Trying to apply the same technique used for directObjects (apparently it ruins the performances)

Hufflepdf.cabal

@@ -17,19 +17,36 @@ cabal-version:       >=1.10
 
 library
   exposed-modules:     PDF
+                     , PDF.CMap
+                     , PDF.Content
                      , PDF.EOL
                      , PDF.Object
+                     , PDF.Object.Navigation
                      , PDF.Output
+                     , PDF.Parser
+                     , PDF.Pages
                      , PDF.Update
   other-modules:       Data.ByteString.Char8.Util
+                     , PDF.Content.Operator
+                     , PDF.Content.Operator.Color
+                     , PDF.Content.Operator.Common
+                     , PDF.Content.Operator.GraphicState
+                     , PDF.Content.Operator.Path
+                     , PDF.Content.Operator.Text
+                     , PDF.Content.Text
+                     , PDF.Encoding
+                     , PDF.Encoding.MacRoman
                      , PDF.Body
-                     , PDF.Parser
+                     , PDF.Font
   -- other-extensions:    
   build-depends:       attoparsec
                      , base >=4.9 && <4.13
                      , bytestring
                      , containers
                      , mtl
+                     , text
+                     , utf8-string
+                     , zlib
   hs-source-dirs:      src
   ghc-options:         -Wall
   default-language:    Haskell2010
@@ -51,3 +68,15 @@ executable             getObj
                      , zlib
   ghc-options:         -Wall
   default-language:    Haskell2010
+
+executable             getText
+  main-is:             examples/getText.hs
+  build-depends:       base
+                     , bytestring
+                     , containers
+                     , Hufflepdf
+                     , mtl
+                     , text
+                     , zlib
+  ghc-options:         -Wall
+  default-language:    Haskell2010

examples/getText.hs

@@ -0,0 +1,38 @@
+import qualified Data.ByteString.Char8 as BS (readFile)
+import qualified Data.Map as Map (toList)
+import qualified Data.Text.IO as Text (putStrLn)
+import PDF (Document(..), parseDocument)
+import PDF.Object (Content)
+import PDF.Pages (Page(..), get, getAll)
+import PDF.Update (unify)
+import System.Environment (getArgs)
+import System.Exit (die)
+import System.IO (BufferMode(..), hSetBuffering, stdout)
+
+onDoc :: FilePath -> (Content -> Either String a) -> IO a
+onDoc inputFile f = do
+  content <- fmap (unify . updates) . parseDocument <$> BS.readFile inputFile
+  case content >>= f of
+    Left someError -> die someError
+    Right value -> return value
+
+displayPage :: Page -> IO ()
+displayPage = mapM_ Text.putStrLn . contents
+
+wholeDoc :: FilePath -> IO ()
+wholeDoc inputFile = do
+  pages <- onDoc inputFile getAll
+  mapM_ (displayPage . snd) $ Map.toList pages
+
+singlePage :: FilePath -> Int -> IO ()
+singlePage inputFile pageNumber =
+  onDoc inputFile (`get` pageNumber) >>= displayPage
+
+main :: IO ()
+main = do
+  hSetBuffering stdout LineBuffering
+  args <- getArgs
+  case args of
+    [inputFile] -> wholeDoc inputFile
+    [inputFile, pageNumber] -> singlePage inputFile (read pageNumber)
+    _ -> die "Syntax: getText INPUT_FILE [PAGE_NUMBER]"

src/Data/ByteString/Char8/Util.hs

@@ -1,16 +1,91 @@
 module Data.ByteString.Char8.Util (
-      previous
+      B16Int(..)
+    , B256Int(..)
+    , b8ToInt
+    , b16ToBytes
+    , b16ToInt
+    , b256ToInt
+    , intToB256
+    , previous
     , subBS
+    , toBytes
+    , unescape
+    , utf16BEToutf8
   ) where
 
-import Data.ByteString.Char8 (ByteString)
-import qualified Data.ByteString.Char8 as BS (drop, index, take)
+import Data.ByteString (ByteString, snoc)
+import qualified Data.ByteString as BS (empty, foldl, length, pack, singleton, splitAt)
+import qualified Data.ByteString.Char8 as Char8 (
+    cons, drop, index, splitAt, take, uncons, unpack
+  )
+import Data.Text (Text)
+import Data.Text.Encoding (decodeUtf16BE)
 import Prelude hiding (length)
+import Text.Printf (printf)
+
+newtype B8Int = B8Int ByteString deriving Show
+newtype B16Int = B16Int ByteString deriving Show
+newtype B256Int = B256Int ByteString deriving Show
 
 previous :: Char -> Int -> ByteString -> Int
 previous char position byteString
-  | BS.index byteString position == char = position
+  | Char8.index byteString position == char = position
   | otherwise = previous char (position - 1) byteString
 
 subBS :: Int -> Int -> ByteString -> ByteString
-subBS offset length = BS.take length . BS.drop offset
+subBS offset length = Char8.take length . Char8.drop offset
+
+intToB256 :: Int -> B256Int
+intToB256 n
+  | n < 0x100 = B256Int . BS.singleton $ toEnum n
+  | otherwise =
+    let B256Int begining = intToB256 (n `div` 0x100) in
+    B256Int $ begining `snoc` (toEnum (n `mod` 0x100))
+
+b256ToInt :: B256Int -> Int
+b256ToInt (B256Int n) = BS.foldl (\k w -> 0x100*k + fromEnum w) 0 n
+
+toBytes :: Int -> Int -> ByteString
+toBytes 0 _ = BS.empty
+toBytes size n =
+  (toBytes (size - 1) (n `div` 0x100)) `snoc` (toEnum (n `mod` 0x100))
+
+b16ToBytes :: B16Int -> ByteString
+b16ToBytes (B16Int n) = BS.pack . fmap b16ToInt $ pairDigits n
+  where
+    pairDigits s =
+      case BS.length s of
+        0 -> []
+        1 -> [B16Int s]
+        _ ->
+          let (twoHexDigits, rest) = BS.splitAt 2 s in
+          (B16Int $ twoHexDigits):(pairDigits rest)
+
+fromBase :: (Num a, Read a) => Char -> ByteString -> a
+fromBase b = read . printf "0%c%s" b . Char8.unpack
+
+b16ToInt :: (Num a, Read a) => B16Int -> a
+b16ToInt (B16Int n) = fromBase 'x' n
+
+b8ToInt :: (Num a, Read a) => B8Int -> a
+b8ToInt (B8Int n) = fromBase 'o' n
+
+unescape :: ByteString -> ByteString
+unescape escapedBS =
+  case Char8.uncons escapedBS of
+    Nothing -> BS.empty
+    Just ('\\', s) -> unescapeChar s
+    Just (c, s) -> Char8.cons c (unescape s)
+  where
+    unescapeChar s =
+      case Char8.uncons s of
+        Nothing -> BS.empty
+        Just (c, s')
+          | c `elem` "()" -> Char8.cons c (unescape s')
+          | c `elem` "nrtbf" -> Char8.cons (read (printf "'\\%c'" c)) (unescape s')
+          | c `elem` ['0'..'7'] -> fromOctal (Char8.splitAt 3 s)
+          | otherwise -> Char8.cons c (unescape s')
+    fromOctal (code, s) = Char8.cons (toEnum $ b8ToInt (B8Int code)) (unescape s)
+
+utf16BEToutf8 :: ByteString -> Text
+utf16BEToutf8 = decodeUtf16BE

