bloodhound-0.19.1.0: src/Database/Bloodhound/Client.hs
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE RecordWildCards #-}
{-# LANGUAGE TupleSections #-}
-------------------------------------------------------------------------------
-- |
-- Module : Database.Bloodhound.Client
-- Copyright : (C) 2014, 2018 Chris Allen
-- License : BSD-style (see the file LICENSE)
-- Maintainer : Chris Allen <cma@bitemyapp.com>
-- Stability : provisional
-- Portability : OverloadedStrings
--
-- Client side functions for talking to Elasticsearch servers.
--
-------------------------------------------------------------------------------
module Database.Bloodhound.Client
( -- * Bloodhound client functions
-- | The examples in this module assume the following code has been run.
-- The :{ and :} will only work in GHCi. You'll only need the data types
-- and typeclass instances for the functions that make use of them.
-- $setup
withBH
-- ** Indices
, createIndex
, createIndexWith
, flushIndex
, deleteIndex
, updateIndexSettings
, getIndexSettings
, forceMergeIndex
, indexExists
, openIndex
, closeIndex
, listIndices
, catIndices
, waitForYellowIndex
-- *** Index Aliases
, updateIndexAliases
, getIndexAliases
, deleteIndexAlias
-- *** Index Templates
, putTemplate
, templateExists
, deleteTemplate
-- ** Mapping
, putMapping
-- ** Documents
, indexDocument
, updateDocument
, getDocument
, documentExists
, deleteDocument
, deleteByQuery
-- ** Searching
, searchAll
, searchByIndex
, searchByIndices
, searchByIndexTemplate
, searchByIndicesTemplate
, scanSearch
, getInitialScroll
, getInitialSortedScroll
, advanceScroll
, pitSearch
, openPointInTime
, closePointInTime
, refreshIndex
, mkSearch
, mkAggregateSearch
, mkHighlightSearch
, mkSearchTemplate
, bulk
, pageSearch
, mkShardCount
, mkReplicaCount
, getStatus
-- ** Templates
, storeSearchTemplate
, getSearchTemplate
, deleteSearchTemplate
-- ** Snapshot/Restore
-- *** Snapshot Repos
, getSnapshotRepos
, updateSnapshotRepo
, verifySnapshotRepo
, deleteSnapshotRepo
-- *** Snapshots
, createSnapshot
, getSnapshots
, deleteSnapshot
-- *** Restoring Snapshots
, restoreSnapshot
-- ** Nodes
, getNodesInfo
, getNodesStats
-- ** Request Utilities
, encodeBulkOperations
, encodeBulkOperation
-- * Authentication
, basicAuthHook
-- * Reply-handling tools
, isVersionConflict
, isSuccess
, isCreated
, parseEsResponse
-- * Count
, countByIndex
)
where
import qualified Blaze.ByteString.Builder as BB
import Control.Applicative as A
import Control.Monad
import Control.Monad.Catch
import Control.Monad.IO.Class
import Data.Aeson
import Data.Aeson.Key
import qualified Data.Aeson.KeyMap as X
import Data.ByteString.Builder
import qualified Data.ByteString.Lazy.Char8 as L
import Data.Foldable (toList)
import Data.Ix
import qualified Data.List as LS (foldl')
import Data.List.NonEmpty (NonEmpty (..))
import Data.Maybe (catMaybes, fromMaybe)
import Data.Monoid
import Data.Text (Text)
import qualified Data.Text as T
import qualified Data.Text.Encoding as T
import Data.Time.Clock
import qualified Data.Vector as V
import Network.HTTP.Client
import qualified Network.HTTP.Types.Method as NHTM
import qualified Network.HTTP.Types.Status as NHTS
import qualified Network.HTTP.Types.URI as NHTU
import qualified Network.URI as URI
import Prelude hiding (filter, head)
import Database.Bloodhound.Types
-- $setup
-- >>> :set -XOverloadedStrings
-- >>> :set -XDeriveGeneric
-- >>> import Database.Bloodhound
-- >>> import Network.HTTP.Client
-- >>> let testServer = (Server "http://localhost:9200")
-- >>> let runBH' = withBH defaultManagerSettings testServer
-- >>> let testIndex = IndexName "twitter"
-- >>> let defaultIndexSettings = IndexSettings (ShardCount 1) (ReplicaCount 0)
-- >>> data TweetMapping = TweetMapping deriving (Eq, Show)
-- >>> _ <- runBH' $ deleteIndex testIndex
-- >>> _ <- runBH' $ deleteIndex (IndexName "didimakeanindex")
-- >>> import GHC.Generics
-- >>> import Data.Time.Calendar (Day (..))
-- >>> import Data.Time.Clock (UTCTime (..), secondsToDiffTime)
-- >>> :{
--instance ToJSON TweetMapping where
-- toJSON TweetMapping =
-- object ["properties" .=
-- object ["location" .=
-- object ["type" .= ("geo_point" :: Text)]]]
--data Location = Location { lat :: Double
-- , lon :: Double } deriving (Eq, Generic, Show)
--data Tweet = Tweet { user :: Text
-- , postDate :: UTCTime
-- , message :: Text
-- , age :: Int
-- , location :: Location } deriving (Eq, Generic, Show)
--exampleTweet = Tweet { user = "bitemyapp"
-- , postDate = UTCTime
-- (ModifiedJulianDay 55000)
-- (secondsToDiffTime 10)
-- , message = "Use haskell!"
-- , age = 10000
-- , location = Location 40.12 (-71.34) }
--instance ToJSON Tweet where
-- toJSON = genericToJSON defaultOptions
--instance FromJSON Tweet where
-- parseJSON = genericParseJSON defaultOptions
--instance ToJSON Location where
-- toJSON = genericToJSON defaultOptions
--instance FromJSON Location where
-- parseJSON = genericParseJSON defaultOptions
--data BulkTest = BulkTest { name :: Text } deriving (Eq, Generic, Show)
--instance FromJSON BulkTest where
-- parseJSON = genericParseJSON defaultOptions
--instance ToJSON BulkTest where
-- toJSON = genericToJSON defaultOptions
-- :}
-- | 'mkShardCount' is a straight-forward smart constructor for 'ShardCount'
-- which rejects 'Int' values below 1 and above 1000.
--
-- >>> mkShardCount 10
-- Just (ShardCount 10)
mkShardCount :: Int -> Maybe ShardCount
mkShardCount n
| n < 1 = Nothing
| n > 1000 = Nothing
| otherwise = Just (ShardCount n)
-- | 'mkReplicaCount' is a straight-forward smart constructor for 'ReplicaCount'
-- which rejects 'Int' values below 0 and above 1000.
--
-- >>> mkReplicaCount 10
-- Just (ReplicaCount 10)
mkReplicaCount :: Int -> Maybe ReplicaCount
mkReplicaCount n
| n < 0 = Nothing
| n > 1000 = Nothing -- ...
| otherwise = Just (ReplicaCount n)
emptyBody :: L.ByteString
emptyBody = L.pack ""
dispatch :: MonadBH m
=> Method
-> Text
-> Maybe L.ByteString
-> m Reply
dispatch dMethod url body = do
initReq <- liftIO $ parseUrl' url
reqHook <- bhRequestHook A.<$> getBHEnv
let reqBody = RequestBodyLBS $ fromMaybe emptyBody body
req <- liftIO
$ reqHook
$ setRequestIgnoreStatus
$ initReq { method = dMethod
, requestHeaders =
-- "application/x-ndjson" for bulk
("Content-Type", "application/json") : requestHeaders initReq
, requestBody = reqBody }
-- req <- liftIO $ reqHook $ setRequestIgnoreStatus $ initReq { method = dMethod
-- , requestBody = reqBody }
mgr <- bhManager <$> getBHEnv
liftIO $ httpLbs req mgr
joinPath' :: [Text] -> Text
joinPath' = T.intercalate "/"
joinPath :: MonadBH m => [Text] -> m Text
joinPath ps = do
Server s <- bhServer <$> getBHEnv
return $ joinPath' (s:ps)
appendSearchTypeParam :: Text -> SearchType -> Text
appendSearchTypeParam originalUrl st = addQuery params originalUrl
where stText = "search_type"
params
| st == SearchTypeDfsQueryThenFetch = [(stText, Just "dfs_query_then_fetch")]
-- used to catch 'SearchTypeQueryThenFetch', which is also the default
| otherwise = []
-- | Severely dumbed down query renderer. Assumes your data doesn't
-- need any encoding
addQuery :: [(Text, Maybe Text)] -> Text -> Text
addQuery q u = u <> rendered
where
rendered =
T.decodeUtf8 $ BB.toByteString $ NHTU.renderQueryText prependQuestionMark q
prependQuestionMark = True
bindM2 :: (Applicative m, Monad m) => (a -> b -> m c) -> m a -> m b -> m c
bindM2 f ma mb = join (f <$> ma <*> mb)
-- | Convenience function that sets up a manager and BHEnv and runs
-- the given set of bloodhound operations. Connections will be
-- pipelined automatically in accordance with the given manager
-- settings in IO. If you've got your own monad transformer stack, you
-- should use 'runBH' directly.
