diff --git a/pdfinfo.cabal b/pdfinfo.cabal --- a/pdfinfo.cabal +++ b/pdfinfo.cabal @@ -1,21 +1,23 @@ name: pdfinfo -version: 0.1.3.3 +version: 1.4.0 synopsis: Wrapper around the pdfinfo command. description: Just a wrapper around the pdfinfo command. See man pdfinfo. license: BSD3 license-file: LICENSE author: Chris Done maintainer: chrisdone@gmail.com -copyright: 2010 Chris Done +copyright: 2013 Chris Done category: Text build-type: Simple -cabal-version: >=1.2 +cabal-version: >=1.8 library - ghc-options: -O2 -Wall + ghc-options: -O2 -Wall exposed-modules: Text.PDF.Info hs-source-dirs: src/ build-depends: base >= 4 && < 5, old-locale >= 1.0.0.0 && < 1.1, time >= 1.1 && < 1.5, - process >= 1.0 && < 1.2, - mtl >= 1.1 && < 2.2 + process-extras, + text, + -- mtl-2.1 contains a severe bug + mtl >= 1.1 && < 2.1 || >= 2.1.1 && < 2.2 diff --git a/src/Text/PDF/Info.hs b/src/Text/PDF/Info.hs --- a/src/Text/PDF/Info.hs +++ b/src/Text/PDF/Info.hs @@ -1,6 +1,9 @@ -{-# LANGUAGE GeneralizedNewtypeDeriving, MultiParamTypeClasses, ViewPatterns #-} {-# LANGUAGE FlexibleContexts #-} +{-# LANGUAGE GeneralizedNewtypeDeriving #-} +{-# LANGUAGE ViewPatterns #-} +{-# LANGUAGE OverloadedStrings #-} {-# OPTIONS -Wall #-} + module Text.PDF.Info (-- * Reading PDF info pdfInfo @@ -16,44 +19,48 @@ ,readRight) where -import Prelude -import Control.Monad.Error -import System.Process -import Control.Applicative -import Control.Arrow -import Data.Char -import Data.Time -import System.Locale -import Control.Exception as E +import Control.Applicative +import Control.Arrow +import Control.Exception as E +import Control.Monad.Error +import Data.Text (Text) +import qualified Data.Text as T +import Data.Time +import Prelude +import System.Exit +import System.Locale +import System.Process.Text + -- | A type representing the output from the pdfinfo command. data PDFInfo = PDFInfo { - pdfInfoTitle :: Maybe String -- ^ Title - , pdfInfoSubject :: Maybe String -- ^ Subject - , pdfInfoAuthor :: Maybe String -- ^ Author: E.g. Chris Done - , pdfInfoCreator :: Maybe String -- ^ Creator: E.g. Microsoft Office Word 2007 - , pdfInfoProducer :: Maybe String -- ^ Producer: E.g. Microsoft Office Word 2007 - , pdfInfoCreationDate :: Maybe UTCTime -- ^ Creation Date - , pdfInfoModDate :: Maybe UTCTime -- ^ Modification Date - , pdfInfoTagged :: Maybe Bool -- ^ Tagged? - , pdfInfoPages :: Maybe Integer -- ^ Pages: E.g. 238 - , pdfInfoEncrypted :: Maybe Bool -- ^ Encrypted? - , pdfInfoPageSize :: Maybe PDFSize -- ^ Page: E.g. 595.32 x 841.92 pts (A4) - , pdfInfoFileSize :: Maybe Integer -- ^ File: E.g. 4061737 bytes - , pdfInfoOptimized :: Maybe Bool -- ^ Optimized? - , pdfInfoPDFVersion :: Maybe Double -- ^ PDF: E.g. 1.5 + pdfInfoTitle :: !(Maybe Text) -- ^ Title + , pdfInfoSubject :: !(Maybe Text) -- ^ Subject + , pdfInfoAuthor :: !(Maybe Text) -- ^ Author: E.g. Chris Done + , pdfInfoCreator :: !(Maybe Text) -- ^ Creator: E.g. Microsoft Office Word 2007 + , pdfInfoProducer :: !(Maybe Text) -- ^ Producer: E.g. Microsoft Office Word 2007 + , pdfInfoCreationDate :: !(Maybe UTCTime) -- ^ Creation Date + , pdfInfoModDate :: !(Maybe UTCTime) -- ^ Modification Date + , pdfInfoTagged :: !(Maybe Bool) -- ^ Tagged? + , pdfInfoPages :: !(Maybe Integer) -- ^ Pages: E.g. 238 + , pdfInfoEncrypted :: !(Maybe Bool) -- ^ Encrypted? + , pdfInfoPageSize :: !(Maybe PDFSize) -- ^ Page: E.g. 595.32 x 841.92 pts (A4) + , pdfInfoFileSize :: !(Maybe Integer) -- ^ File: E.g. 4061737 bytes + , pdfInfoOptimized :: !(Maybe Bool) -- ^ Optimized? + , pdfInfoPDFVersion :: !