >>= (@?= [Only True])+ , testCase "inactive UNION geometry members keep the CRS of their declared type" $+ withConnection ":memory:" \conn -> do+ [Only value] <-+ query_ conn "SELECT union_value(number := 42::BIGINT)::UNION(number BIGINT, shape GEOMETRY('OGC:CRS84'))"+ :: IO [Only (UnionValue FieldValue)]+ (query conn "SELECT typeof(?)" (Only value) :: IO [Only Text]) >>= (@?= [Only "UNION(number BIGINT, shape GEOMETRY('OGC:CRS84'))"])+ (query conn "SELECT ?" (Only value) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only value])+ , testCase "NULL and empty nested geometry collections keep the CRS of their declared type" $+ withConnection ":memory:" \conn -> do+ _ <-+ execute_+ conn+ "CREATE TABLE nested AS SELECT {'null_list': NULL::GEOMETRY('OGC:CRS84')[], 'empty_list': []::GEOMETRY('OGC:CRS84')[], 'null_array': NULL::GEOMETRY('OGC:CRS84')[2], 'array': [NULL, NULL]::GEOMETRY('OGC:CRS84')[2], 'null_map': NULL::MAP(VARCHAR, GEOMETRY('OGC:CRS84')), 'empty_map': map([], [])::MAP(VARCHAR, GEOMETRY('OGC:CRS84')), 'map': map(['one'], [NULL::GEOMETRY('OGC:CRS84')])} AS v"+ [Only value] <- query_ conn "SELECT v FROM nested" :: IO [Only (StructValue FieldValue)]+ (query conn "SELECT ?" (Only value) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only value])+ _ <- execute conn "INSERT INTO nested VALUES (?)" (Only value)+ (query_ conn "SELECT v FROM nested" :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only value, Only value])+ , testCase "custom CRS and recognized definitions retain metadata with extension autoload disabled" $ do+ let crs = "local' íslenska λ"+ withConnectionWithConfig ":memory:" [("autoload_known_extensions", "false")] \conn -> do+ [Only point] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY" :: IO [Only RawGeometry]+ let annotated = point{rawGeometryCRS = Just crs}+ (query conn "SELECT ?" (Only annotated)) >>= (@?= [Only annotated])+ [Only typed] <- query conn "SELECT typeof(?)" (Only (nullShape crs)) :: IO [Only Text]+ assertBool (Text.unpack typed) ("íslenska λ" `Text.isInfixOf` typed)+ (query conn "SELECT ?" (Only (nullShape crs))) >>= (@?= [Only (nullShape crs)])+ (query conn "SELECT typeof(?)" (Only (nullShape "OGC:CRS84")) :: IO [Only Text]) >>= (@?= [Only "STRUCT(shape GEOMETRY('OGC:CRS84'))"])+ (query conn "SELECT ?" (Only (nullShape "OGC:CRS84"))) >>= (@?= [Only (nullShape "OGC:CRS84")])+ , testCase "custom CRS survives nested parameter types and prepared rebinding" $ do+ let crs = "local' íslenska λ"+ value = nestedNullShapes crs+ array = sized [value] :: SizedArray 1 (StructValue FieldValue)+ withConnection ":memory:" \conn -> do+ (query conn "SELECT ?" (Only value)) >>= (@?= [Only value])+ (queryNamed conn "SELECT $shape" ["shape" := value]) >>= (@?= [Only value])+ (query conn "SELECT ?" (Only array)) >>= (@?= [Only array])+ withStatement conn "SELECT ?" \stmt -> do+ bind stmt [toField value]+ nextRow stmt >>= (@?= Just (Only value))+ bind stmt [toField value]+ nextRow stmt >>= (@?= Just (Only value))+ , testCase "a custom CRS bind leaves an active caller stream usable" $ do+ let crs = "local' íslenska λ"+ withConnection ":memory:" \conn ->+ withStatement conn "SELECT ?" \parameter -> do+ withStatement conn "SELECT i::BIGINT FROM range(3) t(i)" \stmt -> do+ Streaming.nextRow stmt >>= (@?= Just (Only (0 :: Int64)))+ bind parameter [toField (nullShape crs)]+ performMajorGC+ Streaming.nextRow stmt >>= (@?= Just (Only (1 :: Int64)))+ nextRow parameter >>= (@?= Just (Only (nullShape crs)))+ , testCase "ignore_unknown_crs cannot silently drop parameter metadata" $+ withConnectionWithConfig ":memory:" [("ignore_unknown_crs", "true")] \conn -> do+ let value = nullShape "local grid"+ (query conn "SELECT ?" (Only value)) >>= (@?= [Only value])+ , testCase "invalid CRS definitions fail during binding" $+ withConnection ":memory:" \conn ->+ forM_ ["", "OGC:CRS84\0bad", " "] \crs -> do+ result <- try (query conn "SELECT ?" (Only (nullShape crs))) :: IO (Either SQLError [Only (StructValue FieldValue)])+ case result of+ Left err -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ sqlErrorMessage err @?= "duckdb-simple: a GEOMETRY CRS must be nonblank and must not contain NUL"+ sqlErrorBody err @?= Nothing+ Right _ -> assertFailure ("expected binding to reject " <> show crs)+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "connection options pass configuration flags" $+ withConnectionWithOptions ":memory:" defaultConnectionOptions{connectionConfig = [("threads", "1")]} \conn ->+ (query_ conn "SELECT current_setting('threads')" :: IO [Only Int64]) >>= (@?= [Only 1])+ , testCase "logical type construction preserves CRS" $+ withConnection ":memory:" \conn -> withConnectionHandle conn \native -> withConnectionFactory native \factory -> do+ let logical = LogicalTypeList (LogicalTypeGeometry (Just "OGC:CRS84"))+ bracket (logicalTypeFromRep factory logical) destroyLogicalType (logicalTypeToRep factory) >>= (@?= logical)+ , testGroup+ "folds cross chunk boundaries and preserve CRS"+ [ testCase mode $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ [Only expected] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ count <- foldRows conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84') FROM range(5000)" (0 :: Int64) \n (Only actual) -> do+ actual @?= expected+ pure (n + 1)+ count @?= 5000+ | (mode, foldRows) <- [("materialized", fold_), ("native streaming", Streaming.fold_)] ]+ , testCase "invalid binary import leaves the connection usable" $+ withConnection ":memory:" \conn -> do+ assertSqlErrorType+ ErrorInputInvalid+ (query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (BS.pack [1, 1, 0, 0, 0])) :: IO [Only RawGeometry])+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "recognized, custom, and complete CRS names bind with their CRS" $ do+ let names = ["OGC:CRS84", "local' íslenska λ", localGridWKT2]+ withConnection ":memory:" \conn ->+ forM_ names \crs -> (query conn "SELECT ?" (Only (nullShape crs))) >>= (@?= [Only (nullShape crs)])+ , testCase "a default connection runs no CRS query, also for a VARIANT read" $+ withLoggedInstance \inst countQueries -> withConnectionOn inst \conn -> do+ countQueries crsQuery >>= (@?= 0)+ countQueries "variant_typeof" >>= (@?= 0)+ (query conn "SELECT ?" (Only (nullShape "OGC:CRS84"))) >>= (@?= [Only (nullShape "OGC:CRS84")])+ (query_ conn "SELECT 42::VARIANT" :: IO [Only Variant]) >>= (@?= [Only (Variant (FieldInt32 42))])+ countQueries crsQuery >>= (@?= 0)+ -- The log records metadata preparation and execution on the helper connection.+ countQueries "variant_typeof" >>= (@?= 2)+ , testCase "a custom CRS resolves at bind time and a VARIANT read runs no CRS query" $+ withLoggedInstance \inst countQueries -> withConnectionOn inst \conn -> do+ countQueries crsQuery >>= (@?= 0)+ let value = nullShape "local grid"+ (query conn "SELECT ?" (Only value)) >>= (@?= [Only value])+ resolved <- countQueries crsQuery+ assertBool "the custom CRS uses the fallback query" (resolved > 0)+ (query_ conn "SELECT 42::VARIANT" :: IO [Only Variant]) >>= (@?= [Only (Variant (FieldInt32 42))])+ countQueries crsQuery >>= (@?= resolved)+ , testCase "a recognized CRS binds in either spelling" $+ withConnection ":memory:" \conn ->+ forM_ ["ogc:crs84", "OGC:CRS84"] \crs -> (query conn "SELECT ?" (Only (nullShape crs))) >>= (@?= [Only (nullShape "OGC:CRS84")])+ ]+ where+ crsQuery = "system.main.ST_SetCRS(" +-- | Run an action on an instance that logs each query. The second argument+-- counts the logged queries whose text contains a part. The count query binds+-- the part as a value, so its own text does not match.+withLoggedInstance :: (Instance -> (Text -> IO Int64) -> IO a) -> IO a+withLoggedInstance action = withEnvironment \env -> withInstance env [] \inst -> withConnectionOn inst \logger -> do+ _ <- execute_ logger "CALL enable_logging(level := 'info')"+ let countQueries part = do+ rows <- query logger "SELECT count(*) FROM duckdb_logs() WHERE type = 'QueryLog' AND contains(message, ?)" (Only part)+ case rows of+ [Only count] -> pure count+ _ -> assertFailure "expected one count"+ action inst countQueries++-- | A complete WKT2 definition that no DuckDB catalog entry identifies.+localGridWKT2 :: Text+localGridWKT2 =+ "ENGCRS[\"Local grid\",EDATUM[\"Local datum\"],CS[Cartesian,2],AXIS[\"x\",east,ORDER[1],LENGTHUNIT[\"metre\",1]],AXIS[\"y\",north,ORDER[2],LENGTHUNIT[\"metre\",1]]]"+ -- | A one-field STRUCT with a NULL GEOMETRY of the given CRS. nullShape :: Text -> StructValue FieldValue-nullShape crs =- StructValue- { structValueFields = listArray (0, 0) [StructField "shape" FieldNull]- , structValueTypes = listArray (0, 0) [StructField "shape" (LogicalTypeGeometry (Just crs))]- , structValueIndex = Map.singleton "shape" 0- }+nullShape crs = geometryStruct (LogicalTypeGeometry (Just crs)) FieldNull +-- | A one-field STRUCT with a value of the given declared GEOMETRY type.+geometryStruct :: LogicalTypeRep -> FieldValue -> StructValue FieldValue+geometryStruct logical value =+ StructValue+ { structValueFields = listArray (0, 0) [StructField "shape" value]+ , structValueTypes = listArray (0, 0) [StructField "shape" logical]+ , structValueIndex = Map.singleton "shape" 0+ }++-- | Retain custom CRS metadata in NULL, empty, and inactive nested values.+nestedNullShapes :: Text -> StructValue FieldValue+nestedNullShapes crs =+ StructValue+ { structValueFields = listArray (0, length fields - 1) [StructField name value | (name, value, _) <- fields]+ , structValueTypes = listArray (0, length fields - 1) [StructField name logical | (name, _, logical) <- fields]+ , structValueIndex = Map.fromList (zip [name | (name, _, _) <- fields] [0 ..])+ }+ where+ geometry = LogicalTypeGeometry (Just crs)+ list = LogicalTypeList geometry+ array = LogicalTypeArray geometry 2+ mapping = LogicalTypeMap (LogicalTypeScalar LogicalTypeIdVarchar) geometry+ members = listArray (0, 1) [UnionMemberType "number" (LogicalTypeScalar LogicalTypeIdBigInt), UnionMemberType "shape" geometry]+ inactive = UnionValue 0 "number" (FieldInt64 42) members+ shape = nullShape crs+ fields =+ [ ("null_list", FieldNull, list)+ , ("empty_list", FieldList [], list)+ , ("null_array", FieldNull, array)+ , ("array", FieldArray (listArray (0, 1) [FieldNull, FieldNull]), array)+ , ("null_map", FieldNull, mapping)+ , ("empty_map", FieldMap [], mapping)+ , ("map", FieldMap [(FieldText "one", FieldNull)], mapping)+ , ("inactive", FieldUnion inactive, LogicalTypeUnion members)+ , ("nested", FieldStruct shape, LogicalTypeStruct (structValueTypes shape))+ ]+ -- | Exercise all coordinate layouts with explicit point and sequence constructors.-shapeRoundTrips :: (G.Coordinate coord) => Connection -> G.Dimensions -> (coord -> G.Point) -> (U.Vector coord -> G.Coordinates) -> coord -> coord -> Assertion+shapeRoundTrips+ :: (G.Coordinate coord) => Connection -> G.Dimensions -> (coord -> G.Point) -> (U.Vector coord -> G.Coordinates) -> coord -> coord -> Assertion shapeRoundTrips conn dimensions point coordinates first second = do- let line = coordinates (U.fromList [first, second])- ring = coordinates (U.fromList [first, second, first, first])- emptyLine = coordinates U.empty- emptyPoint = G.EmptyPoint dimensions- polygon = G.PolygonRings ring V.empty- emptyPolygon = G.PolygonRings emptyLine V.empty- values =- [ G.PointGeometry (point first)- , G.PointGeometry emptyPoint- , G.LineString line- , G.LineString emptyLine- , G.Polygon polygon- , G.Polygon emptyPolygon- , G.MultiPoint (U.fromList [point first, emptyPoint, point second])- , G.MultiPoint U.empty- , G.MultiLineString (V.fromList [line, emptyLine])- , G.MultiPolygon (V.fromList [polygon, emptyPolygon])- , G.GeometryCollection (V.fromList [G.PointGeometry emptyPoint, G.LineString line, G.GeometryCollection V.empty])- ]- forM_ values \shape -> do- rows <- query conn "SELECT ?, ST_CRS(?)" (shape, shape)- rows @?= [(shape, Nothing :: Maybe Text)]- raw <- either assertFailure pure (toRawGeometry shape)- fromRawGeometry raw @?= Right shape- let annotated = raw{rawGeometryCRS = Just "local' íslenska λ"}- (importRaw conn annotated)- >>= (@?= [Only annotated])+ let line = coordinates (U.fromList [first, second])+ ring = coordinates (U.fromList [first, second, first, first])+ emptyLine = coordinates U.empty+ emptyPoint = G.EmptyPoint dimensions+ polygon = G.PolygonRings ring V.empty+ emptyPolygon = G.PolygonRings emptyLine V.empty+ values =+ [ G.PointGeometry (point first)+ , G.PointGeometry emptyPoint+ , G.LineString line+ , G.LineString emptyLine+ , G.Polygon polygon+ , G.Polygon emptyPolygon+ , G.MultiPoint (U.fromList [point first, emptyPoint, point second])+ , G.MultiPoint U.empty+ , G.MultiLineString (V.fromList [line, emptyLine])+ , G.MultiPolygon (V.fromList [polygon, emptyPolygon])+ , G.GeometryCollection (V.fromList [G.PointGeometry emptyPoint, G.LineString line, G.GeometryCollection V.empty])+ ]+ forM_ values \shape -> do+ rows <- query conn "SELECT ?, ST_CRS(?)" (shape, shape)+ rows @?= [(shape, Nothing :: Maybe Text)]+ raw <- either assertFailure pure (toRawGeometry shape)+ fromRawGeometry raw @?= Right shape+ let annotated = raw{rawGeometryCRS = Just "local' íslenska λ"}+ (importRaw conn annotated)+ >>= (@?= [Only annotated]) -- | Import WKB without converting a missing CRS to SQL NULL. importRaw :: Connection -> RawGeometry -> IO [Only RawGeometry] importRaw conn raw = case rawGeometryCRS raw of- Nothing -> query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (rawGeometryWKB raw))- Just crs -> query conn "SELECT system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?)" (rawGeometryWKB raw, crs)+ Nothing -> query conn "SELECT system.main.ST_GeomFromWKB(?)" (Only (rawGeometryWKB raw))+ Just crs -> query conn "SELECT system.main.ST_SetCRS(system.main.ST_GeomFromWKB(?), ?)" (rawGeometryWKB raw, crs) -- | Bind raw bytes and CRS as separate SQL parameters. rawParameters :: RawGeometry -> (BS.ByteString, Maybe Text) rawParameters raw = (rawGeometryWKB raw, rawGeometryCRS raw) --- | Check that generic values cannot silently normalize raw geometry.-assertRawBindingFailure :: IO a -> Assertion-assertRawBindingFailure action = do- result <- try (action >> pure ()) :: IO (Either SomeException ())- case result of- Left err -> assertBool (displayException err) ("raw GEOMETRY binding requires explicit" `isInfixOf` displayException err)- Right () -> assertFailure "expected raw geometry binding rejection"+-- | Check the native rejection without accepting unrelated failures.+assertMixedLayoutFailure :: ErrorType -> IO a -> Assertion+assertMixedLayoutFailure expectedCode action = do+ result <- try (action >> pure ()) :: IO (Either SQLError ())+ case result of+ Left err -> do+ sqlErrorType err @?= Just expectedCode+ assertBool (show err) ("inconsistent Z/M dimensions" `Text.isInfixOf` sqlErrorMessage err)+ Right () -> assertFailure "expected native mixed-layout geometry rejection" +-- | The WKB bytes of each native shape and of a point with NaN payload bits.+shapeBytes :: Connection -> IO [BS.ByteString]+shapeBytes conn = do+ native <- forM shapes \wkt -> do+ [Only raw] <- query conn "SELECT ?::GEOMETRY" (Only wkt) :: IO [Only RawGeometry]+ pure (rawGeometryWKB raw)+ nanPoint <-+ either+ assertFailure+ pure+ (WKB.encodeWKB (G.PointGeometry (G.PointXYZ (G.XYZ (castWord64ToDouble 0x7ff8000000000001) (castWord64ToDouble 0x7ff8000000000002) 7))))+ pure (native <> [nanPoint])+ -- | Native shapes cover all families, dimensions, empty children, and exact doubles. shapes :: [Text] shapes =- [ "POINT (0.10000000000000002 -0.0)"- , "POINT Z (1 2 3)"- , "POINT M (1 2 3)"- , "POINT ZM (1 2 3 4)"- , "POINT EMPTY"- , "LINESTRING EMPTY"- , "LINESTRING (1 2, 3 4)"- , "POLYGON ((0 0, 1 0, 1 1, 0 0))"- , "POLYGON ((0 0, 10 0, 10 10, 0 10, 0 0), (2 2, 2 4, 4 4, 4 2, 2 2))"- , "MULTIPOINT ((1 2), EMPTY, (3 4))"- , "MULTILINESTRING ((1 2, 3 4), EMPTY)"- , "MULTIPOLYGON (((0 0, 1 0, 1 1, 0 0)), EMPTY)"- , "GEOMETRYCOLLECTION (POINT (1 2), LINESTRING (0 0, 1 1))"- , "GEOMETRYCOLLECTION ZM (POINT ZM (1 2 3 4), POINT ZM EMPTY)"- , "POINT Z EMPTY"- , "LINESTRING M EMPTY"- , "GEOMETRYCOLLECTION ZM EMPTY"- ]+ [ "POINT (0.10000000000000002 -0.0)"+ , "POINT Z (1 2 3)"+ , "POINT M (1 2 3)"+ , "POINT ZM (1 2 3 4)"+ , "POINT EMPTY"+ , "LINESTRING EMPTY"+ , "LINESTRING (1 2, 3 4)"+ , "POLYGON ((0 0, 1 0, 1 1, 0 0))"+ , "POLYGON ((0 0, 10 0, 10 10, 0 10, 0 0), (2 2, 2 4, 4 4, 4 2, 2 2))"+ , "MULTIPOINT ((1 2), EMPTY, (3 4))"+ , "MULTILINESTRING ((1 2, 3 4), EMPTY)"+ , "MULTIPOLYGON (((0 0, 1 0, 1 1, 0 0)), EMPTY)"+ , "GEOMETRYCOLLECTION (POINT (1 2), LINESTRING (0 0, 1 1))"+ , "GEOMETRYCOLLECTION ZM (POINT ZM (1 2 3 4), POINT ZM EMPTY)"+ , "POINT Z EMPTY"+ , "LINESTRING M EMPTY"+ , "GEOMETRYCOLLECTION ZM EMPTY"+ ]
test/Properties.hs view
@@ -1,185 +1,219 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-} {-# LANGUAGE ExistentialQuantification #-} {-# LANGUAGE LambdaCase #-} {-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE PatternSynonyms #-} {-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-} +-- |+-- Module : Properties+-- Description : Round-trip properties for parameters and results of each column type. module Properties (roundTripTests) where import Control.Monad (forM)-import Data.Array (Array, elems, listArray)+import Data.Array (elems, listArray)+import Data.Bits (clearBit) import qualified Data.ByteString as BS import Data.Fixed (Fixed (MkFixed)) import Data.Int (Int16, Int32, Int64, Int8) import qualified Data.Map.Strict as Map import Data.Proxy (Proxy (..))+import Data.Ratio ((%)) import qualified Data.Text as Text-import Data.Time.Calendar (Day (..))+import Data.Time.Calendar (Day (..), addDays, fromGregorian) import Data.Time.Clock (UTCTime (..), diffTimeToPicoseconds, picosecondsToDiffTime)-import Data.Time.LocalTime (LocalTime (..), TimeOfDay (..), timeOfDayToTime)+import Data.Time.Clock.POSIX (posixSecondsToUTCTime)+import Data.Time.LocalTime (LocalTime (..), TimeOfDay (..), timeOfDayToTime, utc, utcToLocalTime) import Data.UUID (UUID) import qualified Data.UUID as UUID import Data.Word (Word16, Word32, Word64, Word8)-import Database.DuckDB.FFI (- pattern DuckDBTypeBlob,- pattern DuckDBTypeBoolean,- pattern DuckDBTypeDouble,- pattern DuckDBTypeInteger,- pattern DuckDBTypeVarchar,- )-import Database.DuckDB.Simple (Connection, close, open, query)+import Database.DuckDB.Simple (Connection, SizedArray, close, open, query) import Database.DuckDB.Simple.FromField (- BigNum (..),- BitString (..),- FieldValue (..),- FromField,- LogicalTypeRep (..),- StructField (..),- StructValue (..),- UnionMemberType (..),- UnionValue (..),+ BigNum (..),+ BitString (..),+ FieldValue (..),+ FromField,+ LogicalTypeId (..),+ LogicalTypeRep (..),+ StructField (..),+ StructValue (..),+ UnionMemberType (..),+ UnionValue (..), ) import Database.DuckDB.Simple.ToField (ToField) import Database.DuckDB.Simple.Types (Null (..), Only (..))+import GHC.Float (castWord32ToFloat, castWord64ToDouble) import Numeric.Natural (Natural) import Test.QuickCheck (- Gen,- Property,- arbitrary,- arbitrarySizedNatural,- chooseInt,- chooseInteger,- counterexample,- elements,- forAllShrink,- frequency,- ioProperty,- scale,- shrink,- shrinkIntegral,- sized,- suchThat,- vectorOf,- (===),+ Gen,+ Property,+ arbitrary,+ chooseInt,+ chooseInteger,+ counterexample,+ elements,+ forAllShrink,+ frequency,+ ioProperty,+ shrink,+ shrinkIntegral,+ suchThat,+ vectorOf,+ (===), ) import Test.Tasty (TestTree, testGroup, withResource) import Test.Tasty.QuickCheck (testProperty)+import TestUtils (sized) data SomeRoundTrip- = forall a.- (Eq a, Show a, ToField a, FromField a) =>- SomeRoundTrip String (Proxy a) (Gen a) (a -> [a])+ = forall a.+ (Eq a, Show a, ToField a, FromField a) =>+ SomeRoundTrip String (Proxy a) (Gen a) (a -> [a]) roundTripTests :: TestTree roundTripTests =- withResource- (open ":memory:")- close- \getConn ->- testGroup- "round-trip properties"- (fmap (mkRoundTripTest getConn) specs)+ testGroup+ "round-trip properties"+ (fmap mkRoundTripTest specs) -mkRoundTripTest :: IO Connection -> SomeRoundTrip -> TestTree-mkRoundTripTest getConn (SomeRoundTrip label proxy gen shrinker) =- testProperty label $+mkRoundTripTest :: SomeRoundTrip -> TestTree+mkRoundTripTest (SomeRoundTrip label proxy gen shrinker) =+ withResource+ (open ":memory:")+ close+ \getConn ->+ testProperty label $ forAllShrink gen shrinker (roundTripProperty getConn proxy) -roundTripProperty ::- forall a.- (Eq a, Show a, ToField a, FromField a) =>- IO Connection ->- Proxy a ->- a ->- Property+roundTripProperty+ :: forall a+ . (Eq a, Show a, ToField a, FromField a)+ => IO Connection+ -> Proxy a+ -> a+ -> Property roundTripProperty getConn _ value =- ioProperty do- conn <- getConn- rows <- query conn "SELECT ?" (Only value)- case rows of- [Only out] ->- pure $- counterexample- ("expected " ++ show value ++ ", got " ++ show out)- (out === value)- other ->- fail $- "unexpected row count during round-trip: "- ++ show (length other)+ ioProperty do+ conn <- getConn+ rows <- query conn "SELECT ?" (Only value)+ case rows of+ [Only out] ->+ pure $+ counterexample+ ("expected " ++ show value ++ ", got " ++ show out)+ (out === value)+ other ->+ fail $+ "unexpected row count during round-trip: "+ ++ show (length other) specs :: [SomeRoundTrip] specs =- [ SomeRoundTrip "Null" (Proxy :: Proxy Null) (pure Null) shrinkNone- , SomeRoundTrip "Bool" (Proxy :: Proxy Bool) (arbitrary :: Gen Bool) shrink- , SomeRoundTrip "Int" (Proxy :: Proxy Int) (arbitrary :: Gen Int) shrink- , SomeRoundTrip "Int8" (Proxy :: Proxy Int8) (arbitrary :: Gen Int8) shrink- , SomeRoundTrip "Int16" (Proxy :: Proxy Int16) (arbitrary :: Gen Int16) shrink- , SomeRoundTrip "Int32" (Proxy :: Proxy Int32) (arbitrary :: Gen Int32) shrink- , SomeRoundTrip "Int64" (Proxy :: Proxy Int64) (arbitrary :: Gen Int64) shrink- , SomeRoundTrip "Integer" (Proxy :: Proxy Integer) (arbitrary :: Gen Integer) shrink- , SomeRoundTrip "BigNum" (Proxy :: Proxy BigNum) genBigNum shrinkBigNum- , SomeRoundTrip "Natural" (Proxy :: Proxy Natural) genNatural shrinkNatural- , SomeRoundTrip "Word" (Proxy :: Proxy Word) (arbitrary :: Gen Word) shrink- , SomeRoundTrip "Word8" (Proxy :: Proxy Word8) (arbitrary :: Gen Word8) shrink- , SomeRoundTrip "Word16" (Proxy :: Proxy Word16) (arbitrary :: Gen Word16) shrink- , SomeRoundTrip "Word32" (Proxy :: Proxy Word32) (arbitrary :: Gen Word32) shrink- , SomeRoundTrip "Word64" (Proxy :: Proxy Word64) (arbitrary :: Gen Word64) shrink- , SomeRoundTrip "Float" (Proxy :: Proxy Float) genFiniteFloat shrink- , SomeRoundTrip "Double" (Proxy :: Proxy Double) genFiniteDouble shrink- , SomeRoundTrip "String" (Proxy :: Proxy String) genString shrinkString- , SomeRoundTrip "Text" (Proxy :: Proxy Text.Text) genText shrinkText- , SomeRoundTrip "ByteString" (Proxy :: Proxy BS.ByteString) genByteString shrinkByteString- , SomeRoundTrip "BitString" (Proxy :: Proxy BitString) genBitString shrinkBitString- , SomeRoundTrip "UUID" (Proxy :: Proxy UUID) genUUID shrinkNone- , SomeRoundTrip "Day" (Proxy :: Proxy Day) genDay shrinkDay- , SomeRoundTrip "TimeOfDay" (Proxy :: Proxy TimeOfDay) genTimeOfDay shrinkTimeOfDay- , SomeRoundTrip "LocalTime" (Proxy :: Proxy LocalTime) genLocalTime shrinkLocalTime- , SomeRoundTrip "UTCTime" (Proxy :: Proxy UTCTime) genUTCTime shrinkUTCTime- , SomeRoundTrip "Maybe Int" (Proxy :: Proxy (Maybe Int)) genMaybeInt shrinkMaybeInt- , SomeRoundTrip "Array Int Int" (Proxy :: Proxy (Array Int Int)) genIntArray shrinkIntArray- , SomeRoundTrip "StructValue FieldValue" (Proxy :: Proxy (StructValue FieldValue)) genStructValue shrinkStructValue- , SomeRoundTrip "UnionValue FieldValue" (Proxy :: Proxy (UnionValue FieldValue)) genUnionValue shrinkNone- ]+ [ SomeRoundTrip "Null" (Proxy @Null) (pure Null) shrinkNone+ , SomeRoundTrip "Bool" (Proxy @Bool) (arbitrary :: Gen Bool) shrink+ , SomeRoundTrip "Int" (Proxy @Int) genBounded shrink+ , SomeRoundTrip "Int8" (Proxy @Int8) genBounded shrink+ , SomeRoundTrip "Int16" (Proxy @Int16) genBounded shrink+ , SomeRoundTrip "Int32" (Proxy @Int32) genBounded shrink+ , SomeRoundTrip "Int64" (Proxy @Int64) genBounded shrink+ , SomeRoundTrip "Integer" (Proxy @Integer) genInteger shrink+ , SomeRoundTrip "BigNum" (Proxy @BigNum) genBigNum shrinkBigNum+ , SomeRoundTrip "Natural" (Proxy @Natural) genNatural shrinkNatural+ , SomeRoundTrip "Word" (Proxy @Word) genBounded shrink+ , SomeRoundTrip "Word8" (Proxy @Word8) genBounded shrink+ , SomeRoundTrip "Word16" (Proxy @Word16) genBounded shrink+ , SomeRoundTrip "Word32" (Proxy @Word32) genBounded shrink+ , SomeRoundTrip "Word64" (Proxy @Word64) genBounded shrink+ , SomeRoundTrip "Float" (Proxy @Float) genFiniteFloat shrink+ , SomeRoundTrip "Double" (Proxy @Double) genFiniteDouble shrink+ , SomeRoundTrip "String" (Proxy @String) genString shrinkString+ , SomeRoundTrip "Text" (Proxy @Text.Text) genText shrinkText+ , SomeRoundTrip "ByteString" (Proxy @BS.ByteString) genByteString shrinkByteString+ , SomeRoundTrip "BitString" (Proxy @BitString) genBitString shrinkBitString+ , SomeRoundTrip "UUID" (Proxy @UUID) genUUID shrinkNone+ , SomeRoundTrip "Day" (Proxy @Day) genDay shrinkDay+ , SomeRoundTrip "TimeOfDay" (Proxy @TimeOfDay) genTimeOfDay shrinkTimeOfDay+ , SomeRoundTrip "LocalTime" (Proxy @LocalTime) genLocalTime shrinkLocalTime+ , SomeRoundTrip "UTCTime" (Proxy @UTCTime) genUTCTime shrinkUTCTime+ , SomeRoundTrip "Maybe Int" (Proxy @(Maybe Int)) genMaybeInt shrinkMaybeInt+ , SomeRoundTrip "SizedArray 4 Int" (Proxy @(SizedArray 4 Int)) genIntArray shrinkNone+ , SomeRoundTrip "StructValue FieldValue" (Proxy @(StructValue FieldValue)) genStructValue shrinkStructValue+ , SomeRoundTrip "UnionValue FieldValue" (Proxy @(UnionValue FieldValue)) genUnionValue shrinkNone+ ] shrinkNone :: a -> [a] shrinkNone _ = [] +-- | Choose a value from the whole range of a bounded integral type. One value+-- in four is a boundary of the type: the minimum, the maximum, -1, 0, or 1.+genBounded :: forall a. (Bounded a, Integral a) => Gen a+genBounded =+ frequency+ [ (1, elements [minBound, maxBound, fromInteger (max (-1) (toInteger (minBound :: a))), 0, 1])+ , (3, fromInteger <$> chooseInteger (toInteger (minBound :: a), toInteger (maxBound :: a)))+ ]++-- | Choose an integer of at most 200 bits. The boundaries are those of the+-- 64-bit and 128-bit integer types, where the binding changes its native type.+genInteger :: Gen Integer+genInteger =+ frequency+ [ (1, elements integerBoundaries)+ , (1, chooseInteger (-1000, 1000))+ , (2, chooseInteger (negate (2 ^ (200 :: Int)), 2 ^ (200 :: Int)))+ ]+ where+ integerBoundaries =+ concat [[value, negate value] | bits <- [63, 64, 127, 128 :: Int], value <- [2 ^ bits - 1, 2 ^ bits, 2 ^ bits + 1]] <> [0, 1, -1]+ genNatural :: Gen Natural-genNatural = fromInteger <$> arbitrarySizedNatural+genNatural = fromInteger . abs <$> genInteger shrinkNatural :: Natural -> [Natural] shrinkNatural n = fromInteger <$> shrinkIntegral (toInteger n) genBigNum :: Gen BigNum-genBigNum = BigNum <$> scale (\s -> s * 8) (arbitrary :: Gen Integer)+genBigNum = BigNum <$> genInteger shrinkBigNum :: BigNum -> [BigNum] shrinkBigNum (BigNum n) = BigNum <$> shrinkIntegral n +-- | Choose a finite Float from each bit pattern. The boundaries are the zeros,+-- the largest finite values, the smallest normal value, and the smallest+-- subnormal values. genFiniteFloat :: Gen Float genFiniteFloat =- (arbitrary :: Gen Float)- `suchThat` \x -> not (isNaN x || isInfinite x)+ frequency+ [ (1, elements [0, -0, 1, -1, 3.4028235e38, -3.4028235e38, 1.1754944e-38, 1.0e-45, -1.0e-45])+ , (3, (castWord32ToFloat <$> arbitrary) `suchThat` \x -> not (isNaN x || isInfinite x))+ ] +-- | Choose a finite Double from each bit pattern, with the boundaries of the+-- same kinds as for 'genFiniteFloat'. genFiniteDouble :: Gen Double genFiniteDouble =- (arbitrary :: Gen Double)- `suchThat` \x -> not (isNaN x || isInfinite x)+ frequency+ [ (1, elements [0, -0, 1, -1, 1.7976931348623157e308, -1.7976931348623157e308, 2.2250738585072014e-308, 5.0e-324, -5.0e-324])+ , (3, (castWord64ToDouble <$> arbitrary) `suchThat` \x -> not (isNaN x || isInfinite x))+ ] +-- | Choose a string of at most 300 characters. The boundaries have 0, 1, 11, 12,+-- and 13 ASCII characters, around the 12 bytes that DuckDB stores inline. genString :: Gen String genString =- sized \n -> do- len <- chooseInt (0, max 0 (min n 32))- vectorOf len genChar+ frequency+ [ (1, elements [replicate count 'x' | count <- [0, 1, 11, 12, 13]])+ , (3, chooseInt (0, 300) >>= \len -> vectorOf len genChar)+ ] shrinkString :: String -> [String] shrinkString = shrink genChar :: Gen Char genChar =- frequency [(1, pure '\0'), (9, arbitrary)]+ frequency [(1, pure '\0'), (9, arbitrary)] genText :: Gen Text.Text genText = Text.pack <$> genString@@ -187,221 +221,246 @@ shrinkText :: Text.Text -> [Text.Text] shrinkText txt = Text.pack <$> shrinkString (Text.unpack txt) +-- | Choose a byte string of at most 300 bytes. The boundaries have 0, 12, and+-- 13 bytes, around the 12 bytes that DuckDB stores inline. genByteString :: Gen BS.ByteString genByteString =- sized \n -> do- len <- chooseInt (0, max 0 (min n 64))- BS.pack <$> vectorOf len (arbitrary :: Gen Word8)+ frequency+ [ (1, elements [BS.replicate count byte | count <- [0, 12, 13], byte <- [0, 255]])+ , (3, chooseInt (0, 300) >>= \len -> BS.pack <$> vectorOf len (arbitrary :: Gen Word8))+ ] shrinkByteString :: BS.ByteString -> [BS.ByteString] shrinkByteString bs =- [ BS.pack xs- | xs <- shrink (BS.unpack bs)- , length xs <= 64- ]+ [ BS.pack xs+ | xs <- shrink (BS.unpack bs)+ , length xs <= 300+ ] +-- | Choose a BIT value of 1 to 64 bytes with 0 to 7 bits of padding. The+-- padding bits of the first byte are 0, as the decoder gives them. genBitString :: Gen BitString-genBitString =- sized \n -> do- len <- chooseInt (1, max 1 (min 64 (n * 2)))- bytes <- vectorOf len (arbitrary :: Gen Word8)- pure (BitString 0 (BS.pack bytes))+genBitString = do+ len <- chooseInt (1, 64)+ padding <- chooseInt (0, 7)+ first <- arbitrary :: Gen Word8+ rest <- vectorOf (len - 1) (arbitrary :: Gen Word8)+ pure (BitString (fromIntegral padding) (BS.pack (foldl clearBit first [8 - padding .. 7] : rest))) shrinkBitString :: BitString -> [BitString] shrinkBitString (BitString _ raw) =- [ BitString 0 (BS.pack xs)- | xs <- shrink (BS.unpack raw)- , length xs <= 64- , not (null xs)- ]+ [ BitString 0 (BS.pack xs)+ | xs <- shrink (BS.unpack raw)+ , length xs <= 64+ , not (null xs)+ ] +-- | Choose a UUID from each bit pattern. The boundaries are the nil UUID and+-- the UUID with each bit set. genUUID :: Gen UUID genUUID =- UUID.fromWords- <$> (arbitrary :: Gen Word32)- <*> (arbitrary :: Gen Word32)- <*> (arbitrary :: Gen Word32)- <*> (arbitrary :: Gen Word32)+ frequency+ [ (1, elements [UUID.nil, UUID.fromWords maxBound maxBound maxBound maxBound])+ , (3, UUID.fromWords <$> arbitrary <*> arbitrary <*> arbitrary <*> arbitrary)+ ] genMaybeInt :: Gen (Maybe Int) genMaybeInt =- sized \n ->- let bound = max 1 (min 1000 (n * 10))- in frequency- [ (1, pure Nothing)- , (3, Just <$> chooseInt (-bound, bound))- ]+ frequency+ [ (1, pure Nothing)+ , (3, Just <$> genBounded)+ ] shrinkMaybeInt :: Maybe Int -> [Maybe Int] shrinkMaybeInt Nothing = [] shrinkMaybeInt (Just x) = Nothing : [Just y | y <- shrink x] +-- | Choose a day in the range of DATE that DuckDB accepts: from 5877642-06-25+-- (BC), which is day -2147483646 from the Unix epoch, to 5881580-07-10, which+-- is day 2147483646. The library raises an error for a smaller day. See+-- Note [Time ranges] in "Database.DuckDB.Simple.ToField". The boundaries are+-- the ends of the range and the days next to the epoch. genDay :: Gen Day-genDay = ModifiedJulianDay <$> (arbitrary :: Gen Integer)+genDay =+ (`addDays` fromGregorian 1970 1 1)+ <$> frequency+ [ (1, elements [smallest, -1, 0, 1, largest])+ , (3, chooseInteger (smallest, largest))+ ]+ where+ smallest = -2147483646+ largest = 2147483646 +-- | Choose a count of microseconds from the Unix epoch in the range of TIMESTAMP+-- that DuckDB states: from 290309-12-22 (BC) 00:00:00, which is+-- -9223372022400000000, to 294247-01-10 04:00:54.775806, which is maxBound - 1.+-- DuckDB gives a conversion error for a smaller count.+genTimestampMicros :: Gen Integer+genTimestampMicros =+ frequency+ [ (1, elements [smallest, -1, 0, 1, largest])+ , (3, chooseInteger (smallest, largest))+ ]+ where+ smallest = -9223372022400000000+ largest = toInteger (maxBound :: Int64) - 1+ shrinkDay :: Day -> [Day] shrinkDay (ModifiedJulianDay d) = ModifiedJulianDay <$> shrink d +-- | Choose a time of day in microseconds. The boundaries are midnight, the last+-- microsecond of the day, and 24:00:00, which DuckDB also accepts. A parameter+-- with no known type keeps only microseconds. genTimeOfDay :: Gen TimeOfDay-genTimeOfDay = do- micros <- chooseInteger (0, 86400 * 1000000 - 1)- pure (microsecondsToTimeOfDay micros)+genTimeOfDay =+ microsecondsToTimeOfDay+ <$> frequency+ [ (1, elements [0, 86400 * 1000000 - 1, 86400 * 1000000])+ , (3, chooseInteger (0, 86400 * 1000000 - 1))+ ] shrinkTimeOfDay :: TimeOfDay -> [TimeOfDay] shrinkTimeOfDay tod =- [ microsecondsToTimeOfDay us- | us <- shrinkIntegral (timeOfDayToMicroseconds tod)- , us >= 0- ]+ [ microsecondsToTimeOfDay us+ | us <- shrinkIntegral (timeOfDayToMicroseconds tod)+ , us >= 0+ ] +-- | Choose a local time from the range of TIMESTAMP, with the boundaries of+-- 'genTimestampMicros'. genLocalTime :: Gen LocalTime-genLocalTime = LocalTime <$> genDay <*> genTimeOfDay+genLocalTime = utcToLocalTime utc <$> genUTCTime shrinkLocalTime :: LocalTime -> [LocalTime] shrinkLocalTime LocalTime{localDay, localTimeOfDay} =- [ LocalTime day' localTimeOfDay- | day' <- shrinkDay localDay- ]- ++ [ LocalTime localDay time'- | time' <- shrinkTimeOfDay localTimeOfDay- ]+ [ LocalTime day' localTimeOfDay+ | day' <- shrinkDay localDay+ ]+ ++ [ LocalTime localDay time'+ | time' <- shrinkTimeOfDay localTimeOfDay+ ] +-- | Choose a UTC time from the range of TIMESTAMPTZ, with the boundaries of+-- 'genTimestampMicros'. genUTCTime :: Gen UTCTime genUTCTime = do- day <- genDay- micros <- chooseInteger (0, 86400 * 1000000 - 1)- pure (UTCTime day (picosecondsToDiffTime (micros * 1000000)))+ micros <- genTimestampMicros+ pure (posixSecondsToUTCTime (fromRational (micros % 1000000))) shrinkUTCTime :: UTCTime -> [UTCTime] shrinkUTCTime (UTCTime day diff) =- [UTCTime day' diff | day' <- shrinkDay day]- ++ [UTCTime day diff' | diff' <- shrinkDiff diff]+ [UTCTime day' diff | day' <- shrinkDay day]+ ++ [UTCTime day diff' | diff' <- shrinkDiff diff] where shrinkDiff dt =- [ picosecondsToDiffTime ps- | ps <- shrinkIntegral (diffTimeToPicoseconds dt)- , ps >= 0- , ps `mod` 1000000 == 0- ]--genIntArray :: Gen (Array Int Int)-genIntArray =- sized \n -> do- size <- chooseInt (0, max 0 (min n 8))- values <- vectorOf size (arbitrary :: Gen Int)- pure $- if size == 0- then listArray (1, 0) []- else listArray (0, size - 1) values+ [ picosecondsToDiffTime ps+ | ps <- shrinkIntegral (diffTimeToPicoseconds dt)+ , ps >= 0+ , ps `mod` 1000000 == 0+ ] -shrinkIntArray :: Array Int Int -> [Array Int Int]-shrinkIntArray arr =- let values = elems arr- in [ listArray (0, length xs - 1) xs- | xs <- shrink values- , not (null xs)- ]- ++ if null values then [] else [listArray (1, 0) []]+genIntArray :: Gen (SizedArray 4 Int)+genIntArray = sized <$> vectorOf 4 genBounded genStructValue :: Gen (StructValue FieldValue) genStructValue = do- count <- chooseInt (0, 3)- if count == 0- then pure emptyStruct- else do- types <- vectorOf count genSimpleType- values <- forM types simpleTypeValue- let names = [Text.pack ("field" ++ show i) | i <- [0 .. count - 1]]- fieldArr =- listArray- (0, count - 1)- [ StructField{structFieldName = name, structFieldValue = val}- | (name, val) <- zip names values- ]- typeArr =- listArray- (0, count - 1)- [ StructField{structFieldName = name, structFieldValue = simpleTypeRep ty}- | (name, ty) <- zip names types- ]- indexMap = Map.fromList (zip names [0 ..])- pure- StructValue- { structValueFields = fieldArr- , structValueTypes = typeArr- , structValueIndex = indexMap- }+ count <- chooseInt (0, 3)+ if count == 0+ then pure emptyStruct+ else do+ types <- vectorOf count genSimpleType+ values <- forM types simpleTypeValue+ let names = [Text.pack ("field" ++ show i) | i <- [0 .. count - 1]]+ fieldArr =+ listArray+ (0, count - 1)+ [ StructField{structFieldName = name, structFieldValue = val}+ | (name, val) <- zip names values+ ]+ typeArr =+ listArray+ (0, count - 1)+ [ StructField{structFieldName = name, structFieldValue = simpleTypeRep ty}+ | (name, ty) <- zip names types+ ]+ indexMap = Map.fromList (zip names [0 ..])+ pure+ StructValue+ { structValueFields = fieldArr+ , structValueTypes = typeArr+ , structValueIndex = indexMap+ } shrinkStructValue :: StructValue FieldValue -> [StructValue FieldValue] shrinkStructValue StructValue{structValueFields}- | null (elems structValueFields) = []- | otherwise = [emptyStruct]+ | null (elems structValueFields) = []+ | otherwise = [emptyStruct] genUnionValue :: Gen (UnionValue FieldValue) genUnionValue = do- memberCount <- chooseInt (1, 4)- types <- vectorOf memberCount genSimpleType- let names = [Text.pack ("alt" ++ show i) | i <- [0 .. memberCount - 1]]- members =- listArray- (0, memberCount - 1)- [ UnionMemberType{unionMemberName = name, unionMemberType = simpleTypeRep ty}- | (name, ty) <- zip names types- ]- tag <- chooseInt (0, memberCount - 1)- payload <- simpleTypeValue (types !! tag)- pure- UnionValue- { unionValueIndex = fromIntegral tag- , unionValueLabel = names !! tag- , unionValuePayload = payload- , unionValueMembers = members- }+ memberCount <- chooseInt (1, 4)+ types <- vectorOf memberCount genSimpleType+ let names = [Text.pack ("alt" ++ show i) | i <- [0 .. memberCount - 1]]+ members =+ listArray+ (0, memberCount - 1)+ [ UnionMemberType{unionMemberName = name, unionMemberType = simpleTypeRep ty}+ | (name, ty) <- zip names types+ ]+ tag <- chooseInt (0, memberCount - 1)+ payload <- simpleTypeValue (types !! tag)+ pure+ UnionValue+ { unionValueIndex = fromIntegral tag+ , unionValueLabel = names !! tag+ , unionValuePayload = payload+ , unionValueMembers = members+ } emptyStruct :: StructValue FieldValue emptyStruct =- StructValue- { structValueFields = listArray (1, 0) []- , structValueTypes = listArray (1, 0) []- , structValueIndex = Map.empty- }+ StructValue+ { structValueFields = listArray (1, 0) []+ , structValueTypes = listArray (1, 0) []+ , structValueIndex = Map.empty+ } data SimpleType- = SimpleBool- | SimpleInt32- | SimpleDouble- | SimpleText- | SimpleBlob- deriving (Eq, Show)+ = SimpleBool+ | SimpleInt32+ | SimpleDouble+ | SimpleText+ | SimpleBlob+ deriving (Eq, Show) genSimpleType :: Gen SimpleType genSimpleType =- elements [SimpleBool, SimpleInt32, SimpleDouble, SimpleText, SimpleBlob]+ elements [SimpleBool, SimpleInt32, SimpleDouble, SimpleText, SimpleBlob] simpleTypeRep :: SimpleType -> LogicalTypeRep simpleTypeRep = \case- SimpleBool -> LogicalTypeScalar DuckDBTypeBoolean- SimpleInt32 -> LogicalTypeScalar DuckDBTypeInteger- SimpleDouble -> LogicalTypeScalar DuckDBTypeDouble- SimpleText -> LogicalTypeScalar DuckDBTypeVarchar- SimpleBlob -> LogicalTypeScalar DuckDBTypeBlob+ SimpleBool -> LogicalTypeScalar LogicalTypeIdBoolean+ SimpleInt32 -> LogicalTypeScalar LogicalTypeIdInteger+ SimpleDouble -> LogicalTypeScalar LogicalTypeIdDouble+ SimpleText -> LogicalTypeScalar LogicalTypeIdVarchar+ SimpleBlob -> LogicalTypeScalar LogicalTypeIdBlob simpleTypeValue :: SimpleType -> Gen FieldValue simpleTypeValue = \case- SimpleBool -> FieldBool <$> (arbitrary :: Gen Bool)- SimpleInt32 -> FieldInt32 <$> (arbitrary :: Gen Int32)- SimpleDouble -> FieldDouble <$> genFiniteDouble- SimpleText -> FieldText <$> genText- SimpleBlob -> FieldBlob <$> genByteString+ SimpleBool -> FieldBool <$> (arbitrary :: Gen Bool)+ SimpleInt32 -> FieldInt32 <$> genBounded+ SimpleDouble -> FieldDouble <$> genFiniteDouble+ SimpleText -> FieldText <$> genText+ SimpleBlob -> FieldBlob <$> genByteString timeOfDayToMicroseconds :: TimeOfDay -> Integer timeOfDayToMicroseconds tod =- diffTimeToPicoseconds (timeOfDayToTime tod) `div` 1000000+ diffTimeToPicoseconds (timeOfDayToTime tod) `div` 1000000 microsecondsToTimeOfDay :: Integer -> TimeOfDay microsecondsToTimeOfDay micros =- let (hours, remHour) = micros `divMod` (3600 * 1000000)- (minutes, remMinute) = remHour `divMod` (60 * 1000000)- (seconds, fracMicros) = remMinute `divMod` 1000000- pico = seconds * 1000000000000 + fracMicros * 1000000- in TimeOfDay (fromInteger hours) (fromInteger minutes) (MkFixed pico)+ let (hours, remHour) = micros `divMod` (3600 * 1000000)+ (minutes, remMinute) = remHour `divMod` (60 * 1000000)+ (seconds, fracMicros) = remMinute `divMod` 1000000+ pico = seconds * 1000000000000 + fracMicros * 1000000+ in TimeOfDay (fromInteger hours) (fromInteger minutes) (MkFixed pico)
+ test/RegistrationTests.hs view
@@ -0,0 +1,705 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DeriveGeneric #-}+{-# LANGUAGE DerivingVia #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : RegistrationTests+-- Description : Native integration checks for managed V2 registration.+module RegistrationTests (registrationTests) where++import Control.Exception (IOException, bracket, evaluate, throwIO, try)+import Control.Monad (forM, forM_, void, when)+import Data.IORef (IORef, atomicModifyIORef', mkWeakIORef, modifyIORef', newIORef, readIORef, writeIORef)+import Data.Int (Int64)+import Data.Maybe (isNothing)+import Data.Proxy (Proxy (..))+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.FFI.V2+import Database.DuckDB.Simple hiding (bind, execute)+import Database.DuckDB.Simple.FromField (FieldValue (..))+import Database.DuckDB.Simple.Generic (DuckValue (..), ViaDuckDB (..))+import Database.DuckDB.Simple.Internal (withBytes, withConnectionFactory, withConnectionHandle)+import qualified Database.DuckDB.Simple.Logging as Logging+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructValue, UnionValue)+import qualified Database.DuckDB.Simple.QueryMetrics as Metrics+import qualified Database.DuckDB.Simple.V2.FunctionDocs as Docs+import qualified Database.DuckDB.Simple.V2.Logging as V2Logging+import qualified Database.DuckDB.Simple.V2.Pragma as Pragma+import qualified Database.DuckDB.Simple.V2.Registration as R+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Marshal.Array (advancePtr, allocaArray, peekArray)+import Foreign.Ptr (castPtr, freeHaskellFunPtr, nullPtr)+import Foreign.Storable (peek, peekElemOff, poke, pokeElemOff)+import GHC.Generics (Generic)+import HsBindgen.Runtime.Marshal (staticSizeOf)+import HsBindgen.Runtime.Support.FunPtr (toFunPtr)+import System.Mem (performMajorGC)+import System.Mem.Weak (Weak, deRefWeak)+import Test.Tasty (TestTree, testGroup)+import Test.Tasty.HUnit (Assertion, assertBool, assertEqual, assertFailure, testCase, (@?=))+import TestUtils (assertSqlErrorContaining, writeIntegers, writeNullRows)++-- | Check registration, callback errors, typed signatures, and resource release.+registrationTests :: TestTree+registrationTests =+ testGroup+ "V2 registration"+ [ testCase "scalar named defaults and all parameter kinds" testScalar+ , testCase "scalar type resolution bind state and documentation" testScalarResolution+ , testCase "pragma query execution errors and documentation" testPragma+ , testCase "pragma names must be nonempty and must not contain NUL" testPragmaNames+ , testCase "query metric snapshots copy values and survive later queries" testMetrics+ , testCase "aggregate grouped and window execution" testAggregate+ , testCase "aggregate registration cleanup follows callback ownership" testAggregateCleanup+ , testCase "table state and typed default parameter" testTable+ , testCase "custom type and normal and try cast" testCast+ , testCase "a composite value with a member of a custom type binds with no registered cast" testCastAliasMembers+ , testCase "a second cast for the same pair of types has no effect" testCastSecond+ , testGroup "a failed cast gives a Mismatch Type error and keeps the database usable" $+ [testCase (castPairName castPair) (testCastFailureCode castPair) | castPair <- castFailurePairs]+ <> [ testCase "a query on another connection fails in the callback" testCastFailureQuery+ , testCase "an SQLError of the type Conversion, Invalid Input, or Out of Range in the callback" testCastTypedErrors+ ]+ , testCase "a COMMIT after a failed cast stores no row" testCastFailureCommit+ , testCase "connection scans precede instance scans" testReplacement+ , testCase "callback contexts write log entries" testContextLog+ , testCase "callback failures become SQL errors" testErrors+ , testCase "a native error text that is not valid UTF-8 gives a readable SQL error" testInvalidErrorText+ , testCase "callback closures and query states are released" testRelease+ ]++bigint :: R.TypeDeclaration+bigint = R.StructuredType (LogicalTypeScalar LogicalTypeIdBigInt)++scalarSignature :: [R.FunctionParameter] -> R.FunctionSignature+scalarSignature parameters = R.FunctionSignature parameters (Just bigint)++scalarSum :: R.ContextCallback DuckDBV2ScalarFunctionExecInfoHandle+scalarSum info _ = do+ count <- R.output (c_duckdb_v2_scalar_function_exec_get_row_count info)+ arguments <- allocaArray 4 \counts -> do+ R.checked (c_duckdb_v2_scalar_function_exec_get_arg_count info counts (advancePtr counts 1) (advancePtr counts 2) (advancePtr counts 3))+ fromIntegral . sum <$> peekArray 4 counts+ columns <- forM [0 .. arguments - 1 :: Int] \index ->+ R.output (c_duckdb_v2_scalar_function_exec_get_arg info (fromIntegral index)) >>= readIntegers (fromIntegral count)+ vector <- R.output (c_duckdb_v2_scalar_function_exec_get_result info)+ writeIntegers vector [sum [column !! row | column <- columns] | row <- [0 .. fromIntegral count - 1]]++testScalar :: Assertion+testScalar = withConnection ":memory:" \connection -> do+ let left = (R.parameter "left" bigint){R.parameterDefault = Just (R.defaultValue (7 :: Int64))}+ right =+ (R.parameter "right" bigint)+ { R.parameterDefault = Just (R.DefaultValue (\factory -> R.output (c_duckdb_v2_value_create_bigint factory 35)))+ , R.parameterKind = DuckDBV2FunctionParameterKindNamedOnly+ }+ R.registerScalarFunction connection "managed_default" (scalarSignature [left, right]) [] (R.scalarCallbacks scalarSum)+ query_ connection "SELECT managed_default()" >>= (@?= [Only (42 :: Int64)])+ query_ connection "SELECT managed_default(right := 5, left := 3)" >>= (@?= [Only (8 :: Int64)])+ let target = R.ConnectionTarget connection+ R.withRegistrationFactory target \factory ->+ R.registerScalarFunctionWith+ target+ ( \function -> do+ R.withIdentifier "managed_target_default" (R.checked . c_duckdb_v2_scalar_function_set_name function)+ R.output (c_duckdb_v2_scalar_function_get_signature function) >>= R.configureFunctionSignature factory (scalarSignature [left])+ )+ (R.scalarCallbacks scalarSum)+ query_ connection "SELECT managed_target_default()" >>= (@?= [Only (7 :: Int64)])+ let positional = (R.parameter "first" bigint){R.parameterKind = DuckDBV2FunctionParameterKindPositionalOnly}+ standard = R.parameter "second" bigint+ rest = (R.parameter "rest" bigint){R.parameterKind = DuckDBV2FunctionParameterKindPositionalVariadic}+ named = (R.parameter "named" bigint){R.parameterKind = DuckDBV2FunctionParameterKindNamedOnly}+ kwargs = (R.parameter "kwargs" bigint){R.parameterKind = DuckDBV2FunctionParameterKindNamedVariadic}+ R.registerScalarFunction+ connection+ "managed_variadic"+ (scalarSignature [positional, standard, rest, named, kwargs])+ []+ (R.scalarCallbacks scalarSum)+ query_ connection "SELECT managed_variadic(1, 2, 3, 4, named := 5, extra := 6)" >>= (@?= [Only (21 :: Int64)])++testScalarResolution :: Assertion+testScalarResolution = withConnection ":memory:" \connection -> do+ phases <- newIORef ([] :: [Text.Text])+ let mark phase = modifyIORef' phases (phase :)+ resolve input result _ = do+ mark "resolve"+ bracket (R.output (c_duckdb_v2_function_bind_get_arg_type input 0)) (R.destroy c_duckdb_v2_logical_type_destroy) $ \dtype ->+ R.checked (c_duckdb_v2_scalar_function_resolve_types_set_return_type result dtype)+ bindState _ result _ = do+ mark "bind"+ R.setCallbackState (c_duckdb_v2_scalar_function_bind_set_bind_data result) (42 :: Int)+ executeRows info _ = do+ state <- R.getCallbackState (c_duckdb_v2_scalar_function_exec_get_bind_data info)+ (state :: Int) @?= 42+ mismatch <- try (R.getCallbackState (c_duckdb_v2_scalar_function_exec_get_bind_data info) :: IO Bool)+ either (\(_ :: SQLError) -> pure ()) (\_ -> assertFailure "state read with another type") mismatch+ count <- R.output (c_duckdb_v2_scalar_function_exec_get_row_count info)+ argument <- R.output (c_duckdb_v2_scalar_function_exec_get_arg info 0)+ result <- R.output (c_duckdb_v2_scalar_function_exec_get_result info)+ R.checked (c_duckdb_v2_vector_set_size result count)+ forM_ (takeWhile (< count) [0 ..]) $ \index ->+ bracket (R.output (c_duckdb_v2_vector_get_value argument index)) (R.destroy c_duckdb_v2_value_destroy) $ \value ->+ R.checked (c_duckdb_v2_vector_set_value result index value)+ anyType = R.StructuredType (LogicalTypeScalar LogicalTypeIdAny)+ signature = R.FunctionSignature [R.parameter "value" anyType] (Just anyType)+ target = R.ConnectionTarget connection+ R.withRegistrationFactory target $ \factory ->+ R.registerScalarFunctionWith+ target+ ( \function -> do+ R.withIdentifier "managed_resolved" (R.checked . c_duckdb_v2_scalar_function_set_name function)+ R.output (c_duckdb_v2_scalar_function_get_signature function) >>= R.configureFunctionSignature factory signature+ docs <- R.output (c_duckdb_v2_scalar_function_get_docs function)+ Docs.setFunctionDocs docs (Docs.FunctionDocs (Just "Preserve the input type.") [("Example", "SELECT managed_resolved(42)")] ["test"])+ )+ (R.scalarCallbacks executeRows){R.scalarResolveTypesCallback = Just resolve, R.scalarBindCallback = Just bindState}+ query_ connection "SELECT typeof(managed_resolved(i::SMALLINT)), managed_resolved(i::SMALLINT) FROM range(2) t(i) ORDER BY i"+ >>= (@?= [("SMALLINT" :: Text.Text, 0 :: Int64), ("SMALLINT", 1)])+ seen <- readIORef phases+ assertBool "resolve-types callback" ("resolve" `elem` seen)+ assertBool "bind callback" ("bind" `elem` seen)+ query_ connection "SELECT description FROM duckdb_functions() WHERE function_name='managed_resolved'"+ >>= (@?= [Only ("Preserve the input type." :: Text.Text)])++testPragma :: Assertion+testPragma = withConnection ":memory:" \connection -> do+ let amount = (R.parameter "amount" bigint){R.parameterDefault = Just (R.defaultValue (42 :: Int64))}+ Pragma.registerPragmaFunctionOn (R.ConnectionTarget connection) "managed_query" [amount] $ Pragma.QueryPragma $ \info _ -> do+ value <-+ bracket+ (R.output (c_duckdb_v2_function_bind_get_arg_value info 0))+ (R.destroy c_duckdb_v2_value_destroy)+ (R.output . c_duckdb_v2_value_get_bigint)+ pure ("SELECT " <> Text.pack (show value) <> "::BIGINT")+ query_ connection "PRAGMA managed_query" >>= (@?= [Only (42 :: Int64)])+ query_ connection "PRAGMA managed_query(7)" >>= (@?= [Only (7 :: Int64)])+ calls <- newIORef (0 :: Int)+ Pragma.registerPragmaFunctionWith+ (R.ConnectionTarget connection)+ ( \function -> do+ R.withIdentifier "managed_exec" (R.checked . c_duckdb_v2_pragma_function_set_name function)+ docs <- R.output (c_duckdb_v2_pragma_function_get_docs function)+ Docs.setFunctionDocs docs Docs.defaultFunctionDocs{Docs.functionDescription = Just "Count calls."}+ )+ (Pragma.ExecPragma (\_ _ -> modifyIORef' calls (+ 1)))+ void (execute_ connection "PRAGMA managed_exec")+ readIORef calls >>= (@?= 1)+ Pragma.registerPragmaFunctionOn+ (R.ConnectionTarget connection)+ "managed_pragma_error"+ []+ (Pragma.QueryPragma (\_ _ -> throwIO (userError "managed callback failure")))+ assertSqlError (query_ connection "PRAGMA managed_pragma_error" :: IO [Only Int64])+ query_ connection "SELECT 1::BIGINT" >>= (@?= [Only (1 :: Int64)])++testPragmaNames :: Assertion+testPragmaNames = withConnection ":memory:" \connection -> do+ forM_ ["", "managed\0pragma"] \name -> do+ outcome <- try (Pragma.registerPragmaFunctionOn (R.ConnectionTarget connection) name [] (Pragma.ExecPragma (\_ _ -> pure ())))+ case outcome of+ Left (err :: SQLError) -> assertBool "name diagnostic" ("pragma function name" `Text.isInfixOf` sqlErrorMessage err)+ Right () -> assertFailure "expected SQL error"+ query_ connection "SELECT 1::BIGINT" >>= (@?= [Only (1 :: Int64)])++testMetrics :: Assertion+testMetrics = withConnection ":memory:" \connection -> do+ Metrics.enableProfiling connection True+ query_ connection "SELECT sum(i)::BIGINT FROM range(10) t(i)" >>= (@?= [Only (45 :: Int64)])+ snapshot <- Metrics.queryMetrics connection+ Metrics.metricsNodeName snapshot @?= "QUERY"+ assertBool "root metrics" (not (null (Metrics.metricsValues snapshot)))+ assertBool "operator metrics" (not (null (Metrics.metricsChildren snapshot)))+ let scanRows =+ [ value+ | node <- metricNodes snapshot+ , Metrics.metricsNodeName node == "TABLE_SCAN"+ , Just value <- [lookup "operator.intermediate_rows" (Metrics.metricsValues node)]+ ]+ scanRows @?= [FieldWord64 10]+ Metrics.enableProfiling connection False+ query_ connection "SELECT 2::BIGINT" >>= (@?= [Only (2 :: Int64)])+ empty <- Metrics.queryMetrics connection+ Metrics.metricsValues empty @?= []+ Metrics.metricsChildren empty @?= []+ where+ metricNodes node = node : concatMap metricNodes (Metrics.metricsChildren node)++sumCallbacks :: R.AggregateCallbacks+sumCallbacks = R.aggregateCallbacks stateSize initialize update combine finalize+ where+ stateSize info = R.checked (c_duckdb_v2_aggregate_function_size_set_state_size info (fromIntegral (staticSizeOf (Proxy @Int64))))+ initialize info = do+ count <- R.output (c_duckdb_v2_aggregate_function_init_get_state_count info)+ states <- R.output (c_duckdb_v2_aggregate_function_init_get_states info)+ forM_ [0 .. fromIntegral count - 1] \index -> peekElemOff states index >>= \state -> poke (castPtr state) (0 :: Int64)+ update info = do+ count <- R.output (c_duckdb_v2_aggregate_function_update_get_row_count info)+ states <- R.output (c_duckdb_v2_aggregate_function_update_get_states info)+ values <- R.output (c_duckdb_v2_aggregate_function_update_get_arg info 0) >>= readIntegers (fromIntegral count)+ forM_ (zip [0 ..] values) \(index, value) -> do+ state <- castPtr <$> peekElemOff states index+ total <- peek state+ poke state (total + value :: Int64)+ combine info = do+ count <- R.output (c_duckdb_v2_aggregate_function_combine_get_state_count info)+ sources <- R.output (c_duckdb_v2_aggregate_function_combine_get_sources info)+ targets <- R.output (c_duckdb_v2_aggregate_function_combine_get_targets info)+ forM_ [0 .. fromIntegral count - 1] \index -> do+ source <- peekElemOff sources index >>= peek . castPtr+ target <- castPtr <$> peekElemOff targets index+ total <- peek target+ poke target (total + source :: Int64)+ finalize info = do+ count <- R.output (c_duckdb_v2_aggregate_function_finalize_get_state_count info)+ states <- R.output (c_duckdb_v2_aggregate_function_finalize_get_states info)+ offset <- R.output (c_duckdb_v2_aggregate_function_finalize_get_result_offset info)+ result <- R.output (c_duckdb_v2_aggregate_function_finalize_get_result info)+ raw <- R.output (c_duckdb_v2_vector_get_data_mutable result)+ forM_ [0 .. fromIntegral count - 1] \index -> do+ total <- peekElemOff states index >>= peek . castPtr+ pokeElemOff (castPtr raw) (fromIntegral offset + index) (total :: Int64)++testAggregate :: Assertion+testAggregate = withConnection ":memory:" \connection -> do+ destroyed <- newIORef (0 :: Int)+ let callbacks = sumCallbacks{R.aggregateDestroyCallback = Just (\_ -> modifyIORef' destroyed (+ 1))}+ R.registerAggregateFunction connection "managed_sum" (scalarSignature [R.parameter "value" bigint]) [] callbacks+ query_ connection "SELECT managed_sum(i::BIGINT) FROM range(12) t(i) GROUP BY i % 3 ORDER BY i % 3"+ >>= (@?= map Only ([18, 22, 26] :: [Int64]))+ query_ connection "SELECT managed_sum(i::BIGINT) OVER () FROM range(4096) t(i) LIMIT 1" >>= (@?= [Only (8386560 :: Int64)])+ readIORef destroyed >>= assertBool "aggregate state destruction ran" . (> 0)++-- | Keep registration resources across queries and prepared statements. Release+-- them once when the database closes, or when configuration fails.+testAggregateCleanup :: Assertion+testAggregateCleanup = do+ released <- newIORef ([] :: [Text])+ let release name = atomicModifyIORef' released \names -> (name : names, ())+ withConnection ":memory:" \connection -> do+ let target = R.ConnectionTarget connection+ R.withRegistrationFactory target \factory -> do+ let configure name function = do+ R.withIdentifier name (R.checked . c_duckdb_v2_aggregate_function_set_name function)+ signature <- R.output (c_duckdb_v2_aggregate_function_get_signature function)+ R.configureFunctionSignature factory (scalarSignature [R.parameter "value" bigint]) signature+ R.registerAggregateFunctionWithCleanup target (release "used") (configure "managed_owned_sum") sumCallbacks+ R.registerAggregateFunctionWithCleanup target (release "unused") (configure "managed_unused_sum") sumCallbacks+ failed <-+ try+ ( R.registerAggregateFunctionWithCleanup+ target+ (release "failed")+ (\_ -> throwIO (userError "aggregate configuration failed"))+ sumCallbacks+ )+ case failed of+ Left (_ :: IOException) -> pure ()+ Right () -> assertFailure "aggregate configuration must fail"+ readIORef released >>= (@?= ["failed"])+ forM_ [1 .. 3 :: Int] \_ ->+ query_ connection "SELECT managed_owned_sum(i) FROM range(10) t(i)" >>= (@?= [Only (45 :: Int64)])+ withStatement connection "SELECT managed_owned_sum(i) FROM range(10) t(i)" \statement -> do+ nextRow statement >>= (@?= Just (Only (45 :: Int64)))+ void (executeStatement statement)+ nextRow statement >>= (@?= Just (Only (45 :: Int64)))+ readIORef released >>= (@?= ["failed"])+ names <- readIORef released+ forM_ ["used", "unused", "failed"] \name ->+ assertEqual ("cleanup count for " <> Text.unpack name) 1 (length (filter (== name) names))++tableDefinition :: R.TableCallbacks+tableDefinition = (R.tableCallbacks bind execute){R.tableInitGlobalCallback = Just initialize}+ where+ bind input result context = do+ amount <- bracket (R.output (c_duckdb_v2_function_bind_get_arg_value input 0)) (R.destroy c_duckdb_v2_value_destroy) \value ->+ R.output (c_duckdb_v2_value_get_bigint value)+ R.setCallbackState (c_duckdb_v2_function_bind_set_bind_data input) (amount :: Int64)+ bracket+ ( R.output (c_duckdb_v2_context_get_factory context) >>= \factory -> R.output (c_duckdb_v2_factory_create_type_from_id factory DuckDBV2LogicalTypeIdBigInt (ConstPtr nullPtr) (ConstPtr nullPtr) 0)+ )+ (R.destroy c_duckdb_v2_logical_type_destroy)+ \dtype -> R.withIdentifier "amount" \name -> R.checked (c_duckdb_v2_table_function_bind_add_result_column result name dtype)+ initialize info _ = do+ amount <- R.getCallbackState (c_duckdb_v2_table_function_init_global_get_bind_data info)+ state <- newIORef (Just (amount :: Int64))+ R.setCallbackState (c_duckdb_v2_table_function_init_global_set_global_state info) state+ execute info _ = do+ state <- R.getCallbackState (c_duckdb_v2_table_function_exec_get_global_state info) :: IO (IORef (Maybe Int64))+ amount <- atomicModifyIORef' state (\value -> (Nothing, value))+ chunk <- R.output (c_duckdb_v2_table_function_exec_get_output_chunk info)+ vector <- R.output (c_duckdb_v2_data_chunk_get_vector chunk 0)+ writeIntegers vector (maybe [] (\count -> [42 .. 41 + count]) amount)++testTable :: Assertion+testTable = withConnection ":memory:" \connection -> do+ let count = (R.parameter "count" bigint){R.parameterDefault = Just (R.defaultValue (3 :: Int64))}+ R.registerTableFunction connection "managed_scan" [count] tableDefinition+ query_ connection "SELECT amount FROM managed_scan()" >>= (@?= map Only ([42, 43, 44] :: [Int64]))+ query_ connection "SELECT amount FROM managed_scan(count := 1)" >>= (@?= [Only (42 :: Int64)])++testCast :: Assertion+testCast = withConnection ":memory:" \connection -> do+ R.registerCustomType connection "managed_shifted" bigint+ modes <- newIORef []+ R.registerCast connection bigint (R.TypeExpression "managed_shifted") Nothing \info _ -> do+ mode <- R.output (c_duckdb_v2_cast_function_exec_get_mode info)+ modifyIORef' modes (mode :)+ count <- R.output (c_duckdb_v2_cast_function_exec_get_row_count info)+ values <- R.output (c_duckdb_v2_cast_function_exec_get_input info) >>= readIntegers (fromIntegral count)+ result <- R.output (c_duckdb_v2_cast_function_exec_get_output info)+ writeIntegers result (map (+ 100) values)+ query_ connection "SELECT CAST(i::BIGINT AS managed_shifted)::BIGINT FROM range(3) t(i)" >>= (@?= map Only ([100, 101, 102] :: [Int64]))+ query_ connection "SELECT TRY_CAST(7::BIGINT AS managed_shifted)::BIGINT" >>= (@?= [Only (107 :: Int64)])+ seen <- readIORef modes+ assertBool "normal cast" (DuckDBV2CastModeNormal `elem` seen)+ assertBool "try cast" (DuckDBV2CastModeTry `elem` seen)++-- | A BIGINT of the custom type member_shifted, as a record field.+newtype Shifted = Shifted Int64++instance DuckValue Shifted where+ duckToField (Shifted value) = FieldInt64 value+ duckFromField = fmap Shifted . duckFromField+ duckLogicalType _ = LogicalTypeAlias "member_shifted" (LogicalTypeScalar LogicalTypeIdBigInt)++-- | A record with a field of the custom type member_shifted.+newtype Holder = Holder {held :: Shifted}+ deriving (Generic)+ deriving (ToField) via (ViaDuckDB Holder)++-- | A STRUCT value and a UNION value with a member of a registered custom type+-- bind again with no run of the registered cast from the base type, and so+-- does a record whose field declares that type. The library writes these+-- values with the default cast of DuckDB. See+-- Note [User casts and the casts of the library] in+-- "Database.DuckDB.Simple.Materialize".+testCastAliasMembers :: Assertion+testCastAliasMembers = withConnection ":memory:" \connection -> do+ R.registerCustomType connection "member_shifted" bigint+ calls <- newIORef (0 :: Int)+ R.registerCast connection bigint (R.TypeExpression "member_shifted") Nothing \info _ -> do+ modifyIORef' calls (+ 1)+ count <- R.output (c_duckdb_v2_cast_function_exec_get_row_count info)+ values <- R.output (c_duckdb_v2_cast_function_exec_get_input info) >>= readIntegers (fromIntegral count)+ result <- R.output (c_duckdb_v2_cast_function_exec_get_output info)+ writeIntegers result (map (+ 100) values)+ [(struct, union)] <-+ query_ connection "SELECT {'v': 7::BIGINT::member_shifted}, union_value(v := 7::BIGINT::member_shifted)"+ :: IO [(StructValue FieldValue, UnionValue FieldValue)]+ readIORef calls >>= assertBool "the query runs the registered cast" . (> 0)+ writeIORef calls 0+ query connection "SELECT (?).v::BIGINT, union_extract(?, 'v')::BIGINT, (?).held::BIGINT" (struct, union, Holder (Shifted 7))+ >>= (@?= [(107 :: Int64, 107 :: Int64, 7 :: Int64)])+ query connection "SELECT typeof(?), typeof(?), typeof(?)" (struct, union, Holder (Shifted 7))+ >>= (@?= [("STRUCT(v member_shifted)" :: Text, "UNION(v member_shifted)" :: Text, "STRUCT(held member_shifted)" :: Text)])+ readIORef calls >>= (@?= 0)++-- | A pair of types for 'testCastFailureCode': the setup, the source type, the+-- target type, a constant of the source type, and a column of the source type+-- over @i@. DuckDB plans a cast of the first two pairs as a cast that cannot+-- fail.+data CastPair = CastPair+ { castPairName :: String+ , castPairSetup :: Connection -> IO ()+ , castPairSource :: R.TypeDeclaration+ , castPairTarget :: Text+ , castPairConstant :: Text+ , castPairColumn :: Text+ }++-- | The pairs of 'testCastFailureCode' and 'testCastFailureCommit'.+castFailurePairs :: [CastPair]+castFailurePairs =+ [ CastPair "BIGINT to VARCHAR" (\_ -> pure ()) bigint "VARCHAR" "1::BIGINT" "i::BIGINT"+ , CastPair "INTEGER to BIGINT" (\_ -> pure ()) (R.TypeExpression "INTEGER") "BIGINT" "1::INTEGER" "i::INTEGER"+ , CastPair "BIGINT to a custom type" (\conn -> R.registerCustomType conn "cast_target" bigint) bigint "cast_target" "1::BIGINT" "i::BIGINT"+ ]++-- | Register a cast of the pair that writes NULL to each row and then raises+-- an exception with the text 'castFailure'.+registerFailingCast :: Connection -> CastPair -> IO ()+registerFailingCast conn castPair =+ R.registerCast conn (castPairSource castPair) (R.TypeExpression (castPairTarget castPair)) Nothing \info _ ->+ writeNullRows info >> throwIO (userError castFailure)++-- | The text of the exception of 'registerFailingCast'.+castFailure :: String+castFailure = "the cast refused the value"++-- | Fail a cast of a pair in a constant, in a column of 3000 rows, in an INSERT+-- with the assignment cast, and in an INSERT with CAST. Each failure gives+-- 'ErrorTypeMismatch' with the text of the exception. Each connection of the+-- database runs a query after the failure. TRY_CAST gives the NULL that the+-- callback wrote, and TRY(CAST) raises the error. See+-- Note [The error code of a failed cast] in "Database.DuckDB.Simple.V2.Registration".+testCastFailureCode :: CastPair -> Assertion+testCastFailureCode castPair = withEnvironment \environment ->+ forM_ ["constant", "column", "INSERT", "INSERT with CAST"] \form -> withCastDatabase environment castPair \instance' conn -> do+ let target = castPairTarget castPair+ constant = castPairConstant castPair+ label = castPairName castPair <> ", " <> form+ statement = case form of+ "constant" -> "SELECT CAST(" <> constant <> " AS " <> target <> ")"+ "column" -> "SELECT CAST(c AS " <> target <> ") FROM src"+ "INSERT" -> "INSERT INTO dst SELECT c FROM src"+ _ -> "INSERT INTO dst SELECT CAST(c AS " <> target <> ") FROM src"+ assertMismatch label (Text.pack castFailure) (void (query_ conn (Query statement) :: IO [Only FieldValue]))+ query_ conn "SELECT 42" >>= assertEqual label [Only (42 :: Int64)]+ withConnectionOn instance' \other -> query_ other "SELECT 42" >>= assertEqual label [Only (42 :: Int64)]+ when (form == "constant") do+ query_ conn (Query ("SELECT TRY_CAST(" <> constant <> " AS " <> target <> ") IS NULL")) >>= assertEqual label [Only True]+ assertMismatch+ label+ (Text.pack castFailure)+ (void (query_ conn (Query ("SELECT TRY(CAST(" <> constant <> " AS " <> target <> "))")) :: IO [Only FieldValue]))++-- | A query on another connection fails in the callback of a cast with the+-- type Conversion. The cast error has the type of each cast error.+testCastFailureQuery :: Assertion+testCastFailureQuery = withEnvironment \environment ->+ withCastInstance environment \instance' -> withConnectionOn instance' \conn -> withConnectionOn instance' \other -> do+ R.registerCast conn bigint (R.TypeExpression "VARCHAR") Nothing \_ _ ->+ void (query_ other "SELECT 'abc'::INTEGER" :: IO [Only Int64])+ assertMismatch+ "query"+ "Could not convert string 'abc' to INT32"+ (void (query_ conn "SELECT CAST(i AS VARCHAR) FROM range(10) t(i)" :: IO [Only Text]))+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ query_ other "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | An 'SQLError' of the type Conversion, Invalid Input, or Out of Range from+-- the callback of a cast from BIGINT to VARCHAR gives 'ErrorTypeMismatch', and+-- the database stays usable. DuckDB plans this cast as a cast that cannot fail.+-- With the code of the type of the error, DuckDB would raise an INTERNAL error+-- and invalidate the database. See Note [The error code of a failed cast] in+-- "Database.DuckDB.Simple.V2.Registration".+testCastTypedErrors :: Assertion+testCastTypedErrors = withEnvironment \environment -> forM_ [ErrorTypeConversion, ErrorInputInvalid, ErrorInputOutOfRange] \errorType ->+ withCastInstance environment \instance' -> withConnectionOn instance' \conn -> do+ let label = show errorType+ text = "the cast raised " <> Text.pack label+ void (execute_ conn "CREATE TABLE src AS SELECT i::BIGINT AS c FROM range(3000) t(i)")+ R.registerCast conn bigint (R.TypeExpression "VARCHAR") Nothing \_ _ ->+ throwIO SQLError{sqlErrorMessage = text, sqlErrorType = Just errorType, sqlErrorQuery = Nothing, sqlErrorBody = Just text}+ assertMismatch label text (void (query_ conn "SELECT CAST(c AS VARCHAR) FROM src" :: IO [Only Text]))+ query_ conn "SELECT 42" >>= assertEqual label [Only (42 :: Int64)]+ withConnectionOn instance' \other -> query_ other "SELECT 42" >>= assertEqual label [Only (42 :: Int64)]++-- | A COMMIT after a failed cast in an explicit transaction stores no row,+-- under each transaction invalidation policy. The last cast fails only in its+-- fifth call, the last chunk of 10000 rows, so that the INSERT writes rows+-- before the failure.+testCastFailureCommit :: Assertion+testCastFailureCommit = withEnvironment \environment -> forM_ [Nothing, Just "SYNTACTIC_ERRORS_DO_NOT_INVALIDATE"] \policy -> do+ let commitAfter label instance' conn failing = do+ forM_ (policy :: Maybe Text) \name -> forM_ ["default", "current"] \scope ->+ execute_ conn (Query ("SET " <> scope <> "_transaction_invalidation_policy = '" <> name <> "'"))+ void (execute_ conn "CREATE TABLE marker (m INTEGER)")+ void (execute_ conn "BEGIN TRANSACTION")+ void (execute_ conn "INSERT INTO marker VALUES (1)")+ assertMismatch label (Text.pack castFailure) (void (execute_ conn failing))+ void (try (execute_ conn "COMMIT") :: IO (Either SQLError Int))+ withConnectionOn instance' \other ->+ query_ other "SELECT (SELECT count(*) FROM marker) + (SELECT count(*) FROM dst)"+ >>= assertEqual (label <> ", " <> show policy) [Only (0 :: Int64)]+ forM_ castFailurePairs \castPair -> withCastDatabase environment castPair \instance' conn ->+ commitAfter (castPairName castPair) instance' conn (Query ("INSERT INTO dst SELECT CAST(c AS " <> castPairTarget castPair <> ") FROM src"))+ withCastInstance environment \instance' -> withConnectionOn instance' \conn -> do+ void (execute_ conn "CREATE TABLE src AS SELECT i::BIGINT AS c FROM range(10000) t(i)")+ void (execute_ conn "CREATE TABLE dst (x VARCHAR)")+ calls <- newIORef (0 :: Int)+ R.registerCast conn bigint (R.TypeExpression "VARCHAR") Nothing \info _ -> do+ writeNullRows info+ call <- atomicModifyIORef' calls \count -> (count + 1, count + 1)+ when (call >= 5) (throwIO (userError castFailure))+ commitAfter "BIGINT to VARCHAR over 10000 rows" instance' conn "INSERT INTO dst SELECT CAST(c AS VARCHAR) FROM src"++-- | Require an 'SQLError' of the type 'ErrorTypeMismatch' whose message+-- contains the text.+assertMismatch :: String -> Text -> IO () -> Assertion+assertMismatch label text action =+ try action >>= \case+ Left (err :: SQLError) -> do+ assertEqual label (Just ErrorTypeMismatch) (sqlErrorType err)+ assertBool (label <> ": " <> Text.unpack (sqlErrorMessage err)) (text `Text.isInfixOf` sqlErrorMessage err)+ Right () -> assertFailure (label <> ": expected an SQL error")++-- | Run an action with a new instance that has an in-memory database.+withCastInstance :: Environment -> (Instance -> IO a) -> IO a+withCastInstance environment action = withInstance environment [] \instance' -> do+ attachDatabase instance' ":memory:" "cast_failure" []+ setDefaultDatabase instance' "cast_failure"+ action instance'++-- | Run an action with a new instance, the tables src and dst of a pair, and+-- the failing cast of the pair.+withCastDatabase :: Environment -> CastPair -> (Instance -> Connection -> IO a) -> IO a+withCastDatabase environment castPair action = withCastInstance environment \instance' -> withConnectionOn instance' \conn -> do+ castPairSetup castPair conn+ void (execute_ conn (Query ("CREATE TABLE src AS SELECT " <> castPairColumn castPair <> " AS c FROM range(3000) t(i)")))+ void (execute_ conn (Query ("CREATE TABLE dst (x " <> castPairTarget castPair <> ")")))+ registerFailingCast conn castPair+ action instance' conn++-- | A second registration of a cast for the same pair of types returns with no+-- error, and CAST still runs the callback of the first registration. DuckDB+-- 2.0.0 keeps the first cast for a pair.+testCastSecond :: Assertion+testCastSecond = withConnection ":memory:" \connection -> do+ R.registerCustomType connection "managed_kept" bigint+ let shift amount info _ = do+ count <- R.output (c_duckdb_v2_cast_function_exec_get_row_count info)+ values <- R.output (c_duckdb_v2_cast_function_exec_get_input info) >>= readIntegers (fromIntegral count)+ result <- R.output (c_duckdb_v2_cast_function_exec_get_output info)+ writeIntegers result (map (+ amount) values)+ R.registerCast connection bigint (R.TypeExpression "managed_kept") Nothing (shift 100)+ R.registerCast connection bigint (R.TypeExpression "managed_kept") Nothing (shift 200)+ query_ connection "SELECT CAST(i::BIGINT AS managed_kept)::BIGINT FROM range(3) t(i)" >>= (@?= map Only ([100, 101, 102] :: [Int64]))++replacement :: Int64 -> R.ContextCallback DuckDBV2ReplacementScanInfoHandle+replacement number info _ =+ bracket (R.output (c_duckdb_v2_replacement_scan_get_name info)) (R.destroy c_duckdb_v2_qname_destroy) \name -> do+ count <- R.output (c_duckdb_v2_qname_get_part_count name)+ when (count == 1) do+ part <- R.output (c_duckdb_v2_qname_get_part name 0) >>= R.readIdentifier+ when (part == "managed_missing")+ $ R.withText ("SELECT " <> Text.pack (show number) <> "::BIGINT AS amount")+ $ R.checked . c_duckdb_v2_replacement_scan_set_subquery info++testReplacement :: Assertion+testReplacement = withEnvironment \environment -> withInstance environment [] \instance' -> do+ R.registerInstanceReplacementScan instance' (replacement 7)+ withConnectionOn instance' \first -> withConnectionOn instance' \second -> do+ R.registerReplacementScan first (replacement 42)+ query_ first "SELECT amount FROM managed_missing" >>= (@?= [Only (42 :: Int64)])+ query_ second "SELECT amount FROM managed_missing" >>= (@?= [Only (7 :: Int64)])++testContextLog :: Assertion+testContextLog = do+ entries <- newIORef []+ withConnection ":memory:" \connection -> do+ Logging.registerLogSink connection "context_log" (\entry -> atomicModifyIORef' entries (\values -> (entry : values, ()))) (pure ())+ let logged info context = do+ V2Logging.logMessage context Logging.LogLevelWarning "HaskellContext" "from a callback"+ count <- R.output (c_duckdb_v2_scalar_function_exec_get_row_count info)+ vector <- R.output (c_duckdb_v2_scalar_function_exec_get_result info)+ writeIntegers vector (replicate (fromIntegral count) 1)+ R.registerScalarFunction connection "managed_log" (scalarSignature []) [] (R.scalarCallbacks logged)+ void $ execute_ connection "CALL enable_logging(); SET logging_storage = 'context_log'"+ query_ connection "SELECT managed_log()" >>= (@?= [Only (1 :: Int64)])+ void $ execute_ connection "SET logging_storage = 'memory'"+ received <- filter ((== "HaskellContext") . Logging.logEntryType) <$> readIORef entries+ map Logging.logEntryMessage received @?= ["from a callback"]+ map Logging.logEntryScope received @?= [Logging.LogScopeConnection]++testErrors :: Assertion+testErrors = withConnection ":memory:" \connection -> do+ let failure = throwIO (userError "managed callback failure")+ R.registerScalarFunction connection "managed_error" (scalarSignature []) [] (R.scalarCallbacks (\_ _ -> failure))+ assertSqlError (query_ connection "SELECT managed_error()" :: IO [Only Int64])+ R.registerAggregateFunction+ connection+ "managed_error_sum"+ (scalarSignature [R.parameter "value" bigint])+ []+ sumCallbacks{R.aggregateUpdateCallback = \_ -> failure}+ assertSqlError (query_ connection "SELECT managed_error_sum(i::BIGINT) FROM range(3) t(i)" :: IO [Only Int64])+ R.registerTableFunction connection "managed_error_table" [] (R.tableCallbacks (\_ _ _ -> failure) (\_ _ -> pure ()))+ assertSqlError (query_ connection "SELECT * FROM managed_error_table()" :: IO [Only Int64])+ R.registerReplacementScan connection (\_ _ -> failure)+ assertSqlError (query_ connection "SELECT * FROM managed_unknown" :: IO [Only Int64])++-- | A raw exec callback writes an error text with the byte 0xff, which is not+-- valid UTF-8. The query must fail with an SQL error that can be shown.+testInvalidErrorText :: Assertion+testInvalidErrorText =+ bracket (toFunPtr (DuckDBV2ScalarFunctionExecCallbackFn_Aux invalidText)) freeHaskellFunPtr \exec ->+ withConnection ":memory:" \connection -> do+ withConnectionHandle connection \native -> withConnectionFactory native \factory ->+ bracket (R.output (c_duckdb_v2_scalar_function_create factory)) (R.destroy c_duckdb_v2_scalar_function_destroy) \function -> do+ R.withIdentifier "invalid_error_text" (R.checked . c_duckdb_v2_scalar_function_set_name function)+ R.output (c_duckdb_v2_scalar_function_get_signature function) >>= R.configureFunctionSignature factory (scalarSignature [])+ R.checked (c_duckdb_v2_scalar_function_set_exec_callback function (DuckDBV2ScalarFunctionExecCallbackFn exec))+ R.checked (c_duckdb_v2_connection_register_scalar_function native function)+ outcome <- try (query_ connection "SELECT invalid_error_text()" :: IO [Only Int64])+ case outcome of+ Left (err :: SQLError) -> do+ assertBool "replacement character" ("bad \xFFFD message" `Text.isInfixOf` sqlErrorMessage err)+ sqlErrorQuery err @?= Just "SELECT invalid_error_text()"+ void (evaluate (length (show err)))+ Right _ -> assertFailure "expected SQL error"+ where+ invalidText _ _ err = do+ info <- peek err+ void (c_duckdb_v2_error_info_set_code info DuckDBV2ErrorInputInvalid)+ withBytes "bad \xff message" (void . c_duckdb_v2_error_info_set_text info)++testRelease :: Assertion+testRelease = do+ references <- withConnection ":memory:" \connection -> do+ closure <- newIORef (42 :: Int64)+ closureWeak <- mkWeakIORef closure (pure ())+ states <- newIORef []+ let initialize info _ = do+ state <- newIORef (1 :: Int64)+ weak <- mkWeakIORef state (pure ())+ modifyIORef' states (weak :)+ R.setCallbackState (c_duckdb_v2_scalar_function_init_set_init_data info) state+ execute info _ = do+ state <- R.getCallbackState (c_duckdb_v2_scalar_function_exec_get_init_data info) :: IO (IORef Int64)+ increment <- readIORef state+ value <- (+ increment) <$> readIORef closure+ result <- R.output (c_duckdb_v2_scalar_function_exec_get_result info)+ writeIntegers result [value]+ R.registerScalarFunction+ connection+ "managed_release"+ (scalarSignature [])+ []+ (R.scalarCallbacks execute){R.scalarInitCallback = Just initialize}+ query_ connection "SELECT managed_release()" >>= (@?= [Only (43 :: Int64)])+ stateRefs <- readIORef states+ assertBool "state initialized" (not (null stateRefs))+ pragmaClosure <- newIORef (7 :: Int64)+ pragmaWeak <- mkWeakIORef pragmaClosure (pure ())+ Pragma.registerPragmaFunctionOn+ (R.ConnectionTarget connection)+ "managed_pragma_release"+ []+ (Pragma.ExecPragma (\_ _ -> void (readIORef pragmaClosure)))+ void (execute_ connection "PRAGMA managed_pragma_release")+ failed <- failedRegistration connection+ pure (closureWeak : pragmaWeak : failed : stateRefs)+ performMajorGC+ performMajorGC+ forM_ references \reference -> deRefWeak reference >>= assertBool "native callback state or closure remains live" . isNothing++failedRegistration :: Connection -> IO (Weak (IORef Int64))+failedRegistration connection = do+ retained <- newIORef (1 :: Int64)+ reference <- mkWeakIORef retained (pure ())+ let execute _ _ = void (readIORef retained)+ duplicate = R.parameter "duplicate" bigint+ result <-+ try (R.registerScalarFunction connection "managed_invalid" (scalarSignature [duplicate, duplicate]) [] (R.scalarCallbacks execute))+ case result of+ Left (_ :: SQLError) -> pure ()+ Right () -> assertFailure "duplicate parameter names must fail"+ pure reference++-- | Require an 'SQLError' with the message of a failed managed callback.+assertSqlError :: IO a -> Assertion+assertSqlError = assertSqlErrorContaining "managed callback failure"++readIntegers :: Int -> DuckDBV2VectorHandle -> IO [Int64]+readIntegers count vector = do+ R.checked (c_duckdb_v2_vector_flatten vector)+ raw <- R.output (c_duckdb_v2_vector_get_data_mutable vector)+ mapM (peekElemOff (castPtr raw)) [0 .. count - 1]
+ test/SizedArrayTests.hs view
@@ -0,0 +1,209 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE DeriveGeneric #-}+{-# LANGUAGE DerivingVia #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- | Cases of the sized ARRAY type 'SizedArray'. See Note [Sized arrays] in+-- "Database.DuckDB.Simple.ToField".+module SizedArrayTests (sizedArrayTests) where++import Control.Exception (try)+import Control.Monad (forM_)+import Data.Array (bounds, elems, listArray, (!))+import Data.Foldable (toList)+import Data.Int (Int32, Int64, Int8)+import qualified Data.Map.Strict as Map+import Data.Maybe (fromJust, isNothing)+import Data.Proxy (Proxy (..))+import qualified Data.Text as Text+import Database.DuckDB.Simple+import Database.DuckDB.Simple.FromField (FieldValue (..))+import Database.DuckDB.Simple.Generic (DuckValue (..), ViaDuckDB (..))+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), TypeDeclaration (..))+import Database.DuckDB.Simple.Ok (Ok (..))+import Database.DuckDB.Simple.ToField (ToDuckValue)+import qualified Database.DuckDB.Simple.V2.Registration as R+import GHC.Generics (Generic)+import Test.Tasty (TestTree, testGroup)+import Test.Tasty.HUnit (assertBool, assertFailure, testCase, (@?=))+import TestUtils (assertInputErrorContaining, assertReusable, assertSqlErrorContaining, assertSqlErrorType)++-- | A record with a field of a sized ARRAY type.+newtype Sized = Sized {sizedValues :: SizedArray 3 Int64}+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, ToDuckValue, FromField) via (ViaDuckDB Sized)++-- | A typed NULL must not allow an unsupported ARRAY size in a generic field.+newtype Above = Above {aboveValues :: Maybe (SizedArray 100001 Int64)}+ deriving stock (Show, Generic)+ deriving (DuckDBColumnType, ToField, ToDuckValue, FromField) via (ViaDuckDB Above)++-- | This size would wrap to one element if narrowed directly to Word64.+newtype Overflow = Overflow {overflowValues :: Maybe (SizedArray 18446744073709551617 Int64)}+ deriving stock (Show, Generic)+ deriving (DuckDBColumnType, ToField, ToDuckValue, FromField) via (ViaDuckDB Overflow)++-- | The generic declaration must check ARRAY sizes below each container.+newtype NestedAbove = NestedAbove {nestedValues :: [Map.Map Text.Text (Maybe (SizedArray 100001 Int64))]}+ deriving stock (Show, Generic)+ deriving (DuckDBColumnType, ToField, ToDuckValue, FromField) via (ViaDuckDB NestedAbove)++-- | The cases of 'SizedArray' as a parameter, a column, and a function argument.+sizedArrayTests :: TestTree+sizedArrayTests =+ testGroup+ "sized arrays"+ [ testCase "a SizedArray rejects a longer list without forcing its tail or elements" $ do+ assertBool "an infinite list is too long" (isNothing (sizedArrayFromList @3 ([1 ..] :: [Int64])))+ assertBool+ "a fourth element is enough to reject the list"+ (isNothing (sizedArrayFromList @3 ([undefined, undefined, undefined, undefined] ++ undefined :: [Int64])))+ , testCase "a SizedArray binds as an ARRAY of its size and reads back from a parameter and a column" $ withConnection ":memory:" \conn -> do+ let values = fromJust (sizedArrayFromList @3 [1, 2, 3 :: Int64])+ toList values @?= [1, 2, 3]+ sizedArrayFromList @3 [1, 2 :: Int64] @?= Nothing+ -- An Array with any bounds gives the bounds (0, n - 1) and the same+ -- elements. An Array of another count gives Nothing.+ let shifted = listArray (5, 7) [1, 2, 3 :: Int64]+ fromShifted = fromJust (sizedArray @3 shifted)+ bounds (sizedArrayToArray fromShifted) @?= (0, 2)+ elems (sizedArrayToArray fromShifted) @?= elems shifted+ fromShifted @?= values+ show values @?= "SizedArray (array (0,2) [(0,1),(1,2),(2,3)])"+ -- The count is checked before the Array is built, for an Array and a+ -- list of fewer and of more values.+ sizedArray @3 (listArray (0, 1) [1, 2 :: Int64]) @?= Nothing+ sizedArray @3 (listArray (0, 3) [1, 2, 3, 4 :: Int64]) @?= Nothing+ sizedArrayFromList @3 [1, 2, 3, 4 :: Int64] @?= Nothing+ -- Size 0 has the bounds (0, -1). A size above the range of Int gives+ -- Nothing, and no size wraps into the count of the values.+ bounds (sizedArrayToArray (fromJust (sizedArrayFromList @0 ([] :: [Int64])))) @?= (0, -1)+ sizedArrayFromList @9223372036854775808 (error "the list spine must not be read" :: [Int64]) @?= Nothing+ sizedArrayFromList @18446744073709551619 [1, 2, 3 :: Int64] @?= Nothing+ sizedArray @18446744073709551619 (listArray (0, 2) [1, 2, 3 :: Int64]) @?= Nothing+ query conn "SELECT typeof(?), ?" (values, values) >>= (@?= [("BIGINT[3]" :: Text.Text, values)])+ _ <- execute_ conn "CREATE TABLE stored (v BIGINT[3])"+ _ <- execute conn "INSERT INTO stored VALUES (?)" (Only values)+ [Only stored] <- query_ conn "SELECT v FROM stored"+ stored @?= values+ -- A value that is read has the bounds (0, n - 1).+ bounds (sizedArrayToArray stored) @?= (0, 2)+ sizedArrayToArray stored ! 1 @?= (2 :: Int64)+ -- A LIST of three elements reads, as the cast to BIGINT[3] does.+ query_ conn "SELECT [1, 2, 3]::BIGINT[]" >>= (@?= [Only values])+ , testCase "a SizedArray raises an error for an ARRAY or a LIST of another length" $ withConnection ":memory:" \conn -> do+ let refused expression sqlName count =+ try (query_ conn (Query ("SELECT " <> expression <> " AS v")) :: IO [Only (SizedArray 3 Int64)]) >>= \case+ Left SQLError{sqlErrorMessage} ->+ sqlErrorMessage+ @?= "duckdb-simple: column v has type " <> sqlName <> " but expected SizedArray 3 Int64: the value has " <> count <> " elements"+ Right rows -> assertFailure ("expected Incompatible for " <> Text.unpack expression <> ", got " <> show rows)+ refused "[1, 2]::BIGINT[2]" "BIGINT[2]" "2"+ refused "[1, 2, 3, 4]::BIGINT[4]" "BIGINT[4]" "4"+ refused "[1, 2]::BIGINT[]" "BIGINT[]" "2"+ , testCase "a nested SizedArray and a SizedArray of Maybe elements round-trip" $ withConnection ":memory:" \conn -> do+ let row = fromJust . sizedArrayFromList @3+ nested = fromJust (sizedArrayFromList @2 [row [1, 2, 3 :: Int32], row [4, 5, 6]])+ query conn "SELECT typeof(?), ?" (nested, nested) >>= (@?= [("INTEGER[3][2]" :: Text.Text, nested)])+ let optional = fromJust (sizedArrayFromList @3 [Just 1, Nothing, Just (3 :: Int64)])+ query conn "SELECT typeof(?), ?" (optional, optional) >>= (@?= [("BIGINT[3]" :: Text.Text, optional)])+ , testCase "a SizedArray of size 0 raises an error at the bind, declaration, and read" $ withConnection ":memory:" \conn -> do+ let empty = fromJust (sizedArrayFromList @0 ([] :: [Int64]))+ assertInputErrorContaining "an empty ARRAY parameter is not possible" (query conn "SELECT ?" (Only empty) :: IO [Only (SizedArray 0 Int64)])+ assertInputErrorContaining "a SizedArray of size 0 has no ARRAY type" $+ createFunction conn "sized_empty" \(_ :: SizedArray 0 Int64) -> (1 :: Int64)+ assertSqlErrorContaining "a DuckDB ARRAY has at least one element" $+ (query_ conn "SELECT []::BIGINT[]" :: IO [Only (SizedArray 0 Int64)])+ assertSqlErrorContaining "a DuckDB ARRAY has at least one element" $+ (query_ conn "SELECT [1]::BIGINT[1]" :: IO [Only (SizedArray 0 Int64)])+ assertSqlErrorType ErrorQueryBinder $+ (query_ conn "SELECT []::BIGINT[0]" :: IO [Only Int64])+ runConversion+ (duckFromTypedField (LogicalTypeList (LogicalTypeScalar LogicalTypeIdBigInt)) (FieldList []) :: Conversion (SizedArray 0 Int64))+ >>= \case+ Errors errors -> assertBool (show errors) ("a DuckDB ARRAY has at least one element" `Text.isInfixOf` Text.pack (show errors))+ Ok _ -> assertFailure "expected the generic size error"+ (query_ conn "SELECT NULL::BIGINT[]" :: IO [Only (Maybe (SizedArray 0 Int64))]) >>= (@?= [Only Nothing])+ assertReusable conn+ , testCase "the largest ARRAY size binds and declares, and a larger size fails at the bind, declaration, and read" $ withConnection ":memory:" \conn -> do+ let largest = fromJust (sizedArrayFromList @100000 (replicate 100000 (1 :: Int8)))+ above = fromJust (sizedArrayFromList @100001 (replicate 100001 (1 :: Int8)))+ query conn "SELECT typeof(?), len(?)" (largest, largest) >>= (@?= [("TINYINT[100000]" :: Text.Text, 100000 :: Int64)])+ createFunction conn "sized_largest" \(v :: SizedArray 100000 Int8) -> length v+ query conn "SELECT sized_largest(?)" (Only largest) >>= (@?= [Only (100000 :: Int)])+ -- The value API of DuckDB would bind the larger value, but its SQL has no such type.+ assertInputErrorContaining+ "an ARRAY parameter of 100001 elements is not possible, because a DuckDB ARRAY has at most 100000 elements"+ (query conn "SELECT len(?)" (Only above) :: IO [Only Int64])+ assertInputErrorContaining "a SizedArray of size 100001 has no ARRAY type, because a DuckDB ARRAY has at most 100000 elements" $+ createFunction conn "sized_above" \(v :: SizedArray 100001 Int8) -> length v+ assertSqlErrorContaining "a DuckDB ARRAY has at most 100000 elements" $+ (query_ conn "SELECT range(100001)::BIGINT[]" :: IO [Only (SizedArray 100001 Int64)])+ assertSqlErrorType ErrorQueryBinder $+ (query_ conn "SELECT range(100001)::BIGINT[100001]" :: IO [Only Int64])+ assertSqlErrorContaining "a DuckDB ARRAY has at most 100000 elements" $+ (query_ conn "SELECT {'aboveValues': range(100001)::BIGINT[]}" :: IO [Only Above])+ assertSqlErrorContaining "a DuckDB ARRAY has at most 100000 elements" $+ (query_ conn "SELECT {'aboveValues': [1]::BIGINT[1]}" :: IO [Only Above])+ [Only (Above optional)] <- query_ conn "SELECT {'aboveValues': NULL::BIGINT[]}"+ optional @?= Nothing+ assertReusable conn+ , testCase "generic ARRAY bounds are checked for NULL, nested containers, and a Nat above Word64" $ withConnection ":memory:" \conn -> do+ assertInputErrorContaining "a DuckDB ARRAY has at most 100000 elements" (query conn "SELECT ?" (Only (Above Nothing)) :: IO [Only Above])+ assertInputErrorContaining+ "a DuckDB ARRAY has at most 100000 elements"+ (query conn "SELECT ?" (Only (Overflow Nothing)) :: IO [Only Overflow])+ let above = Above (Just (fromJust (sizedArrayFromList @100001 (replicate 100001 (1 :: Int64)))))+ assertInputErrorContaining "an ARRAY parameter of 100001 elements is not possible" (query conn "SELECT ?" (Only above) :: IO [Only Above])+ assertInputErrorContaining "a DuckDB ARRAY must have 1 to 100000 elements" $+ createFunction conn "generic_above" \(_ :: Above) -> (1 :: Int64)+ assertInputErrorContaining "a DuckDB ARRAY must have 1 to 100000 elements" $+ createFunction conn "generic_overflow" \(_ :: Overflow) -> (1 :: Int64)+ assertInputErrorContaining "a DuckDB ARRAY must have 1 to 100000 elements" $+ createFunction conn "generic_nested_above" \(_ :: NestedAbove) -> (1 :: Int64)+ assertInputErrorContaining "a DuckDB ARRAY has at most 100000 elements" $+ R.registerCustomType+ conn+ "invalid_array"+ (StructuredType (LogicalTypeAlias "invalid_array_base" (LogicalTypeArray (LogicalTypeScalar LogicalTypeIdBigInt) 100001)))+ case duckLogicalType (Proxy @(SizedArray 18446744073709551617 Int64)) of+ LogicalTypeArray _ size -> size @?= maxBound+ logical -> assertFailure ("expected an invalid ARRAY size sentinel, got " <> show logical)+ , testCase "a list goes into an ARRAY column of its length and reads an ARRAY of each size" $ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE fixed (v BIGINT[3])"+ _ <- execute conn "INSERT INTO fixed VALUES (?)" (Only [1, 2, 3 :: Int64])+ query conn "SELECT count(*) FROM fixed WHERE v = ?" (Only [1, 2, 3 :: Int64]) >>= (@?= [Only (1 :: Int64)])+ -- DuckDB casts a list of another length to the column with an error.+ try (execute conn "INSERT INTO fixed VALUES (?)" (Only [1, 2 :: Int64])) >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ sqlErrorType @?= Just ErrorTypeConversion+ assertBool (Text.unpack sqlErrorMessage) ("Cannot cast list with length 2 to array with length 3" `Text.isInfixOf` sqlErrorMessage)+ Right count -> assertFailure ("expected the cast error of DuckDB, got " <> show count)+ query_ conn "SELECT [1, 2]::BIGINT[2], [1, 2, 3]::BIGINT[3]" >>= (@?= [([1, 2], [1, 2, 3]) :: ([Int64], [Int64])])+ , testCase "a FieldArray binds an ARRAY of a size that the program gives at run time" $ withConnection ":memory:" \conn ->+ forM_ [1, 4, 7] \count -> do+ let value = FieldArray (listArray (0, count - 1) (map FieldInt64 [1 .. fromIntegral count]))+ query conn "SELECT typeof(?), len(?)" (value, value) >>= (@?= [(Text.pack ("BIGINT[" <> show count <> "]"), count)])+ , testCase "a SizedArray field of a record declares and reads the ARRAY of its size" $ withConnection ":memory:" \conn -> do+ let row = Sized (fromJust (sizedArrayFromList @3 [1, 2, 3]))+ query conn "SELECT typeof(?), ?" (row, row) >>= (@?= [("STRUCT(sizedValues BIGINT[3])" :: Text.Text, row)])+ createFunction conn "sized_field" \(Sized sized) -> sum sized+ query conn "SELECT typeof(sized_field(?)), sized_field(?)" (row, row) >>= (@?= [("BIGINT" :: Text.Text, 6 :: Int64)])+ try (query_ conn "SELECT {'sizedValues': [1, 2]::BIGINT[2]}" :: IO [Only Sized]) >>= \case+ Left SQLError{sqlErrorMessage} ->+ assertBool+ (Text.unpack sqlErrorMessage)+ ("has type BIGINT[2] but expected SizedArray 3: the value has 2 elements" `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure ("expected the size error, got " <> show rows)+ try (query_ conn "SELECT {'sizedValues': ['x', 'y']::VARCHAR[]}" :: IO [Only Sized]) >>= \case+ Left SQLError{sqlErrorMessage} ->+ assertBool+ (Text.unpack sqlErrorMessage)+ ("has type VARCHAR[] but expected SizedArray 3: the value has 2 elements" `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure ("expected the size error before an element error, got " <> show rows)+ ]