src/PDF.hs

@@ -21,7 +21,7 @@ import PDF.Object (
   )
 import qualified PDF.Output as Output (render, line)
 import PDF.Output (Output(..))
-import PDF.Parser (Parser, runParser, string, takeAll)
+import PDF.Parser (Parser, evalParser, string, takeAll)
 import Text.Printf (printf)
 
 data Document = Document {
@@ -83,7 +83,7 @@ findNextSection offset input =
 
 readStructures :: Int -> ByteString -> Either String [InputStructure]
 readStructures startXref input =
-  runParser structure () (BS.drop startXref input) >>= stopOrFollow
+  evalParser structure () (BS.drop startXref input) >>= stopOrFollow
   where
     stopOrFollow s@(Structure {trailer}) =
       case Map.lookup (Name "Prev") trailer of
@@ -96,7 +96,7 @@ readStructures startXref input =
 
 parseDocument :: ByteString -> Either String Document
 parseDocument input = do
-  (pdfVersion, eolStyle) <- runParser ((,) <$> version <*> EOL.parser) () input
+  (pdfVersion, eolStyle) <- evalParser ((,) <$> version <*> EOL.parser) () input
   startXref <- readStartXref eolStyle input
   structuresRead <- readStructures startXref input
   let updates = populate input <$> structuresRead

src/PDF/Body.hs

@@ -6,6 +6,7 @@ module PDF.Body (
 
 import Control.Applicative ((<|>))
 import Control.Monad.State (get, gets, modify)
+import Data.Attoparsec.ByteString.Char8 (option)
 import Data.ByteString.Char8 (ByteString)
 import qualified Data.ByteString.Char8 as BS (cons, drop, unpack)
 import Data.Map ((!))
@@ -18,7 +19,7 @@ import PDF.Object (
     , blank, dictionary, directObject, integer, line
   )
 import PDF.Output (ObjectId(..), Offset(..))
-import PDF.Parser (Parser, (<?>), block, char, on, option, runParser, takeAll)
+import PDF.Parser (Parser, (<?>), block, char, evalParser, on, takeAll)
 
 data UserState = UserState {
       input :: ByteString
@@ -109,7 +110,7 @@ occurrence =
 populate :: ByteString -> InputStructure -> Content
 populate input structure =
   let bodyInput = BS.drop (startOffset structure) input in
-  case runParser recurseOnOccurrences initialState bodyInput of
+  case evalParser recurseOnOccurrences initialState bodyInput of
     Left _ -> Content {occurrences = [], objects = Map.empty, docStructure}
     Right finalState ->
       let Flow {occurrencesStack, tmpObjects} = flow finalState in

src/PDF/CMap.hs

@@ -0,0 +1,159 @@
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedStrings #-}
+module PDF.CMap (
+      CMap
+    , CMappers
+    , CRange(..)
+    , cMap
+    , emptyCMap
+    , matches
+  ) where
+
+import Control.Applicative ((<|>), many)
+import Control.Monad.Fail (MonadFail(..))
+import Control.Monad.State (modify)
+import Data.Attoparsec.ByteString.Char8 (count)
+import Data.ByteString (ByteString)
+import qualified Data.ByteString as BS (drop, length, null, take)
+import Data.ByteString.Char8 (unpack)
+import Data.ByteString.Char8.Util (
+    B16Int(..), b16ToBytes, b16ToInt, toBytes, utf16BEToutf8
+  )
+import Data.Map (Map, union)
+import qualified Data.Map as Map (
+    adjust, empty, fromList, insertWith, lookup, toList
+  )
+import Data.Text (Text)
+import qualified PDF.EOL as EOL (charset, parser)
+import PDF.Font (Font)
+import PDF.Object (
+      DirectObject(..), Name, StringObject(..)
+    , blank, directObject, integer, line, stringObject
+  )
+import PDF.Parser (MonadParser, Parser, runParser, takeAll)
+import Prelude hiding (fail)
+
+type CMappers = Map Name CMap
+type Mapping = Map ByteString Text
+data CRange = CRange {
+      fromSequence :: ByteString
+    , toSequence :: ByteString
+    , mapping :: Mapping
+  } deriving Show
+type RangeSize = Int
+type CMap = Map RangeSize [CRange]
+
+toFont :: CMap -> Font
+toFont aCMap input
+  | BS.null input = Right ""
+  | otherwise = do
+    (output, remainingInput) <- trySizes input $ Map.toList aCMap
+    mappend output <$> toFont aCMap remainingInput
+  where
+    trySizes s [] = Left $ "No matching code found in font for " ++ unpack s
+    trySizes s ((size, cRanges):others) =
+      let prefix = BS.take size s in
+      case tryRanges prefix cRanges of
+        Nothing -> trySizes s others
+        Just outputSequence -> Right (outputSequence, BS.drop size s)
+    tryRanges :: ByteString -> [CRange] -> Maybe Text
+    tryRanges _ [] = Nothing
+    tryRanges prefix ((CRange {mapping}):cRanges) =
+      case Map.lookup prefix mapping of
+        Nothing -> tryRanges prefix cRanges
+        outputSequence -> outputSequence
+
+emptyCMap :: CMap
+emptyCMap = Map.empty
+
+matches :: ByteString -> CRange -> Bool
+matches code (CRange {fromSequence, toSequence}) =
+  fromSequence <= code && code <= toSequence
+
+cMap :: MonadFail m => ByteString -> m Font
+cMap = either fail (return . toFont . snd) . runParser
+  (many (codeRanges <|> cMapRange <|> cMapChar <|> ignoredLine))
+  emptyCMap
+  where
+    ignoredLine =
+      takeAll (not . (`elem` EOL.charset)) *> EOL.parser *> return ()
+
+codeRanges :: Parser CMap ()
+codeRanges = do
+  size <- integer <* line "begincodespacerange"
+  mapM_ createMapping =<< count size codeRange
+  line "endcodespacerange"
+  where
+    codeRange =
+      (,) <$> stringObject <* blank <*> stringObject <* EOL.parser
+
+createMapping :: (StringObject, StringObject) -> Parser CMap ()
+createMapping (Hexadecimal from, Hexadecimal to) = modify $
+  Map.insertWith (++) size [CRange {fromSequence, toSequence, mapping}]
+  where
+    fromSequence = b16ToBytes from
+    size = BS.length fromSequence
+    toSequence = b16ToBytes to
+    mapping = Map.empty
+createMapping _ = return ()
+
+cMapRange :: Parser CMap ()
+cMapRange = do
+  size <- integer <* line "beginbfrange"
+  mapM_ saveMapping =<< count size rangeMapping
+  line "endbfrange"
+  where
+    rangeMapping = (,,) 
+      <$> (stringObject <* blank)
+      <*> (stringObject <* blank)
+      <*> directObject <* EOL.parser
+      >>= mapFromTo
+
+saveMapping :: [(ByteString, Text)] -> Parser CMap ()
+saveMapping [] = return ()
+saveMapping assoc@((code, _):_) = modify $ Map.adjust insertCRange mappingSize
+  where
+    newMapping = Map.fromList assoc
+    mappingSize = BS.length code
+    appendMapping cRange =
+      cRange {mapping = mapping cRange `union` newMapping}
+    insertCRange = fmap (\cRange ->
+        if code `matches` cRange then appendMapping cRange else cRange
+      )
+
+cMapChar :: Parser CMap ()
+cMapChar = do
+  size <- integer <* line "beginbfchar"
+  saveMapping =<< count size charMapping <* line "endbfchar"
+  where
+    charMapping =
+      (,) <$> stringObject <* blank <*> stringObject <* EOL.parser
+      >>= pairMapping
+
+between :: B16Int -> B16Int -> [ByteString]
+between from@(B16Int s) to =
+  let size = BS.length s `div` 2 in
+  toBytes size <$> [b16ToInt from .. b16ToInt to]
+
+startFrom :: B16Int -> [ByteString]
+startFrom from@(B16Int s) =
+  let size = BS.length s `div` 2 in
+  toBytes size <$> [b16ToInt from .. ]
+
+mapFromTo :: MonadParser m => (StringObject, StringObject, DirectObject) -> m [(ByteString, Text)]
+mapFromTo (Hexadecimal from, Hexadecimal to, StringObject (Hexadecimal dstFrom)) =
+  return $ zip (between from to) (utf16BEToutf8 <$> startFrom dstFrom)
+
+mapFromTo (Hexadecimal from, Hexadecimal to, Array dstPoints) =
+  zip (between from to) <$> (mapM dstByteString dstPoints)
+  where
+    dstByteString (StringObject (Hexadecimal dst)) =
+      return . utf16BEToutf8 $ b16ToBytes dst
+    dstByteString _ = fail "Invalid for a replacement string"
+
+mapFromTo _ = fail "invalid range mapping found"
+
+pairMapping :: MonadParser m => (StringObject, StringObject) -> m (ByteString, Text)
+pairMapping (Hexadecimal from, Hexadecimal to) =
+  return (b16ToBytes  from, utf16BEToutf8 $ b16ToBytes  to)
+pairMapping _ = fail "invalid pair mapping found"