(Maybe Double) -- ^ PDF: E.g. 1.5 } deriving Show -- | Possible things that can go wrong while reading the info. -data PDFInfoError = - ParseError String -- ^ Couldn't parse a property value. - | ProcessError IOException -- ^ Error to do with the pdfinfo process. - | NoMessage -- ^ No message given. - | SomeError String -- ^ Some nonspecific error. +data PDFInfoError + = ParseError !String -- ^ Couldn't parse a property value. + | ProcessFailure !Text -- ^ Process exited with this stderr. + | ProcessError !IOException -- ^ Error to do with the pdfinfo process. + | NoMessage -- ^ No message given. + | SomeError String -- ^ Some nonspecific error. deriving Show -- | Size of the PDF in pts. -data PDFSize = PDFSize { pdfSizeW :: Float, pdfSizeH :: Float } +data PDFSize = PDFSize { pdfSizeW :: !Float, pdfSizeH :: !Float } deriving (Eq,Show) instance Error PDFInfoError where noMsg = NoMessage; strMsg = SomeError @@ -65,11 +72,14 @@ -- running the process. pdfInfo :: MonadIO m => FilePath -> m (Either PDFInfoError PDFInfo) pdfInfo path = liftIO $ loadInfo `E.catch` ioErrorHandler where - loadInfo = parse <$> readProcess "pdfinfo" [path] "" + loadInfo = do (code,out,err) <- readProcessWithExitCode "pdfinfo" ["-enc","UTF-8",path] "" + case code of + ExitSuccess -> return (parse out) + ExitFailure{} -> return (Left (ProcessFailure err)) ioErrorHandler = return . Left . ProcessError -- | Parse PDFInfo's output. -parse :: String -> Either PDFInfoError PDFInfo +parse :: Text -> Either PDFInfoError PDFInfo parse out = runParse $ PDFInfo <$> string "Title" <*> string "Subject" @@ -85,48 +95,41 @@ <*> integer "File size" <*> bool "Optimized" <*> floating "PDF version" - where string = get id date = get (>>= parseDate) size = get (>>= parseSize) bool = get $ fmap $ \yes -> yes == "yes" floating = readIt integer = readIt - readIt :: Read a => String -> ParsePDFInfo (Maybe a) + readIt :: Read a => Text -> ParsePDFInfo (Maybe a) readIt = get (>>= readRight) - - properties = map split . lines $ out + properties = map split . T.lines $ out get f name = case lookup name properties of - Just ok -> catchError (Just <$> (f $ return $ trim ok)) + Just ok -> catchError (Just <$> (f $ return $ T.strip ok)) (\_ -> return Nothing) Nothing -> return Nothing - split = second (drop 2) . span (/=':') - trim = bi reverse (dropWhile isSpace) . dropWhile isSpace + split = second (T.drop 2) . T.span (/=':') -- | Parse a page size. This is loosely defined. -parseSize :: String -> ParsePDFInfo PDFSize +parseSize :: Text -> ParsePDFInfo PDFSize parseSize s = - case words s of + case T.words s of ((readRight -> Right x):"x":(readRight -> Right y):_) -> return $ PDFSize x y _ -> throwError $ ParseError $ "Unable to read size: " ++ show s -- | Parse a date according to pdfinfo's format. -parseDate :: String -> ParsePDFInfo UTCTime +parseDate :: Text -> ParsePDFInfo UTCTime parseDate s = - case parseTime defaultTimeLocale "%a %b %e %H:%M:%S %Y" s of + case parseTime defaultTimeLocale "%a %b %e %H:%M:%S %Y" (T.unpack s) of Just ok -> return ok Nothing -> throwError $ ParseError $ "Unable to parse date: " ++ show s -- | Read a value, maybe, allow misc trailing data. -readRight :: (MonadError PDFInfoError m,Read a) => String -> m a +readRight :: (MonadError PDFInfoError m,Read a) => Text -> m a readRight s = - case reads s of + case reads (T.unpack s) of [(v,_)] -> return v _ -> throwError $ ParseError $ "Couldn't read value: " ++ show s - --- | Untwist a value, apply a function, twist it back. -bi :: (c1 -> c) -> (c -> c1) -> c1 -> c -bi g f = g . f . g