test/Spec.hs view
@@ -1,1258 +1,1733 @@ {-# LANGUAGE BlockArguments #-}-{-# LANGUAGE DeriveAnyClass #-}-{-# LANGUAGE DeriveGeneric #-}-{-# LANGUAGE DerivingVia #-}-{-# LANGUAGE LambdaCase #-}-{-# LANGUAGE NamedFieldPuns #-}-{-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE PatternSynonyms #-}-{-# LANGUAGE ScopedTypeVariables #-}-{-# LANGUAGE TypeOperators #-}---- | Tasty-based test suite for duckdb-simple.-module Main (main) where--import ArrowTests (arrowTests)-import CancellationTests (cancellationTests)-import Control.Applicative ((<|>))-import Control.Exception (ErrorCall, Exception, SomeException, displayException, fromException, try)-import Control.Monad (forM_, replicateM_, when)-import CoreRegressionTests (coreRegressionTests)-import Data.Array (Array, elems, listArray)-import qualified Data.ByteString as BS-import Data.IORef (atomicModifyIORef', newIORef, readIORef)-import Data.Int (Int16, Int32, Int64, Int8)-import Data.List (sortOn)-import qualified Data.Map.Strict as Map-import Data.Maybe (fromJust)-import Data.Proxy (Proxy (..))-import Data.Ratio ((%))-import Data.String (fromString)-import qualified Data.Text as Text-import Data.Time.Calendar (fromGregorian)-import Data.Time.Clock (UTCTime (..), secondsToDiffTime)-import Data.Time.LocalTime (- LocalTime (..),- TimeOfDay (..),- localTimeToUTC,- minutesToTimeZone,- timeOfDayToTime,- utc,- )-import qualified Data.UUID as UUID-import Data.Word (Word16, Word32, Word64, Word8)-import Database.DuckDB.FFI (- pattern DuckDBCatalogEntryTypeTable,- pattern DuckDBFileFlagCreate,- pattern DuckDBFileFlagRead,- pattern DuckDBFileFlagWrite,- )-import Database.DuckDB.Simple-import qualified Database.DuckDB.Simple.Catalog as Catalog-import qualified Database.DuckDB.Simple.Config as Config-import qualified Database.DuckDB.Simple.Copy as Copy-import qualified Database.DuckDB.Simple.FileSystem as FileSystem-import Database.DuckDB.Simple.FromField (- BigNum (..),- BitString (..),- DecimalValue (..),- Field (..),- FieldValue (..),- IntervalValue (..),- TimeWithZone (..),- bsFromBool,- fromBigNumBytes,- returnError,- toBigNumBytes,- )-import Database.DuckDB.Simple.Generic (- ViaDuckDB (..),- genericFromFieldValue,- genericToFieldValue,- genericToStructValue,- )-import qualified Database.DuckDB.Simple.Logging as Logging-import Database.DuckDB.Simple.LogicalRep (- StructField (..),- StructValue (..),- UnionMemberType (..),- UnionValue (..),- )-import Database.DuckDB.Simple.Ok (Ok (..))-import Database.DuckDB.Simple.Time (Unbounded (..))-import ExtensionRegressionTests (extensionRegressionTests)-import GHC.Generics (Generic)-import qualified GeometryRegressionTests-import Numeric.Natural (Natural)-import Properties (roundTripTests)-import StreamingTests (nativeStreamingTests)-import System.Directory (doesFileExist, removeFile)-import Test.Tasty (TestTree, defaultMain, testGroup)-import Test.Tasty.ExpectedFailure (expectFailBecause)-import Test.Tasty.HUnit-import Test.Tasty.QuickCheck (testProperty, (===))-import TimeTests (timeTests)-import ValueRegressionTests (valueRegressionTests)-import qualified VariantRegressionTests--data Person = Person- { personId :: Int- , personName :: Text.Text- }- deriving stock (Eq, Show, Generic)- deriving anyclass (FromRow)--data WithRemaining = WithRemaining Int Int- deriving (Eq, Show)--data GenericRecord = GenericRecord- { grId :: Int- , grName :: Text.Text- }- deriving stock (Eq, Show, Generic)--data GenericSum- = SumInt Int- | SumBool Bool- | SumPair Int Text.Text- deriving stock (Eq, Show, Generic)--data GenericEnum- = EnumRed- | EnumBlue- deriving stock (Eq, Show, Generic)--data ViaUser = ViaUser- { vuId :: Int64- , vuName :: Text.Text- }- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB ViaUser)--data ViaShape- = ViaCircle Double- | ViaSquare Double- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB ViaShape)---- Edge case test types-data TwoFieldStruct = TwoFieldStruct Int Text.Text- deriving stock (Eq, Show, Generic)--data LargeSumType- = LargeC1 Int- | LargeC2 Int- | LargeC3 Int- | LargeC4 Int- | LargeC5 Int- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB LargeSumType)--data EmptyStruct = EmptyStruct- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB EmptyStruct)--data OptionalFields = OptionalFields- { ofInt :: Maybe Int- , ofText :: Maybe Text.Text- }- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB OptionalFields)--instance FromRow WithRemaining where- fromRow = do- firstVal <- field- remaining <- numFieldsRemaining- replicateM_ remaining (fieldWith (const (Ok ())))- pure (WithRemaining firstVal remaining)--newtype YesNo = YesNo Bool- deriving (Eq, Show)--instance FromRow YesNo where- fromRow = parseYes <|> parseNo- where- parseYes = YesNo True <$ fieldWith (match (Text.pack "yes"))- parseNo = YesNo False <$ fieldWith (match (Text.pack "no"))-- match expected fld@Field{} =- case fromField fld :: Ok Text.Text of- Errors err -> Errors err- Ok txt ->- let normalized = Text.toLower txt- in if normalized == expected- then Ok ()- else returnError ConversionFailed fld "failed to match exact string"--newtype NonEmptyText = NonEmptyText Text.Text- deriving (Eq, Show)--nonEmptyTextParser :: FieldParser NonEmptyText-nonEmptyTextParser f@Field{} =- case fromField f of- Errors err -> Errors err- Ok txt- | Text.null txt ->- returnError ConversionFailed f "NonEmptyText requires a non-empty string"- | otherwise -> Ok (NonEmptyText txt)--instance FromField NonEmptyText where- fromField = nonEmptyTextParser-main :: IO ()-main = defaultMain tests--tests :: TestTree-tests =- testGroup- "duckdb-simple"- [ connectionTests- , coreRegressionTests- , arrowTests- , cancellationTests- , nativeStreamingTests- , extensionRegressionTests- , GeometryRegressionTests.tests- , VariantRegressionTests.tests- , valueRegressionTests- , timeTests- , withConnectionTests- , statementTests- , roundTripTests- , typeTests- , streamingTests- , functionsTests- , v15Tests- , transactionTests- ]--connectionTests :: TestTree-connectionTests =- testGroup- "open/close"- [ testCase "opens and closes an in-memory database" $ do- conn <- open ":memory:"- close conn- , testCase "allows closing the same connection twice" $ do- conn <- open ":memory:"- close conn- close conn- , testCase "opens with startup config" $ do- conn <- openWithConfig ":memory:" [("threads", "1")]- mValue <- Config.getConfigOption conn "threads"- close conn- case mValue of- Nothing -> assertFailure "expected threads config value"- Just value -> Config.configValueText value @?= "1"- ]--v15Tests :: TestTree-v15Tests =- testGroup- "DuckDB 1.5 wrappers"- [ testCase "lists config flags" $ do- flags <- Config.listConfigFlags- assertBool "expected at least one config flag" (not (null flags))- assertBool "expected threads flag" (any ((== "threads") . Config.configFlagName) flags)- , testCase "catalog lookup works inside a transaction" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE catalog_probe(id INTEGER)"- withTransaction conn $ do- mType <- Catalog.catalogTypeName conn "memory"- assertBool "expected catalog type name" (mType /= Nothing)- mEntry <- Catalog.lookupCatalogEntry conn "memory" "main" "catalog_probe" DuckDBCatalogEntryTypeTable- case mEntry of- Nothing -> assertFailure "expected catalog entry"- Just entry -> Catalog.catalogEntryName entry @?= "catalog_probe"- , testCase "file system wrappers round-trip bytes" $ do- let path = "/tmp/duckdb-simple-filesystem.bin"- payload = BS.pack [0, 1, 2, 3, 255]- exists <- doesFileExist path- when exists (removeFile path)- withConnection ":memory:" \conn -> do- FileSystem.withFileHandle conn path [DuckDBFileFlagCreate, DuckDBFileFlagWrite] \handle -> do- written <- FileSystem.writeFileHandleBytes handle payload- written @?= fromIntegral (BS.length payload)- FileSystem.fileHandleSync handle- FileSystem.withFileHandle conn path [DuckDBFileFlagRead] \handle -> do- size <- FileSystem.fileHandleSize handle- size @?= fromIntegral (BS.length payload)- bytes <- FileSystem.readFileHandleChunk handle 32- bytes @?= payload- removeFile path- , testCase "copy wrappers receive COPY TO rows" $- withConnection ":memory:" \conn -> do- capturedRef <- newIORef ([] :: [Int])- pathRef <- newIORef Nothing- Copy.registerCopyToFunction- conn- "hs_capture"- (\bindInfo -> pure (length (Copy.copyBindColumnTypes bindInfo)))- ( \initInfo -> do- atomicModifyIORef' pathRef \_ -> (Just (Copy.copyInitFilePath initInfo), ())- pure (Copy.copyInitBindState initInfo)- )- ( \sinkInfo rows -> do- Copy.copySinkBindState sinkInfo @?= 1- Copy.copySinkGlobalState sinkInfo @?= 1- values <- mapM fieldInt rows- atomicModifyIORef' capturedRef \existing -> (existing <> values, ())- )- ( \finalizeInfo -> do- Copy.copyFinalizeBindState finalizeInfo @?= 1- Copy.copyFinalizeGlobalState finalizeInfo @?= 1- )- _ <- execute_ conn "COPY (SELECT * FROM (VALUES (1), (2), (3)) AS t(x)) TO '/tmp/duckdb-simple-copy.out' (FORMAT hs_capture)"- captured <- readIORef capturedRef- filePath <- readIORef pathRef- captured @?= [1, 2, 3]- filePath @?= Just "/tmp/duckdb-simple-copy.out"- , testCase "logging wrapper registers a log sink" $- withConnection ":memory:" \conn -> do- entriesRef <- newIORef ([] :: [Logging.LogEntry])- Logging.registerLogStorage conn "hs_log_capture" \entry ->- atomicModifyIORef' entriesRef \entries -> (entry : entries, ())- _ <- execute_ conn "SELECT write_log('duckdb-simple logging wrapper', level := 'INFO', scope := 'connection')"- _ <- readIORef entriesRef- pure ()- ]--fieldInt :: [Field] -> IO Int-fieldInt [fld] =- case fromField fld of- Errors err -> assertFailure (show err) >> pure 0- Ok value -> pure value-fieldInt fields = assertFailure ("expected one field, got " <> show (length fields)) >> pure 0--withConnectionTests :: TestTree-withConnectionTests =- testGroup- "withConnection"- [ testCase "returns the action result" $ do- result <- withConnection ":memory:" \_ -> pure (21 :: Int)- assertEqual "action result" 21 result- , testCase "propagates exceptions from the action"- $ assertThrowsErrorCall- $ withConnection ":memory:" (\_ -> error "boom" :: IO ())- ]--statementTests :: TestTree-statementTests =- testGroup- "statements"- [ testCase "prepares and closes a simple statement" $- withConnection ":memory:" \conn -> do- stmt <- openStatement conn "SELECT 42"- closeStatement stmt- , testCase "allows closing a statement twice" $- withConnection ":memory:" \conn -> do- stmt <- openStatement conn "SELECT 1"- closeStatement stmt- closeStatement stmt- , testCase "rejects preparing statements on a closed connection" $ do- conn <- open ":memory:"- close conn- assertThrowsSQLError $- openStatement conn "SELECT 1"- , testCase "throws SQLError for invalid SQL" $- withConnection ":memory:" \conn ->- assertThrowsSQLError $- openStatement conn "THIS IS NOT VALID SQL"- , testCase "withStatement closes the statement automatically" $- withConnection ":memory:" \conn -> do- result <- withStatement conn "SELECT 1" \_ -> pure ("done" :: String)- assertEqual "withStatement result" "done" result- , testCase "withStatement closes statements even when the action fails" $- withConnection ":memory:" \conn -> do- assertThrowsErrorCall $- withStatement conn "SELECT 1" (\_ -> error "boom" :: IO ())- stmt <- openStatement conn "SELECT 1"- closeStatement stmt- , testCase "execute returns affected row count" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE test_exec (x INTEGER)"- count <- execute conn "INSERT INTO test_exec VALUES (?)" (Only (1 :: Int))- assertEqual "rows affected" 1 count- , testCase "execute runs with positional parameters" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE exec_params (a INTEGER, b TEXT)"- count <- execute conn "INSERT INTO exec_params VALUES (?, ?)" (5 :: Int, "hi" :: String)- assertEqual "rows affected" 1 count- , testCase "rejects invalid direct execution" $- withConnection ":memory:" \conn ->- assertThrowsSQLError $- execute_ conn "THIS IS NOT SQL"- , testCase "binds positional parameters" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE params (a INTEGER, b TEXT)"- withStatement conn "INSERT INTO params VALUES (?, ?)" \stmt -> do- bind stmt [toField (5 :: Int), toField ("hi" :: String)]- changed <- executeStatement stmt- assertEqual "rows affected" 1 changed- , testCase "executeMany reuses prepared statements" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE params_many (a INTEGER, b TEXT)"- total <- executeMany conn "INSERT INTO params_many VALUES (?, ?)" [(1 :: Int, "x" :: String), (2 :: Int, "y" :: String)]- assertEqual "rows affected" 2 total- , testCase "executeNamed binds named parameters" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE named_params (a INTEGER, b TEXT)"- count <- executeNamed conn "INSERT INTO named_params VALUES ($a, $b)" ["$a" := (1 :: Int), "$b" := ("named" :: String)]- assertEqual "rows affected" 1 count- rows <- queryNamed conn "SELECT a FROM named_params WHERE b = $label" ["$label" := ("named" :: String)]- assertEqual "named query" [Only (1 :: Int)] rows- , testCase "rejects incorrect positional argument counts" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE param_count (a INTEGER, b INTEGER)"- assertThrowsFormatError- (execute conn "INSERT INTO param_count VALUES (?, ?)" (Only (1 :: Int)))- (Text.isInfixOf "parameter(s)" . formatErrorMessage)- , testCase "rejects positional bindings on named statements" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE mixed_mode (a INTEGER)"- assertThrowsFormatError- (execute conn "INSERT INTO mixed_mode VALUES ($a)" (Only (1 :: Int)))- (Text.isInfixOf "named parameters" . formatErrorMessage)- , testCase "reports error when mixing positional and named parameters" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE mixed_params (a INTEGER, b TEXT)"- assertThrows- ( withStatement conn "INSERT INTO mixed_params VALUES (?, $label)" \stmt -> do- bind stmt [toField (1 :: Int)]- bindNamed stmt ["$label" := ("combo" :: String)]- _ <- executeStatement stmt- pure ()- )- ( \(err :: SQLError) ->- Text.isInfixOf "Mixing named and positional parameters" (sqlErrorMessage err)- )- , testCase "generic FromRow derivation works" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE person (id INTEGER, name TEXT)"- _ <- executeMany conn "INSERT INTO person VALUES (?, ?)" [(1 :: Int, "Alice" :: String), (2 :: Int, "Bob" :: String)]- people <- query_ conn "SELECT id, name FROM person ORDER BY id" :: IO [Person]- assertEqual "person rows" [Person 1 (Text.pack "Alice"), Person 2 (Text.pack "Bob")] people- , testCase "(:.) composes row parsing and parameter encoding" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE dot_pair (a INTEGER, b INTEGER, label TEXT)"- let payload = Only (7 :: Int) :. (8 :: Int, Text.pack "hi")- _ <- execute conn "INSERT INTO dot_pair VALUES (?, ?, ?)" payload- rows <- query_ conn "SELECT a, b, label FROM dot_pair" :: IO [Only Int :. (Int, Text.Text)]- case rows of- [Only a :. (b, label)] -> do- assertEqual "(:.) first" 7 a- assertEqual "(:.) second" 8 b- assertEqual "(:.) label" (Text.pack "hi") label- other -> assertFailure ("unexpected (:.) rows: " <> show other)- , testCase "query_ fetches rows" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE query_rows (a INTEGER, b TEXT)"- _ <- executeMany conn "INSERT INTO query_rows VALUES (?, ?)" [(1 :: Int, "x" :: String), (2 :: Int, "y" :: String)]- rows <- query_ conn "SELECT a, b FROM query_rows ORDER BY a"- assertEqual "query rows" [(1 :: Int, "x" :: String), (2 :: Int, "y" :: String)] rows- , testCase "query decodes NULL as Maybe" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE maybe_vals (a TEXT)"- _ <- execute conn "INSERT INTO maybe_vals VALUES (?)" (Only (Just ("present" :: String)))- _ <- execute conn "INSERT INTO maybe_vals VALUES (?)" (Only (Nothing :: Maybe String))- rows <- query_ conn "SELECT a FROM maybe_vals ORDER BY a IS NULL, a"- assertEqual "maybe decoding" [Only (Just ("present" :: String)), Only Nothing] rows- , testCase "query fetches rows with parameters" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE query_params (a INTEGER, b TEXT)"- _ <- executeMany conn "INSERT INTO query_params VALUES (?, ?)" [(1 :: Int, "x" :: String), (2 :: Int, "y" :: String)]- rows <- query conn "SELECT a FROM query_params WHERE b = ?" (Only ("y" :: String))- assertEqual "query result" [Only (2 :: Int)] rows- , testCase "column mismatch surfaces as SQLError" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE mismatch (a INTEGER, b INTEGER)"- _ <- execute_ conn "INSERT INTO mismatch VALUES (1, 2)"- assertThrows- (query_ conn "SELECT a, b FROM mismatch" :: IO [Only Int])- (Text.isInfixOf "column index" . sqlErrorMessage)- , testCase "clears statement bindings without error" $- withConnection ":memory:" \conn -> do- stmt <- openStatement conn "SELECT ?"- clearStatementBindings stmt- closeStatement stmt- , testCase "resolves named parameter indices" $- withConnection ":memory:" \conn -> do- stmt <- openStatement conn "SELECT $named_param"- idx <- namedParameterIndex stmt "named_param"- assertEqual "named parameter index" (Just 1) idx- idxWithPrefix <- namedParameterIndex stmt "$named_param"- assertEqual "named parameter index with prefix" (Just 1) idxWithPrefix- colonIdx <- namedParameterIndex stmt ":named_param"- assertEqual "named parameter index with colon" (Just 1) colonIdx- none <- namedParameterIndex stmt "missing"- assertEqual "missing parameter index" Nothing none- closeStatement stmt- , testCase "reports column metadata for statements" $- withConnection ":memory:" \conn -> do- stmt <- openStatement conn "SELECT 1 AS a, 2 AS b"- count <- columnCount stmt- assertEqual "column count" 2 count- name0 <- columnName stmt 0- name1 <- columnName stmt 1- assertEqual "column names" [Text.pack "a", Text.pack "b"] [name0, name1]- assertThrows- (columnName stmt 2)- (Text.isInfixOf "out of bounds" . sqlErrorMessage)- closeStatement stmt- , testCase "bindNamed rejects statements without named placeholders" $- withConnection ":memory:" \conn -> do- stmt <- openStatement conn "SELECT ?"- assertThrowsFormatError- (bindNamed stmt [":value" := (1 :: Int)])- (Text.isInfixOf "does not define named parameters" . formatErrorMessage)- closeStatement stmt- , testCase "bindNamed rejects unknown parameters" $- withConnection ":memory:" \conn -> do- stmt <- openStatement conn "SELECT $known"- assertThrowsFormatError- (bindNamed stmt ["$missing" := (1 :: Int)])- (Text.isInfixOf "unknown named parameter" . formatErrorMessage)- closeStatement stmt- , testCase "numFieldsRemaining reports remaining columns" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE remaining (a INTEGER, b INTEGER, c INTEGER)"- _ <- execute conn "INSERT INTO remaining VALUES (?, ?, ?)" (1 :: Int, 2 :: Int, 3 :: Int)- rows <- query_ conn "SELECT a, b, c FROM remaining" :: IO [WithRemaining]- assertEqual "remaining count" [WithRemaining 1 2] rows- , testCase "RowParser alternatives fall back" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE yesno (answer TEXT)"- _ <- executeMany conn "INSERT INTO yesno VALUES (?)" [Only ("yes" :: String), Only ("no" :: String)]- rows <- query_ conn "SELECT answer FROM yesno ORDER BY answer" :: IO [YesNo]- assertEqual "yes/no parsing" [YesNo False, YesNo True] rows- ]--data DuckDBExpectation- = ExpectEquals FieldValue- | ExpectSatisfies (FieldValue -> Assertion)- | ExpectException (SomeException -> Assertion)--data DuckDBExpression- = CastExpression Text.Text Text.Text- | DirectExpression Text.Text--data DuckDBCastCase = DuckDBCastCase- { castLabel :: String- , castExpression :: DuckDBExpression- , castExpectation :: DuckDBExpectation- , castExpectFailureReason :: Maybe String- }--duckdbTypeCastTests :: TestTree-duckdbTypeCastTests =- testGroup- "duckdb type casts"- (map buildCase duckdbCastCases)- where- buildCase DuckDBCastCase{castLabel, castExpression, castExpectation, castExpectFailureReason} =- let base =- testCase castLabel $- withConnection ":memory:" \conn -> do- let sql =- Query- ( case castExpression of- CastExpression expressionValue expressionType ->- Text.concat ["SELECT CAST(", expressionValue, " AS ", expressionType, ")"]- DirectExpression expressionSQL ->- Text.concat ["SELECT ", expressionSQL]- )- result <- try (query_ conn sql) :: IO (Either SomeException [Only FieldValue])- runExpectation castLabel castExpectation result- maybe (pure ()) assertFailure castExpectFailureReason- in maybe base (`expectFailBecause` base) castExpectFailureReason--runExpectation :: String -> DuckDBExpectation -> Either SomeException [Only FieldValue] -> Assertion-runExpectation label expectation outcome =- case expectation of- ExpectEquals expected ->- case outcome of- Right [Only fieldValue] ->- assertEqual (label <> " result mismatch") expected fieldValue- Right other ->- assertFailure (label <> " expected single row, got " <> show other)- Left err ->- assertFailure (label <> " expected success, but query failed: " <> displayException err)- ExpectSatisfies predicate ->- case outcome of- Right [Only fieldValue] ->- predicate fieldValue- Right other ->- assertFailure (label <> " expected single row, got " <> show other)- Left err ->- assertFailure (label <> " expected success, but query failed: " <> displayException err)- ExpectException predicate ->- case outcome of- Left err ->- predicate err- Right rows ->- assertFailure (label <> " expected failure, but query succeeded with " <> show rows)--duckdbCastCases :: [DuckDBCastCase]-duckdbCastCases =- [ successCase "BOOLEAN" (quoted "true") "BOOLEAN" (ExpectEquals (FieldBool True))- , successCase "TINYINT" (quotedValue tinyIntValue) "TINYINT" (ExpectEquals (FieldInt8 tinyIntValue))- , successCase "SMALLINT" (quotedValue smallIntValue) "SMALLINT" (ExpectEquals (FieldInt16 smallIntValue))- , successCase "INTEGER" (quotedValue intValue) "INTEGER" (ExpectEquals (FieldInt32 intValue))- , successCase "BIGINT" (quotedValue bigIntValue) "BIGINT" (ExpectEquals (FieldInt64 bigIntValue))- , successCase "UTINYINT" (quotedValue uTinyValue) "UTINYINT" (ExpectEquals (FieldWord8 uTinyValue))- , successCase "USMALLINT" (quotedValue uSmallValue) "USMALLINT" (ExpectEquals (FieldWord16 uSmallValue))- , successCase "UINTEGER" (quotedValue uIntValue) "UINTEGER" (ExpectEquals (FieldWord32 uIntValue))- , successCase "UBIGINT" (quotedValue uBigValue) "UBIGINT" (ExpectEquals (FieldWord64 uBigValue))- , successCase "FLOAT" (quoted "3.25") "FLOAT" (ExpectEquals (FieldFloat floatValue))- , successCase "DOUBLE" (quoted "2.5") "DOUBLE" (ExpectEquals (FieldDouble doubleValue))- , successCase "TIMESTAMP" (quoted "2024-01-02 03:04:05.123456") "TIMESTAMP" (ExpectEquals (FieldTimestamp (Finite timestampLocalTimeMicros)))- , successCase "DATE" (quoted "2024-10-12") "DATE" (ExpectEquals (FieldDate (Finite dateValue)))- , successCase "TIME" (quoted "14:30:15.123456") "TIME" (ExpectEquals (FieldTime timeValue))- , successCase "INTERVAL" (quoted "1 day") "INTERVAL" (ExpectEquals (FieldInterval intervalValue))- , successCase "HUGEINT" (quotedValue hugeIntLiteral) "HUGEINT" (ExpectEquals (FieldHugeInt hugeIntLiteral))- , successCase "UHUGEINT" (quotedValue uHugeIntLiteral) "UHUGEINT" (ExpectEquals (FieldUHugeInt uHugeIntLiteral))- , successCase "VARCHAR" (quoted varcharText) "VARCHAR" (ExpectEquals (FieldText varcharText))- , successCase "BLOB" (quoted "duckdb") "BLOB" (ExpectEquals (FieldBlob blobValue))- , successCase "DECIMAL" (quoted "12345.6789") "DECIMAL(18,4)" (ExpectEquals (FieldDecimal decimalValue))- , successCase "TIMESTAMP_S" (quoted "2024-01-02 03:04:05") "TIMESTAMP_S" (ExpectEquals (FieldTimestamp (Finite timestampLocalTimeSeconds)))- , successCase "TIMESTAMP_MS" (quoted "2024-01-02 03:04:05.123") "TIMESTAMP_MS" (ExpectEquals (FieldTimestamp (Finite timestampLocalTimeMillis)))- , successCase "TIMESTAMP_NS" (quoted "2024-01-02 03:04:05.123456789") "TIMESTAMP_NS" (ExpectEquals (FieldTimestamp (Finite timestampLocalTimeNanos)))- , successCase "ENUM" (quoted "beta") "ENUM('alpha','beta')" (ExpectEquals (FieldEnum 1))- , successCase "LIST" (quoted "[1,2,3]") "INTEGER[]" (ExpectEquals (FieldList listElements))- , successDirect "MAP" "MAP(['a','b'], [1,2])" (expectMapEntries mapPairs)- , successCase "TIMETZ" (quoted "03:04:05+02:30") "TIME WITH TIME ZONE" (ExpectEquals (FieldTimeTZ timeWithZoneValue))- , successCase "TIMESTAMPTZ" (quoted "2024-01-02 03:04:05+02:30") "TIMESTAMP WITH TIME ZONE" (ExpectEquals (FieldTimestampTZ (Finite timestampTzUtc)))- , successCase "BIGNUM" (quoted bigNumLiteralText) "BIGNUM" (ExpectEquals (FieldBigNum (BigNum bigNumLiteral)))- , successCase "UUID" (quoted $ UUID.toText uuid) "UUID" (ExpectEquals (FieldUUID uuid))- , successCase "BIT" (quoted $ Text.pack $ show bits) "BIT" (ExpectEquals (FieldBit bits))- , successCase "ARRAY" (quoted "[1,2,3]") "INTEGER[3]" (ExpectEquals (FieldArray arrayElements))- , -- This one is broken upstream, instead of a DuckDBTypeTimeNs, we get a DuckDBType 0- successCase "TIME_NS" (quoted "03:04:05.123456789") "TIME_NS" (ExpectEquals (FieldTime (TimeOfDay 3 4 5.123456789)))- , successDirect "STRUCT" "{'a': 1, 'b': 2}" (expectStruct structFields)- , successDirect "UNION" "CAST(union_value(a := 42) AS UNION(a INTEGER, b VARCHAR))" (expectUnion 0 (Text.pack "a") (FieldInt32 42))- , -- These can never surface from a query, only internally.- failCaseOK "ANY" (quoted "1") "ANY" (expectSQLErrorContaining "ANY")- , failCaseOK "STRING_LITERAL" (quoted literalText) "STRING_LITERAL" (expectSQLErrorContaining "STRING_LITERAL")- , failCaseOK "INTEGER_LITERAL" (quoted "123") "INTEGER_LITERAL" (expectSQLErrorContaining "INTEGER_LITERAL")- , failCaseOK "INVALID" "0" "INVALID" (expectSQLErrorContaining "INVALID")- , failCaseOK "SQLNULL" "NULL" "SQLNULL" (expectSQLErrorContaining "SQLNULL")- ]- where- bits = BitString 7 $ BS.pack [1, 255]- quoted t = Text.concat ["'", t, "'"]- quotedValue v = quoted (valueText v)- valueText :: (Show a) => a -> Text.Text- valueText = Text.pack . show- successCase label value ty expectation =- DuckDBCastCase- { castLabel = label- , castExpression = CastExpression value ty- , castExpectation = expectation- , castExpectFailureReason = Nothing- }- successDirect label expr expectation =- DuckDBCastCase- { castLabel = label- , castExpression = DirectExpression expr- , castExpectation = expectation- , castExpectFailureReason = Nothing- }- failCaseOK label value ty expectation =- DuckDBCastCase- { castLabel = label- , castExpression = CastExpression value ty- , castExpectation = expectation- , castExpectFailureReason = Nothing- }- expectMapEntries expectedPairs =- ExpectSatisfies $- \case- FieldMap actualPairs ->- let normalize = sortOn (mapKey . fst)- mapKey (FieldText txt) = txt- mapKey other = Text.pack (show other)- in assertEqual "map entries" (normalize expectedPairs) (normalize actualPairs)- other ->- assertFailure ("expected FieldMap, but saw " <> show other)- structFields =- [ StructField{structFieldName = Text.pack "a", structFieldValue = FieldInt32 1}- , StructField{structFieldName = Text.pack "b", structFieldValue = FieldInt32 2}- ]- expectStruct expectedFields =- ExpectSatisfies $- \case- FieldStruct StructValue{structValueFields, structValueTypes, structValueIndex} -> do- let actualFields = elems structValueFields- actualTypes = elems structValueTypes- expectedIndex = Map.fromList (zip (map structFieldName expectedFields) [0 ..])- typeNames = map structFieldName actualTypes- assertEqual "struct field names" (map structFieldName expectedFields) (map structFieldName actualFields)- assertEqual "struct field values" (map structFieldValue expectedFields) (map structFieldValue actualFields)- assertEqual "struct type field names" (map structFieldName expectedFields) typeNames- assertEqual "struct index map" expectedIndex structValueIndex- other ->- assertFailure ("expected FieldStruct, but saw " <> show other)- expectUnion expectedIndex expectedLabel expectedPayload =- ExpectSatisfies $- \case- FieldUnion UnionValue{unionValueIndex, unionValueLabel, unionValuePayload, unionValueMembers} -> do- assertEqual "union tag index" expectedIndex unionValueIndex- assertEqual "union tag label" expectedLabel unionValueLabel- assertEqual "union payload" expectedPayload unionValuePayload- let actualMembers = elems unionValueMembers- expectedMembers = [Text.pack "a", Text.pack "b"]- assertEqual "union member names" expectedMembers (map unionMemberName actualMembers)- other ->- assertFailure ("expected FieldUnion, but saw " <> show other)- expectSQLErrorContaining :: Text.Text -> DuckDBExpectation- expectSQLErrorContaining needle =- ExpectException $ \err ->- case fromException err :: Maybe SQLError of- Just sqlErr ->- let message = sqlErrorMessage sqlErr- in assertBool- ("expected SQLError containing " <> Text.unpack needle <> ", but saw: " <> Text.unpack message)- (Text.isInfixOf needle message)- Nothing ->- assertFailure ("expected SQLError, but saw " <> displayException err)- bsFromString :: String -> BS.ByteString- bsFromString = BS.pack . map (fromIntegral . fromEnum)- secondsWithFraction :: Integer -> Integer -> Integer -> Rational- secondsWithFraction whole numerator denominator =- fromIntegral whole + numerator % denominator- tinyIntValue :: Int8- tinyIntValue = 42- smallIntValue :: Int16- smallIntValue = 32000- intValue :: Int32- intValue = 2147483647- bigIntValue :: Int64- bigIntValue = 9223372036854775807- uTinyValue :: Word8- uTinyValue = 200- uSmallValue :: Word16- uSmallValue = 60000- uIntValue :: Word32- uIntValue = maxBound- uBigValue :: Word64- uBigValue = maxBound- floatValue :: Float- floatValue = 3.25- doubleValue :: Double- doubleValue = 2.5- timestampDay = fromGregorian 2024 1 2- timestampLocalTimeMicros =- LocalTime timestampDay (TimeOfDay 3 4 (fromRational (secondsWithFraction 5 123456 1000000)))- timestampLocalTimeSeconds =- LocalTime timestampDay (TimeOfDay 3 4 5)- timestampLocalTimeMillis =- LocalTime timestampDay (TimeOfDay 3 4 (fromRational (secondsWithFraction 5 123 1000)))- timestampLocalTimeNanos =- LocalTime timestampDay (TimeOfDay 3 4 (fromRational (secondsWithFraction 5 123456789 1000000000)))- dateValue = fromGregorian 2024 10 12- timeValue =- TimeOfDay 14 30 (fromRational (secondsWithFraction 15 123456 1000000))- intervalValue =- IntervalValue{intervalMonths = 0, intervalDays = 1, intervalMicros = 0}- decimalValue =- DecimalValue{decimalWidth = 18, decimalScale = 4, decimalInteger = 123456789}- listElements =- [FieldInt32 1, FieldInt32 2, FieldInt32 3]- arrayElements =- listArray (0, 2) listElements- mapPairs =- [ (FieldText (Text.pack "a"), FieldInt32 1)- , (FieldText (Text.pack "b"), FieldInt32 2)- ]- uuid :: UUID.UUID- uuid = fromJust $ UUID.fromText "123e4567-e89b-12d3-a456-426614174000"- timeWithZoneValue =- TimeWithZone{timeWithZoneTime = TimeOfDay 3 4 5, timeWithZoneZone = minutesToTimeZone 150}- timestampTzUtc =- UTCTime timestampDay (secondsToDiffTime 2045)- bigNumLiteral :: Integer- bigNumLiteral = 1234567890123456789012345678901234567890- bigNumLiteralText = valueText bigNumLiteral- varcharText :: Text.Text- varcharText = "Hello, DuckDB"- blobValue = bsFromString "duckdb"- hugeIntLiteral :: Integer- hugeIntLiteral = (2 ^ (120 :: Int)) + 123456789- uHugeIntLiteral :: Integer- uHugeIntLiteral = (2 ^ (128 :: Int)) - 1- literalText :: Text.Text- literalText = "literal"--typeTests :: TestTree-typeTests =- testGroup- "types"- [ duckdbTypeCastTests- , testCase "round-trips date/time/timestamp" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE temporals (d DATE, t TIME, ts TIMESTAMP)"- let dayVal = fromGregorian 2024 10 12- timeVal = TimeOfDay 14 30 15.123456- tsVal = LocalTime dayVal timeVal- _ <- execute conn "INSERT INTO temporals VALUES (?, ?, ?)" (dayVal, timeVal, tsVal)- [(dRes, tRes, tsRes)] <- query_ conn "SELECT d, t, ts FROM temporals"- assertEqual "date round-trip" dayVal dRes- assertEqual "time round-trip" timeVal tRes- assertEqual "timestamp round-trip" tsVal tsRes- [Only utcRes] <- query_ conn "SELECT ts FROM temporals" :: IO [Only UTCTime]- assertEqual "timestamp as UTC" (localTimeToUTC utc tsVal) utcRes- , testCase "round-trips blob payloads" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE blobs (payload BLOB)"- let payload = BS.pack [0, 1, 2, 3, 255]- _ <- execute conn "INSERT INTO blobs VALUES (?)" (Only payload)- [Only blobOut] <- query_ conn "SELECT payload FROM blobs"- assertEqual "blob round-trip" payload blobOut- , testCase "round-trips unsigned integers" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE unsigneds (u8 UTINYINT, u16 USMALLINT, u32 UINTEGER, u64 UBIGINT)"- let w8 = 200 :: Word8- w16 = 60000 :: Word16- w32 = 4000000000 :: Word32- w64 = maxBound :: Word64- _ <- execute conn "INSERT INTO unsigneds VALUES (?, ?, ?, ?)" (w8, w16, w32, w64)- [(r8, r16, r32, r64)] <- query_ conn "SELECT u8, u16, u32, u64 FROM unsigneds"- assertEqual "Word8 round-trip" w8 r8- assertEqual "Word16 round-trip" w16 r16- assertEqual "Word32 round-trip" w32 r32- assertEqual "Word64 round-trip" w64 r64- [Only (asWord :: Word)] <- query_ conn "SELECT u64 FROM unsigneds"- assertEqual "Word from UBIGINT" (fromIntegral w64) asWord- , testCase "round-trips fixed-length arrays" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE arrays (vals INTEGER[3])"- let inputs =- [ listArray (0, 2) [1, 2, 3]- , listArray (0, 2) [4, 5, 6]- ]- _ <- executeMany conn "INSERT INTO arrays VALUES (?)" (Only <$> inputs)- rows <- query_ conn "SELECT vals FROM arrays ORDER BY rowid" :: IO [Only (Array Int Int)]- let expected =- fmap Only inputs- assertEqual "array round-trip" expected rows- , testCase "decodes huge integers as Integer" $- withConnection ":memory:" \conn -> do- let hugeValue = 170141183460469231731687303715884105727 :: Integer- [Only (hugeOut :: Integer)] <-- query_ conn "SELECT 170141183460469231731687303715884105727::HUGEINT"- assertEqual "hugeint" hugeValue hugeOut- , testCase "decodes huge integers as Natural" $- withConnection ":memory:" \conn -> do- let hugeValue = 170141183460469231731687303715884105727 :: Natural- [Only (naturalOut :: Natural)] <-- query_ conn "SELECT 170141183460469231731687303715884105727::HUGEINT"- assertEqual "hugeint natural" hugeValue naturalOut- , testCase "decodes unsigned huge integers as Natural" $- withConnection ":memory:" \conn -> do- let uhValue = 170141183460469231731687303715884105727 :: Natural- [Only (naturalOut :: Natural)] <-- query_ conn "SELECT 170141183460469231731687303715884105727::UHUGEINT"- assertEqual "uhugeint natural" uhValue naturalOut- , testCase "FromField Integer consumes FieldBigNum" $ do- let original = 1234567899876543210123456789 :: Integer- bigField =- Field- { fieldName = "big"- , fieldIndex = 0- , fieldValue = FieldBigNum (BigNum original)- }- case (fromField bigField :: Ok Integer) of- Ok result -> assertEqual "FieldBigNum Integer" original result- Errors err -> assertFailure ("unexpected parse failure: " <> show err)- , testCase "FromField Natural rejects negative FieldBigNum" $ do- let bigValue = BigNum (-5)- bigField =- Field- { fieldName = "big"- , fieldIndex = 0- , fieldValue = FieldBigNum bigValue- }- case (fromField bigField :: Ok Natural) of- Errors _ -> pure ()- Ok result -> assertFailure ("expected failure, got " <> show result)- , testCase "BigNum conversions round-trip Integer" $ do- let original = (2 ^ (200 :: Int)) + 8675309 :: Integer- converted = toBigNumBytes original- assertEqual "round-trip Integer" original (fromBigNumBytes converted)- , testProperty "fromBigNumBytes . toBigNumBytes is identity" \(n :: Integer) ->- fromBigNumBytes (toBigNumBytes n) === n- , testCase "selects BIGNUM values as BigNum" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE bignums (val BIGNUM)"- let bigValues :: [Integer]- bigValues =- [ (2 ^ (200 :: Int)) + 8675309- , 0- , negate ((2 ^ (180 :: Int)) + 12345)- ]- rows <- concat <$> mapM (\bv -> query_ conn $ fromString $ "SELECT CAST('" <> show bv <> "' AS BIGNUM)") bigValues- -- query_ conn "SELECT val FROM bignums ORDER BY rowid" :: IO [Only BigNum]- assertEqual "BIGNUM results" (fmap (Only . BigNum) bigValues) rows- , testCase "round-trips BIGNUM values through the database" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE bignum_roundtrip (val BIGNUM)"- let ints =- [ (2 ^ (200 :: Int)) + 8675309- , 0- , negate ((2 ^ (180 :: Int)) + 12345)- ]- _ <- executeMany conn "INSERT INTO bignum_roundtrip VALUES (?)" (fmap (Only . BigNum) ints)- rows <- query_ conn "SELECT val FROM bignum_roundtrip ORDER BY rowid" :: IO [Only Integer]- assertEqual "BIGNUM round-trip" (fmap Only ints) rows- , testCase "duckdbColumnType reports BIGNUM for Integer" $- assertEqual- "duckdbColumnType Integer"- (Text.pack "BIGNUM")- (duckdbColumnType (Proxy :: Proxy Integer))- , testCase "duckdbColumnType preserves Maybe parameter" $- assertEqual- "duckdbColumnType Maybe Word16"- (Text.pack "USMALLINT")- (duckdbColumnType (Proxy :: Proxy (Maybe Word16)))- , testCase "decodes interval components" $- withConnection ":memory:" \conn -> do- [Only intervalVal] <-- query_ conn "SELECT INTERVAL '2 months 3 days 04:05:06.007008'"- let IntervalValue{intervalMonths, intervalDays, intervalMicros} = intervalVal- expectedMicros =- (((4 * 60 + 5) * 60) + 6) * 1000000 + 7008- assertEqual "interval months" 2 intervalMonths- assertEqual "interval days" 3 intervalDays- assertEqual "interval micros" expectedMicros intervalMicros- , testCase "decodes decimals via Double conversion" $- withConnection ":memory:" \conn -> do- [Only decimalAsDouble] <-- (query_ conn "SELECT CAST(12345.6789 AS DECIMAL(18,4))" :: IO [Only Double])- let diff = abs (decimalAsDouble - 12345.6789)- assertBool- ("decimal as double: " <> show diff)- (diff < 1e-4)- , testCase "decodes time with time zone" $- withConnection ":memory:" \conn -> do- [Only tzVal] <- query_ conn "SELECT TIMETZ '14:30:15+02:30'"- let expectedTime = TimeOfDay 14 30 15- expectedZone = minutesToTimeZone 150- assertEqual "timetz time" expectedTime (timeWithZoneTime tzVal)- assertEqual "timetz zone" expectedZone (timeWithZoneZone tzVal)- , testCase "decodes list columns into Haskell lists" $- withConnection ":memory:" \conn -> do- [Only (listOut :: [Int])] <-- query_ conn "SELECT LIST_VALUE(1, 2, 3)::INTEGER[]"- assertEqual "list decode" [1, 2, 3] listOut- , testCase "decodes map columns into strict Map" $- withConnection ":memory:" \conn -> do- [Only (mapOut :: Map.Map Text.Text Int)] <-- query_ conn "SELECT MAP(['a', 'b']::TEXT[], [1, 2]::INTEGER[])"- let expected = Map.fromList [(Text.pack "a", 1), (Text.pack "b", 2)]- assertEqual "map decode" expected mapOut- , testCase "decodes timestamp with time zone as UTC" $- withConnection ":memory:" \conn -> do- [Only utcVal] <-- (query_ conn "SELECT TIMESTAMPTZ '2024-10-12 14:30:15+02:30'" :: IO [Only UTCTime])- let expected =- UTCTime- (fromGregorian 2024 10 12)- (timeOfDayToTime (TimeOfDay 12 0 15))- assertEqual "timestamptz utc" expected utcVal- , testCase "custom FieldParser enforces invariants" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE nonempty (name TEXT)"- _ <- execute conn "INSERT INTO nonempty VALUES (?)" (Only (Text.pack "okay"))- rows <- query_ conn "SELECT name FROM nonempty" :: IO [Only NonEmptyText]- assertEqual "non-empty success" [Only (NonEmptyText (Text.pack "okay"))] rows- _ <- execute conn "INSERT INTO nonempty VALUES (?)" (Only (Text.pack ""))- assertThrows- (query_ conn "SELECT name FROM nonempty" :: IO [Only NonEmptyText])- (Text.isInfixOf "NonEmptyText requires a non-empty string" . sqlErrorMessage)- , testCase "round-trips uuid payloads" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE uuids (uuid UUID)"- let uuidTexts =- [ "123e4567-e89b-12d3-a456-426614174000"- , "923e4567-e89b-12d3-a456-426614174000"- , "ffffffff-ffff-ffff-ffff-ffffffffffff"- ]- maybeUUIDs = traverse UUID.fromText uuidTexts- uuids <-- case maybeUUIDs of- Nothing -> assertFailure "invalid UUID literal in test data" >> pure []- Just parsed -> pure parsed- _ <- executeMany conn "INSERT INTO uuids VALUES (?)" (fmap Only uuids)- rows <- query_ conn "SELECT uuid FROM uuids ORDER BY rowid" :: IO [Only UUID.UUID]- assertEqual "uuid round-trip" (fmap Only uuids) rows- , testCase "round-trips bitstring payloads" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE bits (bit BIT)"- let bitValues =- [ bsFromBool [True]- , bsFromBool [False, False, False, True, True, True, False, True, False]- , bsFromBool [False, False, False, False, True, True, True, False, True, False]- , bsFromBool [True, False, False, False, True, True, True, False, True, False]- ]- _ <- executeMany conn "INSERT INTO bits VALUES (?)" (fmap Only bitValues)- rows <- query_ conn "SELECT bit FROM bits ORDER BY rowid" :: IO [Only BitString]- assertEqual "bit round-trip" (fmap Only bitValues) rows- , testCase "round-trips struct and union parameters" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE struct_union (s STRUCT(a INT, b INT), u UNION(a INT, b VARCHAR))"- [(structVal, unionVal)] <-- query_ conn "SELECT {'a': 1, 'b': 2}, CAST(union_value(a := 42) AS UNION(a INTEGER, b VARCHAR))" :: IO [(StructValue FieldValue, UnionValue FieldValue)]- _ <- execute conn "INSERT INTO struct_union VALUES (?, ?)" (structVal, unionVal)- rows <- query_ conn "SELECT s, u FROM struct_union" :: IO [(StructValue FieldValue, UnionValue FieldValue)]- assertEqual "struct/union round-trip" [(structVal, unionVal)] rows- , testCase "generic struct into database" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE generic_struct (payload STRUCT(grId INT, grName TEXT))"- let records = [GenericRecord 1 (Text.pack "alpha"), GenericRecord 2 (Text.pack "beta")]- structParams = traverse genericToStructValue records- case structParams of- Nothing -> assertFailure "expected struct encoding"- Just structs -> do- _ <- executeMany conn "INSERT INTO generic_struct VALUES (?)" (map Only structs)- rows <- query_ conn "SELECT payload FROM generic_struct ORDER BY payload.grId" :: IO [Only (StructValue FieldValue)]- let unwrap (Only s) = s- roundTripped = traverse (genericFromFieldValue . FieldStruct . unwrap) rows- roundTripped @?= Right records- , testCase "generic struct encode/decode" $ do- let rec = GenericRecord 42 (Text.pack "duck")- case genericToStructValue rec of- Just structVal -> do- map structFieldName (elems (structValueFields structVal)) @?= [Text.pack "grId", Text.pack "grName"]- map structFieldValue (elems (structValueFields structVal)) @?= [FieldInt64 42, FieldText (Text.pack "duck")]- genericFromFieldValue (FieldStruct structVal) @?= Right rec- Nothing -> assertFailure "expected struct encoding"- , testCase "generic union encode/decode" $ do- let sumVal = SumBool True- case genericToFieldValue sumVal of- FieldUnion uv -> do- map unionMemberName (elems (unionValueMembers uv))- @?= [ Text.pack "SumInt"- , Text.pack "SumBool"- , Text.pack "SumPair"- ]- unionValueLabel uv @?= Text.pack "SumBool"- genericFromFieldValue (FieldUnion uv) @?= Right sumVal- other -> assertFailure ("expected FieldUnion, got " <> show other)- , testCase "generic union complex constructor" $ do- let sumVal = SumPair 7 (Text.pack "duck")- case genericToFieldValue sumVal of- FieldUnion uv -> do- unionValueLabel uv @?= Text.pack "SumPair"- case unionValuePayload uv of- FieldStruct StructValue{structValueFields} -> do- map structFieldName (elems structValueFields) @?= [Text.pack "field1", Text.pack "field2"]- map structFieldValue (elems structValueFields)- @?= [FieldInt64 7, FieldText (Text.pack "duck")]- other -> assertFailure ("expected struct payload, got " <> show other)- genericFromFieldValue (FieldUnion uv) @?= Right sumVal- other -> assertFailure ("expected FieldUnion, got " <> show other)- , testCase "deriving via generic struct round-trip" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE via_users (payload STRUCT(vuId BIGINT, vuName TEXT))"- let users = [ViaUser 10 (Text.pack "alice"), ViaUser 11 (Text.pack "bob")]- _ <- executeMany conn "INSERT INTO via_users VALUES (?)" (map Only users)- rows <- query_ conn "SELECT payload FROM via_users ORDER BY payload.vuId" :: IO [Only ViaUser]- rows @?= map Only users- , testCase "deriving via generic union round-trip" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE via_shapes (shape UNION(ViaCircle STRUCT(field1 DOUBLE), ViaSquare STRUCT(field1 DOUBLE)))"- let shapes = [ViaCircle 1.5, ViaSquare 3.0]- _ <- executeMany conn "INSERT INTO via_shapes VALUES (?)" (map Only shapes)- rows <- query_ conn "SELECT shape FROM via_shapes ORDER BY rowid" :: IO [Only ViaShape]- rows @?= map Only shapes- , testCase "generic enum null payload" $ do- case genericToFieldValue EnumRed of- FieldUnion uv -> do- unionValuePayload uv @?= FieldNull- genericFromFieldValue (FieldUnion uv) @?= Right EnumRed- other -> assertFailure ("expected FieldUnion, got " <> show other)- , -- Edge case tests- testCase "struct with multiple basic fields" $ do- -- Test a struct with multiple fields- let twoField = TwoFieldStruct 42 (Text.pack "text")- case genericToFieldValue twoField of- FieldStruct sv -> do- genericFromFieldValue (FieldStruct sv) @?= Right twoField- other -> assertFailure ("expected FieldStruct for TwoFieldStruct, got " <> show other)- , testCase "large sum type with many constructors" $ do- let constructors = [LargeC1 1, LargeC2 2, LargeC3 3, LargeC4 4, LargeC5 5]- forM_ constructors $ \val -> do- case genericToFieldValue val of- FieldUnion uv -> do- genericFromFieldValue (FieldUnion uv) @?= Right val- other -> assertFailure ("expected FieldUnion, got " <> show other)- , testCase "empty struct encodes as null" $ do- case genericToFieldValue EmptyStruct of- FieldNull -> pure ()- other -> assertFailure ("expected FieldNull for empty struct, got " <> show other)- , testCase "struct with Maybe fields" $ do- let withJust = OptionalFields (Just 42) (Just (Text.pack "text"))- withNothing = OptionalFields Nothing Nothing- case (genericToFieldValue withJust, genericToFieldValue withNothing) of- (FieldStruct sv1, FieldStruct sv2) -> do- genericFromFieldValue (FieldStruct sv1) @?= Right withJust- genericFromFieldValue (FieldStruct sv2) @?= Right withNothing- (other1, other2) -> assertFailure ("expected FieldStruct, got " <> show (other1, other2))- , -- Error case tests- testCase "decode wrong field type fails" $ do- let wrongType = FieldInt32 42 -- Trying to decode as GenericRecord- case genericFromFieldValue wrongType :: Either String GenericRecord of- Left err -> assertBool "error mentions STRUCT" (Text.pack "STRUCT" `Text.isInfixOf` Text.pack err)- Right _ -> assertFailure "expected decode error for wrong type"- , testCase "decode union as struct fails" $ do- case genericToFieldValue (SumInt 42) of- FieldUnion uv -> do- case genericFromFieldValue (FieldUnion uv) :: Either String GenericRecord of- Left err -> assertBool "error mentions product type" (Text.pack "product type" `Text.isInfixOf` Text.pack err)- Right _ -> assertFailure "expected decode error"- other -> assertFailure ("expected FieldUnion, got " <> show other)- , testCase "decode struct as union fails" $ do- case genericToFieldValue (GenericRecord 1 (Text.pack "test")) of- FieldStruct sv -> do- case genericFromFieldValue (FieldStruct sv) :: Either String GenericSum of- Left err -> assertBool "error mentions sum type" (Text.pack "sum type" `Text.isInfixOf` Text.pack err)- Right _ -> assertFailure "expected decode error"- other -> assertFailure ("expected FieldStruct, got " <> show other)- , testCase "decode struct with wrong field count fails" $- withConnection ":memory:" \conn -> do- -- Get a real struct with 3 fields from the database- _ <- execute_ conn "CREATE TABLE three_field (s STRUCT(a INT, b TEXT, c INT))"- [(Only threeField)] <- query_ conn "SELECT {'a': 1, 'b': 'test', 'c': 99}" :: IO [Only (StructValue FieldValue)]- -- Try to decode it as GenericRecord which only has 2 fields- case genericFromFieldValue (FieldStruct threeField) :: Either String GenericRecord of- Left err -> assertBool "error mentions extra fields" (Text.pack "extra" `Text.isInfixOf` Text.pack err)- Right _ -> assertFailure "expected decode error for extra fields"- , testCase "decode union with wrong tag fails" $- withConnection ":memory:" \conn -> do- -- Get a real union from the database- [(Only unionVal)] <-- query_ conn "SELECT CAST(union_value(x := 42) AS UNION(x INT, y VARCHAR))" :: IO [Only (UnionValue FieldValue)]- -- Manually create a modified version with an out-of-range tag- let wrongTag = unionVal{unionValueIndex = 99, unionValueLabel = Text.pack "Invalid"}- case genericFromFieldValue (FieldUnion wrongTag) :: Either String GenericSum of- Left err -> assertBool "error mentions tag or index" (Text.pack "tag" `Text.isInfixOf` Text.pack err || Text.pack "99" `Text.isInfixOf` Text.pack err)- Right _ -> assertFailure "expected decode error for wrong tag"- ]--streamingTests :: TestTree-streamingTests =- testGroup- "streaming"- [ testCase "fold_ accumulates large result sets" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE stream_fold (n INTEGER)"- let rows = fmap Only [1 .. 1000 :: Int]- _ <- executeMany conn "INSERT INTO stream_fold VALUES (?)" rows- total <- fold_ conn "SELECT n FROM stream_fold ORDER BY n" 0 \acc (Only n) -> pure (acc + n)- assertEqual "folded sum" (sum ([1 .. 1000] :: [Int])) total- , testCase "fold and foldNamed respect parameters" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE stream_filter (n INTEGER)"- let values = fmap Only [1 .. 50 :: Int]- _ <- executeMany conn "INSERT INTO stream_filter VALUES (?)" values- gtTotal <-- fold conn "SELECT n FROM stream_filter WHERE n > ?" (Only (40 :: Int)) 0 $- \acc (Only n) -> pure (acc + n)- let expected = sum ([41 .. 50] :: [Int])- assertEqual "filtered fold sum" expected gtTotal- leCount <-- foldNamed conn "SELECT n FROM stream_filter WHERE n <= $limit" ["$limit" := (10 :: Int)] 0 $- \acc (Only (_ :: Int)) -> pure (acc + 1)- assertEqual "foldNamed count" (10 :: Int) leCount- , testCase "nextRow streams rows sequentially" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE stream_cursor (n INTEGER)"- _ <- executeMany conn "INSERT INTO stream_cursor VALUES (?)" (fmap Only [1 .. 3 :: Int])- withStatement conn "SELECT n FROM stream_cursor ORDER BY n" \stmt -> do- first <- nextRow stmt- second <- nextRow stmt- third <- nextRow stmt- done <- nextRow stmt- assertEqual- "cursor iteration"- [Just (Only (1 :: Int)), Just (Only 2), Just (Only 3), Nothing]- [first, second, third, done]- ]--functionsTests :: TestTree-functionsTests =- testGroup- "user-defined functions"- [ testCase "registers pure scalar function" $- withConnection ":memory:" \conn -> do- createFunction conn "hs_times_two" (\(x :: Int64) -> x * 2)- result <- query_ conn "SELECT hs_times_two(21)" :: IO [Only Int64]- assertEqual "times_two result" [Only 42] result- , testCase "handles nullable arguments and results" $- withConnection ":memory:" \conn -> do- createFunction conn "hs_optional" (\(mx :: Maybe Int64) -> fmap (+ 1) mx)- rows <-- query_ conn "SELECT hs_optional(x) FROM (VALUES (NULL), (41)) AS t(x)" ::- IO [Only (Maybe Int64)]- assertEqual "optional results" [Only Nothing, Only (Just 42)] rows- , testCase "supports IO-based functions" $- withConnection ":memory:" \conn -> do- ref <- newIORef (0 :: Int)- createFunction conn "hs_counter" $ do- atomicModifyIORef' ref \n ->- let next = n + 1- in (next, next)- first <- query_ conn "SELECT hs_counter()" :: IO [Only Int]- second <- query_ conn "SELECT hs_counter()" :: IO [Only Int]- assertEqual "first counter call" [Only 1] first- assertEqual "second counter call" [Only 2] second- , testCase "supports thread-local function state" $ do- conn <- openWithConfig ":memory:" [("threads", "1")]- createFunctionWithState conn "hs_stateful_counter" (newIORef (0 :: Int)) \ref -> do- atomicModifyIORef' ref \n ->- let next = n + 1- in (next, next)- rows <- query_ conn "SELECT hs_stateful_counter() FROM range(3)" :: IO [Only Int]- close conn- assertEqual "stateful rows" [Only 1, Only 2, Only 3] rows- ]--transactionTests :: TestTree-transactionTests =- testGroup- "transactions"- [ testCase "withTransaction commits on success" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE tx_commit (x INTEGER)"- withTransaction conn $ do- _ <- execute conn "INSERT INTO tx_commit VALUES (?)" (Only (1 :: Int))- pure ()- rows <- query_ conn "SELECT COUNT(*) FROM tx_commit" :: IO [Only Int]- assertEqual "commit" [Only 1] rows- , testCase "withTransaction rolls back on exception" $- withConnection ":memory:" \conn -> do- _ <- execute_ conn "CREATE TABLE tx_rollback (x INTEGER)"- assertThrowsErrorCall $- withTransaction conn $ do- _ <- execute conn "INSERT INTO tx_rollback VALUES (?)" (Only (1 :: Int))- error "boom"- rows <- query_ conn "SELECT COUNT(*) FROM tx_rollback" :: IO [Only Int]- assertEqual "rollback" [Only 0] rows- ]--assertThrows :: forall e a. (Exception e) => IO a -> (e -> Bool) -> Assertion-assertThrows action predicate = do- outcome <- try action- case outcome of- Left (err :: e) -> assertBool "unexpected exception" (predicate err)- Right _ -> assertFailure "expected exception, but action succeeded"--assertThrowsSQLError :: IO a -> Assertion-assertThrowsSQLError action =- assertThrows action (const True :: SQLError -> Bool)--assertThrowsErrorCall :: IO a -> Assertion-assertThrowsErrorCall action =- assertThrows action (const True :: ErrorCall -> Bool)--assertThrowsFormatError :: IO a -> (FormatError -> Bool) -> Assertion-assertThrowsFormatError = assertThrows+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE DeriveAnyClass #-}+{-# LANGUAGE DeriveGeneric #-}+{-# LANGUAGE DerivingVia #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}+{-# LANGUAGE TypeOperators #-}++-- |+-- Module : Main+-- Description : Tasty-based test suite for duckdb-simple.+module Main (main) where++import ArrowTests (arrowTests)+import CancellationTests (callbackLockTests, cancellationTests)+import Control.Applicative ((<|>))+import Control.Concurrent (forkFinally, newEmptyMVar, putMVar, readMVar)+import Control.Exception (ErrorCall, Exception, SomeException, bracket, displayException, fromException, throwIO, try)+import Control.Monad (forM_, replicateM_, void, when)+import CopyTests (copyTests)+import CoreRegressionTests (coreRegressionTests)+import Data.Array (elems, listArray)+import qualified Data.ByteString as BS+import Data.IORef (atomicModifyIORef', newIORef, readIORef)+import Data.Int (Int16, Int32, Int64, Int8)+import Data.List (sortOn)+import qualified Data.Map.Strict as Map+import Data.Maybe (fromJust)+import Data.Proxy (Proxy (..))+import Data.Ratio ((%))+import Data.String (fromString)+import qualified Data.Text as Text+import Data.Time.Calendar (Day, fromGregorian)+import Data.Time.Clock (UTCTime (..), secondsToDiffTime)+import Data.Time.LocalTime (+ LocalTime (..),+ TimeOfDay (..),+ localTimeToUTC,+ timeOfDayToTime,+ utc,+ )+import qualified Data.UUID as UUID+import Data.Word (Word16, Word32, Word64, Word8)+import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Catalog as Catalog+import qualified Database.DuckDB.Simple.Config as Config+import qualified Database.DuckDB.Simple.Copy as Copy+import qualified Database.DuckDB.Simple.FileSystem as FileSystem+import Database.DuckDB.Simple.FromField (+ BigNum (..),+ BitString (..),+ DecimalValue (..),+ Field (..),+ FieldValue (..),+ IntervalValue (..),+ TimeWithZone (..),+ bsFromBool,+ fromBigNumBytes,+ returnError,+ toBigNumBytes,+ )+import Database.DuckDB.Simple.Generic (+ DuckValue (..),+ ViaDuckDB (..),+ genericFromFieldValue,+ genericLogicalType,+ genericToFieldValue,+ genericToStructValue,+ )+import qualified Database.DuckDB.Simple.Logging as Logging+import Database.DuckDB.Simple.LogicalRep (+ LogicalTypeRep (..),+ StructField (..),+ StructValue (..),+ UnionMemberType (..),+ UnionValue (..),+ )+import Database.DuckDB.Simple.Ok (Ok (..))+import Database.DuckDB.Simple.Time (Unbounded (..))+import Database.DuckDB.Simple.ToField (ToDuckValue (..))+import Database.DuckDB.Simple.Variant (Variant (..), variantObject)+import ExtensionRegressionTests (extensionRegressionTests)+import GHC.Generics (Generic)+import qualified GeometryRegressionTests+import Numeric.Natural (Natural)+import Properties (roundTripTests)+import RegistrationTests (registrationTests)+import SizedArrayTests (sizedArrayTests)+import qualified SqlFeatureTests+import StreamingTests (nativeStreamingTests)+import System.Directory (doesFileExist, getTemporaryDirectory, removeFile)+import System.IO (hClose, openBinaryTempFile)+import System.Timeout (timeout)+import Test.Tasty (TestTree, Timeout (..), adjustOption, defaultMain, mkTimeout, testGroup)+import Test.Tasty.HUnit+import Test.Tasty.QuickCheck (testProperty, (===))+import TestUtils (sized)+import TimeTests (timeTests)+import TypedRegistrationTests (typedRegistrationTests)+import UserCastTests (geometryArgumentTests, userCastTests)+import ValueRegressionTests (valueRegressionTests)+import qualified VariantRegressionTests++data Person = Person+ { personId :: Int+ , personName :: Text.Text+ }+ deriving stock (Eq, Show, Generic)+ deriving anyclass (FromRow)++data WithRemaining = WithRemaining Int Int+ deriving (Eq, Show)++data GenericRecord = GenericRecord+ { grId :: Int+ , grName :: Text.Text+ }+ deriving stock (Eq, Show, Generic)++data GenericSum+ = SumInt Int+ | SumBool Bool+ | SumPair Int Text.Text+ deriving stock (Eq, Show, Generic)++data GenericEnum+ = EnumRed+ | EnumBlue+ deriving stock (Eq, Show, Generic)++data ViaUser = ViaUser+ { vuId :: Int64+ , vuName :: Text.Text+ }+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToDuckValue, ToField, FromField) via (ViaDuckDB ViaUser)++-- | A newtype with its own type text, for a scalar function argument.+newtype Price = Price DecimalValue++instance DuckDBColumnType Price where+ duckdbColumnTypeFor _ = "DECIMAL(18,3)"++instance FromField Price where+ fromField value = Price <$> fromField value++-- | A newtype with its own type text, for a scalar function result.+newtype Amount = Amount Double++instance DuckDBColumnType Amount where+ duckdbColumnTypeFor _ = "DECIMAL(18,3)"++instance ToDuckValue Amount where+ toDuckValue factory (Amount value) = toDuckValue factory value++data ViaShape+ = ViaCircle Double+ | ViaSquare Double+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToDuckValue, ToField, FromField) via (ViaDuckDB ViaShape)++-- | A UNION with a STRUCT payload that contains a VARIANT.+data VariantChoice = Payload Variant | NoPayload+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB VariantChoice)++-- Edge case test types+data TwoFieldStruct = TwoFieldStruct Int Text.Text+ deriving stock (Eq, Show, Generic)++data LargeSumType+ = LargeC1 Int+ | LargeC2 Int+ | LargeC3 Int+ | LargeC4 Int+ | LargeC5 Int+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB LargeSumType)++data EmptyStruct = EmptyStruct+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB EmptyStruct)++-- | A field of the type EmptyStruct uses the generic encoding.+instance DuckValue EmptyStruct where+ duckToField = genericToFieldValue+ duckFromField = genericFromFieldValue+ duckLogicalType = genericLogicalType++-- | A record with two optional fields of a type without fields.+data OptionalEmpty = OptionalEmpty+ { oeFirst :: Maybe EmptyStruct+ , oeSecond :: Maybe EmptyStruct+ }+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB OptionalEmpty)++-- | A record with a field of a type without fields.+data WithEmpty = WithEmpty+ { weId :: Int64+ , weEmpty :: EmptyStruct+ }+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB WithEmpty)++-- | A sum with a constructor that has a field and a constructor without fields.+data Tally+ = Counted Int64+ | Uncounted+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB Tally)++data OptionalFields = OptionalFields+ { ofInt :: Maybe Int+ , ofText :: Maybe Text.Text+ }+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB OptionalFields)++instance FromRow WithRemaining where+ fromRow = do+ firstVal <- field+ remaining <- numFieldsRemaining+ replicateM_ remaining (fieldWith (const (pure ())))+ pure (WithRemaining firstVal remaining)++newtype YesNo = YesNo Bool+ deriving (Eq, Show)++instance FromRow YesNo where+ fromRow = parseYes <|> parseNo+ where+ parseYes = YesNo True <$ fieldWith (match (Text.pack "yes"))+ parseNo = YesNo False <$ fieldWith (match (Text.pack "no"))++ match expected fld@Field{} = do+ txt <- fromField fld+ if Text.toLower txt == expected+ then pure ()+ else returnError ConversionFailed fld "failed to match exact string"++newtype NonEmptyText = NonEmptyText Text.Text+ deriving (Eq, Show)++nonEmptyTextParser :: FieldParser NonEmptyText+nonEmptyTextParser f@Field{} = do+ txt <- fromField f+ if Text.null txt+ then returnError ConversionFailed f "NonEmptyText requires a non-empty string"+ else pure (NonEmptyText txt)++instance FromField NonEmptyText where+ fromField = nonEmptyTextParser++-- | Run each case with a time limit of 10 minutes. Thus a deadlock fails its+-- case, and not the job. The limit is long, because the suite also runs under+-- Valgrind. The limit cannot stop a thread in a native call. A time limit of+-- the command line (@--timeout@) replaces this limit.+main :: IO ()+main = defaultMain (adjustOption limit tests)+ where+ limit NoTimeout = mkTimeout (10 * 60 * 1000000)+ limit given = given++tests :: TestTree+tests =+ testGroup+ "duckdb-simple"+ [ connectionTests+ , coreRegressionTests+ , arrowTests+ , cancellationTests+ , callbackLockTests+ , nativeStreamingTests+ , extensionRegressionTests+ , registrationTests+ , typedRegistrationTests+ , copyTests+ , GeometryRegressionTests.tests+ , VariantRegressionTests.tests+ , userCastTests+ , geometryArgumentTests+ , valueRegressionTests+ , sizedArrayTests+ , timeTests+ , SqlFeatureTests.tests+ , withConnectionTests+ , statementTests+ , roundTripTests+ , typeTests+ , streamingTests+ , functionsTests+ , v20Tests+ , transactionTests+ ]++connectionTests :: TestTree+connectionTests =+ testGroup+ "open/close"+ [ testCase "opens and closes an in-memory database" $ do+ conn <- open ":memory:"+ close conn+ , testCase "allows closing the same connection twice" $ do+ conn <- open ":memory:"+ close conn+ close conn+ , testCase "opens with startup config" $ do+ conn <- openWithConfig ":memory:" [("threads", "1")]+ mValue <- Config.getConfigOption conn "threads"+ close conn+ case mValue of+ Nothing -> assertFailure "expected threads config value"+ Just value -> Config.configValueText value @?= Just "1"+ ]++v20Tests :: TestTree+v20Tests =+ testGroup+ "DuckDB 2.0 wrappers"+ [ testCase "lists config flags" $ do+ flags <- withConnection ":memory:" Config.listConfigFlags+ assertBool "expected at least one config flag" (not (null flags))+ assertBool "expected threads flag" (any ((== "threads") . Config.configFlagName) flags)+ , testCase "catalog lookup works inside a transaction" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE catalog_probe(id INTEGER)"+ withTransaction conn $ do+ mType <- Catalog.catalogTypeName conn "memory"+ assertBool "expected catalog type name" (mType /= Nothing)+ mEntry <- Catalog.lookupCatalogEntry conn "memory" "main" "catalog_probe" Catalog.CatalogTable+ case mEntry of+ Nothing -> assertFailure "expected catalog entry"+ Just entry -> Catalog.catalogEntryName entry @?= "catalog_probe"+ , testCase "file system wrappers round-trip bytes" $ withTemporaryPath \path -> do+ let payload = BS.pack [0, 1, 2, 3, 255]+ withConnection ":memory:" \conn -> do+ FileSystem.withFileHandle conn path [FileSystem.FileFlagCreate, FileSystem.FileFlagWrite] \handle -> do+ written <- FileSystem.writeFileHandleBytes handle payload+ written @?= fromIntegral (BS.length payload)+ FileSystem.fileHandleSync handle+ FileSystem.withFileHandle conn path [FileSystem.FileFlagRead] \handle -> do+ size <- FileSystem.fileHandleSize handle+ size @?= fromIntegral (BS.length payload)+ bytes <- FileSystem.readFileHandleChunk handle 32+ bytes @?= payload+ removeFile path+ , testCase "copy wrappers receive COPY TO rows" $+ withTemporaryPath \path -> withConnection ":memory:" \conn -> do+ capturedRef <- newIORef ([] :: [Int])+ pathRef <- newIORef Nothing+ Copy.registerCopyToFunction+ conn+ "hs_capture"+ (\bindInfo -> pure (length (Copy.copyBindColumnTypes bindInfo)))+ ( \initInfo -> do+ atomicModifyIORef' pathRef \_ -> (Just (Copy.copyInitFilePath initInfo), ())+ pure (Copy.copyInitBindState initInfo)+ )+ ( \sinkInfo rows -> do+ Copy.copySinkBindState sinkInfo @?= 1+ Copy.copySinkGlobalState sinkInfo @?= 1+ values <- mapM fieldInt rows+ atomicModifyIORef' capturedRef \existing -> (existing <> values, ())+ )+ ( \finalizeInfo -> do+ Copy.copyFinalizeBindState finalizeInfo @?= 1+ Copy.copyFinalizeGlobalState finalizeInfo @?= 1+ )+ _ <- execute conn "COPY (SELECT * FROM (VALUES (1), (2), (3)) AS t(x)) TO ? (FORMAT hs_capture)" (Only (Text.pack path))+ captured <- readIORef capturedRef+ filePath <- readIORef pathRef+ captured @?= [1, 2, 3]+ filePath @?= Just path+ , testCase "custom log storage registers a native sink" do+ entries <- newIORef []+ withConnection ":memory:" \conn -> do+ Logging.registerLogStorage conn "hs_log_capture" (\entry -> atomicModifyIORef' entries (\values -> (entry : values, ())))+ void $ execute_ conn "CALL enable_logging(); SET logging_storage = 'hs_log_capture'; SELECT write_log('registered')"+ -- The sink receives the entry of write_log one time.+ readIORef entries >>= (@?= ["registered"]) . filter (== "registered") . map Logging.logEntryMessage+ ]++fieldInt :: [Field] -> IO Int+fieldInt [fld] =+ runConversion (fromField fld) >>= \case+ Errors err -> assertFailure (show err) >> pure 0+ Ok value -> pure value+fieldInt fields = assertFailure ("expected one field, got " <> show (length fields)) >> pure 0++withConnectionTests :: TestTree+withConnectionTests =+ testGroup+ "withConnection"+ [ testCase "returns the action result" $ do+ result <- withConnection ":memory:" \_ -> pure (21 :: Int)+ assertEqual "action result" 21 result+ , testCase "propagates exceptions from the action"+ $ assertThrowsErrorCall+ $ withConnection ":memory:" (\_ -> error "boom" :: IO ())+ ]++statementTests :: TestTree+statementTests =+ testGroup+ "statements"+ [ testCase "prepares and closes a simple statement" $+ withConnection ":memory:" \conn -> do+ stmt <- openStatement conn "SELECT 42"+ closeStatement stmt+ , testCase "allows closing a statement twice" $+ withConnection ":memory:" \conn -> do+ stmt <- openStatement conn "SELECT 1"+ closeStatement stmt+ closeStatement stmt+ , testCase "rejects preparing statements on a closed connection" $ do+ conn <- open ":memory:"+ close conn+ assertThrowsSQLError $+ openStatement conn "SELECT 1"+ , testCase "throws SQLError for invalid SQL" $+ withConnection ":memory:" \conn ->+ assertThrowsSQLError $+ openStatement conn "THIS IS NOT VALID SQL"+ , testCase "withStatement closes the statement automatically" $+ withConnection ":memory:" \conn -> do+ (result, stmt) <- withStatement conn "SELECT 1" \stmt -> pure ("done" :: String, stmt)+ assertEqual "withStatement result" "done" result+ -- A read of the closed statement raises an error.+ assertThrowsSQLError (nextRow stmt :: IO (Maybe (Only Int)))+ , testCase "withStatement closes statements even when the action fails" $+ withConnection ":memory:" \conn -> do+ used <- newIORef Nothing+ assertThrowsErrorCall $+ withStatement conn "SELECT 1" (\stmt -> atomicModifyIORef' used (const (Just stmt, ())) >> error "boom" :: IO ())+ -- A read of the closed statement raises an error.+ readIORef used+ >>= maybe (assertFailure "the action did not run") (\closed -> assertThrowsSQLError (nextRow closed :: IO (Maybe (Only Int))))+ stmt <- openStatement conn "SELECT 1"+ closeStatement stmt+ , testCase "execute returns affected row count" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE test_exec (x INTEGER)"+ count <- execute conn "INSERT INTO test_exec VALUES (?)" (Only (1 :: Int))+ assertEqual "rows affected" 1 count+ , testCase "execute runs with positional parameters" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE exec_params (a INTEGER, b TEXT)"+ count <- execute conn "INSERT INTO exec_params VALUES (?, ?)" (5 :: Int, "hi" :: String)+ assertEqual "rows affected" 1 count+ , testCase "rejects invalid direct execution" $+ withConnection ":memory:" \conn ->+ assertThrowsSQLError $+ execute_ conn "THIS IS NOT SQL"+ , testCase "binds positional parameters" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE params (a INTEGER, b TEXT)"+ withStatement conn "INSERT INTO params VALUES (?, ?)" \stmt -> do+ bind stmt [toField (5 :: Int), toField ("hi" :: String)]+ changed <- executeStatement stmt+ assertEqual "rows affected" 1 changed+ , testCase "executeMany inserts each row and gives the total count" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE params_many (a INTEGER, b TEXT)"+ total <- executeMany conn "INSERT INTO params_many VALUES (?, ?)" [(1 :: Int, "x" :: String), (2 :: Int, "y" :: String)]+ assertEqual "rows affected" 2 total+ query_ conn "SELECT a, b FROM params_many ORDER BY a" >>= (@?= [(1 :: Int, "x" :: String), (2, "y")])+ , testCase "executeNamed binds named parameters" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE named_params (a INTEGER, b TEXT)"+ count <- executeNamed conn "INSERT INTO named_params VALUES ($a, $b)" ["$a" := (1 :: Int), "$b" := ("named" :: String)]+ assertEqual "rows affected" 1 count+ rows <- queryNamed conn "SELECT a FROM named_params WHERE b = $label" ["$label" := ("named" :: String)]+ assertEqual "named query" [Only (1 :: Int)] rows+ , testCase "rejects incorrect positional argument counts" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE param_count (a INTEGER, b INTEGER)"+ assertThrowsFormatError+ (execute conn "INSERT INTO param_count VALUES (?, ?)" (Only (1 :: Int)))+ (Text.isInfixOf "parameter(s)" . formatErrorMessage)+ , testCase "rejects positional bindings on named statements" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE mixed_mode (a INTEGER)"+ assertThrowsFormatError+ (execute conn "INSERT INTO mixed_mode VALUES ($a)" (Only (1 :: Int)))+ (Text.isInfixOf "named parameters" . formatErrorMessage)+ , testCase "reports error when mixing positional and named parameters" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE mixed_params (a INTEGER, b TEXT)"+ assertThrows+ ( withStatement conn "INSERT INTO mixed_params VALUES (?, $label)" \stmt -> do+ bind stmt [toField (1 :: Int)]+ bindNamed stmt ["$label" := ("combo" :: String)]+ _ <- executeStatement stmt+ pure ()+ )+ ( \(err :: SQLError) ->+ Text.isInfixOf "Mixing named and positional parameters" (sqlErrorMessage err)+ )+ , testCase "generic FromRow derivation works" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE person (id INTEGER, name TEXT)"+ _ <- executeMany conn "INSERT INTO person VALUES (?, ?)" [(1 :: Int, "Alice" :: String), (2 :: Int, "Bob" :: String)]+ people <- query_ conn "SELECT id, name FROM person ORDER BY id" :: IO [Person]+ assertEqual "person rows" [Person 1 (Text.pack "Alice"), Person 2 (Text.pack "Bob")] people+ , testCase "(:.) composes row parsing and parameter encoding" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE dot_pair (a INTEGER, b INTEGER, label TEXT)"+ let payload = Only (7 :: Int) :. (8 :: Int, Text.pack "hi")+ _ <- execute conn "INSERT INTO dot_pair VALUES (?, ?, ?)" payload+ rows <- query_ conn "SELECT a, b, label FROM dot_pair" :: IO [Only Int :. (Int, Text.Text)]+ case rows of+ [Only a :. (b, label)] -> do+ assertEqual "(:.) first" 7 a+ assertEqual "(:.) second" 8 b+ assertEqual "(:.) label" (Text.pack "hi") label+ other -> assertFailure ("unexpected (:.) rows: " <> show other)+ , testCase "query_ fetches rows" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE query_rows (a INTEGER, b TEXT)"+ _ <- executeMany conn "INSERT INTO query_rows VALUES (?, ?)" [(1 :: Int, "x" :: String), (2 :: Int, "y" :: String)]+ rows <- query_ conn "SELECT a, b FROM query_rows ORDER BY a"+ assertEqual "query rows" [(1 :: Int, "x" :: String), (2 :: Int, "y" :: String)] rows+ , testCase "query decodes NULL as Maybe" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE maybe_vals (a TEXT)"+ _ <- execute conn "INSERT INTO maybe_vals VALUES (?)" (Only (Just ("present" :: String)))+ _ <- execute conn "INSERT INTO maybe_vals VALUES (?)" (Only (Nothing :: Maybe String))+ rows <- query_ conn "SELECT a FROM maybe_vals ORDER BY a IS NULL, a"+ assertEqual "maybe decoding" [Only (Just ("present" :: String)), Only Nothing] rows+ , testCase "query fetches rows with parameters" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE query_params (a INTEGER, b TEXT)"+ _ <- executeMany conn "INSERT INTO query_params VALUES (?, ?)" [(1 :: Int, "x" :: String), (2 :: Int, "y" :: String)]+ rows <- query conn "SELECT a FROM query_params WHERE b = ?" (Only ("y" :: String))+ assertEqual "query result" [Only (2 :: Int)] rows+ , testCase "column mismatch surfaces as SQLError" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE mismatch (a INTEGER, b INTEGER)"+ _ <- execute_ conn "INSERT INTO mismatch VALUES (1, 2)"+ assertThrows+ (query_ conn "SELECT a, b FROM mismatch" :: IO [Only Int])+ (Text.isInfixOf "column index" . sqlErrorMessage)+ , testCase "clears statement bindings without error" $+ withConnection ":memory:" \conn -> do+ stmt <- openStatement conn "SELECT ?"+ clearStatementBindings stmt+ closeStatement stmt+ , testCase "resolves named parameter indices" $+ withConnection ":memory:" \conn -> do+ stmt <- openStatement conn "SELECT $named_param"+ idx <- namedParameterIndex stmt "named_param"+ assertEqual "named parameter index" (Just 1) idx+ idxWithPrefix <- namedParameterIndex stmt "$named_param"+ assertEqual "named parameter index with prefix" (Just 1) idxWithPrefix+ colonIdx <- namedParameterIndex stmt ":named_param"+ assertEqual "named parameter index with colon" (Just 1) colonIdx+ none <- namedParameterIndex stmt "missing"+ assertEqual "missing parameter index" Nothing none+ closeStatement stmt+ , testCase "reports column metadata for statements" $+ withConnection ":memory:" \conn -> do+ stmt <- openStatement conn "SELECT 1 AS a, 2 AS b"+ count <- columnCount stmt+ assertEqual "column count" 2 count+ name0 <- columnName stmt 0+ name1 <- columnName stmt 1+ assertEqual "column names" [Text.pack "a", Text.pack "b"] [name0, name1]+ assertThrows+ (columnName stmt 2)+ (Text.isInfixOf "out of bounds" . sqlErrorMessage)+ closeStatement stmt+ , testCase "bindNamed rejects statements without named placeholders" $+ withConnection ":memory:" \conn -> do+ stmt <- openStatement conn "SELECT ?"+ assertThrowsFormatError+ (bindNamed stmt [":value" := (1 :: Int)])+ (Text.isInfixOf "does not define named parameters" . formatErrorMessage)+ closeStatement stmt+ , testCase "bindNamed rejects unknown parameters" $+ withConnection ":memory:" \conn -> do+ stmt <- openStatement conn "SELECT $known"+ assertThrowsFormatError+ (bindNamed stmt ["$missing" := (1 :: Int)])+ (Text.isInfixOf "unknown named parameter" . formatErrorMessage)+ closeStatement stmt+ , testCase "numFieldsRemaining reports remaining columns" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE remaining (a INTEGER, b INTEGER, c INTEGER)"+ _ <- execute conn "INSERT INTO remaining VALUES (?, ?, ?)" (1 :: Int, 2 :: Int, 3 :: Int)+ rows <- query_ conn "SELECT a, b, c FROM remaining" :: IO [WithRemaining]+ assertEqual "remaining count" [WithRemaining 1 2] rows+ , testCase "RowParser alternatives fall back" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE yesno (answer TEXT)"+ _ <- executeMany conn "INSERT INTO yesno VALUES (?)" [Only ("yes" :: String), Only ("no" :: String)]+ rows <- query_ conn "SELECT answer FROM yesno ORDER BY answer" :: IO [YesNo]+ assertEqual "yes/no parsing" [YesNo False, YesNo True] rows+ ]++data DuckDBExpectation+ = ExpectEquals FieldValue+ | ExpectSatisfies (FieldValue -> Assertion)+ | ExpectException (SomeException -> Assertion)++data DuckDBExpression+ = CastExpression Text.Text Text.Text+ | DirectExpression Text.Text++data DuckDBCastCase = DuckDBCastCase+ { castLabel :: String+ , castExpression :: DuckDBExpression+ , castExpectation :: DuckDBExpectation+ }++duckdbTypeCastTests :: TestTree+duckdbTypeCastTests =+ testGroup+ "duckdb type casts"+ (map buildCase duckdbCastCases)+ where+ buildCase DuckDBCastCase{castLabel, castExpression, castExpectation} =+ testCase castLabel $+ withConnection ":memory:" \conn -> do+ let sql =+ Query+ ( case castExpression of+ CastExpression expressionValue expressionType ->+ Text.concat ["SELECT CAST(", expressionValue, " AS ", expressionType, ")"]+ DirectExpression expressionSQL ->+ Text.concat ["SELECT ", expressionSQL]+ )+ result <- try (query_ conn sql) :: IO (Either SomeException [Only FieldValue])+ runExpectation castLabel castExpectation result++runExpectation :: String -> DuckDBExpectation -> Either SomeException [Only FieldValue] -> Assertion+runExpectation label expectation outcome =+ case expectation of+ ExpectEquals expected ->+ case outcome of+ Right [Only fieldValue] ->+ assertEqual (label <> " result mismatch") expected fieldValue+ Right other ->+ assertFailure (label <> " expected single row, got " <> show other)+ Left err ->+ assertFailure (label <> " expected success, but query failed: " <> displayException err)+ ExpectSatisfies predicate ->+ case outcome of+ Right [Only fieldValue] ->+ predicate fieldValue+ Right other ->+ assertFailure (label <> " expected single row, got " <> show other)+ Left err ->+ assertFailure (label <> " expected success, but query failed: " <> displayException err)+ ExpectException predicate ->+ case outcome of+ Left err ->+ predicate err+ Right rows ->+ assertFailure (label <> " expected failure, but query succeeded with " <> show rows)++duckdbCastCases :: [DuckDBCastCase]+duckdbCastCases =+ [ successCase "BOOLEAN" (quoted "true") "BOOLEAN" (ExpectEquals (FieldBool True))+ , successCase "TINYINT" (quotedValue tinyIntValue) "TINYINT" (ExpectEquals (FieldInt8 tinyIntValue))+ , successCase "SMALLINT" (quotedValue smallIntValue) "SMALLINT" (ExpectEquals (FieldInt16 smallIntValue))+ , successCase "INTEGER" (quotedValue intValue) "INTEGER" (ExpectEquals (FieldInt32 intValue))+ , successCase "BIGINT" (quotedValue bigIntValue) "BIGINT" (ExpectEquals (FieldInt64 bigIntValue))+ , successCase "UTINYINT" (quotedValue uTinyValue) "UTINYINT" (ExpectEquals (FieldWord8 uTinyValue))+ , successCase "USMALLINT" (quotedValue uSmallValue) "USMALLINT" (ExpectEquals (FieldWord16 uSmallValue))+ , successCase "UINTEGER" (quotedValue uIntValue) "UINTEGER" (ExpectEquals (FieldWord32 uIntValue))+ , successCase "UBIGINT" (quotedValue uBigValue) "UBIGINT" (ExpectEquals (FieldWord64 uBigValue))+ , successCase "FLOAT" (quoted "3.25") "FLOAT" (ExpectEquals (FieldFloat floatValue))+ , successCase "DOUBLE" (quoted "2.5") "DOUBLE" (ExpectEquals (FieldDouble doubleValue))+ , successCase "TIMESTAMP" (quoted "2024-01-02 03:04:05.123456") "TIMESTAMP" (ExpectEquals (FieldTimestamp (Finite timestampLocalTimeMicros)))+ , successCase "DATE" (quoted "2024-10-12") "DATE" (ExpectEquals (FieldDate (Finite dateValue)))+ , successCase "TIME" (quoted "14:30:15.123456") "TIME" (ExpectEquals (FieldTime timeValue))+ , successCase "INTERVAL" (quoted "1 day") "INTERVAL" (ExpectEquals (FieldInterval intervalValue))+ , successCase "HUGEINT" (quotedValue hugeIntLiteral) "HUGEINT" (ExpectEquals (FieldHugeInt hugeIntLiteral))+ , successCase "UHUGEINT" (quotedValue uHugeIntLiteral) "UHUGEINT" (ExpectEquals (FieldUHugeInt uHugeIntLiteral))+ , successCase "VARCHAR" (quoted varcharText) "VARCHAR" (ExpectEquals (FieldText varcharText))+ , successCase "BLOB" (quoted "duckdb") "BLOB" (ExpectEquals (FieldBlob blobValue))+ , successCase "DECIMAL" (quoted "12345.6789") "DECIMAL(18,4)" (ExpectEquals (FieldDecimal decimalValue))+ , successCase "TIMESTAMP_S" (quoted "2024-01-02 03:04:05") "TIMESTAMP_S" (ExpectEquals (FieldTimestamp (Finite timestampLocalTimeSeconds)))+ , successCase+ "TIMESTAMP_MS"+ (quoted "2024-01-02 03:04:05.123")+ "TIMESTAMP_MS"+ (ExpectEquals (FieldTimestamp (Finite timestampLocalTimeMillis)))+ , successCase+ "TIMESTAMP_NS"+ (quoted "2024-01-02 03:04:05.123456789")+ "TIMESTAMP_NS"+ (ExpectEquals (FieldTimestamp (Finite timestampLocalTimeNanos)))+ , successCase "ENUM" (quoted "beta") "ENUM('alpha','beta')" (ExpectEquals (FieldEnum 1 "beta"))+ , successCase "LIST" (quoted "[1,2,3]") "INTEGER[]" (ExpectEquals (FieldList listElements))+ , successDirect "MAP" "MAP(['a','b'], [1,2])" (expectMapEntries mapPairs)+ , successCase "TIMETZ" (quoted "03:04:05+02:30") "TIME WITH TIME ZONE" (ExpectEquals (FieldTimeTZ timeWithZoneValue))+ , successCase+ "TIMESTAMPTZ"+ (quoted "2024-01-02 03:04:05+02:30")+ "TIMESTAMP WITH TIME ZONE"+ (ExpectEquals (FieldTimestampTZ (Finite timestampTzUtc)))+ , successCase "BIGNUM" (quoted bigNumLiteralText) "BIGNUM" (ExpectEquals (FieldBigNum (BigNum bigNumLiteral)))+ , successCase "UUID" (quoted $ UUID.toText uuid) "UUID" (ExpectEquals (FieldUUID uuid))+ , successCase "BIT" (quoted $ Text.pack $ show bits) "BIT" (ExpectEquals (FieldBit bits))+ , successCase "ARRAY" (quoted "[1,2,3]") "INTEGER[3]" (ExpectEquals (FieldArray arrayElements))+ , successCase "TIME_NS" (quoted "03:04:05.123456789") "TIME_NS" (ExpectEquals (FieldTime (TimeOfDay 3 4 5.123456789)))+ , successDirect "STRUCT" "{'a': 1, 'b': 2}" (expectStruct structFields)+ , successDirect "UNION" "CAST(union_value(a := 42) AS UNION(a INTEGER, b VARCHAR))" (expectUnion 0 (Text.pack "a") (FieldInt32 42))+ , -- These can never surface from a query, only internally.+ failCaseOK "ANY" (quoted "1") "ANY" (expectSQLErrorContaining "ANY")+ , failCaseOK "STRING_LITERAL" (quoted literalText) "STRING_LITERAL" (expectSQLErrorContaining "STRING_LITERAL")+ , failCaseOK "INTEGER_LITERAL" (quoted "123") "INTEGER_LITERAL" (expectSQLErrorContaining "INTEGER_LITERAL")+ , failCaseOK "INVALID" "0" "INVALID" (expectSQLErrorContaining "INVALID")+ , failCaseOK "SQLNULL" "NULL" "SQLNULL" (expectSQLErrorContaining "SQLNULL")+ ]+ where+ bits = BitString 7 $ BS.pack [1, 255]+ quoted t = Text.concat ["'", t, "'"]+ quotedValue v = quoted (valueText v)+ valueText :: (Show a) => a -> Text.Text+ valueText = Text.pack . show+ successCase label value ty expectation =+ DuckDBCastCase+ { castLabel = label+ , castExpression = CastExpression value ty+ , castExpectation = expectation+ }+ successDirect label expr expectation =+ DuckDBCastCase+ { castLabel = label+ , castExpression = DirectExpression expr+ , castExpectation = expectation+ }+ failCaseOK label value ty expectation =+ DuckDBCastCase+ { castLabel = label+ , castExpression = CastExpression value ty+ , castExpectation = expectation+ }+ expectMapEntries expectedPairs =+ ExpectSatisfies $+ \case+ FieldMap actualPairs ->+ let normalize = sortOn (mapKey . fst)+ mapKey (FieldText txt) = txt+ mapKey other = Text.pack (show other)+ in assertEqual "map entries" (normalize expectedPairs) (normalize actualPairs)+ other ->+ assertFailure ("expected FieldMap, but saw " <> show other)+ structFields =+ [ StructField{structFieldName = Text.pack "a", structFieldValue = FieldInt32 1}+ , StructField{structFieldName = Text.pack "b", structFieldValue = FieldInt32 2}+ ]+ expectStruct expectedFields =+ ExpectSatisfies $+ \case+ FieldStruct StructValue{structValueFields, structValueTypes, structValueIndex} -> do+ let actualFields = elems structValueFields+ actualTypes = elems structValueTypes+ expectedIndex = Map.fromList (zip (map structFieldName expectedFields) [0 ..])