withBH :: ManagerSettings -> Server -> BH IO a -> IO a
withBH ms s f = do
mgr <- newManager ms
let env = mkBHEnv s mgr
runBH env f
-- Shortcut functions for HTTP methods
delete :: MonadBH m => Text -> m Reply
delete = flip (dispatch NHTM.methodDelete) Nothing
delete' :: MonadBH m => Text -> Maybe L.ByteString -> m Reply
delete' = dispatch NHTM.methodDelete
get :: MonadBH m => Text -> m Reply
get = flip (dispatch NHTM.methodGet) Nothing
head :: MonadBH m => Text -> m Reply
head = flip (dispatch NHTM.methodHead) Nothing
put :: MonadBH m => Text -> Maybe L.ByteString -> m Reply
put = dispatch NHTM.methodPut
post :: MonadBH m => Text -> Maybe L.ByteString -> m Reply
post = dispatch NHTM.methodPost
-- indexDocument s ix name doc = put (root </> s </> ix </> name </> doc) (Just encode doc)
-- http://hackage.haskell.org/package/http-client-lens-0.1.0/docs/Network-HTTP-Client-Lens.html
-- https://github.com/supki/libjenkins/blob/master/src/Jenkins/Rest/Internal.hs
-- | 'getStatus' fetches the 'Status' of a 'Server'
--
-- >>> serverStatus <- runBH' getStatus
-- >>> fmap tagline (serverStatus)
-- Just "You Know, for Search"
getStatus :: MonadBH m => m (Maybe Status)
getStatus = do
response <- get =<< url
return $ decode (responseBody response)
where url = joinPath []
-- | 'getSnapshotRepos' gets the definitions of a subset of the
-- defined snapshot repos.
getSnapshotRepos
:: ( MonadBH m
, MonadThrow m
)
=> SnapshotRepoSelection
-> m (Either EsError [GenericSnapshotRepo])
getSnapshotRepos sel = fmap (fmap unGSRs) . parseEsResponse =<< get =<< url
where
url = joinPath ["_snapshot", selectorSeg]
selectorSeg = case sel of
AllSnapshotRepos -> "_all"
SnapshotRepoList (p :| ps) -> T.intercalate "," (renderPat <$> (p:ps))
renderPat (RepoPattern t) = t
renderPat (ExactRepo (SnapshotRepoName t)) = t
-- | Wrapper to extract the list of 'GenericSnapshotRepo' in the
-- format they're returned in
newtype GSRs = GSRs { unGSRs :: [GenericSnapshotRepo] }
instance FromJSON GSRs where
parseJSON = withObject "Collection of GenericSnapshotRepo" parse
where
parse = fmap GSRs . mapM (uncurry go) . X.toList
go rawName = withObject "GenericSnapshotRepo" $ \o ->
GenericSnapshotRepo (SnapshotRepoName $ toText rawName) <$> o .: "type"
<*> o .: "settings"
-- | Create or update a snapshot repo
updateSnapshotRepo
:: ( MonadBH m
, SnapshotRepo repo
)
=> SnapshotRepoUpdateSettings
-- ^ Use 'defaultSnapshotRepoUpdateSettings' if unsure
-> repo
-> m Reply
updateSnapshotRepo SnapshotRepoUpdateSettings {..} repo =
bindM2 put url (return (Just body))
where
url = addQuery params <$> joinPath ["_snapshot", snapshotRepoName gSnapshotRepoName]
params
| repoUpdateVerify = []
| otherwise = [("verify", Just "false")]
body = encode $ object [ "type" .= gSnapshotRepoType
, "settings" .= gSnapshotRepoSettings
]
GenericSnapshotRepo {..} = toGSnapshotRepo repo
-- | Verify if a snapshot repo is working. __NOTE:__ this API did not
-- make it into Elasticsearch until 1.4. If you use an older version,
-- you will get an error here.
verifySnapshotRepo
:: ( MonadBH m
, MonadThrow m
)
=> SnapshotRepoName
-> m (Either EsError SnapshotVerification)
verifySnapshotRepo (SnapshotRepoName n) =
parseEsResponse =<< bindM2 post url (return Nothing)
where
url = joinPath ["_snapshot", n, "_verify"]
deleteSnapshotRepo :: MonadBH m => SnapshotRepoName -> m Reply
deleteSnapshotRepo (SnapshotRepoName n) = delete =<< url
where
url = joinPath ["_snapshot", n]
-- | Create and start a snapshot
createSnapshot
:: (MonadBH m)
=> SnapshotRepoName
-> SnapshotName
-> SnapshotCreateSettings
-> m Reply
createSnapshot (SnapshotRepoName repoName)
(SnapshotName snapName)
SnapshotCreateSettings {..} =
bindM2 put url (return (Just body))
where
url = addQuery params <$> joinPath ["_snapshot", repoName, snapName]
params = [("wait_for_completion", Just (boolQP snapWaitForCompletion))]
body = encode $ object prs
prs = catMaybes [ ("indices" .=) . indexSelectionName <$> snapIndices
, Just ("ignore_unavailable" .= snapIgnoreUnavailable)
, Just ("ignore_global_state" .= snapIncludeGlobalState)
, Just ("partial" .= snapPartial)
]
indexSelectionName :: IndexSelection -> Text
indexSelectionName AllIndexes = "_all"
indexSelectionName (IndexList (i :| is)) = T.intercalate "," (renderIndex <$> (i:is))
where
renderIndex (IndexName n) = n
-- | Get info about known snapshots given a pattern and repo name.
getSnapshots
:: ( MonadBH m
, MonadThrow m
)
=> SnapshotRepoName
-> SnapshotSelection
-> m (Either EsError [SnapshotInfo])
getSnapshots (SnapshotRepoName repoName) sel =
fmap (fmap unSIs) . parseEsResponse =<< get =<< url
where
url = joinPath ["_snapshot", repoName, snapPath]
snapPath = case sel of
AllSnapshots -> "_all"
SnapshotList (s :| ss) -> T.intercalate "," (renderPath <$> (s:ss))
renderPath (SnapPattern t) = t
renderPath (ExactSnap (SnapshotName t)) = t
newtype SIs = SIs { unSIs :: [SnapshotInfo] }
instance FromJSON SIs where
parseJSON = withObject "Collection of SnapshotInfo" parse
where
parse o = SIs <$> o .: "snapshots"
-- | Delete a snapshot. Cancels if it is running.
deleteSnapshot :: MonadBH m => SnapshotRepoName -> SnapshotName -> m Reply
deleteSnapshot (SnapshotRepoName repoName) (SnapshotName snapName) =
delete =<< url
where
url = joinPath ["_snapshot", repoName, snapName]
-- | Restore a snapshot to the cluster See
-- <https://www.elastic.co/guide/en/elasticsearch/reference/1.7/modules-snapshots.html#_restore>
-- for more details.
restoreSnapshot
:: MonadBH m
=> SnapshotRepoName
-> SnapshotName
-> SnapshotRestoreSettings
-- ^ Start with 'defaultSnapshotRestoreSettings' and customize
-- from there for reasonable defaults.