src/PDF/Content.hs

@@ -0,0 +1,82 @@
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE FlexibleContexts #-}
+module PDF.Content (
+      Content(..)
+    , ContentUnit(..)
+    , GraphicContextUnit(..)
+    , TextContext
+    , content
+    , parse
+  ) where
+
+import Control.Applicative ((<|>), many)
+import Control.Monad.State (MonadState(..), evalStateT, modify)
+import Data.ByteString (ByteString)
+import PDF.Content.Operator (Instruction, operator)
+import PDF.Object (blank, directObject)
+import PDF.Output (Output(..), line)
+import PDF.Parser (MonadParser, endOfInput, evalParser, peek, string)
+
+data GraphicContextUnit =
+    GraphicInstruction Instruction
+  | ContentUnit ContentUnit
+  deriving Show
+type TextContext = [Instruction]
+data ContentUnit =
+    GraphicContext [GraphicContextUnit]
+  | TextContext TextContext
+  deriving Show
+newtype Content = Content [ContentUnit] deriving Show
+
+content :: MonadParser m => m Content
+content = Content <$> many contentUnit <* blank <* endOfInput
+
+contentUnit :: MonadParser m => m ContentUnit
+contentUnit = next >>= handle
+  where
+    handle 'q' = GraphicContext <$> graphicContext
+    handle 'B' = TextContext <$> textContext
+    handle _ = fail "content unit"
+
+graphicContext :: MonadParser m => m [GraphicContextUnit]
+graphicContext = string "q" *> evalStateT graphicContextUnits []
+  where
+    graphicContextUnits = next >>= handle
+    handle 'Q' = string "Q" *> (reverse <$> get)
+    handle c
+      | c `elem` ['q', 'B'] =
+        (ContentUnit <$> contentUnit >>= push) *> graphicContextUnits
+      | otherwise =
+        (GraphicInstruction <$> instruction >>= push) *> graphicContextUnits
+
+instruction :: MonadParser m => m Instruction
+instruction = evalStateT stackParser []
+  where
+    stackParser = ((directObject <* blank >>= push) *> stackParser) <|> operator
+
+next :: MonadParser m => m Char
+next = blank *> peek
+
+push :: MonadState [a] m => a -> m ()
+push a = modify (a:)
+
+parse :: ByteString -> Either String Content
+parse = evalParser content ()
+
+textContext :: MonadParser m => m TextContext
+textContext = string "BT" *> evalStateT instructions []
+  where
+    instructions = next >>= handle
+    handle 'E' = string "ET" *> (reverse <$> get)
+    handle _ = (instruction >>= push) *> instructions
+
+instance Output Content where
+  output (Content contentUnits) = output contentUnits
+
+instance Output ContentUnit where
+  output (GraphicContext gc) = line "q" `mappend` output gc `mappend` line "Q"
+  output (TextContext tc) = line "BT" `mappend` output tc `mappend` line "ET"
+
+instance Output GraphicContextUnit where
+  output (GraphicInstruction gi) = output gi
+  output (ContentUnit cu) = output cu