+ typeNames = map structFieldName actualTypes+ assertEqual "struct field names" (map structFieldName expectedFields) (map structFieldName actualFields)+ assertEqual "struct field values" (map structFieldValue expectedFields) (map structFieldValue actualFields)+ assertEqual "struct type field names" (map structFieldName expectedFields) typeNames+ assertEqual "struct index map" expectedIndex structValueIndex+ other ->+ assertFailure ("expected FieldStruct, but saw " <> show other)+ expectUnion expectedIndex expectedLabel expectedPayload =+ ExpectSatisfies $+ \case+ FieldUnion UnionValue{unionValueIndex, unionValueLabel, unionValuePayload, unionValueMembers} -> do+ assertEqual "union tag index" expectedIndex unionValueIndex+ assertEqual "union tag label" expectedLabel unionValueLabel+ assertEqual "union payload" expectedPayload unionValuePayload+ let actualMembers = elems unionValueMembers+ expectedMembers = [Text.pack "a", Text.pack "b"]+ assertEqual "union member names" expectedMembers (map unionMemberName actualMembers)+ other ->+ assertFailure ("expected FieldUnion, but saw " <> show other)+ expectSQLErrorContaining :: Text.Text -> DuckDBExpectation+ expectSQLErrorContaining needle =+ ExpectException $ \err ->+ case fromException err :: Maybe SQLError of+ Just sqlErr ->+ let message = sqlErrorMessage sqlErr+ in assertBool+ ("expected SQLError containing " <> Text.unpack needle <> ", but saw: " <> Text.unpack message)+ (Text.isInfixOf needle message)+ Nothing ->+ assertFailure ("expected SQLError, but saw " <> displayException err)+ bsFromString :: String -> BS.ByteString+ bsFromString = BS.pack . map (fromIntegral . fromEnum)+ secondsWithFraction :: Integer -> Integer -> Integer -> Rational+ secondsWithFraction whole numerator denominator =+ fromIntegral whole + numerator % denominator+ tinyIntValue :: Int8+ tinyIntValue = 42+ smallIntValue :: Int16+ smallIntValue = 32000+ intValue :: Int32+ intValue = 2147483647+ bigIntValue :: Int64+ bigIntValue = 9223372036854775807+ uTinyValue :: Word8+ uTinyValue = 200+ uSmallValue :: Word16+ uSmallValue = 60000+ uIntValue :: Word32+ uIntValue = maxBound+ uBigValue :: Word64+ uBigValue = maxBound+ floatValue :: Float+ floatValue = 3.25+ doubleValue :: Double+ doubleValue = 2.5+ timestampDay = fromGregorian 2024 1 2+ timestampLocalTimeMicros =+ LocalTime timestampDay (TimeOfDay 3 4 (fromRational (secondsWithFraction 5 123456 1000000)))+ timestampLocalTimeSeconds =+ LocalTime timestampDay (TimeOfDay 3 4 5)+ timestampLocalTimeMillis =+ LocalTime timestampDay (TimeOfDay 3 4 (fromRational (secondsWithFraction 5 123 1000)))+ timestampLocalTimeNanos =+ LocalTime timestampDay (TimeOfDay 3 4 (fromRational (secondsWithFraction 5 123456789 1000000000)))+ dateValue = fromGregorian 2024 10 12+ timeValue =+ TimeOfDay 14 30 (fromRational (secondsWithFraction 15 123456 1000000))+ intervalValue =+ IntervalValue{intervalMonths = 0, intervalDays = 1, intervalMicros = 0}+ decimalValue =+ DecimalValue{decimalWidth = 18, decimalScale = 4, decimalInteger = 123456789}+ listElements =+ [FieldInt32 1, FieldInt32 2, FieldInt32 3]+ arrayElements =+ listArray (0, 2) listElements+ mapPairs =+ [ (FieldText (Text.pack "a"), FieldInt32 1)+ , (FieldText (Text.pack "b"), FieldInt32 2)+ ]+ uuid :: UUID.UUID+ uuid = fromJust $ UUID.fromText "123e4567-e89b-12d3-a456-426614174000"+ timeWithZoneValue =+ TimeWithZone{timeWithZoneTime = TimeOfDay 3 4 5, timeWithZoneOffsetSeconds = 9000}+ timestampTzUtc =+ UTCTime timestampDay (secondsToDiffTime 2045)+ bigNumLiteral :: Integer+ bigNumLiteral = 1234567890123456789012345678901234567890+ bigNumLiteralText = valueText bigNumLiteral+ varcharText :: Text.Text+ varcharText = "Hello, DuckDB"+ blobValue = bsFromString "duckdb"+ hugeIntLiteral :: Integer+ hugeIntLiteral = (2 ^ (120 :: Int)) + 123456789+ uHugeIntLiteral :: Integer+ uHugeIntLiteral = (2 ^ (128 :: Int)) - 1+ literalText :: Text.Text+ literalText = "literal"++typeTests :: TestTree+typeTests =+ testGroup+ "types"+ [ duckdbTypeCastTests+ , testCase "round-trips date/time/timestamp" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE temporals (d DATE, t TIME, ts TIMESTAMP)"+ let dayVal = fromGregorian 2024 10 12+ timeVal = TimeOfDay 14 30 15.123456+ tsVal = LocalTime dayVal timeVal+ _ <- execute conn "INSERT INTO temporals VALUES (?, ?, ?)" (dayVal, timeVal, tsVal)+ [(dRes, tRes, tsRes)] <- query_ conn "SELECT d, t, ts FROM temporals"+ assertEqual "date round-trip" dayVal dRes+ assertEqual "time round-trip" timeVal tRes+ assertEqual "timestamp round-trip" tsVal tsRes+ -- A TIMESTAMP does not read as UTCTime. AT TIME ZONE 'UTC' gives the instant in UTC.+ [Only utcRes] <- query_ conn "SELECT ts::TIMESTAMP AT TIME ZONE 'UTC' FROM temporals" :: IO [Only UTCTime]+ assertEqual "timestamp at UTC" (localTimeToUTC utc tsVal) utcRes+ (try (query_ conn "SELECT ts FROM temporals" :: IO [Only UTCTime])) >>= \case+ Left SQLError{sqlErrorMessage} -> sqlErrorMessage @?= "duckdb-simple: column ts has type TIMESTAMP but expected UTCTime"+ Right rows -> assertFailure ("expected Incompatible, got " <> show rows)+ , testCase "round-trips blob payloads" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE blobs (payload BLOB)"+ let payload = BS.pack [0, 1, 2, 3, 255]+ _ <- execute conn "INSERT INTO blobs VALUES (?)" (Only payload)+ [Only blobOut] <- query_ conn "SELECT payload FROM blobs"+ assertEqual "blob round-trip" payload blobOut+ , testCase "round-trips unsigned integers" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE unsigneds (u8 UTINYINT, u16 USMALLINT, u32 UINTEGER, u64 UBIGINT)"+ let w8 = 200 :: Word8+ w16 = 60000 :: Word16+ w32 = 4000000000 :: Word32+ w64 = maxBound :: Word64+ _ <- execute conn "INSERT INTO unsigneds VALUES (?, ?, ?, ?)" (w8, w16, w32, w64)+ [(r8, r16, r32, r64)] <- query_ conn "SELECT u8, u16, u32, u64 FROM unsigneds"+ assertEqual "Word8 round-trip" w8 r8+ assertEqual "Word16 round-trip" w16 r16+ assertEqual "Word32 round-trip" w32 r32+ assertEqual "Word64 round-trip" w64 r64+ [Only (asWord :: Word)] <- query_ conn "SELECT u64 FROM unsigneds"+ assertEqual "Word from UBIGINT" (fromIntegral w64) asWord+ , testCase "round-trips fixed-length arrays" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE arrays (vals INTEGER[3])"+ let inputs =+ [ sized [1, 2, 3]+ , sized [4, 5, 6]+ ]+ _ <- executeMany conn "INSERT INTO arrays VALUES (?)" (Only <$> inputs)+ rows <- query_ conn "SELECT vals FROM arrays ORDER BY rowid" :: IO [Only (SizedArray 3 Int)]+ let expected =+ fmap Only inputs+ assertEqual "array round-trip" expected rows+ , testCase "decodes huge integers as Integer" $+ withConnection ":memory:" \conn -> do+ let hugeValue = 170141183460469231731687303715884105727 :: Integer+ [Only (hugeOut :: Integer)] <-+ query_ conn "SELECT 170141183460469231731687303715884105727::HUGEINT"+ assertEqual "hugeint" hugeValue hugeOut+ , testCase "decodes huge integers as Natural" $+ withConnection ":memory:" \conn -> do+ let hugeValue = 170141183460469231731687303715884105727 :: Natural+ [Only (naturalOut :: Natural)] <-+ query_ conn "SELECT 170141183460469231731687303715884105727::HUGEINT"+ assertEqual "hugeint natural" hugeValue naturalOut+ , testCase "decodes unsigned huge integers as Natural" $+ withConnection ":memory:" \conn -> do+ let uhValue = 170141183460469231731687303715884105727 :: Natural+ [Only (naturalOut :: Natural)] <-+ query_ conn "SELECT 170141183460469231731687303715884105727::UHUGEINT"+ assertEqual "uhugeint natural" uhValue naturalOut+ , testCase "FromField Integer consumes FieldBigNum" $ do+ let original = 1234567899876543210123456789 :: Integer+ bigField =+ Field+ { fieldName = "big"+ , fieldIndex = 0+ , fieldValue = FieldBigNum (BigNum original)+ , fieldType = pure (LogicalTypeScalar LogicalTypeIdBigNum)+ }+ runConversion (fromField bigField :: Conversion Integer) >>= \case+ Ok result -> assertEqual "FieldBigNum Integer" original result+ Errors err -> assertFailure ("unexpected parse failure: " <> show err)+ , testCase "FromField Natural rejects negative FieldBigNum" $ do+ let bigValue = BigNum (-5)+ bigField =+ Field+ { fieldName = "big"+ , fieldIndex = 0+ , fieldValue = FieldBigNum bigValue+ , fieldType = pure (LogicalTypeScalar LogicalTypeIdBigNum)+ }+ runConversion (fromField bigField :: Conversion Natural) >>= \case+ Errors _ -> pure ()+ Ok result -> assertFailure ("expected failure, got " <> show result)+ , testCase "BigNum conversions round-trip Integer" $ do+ let original = (2 ^ (200 :: Int)) + 8675309 :: Integer+ converted = toBigNumBytes original+ assertEqual "round-trip Integer" original (fromBigNumBytes converted)+ , testProperty "fromBigNumBytes . toBigNumBytes is identity" \(n :: Integer) ->+ fromBigNumBytes (toBigNumBytes n) === n+ , testCase "selects BIGNUM values as BigNum" $+ withConnection ":memory:" \conn -> do+ rows <- concat <$> mapM (\bv -> query_ conn $ fromString $ "SELECT CAST('" <> show bv <> "' AS BIGNUM)") bigNumValues+ assertEqual "BIGNUM results" (fmap (Only . BigNum) bigNumValues) rows+ , testCase "round-trips BIGNUM values through the database" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE bignum_roundtrip (val BIGNUM)"+ _ <- executeMany conn "INSERT INTO bignum_roundtrip VALUES (?)" (fmap (Only . BigNum) bigNumValues)+ rows <- query_ conn "SELECT val FROM bignum_roundtrip ORDER BY rowid" :: IO [Only Integer]+ assertEqual "BIGNUM round-trip" (fmap Only bigNumValues) rows+ , testCase "duckdbColumnType reports BIGNUM for Integer" $+ assertEqual+ "duckdbColumnType Integer"+ (Text.pack "BIGNUM")+ (duckdbColumnType (Proxy @Integer))+ , testCase "duckdbColumnType preserves Maybe parameter" $+ assertEqual+ "duckdbColumnType Maybe Word16"+ (Text.pack "USMALLINT")+ (duckdbColumnType (Proxy @(Maybe Word16)))+ , testCase "decodes interval components" $+ withConnection ":memory:" \conn -> do+ [Only intervalVal] <-+ query_ conn "SELECT INTERVAL '2 months 3 days 04:05:06.007008'"+ let IntervalValue{intervalMonths, intervalDays, intervalMicros} = intervalVal+ expectedMicros =+ (((4 * 60 + 5) * 60) + 6) * 1000000 + 7008+ assertEqual "interval months" 2 intervalMonths+ assertEqual "interval days" 3 intervalDays+ assertEqual "interval micros" expectedMicros intervalMicros+ , testCase "decodes decimals via Double conversion" $+ withConnection ":memory:" \conn -> do+ [Only decimalAsDouble] <-+ (query_ conn "SELECT CAST(12345.6789 AS DECIMAL(18,4))" :: IO [Only Double])+ let diff = abs (decimalAsDouble - 12345.6789)+ assertBool+ ("decimal as double: " <> show diff)+ (diff < 1e-4)+ , testCase "decodes time with time zone" $+ withConnection ":memory:" \conn -> do+ [Only tzVal] <- query_ conn "SELECT TIMETZ '14:30:15+02:30'"+ let expectedTime = TimeOfDay 14 30 15+ assertEqual "timetz time" expectedTime (timeWithZoneTime tzVal)+ assertEqual "timetz offset" 9000 (timeWithZoneOffsetSeconds tzVal)+ , testCase "decodes list columns into Haskell lists" $+ withConnection ":memory:" \conn -> do+ [Only (listOut :: [Int])] <-+ query_ conn "SELECT LIST_VALUE(1, 2, 3)::INTEGER[]"+ assertEqual "list decode" [1, 2, 3] listOut+ , testCase "decodes map columns into strict Map" $+ withConnection ":memory:" \conn -> do+ [Only (mapOut :: Map.Map Text.Text Int)] <-+ query_ conn "SELECT MAP(['a', 'b']::TEXT[], [1, 2]::INTEGER[])"+ let expected = Map.fromList [(Text.pack "a", 1), (Text.pack "b", 2)]+ assertEqual "map decode" expected mapOut+ , testCase "decodes timestamp with time zone as UTC" $+ withConnection ":memory:" \conn -> do+ [Only utcVal] <-+ (query_ conn "SELECT TIMESTAMPTZ '2024-10-12 14:30:15+02:30'" :: IO [Only UTCTime])+ let expected =+ UTCTime+ (fromGregorian 2024 10 12)+ (timeOfDayToTime (TimeOfDay 12 0 15))+ assertEqual "timestamptz utc" expected utcVal+ , testCase "custom FieldParser enforces invariants" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE nonempty (name TEXT)"+ _ <- execute conn "INSERT INTO nonempty VALUES (?)" (Only (Text.pack "okay"))+ rows <- query_ conn "SELECT name FROM nonempty" :: IO [Only NonEmptyText]+ assertEqual "non-empty success" [Only (NonEmptyText (Text.pack "okay"))] rows+ _ <- execute conn "INSERT INTO nonempty VALUES (?)" (Only (Text.pack ""))+ assertThrows+ (query_ conn "SELECT name FROM nonempty" :: IO [Only NonEmptyText])+ (Text.isInfixOf "NonEmptyText requires a non-empty string" . sqlErrorMessage)+ , testCase "round-trips uuid payloads" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE uuids (uuid UUID)"+ let uuidTexts =+ [ "123e4567-e89b-12d3-a456-426614174000"+ , "923e4567-e89b-12d3-a456-426614174000"+ , "ffffffff-ffff-ffff-ffff-ffffffffffff"+ ]+ maybeUUIDs = traverse UUID.fromText uuidTexts+ uuids <-+ case maybeUUIDs of+ Nothing -> assertFailure "invalid UUID literal in test data" >> pure []+ Just parsed -> pure parsed+ _ <- executeMany conn "INSERT INTO uuids VALUES (?)" (fmap Only uuids)+ rows <- query_ conn "SELECT uuid FROM uuids ORDER BY rowid" :: IO [Only UUID.UUID]+ assertEqual "uuid round-trip" (fmap Only uuids) rows+ , testCase "round-trips bitstring payloads" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE bits (bit BIT)"+ let bitValues =+ [ bsFromBool [True]+ , bsFromBool [False, False, False, True, True, True, False, True, False]+ , bsFromBool [False, False, False, False, True, True, True, False, True, False]+ , bsFromBool [True, False, False, False, True, True, True, False, True, False]+ ]+ _ <- executeMany conn "INSERT INTO bits VALUES (?)" (fmap Only bitValues)+ rows <- query_ conn "SELECT bit FROM bits ORDER BY rowid" :: IO [Only BitString]+ assertEqual "bit round-trip" (fmap Only bitValues) rows+ , testCase "round-trips struct and union parameters" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE struct_union (s STRUCT(a INT, b INT), u UNION(a INT, b VARCHAR))"+ [(structVal, unionVal)] <-+ query_ conn "SELECT {'a': 1, 'b': 2}, CAST(union_value(a := 42) AS UNION(a INTEGER, b VARCHAR))"+ :: IO [(StructValue FieldValue, UnionValue FieldValue)]+ _ <- execute conn "INSERT INTO struct_union VALUES (?, ?)" (structVal, unionVal)+ rows <- query_ conn "SELECT s, u FROM struct_union" :: IO [(StructValue FieldValue, UnionValue FieldValue)]+ assertEqual "struct/union round-trip" [(structVal, unionVal)] rows+ , testCase "generic struct into database" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE generic_struct (payload STRUCT(grId INT, grName TEXT))"+ let records = [GenericRecord 1 (Text.pack "alpha"), GenericRecord 2 (Text.pack "beta")]+ structParams = traverse genericToStructValue records+ case structParams of+ Nothing -> assertFailure "expected struct encoding"+ Just structs -> do+ _ <- executeMany conn "INSERT INTO generic_struct VALUES (?)" (map Only structs)+ rows <- query_ conn "SELECT payload FROM generic_struct ORDER BY payload.grId" :: IO [Only (StructValue FieldValue)]+ let unwrap (Only s) = s+ roundTripped <- runConversion (traverse (genericFromFieldValue . FieldStruct . unwrap) rows)+ roundTripped @?= Ok records+ , testCase "generic struct encode/decode" $ do+ let rec = GenericRecord 42 (Text.pack "duck")+ case genericToStructValue rec of+ Just structVal -> do+ map structFieldName (elems (structValueFields structVal)) @?= [Text.pack "grId", Text.pack "grName"]+ map structFieldValue (elems (structValueFields structVal)) @?= [FieldInt64 42, FieldText (Text.pack "duck")]+ runConversion (genericFromFieldValue (FieldStruct structVal)) >>= (@?= Ok rec)+ Nothing -> assertFailure "expected struct encoding"+ , testCase "generic union encode/decode" $ do+ let sumVal = SumBool True+ case genericToFieldValue sumVal of+ FieldUnion uv -> do+ map unionMemberName (elems (unionValueMembers uv))+ @?= [ Text.pack "SumInt"+ , Text.pack "SumBool"+ , Text.pack "SumPair"+ ]+ unionValueLabel uv @?= Text.pack "SumBool"+ runConversion (genericFromFieldValue (FieldUnion uv)) >>= (@?= Ok sumVal)+ other -> assertFailure ("expected FieldUnion, got " <> show other)+ , testCase "generic union complex constructor" $ do+ let sumVal = SumPair 7 (Text.pack "duck")+ case genericToFieldValue sumVal of+ FieldUnion uv -> do+ unionValueLabel uv @?= Text.pack "SumPair"+ case unionValuePayload uv of+ FieldStruct StructValue{structValueFields} -> do+ map structFieldName (elems structValueFields) @?= [Text.pack "field1", Text.pack "field2"]+ map structFieldValue (elems structValueFields)+ @?= [FieldInt64 7, FieldText (Text.pack "duck")]+ other -> assertFailure ("expected struct payload, got " <> show other)+ runConversion (genericFromFieldValue (FieldUnion uv)) >>= (@?= Ok sumVal)+ other -> assertFailure ("expected FieldUnion, got " <> show other)+ , testCase "deriving via generic struct round-trip" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE via_users (payload STRUCT(vuId BIGINT, vuName TEXT))"+ let users = [ViaUser 10 (Text.pack "alice"), ViaUser 11 (Text.pack "bob")]+ _ <- executeMany conn "INSERT INTO via_users VALUES (?)" (map Only users)+ rows <- query_ conn "SELECT payload FROM via_users ORDER BY payload.vuId" :: IO [Only ViaUser]+ rows @?= map Only users+ , testCase "deriving via generic union round-trip" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE via_shapes (shape UNION(ViaCircle STRUCT(field1 DOUBLE), ViaSquare STRUCT(field1 DOUBLE)))"+ let shapes = [ViaCircle 1.5, ViaSquare 3.0]+ _ <- executeMany conn "INSERT INTO via_shapes VALUES (?)" (map Only shapes)+ rows <- query_ conn "SELECT shape FROM via_shapes ORDER BY rowid" :: IO [Only ViaShape]+ rows @?= map Only shapes+ , testCase "generic enum payload is a STRUCT without fields" $ do+ case genericToFieldValue EnumRed of+ FieldUnion uv -> do+ unionValuePayload uv @?= FieldStruct (StructValue (listArray (0, -1) []) (listArray (0, -1) []) Map.empty)+ runConversion (genericFromFieldValue (FieldUnion uv)) >>= (@?= Ok EnumRed)+ other -> assertFailure ("expected FieldUnion, got " <> show other)+ , -- Edge case tests+ testCase "struct with multiple basic fields" $ do+ -- Test a struct with multiple fields+ let twoField = TwoFieldStruct 42 (Text.pack "text")+ case genericToFieldValue twoField of+ FieldStruct sv -> do+ runConversion (genericFromFieldValue (FieldStruct sv)) >>= (@?= Ok twoField)+ other -> assertFailure ("expected FieldStruct for TwoFieldStruct, got " <> show other)+ , testCase "large sum type with many constructors" $ do+ let constructors = [LargeC1 1, LargeC2 2, LargeC3 3, LargeC4 4, LargeC5 5]+ forM_ constructors $ \val -> do+ case genericToFieldValue val of+ FieldUnion uv -> do+ runConversion (genericFromFieldValue (FieldUnion uv)) >>= (@?= Ok val)+ other -> assertFailure ("expected FieldUnion, got " <> show other)+ , testCase "empty struct encodes as a STRUCT without fields" $ do+ case genericToFieldValue EmptyStruct of+ FieldStruct structVal -> do+ elems (structValueFields structVal) @?= []+ runConversion (genericFromFieldValue (FieldStruct structVal)) >>= (@?= Ok EmptyStruct)+ other -> assertFailure ("expected a STRUCT without fields for empty struct, got " <> show other)+ , testCase "empty struct binds as a STRUCT without fields" $+ withConnection ":memory:" \conn -> do+ query conn "SELECT typeof(?)" (Only EmptyStruct) >>= (@?= [Only (Text.pack "STRUCT")])+ rows <- query conn "SELECT ?" (Only EmptyStruct) :: IO [Only FieldValue]+ case rows of+ [Only (FieldStruct structVal)] -> elems (structValueFields structVal) @?= []+ other -> assertFailure ("expected a STRUCT without fields, got " <> show other)+ , testCase "Just and Nothing of a type without fields stay different in a record" $+ withConnection ":memory:" \conn -> do+ let value = OptionalEmpty (Just EmptyStruct) Nothing+ _ <- execute_ conn "CREATE TABLE optional_empty (r STRUCT(oeFirst STRUCT, oeSecond STRUCT))"+ _ <- execute conn "INSERT INTO optional_empty VALUES (?)" (Only value)+ (query_ conn "SELECT r::VARCHAR, r.oeFirst IS NULL, r.oeSecond IS NULL, r FROM optional_empty" :: IO [(Text.Text, Bool, Bool, OptionalEmpty)])+ >>= (@?= [(Text.pack "{'oeFirst': {}, 'oeSecond': NULL}", False, True, value)])+ , testCase "a constructor without fields binds the STRUCT without fields as its payload" $+ withConnection ":memory:" \conn -> do+ forM_ [(Counted 5, Text.pack "Counted"), (Uncounted, Text.pack "Uncounted")] \(value, tag) ->+ query conn "SELECT union_tag(?)::VARCHAR" (Only value) >>= (@?= [Only tag])+ _ <- execute_ conn "CREATE TABLE tallies (s UNION(Counted STRUCT(field1 BIGINT), Uncounted STRUCT))"+ _ <- executeMany conn "INSERT INTO tallies VALUES (?)" [Only (Counted 5), Only Uncounted]+ (query_ conn "SELECT union_tag(s)::VARCHAR, s::VARCHAR, s FROM tallies ORDER BY rowid" :: IO [(Text.Text, Text.Text, Tally)])+ >>= (@?= [(Text.pack "Counted", Text.pack "{'field1': 5}", Counted 5), (Text.pack "Uncounted", Text.pack "{}", Uncounted)])+ , testCase "the NULL of version 0.3 for a type without fields decodes but does not equal a new parameter" $+ withConnection ":memory:" \conn -> do+ -- Version 0.3 wrote NULL in place of the STRUCT without fields.+ runConversion (genericFromFieldValue FieldNull) >>= (@?= Ok EmptyStruct)+ case genericToFieldValue Uncounted of+ FieldUnion uv -> do+ let oldValue = uv{unionValuePayload = FieldNull}+ runConversion (genericFromFieldValue (FieldUnion oldValue)) >>= (@?= Ok Uncounted)+ _ <- execute_ conn "CREATE TABLE old_tallies (v UNION(Counted STRUCT(field1 BIGINT), Uncounted STRUCT))"+ _ <- execute conn "INSERT INTO old_tallies VALUES (?)" (Only oldValue)+ (query_ conn "SELECT v FROM old_tallies" :: IO [Only Tally]) >>= (@?= [Only Uncounted])+ (query conn "SELECT count(*) FROM old_tallies WHERE v = ?" (Only Uncounted) :: IO [Only Int64]) >>= (@?= [Only 0])+ _ <- execute conn "INSERT INTO old_tallies VALUES (?)" (Only Uncounted)+ (query conn "SELECT count(*) FROM old_tallies WHERE v = ?" (Only Uncounted) :: IO [Only Int64]) >>= (@?= [Only 1])+ other -> assertFailure ("expected FieldUnion, got " <> show other)+ case genericToFieldValue (WithEmpty 7 EmptyStruct) of+ FieldStruct sv -> do+ let oldField item = if structFieldName item == Text.pack "weEmpty" then item{structFieldValue = FieldNull} else item+ oldValue = sv{structValueFields = fmap oldField (structValueFields sv)}+ _ <- execute_ conn "CREATE TABLE old_with_empty (r STRUCT(weId BIGINT, weEmpty STRUCT))"+ _ <- execute conn "INSERT INTO old_with_empty VALUES (?)" (Only oldValue)+ (query_ conn "SELECT r FROM old_with_empty" :: IO [Only WithEmpty]) >>= (@?= [Only (WithEmpty 7 EmptyStruct)])+ (query conn "SELECT count(*) FROM old_with_empty WHERE r = ?" (Only (WithEmpty 7 EmptyStruct)) :: IO [Only Int64]) >>= (@?= [Only 0])+ _ <- execute conn "INSERT INTO old_with_empty VALUES (?)" (Only (WithEmpty 7 EmptyStruct))+ (query conn "SELECT count(*) FROM old_with_empty WHERE r = ?" (Only (WithEmpty 7 EmptyStruct)) :: IO [Only Int64]) >>= (@?= [Only 1])+ other -> assertFailure ("expected FieldStruct, got " <> show other)+ , testCase "a record field of a type without fields round trips" $+ withConnection ":memory:" \conn -> do+ let value = WithEmpty 7 EmptyStruct+ _ <- execute_ conn "CREATE TABLE with_empty (r STRUCT(weId BIGINT, weEmpty STRUCT))"+ _ <- execute conn "INSERT INTO with_empty VALUES (?)" (Only value)+ (query_ conn "SELECT r::VARCHAR, r FROM with_empty" :: IO [(Text.Text, WithEmpty)])+ >>= (@?= [(Text.pack "{'weId': 7, 'weEmpty': {}}", value)])+ , testCase "struct with Maybe fields" $ do+ let withJust = OptionalFields (Just 42) (Just (Text.pack "text"))+ withNothing = OptionalFields Nothing Nothing+ case (genericToFieldValue withJust, genericToFieldValue withNothing) of+ (FieldStruct sv1, FieldStruct sv2) -> do+ runConversion (genericFromFieldValue (FieldStruct sv1)) >>= (@?= Ok withJust)+ runConversion (genericFromFieldValue (FieldStruct sv2)) >>= (@?= Ok withNothing)+ (other1, other2) -> assertFailure ("expected FieldStruct, got " <> show (other1, other2))+ , -- Error case tests+ testCase "decode wrong field type fails" $ do+ let wrongType = FieldInt32 42 -- Trying to decode as GenericRecord+ runConversion (genericFromFieldValue wrongType :: Conversion GenericRecord) >>= \case+ Errors err -> assertBool "error mentions STRUCT" (Text.pack "STRUCT" `Text.isInfixOf` Text.pack (show err))+ Ok _ -> assertFailure "expected decode error for wrong type"+ , testCase "decode union as struct fails" $ do+ case genericToFieldValue (SumInt 42) of+ FieldUnion uv -> do+ runConversion (genericFromFieldValue (FieldUnion uv) :: Conversion GenericRecord) >>= \case+ Errors err -> assertBool "error mentions product type" (Text.pack "product type" `Text.isInfixOf` Text.pack (show err))+ Ok _ -> assertFailure "expected decode error"+ other -> assertFailure ("expected FieldUnion, got " <> show other)+ , testCase "decode struct as union fails" $ do+ case genericToFieldValue (GenericRecord 1 (Text.pack "test")) of+ FieldStruct sv -> do+ runConversion (genericFromFieldValue (FieldStruct sv) :: Conversion GenericSum) >>= \case+ Errors err -> assertBool "error mentions sum type" (Text.pack "sum type" `Text.isInfixOf` Text.pack (show err))+ Ok _ -> assertFailure "expected decode error"+ other -> assertFailure ("expected FieldStruct, got " <> show other)+ , testCase "decode struct with wrong field count fails" $+ withConnection ":memory:" \conn -> do+ -- Get a real struct with 3 fields from the database+ _ <- execute_ conn "CREATE TABLE three_field (s STRUCT(a INT, b TEXT, c INT))"+ [(Only threeField)] <- query_ conn "SELECT {'a': 1, 'b': 'test', 'c': 99}" :: IO [Only (StructValue FieldValue)]+ -- Try to decode it as GenericRecord which only has 2 fields+ runConversion (genericFromFieldValue (FieldStruct threeField) :: Conversion GenericRecord) >>= \case+ Errors err -> assertBool "error mentions extra fields" (Text.pack "extra" `Text.isInfixOf` Text.pack (show err))+ Ok _ -> assertFailure "expected decode error for extra fields"+ , testCase "decode union with wrong tag fails" $+ withConnection ":memory:" \conn -> do+ -- Get a real union from the database+ [(Only unionVal)] <-+ query_ conn "SELECT CAST(union_value(x := 42) AS UNION(x INT, y VARCHAR))" :: IO [Only (UnionValue FieldValue)]+ -- Manually create a modified version with an out-of-range tag+ let wrongTag = unionVal{unionValueIndex = 99, unionValueLabel = Text.pack "Invalid"}+ runConversion (genericFromFieldValue (FieldUnion wrongTag) :: Conversion GenericSum) >>= \case+ Errors err ->+ assertBool+ "error mentions tag or index"+ (Text.pack "tag" `Text.isInfixOf` Text.pack (show err) || Text.pack "99" `Text.isInfixOf` Text.pack (show err))+ Ok _ -> assertFailure "expected decode error for wrong tag"+ ]+ where+ -- 7 ^ 2900 has 1,018 magnitude bytes. A value of at most 9 magnitude bytes+ -- has inline storage.+ bigNumValues :: [Integer]+ bigNumValues =+ [ (2 ^ (200 :: Int)) + 8675309+ , 0+ , negate ((2 ^ (180 :: Int)) + 12345)+ , 7 ^ (2900 :: Int)+ , negate (7 ^ (2900 :: Int))+ , 1+ , -1+ , 255+ , 256+ , -256+ ]++streamingTests :: TestTree+streamingTests =+ testGroup+ "streaming"+ [ testCase "fold_ accumulates large result sets" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE stream_fold (n INTEGER)"+ let rows = fmap Only [1 .. 1000 :: Int]+ _ <- executeMany conn "INSERT INTO stream_fold VALUES (?)" rows+ total <- fold_ conn "SELECT n FROM stream_fold ORDER BY n" 0 \acc (Only n) -> pure (acc + n)+ assertEqual "folded sum" (sum ([1 .. 1000] :: [Int])) total+ , testCase "fold and foldNamed respect parameters" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE stream_filter (n INTEGER)"+ let values = fmap Only [1 .. 50 :: Int]+ _ <- executeMany conn "INSERT INTO stream_filter VALUES (?)" values+ gtTotal <-+ fold conn "SELECT n FROM stream_filter WHERE n > ?" (Only (40 :: Int)) 0 $+ \acc (Only n) -> pure (acc + n)+ let expected = sum ([41 .. 50] :: [Int])+ assertEqual "filtered fold sum" expected gtTotal+ leCount <-+ foldNamed conn "SELECT n FROM stream_filter WHERE n <= $limit" ["$limit" := (10 :: Int)] 0 $+ \acc (Only (_ :: Int)) -> pure (acc + 1)+ assertEqual "foldNamed count" (10 :: Int) leCount+ , testCase "nextRow streams rows sequentially" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE stream_cursor (n INTEGER)"+ _ <- executeMany conn "INSERT INTO stream_cursor VALUES (?)" (fmap Only [1 .. 3 :: Int])+ withStatement conn "SELECT n FROM stream_cursor ORDER BY n" \stmt -> do+ first <- nextRow stmt+ second <- nextRow stmt+ third <- nextRow stmt+ done <- nextRow stmt+ assertEqual+ "cursor iteration"+ [Just (Only (1 :: Int)), Just (Only 2), Just (Only 3), Nothing]+ [first, second, third, done]+ ]++functionsTests :: TestTree+functionsTests =+ testGroup+ "user-defined functions"+ [ testCase "registers pure scalar function" $+ withConnection ":memory:" \conn -> do+ createFunction conn "hs_times_two" (\(x :: Int64) -> x * 2)+ result <- query_ conn "SELECT hs_times_two(21)" :: IO [Only Int64]+ assertEqual "times_two result" [Only 42] result+ , testCase "handles nullable arguments and results" $+ withConnection ":memory:" \conn -> do+ createFunction conn "hs_optional" (\(mx :: Maybe Int64) -> fmap (+ 1) mx)+ rows <-+ query_ conn "SELECT hs_optional(x) FROM (VALUES (NULL), (41)) AS t(x)"+ :: IO [Only (Maybe Int64)]+ assertEqual "optional results" [Only Nothing, Only (Just 42)] rows+ , testCase "keeps the SQL types of the arguments of version 0.3" $+ withConnection ":memory:" \conn -> do+ createFunction conn "hs_id_bool" (id :: Bool -> Bool)+ createFunction conn "hs_id_double" (id :: Double -> Double)+ createFunction conn "hs_id_float" (id :: Float -> Float)+ createFunction conn "hs_id_int" (id :: Int -> Int)+ createFunction conn "hs_id_int16" (id :: Int16 -> Int16)+ createFunction conn "hs_id_int32" (id :: Int32 -> Int32)+ createFunction conn "hs_id_int64" (id :: Int64 -> Int64)+ createFunction conn "hs_id_maybe" (id :: Maybe Word16 -> Maybe Word16)+ createFunction conn "hs_id_string" (id :: String -> String)+ createFunction conn "hs_id_text" (id :: Text.Text -> Text.Text)+ createFunction conn "hs_id_word" (id :: Word -> Word)+ createFunction conn "hs_id_word16" (id :: Word16 -> Word16)+ createFunction conn "hs_id_word32" (id :: Word32 -> Word32)+ createFunction conn "hs_id_word64" (id :: Word64 -> Word64)+ query_+ conn+ "SELECT parameter_types::VARCHAR FROM duckdb_functions() WHERE function_name LIKE 'hs\\_id\\_%' ESCAPE '\\' ORDER BY function_name"+ >>= ( @?=+ map+ Only+ [ "[BOOLEAN]" :: Text.Text+ , "[DOUBLE]"+ , "[FLOAT]"+ , "[BIGINT]"+ , "[SMALLINT]"+ , "[INTEGER]"+ , "[BIGINT]"+ , "[USMALLINT]"+ , "[VARCHAR]"+ , "[VARCHAR]"+ , "[UBIGINT]"+ , "[USMALLINT]"+ , "[UINTEGER]"+ , "[UBIGINT]"+ ]+ )+ , testCase "takes an argument of each type with FromField and DuckDBColumnType" $+ withConnection ":memory:" \conn -> do+ let user = ViaUser 7 "anna"+ createFunction conn "hs_arg_int8" (\(x :: Int8) -> fromIntegral x * 2 :: Int64)+ createFunction conn "hs_arg_integer" (\(x :: Integer) -> Text.pack (show (x * 2)))+ createFunction conn "hs_arg_bytes" (\(x :: BS.ByteString) -> BS.length x)+ createFunction conn "hs_arg_uuid" UUID.toText+ createFunction conn "hs_arg_day" (\(x :: Day) -> Text.pack (show x))+ createFunction conn "hs_arg_local" (\(x :: LocalTime) -> Text.pack (show x))+ createFunction conn "hs_arg_utc" (\(x :: UTCTime) -> Text.pack (show x))+ createFunction conn "hs_arg_list" (\(xs :: [Int64]) -> sum xs)+ createFunction conn "hs_arg_map" (\(pairs :: Map.Map Text.Text Int64) -> sum (Map.elems pairs))+ createFunction conn "hs_arg_interval" (\(IntervalValue months days micros) -> Text.pack (show (months, days, micros)))+ createFunction conn "hs_arg_record" (\(ViaUser identifier name) -> name <> Text.pack (show identifier))+ createFunction conn "hs_arg_records" (\(users :: [ViaUser]) -> length users)+ createFunction conn "hs_arg_maybe_record" (\(found :: Maybe ViaUser) -> maybe "none" vuName found)+ createFunction conn "hs_arg_price" (\(Price DecimalValue{decimalInteger}) -> fromIntegral decimalInteger :: Int64)+ query_+ conn+ "SELECT hs_arg_int8(-3), hs_arg_integer(1267650600228229401496703205376::HUGEINT), hs_arg_bytes('\\x01\\x02'::BLOB), hs_arg_uuid('00000000-0000-0000-0000-000000000001')"+ >>= (@?= [(-6 :: Int64, Text.pack (show (2 * 2 ^ (100 :: Int) :: Integer)), 2 :: Int, "00000000-0000-0000-0000-000000000001" :: Text.Text)])+ query_+ conn+ "SELECT hs_arg_day(DATE '2026-11-15'), hs_arg_local(TIMESTAMP '2026-11-15 12:34:56'), hs_arg_utc(TIMESTAMPTZ '2026-11-15 12:34:56+00')"+ >>= (@?= [("2026-11-15" :: Text.Text, "2026-11-15 12:34:56" :: Text.Text, "2026-11-15 12:34:56 UTC" :: Text.Text)])+ query_ conn "SELECT hs_arg_list([1, 2, 3]), hs_arg_map(MAP {'a': 1, 'b': 2}), hs_arg_interval(INTERVAL '1 month 2 days 3 microseconds')"+ >>= (@?= [(6 :: Int64, 3 :: Int64, "(1,2,3)" :: Text.Text)])+ query conn "SELECT hs_arg_record(?), hs_arg_records([?, ?]), hs_arg_maybe_record(?), hs_arg_maybe_record(NULL)" (user, user, user, user)+ >>= (@?= [("anna7" :: Text.Text, 2 :: Int, "anna" :: Text.Text, "none" :: Text.Text)])+ query_ conn "SELECT hs_arg_price(1.25)" >>= (@?= [Only (1250 :: Int64)])+ query_+ conn+ "SELECT function_name, parameter_types::VARCHAR FROM duckdb_functions() WHERE function_name IN ('hs_arg_int8', 'hs_arg_integer', 'hs_arg_record', 'hs_arg_records', 'hs_arg_price') ORDER BY function_name"+ >>= ( @?=+ [ ("hs_arg_int8" :: Text.Text, "[TINYINT]" :: Text.Text)+ , ("hs_arg_integer", "[HUGEINT]")+ , ("hs_arg_price", "['DECIMAL(18,3)']")+ , ("hs_arg_record", "['STRUCT(vuId BIGINT, vuName VARCHAR)']")+ , ("hs_arg_records", "['STRUCT(vuId BIGINT, vuName VARCHAR)[]']")+ ]+ )+ , testCase "a UTCTime argument with a TIMESTAMP value gets the cast of DuckDB to TIMESTAMPTZ" $+ withConnection ":memory:" \conn -> do+ -- DuckDB casts the argument to the declared TIMESTAMPTZ with the+ -- TimeZone setting of the session. The function reads that instant.+ createFunction conn "hs_arg_utc_text" (\(x :: UTCTime) -> Text.pack (show x))+ _ <- execute_ conn "SET TimeZone = 'Pacific/Auckland'"+ query_ conn "SELECT hs_arg_utc_text(TIMESTAMP '2024-01-02 03:04:05'), (TIMESTAMP '2024-01-02 03:04:05')::TIMESTAMPTZ"+ >>= (@?= [("2024-01-01 14:04:05 UTC" :: Text.Text, UTCTime (fromGregorian 2024 1 1) (secondsToDiffTime 50645))])+ , testCase "declares the SQL type of each result type" $+ withConnection ":memory:" \conn -> do+ createFunction conn "hs_res_bool" True+ createFunction conn "hs_res_int" (1 :: Int)+ createFunction conn "hs_res_int8" (1 :: Int8)+ createFunction conn "hs_res_int16" (1 :: Int16)+ createFunction conn "hs_res_int32" (1 :: Int32)+ createFunction conn "hs_res_int64" (1 :: Int64)+ createFunction conn "hs_res_word" (1 :: Word)+ createFunction conn "hs_res_word8" (1 :: Word8)+ createFunction conn "hs_res_word16" (1 :: Word16)+ createFunction conn "hs_res_word32" (1 :: Word32)+ createFunction conn "hs_res_word64" (1 :: Word64)+ createFunction conn "hs_res_float" (1.5 :: Float)+ createFunction conn "hs_res_double" (1.5 :: Double)+ createFunction conn "hs_res_text" ("x" :: Text.Text)+ createFunction conn "hs_res_string" ("x" :: String)+ createFunction conn "hs_res_io" (pure (1 :: Int16) :: IO Int16)+ createFunction conn "hs_res_integer" (2 ^ (100 :: Int) :: Integer)+ createFunction conn "hs_res_bytes" (BS.pack [1, 2])+ createFunction conn "hs_res_uuid" UUID.nil+ createFunction conn "hs_res_day" (fromGregorian 2026 11 15)+ createFunction conn "hs_res_interval" (IntervalValue 1 2 3)+ createFunction conn "hs_res_amount" (Amount 1.25)+ forM_+ [ ("hs_res_bool", "BOOLEAN")+ , ("hs_res_int", "BIGINT")+ , ("hs_res_int8", "TINYINT")+ , ("hs_res_int16", "SMALLINT")+ , ("hs_res_int32", "INTEGER")+ , ("hs_res_int64", "BIGINT")+ , ("hs_res_word", "UBIGINT")+ , ("hs_res_word8", "UTINYINT")+ , ("hs_res_word16", "USMALLINT")+ , ("hs_res_word32", "UINTEGER")+ , ("hs_res_word64", "UBIGINT")+ , ("hs_res_float", "FLOAT")+ , ("hs_res_double", "DOUBLE")+ , ("hs_res_text", "VARCHAR")+ , ("hs_res_string", "VARCHAR")+ , ("hs_res_io", "SMALLINT")+ , ("hs_res_integer", "HUGEINT")+ , ("hs_res_bytes", "BLOB")+ , ("hs_res_uuid", "UUID")+ , ("hs_res_day", "DATE")+ , ("hs_res_interval", "INTERVAL")+ , ("hs_res_amount", "DECIMAL(18,3)")+ ]+ \(name, expected) -> query_ conn (Query ("SELECT typeof(" <> name <> "())")) >>= (@?= [Only (expected :: Text.Text)])+ -- An Integer argument or result is HUGEINT, and a Natural one is UHUGEINT.+ -- DuckDB casts no BIGNUM or UHUGEINT argument to HUGEINT, and+ -- toDuckValue raises an error for a result outside the range.+ createFunction conn "hs_arg_wide" (\(x :: Integer) -> x)+ assertThrows+ (query_ conn "SELECT hs_arg_wide('5'::BIGNUM)" :: IO [Only Integer])+ (Text.isInfixOf "No function matches" . sqlErrorMessage)+ assertThrows+ (query_ conn "SELECT hs_arg_wide(5::UHUGEINT)" :: IO [Only Integer])+ (Text.isInfixOf "No function matches" . sqlErrorMessage)+ createFunction conn "hs_res_natural" (2 ^ (127 :: Int) :: Natural)+ createFunction conn "hs_res_wide" (2 ^ (127 :: Int) :: Integer)+ query_ conn "SELECT typeof(hs_res_natural()), hs_res_natural()" >>= (@?= [("UHUGEINT" :: Text.Text, 2 ^ (127 :: Int) :: Natural)])+ assertThrows+ (query_ conn "SELECT hs_res_wide()" :: IO [Only Integer])+ (\err -> sqlErrorType err == Just ErrorTypeConversion && Text.isInfixOf "is outside the range of HUGEINT" (sqlErrorMessage err))+ -- DuckDB casts a native result value to the declared type as the SQL cast does.+ createFunction conn "hs_res_rounded" (Amount 1.2345)+ createFunction conn "hs_res_too_large" (Amount 1e20)+ query_ conn "SELECT hs_res_rounded()::VARCHAR, (1.2345::DOUBLE)::DECIMAL(18,3)::VARCHAR"+ >>= (@?= [("1.235" :: Text.Text, "1.235" :: Text.Text)])+ assertThrows+ (query_ conn "SELECT hs_res_too_large()" :: IO [Only Double])+ (Text.isInfixOf "to DECIMAL(18,3)" . sqlErrorMessage)+ query_ conn "SELECT hs_res_int8(), hs_res_word16(), hs_res_float(), hs_res_string(), hs_res_integer(), hs_res_amount()::DOUBLE"+ >>= (@?= [(1 :: Int8, 1 :: Word16, 1.5 :: Float, "x" :: Text.Text, 2 ^ (100 :: Int) :: Integer, 1.25 :: Double)])+ , testCase "writes nested, record, union, and NULL results" $+ withConnection ":memory:" \conn -> do+ createFunction conn "hs_res_user" (\(i :: Int64) -> ViaUser i ("u" <> Text.pack (show i)))+ createFunction conn "hs_res_maybe_user" (\(i :: Int64) -> if odd i then Nothing else Just (ViaUser i "even"))+ createFunction conn "hs_res_shape" (\(i :: Int64) -> if even i then ViaCircle (fromIntegral i) else ViaSquare 1)+ createFunction conn "hs_res_list" (\(i :: Int64) -> [0 .. i])+ createFunction conn "hs_res_map" (\(i :: Int64) -> Map.fromList [(Text.pack (show i), fromIntegral i :: Int16)])+ createFunction conn "hs_res_maybe_small" (\(i :: Int64) -> if odd i then Nothing else Just (fromIntegral i :: Int16))+ query_+ conn+ "SELECT typeof(hs_res_user(1)), typeof(hs_res_maybe_user(1)), typeof(hs_res_shape(1)), typeof(hs_res_list(1)), typeof(hs_res_map(1)), typeof(hs_res_maybe_small(1))"+ >>= ( @?=+ [+ ( "STRUCT(vuId BIGINT, vuName VARCHAR)" :: Text.Text+ , "STRUCT(vuId BIGINT, vuName VARCHAR)" :: Text.Text+ , "UNION(ViaCircle STRUCT(field1 DOUBLE), ViaSquare STRUCT(field1 DOUBLE))" :: Text.Text+ , "BIGINT[]" :: Text.Text+ , "MAP(VARCHAR, SMALLINT)" :: Text.Text+ , "SMALLINT" :: Text.Text+ )+ ]+ )+ rows <-+ query_+ conn+ "SELECT hs_res_user(i), hs_res_maybe_user(i), hs_res_shape(i), hs_res_list(i), hs_res_map(i), hs_res_maybe_small(i) FROM range(5) t(i) ORDER BY i"+ rows+ @?= [ ( ViaUser i ("u" <> Text.pack (show i))+ , if odd i then Nothing else Just (ViaUser i "even")+ , if even i then ViaCircle (fromIntegral i) else ViaSquare 1+ , [0 .. i]+ , Map.fromList [(Text.pack (show i), fromIntegral i :: Int16)]+ , if odd i then Nothing else Just (fromIntegral i :: Int16)+ )+ | i <- [0 .. 4 :: Int64]+ ]+ -- 5000 rows fill more than one chunk. Each odd row is NULL.+ query_ conn "SELECT count(*) FILTER (hs_res_maybe_user(i) IS NULL), count(*) FILTER (hs_res_maybe_small(i) IS NULL) FROM range(5000) t(i)"+ >>= (@?= [(2500 :: Int64, 2500 :: Int64)])+ , testCase "keeps the SQL type of a narrow result in arithmetic" $+ withConnection ":memory:" \conn -> do+ createFunction conn "hs_small" (id :: Int16 -> Int16)+ createFunction conn "hs_unsigned" (id :: Word16 -> Word16)+ -- DuckDB computes SMALLINT * SMALLINT as SMALLINT, and USMALLINT - USMALLINT as USMALLINT.+ assertThrows+ (query_ conn "SELECT hs_small(30000) * hs_small(30000)" :: IO [Only Int64])+ (Text.isInfixOf "Overflow in multiplication of INT16" . sqlErrorMessage)+ assertThrows+ (query_ conn "SELECT hs_unsigned(1) - hs_unsigned(2)" :: IO [Only Int64])+ (Text.isInfixOf "Overflow in subtraction of UINT16" . sqlErrorMessage)+ query_ conn "SELECT hs_small(30000)::INTEGER * hs_small(30000), hs_unsigned(1)::INTEGER - hs_unsigned(2)"+ >>= (@?= [(900000000 :: Int64, -1 :: Int64)])+ , testCase "rejects an argument or result type with no declaration" $+ withConnection ":memory:" \conn -> do+ let rejects expected action = assertThrows action (Text.isInfixOf expected . sqlErrorMessage)+ rejects "hs_bad_struct: argument arg0: StructValue has no static field types" $+ createFunction conn "hs_bad_struct" (\(_ :: StructValue FieldValue) -> True)+ rejects "hs_bad_result: result: StructValue has no static field types" $+ createFunction conn "hs_bad_result" (\(_ :: Int64) -> (undefined :: StructValue FieldValue))+ query_ conn "SELECT count(*) FROM duckdb_functions() WHERE function_name LIKE 'hs\\_bad\\_%' ESCAPE '\\'" >>= (@?= [Only (0 :: Int64)])+ , testCase "scalar callbacks read nested VARIANTs after the registering connection closes" $+ withEnvironment \environment -> withInstance environment [] \instance' -> do+ let payload = Variant (variantObject [("numbers", FieldList [FieldInt64 42, FieldNull])])+ values = Map.singleton "a" [Just payload, Nothing]+ withConnectionOn instance' \conn -> do+ createFunction conn "hs_nested_variant" ((== values) :: Map.Map Text.Text [Maybe Variant] -> Bool)+ createFunction conn "hs_variant_union" ((== Payload payload) :: VariantChoice -> Bool)+ createFunctionWithState conn "hs_stateful_variant" (pure payload) (\expected (value :: Maybe Variant) -> Just expected == value)+ withConnectionOn instance' \conn ->+ query+ conn+ "SELECT hs_nested_variant(?), hs_variant_union(?), hs_variant_union(?), hs_stateful_variant(?), hs_stateful_variant(NULL)"+ (values, Payload payload, NoPayload, payload)+ >>= (@?= [(True, True, False, True, False)])+ , testCase "two connections run the same VARIANT callback concurrently" $+ withEnvironment \environment -> withInstance environment [("threads", "1")] \instance' ->+ withConnectionOn instance' \first -> withConnectionOn instance' \second -> do+ enteredFirst <- newEmptyMVar+ enteredSecond <- newEmptyMVar+ let meet entered other = do+ putMVar entered ()+ ready <- timeout 30000000 (readMVar other)+ when (ready == Nothing) (fail "the other VARIANT callback did not enter")+ createFunction first "hs_variant_read" \case+ Variant (FieldInt64 number) -> do+ when (number == 0) (meet enteredFirst enteredSecond)+ when (number == 5000) (meet enteredSecond enteredFirst)+ pure number :: IO Int64+ other -> fail ("unexpected VARIANT: " <> show other)+ let run conn offset =+ query conn "SELECT sum(hs_variant_read((i + ?::BIGINT)::VARIANT))::BIGINT FROM range(5000) t(i)" (Only offset)+ >>= (@?= [Only (12497500 + 5000 * offset :: Int64)])+ completed <- newEmptyMVar+ bracket (forkFinally (run second 5000) (putMVar completed)) (const (void (readMVar completed))) \_ -> do+ run first 0+ readMVar completed >>= either throwIO pure+ , testCase "rejects an empty name or a name with NUL at registration" $+ withConnection ":memory:" \conn ->+ forM_ ["", "hs_bad\0name"] \name ->+ assertThrows+ (createFunction conn name (id :: Int64 -> Int64))+ (\err -> sqlErrorType err == Just ErrorInputInvalid && sqlErrorMessage err == "duckdb-simple: invalid scalar function name")+ , testCase "raises the name error before the state of a closed connection" $ do+ conn <- open ":memory:"+ close conn+ assertThrows+ (createFunction conn "" (id :: Int64 -> Int64))+ (\err -> sqlErrorType err == Just ErrorInputInvalid && sqlErrorMessage err == "duckdb-simple: invalid scalar function name")+ , testCase "a failing call with constant arguments fails as with column arguments" $+ withConnection ":memory:" \conn -> do+ createFunction conn "hs_fail_negative" (\(x :: Int64) -> (if x < 0 then throwIO (userError "negative") else pure x) :: IO Int64)+ let failsWith expected action = assertThrows action (\err -> sqlErrorType err == Just ErrorInputInvalid && expected `Text.isInfixOf` sqlErrorMessage err)+ failsWith "negative" (query_ conn "SELECT hs_fail_negative(-1)" :: IO [Only Int64])+ failsWith "negative" (query_ conn "SELECT hs_fail_negative(i - 1) FROM range(3) t(i)" :: IO [Only Int64])+ failsWith+ "duckdb-simple: argument arg0 is NULL, but Int64 cannot hold NULL"+ (query_ conn "SELECT hs_fail_negative(NULL::BIGINT)" :: IO [Only Int64])+ -- A failing constant in a CASE branch that is not taken does not raise.+ query_ conn "SELECT CASE WHEN i > 5 THEN hs_fail_negative(-1) ELSE i END FROM range(3) t(i)" >>= (@?= map Only [0, 1, 2 :: Int64])+ , testCase "names the argument that does not convert" $+ withConnection ":memory:" \conn -> do+ createFunction conn "hs_sum" (\(x :: Int64) (y :: Int64) -> x + y)+ assertThrows+ (query_ conn "SELECT hs_sum(1, NULL)" :: IO [Only Int64])+ (Text.isInfixOf "duckdb-simple: argument arg1 is NULL, but Int64 cannot hold NULL" . sqlErrorMessage)+ , testCase "supports IO-based functions" $+ withConnection ":memory:" \conn -> do+ ref <- newIORef (0 :: Int)+ createFunction conn "hs_counter" $ do+ atomicModifyIORef' ref \n ->+ let next = n + 1+ in (next, next)+ first <- query_ conn "SELECT hs_counter()" :: IO [Only Int]+ second <- query_ conn "SELECT hs_counter()" :: IO [Only Int]+ assertEqual "first counter call" [Only 1] first+ assertEqual "second counter call" [Only 2] second+ , testCase "supports thread-local function state" $ do+ conn <- openWithConfig ":memory:" [("threads", "1")]+ createFunctionWithState conn "hs_stateful_counter" (newIORef (0 :: Int)) \ref -> do+ atomicModifyIORef' ref \n ->+ let next = n + 1+ in (next, next)+ rows <- query_ conn "SELECT hs_stateful_counter() FROM range(3)" :: IO [Only Int]+ close conn+ assertEqual "stateful rows" [Only 1, Only 2, Only 3] rows+ , testCase "returns text rows of each length over more than one chunk" $+ withConnection ":memory:" \conn -> do+ let texts =+ [ Just "a"+ , Nothing+ , Just "b"+ , Just ""+ , Just "a\0b"+ , Just "123456789012"+ , Just "1234567890123"+ , Just (Text.replicate 4000 "abcdefghijklmnopqrstuvwxy")+ , Just "liður ð þ ✓ 😀"+ ]+ pick (i :: Int64) = texts !! fromIntegral (i `mod` 9) :: Maybe Text.Text+ createFunction conn "hs_pick" pick+ createFunction conn "hs_echo" (\(text :: Maybe Text.Text) -> text)+ -- 5000 rows fill more than one chunk. A NULL row is between two text rows.+ rows <- query_ conn "SELECT hs_pick(i) FROM range(5000) t(i) ORDER BY i"+ assertEqual "text rows" (map (Only . pick) [0 .. 4999]) rows+ -- DuckDB compares the bytes in the vector, also the inline bytes after a short text.+ query_+ conn+ "SELECT bool_and(hs_pick(i) IS NOT DISTINCT FROM CASE i % 9 WHEN 0 THEN 'a' WHEN 1 THEN NULL WHEN 2 THEN 'b' WHEN 3 THEN '' WHEN 4 THEN 'a' || chr(0) || 'b' WHEN 5 THEN '123456789012' WHEN 6 THEN '1234567890123' WHEN 7 THEN repeat('abcdefghijklmnopqrstuvwxy', 4000) ELSE 'liður ð þ ✓ 😀' END) FROM range(5000) t(i)"+ >>= (@?= [Only True])+ query_ conn "SELECT hs_echo('x'), hs_echo(NULL)" >>= (@?= [(Just "x" :: Maybe Text.Text, Nothing :: Maybe Text.Text)])+ ]++transactionTests :: TestTree+transactionTests =+ testGroup+ "transactions"+ [ testCase "withTransaction commits on success" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE tx_commit (x INTEGER)"+ withTransaction conn $ do+ _ <- execute conn "INSERT INTO tx_commit VALUES (?)" (Only (1 :: Int))+ pure ()+ rows <- query_ conn "SELECT COUNT(*) FROM tx_commit" :: IO [Only Int]+ assertEqual "commit" [Only 1] rows+ , testCase "withTransaction rolls back on exception" $+ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE tx_rollback (x INTEGER)"+ assertThrowsErrorCall $+ withTransaction conn $ do+ _ <- execute conn "INSERT INTO tx_rollback VALUES (?)" (Only (1 :: Int))+ error "boom"+ rows <- query_ conn "SELECT COUNT(*) FROM tx_rollback" :: IO [Only Int]+ assertEqual "rollback" [Only 0] rows+ ]++assertThrows :: forall e a. (Exception e) => IO a -> (e -> Bool) -> Assertion+assertThrows action predicate = do+ outcome <- try action+ case outcome of+ Left (err :: e) -> assertBool "unexpected exception" (predicate err)+ Right _ -> assertFailure "expected exception, but action succeeded"++assertThrowsSQLError :: IO a -> Assertion+assertThrowsSQLError action =+ assertThrows action (const True :: SQLError -> Bool)++assertThrowsErrorCall :: IO a -> Assertion+assertThrowsErrorCall action =+ assertThrows action (const True :: ErrorCall -> Bool)++assertThrowsFormatError :: IO a -> (FormatError -> Bool) -> Assertion+assertThrowsFormatError = assertThrows++-- | Give the action a unique temporary path where no file exists. Remove the+-- file after the action if the action created it. DuckDB writes a COPY TO+-- target that exists through a second file, so the path has no file.+withTemporaryPath :: (FilePath -> IO a) -> IO a+withTemporaryPath = bracket acquire release+ where+ acquire = do+ directory <- getTemporaryDirectory+ (path, handle) <- openBinaryTempFile directory "duckdb-simple-spec"+ hClose handle+ removeFile path+ pure path+ release path = doesFileExist path >>= \exists -> when exists (removeFile path)
+ test/SqlFeatureTests.hs view
@@ -0,0 +1,240 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE OverloadedStrings #-}++-- |+-- Module : SqlFeatureTests+-- Description : Integration checks for DuckDB 2.0 SQL features.+module SqlFeatureTests (tests) where++import Control.Exception (bracket, try)+import Control.Monad (forM_, void, when)+import Data.Int (Int64)+import Data.Text (Text)+import qualified Data.Text as Text+import Data.Time.LocalTime (TimeOfDay (..))+import Database.DuckDB.Simple+import Database.DuckDB.Simple.FromField (BitString (..), FieldValue (..), TimeWithZone (..))+import Database.DuckDB.Simple.Variant (Variant (..), variantObject)+import System.Directory (doesFileExist, getTemporaryDirectory, removeFile)+import System.IO (hClose, openBinaryTempFile)+import Test.Tasty (TestTree, testGroup)+import Test.Tasty.HUnit (Assertion, assertBool, assertFailure, testCase, (@?=))+import TestUtils (assertInputError)++-- | Exercise the public query, binding, cursor, and fold operations.+tests :: TestTree+tests =+ testGroup+ "DuckDB 2.0 SQL features"+ [ testCase "row triggers audit bound inserts and roll back failures" $+ withConnection ":memory:" \conn -> do+ void (execute_ conn "CREATE TABLE target(id BIGINT, label VARCHAR)")+ void (execute_ conn "CREATE TABLE audit(id BIGINT PRIMARY KEY, label VARCHAR)")+ void (execute_ conn "CREATE TRIGGER record_insert AFTER INSERT ON target FOR EACH ROW INSERT INTO audit VALUES (NEW.id, NEW.label)")+ void (executeMany conn "INSERT INTO target VALUES (?, ?)" [(1 :: Int64, "first" :: Text), (2, "O'Reilly")])+ query_ conn "SELECT * FROM audit ORDER BY id" >>= (@?= [(1 :: Int64, "first" :: Text), (2, "O'Reilly")])+ assertDuckDBError+ ErrorDatabaseConstraint+ "Constraint Error: Duplicate key \"id: 1\" violates primary key constraint."+ (execute conn "INSERT INTO target VALUES (?, ?)" (1 :: Int64, "duplicate" :: Text))+ query_ conn "SELECT count(*) FROM target" >>= (@?= [Only (2 :: Int64)])+ , testCase "DML CTEs bind values and deliver RETURNING rows to folds" $+ withConnection ":memory:" \conn -> do+ void (execute_ conn "CREATE TABLE moved(id BIGINT)")+ query conn "WITH ins AS (INSERT INTO moved VALUES (?), (?) RETURNING id) SELECT id FROM ins ORDER BY id" (1 :: Int64, 2 :: Int64)+ >>= (@?= [Only (1 :: Int64), Only 2])+ total <- fold_ conn "WITH del AS (DELETE FROM moved RETURNING id) SELECT id FROM del" (0 :: Int64) \acc (Only value) -> pure (acc + value)+ total @?= 3+ query_ conn "SELECT count(*) FROM moved" >>= (@?= [Only (0 :: Int64)])+ , testCase "nested schemas accept qualified parameterized DML" $+ withConnection ":memory:" \conn -> do+ void (execute_ conn "CREATE SCHEMA parent_schema")+ void (execute_ conn "CREATE SCHEMA parent_schema.child_schema")+ void (execute_ conn "CREATE TABLE parent_schema.child_schema.items(id BIGINT)")+ void (execute conn "INSERT INTO parent_schema.child_schema.items VALUES (?)" (Only (17 :: Int64)))+ query_ conn "SELECT id FROM memory.parent_schema.child_schema.items" >>= (@?= [Only (17 :: Int64)])+ query_ conn "SELECT parent_schema FROM duckdb_schemas() WHERE schema_name='child_schema'" >>= (@?= [Only ("parent_schema" :: Text)])+ , testCase "session variables supply cursor and fold defaults; named bindings override them" $+ withConnection ":memory:" \conn -> do+ void (execute_ conn "SET VARIABLE session_value = 41")+ query_ conn "SELECT $session_value + 1" >>= (@?= [Only (42 :: Int64)])+ withStatement conn "SELECT $session_value + 1" \stmt -> do+ nextRow stmt >>= (@?= Just (Only (42 :: Int64)))+ nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ queryNamed conn "SELECT $session_value + 1" ["session_value" := (9 :: Int64)] >>= (@?= [Only (10 :: Int64)])+ queryNamed conn "SELECT $session_value + $amount" ["amount" := (1 :: Int64)] >>= (@?= [Only (42 :: Int64)])+ queryNamed conn "SELECT $session_value" [] >>= (@?= [Only (41 :: Int64)])+ assertDuckDBError+ ErrorInputInvalid+ "Invalid Input Error: Values were not provided for the following parameters: missing"+ (queryNamed conn "SELECT $session_value + $missing" [] :: IO [Only Int64])+ void (execute_ conn "SET VARIABLE session_value = 83")+ total <- fold_ conn "SELECT $session_value + 1" (0 :: Int64) \acc (Only value) -> pure (acc + value)+ total @?= 84+ void (execute_ conn "RESET VARIABLE session_value")+ assertDuckDBError+ ErrorInputInvalid+ "Invalid Input Error: Values were not provided for the following parameters: session_value"+ (query_ conn "SELECT $session_value" :: IO [Only Int64])+ , testCase "named replacement preserves old values on failure and replaces session overrides on success" $+ withConnection ":memory:" \conn -> do+ void (execute_ conn "SET VARIABLE a = 80; SET VARIABLE b = 90; SET VARIABLE c = 91")+ withStatement conn "SELECT $a::BIGINT, $b::BIGINT, $c::BIGINT" \stmt -> do+ let rejected = bindNamed stmt ["a" := (4 :: Int64), "b" := BitString 8 "x", "c" := (6 :: Int64)]+ assertInputError rejected+ nextRow stmt >>= (@?= Just (80 :: Int64, 90 :: Int64, 91 :: Int64))+ bindNamed stmt ["a" := (1 :: Int64), "b" := (2 :: Int64), "c" := (3 :: Int64)]+ nextRow stmt >>= (@?= Just (1 :: Int64, 2 :: Int64, 3 :: Int64))+ assertInputError rejected+ nextRow stmt >>= (@?= Just (1 :: Int64, 2 :: Int64, 3 :: Int64))+ bindNamed stmt ["a" := (7 :: Int64)]+ nextRow stmt >>= (@?= Just (7 :: Int64, 90 :: Int64, 91 :: Int64))+ , testCase "FETCH FIRST and OFFSET work with bound counts" $+ withConnection ":memory:" \conn ->+ query conn "SELECT i FROM range(10) t(i) ORDER BY i OFFSET ? ROWS FETCH FIRST ? ROWS ONLY" (3 :: Int64, 2 :: Int64)+ >>= (@?= [Only (3 :: Int64), Only 4])+ , testCase "SELECT expressions use the query and cursor paths" $+ withConnection ":memory:" \conn -> do+ query conn "SELECT ?::BIGINT + 1" (Only (41 :: Int64)) >>= (@?= [Only (42 :: Int64)])+ withStatement conn "SELECT 42::BIGINT" \stmt -> do+ nextRow stmt >>= (@?= Just (Only (42 :: Int64)))+ nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ , testCase "NEAREST joins rank candidates for bound input rows" $+ withConnection ":memory:" \conn ->+ query+ conn+ "SELECT q.id::BIGINT, t.value::BIGINT FROM (VALUES (1, ?::BIGINT), (2, ?::BIGINT)) q(id, value) INNER JOIN (VALUES (11), (19), (25), (100)) t(value) EXACT NEAREST 2 BY DISTANCE abs(q.value - t.value) ORDER BY q.id, t.value"+ (10 :: Int64, 20 :: Int64)+ >>= (@?= [(1 :: Int64, 11 :: Int64), (1, 19), (2, 19), (2, 25)])+ , testCase "recursive USING KEY aggregates bound values before folding" $+ withConnection ":memory:" \conn -> do+ total <- fold+ conn+ "WITH RECURSIVE totals(k, total) USING KEY (k, sum(total)) AS (VALUES (1, ?::BIGINT) UNION SELECT k, 1::BIGINT FROM totals WHERE total < ?::BIGINT) SELECT total::BIGINT FROM totals"+ (1 :: Int64, 4 :: Int64)+ (0 :: Int64)+ \acc (Only value) -> pure (acc + value)+ total @?= 4+ , testCase "lambda filter, transform, and reduce accept bound lists" $+ withConnection ":memory:" \conn ->+ query+ conn+ "SELECT list_reduce(list_transform(list_filter(?::BIGINT[], lambda x: x % 2 = 0), lambda x: x * 3), lambda x, y: x + y)"+ (Only [1, 2, 3, 4 :: Int64])+ >>= (@?= [Only (18 :: Int64)])+ , testCase "JSON mutation accepts bound documents, paths, and values" $+ withConnection ":memory:" \conn -> do+ query conn "SELECT json_set(?::JSON, ?, ?::JSON)" ("{\"a\":1}" :: Text, "$.b" :: Text, "[2,3]" :: Text)+ >>= (@?= [Only ("{\"a\":1,\"b\":[2,3]}" :: Text)])+ query conn "SELECT json_set(?::JSON, ?, ?::JSON)::VARCHAR" ("{\"a\":1}" :: Text, "$.b" :: Text, "[2,3]" :: Text)+ >>= (@?= [Only ("{\"a\":1,\"b\":[2,3]}" :: Text)])+ query_ conn "SELECT json_remove(json_replace(json_insert('{\"a\":1}', '$.b', '2'), '$.a', '3'), '$.b')::VARCHAR"+ >>= (@?= [Only ("{\"a\":3}" :: Text)])+ , testCase "default 2.0 storage preserves bound VARIANT values after checkpoint and reopen" $+ withFiles \path -> do+ let value = Variant (variantObject [("n", FieldInt64 7), ("text", FieldText "saved")])+ withConnection path \conn -> do+ void (execute_ conn "CREATE TABLE stored(value VARIANT)")+ void (execute conn "INSERT INTO stored VALUES (?)" (Only value))+ void (execute_ conn "CHECKPOINT")+ query_ conn "SELECT tags['storage_version'] FROM duckdb_databases() WHERE NOT internal" >>= (@?= [Only ("v2.0.0+" :: Text)])+ withConnection path \conn -> do+ query_ conn "SELECT tags['storage_version'] FROM duckdb_databases() WHERE NOT internal" >>= (@?= [Only ("v2.0.0+" :: Text)])+ query_ conn "SELECT value FROM stored" >>= (@?= [Only value])+ , testCase "VARIANT values survive native storage and explicit Parquet shredding" $+ withFiles \path -> do+ let parquet = path <> ".parquet"+ expected = [(i, Text.pack (show i)) | i <- [0, 2048, 4096 :: Int64]]+ withConnectionWithConfig path [("storage_compatibility_version", "v1.5.0")] \conn -> do+ void (execute_ conn "SET variant_minimum_shredding_size = 0")+ void (execute_ conn "CREATE TABLE variants AS SELECT i, {'n': i, 'text': i::VARCHAR}::VARIANT AS v FROM range(5000) t(i)")+ void (execute_ conn "CHECKPOINT")+ withConnectionWithConfig path [("storage_compatibility_version", "v1.5.0")] \conn -> do+ query_ conn "SELECT v.n::BIGINT, v.text::VARCHAR FROM variants WHERE i % 2048 = 0 ORDER BY i" >>= (@?= expected)+ query_ conn "SELECT v FROM variants WHERE i % 2048 = 0 ORDER BY i"+ >>= (@?= [Only (Variant (variantObject [("n", FieldInt64 i), ("text", FieldText text)])) | (i, text) <- expected])+ void (execute conn "COPY variants TO ? (FORMAT PARQUET, SHREDDING {'v': 'STRUCT(n BIGINT, text VARCHAR)'})" (Only (Text.pack parquet)))+ query conn "SELECT v.n::BIGINT, v.text::VARCHAR FROM read_parquet(?) WHERE i % 2048 = 0 ORDER BY i" (Only (Text.pack parquet))+ >>= (@?= expected)+ query conn "SELECT count(*) > 0 FROM parquet_schema(?) WHERE name='typed_value'" (Only (Text.pack parquet)) >>= (@?= [Only True])+ total <- fold conn "SELECT v.n::BIGINT FROM read_parquet(?)" (Only (Text.pack parquet)) (0 :: Int64) \acc (Only value) -> pure (acc + value)+ total @?= 12497500+ , -- Each case of this group runs the SQL form of a limit that the+ -- documentation of duckdb-simple states. The case fails when DuckDB lifts+ -- the limit.+ testGroup+ "limits that DuckDB sets"+ [ testCase "an ARRAY type needs at least one element" $ withConnection ":memory:" \conn ->+ assertDuckDBError+ ErrorQueryBinder+ "Binder Error: ARRAY type size must be at least 1"+ (query_ conn "SELECT []::INTEGER[0]" :: IO [Only Text])+ , testCase "a STRUCT for a VARIANT cannot have two equal keys" $ withConnection ":memory:" \conn ->+ assertDuckDBError+ ErrorQueryBinder+ "Binder Error: Duplicate named argument \"a\""+ (query_ conn "SELECT {'a': 1, 'a': 2}::VARIANT" :: IO [Only Variant])+ , testCase "a VARIANT does not keep the CRS of a GEOMETRY" $ withConnection ":memory:" \conn ->+ query_ conn "SELECT typeof('POINT (1 2)'::GEOMETRY('OGC:CRS84')::VARIANT::GEOMETRY), typeof('POINT (1 2)'::GEOMETRY('OGC:CRS84'))"+ >>= (@?= [("GEOMETRY" :: Text, "GEOMETRY('OGC:CRS84')" :: Text)])+ , testCase "a BIT value needs at least one bit" $ withConnection ":memory:" \conn -> do+ assertDuckDBError+ ErrorTypeConversion+ "Conversion Error: Cannot cast empty string to BIT"+ (query_ conn "SELECT bitstring('', 0)" :: IO [Only Text])+ -- An empty string literal is one zero bit.+ query_ conn "SELECT ''::BIT::VARCHAR" >>= (@?= [Only ("0" :: Text)])+ , testCase "a value that is not NULL has no cast to the type NULL" $ withConnection ":memory:" \conn ->+ assertDuckDBError+ ErrorTypeConversion+ "Conversion Error: Unimplemented type for cast (INTEGER -> \"NULL\")"+ (query_ conn "SELECT 7::\"NULL\"" :: IO [Only Text])+ , testCase "a TIMETZ offset is at most 15:59:59 in each direction" $ withConnection ":memory:" \conn -> do+ forM_ ["12:00:00+16:00:00", "12:00:00-16:00:00"] \value ->+ assertDuckDBError+ ErrorTypeConversion+ ("Conversion Error: time field value out of range: \"" <> value <> "\"")+ (query_ conn (Query ("SELECT '" <> value <> "'::TIMETZ")) :: IO [Only Text])+ query_ conn "SELECT '12:00:00+15:59:59'::TIMETZ, '12:00:00-15:59:59'::TIMETZ"+ >>= (@?= [(TimeWithZone (TimeOfDay 12 0 0) 57599, TimeWithZone (TimeOfDay 12 0 0) (-57599))])+ , testCase "DuckDB has no savepoints" $ withConnection ":memory:" \conn ->+ assertDuckDBError ErrorQueryParser "Parser Error: syntax error at or near \"SAVEPOINT\"" (execute_ conn "SAVEPOINT a")+ , testCase "the CRS of a GEOMETRY cast is a constant" $ withConnection ":memory:" \conn ->+ assertDuckDBError+ ErrorQueryParser+ "Parser Error: Expected a constant as type modifier"+ (withStatement conn "SELECT ?::GEOMETRY(?)" columnCount)+ , testCase "VARIANT columns need the storage version v1.5.0" $ withFiles \path ->+ withConnectionWithConfig path [("storage_compatibility_version", "v1.0.0")] \conn ->+ assertDuckDBError+ ErrorInputInvalid+ "Invalid Input Error: VARIANT columns are not supported in storage versions prior to v1.5.0"+ (execute_ conn "CREATE TABLE old_variant (v VARIANT)")+ ]+ ]++-- | Require an SQL error whose message starts with the text of DuckDB, and+-- whose type is the given one.+assertDuckDBError :: ErrorType -> Text -> IO a -> Assertion+assertDuckDBError expectedType expected action = do+ outcome <- try action+ case outcome of+ Left err -> do+ assertBool (Text.unpack (sqlErrorMessage err)) (expected `Text.isPrefixOf` sqlErrorMessage err)+ sqlErrorType err @?= Just expectedType+ Right _ -> assertFailure ("expected the DuckDB error " <> Text.unpack expected)++-- | Reserve a database path. Remove the database, WAL, and Parquet file on exit.+withFiles :: (FilePath -> IO a) -> IO a+withFiles = bracket acquire release+ where+ acquire = do+ directory <- getTemporaryDirectory+ (path, handle) <- openBinaryTempFile directory "duckdb-sql-feature"+ hClose handle+ removeFile path+ pure path+ release path =+ forM_ [path, path <> ".wal", path <> ".parquet"] \file -> do+ exists <- doesFileExist file+ when exists (removeFile file)
test/StreamingTests.hs view
@@ -1,148 +1,280 @@ {-# LANGUAGE BlockArguments #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-}-{-# OPTIONS_GHC -Wno-deprecations #-} --- | Check the explicit native streaming API and its cursor ownership.+-- |+-- Module : StreamingTests+-- Description : Check the explicit native streaming API and its cursor ownership. module StreamingTests (nativeStreamingTests) where +import Control.Concurrent (forkIO, newEmptyMVar, putMVar, takeMVar) import Control.Exception (IOException, throwIO, try) import Control.Monad (forM_, replicateM_, void, when)+import qualified Data.ByteString as BS+import qualified Data.ByteString.Builder as Builder+import qualified Data.ByteString.Lazy as BL import Data.IORef (atomicWriteIORef, newIORef, readIORef, writeIORef) import Data.Int (Int64)-import Data.List (isInfixOf)+import Data.List (isInfixOf, sort, sortOn) import Data.Text (Text) import qualified Data.Text as Text-import Database.DuckDB.FFI (c_duckdb_vector_size)-import Database.DuckDB.FFI.Deprecated (c_duckdb_result_is_streaming)+import qualified Database.DuckDB.FFI as Retained (c_duckdb_vector_size) import Database.DuckDB.Simple-import qualified Database.DuckDB.Simple.Deprecated.Streaming as Streaming-import Database.DuckDB.Simple.FromField (FieldValue)-import Database.DuckDB.Simple.Internal (Statement (statementStream), StatementStream (statementStreamResult), StatementStreamState (..))+import Database.DuckDB.Simple.FromField (Field (fieldType), FieldValue (..))+import Database.DuckDB.Simple.Geometry (RawGeometry (..))+import Database.DuckDB.Simple.Internal (+ DuckDBV2ResultHandle (..),+ Statement (statementStream),+ StatementStream (statementStreamResult),+ StatementStreamState (..),+ nullHandle,+ )+import qualified Database.DuckDB.Simple.Streaming as Streaming+import Database.DuckDB.Simple.Variant (Variant (..), variantObject)+import System.Timeout (timeout) import Test.Tasty (TestTree, testGroup) import Test.Tasty.HUnit+import TestUtils (assertReusable, filteredQueries) -- | Exercise native streaming, resets, metadata, and cleanup across chunks. nativeStreamingTests :: TestTree nativeStreamingTests =- testGroup- "explicit native streaming"- [ testCase "nextRow starts a native streaming result" $ withDb \conn ->- withStatement conn "SELECT i FROM range(10000) t(i)" \stmt -> do- Streaming.nextRow stmt >>= (@?= Just (Only (0 :: Int64)))- state <- readIORef (statementStream stmt)- case state of- StatementStreamActive stream -> do- flag <- c_duckdb_result_is_streaming (statementStreamResult stream)- assertBool "expected a native streaming result" (flag /= 0)- _ -> assertFailure "expected an active cursor"- , testCase "default execution remains materialized when cursor entry points change" $ withDb \conn ->- withStatement conn "SELECT i FROM range(10000) t(i)" \stmt -> do- nextRow stmt >>= (@?= Just (Only (0 :: Int64)))- Streaming.nextRow stmt >>= (@?= Just (Only (1 :: Int64)))- state <- readIORef (statementStream stmt)- case state of- StatementStreamActive stream -> do- flag <- c_duckdb_result_is_streaming (statementStreamResult stream)- flag @?= 0- _ -> assertFailure "expected an active cursor"- , testCase "nested values survive chunk cleanup" $ withDb \conn ->- forM_- [ "SELECT CASE WHEN i % 7 = 0 THEN NULL ELSE {'n': i, 'text': 'λ' || i::VARCHAR, 'list': [i, NULL], 'union': union_value(value := i)} END FROM range(5000) t(i)"- , "SELECT [{'n': i, 'text': 'λ' || i::VARCHAR}, NULL] FROM range(5000) t(i)"- , "SELECT map([i], [{'values': [i, NULL], 'union': union_value(value := i)}]) FROM range(5000) t(i)"- , "SELECT CASE WHEN i % 2 = 0 THEN union_value(n := i)::UNION(n BIGINT, text VARCHAR) ELSE union_value(text := i::VARCHAR)::UNION(n BIGINT, text VARCHAR) END FROM range(5000) t(i)"- ]- \sql -> do- expected <- query_ conn sql :: IO [Only FieldValue]- actual <- reverse <$> Streaming.fold_ conn sql [] (\acc row -> pure (row : acc))- actual @?= expected- , testCase "fold and foldNamed use executed parameter types" $ withDb \conn -> do- positional <- Streaming.fold conn "SELECT coalesce(?, 'x') FROM range(3)" (Only (5000000000 :: Int64)) [] (\acc row -> pure (row : acc))- positional @?= replicate 3 (Only (5000000000 :: Int64))- named <- Streaming.foldNamed conn "SELECT $value FROM range(3)" ["value" := ("λ" :: Text)] [] (\acc row -> pure (row : acc))- named @?= replicate 3 (Only ("λ" :: Text))- , testCase "rebinding an active cursor replaces its metadata" $ withDb \conn ->- withStatement conn "SELECT ? FROM range(3)" \stmt -> do- bind stmt [toField (5000000000 :: Int64)]- Streaming.nextRow stmt >>= (@?= Just (Only (5000000000 :: Int64)))- bind stmt [toField ("λ" :: Text)]- Streaming.nextRow stmt >>= (@?= Just (Only ("λ" :: Text)))- , testCase "schema changes replace prepare-time metadata" $ withDb \conn -> do- void $ execute_ conn "CREATE TABLE stream_rebind(x VARCHAR)"- withStatement conn "SELECT x FROM stream_rebind" \stmt -> do- void $ execute_ conn "DROP TABLE stream_rebind"- void $ execute_ conn "CREATE TABLE stream_rebind(x BIGINT)"- void $ execute_ conn "INSERT INTO stream_rebind VALUES (5000000000)"- Streaming.nextRow stmt >>= (@?= Just (Only (5000000000 :: Int64)))- , testCase "EOF persists until an explicit binding reset" $ withDb \conn ->- withStatement conn "SELECT ?::BIGINT" \stmt -> do- bind stmt [toField (1 :: Int64)]- Streaming.nextRow stmt >>= (@?= Just (Only (1 :: Int64)))- replicateM_ 3 $ Streaming.nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))- bind stmt [toField (2 :: Int64)]- Streaming.nextRow stmt >>= (@?= Just (Only (2 :: Int64)))- Streaming.nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))- , testCase "clearing bindings discards an active result" $ withDb \conn ->- withStatement conn "SELECT ?::BIGINT FROM range(3)" \stmt -> do- bind stmt [toField (7 :: Int64)]- Streaming.nextRow stmt >>= (@?= Just (Only (7 :: Int64)))- clearStatementBindings stmt- void $ expectSqlError (Streaming.nextRow stmt :: IO (Maybe (Only Int64)))- bind stmt [toField (8 :: Int64)]- Streaming.nextRow stmt >>= (@?= Just (Only (8 :: Int64)))- , testCase "a DML cursor does not execute again after EOF" $ withDb \conn -> do- void $ execute_ conn "CREATE TABLE stream_once(x BIGINT)"- withStatement conn "INSERT INTO stream_once VALUES (1)" \stmt ->- replicateM_ 3 $ Streaming.nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))- query_ conn "SELECT count(*) FROM stream_once" >>= (@?= [Only (1 :: Int64)])- , testCase "nextRowWith runs the supplied parser" $ withDb \conn ->- withStatement conn "SELECT 20, 22" \stmt ->- Streaming.nextRowWith ((+) <$> field <*> field) stmt >>= (@?= Just (42 :: Int64))- , testCase "a fetch failure follows an already delivered batch" $ withDb \conn -> do- void $ execute_ conn "SET streaming_buffer_size = '64KB'"- failNext <- newIORef False- delivered <- newIORef (0 :: Int)- batchSize <- fromIntegral <$> c_duckdb_vector_size- createFunction conn "late_stream_error" \(value :: Int64) -> do- shouldFail <- readIORef failNext- when shouldFail (throwIO (userError "late stream error"))- pure value- err <- expectSqlError $ Streaming.fold_ conn "SELECT late_stream_error(i) FROM range(1000000) t(i)" (0 :: Int) \count (Only (_ :: Int64)) -> do- let next = count + 1- writeIORef delivered next- when (next == batchSize) (atomicWriteIORef failNext True)- pure next- count <- readIORef delivered- assertBool "the first batch must precede the error" (count >= batchSize)- assertBool "fetch error must retain its message" ("late stream error" `Text.isInfixOf` sqlErrorMessage err)- assertReusable conn- , testCase "Arrow reports a fetch failure after a delivered batch" $ withDb \conn -> do- void $ execute_ conn "SET streaming_buffer_size = '64KB'"- failNext <- newIORef False- createFunction conn "late_arrow_error" \(value :: Int64) -> do- shouldFail <- readIORef failNext- when shouldFail (throwIO (userError "late Arrow error"))- pure value- err <- expectSqlError $ Streaming.foldArrow_ conn "SELECT late_arrow_error(i) FROM range(1000000) t(i)" () \() _ _ ->- atomicWriteIORef failNext True- readIORef failNext >>= assertBool "a batch reached the callback"- assertBool "fetch error must retain its message" ("late Arrow error" `Text.isInfixOf` sqlErrorMessage err)- assertReusable conn- , testCase "row decode failure exhausts and releases the cursor" $ withDb \conn -> do- withStatement conn "SELECT {'n': i} FROM range(5000) t(i)" \stmt -> do- void $ expectSqlError (Streaming.nextRow stmt :: IO (Maybe (Only Int64)))- Streaming.nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))- assertReusable conn- , testCase "fold callback failure releases the active chunk" $ withDb \conn -> do- outcome <- try $ Streaming.fold_ conn "SELECT {'n': i, 'list': [i, NULL]} FROM range(5000) t(i)" (0 :: Int) \count (Only (_ :: FieldValue)) ->- if count == 2050 then throwIO (userError "stream step failure") else pure (count + 1)- case outcome of- Left (err :: IOException) -> assertBool "original callback error" ("stream step failure" `isInfixOf` show err)- Right _ -> assertFailure "expected the callback to fail"- assertReusable conn- ]+ testGroup+ "explicit native streaming"+ [ testCase "nextRow starts a native streaming result" $ withDb \conn ->+ withStatement conn "SELECT i FROM range(10000) t(i)" \stmt -> do+ Streaming.nextRow stmt >>= (@?= Just (Only (0 :: Int64)))+ state <- readIORef (statementStream stmt)+ case state of+ StatementStreamActive stream -> do+ assertBool "expected a live native cursor" (statementStreamResult stream /= nullHandle)+ _ -> assertFailure "expected an active cursor"+ , testCase "default execution remains materialized when cursor entry points change" $ withDb \conn ->+ withStatement conn "SELECT i FROM range(10000) t(i)" \stmt -> do+ nextRow stmt >>= (@?= Just (Only (0 :: Int64)))+ Streaming.nextRow stmt >>= (@?= Just (Only (1 :: Int64)))+ state <- readIORef (statementStream stmt)+ case state of+ StatementStreamActive stream -> do+ assertBool "expected a released execution cursor" (statementStreamResult stream == nullHandle)+ _ -> assertFailure "expected an active cursor"+ , testCase "nested values survive chunk cleanup" $ withDb \conn ->+ forM_+ [ "SELECT CASE WHEN i % 7 = 0 THEN NULL ELSE {'n': i, 'text': 'λ' || i::VARCHAR, 'list': [i, NULL], 'union': union_value(value := i)} END FROM range(5000) t(i)"+ , "SELECT [{'n': i, 'text': 'λ' || i::VARCHAR}, NULL] FROM range(5000) t(i)"+ , "SELECT map([i], [{'values': [i, NULL], 'union': union_value(value := i)}]) FROM range(5000) t(i)"+ , "SELECT CASE WHEN i % 2 = 0 THEN union_value(n := i)::UNION(n BIGINT, text VARCHAR) ELSE union_value(text := i::VARCHAR)::UNION(n BIGINT, text VARCHAR) END FROM range(5000) t(i)"+ ]+ \sql -> do+ expected <- query_ conn sql :: IO [Only FieldValue]+ actual <- reverse <$> Streaming.fold_ conn sql [] (\acc row -> pure (row : acc))+ actual @?= expected+ , testCase "fold and foldNamed use executed parameter types" $ withDb \conn -> do+ positional <- Streaming.fold conn "SELECT coalesce(?, 'x') FROM range(3)" (Only (5000000000 :: Int64)) [] (\acc row -> pure (row : acc))+ positional @?= replicate 3 (Only (5000000000 :: Int64))+ named <- Streaming.foldNamed conn "SELECT $value FROM range(3)" ["value" := ("λ" :: Text)] [] (\acc row -> pure (row : acc))+ named @?= replicate 3 (Only ("λ" :: Text))+ , testCase "rebinding an active cursor replaces its metadata" $ withDb \conn ->+ withStatement conn "SELECT ? FROM range(3)" \stmt -> do+ bind stmt [toField (5000000000 :: Int64)]+ Streaming.nextRow stmt >>= (@?= Just (Only (5000000000 :: Int64)))+ bind stmt [toField ("λ" :: Text)]+ Streaming.nextRow stmt >>= (@?= Just (Only ("λ" :: Text)))+ , testCase "schema changes replace prepare-time metadata" $ withDb \conn -> do+ void $ execute_ conn "CREATE TABLE stream_rebind(x VARCHAR)"+ withStatement conn "SELECT x FROM stream_rebind" \stmt -> do+ void $ execute_ conn "DROP TABLE stream_rebind"+ void $ execute_ conn "CREATE TABLE stream_rebind(x BIGINT)"+ void $ execute_ conn "INSERT INTO stream_rebind VALUES (5000000000)"+ Streaming.nextRow stmt >>= (@?= Just (Only (5000000000 :: Int64)))+ , testCase "EOF persists until an explicit binding reset" $ withDb \conn ->+ withStatement conn "SELECT ?::BIGINT" \stmt -> do+ bind stmt [toField (1 :: Int64)]+ Streaming.nextRow stmt >>= (@?= Just (Only (1 :: Int64)))+ replicateM_ 3 $ Streaming.nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ bind stmt [toField (2 :: Int64)]+ Streaming.nextRow stmt >>= (@?= Just (Only (2 :: Int64)))+ Streaming.nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ , testCase "clearing bindings discards an active result" $ withDb \conn ->+ withStatement conn "SELECT ?::BIGINT FROM range(3)" \stmt -> do+ bind stmt [toField (7 :: Int64)]+ Streaming.nextRow stmt >>= (@?= Just (Only (7 :: Int64)))+ clearStatementBindings stmt+ void $ expectSqlError (Streaming.nextRow stmt :: IO (Maybe (Only Int64)))+ bind stmt [toField (8 :: Int64)]+ Streaming.nextRow stmt >>= (@?= Just (Only (8 :: Int64)))+ , testCase "a DML cursor does not execute again after EOF" $ withDb \conn -> do+ void $ execute_ conn "CREATE TABLE stream_once(x BIGINT)"+ withStatement conn "INSERT INTO stream_once VALUES (1)" \stmt -> do+ Streaming.nextRow stmt >>= (@?= Just (Only (1 :: Int64)))+ replicateM_ 3 $ Streaming.nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ query_ conn "SELECT count(*) FROM stream_once" >>= (@?= [Only (1 :: Int64)])+ Streaming.fold_ conn "INSERT INTO stream_once VALUES (2)" [] (\acc row -> pure (row : acc)) >>= (@?= [Only (1 :: Int64)])+ query_ conn "SELECT count(*) FROM stream_once" >>= (@?= [Only (2 :: Int64)])+ , testCase "nextRowWith runs the supplied parser" $ withDb \conn ->+ withStatement conn "SELECT 20, 22" \stmt ->+ Streaming.nextRowWith ((+) <$> field <*> field) stmt >>= (@?= Just (42 :: Int64))+ , testGroup "decoder cursor ownership" $+ map (uncurry decoderCursorTests) [("materialized", nextRowWith), ("streaming", Streaming.nextRowWith)]+ , testCase "eager decoder cannot read uncached native metadata after connection close" $+ withEnvironment \env -> withInstance env [] \inst -> withConnectionOn inst \conn -> do+ let parser = fieldWith \column -> liftConversion (close conn) >> fieldType column+ err <- expectSqlError (queryWith_ parser conn "SELECT 42")+ assertBool "expected closed connection" ("connection is closed" `Text.isInfixOf` sqlErrorMessage err)+ , testCase "a fetch failure follows an already delivered batch" $ withDb \conn -> do+ void $ execute_ conn "SET streaming_buffer_size = '64KB'"+ failNext <- newIORef False+ delivered <- newIORef (0 :: Int)+ batchSize <- fromIntegral <$> Retained.c_duckdb_vector_size+ createFunction conn "late_stream_error" \(value :: Int64) -> do+ shouldFail <- readIORef failNext+ when shouldFail (throwIO (userError "late stream error"))+ pure value+ err <- expectSqlError $ Streaming.fold_ conn "SELECT late_stream_error(i) FROM range(1000000) t(i)" (0 :: Int) \count (Only (_ :: Int64)) -> do+ let next = count + 1+ writeIORef delivered next+ when (next == batchSize) (atomicWriteIORef failNext True)+ pure next+ count <- readIORef delivered+ assertBool "the first batch must precede the error" (count >= batchSize)+ assertBool "fetch error must retain its message" ("late stream error" `Text.isInfixOf` sqlErrorMessage err)+ assertReusable conn+ , testCase "Arrow reports a fetch failure after a delivered batch" $ withDb \conn -> do+ void $ execute_ conn "SET streaming_buffer_size = '64KB'"+ failNext <- newIORef False+ createFunction conn "late_arrow_error" \(value :: Int64) -> do+ shouldFail <- readIORef failNext+ when shouldFail (throwIO (userError "late Arrow error"))+ pure value+ err <- expectSqlError $ Streaming.foldArrow_ conn "SELECT late_arrow_error(i) FROM range(1000000) t(i)" () \() _ _ ->+ atomicWriteIORef failNext True+ readIORef failNext >>= assertBool "a batch reached the callback"+ assertBool "fetch error must retain its message" ("late Arrow error" `Text.isInfixOf` sqlErrorMessage err)+ assertReusable conn+ , testCase "row decode failure exhausts and releases the cursor" $ withDb \conn -> do+ withStatement conn "SELECT {'n': i} FROM range(5000) t(i)" \stmt -> do+ void $ expectSqlError (Streaming.nextRow stmt :: IO (Maybe (Only Int64)))+ cursor <- readIORef (statementStream stmt)+ case cursor of+ StatementStreamExhausted -> pure ()+ _ -> assertFailure "expected an exhausted cursor after the decode failure"+ Streaming.nextRow stmt >>= (@?= (Nothing :: Maybe (Only Int64)))+ assertReusable conn+ -- DuckDB refuses each other query of the connection while a native stream+ -- runs. A query that works while the statement is open shows that the+ -- failure released the native result. The large result keeps the stream+ -- running after the first chunk.+ withStatement conn "SELECT {'n': i} FROM range(1000000) t(i)" \stmt -> do+ void $ expectSqlError (Streaming.nextRow stmt :: IO (Maybe (Only Int64)))+ assertReusable conn+ , -- The case of the leak test for a statement that closes with a live cursor+ -- checks the release of the chunks.+ testCase "a failing step of a streaming fold raises its error and leaves the connection usable" $ withDb \conn -> do+ outcome <- try $ Streaming.fold_ conn "SELECT {'n': i, 'list': [i, NULL]} FROM range(5000) t(i)" (0 :: Int) \count (Only (_ :: FieldValue)) ->+ if count == 2050 then throwIO (userError "stream step failure") else pure (count + 1)+ case outcome of+ Left (err :: IOException) -> assertBool "original callback error" ("stream step failure" `isInfixOf` show err)+ Right _ -> assertFailure "expected the callback to fail"+ assertReusable conn+ , testCase "a query of another thread during an explicit stream fails and the stream continues" $ withDb \conn -> do+ -- The case needs a statement that still runs after the first row, so+ -- the stream reads a large result.+ finished <- timeout 120000000 $ withStatement conn "SELECT i FROM range(1000000) t(i)" \stmt -> do+ Streaming.nextRow stmt >>= (@?= Just (Only (0 :: Int64)))+ other <- newEmptyMVar+ _ <- forkIO (try (query_ conn "SELECT 42" :: IO [Only Int64]) >>= putMVar other)+ outcome <- takeMVar other+ case outcome of+ Left (err :: SQLError) -> sqlErrorType err @?= Just ErrorResourceInUse+ Right _ -> assertFailure "expected a busy connection during the stream"+ let readFrom expected = do+ row <- Streaming.nextRow stmt+ case row of+ Nothing -> pure expected+ Just (Only value) -> do+ when (value /= expected) $ assertFailure ("expected row " <> show expected <> ", got " <> show value)+ readFrom (expected + 1)+ readFrom (1 :: Int64) >>= (@?= 1000000)+ finished @?= Just ()+ assertReusable conn+ , testCase "filtered results give each row through each read function" $+ forM_ ["1", "4"] \threads -> withConnectionWithConfig ":memory:" [("threads", threads)] \conn ->+ forM_ filteredQueries \(sql, expected) ->+ forM_ readFunctions \(name, readRows) -> do+ rows <- map fromOnly <$> readRows conn sql+ -- With more than one thread, DuckDB can give the rows in another order.+ assertEqual (name <> " with " <> Text.unpack threads <> " threads: " <> show sql) expected (if threads == "1" then rows else sort rows)+ , testCase "filtered and joined VARIANT and GEOMETRY columns give the value of their row" $ withDb \conn -> do+ -- A filter or the probe side of a hash join can give a vector with a+ -- selection. Each row holds its own key, so a value of another row fails.+ void $+ execute_ conn "CREATE TABLE shaped AS SELECT i, {'n': i}::VARIANT AS v, ('POINT (' || i || ' 0)')::GEOMETRY AS g FROM range(5000) t(i)"+ forM_+ [ ("SELECT i, v, g FROM shaped WHERE i % 3 = 1", [1, 4 .. 4999])+ , ("SELECT a.i, a.v, a.g FROM shaped a JOIN (SELECT j FROM range(0, 5000, 7) t(j)) b ON a.i = b.j", [0, 7 .. 4999])+ ]+ \(sql, keys) ->+ forM_ readFunctions \(name, readRows) -> do+ rows <- readRows conn sql+ assertEqual+ (name <> ": " <> show sql)+ [(key, Variant (variantObject [("n", FieldInt64 key)]), pointWkb key) | key <- keys]+ (sortOn (\(key, _, _) -> key) [(key, value, rawGeometryWKB geometry) | (key, value, geometry) <- rows])+ ] +-- | Check operations on the same statement from inside its decoder.+decoderCursorTests :: String -> (RowParser Int64 -> Statement -> IO (Maybe Int64)) -> TestTree+decoderCursorTests name next =+ testGroup+ name+ [ testCase "a nested fetch consumes the next row" $ withDb \conn ->+ withStatement conn "SELECT i FROM range(1,4) t(i)" \stmt -> do+ let parser = fieldWith \column -> do+ liftConversion (next field stmt >>= (@?= Just 2))+ fromField column+ next parser stmt >>= (@?= Just 1)+ next field stmt >>= (@?= Just 3)+ , testCase "reset on the last row leaves the statement idle" $ withDb \conn ->+ withStatement conn "SELECT 42" \stmt -> do+ let parser = fieldWith \column -> do+ liftConversion (clearStatementBindings stmt)+ fromField column+ next parser stmt >>= (@?= Just 42)+ next field stmt >>= (@?= Just 42)+ , testCase "decoder failure preserves a replacement cursor" $ withDb \conn ->+ withStatement conn "SELECT ?::BIGINT + i FROM range(2) t(i)" \stmt -> do+ bind stmt [toField (10 :: Int64)]+ let parser = fieldWith \_ -> liftConversion do+ bind stmt [toField (20 :: Int64)]+ next field stmt >>= (@?= Just 20)+ throwIO (userError "decoder failed after rebind")+ outcome <- try (next parser stmt)+ case outcome of+ Left (err :: IOException) -> assertBool "original decoder error" ("decoder failed after rebind" `isInfixOf` show err)+ Right _ -> assertFailure "expected the decoder to fail"+ next field stmt >>= (@?= Just 21)+ ]++-- | Read the rows of a query with each read function of the library.+readFunctions :: (FromRow row) => [(String, Connection -> Query -> IO [row])]+readFunctions =+ [ ("query_", query_)+ , ("fold_", \conn sql -> reverse <$> fold_ conn sql [] (\rows row -> pure (row : rows)))+ , ("nextRow", \conn sql -> withStatement conn sql (readEach nextRow))+ , ("Streaming.fold_", \conn sql -> reverse <$> Streaming.fold_ conn sql [] (\rows row -> pure (row : rows)))+ , ("Streaming.nextRow", \conn sql -> withStatement conn sql (readEach Streaming.nextRow))+ ]+ where+ readEach :: (Statement -> IO (Maybe row)) -> Statement -> IO [row]+ readEach next stmt = go []+ where+ go rows = next stmt >>= maybe (pure (reverse rows)) (\row -> go (row : rows))++-- | The WKB of the point (x, 0) in the little-endian byte order. DuckDB 2.0.0+-- gives the same bytes for 'POINT (x 0)'::GEOMETRY::BLOB.+pointWkb :: Int64 -> BS.ByteString+pointWkb x = BL.toStrict (Builder.toLazyByteString (Builder.word8 1 <> Builder.word32LE 1 <> Builder.doubleLE (fromIntegral x) <> Builder.doubleLE 0))+ -- | Run each case with one native worker so row order is stable. withDb :: (Connection -> IO a) -> IO a withDb = withConnectionWithConfig ":memory:" [("threads", "1")]@@ -150,13 +282,7 @@ -- | Require a SQL or field conversion error and retain its diagnostic fields. expectSqlError :: IO a -> IO SQLError expectSqlError action = do- outcome <- try action- case outcome of- Left err -> pure err- Right _ -> assertFailure "expected SQLError" >> fail "expected SQLError"---- | Verify that cleanup leaves the connection ready for another query.-assertReusable :: Connection -> Assertion-assertReusable conn = do- rows <- query_ conn "SELECT 42" :: IO [Only Int64]- rows @?= [Only 42]+ outcome <- try action+ case outcome of+ Left err -> pure err+ Right _ -> assertFailure "expected SQLError" >> fail "expected SQLError"
test/TestUtils.hs view