-> m Reply
restoreSnapshot (SnapshotRepoName repoName)
(SnapshotName snapName)
SnapshotRestoreSettings {..} = bindM2 post url (return (Just body))
where
url = addQuery params <$> joinPath ["_snapshot", repoName, snapName, "_restore"]
params = [("wait_for_completion", Just (boolQP snapRestoreWaitForCompletion))]
body = encode (object prs)
prs = catMaybes [ ("indices" .=) . indexSelectionName <$> snapRestoreIndices
, Just ("ignore_unavailable" .= snapRestoreIgnoreUnavailable)
, Just ("include_global_state" .= snapRestoreIncludeGlobalState)
, ("rename_pattern" .=) <$> snapRestoreRenamePattern
, ("rename_replacement" .=) . renderTokens <$> snapRestoreRenameReplacement
, Just ("include_aliases" .= snapRestoreIncludeAliases)
, ("index_settings" .= ) <$> snapRestoreIndexSettingsOverrides
, ("ignore_index_settings" .= ) <$> snapRestoreIgnoreIndexSettings
]
renderTokens (t :| ts) = mconcat (renderToken <$> (t:ts))
renderToken (RRTLit t) = t
renderToken RRSubWholeMatch = "$0"
renderToken (RRSubGroup g) = T.pack (show (rrGroupRefNum g))
getNodesInfo
:: ( MonadBH m
, MonadThrow m
)
=> NodeSelection
-> m (Either EsError NodesInfo)
getNodesInfo sel = parseEsResponse =<< get =<< url
where
url = joinPath ["_nodes", selectionSeg]
selectionSeg = case sel of
LocalNode -> "_local"
NodeList (l :| ls) -> T.intercalate "," (selToSeg <$> (l:ls))
AllNodes -> "_all"
selToSeg (NodeByName (NodeName n)) = n
selToSeg (NodeByFullNodeId (FullNodeId i)) = i
selToSeg (NodeByHost (Server s)) = s
selToSeg (NodeByAttribute (NodeAttrName a) v) = a <> ":" <> v
getNodesStats
:: ( MonadBH m
, MonadThrow m
)
=> NodeSelection
-> m (Either EsError NodesStats)
getNodesStats sel = parseEsResponse =<< get =<< url
where
url = joinPath ["_nodes", selectionSeg, "stats"]
selectionSeg = case sel of
LocalNode -> "_local"
NodeList (l :| ls) -> T.intercalate "," (selToSeg <$> (l:ls))
AllNodes -> "_all"
selToSeg (NodeByName (NodeName n)) = n
selToSeg (NodeByFullNodeId (FullNodeId i)) = i
selToSeg (NodeByHost (Server s)) = s
selToSeg (NodeByAttribute (NodeAttrName a) v) = a <> ":" <> v
-- | 'createIndex' will create an index given a 'Server', 'IndexSettings', and an 'IndexName'.
--
-- >>> response <- runBH' $ createIndex defaultIndexSettings (IndexName "didimakeanindex")
-- >>> respIsTwoHunna response
-- True
-- >>> runBH' $ indexExists (IndexName "didimakeanindex")
-- True
createIndex :: MonadBH m => IndexSettings -> IndexName -> m Reply
createIndex indexSettings (IndexName indexName) =
bindM2 put url (return body)
where url = joinPath [indexName]
body = Just $ encode indexSettings
-- | Create an index, providing it with any number of settings. This
-- is more expressive than 'createIndex' but makes is more verbose
-- for the common case of configuring only the shard count and
-- replica count.
createIndexWith :: MonadBH m
=> [UpdatableIndexSetting]
-> Int -- ^ shard count
-> IndexName
-> m Reply
createIndexWith updates shards (IndexName indexName) =
bindM2 put url (return (Just body))
where url = joinPath [indexName]
body = encode $ object
["settings" .= deepMerge
( X.singleton "index.number_of_shards" (toJSON shards) :
[u | Object u <- toJSON <$> updates]
)
]
-- | 'flushIndex' will flush an index given a 'Server' and an 'IndexName'.
flushIndex :: MonadBH m => IndexName -> m Reply
flushIndex (IndexName indexName) = do
path <- joinPath [indexName, "_flush"]
post path Nothing
-- | 'deleteIndex' will delete an index given a 'Server' and an 'IndexName'.
--
-- >>> _ <- runBH' $ createIndex defaultIndexSettings (IndexName "didimakeanindex")
-- >>> response <- runBH' $ deleteIndex (IndexName "didimakeanindex")
-- >>> respIsTwoHunna response
-- True
-- >>> runBH' $ indexExists (IndexName "didimakeanindex")
-- False
deleteIndex :: MonadBH m => IndexName -> m Reply
deleteIndex (IndexName indexName) =
delete =<< joinPath [indexName]
-- | 'updateIndexSettings' will apply a non-empty list of setting updates to an index
--
-- >>> _ <- runBH' $ createIndex defaultIndexSettings (IndexName "unconfiguredindex")
-- >>> response <- runBH' $ updateIndexSettings (BlocksWrite False :| []) (IndexName "unconfiguredindex")
-- >>> respIsTwoHunna response
-- True
updateIndexSettings :: MonadBH m => NonEmpty UpdatableIndexSetting -> IndexName -> m Reply
updateIndexSettings updates (IndexName indexName) =
bindM2 put url (return body)
where
url = joinPath [indexName, "_settings"]
body = Just (encode jsonBody)
jsonBody = Object (deepMerge [u | Object u <- toJSON <$> toList updates])
getIndexSettings :: (MonadBH m, MonadThrow m) => IndexName
-> m (Either EsError IndexSettingsSummary)
getIndexSettings (IndexName indexName) =
parseEsResponse =<< get =<< url
where
url = joinPath [indexName, "_settings"]
-- | 'forceMergeIndex'
--
-- The force merge API allows to force merging of one or more indices through
-- an API. The merge relates to the number of segments a Lucene index holds
-- within each shard. The force merge operation allows to reduce the number of
-- segments by merging them.
--
-- This call will block until the merge is complete. If the http connection is
-- lost, the request will continue in the background, and any new requests will
-- block until the previous force merge is complete.
-- For more information see
-- <https://www.elastic.co/guide/en/elasticsearch/reference/current/indices-forcemerge.html#indices-forcemerge>.
-- Nothing
-- worthwhile comes back in the reply body, so matching on the status
-- should suffice.
--
-- 'forceMergeIndex' with a maxNumSegments of 1 and onlyExpungeDeletes
-- to True is the main way to release disk space back to the OS being
-- held by deleted documents.
--
-- >>> let ixn = IndexName "unoptimizedindex"
-- >>> _ <- runBH' $ deleteIndex ixn >> createIndex defaultIndexSettings ixn
-- >>> response <- runBH' $ forceMergeIndex (IndexList (ixn :| [])) (defaultIndexOptimizationSettings { maxNumSegments = Just 1, onlyExpungeDeletes = True })
-- >>> respIsTwoHunna response
-- True
forceMergeIndex :: MonadBH m => IndexSelection -> ForceMergeIndexSettings -> m Reply
forceMergeIndex ixs ForceMergeIndexSettings {..} =
bindM2 post url (return body)
where url = addQuery params <$> joinPath [indexName, "_forcemerge"]
params = catMaybes [ ("max_num_segments",) . Just . showText <$> maxNumSegments
, Just ("only_expunge_deletes", Just (boolQP onlyExpungeDeletes))
, Just ("flush", Just (boolQP flushAfterOptimize))
]
indexName = indexSelectionName ixs
body = Nothing
deepMerge :: [Object] -> Object
deepMerge = LS.foldl' (X.unionWith merge) mempty
where merge (Object a) (Object b) = Object (deepMerge [a, b])
merge _ b = b
statusCodeIs :: (Int, Int) -> Reply -> Bool
statusCodeIs r resp = inRange r $ NHTS.statusCode (responseStatus resp)
respIsTwoHunna :: Reply -> Bool
respIsTwoHunna = statusCodeIs (200, 299)
existentialQuery :: MonadBH m => Text -> m (Reply, Bool)
existentialQuery url = do
reply <- head url
return (reply, respIsTwoHunna reply)
-- | Tries to parse a response body as the expected type @a@ and
-- failing that tries to parse it as an EsError. All well-formed, JSON
-- responses from elasticsearch should fall into these two
-- categories. If they don't, a 'EsProtocolException' will be
-- thrown. If you encounter this, please report the full body it
-- reports along with your Elasticsearch verison.
parseEsResponse :: ( MonadThrow m
, FromJSON a
)
=> Reply
-> m (Either EsError a)
parseEsResponse reply
| respIsTwoHunna reply = case eitherDecode body of
Right a -> return (Right a)
Left err ->
tryParseError err
| otherwise = tryParseError "Non-200 status code"
where body = responseBody reply
tryParseError originalError
= case eitherDecode body of
Right e -> return (Left e)
-- Failed to parse the error message.