src/PDF/Content/Operator.hs

@@ -0,0 +1,76 @@
+{-# LANGUAGE ConstraintKinds #-}
+{-# LANGUAGE FlexibleContexts #-}
+{-# LANGUAGE TypeSynonymInstances #-}
+{-# LANGUAGE FlexibleInstances #-}
+module PDF.Content.Operator (
+      Instruction
+    , Operator(..)
+    , operator
+  ) where
+
+import Control.Monad.Fail (MonadFail(..))
+import Control.Monad.State (MonadState(..))
+import Data.ByteString.Char8 (ByteString, pack, unpack)
+import Data.Char (toLower)
+import Data.Map (Map, (!?))
+import qualified Data.Map as Map (fromList)
+import qualified PDF.Content.Operator.Color as Color (Operator, signature)
+import PDF.Content.Operator.Common (Signature)
+import qualified PDF.Content.Operator.GraphicState as GraphicState (Operator, signature)
+import qualified PDF.Content.Operator.Path as Path (Operator, signature)
+import qualified PDF.Content.Operator.Text as Text (Operator, signature)
+import PDF.Object (DirectObject, regular)
+import PDF.Output (Output(..), join, line)
+import PDF.Parser (MonadParser, takeAll1)
+import Prelude hiding (fail)
+import Text.Printf (printf)
+
+data Operator =
+    GraphicState GraphicState.Operator
+  | Path Path.Operator
+  | Color Color.Operator
+  | Text Text.Operator 
+
+instance Show Operator where
+  show (GraphicState gcOp) = code gcOp
+  show (Path pOp) = code pOp
+  show (Color cOp) = code cOp
+  show (Text tOp) = code tOp
+
+type Instruction = (Operator, [DirectObject])
+
+instance Output Instruction where
+  output (op, args) = join " " ((output <$> args) ++ [line (show op)])
+
+operatorsTable :: Map ByteString (Signature Operator)
+operatorsTable = Map.fromList (
+      (prepare GraphicState <$> GraphicState.signature)
+    ++ (prepare Path <$> Path.signature)
+    ++ (prepare Color <$> Color.signature)
+    ++ (prepare Text <$> Text.signature)
+  )
+  where
+    prepare constructor (op, sig) = (pack $ code op, (constructor op, sig))
+
+code :: Show a => a -> String
+code = expand . show
+  where
+    expand "" = ""
+    expand (c:'_':s) = toLower c : expand s
+    expand ('s':'t':'a':'r':s) = '*' : expand s
+    expand ('Q':'u':'o':'t':'e':s) = '\'' : expand s
+    expand ('D':'Q':'u':'o':'t':'e':s) = '"' : expand s
+    expand (c:s) = c : expand s
+
+type StackParser m = (MonadState [DirectObject] m, MonadParser m)
+
+operator :: StackParser m => m Instruction
+operator = do
+  chunk <- takeAll1 regular
+  args <- reverse <$> get
+  case operatorsTable !? chunk of
+    Just (op, sig)
+      | sig args -> return (op, args)
+      | otherwise -> 
+        get >>= fail . printf "Operator %s with stack %s" (show op) . show
+    _ -> fail ("Unknown chunk " ++ unpack chunk)

src/PDF/Content/Operator/Color.hs

@@ -0,0 +1,27 @@
+module PDF.Content.Operator.Color (
+      Operator(..)
+    , signature
+  ) where
+
+import PDF.Content.Operator.Common (Signature)
+import PDF.Object (DirectObject(..))
+
+data Operator =
+    CS | C_s | SC | SCN | S_c | S_cn | G | G_ | RG | R_g | K | K_
+  deriving (Bounded, Enum, Show)
+
+signature :: [Signature Operator]
+signature = [
+      (CS, \l -> case l of [NameObject _] -> True ; _ -> False)
+    , (C_s, \l -> case l of [NameObject _] -> True ; _ -> False)
+    , (SC, \_ -> True)
+    , (SCN, \_ -> True)
+    , (S_c, \_ -> True)
+    , (S_cn, \_ -> True)
+    , (G, \l -> case l of [_] -> True ; _ -> False)
+    , (G_, \l -> case l of [_] -> True ; _ -> False)
+    , (RG, \l -> case l of [_, _, _] -> True ; _ -> False)
+    , (R_g, \l -> case l of [_, _, _] -> True ; _ -> False)
+    , (K, \l -> case l of [_, _, _, _] -> True ; _ -> False)
+    , (K_, \l -> case l of [_, _, _, _] -> True ; _ -> False)
+  ]

src/PDF/Content/Operator/Common.hs

@@ -0,0 +1,7 @@
+module PDF.Content.Operator.Common (
+    Signature
+  ) where
+
+import PDF.Object (DirectObject)
+
+type Signature a = (a, [DirectObject] -> Bool)

src/PDF/Content/Operator/GraphicState.hs

@@ -0,0 +1,24 @@
+module PDF.Content.Operator.GraphicState (
+      Operator(..)
+    , signature
+  ) where
+
+import PDF.Content.Operator.Common (Signature)
+import PDF.Object (DirectObject(..))
+
+data Operator =
+  C_m | W_ | J | J_ | M | D_ | R_i | I_ | G_s -- general graphic state
+  deriving (Bounded, Enum, Show)
+
+signature :: [Signature Operator]
+signature = [
+      (C_m, \l -> case l of [_, _, _, _, _, _] -> True ; _ -> False)
+    , (W_, \l -> case l of [_] -> True ; _ -> False)
+    , (J, \l -> case l of [_] -> True ; _ -> False)
+    , (J_, \l -> case l of [_] -> True ; _ -> False)
+    , (M, \l -> case l of [_] -> True ; _ -> False)
+    , (D_, \l -> case l of [_, _] -> True ; _ -> False)
+    , (R_i, \l -> case l of [_] -> True ; _ -> False)
+    , (I_, \l -> case l of [_] -> True ; _ -> False)
+    , (G_s, \l -> case l of [NameObject _] -> True ; _ -> False)
+  ]

src/PDF/Content/Operator/Path.hs

@@ -0,0 +1,35 @@
+module PDF.Content.Operator.Path (
+      Operator(..)
+    , signature
+  ) where
+
+import PDF.Content.Operator.Common (Signature)
+
+data Operator =
+    M_ | L_ | C_ | V_ | Y_ | H_ | R_e -- path construction
+  | S | S_ | F_ | F | Fstar | B | Bstar | B_ | B_star | N_ -- path painting
+  | W | Wstar -- clipping path
+  deriving (Bounded, Enum, Show)
+
+signature :: [Signature Operator]
+signature = [
+      (M_, \l -> case l of [_, _] -> True ; _ -> False)
+    , (L_, \l -> case l of [_, _] -> True ; _ -> False)
+    , (C_, \l -> case l of [_, _, _, _, _, _] -> True ; _ -> False)
+    , (V_, \l -> case l of [_, _, _, _] -> True ; _ -> False)
+    , (Y_, \l -> case l of [_, _, _, _] -> True ; _ -> False)
+    , (H_, \l -> case l of [] -> True ; _ -> False)
+    , (R_e, \l -> case l of [_, _, _, _] -> True ; _ -> False)
+    , (S, \l -> case l of [] -> True ; _ -> False)
+    , (S_, \l -> case l of [] -> True ; _ -> False)
+    , (F_, \l -> case l of [] -> True ; _ -> False)
+    , (F, \l -> case l of [] -> True ; _ -> False)
+    , (Fstar, \l -> case l of [] -> True ; _ -> False)
+    , (B, \l -> case l of [] -> True ; _ -> False)
+    , (Bstar, \l -> case l of [] -> True ; _ -> False)
+    , (B_, \l -> case l of [] -> True ; _ -> False)
+    , (B_star, \l -> case l of [] -> True ; _ -> False)
+    , (N_, \l -> case l of [] -> True ; _ -> False)
+    , (W, \l -> case l of [] -> True ; _ -> False)
+    , (Wstar, \l -> case l of [] -> True ; _ -> False)
+  ]