@@ -1,13 +1,245 @@--- | Assertions that several test modules share.-module TestUtils (assertFailureIO) where+{-# LANGUAGE OverloadedStrings #-} +-- |+-- Module : TestUtils+-- Description : Assertions that several test modules share.+module TestUtils (+ NativeTimestamp (..),+ assertFailureIO,+ assertInputError,+ assertInputErrorContaining,+ assertRangeError,+ assertReleasedAfter,+ assertReusable,+ assertSqlErrorContaining,+ assertSqlErrorType,+ filteredQueries,+ nativeLocalTime,+ raiseCapabilities,+ sized,+ withConstCString,+ writeIntegers,+ writeNullRows,+) where++import Control.Concurrent (MVar, getNumCapabilities, newMVar, setNumCapabilities, withMVar) import Control.Exception (SomeException, try)-import Test.Tasty.HUnit (Assertion, assertFailure)+import Control.Monad (forM_, void, when)+import Data.IORef (IORef, mkWeakIORef, newIORef)+import Data.Int (Int64)+import Data.Maybe (fromMaybe, isNothing)+import Data.Ratio ((%))+import qualified Data.Text as Text+import Data.Time.Clock.POSIX (posixSecondsToUTCTime)+import Data.Time.LocalTime (LocalTime, utc, utcToLocalTime)+import Database.DuckDB.FFI.V2 (+ DuckDBV2CastFunctionExecInfoHandle,+ DuckDBV2VectorHandle,+ c_duckdb_v2_cast_function_exec_get_output,+ c_duckdb_v2_cast_function_exec_get_row_count,+ c_duckdb_v2_value_create_timestamp,+ c_duckdb_v2_value_create_timestamp_ms,+ c_duckdb_v2_value_create_timestamp_ns,+ c_duckdb_v2_value_create_timestamp_sec,+ c_duckdb_v2_vector_flatten,+ c_duckdb_v2_vector_get_data_mutable,+ c_duckdb_v2_vector_set_null,+ c_duckdb_v2_vector_set_size,+ )+import Database.DuckDB.Simple (Connection, ErrorType (..), Only (..), Query, SQLError (..), SizedArray, query_, sizedArrayFromList)+import Database.DuckDB.Simple.ToField (DuckDBColumnType (..), ToDuckValue (..), ToField)+import Database.DuckDB.Simple.V2.Registration (checked, output)+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.C.String (withCString)+import Foreign.C.Types (CChar)+import Foreign.Ptr (castPtr)+import Foreign.Storable (pokeElemOff)+import GHC.TypeLits (KnownNat)+import System.IO.Unsafe (unsafePerformIO)+import System.Mem (performMajorGC)+import System.Mem.Weak (deRefWeak)+import Test.Tasty.HUnit (Assertion, assertBool, assertFailure, (@?=)) +{- Note [Capabilities of the tests]+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+Some cases need more than one capability of the GHC runtime, for example to+run several Haskell threads on one connection at the same time. Such a case+raises the capability count with raiseCapabilities. No case lowers the count,+also not to restore it after the case.++The reason is a defect of the GHC runtime, seen with GHC 9.14.1. After+setNumCapabilities lowers the count, the runtime keeps the other capabilities+as disabled capabilities. A callback from a foreign thread, for example from+a DuckDB worker thread, can then run on a disabled capability. A garbage+collection can then lose the wake-up of that callback. The callback then+waits for ever, and the test run does not end.++The capability count is shared by all cases in the process. capabilityLock+serializes the read and change in raiseCapabilities. Without this lock, two+cases can read the same count, then set a larger count followed by a smaller+count.+-}++-- | Serialize capability changes in this process. See+-- Note [Capabilities of the tests].+capabilityLock :: MVar ()+capabilityLock = unsafePerformIO (newMVar ())+{-# NOINLINE capabilityLock #-}++-- | Raise the capability count of the GHC runtime to the given count. Do not+-- lower it. See Note [Capabilities of the tests].+raiseCapabilities :: Int -> IO ()+raiseCapabilities wanted = withMVar capabilityLock $ \_ -> do+ current <- getNumCapabilities+ when (wanted > current) (setNumCapabilities wanted)+ -- | Require an exception without depending on native error text. assertFailureIO :: IO a -> Assertion assertFailureIO action = do- result <- try (action >> pure ()) :: IO (Either SomeException ())- case result of- Left _ -> pure ()- Right () -> assertFailure "expected an exception"+ result <- try (action >> pure ()) :: IO (Either SomeException ())+ case result of+ Left _ -> pure ()+ Right () -> assertFailure "expected an exception"++-- | A timestamp in the raw unit of one of the four TIMESTAMP types. A+-- parameter of this type stores the raw value, also one that no Haskell time+-- type can bind, such as a value of seconds outside the range of microseconds.+data NativeTimestamp = Seconds Int64 | Milliseconds Int64 | Microseconds Int64 | Nanoseconds Int64+ deriving (Show)++instance DuckDBColumnType NativeTimestamp where+ duckdbColumnTypeFor _ = "TIMESTAMP"++instance ToDuckValue NativeTimestamp where+ toDuckValue factory (Seconds value) = output (c_duckdb_v2_value_create_timestamp_sec factory value)+ toDuckValue factory (Milliseconds value) = output (c_duckdb_v2_value_create_timestamp_ms factory value)+ toDuckValue factory (Microseconds value) = output (c_duckdb_v2_value_create_timestamp factory value)+ toDuckValue factory (Nanoseconds value) = output (c_duckdb_v2_value_create_timestamp_ns factory value)++instance ToField NativeTimestamp++-- | The local time in UTC of a raw timestamp. The computation uses neither+-- DuckDB nor duckdb-simple.+nativeLocalTime :: NativeTimestamp -> LocalTime+nativeLocalTime timestamp = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (toInteger value % units)))+ where+ (value, units) = case timestamp of+ Seconds raw -> (raw, 1)+ Milliseconds raw -> (raw, 1000)+ Microseconds raw -> (raw, 1000000)+ Nanoseconds raw -> (raw, 1000000000)++-- | Require the error of an input check of the library: an 'SQLError' of the+-- type 'ErrorInputInvalid' whose message starts with "duckdb-simple: ".+assertInputError :: IO a -> Assertion+assertInputError = assertInputErrorContaining Text.empty++-- | Require the error of an input check of the library, of the type+-- ErrorInputInvalid, which includes the expected message.+assertInputErrorContaining :: Text.Text -> IO a -> Assertion+assertInputErrorContaining expected action = do+ outcome <- try (void action)+ case outcome of+ Left err -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ -- The prefix shows that a check of the library raised the error, not DuckDB.+ assertBool ("expected an input error containing " <> show expected <> ", got " <> show err) $+ ("duckdb-simple: " `Text.isPrefixOf` sqlErrorMessage err) && expected `Text.isInfixOf` sqlErrorMessage err+ Right () -> assertFailure "expected the error of an input check"++-- | Require the error of a check of the library for a value that its SQL type+-- cannot hold: an 'SQLError' of the type 'ErrorTypeConversion' whose message+-- starts with "duckdb-simple: ".+assertRangeError :: IO a -> Assertion+assertRangeError action = do+ result <- try action+ case result of+ Left err -> do+ sqlErrorType err @?= Just ErrorTypeConversion+ assertBool (Text.unpack (sqlErrorMessage err)) ("duckdb-simple: " `Text.isPrefixOf` sqlErrorMessage err)+ Right _ -> assertFailure "expected an SQLError of the type ErrorTypeConversion"++-- | Require that the connection runs another query. A case uses it after a+-- cleanup or an interrupted query, to show that the connection state is released.+assertReusable :: Connection -> Assertion+assertReusable conn = do+ rows <- query_ conn "SELECT 42" :: IO [Only Int64]+ rows @?= [Only 42]++-- | Require a SQL exception which includes the expected message.+assertSqlErrorContaining :: Text.Text -> IO a -> Assertion+assertSqlErrorContaining expected action = do+ outcome <- try (void action)+ case outcome of+ Left err ->+ assertBool ("expected SQL error containing " <> show expected <> ", got " <> show err) $+ expected `Text.isInfixOf` sqlErrorMessage err+ Right () -> assertFailure "expected SQL error"++-- | Require an 'SQLError' of DuckDB with the given type. An error of+-- duckdb-simple, whose message starts with "duckdb-simple: ", fails the+-- assertion, also when it has the same type.+assertSqlErrorType :: ErrorType -> IO a -> Assertion+assertSqlErrorType expected action = do+ result <- try action+ case result of+ Left err -> do+ assertBool+ ("expected an error of DuckDB, got " <> Text.unpack (sqlErrorMessage err))+ (not ("duckdb-simple: " `Text.isPrefixOf` sqlErrorMessage err))+ sqlErrorType err @?= Just expected+ Right _ -> assertFailure ("expected an SQLError of the type " <> show expected)++-- | Run an action with a new token, and require that no value keeps the token+-- after the action and two major collections. The action gives the token to the+-- callbacks of a registration that fails, is replaced, or ends with its database.+assertReleasedAfter :: String -> (IORef Int64 -> IO ()) -> Assertion+assertReleasedAfter label action = do+ weak <- do+ token <- newIORef 1+ weak <- mkWeakIORef token (pure ())+ action token+ pure weak+ performMajorGC+ performMajorGC+ deRefWeak weak >>= assertBool (label <> " keeps its callback after the case") . isNothing++-- | Supply a constant C string for one native call.+withConstCString :: String -> (ConstPtr CChar -> IO a) -> IO a+withConstCString text action = withCString text (action . ConstPtr)++-- | Set the size of a BIGINT vector and write the values to its rows. The+-- function flattens the vector first, so that each row has its own slot: the+-- header gives the data pointer of a FLAT or CONSTANT vector, and a CONSTANT+-- vector has one value for all rows. A flatten does nothing for a FLAT vector.+writeIntegers :: DuckDBV2VectorHandle -> [Int64] -> IO ()+writeIntegers vector values = do+ checked (c_duckdb_v2_vector_set_size vector (fromIntegral (length values)))+ checked (c_duckdb_v2_vector_flatten vector)+ raw <- output (c_duckdb_v2_vector_get_data_mutable vector)+ forM_ (zip [0 ..] values) $ \(index, value) -> pokeElemOff (castPtr raw) index value++-- | Write NULL to each row of the output vector of a cast call. The function+-- reads no input row.+writeNullRows :: DuckDBV2CastFunctionExecInfoHandle -> IO ()+writeNullRows info = do+ rows <- output (c_duckdb_v2_cast_function_exec_get_row_count info)+ vector <- output (c_duckdb_v2_cast_function_exec_get_output info)+ checked (c_duckdb_v2_vector_set_size vector rows)+ checked (c_duckdb_v2_vector_flatten vector)+ forM_ [0 .. fromIntegral rows - 1 :: Int] $ \row -> checked (c_duckdb_v2_vector_set_null vector (fromIntegral row))++-- | Filtered queries over more than one chunk, each with the rows that DuckDB+-- gives. After the filter, most chunks of the second and the third query have+-- no row.+filteredQueries :: [(Query, [Int64])]+filteredQueries =+ [ ("SELECT i FROM range(100000) t(i) WHERE i < 3 OR i >= 99997", [0, 1, 2, 99997, 99998, 99999])+ , ("SELECT i FROM range(1000000) t(i) WHERE i % 100000 = 99999", [99999, 199999 .. 999999])+ , ("SELECT i FROM range(1000000) t(i) WHERE i = 999999", [999999])+ ]++-- | Make a 'SizedArray' from a list of its length. A list of another length+-- is a defect of the case.+sized :: (KnownNat n) => [a] -> SizedArray n a+sized values = fromMaybe (error ("sized: a list of " <> show (length values) <> " elements")) (sizedArrayFromList values)
test/TimeTests.hs view
@@ -1,92 +1,149 @@+{-# LANGUAGE DataKinds #-} {-# LANGUAGE DeriveGeneric #-} {-# LANGUAGE DerivingVia #-}+{-# LANGUAGE LambdaCase #-} {-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE ScopedTypeVariables #-} --- | Round trips for finite dates and native temporal infinities.+-- |+-- Module : TimeTests+-- Description : Round trips for finite dates and native temporal infinities. module TimeTests (timeTests) where -import Control.Exception (try)-import Control.Monad (forM_)+import Control.Exception (fromException, throwIO, try)+import Control.Monad (forM_, void) import Data.Array (listArray)+import Data.Int (Int64)+import qualified Data.Map.Strict as Map+import Data.Ratio ((%)) import qualified Data.Text as Text-import Data.Time (Day, LocalTime (..), TimeOfDay (..), UTCTime, fromGregorian, localTimeToUTC, utc)+import Data.Time (Day, LocalTime (..), TimeOfDay (..), UTCTime (..), addDays, fromGregorian, localTimeToUTC, utc, utcToLocalTime)+import Data.Time.Clock.POSIX (posixSecondsToUTCTime) import Database.DuckDB.Simple-import Database.DuckDB.Simple.FromField (Field (..), FieldValue (..), returnError)+import Database.DuckDB.Simple.FromField (Field (..), FieldValue (..), TimeWithZone (..), returnError) import Database.DuckDB.Simple.Generic (ViaDuckDB (..))+import Database.DuckDB.Simple.LogicalRep (+ LogicalTypeRep (..),+ StructField (..),+ StructValue (..),+ TupleValue (..),+ UnionMemberType (..),+ UnionValue (..),+ )+import Database.DuckDB.Simple.Ok (Ok (..))+import qualified Database.DuckDB.Simple.Registration as R import Database.DuckDB.Simple.Time+import Database.DuckDB.Simple.Variant (Variant (..)) import GHC.Generics (Generic) import Test.Tasty (TestTree, testGroup) import Test.Tasty.HUnit+import TestUtils (assertInputError, assertRangeError, sized) -- | Record fields exercise the generic scalar and collection instances. data TemporalRecord = TemporalRecord- { dateValue :: Date- , localValue :: LocalTimestamp- , utcValue :: UTCTimestamp- , optionalValue :: Maybe Date- , dates :: [Date]- }- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB TemporalRecord)+ { dateValue :: Date+ , localValue :: LocalTimestamp+ , utcValue :: UTCTimestamp+ , optionalValue :: Maybe Date+ , dates :: [Date]+ }+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB TemporalRecord) -- | Each constructor retains its temporal member type. data TemporalSum = ADate Date | ALocal LocalTimestamp | AUtc UTCTimestamp | ADateNull (Maybe Date)- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB TemporalSum)+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB TemporalSum) -- | A custom decoder must receive infinity before any finite conversion. newtype InfinitySign = InfinitySign Int- deriving (Eq, Show)+ deriving (Eq, Show) instance FromField InfinitySign where- fromField f@Field{fieldValue} = case fieldValue of- FieldDate NegInfinity -> pure (InfinitySign (-1))- FieldDate (Finite _) -> pure (InfinitySign 0)- FieldDate PosInfinity -> pure (InfinitySign 1)- _ -> returnError Incompatible f "expected DATE"+ fromField f@Field{fieldValue} = case fieldValue of+ FieldDate NegInfinity -> pure (InfinitySign (-1))+ FieldDate (Finite _) -> pure (InfinitySign 0)+ FieldDate PosInfinity -> pure (InfinitySign 1)+ _ -> returnError Incompatible f "expected DATE" -- | Exercise native binding, eager queries, cursors, and generic composites. timeTests :: TestTree timeTests =- testGroup- "temporal infinity"- [ testCase "dates bind and decode with their SQL type" $ withConnection ":memory:" $ \conn ->- forM_ [NegInfinity, Finite day, PosInfinity] $ \value -> do- (query conn "SELECT ?, typeof(?)" (value, value) :: IO [(Date, String)]) >>= (@?= [(value, "DATE")])- (query conn "SELECT ?::VARCHAR" (Only value) :: IO [Only String]) >>= (@?= [Only (dateText value)])- , testCase "local timestamps bind and decode with their SQL type" $ withConnection ":memory:" $ \conn ->- forM_ [NegInfinity, Finite local, PosInfinity] $ \value ->- (query conn "SELECT ?, typeof(?)" (value, value) :: IO [(LocalTimestamp, String)]) >>= (@?= [(value, "TIMESTAMP")])- , testCase "UTC timestamps preserve instants and infinities outside UTC" $ withConnection ":memory:" $ \conn -> do- _ <- execute_ conn "SET TimeZone='Pacific/Auckland'"- forM_ [NegInfinity, Finite instant, PosInfinity] $ \value ->- (query conn "SELECT ?, typeof(?)" (value, value) :: IO [(UTCTimestamp, String)]) >>= (@?= [(value, "TIMESTAMP WITH TIME ZONE")])- , testCase "infinity and NULL remain distinct" $ withConnection ":memory:" $ \conn -> do- (query conn "SELECT ?::DATE, ?::TIMESTAMP, ?::TIMESTAMPTZ" (Nothing :: Maybe Date, Nothing :: Maybe LocalTimestamp, Nothing :: Maybe UTCTimestamp) :: IO [(Maybe Date, Maybe LocalTimestamp, Maybe UTCTimestamp)]) >>= (@?= [(Nothing, Nothing, Nothing)])- (query_ conn "SELECT 'infinity'::DATE" :: IO [Only (Maybe Date)]) >>= (@?= [Only (Just PosInfinity)])- , testCase "ordinary finite targets reject infinity with a conversion error" $ withConnection ":memory:" $ \conn -> do- assertConversionError (query_ conn "SELECT 'infinity'::DATE" :: IO [Only Day])- assertConversionError (query_ conn "SELECT '-infinity'::TIMESTAMP" :: IO [Only LocalTime])- assertConversionError (query_ conn "SELECT 'infinity'::TIMESTAMPTZ" :: IO [Only UTCTime])- assertConversionError (query_ conn "SELECT 'infinity'::TIMESTAMP" :: IO [Only TimeOfDay])- (query_ conn "SELECT 42" :: IO [Only Int]) >>= (@?= [Only 42])- , testCase "custom FromField receives native infinity" $ withConnection ":memory:" $ \conn ->- (query_ conn "SELECT value FROM (VALUES ('-infinity'::DATE), (DATE '2000-01-02'), ('infinity'::DATE)) t(value) ORDER BY value" :: IO [Only InfinitySign]) >>= (@?= map (Only . InfinitySign) [-1, 0, 1])- , testCase "cursors preserve infinities" $ withConnection ":memory:" $ \conn -> do- values <- fold_ conn "SELECT value FROM (VALUES ('-infinity'::DATE), (DATE '2000-01-02'), ('infinity'::DATE)) t(value) ORDER BY value" [] (\acc (Only value) -> pure (value : acc))- reverse values @?= [NegInfinity, Finite day, PosInfinity]- , testCase "date arrays preserve infinities" $ withConnection ":memory:" $ \conn -> do- let values = listArray (0 :: Int, 2) [NegInfinity, Finite day, PosInfinity]- query conn "SELECT ?" (Only values) >>= (@?= [Only values])- , testCase "generic records retain finite values infinities and NULL" $ withConnection ":memory:" $ \conn -> do- let values = TemporalRecord NegInfinity (Finite local) PosInfinity Nothing [PosInfinity, Finite day, NegInfinity]- query conn "SELECT ?" (Only values) >>= (@?= [Only values])- , testCase "generic union payloads retain their types" $ withConnection ":memory:" $ \conn ->- forM_ [ADate NegInfinity, ADate (Finite day), ALocal PosInfinity, AUtc NegInfinity, ADateNull Nothing] $ \value ->- query conn "SELECT ?" (Only value) >>= (@?= [Only value])- ]+ testGroup+ "temporal infinity"+ $ [ testCase "dates bind and decode with their SQL type" $ withConnection ":memory:" $ \conn ->+ forM_ [NegInfinity, Finite day, PosInfinity] $ \value -> do+ (query conn "SELECT ?, typeof(?)" (value, value) :: IO [(Date, String)]) >>= (@?= [(value, "DATE")])+ (query conn "SELECT ?::VARCHAR" (Only value) :: IO [Only String]) >>= (@?= [Only (dateText value)])+ , testCase "local timestamps bind and decode with their SQL type" $ withConnection ":memory:" $ \conn ->+ forM_ [NegInfinity, Finite local, PosInfinity] $ \value ->+ (query conn "SELECT ?, typeof(?)" (value, value) :: IO [(LocalTimestamp, String)]) >>= (@?= [(value, "TIMESTAMP")])+ , testCase "UTC timestamps preserve instants and infinities outside UTC" $ withConnection ":memory:" $ \conn -> do+ _ <- execute_ conn "SET TimeZone='Pacific/Auckland'"+ forM_ [NegInfinity, Finite instant, PosInfinity] $ \value ->+ (query conn "SELECT ?, typeof(?)" (value, value) :: IO [(UTCTimestamp, String)]) >>= (@?= [(value, "TIMESTAMP WITH TIME ZONE")])+ , testCase "infinity and NULL remain distinct" $ withConnection ":memory:" $ \conn -> do+ ( query+ conn+ "SELECT ?::DATE, ?::TIMESTAMP, ?::TIMESTAMPTZ"+ (Nothing :: Maybe Date, Nothing :: Maybe LocalTimestamp, Nothing :: Maybe UTCTimestamp)+ :: IO [(Maybe Date, Maybe LocalTimestamp, Maybe UTCTimestamp)]+ )+ >>= (@?= [(Nothing, Nothing, Nothing)])+ (query_ conn "SELECT 'infinity'::DATE" :: IO [Only (Maybe Date)]) >>= (@?= [Only (Just PosInfinity)])+ , testCase "ordinary finite targets reject infinity with a conversion error" $ withConnection ":memory:" $ \conn -> do+ assertConversionError (query_ conn "SELECT 'infinity'::DATE" :: IO [Only Day])+ assertConversionError (query_ conn "SELECT '-infinity'::TIMESTAMP" :: IO [Only LocalTime])+ assertConversionError (query_ conn "SELECT 'infinity'::TIMESTAMPTZ" :: IO [Only UTCTime])+ assertConversionError (query_ conn "SELECT 'infinity'::TIMESTAMP" :: IO [Only TimeOfDay])+ (query_ conn "SELECT 42" :: IO [Only Int]) >>= (@?= [Only 42])+ , testCase "a conversion error names a time zone type as DuckDB does" $ withConnection ":memory:" $ \conn -> do+ [(timestamp, time)] <- query_ conn "SELECT TIMESTAMPTZ '2000-01-02 03:04:05+00', TIMETZ '03:04:05+00'"+ -- A Field of the type VARIANT names the type of its value with the+ -- text of DuckDB.+ forM_ [(timestamp, "TIMESTAMP WITH TIME ZONE"), (time, "TIME WITH TIME ZONE")] $ \(value, name) ->+ runConversion+ ( fromField Field{fieldName = "value", fieldIndex = 0, fieldValue = value, fieldType = pure (LogicalTypeScalar LogicalTypeIdVariant)}+ :: Conversion Int64+ )+ >>= \case+ Errors errors -> map (fmap errSQLType . fromException) errors @?= [Just name]+ Ok number -> assertFailure ("expected a conversion error, got " <> show number)+ -- The message of a query names the column type with the text of DuckDB.+ forM_+ [ ("SELECT TIMESTAMPTZ '2000-01-02 03:04:05+00'", " has type TIMESTAMP WITH TIME ZONE ")+ , ("SELECT TIMETZ '03:04:05+00'", " has type TIME WITH TIME ZONE ")+ ]+ $ \(sql, expected) -> do+ result <- try (query_ conn sql :: IO [Only Int64])+ case result of+ Left SQLError{sqlErrorMessage = message} -> assertBool (Text.unpack message) (expected `Text.isInfixOf` message)+ Right rows -> assertFailure ("expected a conversion error, got " <> show rows)+ , testCase "custom FromField receives native infinity" $ withConnection ":memory:" $ \conn ->+ ( query_ conn "SELECT value FROM (VALUES ('-infinity'::DATE), (DATE '2000-01-02'), ('infinity'::DATE)) t(value) ORDER BY value"+ :: IO [Only InfinitySign]+ )+ >>= (@?= map (Only . InfinitySign) [-1, 0, 1])+ , testCase "cursors preserve infinities" $ withConnection ":memory:" $ \conn -> do+ values <-+ fold_+ conn+ "SELECT value FROM (VALUES ('-infinity'::DATE), (DATE '2000-01-02'), ('infinity'::DATE)) t(value) ORDER BY value"+ []+ (\acc (Only value) -> pure (value : acc))+ reverse values @?= [NegInfinity, Finite day, PosInfinity]+ , testCase "date arrays preserve infinities" $ withConnection ":memory:" $ \conn -> do+ let values = sized [NegInfinity, Finite day, PosInfinity] :: SizedArray 3 (Unbounded Day)+ query conn "SELECT ?" (Only values) >>= (@?= [Only values])+ , testCase "generic records retain finite values infinities and NULL" $ withConnection ":memory:" $ \conn -> do+ let values = TemporalRecord NegInfinity (Finite local) PosInfinity Nothing [PosInfinity, Finite day, NegInfinity]+ query conn "SELECT ?" (Only values) >>= (@?= [Only values])+ , testCase "generic union payloads retain their types" $ withConnection ":memory:" $ \conn ->+ forM_ [ADate NegInfinity, ADate (Finite day), ALocal PosInfinity, AUtc NegInfinity, ADateNull Nothing] $ \value ->+ query conn "SELECT ?" (Only value) >>= (@?= [Only value])+ ]+ <> nanosecondTimeTests where day = fromGregorian 2000 1 2 local = LocalTime day (TimeOfDay 3 4 5)@@ -95,11 +152,723 @@ dateText PosInfinity = "infinity" dateText (Finite _) = "2000-01-02" +-- | Check explicit nanosecond targets without changing ordinary UTC parameter types.+nanosecondTimeTests :: [TestTree]+nanosecondTimeTests =+ [ testCase "nanosecond UTC results and VARIANTs preserve precision and infinity" $ withConnection ":memory:" $ \conn -> do+ _ <- execute_ conn "SET TimeZone='Pacific/Auckland'"+ let precise = localTimeToUTC utc (LocalTime (fromGregorian 2000 1 2) (TimeOfDay 3 4 5.123456789))+ forM_ [("'2000-01-02 03:04:05.123456789+00'", Finite precise), ("'infinity'", PosInfinity), ("'-infinity'", NegInfinity)] $ \(literal, expected) -> do+ let sql = Query ("SELECT " <> literal <> "::TIMESTAMPTZ_NS, (" <> literal <> "::TIMESTAMPTZ_NS)::VARIANT")+ (query_ conn sql :: IO [(UTCTimestamp, Variant)]) >>= (@?= [(expected, Variant (FieldTimestampTZ expected))])+ let value = Variant (FieldTimestampTZ (Finite precise))+ query conn "SELECT ?" (Only value) >>= (@?= [Only value])+ , testCase "explicit nanosecond UTC parameters preserve finite units and infinities" $ withConnection ":memory:" $ \conn -> do+ _ <- execute_ conn "SET TimeZone='Pacific/Auckland'"+ forM_ [minBound, negate (maxBound :: Int64) + 1, -1, 0, 1, 123456789, maxBound - 1] $ \nanos -> do+ let instant = posixSecondsToUTCTime (fromRational (toInteger nanos % 1000000000))+ (query conn "SELECT ?::TIMESTAMPTZ_NS" (Only instant) :: IO [Only UTCTime]) >>= (@?= [Only instant])+ (query conn "SELECT ?::TIMESTAMPTZ_NS" (Only (Finite instant)) :: IO [Only UTCTimestamp]) >>= (@?= [Only (Finite instant)])+ forM_ [NegInfinity, PosInfinity] $ \value ->+ (query conn "SELECT ?::TIMESTAMPTZ_NS" (Only (value :: UTCTimestamp)) :: IO [Only UTCTimestamp]) >>= (@?= [Only value])+ let precise = posixSecondsToUTCTime 0.123456789+ ordinary = posixSecondsToUTCTime 0.123456+ (query conn "SELECT typeof(?), ?" (precise, precise) :: IO [(String, UTCTime)]) >>= (@?= [("TIMESTAMP WITH TIME ZONE", ordinary)])+ , testCase "reused statements refresh nanosecond UTC parameter metadata after ALTER" $ withConnection ":memory:" $ \conn -> do+ void $ execute_ conn "CREATE TABLE utc_rebind(value TIMESTAMPTZ)"+ let coarse = posixSecondsToUTCTime 0.123456+ precise = posixSecondsToUTCTime 0.123456789+ withStatement conn "INSERT INTO utc_rebind VALUES (?)" $ \stmt -> do+ bind stmt [toField coarse]+ void $ executeStatement stmt+ void $ execute_ conn "ALTER TABLE utc_rebind ALTER COLUMN value TYPE TIMESTAMPTZ_NS"+ bind stmt [toField precise]+ void $ executeStatement stmt+ (query_ conn "SELECT epoch_ns(value) FROM utc_rebind ORDER BY value" :: IO [Only Int64]) >>= (@?= [Only 123456000, Only 123456789])+ , testCase "executeMany stores the same UTC values as single execute calls" $ withConnection ":memory:" $ \conn -> do+ let rows = [(t, t, t) | t <- map posixSecondsToUTCTime [0.123456789, 1.5, -0.000000001]]+ columns = " (tz TIMESTAMPTZ, ts TIMESTAMP, ns TIMESTAMPTZ_NS)"+ readBack table = query_ conn (Query ("SELECT epoch_ns(tz), epoch_ns(ts), epoch_ns(ns) FROM " <> table <> " ORDER BY ns")) :: IO [(Int64, Int64, Int64)]+ void $ execute_ conn (Query ("CREATE TABLE utc_many" <> columns))+ void $ execute_ conn (Query ("CREATE TABLE utc_single" <> columns))+ executeMany conn "INSERT INTO utc_many VALUES (?, ?, ?)" rows >>= (@?= 3)+ forM_ rows (void . execute conn "INSERT INTO utc_single VALUES (?, ?, ?)")+ many <- readBack "utc_many"+ readBack "utc_single" >>= (@?= many)+ -- The rows after the first row keep their nanoseconds with the kept parameter types.+ map (\(_, _, ns) -> ns) many @?= [-1, 123456789, 1500000000]+ , testCase "nanosecond UTC arrays retain NULLs and reject empty arrays" $ withConnection ":memory:" $ \conn -> do+ let precise = posixSecondsToUTCTime 0.123456789+ negative = posixSecondsToUTCTime (-0.000000001)+ values = sized [Just precise, Nothing, Just negative] :: SizedArray 3 (Maybe UTCTime)+ empty = sized [] :: SizedArray 0 (Maybe UTCTime)+ nulls = sized [Nothing, Nothing] :: SizedArray 2 (Maybe UTCTime)+ (query conn "SELECT ?::TIMESTAMPTZ_NS[]" (Only values) :: IO [Only [Maybe UTCTime]]) >>= (@?= [Only [Just precise, Nothing, Just negative]])+ assertEmptyArrayUnavailable (query conn "SELECT ?::TIMESTAMPTZ_NS[]" (Only empty) :: IO [Only [Maybe UTCTime]])+ (query conn "SELECT ?::TIMESTAMPTZ_NS[]" (Only nulls) :: IO [Only [Maybe UTCTime]]) >>= (@?= [Only [Nothing, Nothing]])+ (query conn "SELECT ?::TIMESTAMPTZ_NS" (Only (Nothing :: Maybe UTCTime)) :: IO [Only (Maybe UTCTime)]) >>= (@?= [Only Nothing])+ let nested = sized [values, values] :: SizedArray 2 (SizedArray 3 (Maybe UTCTime))+ (query conn "SELECT ?::TIMESTAMPTZ_NS[][]" (Only nested) :: IO [Only [[Maybe UTCTime]]])+ >>= (@?= [Only (replicate 2 [Just precise, Nothing, Just negative])])+ , testCase "nanosecond targets preserve matching STRUCT MAP and UNION leaves" $ withConnection ":memory:" $ \conn -> do+ let precise = Finite (posixSecondsToUTCTime 0.123456789)+ coarse = Finite (localTimeToUTC utc (LocalTime (fromGregorian 5000 1 1) (TimeOfDay 0 0 0)))+ utcType = LogicalTypeScalar LogicalTypeIdTimestampTz+ fields =+ listArray+ (0, 3)+ [ StructField "Precise" (FieldTimestampTZ precise)+ , StructField "Coarse" (FieldTimestampTZ coarse)+ , StructField "items" (FieldList [FieldTimestampTZ precise, FieldNull, FieldTimestampTZ NegInfinity])+ , StructField "lookup" (FieldMap [(FieldInt64 1, FieldTimestampTZ precise)])+ ]+ types =+ listArray+ (0, 3)+ [ StructField "Precise" utcType+ , StructField "Coarse" utcType+ , StructField "items" (LogicalTypeList utcType)+ , StructField "lookup" (LogicalTypeMap (LogicalTypeScalar LogicalTypeIdBigInt) utcType)+ ]+ struct = StructValue fields types (Map.fromList [("Precise", 0), ("Coarse", 1), ("items", 2), ("lookup", 3)])+ sql =+ "WITH bound AS (SELECT ?::STRUCT(coarse TIMESTAMPTZ, items TIMESTAMPTZ_NS[], lookup MAP(BIGINT, TIMESTAMPTZ_NS), precise TIMESTAMPTZ_NS) AS value) SELECT value.precise, value.coarse, value.items, map_values(value.lookup) FROM bound"+ (query conn sql (Only struct) :: IO [(UTCTimestamp, UTCTimestamp, [Maybe UTCTimestamp], [UTCTimestamp])])+ >>= (@?= [(precise, coarse, [Just precise, Nothing, Just NegInfinity], [precise])])+ let union = UnionValue 0 "Value" (FieldTimestampTZ precise) (listArray (0, 0) [UnionMemberType "Value" utcType])+ (query conn "SELECT union_extract(?::UNION(value TIMESTAMPTZ_NS), 'value')" (Only union) :: IO [Only UTCTimestamp]) >>= (@?= [Only precise])+ assertInputError+ (query conn sql (Only struct{structValueFields = listArray (0, 0) [StructField "wrong" (FieldTimestampTZ precise)]}) :: IO [Only FieldValue])+ assertInputError (query conn "SELECT ?::UNION(value TIMESTAMPTZ_NS)" (Only union{unionValueLabel = "wrong"}) :: IO [Only FieldValue])+ , testCase "nanosecond tuple targets match positional fields" $ withConnection ":memory:" $ \conn -> do+ let precise = Finite (posixSecondsToUTCTime 0.123456789)+ utcType = LogicalTypeScalar LogicalTypeIdTimestampTz+ structure names =+ StructValue+ (listArray (3, 5) (zipWith StructField names [FieldTimestampTZ precise, FieldTimestampTZ precise, FieldNull]))+ (listArray (7, 9) [StructField name utcType | name <- names])+ (Map.fromList (zip names [0 ..]))+ tuple = TupleValue (listArray (3, 5) [FieldTimestampTZ precise, FieldTimestampTZ precise, FieldNull]) (listArray (7, 9) (replicate 3 utcType))+ named = structure ["first", "second", "optional"]+ sql =+ "SELECT epoch_ns(value[1]), epoch_ns(value[2]), value[3] IS NULL FROM (SELECT ?::TUPLE(TIMESTAMPTZ,TIMESTAMPTZ_NS,TIMESTAMPTZ_NS) AS value)"+ (query conn sql (Only tuple) :: IO [(Int64, Int64, Bool)]) >>= (@?= [(123456000, 123456789, True)])+ (query conn sql (Only named) :: IO [(Int64, Int64, Bool)]) >>= (@?= [(123456000, 123456789, True)])+ ( query+ conn+ "SELECT epoch_ns(value.first), epoch_ns(value.second), value.optional IS NULL FROM (SELECT ?::STRUCT(first TIMESTAMPTZ,second TIMESTAMPTZ_NS,optional TIMESTAMPTZ_NS) AS value)"+ (Only tuple)+ :: IO [(Int64, Int64, Bool)]+ )+ >>= (@?= [(123456000, 123456789, True)])+ result <- try (query conn "SELECT ?::TUPLE(TIMESTAMPTZ,TIMESTAMPTZ_NS)" (Only tuple) :: IO [Only FieldValue])+ case result of+ Left (_ :: SQLError) -> pure ()+ Right _ -> assertFailure "expected tuple arity mismatch"+ , testCase "nested tuple lists retain independent temporal field metadata" $ withConnection ":memory:" $ \conn -> do+ let precise = Finite (posixSecondsToUTCTime 0.123456789)+ utcType = LogicalTypeScalar LogicalTypeIdTimestampTz+ innerTypes = listArray (4, 4) [StructField "value" utcType]+ inner = StructValue (listArray (2, 2) [StructField "value" (FieldTimestampTZ precise)]) innerTypes (Map.singleton "value" 0)+ tuple = TupleValue (listArray (3, 4) (replicate 2 (FieldStruct inner))) (listArray (6, 7) (replicate 2 (LogicalTypeStruct innerTypes)))+ container =+ StructValue+ (listArray (1, 1) [StructField "items" (FieldList [FieldTuple tuple])])+ (listArray (5, 5) [StructField "items" (LogicalTypeList (LogicalTypeTuple (tupleValueTypes tuple)))])+ (Map.singleton "items" 0)+ sql =+ "SELECT epoch_ns((value.items[1][1]).value), epoch_ns((value.items[1][2]).value) FROM (SELECT ?::STRUCT(items TUPLE(STRUCT(value TIMESTAMPTZ),STRUCT(value TIMESTAMPTZ_NS))[]) AS value)"+ (query conn sql (Only container) :: IO [(Int64, Int64)]) >>= (@?= [(123456000, 123456789)])+ , testCase "a time value keeps its digits below a microsecond only for a nanosecond parameter" $ withConnection ":memory:" $ \conn -> do+ _ <- execute_ conn "SET TimeZone='UTC'"+ let day = fromGregorian 2026 1 2+ time = TimeOfDay 3 4 5.123456999+ local = LocalTime day time+ instant = localTimeToUTC utc local+ stored :: (ToField a) => Text.Text -> Text.Text -> a -> IO [Only Text.Text]+ stored column cast value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ _ <- execute conn (Query ("INSERT INTO stored VALUES (?" <> cast <> ")")) (Only value)+ query_ conn "SELECT v::VARCHAR FROM stored"+ check :: (ToField a) => [(Text.Text, Text.Text, Text.Text)] -> a -> Assertion+ check rows value = forM_ rows $ \(column, cast, expected) ->+ stored column cast value >>= assertEqual (Text.unpack (column <> " ?" <> cast)) [Only expected]+ selected :: (ToField a) => Text.Text -> a -> IO [Only Text.Text]+ selected sql value = query conn (Query sql) (Only value)+ check+ [ ("TIMESTAMP_NS", "", "2026-01-02 03:04:05.123456999")+ , ("TIMESTAMP", "", "2026-01-02 03:04:05.123456")+ , ("TIMESTAMP_NS", "::TIMESTAMP_NS", "2026-01-02 03:04:05.123456999")+ ]+ local+ check [("TIMESTAMP_NS", "", "2026-01-02 03:04:05.123456999"), ("TIMESTAMP", "", "2026-01-02 03:04:05.123456")] (Finite local)+ check [("TIME_NS", "", "03:04:05.123456999"), ("TIME", "", "03:04:05.123456"), ("TIME_NS", "::TIME_NS", "03:04:05.123456999")] time+ check+ [ ("TIMESTAMPTZ_NS", "", "2026-01-02 03:04:05.123456999+00")+ , ("TIMESTAMPTZ", "", "2026-01-02 03:04:05.123456+00")+ , ("TIMESTAMPTZ_NS", "::TIMESTAMPTZ_NS", "2026-01-02 03:04:05.123456999+00")+ ]+ instant+ check [("TIMESTAMPTZ_NS", "", "2026-01-02 03:04:05.123456999+00"), ("TIMESTAMPTZ", "", "2026-01-02 03:04:05.123456+00")] (Finite instant)+ -- A parameter with no known type gets the microsecond value.+ selected "SELECT ?::VARCHAR" local >>= (@?= [Only "2026-01-02 03:04:05.123456"])+ selected "SELECT ?::VARCHAR" time >>= (@?= [Only "03:04:05.123456"])+ selected "SELECT ?::VARCHAR" instant >>= (@?= [Only "2026-01-02 03:04:05.123456+00"])+ -- A comparison with a nanosecond column finds the exact value.+ void $ execute_ conn "CREATE TABLE precise (v TIMESTAMP_NS, t TIME_NS)"+ void $ execute_ conn "INSERT INTO precise VALUES ('2026-01-02 03:04:05.123456999', '03:04:05.123456999')"+ (query conn "SELECT count(*) FROM precise WHERE v = ?" (Only local) :: IO [Only Int64]) >>= (@?= [Only 1])+ (query conn "SELECT count(*) FROM precise WHERE t = ?" (Only time) :: IO [Only Int64]) >>= (@?= [Only 1])+ , testCase "an explicit nanosecond cast for a microsecond column rounds as in SQL text" $ withConnection ":memory:" $ \conn -> do+ -- An explicit cast to a nanosecond type makes a nanosecond parameter. The+ -- cast of DuckDB from the nanosecond type to the microsecond type rounds.+ -- A plain parameter for the microsecond column cuts the digits.+ _ <- execute_ conn "SET TimeZone='UTC'"+ let local = LocalTime (fromGregorian 2026 1 2) (TimeOfDay 3 4 5.123456999)+ time = localTimeOfDay local+ instant = localTimeToUTC utc local+ stored :: (ToField a) => Text.Text -> Text.Text -> a -> IO [Only Text.Text]+ stored column cast value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ _ <- execute conn (Query ("INSERT INTO stored VALUES (?" <> cast <> ")")) (Only value)+ query_ conn "SELECT v::VARCHAR FROM stored"+ stored "TIMESTAMP" "::TIMESTAMP_NS" local >>= (@?= [Only "2026-01-02 03:04:05.123457"])+ stored "TIMESTAMP" "" local >>= (@?= [Only "2026-01-02 03:04:05.123456"])+ stored "TIME" "::TIME_NS" time >>= (@?= [Only "03:04:05.123457"])+ stored "TIME" "" time >>= (@?= [Only "03:04:05.123456"])+ stored "TIMESTAMPTZ" "::TIMESTAMPTZ_NS" instant >>= (@?= [Only "2026-01-02 03:04:05.123457+00"])+ stored "TIMESTAMPTZ" "" instant >>= (@?= [Only "2026-01-02 03:04:05.123456+00"])+ (query conn "SELECT (?::TIMESTAMP_NS + INTERVAL 1 DAY)::VARCHAR" (Only local) :: IO [Only Text.Text])+ >>= (@?= [Only "2026-01-03 03:04:05.123457"])+ (query conn "SELECT (? + INTERVAL 1 DAY)::VARCHAR" (Only local) :: IO [Only Text.Text]) >>= (@?= [Only "2026-01-03 03:04:05.123456"])+ , testCase "nested time values keep their digits below a microsecond only for nanosecond leaves" $ withConnection ":memory:" $ \conn -> do+ let day = fromGregorian 2026 1 2+ precise = LocalTime day (TimeOfDay 3 4 5.123456999)+ coarse = LocalTime day (TimeOfDay 3 4 5.123456)+ time = TimeOfDay 3 4 5.123456999+ stored :: (ToField a) => Text.Text -> a -> IO [Only Text.Text]+ stored column value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ _ <- execute conn "INSERT INTO stored VALUES (?)" (Only value)+ query_ conn "SELECT v::VARCHAR FROM stored"+ struct value logical =+ StructValue (listArray (0, 0) [StructField "v" value]) (listArray (0, 0) [StructField "v" logical]) (Map.singleton "v" 0)+ stored "TIMESTAMP_NS[]" [precise, coarse] >>= (@?= [Only "['2026-01-02 03:04:05.123456999', '2026-01-02 03:04:05.123456']"])+ stored "TIMESTAMP[]" [precise, coarse] >>= (@?= [Only "['2026-01-02 03:04:05.123456', '2026-01-02 03:04:05.123456']"])+ stored "TIMESTAMP_NS[2]" (sized [precise, coarse] :: SizedArray 2 LocalTime)+ >>= (@?= [Only "['2026-01-02 03:04:05.123456999', '2026-01-02 03:04:05.123456']"])+ stored "TIME_NS[]" [time] >>= (@?= [Only "['03:04:05.123456999']"])+ stored "TIME[]" [time] >>= (@?= [Only "['03:04:05.123456']"])+ stored "MAP(VARCHAR, TIMESTAMP_NS)" (Map.fromList [("a" :: Text.Text, precise)]) >>= (@?= [Only "{a='2026-01-02 03:04:05.123456999'}"])+ stored "MAP(VARCHAR, TIMESTAMP)" (Map.fromList [("a" :: Text.Text, precise)]) >>= (@?= [Only "{a='2026-01-02 03:04:05.123456'}"])+ let localStruct = struct (FieldTimestamp (Finite precise)) (LogicalTypeScalar LogicalTypeIdTimestamp)+ timeStruct = struct (FieldTime time) (LogicalTypeScalar LogicalTypeIdTime)+ stored "STRUCT(v TIMESTAMP_NS)" localStruct >>= (@?= [Only "{'v': '2026-01-02 03:04:05.123456999'}"])+ stored "STRUCT(v TIMESTAMP)" localStruct >>= (@?= [Only "{'v': '2026-01-02 03:04:05.123456'}"])+ stored "STRUCT(v TIME_NS)" timeStruct >>= (@?= [Only "{'v': '03:04:05.123456999'}"])+ stored "STRUCT(v TIME)" timeStruct >>= (@?= [Only "{'v': '03:04:05.123456'}"])+ let timestamp = LogicalTypeScalar LogicalTypeIdTimestamp+ list = LogicalTypeList timestamp+ record = LogicalTypeStruct (listArray (0, 0) [StructField "v" timestamp])+ listValue = FieldList [FieldTimestamp (Finite precise)]+ recordValue = FieldStruct localStruct+ forM_+ [ ("time_list", list, listValue, "STRUCT(v TIMESTAMP_NS[])", "{'v': ['2026-01-02 03:04:05.123456999']}")+ , ("time_record", record, recordValue, "STRUCT(v STRUCT(v TIMESTAMP_NS))", "{'v': {'v': '2026-01-02 03:04:05.123456999'}}")+ ]+ $ \(name, base, value, target, expected) -> do+ R.registerCustomType conn name (R.StructuredType base)+ let aliased = struct value (LogicalTypeAlias name base)+ (query conn "SELECT typeof((?).v)" (Only aliased) :: IO [Only Text.Text]) >>= (@?= [Only name])+ stored target aliased >>= (@?= [Only expected])+ let nested = struct (FieldList [recordValue]) (LogicalTypeList (LogicalTypeAlias "time_record" record))+ stored "STRUCT(v STRUCT(v TIMESTAMP_NS)[])" nested+ >>= (@?= [Only "{'v': [{'v': '2026-01-02 03:04:05.123456999'}]}"])+ , testCase "a FieldMap gives its key target to its keys and its value target to its values" $ withConnection ":memory:" $ \conn -> do+ _ <- execute_ conn "CREATE TABLE stored (v MAP(VARCHAR, TIME_NS))"+ _ <-+ execute+ conn+ "INSERT INTO stored VALUES (?)"+ (Only (FieldMap [(FieldText "a", FieldTime (TimeOfDay 3 4 5)), (FieldText "b", FieldTime (TimeOfDay 3 4 5.123456789))]))+ (query_ conn "SELECT v::VARCHAR FROM stored" :: IO [Only Text.Text]) >>= (@?= [Only "{a='03:04:05', b='03:04:05.123456789'}"])+ _ <- execute_ conn "CREATE TABLE stored_keys (v MAP(TIME_NS, VARCHAR))"+ _ <-+ execute+ conn+ "INSERT INTO stored_keys VALUES (?)"+ (Only (FieldMap [(FieldTime (TimeOfDay 3 4 5), FieldText "a"), (FieldTime (TimeOfDay 3 4 5.123456789), FieldText "b")]))+ (query_ conn "SELECT v::VARCHAR FROM stored_keys" :: IO [Only Text.Text]) >>= (@?= [Only "{'03:04:05'=a, '03:04:05.123456789'=b}"])+ , testCase "a FieldList with a microsecond element first keeps the nanosecond element" $ withConnection ":memory:" $ \conn -> do+ _ <- execute_ conn "CREATE TABLE stored (v TIME_NS[])"+ _ <- execute conn "INSERT INTO stored VALUES (?)" (Only (FieldList [FieldTime (TimeOfDay 3 4 5), FieldTime (TimeOfDay 3 4 5.123456789)]))+ (query_ conn "SELECT v::VARCHAR FROM stored" :: IO [Only Text.Text]) >>= (@?= [Only "['03:04:05', '03:04:05.123456789']"])+ , testCase "executeMany binds time rows with and without digits below a microsecond" $ withConnection ":memory:" $ \conn -> do+ let day = fromGregorian 2026 1 2+ rows =+ [ (0 :: Int64, LocalTime day (TimeOfDay 3 4 5.123456999))+ , (1, LocalTime day (TimeOfDay 3 4 5.123456))+ , (2, LocalTime day (TimeOfDay 3 4 5.000000001))+ ]+ forM_+ [ ("TIMESTAMP_NS", ["2026-01-02 03:04:05.123456999", "2026-01-02 03:04:05.123456", "2026-01-02 03:04:05.000000001"])+ , ("TIMESTAMP", ["2026-01-02 03:04:05.123456", "2026-01-02 03:04:05.123456", "2026-01-02 03:04:05"])+ ]+ $ \(column, expected) -> do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE many (i BIGINT, v " <> column <> ")"))+ _ <- executeMany conn "INSERT INTO many VALUES (?, ?)" rows+ (query_ conn "SELECT v::VARCHAR FROM many ORDER BY i" :: IO [Only Text.Text]) >>= assertEqual (Text.unpack column) (map Only expected)+ , testCase "a time value outside the nanosecond range and an infinity for a nanosecond parameter" $ withConnection ":memory:" $ \conn -> do+ let moment year month day hours minutes seconds = LocalTime (fromGregorian year month day) (TimeOfDay hours minutes seconds)+ early = moment 1600 1 2 3 4 5.123456999+ -- DuckDB can cast the microsecond value of this value to TIMESTAMP_NS.+ edge = moment 2262 4 11 23 47 16.854775900+ precise = moment 2026 1 2 3 4 5.123456999+ stamp = FieldTimestamp . Finite+ timestamp = LogicalTypeScalar LogicalTypeIdTimestamp+ pair second =+ StructValue+ (listArray (0, 1) [StructField "a" (stamp precise), StructField "b" (stamp second)])+ (listArray (0, 1) [StructField "a" timestamp, StructField "b" timestamp])+ (Map.fromList [("a", 0), ("b", 1)])+ tuple = TupleValue (listArray (0, 0) [stamp early]) (listArray (0, 0) [timestamp])+ union = UnionValue 0 "v" (stamp early) (listArray (0, 0) [UnionMemberType "v" timestamp])+ insert :: (ToField a) => Text.Text -> a -> IO ()+ insert column value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ void (execute conn "INSERT INTO stored VALUES (?)" (Only value))+ noRows label = (query_ conn "SELECT count(*) FROM stored" :: IO [Only Int64]) >>= assertEqual label [Only 0]+ duckdbError expected label action =+ try action >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertEqual label (Just expected) sqlErrorType+ assertBool label ("Could not convert Timestamp to higher precision" `Text.isInfixOf` sqlErrorMessage)+ noRows label+ Right () -> assertFailure (label <> ": expected the cast error of DuckDB")+ encoderError label action =+ try action >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ assertEqual label (Just ErrorTypeConversion) sqlErrorType+ assertBool label ("duckdb-simple: TIMESTAMP value out of finite range" `Text.isInfixOf` sqlErrorMessage)+ noRows label+ Right () -> assertFailure (label <> ": expected the range error of the encoder")+ -- DuckDB raises the error when it also cannot cast the microsecond value.+ forM_ [early, moment 1600 1 2 3 4 5, moment 2262 4 11 23 47 16.854776001, moment 1677 9 21 0 12 43.145224100] $ \value ->+ duckdbError ErrorTypeConversion (show value) (insert "TIMESTAMP_NS" value)+ duckdbError ErrorTypeConversion "UTCTime with no digits" (insert "TIMESTAMPTZ_NS" (localTimeToUTC utc (moment 1600 1 2 3 4 5)))+ duckdbError ErrorTypeConversion "UTCTime" (insert "TIMESTAMPTZ_NS" (localTimeToUTC utc early))+ -- For an element, the library checks the range before the constructor of+ -- the list, the array, or the map casts the element.+ encoderError "list" (insert "TIMESTAMP_NS[]" [precise, early])+ encoderError "SizedArray" (insert "TIMESTAMP_NS[2]" (sized [precise, early] :: SizedArray 2 LocalTime))+ encoderError "Map key" (insert "MAP(TIMESTAMP_NS, VARCHAR)" (Map.fromList [(moment 2300 1 1 0 0 0, "a" :: Text.Text)]))+ encoderError "Map value" (insert "MAP(VARCHAR, TIMESTAMP_NS)" (Map.fromList [("a" :: Text.Text, precise), ("b", early)]))+ encoderError "list of UTCTime with no digits" (insert "TIMESTAMPTZ_NS[]" [localTimeToUTC utc (moment 1600 1 2 3 4 5)])+ -- The encoder raises the error when the cast would cut the digits.+ encoderError "edge" (insert "TIMESTAMP_NS" edge)+ encoderError "edge list" (insert "TIMESTAMP_NS[]" [precise, edge])+ encoderError "UTCTime edge" (insert "TIMESTAMPTZ_NS" (localTimeToUTC utc edge))+ -- A leaf of a STRUCT, TUPLE, or UNION gets the error of the encoder, also in+ -- TRY_CAST, where a microsecond value would keep the other leaf with its+ -- digits cut.+ encoderError "STRUCT" (insert "STRUCT(a TIMESTAMP_NS, b TIMESTAMP_NS)" (pair early))+ encoderError "STRUCT with no digits" (insert "STRUCT(a TIMESTAMP_NS, b TIMESTAMP_NS)" (pair (moment 1600 1 2 3 4 5)))+ encoderError "TUPLE" (insert "TUPLE(TIMESTAMP_NS)" tuple)+ encoderError "UNION" (insert "UNION(v TIMESTAMP_NS)" union)+ assertRangeError+ (query conn "SELECT TRY_CAST(? AS STRUCT(a TIMESTAMP_NS, b TIMESTAMP_NS))::VARCHAR" (Only (pair early)) :: IO [Only Text.Text])+ -- In TRY_CAST, DuckDB gives NULL for a parameter that it cannot cast, and+ -- the check of a list element raises its error first.+ (query conn "SELECT TRY_CAST(? AS TIMESTAMP_NS) IS NULL" (Only early) :: IO [Only Bool]) >>= (@?= [Only True])+ try (query conn "SELECT TRY_CAST(? AS TIMESTAMP_NS[])::VARCHAR" (Only [precise, early]) :: IO [Only Text.Text]) >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorTypeConversion+ Right rows -> assertFailure ("TRY_CAST list: " <> show rows)+ -- The last values in the range keep their digits.+ insert "TIMESTAMP_NS[]" [moment 2262 4 11 23 47 16.854775806, moment 1677 9 21 0 12 43.145224500]+ (query_ conn "SELECT v::VARCHAR FROM stored" :: IO [Only Text.Text])+ >>= (@?= [Only "['2262-04-11 23:47:16.854775806', '1677-09-21 00:12:43.1452245']"])+ _ <- execute_ conn "CREATE OR REPLACE TABLE stored (v TIMESTAMP_NS, z TIMESTAMPTZ_NS)"+ _ <- execute conn "INSERT INTO stored VALUES (?, ?)" (PosInfinity :: LocalTimestamp, NegInfinity :: UTCTimestamp)+ (query_ conn "SELECT v::VARCHAR, z::VARCHAR FROM stored" :: IO [(Text.Text, Text.Text)]) >>= (@?= [("infinity", "-infinity")])+ , testCase "a time value with digits below a microsecond keeps them for a VARIANT parameter" $ withConnection ":memory:" $ \conn -> do+ _ <- execute_ conn "SET TimeZone = 'UTC'; CREATE TABLE v (x VARIANT); CREATE TABLE l (x VARIANT[])"+ let day = fromGregorian 2026 1 2+ tod = TimeOfDay 3 4 5.123456789+ local = LocalTime day tod+ early = LocalTime (fromGregorian 1600 1 2) tod+ stored :: (ToField a) => a -> IO [(Text.Text, Text.Text)]+ stored value = do+ _ <- execute_ conn "DELETE FROM v"+ _ <- execute conn "INSERT INTO v VALUES (?)" (Only value)+ query_ conn "SELECT variant_typeof(x), x::VARCHAR FROM v"+ -- A value with such digits gets the nanosecond type.+ stored tod >>= (@?= [("TIME_NANOS", "03:04:05.123456789")])+ stored local >>= (@?= [("TIMESTAMP_NANOS", "2026-01-02 03:04:05.123456789")])+ stored (Finite local) >>= (@?= [("TIMESTAMP_NANOS", "2026-01-02 03:04:05.123456789")])+ stored (localTimeToUTC utc local) >>= (@?= [("TIMESTAMP_NANOS_TZ", "2026-01-02 03:04:05.123456789+00")])+ -- A value with no such digits keeps the microsecond type.+ stored (TimeOfDay 3 4 5.123456) >>= (@?= [("TIME_MICROS", "03:04:05.123456")])+ stored (LocalTime day (TimeOfDay 3 4 5.123456)) >>= (@?= [("TIMESTAMP_MICROS", "2026-01-02 03:04:05.123456")])+ -- A value that the nanosecond type cannot hold gets the microsecond value.+ stored early >>= (@?= [("TIMESTAMP_MICROS", "1600-01-02 03:04:05.123456")])+ stored (LocalTime (fromGregorian 2262 4 11) (TimeOfDay 23 47 16.854775900)) >>= (@?= [("TIMESTAMP_MICROS", "2262-04-11 23:47:16.854775")])+ stored (localTimeToUTC utc early) >>= (@?= [("TIMESTAMP_MICROS_TZ", "1600-01-02 03:04:05.123456+00")])+ -- A value outside the range of TIMESTAMP_NS gets the microsecond value, as+ -- a parameter with no known type, also as a Variant payload.+ let far = LocalTime (fromGregorian 3000 1 2) tod+ (query conn "SELECT ?::VARCHAR" (Only far) :: IO [Only Text.Text]) >>= (@?= [Only "3000-01-02 03:04:05.123456"])+ stored far >>= (@?= [("TIMESTAMP_MICROS", "3000-01-02 03:04:05.123456")])+ stored (Variant (FieldTimestamp (Finite far))) >>= (@?= [("TIMESTAMP_MICROS", "3000-01-02 03:04:05.123456")])+ stored (Variant (FieldTimestampTZ (Finite (localTimeToUTC utc far)))) >>= (@?= [("TIMESTAMP_MICROS_TZ", "3000-01-02 03:04:05.123456+00")])+ -- A Variant payload and a FieldValue read from a VARIANT keep the digits.+ stored (Variant (FieldTimestamp (Finite local))) >>= (@?= [("TIMESTAMP_NANOS", "2026-01-02 03:04:05.123456789")])+ _ <-+ execute_+ conn+ "CREATE TABLE src (i INTEGER, x VARIANT); INSERT INTO src VALUES (1, '03:04:05.123456789'::TIME_NS), (2, '2026-01-02 03:04:05.123456789'::TIMESTAMP_NS), (3, '2026-01-02 03:04:05.123456789+00'::TIMESTAMPTZ_NS)"+ values <- query_ conn "SELECT i, x FROM src ORDER BY i" :: IO [(Int64, FieldValue)]+ forM_ values $ \(i, value) -> do+ _ <- execute_ conn "DELETE FROM v"+ _ <- execute conn "INSERT INTO v VALUES (?)" (Only value)+ (query conn "SELECT count(*) FROM src, v WHERE src.i = ? AND src.x IS NOT DISTINCT FROM v.x" (Only i) :: IO [Only Int64])+ >>= assertEqual (show value) [Only 1]+ -- A VARIANT object reads as a FieldStruct with VARIANT fields, so its leaves keep the digits.+ [Only object] <- query_ conn "SELECT {'t': '2026-01-02 03:04:05.123456789'::TIMESTAMP_NS}::VARIANT" :: IO [Only FieldValue]+ stored object >>= (@?= [("OBJECT(t)", "{'t': '2026-01-02 03:04:05.123456789'}")])+ -- A VARIANT array reads as a FieldList, whose elements get the microsecond value.+ [Only array] <- query_ conn "SELECT ['2026-01-02 03:04:05.123456789'::TIMESTAMP_NS]::VARIANT" :: IO [Only FieldValue]+ stored array >>= (@?= [("ARRAY(1)", "['2026-01-02 03:04:05.123456']")])+ -- An element of a VARIANT list keeps the microsecond type, as the list has one element type.+ _ <- execute conn "INSERT INTO l VALUES (?)" (Only [local, LocalTime day (TimeOfDay 3 4 5)])+ (query_ conn "SELECT variant_typeof(x[1]), x::VARCHAR FROM l" :: IO [(Text.Text, Text.Text)])+ >>= (@?= [("TIMESTAMP_MICROS", "[2026-01-02 03:04:05.123456, 2026-01-02 03:04:05]")])+ , testCase "a UTCTime outside the nanosecond range binds as a LocalTime does" $ withConnection ":memory:" $ \conn -> do+ -- See Note [Nanosecond parameters] in "Database.DuckDB.Simple.ToField".+ let at year time = LocalTime (fromGregorian year 1 1) time+ digits = TimeOfDay 0 0 0.000000001+ plain = TimeOfDay 0 0 0.000001+ typeOf :: (ToField a) => Text.Text -> a -> IO (Maybe (Maybe ErrorType))+ typeOf column value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ try (execute conn "INSERT INTO stored VALUES (?)" (Only value)) >>= \case+ Left SQLError{sqlErrorType} -> pure (Just sqlErrorType)+ Right _ -> pure Nothing+ -- DuckDB does not cast a parameter that it does not evaluate, so the+ -- microsecond value of each type gives the row.+ _ <- execute_ conn "CREATE TABLE present (c TIMESTAMP_NS, z TIMESTAMPTZ_NS)"+ _ <- execute_ conn "INSERT INTO present VALUES ('2000-01-01 00:00:00', '2000-01-01 00:00:00+00')"+ (query conn "SELECT coalesce(c, ?)::VARCHAR FROM present" (Only (at 2300 digits)) :: IO [Only Text.Text])+ >>= (@?= [Only "2000-01-01 00:00:00"])+ (query conn "SELECT count(*) FROM present WHERE z = ? OR true" (Only (localTimeToUTC utc (at 2300 digits))) :: IO [Only Int64])+ >>= (@?= [Only 1])+ -- DuckDB does not mix TIMESTAMPTZ_NS and TIMESTAMPTZ in coalesce, so a+ -- UTCTime there gets the Binder Error of DuckDB with and without digits.+ let coalesced instant =+ try (query conn "SELECT coalesce(z, ?) FROM present" (Only instant) :: IO [Only UTCTime]) >>= \case+ Left SQLError{sqlErrorType} -> pure sqlErrorType+ Right _ -> pure Nothing+ coalesced (localTimeToUTC utc (at 2300 digits)) >>= (@?= Just ErrorQueryBinder)+ coalesced (localTimeToUTC utc (at 2300 plain)) >>= (@?= Just ErrorQueryBinder)+ -- A plain parameter and a list element get the same error for each type.+ forM_ [(2300, digits), (2300, plain), (300000, digits), (300000, plain)] $ \(year, time) -> do+ let local = at year time+ instant = localTimeToUTC utc local+ label position = show year <> " " <> show time <> " " <> position+ localPlain <- typeOf "TIMESTAMP_NS" local+ utcPlain <- typeOf "TIMESTAMPTZ_NS" instant+ assertEqual (label "plain parameter") localPlain utcPlain+ localList <- typeOf "TIMESTAMP_NS[]" [local]+ utcList <- typeOf "TIMESTAMPTZ_NS[]" [instant]+ assertEqual (label "list element") localList utcList+ assertEqual (label "plain parameter type") (Just (Just ErrorTypeConversion)) utcPlain+ assertEqual (label "list element type") (Just (Just ErrorTypeConversion)) utcList+ let rejects :: (ToField a) => String -> Text.Text -> a -> ErrorType -> Text.Text -> Assertion+ rejects position column value expectedType expectedText = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ try (execute conn "INSERT INTO stored VALUES (?)" (Only value)) >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ assertEqual (position <> " type") (Just expectedType) sqlErrorType+ assertBool (position <> ": " <> Text.unpack sqlErrorMessage) (expectedText `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure (position <> ": expected an error")+ late = localTimeToUTC utc (at 2300 digits)+ -- A Map value and an Array element get the range check of an element,+ -- before the native constructor casts them. A FieldTimestampTZ binds the+ -- microsecond value, and DuckDB raises its error.+ rejects+ "Map value"+ "MAP(VARCHAR, TIMESTAMPTZ_NS)"+ (Map.singleton ("k" :: Text.Text) late)+ ErrorTypeConversion+ "TIMESTAMP value out of finite range"+ rejects+ "SizedArray element"+ "TIMESTAMPTZ_NS[1]"+ (sized [late] :: SizedArray 1 UTCTime)+ ErrorTypeConversion+ "TIMESTAMP value out of finite range"+ rejects+ "FieldTimestampTZ"+ "TIMESTAMPTZ_NS"+ (FieldTimestampTZ (Finite late))+ ErrorTypeConversion+ "Could not convert Timestamp to higher precision"+ -- The 2262 zone keeps the error of the encoder, from its first to its last+ -- nanosecond. The nanosecond before the zone binds with its digits.+ let near nanos = localTimeToUTC utc (LocalTime (fromGregorian 2262 4 11) (TimeOfDay 23 47 nanos))+ forM_ [16.854775807, 16.8547759, 16.854775999] $ \nanos ->+ rejects ("2262 zone " <> show nanos) "TIMESTAMPTZ_NS" (near nanos) ErrorTypeConversion "TIMESTAMP value out of finite range"+ _ <- execute conn "INSERT INTO stored VALUES (?)" (Only (near 16.854775806))+ (query_ conn "SELECT epoch_ns(v) FROM stored" :: IO [Only Int64]) >>= (@?= [Only (maxBound - 1)])+ -- A microsecond target and a VARIANT keep their values.+ typeOf "TIMESTAMPTZ" (localTimeToUTC utc (at 2300 digits)) >>= (@?= Nothing)+ typeOf "VARIANT" (localTimeToUTC utc (at 2300 digits)) >>= (@?= Nothing)+ , testCase "a FieldList or a FieldMap of time values for VARIANT elements binds microsecond elements" $ withConnection ":memory:" $ \conn -> do+ let precise = FieldTimestamp (Finite (LocalTime (fromGregorian 2026 1 2) (TimeOfDay 3 4 5.123456789)))+ (query conn "SELECT [variant_typeof(e) FOR e IN ?::VARIANT[]]" (Only (FieldList [precise])) :: IO [Only [Text.Text]])+ >>= (@?= [Only ["TIMESTAMP_MICROS"]])+ (query conn "SELECT variant_typeof((?::MAP(VARCHAR, VARIANT))['a'])" (Only (FieldMap [(FieldText "a", precise)])) :: IO [Only Text.Text])+ >>= (@?= [Only "TIMESTAMP_MICROS"])+ , testCase "a Map or an Array of time values for VARIANT elements binds microsecond values" $ withConnection ":memory:" $ \conn -> do+ let precise = LocalTime (fromGregorian 2026 1 2) (TimeOfDay 3 4 5.123456789)+ (query conn "SELECT (?::MAP(VARCHAR, VARIANT))::VARCHAR" (Only (Map.singleton ("a" :: Text.Text) precise)) :: IO [Only Text.Text])+ >>= (@?= [Only "{a=2026-01-02 03:04:05.123456}"])+ (query conn "SELECT (?::VARIANT[1])::VARCHAR" (Only (sized [precise] :: SizedArray 1 LocalTime)) :: IO [Only Text.Text])+ >>= (@?= [Only "[2026-01-02 03:04:05.123456]"])+ , testCase "nanosecond UTC targets reject finite overflow and sentinel collisions" $ withConnection ":memory:" $ \conn -> do+ forM_+ [ toInteger (minBound :: Int64) - 1+ , negate (toInteger (maxBound :: Int64))+ , toInteger (maxBound :: Int64)+ , toInteger (maxBound :: Int64) + 1+ ]+ $ \nanos -> do+ -- A UTCTime gets the error of a LocalTime of the same instant in UTC:+ -- below the range, the cast of DuckDB raises; at the top, the encoder+ -- raises, because the cast would cut the digits. See+ -- Note [Nanosecond parameters] in "Database.DuckDB.Simple.ToField".+ let instant = posixSecondsToUTCTime (fromRational (nanos % 1000000000))+ errorType :: (ToField a) => Query -> a -> IO (Maybe ErrorType)+ errorType sql value =+ try (query conn sql (Only value) :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorType} -> pure sqlErrorType+ Right rows -> assertFailure ("expected an error, got " <> show rows)+ expected = Just ErrorTypeConversion+ errorType "SELECT ?::TIMESTAMPTZ_NS" instant >>= (@?= expected)+ errorType "SELECT ?::TIMESTAMPTZ_NS" (Finite instant :: UTCTimestamp) >>= (@?= expected)+ errorType "SELECT ?::TIMESTAMP_NS" (utcToLocalTime utc instant) >>= (@?= expected)+ (query_ conn "SELECT 42" :: IO [Only Int]) >>= (@?= [Only 42])+ , testCase "dates and timestamps at the ends of the range of DuckDB bind, and a value outside raises an error" $ withConnection ":memory:" $ \conn -> do+ -- See Note [Time ranges] in "Database.DuckDB.Simple.ToField".+ let epoch = fromGregorian 1970 1 1+ firstDay = addDays (-2147483646) epoch+ -- Day -2147483647 is -infinity, so -2147483648 is the only INT32 day before the range.+ beforeDay = addDays (-2147483648) epoch+ lastDay = addDays 2147483646 epoch+ -- Day 2147483647 is infinity.+ afterDay = addDays 2147483647 epoch+ local units count = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (count % units)))+ firstTime = local 1000000 (-9223372022400000000)+ beforeTime = local 1000000 (-9223372022400000001)+ lastTime = local 1000000 9223372036854775806+ -- The microsecond value 2^63 - 1 is infinity.+ afterTime = local 1000000 9223372036854775807+ structOf logical value =+ StructValue+ (listArray (0, 0) [StructField "a" value])+ (listArray (0, 0) [StructField "a" (LogicalTypeScalar logical)])+ (Map.fromList [("a", 0)])+ create column = void (execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")")))+ stored :: forall a. (ToField a, FromField a, Eq a, Show a) => Text.Text -> a -> IO ()+ stored column value = do+ create column+ void (execute conn "INSERT INTO stored VALUES (?)" (Only value))+ (query_ conn "SELECT v FROM stored" :: IO [Only a]) >>= assertEqual (Text.unpack column) [Only value]+ rejected :: (ToField a) => Text.Text -> a -> IO ()+ rejected column value = do+ create column+ try (execute conn "INSERT INTO stored VALUES (?)" (Only value)) >>= \case+ Left SQLError{sqlErrorType} -> assertEqual (Text.unpack column) (Just ErrorTypeConversion) sqlErrorType+ Right _ -> assertFailure (Text.unpack column <> ": expected the range error")+ (query_ conn "SELECT count(*) FROM stored" :: IO [Only Int64]) >>= assertEqual (Text.unpack column) [Only 0]+ stored "DATE" firstDay+ rejected "DATE" beforeDay+ stored "DATE" lastDay+ rejected "DATE" afterDay+ -- A value at -infinity, and a value below the storage, get the same error.+ rejected "DATE" (addDays (-2147483647) epoch)+ rejected "DATE" (addDays (-3000000000) epoch)+ stored "DATE" (Finite firstDay :: Date)+ rejected "DATE" (Finite beforeDay :: Date)+ stored "DATE[]" [firstDay]+ rejected "DATE[]" [firstDay, beforeDay]+ stored "STRUCT(a DATE)" (structOf LogicalTypeIdDate (FieldDate (Finite firstDay)))+ rejected "STRUCT(a DATE)" (structOf LogicalTypeIdDate (FieldDate (Finite beforeDay)))+ stored "VARIANT" (Variant (FieldDate (Finite firstDay)))+ rejected "VARIANT" (Variant (FieldDate (Finite beforeDay)))+ stored "TIMESTAMP" firstTime+ rejected "TIMESTAMP" beforeTime+ stored "TIMESTAMP" lastTime+ rejected "TIMESTAMP" afterTime+ rejected "TIMESTAMP" (local 1000000 (-9223372036854775807))+ rejected "TIMESTAMP" (local 1000000 (-9223372036854775809))+ stored "TIMESTAMP" (Finite firstTime :: LocalTimestamp)+ rejected "TIMESTAMP" (Finite beforeTime :: LocalTimestamp)+ stored "TIMESTAMP[]" [firstTime]+ rejected "TIMESTAMP[]" [firstTime, beforeTime]+ stored "STRUCT(a TIMESTAMP)" (structOf LogicalTypeIdTimestamp (FieldTimestamp (Finite firstTime)))+ rejected "STRUCT(a TIMESTAMP)" (structOf LogicalTypeIdTimestamp (FieldTimestamp (Finite beforeTime)))+ stored "VARIANT" (Variant (FieldTimestamp (Finite firstTime)))+ rejected "VARIANT" (Variant (FieldTimestamp (Finite beforeTime)))+ stored "TIMESTAMPTZ" (localTimeToUTC utc firstTime)+ rejected "TIMESTAMPTZ" (localTimeToUTC utc beforeTime)+ stored "TIMESTAMPTZ" (localTimeToUTC utc lastTime)+ rejected "TIMESTAMPTZ" (localTimeToUTC utc afterTime)+ stored "TIMESTAMPTZ" (Finite (localTimeToUTC utc firstTime) :: UTCTimestamp)+ rejected "TIMESTAMPTZ" (Finite (localTimeToUTC utc beforeTime) :: UTCTimestamp)+ -- A second or a millisecond value has the range of a microsecond value at+ -- both ends. A STRUCT field can have these types.+ forM_+ [ (LogicalTypeIdTimestampSec, "STRUCT(a TIMESTAMP_S)", 1, 9223372036854)+ , (LogicalTypeIdTimestampMs, "STRUCT(a TIMESTAMP_MS)", 1000, 9223372036854775)+ ]+ $ \(logical, column, units, largest) -> do+ let at count = structOf logical (FieldTimestamp (Finite (local units count)))+ stored column (at (-9223372022400000000 * units `div` 1000000))+ stored column (at largest)+ rejected column (at (-9223372022400000000 * units `div` 1000000 - 1))+ rejected column (at (largest + 1))+ , testCase "each bind path checks the range of DuckDB for a date and a timestamp" $ withConnection ":memory:" $ \conn -> do+ -- See Note [Time ranges] in "Database.DuckDB.Simple.ToField".+ let epoch = fromGregorian 1970 1 1+ local micros = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (micros % 1000000)))+ rejected :: (ToField a) => Text.Text -> String -> Text.Text -> a -> IO ()+ rejected bound label column value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ try (execute conn "INSERT INTO stored VALUES (?)" (Only value)) >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ assertEqual label (Just ErrorTypeConversion) sqlErrorType+ assertBool (label <> ": " <> Text.unpack sqlErrorMessage) (bound `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure (label <> ": expected the range error")+ forM_+ [ ("below the smallest", addDays (-2147483648) epoch, local (-9223372022400000001))+ , ("above the largest", addDays 2147483647 epoch, local 9223372036854775807)+ ]+ $ \(bound, day, time) -> do+ forM_ [("DATE", FieldDate (Finite day)), ("TIMESTAMP", FieldTimestamp (Finite time))] $ \(sqlType, value) -> do+ let label part = Text.unpack (bound <> " " <> sqlType) <> " " <> part+ rejected bound (label "FieldValue") sqlType value+ rejected bound (label "FieldValue for a VARIANT") "VARIANT" value+ rejected bound (label "FieldList") (sqlType <> "[]") (FieldList [FieldNull, value])+ rejected bound (label "FieldArray") (sqlType <> "[2]") (FieldArray (listArray (0, 1) [FieldNull, value]))+ rejected bound (label "FieldMap key") ("MAP(" <> sqlType <> ", VARCHAR)") (FieldMap [(value, FieldText "v")])+ rejected bound (label "FieldMap value") ("MAP(VARCHAR, " <> sqlType <> ")") (FieldMap [(FieldText "k", value)])+ rejected bound (label "Variant payload in a list") "VARIANT" (Variant (FieldList [value]))+ let label part = Text.unpack bound <> " " <> part+ rejected bound (label "list of Maybe Day") "DATE[]" [Nothing, Just day]+ rejected bound (label "Map key of Day") "MAP(DATE, VARCHAR)" (Map.fromList [(day, "v" :: Text.Text)])+ rejected bound (label "Map of Day") "MAP(VARCHAR, DATE)" (Map.fromList [("k" :: Text.Text, day)])+ rejected bound (label "SizedArray of LocalTime") "TIMESTAMP[1]" (sized [time] :: SizedArray 1 LocalTime)+ rejected bound (label "Map key of LocalTime") "MAP(TIMESTAMP, VARCHAR)" (Map.fromList [(time, "v" :: Text.Text)])+ rejected bound (label "Map of LocalTime") "MAP(VARCHAR, TIMESTAMP)" (Map.fromList [("k" :: Text.Text, time)])+ rejected bound (label "LocalTime for a VARIANT") "VARIANT" time+ , testCase "a leap second binds as the next second, as make_time and make_timestamp give it" $ withConnection ":memory:" $ \conn -> do+ let leap = TimeOfDay 23 59 60+ next = LocalTime (fromGregorian 2017 1 1) (TimeOfDay 0 0 0)+ -- The SQL literal '23:59:60' raises an error. See Note [Time ranges] in+ -- "Database.DuckDB.Simple.ToField".+ ( query+ conn+ "SELECT ?, make_time(23, 59, 60), ?, make_timestamp(2016, 12, 31, 23, 59, 60), ?, ?"+ (leap, LocalTime (fromGregorian 2016 12 31) leap, TimeWithZone leap 0, UTCTime (fromGregorian 2016 12 31) 86400)+ :: IO [(TimeOfDay, TimeOfDay, LocalTime, LocalTime, TimeWithZone, UTCTime)]+ )+ >>= (@?= [(TimeOfDay 24 0 0, TimeOfDay 24 0 0, next, next, TimeWithZone (TimeOfDay 24 0 0) 0, localTimeToUTC utc next)])+ -- A TIME_NS value with 60 seconds also binds as the next second.+ (query conn "SELECT ?::TIME_NS, ?::TIME_NS" (TimeOfDay 23 59 60, TimeOfDay 12 59 60.000000001) :: IO [(TimeOfDay, TimeOfDay)])+ >>= (@?= [(TimeOfDay 24 0 0, TimeOfDay 13 0 0.000000001)])+ -- A time after 24:00:00 raises an error, for example 60.5 seconds in the last minute of a day.+ assertRangeError (query conn "SELECT ?" (Only (TimeOfDay 23 59 60.5)) :: IO [Only TimeOfDay])+ assertRangeError (query conn "SELECT ?" (Only (LocalTime (fromGregorian 2016 12 31) (TimeOfDay 23 59 60.5))) :: IO [Only LocalTime])+ assertRangeError (query conn "SELECT ?" (Only (UTCTime (fromGregorian 2016 12 31) 86400.5)) :: IO [Only UTCTime])+ assertRangeError (query conn "SELECT ?::TIME_NS" (Only (TimeOfDay 23 59 60.000000001)) :: IO [Only TimeOfDay])+ , testCase "each bind path raises the range error for a time of day that is not a TIME value" $ withConnection ":memory:" $ \conn -> do+ -- See timeOfDayUnits in "Database.DuckDB.Simple.ToField".+ let rejected :: (ToField a) => String -> Text.Text -> Text.Text -> a -> IO ()+ rejected label expected column value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE stored (v " <> column <> ")"))+ outcome <- try (execute conn "INSERT INTO stored VALUES (?)" (Only value))+ either (\err -> assertBool (label <> ": " <> show err) (expected `Text.isInfixOf` sqlErrorMessage err)) (const (pure ())) outcome+ assertRangeError (either throwIO pure outcome)+ forM_+ [ ("24:00:00 + 1 ps", TimeOfDay 24 0 0.000000000001, "time of day out of range")+ , ("24:00:01", TimeOfDay 24 0 1, "time of day out of range")+ , ("23:59:60.5", TimeOfDay 23 59 60.5, "time of day out of range")+ , ("25:00:00", TimeOfDay 25 0 0, "time of day out of range")+ , ("23:61:00", TimeOfDay 23 61 0, "invalid time of day")+ , ("-01:00:00", TimeOfDay (-1) 0 0, "invalid time of day")+ , ("23:59:61", TimeOfDay 23 59 61, "invalid time of day")+ ]+ $ \(name, time, expected) -> do+ let label part = name <> " " <> part+ timeType = LogicalTypeScalar LogicalTypeIdTime+ leaf =+ FieldStruct+ ( StructValue+ (listArray (0, 0) [StructField "a" (FieldTime time)])+ (listArray (0, 0) [StructField "a" timeType])+ (Map.fromList [("a", 0)])+ )+ stamp = LocalTime (fromGregorian 2026 1 2) time+ rejected (label "TimeOfDay") expected "TIME" time+ rejected (label "TimeOfDay for TIME_NS") expected "TIME_NS" time+ rejected (label "TimeWithZone") expected "TIMETZ" (TimeWithZone time 0)+ rejected (label "list of TimeOfDay") expected "TIME[]" [time]+ rejected (label "list of TimeOfDay for TIME_NS[]") expected "TIME_NS[]" [time]+ rejected (label "list of TimeWithZone") expected "TIMETZ[]" [TimeWithZone time 0]+ rejected (label "Map of TimeOfDay") expected "MAP(VARCHAR, TIME)" (Map.fromList [("k" :: Text.Text, time)])+ rejected (label "FieldTime") expected "TIME" (FieldTime time)+ rejected (label "FieldTimeTZ") expected "TIMETZ" (FieldTimeTZ (TimeWithZone time 0))+ rejected (label "FieldStruct leaf") expected "STRUCT(a TIME)" leaf+ rejected (label "Variant payload") expected "VARIANT" (Variant (FieldTime time))+ rejected (label "Variant list payload") expected "VARIANT" (Variant (FieldList [FieldTime time]))+ rejected (label "SizedArray of TimeOfDay") expected "TIME[1]" (sized [time] :: SizedArray 1 TimeOfDay)+ rejected+ (label "FieldTuple leaf")+ expected+ "TUPLE(TIME)"+ (FieldTuple (TupleValue (listArray (0, 0) [FieldTime time]) (listArray (0, 0) [timeType])))+ rejected+ (label "FieldUnion leaf")+ expected+ "UNION(a TIME)"+ (FieldUnion (UnionValue 0 "a" (FieldTime time) (listArray (0, 0) [UnionMemberType "a" timeType])))+ rejected (label "LocalTime") expected "TIMESTAMP" stamp+ rejected (label "FieldTimestamp") expected "TIMESTAMP" (FieldTimestamp (Finite stamp))+ -- A UTCTime holds a time after 24:00:00 only in a leap second.+ rejected "UTCTime at 23:59:60.5" "time of day out of range" "TIMESTAMPTZ" (UTCTime (fromGregorian 2026 1 2) 86400.5)+ rejected "UTCTime at 23:59:61.5" "invalid time of day" "TIMESTAMPTZ" (UTCTime (fromGregorian 2026 1 2) 86401.5)+ -- A time with digits below a microsecond binds through the nanosecond+ -- encoder for a TIME_NS target.+ forM_ [TimeOfDay 24 0 0.000000001, TimeOfDay 23 59 60.000000001] $ \time -> do+ rejected (show time <> " TimeOfDay for TIME_NS") "time of day out of range" "TIME_NS" time+ rejected (show time <> " list for TIME_NS[]") "time of day out of range" "TIME_NS[]" [time]+ ]+ -- | Require a field conversion error, rather than an unrelated exception. assertConversionError :: IO a -> Assertion assertConversionError action = do- result <- try action- case result of- Left SQLError{sqlErrorMessage = message} ->- assertBool "error must identify infinity" ("infinity" `Text.isInfixOf` message)- Right _ -> assertFailure "expected a conversion error"+ result <- try action+ case result of+ Left SQLError{sqlErrorMessage = message} ->+ assertBool "error must identify infinity" ("infinity" `Text.isInfixOf` message)+ Right _ -> assertFailure "expected a conversion error"++-- | Require the error of the library for an ARRAY parameter with no elements.+assertEmptyArrayUnavailable :: IO a -> Assertion+assertEmptyArrayUnavailable action = do+ result <- try action+ case result of+ Left SQLError{sqlErrorMessage = message, sqlErrorType} -> do+ assertBool "must identify empty ARRAY" ("empty ARRAY" `Text.isInfixOf` message)+ sqlErrorType @?= Just ErrorInputInvalid+ Right _ -> assertFailure "expected the empty ARRAY error"
+ test/TypedRegistrationTests.hs view
@@ -0,0 +1,2023 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE DeriveGeneric #-}+{-# LANGUAGE DerivingVia #-}+{-# LANGUAGE GeneralizedNewtypeDeriving #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : TypedRegistrationTests+-- Description : Integration checks for typed registration through SQL.+module TypedRegistrationTests (typedRegistrationTests) where++import Control.Exception (SomeException, bracket, displayException, throwIO, try)+import Control.Monad (forM, forM_, void, when)+import Data.Array (listArray)+import qualified Data.Geometry as G+import Data.IORef (IORef, atomicModifyIORef', mkWeakIORef, modifyIORef', newIORef, readIORef, writeIORef)+import Data.Int (Int16, Int32, Int64)+import qualified Data.Map.Strict as Map+import Data.Maybe (fromJust, isJust, isNothing)+import Data.Text (Text)+import qualified Data.Text as Text+import qualified Data.Text.IO as TextIO+import Data.Time.Calendar (Day, addDays, fromGregorian)+import Data.Time.LocalTime (TimeOfDay (..))+import Data.Word (Word64)+import Database.DuckDB.Simple+import Database.DuckDB.Simple.Config (+ CustomOption (..),+ SettingScope (..),+ customOption,+ getConfigValue,+ registerCustomOption,+ setConfigValue,+ )+import Database.DuckDB.Simple.Copy (+ CopyBindInfo (..),+ CopyFromInfo (..),+ CopyInitInfo (..),+ CopySinkInfo (..),+ CopyTo (..),+ defaultMultiFileOptions,+ registerCopyFromFunction,+ registerCopyFunction,+ registerCopyToFunction,+ registerMultiFileFunction,+ )+import Database.DuckDB.Simple.FromField (BigNum (..), Field (..), FieldValue (..), IntervalValue (..), TimeWithZone (..))+import Database.DuckDB.Simple.Function (FunctionResult)+import Database.DuckDB.Simple.Generic (DuckValue (..), ViaDuckDB (..))+import Database.DuckDB.Simple.Internal (Connection (..))+import qualified Database.DuckDB.Simple.Logging as Logging+import Database.DuckDB.Simple.LogicalRep (+ LogicalTypeRep (..),+ StructValue (..),+ TupleValue,+ UnionValue,+ )+import Database.DuckDB.Simple.Ok (Ok (..))+import Database.DuckDB.Simple.Pragma (PragmaCallback (..), registerPragmaFunction)+import Database.DuckDB.Simple.Registration (+ FunctionParameter (..),+ Replacement (..),+ TypeDeclaration (..),+ defaultValue,+ parameter,+ registerCustomType,+ registerInstanceReplacementScan,+ registerReplacementScan,+ )+import Database.DuckDB.Simple.ToField (ToDuckValue (..))+import qualified Database.DuckDB.Simple.V2.Registration as R+import Database.DuckDB.Simple.Variant (Variant (..))+import GHC.Generics (Generic)+import Numeric.Natural (Natural)+import System.Directory (createDirectory, getTemporaryDirectory, removeFile, removePathForcibly)+import System.FilePath ((</>))+import System.IO (hClose, openTempFile)+import System.IO.Unsafe (unsafeInterleaveIO)+import System.Mem (performMajorGC)+import System.Mem.Weak (Weak, deRefWeak)+import Test.Tasty (TestTree, localOption, mkTimeout, testGroup)+import Test.Tasty.HUnit (Assertion, assertBool, assertEqual, assertFailure, testCase, (@?=))+import TestUtils (assertInputErrorContaining, assertReleasedAfter, writeNullRows)++-- | Check the typed registrations through SQL.+typedRegistrationTests :: TestTree+typedRegistrationTests =+ testGroup+ "typed registration"+ [ testCase "a scalar function that changes a value gives the rows of its Haskell result in a filter and a join" testScalarFilterJoin+ , testCase "replacement scan reads the rows of a query" testReplacementQuery+ , testCase "replacement scan that declines gives the unknown table error" testReplacementDecline+ , testCase "replacement scan gets the name parts as written" testReplacementNames+ , testCase "replacement scan errors fail the query" testReplacementErrors+ , testCase "connection replacement scan serves only its connection" testReplacementConnectionScope+ , testCase "instance replacement scan serves each connection" testReplacementInstance+ , testCase "replacement scans run in the order of registration" testReplacementOrder+ , testCase "query pragma with no parameter" testPragmaNoParameter+ , testCase "query pragma with a default parameter" testPragmaDefault+ , testCase "query pragma with an Integer default for each integer type" testPragmaIntegerDefault+ , testCase "pragma row has the order of the declaration" testPragmaRow+ , testCase "exec pragma counts its calls" testPragmaExec+ , testCase "pragma argument rows that do not decode" testPragmaDecodeErrors+ , testCase "pragma parameters read VARIANT in declared and ANY arguments" testPragmaVariantParameters+ , testCase "pragma exceptions become SQL errors" testPragmaExceptions+ , testCase "pragma registration checks the name before the state of a closed connection" testPragmaClosedName+ , testCase "an SQL error in each callback kind keeps its type and one prefix" testCallbackErrorTypes+ , testCase "an SQL error of each type in a callback under the syntactic invalidation policy" testCallbackErrorPolicy+ , testCase "custom option update accepts and rejects BIGINT values" testOptionUpdate+ , testCase "custom option update gets VARCHAR values and can reject RESET" testOptionText+ , testCase "table function with one parameter over several chunks" testTableSquares+ , testCase "table function with no, two, and NULL parameters" testTableParameters+ , testCase "table function columns of several types and a record row" testTableTypes+ , testCase "table function with Integer and Natural columns" testTableIntegers+ , testCase "table function with INTERVAL and TIMETZ columns" testTableIntervals+ , testCase "table function with a MAP column" testTableMaps+ , testCase "table function and aggregate with record, list, map, and union types" testTableRecords+ , testCase "a record field of an alias type declares the alias" testAliasDeclaration+ , testCase "a scalar function argument gives its type to FromField" testArgumentType+ , testCase "scalar callback metadata expires after success and failure" testArgumentTypeScope+ , testCase "table argument metadata expires before its deferred scan action" testTableArgumentTypeScope+ , testCase "typed registration rejects a type with no declaration" testNoDeclaration+ , testCase "a newtype with a sized ARRAY type text declares an ARRAY in each position" testSizedArray+ , testCase "record field outside the range of its declared type" testRecordFieldRange+ , testCase "table function runs one time for each scan" testTableRuns+ , -- A defect can make the scan run without end. Then the time limit fails+ -- the case. The limit is long, because the suite also runs under Valgrind.+ localOption (mkTimeout (10 * 60 * 1000000)) (testCase "table function reads an infinite list with LIMIT" testTableLazy)+ , testCase "table function registration and argument errors" testTableArgumentErrors+ , testCase "table function exceptions become SQL errors" testTableExceptions+ , testCase "table function evaluates only the columns that the query uses" testTableProjection+ , testCase "table function releases the list of each scan" testTableRelease+ , testCase "multi-file function gives each row of a typed table function one time" testTableMultiFile+ , testCase "COPY format with both directions under one name" testCopyBothDirections+ , testCase "COPY keeps the first registration of a format name" testCopyFirstRegistration+ , testCase "COPY TO registration checks the name before the state of a closed connection" testCopyClosedName+ , testCase "COPY errors in each function fail only their statement" testCopyErrors+ , testCase "COPY format on two connections of one database" testCopySharedDatabase+ , testCase "COPY FROM gives the name and the type of each column" testCopyFromColumns+ , testCase "aggregate over one group and over many groups" testAggregateGroups+ , testCase "aggregate with other state table stripe counts" testAggregateStripes+ , testCase "aggregate registration does not keep its connection" testAggregateConnectionRelease+ , testCase "aggregate skips NULL rows and gives the initial result" testAggregateNulls+ , testCase "aggregate arguments, states, and result types" testAggregateTypes+ , testCase "aggregate exceptions fail the query only" testAggregateErrors+ , testCase "aggregate as a window function and with modifiers" testAggregateModifiers+ , testCase "aggregate result types and custom result encoders" testAggregateResultTypes+ , testCase "aggregate native results preserve nullable values in groups and windows" testAggregateNativeResults+ , testCase "aggregate decodes VARIANT list elements across chunks" testAggregateVariant+ , testCase "aggregate states are released" testAggregateRelease+ , testGroup+ "the callback closure of each registration kind is released when its database closes"+ [testCase kind (assertReleasedAfter kind use) | (kind, use) <- closureKinds]+ ]++-- | Run the SQL of a query pragma without a parameter.+testPragmaNoParameter :: Assertion+testPragmaNoParameter = withConnection ":memory:" \conn -> do+ registerPragmaFunction conn "typed_answer" [] (QueryPragma \() -> pure "SELECT 42::BIGINT")+ query_ conn "PRAGMA typed_answer" >>= (@?= [Only (42 :: Int64)])++-- | Give the default value to the callback when the call omits the argument.+testPragmaDefault :: Assertion+testPragmaDefault = withConnection ":memory:" \conn -> do+ let amount = (parameter "amount" (TypeExpression "BIGINT")){parameterDefault = Just (defaultValue (42 :: Int64))}+ registerPragmaFunction conn "typed_amount" [amount] $+ QueryPragma \(Only (value :: Maybe Int64)) -> pure (maybe "SELECT NULL::BIGINT" (\number -> "SELECT " <> Text.pack (show number) <> "::BIGINT") value)+ query_ conn "PRAGMA typed_amount" >>= (@?= [Only (Just (42 :: Int64))])+ query_ conn "PRAGMA typed_amount(7)" >>= (@?= [Only (Just (7 :: Int64))])+ query_ conn "PRAGMA typed_amount(amount = 9)" >>= (@?= [Only (Just (9 :: Int64))])+ query_ conn "PRAGMA typed_amount(NULL)" >>= (@?= [Only (Nothing :: Maybe Int64)])++-- | Give an Integer default to parameters of several integer types. DuckDB+-- casts the default to the type of the parameter and checks the range.+testPragmaIntegerDefault :: Assertion+testPragmaIntegerDefault = withConnection ":memory:" \conn -> do+ seen <- newIORef FieldNull+ let register name dtype (value :: Integer) =+ registerPragmaFunction conn name [(parameter "amount" (TypeExpression dtype)){parameterDefault = Just (defaultValue value)}] $+ QueryPragma \(Only received) -> writeIORef seen received >> pure "SELECT 1::BIGINT"+ omitted name = do+ query_ conn (Query ("PRAGMA " <> name)) >>= (@?= [Only (1 :: Int64)])+ readIORef seen+ register "typed_default_bigint" "BIGINT" 42+ omitted "typed_default_bigint" >>= (@?= FieldInt64 42)+ register "typed_default_hugeint" "HUGEINT" 42+ omitted "typed_default_hugeint" >>= (@?= FieldHugeInt 42)+ register "typed_default_bignum" "BIGNUM" 42+ omitted "typed_default_bignum" >>= (@?= FieldBigNum (BigNum 42))+ register "typed_default_unsigned" "UBIGINT" (-5)+ assertSqlError "out of range for the destination type UINT64" (query_ conn "PRAGMA typed_default_unsigned" :: IO [Only Int64])++-- | Decode two parameters of different types as a tuple. The row has the order+-- of the declaration when the call gives the arguments by name in another order.+-- An omitted argument has its default value.+testPragmaRow :: Assertion+testPragmaRow = withConnection ":memory:" \conn -> do+ let label = (parameter "label" (TypeExpression "VARCHAR")){parameterDefault = Just (defaultValue ("none" :: Text))}+ registerPragmaFunction conn "typed_pair" [parameter "amount" (TypeExpression "BIGINT"), label] $+ QueryPragma \(amount :: Int64, text :: Text) -> pure ("SELECT " <> Text.pack (show amount) <> "::BIGINT, '" <> Text.replace "'" "''" text <> "'")+ query_ conn "PRAGMA typed_pair(3, 'liður ''x''')" >>= (@?= [(3 :: Int64, "liður 'x'" :: Text)])+ query_ conn "PRAGMA typed_pair(label = 'named', amount = 4)" >>= (@?= [(4 :: Int64, "named" :: Text)])+ query_ conn "PRAGMA typed_pair(5)" >>= (@?= [(5 :: Int64, "none" :: Text)])++-- | Run the action of an exec pragma once for each statement. The action gets+-- the argument or the default value.+testPragmaExec :: Assertion+testPragmaExec = withConnection ":memory:" \conn -> do+ calls <- newIORef (0 :: Int64)+ let step = (parameter "step" (TypeExpression "BIGINT")){parameterDefault = Just (defaultValue (1 :: Int64))}+ registerPragmaFunction conn "typed_count" [step] (ExecPragma \(Only increment) -> modifyIORef' calls (+ increment))+ void (execute_ conn "PRAGMA typed_count")+ void (execute_ conn "PRAGMA typed_count")+ readIORef calls >>= (@?= 2)+ void (execute_ conn "PRAGMA typed_count(5)")+ readIORef calls >>= (@?= 7)++-- | Name the pragma when the row has another type, another size, or a NULL.+testPragmaDecodeErrors :: Assertion+testPragmaDecodeErrors = withConnection ":memory:" \conn -> do+ let amount = parameter "amount" (TypeExpression "BIGINT")+ registerPragmaFunction+ conn+ "typed_text"+ [parameter "label" (TypeExpression "VARCHAR")]+ (QueryPragma \(Only (_ :: Int64)) -> pure "SELECT 1")+ assertSqlError+ "cannot decode the arguments of pragma typed_text: column label has type"+ (query_ conn "PRAGMA typed_text('x')" :: IO [Only Int64])+ registerPragmaFunction conn "typed_size" [amount] (QueryPragma \() -> pure "SELECT 1")+ assertSqlError "cannot decode the arguments of pragma typed_size" (query_ conn "PRAGMA typed_size(1)" :: IO [Only Int64])+ registerPragmaFunction conn "typed_strict" [amount] (ExecPragma \(Only (_ :: Int64)) -> pure ())+ assertSqlError "cannot decode the arguments of pragma typed_strict: column amount is NULL" (execute_ conn "PRAGMA typed_strict(NULL)")+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | Compare pragma arguments with the same expressions read by SELECT.+testPragmaVariantParameters :: Assertion+testPragmaVariantParameters = withConnection ":memory:" \conn -> do+ seen <- newIORef Nothing+ let check (name, declaration, expression) = do+ registerPragmaFunction conn name [parameter "value" declaration] $+ QueryPragma \(Only value) -> writeIORef seen (Just value) >> pure "SELECT 1"+ [Only expected] <- query_ conn (Query ("SELECT " <> expression)) :: IO [Only FieldValue]+ query_ conn (Query ("PRAGMA " <> name <> "(" <> expression <> ")")) >>= (@?= [Only (1 :: Int64)])+ readIORef seen >>= (@?= Just expected)+ mapM_+ check+ [ ("typed_variant_scalar", TypeExpression "VARIANT", "42::VARIANT")+ , ("typed_variant_list", TypeExpression "VARIANT[]", "[1::VARIANT, NULL]")+ , ("typed_variant_struct", TypeExpression "STRUCT(v VARIANT)", "{'v': [2,3]::VARIANT}")+ , ("typed_variant_map", TypeExpression "MAP(VARCHAR, VARIANT)", "map(['n'], [4::VARIANT])")+ , ("typed_variant_union", TypeExpression "UNION(v VARIANT)", "union_value(v := 5::VARIANT)")+ , ("typed_any_parameter", StructuredType (LogicalTypeScalar LogicalTypeIdAny), "{'a': 6}::VARIANT")+ , ("typed_geometry_parameter", TypeExpression "GEOMETRY", "'POINT (1 2)'::GEOMETRY")+ ]++-- | Raise the input error of an empty pragma name, also on a closed+-- connection. The name check comes before the read of the state of the+-- connection.+testPragmaClosedName :: Assertion+testPragmaClosedName = do+ closed <- open ":memory:"+ close closed+ assertInputErrorContaining "pragma function name must be nonempty" $+ registerPragmaFunction closed "" [] (QueryPragma \() -> pure "SELECT 1")++-- | Make an SQL error from an exception of each callback kind.+testPragmaExceptions :: Assertion+testPragmaExceptions = withConnection ":memory:" \conn -> do+ registerPragmaFunction conn "typed_query_failure" [] (QueryPragma \() -> throwIO (userError "typed query pragma failure"))+ assertSqlError "typed query pragma failure" (query_ conn "PRAGMA typed_query_failure" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ registerPragmaFunction conn "typed_exec_failure" [] (ExecPragma \() -> throwIO (userError "typed exec pragma failure"))+ assertSqlError "typed exec pragma failure" (execute_ conn "PRAGMA typed_exec_failure")+ query_ conn "SELECT 43" >>= (@?= [Only (43 :: Int64)])++-- | Give the type and the text of an error in a callback to the outer+-- statement. The failing statement runs on a connection of another database.+-- See Note [Errors that leave a callback] in "Database.DuckDB.Simple.Callback".+testCallbackErrorTypes :: Assertion+testCallbackErrorTypes = withDirectory \directory -> withConnection ":memory:" \conn -> withConnection ":memory:" \other -> withConnection ":memory:" \third -> do+ let failing = void (query_ other "SELECT 'abc'::INTEGER" :: IO [Only Int64])+ run sql = void (query_ conn sql :: IO [[FieldValue]])+ file = Text.pack (directory </> "rows.txt")+ conversion = "Conversion Error: Could not convert string 'abc' to INT32"+ -- The body of a native error is its text without the prefix.+ try failing >>= \case+ Left (err :: SQLError) -> ("Conversion Error: " <>) <$> sqlErrorBody err @?= Just (sqlErrorMessage err)+ Right () -> assertFailure "expected a Conversion Error"+ writeFile (directory </> "rows.txt") ""+ createFunction conn "error_scalar" (failing >> pure (0 :: Int64))+ createFunctionWithState conn "error_state" failing \() -> (0 :: Int64)+ createAggregate conn "error_update" sumAggregate{aggregateStep = \_ (Only (_ :: Int64)) -> failing >> pure 0}+ createAggregate conn "error_combine" sumAggregate{aggregateCombine = \_ _ -> failing >> pure 0}+ createAggregate conn "error_finalize" sumAggregate{aggregateResult = \_ -> failing >> pure (0 :: Int64)}+ createTableFunction conn "error_init" ["n"] \() -> failing >> pure [Only (1 :: Int64)]+ createTableFunction conn "error_exec" ["n"] \() -> unsafeInterleaveIO (failing >> pure [Only (1 :: Int64)])+ registerCustomType conn "error_target" (TypeExpression "BIGINT")+ R.registerCast conn (TypeExpression "BIGINT") (TypeExpression "error_target") Nothing \_ _ -> failing+ registerPragmaFunction conn "error_query_pragma" [] (QueryPragma \() -> failing >> pure "SELECT 1")+ registerPragmaFunction conn "error_exec_pragma" [] (ExecPragma \() -> failing)+ registerCustomOption conn (customOption "error_option" (TypeExpression "BIGINT")){customOptionUpdate = Just \_ -> failing}+ registerCopyToFunction conn "error_sink" (\_ -> pure ()) (\_ -> pure ()) (\_ _ -> failing) (\_ -> pure ())+ registerCopyToFunction conn "error_final" (\_ -> pure ()) (\_ -> pure ()) (\_ _ -> pure ()) (\_ -> failing)+ registerCopyFromFunction conn "error_from" \_ -> failing >> pure [Only (1 :: Int64)]+ void (execute_ conn "CREATE TABLE error_rows (value BIGINT)")+ registerReplacementScan conn \parts -> if parts == ["error_missing"] then failing >> pure Nothing else pure Nothing+ forM_+ [ ("scalar function", run "SELECT error_scalar()")+ , ("state of a scalar function", run "SELECT error_state() FROM range(3)")+ , ("aggregate update", run "SELECT error_update(i) FROM range(10) t(i)")+ , ("aggregate combine", run "SELECT error_combine(i) OVER (ORDER BY i ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING) FROM range(32) t(i)")+ , ("aggregate finalize", run "SELECT error_finalize(i) FROM range(10) t(i)")+ , ("table function init", run "SELECT * FROM error_init()")+ , ("table function exec", run "SELECT * FROM error_exec()")+ , ("query pragma", run "PRAGMA error_query_pragma")+ , ("exec pragma", void (execute_ conn "PRAGMA error_exec_pragma"))+ , ("option update", void (execute_ conn "SET error_option = 1"))+ , ("COPY TO sink", void (execute conn "COPY (SELECT 1) TO ? (FORMAT error_sink)" (Only file)))+ , ("COPY TO finalize", void (execute conn "COPY (SELECT 1) TO ? (FORMAT error_final)" (Only file)))+ , ("COPY FROM", void (execute conn "COPY error_rows FROM ? (FORMAT error_from)" (Only file)))+ , ("replacement scan", run "SELECT * FROM error_missing")+ ]+ \(label, action) -> assertErrorType label ErrorTypeConversion conversion action+ -- A cast writes the code of a failed cast and the whole message. See+ -- Note [The error code of a failed cast] in+ -- "Database.DuckDB.Simple.V2.Registration".+ assertErrorType "cast" ErrorTypeMismatch ("Mismatch Type Error: " <> conversion) (run "SELECT CAST(1::BIGINT AS error_target)::BIGINT")+ -- An error that passes through two callbacks keeps its type.+ createFunction other "error_middle" (void (query_ third "SELECT 'abc'::INTEGER" :: IO [Only Int64]) >> pure (0 :: Int64))+ createFunction conn "error_outer" (void (query_ other "SELECT error_middle()" :: IO [Only Int64]) >> pure (0 :: Int64))+ assertErrorType "two callbacks" ErrorTypeConversion conversion (run "SELECT error_outer()")+ -- An error of the library keeps its type and gets one prefix.+ createFunction+ conn+ "error_range"+ (void (query other "SELECT ?::TINYINT" (Only (2 ^ (200 :: Int) :: Integer)) :: IO [Only Int64]) >> pure (0 :: Int64))+ assertErrorType "an error of the library" ErrorTypeConversion "Conversion Error: duckdb-simple: the value " (run "SELECT error_range()")+ -- An INTERNAL error does not keep its type, because DuckDB then invalidates+ -- the database.+ createFunction+ conn+ "error_internal"+ (throwIO (SQLError "INTERNAL Error: inner" (Just ErrorRuntimeInternal) Nothing (Just "inner")) :: IO Int64)+ assertErrorType "an INTERNAL error" ErrorInputInvalid "Invalid Input Error: INTERNAL Error: inner" (run "SELECT error_internal()")+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | Raise an 'SQLError' of each type in a scalar function inside an INSERT+-- that has written rows, in an explicit transaction, under the DuckDB setting+-- SYNTACTIC_ERRORS_DO_NOT_INVALIDATE. Check the type of the outer error, the+-- rows of that INSERT after COMMIT, and that the same connection and a new+-- connection of the database still run a query. See+-- Note [Errors that leave a callback] in "Database.DuckDB.Simple.Callback".+testCallbackErrorPolicy :: Assertion+testCallbackErrorPolicy = do+ outcomes <- forM [minBound .. maxBound] \errorType -> withEnvironment \environment -> withInstance environment [] \instance' -> do+ attachDatabase instance' ":memory:" "typed_policy" []+ setDefaultDatabase instance' "typed_policy"+ (outerType, rows, same) <- withConnectionOn instance' \conn -> do+ -- With one thread, the INSERT writes two chunks of 2048 rows before the+ -- chunk with row 5000 fails.+ void (execute_ conn "SET threads = 1")+ void (execute_ conn "SET default_transaction_invalidation_policy = 'SYNTACTIC_ERRORS_DO_NOT_INVALIDATE'")+ void (execute_ conn "SET current_transaction_invalidation_policy = 'SYNTACTIC_ERRORS_DO_NOT_INVALIDATE'")+ void (execute_ conn "CREATE TABLE typed_policy_rows (v BIGINT)")+ createFunction conn "typed_policy_fail" \(i :: Int64) ->+ if i < 5000+ then pure i+ else+ throwIO+ SQLError{sqlErrorMessage = "typed failure", sqlErrorType = Just errorType, sqlErrorQuery = Nothing, sqlErrorBody = Just "typed failure"}+ :: IO Int64+ void (execute_ conn "BEGIN")+ void (execute_ conn "INSERT INTO typed_policy_rows VALUES (-1)")+ outer <- try (execute_ conn "INSERT INTO typed_policy_rows SELECT typed_policy_fail(i) FROM range(10000) t(i)")+ void (execute_ conn "COMMIT")+ rows <- query_ conn "SELECT count(*) FROM typed_policy_rows WHERE v >= 0"+ same <- query_ conn "SELECT 42"+ pure (either sqlErrorType (const Nothing) (outer :: Either SQLError Int), rows, same)+ new <- withConnectionOn instance' \other -> query_ other "SELECT 42"+ pure (errorType, outerType, rows :: [Only Int64], same :: [Only Int64], new :: [Only Int64])+ let invalidInput =+ [ ErrorQueryBinder+ , ErrorDatabaseCatalog+ , ErrorResourceConnection+ , ErrorQueryParameterNotAllowed+ , ErrorQueryParser+ , ErrorConfigurationPermission+ , ErrorResourceInUse+ , ErrorApi+ , ErrorIOFileNotFound+ , ErrorIOReadFailure+ , ErrorIOEof+ , ErrorInputParameterInvalid+ , ErrorInputMetadataDeferred+ , ErrorIONetwork+ , ErrorRuntimeInternal+ , ErrorRuntimeFatal+ ]+ expectedType errorType+ | errorType `elem` invalidInput = ErrorInputInvalid+ | otherwise = errorType+ outcomes+ @?= [ (errorType, Just (expectedType errorType), [Only 0], [Only 42], [Only 42])+ | errorType <- [minBound .. maxBound]+ ]++-- | Require an 'SQLError' of the type, with a message that starts with the+-- text.