Left err -> explode ("Original error was: " <> originalError <> " Error parse failure was: " <> err)
explode errorMsg = throwM (EsProtocolException (T.pack errorMsg) body)
-- | 'indexExists' enables you to check if an index exists. Returns 'Bool'
-- in IO
--
-- >>> exists <- runBH' $ indexExists testIndex
indexExists :: MonadBH m => IndexName -> m Bool
indexExists (IndexName indexName) = do
(_, exists) <- existentialQuery =<< joinPath [indexName]
return exists
-- | 'refreshIndex' will force a refresh on an index. You must
-- do this if you want to read what you wrote.
--
-- >>> _ <- runBH' $ createIndex defaultIndexSettings testIndex
-- >>> _ <- runBH' $ refreshIndex testIndex
refreshIndex :: MonadBH m => IndexName -> m Reply
refreshIndex (IndexName indexName) =
bindM2 post url (return Nothing)
where url = joinPath [indexName, "_refresh"]
-- | Block until the index becomes available for indexing
-- documents. This is useful for integration tests in which
-- indices are rapidly created and deleted.
waitForYellowIndex :: MonadBH m => IndexName -> m Reply
waitForYellowIndex (IndexName indexName) = get =<< url
where url = addQuery q <$> joinPath ["_cluster","health",indexName]
q = [("wait_for_status",Just "yellow"),("timeout",Just "10s")]
stringifyOCIndex :: OpenCloseIndex -> Text
stringifyOCIndex oci = case oci of
OpenIndex -> "_open"
CloseIndex -> "_close"
openOrCloseIndexes :: MonadBH m => OpenCloseIndex -> IndexName -> m Reply
openOrCloseIndexes oci (IndexName indexName) =
bindM2 post url (return Nothing)
where ociString = stringifyOCIndex oci
url = joinPath [indexName, ociString]
-- | 'openIndex' opens an index given a 'Server' and an 'IndexName'. Explained in further detail at
-- <http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/indices-open-close.html>
--
-- >>> reply <- runBH' $ openIndex testIndex
openIndex :: MonadBH m => IndexName -> m Reply
openIndex = openOrCloseIndexes OpenIndex
-- | 'closeIndex' closes an index given a 'Server' and an 'IndexName'. Explained in further detail at
-- <http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/indices-open-close.html>
--
-- >>> reply <- runBH' $ closeIndex testIndex
closeIndex :: MonadBH m => IndexName -> m Reply
closeIndex = openOrCloseIndexes CloseIndex
-- | 'listIndices' returns a list of all index names on a given 'Server'
listIndices :: (MonadThrow m, MonadBH m) => m [IndexName]
listIndices =
parse . responseBody =<< get =<< url
where
url = joinPath ["_cat/indices?format=json"]
parse body = either (\msg -> (throwM (EsProtocolException (T.pack msg) body))) return $ do
vals <- eitherDecode body
forM vals $ \val ->
case val of
Object obj ->
case X.lookup "index" obj of
(Just (String txt)) -> Right (IndexName txt)
v -> Left $ "indexVal in listIndices failed on non-string, was: " <> show v
v -> Left $ "One of the values parsed in listIndices wasn't an object, it was: " <> show v
-- | 'catIndices' returns a list of all index names on a given 'Server' as well as their doc counts
catIndices :: (MonadThrow m, MonadBH m) => m [(IndexName, Int)]
catIndices =
parse . responseBody =<< get =<< url
where
url = joinPath ["_cat/indices?format=json"]
parse body = either (\msg -> (throwM (EsProtocolException (T.pack msg) body))) return $ do
vals <- eitherDecode body
forM vals $ \val ->
case val of
Object obj ->
case (X.lookup "index" obj, X.lookup "docs.count" obj) of
(Just (String txt), Just (String docs)) -> Right ((IndexName txt), read (T.unpack docs))
v -> Left $ "indexVal in catIndices failed on non-string, was: " <> show v
v -> Left $ "One of the values parsed in catIndices wasn't an object, it was: " <> show v
-- | 'updateIndexAliases' updates the server's index alias
-- table. Operations are atomic. Explained in further detail at
-- <https://www.elastic.co/guide/en/elasticsearch/reference/current/indices-aliases.html>
--
-- >>> let src = IndexName "a-real-index"
-- >>> let aliasName = IndexName "an-alias"
-- >>> let iAlias = IndexAlias src (IndexAliasName aliasName)
-- >>> let aliasCreate = IndexAliasCreate Nothing Nothing
-- >>> _ <- runBH' $ deleteIndex src
-- >>> respIsTwoHunna <$> runBH' (createIndex defaultIndexSettings src)
-- True
-- >>> runBH' $ indexExists src
-- True
-- >>> respIsTwoHunna <$> runBH' (updateIndexAliases (AddAlias iAlias aliasCreate :| []))
-- True
-- >>> runBH' $ indexExists aliasName
-- True
updateIndexAliases :: MonadBH m => NonEmpty IndexAliasAction -> m Reply
updateIndexAliases actions = bindM2 post url (return body)
where url = joinPath ["_aliases"]
body = Just (encode bodyJSON)
bodyJSON = object [ "actions" .= toList actions]
-- | Get all aliases configured on the server.
getIndexAliases :: (MonadBH m, MonadThrow m)
=> m (Either EsError IndexAliasesSummary)
getIndexAliases = parseEsResponse =<< get =<< url
where url = joinPath ["_aliases"]
-- | Delete a single alias, removing it from all indices it
-- is currently associated with.
deleteIndexAlias :: MonadBH m => IndexAliasName -> m Reply
deleteIndexAlias (IndexAliasName (IndexName name)) = delete =<< url
where url = joinPath ["_all","_alias",name]
-- | 'putTemplate' creates a template given an 'IndexTemplate' and a 'TemplateName'.
-- Explained in further detail at
-- <https://www.elastic.co/guide/en/elasticsearch/reference/1.7/indices-templates.html>
--
-- >>> let idxTpl = IndexTemplate [IndexPattern "tweet-*"] (Just (IndexSettings (ShardCount 1) (ReplicaCount 1))) [toJSON TweetMapping]
-- >>> resp <- runBH' $ putTemplate idxTpl (TemplateName "tweet-tpl")
putTemplate :: MonadBH m => IndexTemplate -> TemplateName -> m Reply
putTemplate indexTemplate (TemplateName templateName) =
bindM2 put url (return body)
where url = joinPath ["_template", templateName]
body = Just $ encode indexTemplate
-- | 'templateExists' checks to see if a template exists.
--
-- >>> exists <- runBH' $ templateExists (TemplateName "tweet-tpl")
templateExists :: MonadBH m => TemplateName -> m Bool
templateExists (TemplateName templateName) = do
(_, exists) <- existentialQuery =<< joinPath ["_template", templateName]
return exists
-- | 'deleteTemplate' is an HTTP DELETE and deletes a template.
