X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=Network%2FHTTP%2FLucu%2FMIMEType%2FGuess.hs;h=d8bca8e785658efa5390862f8baa97c071a93f58;hb=a362be1c8664306b970c32e1df9b62081498feb1;hp=12f19e0684748eb7bccfa229e670fb80f9b91e57;hpb=854f6d1709d468f7e1bf0db2dcd2d30de6bd5f5e;p=Lucu.git diff --git a/Network/HTTP/Lucu/MIMEType/Guess.hs b/Network/HTTP/Lucu/MIMEType/Guess.hs index 12f19e0..d8bca8e 100644 --- a/Network/HTTP/Lucu/MIMEType/Guess.hs +++ b/Network/HTTP/Lucu/MIMEType/Guess.hs @@ -1,117 +1,156 @@ +{-# LANGUAGE + UnicodeSyntax + #-} +-- |Guessing MIME Types by file extensions. It's not always accurate +-- but simple and fast. +-- +-- In general you don't have to use this module directly. module Network.HTTP.Lucu.MIMEType.Guess ( ExtMap - , guessTypeByFileName -- ExtMap -> FilePath -> Maybe MIMEType + , guessTypeByFileName - , parseExtMapFile -- FilePath -> IO ExtMap - , outputExtMapAsHS -- ExtMap -> FilePath -> IO () + , parseExtMapFile + , serializeExtMap ) where - +import Control.Applicative +import Control.Monad +import qualified Data.Ascii as A +import Data.Attoparsec.Char8 as P +import qualified Data.Attoparsec.Lazy as LP import qualified Data.ByteString.Lazy.Char8 as B -import Data.ByteString.Lazy.Char8 (ByteString) import qualified Data.Map as M -import Data.Map (Map) -import Data.Maybe -import Language.Haskell.Pretty -import Language.Haskell.Syntax -import Network.HTTP.Lucu.MIMEType -import Network.HTTP.Lucu.Parser -import Network.HTTP.Lucu.Parser.Http -import Network.HTTP.Lucu.Utils -import System.IO - -type ExtMap = Map String MIMEType - - -guessTypeByFileName :: ExtMap -> FilePath -> Maybe MIMEType +import Data.Map (Map) +import Data.Maybe +import Data.Monoid.Unicode +import Data.Text (Text) +import qualified Data.Text as T +import Data.Text.Encoding +import Language.Haskell.Exts.Build +import Language.Haskell.Exts.Extension +import Language.Haskell.Exts.Pretty +import Language.Haskell.Exts.Syntax +import Network.HTTP.Lucu.MIMEType +import Prelude.Unicode +import System.FilePath + +-- |A 'Map' from file extensions to 'MIMEType's. +type ExtMap = Map Text MIMEType + +-- |Guess the MIME Type of a file. +guessTypeByFileName ∷ ExtMap → FilePath → Maybe MIMEType guessTypeByFileName extMap fpath - = let ext = last $ splitBy (== '.') fpath - in - M.lookup ext extMap >>= return - + = case takeExtension fpath of + [] → Nothing + (_:ext) → M.lookup (T.pack ext) extMap -parseExtMapFile :: FilePath -> IO ExtMap +-- |Read an Apache mime.types and parse it. +parseExtMapFile ∷ FilePath → IO ExtMap parseExtMapFile fpath - = do file <- B.readFile fpath - case parse (allowEOF extMapP) file of - (Success xs, _) -> return $ compile xs - (_, input') -> let near = B.unpack $ B.take 100 input' - in - fail ("Failed to parse: " ++ fpath ++ " (near: " ++ near ++ ")") - - -extMapP :: Parser [ (MIMEType, [String]) ] -extMapP = do xs <- many (comment <|> validLine <|> emptyLine) - eof + = do file ← B.readFile fpath + case LP.parse extMapP file of + LP.Done _ xs + → case compile xs of + Right m → return m + Left e → fail (concat [ "Duplicate extension \"" + , show e + , "\" in: " + , fpath + ]) + LP.Fail _ _ e + → fail ("Failed to parse: " ⧺ fpath ⧺ ": " ⧺ e) + +extMapP ∷ Parser [ (MIMEType, [Text]) ] +extMapP = do xs ← P.many (try comment <|> try validLine <|> emptyLine) + endOfInput return $ catMaybes xs where - spc = oneOf " \t" + isSpc ∷ Char → Bool + isSpc c = c ≡ '\x20' ∨ c ≡ '\x09' - comment = do many spc - char '#' - many $ satisfy (/= '\n') + comment ∷ Parser (Maybe (MIMEType, [Text])) + comment = do skipWhile isSpc + void $ char '#' + skipWhile (≢ '\x0A') return Nothing - validLine = do many spc - mime <- mimeTypeP - many spc - exts <- sepBy token (many spc) + validLine ∷ Parser (Maybe (MIMEType, [Text])) + validLine = do skipWhile isSpc + mime ← mimeType + skipWhile isSpc + exts ← sepBy extP (skipWhile isSpc) return $ Just (mime, exts) - emptyLine = oneOf " \t\n" >> return Nothing + extP ∷ Parser Text + extP = decodeUtf8 <$> takeWhile1 (\c → (¬) (isSpc c ∨ c ≡ '\x0A')) + emptyLine ∷ Parser (Maybe (MIMEType, [Text])) + emptyLine = do skipWhile isSpc + void $ char '\x0A' + return Nothing -compile :: [ (MIMEType, [String]) ] -> Map String MIMEType -compile = M.fromList . foldr (++) [] . map tr +compile ∷ Ord k ⇒ [(v, [k])] → Either (k, v, v) (Map k v) +compile = go (∅) ∘ concat ∘ map tr where - tr :: (MIMEType, [String]) -> [ (String, MIMEType) ] - tr (mime, exts) = [ (ext, mime) | ext <- exts ] - - -outputExtMapAsHS :: ExtMap -> FilePath -> IO () -outputExtMapAsHS extMap fpath - = let hsModule = HsModule undefined modName (Just exports) imports decls - modName = Module "Network.HTTP.Lucu.MIMEType.DefaultExtensionMap" - exports = [HsEVar (UnQual (HsIdent "defaultExtensionMap"))] - imports = [ HsImportDecl undefined (Module "Network.HTTP.Lucu.MIMEType") False Nothing Nothing - , HsImportDecl undefined (Module "Data.Map") True (Just (Module "M")) Nothing - , HsImportDecl undefined (Module "Data.Map") False Nothing (Just (False, [HsIAbs (HsIdent "Map")])) - ] - decls = [ HsTypeSig undefined [HsIdent "defaultExtensionMap"] - (HsQualType [] (HsTyApp (HsTyApp (HsTyCon (UnQual (HsIdent "Map"))) - (HsTyCon (UnQual (HsIdent "String")))) - (HsTyCon (UnQual (HsIdent "MIMEType"))))) - , HsFunBind [HsMatch undefined (HsIdent "defaultExtensionMap") - [] (HsUnGuardedRhs extMapExp) []] - ] - extMapExp = HsApp (HsVar (Qual (Module "M") (HsIdent "fromList"))) (HsList records) - comment = "{- !!! WARNING !!!\n" - ++ " This file is automatically generated from data/mime.types.\n" - ++ " DO NOT EDIT BY HAND OR YOU WILL REGRET -}\n\n" + tr ∷ (v, [k]) → [(k, v)] + tr (v, ks) = [(k, v) | k ← ks] + + go ∷ Ord k ⇒ Map k v → [(k, v)] → Either (k, v, v) (Map k v) + go m [] = Right m + go m ((k, v):xs) + = case M.insertLookupWithKey' f k v m of + (Nothing, m') → go m' xs + (Just v0, _ ) → Left (k, v0, v) + + f ∷ k → v → v → v + f _ _ = id + +-- |@'serializeExtMap' extMap moduleName variableName@ generates a +-- Haskell source code which contains the following things: +-- +-- * A definition of module named @moduleName@. +-- +-- * @variableName :: 'ExtMap'@ whose content is the serialised +-- @extMap@. +-- +-- The module "Network.HTTP.Lucu.MIMEType.DefaultExtensionMap" is +-- surely generated using this function. +serializeExtMap ∷ ExtMap → String → String → String +serializeExtMap extMap moduleName variableName + = let hsModule = Module (⊥) (ModuleName moduleName) modPragma + Nothing (Just exports) imports decls + modPragma = [ LanguagePragma (⊥) [ name (show OverloadedStrings) ] ] + exports = [ EVar (UnQual (name variableName)) ] + imports = [ ImportDecl (⊥) (ModuleName "Network.HTTP.Lucu.MIMEType") + False False Nothing Nothing Nothing + , ImportDecl (⊥) (ModuleName "Network.HTTP.Lucu.MIMEType.Guess") + False False Nothing Nothing Nothing + , ImportDecl (⊥) (ModuleName "Data.Ascii") + False False Nothing Nothing (Just (False, [])) + , ImportDecl (⊥) (ModuleName "Data.Map") + True False Nothing (Just (ModuleName "M")) Nothing + ] + decls = [ TypeSig (⊥) [name variableName] + (TyCon (UnQual (name "ExtMap"))) + , nameBind (⊥) (name variableName) extMapExp + , InlineSig (⊥) False AlwaysActive (UnQual (name variableName)) + ] + comment = concat [ "{- !!! WARNING !!!\n" + , " This file is automatically generated.\n" + , " DO NOT EDIT BY HAND OR YOU WILL REGRET -}\n\n" + ] + extMapExp = qvar (ModuleName "M") (name "fromList") `app` listE records in - writeFile fpath $ comment ++ prettyPrint hsModule ++ "\n" + comment ⧺ prettyPrint hsModule ⧺ "\n" where - records :: [HsExp] + records ∷ [Exp] records = map record $ M.assocs extMap - record :: (String, MIMEType) -> HsExp + record ∷ (Text, MIMEType) → Exp record (ext, mime) - = HsTuple [HsLit (HsString ext), mimeToExp mime] - - mimeToExp :: MIMEType -> HsExp - mimeToExp (MIMEType maj min params) - = foldl appendParam (HsInfixApp - (HsLit (HsString maj)) - (HsQVarOp (UnQual (HsSymbol "+/+"))) - (HsLit (HsString min))) params - - appendParam :: HsExp -> (String, String) -> HsExp - appendParam x param - = HsInfixApp x (HsQVarOp (UnQual (HsSymbol "+:+"))) $ paramToExp param + = tuple [ strE (T.unpack ext) + , function "parseMIMEType" `app` strE (mimeToString mime) + ] - paramToExp :: (String, String) -> HsExp - paramToExp (name, value) - = HsInfixApp - (HsLit (HsString name)) - (HsQVarOp (UnQual (HsSymbol "+=+"))) - (HsLit (HsString value)) \ No newline at end of file + mimeToString ∷ MIMEType → String + mimeToString = A.toString ∘ A.fromAsciiBuilder ∘ printMIMEType