src/PDF/Content/Operator/Text.hs

@@ -0,0 +1,32 @@
+module PDF.Content.Operator.Text (
+      Operator(..)
+    , signature
+  ) where
+
+import PDF.Content.Operator.Common (Signature)
+import PDF.Object (DirectObject(..))
+
+data Operator =
+    Td | TD | Tm | Tstar -- text positioning
+  | TJ | Tj | Quote | DQuote -- text showing
+  | Tc | Tw | Tz | TL | Tf | Tr | Ts -- text state
+  deriving (Bounded, Enum, Show)
+
+signature :: [Signature Operator]
+signature = [
+      (Td, \l -> case l of [_, _] -> True ; _ -> False)
+    , (TD, \l -> case l of [_, _] -> True ; _ -> False)
+    , (Tm, \l -> case l of [_, _, _, _, _, _] -> True ; _ -> False)
+    , (Tstar, \l -> case l of [] -> True ; _ -> False)
+    , (TJ, \l -> case l of [Array _] -> True ; _ -> False)
+    , (Tj, \l -> case l of [StringObject _] -> True ; _ -> False)
+    , (Quote, \l -> case l of [StringObject _] -> True ; _ -> False)
+    , (DQuote, \l -> case l of [StringObject _] -> True ; _ -> False)
+    , (Tc, \l -> case l of [_] -> True ; _ -> False)
+    , (Tw, \l -> case l of [_] -> True ; _ -> False)
+    , (Tz, \l -> case l of [_] -> True ; _ -> False)
+    , (TL, \l -> case l of [_] -> True ; _ -> False)
+    , (Tf, \l -> case l of [NameObject _, _] -> True ; _ -> False)
+    , (Tr, \l -> case l of [_] -> True ; _ -> False)
+    , (Ts, \l -> case l of [_] -> True ; _ -> False)
+  ]

src/PDF/Content/Text.hs

@@ -0,0 +1,75 @@
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE ConstraintKinds #-}
+{-# LANGUAGE FlexibleContexts #-}
+module PDF.Content.Text (
+      format
+    , renderText
+  ) where
+
+import Control.Monad (foldM)
+import Control.Monad.Fail (MonadFail(..))
+import Control.Monad.Reader (MonadReader(..), asks, runReaderT)
+import Control.Monad.State (MonadState(..), evalStateT)
+import Data.ByteString.Char8 (pack)
+import Data.Map ((!))
+import Data.Text (Text)
+import PDF.Content (Content(..), ContentUnit(..), GraphicContextUnit(..))
+import PDF.Content.Operator (Instruction, Operator(..))
+import PDF.Content.Operator.Text (Operator(..))
+import PDF.Font (Font, FontSet, emptyFont)
+import PDF.Object (DirectObject(..), StringObject(..), toByteString)
+import Prelude hiding (fail)
+
+type TextContent m = (MonadReader FontSet m, MonadFail m)
+type FontContext m = (MonadState Font m, TextContent m)
+
+decodeString :: FontContext m => StringObject -> m Text
+decodeString input = do
+  font <- get
+  either fail return . font $ toByteString input
+
+renderText :: MonadFail m => FontSet -> Content -> m [Text]
+renderText fontSet (Content contentUnits) =
+  runReaderT (concat <$> mapM renderContentUnit contentUnits) fontSet
+
+renderContentUnit :: TextContent m => ContentUnit -> m [Text]
+renderContentUnit (GraphicContext graphicContextUnits) =
+  concat <$> mapM renderGraphicContextUnit graphicContextUnits
+renderContentUnit (TextContext instructions) =
+  evalStateT (concat <$> mapM renderInstruction instructions) emptyFont
+
+renderGraphicContextUnit :: TextContent m => GraphicContextUnit -> m [Text]
+renderGraphicContextUnit (GraphicInstruction _) = return []
+renderGraphicContextUnit (ContentUnit contentUnit) =
+  renderContentUnit contentUnit
+
+renderInstruction :: FontContext m => Instruction -> m [Text]
+renderInstruction (Text Tf, [NameObject fontName, _]) =
+  asks (! fontName) >>= put >> return []
+
+renderInstruction (Text Tstar, []) = return ["\n"]
+
+renderInstruction (Text TJ, [Array arrayObject]) =
+  replicate 1 <$> foldM appendText "" arrayObject
+  where
+    appendText t (StringObject outputString) =
+      mappend t <$> decodeString outputString
+    appendText t _ = return t
+
+renderInstruction (Text Tj, [StringObject outputString]) =
+  replicate 1 <$> decodeString outputString
+
+renderInstruction (Text Quote, [StringObject outputString]) =
+  (\t -> ["\n", t]) <$> decodeString outputString
+
+renderInstruction (Text DQuote, [StringObject outputString]) =
+  (\t -> ["\n", t]) <$> decodeString outputString
+
+renderInstruction _ = return []
+
+format :: String -> [Instruction]
+format s =
+  case break (== '\n') s of
+    ("", "") -> []
+    ("", left) -> (Text Tstar, []) : format (drop 1 left)
+    (line, left) -> (Text Tj, [StringObject . Literal $ pack line]) : format left

src/PDF/EOL.hs

@@ -6,14 +6,14 @@ module PDF.EOL (
   ) where
 
 import Control.Applicative ((<|>))
-import PDF.Parser (Parser, string)
+import PDF.Parser (MonadParser, string)
 
 data Style = CR | LF | CRLF deriving Show
 
 charset :: String
 charset = "\r\n"
 