--
-- >>> let idxTpl = IndexTemplate [IndexPattern "tweet-*"] (Just (IndexSettings (ShardCount 1) (ReplicaCount 1))) [toJSON TweetMapping]
-- >>> _ <- runBH' $ putTemplate idxTpl (TemplateName "tweet-tpl")
-- >>> resp <- runBH' $ deleteTemplate (TemplateName "tweet-tpl")
deleteTemplate :: MonadBH m => TemplateName -> m Reply
deleteTemplate (TemplateName templateName) =
delete =<< joinPath ["_template", templateName]
-- | 'putMapping' is an HTTP PUT and has upsert semantics. Mappings are schemas
-- for documents in indexes.
--
-- >>> _ <- runBH' $ createIndex defaultIndexSettings testIndex
-- >>> resp <- runBH' $ putMapping testIndex TweetMapping
-- >>> print resp
-- Response {responseStatus = Status {statusCode = 200, statusMessage = "OK"}, responseVersion = HTTP/1.1, responseHeaders = [("content-type","application/json; charset=UTF-8"),("content-encoding","gzip"),("transfer-encoding","chunked")], responseBody = "{\"acknowledged\":true}", responseCookieJar = CJ {expose = []}, responseClose' = ResponseClose}
putMapping :: (MonadBH m, ToJSON a) => IndexName -> a -> m Reply
putMapping (IndexName indexName) mapping =
bindM2 put url (return body)
where url = joinPath [indexName, "_mapping"]
-- "_mapping" above is originally transposed
-- erroneously. The correct API call is: "/INDEX/_mapping"
body = Just $ encode mapping
versionCtlParams :: IndexDocumentSettings -> [(Text, Maybe Text)]
versionCtlParams cfg =
case idsVersionControl cfg of
NoVersionControl -> []
InternalVersion v -> versionParams v "internal"
ExternalGT (ExternalDocVersion v) -> versionParams v "external_gt"
ExternalGTE (ExternalDocVersion v) -> versionParams v "external_gte"
ForceVersion (ExternalDocVersion v) -> versionParams v "force"
where
vt = showText . docVersionNumber
versionParams v t = [ ("version", Just $ vt v)
, ("version_type", Just t)
]
-- | 'indexDocument' is the primary way to save a single document in
-- Elasticsearch. The document itself is simply something we can
-- convert into a JSON 'Value'. The 'DocId' will function as the
-- primary key for the document. You are encouraged to generate
-- your own id's and not rely on Elasticsearch's automatic id
-- generation. Read more about it here:
-- https://github.com/bitemyapp/bloodhound/issues/107
--
-- >>> resp <- runBH' $ indexDocument testIndex defaultIndexDocumentSettings exampleTweet (DocId "1")
-- >>> print resp
-- Response {responseStatus = Status {statusCode = 200, statusMessage = "OK"}, responseVersion = HTTP/1.1, responseHeaders = [("content-type","application/json; charset=UTF-8"),("content-encoding","gzip"),("content-length","152")], responseBody = "{\"_index\":\"bloodhound-tests-twitter-1\",\"_type\":\"_doc\",\"_id\":\"1\",\"_version\":2,\"result\":\"updated\",\"_shards\":{\"total\":1,\"successful\":1,\"failed\":0},\"_seq_no\":1,\"_primary_term\":1}", responseCookieJar = CJ {expose = []}, responseClose' = ResponseClose}
indexDocument :: (ToJSON doc, MonadBH m) => IndexName
-> IndexDocumentSettings -> doc -> DocId -> m Reply
indexDocument (IndexName indexName) cfg document (DocId docId) =
bindM2 put url (return body)
where url = addQuery (indexQueryString cfg (DocId docId)) <$> joinPath [indexName, "_doc", docId]
jsonBody = encodeDocument cfg document
body = Just (encode jsonBody)
-- | 'updateDocument' provides a way to perform an partial update of a
-- an already indexed document.
updateDocument :: (ToJSON patch, MonadBH m) => IndexName
-> IndexDocumentSettings -> patch -> DocId -> m Reply
updateDocument (IndexName indexName) cfg patch (DocId docId) =
bindM2 post url (return body)
where url = addQuery (indexQueryString cfg (DocId docId)) <$>
joinPath [indexName, "_update", docId]
jsonBody = encodeDocument cfg patch
body = Just (encode $ object ["doc" .= jsonBody])
{- From ES docs:
Parent and child documents must be indexed on the same shard.
This means that the same routing value needs to be provided when getting, deleting, or updating a child document.
Parent/Child support in Bloodhound requires MUCH more love.
To work it around for now (and to support the existing unit test) we route "parent" documents to their "_id"
(which is the default strategy for the ES), and route all child documents to their parens' "_id"
However, it may not be flexible enough for some corner cases.
Buld operations are completely unaware of "routing" and are probably broken in that matter.
Or perhaps they always were, because the old "_parent" would also have this requirement.
-}
indexQueryString :: IndexDocumentSettings -> DocId -> [(Text, Maybe Text)]
indexQueryString cfg (DocId docId) =
versionCtlParams cfg <> routeParams
where
routeParams = case idsJoinRelation cfg of
Nothing -> []
Just (ParentDocument _ _) -> [("routing", Just docId)]
Just (ChildDocument _ _ (DocId pid)) -> [("routing", Just pid)]
encodeDocument :: ToJSON doc => IndexDocumentSettings -> doc -> Value
encodeDocument cfg document =
case idsJoinRelation cfg of
Nothing -> toJSON document
Just (ParentDocument (FieldName field) name) ->
mergeObjects (toJSON document) (object [fromText field .= name])
Just (ChildDocument (FieldName field) name parent) ->
mergeObjects (toJSON document) (object [fromText field .= object ["name" .= name, "parent" .= parent]])
where
mergeObjects (Object a) (Object b) = Object (a <> b)
mergeObjects _ _ = error "Impossible happened: both document body and join parameters must be objects"
-- | 'deleteDocument' is the primary way to delete a single document.
--
-- >>> _ <- runBH' $ deleteDocument testIndex (DocId "1")
deleteDocument :: MonadBH m => IndexName -> DocId -> m Reply
deleteDocument (IndexName indexName) (DocId docId) =
delete =<< joinPath [indexName, "_doc", docId]
-- | 'deleteByQuery' performs a deletion on every document that matches a query.
--
-- >>> let query = TermQuery (Term "user" "bitemyapp") Nothing
-- >>> _ <- runBH' $ deleteDocument testIndex query
deleteByQuery :: MonadBH m => IndexName -> Query -> m Reply
deleteByQuery (IndexName indexName) query =
bindM2 post url (return body)
where
url = joinPath [indexName, "_delete_by_query"]
body = Just (encode $ object [ "query" .= query ])
-- | 'bulk' uses
-- <http://www.elasticsearch.org/guide/en/elasticsearch/reference/current/docs-bulk.html Elasticsearch's bulk API>
-- to perform bulk operations. The 'BulkOperation' data type encodes the
-- index\/update\/delete\/create operations. You pass a 'V.Vector' of 'BulkOperation's
-- and a 'Server' to 'bulk' in order to send those operations up to your Elasticsearch
-- server to be performed. I changed from [BulkOperation] to a Vector due to memory overhead.
