diff --git a/hackage-db.cabal b/hackage-db.cabal --- a/hackage-db.cabal +++ b/hackage-db.cabal @@ -1,5 +1,5 @@ Name: hackage-db -Version: 1.8 +Version: 1.9 Copyright: Peter Simons License: BSD3 License-File: LICENSE @@ -39,7 +39,9 @@ quickly, but the inital cost of reading @00-index.tar@ is fairly high. . - This package is known to work on Linux and Mac OS X, but not Windows. + This package is known to work on Linux and Mac OS X, but it's + probably not going to work on Windows (because no-one tested it, as + far as I know). Source-Repository head Type: git @@ -49,5 +51,6 @@ Build-Depends: base >= 3 && < 5, tar >= 0.4, Cabal, containers, directory, filepath, bytestring, utf8-string hs-source-dirs: src - Ghc-Options: -Wall - Exposed-Modules: Distribution.Hackage.DB + Exposed-Modules: Distribution.Hackage.DB.Unparsed + Distribution.Hackage.DB.Parsed + Distribution.Hackage.DB diff --git a/src/Distribution/Hackage/DB.hs b/src/Distribution/Hackage/DB.hs --- a/src/Distribution/Hackage/DB.hs +++ b/src/Distribution/Hackage/DB.hs @@ -20,70 +20,7 @@ import Data.Map import Data.Version +import Distribution.Hackage.DB.Parsed +import Distribution.Hackage.DB.Path import Distribution.Package import Distribution.PackageDescription - -import qualified Codec.Archive.Tar as Tar -import Data.ByteString.Lazy.Char8 ( ByteString ) -import qualified Data.ByteString.Lazy.Char8 as BS8 ( readFile ) -import qualified Data.ByteString.Lazy as BSC ( unpack ) -import Data.String.UTF8 ( toString, fromRep ) -import Data.Maybe ( fromMaybe ) -import System.Directory ( getHomeDirectory ) -import System.FilePath ( joinPath, splitDirectories ) -import Distribution.Text ( simpleParse ) -import Distribution.PackageDescription.Parse ( parsePackageDescription, ParseResult(..) ) - --- | A 'Map' representation of the Hackage database. For sake of --- simplicity, we use 'String' rather than 'PackageName' to represent --- the name of a package. - -type Hackage = Map String (Map Version GenericPackageDescription) - --- | Read the Hackage database from the location determined by 'hackagePath' --- and return a 'Map' that provides fast access to its contents. - -readHackage :: IO Hackage -readHackage = hackagePath >>= readHackage' - --- | Read the Hackage database from the given 'FilePath' and return a --- 'Hackage' map that provides fast access to its contents. - -readHackage' :: FilePath -> IO Hackage -readHackage' = fmap parseHackage . BS8.readFile - --- | Parse the contents of Hackage's @00-index.tar@ into a 'Hackage' map. - -parseHackage :: ByteString -> Hackage -parseHackage = Tar.foldEntries addEntry empty (error . show) . Tar.read - where - decodeUTF8 :: ByteString -> String - decodeUTF8 = toString . fromRep . BSC.unpack - - addEntry :: Tar.Entry -> Hackage -> Hackage - addEntry e db = case splitDirectories (Tar.entryPath e) of - [".",".","@LongLink"] -> db - path@[name,vers,_] -> case Tar.entryContent e of - Tar.NormalFile buf _ -> add name vers buf db - _ -> error ("Hackage.DB.parseHackage: unexpected content type for " ++ show path) - _ -> db - - add :: String -> String -> ByteString -> Hackage -> Hackage - add name version pkg = insertWith union name (singleton (pVersion version) (pPackage name pkg)) - - pPackage :: String -> ByteString -> GenericPackageDescription - pPackage name buf = case parsePackageDescription (decodeUTF8 buf) of - ParseOk _ a -> a - ParseFailed err -> error ("Hackage.DB.parseHackage: cannot parse cabal file " ++ show name ++ ": " ++ show err) - - pVersion :: String -> Version - pVersion str = fromMaybe (error $ "Hackage.DB.parseHackage: cannot parse version " ++ show str) (simpleParse str) - --- | Determine the default path of the Hackage database, which typically --- resides at @"$HOME\/.cabal\/packages\/hackage.haskell.org\/00-index.tar"@. --- Running the command @"cabal update"@ will keep that file up-to-date. - -hackagePath :: IO FilePath -hackagePath = do - homedir <- getHomeDirectory - return $ joinPath [homedir, ".cabal", "packages", "hackage.haskell.org", "00-index.tar"] diff --git a/src/Distribution/Hackage/DB/Parsed.hs b/src/Distribution/Hackage/DB/Parsed.hs new file mode 100644 --- /dev/null +++ b/src/Distribution/Hackage/DB/Parsed.hs @@ -0,0 +1,65 @@ +{- | + Module : Distribution.Hackage.DB.Parsed + License : BSD3 + Maintainer : simons@cryp.to + Stability : provisional + Portability : portable + + This module provides simple access to the Hackage database by means + of 'Map'. + -} + +module Distribution.Hackage.DB.Parsed + ( Hackage, readHackage, readHackage', parseHackage + , parseUnparsedHackage + ) + where + +import qualified Data.ByteString.Lazy as BSC ( unpack ) +import Data.ByteString.Lazy.Char8 ( ByteString ) +import Data.Map +import Data.String.UTF8 ( toString, fromRep ) +import Data.Version +import qualified Distribution.Hackage.DB.Unparsed as Unparsed +import Distribution.PackageDescription +import Distribution.PackageDescription.Parse ( parsePackageDescription, ParseResult(..) ) +import Distribution.Text ( display ) + +-- | A 'Map' representation of the Hackage database. For sake of +-- simplicity, we use 'String' rather than 'PackageName' to represent +-- the name of a package. + +type Hackage = Map String (Map Version GenericPackageDescription) + +-- | Read the Hackage database from the location determined by 'hackagePath' +-- and return a 'Map' that provides fast access to its contents. + +readHackage :: IO Hackage +readHackage = fmap parseUnparsedHackage Unparsed.readHackage + +-- | Read the Hackage database from the given 'FilePath' and return a +-- 'Hackage' map that provides fast access to its contents. + +readHackage' :: FilePath -> IO Hackage +readHackage' = fmap parseUnparsedHackage . Unparsed.readHackage' + +-- | Parse the contents of Hackage's @00-index.tar@ into a 'Hackage' map. + +parseHackage :: ByteString -> Hackage +parseHackage = parseUnparsedHackage . Unparsed.parseHackage + +-- | Convert an 'Unparsed.Hackage' map into a parsed 'Hackage' map. + +parseUnparsedHackage :: Unparsed.Hackage -> Hackage +parseUnparsedHackage = Data.Map.mapWithKey parsePackages + where + parsePackages :: String -> Map Version ByteString -> Map Version GenericPackageDescription + parsePackages name = Data.Map.mapWithKey (parsePackage name) + + parsePackage :: String -> Version -> ByteString -> GenericPackageDescription + parsePackage name version buf = case parsePackageDescription (decodeUTF8 buf) of + ParseOk _ a -> a + ParseFailed err -> error ("Hackage.DB.parseHackage: cannot parse cabal file " ++ show name ++ "-" ++ display version ++ ": " ++ show err) + + decodeUTF8 :: ByteString -> String + decodeUTF8 = toString . fromRep . BSC.unpack diff --git a/src/Distribution/Hackage/DB/Unparsed.hs b/src/Distribution/Hackage/DB/Unparsed.hs new file mode 100644 --- /dev/null +++ b/src/Distribution/Hackage/DB/Unparsed.hs @@ -0,0 +1,71 @@ +{- | + Module : Distribution.Hackage.DB.Unparsed + License : BSD3 + Maintainer : simons@cryp.to + Stability : provisional + Portability : portable + + This module provides simple access to the Hackage database by means + of 'Map'. + -} + +module Distribution.Hackage.DB.Unparsed + ( Hackage, readHackage, readHackage', parseHackage, hackagePath + ) + where + +import qualified Codec.Archive.Tar as Tar +import Data.ByteString.Lazy.Char8 ( ByteString ) +import qualified Data.ByteString.Lazy.Char8 as BS8 ( readFile ) +import Data.Map +import Data.Maybe ( fromMaybe ) +import Data.Version +import Distribution.Text ( simpleParse ) +import System.Directory ( getHomeDirectory ) +import System.FilePath ( joinPath, splitDirectories ) + +-- | A 'Map' representation of the Hackage database. Every package name +-- maps to a non-empty set of version, and for every version there is a +-- Cabal file stored as a (lazy) 'ByteString'. + +type Hackage = Map String (Map Version ByteString) + +-- | Read the Hackage database from the location determined by 'hackagePath' +-- and return a 'Map' that provides fast access to its contents. + +readHackage :: IO Hackage +readHackage = hackagePath >>= readHackage' + +-- | Read the Hackage database from the given 'FilePath' and return a +-- 'Hackage' map that provides fast access to its contents. + +readHackage' :: FilePath -> IO Hackage +readHackage' = fmap parseHackage . BS8.readFile + +-- | Parse the contents of Hackage's @00-index.tar@ into a 'Hackage' map. + +parseHackage :: ByteString -> Hackage +parseHackage = Tar.foldEntries addEntry empty (error . show) . Tar.read + where + addEntry :: Tar.Entry -> Hackage -> Hackage + addEntry e db = case splitDirectories (Tar.entryPath e) of + [".",".","@LongLink"] -> db + path@[name,vers,_] -> case Tar.entryContent e of + Tar.NormalFile buf _ -> add name vers buf db + _ -> error ("Hackage.DB.parseHackage: unexpected content type for " ++ show path) + _ -> db + + add :: String -> String -> ByteString -> Hackage -> Hackage + add name version pkg = insertWith union name (singleton (pVersion version) pkg) + + pVersion :: String -> Version + pVersion str = fromMaybe (error $ "Hackage.DB.parseHackage: cannot parse version " ++ show str) (simpleParse str) + +-- | Determine the default path of the Hackage database, which typically +-- resides at @"$HOME\/.cabal\/packages\/hackage.haskell.org\/00-index.tar"@. +-- Running the command @"cabal update"@ will keep that file up-to-date. + +hackagePath :: IO FilePath +hackagePath = do + homedir <- getHomeDirectory + return $ joinPath [homedir, ".cabal", "packages", "hackage.haskell.org", "00-index.tar"]