+assertErrorType :: String -> ErrorType -> Text -> IO () -> Assertion+assertErrorType label expected prefix action =+ try action >>= \case+ Left (err :: SQLError) -> do+ assertEqual label (Just expected) (sqlErrorType err)+ assertBool (label <> ": " <> show (sqlErrorMessage err)) (prefix `Text.isPrefixOf` sqlErrorMessage err)+ Right () -> assertFailure (label <> ": expected an SQL error")++-- | Accept and reject the values of a BIGINT option from SET, RESET, and+-- 'setConfigValue'. The function gets each value after the cast to BIGINT. A+-- rejected value does not change the option.+testOptionUpdate :: Assertion+testOptionUpdate = withConnection ":memory:" \conn -> do+ seen <- newIORef []+ registerCustomOption+ conn+ (customOption "typed_limit" (TypeExpression "BIGINT"))+ { customOptionDefault = Just (defaultValue (3 :: Int64))+ , customOptionUpdate = Just \value -> do+ modifyIORef' seen (value :)+ case value of+ FieldInt64 number | number < 0 -> throwIO (userError "typed_limit must not be negative")+ _ -> pure ()+ }+ let current = fst <$> getConfigValue conn "typed_limit"+ void (execute_ conn "SET typed_limit = 7")+ current >>= (@?= FieldInt64 7)+ void (execute_ conn "SET typed_limit = '5'")+ current >>= (@?= FieldInt64 5)+ assertSqlError "typed_limit must not be negative" (execute_ conn "SET typed_limit = -1")+ current >>= (@?= FieldInt64 5)+ void (execute_ conn "RESET typed_limit")+ current >>= (@?= FieldInt64 3)+ setConfigValue conn SettingScopeDefault "typed_limit" (11 :: Int64)+ current >>= (@?= FieldInt64 11)+ assertSqlError "typed_limit must not be negative" (setConfigValue conn SettingScopeDefault "typed_limit" (-2 :: Int64))+ current >>= (@?= FieldInt64 11)+ readIORef seen >>= (@?= reverse (map FieldInt64 [7, 5, -1, 3, 11, -2]))++-- | Give the values of a VARCHAR option to the function. A RESET of an option+-- without a default gives NULL. A function that rejects NULL rejects the RESET.+testOptionText :: Assertion+testOptionText = withConnection ":memory:" \conn -> do+ seen <- newIORef []+ registerCustomOption+ conn+ (customOption "typed_option_label" (TypeExpression "VARCHAR")){customOptionUpdate = Just \value -> modifyIORef' seen (value :)}+ void (execute_ conn "SET typed_option_label = 'liður'")+ void (execute_ conn "SET typed_option_label = 42")+ setConfigValue conn SettingScopeDefault "typed_option_label" ("set" :: Text)+ void (execute_ conn "RESET typed_option_label")+ readIORef seen >>= (@?= reverse [FieldText "liður", FieldText "42", FieldText "set", FieldNull])+ let required value = case value of+ FieldNull -> throwIO (userError "typed_option_required must have a value")+ _ -> pure ()+ registerCustomOption conn (customOption "typed_option_required" (TypeExpression "VARCHAR")){customOptionUpdate = Just required}+ void (execute_ conn "SET typed_option_required = 'kept'")+ assertSqlError "typed_option_required must have a value" (execute_ conn "RESET typed_option_required")+ fst <$> getConfigValue conn "typed_option_required" >>= (@?= FieldText "kept")++-- | A scalar function that changes its argument gives the rows of its Haskell+-- result in a filter, in an equality filter, and in a join condition. A+-- registered cast between two built-in types can give other rows there, so+-- the README names a scalar function for such a conversion.+testScalarFilterJoin :: Assertion+testScalarFilterJoin = withConnection ":memory:" \conn -> do+ let bend x = 100 - x * x :: Int64+ values = [-20 .. 20]+ targets = [100, 99, 64, 0, -300]+ createFunction conn "typed_bend" bend+ void (execute_ conn "CREATE TABLE bend_values AS SELECT i::BIGINT AS c FROM range(-20, 21) t(i)")+ void (execute_ conn "CREATE TABLE bend_targets AS SELECT d::BIGINT AS d FROM (VALUES (100), (99), (64), (0), (-300)) t(d)")+ query_ conn "SELECT c FROM bend_values WHERE typed_bend(c) > 50 ORDER BY c" >>= (@?= [Only c | c <- values, bend c > 50])+ query_ conn "SELECT c FROM bend_values WHERE typed_bend(c) = 100 ORDER BY c" >>= (@?= [Only c | c <- values, bend c == 100])+ query_ conn "SELECT a.c, u.d FROM bend_values a JOIN bend_targets u ON typed_bend(a.c) = u.d ORDER BY a.c"+ >>= (@?= [(c, d) | c <- values, d <- targets, bend c == d])++-- | Replace one name with a query. Read the rows under the table name and an alias.+testReplacementQuery :: Assertion+testReplacementQuery = withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TABLE typed_present AS SELECT 9::BIGINT AS amount"+ registerReplacementScan conn \parts ->+ pure+ (if parts `elem` [["typed_missing"], ["typed_present"]] then Just (ReplaceWithQuery "SELECT i AS amount FROM range(3) t(i)") else Nothing)+ query_ conn "SELECT amount FROM typed_missing ORDER BY amount" >>= (@?= map Only [0, 1, 2 :: Int64])+ query_ conn "SELECT typed_missing.amount FROM typed_missing ORDER BY 1" >>= (@?= map Only [0, 1, 2 :: Int64])+ query_ conn "SELECT x.amount FROM typed_missing AS x ORDER BY 1" >>= (@?= map Only [0, 1, 2 :: Int64])+ -- DuckDB does not call the scan for a table that the catalog has.+ query_ conn "SELECT amount FROM typed_present" >>= (@?= [Only (9 :: Int64)])++-- | Decline a name. Require the same error as on a connection with no scan.+testReplacementDecline :: Assertion+testReplacementDecline = withConnection ":memory:" \conn -> do+ before <- sqlErrorOf (query_ conn "SELECT * FROM typed_unknown" :: IO [Only Int64])+ assertBool (Text.unpack before) ("Table with name typed_unknown does not exist" `Text.isInfixOf` before)+ registerReplacementScan conn \_ -> pure Nothing+ sqlErrorOf (query_ conn "SELECT * FROM typed_unknown" :: IO [Only Int64]) >>= (@?= before)++-- | Return the parts of each name. Compare them with the query text.+testReplacementNames :: Assertion+testReplacementNames = withConnection ":memory:" \conn -> do+ registerReplacementScan conn \parts ->+ pure (Just (ReplaceWithQuery ("SELECT '" <> Text.replace "'" "''" (Text.intercalate "|" parts) <> "' AS parts")))+ let check name expected = query_ conn (Query ("SELECT parts FROM " <> name)) >>= assertEqual (Text.unpack name) [Only (Text.intercalate "|" expected)]+ check "MiXeD" ["MiXeD"]+ check "\"Quoted Name\"" ["Quoted Name"]+ check "\"has.dot\"" ["has.dot"]+ check "\"a\"\"b\"" ["a\"b"]+ -- The catalog "memory" and the schema "main" exist. The others do not.+ check "main.typed_t" ["main", "typed_t"]+ check "MAIN.Typed_T" ["MAIN", "Typed_T"]+ check "typed_schema.typed_t" ["typed_schema", "typed_t"]+ check "memory.main.typed_t" ["memory", "main", "typed_t"]+ check "typed_catalog.main.typed_t" ["typed_catalog", "main", "typed_t"]+ check "'typed file.csv'" ["typed file.csv"]+ check "'typed''quote.parquet'" ["typed'quote.parquet"]++-- | Fail the query with an exception and with query text that is not one SELECT.+testReplacementErrors :: Assertion+testReplacementErrors = withConnection ":memory:" \conn -> do+ registerReplacementScan conn \case+ ["typed_reserved"] -> ioError (userError "typed_reserved: the table name is reserved")+ ["typed_two"] -> pure (Just (ReplaceWithQuery "SELECT 1; SELECT 2"))+ ["typed_create"] -> pure (Just (ReplaceWithQuery "CREATE TABLE typed_created (i BIGINT)"))+ _ -> pure Nothing+ assertSqlError "typed_reserved: the table name is reserved" (query_ conn "SELECT * FROM typed_reserved" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ assertSqlError "exactly one SELECT statement" (query_ conn "SELECT * FROM typed_two" :: IO [Only Int64])+ assertSqlError "must be a SELECT statement" (query_ conn "SELECT * FROM typed_create" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | Register a scan on one connection of an instance. The other connection does+-- not use it. The scan replaces a file path before the built-in file scan.+testReplacementConnectionScope :: Assertion+testReplacementConnectionScope = withEnvironment \environment -> withInstance environment [] \instance' ->+ withConnectionOn instance' \first -> withConnectionOn instance' \second -> do+ registerReplacementScan first \parts ->+ pure (if parts `elem` [["typed_scoped"], ["typed_missing.csv"]] then Just (ReplaceWithQuery "SELECT 'first' AS source") else Nothing)+ query_ first "SELECT source FROM typed_scoped" >>= (@?= [Only ("first" :: Text)])+ assertSqlError "Table with name typed_scoped does not exist" (query_ second "SELECT source FROM typed_scoped" :: IO [Only Text])+ query_ first "SELECT source FROM 'typed_missing.csv'" >>= (@?= [Only ("first" :: Text)])+ assertSqlError "No files found" (query_ second "SELECT source FROM 'typed_missing.csv'" :: IO [Only Text])++-- | Register a scan on an instance before its connections. Each connection uses+-- it. A connection scan comes first and passes a declined name to it.+testReplacementInstance :: Assertion+testReplacementInstance = withEnvironment \environment -> withInstance environment [] \instance' -> do+ registerInstanceReplacementScan instance' \parts -> pure (if parts `elem` [["typed_shared"], ["typed_instance"]] then Just (ReplaceWithQuery "SELECT 'instance' AS source") else Nothing)+ withConnectionOn instance' \first -> withConnectionOn instance' \second -> do+ query_ first "SELECT source FROM typed_shared" >>= (@?= [Only ("instance" :: Text)])+ query_ second "SELECT source FROM typed_shared" >>= (@?= [Only ("instance" :: Text)])+ registerReplacementScan first \parts -> pure (if parts == ["typed_shared"] then Just (ReplaceWithQuery "SELECT 'connection' AS source") else Nothing)+ query_ first "SELECT source FROM typed_shared" >>= (@?= [Only ("connection" :: Text)])+ query_ first "SELECT source FROM typed_instance" >>= (@?= [Only ("instance" :: Text)])+ query_ second "SELECT source FROM typed_shared" >>= (@?= [Only ("instance" :: Text)])++-- | Register two scans on one connection. The first scan that replaces a name wins.+testReplacementOrder :: Assertion+testReplacementOrder = withConnection ":memory:" \conn -> do+ registerReplacementScan conn \parts -> pure (if parts == ["typed_both"] then Just (ReplaceWithQuery "SELECT 'first' AS source") else Nothing)+ registerReplacementScan conn \parts -> pure (if parts `elem` [["typed_both"], ["typed_later"]] then Just (ReplaceWithQuery "SELECT 'second' AS source") else Nothing)+ query_ conn "SELECT source FROM typed_both" >>= (@?= [Only ("first" :: Text)])+ query_ conn "SELECT source FROM typed_later" >>= (@?= [Only ("second" :: Text)])++-- | Require an 'SQLError' and return its message.+sqlErrorOf :: IO a -> IO Text+sqlErrorOf action =+ try action >>= \case+ Left (err :: SQLError) -> pure (sqlErrorMessage err)+ Right _ -> assertFailure "expected an SQL error"++-- | Add BIGINT values.+sumAggregate :: Aggregate Int64 (Only Int64) Int64+sumAggregate =+ Aggregate+ { aggregateInitial = 0+ , aggregateStep = \total (Only value) -> pure (total + value)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }++-- | Compare the two columns of each row.+assertColumnsEqual :: Connection -> Query -> Assertion+assertColumnsEqual conn sql = do+ rows <- query_ conn sql+ assertBool "rows" (not (null rows))+ forM_ rows \(typed, builtin) -> assertEqual (Text.unpack (fromQuery sql)) (builtin :: Maybe Int64) typed++-- | Fold one group and many groups, also with partial states from 4 threads.+testAggregateGroups :: Assertion+testAggregateGroups = withConnection ":memory:" \conn -> do+ combines <- newIORef (0 :: Int)+ createAggregate+ conn+ "typed_sum"+ sumAggregate{aggregateCombine = \a b -> atomicModifyIORef' combines (\count -> (count + 1, ())) >> pure (a + b)}+ query_ conn "SELECT typed_sum(i) FROM range(10) t(i)" >>= (@?= [Only (45 :: Int64)])+ -- A chunk has at most 2048 rows. Thus 5000 rows need more than one update call.+ assertColumnsEqual conn "SELECT typed_sum(i), sum(i)::BIGINT FROM range(5000) t(i) GROUP BY i % 50"+ query_ conn "SELECT count(*) FROM (SELECT i % 50 AS g, typed_sum(i) FROM range(5000) t(i) GROUP BY g)" >>= (@?= [Only (50 :: Int64)])+ -- A table of 3 row groups lets 4 threads make partial states. DuckDB then combines them.+ void (execute_ conn "SET threads = 4")+ void (execute_ conn "CREATE TABLE typed_numbers AS SELECT i FROM range(300000) t(i)")+ assertColumnsEqual conn "SELECT typed_sum(i), sum(i)::BIGINT FROM typed_numbers GROUP BY i % 1000"+ -- An ungrouped aggregate combines the state of each thread into one state.+ -- Thus the combine function runs at least one time.+ writeIORef combines 0+ assertColumnsEqual conn "SELECT typed_sum(i), sum(i)::BIGINT FROM typed_numbers"+ readIORef combines >>= \count -> assertBool ("combine calls: " <> show count) (count > 0)+ query_ conn "SELECT typed_sum(i), typed_sum(i + 1) FROM range(10) t(i)" >>= (@?= [(45 :: Int64, 55 :: Int64)])++-- | One stripe gives the same results as the default stripes, also with+-- partial states from 4 threads. More stripes than states give the same+-- results. A connection with another stripe count calls the aggregate of a+-- first connection. A stripe count of 0 fails the open.+testAggregateStripes :: Assertion+testAggregateStripes = do+ withConnectionWithOptions ":memory:" defaultConnectionOptions{connectionAggregateStripes = 1} \conn -> do+ createAggregate conn "typed_sum" sumAggregate+ void (execute_ conn "SET threads = 4")+ void (execute_ conn "CREATE TABLE typed_numbers AS SELECT i FROM range(300000) t(i)")+ assertColumnsEqual conn "SELECT typed_sum(i), sum(i)::BIGINT FROM typed_numbers GROUP BY i % 1000"+ assertColumnsEqual conn "SELECT typed_sum(i), sum(i)::BIGINT FROM typed_numbers"+ withEnvironment \environment -> withInstance environment [] \instance' ->+ bracket (connectWithOptions instance' defaultConnectionOptions{connectionAggregateStripes = 1000}) close \first ->+ bracket (connectWithOptions instance' defaultConnectionOptions{connectionAggregateStripes = 1}) close \second -> do+ createAggregate first "typed_wide_sum" sumAggregate+ forM_ [first, second] \conn -> do+ assertColumnsEqual conn "SELECT typed_wide_sum(i), sum(i)::BIGINT FROM range(10) t(i) GROUP BY i % 3"+ assertColumnsEqual conn "SELECT typed_wide_sum(i), sum(i)::BIGINT FROM range(10) t(i)"+ try (withConnectionWithOptions ":memory:" defaultConnectionOptions{connectionAggregateStripes = 0} (const (pure ()))) >>= \case+ Left (err :: SQLError) -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ assertBool ("unexpected message: " <> show err) ("connectionAggregateStripes" `Text.isInfixOf` sqlErrorMessage err)+ Right () -> assertFailure "a stripe count of 0 opened a connection"++-- | The callbacks of an aggregate do not keep the connection that registered+-- it. A connection with no close and no query becomes garbage, and its+-- finalizer closes it. The aggregate still runs on another connection of the+-- database.+testAggregateConnectionRelease :: Assertion+testAggregateConnectionRelease = withEnvironment \environment -> withInstance environment [] \instance' -> do+ weak <- do+ conn <- connect instance'+ createAggregate conn "typed_dropped_sum" sumAggregate{aggregateResult = pure . Variant . FieldInt64}+ mkWeakIORef (connectionState conn) (pure ())+ performMajorGC+ performMajorGC+ deRefWeak weak >>= assertBool "the aggregate keeps the connection that registered it" . isNothing+ withConnectionOn instance' \other ->+ query_ other "SELECT typed_dropped_sum(i)::BIGINT FROM range(10) t(i)" >>= (@?= [Only (45 :: Int64)])++-- | Skip the rows with NULL. A group without other rows gives the initial result.+testAggregateNulls :: Assertion+testAggregateNulls = withConnection ":memory:" \conn -> do+ createAggregate conn "typed_sum" sumAggregate+ query_ conn "SELECT typed_sum(v) FROM (VALUES (1::BIGINT), (NULL), (2)) t(v)" >>= (@?= [Only (3 :: Int64)])+ -- The built-in sum gives NULL for group 2. The initial state gives 0.+ query_ conn "SELECT g, typed_sum(v) FROM (VALUES (1::BIGINT, 10::BIGINT), (1, NULL), (2, NULL)) t(g, v) GROUP BY g ORDER BY g"+ >>= (@?= [(1 :: Int64, 10 :: Int64), (2, 0)])+ query_ conn "SELECT typed_sum(i) FROM range(0) t(i)" >>= (@?= [Only (0 :: Int64)])+ query_ conn "SELECT typed_sum(NULL::BIGINT)" >>= (@?= [Only (0 :: Int64)])+ query_ conn "SELECT count(*) FROM (SELECT typed_sum(i) FROM range(0) t(i) GROUP BY i)" >>= (@?= [Only (0 :: Int64)])++-- | Count the words of each group in a 'Map'. The result lists each word and+-- its count, and an empty map gives NULL.+tallyAggregate :: Aggregate (Map.Map Text Int64) (Text, Int64) (Maybe Text)+tallyAggregate =+ Aggregate+ { aggregateInitial = Map.empty+ , aggregateStep = \tally (word, count) -> pure (Map.insertWith (+) word count tally)+ , aggregateCombine = \a b -> pure (Map.unionWith (+) a b)+ , aggregateResult = \tally ->+ pure+ if Map.null tally then Nothing else Just (Text.intercalate "," [word <> "=" <> Text.pack (show count) | (word, count) <- Map.toList tally])+ }++-- | Use a tuple of two argument types, a 'Map' state, and each result type.+testAggregateTypes :: Assertion+testAggregateTypes = withConnection ":memory:" \conn -> do+ createAggregate conn "typed_tally" tallyAggregate+ -- Group 2 has the 12 bytes that fit inline. Group 3 has 13 bytes. Group 4 has 14+ -- bytes in 8 characters. Group 5 has NULL in each row.+ _ <-+ execute_+ conn+ "CREATE TABLE typed_words AS SELECT * FROM (VALUES (1, 'a', 1::BIGINT), (1, 'a', 2), (1, NULL, 5), (1, 'b', NULL), (2, 'abcdefghij', 1), (3, 'abcdefghijk', 1), (4, '\240\240\240\240\240\240', 1), (5, NULL, 1), (5, 'c', NULL), (6, 'cherry', 2), (6, 'apple', 3), (6, 'banana', 5)) t(g, w, n)"+ let expected =+ [ Just "a=3"+ , Just "abcdefghij=1"+ , Just "abcdefghijk=1"+ , Just "\240\240\240\240\240\240=1"+ , Nothing+ , Just "apple=3,banana=5,cherry=2" :: Maybe Text+ ]+ query_ conn "SELECT typed_tally(w, n) FROM typed_words GROUP BY g ORDER BY g" >>= (@?= map Only expected)+ -- DuckDB compares the bytes of the results with the bytes of the literals.+ query_+ conn+ "SELECT typed_tally(w, n) = 'abcdefghij=1', typed_tally(w, n) = 'abcdefghijk=1' FROM typed_words WHERE g IN (2, 3) GROUP BY g ORDER BY g"+ >>= (@?= [(Just True, Just False), (Just False, Just True :: Maybe Bool)])+ query_ conn "SELECT typed_tally(w, n) FROM typed_words"+ >>= (@?= [Only (Just "a=3,abcdefghij=1,abcdefghijk=1,apple=3,banana=5,cherry=2,\240\240\240\240\240\240=1" :: Maybe Text)])+ query_ conn "SELECT typed_tally(w, n) FROM typed_words WHERE g = 5" >>= (@?= [Only (Nothing :: Maybe Text)])+ query_+ conn+ "SELECT typed_tally(w, n) OVER (ORDER BY g ROWS BETWEEN CURRENT ROW AND 1 FOLLOWING) FROM (VALUES (1, 'abcdefghij', 1::BIGINT), (2, 'abcdefghijk', 1), (3, NULL, 1)) t(g, w, n) ORDER BY g"+ >>= (@?= map Only [Just "abcdefghij=1,abcdefghijk=1", Just "abcdefghijk=1", Nothing :: Maybe Text])+ createAggregate+ conn+ "typed_any_odd"+ Aggregate+ { aggregateInitial = False+ , aggregateStep = \seen (Only (value :: Int64)) -> pure (seen || odd value)+ , aggregateCombine = \a b -> pure (a || b)+ , aggregateResult = pure+ }+ createAggregate+ conn+ "typed_mean"+ Aggregate+ { aggregateInitial = (0, 0)+ , aggregateStep = \(total, count) (Only value) -> pure (total + value, count + 1 :: Int64)+ , aggregateCombine = \(t1, c1) (t2, c2) -> pure (t1 + t2, c1 + c2)+ , aggregateResult = \(total, count) -> pure (if count == 0 then Nothing else Just (total / fromIntegral count :: Double))+ }+ createAggregate+ conn+ "typed_count"+ Aggregate+ { aggregateInitial = 0+ , aggregateStep = \count (Only (_ :: Int32)) -> pure (count + 1)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure :: Word64 -> IO Word64+ }+ createAggregate+ conn+ "typed_small"+ Aggregate+ { aggregateInitial = 0+ , aggregateStep = \total (Only value) -> pure (total + value)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure :: Int32 -> IO Int32+ }+ createAggregate+ conn+ "typed_label"+ Aggregate+ { aggregateInitial = 0 :: Int+ , aggregateStep = \count (Only (_ :: Text)) -> pure (count + 1)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = \count -> pure (Text.replicate count "ab")+ }+ query_+ conn+ "SELECT typed_any_odd(i), typed_mean(i::DOUBLE), typed_count(i::INTEGER), typed_small(i::INTEGER), typed_label(i::VARCHAR) FROM range(8) t(i) GROUP BY i < 1 ORDER BY i < 1"+ >>= (@?= [(True, Just (4 :: Double), 7 :: Word64, 28 :: Int32, "ababababababab" :: Text), (False, Just 0, 1, 0, "ab")])+ query_ conn "SELECT typed_any_odd(i), typed_mean(i::DOUBLE), typed_count(i::INTEGER), typed_label(i::VARCHAR) FROM range(0) t(i)"+ >>= (@?= [(False, Nothing :: Maybe Double, 0 :: Word64, "" :: Text)])+ -- An Int32 argument and an Int32 result are INTEGER.+ query_+ conn+ "SELECT typeof(typed_any_odd(1)), typeof(typed_mean(1)), typeof(typed_count(1)), typeof(typed_small(1)), typeof(typed_label('x'))"+ >>= (@?= [("BOOLEAN" :: Text, "DOUBLE" :: Text, "UBIGINT" :: Text, "INTEGER" :: Text, "VARCHAR" :: Text)])+ query_+ conn+ "SELECT function_name, parameters::VARCHAR, parameter_types::VARCHAR FROM duckdb_functions() WHERE function_name IN ('typed_small', 'typed_tally') ORDER BY function_name"+ >>= (@?= [("typed_small" :: Text, "[arg0]" :: Text, "[INTEGER]" :: Text), ("typed_tally", "[arg0, arg1]", "[VARCHAR, BIGINT]")])++-- | Fail the query for an exception in each function. The next query works.+testAggregateErrors :: Assertion+testAggregateErrors = withConnection ":memory:" \conn -> do+ createAggregate+ conn+ "typed_bad_step"+ sumAggregate{aggregateStep = \total (Only value) -> if value == 13 then throwIO (userError "typed_bad_step: 13") else pure (total + value)}+ assertSqlError "typed_bad_step: 13" (query_ conn "SELECT typed_bad_step(i) FROM range(20) t(i) GROUP BY i % 3" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ -- The step evaluates its state, so the error belongs to the step.+ createAggregate+ conn+ "typed_lazy_step"+ sumAggregate{aggregateStep = \total (Only value) -> pure (if value == 7 then error "typed_lazy_step: 7" else total + value)}+ assertSqlError "typed_lazy_step: 7" (query_ conn "SELECT typed_lazy_step(i) FROM range(20) t(i)" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ -- DuckDB adds rows 0 to 15 and rows 16 to 31 of the frame of row 0 in separate+ -- states. Then it combines them.+ createAggregate conn "typed_bad_combine" sumAggregate{aggregateCombine = \_ _ -> throwIO (userError "typed_bad_combine: combine")}+ assertSqlError+ "typed_bad_combine: combine"+ ( query_ conn "SELECT typed_bad_combine(i) OVER (ORDER BY i ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING) FROM range(32) t(i)"+ :: IO [Only Int64]+ )+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ -- A Natural argument is UHUGEINT. DuckDB casts a BIGINT to it only with an+ -- explicit cast, and the cast of a negative value fails.+ createAggregate+ conn+ "typed_natural_sum"+ Aggregate+ { aggregateInitial = 0+ , aggregateStep = \total (value :: Int64, extra :: Natural) -> pure (total + value + fromIntegral extra)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure :: Int64 -> IO Int64+ }+ query_ conn "SELECT typed_natural_sum(i, i::UHUGEINT) FROM range(10) t(i)" >>= (@?= [Only (90 :: Int64)])+ assertSqlError "No function matches" (query_ conn "SELECT typed_natural_sum(i, i) FROM range(10) t(i)" :: IO [Only Int64])+ assertSqlError+ "out of range for the destination type UINT128"+ (query_ conn "SELECT typed_natural_sum(i, (i - 5)::UHUGEINT) FROM range(10) t(i)" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ createAggregate+ conn+ "typed_bad_result"+ sumAggregate{aggregateResult = \total -> if total > 100 then throwIO (userError "typed_bad_result: too large") else pure total}+ query_ conn "SELECT typed_bad_result(i) FROM range(10) t(i)" >>= (@?= [Only (45 :: Int64)])+ assertSqlError+ "typed_bad_result: too large"+ (query_ conn "SELECT typed_bad_result(i) FROM range(100) t(i) GROUP BY i % 2" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ createAggregate+ conn+ "typed_lazy_result"+ sumAggregate{aggregateResult = \total -> pure (if total == 0 then error "typed_lazy_result: 0" else total)}+ assertSqlError "typed_lazy_result: 0" (query_ conn "SELECT typed_lazy_result(i) FROM range(1) t(i)" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | Declare the true type of a SMALLINT and a FLOAT result. Write them also in+-- an ordered window, where the result vector of a finalize call has the size 0.+-- Use the native value encoder for results without a raw writer.+testAggregateResultTypes :: Assertion+testAggregateResultTypes = withConnection ":memory:" \conn -> do+ createAggregate+ conn+ "typed_small_count"+ Aggregate+ { aggregateInitial = 0 :: Int16+ , aggregateStep = \count (Only (_ :: Int64)) -> pure (count + 1)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }+ createAggregate+ conn+ "typed_float_mean"+ Aggregate+ { aggregateInitial = (0, 0 :: Int)+ , aggregateStep = \(total, count) (Only value) -> pure (total + value, count + 1)+ , aggregateCombine = \(t1, c1) (t2, c2) -> pure (t1 + t2, c1 + c2)+ , aggregateResult = \(total, count) -> pure (if count == 0 then Nothing else Just (realToFrac (total / fromIntegral count :: Double) :: Float))+ }+ query_ conn "SELECT typeof(typed_small_count(1)), typeof(typed_float_mean(1.0::DOUBLE))" >>= (@?= [("SMALLINT" :: Text, "FLOAT" :: Text)])+ query_ conn "SELECT min(c), max(c), count(*) FROM (SELECT typed_small_count(i) AS c FROM range(100000) t(i) GROUP BY i % 5000)"+ >>= (@?= [(20 :: Int16, 20 :: Int16, 5000 :: Int64)])+ query_+ conn+ "SELECT min(c), max(c), count(*) FROM (SELECT typed_small_count(i) OVER (ORDER BY i ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) AS c FROM range(5000) t(i))"+ >>= (@?= [(1 :: Int16, 3 :: Int16, 5000 :: Int64)])+ query_+ conn+ "SELECT bool_and(m = CASE WHEN i = 0 THEN 0 ELSE i - 0.5 END), typeof(any_value(m)) FROM (SELECT i, typed_float_mean(i::DOUBLE) OVER (ORDER BY i ROWS BETWEEN 1 PRECEDING AND CURRENT ROW) AS m FROM range(5000) t(i))"+ >>= (@?= [(True, "FLOAT" :: Text)])+ -- A newtype over a raw type with a derived FunctionResult has the raw writer.+ createAggregate+ conn+ "typed_oldest"+ Aggregate+ { aggregateInitial = Age 0+ , aggregateStep = \oldest (Only value) -> pure (max oldest (Age value))+ , aggregateCombine = \a b -> pure (max a b)+ , aggregateResult = pure+ }+ query_ conn "SELECT typeof(typed_oldest(1::SMALLINT)), typed_oldest(i::SMALLINT) FROM range(10) t(i)"+ >>= (@?= [("SMALLINT" :: Text, Age 9)])+ createFunction conn "typed_age" (Age 5)+ query_ conn "SELECT typeof(typed_age()), typed_age()" >>= (@?= [("SMALLINT" :: Text, Age 5)])+ createAggregate conn "typed_years" sumAggregate{aggregateResult = pure . Years . fromIntegral}+ query_ conn "SELECT typeof(typed_years(i)), typed_years(i) FROM range(10) t(i)"+ >>= (@?= [("SMALLINT" :: Text, 45 :: Int16)])+ createAggregate conn "typed_peak_result" sumAggregate{aggregateResult = \total -> pure (Peak "total" (fromIntegral total) "")}+ query_ conn "SELECT typed_peak_result(i) FROM range(10) t(i)"+ >>= (@?= [Only (Peak "total" 45 "")])+ createAggregate conn "typed_integer_result" sumAggregate{aggregateResult = pure . toInteger}+ query_ conn "SELECT typed_integer_result(i) FROM range(10) t(i)"+ >>= (@?= [Only (45 :: Integer)])+ registerCustomType conn "aggregate_total" (TypeExpression "BIGINT")+ createAggregate conn "typed_alias_result" sumAggregate{aggregateResult = pure . AggregateTotal}+ query_ conn "SELECT typeof(typed_alias_result(i)), typed_alias_result(i)::BIGINT FROM range(10) t(i)"+ >>= (@?= [("aggregate_total" :: Text, 45 :: Int64)])+ createAggregate conn "typed_catalog_encoder" sumAggregate{aggregateResult = pure . CatalogTotal}+ outcome <- try (query_ conn "SELECT typed_catalog_encoder(i) FROM range(10) t(i)" :: IO [Only Int64])+ case outcome of+ Left (err :: SQLError) ->+ assertBool (show err) ("aggregate_total" `Text.isInfixOf` Text.toLower (sqlErrorMessage err))+ Right rows -> assertFailure ("catalog lookup in an aggregate encoder succeeded: " <> show rows)+ query_ conn "SELECT typed_alias_result(i)::BIGINT FROM range(10) t(i)" >>= (@?= [Only (45 :: Int64)])+ assertSqlError+ "typed_struct_result: result: StructValue has no static field types"+ (createAggregate conn "typed_struct_result" sumAggregate{aggregateResult = \_ -> pure (undefined :: StructValue FieldValue)})+ assertInputErrorContaining "aggregate function name must be nonempty" (createAggregate conn "" sumAggregate)++-- | Check native values across chunks and finalize calls with result offsets.+-- Compare geometry bytes in SQL so that output conversion cannot hide changes.+testAggregateNativeResults :: Assertion+testAggregateNativeResults = withConnection ":memory:" \conn -> do+ let nullable n value = if n `mod` 7 == 0 then Nothing else Just value+ createAggregate conn "typed_native_variant" countAggregate{aggregateResult = \n -> pure (nullable n (Variant (FieldInt64 n)))}+ createAggregate+ conn+ "typed_native_geometry"+ countAggregate{aggregateResult = \n -> pure (nullable n (G.PointGeometry (G.PointXY (G.XY (fromIntegral n) 0))))}+ createAggregate conn "typed_native_nested" countAggregate{aggregateResult = \n -> pure [Just (Variant (FieldInt64 n)), Nothing]}+ forM_+ [ ("typed_native_variant", "q::BIGINT IS DISTINCT FROM e")+ , ("typed_native_geometry", "q::BLOB IS DISTINCT FROM (('POINT (' || e || ' 0)')::GEOMETRY)::BLOB")+ ]+ \(name, differs) ->+ forM_+ [ (name <> "(i)", "GROUP BY i % 3001", "count(*)", 3001)+ , (name <> "(i ORDER BY i DESC)", "GROUP BY i % 13", "count(*)", 13)+ , (name <> "(i) OVER w", "WINDOW w AS (PARTITION BY i % 3 ORDER BY i ROWS BETWEEN 9 PRECEDING AND CURRENT ROW)", "count(*) OVER w", 8192)+ ]+ \(call, suffix, expected, rows) -> do+ let sql =+ "SELECT count(*)::BIGINT, count(*) FILTER (WHERE "+ <> differs+ <> ")::BIGINT FROM (SELECT q, CASE WHEN n % 7 <> 0 THEN n END AS e FROM (SELECT "+ <> call+ <> " AS q, "+ <> expected+ <> " AS n FROM range(8192) t(i) "+ <> suffix+ <> "))"+ query_ conn (Query sql) >>= (@?= [(rows :: Int64, 0 :: Int64)])+ query_ conn "SELECT typed_native_nested(i) FROM range(10) t(i)"+ >>= (@?= [Only [Just (Variant (FieldInt64 10)), Nothing]])+ query_ conn "SELECT typed_native_variant(i) FROM range(0) t(i)"+ >>= (@?= [Only (Nothing :: Maybe Variant)])+ createAggregate+ conn+ "typed_native_order"+ Aggregate+ { aggregateInitial = []+ , aggregateStep = \seen (Only (value :: Int64)) -> pure (value : seen)+ , aggregateCombine = \a b -> pure (b <> a)+ , aggregateResult = pure . reverse+ }+ query_+ conn+ "SELECT bool_and(actual = expected) FROM (SELECT typed_native_order(i ORDER BY i DESC) AS actual, list(i ORDER BY i DESC) AS expected FROM range(8192) t(i) GROUP BY i % 13)"+ >>= (@?= [Only True])++-- | Read each BIGINT and NULL from a VARIANT list in aggregate update.+testAggregateVariant :: Assertion+testAggregateVariant = withEnvironment \environment -> withInstance environment [("threads", "4")] \instance' -> do+ let step total (Only values) = case values of+ [Just (Variant (FieldInt64 number)), Nothing] -> pure (total + number)+ _ -> fail ("unexpected VARIANT list: " <> show values)+ withConnectionOn instance' \registering -> createAggregate registering "variant_sum" countAggregate{aggregateStep = step}+ withConnectionOn instance' \other ->+ query_ other "SELECT variant_sum([i::VARIANT, NULL::VARIANT]), sum(i)::BIGINT FROM range(5000) t(i)"+ >>= (@?= [(12497500 :: Int64, 12497500 :: Int64)])++-- | An aggregate that counts its rows. Change 'aggregateStep' for another+-- argument type.+countAggregate :: Aggregate Int64 (Only Int64) Int64+countAggregate =+ Aggregate+ { aggregateInitial = 0+ , aggregateStep = \count _ -> pure (count + 1)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }++-- | Compare window, DISTINCT, and FILTER uses with the built-in sum. Use ORDER BY in the call.+testAggregateModifiers :: Assertion+testAggregateModifiers = withConnection ":memory:" \conn -> do+ createAggregate conn "typed_sum" sumAggregate+ forM_+ [ "ROWS BETWEEN 2 PRECEDING AND CURRENT ROW"+ , "ROWS UNBOUNDED PRECEDING"+ , "ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING"+ , "PARTITION BY i % 3"+ , "PARTITION BY i % 3 ORDER BY i"+ , ""+ ]+ \frame -> do+ let window = if Text.null frame || "PARTITION" `Text.isPrefixOf` frame then frame else "ORDER BY i " <> frame+ -- The built-in sum gives NULL for a frame without values. The initial state gives 0.+ assertColumnsEqual+ conn+ ( Query+ ( "SELECT typed_sum(v) OVER ("+ <> window+ <> "), coalesce(sum(v) OVER ("+ <> window+ <> "), 0)::BIGINT FROM (SELECT i, CASE WHEN i % 7 = 0 THEN NULL ELSE i END AS v FROM range(100) t(i))"+ )+ )+ assertColumnsEqual conn "SELECT typed_sum(DISTINCT i % 5), sum(DISTINCT i % 5)::BIGINT FROM range(100) t(i)"+ assertColumnsEqual conn "SELECT typed_sum(DISTINCT i % 5), sum(DISTINCT i % 5)::BIGINT FROM range(100) t(i) GROUP BY i % 2"+ assertColumnsEqual conn "SELECT typed_sum(i) FILTER (WHERE i % 2 = 0), sum(i) FILTER (WHERE i % 2 = 0)::BIGINT FROM range(10) t(i)"+ createAggregate+ conn+ "typed_words"+ Aggregate+ { aggregateInitial = []+ , aggregateStep = \seen (Only word) -> pure (word : seen)+ , aggregateCombine = \a b -> pure (b <> a)+ , aggregateResult = pure . Text.intercalate "," . reverse+ }+ query_ conn "SELECT typed_words(w ORDER BY k DESC) FROM (VALUES (1, 'a'), (2, 'b'), (3, 'c')) t(k, w)" >>= (@?= [Only ("c,b,a" :: Text)])++-- | Release each Haskell state after a query, also after a failed query and+-- after a query that stops early. The step makes a new 'IORef' for each state and+-- keeps a weak pointer to it.+testAggregateRelease :: Assertion+testAggregateRelease = withConnection ":memory:" \conn -> do+ cells <- newIORef []+ failure <- newIORef ""+ let newCell value = do+ cell <- newIORef value+ weak <- mkWeakIORef cell (pure ())+ atomicModifyIORef' cells \weaks -> (weak : weaks, ())+ pure (Just cell)+ failIn phase = readIORef failure >>= \current -> if current == phase then throwIO (userError ("typed_release: " <> phase)) else pure ()+ step state (Only value) = do+ failIn "step"+ maybe (newCell value) (\cell -> Just cell <$ atomicModifyIORef' cell \total -> (total + value, ())) state+ combine a b = do+ failIn "combine"+ totals <- mapM readIORef [cell | Just cell <- [a, b]]+ if null totals then pure Nothing else newCell (sum totals)+ result state = do+ failIn "result"+ value <- maybe (pure 0) readIORef state+ phase <- readIORef failure+ pure (EncodedTotal value (phase == "encode" && value == 3))+ run phase sql = do+ writeIORef failure phase+ outcome <- try (query_ conn sql :: IO [Only Int64])+ case (phase, outcome) of+ ("", Left err) -> assertFailure (show err)+ ("", Right _) -> pure ()+ (_, Left (err :: SQLError)) ->+ assertBool (Text.unpack (sqlErrorMessage err)) (Text.pack ("typed_release: " <> phase) `Text.isInfixOf` sqlErrorMessage err)+ (_, Right _) -> assertFailure ("expected an error in " <> Text.unpack (fromQuery sql))+ createAggregate+ conn+ "typed_release"+ Aggregate+ { aggregateInitial = Nothing+ , aggregateStep = step+ , aggregateCombine = combine+ , aggregateResult = result :: Maybe (IORef Int64) -> IO EncodedTotal+ }+ run "" "SELECT typed_release(i) FROM range(5000) t(i) GROUP BY i % 50"+ run "" "SELECT typed_release(i) OVER (ORDER BY i ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) FROM range(100) t(i)"+ run "" "SELECT typed_release(DISTINCT i % 5) + typed_release(i) FROM range(100) t(i)"+ -- A prepared statement keeps its bind data over its runs. DuckDB keeps the+ -- states of the last run until the statement runs again or closes.+ withStatement conn "SELECT typed_release(i) FROM range(10) t(i)" \statement -> do+ nextRow statement >>= (@?= Just (Only (45 :: Int64)))+ void (executeStatement statement)+ earlier <- readIORef cells+ nextRow statement >>= (@?= Just (Only (45 :: Int64)))+ assertReleased "the earlier runs of a prepared statement" earlier+ run "step" "SELECT typed_release(i) FROM range(5000) t(i) GROUP BY i % 50"+ run "combine" "SELECT typed_release(i) OVER (ORDER BY i ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING) FROM range(32) t(i)"+ -- DuckDB does not destroy these states. The release of the bind data releases them.+ run "result" "SELECT typed_release(i) FROM range(20000) t(i) GROUP BY i % 10000"+ run "result" "SELECT typed_release(i) OVER (ORDER BY i ROWS BETWEEN 2 PRECEDING AND CURRENT ROW) FROM range(100) t(i)"+ -- Earlier rows have already been written when DuckDB fails to cast row 3.+ run "encode" "SELECT typed_release(i) OVER (ORDER BY i ROWS BETWEEN CURRENT ROW AND CURRENT ROW) FROM range(10) t(i)"+ run "" "SELECT typed_release(i) FROM range(10) t(i)"+ void (execute_ conn "SET threads = 4")+ void (execute_ conn "CREATE TABLE typed_numbers AS SELECT i FROM range(300000) t(i)")+ run "" "SELECT typed_release(i) FROM range(20000) t(i) GROUP BY i % 10000 LIMIT 1"+ run "" "SELECT typed_release(i) FROM typed_numbers GROUP BY i % 5000 LIMIT 1"+ run "combine" "SELECT typed_release(i) FROM typed_numbers GROUP BY i % 50"+ run "result" "SELECT typed_release(i) FROM typed_numbers GROUP BY i % 5000"+ readIORef cells >>= assertReleased "the queries"++-- | Require that no weak pointer of the list is alive after a major collection.+assertReleased :: String -> [Weak (IORef Int64)] -> Assertion+assertReleased label weaks = do+ assertBool ("states made in " <> label) (not (null weaks))+ performMajorGC+ performMajorGC+ alive <- length . filter isJust <$> mapM deRefWeak weaks+ assertEqual ("live states after " <> label) 0 alive++-- | Require an 'SQLError' with a message that contains the text.+assertSqlError :: Text -> IO a -> Assertion+assertSqlError expected action = do+ outcome <- try action+ case outcome of+ Left (err :: SQLError) -> do+ let message = sqlErrorMessage err+ assertBool (Text.unpack message) (expected `Text.isInfixOf` message)+ -- The SQL error has the plain message of the Haskell exception.+ assertBool (Text.unpack message) (not (any (`Text.isInfixOf` message) ["user error", "SQLError {"]))+ Right _ -> assertFailure "expected an SQL error"++-- | Read one column, both columns, and more than one chunk of the example.+testTableSquares :: Assertion+testTableSquares = withConnection ":memory:" \conn -> do+ createTableFunction conn "typed_squares" ["n", "square"] \(Only count) ->+ pure [(n, n * n) | n <- [0 .. count - 1 :: Int64]]+ query_ conn "SELECT * FROM typed_squares(3)" >>= (@?= [(0 :: Int64, 0 :: Int64), (1, 1), (2, 4)])+ query_ conn "SELECT square FROM typed_squares(3)" >>= (@?= map Only [0, 1, 4 :: Int64])+ query_ conn "SELECT * FROM typed_squares(0)" >>= (@?= ([] :: [(Int64, Int64)]))+ -- A chunk has at most 2048 rows. Thus 5000 rows need three chunks.+ query_ conn "SELECT n, square FROM typed_squares(5000)" >>= (@?= [(n, n * n) | n <- [0 .. 4999 :: Int64]])++-- | Call table functions with no parameter, two parameters, and NULL.+testTableParameters :: Assertion+testTableParameters = withConnection ":memory:" \conn -> do+ createTableFunction conn "typed_words" ["word"] \() -> pure (map Only ["eitt", "tvö" :: Text])+ query_ conn "SELECT word FROM typed_words()" >>= (@?= map Only ["eitt", "tvö" :: Text])+ createTableFunction conn "typed_labels" ["label", "n"] \(prefix :: Text, start :: Int64) ->+ pure [(prefix <> Text.pack (show n), n) | n <- [start .. start + 2]]+ query_ conn "SELECT * FROM typed_labels('x', 5)" >>= (@?= [("x5" :: Text, 5 :: Int64), ("x6", 6), ("x7", 7)])+ -- DuckDB casts an argument to the parameter type.+ query_ conn "SELECT n FROM typed_labels('y', 1::INTEGER)" >>= (@?= map Only [1, 2, 3 :: Int64])+ query_ conn "SELECT parameters::VARCHAR FROM duckdb_functions() WHERE function_name = 'typed_labels'"+ >>= (@?= [Only ("[arg0, arg1]" :: Text)])+ -- A NULL argument converts to Nothing.+ createTableFunction conn "typed_optional" ["value"] \(Only (value :: Maybe Int64)) -> pure [Only value]+ query_ conn "SELECT value FROM typed_optional(NULL)" >>= (@?= [Only (Nothing :: Maybe Int64)])+ query_ conn "SELECT value FROM typed_optional(4)" >>= (@?= [Only (Just 4 :: Maybe Int64)])+ createTableFunction conn "typed_variant" ["value"] \(Only (value :: Maybe Variant)) -> pure [Only value]+ query_ conn "SELECT value FROM typed_variant(NULL)" >>= (@?= [Only (Nothing :: Maybe Variant)])+ expected <- query_ conn "SELECT {'a': [1,2]}::VARIANT" :: IO [Only Variant]+ query_ conn "SELECT value FROM typed_variant({'a': [1,2]}::VARIANT)" >>= (@?= expected)++-- | A row type with a 'Generic' instance.+data Reading = Reading+ { readingStation :: Text+ , readingValue :: Double+ , readingDay :: Day+ , readingNote :: Maybe Text+ }+ deriving (Generic)++instance TypedRow Reading++-- | Write columns of several types and NULLs from tuples and from a record.+testTableTypes :: Assertion+testTableTypes = withConnection ":memory:" \conn -> do+ let start = fromGregorian 2026 11 12+ mixed = [("a" :: Text, 0.5 :: Double, start, Just (1 :: Int64)), ("liður", -1.25, addDays 1 start, Nothing)]+ createTableFunction conn "typed_mixed" ["label", "value", "day", "amount"] \() -> pure mixed+ query_ conn "SELECT * FROM typed_mixed()" >>= (@?= mixed)+ query_ conn "SELECT typeof(label), typeof(value), typeof(day), typeof(amount) FROM typed_mixed() LIMIT 1"+ >>= (@?= [("VARCHAR" :: Text, "DOUBLE" :: Text, "DATE" :: Text, "BIGINT" :: Text)])+ createTableFunction conn "typed_readings" ["station", "value", "day", "note"] \(Only count) ->+ pure+ [ Reading ("s" <> Text.pack (show n)) (fromIntegral n / 2) (addDays (fromIntegral n) start) (if even n then Nothing else Just "odd")+ | n <- [0 .. count - 1 :: Int64]+ ]+ query_ conn "SELECT * FROM typed_readings(3)"+ >>= ( @?=+ [("s0" :: Text, 0 :: Double, start, Nothing :: Maybe Text), ("s1", 0.5, addDays 1 start, Just "odd"), ("s2", 1, addDays 2 start, Nothing)]+ )+ query_ conn "SELECT count(*) FROM typed_readings(3000) WHERE note IS NULL" >>= (@?= [Only (1500 :: Int64)])++-- | Give Integer and Natural columns. The declared column type is HUGEINT for+-- Integer and UHUGEINT for Natural, also for the elements of a list and the+-- values of a MAP. A value outside that range raises an error.+testTableIntegers :: Assertion+testTableIntegers = withConnection ":memory:" \conn -> do+ let integers = [5, -5, 2 ^ (100 :: Int), 2 ^ (127 :: Int) - 1 :: Integer]+ createTableFunction conn "typed_integers" ["v"] \() -> pure (map Only integers)+ query_ conn "SELECT typeof(v), v FROM typed_integers()" >>= (@?= [("HUGEINT" :: Text, value) | value <- integers])+ let naturals = [("small" :: Text, 7 :: Natural), ("large", 2 ^ (128 :: Int) - 1)]+ createTableFunction conn "typed_naturals" ["label", "v"] \() -> pure naturals+ query_ conn "SELECT label, typeof(v), v FROM typed_naturals()"+ >>= (@?= [(label, "UHUGEINT" :: Text, value) | (label, value) <- naturals])+ let nested = [([5, 2 ^ (100 :: Int)], Map.fromList [("a", 2 ^ (127 :: Int) :: Natural)])] :: [([Integer], Map.Map Text Natural)]+ createTableFunction conn "typed_integer_lists" ["vs", "m"] \() -> pure nested+ query_ conn "SELECT typeof(vs), typeof(m), vs, m FROM typed_integer_lists()"+ >>= (@?= [("HUGEINT[]" :: Text, "MAP(VARCHAR, UHUGEINT)" :: Text, vs, m) | (vs, m) <- nested])+ -- A MAP of a type text and an Integer or a Natural, also in a list, declares+ -- the text MAP(K, V) with the type text of the other part.+ let keyed = Map.fromList [([1, 2], 2 ^ (100 :: Int))] :: Map.Map [Int64] Integer+ listed = Map.fromList [([3], [5, -5])] :: Map.Map [Int64] [Integer]+ nestedMaps = [[Map.fromList [([4], 2 ^ (127 :: Int))]], []] :: [[Map.Map [Int64] Natural]]+ createTableFunction conn "typed_text_maps" ["k", "l", "n", "p"] \() ->+ pure [(keyed, listed, maps, Map.fromList [(7 :: Natural, Price 1.5)]) | maps <- nestedMaps]+ query_ conn "SELECT DISTINCT typeof(k), typeof(l), typeof(n), typeof(p) FROM typed_text_maps()"+ >>= ( @?=+ [+ ( "MAP(BIGINT[], HUGEINT)" :: Text+ , "MAP(BIGINT[], HUGEINT[])" :: Text+ , "MAP(BIGINT[], UHUGEINT)[]" :: Text+ , "MAP(UHUGEINT, DECIMAL(18,3))" :: Text+ )+ ]+ )+ query_ conn "SELECT k, l, n, p::VARCHAR FROM typed_text_maps()"+ >>= (@?= [(keyed, listed, maps, "{7=1.500}" :: Text) | maps <- nestedMaps])+ -- DuckDB sums a UHUGEINT column as a DOUBLE.+ query_ conn "SELECT typeof(sum(v)), sum(v) FROM typed_naturals()" >>= (@?= [("DOUBLE" :: Text, 2 ^ (128 :: Int) :: Double)])+ -- toDuckValue raises an error for a value outside the range. A BIGNUM value+ -- would reach the HUGEINT column, and DuckDB casts the BIGNUM 2^128 to 0.+ let outside typeName = "Conversion Error: duckdb-simple: the value 340282366920938463463374607431768211456 is outside the range of " <> typeName+ createTableFunction conn "typed_wide_integer" ["v"] \() -> pure [Only (2 ^ (128 :: Int) :: Integer)]+ assertErrorType+ "Integer"+ ErrorTypeConversion+ (outside "HUGEINT")+ (void (query_ conn "SELECT v FROM typed_wide_integer()" :: IO [Only Integer]))+ createTableFunction conn "typed_wide_list" ["v"] \() -> pure [Only [5, 2 ^ (128 :: Int) :: Integer]]+ assertErrorType+ "Integer list"+ ErrorTypeConversion+ (outside "HUGEINT")+ (void (query_ conn "SELECT v FROM typed_wide_list()" :: IO [Only [Integer]]))+ createTableFunction conn "typed_wide_natural" ["v"] \() -> pure [Only (2 ^ (128 :: Int) :: Natural)]+ assertErrorType+ "Natural"+ ErrorTypeConversion+ (outside "UHUGEINT")+ (void (query_ conn "SELECT v FROM typed_wide_natural()" :: IO [Only Natural]))+ -- A BigNum column holds each value.+ createTableFunction conn "typed_bignums" ["v"] \() -> pure [Only (BigNum (2 ^ (130 :: Int)))]+ query_ conn "SELECT typeof(v), v FROM typed_bignums()" >>= (@?= [("BIGNUM" :: Text, BigNum (2 ^ (130 :: Int)))])++-- | Give IntervalValue and TimeWithZone columns.+testTableIntervals :: Assertion+testTableIntervals = withConnection ":memory:" \conn -> do+ let rows =+ [ (IntervalValue 1 2 3, TimeWithZone (TimeOfDay 12 34 0) 19800)+ , (IntervalValue (-1) 0 0, TimeWithZone (TimeOfDay 0 0 0) 3630)+ ]+ createTableFunction conn "typed_intervals" ["duration", "moment"] \() -> pure rows+ query_ conn "SELECT typeof(duration), typeof(moment) FROM typed_intervals() LIMIT 1"+ >>= (@?= [("INTERVAL" :: Text, "TIME WITH TIME ZONE" :: Text)])+ query_ conn "SELECT duration, moment FROM typed_intervals()" >>= (@?= rows)++-- | Give a MAP column. The pairs go in ascending key order.+testTableMaps :: Assertion+testTableMaps = withConnection ":memory:" \conn -> do+ let rows = map Only [Map.fromList [("b", 2), ("a", 1)], Map.empty :: Map.Map Text Int64]+ createTableFunction conn "typed_maps" ["pairs"] \() -> pure rows+ query_ conn "SELECT typeof(pairs), map_keys(pairs) FROM typed_maps()"+ >>= (@?= [("MAP(VARCHAR, BIGINT)" :: Text, ["a", "b"] :: [Text]), ("MAP(VARCHAR, BIGINT)", [])])+ query_ conn "SELECT pairs FROM typed_maps()" >>= (@?= rows)++-- | A record for a STRUCT column. A String field is VARCHAR.+data Peak = Peak {peak :: Text, metres :: Int32, remark :: String}+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToDuckValue, ToField, FromField) via (ViaDuckDB Peak)++-- | A text of the type JSON. A record field of this type has the type JSON.+newtype JsonText = JsonText Text+ deriving newtype (Eq, Show, FromField, ToField)++instance DuckValue JsonText where+ duckToField (JsonText text) = duckToField text+ duckFromField = fmap JsonText . duckFromField+ duckLogicalType _ = LogicalTypeAlias "JSON" (LogicalTypeScalar LogicalTypeIdVarchar)++-- | A record with a field of the type JSON.+data Doc = Doc {serial :: Int64, body :: JsonText}+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToDuckValue, ToField, FromField) via (ViaDuckDB Doc)++-- | A sum type for a UNION column.+data Weather = Sunny | Rain Double+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToDuckValue, FromField) via (ViaDuckDB Weather)++-- | A record with an Integer field, of the type HUGEINT, and a Natural field,+-- of the type UHUGEINT.+data Counts = Counts {signed :: Integer, unsigned :: Natural}+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToDuckValue, ToField, FromField) via (ViaDuckDB Counts)++-- | A list of three BIGINT values. Its type text gives the size.+newtype Triple = Triple [Int64]+ deriving newtype (Eq, Show, FromField, ToField, ToDuckValue)++instance DuckDBColumnType Triple where+ duckdbColumnTypeFor _ = "BIGINT[3]"++-- | A record with a field of a sized ARRAY type.+newtype Boxed = Boxed {boxed :: SizedArray 3 Int64}+ deriving stock (Generic)+ deriving (DuckDBColumnType, ToDuckValue, FromField) via (ViaDuckDB Boxed)++-- | A newtype over a raw result type. It derives the raw result writer.+newtype Age = Age Int16+ deriving newtype (Eq, Ord, Show, DuckDBColumnType, ToDuckValue, FromField, FunctionResult)++-- | A newtype over a raw result type, with the catch-all 'FunctionResult'.+newtype Years = Years Int16+ deriving newtype (DuckDBColumnType, ToDuckValue)++-- | Produce an uncastable value to exercise cleanup after a native write fails.+data EncodedTotal = EncodedTotal Int64 Bool++instance DuckDBColumnType EncodedTotal where+ duckdbColumnTypeFor _ = "BIGINT"++instance ToDuckValue EncodedTotal where+ toDuckValue factory (EncodedTotal value failing)+ | failing = toDuckValue factory ("typed_release: encode" :: Text)+ | otherwise = toDuckValue factory value++-- | A catalog type whose encoder needs only the built-in BIGINT factory.+newtype AggregateTotal = AggregateTotal Int64+ deriving newtype (ToDuckValue)++instance DuckDBColumnType AggregateTotal where+ duckdbColumnTypeFor _ = "aggregate_total"++-- | An encoder that requires catalog lookup from its factory.+newtype CatalogTotal = CatalogTotal Int64++instance DuckDBColumnType CatalogTotal where+ duckdbColumnTypeFor _ = "aggregate_total"++instance ToDuckValue CatalogTotal where+ toDuckValue factory (CatalogTotal value) =+ R.withTypeDeclaration factory (TypeExpression "aggregate_total") \_ -> toDuckValue factory value++-- | A newtype with its own type text.+newtype Price = Price Double++instance DuckDBColumnType Price where+ duckdbColumnTypeFor _ = "DECIMAL(18,3)"++instance ToDuckValue Price where+ toDuckValue factory (Price value) = toDuckValue factory value++-- | Declare record, list, MAP, and UNION columns and a record aggregate+-- argument from the generic types. An empty list and an empty MAP of records+-- take the element type from the declaration.+testTableRecords :: Assertion+testTableRecords = withConnection ":memory:" \conn -> do+ let peaks = [Peak "Hekla" 1491 "volcano", Peak "Esja" 914 ""]+ rows =+ [(row, peaks, Map.fromList [(peak p, p) | p <- peaks], weather) | (row, weather) <- zip peaks [Sunny, Rain 2.5]]+ <> [(Peak "Snæfell" 1833 "", [], Map.empty, Rain 0)]+ createTableFunction conn "typed_peaks" ["peak", "nearby", "by_name", "weather"] \() -> pure rows+ query_ conn "SELECT typeof(peak), typeof(nearby), typeof(by_name), typeof(weather) FROM typed_peaks() LIMIT 1"+ >>= ( @?=+ [+ ( "STRUCT(peak VARCHAR, metres INTEGER, remark VARCHAR)" :: Text+ , "STRUCT(peak VARCHAR, metres INTEGER, remark VARCHAR)[]" :: Text+ , "MAP(VARCHAR, STRUCT(peak VARCHAR, metres INTEGER, remark VARCHAR))" :: Text+ , "UNION(Sunny STRUCT, Rain STRUCT(field1 DOUBLE))" :: Text+ )+ ]+ )+ query_ conn "SELECT peak, nearby, by_name, weather FROM typed_peaks()" >>= (@?= rows)+ createAggregate+ conn+ "typed_highest"+ Aggregate+ { aggregateInitial = 0+ , aggregateStep = \highest (Only p) -> pure (max highest (metres p))+ , aggregateCombine = \a b -> pure (max a b)+ , aggregateResult = pure :: Int32 -> IO Int32+ }+ query_ conn "SELECT typed_highest(peak) FROM typed_peaks()" >>= (@?= [Only (1833 :: Int32)])+ query_ conn "SELECT parameter_types::VARCHAR FROM duckdb_functions() WHERE function_name = 'typed_highest'"+ >>= (@?= [Only ("['STRUCT(peak VARCHAR, metres INTEGER, remark VARCHAR)']" :: Text)])+ query conn "SELECT ?, typeof(?)" ([] :: [Peak], [] :: [Peak])+ >>= (@?= [([] :: [Peak], "STRUCT(peak VARCHAR, metres INTEGER, remark VARCHAR)[]" :: Text)])+ -- With no target type, a list of NULL records also takes the declaration.+ -- With a target type, the empty list and the list of NULL records take the+ -- target type, as for each element type.+ query conn "SELECT typeof(?)" (Only [Nothing :: Maybe Peak])+ >>= (@?= [Only ("STRUCT(peak VARCHAR, metres INTEGER, remark VARCHAR)[]" :: Text)])+ _ <- execute_ conn "CREATE TABLE typed_short_peaks (ps STRUCT(peak VARCHAR, metres INTEGER)[])"+ _ <- execute conn "INSERT INTO typed_short_peaks VALUES (?), (?)" ([] :: [Peak], [Nothing :: Maybe Peak])+ query_ conn "SELECT len(ps) FROM typed_short_peaks ORDER BY len(ps)" >>= (@?= [Only (0 :: Int64), Only 1])++-- | The type of a DECIMAL(4,1) argument, as its text.+newtype ArgumentType = ArgumentType Text++instance DuckDBColumnType ArgumentType where+ duckdbColumnTypeFor _ = "DECIMAL(4,1)"++instance FromField ArgumentType where+ fromField Field{fieldType} = ArgumentType . Text.pack . show <$> fieldType++-- | The argument of a scalar function gives the type of its vector.+testArgumentType :: Assertion+testArgumentType = withConnection ":memory:" \conn -> do+ createFunction conn "argument_type" \(ArgumentType text) -> text+ query_ conn "SELECT argument_type(1.5) FROM range(3000) t(i) LIMIT 1" >>= (@?= [Only ("LogicalTypeDecimal 4 1" :: Text)])++-- | A scalar argument with its value and native metadata action.+data ScopedArgument = ScopedArgument Double (Conversion LogicalTypeRep)++instance DuckDBColumnType ScopedArgument where+ duckdbColumnTypeFor _ = "DOUBLE"++instance FromField ScopedArgument where+ fromField entry = ScopedArgument <$> fromField entry <*> pure (fieldType entry)++instance ToDuckValue ScopedArgument where+ toDuckValue factory (ScopedArgument value _) = toDuckValue factory value++-- | Pure snapshots survive a scalar callback; its native actions expire.+testArgumentTypeScope :: Assertion+testArgumentTypeScope = forM_ [False, True] \cached -> forM_ [False, True] \fails -> withConnection ":memory:" \conn -> do+ captured <- newIORef Nothing+ createFunction conn "argument_scope" \(ScopedArgument value logical) -> do+ value @?= 1.5+ snapshot <-+ if cached+ then+ runConversion logical >>= \case+ Ok rep -> pure (Just rep)+ Errors errors -> assertFailure (show errors) >> pure Nothing+ else pure Nothing+ writeIORef captured (Just (logical, snapshot))+ when fails (ioError (userError "scope callback failed"))+ pure (7 :: Int64)+ try (query_ conn "SELECT argument_scope(1.5)" :: IO [Only Int64]) >>= \case+ Left SQLError{sqlErrorMessage} -> assertBool (Text.unpack sqlErrorMessage) (fails && "scope callback failed" `Text.isInfixOf` sqlErrorMessage)+ Right rows -> do+ assertBool "the callback should have failed" (not fails)+ rows @?= [Only 7]+ performMajorGC+ readIORef captured >>= \case+ Nothing -> assertFailure "the scalar callback did not capture its argument type"+ Just (logical, snapshot) -> do+ snapshot @?= if cached then Just (LogicalTypeScalar LogicalTypeIdDouble) else Nothing+ try (runConversion logical) >>= \case+ Left SQLError{sqlErrorMessage} -> sqlErrorMessage @?= "duckdb-simple: field type scope is closed"+ Right _ -> assertFailure "scalar metadata survived its callback"++-- | A deferred table scan cannot use native metadata from its bind callback.+testTableArgumentTypeScope :: Assertion+testTableArgumentTypeScope = withConnection ":memory:" \conn -> do+ createTableFunction conn "table_argument_scope" ["value"] \(Only (ScopedArgument value logical)) -> do+ try (runConversion logical) >>= \case+ Left SQLError{sqlErrorMessage} -> sqlErrorMessage @?= "duckdb-simple: field type scope is closed"+ Right _ -> assertFailure "table metadata survived its bind callback"+ pure [Only value]+ query_ conn "SELECT value FROM table_argument_scope(1.5)" >>= (@?= [Only (1.5 :: Double)])++-- | A record field of an alias type declares the alias in the STRUCT type of a+-- table function column. A record parameter, a list of records, and a list of+-- NULL records have the same type. A value of the field reads as its base type.+testAliasDeclaration :: Assertion+testAliasDeclaration = withConnection ":memory:" \conn -> do+ let docs = [Doc 1 (JsonText "{\"a\": 1}"), Doc 2 (JsonText "[2]")]+ createTableFunction conn "typed_docs" ["doc"] \() -> pure (map Only docs)+ query_ conn "SELECT typeof(doc), typeof(doc.body) FROM typed_docs() LIMIT 1"+ >>= (@?= [("STRUCT(serial BIGINT, body JSON)" :: Text, "JSON" :: Text)])+ query_ conn "SELECT doc FROM typed_docs()" >>= (@?= map Only docs)+ query conn "SELECT typeof(?), typeof(?), typeof(?)" (Doc 3 (JsonText "{}"), docs, [Nothing :: Maybe Doc])+ >>= ( @?=+ [("STRUCT(serial BIGINT, body JSON)" :: Text, "STRUCT(serial BIGINT, body JSON)[]" :: Text, "STRUCT(serial BIGINT, body JSON)[]" :: Text)]+ )++-- | Reject a column, a parameter, or an argument type with no declaration at+-- registration, also inside a list, and a MAP of a type text and a record.+testNoDeclaration :: Assertion+testNoDeclaration = withConnection ":memory:" \conn -> do+ assertSqlError+ "typed_struct: column s: StructValue has no static field types"+ (createTableFunction conn "typed_struct" ["s"] \() -> pure ([] :: [Only (StructValue FieldValue)]))+ assertSqlError+ "typed_tuple: column t: TupleValue has no static field types"+ (createTableFunction conn "typed_tuple" ["t"] \() -> pure ([] :: [Only (TupleValue FieldValue)]))+ assertSqlError+ "typed_union: column u: UnionValue has no static field types"+ (createTableFunction conn "typed_union" ["u"] \() -> pure ([] :: [Only (UnionValue FieldValue)]))+ assertSqlError+ "typed_structs: column s: StructValue has no static field types"+ (createTableFunction conn "typed_structs" ["n", "s"] \() -> pure ([] :: [(Int64, [Maybe (StructValue FieldValue)])]))+ assertSqlError+ "typed_struct_parameter: parameter arg0: StructValue has no static field types"+ (createTableFunction conn "typed_struct_parameter" ["n"] \(Only (_ :: StructValue FieldValue)) -> pure [Only (1 :: Int64)])+ assertSqlError+ "typed_struct_count: argument arg0: StructValue has no static field types"+ ( createAggregate+ conn+ "typed_struct_count"+ Aggregate+ { aggregateInitial = 0 :: Int64+ , aggregateStep = \count (Only (_ :: StructValue FieldValue)) -> pure (count + 1)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }+ )+ assertSqlError+ "typed_priced: column p: a MAP with the type text DECIMAL(18,3) and a structured type has no declaration"+ (createTableFunction conn "typed_priced" ["p"] \() -> pure ([] :: [Only (Map.Map Price Peak)]))+ query_ conn "SELECT count(*) FROM duckdb_functions() WHERE function_name LIKE 'typed\\_%' ESCAPE '\\'" >>= (@?= [Only (0 :: Int64)])++-- | Use a SizedArray of size 3 in each position, and a newtype with a sized ARRAY type+-- text as an argument. DuckDB rejects an ARRAY of another size when it binds+-- the call, casts a LIST of the same length to the ARRAY, and rejects a LIST+-- of another length when the statement runs. See Note [Sized arrays] in+-- "Database.DuckDB.Simple.ToField".+testSizedArray :: Assertion+testSizedArray = withConnection ":memory:" \conn -> do+ let total = sum :: SizedArray 3 Int64 -> Int64+ triple x = fromJust (sizedArrayFromList @3 [x, x, x :: Int64])+ createFunction conn "sized_argument" total+ createFunction conn "sized_newtype" \(Triple values) -> sum values+ createFunction conn "sized_maybe" (maybe (-1) total :: Maybe (SizedArray 3 Int64) -> Int64)+ createFunction conn "sized_list" (sum . map total :: [SizedArray 3 Int64] -> Int64)+ createFunction conn "sized_record" \(Boxed values) -> total values+ createAggregate+ conn+ "sized_aggregate"+ Aggregate+ { aggregateInitial = 0 :: Int64+ , aggregateStep = \sum' (Only values) -> pure (sum' + total values)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }+ createTableFunction conn "sized_table" ["total"] \(Only values) -> pure [Only (total values)]+ -- The third part is the type of the error for a LIST of another length.+ -- DuckDB casts the parameter of a table function as a constant value, and+ -- that cast raises ErrorInputInvalid "Failed to cast value".+ let positions :: [(String, Text -> Text, ErrorType)]+ positions =+ [ ("scalar argument", \v -> "SELECT sized_argument(" <> v <> ")", ErrorTypeConversion)+ , ("newtype argument", \v -> "SELECT sized_newtype(" <> v <> ")", ErrorTypeConversion)+ , ("Maybe argument", \v -> "SELECT sized_maybe(" <> v <> ")", ErrorTypeConversion)+ , ("list argument", \v -> "SELECT sized_list([" <> v <> "])", ErrorTypeConversion)+ , ("record field", \v -> "SELECT sized_record({'boxed': " <> v <> "})", ErrorTypeConversion)+ , ("aggregate argument", \v -> "SELECT sized_aggregate(" <> v <> ")", ErrorTypeConversion)+ , ("table function parameter", \v -> "SELECT * FROM sized_table(" <> v <> ")", ErrorInputInvalid)+ ]+ values :: ErrorType -> [(String, Text, Either (ErrorType, Text) Int64)]+ values lengthError =+ [ ("ARRAY of 3", "[1, 2, 3]::BIGINT[3]", Right 6)+ , ("ARRAY of 4", "[1, 2, 3, 4]::BIGINT[4]", Left (ErrorQueryBinder, "No function matches"))+ , ("LIST of 3", "[1, 2, 3]", Right 6)+ , ("LIST of 2", "[1, 2]", Left (lengthError, "Cannot cast list with length 2 to array with length 3"))+ ]+ check :: String -> IO [Only Int64] -> Either (ErrorType, Text) Int64 -> Assertion+ check name action expected = do+ outcome <- try action+ case (outcome, expected) of+ (Right rows, Right value) -> assertEqual name [Only value] rows+ (Left err, Left (errorType, text)) -> do+ assertEqual (name <> ": " <> Text.unpack (sqlErrorMessage err)) (Just errorType) (sqlErrorType err)+ assertBool (name <> ": " <> Text.unpack (sqlErrorMessage err)) (text `Text.isInfixOf` sqlErrorMessage err)+ (Right rows, Left _) -> assertFailure (name <> ": expected an error, got " <> show rows)+ (Left (err :: SQLError), Right _) -> assertFailure (name <> ": " <> show err)+ forM_ positions \(position, sql, lengthError) -> forM_ (values lengthError) \(label, value, expected) ->+ check (position <> ", " <> label) (query_ conn (Query (sql value))) expected+ -- A bound SizedArray binds as an ARRAY of its size, and a FieldArray as an+ -- ARRAY of its own size. DuckDB casts a bound list of the same length.+ let fieldArray items = FieldArray (listArray (0, length items - 1) (map FieldInt64 items))+ check "bound SizedArray of 3" (query conn "SELECT sized_argument(?)" (Only (triple 1))) (Right 3)+ check "bound FieldArray of 3" (query conn "SELECT sized_argument(?)" (Only (fieldArray [1, 2, 3]))) (Right 6)+ check+ "bound FieldArray of 4"+ (query conn "SELECT sized_argument(?)" (Only (fieldArray [1, 2, 3, 4])))+ (Left (ErrorQueryBinder, "No function matches"))+ check "bound list of 3" (query conn "SELECT sized_argument(?)" (Only [1, 2, 3 :: Int64])) (Right 6)+ -- A result and a column of a table function are ARRAY values of the size.+ createFunction conn "sized_result" triple+ createTableFunction conn "sized_rows" ["t"] \(Only (x :: Int64)) -> pure [Only (triple x)]+ query_ conn "SELECT typeof(sized_result(1)), sized_result(1)::VARCHAR" >>= (@?= [("BIGINT[3]" :: Text, "[1, 1, 1]" :: Text)])+ query_ conn "SELECT typeof(t), t::VARCHAR FROM sized_rows(2)" >>= (@?= [("BIGINT[3]" :: Text, "[2, 2, 2]" :: Text)])+ -- A result of another size raises the cast error of DuckDB when the function+ -- runs. A SizedArray has no value of another size, but a hand-written newtype has.+ createFunction conn "sized_wrong_result" \(x :: Int64) -> Triple [x, x, x, x]+ check+ "result of 4"+ (query_ conn "SELECT len(sized_wrong_result(1))")+ (Left (ErrorInputInvalid, "Cannot cast list with length 4 to array with length 3"))++-- | Give an error, and no other value, for a record field outside the range+-- of its declared type, as a parameter, as a column, and as a scalar result.+testRecordFieldRange :: Assertion+testRecordFieldRange = withConnection ":memory:" \conn -> do+ let largest = Counts (2 ^ (127 :: Int) - 1) (2 ^ (128 :: Int) - 1)+ outside = [(Counts (2 ^ (127 :: Int)) 0, "HUGEINT value out of range"), (Counts 0 (2 ^ (128 :: Int)), "UHUGEINT value out of range")]+ cases = largest : map fst outside+ query conn "SELECT ?" (Only largest) >>= (@?= [Only largest])+ createFunction conn "typed_counts_result" (\(index :: Int) -> cases !! index)+ query_ conn "SELECT typeof(typed_counts_result(0)), typed_counts_result(0)"+ >>= (@?= [("STRUCT(signed HUGEINT, unsigned UHUGEINT)" :: Text, largest)])+ createTableFunction conn "typed_counts" ["c"] \(Only (index :: Int)) -> pure [Only (cases !! index)]+ query_ conn "SELECT typeof(c), c FROM typed_counts(0)" >>= (@?= [("STRUCT(signed HUGEINT, unsigned UHUGEINT)" :: Text, largest)])+ forM_ (zip [1 :: Int ..] outside) \(index, (counts, message)) -> do+ outcome <- try (query conn "SELECT ?" (Only counts) :: IO [Only Counts])+ case outcome of+ Left (err :: SomeException) -> assertBool (displayException err) (message `Text.isInfixOf` Text.pack (displayException err))+ Right rows -> assertFailure ("expected an error, got " <> show rows)+ assertSqlError message (query_ conn (Query ("SELECT c FROM typed_counts(" <> Text.pack (show index) <> ")")) :: IO [Only Counts])+ assertSqlError message (query_ conn (Query ("SELECT typed_counts_result(" <> Text.pack (show index) <> ")")) :: IO [Only Counts])++-- | Run the function for each scan. Do not run it at registration or preparation.+testTableRuns :: Assertion+testTableRuns = withConnection ":memory:" \conn -> do+ calls <- newIORef (0 :: Int)+ createTableFunction conn "typed_counted" ["n"] \(Only count) -> do+ modifyIORef' calls (+ 1)+ pure (map Only [1 .. count :: Int64])+ readIORef calls >>= (@?= 0)+ withStatement conn "SELECT n FROM typed_counted(2)" \statement -> do+ readIORef calls >>= (@?= 0)+ _ <- executeStatement statement+ readIORef calls >>= (@?= 1)+ _ <- executeStatement statement+ readIORef calls >>= (@?= 2)+ query_ conn "SELECT a.n, b.n FROM typed_counted(2) a JOIN typed_counted(3) b ON a.n = b.n ORDER BY a.n"+ >>= (@?= [(1 :: Int64, 1 :: Int64), (2, 2)])+ readIORef calls >>= (@?= 4)++-- | Read the first rows of an infinite list.+testTableLazy :: Assertion+testTableLazy = withConnection ":memory:" \conn -> do+ createTableFunction conn "typed_naturals" ["n"] \() -> pure (map Only [0 :: Int64 ..])+ query_ conn "SELECT n FROM typed_naturals() LIMIT 3" >>= (@?= map Only [0, 1, 2 :: Int64])+ -- 10000 rows need more than one chunk.+ query_ conn "SELECT sum(n)::BIGINT FROM (SELECT n FROM typed_naturals() LIMIT 10000)" >>= (@?= [Only (49995000 :: Int64)])++-- | Reject wrong column names at registration. Fail a query with an argument that does not convert.+testTableArgumentErrors :: Assertion+testTableArgumentErrors = withConnection ":memory:" \conn -> do+ assertSqlError+ "1 column names for 2 result columns"+ (createTableFunction conn "typed_bad_names" ["a"] \() -> pure [(1 :: Int64, 2 :: Int64)])+ assertSqlError "at least one result column" (createTableFunction conn "typed_no_columns" [] \() -> pure [()])+ -- A Natural parameter is UHUGEINT. DuckDB does not cast a negative literal to it.+ createTableFunction conn "typed_natural" ["n"] \(Only (value :: Natural)) -> pure [Only (fromIntegral value :: Int64)]+ query_ conn "SELECT n FROM typed_natural(7)" >>= (@?= [Only (7 :: Int64)])+ assertSqlError "No function matches" (query_ conn "SELECT n FROM typed_natural(-7)" :: IO [Only Int64])+ assertSqlError "typed_natural: argument arg0 is NULL" (query_ conn "SELECT n FROM typed_natural(NULL)" :: IO [Only Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | Fail a query with an exception in the function and in a row of the second chunk.+testTableExceptions :: Assertion+testTableExceptions = withConnection ":memory:" \conn -> do+ createTableFunction conn "typed_checked" ["n"] \(Only count) ->+ if count < 0 then throwIO (userError "typed_checked: the count is negative") else pure (map Only [1 .. count :: Int64])+ assertSqlError "typed_checked: the count is negative" (query_ conn "SELECT n FROM typed_checked(-1)" :: IO [Only Int64])+ query_ conn "SELECT n FROM typed_checked(2)" >>= (@?= map Only [1, 2 :: Int64])+ -- Row 3000 is in the second chunk.+ createTableFunction conn "typed_partial" ["n"] \() ->+ pure [Only (if n == 3000 then error "typed_partial: row 3000 fails" else n) | n <- [0 :: Int64 ..]]+ assertSqlError "typed_partial: row 3000 fails" (query_ conn "SELECT count(*) FROM typed_partial()" :: IO [Only Int64])+ query_ conn "SELECT n FROM typed_partial() LIMIT 2" >>= (@?= map Only [0, 1 :: Int64])+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | Evaluate only the columns that the query uses. A column that the query does+-- not use can raise an exception, and the query does not fail. A column in+-- WHERE or in ORDER BY is a column that the query uses. SELECT * reads all+-- columns, and count(*) reads the first column. The row is evaluated with each+-- strict field.+testTableProjection :: Assertion+testTableProjection = withConnection ":memory:" \conn -> do+ createTableFunction conn "typed_pairs" ["n", "failing"] \(Only count) ->+ pure [(n, if n == 2500 then error "typed_pairs: the column fails" else n * 2) | n <- [0 .. count - 1 :: Int64]]+ query_ conn "SELECT n FROM typed_pairs(3000) WHERE n % 1000 = 1" >>= (@?= map Only [1, 1001, 2001 :: Int64])+ query_ conn "SELECT sum(n)::BIGINT FROM typed_pairs(3000)" >>= (@?= [Only (sum [0 .. 2999 :: Int64])])+ assertSqlError "typed_pairs: the column fails" (query_ conn "SELECT failing FROM typed_pairs(3000)" :: IO [Only Int64])+ query_ conn "SELECT failing FROM typed_pairs(3)" >>= (@?= map Only [0, 2, 4 :: Int64])+ query_ conn "SELECT * FROM typed_pairs(3)" >>= (@?= ([(0, 0), (1, 2), (2, 4)] :: [(Int64, Int64)]))+ query_ conn "SELECT failing, n FROM typed_pairs(3)" >>= (@?= ([(0, 0), (2, 1), (4, 2)] :: [(Int64, Int64)]))+ assertSqlError "typed_pairs: the column fails" (query_ conn "SELECT failing, n FROM typed_pairs(3000)" :: IO [(Int64, Int64)])+ query_ conn "SELECT n, n FROM typed_pairs(3000) WHERE n % 1000 = 1" >>= (@?= ([(1, 1), (1001, 1001), (2001, 2001)] :: [(Int64, Int64)]))+ query_ conn "SELECT failing, failing FROM typed_pairs(3)" >>= (@?= ([(0, 0), (2, 2), (4, 4)] :: [(Int64, Int64)]))+ query_ conn "SELECT n FROM typed_pairs(3) WHERE failing > 2" >>= (@?= [Only (2 :: Int64)])+ query_ conn "SELECT n FROM typed_pairs(3) ORDER BY failing DESC" >>= (@?= map Only [2, 1, 0 :: Int64])+ assertSqlError "typed_pairs: the column fails" (query_ conn "SELECT n FROM typed_pairs(3000) WHERE failing >= 0" :: IO [Only Int64])+ assertSqlError "typed_pairs: the column fails" (query_ conn "SELECT n FROM typed_pairs(3000) ORDER BY failing" :: IO [Only Int64])+ query_ conn "SELECT count(*) FROM typed_pairs(3000)" >>= (@?= [Only (3000 :: Int64)])+ createTableFunction conn "typed_pairs_first" ["failing", "n"] \(Only count) ->+ pure [(if n == 2500 then error "typed_pairs_first: the column fails" else n * 2, n) | n <- [0 .. count - 1 :: Int64]]+ assertSqlError "typed_pairs_first: the column fails" (query_ conn "SELECT count(*) FROM typed_pairs_first(3000)" :: IO [Only Int64])+ query_ conn "SELECT n FROM typed_pairs_first(3000) WHERE n % 1000 = 1" >>= (@?= map Only [1, 1001, 2001 :: Int64])+ createTableFunction conn "typed_strict_pairs" ["n", "failing"] \(Only count) ->+ pure [StrictPair n (if n == 2500 then error "typed_strict_pairs: the field fails" else n * 2) | n <- [0 .. count - 1 :: Int64]]+ query_ conn "SELECT n FROM typed_strict_pairs(3)" >>= (@?= map Only [0, 1, 2 :: Int64])+ assertSqlError "typed_strict_pairs: the field fails" (query_ conn "SELECT n FROM typed_strict_pairs(3000)" :: IO [Only Int64])++-- | A row type with strict fields.+data StrictPair = StrictPair !Int64 !Int64+ deriving (Generic)++instance TypedRow StrictPair++-- | Release the list of each scan after a full scan, after LIMIT, and after an+-- error. Each call makes a token. The rows that the scan did not read refer to+-- the token. Release the function when the connection closes.+testTableRelease :: Assertion+testTableRelease = do+ closure <- withConnection ":memory:" \conn -> do+ tokens <- newIORef []+ offset <- newIORef (0 :: Int64)+ offsetWeak <- mkWeakIORef offset (pure ())+ createTableFunction conn "typed_tokens" ["n"] \(count :: Int64, failure :: Int64) -> do+ token <- readIORef offset >>= newIORef+ weak <- mkWeakIORef token (pure ())+ modifyIORef' tokens (weak :)+ let rowsFrom n+ | n == count = pure []+ | otherwise = unsafeInterleaveIO do+ value <- readIORef token+ when (n == failure) $ throwIO (userError "typed_tokens: the row fails")+ (Only (value + n) :) <$> rowsFrom (n + 1)+ rowsFrom 0+ -- Check the token of the last scan before the next statement.+ let released scan = do+ performMajorGC+ performMajorGC+ scans <- readIORef tokens+ case scans of+ weak : _ -> deRefWeak weak >>= assertBool (scan <> " keeps its list after the scan ends") . isNothing+ [] -> assertFailure (scan <> " did not run the function")+ query_ conn "SELECT count(*) FROM typed_tokens(3000, -1)" >>= (@?= [Only (3000 :: Int64)])+ released "a full scan"+ query_ conn "SELECT n FROM typed_tokens(-1, -1) LIMIT 3" >>= (@?= map Only [0, 1, 2 :: Int64])+ released "a scan with LIMIT"+ assertSqlError "typed_tokens: the row fails" (query_ conn "SELECT count(*) FROM typed_tokens(-1, 2500)" :: IO [Only Int64])+ released "a failed scan"+ readIORef tokens >>= (@?= 3) . length+ pure offsetWeak+ performMajorGC+ performMajorGC+ deRefWeak closure >>= assertBool "the table function remains after the connection closes" . isNothing++-- | Write 3000 rows with NULLs to a file and read them back with one format.+-- The rows need more than one chunk in each direction.+testCopyBothDirections :: Assertion+testCopyBothDirections = withDirectory \directory -> withConnection ":memory:" \conn -> do+ calls <- newIORef ([] :: [CopyFromInfo])+ registerCopyFunction conn "typed_lines" linesTo \info -> do+ modifyIORef' calls (info :)+ readLines (copyFromPath info)+ let file = directory </> "lines.txt"+ expected = [if i `mod` 1000 == 7 then Nothing else Just i | i <- [0 .. 2999 :: Int64]]+ void+ ( execute+ conn+ "COPY (SELECT CASE WHEN i % 1000 = 7 THEN NULL ELSE i END AS value FROM range(3000) t(i)) TO ? (FORMAT typed_lines)"+ (Only (Text.pack file))+ )+ readFile file >>= (@?= unlines (map (maybe "" show) expected))+ void (execute_ conn "CREATE TABLE typed_copied (value BIGINT)")+ -- The function runs one time for each run of a COPY FROM statement.+ forM_ [1 :: Int, 2] \_ -> void (execute conn "COPY typed_copied FROM ? (FORMAT typed_lines)" (Only (Text.pack file)))+ query_ conn "SELECT value FROM typed_copied ORDER BY rowid" >>= (@?= map Only (expected <> expected))+ infos <- readIORef calls+ length infos @?= 2+ forM_ infos \info -> do+ copyFromPath info @?= file+ copyFromColumnNames info @?= ["value"]+ length (copyFromColumnTypes info) @?= 1++-- | DuckDB keeps the first registration of a format name. A later+-- registration has no effect, so it does not add a COPY FROM side.+testCopyFirstRegistration :: Assertion+testCopyFirstRegistration = withDirectory \directory -> withConnection ":memory:" \conn -> do+ sinks <- newIORef ([] :: [Text])+ let file = Text.pack (directory </> "first.txt")+ register label =+ registerCopyToFunction+ conn+ "typed_first"+ (\_ -> pure ())+ (\info -> writeFile (copyInitFilePath info) "")+ (\_ _ -> modifyIORef' sinks (label :))+ (\_ -> pure ())+ register "first"+ register "second"+ registerCopyFromFunction conn "typed_first" \_ -> pure [Only (1 :: Int64)]+ void (execute conn "COPY (SELECT 1) TO ? (FORMAT typed_first)" (Only file))+ readIORef sinks >>= (@?= ["first"])+ void (execute_ conn "CREATE TABLE typed_first_rows (value BIGINT)")+ assertSqlError+ "COPY FROM is not supported for FORMAT \"typed_first\""+ (execute conn "COPY typed_first_rows FROM ? (FORMAT typed_first)" (Only file))+ -- A format with a COPY FROM side only rejects COPY TO.+ registerCopyFromFunction conn "typed_from_only" \_ -> pure [Only (2 :: Int64)]+ void (execute conn "COPY typed_first_rows FROM ? (FORMAT typed_from_only)" (Only file))+ query_ conn "SELECT value FROM typed_first_rows" >>= (@?= [Only (2 :: Int64)])+ assertSqlError+ "COPY TO is not supported for FORMAT \"typed_from_only\""+ (execute conn "COPY (SELECT 1) TO ? (FORMAT typed_from_only)" (Only file))+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])++-- | Give the reader of COPY FROM the name and the type of each column, in the+-- order of the table. DuckDB gives the same names and types.+testCopyFromColumns :: Assertion+testCopyFromColumns = withDirectory \directory -> withConnection ":memory:" \conn -> do+ calls <- newIORef ([] :: [CopyFromInfo])+ registerCopyFromFunction conn "typed_columns" \info -> do+ modifyIORef' calls (info :)+ pure [(7 :: Int64, "seven" :: Text, fromGregorian 2026 11 15)]+ let file = directory </> "columns.txt"+ writeFile file ""+ void (execute_ conn "CREATE TABLE typed_three (a BIGINT, b VARCHAR, c DATE)")+ void (execute conn "COPY typed_three FROM ? (FORMAT typed_columns)" (Only (Text.pack file)))+ query_ conn "SELECT column_name, data_type FROM information_schema.columns WHERE table_name = 'typed_three' ORDER BY ordinal_position"+ >>= (@?= [("a" :: Text, "BIGINT" :: Text), ("b", "VARCHAR"), ("c", "DATE")])+ readIORef calls >>= (@?= [CopyFromInfo file ["a", "b", "c"] [LogicalTypeIdBigInt, LogicalTypeIdVarchar, LogicalTypeIdDate]])+ query_ conn "SELECT a, b, c FROM typed_three" >>= (@?= [(7 :: Int64, "seven" :: Text, fromGregorian 2026 11 15)])++-- | Raise an error in each COPY TO function, in the COPY FROM function, and in+-- a lazy row after the first chunk. Each error fails its statement. The next+-- statement runs.+testCopyErrors :: Assertion+testCopyErrors = withDirectory \directory -> withConnection ":memory:" \conn -> do+ let file = Text.pack (directory </> "errors.txt")+ void (execute_ conn "CREATE TABLE typed_failed (value BIGINT)")+ forM_ ["bind", "init", "sink", "finalize"] \phase -> do+ registerFailingFormat conn phase+ assertSqlError+ ("typed_copy_error: " <> phase)+ (execute conn (Query ("COPY (SELECT 1) TO ? (FORMAT typed_error_" <> phase <> ")")) (Only file))+ query_ conn "SELECT 42" >>= (@?= [Only (42 :: Int64)])+ forM_ ["from", "row"] \phase -> do+ registerFailingFormat conn phase+ assertSqlError+ ("typed_copy_error: " <> phase)+ (execute conn (Query ("COPY typed_failed FROM ? (FORMAT typed_error_" <> phase <> ")")) (Only file))+ query_ conn "SELECT count(*) FROM typed_failed" >>= (@?= [Only (0 :: Int64)])+ -- The rows of a format must have one column for each column of the table.+ void (execute_ conn "CREATE TABLE typed_wide (a BIGINT, b BIGINT)")+ registerFailingFormat conn "none"+ assertSqlError+ "the table of COPY FROM has 2 columns, but a row of typed_error_none has 1"+ (execute conn "COPY typed_wide FROM ? (FORMAT typed_error_none)" (Only file))+ void (execute conn "COPY typed_failed FROM ? (FORMAT typed_error_none)" (Only file))+ query_ conn "SELECT count(*) FROM typed_failed" >>= (@?= [Only (3000 :: Int64)])++-- | Register a format on one connection. Use it from another connection.+testCopySharedDatabase :: Assertion+testCopySharedDatabase = withDirectory \directory -> withEnvironment \environment -> withInstance environment [] \instance' -> do+ -- An instance starts with no database. The case creates a table, so it+ -- attaches an in-memory database and makes it the default.+ attachDatabase instance' ":memory:" "typed_shared" []+ setDefaultDatabase instance' "typed_shared"+ let file = Text.pack (directory </> "shared.txt")+ withConnectionOn instance' \first -> withConnectionOn instance' \second -> do+ registerCopyFunction first "typed_shared_lines" linesTo (readLines . copyFromPath)+ registerCopyFunction second "typed_shared_lines" linesTo \_ -> ioError (userError "the second registration runs") :: IO [Only Int64]+ void (execute second "COPY (SELECT i FROM range(5) t(i)) TO ? (FORMAT typed_shared_lines)" (Only file))+ void (execute_ first "CREATE TABLE typed_shared_rows (value BIGINT)")+ void (execute second "COPY typed_shared_rows FROM ? (FORMAT typed_shared_lines)" (Only file))+ void (execute first "COPY typed_shared_rows FROM ? (FORMAT typed_shared_lines)" (Only file))+ query_ first "SELECT value FROM typed_shared_rows ORDER BY rowid" >>= (@?= map (Only . Just) ([0 .. 4] <> [0 .. 4 :: Int64]))++-- | Raise the input error of an empty format name, also on a closed connection.+-- The name check comes before the read of the state of the connection.+testCopyClosedName :: Assertion+testCopyClosedName = do+ closed <- open ":memory:"+ close closed+ assertInputErrorContaining "invalid copy function name" $+ registerCopyToFunction closed "" (\_ -> pure ()) (\_ -> pure ()) (\_ _ -> pure ()) (\_ -> pure ())+ assertInputErrorContaining "invalid copy function name" $+ registerCopyFunction closed "" linesTo (\_ -> pure [Only (1 :: Int64)])++-- | A COPY TO side that writes one BIGINT column as lines. A NULL is an empty line.+linesTo :: CopyTo () FilePath+linesTo =+ CopyTo+ { copyToBind = \info -> when (length (copyBindColumnTypes info) /= 1) (ioError (userError "typed_lines writes one column"))+ , copyToInit = \info -> writeFile (copyInitFilePath info) "" >> pure (copyInitFilePath info)+ , copyToSink = \info rows -> appendFile (copySinkGlobalState info) (unlines (map line rows))+ , copyToFinalize = \_ -> pure ()+ }+ where+ line [Field{fieldValue = FieldInt64 value}] = show value+ line [Field{fieldValue = FieldNull}] = ""+ line row = error ("typed_lines: unexpected row " <> show (map fieldValue row))++-- | Read a file of 'linesTo'. An empty line is NULL.+readLines :: FilePath -> IO [Only (Maybe Int64)]+readLines path = map (Only . parse . Text.unpack) . Text.lines <$> TextIO.readFile path+ where+ parse "" = Nothing+ parse line = Just (read line)++-- | Register the format @typed_error_@ with the phase as suffix. The function+-- of that phase raises an error. The phase @row@ makes the row at index 2500 raise+-- an error. That row is in the second chunk.+registerFailingFormat :: Connection -> Text -> IO ()+registerFailingFormat conn phase =+ registerCopyFunction conn ("typed_error_" <> phase) failing \_ -> failIn "from" >> pure rows+ where+ failIn name = when (phase == name) (ioError (userError ("typed_copy_error: " <> Text.unpack name)))+ failing =+ CopyTo+ (\_ -> failIn "bind")+ (\info -> failIn "init" >> writeFile (copyInitFilePath info) "")+ (\_ _ -> failIn "sink")+ (\_ -> failIn "finalize")+ rows = [Only (if phase == "row" && i == 2500 then error "typed_copy_error: row" else i) | i <- [0 .. 2999 :: Int64]]++-- | Read 8 files through a multi-file function over a typed table function.+-- DuckDB runs exec callbacks of one scan at the same time for a multi-file+-- function. Each row must come one time. The case sets 4 threads, so that it+-- does not use the default number of threads of the machine.+testTableMultiFile :: Assertion+testTableMultiFile = withDirectory \directory -> withConnection ":memory:" \conn -> do+ void (execute_ conn "SET threads = 4")+ createTableFunction conn "typed_numbers" ["value"] \(Only path) ->+ map (Only . read) . lines <$> readFile (Text.unpack path) :: IO [Only Int64]+ registerMultiFileFunction conn "typed_all_numbers" "typed_numbers" defaultMultiFileOptions+ -- A file of 10000 rows needs 5 output chunks. The multi-file function can+ -- then run more than one exec callback for the scan of one file.+ let numbers file = [file * 100000 + row | row <- [0 .. 9999]] :: [Int64]+ forM_ [0 .. 7] \file -> writeFile (directory </> ("numbers" <> show file <> ".txt")) (unlines (map show (numbers file)))+ forM_ [1 :: Int .. 5] \_ ->+ query conn "SELECT value FROM typed_all_numbers(?) ORDER BY value" (Only (Text.pack (directory </> "numbers*.txt")))+ >>= (@?= map Only (concatMap numbers [0 .. 7]))+ -- count(*) uses no column. The typed table function declares projection+ -- pushdown, so the multi-file function can run it.+ query conn "SELECT count(*) FROM typed_all_numbers(?)" (Only (Text.pack (directory </> "numbers*.txt")))+ >>= (@?= [Only (80000 :: Int64)])++-- | Create and remove a unique directory for the files of a test.+withDirectory :: (FilePath -> IO a) -> IO a+withDirectory = bracket create removePathForcibly+ where+ create = do+ root <- getTemporaryDirectory+ (path, handle) <- openTempFile root "duckdb-typed-copy"+ hClose handle+ removeFile path+ createDirectory path+ pure path++-- | The registration kinds that have no other check of the release of their+-- callback closure. Each action registers one callback that reads the token,+-- runs the callback one time, and closes the database.+closureKinds :: [(String, IORef Int64 -> IO ())]+closureKinds =+ [+ ( "typed aggregate"+ , \token -> withConnection ":memory:" \conn -> do+ createAggregate+ conn+ "closure_sum"+ sumAggregate{aggregateStep = \total (Only value) -> (\factor -> total + factor * value) <$> readIORef token}+ query_ conn "SELECT closure_sum(i) FROM range(10) t(i)" >>= (@?= [Only (45 :: Int64)])+ )+ ,+ ( "cast"+ , \token -> withConnection ":memory:" \conn -> do+ registerCustomType conn "closure_type" (TypeExpression "BIGINT")+ -- The callback reads the token in each call, so its closure keeps the token.+ R.registerCast conn (TypeExpression "BIGINT") (TypeExpression "closure_type") Nothing \info _ ->+ readIORef token >> writeNullRows info+ query_ conn "SELECT CAST(41::BIGINT AS closure_type) IS NULL" >>= (@?= [Only True])+ )+ ,+ ( "connection replacement scan"+ , \token -> withConnection ":memory:" \conn -> do+ registerReplacementScan conn (closureScan token)+ query_ conn "SELECT value FROM closure_table" >>= (@?= [Only (42 :: Int64)])+ )+ ,+ ( "instance replacement scan"+ , \token -> withEnvironment \environment -> withInstance environment [] \instance' -> do+ registerInstanceReplacementScan instance' (closureScan token)+ withConnectionOn instance' \conn -> query_ conn "SELECT value FROM closure_table" >>= (@?= [Only (42 :: Int64)])+ )+ ,+ ( "custom option update"+ , \token -> withConnection ":memory:" \conn -> do+ registerCustomOption conn (customOption "closure_option" (TypeExpression "BIGINT")){customOptionUpdate = Just \_ -> void (readIORef token)}+ void (execute_ conn "SET closure_option = 5")+ )+ ,+ ( "COPY format"+ , \token -> withConnection ":memory:" \conn -> do+ let use = void (readIORef token)+ registerCopyFunction+ conn+ "closure_copy"+ CopyTo{copyToBind = const use, copyToInit = const use, copyToSink = \_ _ -> use, copyToFinalize = const use}+ (\_ -> use >> pure [Only (42 :: Int64)])+ void (execute_ conn "COPY (SELECT 1::BIGINT) TO 'unused' (FORMAT closure_copy)")+ void (execute_ conn "CREATE TABLE closure_rows (value BIGINT)")+ void (execute_ conn "COPY closure_rows FROM 'unused' (FORMAT closure_copy)")+ query_ conn "SELECT value FROM closure_rows" >>= (@?= [Only (42 :: Int64)])+ )+ ,+ ( "typed pragma"+ , \token -> withConnection ":memory:" \conn -> do+ registerPragmaFunction conn "closure_pragma" [] (ExecPragma \() -> void (readIORef token))+ void (execute_ conn "PRAGMA closure_pragma")+ )+ ,+ ( "log sink"+ , \token -> withConnection ":memory:" \conn -> do+ Logging.registerLogSink conn "closure_log" (\_ -> void (readIORef token)) (void (readIORef token))+ void+ (execute_ conn "CALL enable_logging(); SET logging_storage = 'closure_log'; SELECT write_log('closure'); SET logging_storage = 'memory'")+ )+ ]++-- | A replacement scan that reads the token and replaces @closure_table@.+closureScan :: IORef Int64 -> [Text] -> IO (Maybe Replacement)+closureScan token parts = do+ void (readIORef token)+ pure (if parts == ["closure_table"] then Just (ReplaceWithQuery "SELECT 42::BIGINT AS value") else Nothing)
+ test/UserCastTests.hs view