--
-- >>> let stream = V.fromList [BulkIndex testIndex (DocId "2") (toJSON (BulkTest "blah"))]
-- >>> _ <- runBH' $ bulk stream
-- >>> _ <- runBH' $ refreshIndex testIndex
bulk :: MonadBH m => V.Vector BulkOperation -> m Reply
bulk bulkOps =
bindM2 post url (return body)
where
url = joinPath ["_bulk"]
body = Just $ encodeBulkOperations bulkOps
-- | 'encodeBulkOperations' is a convenience function for dumping a vector of 'BulkOperation'
-- into an 'L.ByteString'
--
-- >>> let bulkOps = V.fromList [BulkIndex testIndex (DocId "2") (toJSON (BulkTest "blah"))]
-- >>> encodeBulkOperations bulkOps
-- "\n{\"index\":{\"_id\":\"2\",\"_index\":\"twitter\"}}\n{\"name\":\"blah\"}\n"
encodeBulkOperations :: V.Vector BulkOperation -> L.ByteString
encodeBulkOperations stream = collapsed where
blobs =
fmap encodeBulkOperation stream
mashedTaters =
mash (mempty :: Builder) blobs
collapsed =
toLazyByteString $ mappend mashedTaters (byteString "\n")
mash :: Builder -> V.Vector L.ByteString -> Builder
mash = V.foldl' (\b x -> b <> byteString "\n" <> lazyByteString x)
mkBulkStreamValue :: Text -> Text -> Text -> Value
mkBulkStreamValue operation indexName docId =
object [fromText operation .=
object [ "_index" .= indexName
, "_id" .= docId]]
mkBulkStreamValueAuto :: Text -> Text -> Value
mkBulkStreamValueAuto operation indexName =
object [fromText operation .=
object [ "_index" .= indexName ]]
mkBulkStreamValueWithMeta :: [UpsertActionMetadata] -> Text -> Text -> Text -> Value
mkBulkStreamValueWithMeta meta operation indexName docId =
object [ fromText operation .=
object ([ "_index" .= indexName
, "_id" .= docId]
<> (buildUpsertActionMetadata <$> meta))]
-- | 'encodeBulkOperation' is a convenience function for dumping a single 'BulkOperation'
-- into an 'L.ByteString'
--
-- >>> let bulkOp = BulkIndex testIndex (DocId "2") (toJSON (BulkTest "blah"))
-- >>> encodeBulkOperation bulkOp
-- "{\"index\":{\"_id\":\"2\",\"_index\":\"twitter\"}}\n{\"name\":\"blah\"}"
encodeBulkOperation :: BulkOperation -> L.ByteString
encodeBulkOperation (BulkIndex (IndexName indexName) (DocId docId) value) = blob
where metadata = mkBulkStreamValue "index" indexName docId
blob = encode metadata `mappend` "\n" `mappend` encode value
encodeBulkOperation (BulkIndexAuto (IndexName indexName) value) = blob
where metadata = mkBulkStreamValueAuto "index" indexName
blob = encode metadata `mappend` "\n" `mappend` encode value
encodeBulkOperation (BulkIndexEncodingAuto (IndexName indexName) encoding) = toLazyByteString blob
where metadata = toEncoding (mkBulkStreamValueAuto "index" indexName)
blob = fromEncoding metadata <> "\n" <> fromEncoding encoding
encodeBulkOperation (BulkCreate (IndexName indexName) (DocId docId) value) = blob
where metadata = mkBulkStreamValue "create" indexName docId
blob = encode metadata `mappend` "\n" `mappend` encode value
encodeBulkOperation (BulkDelete (IndexName indexName) (DocId docId)) = blob
where metadata = mkBulkStreamValue "delete" indexName docId
blob = encode metadata
encodeBulkOperation (BulkUpdate (IndexName indexName) (DocId docId) value) = blob
where metadata = mkBulkStreamValue "update" indexName docId
doc = object ["doc" .= value]
blob = encode metadata `mappend` "\n" `mappend` encode doc
encodeBulkOperation (BulkUpsert (IndexName indexName)
(DocId docId)
payload
actionMeta) = blob
where metadata = mkBulkStreamValueWithMeta actionMeta "update" indexName docId
blob = encode metadata <> "\n" <> encode doc
doc = case payload of
UpsertDoc value -> object ["doc" .= value, "doc_as_upsert" .= True]
UpsertScript scriptedUpsert script value ->
let scup = if scriptedUpsert then ["scripted_upsert" .= True] else []
upsert = ["upsert" .= value]
in
case (object (scup <> upsert), toJSON script) of
(Object obj, Object jscript) -> Object $ jscript <> obj
_ -> error "Impossible happened: serialising Script to Json should always be Object"
encodeBulkOperation (BulkCreateEncoding (IndexName indexName) (DocId docId) encoding) =
toLazyByteString blob
where metadata = toEncoding (mkBulkStreamValue "create" indexName docId)
blob = fromEncoding metadata <> "\n" <> fromEncoding encoding
-- | 'getDocument' is a straight-forward way to fetch a single document from
-- Elasticsearch using a 'Server', 'IndexName', and a 'DocId'.
-- The 'DocId' is the primary key for your Elasticsearch document.
--
-- >>> yourDoc <- runBH' $ getDocument testIndex (DocId "1")
getDocument :: MonadBH m => IndexName -> DocId -> m Reply
getDocument (IndexName indexName) (DocId docId) =
get =<< joinPath [indexName, "_doc", docId]
-- | 'documentExists' enables you to check if a document exists.
documentExists :: MonadBH m => IndexName -> DocId -> m Bool
documentExists (IndexName indexName) (DocId docId) =
fmap snd (existentialQuery =<< joinPath [indexName, "_doc", docId])
dispatchSearch :: MonadBH m => Text -> Search -> m Reply
dispatchSearch url search = post url' (Just (encode search))
where url' = appendSearchTypeParam url (searchType search)
-- | 'searchAll', given a 'Search', will perform that search against all indexes
-- on an Elasticsearch server. Try to avoid doing this if it can be helped.
--
-- >>> let query = TermQuery (Term "user" "bitemyapp") Nothing
-- >>> let search = mkSearch (Just query) Nothing
-- >>> reply <- runBH' $ searchAll search
searchAll :: MonadBH m => Search -> m Reply
searchAll = bindM2 dispatchSearch url . return
where url = joinPath ["_search"]
-- | 'searchByIndex', given a 'Search' and an 'IndexName', will perform that search
-- within an index on an Elasticsearch server.
--
-- >>> let query = TermQuery (Term "user" "bitemyapp") Nothing
-- >>> let search = mkSearch (Just query) Nothing
-- >>> reply <- runBH' $ searchByIndex testIndex search
searchByIndex :: MonadBH m => IndexName -> Search -> m Reply
searchByIndex (IndexName indexName) = bindM2 dispatchSearch url . return
where url = joinPath [indexName, "_search"]
-- | 'searchByIndices' is a variant of 'searchByIndex' that executes a
-- 'Search' over many indices. This is much faster than using
-- 'mapM' to 'searchByIndex' over a collection since it only
-- causes a single HTTP request to be emitted.
searchByIndices :: MonadBH m => NonEmpty IndexName -> Search -> m Reply
searchByIndices ixs = bindM2 dispatchSearch url . return
where url = joinPath [renderedIxs, "_search"]
renderedIxs = T.intercalate (T.singleton ',') (map (\(IndexName t) -> t) (toList ixs))
dispatchSearchTemplate :: MonadBH m => Text -> SearchTemplate -> m Reply
dispatchSearchTemplate url search = post url (Just (encode search))
-- | 'searchByIndexTemplate', given a 'SearchTemplate' and an 'IndexName', will perform that search
-- within an index on an Elasticsearch server.
--
-- >>> let query = SearchTemplateSource "{\"query\": { \"match\" : { \"{{my_field}}\" : \"{{my_value}}\" } }, \"size\" : \"{{my_size}}\"}"
-- >>> let search = mkSearchTemplate (Right query) Nothing
-- >>> reply <- runBH' $ searchByIndexTemplate testIndex search
searchByIndexTemplate :: MonadBH m => IndexName -> SearchTemplate -> m Reply
searchByIndexTemplate (IndexName indexName) = bindM2 dispatchSearchTemplate url . return
where url = joinPath [indexName, "_search", "template"]
-- | 'searchByIndicesTemplate' is a variant of 'searchByIndexTemplate' that executes a
-- 'SearchTemplate' over many indices. This is much faster than using
-- 'mapM' to 'searchByIndexTemplate' over a collection since it only
-- causes a single HTTP request to be emitted.
searchByIndicesTemplate :: MonadBH m => NonEmpty IndexName -> SearchTemplate -> m Reply
searchByIndicesTemplate ixs = bindM2 dispatchSearchTemplate url . return
where url = joinPath [renderedIxs, "_search", "template"]
renderedIxs = T.intercalate (T.singleton ',') (map (\(IndexName t) -> t) (toList ixs))
-- | 'storeSearchTemplate', saves a 'SearchTemplateSource' to be used later.