-parser :: Parser s Style
+parser :: MonadParser m => m Style
 parser =
     (string "\r\n" >> return CRLF)
   <|> (string "\r" >> return CR)

src/PDF/Encoding.hs

@@ -0,0 +1,12 @@
+module PDF.Encoding (
+    encoding
+  ) where
+
+import Control.Monad.Fail (MonadFail(..))
+import PDF.Encoding.MacRoman (macRomanEncoding)
+import PDF.Font (Font)
+import Prelude hiding (fail)
+
+encoding :: MonadFail m => String -> m Font
+encoding "MacRomanEncoding" = return macRomanEncoding
+encoding s = fail $ "Unknown encoding " ++ s

src/PDF/Encoding/MacRoman.hs

@@ -0,0 +1,141 @@
+module PDF.Encoding.MacRoman (
+    macRomanEncoding
+  ) where
+
+import Data.ByteString.Char8 (unpack)
+import Data.Text (pack)
+import PDF.Font (Font)
+
+macRomanEncoding :: Font
+macRomanEncoding = Right . pack . fmap decode . unpack
+
+decode :: Char -> Char
+decode '\x80' = '\x00C4' -- LATIN CAPITAL LETTER A WITH DIAERESIS
+decode '\x81' = '\x00C5' -- LATIN CAPITAL LETTER A WITH RING ABOVE
+decode '\x82' = '\x00C7' -- LATIN CAPITAL LETTER C WITH CEDILLA
+decode '\x83' = '\x00C9' -- LATIN CAPITAL LETTER E WITH ACUTE
+decode '\x84' = '\x00D1' -- LATIN CAPITAL LETTER N WITH TILDE
+decode '\x85' = '\x00D6' -- LATIN CAPITAL LETTER O WITH DIAERESIS
+decode '\x86' = '\x00DC' -- LATIN CAPITAL LETTER U WITH DIAERESIS
+decode '\x87' = '\x00E1' -- LATIN SMALL LETTER A WITH ACUTE
+decode '\x88' = '\x00E0' -- LATIN SMALL LETTER A WITH GRAVE
+decode '\x89' = '\x00E2' -- LATIN SMALL LETTER A WITH CIRCUMFLEX
+decode '\x8A' = '\x00E4' -- LATIN SMALL LETTER A WITH DIAERESIS
+decode '\x8B' = '\x00E3' -- LATIN SMALL LETTER A WITH TILDE
+decode '\x8C' = '\x00E5' -- LATIN SMALL LETTER A WITH RING ABOVE
+decode '\x8D' = '\x00E7' -- LATIN SMALL LETTER C WITH CEDILLA
+decode '\x8E' = '\x00E9' -- LATIN SMALL LETTER E WITH ACUTE
+decode '\x8F' = '\x00E8' -- LATIN SMALL LETTER E WITH GRAVE
+decode '\x90' = '\x00EA' -- LATIN SMALL LETTER E WITH CIRCUMFLEX
+decode '\x91' = '\x00EB' -- LATIN SMALL LETTER E WITH DIAERESIS
+decode '\x92' = '\x00ED' -- LATIN SMALL LETTER I WITH ACUTE
+decode '\x93' = '\x00EC' -- LATIN SMALL LETTER I WITH GRAVE
+decode '\x94' = '\x00EE' -- LATIN SMALL LETTER I WITH CIRCUMFLEX
+decode '\x95' = '\x00EF' -- LATIN SMALL LETTER I WITH DIAERESIS
+decode '\x96' = '\x00F1' -- LATIN SMALL LETTER N WITH TILDE
+decode '\x97' = '\x00F3' -- LATIN SMALL LETTER O WITH ACUTE
+decode '\x98' = '\x00F2' -- LATIN SMALL LETTER O WITH GRAVE
+decode '\x99' = '\x00F4' -- LATIN SMALL LETTER O WITH CIRCUMFLEX
+decode '\x9A' = '\x00F6' -- LATIN SMALL LETTER O WITH DIAERESIS
+decode '\x9B' = '\x00F5' -- LATIN SMALL LETTER O WITH TILDE
+decode '\x9C' = '\x00FA' -- LATIN SMALL LETTER U WITH ACUTE
+decode '\x9D' = '\x00F9' -- LATIN SMALL LETTER U WITH GRAVE
+decode '\x9E' = '\x00FB' -- LATIN SMALL LETTER U WITH CIRCUMFLEX
+decode '\x9F' = '\x00FC' -- LATIN SMALL LETTER U WITH DIAERESIS
+decode '\xA0' = '\x2020' -- DAGGER
+decode '\xA1' = '\x00B0' -- DEGREE SIGN
+decode '\xA2' = '\x00A2' -- CENT SIGN
+decode '\xA3' = '\x00A3' -- POUND SIGN
+decode '\xA4' = '\x00A7' -- SECTION SIGN
+decode '\xA5' = '\x2022' -- BULLET
+decode '\xA6' = '\x00B6' -- PILCROW SIGN
+decode '\xA7' = '\x00DF' -- LATIN SMALL LETTER SHARP S
+decode '\xA8' = '\x00AE' -- REGISTERED SIGN
+decode '\xA9' = '\x00A9' -- COPYRIGHT SIGN
+decode '\xAA' = '\x2122' -- TRADE MARK SIGN
+decode '\xAB' = '\x00B4' -- ACUTE ACCENT
+decode '\xAC' = '\x00A8' -- DIAERESIS
+decode '\xAD' = '\x2260' -- NOT EQUAL TO
+decode '\xAE' = '\x00C6' -- LATIN CAPITAL LETTER AE
+decode '\xAF' = '\x00D8' -- LATIN CAPITAL LETTER O WITH STROKE
+decode '\xB0' = '\x221E' -- INFINITY
+decode '\xB1' = '\x00B1' -- PLUS-MINUS SIGN
+decode '\xB2' = '\x2264' -- LESS-THAN OR EQUAL TO
+decode '\xB3' = '\x2265' -- GREATER-THAN OR EQUAL TO
+decode '\xB4' = '\x00A5' -- YEN SIGN
+decode '\xB5' = '\x00B5' -- MICRO SIGN
+decode '\xB6' = '\x2202' -- PARTIAL DIFFERENTIAL
+decode '\xB7' = '\x2211' -- N-ARY SUMMATION