@@ -0,0 +1,350 @@+{-# LANGUAGE BlockArguments #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}++-- |+-- Module : UserCastTests+-- Description : Check that a cast that a user registers does not change what the library reads.+--+-- Each case registers a cast that writes NULL for each row and counts its+-- calls. Then the case reads a value whose reader uses the built-in cast for+-- the same pair of types. The value must be correct, and the user cast must+-- not run. See Note [User casts and the casts of the library] in+-- "Database.DuckDB.Simple.Materialize".+module UserCastTests (userCastTests, geometryArgumentTests) where++import Control.Exception (throwIO, try)+import Control.Monad (void)+import Data.Array (listArray, (!))+import qualified Data.ByteString as BS+import qualified Data.Geometry as G+import Data.IORef (IORef, atomicModifyIORef', modifyIORef', newIORef, readIORef, writeIORef)+import Data.Int (Int64)+import qualified Data.Map.Strict as Map+import Data.Maybe (fromMaybe)+import Data.Text (Text)+import qualified Data.Text as Text+import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Config as Config+import qualified Database.DuckDB.Simple.Copy as Copy+import Database.DuckDB.Simple.FromField (Field (..), FieldValue (..), returnError)+import Database.DuckDB.Simple.Geometry (RawGeometry (..))+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructField (..), StructValue (..))+import Database.DuckDB.Simple.Pragma (PragmaCallback (..), registerPragmaFunction)+import qualified Database.DuckDB.Simple.Streaming as Streaming+import Database.DuckDB.Simple.ToField (ToDuckValue (..))+import qualified Database.DuckDB.Simple.V2.Registration as R+import Database.DuckDB.Simple.Variant (Variant (..), variantObject)+import Test.Tasty (TestTree, testGroup)+import Test.Tasty.HUnit+import TestUtils (writeNullRows)++-- | Check the readers and binders of the library under user casts.+userCastTests :: TestTree+userCastTests =+ testGroup+ "user casts"+ [ testCase "a cast from GEOMETRY to BLOB does not change a GEOMETRY result" $ withConnection ":memory:" \conn -> do+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ (query_ conn "SELECT 'POINT (1 2)'::GEOMETRY" :: IO [Only (Maybe RawGeometry)]) >>= (@?= [Only (Just point)])+ (query_ conn "SELECT 'POINT (1 2)'::GEOMETRY" :: IO [Only (Maybe G.Geometry)]) >>= (@?= [Only (Just shape)])+ (query_ conn "SELECT {'g': 'POINT (1 2)'::GEOMETRY}" :: IO [Only FieldValue])+ >>= (@?= [Only (FieldStruct (oneField "g" (LogicalTypeGeometry Nothing) (FieldGeometry point)))])+ Streaming.fold_ conn "SELECT 'POINT (1 2)'::GEOMETRY FROM range(3)" [] (\seen (Only value) -> pure (value : seen))+ >>= (@?= replicate 3 point)+ -- A GEOMETRY with a CRS has another type, so this cast is not used for it.+ (query conn "SELECT ST_SetCRS(ST_GeomFromWKB(?), 'local grid')" (Only pointWKB) :: IO [Only RawGeometry])+ >>= (@?= [Only point{rawGeometryCRS = Just "local grid"}])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from GEOMETRY('OGC:CRS84') to BLOB does not change a GEOMETRY('OGC:CRS84') result" $+ withConnection ":memory:" \conn -> do+ calls <- nullCast conn (R.StructuredType (LogicalTypeGeometry (Just "OGC:CRS84"))) (R.TypeExpression "BLOB")+ (query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry])+ >>= (@?= [Only point{rawGeometryCRS = Just "OGC:CRS84"}])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from GEOMETRY to BLOB does not change the value of a GEOMETRY option" $+ withConnection ":memory:" \conn -> do+ Config.registerCustomOption+ conn+ (Config.customOption "user_cast_shape" (R.TypeExpression "GEOMETRY")){Config.customOptionDefault = Just (R.defaultValue shape)}+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ Config.getConfigDefaultValue conn "user_cast_shape" >>= (@?= FieldGeometry point)+ (fst <$> Config.getConfigValue conn "user_cast_shape") >>= (@?= FieldGeometry point)+ readIORef calls >>= (@?= 0)+ , testCase "a cast from GEOMETRY to BLOB does not change a GEOMETRY argument of a table function" $+ withConnection ":memory:" \conn -> do+ createTableFunction conn "user_cast_shapes" ["same"] \(Only argument) -> pure [Only (argument == Just shape)]+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ query_ conn "SELECT same FROM user_cast_shapes('POINT (1 2)'::GEOMETRY)" >>= (@?= [Only True])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from GEOMETRY to BLOB does not change a GEOMETRY row of a COPY TO function" $+ withConnection ":memory:" \conn -> do+ rows <- newIORef []+ Copy.registerCopyToFunction conn "user_cast_copy" (\_ -> pure ()) (\_ -> pure ()) (\_ batch -> modifyIORef' rows (<> batch)) (\_ -> pure ())+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ void (execute_ conn "COPY (SELECT 'POINT (1 2)'::GEOMETRY AS g) TO 'unused' (FORMAT user_cast_copy)")+ (map (map fieldValue) <$> readIORef rows) >>= (@?= [[FieldGeometry point]])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from GEOMETRY to BLOB does not change a GEOMETRY argument of a pragma" $+ withConnection ":memory:" \conn -> do+ seen <- newIORef Nothing+ registerPragmaFunction+ conn+ "user_cast_pragma"+ [R.parameter "shape" (R.TypeExpression "GEOMETRY")]+ (ExecPragma \(Only argument) -> writeIORef seen argument)+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ void (execute_ conn "PRAGMA user_cast_pragma('POINT (1 2)')")+ readIORef seen >>= (@?= Just shape)+ readIORef calls >>= (@?= 0)+ , testCase "a cast from GEOMETRY to BLOB does not change the value of a GEOMETRY option update" $+ withConnection ":memory:" \conn -> do+ seen <- newIORef []+ Config.registerCustomOption+ conn+ (Config.customOption "user_cast_update" (R.TypeExpression "GEOMETRY"))+ { Config.customOptionUpdate = Just \value -> modifyIORef' seen (value :)+ }+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ void (execute_ conn "SET user_cast_update = 'POINT (1 2)'")+ readIORef seen >>= (@?= [FieldGeometry point])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from VARIANT to BIGINT does not change a BIGINT payload of a VARIANT result" $+ withConnection ":memory:" \conn -> do+ calls <- nullCast conn (R.TypeExpression "VARIANT") (R.TypeExpression "BIGINT")+ createFunction conn "user_cast_variant_arg" ((== Variant (FieldInt64 42)) :: Variant -> Bool)+ query_ conn "SELECT user_cast_variant_arg(42::BIGINT::VARIANT)" >>= (@?= [Only True])+ query_ conn "SELECT 42::BIGINT::VARIANT" >>= (@?= [Only (Variant (FieldInt64 42))])+ query_ conn "SELECT {'n': 42::BIGINT}::VARIANT" >>= (@?= [Only (Variant (variantObject [("n", FieldInt64 42)]))])+ query_ conn "SELECT [1, 2]::BIGINT[]::VARIANT" >>= (@?= [Only (Variant (FieldList [FieldInt64 1, FieldInt64 2]))])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from GEOMETRY to BLOB does not change a GEOMETRY payload of a VARIANT result" $+ withConnection ":memory:" \conn -> do+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ query_ conn "SELECT 'POINT (1 2)'::GEOMETRY::VARIANT" >>= (@?= [Only (Variant (FieldGeometry point))])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from a VARIANT list to a STRUCT list does not change a VARIANT object" $+ withConnection ":memory:" \conn -> do+ let variant = LogicalTypeScalar LogicalTypeIdVariant+ objects = LogicalTypeList (LogicalTypeStruct (listArray (0, 0) [StructField "a" variant]))+ calls <- nullCast conn (R.StructuredType (LogicalTypeList variant)) (R.StructuredType objects)+ query_ conn "SELECT {'a': 1}::VARIANT" >>= (@?= [Only (Variant (variantObject [("a", FieldInt32 1)]))])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from VARIANT to BIGINT does not change a VARIANT row of a COPY TO function" $+ withConnection ":memory:" \conn -> do+ rows <- newIORef []+ Copy.registerCopyToFunction+ conn+ "user_cast_variant_copy"+ (\_ -> pure ())+ (\_ -> pure ())+ (\_ batch -> modifyIORef' rows (<> batch))+ (\_ -> pure ())+ calls <- nullCast conn (R.TypeExpression "VARIANT") (R.TypeExpression "BIGINT")+ void (execute_ conn "COPY (SELECT 42::BIGINT::VARIANT AS v) TO 'unused' (FORMAT user_cast_variant_copy)")+ (map (map fieldValue) <$> readIORef rows) >>= (@?= [[FieldInt64 42]])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from VARCHAR to GEOMETRY does not run when a custom CRS binds" $+ withEnvironment \environment -> withInstance environment [] \instance' -> withConnectionOn instance' \conn -> do+ calls <- nullCast conn (R.TypeExpression "VARCHAR") (R.TypeExpression "GEOMETRY")+ -- The bind resolves the custom CRS without a VARCHAR cast.+ withConnectionOn instance' \other -> do+ let grid = point{rawGeometryCRS = Just "local grid"}+ struct = oneField "g" (LogicalTypeGeometry (Just "local grid")) (FieldGeometry grid)+ query other "SELECT ?" (Only struct) >>= (@?= [Only struct])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from BLOB to GEOMETRY does not change a bound raw geometry" $ withConnection ":memory:" \conn -> do+ calls <- nullCast conn (R.TypeExpression "BLOB") (R.TypeExpression "GEOMETRY")+ let struct = oneField "g" (LogicalTypeGeometry Nothing) (FieldGeometry point)+ query conn "SELECT ?" (Only point) >>= (@?= [Only point])+ query conn "SELECT ?" (Only struct) >>= (@?= [Only struct])+ query conn "SELECT ?" (Only (Variant (FieldGeometry point))) >>= (@?= [Only (Variant (FieldGeometry point))])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from BLOB to GEOMETRY('OGC:CRS84') does not change a bound raw geometry with that CRS" $+ withConnection ":memory:" \conn -> do+ let crs84 = LogicalTypeGeometry (Just "OGC:CRS84")+ struct = oneField "g" crs84 (FieldGeometry point{rawGeometryCRS = Just "OGC:CRS84"})+ calls <- nullCast conn (R.TypeExpression "BLOB") (R.StructuredType crs84)+ query conn "SELECT ?" (Only struct) >>= (@?= [Only struct])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from VARCHAR to GEOMETRY does not change a bound Geometry" $ withConnection ":memory:" \conn -> do+ calls <- nullCast conn (R.TypeExpression "VARCHAR") (R.TypeExpression "GEOMETRY")+ query conn "SELECT ?" (Only shape) >>= (@?= [Only shape])+ query conn "SELECT ?" (Only [shape, shape]) >>= (@?= [Only [shape, shape]])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from VARCHAR to an ENUM type does not change a bound ENUM value" $ withConnection ":memory:" \conn -> do+ let entries = LogicalTypeEnum (listArray (0, 1) ["x", "y"])+ struct = oneField "e" entries (FieldEnum 1 "y")+ calls <- nullCast conn (R.TypeExpression "VARCHAR") (R.StructuredType entries)+ query conn "SELECT ?" (Only struct) >>= (@?= [Only struct])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from BIGINT to VARIANT does not change a bound Variant" $ withConnection ":memory:" \conn -> do+ calls <- nullCast conn (R.TypeExpression "BIGINT") (R.TypeExpression "VARIANT")+ let list = Variant (FieldList [FieldInt64 1, FieldInt64 2])+ query conn "SELECT ?" (Only (Variant (FieldInt64 5))) >>= (@?= [Only (Variant (FieldInt64 5))])+ query conn "SELECT ?" (Only list) >>= (@?= [Only list])+ readIORef calls >>= (@?= 0)+ , testCase "a cast from VARCHAR to GEOMETRY does not change a Geometry row of a table function" $+ withConnection ":memory:" \conn -> do+ createTableFunction conn "user_cast_rows" ["g"] \() -> pure [Only shape]+ calls <- nullCast conn (R.TypeExpression "VARCHAR") (R.TypeExpression "GEOMETRY")+ query_ conn "SELECT g FROM user_cast_rows()" >>= (@?= [Only (Just shape)])+ readIORef calls >>= (@?= 0)+ , testCase "a bound raw geometry keeps its typed NULL and its error for bytes that are not WKB" $+ withConnection ":memory:" \conn -> do+ let crs84 = LogicalTypeGeometry (Just "OGC:CRS84")+ struct = oneField "g" crs84 FieldNull+ query conn "SELECT ?" (Only struct) >>= (@?= [Only struct])+ try (query conn "SELECT ?" (Only (RawGeometry (BS.pack [1, 2, 3]) Nothing)) :: IO [Only RawGeometry]) >>= \case+ Left SQLError{sqlErrorMessage} ->+ assertBool (show sqlErrorMessage) ("Could not convert BLOB to GEOMETRY" `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure ("expected an error, got " <> show rows)+ ]++-- | Check the GEOMETRY arguments of a scalar function and a typed aggregate.+-- Each case also registers a cast from GEOMETRY to BLOB that writes NULL, and+-- that cast must not run.+geometryArgumentTests :: TestTree+geometryArgumentTests =+ testGroup+ "geometry arguments"+ [ testCase "a scalar function reads a Geometry argument" $ withConnection ":memory:" \conn -> do+ createFunction conn "hs_point_x" pointX+ createFunction conn "hs_maybe_x" (fmap pointX :: Maybe G.Geometry -> Maybe Double)+ createFunction conn "hs_count_shapes" (\(shapes :: [G.Geometry]) -> fromIntegral (length shapes) :: Int64)+ createFunction conn "hs_placed_x" (\(Placed g) -> pointX g)+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ query_ conn "SELECT hs_point_x('POINT (1 2)'::GEOMETRY)" >>= (@?= [Only (1 :: Double)])+ query_ conn "SELECT hs_point_x('POINT (3 4)'::GEOMETRY('OGC:CRS84'))" >>= (@?= [Only (3 :: Double)])+ query_ conn "SELECT hs_maybe_x(NULL::GEOMETRY)" >>= (@?= [Only (Nothing :: Maybe Double)])+ query_ conn "SELECT hs_count_shapes(['POINT (1 2)'::GEOMETRY, 'POINT (3 4)'::GEOMETRY])" >>= (@?= [Only (2 :: Int64)])+ query_ conn "SELECT hs_placed_x({'name': 'a', 'shape': 'POINT (1 2)'::GEOMETRY})" >>= (@?= [Only (1 :: Double)])+ query_ conn "SELECT sum(hs_point_x(('POINT (' || i || ' 0)')::GEOMETRY)) FROM range(5000) t(i)" >>= (@?= [Only (12497500 :: Double)])+ readIORef calls >>= (@?= 0)+ withConnection ":memory:" \grid -> do+ createFunction grid "hs_point_x" pointX+ query grid "SELECT hs_point_x(ST_SetCRS(ST_GeomFromWKB(?), 'local grid'))" (Only pointWKB) >>= (@?= [Only (1 :: Double)])+ , testCase "a scalar function reads its arguments through a selection and from a constant vector" $ withConnection ":memory:" \conn -> do+ createFunction conn "hs_point_x" pointX+ createFunction conn "hs_maybe_x" (fmap pointX :: Maybe G.Geometry -> Maybe Double)+ createFunction conn "hs_count_shapes" (\(shapes :: [G.Geometry]) -> fromIntegral (length shapes) :: Int64)+ _ <- execute_ conn "CREATE TABLE selected AS SELECT i, ('POINT (' || i || ' 2)')::GEOMETRY AS g FROM range(3000) t(i)"+ let kept = [i | i <- [0 .. 2999 :: Int64], i `mod` 7 == 0]+ -- A WHERE filter gives the function a DICTIONARY vector.+ query_ conn "SELECT count(*), count(DISTINCT hs_point_x(g)), sum(hs_point_x(g)) FROM selected WHERE i % 7 = 0"+ >>= (@?= [(fromIntegral (length kept) :: Int64, fromIntegral (length kept) :: Int64, fromIntegral (sum kept) :: Double)])+ query_ conn "SELECT sum(hs_count_shapes([g, g])) FROM selected WHERE i % 7 = 0" >>= (@?= [Only (2 * fromIntegral (length kept) :: Int64)])+ query_ conn "SELECT count(hs_maybe_x(CASE WHEN i = 7 THEN NULL ELSE g END)) FROM selected WHERE i % 7 = 0"+ >>= (@?= [Only (fromIntegral (length kept) - 1 :: Int64)])+ -- A constant argument over FROM gives the function a CONSTANT vector.+ query_ conn "SELECT hs_point_x('POINT (7 8)'::GEOMETRY) FROM range(3)" >>= (@?= replicate 3 (Only (7 :: Double)))+ , testCase "a registration cannot declare a custom CRS of the connection" $+ withConnection ":memory:" \grid -> do+ try+ ( R.registerCast grid (R.StructuredType (LogicalTypeGeometry (Just "local grid"))) (R.TypeExpression "BLOB") Nothing \info _ -> writeNullRows info+ )+ >>= \case+ Left SQLError{sqlErrorMessage} ->+ assertBool (show sqlErrorMessage) ("unrecognized coordinate system 'local grid'" `Text.isInfixOf` sqlErrorMessage)+ Right () -> assertFailure "expected an error at registration"+ -- A plain GEOMETRY argument reads the value without its CRS.+ createFunction grid "hs_crs" (\(PlainGeometry raw) -> fromMaybe "no CRS" (rawGeometryCRS raw) :: Text)+ query grid "SELECT hs_crs(ST_SetCRS(ST_GeomFromWKB(?), 'local grid'))" (Only pointWKB) >>= (@?= [Only ("no CRS" :: Text)])+ , testCase "a typed aggregate reads a Geometry argument on four threads" $+ withConnectionWithConfig ":memory:" [("threads", "4")] \conn -> do+ _ <-+ execute_ conn "CREATE TABLE shapes AS SELECT i % 100 AS k, ('POINT (' || i || ' ' || (i + 1) || ')')::GEOMETRY AS g FROM range(200000) t(i)"+ let total = 19999900000 :: Double+ createAggregate+ conn+ "hs_sum_x"+ Aggregate+ { aggregateInitial = 0 :: Double+ , aggregateStep = \sumX (Only g) -> pure (sumX + pointX g)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }+ createAggregate+ conn+ "hs_sum_placed_x"+ Aggregate+ { aggregateInitial = 0 :: Double+ , aggregateStep = \sumX (Only (Placed g)) -> pure (sumX + pointX g)+ , aggregateCombine = \a b -> pure (a + b)+ , aggregateResult = pure+ }+ calls <- nullCast conn (R.TypeExpression "GEOMETRY") (R.TypeExpression "BLOB")+ query_ conn "SELECT hs_sum_x(g) FROM shapes" >>= (@?= [Only total])+ query_ conn "SELECT hs_sum_placed_x({'name': 'p', 'shape': g}) FROM shapes" >>= (@?= [Only total])+ query_ conn "SELECT count(*), sum(s) FROM (SELECT k, hs_sum_x(g) AS s FROM shapes GROUP BY k)" >>= (@?= [(100 :: Int64, total)])+ query_ conn "SELECT DISTINCT hs_sum_x(g) OVER () FROM shapes" >>= (@?= [Only total])+ query_ conn "SELECT hs_sum_x(DISTINCT g) FROM (SELECT g FROM shapes UNION ALL SELECT g FROM shapes)" >>= (@?= [Only total])+ query_ conn "SELECT hs_sum_x(g) FROM (VALUES (NULL::GEOMETRY), ('POINT (5 6)'::GEOMETRY)) t(g)" >>= (@?= [Only (5 :: Double)])+ query_ conn "SELECT hs_sum_x('POINT (7 8)'::GEOMETRY('OGC:CRS84'))" >>= (@?= [Only (7 :: Double)])+ readIORef calls >>= (@?= 0)+ ]++-- | The x coordinate of a point, or -1 for another shape.+pointX :: G.Geometry -> Double+pointX = \case+ G.PointGeometry (G.PointXY (G.XY x _)) -> x+ _ -> -1++-- | A STRUCT argument with a GEOMETRY field. A record through ViaDuckDB cannot+-- hold a Geometry field, because Geometry has no DuckValue instance. Thus the+-- instances are written here. The reader reads only the field shape.+newtype Placed = Placed G.Geometry++instance DuckDBColumnType Placed where+ duckdbColumnTypeFor _ = "STRUCT(name VARCHAR, shape GEOMETRY)"++instance FromField Placed where+ fromField f = case fieldValue f of+ FieldStruct struct+ | Just index <- Map.lookup "shape" (structValueIndex struct) ->+ Placed <$> fromField f{fieldValue = structFieldValue (structValueFields struct ! index)}+ _ -> returnError Incompatible f "expected STRUCT(name VARCHAR, shape GEOMETRY)"++-- | An aggregate gets the SQL type of an argument column from this class. The+-- cases do not write a Placed value.+instance ToDuckValue Placed where+ toDuckValue _ _ = throwIO (userError "a Placed value is only an argument")++-- | A GEOMETRY argument that reads the CRS of the value. RawGeometry has no+-- DuckDBColumnType instance, so the instances are written here.+newtype PlainGeometry = PlainGeometry RawGeometry++instance DuckDBColumnType PlainGeometry where+ duckdbColumnTypeFor _ = "GEOMETRY"++instance FromField PlainGeometry where+ fromField f = PlainGeometry <$> fromField f++-- | Register a cast that writes NULL for each row, and give its call count.+nullCast :: Connection -> R.TypeDeclaration -> R.TypeDeclaration -> IO (IORef Int)+nullCast conn source target = do+ calls <- newIORef 0+ R.registerCast conn source target Nothing \info _ -> do+ atomicModifyIORef' calls (\count -> (count + 1, ()))+ writeNullRows info+ pure calls++-- | WKB of POINT (1 2).+pointWKB :: BS.ByteString+pointWKB = BS.pack [1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 240, 63, 0, 0, 0, 0, 0, 0, 0, 64]++point :: RawGeometry+point = RawGeometry pointWKB Nothing++shape :: G.Geometry+shape = G.PointGeometry (G.PointXY (G.XY 1 2))++oneField :: Text -> LogicalTypeRep -> FieldValue -> StructValue FieldValue+oneField name rep value =+ StructValue+ { structValueFields = listArray (0, 0) [StructField name value]+ , structValueTypes = listArray (0, 0) [StructField name rep]+ , structValueIndex = Map.singleton name 0+ }
test/ValueRegressionTests.hs view
@@ -1,363 +1,2771 @@ {-# LANGUAGE BlockArguments #-}-{-# LANGUAGE DeriveGeneric #-}-{-# LANGUAGE DerivingVia #-}-{-# LANGUAGE NamedFieldPuns #-}-{-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE ScopedTypeVariables #-}---- | Regression tests for exact values and controlled conversion failures.-module ValueRegressionTests (main, valueRegressionTests) where--import Control.Exception (IOException, bracket, try)-import Control.Monad (forM_)-import Data.Array (Array, listArray)-import qualified Data.ByteString as BS-import Data.Int (Int32, Int64, Int8)-import qualified Data.Map.Strict as Map-import Data.Ratio ((%))-import Data.String (fromString)-import Data.Text (Text)-import Data.Time.Calendar (Day, addDays, fromGregorian)-import Data.Time.Clock (UTCTime (..))-import Data.Time.Clock.POSIX (posixSecondsToUTCTime)-import Data.Time.LocalTime (LocalTime (..), TimeOfDay (..), minutesToTimeZone, utc, utcToLocalTime)-import Data.Word (Word16, Word32, Word8)-import Database.DuckDB.FFI-import Database.DuckDB.Simple-import Database.DuckDB.Simple.FromField (BitString (..), DecimalValue (..), FieldValue (..), TimeWithZone (..), bsFromBool)-import Database.DuckDB.Simple.Generic (ViaDuckDB (..), genericFromFieldValue, genericToStructValue)-import Database.DuckDB.Simple.Internal (destroyValue, withConnectionHandle)-import Database.DuckDB.Simple.LogicalRep-import Database.DuckDB.Simple.Time (Unbounded (..))-import Database.DuckDB.Simple.ToField (ToDuckValue (..))-import Foreign.C.String (withCString)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (nullPtr)-import Foreign.Storable (peek, poke)-import GHC.Generics (Generic)-import Test.Tasty (TestTree, defaultMain, testGroup)-import Test.Tasty.HUnit-import TestUtils (assertFailureIO)---- | Native timestamps used to test the full storage range.-data NativeTimestamp = Seconds Int64 | Milliseconds Int64 | Microseconds Int64 | Nanoseconds Int64- deriving (Show)--instance DuckDBColumnType NativeTimestamp where- duckdbColumnTypeFor _ = "TIMESTAMP"--instance ToDuckValue NativeTimestamp where- toDuckValue (Seconds value) = c_duckdb_create_timestamp_s (DuckDBTimestampS value)- toDuckValue (Milliseconds value) = c_duckdb_create_timestamp_ms (DuckDBTimestampMs value)- toDuckValue (Microseconds value) = c_duckdb_create_timestamp (DuckDBTimestamp value)- toDuckValue (Nanoseconds value) = c_duckdb_create_timestamp_ns (DuckDBTimestampNs value)--instance ToField NativeTimestamp---- | An array element with a ToField instance and no ToDuckValue instance.-newtype DelegatedInteger = DelegatedInteger Int64--instance DuckDBColumnType DelegatedInteger where- duckdbColumnTypeFor _ = "BIGINT"--instance ToField DelegatedInteger where- toField (DelegatedInteger value) = toField value---- | Bind an array through its connection-free value.-newtype Scores = Scores (Array Int Int64)- deriving (Show)--instance DuckDBColumnType Scores where- duckdbColumnTypeFor _ = "BIGINT[]"--instance ToDuckValue Scores where- toDuckValue (Scores values) = toDuckValue values--instance ToField Scores---- | Record with identical field types to detect positional decoding.-data NamedRecord = NamedRecord {firstValue :: Int64, secondValue :: Int64}- deriving (Eq, Show, Generic)---- | Sum with a payload to test NULL member handling.-data NullableSum = EmptyMember | DataMember Int64- deriving (Eq, Show, Generic)---- | A generic record for array parameters.-data Pair = Pair {pairName :: Text, pairValue :: Int64}- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB Pair)---- | Generic nullary constructors retain their existing UNION schema.-data Colour = Red | Blue- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB Colour)---- | Run this module without the main integration suite.-main :: IO ()-main = defaultMain valueRegressionTests---- | Focused value regressions that also run against the baseline library.-valueRegressionTests :: TestTree-valueRegressionTests =- testGroup "value regressions" $- [ testCase "REAL decodes to Float and Double" $ withConnection ":memory:" \conn -> do- (query_ conn "SELECT 1.25::REAL" :: IO [Only Float]) >>= (@?= [Only 1.25])- (query_ conn "SELECT 1.25::REAL" :: IO [Only Double]) >>= (@?= [Only 1.25])- , testCase "Float parameters still decode as Text" $ withConnection ":memory:" \conn ->- (query conn "SELECT ?" (Only (1.25 :: Float)) :: IO [Only Text]) >>= (@?= [Only "1.25"])- , testCase "generic nullary UNION constructors round trip through DuckDB" $ withConnection ":memory:" \conn ->- forM_ [Red, Blue] \colour ->- (query conn "SELECT ?" (Only colour) :: IO [Only Colour]) >>= (@?= [Only colour])- , testCase "UNION NULL payload retains its member type" $ withConnection ":memory:" \conn -> do- let members = listArray (0, 1) [UnionMemberType "number" (LogicalTypeScalar DuckDBTypeBigInt), UnionMemberType "text" (LogicalTypeScalar DuckDBTypeVarchar)]- original = UnionValue 0 "number" FieldNull members- (query conn "SELECT ?" (Only original) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only original])- , testCase "array elements need only a ToField instance" $ withConnection ":memory:" \conn -> do- let values = listArray (0 :: Int, 2) [Nothing, Just (DelegatedInteger 41), Just (DelegatedInteger 99)]- expected = listArray (0, 2) [Nothing, Just 41, Just 99] :: Array Int (Maybe Int64)- (query conn "SELECT ?" (Only values) :: IO [Only (Array Int (Maybe Int64))]) >>= (@?= [Only expected])- , testCase "array children use default ToDuckValue instances" $ withConnection ":memory:" \conn -> do- let units = [-1, 0, 1234567]- values = listArray (0 :: Int, 2) (map Microseconds units)- expected = listArray (0, 2) (map (utcToLocalTime utc . posixSecondsToUTCTime . fromRational . (% 1000000) . toInteger) units)- (query conn "SELECT ?" (Only values) :: IO [Only (Array Int LocalTime)]) >>= (@?= [Only expected])- , testCase "arrays bind through ToDuckValue without a connection" $ withConnection ":memory:" \conn -> do- let values = listArray (0, 2) [1, -2, 3]- (query conn "SELECT typeof(?), ?" (Scores values, Scores values) :: IO [(Text, Array Int Int64)]) >>= (@?= [("BIGINT[3]", values)])- , testCase "arrays of STRUCT, record, and ARRAY elements take the type of a present element" $ withConnection ":memory:" \conn -> do- [Only struct] <- query_ conn "SELECT {'a': 1, 'b': 'x'}" :: IO [Only (StructValue FieldValue)]- let structs = listArray (0, 1) [struct, struct]- (query conn "SELECT typeof(?), ?" (structs, structs) :: IO [(Text, Array Int (StructValue FieldValue))]) >>= (@?= [("STRUCT(a INTEGER, b VARCHAR)[2]", structs)])- let pairs = listArray (0, 2) [Nothing, Just (Pair "x" 1), Just (Pair "y" 2)]- (query conn "SELECT typeof(?), ?" (pairs, pairs) :: IO [(Text, Array Int (Maybe Pair))]) >>= (@?= [("STRUCT(pairName VARCHAR, pairValue BIGINT)[3]", pairs)])- let nested = listArray (0, 1) [listArray (0, 1) [1, 2], listArray (0, 1) [3, 4]] :: Array Int (Array Int Int64)- (query conn "SELECT typeof(?), ?" (nested, nested) :: IO [(Text, Array Int (Array Int Int64))]) >>= (@?= [("BIGINT[2][2]", nested)])- , testCase "empty arrays of record elements have no element type" $ withConnection ":memory:" \conn ->- assertFailureIO (query conn "SELECT ?" (Only (listArray (0, -1) [] :: Array Int Pair)) :: IO [Only Text])- , testCase "composite arrays reject schemas that would discard fields or digits" $ withConnection ":memory:" \conn -> do- forM_- [ ("SELECT {'a': 1::INTEGER}", "SELECT {'a': 2::INTEGER, 'b': 3::INTEGER}")- , ("SELECT {'a': 1::INTEGER}", "SELECT {'a': 1.6::DOUBLE}")- , ("SELECT {'a': 1.2::DECIMAL(4,1)}", "SELECT {'a': 1.26::DECIMAL(4,2)}")- , ("SELECT {'a': [1.2::DECIMAL(4,1)]}", "SELECT {'a': [1.26::DECIMAL(4,2)]}")- ]- \(firstQuery, secondQuery) -> do- [Only first] <- query_ conn firstQuery :: IO [Only (StructValue FieldValue)]- [Only second] <- query_ conn secondQuery :: IO [Only (StructValue FieldValue)]- forM_ [[Nothing, Just first, Just second], [Just second, Nothing, Just first]] \elements -> do- let values = listArray (0 :: Int, 2) elements- assertFailureIO (query conn "SELECT ?" (Only values) :: IO [Only FieldValue])- assertFailureIO (bracket (toDuckValue values) destroyValue (const (pure ())))- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])- , testCase "GEOMETRY decodes as well-known binary" $ withConnection ":memory:" \conn -> do- [Only bytes] <- query_ conn "SELECT 'POINT(1 2)'::GEOMETRY" :: IO [Only BS.ByteString]- BS.length bytes @?= 21- (query conn "SELECT ST_AsText(ST_GeomFromWKB(?))" (Only bytes) :: IO [Only Text]) >>= (@?= [Only "POINT (1 2)"])- , testCase "Float parameter retains FLOAT type" $ withConnection ":memory:" \conn -> do- (query conn "SELECT typeof(?), ?" (1.25 :: Float, 1.25 :: Float) :: IO [(Text, Float)]) >>= (@?= [("FLOAT", 1.25)])- , testCase "Float special values survive decoding" $ withConnection ":memory:" \conn -> do- [Only value] <- query_ conn "SELECT 'NaN'::REAL" :: IO [Only Float]- assertBool "expected NaN" (isNaN value)- [Only value'] <- query_ conn "SELECT 'Infinity'::REAL" :: IO [Only Double]- assertBool "expected positive infinity" (isInfinite value' && value' > 0)- , testCase "finite Double overflow to Float fails" $ withConnection ":memory:" \conn ->- assertConversionError (query_ conn "SELECT 1e300::DOUBLE" :: IO [Only Float])- , testCase "embedded NUL and Unicode Text round trip" $ withConnection ":memory:" \conn -> do- let value = "before\0after íslenska λ 😀" :: Text- (query conn "SELECT ?" (Only value) :: IO [Only Text]) >>= (@?= [Only value])- , testCase "Int8 rejects both overflow directions" $ withConnection ":memory:" \conn -> do- forM_ ["SELECT 128::BIGINT", "SELECT -129::SMALLINT"] \sql ->- assertConversionError (query_ conn sql :: IO [Only Int8])- (query_ conn "SELECT -128::BIGINT UNION ALL SELECT 127" :: IO [Only Int8]) >>= (@?= [Only (-128), Only 127])- , testCase "unsigned narrowing rejects overflow" $ withConnection ":memory:" \conn -> do- assertConversionError (query_ conn "SELECT 256::USMALLINT" :: IO [Only Word8])- assertConversionError (query_ conn "SELECT 65536::UINTEGER" :: IO [Only Word16])- assertConversionError (query_ conn "SELECT 4294967296::UBIGINT" :: IO [Only Word32])- , testCase "finite timestamp extrema preserve all units" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- forM_ [(Seconds, 1, "TIMESTAMP_S"), (Milliseconds, 1000, "TIMESTAMP_MS"), (Microseconds, 1000000, "TIMESTAMP"), (Nanoseconds, 1000000000, "TIMESTAMP_NS")] \(constructor, units, dtype) ->- forM_ [minBound, negate (maxBound :: Int64) + 1, -1, 0, maxBound - 1] \value -> do- let expected = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (toInteger value % units)))- appendTimestamp conn dtype (toDuckValue (constructor value))- (query_ conn "SELECT value FROM native_timestamp" :: IO [Only LocalTime]) >>= (@?= [Only expected])- [Only original] <- query_ conn "SELECT {'value': value} FROM native_timestamp" :: IO [Only (StructValue FieldValue)]- -- DuckDB needs an explicit parameter type for extreme S/MS values.- query conn (fromString ("SELECT ?::STRUCT(value " <> dtype <> ")")) (Only original) >>= (@?= [Only original])- _ <- execute_ conn "DROP TABLE native_timestamp"- pure ()- , testCase "composite timestamps reject infinity and storage overflow" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- forM_ [(DuckDBTypeTimestampS, 1), (DuckDBTypeTimestampMs, 1000), (DuckDBTypeTimestamp, 1000000), (DuckDBTypeTimestampNs, 1000000000)] \(dtype, units) ->- forM_ [toInteger (minBound :: Int64) - 1, negate (toInteger (maxBound :: Int64)), toInteger (maxBound :: Int64), toInteger (maxBound :: Int64) + 1] \value -> do- let timestamp = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (value % units)))- struct = singleField (LogicalTypeScalar dtype) (FieldTimestamp (Finite timestamp))- assertIOError (query conn "SELECT ?" (Only struct) :: IO [Only FieldValue])- , testCase "composite timestamps floor fractional units before the epoch" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- forM_ [(DuckDBTypeTimestampS, 1), (DuckDBTypeTimestampMs, 1000), (DuckDBTypeTimestamp, 1000000), (DuckDBTypeTimestampNs, 1000000000)] \(dtype, units) -> do- let timestamp = utcToLocalTime utc (posixSecondsToUTCTime (fromRational ((-1) % (2 * units))))- expected = utcToLocalTime utc (posixSecondsToUTCTime (fromRational ((-1) % units)))- struct = singleField (LogicalTypeScalar dtype) (FieldTimestamp (Finite timestamp))- (query conn "SELECT (?).value" (Only struct) :: IO [Only LocalTime]) >>= (@?= [Only expected])- , testCase "composite TIME_NS rejects invalid clock components" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- forM_ [TimeOfDay (-1) 0 0, TimeOfDay 25 0 0, TimeOfDay 0 60 0, TimeOfDay 0 0 (-1), TimeOfDay 24 0 0.000000001, TimeOfDay 23 59 60.000000001] \value ->- assertIOError (query conn "SELECT ?" (Only (singleField (LogicalTypeScalar DuckDBTypeTimeNs) (FieldTime value))) :: IO [Only FieldValue])- , testCase "microsecond timestamps retain finite extrema" $ withConnection ":memory:" \conn ->- forM_ [minBound, negate (maxBound :: Int64) + 1, -1, 0, maxBound - 1] \value -> do- let expected = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (toInteger value % 1000000)))- _ <- execute_ conn "CREATE TABLE microsecond_timestamp(value TIMESTAMP)"- _ <- execute conn "INSERT INTO microsecond_timestamp VALUES (?)" (Only expected)- (query_ conn "SELECT value FROM microsecond_timestamp" :: IO [Only LocalTime]) >>= (@?= [Only expected])- _ <- execute_ conn "DROP TABLE microsecond_timestamp"- pure ()- , testCase "out-of-range dates and timestamps fail without abort" $ withConnection ":memory:" \conn -> do- let hugeDay = fromGregorian 10000000 1 1- local = LocalTime hugeDay (TimeOfDay 0 0 0)- assertIOError (query conn "SELECT ?" (Only hugeDay) :: IO [Only Day])- assertIOError (query conn "SELECT ?" (Only local) :: IO [Only LocalTime])- assertIOError (query conn "SELECT ?" (Only (UTCTime hugeDay 0)) :: IO [Only UTCTime])- , testCase "date finite storage extrema round trip" $ withConnection ":memory:" \conn ->- forM_ [minBound, negate (maxBound :: Int32) + 1, -1, 0, maxBound - 1] \value -> do- let day = addDays (toInteger value) (fromGregorian 1970 1 1)- (query conn "SELECT ?" (Only day) :: IO [Only Day]) >>= (@?= [Only day])- , testCase "input infinity sentinels are rejected" $ withConnection ":memory:" \conn -> do- forM_ [toInteger (maxBound :: Int32), negate (toInteger (maxBound :: Int32))] \days ->- assertIOError (query conn "SELECT ?" (Only (addDays days (fromGregorian 1970 1 1))) :: IO [Only Day])- forM_ [toInteger (maxBound :: Int64), negate (toInteger (maxBound :: Int64))] \micros -> do- let value = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (micros % 1000000)))- assertIOError (query conn "SELECT ?" (Only value) :: IO [Only LocalTime])- , testCase "UTC binding is TIMESTAMPTZ under a non-UTC zone" $ withConnection ":memory:" \conn -> do- _ <- execute_ conn "SET TimeZone = 'Pacific/Auckland'"- let value = UTCTime (fromGregorian 2000 1 2) 12345- (query conn "SELECT typeof(?), ?" (value, value) :: IO [(Text, UTCTime)]) >>= (@?= [("TIMESTAMP WITH TIME ZONE", value)])- _ <- execute_ conn "CREATE TABLE utc_binding(value TIMESTAMPTZ)"- _ <- execute conn "INSERT INTO utc_binding VALUES (?)" (Only value)- (query_ conn "SELECT value FROM utc_binding" :: IO [Only UTCTime]) >>= (@?= [Only value])- , testCase "DECIMAL(38) retains exact integer precision" $ withConnection ":memory:" \conn -> do- let decimal = DecimalValue 38 9 12345678901234567890123456789012345678- struct = singleField (LogicalTypeDecimal 38 9) (FieldDecimal decimal)- [Only result] <- query conn "SELECT ?" (Only struct) :: IO [Only (StructValue FieldValue)]- structValueFields result @?= structValueFields struct- , testCase "invalid DECIMAL width scale and magnitude are rejected" $ withConnection ":memory:" \conn ->- forM_ [DecimalValue 0 0 0, DecimalValue 39 0 0, DecimalValue 2 3 1, DecimalValue 2 0 100, DecimalValue 2 0 (-100)] \decimal ->- assertIOError (query conn "SELECT ?" (Only (singleField (LogicalTypeDecimal (decimalWidth decimal) (decimalScale decimal)) (FieldDecimal decimal))) :: IO [Only FieldValue])- , testCase "nested conversion failure leaves the connection usable" $ withConnection ":memory:" \conn -> do- let days = listArray (0, 1) [fromGregorian 2000 1 1, fromGregorian 10000000 1 1] :: Array Int Day- assertIOError (query conn "SELECT ?" (Only days) :: IO [Only FieldValue])- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])- , testCase "generic NULL product and payload return Left" $ do- assertLeft (genericFromFieldValue FieldNull :: Either String NamedRecord)- let member = UnionMemberType "DataMember" (LogicalTypeScalar DuckDBTypeBigInt)- union = UnionValue 0 "DataMember" FieldNull (listArray (0, 0) [member])- assertLeft (genericFromFieldValue (FieldUnion union) :: Either String NullableSum)- , testCase "generic records decode by name" $ withConnection ":memory:" \conn -> do- [Only struct] <- query_ conn "SELECT {'secondValue': 2::BIGINT, 'firstValue': 1::BIGINT}" :: IO [Only (StructValue FieldValue)]- genericFromFieldValue (FieldStruct struct) @?= Right (NamedRecord 1 2)- , testCase "generic sums decode by member name" $ withConnection ":memory:" \conn -> do- [Only union] <- query_ conn "SELECT union_value(DataMember := {'field1': 42::BIGINT})" :: IO [Only (UnionValue FieldValue)]- genericFromFieldValue (FieldUnion union) @?= Right (DataMember 42)- , testCase "union binding rejects inconsistent member name" $ withConnection ":memory:" \conn -> do- let member = UnionMemberType "right" (LogicalTypeScalar DuckDBTypeBigInt)- union = UnionValue 0 "wrong" (FieldInt64 42) (listArray (0, 0) [member])- assertIOError (query conn "SELECT ?" (Only union) :: IO [Only FieldValue])- , testCase "generic records reject wrong names" $ do- case genericToStructValue (NamedRecord 1 2) of- Nothing -> assertFailure "missing generic struct"- Just struct -> do- let fields = listArray (0, 1) [StructField "wrong" (FieldInt64 1), StructField "secondValue" (FieldInt64 2)]- assertLeft (genericFromFieldValue (FieldStruct struct{structValueFields = fields}) :: Either String NamedRecord)- , testCase "logical type Unicode names round trip" $ do- let logical = LogicalTypeStruct (listArray (0, 0) [StructField "íslenska_λ_😀" (LogicalTypeScalar DuckDBTypeBigInt)])- bracket (logicalTypeFromRep logical) destroyLogicalType \handle ->- logicalTypeToRep handle >>= (@?= logical)- , testCase "logical type names reject embedded NUL" $ do- let logical = LogicalTypeStruct (listArray (0, 0) [StructField "before\0after" (LogicalTypeScalar DuckDBTypeBigInt)])- assertIOError (bracket (logicalTypeFromRep logical) destroyLogicalType (const (pure ())))- , testCase "invalid native composite constructor returns controlled error" $ withConnection ":memory:" \conn -> do- let struct = singleField (LogicalTypeMap (LogicalTypeScalar DuckDBTypeBigInt) (LogicalTypeScalar DuckDBTypeBigInt)) (FieldMap [(FieldNull, FieldInt64 1)])- assertIOError (query conn "SELECT ?" (Only struct) :: IO [Only FieldValue])- , testCase "BIT padding preserves SQL bit_count" $ withConnection ":memory:" \conn -> do- let bits = bsFromBool [True, False, True]- (query conn "SELECT bit_count(?), ?" (bits, bits) :: IO [(Int64, BitString)]) >>= (@?= [(2, bits)])- , testCase "unsupported BIT input fails before native use" $ withConnection ":memory:" \conn ->- forM_ [BitString 0 BS.empty, BitString 8 (BS.singleton 1)] \bits ->- assertIOError (query conn "SELECT ?" (Only bits) :: IO [Only BitString])- , testCase "TIMETZ second offsets fail without rounding" $ withConnection ":memory:" \conn ->- assertIOError (query_ conn "SELECT '12:00:00+01:23:45'::TIMETZ" :: IO [Only TimeWithZone])- , testCase "ENUM payload index must be in its dictionary" $ withConnection ":memory:" \conn ->- assertIOError (query conn "SELECT ?" (Only (singleField (LogicalTypeEnum (listArray (0, 1) ["a", "b"])) (FieldEnum 2))) :: IO [Only FieldValue])- , testCase "TIMETZ input offset overflow fails before native use" $ withConnection ":memory:" \conn -> do- let value = TimeWithZone (TimeOfDay 12 0 0) (minutesToTimeZone maxBound)- struct = singleField (LogicalTypeScalar DuckDBTypeTimeTz) (FieldTimeTZ value)- assertIOError (query conn "SELECT ?" (Only struct) :: IO [Only FieldValue])- , testCase "invalid TIME input returns controlled error" $ withConnection ":memory:" \conn ->- assertIOError (query conn "SELECT ?" (Only (TimeOfDay 1000000000 0 0)) :: IO [Only TimeOfDay])- ]- <> [ testCase ("composite temporal round trip: " <> dtype) $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->- forM_ ("NULL" : map (\value -> "'" <> value <> "'") values) \value ->- assertTemporalRoundTrip conn dtype (value <> "::" <> dtype)- | (dtype, values) <-- [(dtype, ["-infinity", "infinity", "1969-12-31 23:59:59.123456789", "2000-01-01 12:34:56.123456789"]) | dtype <- ["TIMESTAMP", "TIMESTAMP_S", "TIMESTAMP_MS", "TIMESTAMP_NS", "TIMESTAMPTZ"]]- <> [("DATE", ["-infinity", "infinity", "1969-12-31", "2000-01-01"])]- <> [("TIME_NS", ["00:00:00", "12:34:56.123456789", "23:59:59.999999999", "24:00:00"])]- ]- <> [ testCase (dtype <> " " <> value <> " rejects a finite result type") $ withConnection ":memory:" \conn ->- assertConversionError (query_ conn (fromString ("SELECT '" <> value <> "'::" <> dtype)) :: IO [Only LocalTime])- | dtype <- ["DATE", "TIMESTAMP", "TIMESTAMP_S", "TIMESTAMP_MS", "TIMESTAMP_NS", "TIMESTAMPTZ"]- , value <- ["infinity", "-infinity"]- ]---- | Rebind native temporal values in structs, unions, and nested collections.-assertTemporalRoundTrip :: Connection -> String -> String -> Assertion-assertTemporalRoundTrip conn dtype expression = do- let sql =- "WITH temporal AS (SELECT "- <> expression- <> " AS value) "- <> "SELECT {'scalar': value, 'list': [value, NULL], 'array': [value, NULL]::"- <> dtype- <> "[2], 'map': map([1], [value])}, union_value(value := value) FROM temporal"- [original] <- query_ conn (fromString sql) :: IO [(StructValue FieldValue, UnionValue FieldValue)]- query conn "SELECT ?, ?" original >>= (@?= [original])---- | Store raw timestamp units without the prepared-parameter type conversion.-appendTimestamp :: Connection -> String -> IO DuckDBValue -> IO ()-appendTimestamp conn dtype createValue = do- _ <- execute_ conn (fromString ("CREATE TABLE native_timestamp(value " <> dtype <> ")"))- withConnectionHandle conn \handle ->- alloca \appenderPtr -> do- poke appenderPtr nullPtr- withCString "native_timestamp" \table ->- c_duckdb_appender_create handle nullPtr table appenderPtr >>= (@?= DuckDBSuccess)- bracket (peek appenderPtr) (const (c_duckdb_appender_destroy appenderPtr >> pure ())) \appender -> do- bracket createValue (\value -> alloca \ptr -> poke ptr value >> c_duckdb_destroy_value ptr) \value ->- c_duckdb_append_value appender value >>= (@?= DuckDBSuccess)- c_duckdb_appender_end_row appender >>= (@?= DuckDBSuccess)- c_duckdb_appender_flush appender >>= (@?= DuckDBSuccess)---- | Build a single-field struct for nested binding tests.-singleField :: LogicalTypeRep -> FieldValue -> StructValue FieldValue-singleField logical value =- StructValue- (listArray (0, 0) [StructField "value" value])- (listArray (0, 0) [StructField "value" logical])- (Map.singleton "value" 0)---- | Assert a controlled input or materialization error.-assertIOError :: IO a -> Assertion-assertIOError action = do- result <- try action- case result of- Left (_ :: IOException) -> pure ()- Right _ -> assertFailure "expected IOException"---- | Assert a controlled FromField conversion error.-assertConversionError :: IO a -> Assertion-assertConversionError action = do- result <- try action- case result of- Left (_ :: SQLError) -> pure ()- Right _ -> assertFailure "expected ResultError"---- | Assert rejection without forcing a partial generic value.-assertLeft :: Either String a -> Assertion-assertLeft (Left _) = pure ()-assertLeft (Right _) = assertFailure "expected Left"+{-# LANGUAGE DataKinds #-}+{-# LANGUAGE DeriveGeneric #-}+{-# LANGUAGE DerivingVia #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-}+{-# LANGUAGE OverloadedStrings #-}+{-# LANGUAGE ScopedTypeVariables #-}+{-# LANGUAGE TypeApplications #-}++-- |+-- Module : ValueRegressionTests+-- Description : Regression tests for exact values and controlled conversion failures.+module ValueRegressionTests (main, valueRegressionTests) where++import Control.Applicative ((<|>))+import Control.Concurrent (forkIO, killThread, newEmptyMVar, putMVar, readMVar, takeMVar, threadDelay)+import Control.Exception (+ AsyncException (ThreadKilled),+ ErrorCall,+ MaskingState (..),+ SomeException,+ bracket,+ displayException,+ fromException,+ getMaskingState,+ mask,+ mask_,+ throwIO,+ try,+ uninterruptibleMask_,+ )+import Control.Monad (forM, forM_, replicateM, void, when)+import Data.Array (elems, listArray)+import Data.Bits (shiftR, xor)+import qualified Data.ByteString as BS+import Data.IORef (atomicModifyIORef', newIORef, readIORef)+import Data.Int (Int16, Int32, Int64, Int8)+import qualified Data.Map.Strict as Map+import Data.Maybe (fromMaybe, isJust)+import Data.Proxy (Proxy (..), asProxyTypeOf)+import Data.Ratio ((%))+import Data.String (fromString)+import Data.Text (Text)+import qualified Data.Text as Text+import qualified Data.Text.Encoding as TextEncoding+import Data.Time.Calendar (Day, addDays, diffDays, fromGregorian)+import Data.Time.Clock (UTCTime (..))+import Data.Time.Clock.POSIX (posixSecondsToUTCTime)+import Data.Time.LocalTime (LocalTime (..), TimeOfDay (..), hoursToTimeZone, minutesToTimeZone, utc, utcToLocalTime)+import qualified Data.UUID as UUID+import Data.Word (Word16, Word32, Word64, Word8)+import Database.DuckDB.FFI.V2 (+ DuckDBV2LogicalTypeHandle,+ c_duckdb_v2_factory_create_common_type,+ c_duckdb_v2_factory_create_type_from_text,+ c_duckdb_v2_logical_type_is_equal,+ c_duckdb_v2_logical_type_to_text,+ c_duckdb_v2_value_cast,+ c_duckdb_v2_value_create_null,+ c_duckdb_v2_vector_set_size,+ c_duckdb_v2_vector_set_value,+ )+import Database.DuckDB.Simple+import qualified Database.DuckDB.Simple.Catalog as Catalog+import Database.DuckDB.Simple.Conversion (conversionErrorWith)+import Database.DuckDB.Simple.FromField (+ BigNum (..),+ BitString (..),+ DecimalValue (..),+ Field (..),+ FieldValue (..),+ IntervalValue (..),+ TimeWithZone (..),+ bsFromBool,+ returnError,+ timeWithZone,+ timeWithZoneZone,+ )+import Database.DuckDB.Simple.Generic (DuckValue (..), ViaDuckDB (..), genericFromFieldValue, genericToStructValue, typedFromField)+import Database.DuckDB.Simple.Internal (destroyValue, withConnectionFactory, withConnectionHandle, withOwnedVector)+import Database.DuckDB.Simple.LogicalRep+import Database.DuckDB.Simple.Ok (Ok (..))+import qualified Database.DuckDB.Simple.Streaming as Streaming+import Database.DuckDB.Simple.Time (Unbounded (..))+import Database.DuckDB.Simple.ToField (ToDuckValue (..))+import Database.DuckDB.Simple.V2.LogicalRep (+ destroyLogicalType,+ logicalTypeFromRep,+ logicalTypeToBaseRep,+ logicalTypeToRep,+ )+import qualified Database.DuckDB.Simple.V2.Registration as R+import Database.DuckDB.Simple.Variant (Variant (..))+import Foreign.C.ConstPtr (ConstPtr (..))+import Foreign.Marshal.Alloc (alloca, allocaBytes)+import Foreign.Marshal.Array (withArray)+import Foreign.Ptr (castPtr, nullPtr)+import Foreign.Storable (peek)+import GHC.Float (castDoubleToWord64, castFloatToWord32)+import GHC.Generics (Generic)+import Numeric.Natural (Natural)+import System.Mem (getAllocationCounter, performMajorGC)+import System.Timeout (timeout)+import Test.Tasty (TestTree, defaultMain, testGroup)+import Test.Tasty.HUnit+import TestUtils (NativeTimestamp (..), assertInputError, assertRangeError, sized)++-- | The type requested by a field parser.+data FieldTypeOf = FieldTypeOf LogicalTypeRep+ deriving (Eq, Show)++instance FromField FieldTypeOf where+ fromField Field{fieldType} = FieldTypeOf <$> fieldType++-- | An Int64, or else a Text. A read of an ENUM value fails as Int64, and no+-- one renders that error.+newtype IntOrText = IntOrText (Either Int64 Text)+ deriving (Eq, Show)++instance FromField IntOrText where+ fromField f = IntOrText <$> ((Left <$> fromField f) <|> (Right <$> fromField f))++-- | The text of a native type, from duckdb_v2_logical_type_to_text.+nativeTypeText :: DuckDBV2LogicalTypeHandle -> IO Text+nativeTypeText logical = alloca \lengthSlot -> do+ R.checked (c_duckdb_v2_logical_type_to_text logical nullPtr 0 lengthSlot)+ size <- peek lengthSlot+ allocaBytes (fromIntegral size + 1) \buffer -> do+ R.checked (c_duckdb_v2_logical_type_to_text logical buffer (size + 1) lengthSlot)+ TextEncoding.decodeUtf8 <$> BS.packCStringLen (castPtr buffer, fromIntegral size)++-- | A value whose read checks the column type, as a user instance can: a+-- DECIMAL with the scale 2, a TIMESTAMP_NS, an ARRAY of three elements, or+-- JSON. The value names the accepted kind.+newtype Checked = Checked Text+ deriving (Eq, Ord, Show)++instance FromField Checked where+ fromField f@Field{fieldType} = do+ logical <- fieldType+ case logical of+ LogicalTypeDecimal _ 2 -> pure (Checked "decimal")+ LogicalTypeScalar LogicalTypeIdTimestampNs -> pure (Checked "timestamp_ns")+ LogicalTypeArray _ 3 -> pure (Checked "array")+ LogicalTypeAlias "JSON" _ -> pure (Checked "json")+ _ -> returnError Incompatible f "expected DECIMAL(_,2), TIMESTAMP_NS, an ARRAY of 3 elements, or JSON"++-- | A record field reads with its type through 'typedFromField'.+instance DuckValue Checked where+ duckToField (Checked text) = FieldText text+ duckLogicalType _ = LogicalTypeScalar LogicalTypeIdVarchar+ duckFromTypedField = typedFromField++-- | A record with a field that checks its type.+newtype Priced = Priced {price :: Checked}+ deriving (Eq, Show, Generic)+ deriving (FromField) via (ViaDuckDB Priced)++-- | An array element with a ToField instance and no ToDuckValue instance.+newtype DelegatedInteger = DelegatedInteger Int64++instance DuckDBColumnType DelegatedInteger where+ duckdbColumnTypeFor _ = "BIGINT"++instance ToField DelegatedInteger where+ toField (DelegatedInteger value) = toField value++-- | Bind an array through a value built with the explicit factory.+newtype Scores = Scores (SizedArray 3 Int64)+ deriving (Show)++instance DuckDBColumnType Scores where+ duckdbColumnTypeFor _ = "BIGINT[]"++instance ToDuckValue Scores where+ toDuckValue factory (Scores values) = toDuckValue factory values++instance ToField Scores++-- | Bind declared child types without adapting them to a parameter target.+data UnadaptedStruct+ = UnadaptedStruct (StructValue FieldValue)+ | TypedStructNull LogicalTypeRep+ deriving (Show)++instance DuckDBColumnType UnadaptedStruct where+ duckdbColumnTypeFor _ = "STRUCT"+ duckdbTypeDeclaration _ = Left "the fields depend on the value"++instance ToDuckValue UnadaptedStruct where+ toDuckValue factory (UnadaptedStruct value) = toDuckValue factory value+ toDuckValue factory (TypedStructNull rep) =+ bracket (logicalTypeFromRep factory rep) destroyLogicalType \logical ->+ R.output (c_duckdb_v2_value_create_null factory logical)++instance ToField UnadaptedStruct++-- | Record with identical field types to detect positional decoding.+data NamedRecord = NamedRecord {firstValue :: Int64, secondValue :: Int64}+ deriving (Eq, Show, Generic)++-- | Sum with a payload to test NULL member handling.+data NullableSum = EmptyMember | DataMember Int64+ deriving (Eq, Show, Generic)++-- | A generic record with a UTCTime field.+newtype Stamped = Stamped {stampedAt :: UTCTime}+ deriving stock (Eq, Show, Generic)+ deriving (FromField) via (ViaDuckDB Stamped)++-- | A generic record with a LIST, an ARRAY, and a MAP field.+data Holder = Holder {listField :: [Int32], arrayField :: SizedArray 2 Int32, mapField :: Map.Map Text Int32}+ deriving stock (Eq, Show, Generic)+ deriving (FromField) via (ViaDuckDB Holder)++-- | A generic record for array parameters.+data Pair = Pair {pairName :: Text, pairValue :: Int64}+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB Pair)++-- | A generic record with an ARRAY field.+data Scored = Scored {scoredName :: Text, scoredValues :: SizedArray 3 Int32}+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB Scored)++-- | Generic nullary constructors retain their existing UNION schema.+data Colour = Red | Blue+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB Colour)++-- | Run this module without the main integration suite.+main :: IO ()+main = defaultMain valueRegressionTests++-- | Focused value regressions that also run against the baseline library.+valueRegressionTests :: TestTree+valueRegressionTests =+ testGroup "value regressions" $+ [ testCase "REAL decodes to Float and Double" $ withConnection ":memory:" \conn -> do+ (query_ conn "SELECT 1.25::REAL" :: IO [Only Float]) >>= (@?= [Only 1.25])+ (query_ conn "SELECT 1.25::REAL" :: IO [Only Double]) >>= (@?= [Only 1.25])+ , testCase "a Float parameter does not read as Text, and its cast to VARCHAR does" $ withConnection ":memory:" \conn -> do+ assertIncompatible conn "?::FLOAT" (Only (1.25 :: Float)) "FLOAT" (Proxy @Text) "Text"+ (query conn "SELECT ?::VARCHAR" (Only (1.25 :: Float)) :: IO [Only Text]) >>= (@?= [Only "1.25"])+ , testCase "BOOLEAN read as Text and String gives the cast to VARCHAR" $ withConnection ":memory:" \conn ->+ (query_ conn "SELECT b, b::VARCHAR, b FROM (VALUES (true), (false)) t(b)" :: IO [(Text, Text, String)])+ >>= (@?= [("true", "true", "true"), ("false", "false", "false")])+ , testCase "generic nullary UNION constructors round trip through DuckDB" $ withConnection ":memory:" \conn ->+ forM_ [Red, Blue] \colour ->+ (query conn "SELECT ?" (Only colour) :: IO [Only Colour]) >>= (@?= [Only colour])+ , testCase "UNION NULL payload retains its member type" $ withConnection ":memory:" \conn -> do+ let members =+ listArray+ (0, 1)+ [UnionMemberType "number" (LogicalTypeScalar LogicalTypeIdBigInt), UnionMemberType "text" (LogicalTypeScalar LogicalTypeIdVarchar)]+ original = UnionValue 0 "number" FieldNull members+ (query conn "SELECT ?" (Only original) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only original])+ , testCase "a UNION value binds the member of its tag up to the member limit of DuckDB" $ withConnection ":memory:" \conn -> do+ let lastMember :: Int -> UnionValue FieldValue+ lastMember count =+ let name index = "m" <> Text.pack (show index)+ members = listArray (0, count - 1) [UnionMemberType (name index) (LogicalTypeScalar LogicalTypeIdInteger) | index <- [0 .. count - 1]]+ in UnionValue (fromIntegral (count - 1)) (name (count - 1)) (FieldInt32 7) members+ accepted = lastMember 255+ (query conn "SELECT union_tag(?)::VARCHAR, ?" (accepted, accepted) :: IO [(Text, UnionValue FieldValue)]) >>= (@?= [("m254", accepted)])+ -- DuckDB 2.0.0 permits at most 255 members in a UNION type.+ forM_ [256, 300] \count ->+ try (query conn "SELECT ?" (Only (lastMember count)) :: IO [Only (UnionValue FieldValue)]) >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorMessage @?= "Binder Error: UNION type supports at most 255 type modifiers"+ sqlErrorType @?= Just ErrorQueryBinder+ Right rows -> assertFailure (show count <> " members: expected the DuckDB member limit error, got " <> show rows)+ , testCase "array elements need only a ToField instance" $ withConnection ":memory:" \conn -> do+ let values = sized [Nothing, Just (DelegatedInteger 41), Just (DelegatedInteger 99)] :: SizedArray 3 (Maybe DelegatedInteger)+ expected = sized [Nothing, Just 41, Just 99] :: SizedArray 3 (Maybe Int64)+ (query conn "SELECT ?" (Only values) :: IO [Only (SizedArray 3 (Maybe Int64))]) >>= (@?= [Only expected])+ , testCase "array children use default ToDuckValue instances" $ withConnection ":memory:" \conn -> do+ let units = [-1, 0, 1234567]+ values = sized (map Microseconds units) :: SizedArray 3 NativeTimestamp+ expected = sized (map (utcToLocalTime utc . posixSecondsToUTCTime . fromRational . (% 1000000) . toInteger) units)+ (query conn "SELECT ?" (Only values) :: IO [Only (SizedArray 3 LocalTime)]) >>= (@?= [Only expected])+ , testCase "arrays bind through ToDuckValue with an explicit connection" $ withConnection ":memory:" \conn -> do+ let values = sized [1, -2, 3]+ (query conn "SELECT typeof(?), ?" (Scores values, Scores values) :: IO [(Text, SizedArray 3 Int64)]) >>= (@?= [("BIGINT[3]", values)])+ , testCase "arrays of STRUCT, record, and ARRAY elements take the type of a present element" $ withConnection ":memory:" \conn -> do+ [Only struct] <- query_ conn "SELECT {'a': 1, 'b': 'x'}" :: IO [Only (StructValue FieldValue)]+ let structs = sized [struct, struct]+ (query conn "SELECT typeof(?), ?" (structs, structs) :: IO [(Text, SizedArray 2 (StructValue FieldValue))])+ >>= (@?= [("STRUCT(a INTEGER, b VARCHAR)[2]", structs)])+ let pairs = sized [Nothing, Just (Pair "x" 1), Just (Pair "y" 2)]+ (query conn "SELECT typeof(?), ?" (pairs, pairs) :: IO [(Text, SizedArray 3 (Maybe Pair))])+ >>= (@?= [("STRUCT(pairName VARCHAR, pairValue BIGINT)[3]", pairs)])+ let nested = sized [sized [1, 2], sized [3, 4]] :: SizedArray 2 (SizedArray 2 Int64)+ (query conn "SELECT typeof(?), ?" (nested, nested) :: IO [(Text, SizedArray 2 (SizedArray 2 Int64))]) >>= (@?= [("BIGINT[2][2]", nested)])+ , testCase "generic records with ARRAY fields bind with the size of the type" $ withConnection ":memory:" \conn -> do+ let first = Scored "x" (sized [1, 2, 3])+ second = Scored "y" (sized [4, 5, 6])+ rowType = "STRUCT(scoredName VARCHAR, scoredValues INTEGER[3])"+ (query conn "SELECT typeof(?), ?" (first, first) :: IO [(Text, Scored)]) >>= (@?= [(rowType, first)])+ _ <- execute_ conn "CREATE TABLE scored (s STRUCT(scoredName VARCHAR, scoredValues INTEGER[3]))"+ _ <- executeMany conn "INSERT INTO scored VALUES (?)" [Only first, Only second]+ (query_ conn "SELECT s FROM scored ORDER BY s.scoredName" :: IO [Only Scored]) >>= (@?= [Only first, Only second])+ let records = sized [first, second] :: SizedArray 2 Scored+ (query conn "SELECT typeof(?), ?" (records, records) :: IO [(Text, SizedArray 2 Scored)]) >>= (@?= [(rowType <> "[2]", records)])+ , testCase "empty arrays of record elements raise the empty ARRAY error" $ withConnection ":memory:" \conn -> do+ result <- try (query conn "SELECT ?" (Only (sized [] :: SizedArray 0 Pair)))+ case result of+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertBool (Text.unpack sqlErrorMessage) ("duckdb-simple: an empty ARRAY parameter is not possible" `Text.isInfixOf` sqlErrorMessage)+ sqlErrorType @?= Just ErrorInputInvalid+ Right (_ :: [Only Text]) -> assertFailure "expected the empty ARRAY error"+ , testCase "composite arrays with different element types take the type of array_value" $ withConnection ":memory:" \conn -> do+ forM_+ [ ("{'a': 1::INTEGER}", "{'a': 2::INTEGER, 'b': 3::INTEGER}", "STRUCT(a INTEGER, b INTEGER)[3]")+ , ("{'a': 1::INTEGER}", "{'a': 1.6::DOUBLE}", "STRUCT(a DOUBLE)[3]")+ , ("{'a': 1.2::DECIMAL(4,1)}", "{'a': 1.26::DECIMAL(4,2)}", "STRUCT(a DECIMAL(5,2))[3]")+ , ("{'a': [1.2::DECIMAL(4,1)]}", "{'a': [1.26::DECIMAL(4,2)]}", "STRUCT(a DECIMAL(5,2)[])[3]")+ ]+ \(firstSql, secondSql, typeName) -> do+ [Only first] <- query_ conn (Query ("SELECT " <> firstSql)) :: IO [Only (StructValue FieldValue)]+ [Only second] <- query_ conn (Query ("SELECT " <> secondSql)) :: IO [Only (StructValue FieldValue)]+ forM_+ [([Nothing, Just first, Just second], ["NULL", firstSql, secondSql]), ([Just second, Nothing, Just first], [secondSql, "NULL", firstSql])]+ \(elements, sqls) -> do+ let values = sized elements :: SizedArray 3 (Maybe (StructValue FieldValue))+ literal = "array_value(" <> Text.intercalate ", " sqls <> ")"+ (query conn (Query ("SELECT typeof(?), ?::VARCHAR, (" <> literal <> ")::VARCHAR")) (values, values) :: IO [(Text, Text, Text)]) >>= \case+ [(actualType, text, literalText)] -> do+ actualType @?= typeName+ text @?= literalText+ rows -> assertFailure (show rows)+ withConnectionHandle conn \native -> withConnectionFactory native \factory -> bracket (toDuckValue factory values) destroyValue (const (pure ()))+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "a FieldValue binds back as the value that it was read from" $ withConnection ":memory:" \conn -> do+ -- The third item is the type of the bound value for SELECT ?, when it+ -- is not the type of the source.+ forM_+ [ ("BOOLEAN", "true", Nothing)+ , ("TINYINT", "-7", Nothing)+ , ("SMALLINT", "-300", Nothing)+ , ("INTEGER", "-70000", Nothing)+ , ("BIGINT", "-9007199254740993", Nothing)+ , ("UTINYINT", "200", Nothing)+ , ("USMALLINT", "60000", Nothing)+ , ("UINTEGER", "4000000000", Nothing)+ , ("UBIGINT", "18446744073709551615", Nothing)+ , ("HUGEINT", "-170141183460469231731687303715884105728", Nothing)+ , ("UHUGEINT", "5", Nothing)+ , ("UHUGEINT", "340282366920938463463374607431768211455", Nothing)+ , ("FLOAT", "1.25", Nothing)+ , ("DOUBLE", "0.1", Nothing)+ , ("DECIMAL(4,1)", "123.4", Nothing)+ , ("DECIMAL(18,3)", "123456789012345.678", Nothing)+ , ("DECIMAL(38,10)", "1234567890123456789012345678.0123456789", Nothing)+ , ("VARCHAR", "'text'", Nothing)+ , ("VARCHAR COLLATE NOCASE", "'Text'", Nothing)+ , ("JSON", "'{\"a\": 1}'", Just "VARCHAR")+ , ("BLOB", "'\\x00\\xFF'::BLOB", Nothing)+ , ("UUID", "'00112233-4455-6677-8899-aabbccddeeff'", Nothing)+ , ("BIT", "'10110'", Nothing)+ , ("DATE", "'2026-01-02'", Nothing)+ , ("DATE", "'infinity'", Nothing)+ , ("TIME", "'03:04:05.123456'", Nothing)+ , ("TIME_NS", "'03:04:05.123456789'", Just "TIME")+ , ("TIMETZ", "'03:04:05+01:30:15'", Nothing)+ , ("TIMESTAMP", "'2026-01-02 03:04:05.123456'", Nothing)+ , ("TIMESTAMP", "'-infinity'", Nothing)+ , ("TIMESTAMP_S", "'2026-01-02 03:04:05'", Just "TIMESTAMP")+ , ("TIMESTAMP_MS", "'2026-01-02 03:04:05.123'", Just "TIMESTAMP")+ , ("TIMESTAMP_NS", "'2026-01-02 03:04:05.123456789'", Just "TIMESTAMP")+ , ("TIMESTAMPTZ", "'2026-01-02 03:04:05.123456+00'", Nothing)+ , ("TIMESTAMPTZ_NS", "'2026-01-02 03:04:05.123456789+00'", Just "TIMESTAMP WITH TIME ZONE")+ , ("INTERVAL", "INTERVAL '1 month 2 days 3 microseconds'", Nothing)+ , ("BIGNUM", "'123456789012345678901234567890123456789012345678901234567890'", Nothing)+ , ("GEOMETRY", "'POINT (1 2)'", Nothing)+ , ("GEOMETRY('OGC:CRS84')", "'POINT (1 2)'", Nothing)+ , ("VARIANT", "42::BIGINT", Just "BIGINT")+ , ("VARIANT", "{'k': 1}", Just "STRUCT(k VARIANT)")+ , ("INTEGER", "NULL", Just "\"NULL\"")+ , ("INTEGER[]", "[1, NULL, 3]", Nothing)+ , ("TIMESTAMP_NS[]", "['2026-01-02 03:04:05.123456789', '2026-01-02 03:04:05']", Just "TIMESTAMP[]")+ , ("TIME_NS[]", "['03:04:05.123456789', '03:04:05']", Just "TIME[]")+ , ("TIMESTAMPTZ_NS[]", "['2026-01-02 03:04:05.123456789+00', '2026-01-02 03:04:05+00']", Just "TIMESTAMP WITH TIME ZONE[]")+ , ("DECIMAL(4,1)[]", "[1.5, 2.5]", Nothing)+ , ("INTEGER[2]", "[1, 2]", Nothing)+ , ("VARCHAR[2][]", "[['a', 'b'], ['c', NULL]]", Nothing)+ , ("MAP(VARCHAR, INTEGER)", "MAP {'a': 1, 'b': NULL}", Nothing)+ , ("MAP(INTEGER[], STRUCT(x DOUBLE))", "MAP {[1]: {'x': 1.5}}", Nothing)+ , ("STRUCT(a INTEGER, b VARCHAR)", "{'a': 1, 'b': 'x'}", Nothing)+ , ("STRUCT(e ENUM('a', 'b'), n INTEGER[])", "{'e': 'b', 'n': []}", Nothing)+ , ("ENUM('a', 'b')", "'b'", Just "VARCHAR")+ , ("ENUM('a', 'b')[]", "['a', NULL, 'b']", Just "VARCHAR[]")+ , ("TUPLE(INTEGER, VARCHAR)", "(1, 'x')", Nothing)+ , ("UNION(a INTEGER, b VARCHAR)", "union_value(b := 'x')", Nothing)+ , ("STRUCT(a INTEGER)[]", "[{'a': 1}, NULL]", Nothing)+ , ("VARIANT", "'2026-01-02 03:04:05.123456789'::TIMESTAMP_NS", Just "TIMESTAMP")+ , ("VARIANT", "'03:04:05.123456789'::TIME_NS", Just "TIME")+ , ("VARIANT", "123.4::DECIMAL(4,1)", Just "DECIMAL(4,1)")+ , ("VARIANT", "'00112233-4455-6677-8899-aabbccddeeff'::UUID", Just "UUID")+ , ("VARIANT", "'\\x00\\xFF'::BLOB", Just "BLOB")+ , ("VARIANT", "'03:04:05+01:30'::TIMETZ", Just "TIME WITH TIME ZONE")+ , ("VARIANT", "INTERVAL '1 month 2 days 3 microseconds'", Just "INTERVAL")+ , ("GEOMETRY('OGC:CRS84')[]", "['POINT (1 2)', NULL]", Nothing)+ , ("STRUCT(g GEOMETRY('OGC:CRS84'))", "{'g': 'POINT (1 2)'}", Nothing)+ , ("TIMESTAMPTZ", "'infinity'", Nothing)+ , ("TIME", "'24:00:00'", Nothing)+ , ("DOUBLE", "'NaN'", Nothing)+ , ("DOUBLE", "'-Infinity'", Nothing)+ , ("FLOAT", "'NaN'", Nothing)+ , ("FLOAT", "'Infinity'", Nothing)+ , ("UNION(a INTEGER, b VARCHAR)", "union_value(a := NULL::INTEGER)", Nothing)+ , ("INTEGER[][2]", "[[1, 2], [3]]", Nothing)+ , ("MAP(STRUCT(k INTEGER), VARCHAR)", "MAP {{'k': 1}: 'a'}", Nothing)+ , ("MAP(VARCHAR, INTEGER)", "MAP {}", Just "MAP(\"NULL\", \"NULL\")")+ , ("INTEGER[]", "[]", Just "\"NULL\"[]")+ , ("INTEGER[]", "[NULL, NULL]", Just "\"NULL\"[]")+ , ("BIT[]", "['101', '0']", Nothing)+ , ("TIMETZ[]", "['03:04:05+01:30']", Nothing)+ , ("UUID[]", "['00112233-4455-6677-8899-aabbccddeeff']", Nothing)+ , ("TIME_NS[][]", "[['03:04:05.123456789', '03:04:05']]", Just "TIME[][]")+ ]+ \(sqlType, literal, boundType) -> do+ let source = "(" <> literal <> ")::" <> sqlType+ label = Text.unpack source+ [Only value] <- query_ conn (Query ("SELECT " <> source)) :: IO [Only FieldValue]+ rows <- query conn (Query ("SELECT typeof(?), typeof(" <> source <> "), ? IS NOT DISTINCT FROM " <> source)) (value, value)+ case (rows, boundType) of+ ([(actual, expected, same)], Nothing) -> do+ assertEqual label expected actual+ assertBool (label <> ": SELECT ? gives another value") same+ ([(actual, _, _ :: Bool)], Just expected) -> assertEqual label expected (actual :: Text)+ _ -> assertFailure (label <> ": " <> show rows)+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE back (v " <> sqlType <> ")"))+ _ <- execute conn "INSERT INTO back VALUES (?)" (Only value)+ (query_ conn (Query ("SELECT v IS NOT DISTINCT FROM " <> source <> " FROM back")) :: IO [Only Bool]) >>= assertEqual label [Only True]+ -- ST_SetCRS supplies the custom CRS in the result type.+ withConnection ":memory:" \custom -> do+ _ <- execute_ custom "CREATE TABLE shapes AS SELECT ST_SetCRS('POINT (1 2)'::GEOMETRY, 'local grid') AS v"+ [Only shape] <- query_ custom "SELECT v FROM shapes" :: IO [Only FieldValue]+ ( query custom "SELECT typeof(?), (SELECT typeof(v) FROM shapes), ? IS NOT DISTINCT FROM (SELECT v FROM shapes)" (shape, shape)+ :: IO [(Text, Text, Bool)]+ )+ >>= \case+ [(actual, expected, same)] -> do+ actual @?= expected+ assertBool "custom CRS: SELECT ? gives another value" same+ rows -> assertFailure (show rows)+ _ <- execute custom "INSERT INTO shapes VALUES (?)" (Only shape)+ (query_ custom "SELECT count(*) FROM shapes WHERE v IS NOT DISTINCT FROM (SELECT v FROM shapes LIMIT 1)" :: IO [Only Int64])+ >>= (@?= [Only 2])+ -- A row read as a list of FieldValue binds back as a row.+ _ <-+ execute_+ conn+ "CREATE TABLE source_row (i INTEGER, t VARCHAR, s STRUCT(a INTEGER), l DOUBLE[]); INSERT INTO source_row VALUES (1, 'x', {'a': 2}, [1.5, NULL]); CREATE TABLE copied AS FROM source_row LIMIT 0"+ [row] <- query_ conn "SELECT * FROM source_row" :: IO [[FieldValue]]+ _ <- execute conn "INSERT INTO copied VALUES (?, ?, ?, ?)" row+ (query_ conn "SELECT (SELECT count(*) FROM copied), (SELECT count(*) FROM (FROM source_row EXCEPT FROM copied))" :: IO [(Int64, Int64)])+ >>= (@?= [(1, 0)])+ , testCase "a FieldValue with no known type raises an error" $ withConnection ":memory:" \conn -> do+ let untyped :: String -> FieldValue -> Text -> IO ()+ untyped label value part =+ try (query conn "SELECT ?" (Only value) :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ assertEqual label (Just ErrorInputInvalid) sqlErrorType+ assertBool (Text.unpack sqlErrorMessage) (part `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure (label <> ": " <> show rows)+ untyped "empty array" (FieldArray (listArray (0, -1) [])) "an empty ARRAY parameter is not possible"+ -- Empty values and NULL elements bind as the SQL literal does.+ forM_+ [ ("empty list", FieldList [], "\"NULL\"[]", "[]")+ , ("empty inner list", FieldList [FieldList [FieldInt32 1], FieldList []], "INTEGER[][]", "[[1], []]")+ , ("empty map", FieldMap [], "MAP(\"NULL\", \"NULL\")", "{}")+ , ("NULL list", FieldList [FieldNull, FieldNull], "\"NULL\"[]", "[NULL, NULL]")+ , ("NULL array", FieldArray (listArray (0, 1) [FieldNull, FieldNull]), "\"NULL\"[2]", "[NULL, NULL]")+ , ("NULL map values", FieldMap [(FieldText "a", FieldNull)], "MAP(VARCHAR, \"NULL\")", "{a=NULL}")+ ]+ \(label, value, typeName, text) ->+ (query conn "SELECT typeof(?), ?::VARCHAR" (value, value) :: IO [(Text, Text)]) >>= assertEqual label [(typeName, text)]+ forM_+ [ ("STRUCT(a INTEGER)[]", FieldList [], "[]")+ , ("STRUCT(a INTEGER)[]", FieldList [FieldNull], "[NULL]")+ , ("INTEGER[][]", FieldList [FieldList [FieldInt32 1], FieldList []], "[[1], []]")+ , ("MAP(VARCHAR, STRUCT(a INTEGER))", FieldMap [], "{}")+ , ("MAP(VARCHAR, STRUCT(a INTEGER))", FieldMap [(FieldText "k", FieldNull)], "{k=NULL}")+ ]+ \(sqlType, value, text) -> do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE typed (v " <> sqlType <> ")"))+ _ <- execute conn "INSERT INTO typed VALUES (?)" (Only value)+ (query_ conn "SELECT v::VARCHAR FROM typed" :: IO [Only Text]) >>= assertEqual (Text.unpack sqlType) [Only text]+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "a FieldList of scalar values with different types follows the SQL list literal" $ withConnection ":memory:" \conn -> do+ forM_+ [ ([FieldInt32 1, FieldInt64 2], "[1::INTEGER, 2::BIGINT]", "BIGINT[]")+ , ([FieldInt64 9007199254740993, FieldDouble 1.5], "[9007199254740993::BIGINT, 1.5::DOUBLE]", "DOUBLE[]")+ , ([FieldList [FieldInt32 1], FieldList [FieldInt64 2]], "[[1::INTEGER], [2::BIGINT]]", "BIGINT[][]")+ ]+ \(items, literal, typeName) ->+ ( query conn (Query ("SELECT typeof(?), ?::VARCHAR, (" <> literal <> ")::VARCHAR")) (FieldList items, FieldList items)+ :: IO [(Text, Text, Text)]+ )+ >>= \case+ [(actualType, text, literalText)] -> do+ assertEqual (Text.unpack literal) typeName actualType+ assertEqual (Text.unpack literal) literalText text+ rows -> assertFailure (show rows)+ (query conn "SELECT ?::VARCHAR" (Only (FieldList [FieldInt64 9007199254740993, FieldDouble 1.5])) :: IO [Only Text])+ >>= (@?= [Only "[9007199254740992.0, 1.5]"])+ [Only mixedVariant] <- query_ conn "SELECT [1::VARIANT, 'x'::VARIANT]::VARIANT" :: IO [Only FieldValue]+ forM_+ [ ("scalars", FieldList [FieldInt32 1, FieldText "a"], "Cannot combine types INTEGER and VARCHAR")+ , ("lists", FieldList [FieldList [FieldInt32 1], FieldList [FieldText "a"]], "Cannot combine types INTEGER[] and VARCHAR[]")+ , ("array", FieldArray (listArray (0, 1) [FieldInt32 1, FieldText "a"]), "Cannot combine types INTEGER and VARCHAR")+ , ("VARIANT array", mixedVariant, "Cannot combine types INTEGER and VARCHAR")+ ]+ \(label, value, part) ->+ try (query conn "SELECT ?" (Only value) :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ assertEqual label (Just ErrorQueryNotImplemented) sqlErrorType+ assertBool (Text.unpack sqlErrorMessage) (part `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure (label <> ": " <> show rows)+ , testCase "a list or a FieldArray of composite elements with different types follows the SQL list literal" $ withConnection ":memory:" \conn -> do+ let elementOf :: Text -> IO (Maybe (StructValue FieldValue))+ elementOf "NULL" = pure Nothing+ elementOf sql = do+ [Only value] <- query_ conn (Query ("SELECT " <> sql))+ pure (Just value)+ sameAs :: (ToField a) => a -> Text -> Text -> IO ()+ sameAs value literal typeName =+ (query conn (Query ("SELECT typeof(?), ?::VARCHAR, (" <> literal <> ")::VARCHAR")) (value, value) :: IO [(Text, Text, Text)]) >>= \case+ [(actualType, text, literalText)] -> do+ assertEqual (Text.unpack literal) typeName actualType+ assertEqual (Text.unpack literal) literalText text+ rows -> assertFailure (show rows)+ duckdbError :: (ToField a) => String -> a -> ErrorType -> Text -> IO ()+ duckdbError label value errorType part =+ try (query conn "SELECT ?" (Only value) :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ assertEqual label (Just errorType) sqlErrorType+ assertBool (Text.unpack sqlErrorMessage) (part `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure (label <> ": " <> show rows)+ forM_+ [ (["{'a': 1}", "{'b': 2}"], "STRUCT(a INTEGER, b INTEGER)")+ , (["{'a': 1::INTEGER}", "{'a': 2::BIGINT}"], "STRUCT(a BIGINT)")+ , (["{'a': 9007199254740993::BIGINT}", "{'a': 1.5::DOUBLE}"], "STRUCT(a DOUBLE)")+ , (["{'a': [1::INTEGER]}", "{'a': [2::BIGINT]}"], "STRUCT(a BIGINT[])")+ , (["{'a': 1}", "NULL"], "STRUCT(a INTEGER)")+ , (["NULL", "{'a': 1}", "{'b': 2}"], "STRUCT(a INTEGER, b INTEGER)")+ ]+ \(sqls, elementType) -> do+ elements <- mapM elementOf sqls+ let literal = Text.intercalate ", " sqls+ count = length elements+ sameAs elements ("[" <> literal <> "]") (elementType <> "[]")+ sameAs+ (FieldArray (listArray (0, count - 1) (map (maybe FieldNull FieldStruct) elements)))+ ("array_value(" <> literal <> ")")+ (elementType <> "[" <> Text.pack (show count) <> "]")+ -- The rule can round, as in SQL.+ large <- mapM elementOf ["{'a': 9007199254740993::BIGINT}", "{'a': 1.5::DOUBLE}"]+ (query conn "SELECT ?::VARCHAR" (Only large) :: IO [Only Text]) >>= (@?= [Only "[{'a': 9007199254740992.0}, {'a': 1.5}]"])+ mismatch <- mapM elementOf ["{'a': 1}", "{'a': 'x'}"]+ duckdbError "list" mismatch ErrorQueryNotImplemented "Cannot combine types STRUCT(a INTEGER) and STRUCT(a VARCHAR)"+ duckdbError+ "SizedArray"+ (sized mismatch :: SizedArray 2 (Maybe (StructValue FieldValue)))+ ErrorQueryNotImplemented+ "Cannot combine types"+ unions <-+ mapM+ (\sql -> query_ conn sql :: IO [Only (UnionValue FieldValue)])+ ["SELECT union_value(a := 1)::UNION(a INTEGER)", "SELECT union_value(b := 'x')::UNION(b VARCHAR)"]+ duckdbError "UNION list" (map fromOnly (concat unions)) ErrorTypeConversion "is not present in target union"+ duckdbError+ "ARRAY sizes"+ ( FieldList+ [FieldArray (listArray (0, 1) [FieldInt32 1, FieldInt32 2]), FieldArray (listArray (0, 2) [FieldInt32 1, FieldInt32 2, FieldInt32 3])]+ )+ ErrorInputInvalid+ "Cannot cast array of size 2 to array of size 3"+ [Just first, Just second] <- mapM elementOf ["{'a': 1}", "{'b': 2}"]+ let mapped = Map.fromList [("k1" :: Text, first), ("k2", second)]+ (query conn "SELECT typeof(?), ?::VARCHAR" (mapped, mapped) :: IO [(Text, Text)])+ >>= (@?= [("MAP(VARCHAR, STRUCT(a INTEGER, b INTEGER))", "{k1={'a': 1, 'b': NULL}, k2={'a': NULL, 'b': 2}}")])+ -- The column type does not change the type of the list, as in SQL.+ _ <- execute_ conn "CREATE TABLE wide (v STRUCT(a BIGINT, b BIGINT)[]); CREATE TABLE named (v STRUCT(a VARCHAR)[])"+ _ <- execute conn "INSERT INTO wide VALUES (?)" (Only [first, second])+ (query_ conn "SELECT v::VARCHAR FROM wide" :: IO [Only Text]) >>= (@?= [Only "[{'a': 1, 'b': NULL}, {'a': NULL, 'b': 2}]"])+ try (execute conn "INSERT INTO named VALUES (?)" (Only mismatch)) >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorQueryNotImplemented+ Right count -> assertFailure ("expected the error of DuckDB, stored " <> show count)+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "collection inference keeps aliases and collations as native common types do" $ withConnection ":memory:" \conn -> do+ let check :: (ToField a) => String -> a -> Text -> Assertion+ check label value literal =+ (query conn (Query ("SELECT ?, " <> literal)) (Only value) :: IO [(FieldValue, FieldValue)]) >>= \case+ [(actual, expected)] -> assertEqual label expected actual+ rows -> assertFailure (show rows)+ forM_+ [ ("{'k': '{\"n\":1}'::VARCHAR}", "{'k': '{\"n\":2}'::JSON}", True)+ , ("{'k': 'A'::VARCHAR}", "{'k': 'B' COLLATE nocase}", False)+ ]+ \(plainSql, annotatedSql, compareSQL) -> do+ [Only plain] <- query_ conn (Query ("SELECT " <> plainSql)) :: IO [Only (StructValue FieldValue)]+ [Only annotated] <- query_ conn (Query ("SELECT " <> annotatedSql)) :: IO [Only (StructValue FieldValue)]+ forM_ [(plain, annotated, plainSql, annotatedSql), (annotated, plain, annotatedSql, plainSql)] \(first, second, firstSql, secondSql) -> do+ let items = firstSql <> ", " <> secondSql+ array = FieldArray (listArray (0, 1) [FieldStruct first, FieldStruct second])+ mapping = Map.fromList [("a" :: Text, first), ("b", second)]+ when compareSQL do+ check "LIST" [first, second] ("[" <> items <> "]")+ check "ARRAY" array ("array_value(" <> items <> ")")+ check "MAP" mapping ("map(['a', 'b'], [" <> items <> "])")+ -- SQL expression binding can discard collations before common+ -- type inference. Compare the preserved input types directly.+ common <- withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ bracket (logicalTypeFromRep factory (structValueTypeRep first)) destroyLogicalType \left ->+ bracket (logicalTypeFromRep factory (structValueTypeRep second)) destroyLogicalType \right ->+ withArray [left, right] \types ->+ bracket (R.output (c_duckdb_v2_factory_create_common_type factory (ConstPtr types) 2)) destroyLogicalType (logicalTypeToRep factory)+ let checkType :: (ToField a) => a -> LogicalTypeRep -> Assertion+ checkType value expected =+ (query conn "SELECT ?" (Only value) :: IO [Only FieldTypeOf]) >>= (@?= [Only (FieldTypeOf expected)])+ checkType [first, second] (LogicalTypeList common)+ checkType array (LogicalTypeArray common 2)+ checkType mapping (LogicalTypeMap (LogicalTypeScalar LogicalTypeIdVarchar) common)+ , testCase "collection inference preserves exact metadata at native equality boundaries" $ withConnection ":memory:" \conn -> do+ let integer = LogicalTypeScalar LogicalTypeIdInteger+ varchar = LogicalTypeScalar LogicalTypeIdVarchar+ record name = LogicalTypeStruct (listArray (0, 0) [StructField name integer])+ member name rep = LogicalTypeUnion (listArray (0, 0) [UnionMemberType name rep])+ tuple rep = LogicalTypeTuple (listArray (0, 0) [rep])+ enum labels = LogicalTypeEnum (listArray (0, length labels - 1) labels)+ wkt =+ "ENGCRS[\"Local grid\",EDATUM[\"Local datum\"],CS[Cartesian,2],AXIS[\"x\",east,ORDER[1],LENGTHUNIT[\"metre\",1]],AXIS[\"y\",north,ORDER[2],LENGTHUNIT[\"metre\",1]]]"+ geometry = LogicalTypeGeometry . Just+ wrappers =+ [ ("STRUCT", id)+ , ("LIST", LogicalTypeList)+ , ("ARRAY", \rep -> LogicalTypeArray rep 1)+ , ("MAP key", \rep -> LogicalTypeMap rep integer)+ , ("MAP value", LogicalTypeMap varchar)+ , ("TUPLE", tuple)+ , ("UNION", member "value")+ , ("alias", LogicalTypeAlias "nested_alias")+ ]+ cases =+ [(label <> " field spelling", wrap (record "a"), wrap (record "A"), True, False) | (label, wrap) <- wrappers]+ <> [ ("collation", tuple varchar, tuple (LogicalTypeCollatedVarchar "nocase"), True, False)+ , ("UNION spelling", member "a" integer, member "A" integer, True, False)+ , ("CRS definition", geometry wkt, geometry (Text.replace ",EDATUM" ", EDATUM" wkt), True, False)+ , ("ENUM label", enum ["A", "b"], enum ["a", "b"], False, False)+ , ("ENUM order", enum ["a", "b"], enum ["b", "a"], False, False)+ , ("same ENUM", enum ["a", "b"], enum ["a", "b"], True, True)+ , ("alias spelling", LogicalTypeAlias "alias" integer, LogicalTypeAlias "Alias" integer, False, True)+ , ("empty STRUCT", LogicalTypeStruct (listArray (0, -1) []), LogicalTypeStruct (listArray (0, -1) []), True, True)+ , ("empty TUPLE", LogicalTypeTuple (listArray (0, -1) []), LogicalTypeTuple (listArray (0, -1) []), True, True)+ ]+ check :: (ToField a) => String -> a -> LogicalTypeRep -> Assertion+ check label value expected =+ (query conn "SELECT ?" (Only value) :: IO [Only FieldTypeOf]) >>= assertEqual label [Only (FieldTypeOf expected)]+ forM_ cases \(label, first, second, nativeEqual, baseEqual) ->+ forM_ [(first, second), (second, first)] \(leftRep, rightRep) -> do+ expected <- withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ bracket (logicalTypeFromRep factory leftRep) destroyLogicalType \left ->+ bracket (logicalTypeFromRep factory rightRep) destroyLogicalType \right -> do+ same <- (/= 0) <$> R.output (c_duckdb_v2_logical_type_is_equal left right)+ assertEqual (label <> " native equality") nativeEqual same+ sameBase <- (==) <$> logicalTypeToBaseRep left <*> logicalTypeToBaseRep right+ assertEqual (label <> " base equality") baseEqual sameBase+ -- Equal native types retain their exact metadata. Other types use native inference.+ if same && sameBase+ then logicalTypeToRep factory left+ else withArray [left, right] \types ->+ bracket (R.output (c_duckdb_v2_factory_create_common_type factory (ConstPtr types) 2)) destroyLogicalType (logicalTypeToRep factory)+ let items = [TypedStructNull leftRep, TypedStructNull rightRep]+ check label items (LogicalTypeList expected)+ check label (sized items :: SizedArray 2 UnadaptedStruct) (LogicalTypeArray expected 2)+ check label (Map.fromList (zip ["a" :: Text, "b"] items)) (LogicalTypeMap varchar expected)+ , testCase "collection inference keeps typed NULL aliases and collations" $ withConnection ":memory:" \conn -> do+ let varchar = LogicalTypeScalar LogicalTypeIdVarchar+ plain = singleField varchar (FieldText "{}")+ plainRep = structValueTypeRep plain+ check :: (ToField a) => a -> LogicalTypeRep -> Assertion+ check value expected =+ (query conn "SELECT ?" (Only value) :: IO [Only FieldTypeOf]) >>= (@?= [Only (FieldTypeOf expected)])+ cases <- withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ bracket (logicalTypeFromRep factory plainRep) destroyLogicalType \left ->+ fmap concat $ forM+ [ structValueTypeRep (singleField (LogicalTypeAlias "JSON" varchar) FieldNull)+ , structValueTypeRep (singleField (LogicalTypeCollatedVarchar "nocase") FieldNull)+ , LogicalTypeAlias "STRUCT" plainRep+ ]+ \rep -> bracket (logicalTypeFromRep factory rep) destroyLogicalType \right ->+ forM+ [ ([UnadaptedStruct plain, TypedStructNull rep], [left, right])+ , ([TypedStructNull rep, UnadaptedStruct plain], [right, left])+ , ([TypedStructNull plainRep, TypedStructNull rep], [left, right])+ , ([TypedStructNull rep, TypedStructNull plainRep], [right, left])+ , ([TypedStructNull rep], [right])+ ]+ \(items, types) -> do+ common <- withArray types \nativeTypes ->+ bracket+ (R.output (c_duckdb_v2_factory_create_common_type factory (ConstPtr nativeTypes) (fromIntegral (length types))))+ destroyLogicalType+ (logicalTypeToRep factory)+ pure (items, common)+ forM_ cases \(items, common) -> do+ check items (LogicalTypeList common)+ if length items == 2+ then check (sized items :: SizedArray 2 UnadaptedStruct) (LogicalTypeArray common 2)+ else check (sized items :: SizedArray 1 UnadaptedStruct) (LogicalTypeArray common 1)+ check (Map.fromList (zip ["a" :: Text, "b"] items)) (LogicalTypeMap varchar common)+ -- SQLNULL can itself have its canonical name as an alias. It must not+ -- be discarded with an ordinary untyped NULL.+ let canonicalNull = LogicalTypeAlias "NULL" (LogicalTypeScalar LogicalTypeIdSQLNull)+ check [Just (TypedStructNull canonicalNull), Nothing] (LogicalTypeList canonicalNull)+ , testCase "collection inference keeps canonical aliases in both element orders" $ withConnection ":memory:" \conn -> do+ let integer = LogicalTypeScalar LogicalTypeIdInteger+ canonical = LogicalTypeAlias "INTEGER" integer+ plain = singleField integer (FieldInt32 7)+ annotated = singleField canonical (FieldInt32 7)+ element = structValueTypeRep annotated+ check :: (ToField a) => a -> LogicalTypeRep -> Assertion+ check value expected = do+ [Only actual] <- query conn "SELECT ?" (Only value) :: IO [Only FieldTypeOf]+ actual @?= FieldTypeOf expected+ forM_ [[plain, annotated], [annotated, plain]] \items -> do+ check items (LogicalTypeList element)+ check (FieldArray (listArray (0, 1) (map FieldStruct items))) (LogicalTypeArray element 2)+ check (Map.fromList (zip ["a" :: Text, "b"] items)) (LogicalTypeMap (LogicalTypeScalar LogicalTypeIdVarchar) element)+ , testCase "mixed temporal inference preserves aliases on unchanged nested leaves" $ withConnection ":memory:" \conn -> do+ let varchar = LogicalTypeScalar LogicalTypeIdVarchar+ json = LogicalTypeAlias "JSON" varchar+ time = LogicalTypeScalar LogicalTypeIdTime+ nanos = LogicalTypeScalar LogicalTypeIdTimeNs+ structure textType timeType value =+ structOf [("k", FieldText "{}", textType), ("t", FieldTime value, timeType)]+ coarse textType = structure textType time (TimeOfDay 3 4 5)+ precise textType = structure textType nanos (TimeOfDay 3 4 5.123456789)+ aliasedCoarse = structure json (LogicalTypeAlias "TIME" time) (TimeOfDay 3 4 5)+ target = "STRUCT(k JSON, t TIME_NS)[]"+ check items =+ (query conn (Query ("SELECT ?::" <> target)) (Only (map UnadaptedStruct items)) :: IO [Only [StructValue FieldValue]])+ forM_+ [ ([coarse json], [structure json nanos (TimeOfDay 3 4 5)])+ , ([precise json], [precise json])+ , ([coarse json, coarse json], replicate 2 (structure json nanos (TimeOfDay 3 4 5)))+ , ([precise json, precise json], [precise json, precise json])+ , ([coarse json, precise json], [structure json nanos (TimeOfDay 3 4 5), precise json])+ , ([precise json, coarse json], [precise json, structure json nanos (TimeOfDay 3 4 5)])+ ]+ \(items, expected) -> check items >>= (@?= [Only expected])+ -- Keep the native cast failure as a control for public vector writes.+ nativeAlias <- try $ withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ bracket (logicalTypeFromRep factory (LogicalTypeAlias "TIME" time)) destroyLogicalType \logical ->+ withOwnedVector factory logical \vector -> do+ R.checked (c_duckdb_v2_vector_set_size vector 1)+ bracket (toDuckValue factory (TimeOfDay 3 4 5)) destroyValue \value ->+ R.checked (c_duckdb_v2_vector_set_value vector 0 value)+ let rejectsAlias :: Either SQLError a -> Assertion+ rejectsAlias result = case result of+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorType @?= Just ErrorInputInvalid+ assertBool (Text.unpack sqlErrorMessage) ("Unimplemented type for cast (TIME -> TIME)" `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure "expected the native TIME alias cast limit"+ rejectsAlias nativeAlias+ nativeCast <- try $ withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ bracket (logicalTypeFromRep factory (LogicalTypeAlias "TIME" time)) destroyLogicalType \logical ->+ bracket (toDuckValue factory (TimeOfDay 3 4 5)) destroyValue \value ->+ bracket (R.output (c_duckdb_v2_value_cast factory value logical)) destroyValue (const (pure ()))+ rejectsAlias nativeCast+ check [aliasedCoarse, precise json] >>= (@?= [Only [structure json nanos (TimeOfDay 3 4 5), precise json]])+ check [precise json, aliasedCoarse] >>= (@?= [Only [precise json, structure json nanos (TimeOfDay 3 4 5)]])+ let plainSql = "{'k': '{}'::VARCHAR, 't': TIME '03:04:05'}"+ annotatedSql = "{'k': '{}'::JSON, 't': '03:04:05.123456789'::TIME_NS}"+ forM_+ [ ([coarse varchar, precise json], plainSql <> ", " <> annotatedSql)+ , ([precise json, coarse varchar], annotatedSql <> ", " <> plainSql)+ ]+ \(items, sqls) -> do+ native <- try (query_ conn (Query ("SELECT array_value(" <> sqls <> ")")) :: IO [Only FieldValue])+ bound <- try (check items)+ let rejected label result = case result of+ Left SQLError{sqlErrorMessage} ->+ assertBool (label <> ": " <> Text.unpack sqlErrorMessage) ("Cannot combine types" `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure (label <> ": expected the native common-type error")+ rejected "native SQL" native+ rejected "bound list" bound+ , testCase "TIME aliases bind exact scalar and nested payloads under session settings" $ withConnection ":memory:" \conn -> do+ let time = LogicalTypeScalar LogicalTypeIdTime+ varchar = LogicalTypeScalar LogicalTypeIdVarchar+ check rep payload = do+ let original = singleField rep payload+ (query conn "SELECT ?" (Only original) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only original])+ forM_ ["UTC", "Europe/Stockholm", "America/New_York"] \zone -> do+ _ <- execute conn "SET TimeZone = ?" (Only (zone :: Text))+ forM_ ["TIME", "clock_alias"] \name -> do+ let alias = LogicalTypeAlias name time+ members = listArray (0, 0) [UnionMemberType "t" alias]+ forM_+ [ FieldNull+ , FieldTime (TimeOfDay 0 0 0)+ , FieldTime (TimeOfDay 0 0 0.000001)+ , FieldTime (TimeOfDay 12 34 56.123456)+ , FieldTime (TimeOfDay 23 59 59.999999)+ , FieldTime (TimeOfDay 24 0 0)+ ]+ \payload -> do+ check alias payload+ check (LogicalTypeList alias) (FieldList [payload, FieldNull])+ check (LogicalTypeArray alias 2) (FieldArray (listArray (0, 1) [payload, FieldNull]))+ check (LogicalTypeMap varchar alias) (FieldMap [(FieldText "t", payload)])+ let nested = singleField alias payload+ check (structValueTypeRep nested) (FieldStruct nested)+ let union = UnionValue 0 "t" payload members+ check (unionValueTypeRep union) (FieldUnion union)+ let precise = singleField alias (FieldTime (TimeOfDay 12 34 56.123456789))+ truncated = singleField alias (FieldTime (TimeOfDay 12 34 56.123456))+ (query conn "SELECT ?" (Only precise) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only truncated])+ forM_ [TimeOfDay 25 0 0, TimeOfDay 24 0 0.000001, TimeOfDay 12 60 0] \invalid ->+ assertRangeError (query conn "SELECT ?" (Only (singleField alias (FieldTime invalid))) :: IO [Only FieldValue])+ , testCase "a LIST parameter fails at an element of the wrong type after earlier elements" $ withConnection ":memory:" \conn -> do+ let struct =+ singleField (LogicalTypeList (LogicalTypeScalar LogicalTypeIdBigInt)) (FieldList [FieldInt64 1, FieldInt64 2, FieldText "x", FieldInt64 4])+ result <- try (query conn "SELECT ?" (Only struct) :: IO [Only FieldValue])+ case result of+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ sqlErrorType @?= Just ErrorInputInvalid+ assertBool+ (Text.unpack sqlErrorMessage)+ ("unsupported scalar conversion for BIGINT from a value of type VARCHAR" `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure "expected a conversion error for the third element"+ -- A value of another kind than the declared type names its SQL type.+ try+ (query conn "SELECT ?" (Only (singleField (LogicalTypeList (LogicalTypeScalar LogicalTypeIdBigInt)) (FieldInt32 7))) :: IO [Only FieldValue])+ >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ sqlErrorType @?= Just ErrorInputInvalid+ assertBool (Text.unpack sqlErrorMessage) ("cannot encode a value of type INTEGER as BIGINT[]" `Text.isSuffixOf` sqlErrorMessage)+ Right rows -> assertFailure ("expected the encode error, got " <> show rows)+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "a declared NULL type binds NULL and rejects a value that is not NULL" $ withConnection ":memory:" \conn -> do+ let sqlNull = LogicalTypeScalar LogicalTypeIdSQLNull+ nullStruct = singleField sqlNull FieldNull+ (query conn "SELECT typeof(?), CAST(? AS VARCHAR)" (nullStruct, nullStruct) :: IO [(Text, Text)])+ >>= (@?= [("STRUCT(\"value\" \"NULL\")", "{'value': NULL}")])+ result <- try (query conn "SELECT ?" (Only (singleField sqlNull (FieldInt32 7))) :: IO [Only FieldValue])+ case result of+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ sqlErrorType @?= Just ErrorInputInvalid+ assertBool+ (Text.unpack sqlErrorMessage)+ ("unsupported scalar conversion for \"NULL\" from a value of type INTEGER" `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure "expected a conversion error for a value that is not NULL"+ , testCase "GEOMETRY decodes as well-known binary" $ withConnection ":memory:" \conn -> do+ [Only bytes] <- query_ conn "SELECT 'POINT(1 2)'::GEOMETRY" :: IO [Only BS.ByteString]+ BS.length bytes @?= 21+ (query conn "SELECT ST_AsText(ST_GeomFromWKB(?))" (Only bytes) :: IO [Only Text]) >>= (@?= [Only "POINT (1 2)"])+ , testCase "Float parameter retains FLOAT type" $ withConnection ":memory:" \conn -> do+ (query conn "SELECT typeof(?), ?" (1.25 :: Float, 1.25 :: Float) :: IO [(Text, Float)]) >>= (@?= [("FLOAT", 1.25)])+ , testCase "Float special values survive decoding" $ withConnection ":memory:" \conn -> do+ [Only value] <- query_ conn "SELECT 'NaN'::REAL" :: IO [Only Float]+ assertBool "expected NaN" (isNaN value)+ [Only value'] <- query_ conn "SELECT 'Infinity'::REAL" :: IO [Only Double]+ assertBool "expected positive infinity" (isInfinite value' && value' > 0)+ , testCase "finite Double overflow to Float fails" $ withConnection ":memory:" \conn ->+ assertConversionError (query_ conn "SELECT 1e300::DOUBLE" :: IO [Only Float])+ , testCase "embedded NUL and Unicode Text round trip" $ withConnection ":memory:" \conn -> do+ let value = "before\0after íslenska λ 😀" :: Text+ (query conn "SELECT ?" (Only value) :: IO [Only Text]) >>= (@?= [Only value])+ , testCase "VARCHAR, BLOB, and BIT values on both sides of the inline length read back" $ withConnection ":memory:" \conn -> do+ -- DuckDB keeps a value of at most 12 bytes inline and a longer value+ -- behind a pointer. A BIT value has one more byte for its padding count.