storeSearchTemplate :: MonadBH m => SearchTemplateId -> SearchTemplateSource -> m Reply
storeSearchTemplate (SearchTemplateId tid) ts =
url >>= flip post (Just (encode json_))
where
url = joinPath ["_scripts", tid]
json_ = Object $ X.fromList ["script" .= Object ("lang" .= String "mustache" <> "source" .= ts) ]
-- | 'getSearchTemplate', get info of an stored 'SearchTemplateSource'.
getSearchTemplate :: MonadBH m => SearchTemplateId -> m Reply
getSearchTemplate (SearchTemplateId tid) =
url >>= get
where
url = joinPath ["_scripts", tid]
-- | 'storeSearchTemplate',
deleteSearchTemplate :: MonadBH m => SearchTemplateId -> m Reply
deleteSearchTemplate (SearchTemplateId tid) =
url >>= delete
where
url = joinPath ["_scripts", tid]
-- | For a given search, request a scroll for efficient streaming of
-- search results. Note that the search is put into 'SearchTypeScan'
-- mode and thus results will not be sorted. Combine this with
-- 'advanceScroll' to efficiently stream through the full result set
getInitialScroll ::
(FromJSON a, MonadThrow m, MonadBH m) => IndexName ->
Search ->
m (Either EsError (SearchResult a))
getInitialScroll (IndexName indexName) search' = do
let url = addQuery params <$> joinPath [indexName, "_search"]
params = [("scroll", Just "1m")]
sorting = Just [DefaultSortSpec $ mkSort (FieldName "_doc") Descending]
search = search' { sortBody = sorting }
resp' <- bindM2 dispatchSearch url (return search)
parseEsResponse resp'
-- | For a given search, request a scroll for efficient streaming of
-- search results. Combine this with 'advanceScroll' to efficiently
-- stream through the full result set. Note that this search respects
-- sorting and may be less efficient than 'getInitialScroll'.
getInitialSortedScroll ::
(FromJSON a, MonadThrow m, MonadBH m) => IndexName ->
Search ->
m (Either EsError (SearchResult a))
getInitialSortedScroll (IndexName indexName) search = do
let url = addQuery params <$> joinPath [indexName, "_search"]
params = [("scroll", Just "1m")]
resp' <- bindM2 dispatchSearch url (return search)
parseEsResponse resp'
scroll' :: (FromJSON a, MonadBH m, MonadThrow m) => Maybe ScrollId ->
m ([Hit a], Maybe ScrollId)
scroll' Nothing = return ([], Nothing)
scroll' (Just sid) = do
res <- advanceScroll sid 60
case res of
Right SearchResult {..} -> return (hits searchHits, scrollId)
Left _ -> return ([], Nothing)
-- | Use the given scroll to fetch the next page of documents. If there are no
-- further pages, 'SearchResult.searchHits.hits' will be '[]'.
advanceScroll
:: ( FromJSON a
, MonadBH m
, MonadThrow m
)
=> ScrollId
-> NominalDiffTime
-- ^ How long should the snapshot of data be kept around? This timeout is updated every time 'advanceScroll' is used, so don't feel the need to set it to the entire duration of your search processing. Note that durations < 1s will be rounded up. Also note that 'NominalDiffTime' is an instance of Num so literals like 60 will be interpreted as seconds. 60s is a reasonable default.
-> m (Either EsError (SearchResult a))
advanceScroll (ScrollId sid) scroll = do
url <- joinPath ["_search", "scroll"]
resp <- post url (Just $ encode scrollObject)
parseEsResponse resp
where scrollTime = showText secs <> "s"
secs :: Integer
secs = round scroll
scrollObject = object [ "scroll" .= scrollTime
, "scroll_id" .= sid
]
scanAccumulator ::
(FromJSON a, MonadBH m, MonadThrow m) =>
[Hit a] ->
([Hit a], Maybe ScrollId) ->
m ([Hit a], Maybe ScrollId)
scanAccumulator oldHits (newHits, Nothing) = return (oldHits ++ newHits, Nothing)
scanAccumulator oldHits ([], _) = return (oldHits, Nothing)
scanAccumulator oldHits (newHits, msid) = do
(newHits', msid') <- scroll' msid
scanAccumulator (oldHits ++ newHits) (newHits', msid')
-- | 'scanSearch' uses the 'scroll' API of elastic,
-- for a given 'IndexName'. Note that this will
-- consume the entire search result set and will be doing O(n) list
-- appends so this may not be suitable for large result sets. In that
-- case, 'getInitialScroll' and 'advanceScroll' are good low level
-- tools. You should be able to hook them up trivially to conduit,
-- pipes, or your favorite streaming IO abstraction of choice. Note
-- that ordering on the search would destroy performance and thus is
-- ignored.
scanSearch :: (FromJSON a, MonadBH m, MonadThrow m) => IndexName
-> Search
-> m [Hit a]
scanSearch indexName search = do
initialSearchResult <- getInitialScroll indexName search
let (hits', josh) = case initialSearchResult of
Right SearchResult {..} -> (hits searchHits, scrollId)
Left _ -> ([], Nothing)
(totalHits, _) <- scanAccumulator [] (hits', josh)
return totalHits
pitAccumulator
:: (FromJSON a, MonadBH m, MonadThrow m) => Search -> [Hit a] -> m [Hit a]
pitAccumulator search oldHits = do
resp <- searchAll search
parsed <- parseEsResponse resp
case parsed of
Left _ -> return []
Right searchResult -> case hits (searchHits searchResult) of
[] -> return oldHits
newHits -> case (hitSort $ last newHits, pitId searchResult) of
(Nothing, Nothing) ->
error "no point in time (PIT) ID or last sort value"
(Just _ , Nothing ) -> error "no point in time (PIT) ID"
(Nothing , _ ) -> return (oldHits <> newHits)
(Just lastSort, Just pitId') -> do
let newSearch = search { pointInTime = Just (PointInTime pitId' "1m")
, searchAfterKey = Just lastSort
}
pitAccumulator newSearch (oldHits <> newHits)
-- | 'pitSearch' uses the point in time (PIT) API of elastic, for a given
-- 'IndexName'. Requires Elasticsearch >=7.10. Note that this will consume the
-- entire search result set and will be doing O(n) list appends so this may
-- not be suitable for large result sets. In that case, the point in time API
-- should be used directly with `openPointInTime` and `closePointInTime`.
--
-- Note that 'pitSearch' utilizes the 'search_after' parameter under the hood,
-- which requires a non-empty 'sortBody' field in the provided 'Search' value.
-- Otherwise, 'pitSearch' will fail to return all matching documents.
--
-- For more information see
-- <https://www.elastic.co/guide/en/elasticsearch/reference/current/point-in-time-api.html>.
pitSearch
:: (FromJSON a, MonadBH m, MonadThrow m) => IndexName -> Search -> m [Hit a]
pitSearch indexName search = do
openResp <- openPointInTime indexName
case openResp of
Left _ -> return []
Right OpenPointInTimeResponse {..} -> do
let searchPIT = search { pointInTime = Just (PointInTime oPitId "1m") }
hits <- pitAccumulator searchPIT []
closeResp <- closePointInTime $ ClosePointInTime oPitId
case closeResp of
Left _ -> return []
Right (ClosePointInTimeResponse False _) ->
error "failed to close point in time (PIT)"
Right (ClosePointInTimeResponse True _) -> return hits
-- | 'mkSearch' is a helper function for defaulting additional fields of a 'Search'
-- to Nothing in case you only care about your 'Query' and 'Filter'. Use record update
-- syntax if you want to add things like aggregations or highlights while still using
-- this helper function.