+decode '\xB8' = '\x220F' -- N-ARY PRODUCT
+decode '\xB9' = '\x03C0' -- GREEK SMALL LETTER PI
+decode '\xBA' = '\x222B' -- INTEGRAL
+decode '\xBB' = '\x00AA' -- FEMININE ORDINAL INDICATOR
+decode '\xBC' = '\x00BA' -- MASCULINE ORDINAL INDICATOR
+decode '\xBD' = '\x03A9' -- GREEK CAPITAL LETTER OMEGA
+decode '\xBE' = '\x00E6' -- LATIN SMALL LETTER AE
+decode '\xBF' = '\x00F8' -- LATIN SMALL LETTER O WITH STROKE
+decode '\xC0' = '\x00BF' -- INVERTED QUESTION MARK
+decode '\xC1' = '\x00A1' -- INVERTED EXCLAMATION MARK
+decode '\xC2' = '\x00AC' -- NOT SIGN
+decode '\xC3' = '\x221A' -- SQUARE ROOT
+decode '\xC4' = '\x0192' -- LATIN SMALL LETTER F WITH HOOK
+decode '\xC5' = '\x2248' -- ALMOST EQUAL TO
+decode '\xC6' = '\x2206' -- INCREMENT
+decode '\xC7' = '\x00AB' -- LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
+decode '\xC8' = '\x00BB' -- RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
+decode '\xC9' = '\x2026' -- HORIZONTAL ELLIPSIS
+decode '\xCA' = '\x00A0' -- NO-BREAK SPACE
+decode '\xCB' = '\x00C0' -- LATIN CAPITAL LETTER A WITH GRAVE
+decode '\xCC' = '\x00C3' -- LATIN CAPITAL LETTER A WITH TILDE
+decode '\xCD' = '\x00D5' -- LATIN CAPITAL LETTER O WITH TILDE
+decode '\xCE' = '\x0152' -- LATIN CAPITAL LIGATURE OE
+decode '\xCF' = '\x0153' -- LATIN SMALL LIGATURE OE
+decode '\xD0' = '\x2013' -- EN DASH
+decode '\xD1' = '\x2014' -- EM DASH
+decode '\xD2' = '\x201C' -- LEFT DOUBLE QUOTATION MARK
+decode '\xD3' = '\x201D' -- RIGHT DOUBLE QUOTATION MARK
+decode '\xD4' = '\x2018' -- LEFT SINGLE QUOTATION MARK
+decode '\xD5' = '\x2019' -- RIGHT SINGLE QUOTATION MARK
+decode '\xD6' = '\x00F7' -- DIVISION SIGN
+decode '\xD7' = '\x25CA' -- LOZENGE
+decode '\xD8' = '\x00FF' -- LATIN SMALL LETTER Y WITH DIAERESIS
+decode '\xD9' = '\x0178' -- LATIN CAPITAL LETTER Y WITH DIAERESIS
+decode '\xDA' = '\x2044' -- FRACTION SLASH
+decode '\xDB' = '\x20AC' -- EURO SIGN
+decode '\xDC' = '\x2039' -- SINGLE LEFT-POINTING ANGLE QUOTATION MARK
+decode '\xDD' = '\x203A' -- SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
+decode '\xDE' = '\xFB01' -- LATIN SMALL LIGATURE FI
+decode '\xDF' = '\xFB02' -- LATIN SMALL LIGATURE FL
+decode '\xE0' = '\x2021' -- DOUBLE DAGGER
+decode '\xE1' = '\x00B7' -- MIDDLE DOT
+decode '\xE2' = '\x201A' -- SINGLE LOW-9 QUOTATION MARK
+decode '\xE3' = '\x201E' -- DOUBLE LOW-9 QUOTATION MARK
+decode '\xE4' = '\x2030' -- PER MILLE SIGN
+decode '\xE5' = '\x00C2' -- LATIN CAPITAL LETTER A WITH CIRCUMFLEX
+decode '\xE6' = '\x00CA' -- LATIN CAPITAL LETTER E WITH CIRCUMFLEX
+decode '\xE7' = '\x00C1' -- LATIN CAPITAL LETTER A WITH ACUTE
+decode '\xE8' = '\x00CB' -- LATIN CAPITAL LETTER E WITH DIAERESIS
+decode '\xE9' = '\x00C8' -- LATIN CAPITAL LETTER E WITH GRAVE
+decode '\xEA' = '\x00CD' -- LATIN CAPITAL LETTER I WITH ACUTE
+decode '\xEB' = '\x00CE' -- LATIN CAPITAL LETTER I WITH CIRCUMFLEX
+decode '\xEC' = '\x00CF' -- LATIN CAPITAL LETTER I WITH DIAERESIS
+decode '\xED' = '\x00CC' -- LATIN CAPITAL LETTER I WITH GRAVE
+decode '\xEE' = '\x00D3' -- LATIN CAPITAL LETTER O WITH ACUTE
+decode '\xEF' = '\x00D4' -- LATIN CAPITAL LETTER O WITH CIRCUMFLEX
+decode '\xF0' = '\xF8FF' -- Apple logo
+decode '\xF1' = '\x00D2' -- LATIN CAPITAL LETTER O WITH GRAVE
+decode '\xF2' = '\x00DA' -- LATIN CAPITAL LETTER U WITH ACUTE
+decode '\xF3' = '\x00DB' -- LATIN CAPITAL LETTER U WITH CIRCUMFLEX
+decode '\xF4' = '\x00D9' -- LATIN CAPITAL LETTER U WITH GRAVE
+decode '\xF5' = '\x0131' -- LATIN SMALL LETTER DOTLESS I
+decode '\xF6' = '\x02C6' -- MODIFIER LETTER CIRCUMFLEX ACCENT
+decode '\xF7' = '\x02DC' -- SMALL TILDE
+decode '\xF8' = '\x00AF' -- MACRON
+decode '\xF9' = '\x02D8' -- BREVE
+decode '\xFA' = '\x02D9' -- DOT ABOVE
+decode '\xFB' = '\x02DA' -- RING ABOVE
+decode '\xFC' = '\x00B8' -- CEDILLA
+decode '\xFD' = '\x02DD' -- DOUBLE ACUTE ACCENT
+decode '\xFE' = '\x02DB' -- OGONEK
+decode '\xFF' = '\x02C7' -- CARON
+decode c = c             -- The rest is ASCII