+ let lengths = [0, 1, 11, 12, 13, 64]+ texts = [Text.pack (take n (cycle ['a' .. 'z'])) | n <- lengths]+ blobs = [BS.pack (take n (cycle [0, 255, 1, 128])) | n <- lengths]+ bits = [bsFromBool (take n (cycle [True, False, False])) | n <- [1, 8, 87, 88, 89, 200]]+ present = [(Just text, Just blob, Just bit) | (text, blob, bit) <- zip3 texts blobs bits]+ -- The NULL row is between the values of 11 and 12 bytes.+ rows = take 3 present <> [(Nothing, Nothing, Nothing)] <> drop 3 present+ _ <- execute_ conn "CREATE TABLE inline_boundary (i INTEGER, t VARCHAR, b BLOB, bits BIT)"+ _ <-+ executeMany+ conn+ "INSERT INTO inline_boundary VALUES (?, ?, ?, ?)"+ [(i, text, blob, bit) | (i, (text, blob, bit)) <- zip [0 :: Int32 ..] rows]+ (query_ conn "SELECT t, b, bits FROM inline_boundary ORDER BY i" :: IO [(Maybe Text, Maybe BS.ByteString, Maybe BitString)])+ >>= (@?= rows)+ (query_ conn "SELECT list(t ORDER BY i) FROM inline_boundary" :: IO [Only [Maybe Text]])+ >>= (@?= [Only [text | (text, _, _) <- rows]])+ , testCase "Int8 rejects both overflow directions" $ withConnection ":memory:" \conn -> do+ forM_ ["SELECT 128::BIGINT", "SELECT -129::SMALLINT"] \sql ->+ assertConversionError (query_ conn sql :: IO [Only Int8])+ (query_ conn "SELECT -128::BIGINT UNION ALL SELECT 127" :: IO [Only Int8]) >>= (@?= [Only (-128), Only 127])+ , testCase "unsigned narrowing rejects overflow" $ withConnection ":memory:" \conn -> do+ assertConversionError (query_ conn "SELECT 256::USMALLINT" :: IO [Only Word8])+ assertConversionError (query_ conn "SELECT 65536::UINTEGER" :: IO [Only Word16])+ assertConversionError (query_ conn "SELECT 4294967296::UBIGINT" :: IO [Only Word32])+ , testCase "a parameter for a column of a type from CREATE TYPE follows the base type" $ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TYPE tsns AS TIMESTAMP_NS; CREATE TYPE wide AS UHUGEINT; CREATE TABLE aliased (t tsns, w wide)"+ let precise = LocalTime (fromGregorian 2024 1 2) (TimeOfDay 3 4 5.123456789)+ _ <- execute conn "INSERT INTO aliased VALUES (?, ?)" (precise, 2 ^ (127 :: Int) + 5 :: Integer)+ (query_ conn "SELECT t::VARCHAR, w::VARCHAR FROM aliased" :: IO [(Text, Text)])+ >>= (@?= [("2024-01-02 03:04:05.123456789", "170141183460469231731687303715884105733")])+ , testCase "Integer and Natural parameters store exact values for each numeric target" $ withConnection ":memory:" \conn -> do+ let store :: (Show a, ToField a) => Text -> a -> Text -> Assertion+ store dtype value expected = do+ let label = Text.unpack dtype <> " " <> show value+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE exact_value (v " <> dtype <> ")"))+ _ <- execute conn "INSERT INTO exact_value VALUES (?)" (Only value)+ (query_ conn "SELECT v::VARCHAR FROM exact_value" :: IO [Only Text]) >>= assertEqual label [Only expected]+ (query conn "SELECT count(*) FROM exact_value WHERE v = ?" (Only value) :: IO [Only Int64]) >>= assertEqual label [Only 1]+ huge = 2 ^ (128 :: Int) + 5 :: Integer+ forM_ ["TINYINT", "SMALLINT", "INTEGER", "BIGINT", "UBIGINT", "HUGEINT", "UHUGEINT", "DECIMAL(38,0)", "DOUBLE", "VARCHAR", "BIGNUM"] \dtype -> do+ let expected = if dtype == "DOUBLE" then "5.0" else "5"+ store dtype (5 :: Integer) expected+ store dtype (5 :: Natural) expected+ forM_ ["TINYINT", "SMALLINT", "INTEGER", "BIGINT", "HUGEINT", "DECIMAL(38,0)", "DOUBLE", "VARCHAR", "BIGNUM"] \dtype ->+ store dtype (-5 :: Integer) (if dtype == "DOUBLE" then "-5.0" else "-5")+ forM_ ["HUGEINT", "UHUGEINT", "DECIMAL(38,0)", "BIGNUM", "VARCHAR"] \dtype ->+ store dtype (2 ^ (100 :: Int) :: Integer) "1267650600228229401496703205376"+ store "HUGEINT" (negate (2 ^ (127 :: Int)) :: Integer) "-170141183460469231731687303715884105728"+ store "HUGEINT" (2 ^ (127 :: Int) - 1 :: Integer) "170141183460469231731687303715884105727"+ forM_ ["UHUGEINT", "BIGNUM"] \dtype -> do+ store dtype (2 ^ (127 :: Int) :: Integer) "170141183460469231731687303715884105728"+ store dtype (2 ^ (127 :: Int) :: Natural) "170141183460469231731687303715884105728"+ store dtype (2 ^ (128 :: Int) - 1 :: Integer) "340282366920938463463374607431768211455"+ store dtype (2 ^ (128 :: Int) - 1 :: Natural) "340282366920938463463374607431768211455"+ forM_ ["BIGNUM", "VARCHAR"] \dtype -> forM_ [huge, negate huge] \value ->+ store dtype value (Text.pack (show value))+ , testCase "an Integer parameter that its target cannot hold raises an error and stores no row" $ withConnection ":memory:" \conn -> do+ let insertValue :: (Show a, ToField a) => Text -> a -> IO (Either SQLError Int)+ insertValue dtype value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE rejected_value (v " <> dtype <> ")"))+ result <- try (execute conn "INSERT INTO rejected_value VALUES (?)" (Only value))+ (query_ conn "SELECT count(*) FROM rejected_value" :: IO [Only Int64]) >>= assertEqual (Text.unpack dtype <> " " <> show value) [Only 0]+ pure result+ insert :: Text -> Integer -> IO (Either SQLError Int)+ insert = insertValue+ huge = 2 ^ (128 :: Int) + 5 :: Integer+ -- DuckDB checks the range of a cast from HUGEINT.+ forM_+ ( [(dtype, -5) | dtype <- ["UTINYINT", "USMALLINT", "UINTEGER", "UBIGINT", "UHUGEINT"]]+ <> [("BIGINT", 2 ^ (70 :: Int)), ("DECIMAL(10,0)", 2 ^ (100 :: Int))]+ )+ \(dtype, value) ->+ insert dtype value >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertBool (Text.unpack sqlErrorMessage) (not ("duckdb-simple:" `Text.isPrefixOf` sqlErrorMessage))+ sqlErrorType @?= Just ErrorTypeConversion+ Right _ -> assertFailure ("expected a DuckDB error for " <> Text.unpack dtype <> " " <> show value)+ -- duckdb-simple checks the range of a value outside the HUGEINT range.+ forM_+ ( [("HUGEINT", 2 ^ (127 :: Int)), ("BIGINT", 2 ^ (127 :: Int)), ("UHUGEINT", 2 ^ (128 :: Int))]+ <> [(dtype, value) | dtype <- ["TINYINT", "BIGINT", "UBIGINT", "HUGEINT", "UHUGEINT", "DECIMAL(38,0)"], value <- [huge, negate huge]]+ )+ \(dtype, value) ->+ insert dtype value >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorType @?= Just ErrorTypeConversion+ sqlErrorMessage @?= "duckdb-simple: the value " <> Text.pack (show value) <> " is out of range for a parameter of the type " <> dtype+ Right _ -> assertFailure ("expected the duckdb-simple range error for " <> Text.unpack dtype <> " " <> show value)+ insertValue "BIGINT" (2 ^ (128 :: Int) + 5 :: Natural) >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorType @?= Just ErrorTypeConversion+ sqlErrorMessage @?= "duckdb-simple: the value 340282366920938463463374607431768211461 is out of range for a parameter of the type BIGINT"+ Right _ -> assertFailure "expected the duckdb-simple range error for a Natural BIGINT value"+ _ <- execute_ conn "CREATE TABLE wrapped_value (v UBIGINT)"+ _ <- execute_ conn "INSERT INTO wrapped_value VALUES (18446744073709551611), (5)"+ (query conn "SELECT count(*) FROM wrapped_value WHERE v = ?" (Only (-5 :: Integer)) :: IO [Only Int64]) >>= (@?= [Only 0])+ , testCase "the SQL type of an Integer parameter follows its range" $ withConnection ":memory:" \conn -> do+ let typed :: (ToField a) => a -> Integer -> Text -> Assertion+ typed value expected dtype = (query conn "SELECT typeof(?), ?" (value, value) :: IO [(Text, Integer)]) >>= (@?= [(dtype, expected)])+ typed (5 :: Integer) 5 "HUGEINT"+ typed (negate (2 ^ (127 :: Int)) :: Integer) (negate (2 ^ (127 :: Int))) "HUGEINT"+ typed (5 :: Natural) 5 "HUGEINT"+ typed (2 ^ (127 :: Int) :: Integer) (2 ^ (127 :: Int)) "BIGNUM"+ typed (2 ^ (130 :: Int) :: Integer) (2 ^ (130 :: Int)) "BIGNUM"+ typed (BigNum 5) 5 "BIGNUM"+ , testCase "executeMany stores Integer rows of each range" $ withConnection ":memory:" \conn ->+ forM_ [("BIGNUM", [5, 2 ^ (127 :: Int), 2 ^ (130 :: Int)]), ("HUGEINT", [5, 2 ^ (100 :: Int)])] \(dtype, values :: [Integer]) -> do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE many_values (v " <> dtype <> ")"))+ _ <- executeMany conn "INSERT INTO many_values VALUES (?)" (map Only values)+ (query_ conn "SELECT v FROM many_values ORDER BY rowid" :: IO [Only Integer]) >>= (@?= map Only values)+ , testCase "an array of Integer values takes the type that its elements give" $ withConnection ":memory:" \conn ->+ -- See Note [Integer parameters] in "Database.DuckDB.Simple.ToField".+ forM_ [(sized [5, 2 ^ (130 :: Int)], "BIGNUM[2]"), (sized [5, 6], "HUGEINT[2]")] \(values :: SizedArray 2 Integer, dtype) ->+ (query conn "SELECT typeof(?), ?" (values, values) :: IO [(Text, SizedArray 2 Integer)]) >>= (@?= [(dtype, values)])+ , testCase "an array of Integer values takes the element type of its target" $ withConnection ":memory:" \conn -> do+ let store :: (Show a, ToField a) => Text -> a -> Text -> Assertion+ store dtype value expected = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE array_value (v " <> dtype <> ")"))+ _ <- execute conn "INSERT INTO array_value VALUES (?)" (Only value)+ (query_ conn "SELECT v::VARCHAR FROM array_value" :: IO [Only Text])+ >>= assertEqual (Text.unpack dtype <> " " <> show value) [Only expected]+ pair :: Integer -> Integer -> SizedArray 2 Integer+ pair first second = sized [first, second]+ huge = 2 ^ (128 :: Int) + 7 :: Integer+ forM_ ["BIGINT[2]", "UBIGINT[2]", "HUGEINT[2]", "UHUGEINT[2]", "DECIMAL(38,0)[2]", "BIGNUM[2]", "VARCHAR[2]"] \dtype ->+ store dtype (pair 5 6) "[5, 6]"+ forM_ ["BIGINT[2]", "HUGEINT[2]", "BIGNUM[2]"] \dtype ->+ store dtype (pair 5 (-6)) "[5, -6]"+ store "UHUGEINT[2]" (sized [5, 2 ^ (127 :: Int)] :: SizedArray 2 Natural) "[5, 170141183460469231731687303715884105728]"+ forM_ ["BIGNUM[2]", "VARCHAR[2]"] \dtype ->+ store dtype (pair 5 huge) "[5, 340282366920938463463374607431768211463]"+ store "BIGINT[2][2]" (sized [pair 1 2, pair 3 (-4)] :: SizedArray 2 (SizedArray 2 Integer)) "[[1, 2], [3, -4]]"+ store "BIGINT[]" (pair 5 6) "[5, 6]"+ , testCase "an array of Integer values with an element that its target cannot hold raises an error and stores no row" $ withConnection ":memory:" \conn -> do+ let insert :: Text -> Integer -> IO SQLError+ insert dtype second = do+ let label = Text.unpack dtype <> " [5, " <> show second <> "]"+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE rejected_array (v " <> dtype <> ")"))+ result <- try (execute conn "INSERT INTO rejected_array VALUES (?)" (Only (sized [5, second] :: SizedArray 2 Integer)))+ (query_ conn "SELECT count(*) FROM rejected_array" :: IO [Only Int64]) >>= assertEqual label [Only 0]+ either pure (\_ -> assertFailure ("expected an error for " <> label)) result+ huge = 2 ^ (128 :: Int) + 7 :: Integer+ -- DuckDB checks the range of each element when it builds the array.+ forM_ ([(dtype, -6) | dtype <- ["UTINYINT[2]", "UBIGINT[2]", "UHUGEINT[2]", "UBIGINT[]"]] <> [("TINYINT[2]", 300)]) \(dtype, second) -> do+ SQLError{sqlErrorMessage, sqlErrorType} <- insert dtype second+ assertBool (Text.unpack sqlErrorMessage) ("is out of range for the destination type" `Text.isInfixOf` sqlErrorMessage)+ sqlErrorType @?= Just ErrorInputInvalid+ -- duckdb-simple checks the range of an element outside the HUGEINT range.+ forM_ [(dtype, second) | dtype <- ["BIGINT[2]", "UBIGINT[2]", "HUGEINT[2]", "DECIMAL(38,0)[2]"], second <- [huge, negate huge]] \(dtype, second) -> do+ SQLError{sqlErrorType} <- insert dtype second+ sqlErrorType @?= Just ErrorTypeConversion+ , testCase "a list binds as a LIST parameter in each SQL form" $ withConnection ":memory:" \conn -> do+ let ints = [1, 2, 3] :: [Int64]+ none = [] :: [Int64]+ count :: (ToRow p) => Query -> p -> IO [Only Int64]+ count = query conn+ forM_ [ints, none] \values ->+ (query conn "SELECT typeof(?), ?" (values, values) :: IO [(Text, [Int64])]) >>= (@?= [("BIGINT[]", values)])+ _ <- execute_ conn "CREATE TABLE list_value (i BIGINT, v BIGINT[])"+ _ <- executeMany conn "INSERT INTO list_value VALUES (?, ?)" [(0 :: Int64, ints), (1, none)]+ (query_ conn "SELECT v FROM list_value ORDER BY i" :: IO [Only [Int64]]) >>= (@?= [Only ints, Only none])+ count "SELECT count(*) FROM list_value WHERE v = ?" (Only ints) >>= (@?= [Only 1])+ count "SELECT count(*) FROM list_value WHERE v = ?" (Only none) >>= (@?= [Only 1])+ -- DuckDB casts a LIST to an ARRAY of the same length only.+ _ <- execute_ conn "CREATE TABLE array_column (v BIGINT[3])"+ _ <- execute conn "INSERT INTO array_column VALUES (?)" (Only ints)+ (query_ conn "SELECT v FROM array_column" :: IO [Only [Int64]]) >>= (@?= [Only ints])+ count "SELECT count(*) FROM array_column WHERE v = ?" (Only ints) >>= (@?= [Only 1])+ forM_ [[1, 2], none] \values -> do+ result <- try (execute conn "INSERT INTO array_column VALUES (?)" (Only values))+ case result of+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertBool (Text.unpack sqlErrorMessage) ("Conversion Error: Cannot cast list with length " `Text.isPrefixOf` sqlErrorMessage)+ sqlErrorType @?= Just ErrorTypeConversion+ Right (_ :: Int) -> assertFailure ("expected the DuckDB length error for " <> show values)+ (query_ conn "SELECT count(*) FROM array_column" :: IO [Only Int64]) >>= (@?= [Only 1])+ _ <- execute_ conn "CREATE TABLE searched (v BIGINT)"+ _ <- execute_ conn "INSERT INTO searched VALUES (1), (2), (5)"+ forM_+ [ "SELECT count(*) FROM searched WHERE v = ANY(?)"+ , "SELECT count(*) FROM searched WHERE v IN (SELECT unnest(?))"+ , "SELECT count(*) FROM searched WHERE list_contains(?, v)"+ ]+ \sql -> do+ count sql (Only ([1, 5] :: [Int64])) >>= (@?= [Only 2])+ count sql (Only none) >>= (@?= [Only 0])+ , testCase "a list binds Maybe, nested, Text, String, and UTC elements" $ withConnection ":memory:" \conn -> do+ (query conn "SELECT typeof(?), (?)::VARCHAR" ([Just 1, Nothing] :: [Maybe Int64], [Just 1, Nothing] :: [Maybe Int64]) :: IO [(Text, Text)])+ >>= (@?= [("BIGINT[]", "[1, NULL]")])+ let nested = [[1, 2], [], [3]] :: [[Int64]]+ (query conn "SELECT typeof(?), ?" (nested, nested) :: IO [(Text, [[Int64]])]) >>= (@?= [("BIGINT[][]", nested)])+ -- A String stays one VARCHAR parameter next to a list of Text.+ let labels = ["a", "b"] :: [Text]+ label = "ab" :: String+ (query conn "SELECT typeof(?), ?, typeof(?), ?" (labels, labels, label, label) :: IO [(Text, [Text], Text, String)])+ >>= (@?= [("VARCHAR[]", labels, "VARCHAR", label)])+ (query conn "SELECT typeof(?), typeof(?)" (["x"] :: [String], [] :: [String]) :: IO [(Text, Text)]) >>= (@?= [("VARCHAR[]", "VARCHAR[]")])+ -- A UTC element keeps its nanoseconds for a nanosecond target.+ let precise = posixSecondsToUTCTime 0.123456789+ (query conn "SELECT ?::TIMESTAMPTZ_NS[]" (Only [Just precise, Nothing]) :: IO [Only [Maybe UTCTime]])+ >>= (@?= [Only [Just precise, Nothing]])+ , testCase "a list of Integer values takes the element type of its target" $ withConnection ":memory:" \conn -> do+ let insert :: Text -> [Integer] -> IO (Either SQLError Int)+ insert dtype values = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE integer_list (v " <> dtype <> ")"))+ try (execute conn "INSERT INTO integer_list VALUES (?)" (Only values))+ stored = query_ conn "SELECT v::VARCHAR FROM integer_list" :: IO [Only Text]+ insert "UBIGINT[]" [5, 6] >>= either (assertFailure . show) (const (pure ()))+ stored >>= (@?= [Only "[5, 6]"])+ insert "UBIGINT[]" [5, -6] >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorInputInvalid+ Right _ -> assertFailure "expected the DuckDB range error for -6"+ insert "UBIGINT[]" [5, 2 ^ (130 :: Int)] >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorTypeConversion+ Right _ -> assertFailure "expected the duckdb-simple range error for 2^130"+ stored >>= (@?= [])+ -- With no integer target, a list takes the type that its elements give,+ -- as a SQL list literal of the HUGEINT and BIGNUM values. A list with+ -- no element, or with only NULL elements, gets the type NULL. See+ -- Note [Integer parameters] in "Database.DuckDB.Simple.ToField".+ let typeOf :: (ToField a) => a -> IO [Only Text]+ typeOf value = query conn "SELECT typeof(?)" (Only value)+ typeOf ([5, 6] :: [Integer]) >>= (@?= [Only "HUGEINT[]"])+ typeOf ([5, 2 ^ (130 :: Int)] :: [Integer]) >>= (@?= [Only "BIGNUM[]"])+ typeOf ([] :: [Integer]) >>= (@?= [Only "\"NULL\"[]"])+ typeOf ([Nothing] :: [Maybe Integer]) >>= (@?= [Only "\"NULL\"[]"])+ typeOf ([5, 2 ^ (64 :: Int)] :: [Natural]) >>= (@?= [Only "HUGEINT[]"])+ typeOf (Map.fromList [(5, "a")] :: Map.Map Integer Text) >>= (@?= [Only "MAP(HUGEINT, VARCHAR)"])+ typeOf ([[1], []] :: [[Integer]]) >>= (@?= [Only "HUGEINT[][]"])+ typeOf ([[]] :: [[Integer]]) >>= (@?= [Only "\"NULL\"[][]"])+ -- BigNum has the same column type name. Its elements are BIGNUM values.+ typeOf [BigNum 5, BigNum 6] >>= (@?= [Only "BIGNUM[]"])+ typeOf ([Nothing] :: [Maybe BigNum]) >>= (@?= [Only "\"NULL\"[]"])+ -- An empty list for an integer target takes the element type of the target.+ _ <- execute_ conn "CREATE OR REPLACE TABLE empty_integers (v INTEGER[])"+ _ <- execute conn "INSERT INTO empty_integers VALUES (?)" (Only ([] :: [Integer]))+ (query_ conn "SELECT typeof(v), v::VARCHAR FROM empty_integers" :: IO [(Text, Text)]) >>= (@?= [("INTEGER[]", "[]")])+ -- In a VARIANT, the elements keep the HUGEINT values, so a cast to an+ -- unsigned type checks their range. A BIGNUM element gave+ -- 18446744073709551611 for -5 with no error.+ let variantCast :: (ToField a) => String -> a -> Text -> Assertion+ variantCast label value target = do+ _ <- execute_ conn "CREATE OR REPLACE TABLE variant_integers (v VARIANT)"+ _ <- execute conn "INSERT INTO variant_integers VALUES (?)" (Only value)+ try (query_ conn (Query ("SELECT (v::" <> target <> ")::VARCHAR FROM variant_integers")) :: IO [Only Text]) >>= \case+ Left SQLError{sqlErrorType} -> assertEqual label (Just ErrorTypeConversion) sqlErrorType+ Right rows -> assertFailure (label <> ": expected the error of the cast, got " <> show rows)+ variantCast "[-5]" ([-5] :: [Integer]) "UBIGINT[]"+ variantCast "Map {a: -5}" (Map.fromList [("a", -5)] :: Map.Map Text Integer) "MAP(VARCHAR, UBIGINT)"+ variantCast "[[-5], []]" ([[-5], []] :: [[Integer]]) "UBIGINT[][]"+ , testCase "a list and a SizedArray with the same elements store the same LIST value" $ withConnection ":memory:" \conn -> do+ let ints = [1, 2, 3] :: [Int64]+ arr = sized ints :: SizedArray 3 Int64+ (query conn "SELECT typeof(?), typeof(?)" (ints, arr) :: IO [(Text, Text)]) >>= (@?= [("BIGINT[]", "BIGINT[3]")])+ _ <- execute_ conn "CREATE TABLE same_value (i BIGINT, v BIGINT[])"+ _ <- execute conn "INSERT INTO same_value VALUES (0, ?), (1, ?)" (ints, arr)+ (query_ conn "SELECT v FROM same_value ORDER BY i" :: IO [Only [Int64]]) >>= (@?= [Only ints, Only ints])+ (query_ conn "SELECT count(DISTINCT v) FROM same_value" :: IO [Only Int64]) >>= (@?= [Only 1])+ , testCase "an empty list of composite elements binds as the SQL literal []" $ withConnection ":memory:" \conn -> do+ let literal :: (ToField a) => String -> a -> Assertion+ literal label value =+ (query conn "SELECT typeof(?), ?::VARCHAR, typeof([])" (value, value) :: IO [(Text, Text, Text)])+ >>= assertEqual label [("\"NULL\"[]", "[]", "\"NULL\"[]")]+ literal "STRUCT" ([] :: [StructValue FieldValue])+ literal "BIGINT[]" ([] :: [[Int64]])+ -- A StructValue has no declaration, so a list of NULL elements and an+ -- empty MAP of them get the type of the SQL literal. See+ -- Note [Element types of a list] in "Database.DuckDB.Simple.ToField".+ (query conn "SELECT typeof(?)" (Only [Nothing :: Maybe (StructValue FieldValue)]) :: IO [Only Text]) >>= (@?= [Only "\"NULL\"[]"])+ (query conn "SELECT typeof(?)" (Only (Map.empty :: Map.Map Text (StructValue FieldValue))) :: IO [Only Text])+ >>= (@?= [Only "MAP(VARCHAR, \"NULL\")"])+ -- A SizedArray of NULL STRUCT values gets the type of array_value(NULL).+ let nullArray = sized [Nothing] :: SizedArray 1 (Maybe (StructValue FieldValue))+ (query conn "SELECT typeof(?), ?::VARCHAR" (nullArray, nullArray) :: IO [(Text, Text)]) >>= (@?= [("\"NULL\"[1]", "[NULL]")])+ , testCase "a list of NULL composite elements takes the element type of a typed column" $ withConnection ":memory:" \conn -> do+ let nulls = [Nothing, Nothing] :: [Maybe (StructValue FieldValue)]+ _ <- execute_ conn "CREATE TABLE null_list (v STRUCT(a INTEGER)[]); CREATE TABLE null_array (v STRUCT(a INTEGER)[2])"+ _ <- execute conn "INSERT INTO null_list VALUES (?)" (Only nulls)+ _ <- execute conn "INSERT INTO null_array VALUES (?)" (Only (sized nulls :: SizedArray 2 (Maybe (StructValue FieldValue))))+ (query_ conn "SELECT typeof(v), v::VARCHAR FROM null_list UNION ALL SELECT typeof(v), v::VARCHAR FROM null_array" :: IO [(Text, Text)])+ >>= (@?= [("STRUCT(a INTEGER)[]", "[NULL, NULL]"), ("STRUCT(a INTEGER)[2]", "[NULL, NULL]")])+ (query conn "SELECT typeof(?::STRUCT(a INTEGER)[])" (Only nulls) :: IO [Only Text]) >>= (@?= [Only "STRUCT(a INTEGER)[]"])+ -- An empty list takes the element type of a typed column too.+ _ <- execute_ conn "DELETE FROM null_list"+ _ <- execute conn "INSERT INTO null_list VALUES (?)" (Only ([] :: [StructValue FieldValue]))+ (query_ conn "SELECT v::VARCHAR FROM null_list" :: IO [Only Text]) >>= (@?= [Only "[]"])+ -- With no known type, a list of NULL elements gets the type of the SQL literal [NULL, NULL].+ (query conn "SELECT typeof(?)" (Only nulls) :: IO [Only Text]) >>= (@?= [Only "\"NULL\"[]"])+ , testCase "an IntervalValue binds as an INTERVAL and keeps its three fields" $ withConnection ":memory:" \conn -> do+ let value = IntervalValue 1 2 3+ negative = IntervalValue (-1) (-2) (-3)+ month = IntervalValue 1 0 0+ days30 = IntervalValue 0 30 0+ forM_ [value, negative] \interval ->+ (query conn "SELECT typeof(?), ?" (interval, interval) :: IO [(Text, IntervalValue)]) >>= (@?= [("INTERVAL", interval)])+ _ <- execute_ conn "CREATE TABLE intervals (i BIGINT, v INTERVAL)"+ _ <- executeMany conn "INSERT INTO intervals VALUES (?, ?)" [(0 :: Int64, days30), (1, month), (2, negative)]+ (query_ conn "SELECT v FROM intervals ORDER BY i" :: IO [Only IntervalValue]) >>= (@?= map Only [days30, month, negative])+ -- DuckDB compares intervals with 30 days for each month.+ (query conn "SELECT count(*) FROM intervals WHERE v = ?" (Only days30) :: IO [Only Int64]) >>= (@?= [Only 2])+ (query conn "SELECT count(*) FROM intervals WHERE v < ?" (Only (IntervalValue 0 31 0)) :: IO [Only Int64]) >>= (@?= [Only 3])+ (query conn "SELECT typeof(TIMESTAMP '2026-01-31' + ?), TIMESTAMP '2026-01-31' + ?" (month, month) :: IO [(Text, LocalTime)])+ >>= (@?= [("TIMESTAMP", LocalTime (fromGregorian 2026 2 28) (TimeOfDay 0 0 0))])+ (query conn "SELECT ?" (Only (Nothing :: Maybe IntervalValue)) :: IO [Only (Maybe IntervalValue)]) >>= (@?= [Only Nothing])+ (query conn "SELECT typeof(?), ?" ([value, negative], [value, negative]) :: IO [(Text, [IntervalValue])])+ >>= (@?= [("INTERVAL[]", [value, negative])])+ let array = sized [value, negative] :: SizedArray 2 IntervalValue+ (query conn "SELECT typeof(?), ?" (array, array) :: IO [(Text, [IntervalValue])]) >>= (@?= [("INTERVAL[2]", [value, negative])])+ , testCase "MAP keys must remain distinct in field and generic decoding" $ withConnection ":memory:" \conn -> do+ let genericKeyMap entry = fieldType entry >>= \logical -> duckFromTypedField logical (fieldValue entry)+ check parser = do+ let get sql = queryWith_ (fieldWith parser) conn sql :: IO [Map.Map Bool Int]+ get "SELECT MAP {0:10, 1:20}" >>= (@?= [Map.fromList [(False, 10), (True, 20)]])+ try (get "SELECT MAP {1:10, 2:20}") >>= \case+ Left SQLError{sqlErrorMessage} -> assertBool (Text.unpack sqlErrorMessage) ("duplicate MAP key after conversion" `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure ("expected duplicate MAP key error, got " <> show rows)+ check fromField+ check genericKeyMap+ , testCase "a JSON alias on an INTEGER reads as a VARIANT integer" $ withConnection ":memory:" \conn -> do+ let value = singleField (LogicalTypeAlias "JSON" (LogicalTypeScalar LogicalTypeIdInteger)) (FieldInt32 17)+ (query conn "SELECT (?).value, (?).value::VARIANT" (value, value) :: IO [(Variant, Variant)])+ >>= (@?= [(Variant (FieldInt32 17), Variant (FieldInt32 17))])+ , testCase "a TimeWithZone from a Data.Time zone reads back with that zone" $ withConnection ":memory:" \conn -> do+ let zone = hoursToTimeZone 2+ value = timeWithZone (TimeOfDay 12 34 0) zone+ timeWithZoneOffsetSeconds value @?= 7200+ [Only stored] <- query conn "SELECT ?" (Only value)+ stored @?= value+ timeWithZoneZone stored @?= zone+ -- A TimeZone holds whole minutes, so the seconds of an offset go.+ [Only withSeconds] <- query_ conn "SELECT '12:00:00+01:00:30'::TIMETZ"+ timeWithZoneOffsetSeconds withSeconds @?= 3630+ timeWithZoneZone withSeconds @?= hoursToTimeZone 1+ , testCase "a TimeWithZone binds as a TIMETZ with its time and offset in seconds" $ withConnection ":memory:" \conn -> do+ let at hours minutes offset = TimeWithZone (TimeOfDay hours minutes 0) offset+ value = at 12 34 19800+ -- The offset +01:00:30.+ seconds = at 12 0 3630+ roundTrip time = (query conn "SELECT ?" (Only time) :: IO [Only TimeWithZone]) >>= (@?= [Only time])+ (query conn "SELECT typeof(?), ?" (value, value) :: IO [(Text, TimeWithZone)]) >>= (@?= [("TIME WITH TIME ZONE", value)])+ -- An offset with seconds reads, binds, and prints exactly.+ ( query_ conn "SELECT '12:00:00+01:00:30'::TIMETZ, '12:00:00-15:59:59'::TIMETZ, '24:00:00+15:59:59'::TIMETZ"+ :: IO [(TimeWithZone, TimeWithZone, TimeWithZone)]+ )+ >>= (@?= [(seconds, at 12 0 (-57599), TimeWithZone (TimeOfDay 24 0 0) 57599)])+ (query conn "SELECT ?::VARCHAR" (Only seconds) :: IO [Only Text]) >>= (@?= [Only "12:00:00+01:00:30"])+ mapM_+ roundTrip+ [seconds, at 12 0 57599, at 12 0 (-57599), TimeWithZone (TimeOfDay 24 0 0) 0, TimeWithZone (TimeOfDay 23 59 59.999999) (-3630)]+ -- DuckDB stores an offset from -15:59:59 to +15:59:59.+ forM_ [57600, -57600] \offset ->+ try (query conn "SELECT ?" (Only (at 12 0 offset)) :: IO [Only TimeWithZone]) >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ sqlErrorType @?= Just ErrorTypeConversion+ assertBool+ (Text.unpack sqlErrorMessage)+ (("duckdb-simple: the TIMETZ offset " <> Text.pack (show offset) <> " seconds is out of range") `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure ("expected the offset error, got " <> show rows)+ -- A TIME column removes the offset and keeps the local time.+ _ <- execute_ conn "CREATE TABLE times (a TIMETZ, b TIME)"+ _ <- execute conn "INSERT INTO times VALUES (?, ?)" (seconds, seconds)+ (query_ conn "SELECT a, b FROM times" :: IO [(TimeWithZone, TimeOfDay)]) >>= (@?= [(seconds, TimeOfDay 12 0 0)])+ -- DuckDB compares the instant first, and then the offset.+ (query conn "SELECT count(*) FROM times WHERE a = ?" (Only seconds) :: IO [Only Int64]) >>= (@?= [Only 1])+ (query conn "SELECT count(*) FROM times WHERE a = ?" (Only (at 12 0 3631)) :: IO [Only Int64]) >>= (@?= [Only 0])+ (query conn "SELECT count(*) FROM times WHERE ? < a" (Only (at 12 0 3631)) :: IO [Only Int64]) >>= (@?= [Only 1])+ (query conn "SELECT count(*) FROM times WHERE a = ?" (Only (TimeWithZone (TimeOfDay 10 59 30) 0)) :: IO [Only Int64]) >>= (@?= [Only 0])+ -- The digits below one microsecond go away, as in SQL text.+ (query conn "SELECT ?" (Only seconds{timeWithZoneTime = TimeOfDay 1 2 3.0000005}) :: IO [Only TimeWithZone])+ >>= (@?= [Only seconds{timeWithZoneTime = TimeOfDay 1 2 3}])+ (query conn "SELECT ?" (Only (Nothing :: Maybe TimeWithZone)) :: IO [Only (Maybe TimeWithZone)]) >>= (@?= [Only Nothing])+ let times = [value, seconds, at 0 0 (-57599)]+ (query conn "SELECT typeof(?), ?" (times, times) :: IO [(Text, [TimeWithZone])]) >>= (@?= [("TIME WITH TIME ZONE[]", times)])+ let array = sized times :: SizedArray 3 TimeWithZone+ (query conn "SELECT typeof(?), ?" (array, array) :: IO [(Text, [TimeWithZone])]) >>= (@?= [("TIME WITH TIME ZONE[3]", times)])+ let nested = singleField (LogicalTypeScalar LogicalTypeIdTimeTz) (FieldTimeTZ seconds)+ (query conn "SELECT ?, (?).value::VARCHAR" (nested, nested) :: IO [(StructValue FieldValue, Text)])+ >>= (@?= [(nested, "12:00:00+01:00:30")])+ , testCase "a DecimalValue binds as a DECIMAL with its own width and scale" $ withConnection ":memory:" \conn -> do+ let value = DecimalValue 5 3 2125+ roundTrip decimal = (query conn "SELECT ?" (Only decimal) :: IO [Only DecimalValue]) >>= (@?= [Only decimal])+ stored column decimal = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE decimals (v " <> column <> ")"))+ _ <- execute conn "INSERT INTO decimals VALUES (?)" (Only decimal)+ query_ conn "SELECT v FROM decimals" :: IO [Only DecimalValue]+ (query conn "SELECT typeof(?), ?" (value, value) :: IO [(Text, DecimalValue)]) >>= (@?= [("DECIMAL(5,3)", value)])+ -- A column of another width and scale gives its own type. DuckDB rounds+ -- half away from zero.+ stored "DECIMAL(4,2)" value >>= (@?= [Only (DecimalValue 4 2 213)])+ stored "DECIMAL(4,2)" (DecimalValue 5 3 2135) >>= (@?= [Only (DecimalValue 4 2 214)])+ stored "DECIMAL(4,2)" (DecimalValue 5 3 (-2125)) >>= (@?= [Only (DecimalValue 4 2 (-213))])+ stored "DECIMAL(10,4)" value >>= (@?= [Only (DecimalValue 10 4 21250)])+ try (stored "DECIMAL(4,2)" (DecimalValue 8 3 12345678)) >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertBool+ (Text.unpack sqlErrorMessage)+ ("Casting value \"12345.678\" to type DECIMAL(4,2) failed: value is out of range!" `Text.isInfixOf` sqlErrorMessage)+ sqlErrorType @?= Just ErrorTypeConversion+ Right rows -> assertFailure ("expected the DECIMAL range error, got " <> show rows)+ stored "DECIMAL(3,2)" (DecimalValue 3 2 210) >>= (@?= [Only (DecimalValue 3 2 210)])+ (query conn "SELECT count(*) FROM decimals WHERE v = ?" (Only (DecimalValue 2 1 21)) :: IO [Only Int64]) >>= (@?= [Only 1])+ (query conn "SELECT count(*) FROM decimals WHERE v < ?" (Only (DecimalValue 5 2 211)) :: IO [Only Int64]) >>= (@?= [Only 1])+ (query conn "SELECT typeof(? * 2), ? * 2" (value, value) :: IO [(Text, DecimalValue)]) >>= (@?= [("DECIMAL(15,3)", DecimalValue 15 3 4250)])+ mapM_+ roundTrip+ [ DecimalValue 38 0 (10 ^ (38 :: Int) - 1)+ , DecimalValue 38 0 (negate (10 ^ (38 :: Int) - 1))+ , DecimalValue 38 38 1+ , DecimalValue 2 1 (-15)+ , DecimalValue 1 0 7+ ]+ -- The encoder raises an error for a value that it cannot encode.+ assertRangeError (query conn "SELECT ?" (Only (DecimalValue 2 1 100)) :: IO [Only DecimalValue])+ forM_ [DecimalValue 39 0 1, DecimalValue 2 3 1, DecimalValue 0 0 0] \decimal ->+ assertInputError (query conn "SELECT ?" (Only decimal) :: IO [Only DecimalValue])+ (query conn "SELECT ?" (Only (Nothing :: Maybe DecimalValue)) :: IO [Only (Maybe DecimalValue)]) >>= (@?= [Only Nothing])+ , testCase "a Map binds as a MAP parameter in each SQL form" $ withConnection ":memory:" \conn -> do+ let pairs = Map.fromList [("a", 1), ("b", 2)] :: Map.Map Text Int64+ none = Map.empty :: Map.Map Text Int64+ rejected :: (ToField a) => Text -> a -> Assertion+ rejected message value =+ try (query conn "SELECT ?::VARCHAR" (Only value) :: IO [Only Text]) >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorMessage @?= message+ sqlErrorType @?= Just ErrorInputInvalid+ Right rows -> assertFailure ("expected " <> Text.unpack message <> ", got " <> show rows)+ forM_ [pairs, none] \value ->+ (query conn "SELECT typeof(?), ?" (value, value) :: IO [(Text, Map.Map Text Int64)]) >>= (@?= [("MAP(VARCHAR, BIGINT)", value)])+ -- DuckDB casts the keys and the values to the types of the column.+ forM_ ["MAP(VARCHAR, BIGINT)", "MAP(VARCHAR, INTEGER)"] \dtype -> do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE map_value (i BIGINT, m " <> dtype <> ")"))+ _ <- executeMany conn "INSERT INTO map_value VALUES (?, ?)" [(0 :: Int64, pairs), (1, none)]+ (query_ conn "SELECT m FROM map_value ORDER BY i" :: IO [Only (Map.Map Text Int64)])+ >>= assertEqual (Text.unpack dtype) [Only pairs, Only none]+ _ <- execute_ conn "CREATE TABLE integer_keys (m MAP(INTEGER, BIGINT))"+ _ <- execute conn "INSERT INTO integer_keys VALUES (?)" (Only (Map.fromList [("1", 1)] :: Map.Map Text Int64))+ (query_ conn "SELECT m FROM integer_keys" :: IO [Only (Map.Map Int32 Int64)]) >>= (@?= [Only (Map.fromList [(1, 1)])])+ try (execute conn "INSERT INTO integer_keys VALUES (?)" (Only pairs)) >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertBool (Text.unpack sqlErrorMessage) ("Conversion Error: Could not convert string 'a' to INT32" `Text.isPrefixOf` sqlErrorMessage)+ sqlErrorType @?= Just ErrorTypeConversion+ Right (_ :: Int) -> assertFailure "expected the DuckDB key conversion error"+ -- DuckDB compares two MAP values pair by pair in their order. A map binds+ -- its pairs in ascending key order.+ _ <- execute_ conn "CREATE TABLE ordered_maps (m MAP(VARCHAR, BIGINT))"+ _ <- execute_ conn "INSERT INTO ordered_maps VALUES (MAP {'b': 2, 'a': 1}), (MAP {'a': 1, 'b': 2})"+ (query conn "SELECT count(*) FROM ordered_maps WHERE m = ?" (Only pairs) :: IO [Only Int64]) >>= (@?= [Only 1])+ (query_ conn "SELECT m FROM ordered_maps" :: IO [Only (Map.Map Text Int64)]) >>= (@?= [Only pairs, Only pairs])+ ( query conn "SELECT map_extract(?, 'a'), (?)['a'], map_keys(?), element_at(?, 'b')" (pairs, pairs, pairs, pairs)+ :: IO [([Int64], Int64, [Text], [Int64])]+ )+ >>= (@?= [([1], 1, ["a", "b"], [2])])+ -- A Haskell map can hold two NaN keys. DuckDB rejects them and a NULL key.+ let nan = 0 / 0 :: Double+ rejected "Invalid Input Error: Map keys must be unique." (Map.insert nan 2 (Map.fromList [(nan, 1)]) :: Map.Map Double Int64)+ rejected "Invalid Input Error: Map keys can not be NULL." (Map.fromList [(Nothing, 1), (Just "a", 2)] :: Map.Map (Maybe Text) Int64)+ , testCase "a Map binds NULL, nested, Integer, and UTC values" $ withConnection ":memory:" \conn -> do+ let withNull = Map.fromList [("a", Just 1), ("b", Nothing)] :: Map.Map Text (Maybe Int64)+ nested = Map.fromList [("a", [1, 2]), ("b", [])] :: Map.Map Text [Int64]+ pairs = Map.fromList [("a", 1)] :: Map.Map Text Int64+ (query conn "SELECT typeof(?), ?" (withNull, withNull) :: IO [(Text, Map.Map Text (Maybe Int64))])+ >>= (@?= [("MAP(VARCHAR, BIGINT)", withNull)])+ (query conn "SELECT typeof(?), ?" (nested, nested) :: IO [(Text, Map.Map Text [Int64])]) >>= (@?= [("MAP(VARCHAR, BIGINT[])", nested)])+ (query conn "SELECT ?" (Only (Nothing :: Maybe (Map.Map Text Int64))) :: IO [Only (Maybe (Map.Map Text Int64))]) >>= (@?= [Only Nothing])+ (query conn "SELECT typeof(?), ?" ([pairs, Map.empty], [pairs, Map.empty]) :: IO [(Text, [Map.Map Text Int64])])+ >>= (@?= [("MAP(VARCHAR, BIGINT)[]", [pairs, Map.empty])])+ -- An Integer value takes the value type of its target.+ let insert :: Map.Map Text Integer -> IO (Either SQLError Int)+ insert value = do+ _ <- execute_ conn "CREATE OR REPLACE TABLE unsigned_map (m MAP(VARCHAR, UBIGINT))"+ try (execute conn "INSERT INTO unsigned_map VALUES (?)" (Only value))+ insert (Map.fromList [("a", 5), ("b", 6)]) >>= either (assertFailure . show) (const (pure ()))+ (query_ conn "SELECT m::VARCHAR FROM unsigned_map" :: IO [Only Text]) >>= (@?= [Only "{a=5, b=6}"])+ insert (Map.fromList [("a", -5)]) >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorInputInvalid+ Right _ -> assertFailure "expected the DuckDB range error for -5"+ insert (Map.fromList [("a", 2 ^ (130 :: Int))]) >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorTypeConversion+ Right _ -> assertFailure "expected the duckdb-simple range error for 2^130"+ (query conn "SELECT typeof(?)" (Only (Map.fromList [("a", 5)] :: Map.Map Text Integer)) :: IO [Only Text])+ >>= (@?= [Only "MAP(VARCHAR, HUGEINT)"])+ -- A UTC value keeps its nanoseconds for a nanosecond target.+ let times = Map.fromList [("t", posixSecondsToUTCTime 0.123456789)] :: Map.Map Text UTCTime+ (query conn "SELECT ?::MAP(VARCHAR, TIMESTAMPTZ_NS)" (Only times) :: IO [Only (Map.Map Text UTCTime)]) >>= (@?= [Only times])+ -- An empty map of nested values has the value type NULL, as in MAP {}.+ (query conn "SELECT typeof(?), ?::VARCHAR" (Map.empty :: Map.Map Text [Int64], Map.empty :: Map.Map Text [Int64]) :: IO [(Text, Text)])+ >>= (@?= [("MAP(VARCHAR, \"NULL\")", "{}")])+ , testCase "an array of BigNum values keeps the BIGNUM casts of DuckDB" $ withConnection ":memory:" \conn -> do+ -- DuckDB 2.0.0 casts the BIGNUM -6 to UBIGINT as a wrapped value with no+ -- error. See Note [Integer parameters] in "Database.DuckDB.Simple.ToField".+ forM_+ [ ("BIGINT[2]", [BigNum 5, BigNum 6], "[5, 6]")+ , ("BIGNUM[2]", [BigNum 5, BigNum 6], "[5, 6]")+ , ("UBIGINT[2]", [BigNum 5, BigNum (-6)], "[5, 18446744073709551610]")+ ]+ \(dtype, values, expected) -> do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE bignum_array (v " <> dtype <> ")"))+ _ <- execute conn "INSERT INTO bignum_array VALUES (?)" (Only (sized values :: SizedArray 2 BigNum))+ (query_ conn "SELECT v::VARCHAR FROM bignum_array" :: IO [Only Text]) >>= assertEqual (Text.unpack dtype) [Only expected]+ , testCase "native timestamp extrema read back in all units, and bind back only in the range of DuckDB" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ let checkTemporalFailure :: forall a. (FromField a) => Query -> Proxy a -> Text -> IO ()+ checkTemporalFailure sql _ target = do+ assertConversionError (query_ conn sql :: IO [Only a])+ results <- queryWith_ (fieldWith (liftConversion . runConversion . (fromField :: Field -> Conversion a))) conn sql+ case results of+ [Errors [failure]] -> case fromException failure of+ Just ConversionFailed{errHaskellType} -> errHaskellType @?= target+ _ -> assertFailure ("expected ConversionFailed, got " <> show failure)+ _ -> assertFailure ("expected one conversion failure for " <> Text.unpack target)+ let epoch = fromGregorian 1970 1 1+ lowerDay = toInteger (minBound :: Int64) `div` 86400000000+ upperDay = toInteger (maxBound :: Int64) `div` 86400000000+ dates =+ [fromGregorian 2000 1 2, fromGregorian (-500000) 1 1, fromGregorian 500000 1 1]+ <> map (`addDays` epoch) [-2147483646, 2147483646, lowerDay, lowerDay + 1, upperDay, upperDay + 1]+ -- DATE itself has a wider range than the intermediate of a timestamp cast.+ _ <- execute_ conn "CREATE TABLE native_date(value DATE)"+ forM_ dates \day -> do+ _ <- execute conn "INSERT INTO native_date VALUES (?)" (Only day)+ (query_ conn "SELECT value FROM native_date" :: IO [Only Day]) >>= (@?= [Only day])+ (query_ conn "SELECT value FROM native_date" :: IO [Only (Unbounded Day)]) >>= (@?= [Only (Finite day)])+ let expected = LocalTime day (TimeOfDay 0 0 0)+ micros = diffDays day epoch * 86400000000+ if micros >= toInteger (minBound :: Int64) && micros <= toInteger (maxBound :: Int64)+ then do+ (query_ conn "SELECT value::TIMESTAMP FROM native_date" :: IO [Only LocalTime]) >>= (@?= [Only expected])+ (query_ conn "SELECT value FROM native_date" :: IO [Only LocalTime]) >>= (@?= [Only expected])+ (query_ conn "SELECT value FROM native_date" :: IO [Only (Unbounded LocalTime)]) >>= (@?= [Only (Finite expected)])+ else do+ checkTemporalFailure "SELECT value FROM native_date" (Proxy @LocalTime) "LocalTime"+ checkTemporalFailure "SELECT value FROM native_date" (Proxy @(Unbounded LocalTime)) "Unbounded LocalTime"+ forM_ ["TIMESTAMP", "TIMESTAMP_S", "TIMESTAMP_MS", "TIMESTAMP_NS"] \dtype ->+ try (query_ conn (fromString ("SELECT value::" <> dtype <> " FROM native_date")) :: IO [Only LocalTime]) >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorTypeConversion+ Right rows -> assertFailure ("expected the native timestamp cast to fail, got " <> show rows)+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ _ <- execute_ conn "DELETE FROM native_date"+ pure ()+ _ <- execute_ conn "DROP TABLE native_date"+ forM_ [("-infinity", NegInfinity), ("infinity", PosInfinity)] \(literal, expected) -> do+ let sql = fromString ("SELECT '" <> literal <> "'::DATE")+ cast = fromString ("SELECT '" <> literal <> "'::DATE::TIMESTAMP")+ assertConversionError (query_ conn sql :: IO [Only LocalTime])+ (query_ conn sql :: IO [Only (Unbounded LocalTime)]) >>= (@?= [Only expected])+ (query_ conn cast :: IO [Only (Unbounded LocalTime)]) >>= (@?= [Only expected])+ forM_+ [ (Seconds, 1, "TIMESTAMP_S")+ , (Milliseconds, 1000, "TIMESTAMP_MS")+ , (Microseconds, 1000000, "TIMESTAMP")+ , (Nanoseconds, 1000000000, "TIMESTAMP_NS")+ ]+ \(constructor, units, dtype) -> do+ let extraValues+ | units <= 1000 =+ let limit = toInteger (maxBound :: Int64) `div` (1000000 `div` units)+ in map+ fromInteger+ ( [-limit - 1, -limit, limit, limit + 1]+ <> [days * 86400 * units | days <- [-2147483648, -2147483647, -2147483646, 2147483646, 2147483647, 2147483648]]+ )+ | units == 1000000000 =+ let day = 86400000000000+ in [-1, -499, -500, -501, day - 1, day - 499, day - 500, day - 501]+ | otherwise = []+ forM_ ([minBound, negate (maxBound :: Int64) + 1, -1, 0, maxBound - 1] <> extraValues) \value -> do+ let expected = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (toInteger value % units)))+ micros = toInteger value * 1000000 `div` units+ appendTimestamp conn dtype (constructor value)+ (query_ conn "SELECT value FROM native_timestamp" :: IO [Only LocalTime]) >>= (@?= [Only expected])+ (query_ conn "SELECT value FROM native_timestamp" :: IO [Only (Unbounded LocalTime)]) >>= (@?= [Only (Finite expected)])+ -- S/MS casts check a microsecond intermediate. NS casts round it+ -- to the nearest microsecond before they extract the date.+ if micros >= toInteger (minBound :: Int64) && micros <= toInteger (maxBound :: Int64)+ then do+ [Only day] <- query_ conn "SELECT value::DATE FROM native_timestamp" :: IO [Only Day]+ (query_ conn "SELECT value FROM native_timestamp" :: IO [Only Day]) >>= (@?= [Only day])+ (query_ conn "SELECT value FROM native_timestamp" :: IO [Only (Unbounded Day)]) >>= (@?= [Only (Finite day)])+ let timeType = if units == 1000000000 then "TIME_NS" else "TIME"+ [Only clock] <- query_ conn (fromString ("SELECT value::" <> timeType <> " FROM native_timestamp")) :: IO [Only TimeOfDay]+ (query_ conn "SELECT value FROM native_timestamp" :: IO [Only TimeOfDay]) >>= (@?= [Only clock])+ else do+ checkTemporalFailure "SELECT value FROM native_timestamp" (Proxy @Day) "Day"+ checkTemporalFailure "SELECT value FROM native_timestamp" (Proxy @(Unbounded Day)) "Unbounded Day"+ checkTemporalFailure "SELECT value FROM native_timestamp" (Proxy @TimeOfDay) "TimeOfDay"+ try (query_ conn "SELECT value::DATE FROM native_timestamp" :: IO [Only Day]) >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorTypeConversion+ Right rows -> assertFailure ("expected the native DATE cast to fail, got " <> show rows)+ try (query_ conn "SELECT value::TIME FROM native_timestamp" :: IO [Only TimeOfDay]) >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorTypeConversion+ Right rows -> assertFailure ("expected the native TIME cast to fail, got " <> show rows)+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ [Only original] <- query_ conn "SELECT {'value': value} FROM native_timestamp" :: IO [Only (StructValue FieldValue)]+ -- DuckDB needs an explicit parameter type for extreme S/MS values.+ let rebind = query conn (fromString ("SELECT ?::STRUCT(value " <> dtype <> ")")) (Only original)+ -- A second, millisecond, or microsecond value outside the range of+ -- DuckDB raises an error. See Note [Time ranges] in+ -- "Database.DuckDB.Simple.ToField".+ if units == 1000000000 || (micros >= -9223372022400000000 && micros <= 9223372036854775806)+ then rebind >>= (@?= [Only original])+ else assertOutsideTimeRange rebind+ _ <- execute_ conn "DROP TABLE native_timestamp"+ pure ()+ forM_ [("-infinity", NegInfinity), ("infinity", PosInfinity)] \(literal, expected) -> do+ let sql = fromString ("SELECT '" <> literal <> "'::" <> dtype)+ cast = fromString ("SELECT '" <> literal <> "'::" <> dtype <> "::DATE")+ assertConversionError (query_ conn sql :: IO [Only Day])+ (query_ conn sql :: IO [Only (Unbounded Day)]) >>= (@?= [Only expected])+ (query_ conn cast :: IO [Only (Unbounded Day)]) >>= (@?= [Only expected])+ , testCase "composite timestamps reject infinity and storage overflow" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ -- Only a nanosecond value has the range of its storage. See+ -- Note [Time ranges] in "Database.DuckDB.Simple.ToField".+ forM_+ [ (LogicalTypeIdTimestampSec, 1, assertOutsideTimeRange)+ , (LogicalTypeIdTimestampMs, 1000, assertOutsideTimeRange)+ , (LogicalTypeIdTimestamp, 1000000, assertOutsideTimeRange)+ , (LogicalTypeIdTimestampNs, 1000000000, assertRangeError)+ ]+ \(dtype, units, rejected) ->+ forM_+ [ toInteger (minBound :: Int64) - 1+ , negate (toInteger (maxBound :: Int64))+ , toInteger (maxBound :: Int64)+ , toInteger (maxBound :: Int64) + 1+ ]+ \value -> do+ let timestamp = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (value % units)))+ struct = singleField (LogicalTypeScalar dtype) (FieldTimestamp (Finite timestamp))+ rejected (query conn "SELECT ?" (Only struct) :: IO [Only FieldValue])+ , testCase "composite timestamps floor fractional units before the epoch" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ forM_+ [ (LogicalTypeIdTimestampSec, 1)+ , (LogicalTypeIdTimestampMs, 1000)+ , (LogicalTypeIdTimestamp, 1000000)+ , (LogicalTypeIdTimestampNs, 1000000000)+ ]+ \(dtype, units) -> do+ let timestamp = utcToLocalTime utc (posixSecondsToUTCTime (fromRational ((-1) % (2 * units))))+ expected = utcToLocalTime utc (posixSecondsToUTCTime (fromRational ((-1) % units)))+ struct = singleField (LogicalTypeScalar dtype) (FieldTimestamp (Finite timestamp))+ (query conn "SELECT (?).value" (Only struct) :: IO [Only LocalTime]) >>= (@?= [Only expected])+ , testCase "composite TIME_NS rejects invalid clock components" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ -- A field outside its own range and a time with valid fields after+ -- 24:00:00 are both outside the range of TIME_NS.+ forM_+ [ TimeOfDay (-1) 0 0+ , TimeOfDay 0 60 0+ , TimeOfDay 0 0 (-1)+ , TimeOfDay 25 0 0+ , TimeOfDay 24 0 0.000000001+ , TimeOfDay 23 59 60.000000001+ ]+ \value ->+ assertRangeError+ (query conn "SELECT ?" (Only (singleField (LogicalTypeScalar LogicalTypeIdTimeNs) (FieldTime value))) :: IO [Only FieldValue])+ , testCase "microsecond timestamps round trip at the ends of the range of DuckDB" $ withConnection ":memory:" \conn -> do+ let local (micros :: Int64) = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (toInteger micros % 1000000)))+ forM_ [-9223372022400000000, -1, 0, maxBound - 1] \value -> do+ _ <- execute_ conn "CREATE TABLE microsecond_timestamp(value TIMESTAMP)"+ _ <- execute conn "INSERT INTO microsecond_timestamp VALUES (?)" (Only (local value))+ (query_ conn "SELECT value FROM microsecond_timestamp" :: IO [Only LocalTime]) >>= (@?= [Only (local value)])+ _ <- execute_ conn "DROP TABLE microsecond_timestamp"+ pure ()+ -- The INT64 storage also holds the four hours before the range, but+ -- DuckDB does not accept them.+ forM_ [minBound, negate (maxBound :: Int64) + 1, -9223372022400000001] \value ->+ assertOutsideTimeRange (query conn "SELECT ?" (Only (local value)) :: IO [Only LocalTime])+ , testCase "out-of-range dates and timestamps fail without abort" $ withConnection ":memory:" \conn -> do+ let hugeDay = fromGregorian 10000000 1 1+ local = LocalTime hugeDay (TimeOfDay 0 0 0)+ assertOutsideTimeRange (query conn "SELECT ?" (Only hugeDay) :: IO [Only Day])+ assertOutsideTimeRange (query conn "SELECT ?" (Only local) :: IO [Only LocalTime])+ assertOutsideTimeRange (query conn "SELECT ?" (Only (UTCTime hugeDay 0)) :: IO [Only UTCTime])+ , testCase "dates round trip at the ends of the range of DuckDB" $ withConnection ":memory:" \conn -> do+ forM_ [negate (maxBound :: Int32) + 1, -1, 0, maxBound - 1] \value -> do+ let day = addDays (toInteger value) (fromGregorian 1970 1 1)+ (query conn "SELECT ?" (Only day) :: IO [Only Day]) >>= (@?= [Only day])+ -- The INT32 storage also holds day -2147483648, but DuckDB does not accept it.+ assertOutsideTimeRange (query conn "SELECT ?" (Only (addDays (toInteger (minBound :: Int32)) (fromGregorian 1970 1 1))) :: IO [Only Day])+ , testCase "input infinity sentinels are rejected" $ withConnection ":memory:" \conn -> do+ forM_ [toInteger (maxBound :: Int32), negate (toInteger (maxBound :: Int32))] \days ->+ assertOutsideTimeRange (query conn "SELECT ?" (Only (addDays days (fromGregorian 1970 1 1))) :: IO [Only Day])+ forM_ [toInteger (maxBound :: Int64), negate (toInteger (maxBound :: Int64))] \micros -> do+ let value = utcToLocalTime utc (posixSecondsToUTCTime (fromRational (micros % 1000000)))+ assertOutsideTimeRange (query conn "SELECT ?" (Only value) :: IO [Only LocalTime])+ , testCase "UTC binding is TIMESTAMPTZ under a non-UTC zone" $ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "SET TimeZone = 'Pacific/Auckland'"+ let value = UTCTime (fromGregorian 2000 1 2) 12345+ (query conn "SELECT typeof(?), ?" (value, value) :: IO [(Text, UTCTime)]) >>= (@?= [("TIMESTAMP WITH TIME ZONE", value)])+ _ <- execute_ conn "CREATE TABLE utc_binding(value TIMESTAMPTZ)"+ _ <- execute conn "INSERT INTO utc_binding VALUES (?)" (Only value)+ (query_ conn "SELECT value FROM utc_binding" :: IO [Only UTCTime]) >>= (@?= [Only value])+ , testCase "DECIMAL(38) retains exact integer precision" $ withConnection ":memory:" \conn -> do+ let decimal = DecimalValue 38 9 12345678901234567890123456789012345678+ struct = singleField (LogicalTypeDecimal 38 9) (FieldDecimal decimal)+ [Only result] <- query conn "SELECT ?" (Only struct) :: IO [Only (StructValue FieldValue)]+ structValueFields result @?= structValueFields struct+ , testCase "DECIMAL and ENUM columns decode at each storage width" $ withConnection ":memory:" \conn -> do+ -- Each column has a NULL between values, so one reader serves valid and invalid rows.+ forM_ [(4, 1), (9, 3), (18, 3), (38, 9)] \(width, scale) -> do+ let dtype = "DECIMAL(" <> Text.pack (show width) <> "," <> Text.pack (show scale) <> ")"+ decimal units = FieldDecimal (DecimalValue width scale (units * 10 ^ scale))+ rows <-+ query_ conn (Query ("SELECT CASE WHEN i = 1 THEN NULL ELSE (i - 2)::" <> dtype <> " END FROM range(4) t(i)")) :: IO [Only FieldValue]+ rows @?= map Only [decimal (-2), FieldNull, decimal 0, decimal 1]+ forM_ [200, 300, 70000 :: Int] \size -> do+ let label index = "v" <> Text.pack (show index)+ dtype = "enum_width_" <> Text.pack (show size)+ _ <-+ execute_+ conn+ (Query ("CREATE TYPE " <> dtype <> " AS ENUM (" <> Text.intercalate "," ["'" <> label index <> "'" | index <- [0 .. size - 1]] <> ")"))+ rows <-+ query_ conn (Query ("SELECT x::" <> dtype <> " FROM (VALUES ('v0'), (NULL), ('" <> label (size - 1) <> "')) t(x)")) :: IO [Only FieldValue]+ rows @?= map Only [FieldEnum 0 (label (0 :: Int)), FieldNull, FieldEnum (fromIntegral (size - 1)) (label (size - 1))]+ , testCase "DECIMAL and BIGINT values read as Double and Float give the casts of DuckDB" $ withConnection ":memory:" \conn -> do+ -- Each row has the read of v as Double and Float, and the casts of v+ -- to DOUBLE and FLOAT read at their own types. The value lists are+ -- fixed. See Note [DECIMAL as DOUBLE and FLOAT] in+ -- "Database.DuckDB.Simple.FromField". For scale 0 and a width of 19+ -- to 38, the library copies the cast of DuckDB 2.0.0, which is not+ -- always the nearest value. If a later DuckDB version changes the+ -- rounding of that cast, the first assertion fails and shows the new+ -- values of DuckDB. DuckDB 2.0.0 gives these values. The nearest values+ -- are -9007199254740994 and 3.3569516755834854e36.+ query_ conn "SELECT (-9007199254740994)::DECIMAL(19,0)::DOUBLE, 3356951675583485076575379654993209529::DECIMAL(38,0)::DOUBLE"+ >>= (@?= [(-9007199254740992 :: Double, 3.356951675583485e36 :: Double)])+ let bigints =+ filter (\v -> v >= toInteger (minBound :: Int64) && v <= toInteger (maxBound :: Int64)) (toInteger (minBound :: Int64) : decimalIntegers 19)+ kinds =+ [ ("DECIMAL(" <> show width <> "," <> show scale <> ")", map (decimalText scale) (decimalIntegers width <> roundingTies width scale))+ | (width, scale) <- decimalKinds+ ]+ <> [("BIGINT", map (Text.pack . show) bigints)]+ forM_ kinds \(sqlType, texts) -> do+ rows <-+ query+ conn+ (fromString ("SELECT s, v, v::DOUBLE, v, v::FLOAT FROM (SELECT s, s::" <> sqlType <> " AS v FROM unnest(?::VARCHAR[]) t(s))"))+ (Only texts)+ :: IO [(Text, Double, Double, Float, Float)]+ length rows @?= length texts+ [(sqlType, text, value, cast) | (text, value, cast, _, _) <- rows, castDoubleToWord64 value /= castDoubleToWord64 cast] @?= []+ if sqlType /= "BIGINT"+ then [(sqlType, text, value, cast) | (text, _, _, value, cast) <- rows, castFloatToWord32 value /= castFloatToWord32 cast] @?= []+ else do+ -- CI on macOS (Apple Silicon) showed a cast of DuckDB 2.0.0 from+ -- BIGINT to FLOAT that rounds two times, through DOUBLE, for some+ -- values above 2^53: 9007199791611905 gave 9.007199e15, and the+ -- nearest FLOAT is 9.0072e15. On Linux x86-64 and aarch64 the cast+ -- gave the nearest value. The read of the library is the nearest+ -- value on each platform. The cast of DuckDB is one of the two.+ let integer text = read (Text.unpack text) :: Integer+ nearest text = fromRational (toRational (integer text)) :: Float+ throughDouble text = realToFrac (fromInteger (integer text) :: Double) :: Float+ bits = castFloatToWord32+ [(text, value) | (text, _, _, value, _) <- rows, bits value /= bits (nearest text)] @?= []+ [(text, cast) | (text, _, _, _, cast) <- rows, bits cast `notElem` [bits (nearest text), bits (throughDouble text)]] @?= []+ , testCase "a read of another SQL type gives the value of the cast of DuckDB" $ withConnection ":memory:" \conn -> do+ -- Each value reads as the Haskell type, next to its cast to the SQL+ -- type of the Haskell type in the same row. See+ -- Note [Reads of another SQL type] in "Database.DuckDB.Simple.FromField".+ let signedValues = ["-100", "0", "7", "100"]+ unsignedValues = ["0", "7", "100"]+ signedSources = ["TINYINT", "SMALLINT", "INTEGER", "BIGINT", "HUGEINT"]+ unsignedSources = ["UTINYINT", "USMALLINT", "UINTEGER", "UBIGINT", "UHUGEINT"]+ signedTarget :: (FromField a, Eq a, Show a) => Proxy a -> Text -> Assertion+ signedTarget proxy target = do+ forM_ signedSources \source -> assertReadsAsCast conn proxy source target signedValues+ assertReadsAsCast conn proxy "UHUGEINT" target unsignedValues+ unsignedTarget :: (FromField a, Eq a, Show a) => Proxy a -> Text -> Assertion+ unsignedTarget proxy target = forM_ (signedSources <> unsignedSources) \source -> assertReadsAsCast conn proxy source target unsignedValues+ timestamps = ["2024-01-02 03:04:05", "1969-12-31 23:59:59"]+ timePairs = do+ forM_ ["TIMESTAMP", "TIMESTAMP_S", "TIMESTAMP_MS", "TIMESTAMP_NS"] \source -> do+ assertReadsAsCast conn (Proxy @Day) source "DATE" timestamps+ assertReadsAsCast conn (Proxy @(Unbounded Day)) source "DATE" timestamps+ assertReadsAsCast conn (Proxy @(Unbounded Day)) "TIMESTAMP" "DATE" ["infinity", "-infinity"]+ forM_ ["TIMESTAMP", "TIMESTAMP_S", "TIMESTAMP_MS"] \source -> assertReadsAsCast conn (Proxy @TimeOfDay) source "TIME" timestamps+ assertReadsAsCast conn (Proxy @TimeOfDay) "TIMESTAMP" "TIME" ["2024-01-02 03:04:05.123456"]+ assertReadsAsCast+ conn+ (Proxy @TimeOfDay)+ "TIMESTAMP_NS"+ "TIME_NS"+ ["2024-01-02 03:04:05.123456789", "1969-12-31 23:59:59.999999999"]+ forM_ [("DATE", ["2024-01-02", "1969-12-31"]), ("TIMESTAMP_S", timestamps), ("TIMESTAMP_MS", timestamps)] \(source, values) -> do+ assertReadsAsCast conn (Proxy @LocalTime) source "TIMESTAMP" values+ assertReadsAsCast conn (Proxy @(Unbounded LocalTime)) source "TIMESTAMP" values+ assertReadsAsCast conn (Proxy @(Unbounded LocalTime)) "DATE" "TIMESTAMP" ["infinity", "-infinity"]+ signedTarget (Proxy @Int8) "TINYINT"+ signedTarget (Proxy @Int16) "SMALLINT"+ signedTarget (Proxy @Int32) "INTEGER"+ signedTarget (Proxy @Int64) "BIGINT"+ signedTarget (Proxy @Int) "BIGINT"+ signedTarget (Proxy @Integer) "BIGNUM"+ forM_ unsignedSources \source -> assertReadsAsCast conn (Proxy @Integer) source "BIGNUM" unsignedValues+ unsignedTarget (Proxy @Natural) "BIGNUM"+ unsignedTarget (Proxy @Word8) "UTINYINT"+ unsignedTarget (Proxy @Word16) "USMALLINT"+ unsignedTarget (Proxy @Word32) "UINTEGER"+ unsignedTarget (Proxy @Word64) "UBIGINT"+ unsignedTarget (Proxy @Word) "UBIGINT"+ forM_ (take 4 signedSources) \source -> do+ assertReadsAsCast conn (Proxy @Bool) source "BOOLEAN" ["-2", "0", "1", "7"]+ assertReadsAsCast conn (Proxy @Double) source "DOUBLE" signedValues+ assertReadsAsCast conn (Proxy @Float) source "FLOAT" signedValues+ assertReadsAsCast conn (Proxy @Text) source "VARCHAR" signedValues+ assertReadsAsCast conn (Proxy @String) source "VARCHAR" signedValues+ assertReadsAsCast conn (Proxy @Double) "BIGINT" "DOUBLE" ["9007199254740993", "-9223372036854775808"]+ assertReadsAsCast conn (Proxy @Double) "FLOAT" "DOUBLE" ["1.25", "0.1", "-3.5"]+ assertReadsAsCast conn (Proxy @Float) "DOUBLE" "FLOAT" ["1.25", "0.1", "16777217"]+ assertReadsAsCast conn (Proxy @Text) "BOOLEAN" "VARCHAR" ["true", "false"]+ assertReadsAsCast conn (Proxy @String) "BOOLEAN" "VARCHAR" ["true", "false"]+ assertReadsAsCast conn (Proxy @BS.ByteString) "BIT" "BLOB" ["101", "10110011101", "00000000"]+ assertReadsAsCast conn (Proxy @BS.ByteString) "GEOMETRY" "BLOB" ["POINT (1 2)", "LINESTRING (0 0, 1 1)"]+ assertReadsAsCast+ conn+ (Proxy @UUID.UUID)+ "VARCHAR"+ "UUID"+ ["00000000-0000-0000-0000-000000000000", "a0eebc99-9c0b-4ef8-bb6d-6bb9bd380a11"]+ assertReadsAsCast conn (Proxy @[Int32]) "INTEGER[3]" "INTEGER[]" ["[1, 2, 3]", "[-1, 2, 0]"]+ -- An ENUM value reads as its label, the text of the cast to VARCHAR.+ assertReadsAsCast conn (Proxy @Text) "ENUM('10', 'beta')" "VARCHAR" ["10", "beta"]+ assertReadsAsCast conn (Proxy @String) "ENUM('10', 'beta')" "VARCHAR" ["10", "beta"]+ -- The time pairs do not use the session time zone.+ timePairs+ _ <- execute_ conn "SET TimeZone = 'Pacific/Auckland'"+ timePairs+ , testCase "a field of a generic record reads the SQL types of the top-level read" $ withConnection ":memory:" \conn -> do+ -- See Note [Reads of another SQL type] in "Database.DuckDB.Simple.FromField".+ [Only column] <- query_ conn "SELECT [1, 2, 3]::INTEGER[3]" :: IO [Only [Int32]]+ let holder list array map' = "SELECT {'listField': " <> list <> ", 'arrayField': " <> array <> ", 'mapField': " <> map' <> "}"+ (query_ conn (Query (holder "[1, 2, 3]::INTEGER[3]" "[4, 5]::INTEGER[2]" "MAP {'k': 6}")) :: IO [Only Holder])+ >>= (@?= [Only (Holder column (sized [4, 5]) (Map.singleton "k" 6))])+ -- The error of a value that a decoder does not read names its SQL type.+ let refused :: forall a. (FromField a, Show a) => Proxy a -> Text -> Text -> Assertion+ refused _ sql expected =+ try (query_ conn (Query sql) :: IO [Only a]) >>= \case+ Left SQLError{sqlErrorMessage} -> sqlErrorMessage @?= expected+ Right rows -> assertFailure ("expected an error for " <> Text.unpack sql <> ", got " <> show rows)+ refused+ (Proxy @Holder)+ (holder "MAP {'k': 1}" "[4, 5]::INTEGER[2]" "MAP {'k': 6}")+ "duckdb-simple: field listField: a value has type MAP(VARCHAR, INTEGER) but expected LIST or ARRAY"+ refused+ (Proxy @Holder)+ (holder "[1]" "[4, 5, 6]" "MAP {'k': 6}")+ "duckdb-simple: field arrayField: a value has type INTEGER[] but expected SizedArray 2: the value has 3 elements"+ refused+ (Proxy @Holder)+ (holder "[1]" "[4, 5]::INTEGER[2]" "[6]")+ "duckdb-simple: field mapField: a value has type INTEGER[] but expected MAP"+ refused (Proxy @Holder) "SELECT 7" "duckdb-simple: a value has type INTEGER but expected STRUCT for a product type"+ refused (Proxy @Colour) "SELECT {'a': 1}" "duckdb-simple: a value has type STRUCT but expected UNION for a sum type"+ refused+ (Proxy @Colour)+ "SELECT union_value(Red := 1)::UNION(Red INTEGER, Blue INTEGER)"+ "duckdb-simple: a value has type INTEGER but expected STRUCT for a member of a sum type"+ , testCase "a refused field of a generic record names the field and the two types" $ withConnection ":memory:" \conn -> do+ try (query_ conn "SELECT {'stampedAt': TIMESTAMP '2024-01-02 03:04:05'} AS s" :: IO [Only Stamped]) >>= \case+ Left SQLError{sqlErrorMessage} -> sqlErrorMessage @?= "duckdb-simple: field stampedAt: a value has type TIMESTAMP but expected UTCTime"+ Right rows -> assertFailure ("expected the error of the field, got " <> show rows)+ (query_ conn "SELECT {'stampedAt': TIMESTAMPTZ '2024-01-02 03:04:05+00'} AS s" :: IO [Only Stamped])+ >>= (@?= [Only (Stamped (UTCTime (fromGregorian 2024 1 2) 11045))])+ , testCase "a read of another SQL type that needs the time zone or another text raises Incompatible" $ withConnection ":memory:" \conn -> do+ -- See Note [Reads of another SQL type] in "Database.DuckDB.Simple.FromField".+ let refusedPairs = do+ forM_ ["TIMESTAMP", "TIMESTAMP_S", "TIMESTAMP_MS", "TIMESTAMP_NS"] \source -> do+ assertIncompatible conn ("'2024-01-02 03:04:05'::" <> source) () source (Proxy @UTCTime) "UTCTime"+ assertIncompatible conn ("'2024-01-02 03:04:05'::" <> source) () source (Proxy @(Unbounded UTCTime)) "Unbounded UTCTime"+ assertIncompatible conn "DATE '2024-01-02'" () "DATE" (Proxy @UTCTime) "UTCTime"+ assertIncompatible conn "DATE '2024-01-02'" () "DATE" (Proxy @(Unbounded UTCTime)) "Unbounded UTCTime"+ forM_ [("TIMESTAMPTZ", "TIMESTAMP WITH TIME ZONE"), ("TIMESTAMPTZ_NS", "TIMESTAMPTZ_NS")] \(source, sqlName) -> do+ assertIncompatible conn ("'2024-01-02 03:04:05+00'::" <> source) () sqlName (Proxy @LocalTime) "LocalTime"+ assertIncompatible+ conn+ ("'2024-01-02 03:04:05+00'::" <> source)+ ()+ sqlName+ (Proxy @(Unbounded LocalTime))+ "Unbounded LocalTime"+ forM_ [("FLOAT", "0.01::FLOAT"), ("DOUBLE", "0.01::DOUBLE")] \(sqlName, expression) -> do+ assertIncompatible conn expression () sqlName (Proxy @Text) "Text"+ assertIncompatible conn expression () sqlName (Proxy @String) "Text"+ assertIncompatible conn "'a\\x41'::VARCHAR" () "VARCHAR" (Proxy @BS.ByteString) "ByteString"+ -- An integer type does not read an ENUM value. The cast of DuckDB+ -- parses the label, and enum_code gives the index.+ let enumAs :: (FromField a, Show a) => Proxy a -> Text -> Assertion+ enumAs = assertIncompatible conn "'10'::ENUM('10', '20')" () "ENUM('10', '20')"+ enumAs (Proxy @Int8) "Int8"+ enumAs (Proxy @Int16) "Int16"+ enumAs (Proxy @Int32) "Int32"+ enumAs (Proxy @Int64) "Int64"+ enumAs (Proxy @Int) "Int"+ enumAs (Proxy @Integer) "Integer"+ enumAs (Proxy @Natural) "Natural"+ enumAs (Proxy @Word8) "Word8"+ enumAs (Proxy @Word16) "Word16"+ enumAs (Proxy @Word32) "Word32"+ enumAs (Proxy @Word64) "Word64"+ enumAs (Proxy @Word) "Word"+ -- The SQL in place of the refused read: enum_code(c)::BIGINT gives+ -- the index, and c::INTEGER parses the label.+ ( query_+ conn+ "SELECT enum_code(c)::BIGINT, enum_code(c)::BIGINT, enum_code(c)::BIGINT, c::INTEGER FROM (VALUES ('20'::ENUM('10', '20'))) t(c)"+ :: IO [(Int8, Word8, Natural, Int32)]+ )+ >>= (@?= [(1, 1, 1, 20)])+ -- The SQL in place of each refused read, also for the year 1500,+ -- which a nanosecond value cannot hold.+ forM_ [2024, 1500] \year -> do+ let day = fromGregorian year 1 2+ text = Text.pack (show day)+ (query_ conn (Query ("SELECT c::TIMESTAMP AT TIME ZONE 'UTC' FROM (VALUES (TIMESTAMP '" <> text <> " 03:04:05')) t(c)")) :: IO [Only UTCTime])+ >>= (@?= [Only (UTCTime day 11045)])+ forM_ ["TIMESTAMP_S", "TIMESTAMP_MS"] \source ->+ ( query_ conn (Query ("SELECT c::TIMESTAMP AT TIME ZONE 'UTC' FROM (VALUES ('" <> text <> " 03:04:05'::" <> source <> ")) t(c)"))+ :: IO [Only UTCTime]+ )+ >>= (@?= [Only (UTCTime day 11045)])+ (query_ conn (Query ("SELECT c::TIMESTAMP AT TIME ZONE 'UTC' FROM (VALUES (DATE '" <> text <> "')) t(c)")) :: IO [Only UTCTime])+ >>= (@?= [Only (UTCTime day 0)])+ (query_ conn (Query ("SELECT c AT TIME ZONE 'UTC' FROM (VALUES (TIMESTAMPTZ '" <> text <> " 03:04:05+00')) t(c)")) :: IO [Only LocalTime])+ >>= (@?= [Only (LocalTime day (TimeOfDay 3 4 5))])+ (query_ conn "SELECT c AT TIME ZONE 'UTC' FROM (VALUES ('2024-01-02 03:04:05.123456789'::TIMESTAMP_NS)) t(c)" :: IO [Only UTCTime])+ >>= (@?= [Only (UTCTime (fromGregorian 2024 1 2) 11045.123456789)])+ (query_ conn "SELECT 0.01::DOUBLE::VARCHAR, encode('a\\x41')" :: IO [(Text, BS.ByteString)]) >>= (@?= [("0.01", "a\\x41")])+ refusedPairs+ _ <- execute_ conn "SET TimeZone = 'Pacific/Auckland'"+ refusedPairs+ , testCase "invalid DECIMAL width scale and magnitude are rejected" $ withConnection ":memory:" \conn ->+ forM_+ [ (DecimalValue 0 0 0, assertInputError)+ , (DecimalValue 39 0 0, assertInputError)+ , (DecimalValue 2 3 1, assertInputError)+ , (DecimalValue 2 0 100, assertRangeError)+ , (DecimalValue 2 0 (-100), assertRangeError)+ ]+ \(decimal, rejected) ->+ rejected+ ( query conn "SELECT ?" (Only (singleField (LogicalTypeDecimal (decimalWidth decimal) (decimalScale decimal)) (FieldDecimal decimal)))+ :: IO [Only FieldValue]+ )+ , testCase "nested conversion failure leaves the connection usable" $ withConnection ":memory:" \conn -> do+ let days = sized [fromGregorian 2000 1 1, fromGregorian 10000000 1 1] :: SizedArray 2 Day+ assertOutsideTimeRange (query conn "SELECT ?" (Only days) :: IO [Only FieldValue])+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "generic NULL product and payload fail conversion" $ do+ assertConversionFails (genericFromFieldValue FieldNull :: Conversion NamedRecord)+ let member = UnionMemberType "DataMember" (LogicalTypeScalar LogicalTypeIdBigInt)+ union = UnionValue 0 "DataMember" FieldNull (listArray (0, 0) [member])+ assertConversionFails (genericFromFieldValue (FieldUnion union) :: Conversion NullableSum)+ , testCase "generic records decode by name" $ withConnection ":memory:" \conn -> do+ [Only struct] <- query_ conn "SELECT {'secondValue': 2::BIGINT, 'firstValue': 1::BIGINT}" :: IO [Only (StructValue FieldValue)]+ runConversion (genericFromFieldValue (FieldStruct struct)) >>= (@?= Ok (NamedRecord 1 2))+ , testCase "generic sums decode by member name" $ withConnection ":memory:" \conn -> do+ [Only union] <- query_ conn "SELECT union_value(DataMember := {'field1': 42::BIGINT})" :: IO [Only (UnionValue FieldValue)]+ runConversion (genericFromFieldValue (FieldUnion union)) >>= (@?= Ok (DataMember 42))+ , testCase "union binding rejects inconsistent member name" $ withConnection ":memory:" \conn -> do+ let member = UnionMemberType "right" (LogicalTypeScalar LogicalTypeIdBigInt)+ union = UnionValue 0 "wrong" (FieldInt64 42) (listArray (0, 0) [member])+ assertInputError (query conn "SELECT ?" (Only union) :: IO [Only FieldValue])+ , testCase "union binding rejects a tag out of range" $ withConnection ":memory:" \conn -> do+ let member = UnionMemberType "only" (LogicalTypeScalar LogicalTypeIdBigInt)+ union = UnionValue 1 "only" (FieldInt64 42) (listArray (0, 0) [member])+ try (query conn "SELECT ?" (Only union) :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ sqlErrorType @?= Just ErrorInputInvalid+ sqlErrorMessage @?= "duckdb-simple: union value tag out of range"+ Right rows -> assertFailure ("expected the tag error, got " <> show rows)+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "generic records reject wrong names" $ do+ case genericToStructValue (NamedRecord 1 2) of+ Nothing -> assertFailure "missing generic struct"+ Just struct -> do+ let fields = listArray (0, 1) [StructField "wrong" (FieldInt64 1), StructField "secondValue" (FieldInt64 2)]+ assertConversionFails (genericFromFieldValue (FieldStruct struct{structValueFields = fields}) :: Conversion NamedRecord)+ , testCase "unnamed tuples preserve scalar and nested fields on round trip" $ withConnection ":memory:" \conn -> do+ let tuple =+ TupleValue+ (listArray (0, 1) [FieldInt64 42, FieldText "λ"])+ (listArray (0, 1) [LogicalTypeScalar LogicalTypeIdBigInt, LogicalTypeScalar LogicalTypeIdVarchar])+ (query_ conn "SELECT row(42::BIGINT, 'λ')" :: IO [Only (TupleValue FieldValue)]) >>= (@?= [Only tuple])+ (query_ conn "SELECT [row(42::BIGINT, 'λ')]" :: IO [Only [TupleValue FieldValue]]) >>= (@?= [Only [tuple]])+ (query conn "SELECT ?" (Only tuple) :: IO [Only (TupleValue FieldValue)]) >>= (@?= [Only tuple])+ , testCase "empty tuples retain their type independently of empty structs" $ withConnection ":memory:" \conn -> do+ let tuple = TupleValue (listArray (0, -1) []) (listArray (0, -1) []) :: TupleValue FieldValue+ struct = StructValue (listArray (0, -1) []) (listArray (0, -1) []) Map.empty :: StructValue FieldValue+ (query_ conn "SELECT row(), [row()]" :: IO [(TupleValue FieldValue, [TupleValue FieldValue])]) >>= (@?= [(tuple, [tuple])])+ (query conn "SELECT typeof(?), ?" (tuple, tuple) :: IO [(Text, TupleValue FieldValue)]) >>= (@?= [("TUPLE", tuple)])+ (query conn "SELECT typeof(?), ?" (struct, struct) :: IO [(Text, StructValue FieldValue)]) >>= (@?= [("STRUCT", struct)])+ withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ forM_ [tupleValueTypeRep tuple, structValueTypeRep struct] \rep ->+ bracket (logicalTypeFromRep factory rep) destroyLogicalType \logical -> do+ logicalTypeToRep factory logical >>= (@?= rep)+ nativeTypeText logical >>= (@?= logicalTypeSQL rep)+ assertInputError (query conn "SELECT ?" (Only tuple{tupleValueFields = listArray (4, 4) [FieldNull]}) :: IO [Only FieldValue])+ , testCase "filtered tuple cursors retain positional fields" $ withConnection ":memory:" \conn -> do+ rows <- Streaming.fold_ conn "SELECT row(i::BIGINT, 'λ') FROM range(6) t(i) WHERE i % 2 = 1" [] $ \values (Only tuple) ->+ pure (values <> [elems (tupleValueFields tuple)])+ rows @?= [[FieldInt64 value, FieldText "λ"] | value <- [1, 3, 5]]+ , testCase "composite VARCHAR collations retain native comparison semantics" $ withConnection ":memory:" \conn -> do+ let collated = LogicalTypeCollatedVarchar "nocase"+ [Only original] <-+ query_ conn "SELECT {'text':'A' COLLATE nocase,'items':['B' COLLATE nocase], 'optional':NULL::VARCHAR COLLATE nocase}"+ :: IO [Only (StructValue FieldValue)]+ map structFieldValue (elems (structValueTypes original)) @?= [collated, LogicalTypeList collated, collated]+ ( query conn "SELECT value.text='a', value.items[1]='b', value.optional IS NULL FROM (SELECT ? AS value)" (Only original)+ :: IO [(Bool, Bool, Bool)]+ )+ >>= (@?= [(True, True, True)])+ (query conn "SELECT ?" (Only original) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only original])+ [Only tuple] <- query_ conn "SELECT row('A' COLLATE nocase)" :: IO [Only (TupleValue FieldValue)]+ (query conn "SELECT value[1]='a' FROM (SELECT ? AS value)" (Only tuple) :: IO [Only Bool]) >>= (@?= [Only True])+ withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ bracket (logicalTypeFromRep factory collated) destroyLogicalType \logical -> logicalTypeToRep factory logical >>= (@?= collated)+ , testCase "logical type Unicode names round trip" $ withConnection ":memory:" \conn -> withConnectionHandle conn \native -> withConnectionFactory native \factory -> do+ let logical = LogicalTypeStruct (listArray (0, 0) [StructField "íslenska_λ_😀" (LogicalTypeScalar LogicalTypeIdBigInt)])+ bracket (logicalTypeFromRep factory logical) destroyLogicalType \handle ->+ logicalTypeToRep factory handle >>= (@?= logical)+ , testCase "a column type keeps the alias JSON at each depth, and reads as its base type" $ withConnection ":memory:" \conn -> do+ let json = LogicalTypeAlias "JSON" (LogicalTypeScalar LogicalTypeIdVarchar)+ varchar = LogicalTypeScalar LogicalTypeIdVarchar+ structType =+ LogicalTypeStruct+ (listArray (0, 2) [StructField "k" json, StructField "js" (LogicalTypeList json), StructField "m" (LogicalTypeMap varchar json)])+ _ <- execute_ conn "CREATE TABLE aliased (j JSON, js JSON[], s STRUCT(k JSON, js JSON[], m MAP(VARCHAR, JSON)), m MAP(VARCHAR, JSON))"+ description <- Catalog.describeTable conn "aliased"+ map Catalog.columnDescriptionType (Catalog.tableDescriptionColumns description)+ @?= [json, LogicalTypeList json, structType, LogicalTypeMap varchar json]+ _ <- execute_ conn "INSERT INTO aliased VALUES ('{\"a\": 1}', ['[1]'], {'k': '2', 'js': ['[3]'], 'm': MAP {'y': '4'}}, MAP {'x': '3'})"+ query_ conn "SELECT j, js, m FROM aliased"+ >>= (@?= [("{\"a\": 1}" :: Text, ["[1]"] :: [Text], Map.fromList [("x" :: Text, "3" :: Text)])])+ -- The type fields of a STRUCT value keep the alias, so the value binds+ -- again with its type.+ [Only struct] <- query_ conn "SELECT s FROM aliased" :: IO [Only (StructValue FieldValue)]+ structValueTypeRep struct @?= structType+ map structFieldValue (elems (structValueFields struct))+ @?= [FieldText "2", FieldList [FieldText "[3]"], FieldMap [(FieldText "y", FieldText "4")]]+ query conn "SELECT typeof(?), ?" (struct, struct)+ >>= (@?= [("STRUCT(k JSON, js JSON[], m MAP(VARCHAR, JSON))" :: Text, struct)])+ , testCase "a column of a type from CREATE TYPE has the base type and no name" $ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TYPE meters AS BIGINT"+ _ <- execute_ conn "CREATE TYPE mood AS ENUM ('sad', 'happy')"+ _ <- execute_ conn "CREATE TABLE created (m meters, mo mood)"+ description <- Catalog.describeTable conn "created"+ map Catalog.columnDescriptionType (Catalog.tableDescriptionColumns description)+ @?= [LogicalTypeScalar LogicalTypeIdBigInt, LogicalTypeEnum (listArray (0, 1) ["sad", "happy"])]+ , testCase "a registered custom type keeps its name, and reads and binds as its base type" $ withConnection ":memory:" \conn -> do+ let decimal = LogicalTypeDecimal 9 2+ point =+ LogicalTypeStruct+ (listArray (0, 1) [StructField "x" (LogicalTypeScalar LogicalTypeIdInteger), StructField "label" (LogicalTypeScalar LogicalTypeIdVarchar)])+ custom =+ [ ("money", decimal)+ , ("point", point)+ , ("tags", LogicalTypeList (LogicalTypeScalar LogicalTypeIdVarchar))+ , ("level", LogicalTypeEnum (listArray (0, 1) ["low", "high"]))+ , ("meters", LogicalTypeScalar LogicalTypeIdBigInt)+ ]+ forM_ custom \(name, base) -> R.registerCustomType conn name (StructuredType base)+ _ <- execute_ conn "CREATE TABLE custom (a money, b point, c tags, d level, e meters, f money[])"+ description <- Catalog.describeTable conn "custom"+ map Catalog.columnDescriptionType (Catalog.tableDescriptionColumns description)+ @?= [LogicalTypeAlias name base | (name, base) <- custom] <> [LogicalTypeList (LogicalTypeAlias "money" decimal)]+ _ <- execute_ conn "INSERT INTO custom VALUES (1.25, {'x': 1, 'label': 'a'}, ['t'], 'high', 7, [2.5])"+ query_ conn "SELECT a, c, d, e, f FROM custom"+ >>= (@?= [(DecimalValue 9 2 125, ["t"] :: [Text], "high" :: Text, 7 :: Int64, [DecimalValue 9 2 250])])+ [Only struct] <- query_ conn "SELECT b FROM custom" :: IO [Only (StructValue FieldValue)]+ map structFieldValue (elems (structValueFields struct)) @?= [FieldInt32 1, FieldText "a"]+ -- The binding uses the range of the base type.+ try (execute conn "INSERT INTO custom (e) VALUES (?)" (Only (2 ^ (127 :: Int) :: Integer))) >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorType @?= Just ErrorTypeConversion+ sqlErrorMessage @?= "duckdb-simple: the value 170141183460469231731687303715884105728 is out of range for a parameter of the type BIGINT"+ Right _ -> assertFailure "expected the duckdb-simple range error for a custom type over BIGINT"+ , testCase "an alias type round trips through a native type" $ withConnection ":memory:" \conn -> do+ let json = LogicalTypeAlias "JSON" (LogicalTypeScalar LogicalTypeIdVarchar)+ money = LogicalTypeAlias "money" (LogicalTypeDecimal 9 2)+ R.registerCustomType conn "money" (StructuredType (LogicalTypeDecimal 9 2))+ withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ forM_ [(json, "JSON"), (money, "money"), (LogicalTypeList json, "JSON[]")] \(logical, text) ->+ bracket (logicalTypeFromRep factory logical) destroyLogicalType \handle -> do+ logicalTypeToRep factory handle >>= (@?= logical)+ -- The type is the type that DuckDB gives for the type text.+ bracket (R.withText text (R.output . c_duckdb_v2_factory_create_type_from_text factory)) destroyLogicalType \parsed ->+ R.output (c_duckdb_v2_logical_type_is_equal handle parsed) >>= (@?= 1)+ , testCase "aliases equal to their type id names retain native type identity" $ withConnection ":memory:" \conn ->+ withConnectionHandle conn \native -> withConnectionFactory native \factory -> do+ let integer = LogicalTypeScalar LogicalTypeIdInteger+ decimal = LogicalTypeDecimal 9 2+ struct = LogicalTypeStruct (listArray (0, 0) [StructField "a" integer])+ integerAlias = LogicalTypeAlias "INTEGER" integer+ decimalAlias = LogicalTypeAlias "DECIMAL" decimal+ structAlias = LogicalTypeAlias "STRUCT" struct+ nested =+ LogicalTypeStruct+ (listArray (0, 2) [StructField "i" integerAlias, StructField "d" decimalAlias, StructField "s" structAlias])+ forM_+ [ integerAlias+ , decimalAlias+ , structAlias+ , LogicalTypeAlias "STRUCT" (LogicalTypeStruct (listArray (0, -1) []))+ , LogicalTypeAlias "TUPLE" (LogicalTypeTuple (listArray (0, -1) []))+ , LogicalTypeAlias "ENUM" (LogicalTypeEnum (listArray (0, 1) ["first", "it's second"]))+ , LogicalTypeAlias "VARCHAR" (LogicalTypeCollatedVarchar "nocase")+ , LogicalTypeList integerAlias+ , nested+ ]+ \rep -> bracket (logicalTypeFromRep factory rep) destroyLogicalType \logical -> do+ copied <- logicalTypeToRep factory logical+ copied @?= rep+ logicalTypeToBaseRep logical >>= (@?= withoutAliases rep)+ nativeTypeText logical >>= (@?= logicalTypeSQL copied)+ bracket (logicalTypeFromRep factory copied) destroyLogicalType \rebuilt ->+ R.output (c_duckdb_v2_logical_type_is_equal logical rebuilt) >>= (@?= 1)+ , testCase "logicalTypeSQL gives the type text of DuckDB for each kind, and a read error names that text" $ withConnection ":memory:" \conn -> do+ R.registerCustomType conn "money" (StructuredType (LogicalTypeDecimal 9 2))+ keywords <- map fromOnly <$> (query_ conn "SELECT DISTINCT keyword_name FROM duckdb_keywords() ORDER BY 1" :: IO [Only Text])+ let keywordStruct = "STRUCT(" <> Text.intercalate ", " ["\"" <> keyword <> "\" INTEGER" | keyword <- keywords] <> ")"+ withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ forM_+ [ "DECIMAL(18,3)"+ , "TIMESTAMP_NS"+ , "TIMESTAMPTZ"+ , "TIMETZ"+ , "\"NULL\""+ , "ENUM('sad', 'it''s')"+ , "BIGINT[]"+ , "INTEGER[3]"+ , "INTEGER[3][]"+ , "MAP(VARCHAR, BIGINT)"+ , "STRUCT(a INTEGER, B VARCHAR, \"c d\" BIGINT[], \"x\"\"y\" DOUBLE, \"1a\" INTEGER, \"Name\" INTEGER, \"_1\" INTEGER, _a INTEGER)"+ , "UNION(n INTEGER, s VARCHAR)"+ , "TUPLE(INTEGER, VARCHAR)"+ , "GEOMETRY"+ , "GEOMETRY('OGC:CRS84')"+ , "VARCHAR COLLATE nocase"+ , "VARIANT"+ , "JSON"+ , "JSON[]"+ , "STRUCT(k JSON)"+ , "money"+ , keywordStruct+ ]+ \text ->+ bracket (R.withText text (R.output . c_duckdb_v2_factory_create_type_from_text factory)) destroyLogicalType \logical -> do+ expected <- nativeTypeText logical+ rep <- logicalTypeToRep factory logical+ assertEqual (Text.unpack (Text.take 80 text)) expected (logicalTypeSQL rep)+ -- A read error names the column type. Under VARIANT it names the type of+ -- the stored value, in the same text.+ assertIncompatible conn "1.5::DECIMAL(18,3)" () "DECIMAL(18,3)" (Proxy @Text) "Text"+ assertIncompatible conn "(1.5::DECIMAL(18,3))::VARIANT" () "DECIMAL(18,3)" (Proxy @Text) "Text"+ assertIncompatible conn "(TIMESTAMPTZ '2024-01-02 03:04:05+00')::VARIANT" () "TIMESTAMP WITH TIME ZONE" (Proxy @Text) "Text"+ , testCase "logicalTypeSQL quotes an empty named STRUCT field" $ do+ let integer = LogicalTypeScalar LogicalTypeIdInteger+ rep = LogicalTypeStruct (listArray (0, 1) [StructField "" integer, StructField "normal" integer])+ logicalTypeSQL rep @?= "STRUCT(\"\" INTEGER, normal INTEGER)"+ , testCase "a user FromField checks the type of a column, an element, a MAP key and value, and a record field" $ withConnection ":memory:" \conn -> do+ query_ conn "SELECT 1.25::DECIMAL(9,2), TIMESTAMP_NS '2024-01-02 03:04:05.123456789', [1, 2, 3]::INTEGER[3], '{\"a\": 1}'::JSON"+ >>= (@?= [(Checked "decimal", Checked "timestamp_ns", Checked "array", Checked "json")])+ forM_ ["1.25::DECIMAL(9,3)", "TIMESTAMP '2024-01-02 03:04:05'", "[1, 2]::INTEGER[2]", "'{}'::VARCHAR"] \value ->+ try (query_ conn (Query ("SELECT " <> value)) :: IO [Only Checked]) >>= \case+ Left SQLError{sqlErrorMessage} -> assertBool (Text.unpack sqlErrorMessage) ("expected DECIMAL(_,2)" `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure (Text.unpack value <> " reads as " <> show rows)+ query_ conn "SELECT [1.25::DECIMAL(9,2)], MAP {TIMESTAMP_NS '2024-01-02': '{}'::JSON}, {'price': 2.5::DECIMAL(9,2)}"+ >>= (@?= [([Checked "decimal"], Map.fromList [(Checked "timestamp_ns", Checked "json")], Priced (Checked "decimal"))])+ , testCase "an element and a MAP value have the child type, and a value in a VARIANT has the type VARIANT" $ withConnection ":memory:" \conn -> do+ let decimal = LogicalTypeDecimal 4 1+ json = LogicalTypeAlias "JSON" (LogicalTypeScalar LogicalTypeIdVarchar)+ variant = LogicalTypeScalar LogicalTypeIdVariant+ query_ conn "SELECT [1.5::DECIMAL(4,1)], [1.5::DECIMAL(4,1)], MAP {'k': '{}'::JSON}, [1, 2]::VARIANT"+ >>= ( @?=+ [+ ( FieldTypeOf (LogicalTypeList decimal)+ , [FieldTypeOf decimal]+ , Map.fromList [("k" :: Text, FieldTypeOf json)]+ , [FieldTypeOf variant, FieldTypeOf variant]+ )+ ]+ )+ , testCase "a column whose type no read forces converts no ENUM dictionary" $ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TYPE wide AS ENUM (SELECT 'label' || i FROM range(100000) t(i))"+ let readLabel = query_ conn "SELECT 'label7'::wide" :: IO [Only Text]+ allocation action = do+ before <- getAllocationCounter+ result <- action+ after <- getAllocationCounter+ -- The counter counts down.+ pure (result, before - after)+ _ <- readLabel+ (labels, withoutType) <- allocation readLabel+ labels @?= [Only "label7"]+ -- A failed read whose error no one renders converts no type.+ (alternative, withoutRender) <- allocation (query_ conn "SELECT 'label7'::wide" :: IO [Only IntOrText])+ alternative @?= [Only (IntOrText (Right "label7"))]+ assertBool ("the read with a failed alternative allocated " <> show withoutRender <> " bytes") (withoutRender < 1000000)+ [(size, conversion)] <-+ queryWith_+ (fieldWith (\entry -> liftConversion (allocation (runConversion (enumSize <$> fieldType entry)))))+ conn+ "SELECT 'label7'::wide"+ size @?= Ok 100000+ -- Each of the 100,000 labels is a Text of at least 32 bytes.+ assertBool ("the read without the type allocated " <> show withoutType <> " bytes") (withoutType < 1000000)+ assertBool ("the conversion allocated " <> show conversion <> " bytes") (conversion > 3200000)+ , testCase "a pure column type snapshot survives all database scopes" $ do+ snapshot <- withEnumType runConversion+ performMajorGC+ snapshot @?= Ok (LogicalTypeEnum (listArray (0, 99999) ["label" <> Text.pack (show index) | index <- [0 .. 99999 :: Int]]))+ , testCase "unused and cached native type actions reject after the query returns" $+ forM_ [False, True] \cached -> do+ logical <- withEnumType \action -> do+ when cached (runConversion (enumSize <$> action) >>= (@?= Ok 100000))+ pure action+ performMajorGC+ assertClosedMetadata logical+ , testCase "a parser can read metadata around a nested query on its connection" $ withConnection ":memory:" \conn -> do+ rows <-+ queryWith_+ ( fieldWith \entry -> do+ before <- fieldType entry+ liftConversion (query_ conn "SELECT 7::INTEGER" :: IO [Only Int32]) >>= liftConversion . (@?= [Only 7])+ after <- fieldType entry+ pure (before, after)+ )+ conn+ "SELECT 1.5::DECIMAL(4,1)"+ rows @?= [(LogicalTypeDecimal 4 1, LogicalTypeDecimal 4 1)]+ , testCase "a cursor keeps metadata until exhaustion reset or statement close" $ withEnumConnection \conn ->+ forM_ [False, True] \cached ->+ forM_ (["exhaustion", "reset", "close"] :: [String]) \end ->+ withStatement conn "SELECT 'label7'::scoped_wide FROM range(2)" \stmt -> do+ Just logical <- nextRowWith (fieldWith (pure . fieldType)) stmt+ when cached (runConversion (enumSize <$> logical) >>= (@?= Ok 100000))+ case end of+ "exhaustion" -> do+ Just (Only label) <- nextRow stmt+ (label :: Text) @?= "label7"+ -- The first field's metadata is still valid after another row.+ when cached (runConversion (enumSize <$> logical) >>= (@?= Ok 100000))+ nextRow stmt >>= (@?= (Nothing :: Maybe (Only Text)))+ "reset" -> clearStatementBindings stmt+ _ -> closeStatement stmt+ assertClosedMetadata logical+ , testCase "an ENUM type retries cancellation while its query scope remains open" $+ forM_ [1 .. 3 :: Int] \_ -> withEnumType \logical -> do+ started <- newEmptyMVar+ finished <- newEmptyMVar+ worker <- forkIO $ mask \restore -> do+ outcome <- try $ restore $ do+ putMVar started ()+ runConversion (enumSize <$> logical)+ putMVar finished (outcome :: Either SomeException (Ok Int))+ awaitMetadata "metadata worker did not start" (takeMVar started)+ -- Completion can win this cancellation race.+ threadDelay 1000+ awaitMetadata "metadata worker did not finish" (killThread worker >> takeMVar finished) >>= \case+ Left failure -> fromException failure @?= Just ThreadKilled+ Right result -> result @?= Ok 100000+ runConversion (enumSize <$> logical) >>= (@?= Ok 100000)+ , testCase "conversion steps force successful values inside IO" $ do+ let marker = "conversion result failure"+ bad = error marker :: Int+ expectFailure conversion = do+ outcome <- try (runConversion conversion)+ -- Match the outcome after try has left the IO exception boundary.+ case outcome of+ Left (failure :: ErrorCall) -> assertBool (displayException failure) (Text.pack marker `Text.isInfixOf` Text.pack (displayException failure))+ Right _ -> assertFailure "the result exception escaped the conversion boundary"+ expectFailure (pure bad)+ expectFailure ((,) <$> pure (1 :: Int) <*> pure bad)+ expectFailure ((,) <$> pure (1 :: Int) <*> fmap (const bad) (pure ()))+ expectFailure ((,) <$> pure (1 :: Int) <*> liftConversion (pure bad))+ expectFailure (pure bad >> pure ())+ , testCase "a conversion propagates cancellation and retries its explicit IO" $ withEnumType \logical -> do+ attempts <- newIORef (0 :: Int)+ let conversion = do+ attempt <- liftConversion (atomicModifyIORef' attempts (\count -> (count + 1, count)))+ if attempt == 0+ then liftConversion (throwIO ThreadKilled)+ else enumSize <$> logical+ try (runConversion conversion) >>= \case+ Left failure -> (failure :: AsyncException) @?= ThreadKilled+ Right result -> assertFailure ("expected cancellation, got " <> show result)+ runConversion conversion >>= (@?= Ok 100000)+ readIORef attempts >>= (@?= 2)+ , testCase "concurrent demands share an ENUM snapshot within its query scope" $ withEnumType \logical -> do+ start <- newEmptyMVar+ completed <- replicateM 8 newEmptyMVar+ forM_ completed \done -> void $ forkIO $ mask \restore -> do+ outcome <- try $ restore $ readMVar start >> runConversion logical+ putMVar done (outcome :: Either SomeException (Ok LogicalTypeRep))+ putMVar start ()+ outcomes <- awaitMetadata "concurrent metadata workers did not finish" (mapM takeMVar completed)+ forM_ outcomes \case+ Left failure -> throwIO failure+ Right result -> result @?= Ok (LogicalTypeEnum (listArray (0, 99999) ["label" <> Text.pack (show index) | index <- [0 .. 99999 :: Int]]))+ case outcomes of+ Right first : rest -> forM_ rest \case+ Right result -> result @?= first+ Left failure -> throwIO failure+ _ -> assertFailure "expected completed metadata demands"+ , testCase "concurrent type demand and cursor closure complete without deadlock" $ withEnumConnection \conn ->+ forM_ [1 .. 3 :: Int] \_ -> withStatement conn "SELECT 'label7'::scoped_wide" \stmt -> do+ Just logical <- nextRowWith (fieldWith (pure . fieldType)) stmt+ started <- newEmptyMVar+ readFinished <- newEmptyMVar+ closeFinished <- newEmptyMVar+ void $ forkIO $ mask \restore -> do+ outcome <- try $ restore (putMVar started () >> runConversion (enumSize <$> logical))+ putMVar readFinished (outcome :: Either SQLError (Ok Int))+ awaitMetadata "metadata reader did not start" (takeMVar started)+ void $ forkIO $ do+ outcome <- try (closeStatement stmt)+ putMVar closeFinished (outcome :: Either SomeException ())+ awaitMetadata "cursor closure blocked on metadata" (takeMVar closeFinished) >>= either throwIO pure+ awaitMetadata "metadata reader blocked on cursor closure" (takeMVar readFinished) >>= \case+ Left failure -> sqlErrorMessage failure @?= "duckdb-simple: field type scope is closed"+ Right result -> result @?= Ok 100000+ assertClosedMetadata logical+ , testCase "conversion actions preserve the caller's masking state" $ do+ let check expected = withEnumType \logical -> do+ runConversion (liftConversion getMaskingState) >>= (@?= Ok expected)+ runConversion (logical >> liftConversion getMaskingState) >>= (@?= Ok expected)+ getMaskingState >>= (@?= expected)+ initialState <- getMaskingState+ check initialState+ mask_ (check MaskedInterruptible)+ uninterruptibleMask_ (check MaskedUninterruptible)+ , testCase "a discarded diagnostic performs no metadata read or diagnostic IO" $ withEnumType \logical -> do+ diagnosticReads <- newIORef (0 :: Int)+ let failure :: Conversion Int+ failure = conversionErrorWith $ do+ liftConversion (atomicModifyIORef' diagnosticReads (\count -> (count + 1, ())))+ size <- enumSize <$> logical+ pure (userError ("ENUM dictionary size " <> show size))+ runConversion (failure <|> pure 42) >>= (@?= Ok 42)+ readIORef diagnosticReads >>= (@?= 0)+ runConversion failure >>= \case+ Errors [problem] -> assertBool (show problem) ("ENUM dictionary size 100000" `Text.isInfixOf` Text.pack (show problem))+ other -> assertFailure ("expected the final diagnostic, got " <> show other)+ readIORef diagnosticReads >>= (@?