--
-- >>> let query = TermQuery (Term "user" "bitemyapp") Nothing
-- >>> mkSearch (Just query) Nothing
-- Search {queryBody = Just (TermQuery (Term {termField = "user", termValue = "bitemyapp"}) Nothing), filterBody = Nothing, searchAfterKey = Nothing, sortBody = Nothing, aggBody = Nothing, highlight = Nothing, trackSortScores = False, from = From 0, size = Size 10, searchType = SearchTypeQueryThenFetch, fields = Nothing, source = Nothing}
mkSearch :: Maybe Query -> Maybe Filter -> Search
mkSearch query filter = Search
{ queryBody = query
, filterBody = filter
, sortBody = Nothing
, aggBody = Nothing
, highlight = Nothing
, trackSortScores = False
, from = From 0
, size = Size 10
, searchType = SearchTypeQueryThenFetch
, searchAfterKey = Nothing
, fields = Nothing
, scriptFields = Nothing
, source = Nothing
, suggestBody = Nothing
, pointInTime = Nothing
}
-- | 'mkAggregateSearch' is a helper function that defaults everything in a 'Search' except for
-- the 'Query' and the 'Aggregation'.
--
-- >>> let terms = TermsAgg $ (mkTermsAggregation "user") { termCollectMode = Just BreadthFirst }
-- >>> terms
-- TermsAgg (TermsAggregation {term = Left "user", termInclude = Nothing, termExclude = Nothing, termOrder = Nothing, termMinDocCount = Nothing, termSize = Nothing, termShardSize = Nothing, termCollectMode = Just BreadthFirst, termExecutionHint = Nothing, termAggs = Nothing})
-- >>> let myAggregation = mkAggregateSearch Nothing $ mkAggregations "users" terms
mkAggregateSearch :: Maybe Query -> Aggregations -> Search
mkAggregateSearch query mkSearchAggs = Search
{ queryBody = query
, filterBody = Nothing
, sortBody = Nothing
, aggBody = Just mkSearchAggs
, highlight = Nothing
, trackSortScores = False
, from = From 0
, size = Size 0
, searchType = SearchTypeQueryThenFetch
, searchAfterKey = Nothing
, fields = Nothing
, scriptFields = Nothing
, source = Nothing
, suggestBody = Nothing
, pointInTime = Nothing
}
-- | 'mkHighlightSearch' is a helper function that defaults everything in a 'Search' except for
-- the 'Query' and the 'Aggregation'.
--
-- >>> let query = QueryMatchQuery $ mkMatchQuery (FieldName "_all") (QueryString "haskell")
-- >>> let testHighlight = Highlights Nothing [FieldHighlight (FieldName "message") Nothing]
-- >>> let search = mkHighlightSearch (Just query) testHighlight
mkHighlightSearch :: Maybe Query -> Highlights -> Search
mkHighlightSearch query searchHighlights = Search
{ queryBody = query
, filterBody = Nothing
, sortBody = Nothing
, aggBody = Nothing
, highlight = Just searchHighlights
, trackSortScores = False
, from = From 0
, size = Size 10
, searchType = SearchTypeDfsQueryThenFetch
, searchAfterKey = Nothing
, fields = Nothing
, scriptFields = Nothing
, source = Nothing
, suggestBody = Nothing
, pointInTime = Nothing
}
-- | 'mkSearchTemplate' is a helper function for defaulting additional fields of a 'SearchTemplate'
-- to Nothing. Use record update syntax if you want to add things.
mkSearchTemplate :: Either SearchTemplateId SearchTemplateSource -> TemplateQueryKeyValuePairs -> SearchTemplate
mkSearchTemplate id_ params = SearchTemplate id_ params Nothing Nothing
-- | 'pageSearch' is a helper function that takes a search and assigns the from
-- and size fields for the search. The from parameter defines the offset
-- from the first result you want to fetch. The size parameter allows you to
-- configure the maximum amount of hits to be returned.
--
-- >>> let query = QueryMatchQuery $ mkMatchQuery (FieldName "_all") (QueryString "haskell")
-- >>> let search = mkSearch (Just query) Nothing
-- >>> search
-- Search {queryBody = Just (QueryMatchQuery (MatchQuery {matchQueryField = FieldName "_all", matchQueryQueryString = QueryString "haskell", matchQueryOperator = Or, matchQueryZeroTerms = ZeroTermsNone, matchQueryCutoffFrequency = Nothing, matchQueryMatchType = Nothing, matchQueryAnalyzer = Nothing, matchQueryMaxExpansions = Nothing, matchQueryLenient = Nothing, matchQueryBoost = Nothing})), filterBody = Nothing, sortBody = Nothing, aggBody = Nothing, highlight = Nothing, trackSortScores = False, from = From 0, size = Size 10, searchType = SearchTypeQueryThenFetch, fields = Nothing, source = Nothing}
-- >>> pageSearch (From 10) (Size 100) search
-- Search {queryBody = Just (QueryMatchQuery (MatchQuery {matchQueryField = FieldName "_all", matchQueryQueryString = QueryString "haskell", matchQueryOperator = Or, matchQueryZeroTerms = ZeroTermsNone, matchQueryCutoffFrequency = Nothing, matchQueryMatchType = Nothing, matchQueryAnalyzer = Nothing, matchQueryMaxExpansions = Nothing, matchQueryLenient = Nothing, matchQueryBoost = Nothing})), filterBody = Nothing, sortBody = Nothing, aggBody = Nothing, highlight = Nothing, trackSortScores = False, from = From 10, size = Size 100, searchType = SearchTypeQueryThenFetch, fields = Nothing, source = Nothing}
pageSearch :: From -- ^ The result offset
-> Size -- ^ The number of results to return
-> Search -- ^ The current seach
-> Search -- ^ The paged search
pageSearch resultOffset pageSize search = search { from = resultOffset, size = pageSize }
parseUrl' :: MonadThrow m => Text -> m Request
parseUrl' t = parseRequest (URI.escapeURIString URI.isAllowedInURI (T.unpack t))
-- | Was there an optimistic concurrency control conflict when
-- indexing a document?
isVersionConflict :: Reply -> Bool
isVersionConflict = statusCheck (== 409)
isSuccess :: Reply -> Bool
isSuccess = statusCheck (inRange (200, 299))
isCreated :: Reply -> Bool
isCreated = statusCheck (== 201)
statusCheck :: (Int -> Bool) -> Reply -> Bool
statusCheck prd = prd . NHTS.statusCode . responseStatus
-- | This is a hook that can be set via the 'bhRequestHook' function
-- that will authenticate all requests using an HTTP Basic
-- Authentication header. Note that it is *strongly* recommended that
-- this option only be used over an SSL connection.
--
-- >> (mkBHEnv myServer myManager) { bhRequestHook = basicAuthHook (EsUsername "myuser") (EsPassword "mypass") }
basicAuthHook :: Monad m => EsUsername -> EsPassword -> Request -> m Request
basicAuthHook (EsUsername u) (EsPassword p) = return . applyBasicAuth u' p'
where u' = T.encodeUtf8 u
p' = T.encodeUtf8 p
boolQP :: Bool -> Text
boolQP True = "true"
boolQP False = "false"
countByIndex :: (MonadBH m, MonadThrow m) => IndexName -> CountQuery -> m (Either EsError CountResponse)
countByIndex (IndexName indexName) q = do
url <- joinPath [indexName, "_count"]
parseEsResponse =<< post url (Just (encode q))
-- | 'openPointInTime' opens a point in time for an index given an 'IndexName'.
-- Note that the point in time should be closed with 'closePointInTime' as soon
-- as it is no longer needed.
--
-- For more information see
-- <https://www.elastic.co/guide/en/elasticsearch/reference/current/point-in-time-api.html>.
openPointInTime
:: (MonadBH m, MonadThrow m)
=> IndexName
-> m (Either EsError OpenPointInTimeResponse)
openPointInTime (IndexName indexName) = do
url <- joinPath [indexName, "_pit?keep_alive=1m"]
parseEsResponse =<< post url Nothing
-- | 'closePointInTime' closes a point in time given a 'ClosePointInTime'.
--
-- For more information see
-- <https://www.elastic.co/guide/en/elasticsearch/reference/current/point-in-time-api.html>.
closePointInTime
:: (MonadBH m, MonadThrow m)
=> ClosePointInTime
-> m (Either EsError ClosePointInTimeResponse)
closePointInTime q = do
url <- joinPath ["_pit"]
parseEsResponse =<< delete' url (Just (encode q))