= 1)+ , testCase "logical type names reject embedded NUL" $ withConnection ":memory:" \conn -> withConnectionHandle conn \native -> withConnectionFactory native \factory -> do+ let logical = LogicalTypeStruct (listArray (0, 0) [StructField "before\0after" (LogicalTypeScalar LogicalTypeIdBigInt)])+ assertInputError (bracket (logicalTypeFromRep factory logical) destroyLogicalType (const (pure ())))+ , testCase "invalid native composite constructor returns controlled error" $ withConnection ":memory:" \conn -> do+ let struct =+ singleField+ (LogicalTypeMap (LogicalTypeScalar LogicalTypeIdBigInt) (LogicalTypeScalar LogicalTypeIdBigInt))+ (FieldMap [(FieldNull, FieldInt64 1)])+ result <- try (query conn "SELECT ?" (Only struct) :: IO [Only FieldValue])+ case result of+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorInputInvalid+ Right _ -> assertFailure "expected a V2 native constructor error"+ , testCase "BIT padding preserves SQL bit_count" $ withConnection ":memory:" \conn -> do+ let bits = bsFromBool [True, False, True]+ (query conn "SELECT bit_count(?), ?" (bits, bits) :: IO [(Int64, BitString)]) >>= (@?= [(2, bits)])+ , testCase "unsupported BIT input fails before native use" $ withConnection ":memory:" \conn ->+ forM_ [BitString 0 BS.empty, BitString 8 (BS.singleton 1)] \bits ->+ assertInputError (query conn "SELECT ?" (Only bits) :: IO [Only BitString])+ , testCase "TIMETZ offsets with seconds read without rounding" $ withConnection ":memory:" \conn ->+ (query_ conn "SELECT '12:00:00+01:23:45'::TIMETZ" :: IO [Only TimeWithZone]) >>= (@?= [Only (TimeWithZone (TimeOfDay 12 0 0) 5025)])+ , testCase "a failure of the reader of a chunk gives the error of DuckDB and releases each chunk" $ withConnection ":memory:" \conn -> do+ -- The VARIANT reader runs its metadata query on a helper connection, and+ -- that query needs an expression depth of 11. A GLOBAL depth of 10 thus+ -- fails the reader at the bind of that query, and the SESSION depth of+ -- the caller lets the query of the case run.+ _ <- execute_ conn "SET GLOBAL max_expression_depth = 10"+ _ <- execute_ conn "SET SESSION max_expression_depth = 1000"+ let rows = "SELECT 42::BIGINT::VARIANT FROM range(10000)"+ try (query_ conn rows :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertBool (Text.unpack sqlErrorMessage) ("expression depth" `Text.isInfixOf` sqlErrorMessage)+ sqlErrorType @?= Just ErrorQueryBinder+ Right result -> assertFailure ("expected the error of the VARIANT reader, got " <> show (length result) <> " rows")+ (query_ conn "SELECT 42" :: IO [Only Int]) >>= (@?= [Only 42])+ _ <- execute_ conn "SET GLOBAL max_expression_depth = 1000"+ (length <$> (query_ conn rows :: IO [Only FieldValue])) >>= (@?= 10000)+ , testCase "a failure of the reader of a later chunk of a stream gives the error of DuckDB" $+ withEnvironment \environment -> withInstance environment [] \instance' ->+ withConnectionOn instance' \conn -> withConnectionOn instance' \other -> do+ _ <- execute_ conn "SET SESSION max_expression_depth = 1000"+ -- Metadata reuses its prepared plan. The unusual key needs the+ -- extraction query, which binds again for each chunk. A lower+ -- GLOBAL depth fails that bind after the first chunk.+ let step count (Only (_ :: FieldValue)) = do+ when (count == 0) $ do+ _ <- execute_ other "SET GLOBAL max_expression_depth = 3"+ pure ()+ pure (count + 1)+ let rows = "SELECT json_object('', i)::VARIANT FROM range(10000) t(i)"+ try (Streaming.fold_ conn rows (0 :: Int) step) >>= \case+ Left SQLError{sqlErrorMessage} -> assertBool (Text.unpack sqlErrorMessage) ("expression depth" `Text.isInfixOf` sqlErrorMessage)+ Right count -> assertFailure ("expected the error of the VARIANT reader, got " <> show count <> " rows")+ (query_ conn "SELECT 42" :: IO [Only Int]) >>= (@?= [Only 42])+ _ <- execute_ conn "SET GLOBAL max_expression_depth = 1000"+ (length <$> (query_ conn rows :: IO [Only FieldValue])) >>= (@?= 10000)+ , testCase "ENUM labels read as the cast to VARCHAR in each chunk, execution, list, and map key" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ -- 5000 rows are three chunks. Each row has another label of a+ -- dictionary of 16-bit or 32-bit indexes. See Note [ENUM labels] in+ -- "Database.DuckDB.Simple.Materialize".+ forM_ [300, 70000 :: Int] \size -> do+ let dtype = "labels_" <> Text.pack (show size)+ count = Text.pack (show size)+ _ <- execute_ conn (Query ("CREATE TYPE " <> dtype <> " AS ENUM (SELECT 'v' || i FROM range(" <> count <> ") t(i))"))+ let sql =+ Query+ ( "SELECT e, [e, NULL], map([coalesce(e, 'v1'::"+ <> dtype+ <> ")], [i]), e::VARCHAR, [e, NULL]::VARCHAR[] FROM (SELECT i, CASE WHEN i % 1000 = 999 THEN NULL ELSE ('v' || ((i * 7) % "+ <> count+ <> "))::"+ <> dtype+ <> " END AS e FROM range(5000) t(i))"+ )+ check :: [(Maybe Text, [Maybe Text], Map.Map Text Int64, Maybe Text, [Maybe Text])] -> Assertion+ check rows = do+ length rows @?= 5000+ forM_ rows \(label, list, keys, cast, castList) -> do+ label @?= cast+ list @?= castList+ Map.keys keys @?= [fromMaybe "v1" cast]+ forM_ [1 :: Int, 2] \_ ->+ query_ conn sql >>= check+ withStatement conn sql (readAllRows Streaming.nextRow) >>= check+ , testCase "a prepared statement reads the labels of an ENUM type that DROP TYPE and CREATE TYPE replaced" $ withConnection ":memory:" \conn -> do+ _ <- execute_ conn "CREATE TYPE swapped AS ENUM ('a', 'b')"+ withStatement conn "SELECT ?::swapped, (?::swapped)::VARCHAR" \stmt -> do+ let run label = do+ bind stmt [toField (label :: Text), toField label]+ readAllRows nextRow stmt+ run "b" >>= (@?= [(FieldEnum 1 "b", "b" :: Text)])+ _ <- execute_ conn "DROP TYPE swapped"+ _ <- execute_ conn "CREATE TYPE swapped AS ENUM ('x', 'y', 'b')"+ -- Index 1 is 'y' in the new type. A cache of the first execution+ -- would give 'b'.+ run "y" >>= (@?= [(FieldEnum 1 "y", "y" :: Text)])+ , testCase "ENUM members of a UNION and the key and value of a MAP read the labels of their own dictionaries" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ _ <- execute_ conn "CREATE TYPE first_labels AS ENUM ('x', 'y'); CREATE TYPE second_labels AS ENUM ('p', 'q', 'r')"+ -- 3000 rows are two chunks.+ rows <-+ query_+ conn+ "SELECT CASE WHEN i % 2 = 0 THEN union_value(a := (['x', 'y'])[i % 4 // 2 + 1]::first_labels)::UNION(a first_labels, b second_labels) \+ \ELSE union_value(b := (['p', 'q', 'r'])[i % 3 + 1]::second_labels)::UNION(a first_labels, b second_labels) END, \+ \map([(['x', 'y'])[i % 2 + 1]::first_labels], [(['p', 'q', 'r'])[i % 3 + 1]::second_labels]), \+ \(['x', 'y'])[i % 4 // 2 + 1], (['p', 'q', 'r'])[i % 3 + 1], (['x', 'y'])[i % 2 + 1] \+ \FROM range(3000) t(i)"+ :: IO [(UnionValue FieldValue, Map.Map Text Text, Text, Text, Text)]+ length rows @?= 3000+ forM_ (zip [0 :: Int ..] rows) \(i, (union, pairs, first, second, key)) -> do+ unionValuePayload union+ @?= if even i+ then FieldEnum (if first == "x" then 0 else 1) first+ else FieldEnum (if second == "p" then 0 else if second == "q" then 1 else 2) second+ pairs @?= Map.singleton key second+ , testCase "ENUM payload label must be in its dictionary" $ withConnection ":memory:" \conn -> do+ try (query conn "SELECT ?" (Only (singleField (LogicalTypeEnum (listArray (0, 1) ["a", "b"])) (FieldEnum 0 "c"))) :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorType @?= Just ErrorTypeConversion+ sqlErrorMessage @?= "duckdb-simple: the ENUM label 'c' is not in the dictionary of the type"+ Right rows -> assertFailure ("expected the label error, got " <> show rows)+ -- DuckDB gives the same type for the same label in SQL.+ try (query_ conn "SELECT {'value': 'c'}::STRUCT(value ENUM('a', 'b'))" :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorTypeConversion+ Right rows -> assertFailure ("expected the cast error of DuckDB, got " <> show rows)+ -- A label with a single quote prints as it is, as in the cast error of DuckDB.+ try+ (query conn "SELECT ?" (Only (singleField (LogicalTypeEnum (listArray (0, 1) ["a", "b"])) (FieldEnum 0 "it's"))) :: IO [Only FieldValue])+ >>= \case+ Left SQLError{sqlErrorMessage} -> sqlErrorMessage @?= "duckdb-simple: the ENUM label 'it's' is not in the dictionary of the type"+ Right rows -> assertFailure ("expected the label error, got " <> show rows)+ try (query_ conn "SELECT {'value': 'it''s'}::STRUCT(value ENUM('a', 'b'))" :: IO [Only FieldValue]) >>= \case+ Left SQLError{sqlErrorMessage} -> Text.takeWhile (/= '\n') sqlErrorMessage @?= "Conversion Error: Could not convert string 'it's' to ENUM('a', 'b')"+ Right rows -> assertFailure ("expected the cast error of DuckDB, got " <> show rows)+ , testCase "ENUM payload binds its label, not its index" $ withConnection ":memory:" \conn -> do+ let struct = singleField (LogicalTypeEnum (listArray (0, 2) ["a", "b", "c"])) (FieldEnum 1 "b")+ (query conn "SELECT typeof((?).value), (?).value::VARCHAR, ?" (struct, struct, struct) :: IO [(Text, Text, StructValue FieldValue)])+ >>= (@?= [("ENUM('a', 'b', 'c')", "b", struct)])+ -- The index of another dictionary does not change the value.+ let other = singleField (LogicalTypeEnum (listArray (0, 2) ["a", "b", "c"])) (FieldEnum 0 "c")+ (query conn "SELECT (?).value::VARCHAR, ?" (other, other) :: IO [(Text, StructValue FieldValue)])+ >>= (@?= [("c", singleField (LogicalTypeEnum (listArray (0, 2) ["a", "b", "c"])) (FieldEnum 2 "c"))])+ , testCase "ENUM values of each dictionary size bind their labels" $ withConnection ":memory:" \conn ->+ -- DuckDB stores the index of a dictionary of 200, 300, and 70000 entries+ -- in 8, 16, and 32 bits.+ forM_ [200, 300, 70000 :: Int] \size -> do+ let label index = "v" <> Text.pack (show index)+ entries = listArray (0, size - 1) (map label [0 .. size - 1])+ forM_ [0, size - 1] \index -> do+ let struct = singleField (LogicalTypeEnum entries) (FieldEnum (fromIntegral index) (label index))+ (query conn "SELECT (?).value::VARCHAR, ?" (struct, struct) :: IO [(Text, StructValue FieldValue)])+ >>= assertEqual (show size <> " " <> show index) [(label index, struct)]+ , testCase "TIMETZ input offset overflow fails before native use" $ withConnection ":memory:" \conn -> do+ let value = TimeWithZone (TimeOfDay 12 0 0) maxBound+ struct = singleField (LogicalTypeScalar LogicalTypeIdTimeTz) (FieldTimeTZ value)+ assertRangeError (query conn "SELECT ?" (Only struct) :: IO [Only FieldValue])+ forM_ [71582789, -71582789, minBound, maxBound] \minutes ->+ assertRangeError+ (query conn "SELECT ?::TIMETZ" (Only (timeWithZone (TimeOfDay 12 0 0) (minutesToTimeZone minutes))) :: IO [Only TimeWithZone])+ , testCase "invalid TIME input returns controlled error" $ withConnection ":memory:" \conn ->+ assertRangeError (query conn "SELECT ?" (Only (TimeOfDay 1000000000 0 0)) :: IO [Only TimeOfDay])+ , testCase "nested types decode the same in each chunk and each read mode" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ _ <- execute_ conn "CREATE TYPE mood AS ENUM ('calm', 'busy', 'tired')"+ let sql =+ "SELECT st, [st, st] AS l, [st, st]::STRUCT(n BIGINT, e mood)[2] AS a, tiers, [tiers, tiers] AS tl, \+ \CASE WHEN i % 2 = 0 THEN union_value(d := ((i % 1000) / 10)::DECIMAL(9, 2))::UNION(d DECIMAL(9, 2), s STRUCT(n BIGINT, e mood)) \+ \ELSE union_value(s := st)::UNION(d DECIMAL(9, 2), s STRUCT(n BIGINT, e mood)) END AS u, \+ \map([st.e], [((i % 1000) / 10)::DECIMAL(9, 2)]) AS m, i::DECIMAL(18, 3) AS d \+ \FROM (SELECT i, {'n': i, 'e': (['calm', 'busy', 'tired'])[i % 3 + 1]::mood} AS st, \+ \{'small': ((i % 1000) / 10)::DECIMAL(4, 1), 'large': (i * 1000 + 7)::DECIMAL(18, 3)} AS tiers FROM range(5000) r(i))"+ mood = LogicalTypeEnum (listArray (0, 2) ["calm", "busy", "tired"])+ bigint = LogicalTypeScalar LogicalTypeIdBigInt+ stType = LogicalTypeStruct (listArray (0, 1) [StructField "n" bigint, StructField "e" mood])+ members = listArray (0, 1) [UnionMemberType "d" (LogicalTypeDecimal 9 2), UnionMemberType "s" stType]+ decimal width scale integer = FieldDecimal (DecimalValue width scale integer)+ moodValue i = FieldEnum (fromIntegral (i `mod` 3)) (["calm", "busy", "tired"] !! fromIntegral (i `mod` 3))+ expected :: Int64 -> [FieldValue]+ expected i =+ let st = FieldStruct (structOf [("n", FieldInt64 i, bigint), ("e", moodValue i, mood)])+ tenths = toInteger (i `mod` 1000)+ tiers =+ FieldStruct+ ( structOf+ [ ("small", decimal 4 1 tenths, LogicalTypeDecimal 4 1)+ , ("large", decimal 18 3 ((toInteger i * 1000 + 7) * 1000), LogicalTypeDecimal 18 3)+ ]+ )+ union+ | even i = UnionValue 0 "d" (decimal 9 2 (tenths * 10)) members+ | otherwise = UnionValue 1 "s" st members+ in [ st+ , FieldList [st, st]+ , FieldArray (listArray (0, 1) [st, st])+ , tiers+ , FieldList [tiers, tiers]+ , FieldUnion union+ , FieldMap [(moodValue i, decimal 9 2 (tenths * 10))]+ , decimal 18 3 (toInteger i * 1000)+ ]+ materialized <- query_ conn sql :: IO [[FieldValue]]+ cursor <- withStatement conn sql (readAllRows nextRow)+ streamed <- withStatement conn sql (readAllRows Streaming.nextRow)+ length materialized @?= 5000+ cursor @?= materialized+ streamed @?= materialized+ forM_ [0, 2047, 2048, 4095, 4999 :: Int] \i -> (materialized !! i) @?= expected (fromIntegral i)+ , testCase "a statement that runs again with another type reads the new type" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ withStatement conn "SELECT ? AS s FROM range(3000) r(i)" \stmt ->+ forM_ ([nextRow, Streaming.nextRow] :: [Statement -> IO (Maybe (Only (StructValue FieldValue)))]) \next ->+ forM_ [DecimalValue 4 1 123, DecimalValue 18 3 123456789012, DecimalValue 4 1 (-45)] \value -> do+ let struct = singleField (LogicalTypeDecimal (decimalWidth value) (decimalScale value)) (FieldDecimal value)+ bind stmt [toField struct]+ readAllRows next stmt >>= (@?= replicate 3000 (Only struct))+ , testCase "an Integer outside the HUGEINT range raises the range error for each integer type" $ withConnection ":memory:" \conn -> do+ -- A BIGNUM value for these types stores a wrong value with no error.+ -- See Note [Integer parameters] in "Database.DuckDB.Simple.ToField".+ let huge = 2 ^ (128 :: Int) + 5 :: Integer+ insert :: Text -> Integer -> IO (Either SQLError Int)+ insert dtype value = do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE wide_integer (v " <> dtype <> ")"))+ result <- try (execute conn "INSERT INTO wide_integer VALUES (?)" (Only value))+ (query_ conn "SELECT count(*) FROM wide_integer" :: IO [Only Int64]) >>= assertEqual (Text.unpack dtype <> " " <> show value) [Only 0]+ pure result+ forM_ ["TINYINT", "SMALLINT", "INTEGER", "BIGINT", "HUGEINT", "UTINYINT", "USMALLINT", "UINTEGER", "UBIGINT", "UHUGEINT"] \dtype -> do+ -- DuckDB also rejects the value for the type.+ try (query_ conn (Query ("SELECT '" <> Text.pack (show huge) <> "'::" <> dtype)) :: IO [Only Text]) >>= \case+ Left SQLError{sqlErrorMessage} ->+ assertBool (Text.unpack sqlErrorMessage) ("Conversion Error: Could not convert string" `Text.isPrefixOf` sqlErrorMessage)+ Right rows -> assertFailure ("DuckDB accepts " <> show huge <> " as " <> Text.unpack dtype <> ": " <> show rows)+ forM_ [huge, negate huge] \value ->+ insert dtype value >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorType @?= Just ErrorTypeConversion+ sqlErrorMessage @?= "duckdb-simple: the value " <> Text.pack (show value) <> " is out of range for a parameter of the type " <> dtype+ Right _ -> assertFailure ("expected the duckdb-simple range error for " <> Text.unpack dtype <> " " <> show value)+ -- An element of an array takes the element type of its target. DuckDB+ -- then checks the range of each element.+ forM_ ["UTINYINT[2]", "USMALLINT[2]", "UINTEGER[2]", "UBIGINT[2]"] \dtype -> do+ _ <- execute_ conn (Query ("CREATE OR REPLACE TABLE wide_integer (v " <> dtype <> ")"))+ result <- try (execute conn "INSERT INTO wide_integer VALUES (?)" (Only (sized [5, -6] :: SizedArray 2 Integer)))+ (query_ conn "SELECT count(*) FROM wide_integer" :: IO [Only Int64]) >>= assertEqual (Text.unpack dtype) [Only 0]+ case result of+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertBool+ (Text.unpack sqlErrorMessage)+ ("-6 can't be cast because the value is out of range for the destination type" `Text.isInfixOf` sqlErrorMessage)+ sqlErrorType @?= Just ErrorInputInvalid+ Right _ -> assertFailure ("expected the DuckDB range error for " <> Text.unpack dtype)+ , testCase "DECIMAL columns decode the largest value at each storage boundary" $ withConnection ":memory:" \conn ->+ -- DuckDB stores a DECIMAL of the width 1 to 4 in 16 bits, 5 to 9 in 32 bits,+ -- 10 to 18 in 64 bits, and 19 to 38 in 128 bits. The text comes from DuckDB.+ forM_ [4, 5, 9, 10, 18, 19, 38] \width -> do+ let largest = 10 ^ width - 1 :: Integer+ dtype = "DECIMAL(" <> Text.pack (show width) <> ",0)"+ sql =+ "SELECT x::"+ <> dtype+ <> ", x::"+ <> dtype+ <> "::VARCHAR FROM (VALUES ("+ <> Text.pack (show (negate largest))+ <> "), (NULL), ("+ <> Text.pack (show largest)+ <> ")) t(x)"+ rows <- query_ conn (Query sql) :: IO [(FieldValue, Maybe Text)]+ assertEqual+ (Text.unpack dtype)+ [ (FieldDecimal (DecimalValue width 0 (negate largest)), Just (Text.pack (show (negate largest))))+ , (FieldNull, Nothing)+ , (FieldDecimal (DecimalValue width 0 largest), Just (Text.pack (show largest)))+ ]+ rows+ , testCase "ENUM columns decode the last label at each storage boundary" $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ -- DuckDB stores the index of an ENUM with at most 255 labels in 8 bits, with+ -- at most 65535 labels in 16 bits, and with more labels in 32 bits.+ forM_ [255, 256, 65535, 65536 :: Int] \size -> do+ let dtype = "enum_boundary_" <> Text.pack (show size)+ lastLabel = "v" <> Text.pack (show (size - 1))+ _ <- execute_ conn (Query ("CREATE TYPE " <> dtype <> " AS ENUM (SELECT 'v' || i FROM range(" <> Text.pack (show size) <> ") t(i))"))+ rows <-+ query_+ conn+ ( Query+ ( "SELECT x::"+ <> dtype+ <> ", enum_code(x::"+ <> dtype+ <> ")::BIGINT, x::"+ <> dtype+ <> "::VARCHAR FROM (VALUES ('v0'), (NULL), ('"+ <> lastLabel+ <> "')) t(x)"+ )+ )+ :: IO [(FieldValue, Maybe Int64, Maybe Text)]+ -- DuckDB gives the index and the label of each row.+ assertEqual+ (Text.unpack dtype)+ [(Just 0, Just "v0"), (Nothing, Nothing), (Just (fromIntegral (size - 1)), Just lastLabel)]+ [(index, label) | (_, index, label) <- rows]+ forM_ rows \(value, index, label) -> assertEqual (Text.unpack dtype) (enumField index label) value+ , testCase "NULL TUPLE and UNION rows decode as NULL in each chunk" $ withConnection ":memory:" \conn -> do+ rows <-+ query_+ conn+ "SELECT t, u, t IS NULL, u IS NULL FROM (SELECT CASE WHEN i % 7 = 1 THEN NULL ELSE row(i, 'x') END AS t, CASE WHEN i % 7 = 2 THEN NULL ELSE union_value(n := i)::UNION(n BIGINT, s VARCHAR) END AS u FROM range(3000) r(i))"+ :: IO [(FieldValue, FieldValue, Bool, Bool)]+ length rows @?= 3000+ -- DuckDB gives the NULL rows in the last two columns.+ forM_ (zip [0 :: Int ..] rows) \(row, (tuple, member, tupleIsNull, memberIsNull)) -> do+ assertEqual ("TUPLE row " <> show row) tupleIsNull (tuple == FieldNull)+ assertEqual ("UNION row " <> show row) memberIsNull (member == FieldNull)+ , testCase "narrowing reads reject the values that DuckDB rejects for the type" $ withConnection ":memory:" \conn -> do+ let rejects :: forall a. (FromField a, Show a) => Proxy a -> Text -> Text -> Assertion+ rejects _ value sqlType = do+ -- DuckDB raises an error for the same cast.+ try (query_ conn (Query ("SELECT (" <> value <> ")::" <> sqlType)) :: IO [Only Text]) >>= \case+ Left SQLError{sqlErrorMessage} ->+ assertBool+ (Text.unpack sqlErrorMessage)+ ("can't be cast because the value is out of range for the destination type" `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure ("DuckDB accepts " <> Text.unpack value <> " as " <> Text.unpack sqlType <> ": " <> show rows)+ -- The read raises the conversion error of the query.+ let sql = Query ("SELECT " <> value)+ try (query_ conn sql :: IO [Only a]) >>= \case+ Left SQLError{sqlErrorType, sqlErrorQuery} -> do+ assertEqual (Text.unpack value) Nothing sqlErrorType+ assertEqual (Text.unpack value) (Just sql) sqlErrorQuery+ Right rows -> assertFailure (Text.unpack value <> " decodes as " <> show rows)+ -- The cause of that error is a ResultError.+ [Only native] <- query_ conn sql :: IO [Only FieldValue]+ runConversion+ ( fromField Field{fieldName = "value", fieldIndex = 0, fieldValue = native, fieldType = pure (LogicalTypeScalar LogicalTypeIdVariant)}+ :: Conversion a+ )+ >>= \case+ Errors errors ->+ assertBool ("expected a ResultError, got " <> show errors) (not (null errors) && all isResultError errors)+ Ok decoded -> assertFailure (Text.unpack value <> " converts to " <> show decoded)+ isResultError failure = isJust (fromException failure :: Maybe ResultError)+ rejects (Proxy @Word64) "-1::BIGINT" "UBIGINT"+ rejects (Proxy @Word) "-1::BIGINT" "UBIGINT"+ rejects (Proxy @Word32) "-1::INTEGER" "UINTEGER"+ rejects (Proxy @Int16) "32768::INTEGER" "SMALLINT"+ rejects (Proxy @Int32) "2147483648::BIGINT" "INTEGER"+ rejects (Proxy @Int64) "1180591620717411303424::HUGEINT" "BIGINT"+ ]+ <> [ testCase ("composite temporal round trip: " <> dtype) $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn ->+ forM_ ("NULL" : map (\value -> "'" <> value <> "'") values) \value ->+ assertTemporalRoundTrip conn dtype (value <> "::" <> dtype)+ | (dtype, values) <-+ [ (dtype, ["-infinity", "infinity", "1969-12-31 23:59:59.123456789", "2000-01-01 12:34:56.123456789"])+ | dtype <- ["TIMESTAMP", "TIMESTAMP_S", "TIMESTAMP_MS", "TIMESTAMP_NS", "TIMESTAMPTZ"]+ ]+ <> [("DATE", ["-infinity", "infinity", "1969-12-31", "2000-01-01"])]+ <> [("TIME_NS", ["00:00:00", "12:34:56.123456789", "23:59:59.999999999", "24:00:00"])]+ ]+ <> [ testCase (dtype <> " " <> value <> " rejects a finite result type") $ withConnection ":memory:" \conn ->+ assertConversionError (query_ conn (fromString ("SELECT '" <> value <> "'::" <> dtype)) :: IO [Only LocalTime])+ | dtype <- ["DATE", "TIMESTAMP", "TIMESTAMP_S", "TIMESTAMP_MS", "TIMESTAMP_NS", "TIMESTAMPTZ"]+ , value <- ["infinity", "-infinity"]+ ]++-- | Rebind native temporal values in structs, unions, and nested collections.+assertTemporalRoundTrip :: Connection -> String -> String -> Assertion+assertTemporalRoundTrip conn dtype expression = do+ let sql =+ "WITH temporal AS (SELECT "+ <> expression+ <> " AS value) "+ <> "SELECT {'scalar': value, 'list': [value, NULL], 'array': [value, NULL]::"+ <> dtype+ <> "[2], 'map': map([1], [value])}, union_value(value := value) FROM temporal"+ [original] <- query_ conn (fromString sql) :: IO [(StructValue FieldValue, UnionValue FieldValue)]+ query conn "SELECT ?, ?" original >>= (@?= [original])++-- | Store raw V2 timestamp units through the parameter API.+appendTimestamp :: Connection -> String -> NativeTimestamp -> IO ()+appendTimestamp conn dtype value = do+ _ <- execute_ conn (fromString ("CREATE TABLE native_timestamp(value " <> dtype <> ")"))+ _ <- execute conn "INSERT INTO native_timestamp VALUES (?)" (Only value)+ pure ()++-- | Build a struct from the name, the value, and the type of each field.+structOf :: [(Text, FieldValue, LogicalTypeRep)] -> StructValue FieldValue+structOf fields =+ StructValue+ (listArray (0, length fields - 1) [StructField name value | (name, value, _) <- fields])+ (listArray (0, length fields - 1) [StructField name logical | (name, _, logical) <- fields])+ (Map.fromList (zip [name | (name, _, _) <- fields] [0 ..]))++-- | Read the rows of a statement until the cursor gives no row.+readAllRows :: (Statement -> IO (Maybe row)) -> Statement -> IO [row]+readAllRows next stmt = go []+ where+ go rows = next stmt >>= maybe (pure (reverse rows)) (\row -> go (row : rows))++-- | Build a single-field struct for nested binding tests.+singleField :: LogicalTypeRep -> FieldValue -> StructValue FieldValue+singleField logical value =+ StructValue+ (listArray (0, 0) [StructField "value" value])+ (listArray (0, 0) [StructField "value" logical])+ (Map.singleton "value" 0)++-- | The field of an ENUM row with the index and the label that DuckDB gives.+enumField :: Maybe Int64 -> Maybe Text -> FieldValue+enumField (Just index) (Just label) = FieldEnum (fromIntegral index) label+enumField _ _ = FieldNull++-- | Require the range error of the library for a date or a timestamp outside+-- the range of DuckDB. See Note [Time ranges] in "Database.DuckDB.Simple.ToField".+assertOutsideTimeRange :: IO a -> Assertion+assertOutsideTimeRange action =+ try action >>= \case+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ sqlErrorType @?= Just ErrorTypeConversion+ assertBool (Text.unpack sqlErrorMessage) ("of DuckDB" `Text.isInfixOf` sqlErrorMessage)+ Right _ -> assertFailure "expected the range error of the library"++-- | Assert a controlled FromField conversion error.+assertConversionError :: IO a -> Assertion+assertConversionError action = do+ result <- try action+ case result of+ Left (_ :: SQLError) -> pure ()+ Right _ -> assertFailure "expected ResultError"++-- | The DECIMAL widths and scales of the case on Double and Float reads. They+-- have each storage type of a DECIMAL, each path of+-- Note [DECIMAL as DOUBLE and FLOAT] in "Database.DuckDB.Simple.FromField",+-- and the scales on both sides of the limits 10 and 22 of path 2.+decimalKinds :: [(Int, Int)]+decimalKinds =+ [(4, 0), (4, 2), (4, 4), (9, 0), (9, 2), (9, 5), (9, 9), (18, 0), (18, 2), (18, 6), (18, 12), (18, 18)]+ <> [(19, 0), (19, 4), (38, 0), (38, 2), (38, 5), (38, 10), (38, 11), (38, 20), (38, 22), (38, 23), (38, 30), (38, 38)]++-- | A fixed sequence of 64-bit words: SplitMix64 with seed 1.+splitMix64 :: [Word64]+splitMix64 = map mix (drop 1 (iterate (+ 0x9e3779b97f4a7c15) 1))+ where+ mix z0 =+ let z1 = (z0 `xor` (z0 `shiftR` 30)) * 0xbf58476d1ce4e5b9+ z2 = (z1 `xor` (z1 `shiftR` 27)) * 0x94d049bb133111eb+ in z2 `xor` (z2 `shiftR` 31)++-- | Unscaled integers that a DECIMAL of a width holds, with both signs: zero+-- and the extremes; two values that DuckDB 2.0.0 does not cast to the nearest+-- DOUBLE at scale 0; a value that two roundings change at scale 2; values next+-- to each power of two and next to the DOUBLE and FLOAT midpoints above it; and+-- 400 values of 'splitMix64' with 1 to width digits.+decimalIntegers :: Int -> [Integer]+decimalIntegers width = filter (\v -> abs v <= limit) (special <> near <> spread)+ where+ limit = 10 ^ width - 1+ special = [0, limit, negate limit, -9007199254740994, 270628455292663770, 3356951675583485076575379654993209529]+ near =+ [ sign * (2 ^ k + base + delta)+ | k <- [0 .. 127 :: Int]+ , base <- 0 : [m * 2 ^ (k - 53) | k >= 53, m <- [1, 3]] <> [m * 2 ^ (k - 24) | k >= 24, m <- [1, 3]]+ , delta <- [-1, 0, 1]+ , sign <- [1, -1]+ ]+ spread =+ [ (if odd n then negate else id) ((toInteger high * 2 ^ (64 :: Int) + toInteger low) `mod` (10 ^ (1 + n `mod` width)))+ | (n, high, low) <- zip3 [0 .. 399 :: Int] (everyOther splitMix64) (everyOther (drop 1 splitMix64))+ ]+ everyOther (a : _ : rest) = a : everyOther rest+ everyOther rest = rest++-- | Unscaled integers of path 3 in Note [DECIMAL as DOUBLE and FLOAT] whose+-- value is a midpoint of two DOUBLE values (scale 5) or of two FLOAT values+-- (scale 11), and the integers next to them, with both signs. A midpoint rounds+-- to the even value.+roundingTies :: Int -> Int -> [Integer]+roundingTies 38 5 = [sign * ((2 ^ (100 :: Int) + 2 ^ (47 :: Int)) * 10 ^ (5 :: Int) + delta) | delta <- [-1, 0, 1], sign <- [1, -1]]+roundingTies 38 11 = [sign * ((2 ^ (60 :: Int) + 2 ^ (36 :: Int)) * 10 ^ (11 :: Int) + delta) | delta <- [-1, 0, 1], sign <- [1, -1]]+roundingTies _ _ = []++-- | The text of an unscaled DECIMAL integer with a scale.+decimalText :: Int -> Integer -> Text+decimalText 0 v = Text.pack (show v)+decimalText scale v = Text.pack ((if v < 0 then "-" else "") <> whole <> "." <> fraction)+ where+ digits = show (abs v)+ padded = replicate (scale + 1 - length digits) '0' <> digits+ (whole, fraction) = splitAt (length padded - scale) padded++-- | Assert that each text value, cast to the source type, reads as @a@ with the+-- value of its cast from the source type to the target type. The target type is+-- the SQL type of @a@. The two reads are in the same row.+assertReadsAsCast :: (FromField a, Eq a, Show a) => Connection -> Proxy a -> Text -> Text -> [Text] -> Assertion+assertReadsAsCast conn proxy source target values = do+ rows <-+ query conn (Query ("SELECT v, v::" <> source <> ", v::" <> source <> "::" <> target <> " FROM unnest(?::VARCHAR[]) t(v)")) (Only values)+ length rows @?= length values+ forM_ rows \(value, actual, cast) ->+ assertEqual (Text.unpack ("'" <> value <> "'::" <> source <> " as " <> target)) (cast `asProxyTypeOf` proxy) actual++-- | Assert that a read of an expression as @a@ raises Incompatible with the SQL+-- type name and the Haskell type name. A String reads through Text, so its+-- error names Text.+assertIncompatible :: forall a q. (FromField a, Show a, ToRow q) => Connection -> Text -> q -> Text -> Proxy a -> Text -> Assertion+assertIncompatible conn expression parameters sqlName _ haskellName =+ try (query conn (Query ("SELECT " <> expression <> " AS c")) parameters :: IO [Only a]) >>= \case+ Left SQLError{sqlErrorMessage} ->+ sqlErrorMessage @?= "duckdb-simple: column c has type " <> sqlName <> " but expected " <> haskellName+ Right rows -> assertFailure ("expected Incompatible for " <> Text.unpack expression <> ", got " <> show rows)++-- | Assert rejection without forcing a partial generic value.+assertConversionFails :: Conversion a -> Assertion+assertConversionFails conversion =+ runConversion conversion >>= \case+ Errors _ -> pure ()+ Ok _ -> assertFailure "expected a conversion failure"++-- | Run a parser action while the ENUM column's metadata scope is open.+withEnumType :: (Conversion LogicalTypeRep -> IO a) -> IO a+withEnumType action = withEnumConnection \connection -> do+ [result] <- queryWith_ (fieldWith (liftConversion . action . fieldType)) connection "SELECT 'label7'::scoped_wide"+ pure result++-- | Create an ENUM in an explicitly attached in-memory database.+withEnumConnection :: (Connection -> IO a) -> IO a+withEnumConnection action = withEnvironment \environment -> withInstance environment [] \instance_ ->+ withConnectionOn instance_ \connection -> do+ _ <- execute_ connection "ATTACH ':memory:' AS metadata"+ _ <- execute_ connection "USE metadata"+ _ <- execute_ connection "CREATE TYPE scoped_wide AS ENUM (SELECT 'label' || i FROM range(100000) t(i) ORDER BY i)"+ action connection++-- | Reject native metadata reads after their query or callback scope closes.+assertClosedMetadata :: Conversion a -> Assertion+assertClosedMetadata conversion =+ try (runConversion conversion) >>= \case+ Left SQLError{sqlErrorMessage} -> sqlErrorMessage @?= "duckdb-simple: field type scope is closed"+ Right _ -> assertFailure "metadata read succeeded after its scope closed"++-- | Read the size of an ENUM dictionary.+enumSize :: LogicalTypeRep -> Int+enumSize (LogicalTypeEnum entries) = length (elems entries)+enumSize other = error ("expected an ENUM type, got " <> take 100 (show other))++-- | Bound waits for metadata workers, including a worker that fails.+awaitMetadata :: String -> IO a -> IO a+awaitMetadata message action =+ timeout 60000000 action >>= \case+ Just result -> pure result+ Nothing -> assertFailure message >> fail message
test/VariantRegressionTests.hs view
@@ -1,282 +1,645 @@ {-# LANGUAGE BlockArguments #-}+{-# LANGUAGE DataKinds #-} {-# LANGUAGE DeriveGeneric #-} {-# LANGUAGE DerivingVia #-}+{-# LANGUAGE LambdaCase #-}+{-# LANGUAGE NamedFieldPuns #-} {-# LANGUAGE OverloadedStrings #-}-{-# LANGUAGE PatternSynonyms #-}-{-# OPTIONS_GHC -Wno-deprecations #-}+{-# LANGUAGE ScopedTypeVariables #-} +-- |+-- Module : VariantRegressionTests+-- Description : Lossless VARIANT binding and result conversion through the V2 API. module VariantRegressionTests (tests) where -import Control.Exception (SomeException, bracket, displayException, try)-import Control.Monad (forM_, void, when)-import Data.Array (Array, listArray)+import Control.Concurrent (forkFinally, killThread, myThreadId, newEmptyMVar, putMVar, readMVar, takeMVar, throwTo)+import Control.Exception (AsyncException (..), bracket, finally, throwIO, try)+import Control.Monad (forM, forM_, void, when)+import Data.Array (elems, listArray) import qualified Data.ByteString as BS import qualified Data.Geometry as G import Data.Int (Int64)-import Data.List (isInfixOf) import Data.Text (Text) import qualified Data.Text as Text import qualified Data.Vector as V-import Database.DuckDB.FFI import Database.DuckDB.Simple-import qualified Database.DuckDB.Simple.Deprecated.Streaming as Streaming-import Database.DuckDB.Simple.FromField (BitString (..), DecimalValue (..), FieldValue (..), StructValue (..), UnionValue)+import Database.DuckDB.Simple.FromField (BitString (..), DecimalValue (..), FieldValue (..), StructValue (..), UnionValue (..)) import Database.DuckDB.Simple.Generic (ViaDuckDB (..)) import Database.DuckDB.Simple.Geometry (RawGeometry (..), toRawGeometry)-import Database.DuckDB.Simple.Internal (destroyValue, withConnectionHandle)-import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), destroyLogicalType, logicalTypeFromRep)+import Database.DuckDB.Simple.Internal (withConnectionFactory, withConnectionHandle)+import Database.DuckDB.Simple.LogicalRep (LogicalTypeRep (..), StructField (..), UnionMemberType (..))+import qualified Database.DuckDB.Simple.Streaming as Streaming+import Database.DuckDB.Simple.V2.LogicalRep (destroyLogicalType, logicalTypeFromRep, logicalTypeToRep) import Database.DuckDB.Simple.Variant-import Foreign.C.String (withCString)-import Foreign.Marshal.Alloc (alloca)-import Foreign.Ptr (nullPtr)-import Foreign.Storable (peek, poke) import GHC.Float (castDoubleToWord64, castFloatToWord32, castWord64ToDouble) import GHC.Generics (Generic) import System.Directory (doesFileExist, getTemporaryDirectory, removeFile) import System.IO (hClose, openBinaryTempFile)+import System.Timeout (timeout) import Test.Tasty (TestTree, testGroup) import Test.Tasty.HUnit-import TestUtils (assertFailureIO)+import TestUtils (NativeTimestamp (..), assertFailureIO, nativeLocalTime, sized) -- | A generic record checks the bridge to composite type metadata. newtype VariantRecord = VariantRecord {payload :: Variant}- deriving stock (Eq, Show, Generic)- deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB VariantRecord)+ deriving stock (Eq, Show, Generic)+ deriving (DuckDBColumnType, ToField, FromField) via (ViaDuckDB VariantRecord) -- | Check VARIANT payloads against native decoding and parameter round trips. tests :: TestTree tests =- testGroup- "variant integration"- [ testGroup- "scalar payloads decode like native values"- [ testCase (Text.unpack sql) $ withConnection ":memory:" \conn -> do- [(Variant actual, native)] <- query_ conn (Query ("SELECT (" <> sql <> ")::VARIANT, " <> sql)) :: IO [(Variant, FieldValue)]- actual @?= native- roundTrip conn native- | sql <- scalarCases- ]- , testCase "explicit casts bind plain parameters" $ withConnection ":memory:" \conn -> do- [Only record] <- query_ conn "SELECT {'a': 1, 'b': 'x'}" :: IO [Only (StructValue FieldValue)]- (query conn "SELECT ?::VARIANT, ?::VARIANT, ?::VARIANT, ?::VARIANT" (42 :: Int64, "two" :: Text, record, listArray (0, 1) [1, 2] :: Array Int Int64) :: IO [(Variant, Variant, Variant, Variant)])- >>= (@?= [(Variant (FieldInt64 42), Variant (FieldText "two"), Variant (variantObject [("a", FieldInt32 1), ("b", FieldText "x")]), Variant (FieldList [FieldInt64 1, FieldInt64 2]))])- (query conn "SELECT [NULL::VARIANT, ?::VARIANT, ?::VARIANT]" (42 :: Int64, "two" :: Text) :: IO [Only [Variant]])- >>= (@?= [Only [Variant FieldNull, Variant (FieldInt64 42), Variant (FieldText "two")]])- _ <- execute_ conn "CREATE TABLE variants (v VARIANT)"- _ <- executeMany conn "INSERT INTO variants VALUES (?)" [Only (7 :: Int64), Only 8]- _ <- execute conn "INSERT INTO variants VALUES (?)" (Only record)- (query_ conn "SELECT v FROM variants" :: IO [Only Variant])- >>= (@?= [Only (Variant (FieldInt64 7)), Only (Variant (FieldInt64 8)), Only (Variant (variantObject [("a", FieldInt32 1), ("b", FieldText "x")]))])- , testCase "existing FromField instances read VARIANT payloads" $ withConnection ":memory:" \conn -> do- (query_ conn "SELECT 42::BIGINT::VARIANT, 'x'::VARIANT, [1, 2, 3]::VARIANT" :: IO [(Int64, Text, [Int64])])- >>= (@?= [(42, "x", [1, 2, 3])])- (query_ conn "SELECT {'payload': {'x': 18446744073709551615::UBIGINT}::VARIANT}" :: IO [Only VariantRecord])- >>= (@?= [Only (VariantRecord (Variant (variantObject [("x", FieldWord64 maxBound)])))])- , testCase "bound payloads keep integer widths and decimal scale" $ withConnection ":memory:" \conn ->- forM_ [("'-128'::TINYINT", FieldInt8 minBound), ("12.34::DECIMAL(4,2)", FieldDecimal (DecimalValue 4 2 1234)), ("255::UTINYINT", FieldWord8 maxBound)] \(sql, value) ->- (query conn (Query ("SELECT variant_typeof(?) = variant_typeof((" <> sql <> ")::VARIANT)")) (Only (Variant value)) :: IO [Only Bool])- >>= (@?= [Only True])- , testCase "floating special values retain type and bits" $ withConnection ":memory:" \conn -> do- forM_ [0, -0.0, 1 / 0, -1 / 0, 0 / 0] \value -> do- [Only (Variant actual)] <- query conn "SELECT ?" (Only (Variant (FieldDouble value)))- case actual of- FieldDouble decoded- | isNaN value -> assertBool "expected Double NaN" (isNaN decoded)- | otherwise -> castDoubleToWord64 decoded @?= castDoubleToWord64 value- _ -> assertFailure (show actual)- forM_ [0, -0.0, 1 / 0, -1 / 0, 0 / 0] \value -> do- [Only (Variant actual)] <- query conn "SELECT ?" (Only (Variant (FieldFloat value)))- case actual of- FieldFloat decoded- | isNaN value -> assertBool "expected Float NaN" (isNaN decoded)- | otherwise -> castFloatToWord32 decoded @?= castFloatToWord32 value- _ -> assertFailure (show actual)- , testCase "TIMETZ offsets decode like native values" $ withConnection ":memory:" \conn -> do- forM_ ["'12:00:00+01:23'::TIMETZ", "'23:59:59.999999-15:59'::TIMETZ", "'00:00:00+00'::TIMETZ"] \sql -> do- [(Variant actual, native)] <- query_ conn (Query ("SELECT (" <> sql <> ")::VARIANT, " <> sql)) :: IO [(Variant, FieldValue)]- actual @?= native- roundTrip conn native- assertFailureIO (query_ conn "SELECT '12:00:00+01:23:45'::TIMETZ::VARIANT" :: IO [Only Variant])- , testCase "wide finite timestamp payloads round trip without losing digits" $ withConnection ":memory:" \conn -> do- void (execute_ conn "CREATE TABLE wide_timestamps (seconds TIMESTAMP_S, millis TIMESTAMP_MS)")- appendWideTimestamps conn [(30000000000000, 30000000000000001), (-30000000000000, -30000000000000001), (maxBound - 1, maxBound - 1), (minBound, minBound)]- rows <- query_ conn "SELECT seconds::VARIANT, seconds, millis::VARIANT, millis FROM wide_timestamps" :: IO [(Variant, FieldValue, Variant, FieldValue)]- length rows @?= 4- forM_ rows \(seconds, nativeSeconds, millis, nativeMillis) ->- forM_ [(seconds, nativeSeconds), (millis, nativeMillis)] \(Variant actual, native) -> do- actual @?= native- roundTrip conn actual- , testCase "text and binary values preserve embedded NUL" $ withConnection ":memory:" \conn -> do- roundTrip conn (FieldText "before\0after íslenska λ 😀")- roundTrip conn (FieldBlob (BS.pack [0, 1, 127, 128, 255, 0]))- , testCase "objects preserve case-sensitive names and heterogeneous children" $ withConnection ":memory:" \conn -> do- let value =- variantObject- [ ("Case", FieldInt64 9007199254740993)- , ("case", FieldDecimal (DecimalValue 38 9 1234567890123456789))- , ("quote' íslenska λ", FieldList [FieldNull, FieldText "x", FieldList [], variantObject []])- ]- roundTrip conn value- (query conn "SELECT variant_extract(?, 'Case')::BIGINT, variant_extract(?, 'case')::DECIMAL(38,9)::VARCHAR" (Variant value, Variant value) :: IO [(Int64, Text)])- >>= (@?= [(9007199254740993, "1234567890.123456789")])- , testCase "empty containers and NULL stay distinct" $ withConnection ":memory:" \conn -> do- forM_ [FieldNull, FieldList [], variantObject [], FieldList [FieldNull], variantObject [("x", FieldNull)]] (roundTrip conn)- (query_ conn "SELECT NULL::VARIANT" :: IO [Only (Maybe Variant)]) >>= (@?= [Only Nothing])- (query conn "SELECT ?" (Only (Variant FieldNull)) :: IO [Only FieldValue]) >>= (@?= [Only FieldNull])- , testCase "deep native arrays decode without a library depth limit" $ withConnection ":memory:" \conn ->- forM_ [128, 256 :: Int] \depth -> do- let expected = Variant (foldr (const (FieldList . pure)) (FieldWord64 7) [1 .. depth])- (query conn "SELECT (repeat('[', ?) || '7' || repeat(']', ?))::JSON::VARIANT" (depth, depth) :: IO [Only Variant])- >>= (@?= [Only expected])- , testCase "ARRAY parameters and generic records contain VARIANT" $ withConnection ":memory:" \conn -> do- let array = listArray (0, 2) [Variant (FieldInt8 1), Variant (FieldText "two"), Variant FieldNull]- (query conn "SELECT ?" (Only array) :: IO [Only (Array Int Variant)]) >>= (@?= [Only array])- let record = VariantRecord (Variant (variantObject [("x", FieldWord64 maxBound)]))- (query conn "SELECT ?" (Only record) :: IO [Only VariantRecord]) >>= (@?= [Only record])- , testCase "nullable ARRAY parameters preserve heterogeneous VARIANT values" $ withConnection ":memory:" \conn -> do- let values = listArray (0, 3) [Nothing, Just (Variant (FieldInt64 42)), Just (Variant (FieldText "two")), Just (Variant (variantObject [("xs", FieldList [FieldNull, FieldBool True])]))]- (query conn "SELECT typeof(?), ?" (values, values) :: IO [(Text, Array Int (Maybe Variant))]) >>= (@?= [("VARIANT[4]", values)])- , testCase "empty ARRAY parameters retain VARIANT element type" $ withConnection ":memory:" \conn -> do- let values = listArray (0, -1) [] :: Array Int (Maybe Variant)- (query conn "SELECT typeof(?)" (Only values) :: IO [Only Text]) >>= (@?= [Only "VARIANT[ANY]"])- , testCase "parameter binding supplies a complete VARIANT type" $ withConnection ":memory:" \conn -> do- let value = variantObject [("xs", FieldList [FieldNull, FieldInt64 42])]- roundTrip conn value- (query conn "SELECT typeof(?)" (Only (Variant value)) :: IO [Only Text]) >>= (@?= [Only "VARIANT"])- , testCase "the first VARIANT parameter does not end a streaming result" $ withConnection ":memory:" \conn ->- withStatement conn "SELECT ?" \stmt -> do- total <- Streaming.fold_ conn "SELECT i FROM range(100000) t(i)" 0 \acc (Only i) -> do- when (i == 0) (bind stmt [toField (Variant (FieldInt64 42))])- pure (acc + i)- total @?= (sum [0 .. 99999] :: Int64)- , testCase "VARIANT type construction raises an error" $ do- result <- try (bracket (logicalTypeFromRep (LogicalTypeScalar DuckDBTypeVariant)) destroyLogicalType (const (pure ())))+ testGroup+ "variant integration"+ [ testCase "separate connections decode VARIANT concurrently" $+ withEnvironment \env -> withInstance env [] \inst ->+ withConnectionOn inst \left -> withConnectionOn inst \right -> do+ start <- newEmptyMVar+ let worker conn input = do+ readMVar start+ forM_ [1 .. 20 :: Int] \_ -> roundTrip conn input+ completions <- forM [(left, FieldList [FieldInt64 42, FieldNull]), (right, variantObject [("", FieldText "right"), ("n", FieldInt64 7)])] \(conn, input) -> do+ done <- newEmptyMVar+ thread <- forkFinally (worker conn input) (putMVar done)+ pure (thread, done)+ putMVar start ()+ result <- timeout 10000000 (mapM (takeMVar . snd) completions) `finally` mapM_ (killThread . fst) completions case result of- Left err -> assertBool (displayException (err :: SomeException)) ("?::VARIANT" `isInfixOf` displayException err)- Right () -> assertFailure "expected VARIANT type rejection"- , testCase "VARIANT payloads and GEOMETRY CRS metadata bind together" $ withConnection ":memory:" \conn -> do- [Only value] <- query_ conn "SELECT {'payload': 42::VARIANT, 'shape': NULL::GEOMETRY('OGC:CRS84')}" :: IO [Only (StructValue FieldValue)]- (query conn "SELECT typeof(?)" (Only value) :: IO [Only Text]) >>= (@?= [Only "STRUCT(payload VARIANT, shape GEOMETRY('OGC:CRS84'))"])- (query conn "SELECT ?" (Only value) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only value])- , testCase "inactive VARIANT UNION members have complete types" $ withConnection ":memory:" \conn -> do- [Only value] <- query_ conn "SELECT union_value(number := 42::BIGINT)::UNION(number BIGINT, payload VARIANT)" :: IO [Only (UnionValue FieldValue)]- (query conn "SELECT ?" (Only value) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only value])- , testCase "native LIST, ARRAY, MAP and STRUCT metadata round trip" $ withConnection ":memory:" \conn -> do- [Only value] <-- query_- conn- "SELECT {'xs': [1::VARIANT, 'two'::VARIANT, NULL], 'fixed': [42::VARIANT]::VARIANT[1], 'map': MAP {'x': {'a': 9}::VARIANT}}" ::- IO [Only (StructValue FieldValue)]- (query conn "SELECT ?" (Only value) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only value])- , testCase "UNION payloads retain VARIANT type, including NULL" $ withConnection ":memory:" \conn ->- forM_ ["SELECT union_value(v := {'a': 42}::VARIANT)", "SELECT union_value(v := NULL::VARIANT)"] \sql -> do- [Only value] <- query_ conn sql :: IO [Only (UnionValue FieldValue)]- (query conn "SELECT ?" (Only value) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only value])- , testGroup- "heterogeneous folds cross chunk boundaries"- [ testCase mode $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do- count <- foldRows- conn- "SELECT CASE WHEN i % 3 = 0 THEN i::VARIANT WHEN i % 3 = 1 THEN 'text'::VARIANT ELSE {'n': i, 'xs': [NULL, i]}::VARIANT END FROM range(5000) t(i)"- (0 :: Int64)- \n (Only (Variant actual)) -> do- let expected = case n `mod` 3 of- 0 -> FieldInt64 n- 1 -> FieldText "text"- _ -> variantObject [("n", FieldInt64 n), ("xs", FieldList [FieldNull, FieldInt64 n])]- actual @?= expected- pure (n + 1)- count @?= 5000- | (mode, foldRows) <- [("materialized", fold_), ("deprecated streaming", Streaming.fold_)]- ]- , testCase "filtered and reordered rows use the right child offsets" $ withConnection ":memory:" \conn -> do- rows <- query_ conn "SELECT {'n': i, 'xs': [i, i + 1]}::VARIANT FROM range(10000) t(i) WHERE i % 97 = 0 ORDER BY i DESC LIMIT 40"- let expected = [Only (Variant (variantObject [("n", FieldInt64 n), ("xs", FieldList [FieldInt64 n, FieldInt64 (n + 1)])])) | n <- take 40 (reverse [0, 97 .. 9999])]- rows @?= expected- , testCase "file-backed values survive checkpoint and reopen" $- bracket newDatabase removeDatabase \path -> do- withConnectionWithConfig path [("storage_compatibility_version", "v1.5.0")] \conn -> do- void (execute_ conn "CREATE TABLE stored AS SELECT i, {'n': i, 'xs': [i, NULL], 'text': i::VARCHAR}::VARIANT AS value, 'POINT (1 2)'::GEOMETRY('OGC:CRS84') AS geometry FROM range(10000) t(i)")- void (execute_ conn "CHECKPOINT")- withConnectionWithConfig path [("storage_compatibility_version", "v1.5.0")] \conn -> do- rows <- query_ conn "SELECT value FROM stored WHERE i % 97 = 0 ORDER BY i DESC LIMIT 40" :: IO [Only Variant]- let expected =- [ Only (Variant (variantObject [("n", FieldInt64 n), ("xs", FieldList [FieldInt64 n, FieldNull]), ("text", FieldText (Text.pack (show n)))]))- | n <- take 40 (reverse [0, 97 .. 9999])- ]- rows @?= expected- [Only geometry] <- query_ conn "SELECT geometry FROM stored LIMIT 1" :: IO [Only RawGeometry]- rawGeometryCRS geometry @?= Just "OGC:CRS84"- BS.length (rawGeometryWKB geometry) @?= 21- , testCase "geometry payload keeps WKB; native VARIANT does not retain CRS" $ withConnection ":memory:" \conn -> do- [Only geometry] <- query_ conn "SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]- [Only value] <- query_ conn "SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84')::VARIANT" :: IO [Only Variant]- value @?= geometryPayload (rawGeometryWKB geometry)- (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB geometry)) :: IO [Only Variant]) >>= (@?= [Only value])- (query conn "SELECT system.main.ST_CRS((system.main.ST_GeomFromWKB(?)::VARIANT)::GEOMETRY)" (Only (rawGeometryWKB geometry)) :: IO [Only (Maybe Text)]) >>= (@?= [Only Nothing])- forM_ [variantPayload value, FieldList [variantPayload value], variantObject [("shape", variantPayload value)]] \input -> do- result <- try (query conn "SELECT ?" (Only (Variant input)) :: IO [Only Variant]) :: IO (Either SomeException [Only Variant])- case result of- Left err -> assertBool (displayException err) ("raw GEOMETRY binding requires explicit" `isInfixOf` displayException err)- Right _ -> assertFailure "expected raw geometry binding rejection"- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])- , testCase "geometry payload retains empty layout tags and native NaN points" $ withConnection ":memory:" \conn ->- forM_ ["GEOMETRYCOLLECTION ZM EMPTY", "MULTIPOLYGON M EMPTY", "POINT Z (NaN NaN 7)"] \wkt -> do- [Only geometry] <- query conn "SELECT ?::GEOMETRY" (Only (wkt :: Text)) :: IO [Only RawGeometry]- (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB geometry)) :: IO [Only Variant]) >>= (@?= [Only (geometryPayload (rawGeometryWKB geometry))])- , testCase "geometry payload conversion ignores user macros with built-in names" $ withConnection ":memory:" \conn -> do- [Only geometry] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY" :: IO [Only RawGeometry]- _ <- execute_ conn "CREATE MACRO ST_AsText(x) AS 42"- _ <- execute_ conn "CREATE MACRO ST_GeomFromWKB(x) AS 'POINT (9 9)'::GEOMETRY"- (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB geometry)) :: IO [Only Variant]) >>= (@?= [Only (geometryPayload (rawGeometryWKB geometry))])- , testCase "binary geometry payloads retain NaN bits and mixed member layouts" $ withConnection ":memory:" \conn ->- forM_- [ G.PointGeometry (G.PointXYZ (G.XYZ (castWord64ToDouble 0x7ff8000000000001) (castWord64ToDouble 0x7ff8000000000002) 7))- , G.GeometryCollection (V.fromList [G.PointGeometry (G.PointXY (G.XY 1 2)), G.PointGeometry (G.PointXYZ (G.XYZ 3 4 5))])+ Nothing -> assertFailure "concurrent VARIANT decoding timed out"+ Just outcomes -> mapM_ (either throwIO pure) outcomes+ , testGroup+ "scalar payloads decode like native values"+ [ testCase (Text.unpack sql) $ withConnection ":memory:" \conn -> do+ -- Two rows with one tag use the list cast. The round trip uses the cast of one value.+ rows <-+ query_ conn (Query ("SELECT (" <> sql <> ")::VARIANT, {'v': " <> sql <> "}::VARIANT, " <> sql <> " FROM range(2)"))+ :: IO [(Variant, Variant, FieldValue)]+ length rows @?= 2+ forM_ rows \(Variant actual, Variant object, native) -> do+ actual @?= native+ object @?= variantObject [("v", native)]+ roundTrip conn native+ | sql <- scalarCases+ ]+ , testCase "explicit casts bind plain parameters" $ withConnection ":memory:" \conn -> do+ [Only record] <- query_ conn "SELECT {'a': 1, 'b': 'x'}" :: IO [Only (StructValue FieldValue)]+ ( query+ conn+ "SELECT ?::VARIANT, ?::VARIANT, ?::VARIANT, ?::VARIANT"+ (42 :: Int64, "two" :: Text, record, sized [1, 2] :: SizedArray 2 Int64)+ :: IO [(Variant, Variant, Variant, Variant)]+ )+ >>= ( @?=+ [+ ( Variant (FieldInt64 42)+ , Variant (FieldText "two")+ , Variant (variantObject [("a", FieldInt32 1), ("b", FieldText "x")])+ , Variant (FieldList [FieldInt64 1, FieldInt64 2])+ )+ ]+ )+ (query conn "SELECT [NULL::VARIANT, ?::VARIANT, ?::VARIANT]" (42 :: Int64, "two" :: Text) :: IO [Only [Variant]])+ >>= (@?= [Only [Variant FieldNull, Variant (FieldInt64 42), Variant (FieldText "two")]])+ _ <- execute_ conn "CREATE TABLE variants (v VARIANT)"+ _ <- executeMany conn "INSERT INTO variants VALUES (?)" [Only (7 :: Int64), Only 8]+ _ <- execute conn "INSERT INTO variants VALUES (?)" (Only record)+ (query_ conn "SELECT v FROM variants" :: IO [Only Variant])+ >>= ( @?=+ [Only (Variant (FieldInt64 7)), Only (Variant (FieldInt64 8)), Only (Variant (variantObject [("a", FieldInt32 1), ("b", FieldText "x")]))]+ )+ , testCase "existing FromField instances read VARIANT payloads" $ withConnection ":memory:" \conn -> do+ (query_ conn "SELECT 42::BIGINT::VARIANT, 'x'::VARIANT, [1, 2, 3]::VARIANT" :: IO [(Int64, Text, [Int64])])+ >>= (@?= [(42, "x", [1, 2, 3])])+ (query_ conn "SELECT {'payload': {'x': 18446744073709551615::UBIGINT}::VARIANT}" :: IO [Only VariantRecord])+ >>= (@?= [Only (VariantRecord (Variant (variantObject [("x", FieldWord64 maxBound)])))])+ , testCase "bound payloads keep integer widths and decimal scale" $ withConnection ":memory:" \conn ->+ forM_+ [ ("'-128'::TINYINT", FieldInt8 minBound)+ , ("12.34::DECIMAL(4,2)", FieldDecimal (DecimalValue 4 2 1234))+ , ("255::UTINYINT", FieldWord8 maxBound)+ ]+ \(sql, value) ->+ (query conn (Query ("SELECT variant_typeof(?) = variant_typeof((" <> sql <> ")::VARIANT)")) (Only (Variant value)) :: IO [Only Bool])+ >>= (@?= [Only True])+ , testCase "floating special values retain type and bits" $ withConnection ":memory:" \conn -> do+ forM_ [0, -0.0, 1 / 0, -1 / 0, 0 / 0] \value -> do+ [Only (Variant actual)] <- query conn "SELECT ?" (Only (Variant (FieldDouble value)))+ case actual of+ FieldDouble decoded+ | isNaN value -> assertBool "expected Double NaN" (isNaN decoded)+ | otherwise -> castDoubleToWord64 decoded @?= castDoubleToWord64 value+ _ -> assertFailure (show actual)+ forM_ [0, -0.0, 1 / 0, -1 / 0, 0 / 0] \value -> do+ [Only (Variant actual)] <- query conn "SELECT ?" (Only (Variant (FieldFloat value)))+ case actual of+ FieldFloat decoded+ | isNaN value -> assertBool "expected Float NaN" (isNaN decoded)+ | otherwise -> castFloatToWord32 decoded @?= castFloatToWord32 value+ _ -> assertFailure (show actual)+ , testCase "TIMETZ offsets decode like native values" $ withConnection ":memory:" \conn -> do+ forM_+ [ "'12:00:00+01:23'::TIMETZ"+ , "'23:59:59.999999-15:59'::TIMETZ"+ , "'00:00:00+00'::TIMETZ"+ , "'12:00:00+01:23:45'::TIMETZ"+ , "'00:00:00-15:59:59'::TIMETZ"+ ]+ \sql -> do+ rows <- query_ conn (Query ("SELECT (" <> sql <> ")::VARIANT, " <> sql <> " FROM range(2)")) :: IO [(Variant, FieldValue)]+ length rows @?= 2+ forM_ rows \(Variant actual, native) -> do+ actual @?= native+ roundTrip conn native+ , testCase "wide finite timestamp payloads read without losing digits, and do not bind" $ withConnection ":memory:" \conn -> do+ void (execute_ conn "CREATE TABLE wide_timestamps (seconds TIMESTAMP_S, millis TIMESTAMP_MS)")+ let units =+ [(30000000000000, 30000000000000001), (-30000000000000, -30000000000000001), (maxBound - 1, maxBound - 1), (minBound, minBound)]+ :: [(Int64, Int64)]+ void (executeMany conn "INSERT INTO wide_timestamps VALUES (?, ?)" [(Seconds seconds, Milliseconds millis) | (seconds, millis) <- units])+ -- DuckDB sorts the raw values. A cast of these values to VARCHAR gives an+ -- internal error in DuckDB 2.0.0, so the case does not read their text.+ forM_ ["seconds", "millis"] \column ->+ query_ conn (Query ("SELECT rowid FROM wide_timestamps ORDER BY " <> column)) >>= (@?= map Only [3, 1, 0, 2 :: Int64])+ query_ conn "SELECT seconds, millis FROM wide_timestamps ORDER BY rowid"+ >>= (@?= [(nativeLocalTime (Seconds seconds), nativeLocalTime (Milliseconds millis)) | (seconds, millis) <- units])+ rows <-+ query_ conn "SELECT seconds::VARIANT, seconds, millis::VARIANT, millis FROM wide_timestamps"+ :: IO [(Variant, FieldValue, Variant, FieldValue)]+ length rows @?= 4+ forM_ rows \(seconds, nativeSeconds, millis, nativeMillis) ->+ forM_ [(seconds, nativeSeconds), (millis, nativeMillis)] \(Variant actual, native) -> do+ actual @?= native+ -- The value is outside the range of TIMESTAMP. See+ -- Note [Time ranges] in "Database.DuckDB.Simple.ToField".+ outcome <- try (query conn "SELECT ?" (Only (Variant actual)) :: IO [Only Variant])+ case outcome of+ Left (err :: SQLError) -> do+ sqlErrorType err @?= Just ErrorTypeConversion+ assertBool (show err) ("TIMESTAMP of DuckDB" `Text.isInfixOf` sqlErrorMessage err)+ Right bound -> assertFailure ("expected the range error, got " <> show bound)+ , testCase "text and binary values preserve embedded NUL" $ withConnection ":memory:" \conn -> do+ roundTrip conn (FieldText "before\0after íslenska λ 😀")+ roundTrip conn (FieldBlob (BS.pack [0, 1, 127, 128, 255, 0]))+ , testCase "objects preserve case-sensitive names and heterogeneous children" $ withConnection ":memory:" \conn -> do+ let value =+ variantObject+ [ ("Case", FieldInt64 9007199254740993)+ , ("case", FieldDecimal (DecimalValue 38 9 1234567890123456789))+ , ("quote' íslenska λ", FieldList [FieldNull, FieldText "x", FieldList [], variantObject []]) ]- \shape -> do- raw <- either assertFailure pure (toRawGeometry shape)- (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB raw)) :: IO [Only Variant]) >>= (@?= [Only (geometryPayload (rawGeometryWKB raw))])- , testCase "invalid values fail before binding and connection reuse succeeds" $ withConnection ":memory:" \conn -> do- forM_- [ FieldHugeInt (2 ^ (127 :: Int))- , FieldUHugeInt (-1)- , FieldUHugeInt (2 ^ (128 :: Int))- , FieldDecimal (DecimalValue 0 0 0)- , FieldDecimal (DecimalValue 5 6 1)- , FieldDecimal (DecimalValue 3 0 1000)- , FieldBit (BitString 8 (BS.singleton 0))- , variantObject [("same", FieldNull), ("same", FieldBool True)]- , variantObject [("", FieldBool True)]- , variantObject [("first", FieldBool False), ("", FieldBool True)]- , variantObject [("before\0after", FieldNull)]+ roundTrip conn value+ ( query conn "SELECT variant_extract(?, 'Case')::BIGINT, variant_extract(?, 'case')::DECIMAL(38,9)::VARCHAR" (Variant value, Variant value)+ :: IO [(Int64, Text)]+ )+ >>= (@?= [(9007199254740993, "1234567890.123456789")])+ , testCase "native objects use lexical key order and arrays retain element order" $ withConnection ":memory:" \conn ->+ (query_ conn "SELECT {'z':1, 'a':2, 'm':[3,1,2]}::VARIANT" :: IO [Only Variant])+ >>= (@?= [Only (Variant (variantObject [("a", FieldInt32 2), ("m", FieldList [FieldInt32 3, FieldInt32 1, FieldInt32 2]), ("z", FieldInt32 1)]))])+ , testCase "empty containers and NULL stay distinct" $ withConnection ":memory:" \conn -> do+ forM_ [FieldNull, FieldList [], variantObject [], FieldList [FieldNull], variantObject [("x", FieldNull)]] (roundTrip conn)+ (query_ conn "SELECT NULL::VARIANT" :: IO [Only (Maybe Variant)]) >>= (@?= [Only Nothing])+ (query conn "SELECT ?" (Only (Variant FieldNull)) :: IO [Only FieldValue]) >>= (@?= [Only FieldNull])+ , testCase "native objects preserve empty and embedded NUL keys" $ withConnection ":memory:" \conn -> do+ (query_ conn "SELECT '{\"\":42}'::JSON::VARIANT" :: IO [Only Variant])+ >>= (@?= [Only (Variant (variantObject [("", FieldWord64 42)]))])+ (query_ conn "SELECT '{\"A\":1,\"a\":2,\"a\\u0000b\":3}'::JSON::VARIANT" :: IO [Only Variant])+ >>= (@?= [Only (Variant (variantObject [("A", FieldWord64 1), ("a", FieldWord64 2), ("a\0b", FieldWord64 3)]))])+ , testCase "empty object keys survive parameter binding" $ withConnection ":memory:" \conn -> do+ [Only value] <- query_ conn "SELECT '{\"named\": 1, \"\": 42}'::JSON::VARIANT" :: IO [Only Variant]+ query conn "SELECT ?" (Only value) >>= (@?= [Only value])+ -- Decoded objects list their keys in DuckDB's lexicographic order.+ roundTrip+ conn+ (variantObject [("", FieldList [FieldNull, variantObject [("", FieldText "λ"), ("named", FieldInt64 1)]]), ("named", FieldBool True)])+ sole <- try (query conn "SELECT ?" (Only (Variant (variantObject [("", FieldInt64 42)])))) :: IO (Either SQLError [Only Variant])+ case sole of+ Left err -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ assertBool (Text.unpack (sqlErrorMessage err)) ("only an empty key" `Text.isInfixOf` sqlErrorMessage err)+ Right _ -> assertFailure "expected the native constructor limit for an object with only an empty key"+ , testCase "objects with two empty keys fail the unique key check" $ withConnection ":memory:" \conn -> do+ result <-+ try (query conn "SELECT ?" (Only (Variant (variantObject [("", FieldInt64 1), ("", FieldInt64 2)]))))+ :: IO (Either SQLError [Only Variant])+ case result of+ Left err -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ sqlErrorMessage err @?= "duckdb-simple: VARIANT objects need unique keys"+ Right _ -> assertFailure "expected the unique key check to reject two empty keys"+ , testCase "metadata conversion recovers after preparation failure and ignores shadowing macros" $ withConnection ":memory:" \conn -> do+ void (execute_ conn "SET GLOBAL max_expression_depth = 10")+ failed <- try (query_ conn "SELECT 42::VARIANT" :: IO [Only Variant])+ void (execute_ conn "SET GLOBAL max_expression_depth = 1000")+ case failed of+ Left SQLError{sqlErrorType} -> sqlErrorType @?= Just ErrorQueryBinder+ Right _ -> assertFailure "expected the metadata preparation depth error"+ -- The connection keeps the prepared metadata statement of the first successful decode.+ (query_ conn "SELECT {'a': 42::TINYINT}::VARIANT" :: IO [Only Variant]) >>= (@?= [Only (Variant (variantObject [("a", FieldInt8 42)]))])+ void (execute_ conn "CREATE MACRO variant_typeof(x) AS 'VARCHAR'")+ void (execute_ conn "CREATE MACRO list_transform(x,y) AS [42]")+ void (execute_ conn "CREATE MACRO variant_keys(x) AS ['wrong']")+ (query_ conn "SELECT [42::TINYINT::VARIANT, 'two'::VARIANT]::VARIANT" :: IO [Only Variant])+ >>= (@?= [Only (Variant (FieldList [FieldInt8 42, FieldText "two"]))])+ (query_ conn "SELECT {'a': 42::TINYINT}::VARIANT" :: IO [Only Variant]) >>= (@?= [Only (Variant (variantObject [("a", FieldInt8 42)]))])+ , testCase "deep native arrays decode without a library depth limit" $ withConnection ":memory:" \conn ->+ forM_ [128, 256 :: Int] \depth -> do+ let expected = Variant (foldr (const (FieldList . pure)) (FieldWord64 7) [1 .. depth])+ (query conn "SELECT (repeat('[', ?) || '7' || repeat(']', ?))::JSON::VARIANT AS payload" (depth, depth) :: IO [Only Variant])+ >>= (@?= [Only expected])+ let object = Variant (variantObject [("nested", variantPayload expected)])+ ( query conn "SELECT ('{\"nested\":' || repeat('[', ?) || '7' || repeat(']', ?) || '}')::JSON::VARIANT AS payload" (depth, depth)+ :: IO [Only Variant]+ )+ >>= (@?= [Only object])+ , testCase "ARRAY parameters and generic records contain VARIANT" $ withConnection ":memory:" \conn -> do+ let array = sized [Variant (FieldInt8 1), Variant (FieldText "two"), Variant FieldNull] :: SizedArray 3 Variant+ (query conn "SELECT ?" (Only array) :: IO [Only (SizedArray 3 Variant)]) >>= (@?= [Only array])+ let record = VariantRecord (Variant (variantObject [("x", FieldWord64 maxBound)]))+ (query conn "SELECT ?" (Only record) :: IO [Only VariantRecord]) >>= (@?= [Only record])+ , testCase "nullable ARRAY parameters preserve heterogeneous VARIANT values" $ withConnection ":memory:" \conn -> do+ let values =+ sized+ [ Nothing+ , Just (Variant (FieldInt64 42))+ , Just (Variant (FieldText "two"))+ , Just (Variant (variantObject [("xs", FieldList [FieldNull, FieldBool True])])) ]- \value -> assertFailureIO (query conn "SELECT ?" (Only (Variant value)) :: IO [Only Variant])- (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ (query conn "SELECT typeof(?), ?" (values, values) :: IO [(Text, SizedArray 4 (Maybe Variant))]) >>= (@?= [("VARIANT[4]", values)])+ , testCase "empty ARRAY parameters raise an error" $ withConnection ":memory:" \conn -> do+ let values = sized [] :: SizedArray 0 (Maybe Variant)+ result <- try (query conn "SELECT typeof(?)" (Only values))+ case result of+ Left SQLError{sqlErrorMessage, sqlErrorType} -> do+ assertBool "must identify empty ARRAY" ("empty ARRAY" `Text.isInfixOf` sqlErrorMessage)+ sqlErrorType @?= Just ErrorInputInvalid+ Right (_ :: [Only Text]) -> assertFailure "expected the empty ARRAY error"+ , testCase "parameter binding supplies a complete VARIANT type" $ withConnection ":memory:" \conn -> do+ let value = variantObject [("xs", FieldList [FieldNull, FieldInt64 42])]+ roundTrip conn value+ (query conn "SELECT typeof(?)" (Only (Variant value)) :: IO [Only Text]) >>= (@?= [Only "VARIANT"])+ , testCase "the first VARIANT parameter does not end a streaming result" $ withConnection ":memory:" \conn ->+ withStatement conn "SELECT ?" \stmt -> do+ total <- Streaming.fold_ conn "SELECT i FROM range(100000) t(i)" 0 \acc (Only i) -> do+ when (i == 0) (bind stmt [toField (Variant (FieldInt64 42))])+ pure (acc + i)+ total @?= (sum [0 .. 99999] :: Int64)+ , testCase "VARIANT type construction uses the supplied factory" $ withConnection ":memory:" \conn ->+ withConnectionHandle conn \native -> withConnectionFactory native \factory ->+ bracket (logicalTypeFromRep factory (LogicalTypeScalar LogicalTypeIdVariant)) destroyLogicalType \logical ->+ logicalTypeToRep factory logical >>= (@?= LogicalTypeScalar LogicalTypeIdVariant)+ , testCase "VARIANT payloads and GEOMETRY CRS metadata bind together" $ withConnection ":memory:" \conn -> do+ [Only value] <- query_ conn "SELECT {'payload': 42::VARIANT, 'shape': NULL::GEOMETRY('OGC:CRS84')}" :: IO [Only (StructValue FieldValue)]+ (query conn "SELECT typeof(?)" (Only value) :: IO [Only Text]) >>= (@?= [Only "STRUCT(payload VARIANT, shape GEOMETRY('OGC:CRS84'))"])+ (query conn "SELECT ?" (Only value) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only value])+ , testCase "inactive VARIANT UNION members have complete types" $ withConnection ":memory:" \conn -> do+ [Only value] <-+ query_ conn "SELECT union_value(number := 42::BIGINT)::UNION(number BIGINT, payload VARIANT)" :: IO [Only (UnionValue FieldValue)]+ (query conn "SELECT ?" (Only value) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only value])+ , testCase "native LIST, ARRAY, MAP and STRUCT metadata round trip" $ withConnection ":memory:" \conn -> do+ [Only value] <-+ query_+ conn+ "SELECT {'xs': [1::VARIANT, 'two'::VARIANT, NULL], 'fixed': [42::VARIANT]::VARIANT[1], 'map': MAP {'x': {'a': 9}::VARIANT}}"+ :: IO [Only (StructValue FieldValue)]+ (query conn "SELECT ?" (Only value) :: IO [Only (StructValue FieldValue)]) >>= (@?= [Only value])+ , testCase "UNION payloads retain VARIANT type, including NULL" $ withConnection ":memory:" \conn ->+ forM_ ["SELECT union_value(v := {'a': 42}::VARIANT)", "SELECT union_value(v := NULL::VARIANT)"] \sql -> do+ [Only value] <- query_ conn sql :: IO [Only (UnionValue FieldValue)]+ (query conn "SELECT ?" (Only value) :: IO [Only (UnionValue FieldValue)]) >>= (@?= [Only value])+ , testGroup+ "heterogeneous folds cross chunk boundaries"+ [ testCase mode $ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ -- Cancel after metadata and the first chunk have been decoded.+ cancelled <- try $ foldRows conn "SELECT i::VARIANT FROM range(5000) t(i)" (0 :: Int64) \n (Only (Variant actual)) -> do+ actual @?= FieldInt64 0+ self <- myThreadId+ throwTo self UserInterrupt+ pure n+ (cancelled :: Either AsyncException Int64) @?= Left UserInterrupt+ count <- foldRows+ conn+ "SELECT CASE WHEN i % 3 = 0 THEN i::VARIANT WHEN i % 3 = 1 THEN 'text'::VARIANT ELSE {'n': i, 'xs': [NULL, i]}::VARIANT END FROM range(5000) t(i)"+ (0 :: Int64)+ \n (Only (Variant actual)) -> do+ let expected = case n `mod` 3 of+ 0 -> FieldInt64 n+ 1 -> FieldText "text"+ _ -> variantObject [("n", FieldInt64 n), ("xs", FieldList [FieldNull, FieldInt64 n])]+ actual @?= expected+ pure (n + 1)+ count @?= 5000+ | (mode, foldRows) <- [("materialized", fold_), ("native streaming", Streaming.fold_)] ]+ , testCase "one column with many payload tags decodes each row exactly" $ withConnection ":memory:" \conn -> do+ rows <-+ query_+ conn+ "SELECT i, CASE i % 12 WHEN 0 THEN NULL WHEN 1 THEN (i % 24 = 1)::VARIANT WHEN 2 THEN (i % 100)::TINYINT::VARIANT WHEN 3 THEN i::VARIANT WHEN 4 THEN (i * 0.5)::DOUBLE::VARIANT WHEN 5 THEN i::DECIMAL(9,2)::VARIANT WHEN 6 THEN ('t' || i)::VARIANT WHEN 7 THEN [i, NULL, i + 1]::VARIANT WHEN 8 THEN {'n': i, 'xs': [i]}::VARIANT WHEN 9 THEN {'k': {'deep': 't' || i}}::VARIANT WHEN 10 THEN []::VARIANT ELSE i::UBIGINT::VARIANT END FROM range(5000) t(i)"+ :: IO [(Int64, Maybe Variant)]+ let text n = FieldText ("t" <> Text.pack (show n))+ expected n = case n `mod` 12 of+ 0 -> Nothing+ 1 -> Just (FieldBool (n `mod` 24 == 1))+ 2 -> Just (FieldInt8 (fromIntegral (n `mod` 100)))+ 3 -> Just (FieldInt64 n)+ 4 -> Just (FieldDouble (fromIntegral n * 0.5))+ 5 -> Just (FieldDecimal (DecimalValue 9 2 (toInteger n * 100)))+ 6 -> Just (text n)+ 7 -> Just (FieldList [FieldInt64 n, FieldNull, FieldInt64 (n + 1)])+ 8 -> Just (variantObject [("n", FieldInt64 n), ("xs", FieldList [FieldInt64 n])])+ 9 -> Just (variantObject [("k", variantObject [("deep", text n)])])+ 10 -> Just (FieldList [])+ _ -> Just (FieldWord64 (fromIntegral n))+ length rows @?= 5000+ forM_ rows \(n, actual) -> fmap variantPayload actual @?= expected n+ , testCase "GEOMETRY, BLOB and BIGNUM payloads in one column decode like native values" $ withConnection ":memory:" \conn -> do+ rows <-+ query_+ conn+ "SELECT i, CASE i % 3 WHEN 0 THEN g::VARIANT WHEN 1 THEN b::VARIANT ELSE n::VARIANT END, g, b, n FROM (SELECT i, ('POINT (' || i || ' ' || (i + 1) || ')')::GEOMETRY AS g, encode('b' || i) AS b, (i::VARCHAR || '12345678901234567890')::BIGNUM AS n FROM range(5000) t(i))"+ :: IO [(Int64, Variant, FieldValue, FieldValue, FieldValue)]+ length rows @?= 5000+ forM_ rows \(n, Variant actual, geometry, blob, bignum) ->+ actual @?= case n `mod` 3 of+ 0 -> geometry+ 1 -> blob+ _ -> bignum+ , testCase "objects with many key layouts and many keys decode exactly" $ withConnection ":memory:" \conn -> do+ let digits = Text.pack . show :: Int64 -> Text+ layouts <- query_ conn "SELECT i, json_object('k' || i::VARCHAR, i)::VARIANT FROM range(3000) t(i)" :: IO [(Int64, Variant)]+ length layouts @?= 3000+ forM_ layouts \(n, Variant actual) -> actual @?= variantObject [("k" <> digits n, FieldWord64 (fromIntegral n))]+ (query_ conn "SELECT json_group_object('k' || lpad(i::VARCHAR, 3, '0'), i)::VARIANT FROM range(300) t(i)" :: IO [Only Variant])+ >>= (@?= [Only (Variant (variantObject [("k" <> Text.justifyRight 3 '0' (digits n), FieldWord64 (fromIntegral n)) | n <- [0 .. 299]]))])+ , testCase "one column mixes object layouts with special keys" $ withConnection ":memory:" \conn -> do+ rows <-+ query_+ conn+ "SELECT i, CASE i % 4 WHEN 0 THEN json_object('A', i, 'a', i + 1)::VARIANT WHEN 1 THEN json_object('', i, 'n', i + 1)::VARIANT WHEN 2 THEN json_object('É', i, 'é', i + 1, 'a\"b', i + 2, 'it''s λ', i + 3)::VARIANT ELSE json_object('n', i, 'a', i + 1)::VARIANT END FROM range(3000) t(i)"+ :: IO [(Int64, Variant)]+ -- Decoded objects list their keys in DuckDB's lexicographic order.+ let expected n = variantObject [(key, FieldWord64 (fromIntegral (n + offset))) | (key, offset) <- layout]+ where+ layout = case n `mod` 4 of+ 0 -> [("A", 0), ("a", 1)]+ 1 -> [("", 0), ("n", 1)]+ 2 -> [("a\"b", 2), ("it's λ", 3), ("É", 0), ("é", 1)]+ _ -> [("a", 1), ("n", 0)]+ length rows @?= 3000+ forM_ rows \(n, Variant actual) -> actual @?= expected n+ , testCase "VARIANT objects with keys that a STRUCT cannot name decode in each chunk" $+ withConnectionWithConfig ":memory:" [("threads", "1")] \conn -> do+ -- The empty key needs the extraction query. The result has three chunks.+ let sql = "SELECT i, json_object('', i, 'n', i + 1)::VARIANT FROM range(5000) t(i)" :: Query+ expected n = variantObject [("", FieldWord64 (fromIntegral n)), ("n", FieldWord64 (fromIntegral (n + 1)))]+ rows <- query_ conn sql :: IO [(Int64, Variant)]+ rows @?= [(n, Variant (expected n)) | n <- [0 .. 4999]]+ count <- fold_ conn sql (0 :: Int64) \n (i, Variant actual) -> do+ (i, actual) @?= (n, expected n)+ pure (n + 1)+ count @?= 5000+ , testCase "VARIANT objects with different key counts decode after each other" $ withConnection ":memory:" \conn -> do+ -- Each object has an empty key, so each object group uses the extraction+ -- query. The key count changes between the queries and in the last query.+ let two n = variantObject [("", FieldWord64 n), ("n", FieldWord64 (n + 1))]+ three n = variantObject [("", FieldWord64 n), ("a", FieldWord64 (n + 1)), ("b", FieldWord64 (n + 2))]+ twoKeys = "SELECT json_object('', i, 'n', i + 1)::VARIANT FROM range(3) t(i)"+ threeKeys = "SELECT json_object('', i, 'a', i + 1, 'b', i + 2)::VARIANT FROM range(3) t(i)"+ check sql expected = (query_ conn sql :: IO [Only Variant]) >>= (@?= map (Only . Variant) expected)+ check twoKeys (map two [0 .. 2])+ check threeKeys (map three [0 .. 2])+ check twoKeys (map two [0 .. 2])+ check+ "SELECT CASE WHEN i % 2 = 0 THEN json_object('', i, 'n', i + 1) ELSE json_object('', i, 'a', i + 1, 'b', i + 2) END::VARIANT FROM range(4) t(i)"+ [two 0, three 1, two 2, three 3]+ forM_+ [ []+ , [FieldNull, FieldInt8 3, FieldText "two"]+ , [FieldList [], variantObject []]+ , [FieldList [FieldInt64 7, FieldNull], three 4]+ , []+ , [FieldInt64 9]+ ]+ \payloads ->+ (query conn "SELECT unnest(?::VARIANT[])" (Only (map Variant payloads)) :: IO [Only Variant])+ >>= (@?= map (Only . Variant) payloads)+ , testCase "filtered and reordered rows use the right child offsets" $ withConnection ":memory:" \conn -> do+ rows <- query_ conn "SELECT {'n': i, 'xs': [i, i + 1]}::VARIANT FROM range(10000) t(i) WHERE i % 97 = 0 ORDER BY i DESC LIMIT 40"+ let expected =+ [ Only (Variant (variantObject [("n", FieldInt64 n), ("xs", FieldList [FieldInt64 n, FieldInt64 (n + 1)])]))+ | n <- take 40 (reverse [0, 97 .. 9999])+ ]+ rows @?= expected+ , testCase "file-backed values survive checkpoint and reopen" $+ bracket newDatabase removeDatabase \path -> do+ withConnectionWithConfig path [("storage_compatibility_version", "v1.5.0")] \conn -> do+ void+ ( execute_+ conn+ "CREATE TABLE stored AS SELECT i, {'n': i, 'xs': [i, NULL], 'text': i::VARCHAR}::VARIANT AS value, 'POINT (1 2)'::GEOMETRY('OGC:CRS84') AS geometry FROM range(10000) t(i)"+ )+ void (execute_ conn "CHECKPOINT")+ withConnectionWithConfig path [("storage_compatibility_version", "v1.5.0")] \conn -> do+ rows <- query_ conn "SELECT value FROM stored WHERE i % 97 = 0 ORDER BY i DESC LIMIT 40" :: IO [Only Variant]+ let expected =+ [ Only (Variant (variantObject [("n", FieldInt64 n), ("text", FieldText (Text.pack (show n))), ("xs", FieldList [FieldInt64 n, FieldNull])]))+ | n <- take 40 (reverse [0, 97 .. 9999])+ ]+ rows @?= expected+ [Only nativeKeys] <- query_ conn "SELECT system.main.variant_keys(value) FROM stored LIMIT 1" :: IO [Only [Text]]+ nativeKeys @?= ["n", "text", "xs"]+ forM_ rows \(Only (Variant payloadValue)) -> case payloadValue of+ FieldStruct object -> map structFieldName (elems (structValueFields object)) @?= nativeKeys+ other -> assertFailure (show other)+ [Only geometry] <- query_ conn "SELECT geometry FROM stored LIMIT 1" :: IO [Only RawGeometry]+ rawGeometryCRS geometry @?= Just "OGC:CRS84"+ BS.length (rawGeometryWKB geometry) @?= 21+ , testCase "geometry payload keeps WKB; native VARIANT does not retain CRS" $ withConnection ":memory:" \conn -> do+ [Only geometry] <- query_ conn "SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ [Only value] <- query_ conn "SELECT 'POINT ZM (1 2 3 4)'::GEOMETRY('OGC:CRS84')::VARIANT" :: IO [Only Variant]+ value @?= geometryPayload (rawGeometryWKB geometry)+ (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB geometry)) :: IO [Only Variant]) >>= (@?= [Only value])+ ( query conn "SELECT system.main.ST_CRS((system.main.ST_GeomFromWKB(?)::VARIANT)::GEOMETRY)" (Only (rawGeometryWKB geometry))+ :: IO [Only (Maybe Text)]+ )+ >>= (@?= [Only Nothing])+ forM_ [variantPayload value, FieldList [variantPayload value], variantObject [("shape", variantPayload value)]] (roundTrip conn)+ , testCase "raw geometry without a CRS binds in a VARIANT, and a CRS raises an error" $ withConnection ":memory:" \conn -> do+ nanPoint <-+ either assertFailure pure $+ toRawGeometry (G.PointGeometry (G.PointXYZ (G.XYZ (castWord64ToDouble 0x7ff8000000000001) (castWord64ToDouble 0x7ff8000000000002) 7)))+ native <- forM ["GEOMETRYCOLLECTION ZM EMPTY", "MULTIPOLYGON M EMPTY", "POINT Z (NaN NaN 7)", "POLYGON ((0 0, 1 0, 1 1, 0 0))"] \wkt -> do+ [Only geometry] <- query conn "SELECT ?::GEOMETRY" (Only (wkt :: Text)) :: IO [Only RawGeometry]+ pure geometry+ forM_ (map rawGeometryWKB (nanPoint : native)) \wkb -> do+ let value = geometryPayload wkb+ roundTrip conn (variantPayload value)+ (query conn "SELECT variant_typeof(?), (?)::GEOMETRY::BLOB" (value, value) :: IO [(Text, BS.ByteString)]) >>= (@?= [("GEOMETRY", wkb)])+ [Only withCRS] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY('OGC:CRS84')" :: IO [Only RawGeometry]+ forM_ [FieldGeometry withCRS, FieldList [FieldGeometry withCRS], variantObject [("shape", FieldGeometry withCRS)]] \input -> do+ result <- try (query conn "SELECT ?" (Only (Variant input)) :: IO [Only Variant])+ case result of+ Left err -> do+ sqlErrorMessage err+ @?= "duckdb-simple: a VARIANT does not keep the CRS of a GEOMETRY, so the GEOMETRY value with the CRS 'OGC:CRS84' cannot bind as a VARIANT"+ sqlErrorType err @?= Just ErrorInputInvalid+ Right _ -> assertFailure "expected a VARIANT CRS error"+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "UNION, MAP, and ENUM payloads raise an error at each depth" $ withConnection ":memory:" \conn -> do+ let members =+ listArray+ (0, 1)+ [UnionMemberType "a" (LogicalTypeScalar LogicalTypeIdInteger), UnionMemberType "b" (LogicalTypeScalar LogicalTypeIdVarchar)]+ depths payload = [payload, FieldList [payload], variantObject [("u", payload)], variantObject [("u", FieldList [payload])]]+ bindPayload input = try (query conn "SELECT ?" (Only (Variant input)) :: IO [Only Variant])+ forM_ [FieldInt32 1, FieldNull] \member ->+ forM_ (depths (FieldUnion (UnionValue 0 "a" member members))) \input ->+ bindPayload input >>= \result -> case result of+ Left err -> do+ sqlErrorType err @?= Just ErrorInputInvalid+ assertBool (Text.unpack (sqlErrorMessage err)) ("a VARIANT keeps only the member value of a UNION" `Text.isInfixOf` sqlErrorMessage err)+ Right rows -> assertFailure ("expected the UNION error, got " <> show rows)+ forM_+ [ (FieldMap [(FieldText "k", FieldInt32 1)], "a VARIANT stores a MAP as a list of key and value objects")+ , (FieldEnum 0 "a", "a VARIANT stores an ENUM as its text")+ ]+ \(payload, reason) -> forM_ (depths payload) \input -> do+ result <- try (query conn "SELECT ?" (Only (Variant input)) :: IO [Only Variant])+ case result of+ Left SQLError{sqlErrorType, sqlErrorMessage} -> do+ sqlErrorType @?= Just ErrorInputInvalid+ assertBool (Text.unpack sqlErrorMessage) (Text.pack reason `Text.isInfixOf` sqlErrorMessage)+ Right rows -> assertFailure ("expected the " <> reason <> " error, got " <> show rows)+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "a Variant read of a JSON type raises an error at each depth, and the cast reads the parsed value" $ withConnection ":memory:" \conn -> do+ let refused column expression sqlName =+ try (query_ conn (Query ("SELECT " <> expression <> " AS " <> column)) :: IO [Only Variant]) >>= \case+ Left SQLError{sqlErrorMessage} ->+ sqlErrorMessage+ @?= "duckdb-simple: column "+ <> column+ <> " has type "+ <> sqlName+ <> " but expected Variant: DuckDB casts JSON to VARIANT by parsing the text; write "+ <> column+ <> "::VARIANT"+ Right rows -> assertFailure ("expected Incompatible for " <> Text.unpack expression <> ", got " <> show rows)+ refused "j" "'{\"a\": 1}'::JSON" "JSON"+ refused "l" "['{\"a\": 1}']::JSON[]" "JSON[]"+ refused "s" "{'k': '{\"a\": 1}'::JSON}" "STRUCT(k JSON)"+ refused "m" "MAP {'k': '{\"a\": 1}'::JSON}" "MAP(VARCHAR, JSON)"+ (query_ conn "SELECT j::VARIANT FROM (SELECT '{\"a\": 1}'::JSON AS j)" :: IO [Only Variant])+ >>= (@?= [Only (Variant (variantObject [("a", FieldWord64 1)]))])+ (query_ conn "SELECT l::VARIANT FROM (SELECT ['{\"a\": 1}']::JSON[] AS l)" :: IO [Only Variant])+ >>= (@?= [Only (Variant (FieldList [variantObject [("a", FieldWord64 1)]]))])+ , testCase "a Variant reads only a value that a VARIANT can hold" $ withConnection ":memory:" \conn -> do+ -- The cast to VARIANT changes a value of another shape, so the read+ -- raises the error of an incompatible column at each depth. The cast in+ -- SQL reads.+ forM_+ [ ("'a'::ENUM('a', 'b')", "ENUM('a', 'b')")+ , ("MAP {'k': 1}", "MAP(VARCHAR, INTEGER)")+ , ("union_value(n := 1)", "UNION(n INTEGER)")+ , ("{'b': 1, 'a': 2}", "STRUCT(b INTEGER, a INTEGER)")+ , ("[1, 2]::INTEGER[2]", "INTEGER[2]")+ , ("row(1, 'x')", "TUPLE(INTEGER, VARCHAR)")+ , ("'POINT (1 2)'::GEOMETRY('OGC:CRS84')", "GEOMETRY('OGC:CRS84')")+ , ("['a'::ENUM('a', 'b')]", "ENUM('a', 'b')[]")+ , ("[{'a': 1}]", "STRUCT(a INTEGER)[]")+ , ("{'v': 1::VARIANT, 'm': MAP {'k': 1}}", "STRUCT(v VARIANT, m MAP(VARCHAR, INTEGER))")+ , ("{'b': 1::VARIANT, 'a': 2::VARIANT}", "STRUCT(b VARIANT, a VARIANT)")+ , ("{'é': 1::VARIANT, 'z': 2::VARIANT}", "STRUCT(\"é\" VARIANT, z VARIANT)")+ ]+ \(expression, sqlName) -> do+ result <- try (query_ conn (Query ("SELECT " <> expression <> " AS c")) :: IO [Only Variant])+ case result of+ Left SQLError{sqlErrorMessage} ->+ sqlErrorMessage+ @?= "duckdb-simple: column c has type "+ <> sqlName+ <> " but expected Variant: a VARIANT does not hold a value of this shape; write c::VARIANT"+ Right rows -> assertFailure ("expected Incompatible for " <> Text.unpack expression <> ", got " <> show rows)+ (query_ conn (Query ("SELECT (" <> expression <> ")::VARIANT")) :: IO [Only Variant]) >>= \rows -> length rows @?= 1+ ( query_ conn "SELECT ('a'::ENUM('a', 'b'))::VARIANT, (MAP {'k': 1})::VARIANT, {'b': 1, 'a': 2}::VARIANT, ([1, 2]::INTEGER[2])::VARIANT"+ :: IO [(Variant, Variant, Variant, Variant)]+ )+ >>= ( @?=+ [+ ( Variant (FieldText "a")+ , Variant (FieldList [variantObject [("key", FieldText "k"), ("value", FieldInt32 1)]])+ , Variant (variantObject [("a", FieldInt32 2), ("b", FieldInt32 1)])+ , Variant (FieldList [FieldInt32 1, FieldInt32 2])+ )+ ]+ )+ -- A value of the VARIANT shape reads: a scalar, a LIST, a VARIANT in a+ -- LIST or in a STRUCT field, a STRUCT with VARIANT fields and keys in+ -- byte order, and a VARIANT object whose keys were given out of order.+ -- Each equals its cast.+ forM_+ [ "7::INTEGER"+ , "[1, 2]"+ , "[1::VARIANT, 'x'::VARIANT]"+ , "{'v': 1::VARIANT}"+ , "{'a': 1::VARIANT, 'b': 2::VARIANT}"+ , "{'B': 1::VARIANT, 'a': NULL::VARIANT, 'z': 2::VARIANT, 'é': 3::VARIANT}"+ , "{'é': 3, 'b': 1, 'Z': 2}::VARIANT"+ , "{'a': [1, 2]}::VARIANT"+ , "'POINT (1 2)'::GEOMETRY"+ ]+ \expression ->+ (query_ conn (Query ("SELECT " <> expression <> ", (" <> expression <> ")::VARIANT")) :: IO [(Variant, Variant)])+ >>= \case+ [(value, cast)] -> assertEqual (Text.unpack expression) cast value+ rows -> assertFailure (show rows)+ -- The cast of a GEOMETRY with a CRS drops the CRS.+ [(withCrs, wkb)] <-+ query_ conn "SELECT ('POINT (1 2)'::GEOMETRY('OGC:CRS84'))::VARIANT, ST_AsWKB('POINT (1 2)'::GEOMETRY)" :: IO [(Variant, BS.ByteString)]+ withCrs @?= geometryPayload wkb+ , testCase "geometry payload retains empty layout tags and native NaN points" $ withConnection ":memory:" \conn ->+ forM_ ["GEOMETRYCOLLECTION ZM EMPTY", "MULTIPOLYGON M EMPTY", "POINT Z (NaN NaN 7)"] \wkt -> do+ [Only geometry] <- query conn "SELECT ?::GEOMETRY" (Only (wkt :: Text)) :: IO [Only RawGeometry]+ (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB geometry)) :: IO [Only Variant])+ >>= (@?= [Only (geometryPayload (rawGeometryWKB geometry))])+ , testCase "geometry payload conversion ignores user macros with built-in names" $ withConnection ":memory:" \conn -> do+ [Only geometry] <- query_ conn "SELECT 'POINT (1 2)'::GEOMETRY" :: IO [Only RawGeometry]+ _ <- execute_ conn "CREATE MACRO ST_AsText(x) AS 42"+ _ <- execute_ conn "CREATE MACRO ST_GeomFromWKB(x) AS 'POINT (9 9)'::GEOMETRY"+ (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB geometry)) :: IO [Only Variant])+ >>= (@?= [Only (geometryPayload (rawGeometryWKB geometry))])+ , testCase "binary geometry preserves NaN bits and rejects mixed member layouts" $ withConnection ":memory:" \conn -> do+ let nanPoint = G.PointGeometry (G.PointXYZ (G.XYZ (castWord64ToDouble 0x7ff8000000000001) (castWord64ToDouble 0x7ff8000000000002) 7))+ raw <- either assertFailure pure (toRawGeometry nanPoint)+ (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB raw)) :: IO [Only Variant])+ >>= (@?= [Only (geometryPayload (rawGeometryWKB raw))])+ let mixed = G.GeometryCollection (V.fromList [G.PointGeometry (G.PointXY (G.XY 1 2)), G.PointGeometry (G.PointXYZ (G.XYZ 3 4 5))])+ mixedRaw <- either assertFailure pure (toRawGeometry mixed)+ outcome <- try (query conn "SELECT system.main.ST_GeomFromWKB(?)::VARIANT" (Only (rawGeometryWKB mixedRaw)) :: IO [Only Variant])+ case outcome of+ Left (err :: SQLError) -> sqlErrorType err @?= Just ErrorInputInvalid+ Right _ -> assertFailure "native GEOMETRY must reject mixed Z/M dimensions"+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ , testCase "invalid values fail before binding and connection reuse succeeds" $ withConnection ":memory:" \conn -> do+ forM_+ [ FieldHugeInt (2 ^ (127 :: Int))+ , FieldUHugeInt (-1)+ , FieldUHugeInt (2 ^ (128 :: Int))+ , FieldDecimal (DecimalValue 0 0 0)+ , FieldDecimal (DecimalValue 5 6 1)+ , FieldDecimal (DecimalValue 3 0 1000)+ , FieldBit (BitString 8 (BS.singleton 0))+ , variantObject [("same", FieldNull), ("same", FieldBool True)]+ , variantObject [("", FieldBool True)]+ , variantObject [("before\0after", FieldNull)]+ ]+ \value -> assertFailureIO (query conn "SELECT ?" (Only (Variant value)) :: IO [Only Variant])+ (query_ conn "SELECT 42" :: IO [Only Int64]) >>= (@?= [Only 42])+ ] -- | Bind a payload as VARIANT without a cast and read it back. roundTrip :: Connection -> FieldValue -> Assertion roundTrip conn expected =- (query conn "SELECT typeof(?) AS kind, ? AS value" (Variant expected, Variant expected) :: IO [(Text, Variant)]) >>= (@?= [("VARIANT", Variant expected)])--{- | Insert native timestamp units without the parameter API, which converts-TIMESTAMP_S and TIMESTAMP_MS parameters to microseconds before execution.--}-appendWideTimestamps :: Connection -> [(Int64, Int64)] -> IO ()-appendWideTimestamps conn rows =- withConnectionHandle conn \native ->- withCString "wide_timestamps" \table ->- alloca \appenderPtr -> do- poke appenderPtr nullPtr- bracket (c_duckdb_appender_create native nullPtr table appenderPtr) (const (void (c_duckdb_appender_destroy appenderPtr))) \created -> do- created @?= DuckDBSuccess- appender <- peek appenderPtr- forM_ rows \(seconds, millis) -> do- bracket (c_duckdb_create_timestamp_s (DuckDBTimestampS seconds)) destroyValue (c_duckdb_append_value appender) >>= (@?= DuckDBSuccess)- bracket (c_duckdb_create_timestamp_ms (DuckDBTimestampMs millis)) destroyValue (c_duckdb_append_value appender) >>= (@?= DuckDBSuccess)- c_duckdb_appender_end_row appender >>= (@?= DuckDBSuccess)- c_duckdb_appender_flush appender >>= (@?= DuckDBSuccess)+ (query conn "SELECT typeof(?) AS kind, ? AS value" (Variant expected, Variant expected) :: IO [(Text, Variant)])+ >>= (@?= [("VARIANT", Variant expected)]) -- | A raw geometry payload has no CRS inside a VARIANT. geometryPayload :: BS.ByteString -> Variant@@ -285,58 +648,61 @@ -- | SQL constructors for each scalar payload tag. scalarCases :: [Text] scalarCases =- [ "NULL"- , "TRUE"- , "FALSE"- , "'-128'::TINYINT"- , "'-32768'::SMALLINT"- , "'-2147483648'::INTEGER"- , "9007199254740993::BIGINT"- , "'-170141183460469231731687303715884105728'::HUGEINT"- , "255::UTINYINT"- , "65535::USMALLINT"- , "4294967295::UINTEGER"- , "18446744073709551615::UBIGINT"- , "'340282366920938463463374607431768211455'::UHUGEINT"- , "1.25::FLOAT"- , "1.25::DOUBLE"- , "12.34::DECIMAL(4,2)"- , "-1234.56::DECIMAL(9,2)"- , "123456789012.345::DECIMAL(18,3)"- , "1234567890.123456789::DECIMAL(38,9)"- , "'text λ'::VARCHAR"- , "'\\x00\\xFF'::BLOB"- , "'01234567-89ab-cdef-fedc-ba9876543210'::UUID"- , "DATE '1970-01-02'"- , "DATE 'infinity'"- , "DATE '-infinity'"- , "TIME '01:02:03.456789'"- , "'01:02:03.456789012'::TIME_NS"- , "'1970-01-01 00:00:01'::TIMESTAMP_S"- , "'1969-12-31 23:59:59.999'::TIMESTAMP_MS"- , "'1970-01-01 00:00:01.234567'::TIMESTAMP"- , "'1970-01-01 00:00:01.234567891'::TIMESTAMP_NS"- , "'infinity'::TIMESTAMP_NS"- , "'1970-01-01 00:00:01.234567+00'::TIMESTAMPTZ"- , "INTERVAL '-4 MONTHS 8 DAYS 123456789 MICROSECONDS'"- , "'12345678901234567890123456789012345678901234567890'::BIGNUM"- , "'-12345678901234567890123456789012345678901234567890'::BIGNUM"- , "0::BIGNUM"- , "'10101'::BIT"- ]+ [ "NULL"+ , "TRUE"+ , "FALSE"+ , "'-128'::TINYINT"+ , "'-32768'::SMALLINT"+ , "'-2147483648'::INTEGER"+ , "9007199254740993::BIGINT"+ , "'-9223372036854775808'::BIGINT"+ , "'9223372036854775807'::BIGINT"+ , "'-170141183460469231731687303715884105728'::HUGEINT"+ , "255::UTINYINT"+ , "65535::USMALLINT"+ , "4294967295::UINTEGER"+ , "18446744073709551615::UBIGINT"+ , "'340282366920938463463374607431768211455'::UHUGEINT"+ , "1.25::FLOAT"+ , "1.25::DOUBLE"+ , "12.34::DECIMAL(4,2)"+ , "-1234.56::DECIMAL(9,2)"+ , "123456789012.345::DECIMAL(18,3)"+ , "1234567890.123456789::DECIMAL(38,9)"+ , "'text λ'::VARCHAR"+ , "'\\x00\\xFF'::BLOB"+ , "'01234567-89ab-cdef-fedc-ba9876543210'::UUID"+ , "DATE '1970-01-02'"+ , "DATE 'infinity'"+ , "DATE '-infinity'"+ , "TIME '01:02:03.456789'"+ , "'01:02:03.456789012'::TIME_NS"+ , "'1970-01-01 00:00:01'::TIMESTAMP_S"+ , "'1969-12-31 23:59:59.999'::TIMESTAMP_MS"+ , "'1970-01-01 00:00:01.234567'::TIMESTAMP"+ , "'1970-01-01 00:00:01.234567891'::TIMESTAMP_NS"+ , "'infinity'::TIMESTAMP_NS"+ , "'1970-01-01 00:00:01.234567+00'::TIMESTAMPTZ"+ , "'1970-01-01 00:00:01.234567891+00'::TIMESTAMPTZ_NS"+ , "INTERVAL '-4 MONTHS 8 DAYS 123456789 MICROSECONDS'"+ , "'12345678901234567890123456789012345678901234567890'::BIGNUM"+ , "'-12345678901234567890123456789012345678901234567890'::BIGNUM"+ , "0::BIGNUM"+ , "'10101'::BIT"+ ] -- | Reserve a unique path and let DuckDB create its file. newDatabase :: IO FilePath newDatabase = do- directory <- getTemporaryDirectory- (path, handle) <- openBinaryTempFile directory "duckdb-variant"- hClose handle- removeFile path- pure path+ directory <- getTemporaryDirectory+ (path, handle) <- openBinaryTempFile directory "duckdb-variant"+ hClose handle+ removeFile path+ pure path -- | Remove files created by this test after DuckDB closes them. removeDatabase :: FilePath -> IO () removeDatabase path =- forM_ [path, path <> ".wal"] \file -> do- exists <- doesFileExist file- when exists (removeFile file)+ forM_ [path, path <> ".wal"] \file -> do+ exists <- doesFileExist file+ when exists (removeFile file)