src/PDF/Font.hs

@@ -0,0 +1,16 @@
+module PDF.Font (
+      Font
+    , FontSet
+    , emptyFont
+  ) where
+
+import Data.ByteString (ByteString)
+import Data.Map (Map)
+import Data.Text (Text)
+import PDF.Object (Name)
+
+type Font = ByteString -> Either String Text
+type FontSet = Map Name Font
+
+emptyFont :: Font
+emptyFont _ = Left "No fond loaded"

src/PDF/Object.hs

@@ -3,6 +3,7 @@
 {-# LANGUAGE FlexibleInstances #-}
 module PDF.Object (
       Content(..)
+    , Dictionary
     , DirectObject(..)
     , Flow(..)
     , IndexedObjects
@@ -12,9 +13,11 @@ module PDF.Object (
     , Number(..)
     , Object(..)
     , Occurrence(..)
+    , StringObject(..)
     , Structure(..)
     , XRefEntry(..)
     , XRefSection
+    , array
     , blank
     , dictionary
     , directObject
@@ -22,34 +25,36 @@ module PDF.Object (
     , integer
     , line
     , magicNumber
+    , name
+    , number
+    , regular
+    , stringObject
     , structure
+    , toByteString
   ) where
 
-import Control.Applicative ((<|>))
-import Data.ByteString.Char8 (ByteString)
-import qualified Data.ByteString.Char8 as BS (
-    concat, cons, pack, singleton, unpack
-  )
+import Control.Applicative ((<|>), many)
+import Data.Attoparsec.ByteString.Char8 (choice, count, option, sepBy)
+import Data.ByteString (ByteString)
+import qualified Data.ByteString as BS (concat)
+import qualified Data.ByteString.Char8 as Char8 (cons, pack, singleton, unpack)
+import Data.ByteString.Char8.Util (B16Int(..), b16ToBytes, unescape)
 import Data.Map (Map, (!), mapWithKey)
 import qualified Data.Map as Map (
     delete, empty, fromList, lookup, minViewWithKey, toList, union
   )
 import qualified PDF.EOL as EOL (charset, parser)
-import qualified PDF.Output as Output (concat, line, string)
+import qualified PDF.Output as Output (line, string)
 import PDF.Output (
       OBuilder, ObjectId(..), Offset(..), Output(..), Resource(..)
     , byteString, getObjectId, getOffset, getOffsets, join, newLine
     , saveOffset
   )
-import PDF.Parser (
-      Parser, (<?>)
-    , char, choice, count, decNumber, hexNumber, many, octDigit, oneOf, option
-    , sepBy, string, takeAll, takeAll1
-  )
+import PDF.Parser (MonadParser(..), Parser, (<?>), octDigit, oneOf)
 import Text.Printf (printf)
 
-line :: String -> Parser u ()
-line l = (string (BS.pack l) *> blank *> return ()) <?> printf "line «%s»" l
+line :: MonadParser m => String -> m ()
+line l = (string (Char8.pack l) *> blank *> return ()) <?> printf "line «%s»" l
 
 magicNumber :: ByteString
 magicNumber = "%PDF-"
@@ -60,8 +65,8 @@ eofMarker = "%%EOF"
 whiteSpaceCharset :: String
 whiteSpaceCharset = "\0\t\12 "
 
-blank :: Parser u ()
-blank = takeAll (`elem` (EOL.charset ++ whiteSpaceCharset)) *> return ()
+blank :: MonadParser m => m ()
+blank = takeAll (`elem` (EOL.charset ++ whiteSpaceCharset)) *> pure ()
 
 delimiterCharset :: String
 delimiterCharset = "()<>[]{}/%"
@@ -69,8 +74,8 @@ delimiterCharset = "()<>[]{}/%"
 regular :: Char -> Bool
 regular = not . (`elem` (EOL.charset ++ whiteSpaceCharset ++ delimiterCharset))
 
-integer :: (Read a, Num a) => Parser u a
-integer = read . BS.unpack <$> decNumber <* blank <?> "decimal integer"
+integer :: (Read a, Num a, MonadParser m) => m a
+integer = read . Char8.unpack <$> decNumber <* blank <?> "decimal integer"
 
 -------------------------------------
 --          OBJECTS
@@ -81,7 +86,7 @@ type IndexedObjects = Map ObjectId Object
 --
 -- Boolean
 --
-boolean :: Parser u Bool
+boolean :: MonadParser m => m Bool
 boolean =
   (string "true" *> return True) <|> (string "false" *> return False) <?> "boolean"
 
@@ -96,38 +101,42 @@ instance Output Number where
       (n, 0) -> printf "%d" (n :: Int)
       _ -> printf "%f" f
 
-number :: Parser u Number
-number = Number . read . BS.unpack <$>
-  (mappend <$> sign <*> (integerPart <|> BS.cons '0' <$> floatPart))
+number :: MonadParser m => m Number
+number = Number . read . Char8.unpack <$>
+  (mappend <$> sign <*> (integerPart <|> Char8.cons '0' <$> floatPart))
   <?> "number"
   where
     sign = string "-" <|> option "" (char '+' >> return "")
     integerPart = mappend <$> decNumber <*> option "" floatPart
-    floatPart = BS.cons <$> char '.' <*> (option "0" $ decNumber)
+    floatPart = Char8.cons <$> char '.' <*> (option "0" $ decNumber)
 
 --
 -- StringObject
 --
-data StringObject = Literal String | Hexadecimal String deriving Show
+data StringObject = Literal ByteString | Hexadecimal B16Int deriving Show
 
 instance Output StringObject where
-  output (Literal s) = Output.string (printf "(%s)" s)
-  output (Hexadecimal s) = Output.string (printf "<%s>" s)
+  output (Literal s) = Output.string (printf "(%s)" (Char8.unpack s))
+  output (Hexadecimal (B16Int n)) = Output.string (printf "<%s>" (Char8.unpack n))
 
-stringObject :: Parser u StringObject
+stringObject :: MonadParser m => m StringObject
 stringObject =
-    Literal . BS.unpack <$> (char '(' *> (BS.concat <$> literalString) <* char ')')
-  <|> Hexadecimal . BS.unpack <$> (char '<' *> hexNumber <* char '>')
+    Literal <$> (char '(' *> (BS.concat <$> literalString) <* char ')')
+  <|> Hexadecimal <$> (char '<' *> hexNumber <* char '>')
   <?> "string object (literal or hexadecimal)"
   where
     literalString = many literalStringBlock
     literalStringBlock = takeAll1 normalChar <|> matchingParenthesis <|> escapedChar
     normalChar = not . (`elem` ("\\()" :: String))
     matchingParenthesis =
-      mappend <$> (BS.cons <$> char '(' <*> literalStringBlock) <*> string ")"
+      mappend <$> (Char8.cons <$> char '(' <*> literalStringBlock) <*> string ")"
     escapedChar =
-      BS.cons <$> char '\\' <*> (BS.singleton <$> oneOf "nrtbf()\\" <|> octalCode)
-    octalCode = choice $ (\n -> BS.pack <$> count n octDigit) <$> [1..3]
+      Char8.cons <$> char '\\' <*> (Char8.singleton <$> oneOf "nrtbf()\\\n" <|> octalCode)
+    octalCode = choice $ (\n -> Char8.pack <$> count n octDigit) <$> [1..3]
+
+toByteString :: StringObject -> ByteString
+toByteString (Hexadecimal h) = b16ToBytes h
+toByteString (Literal s) = unescape s
 
 --
 -- Name
@@ -137,13 +146,13 @@ newtype Name = Name String deriving (Eq, Ord, Show)
 instance Output Name where
   output (Name n) = Output.string ('/':n)
 
-name :: Parser u Name
-name = Name . BS.unpack <$> (char '/' *> takeAll regular) <?> "name"
+name :: MonadParser m => m Name
+name = Name . Char8.unpack <$> (char '/' *> takeAll regular) <?> "name"
 
 --
 -- Array
 --
-array :: Parser u [DirectObject]
+array :: MonadParser m => m [DirectObject]
 array =
   char '[' *> blank *> directObject `sepBy` blank <* blank <* char ']' <?> "array"
 
@@ -153,14 +162,14 @@ array =
 type Dictionary = Map Name DirectObject
 
 instance Output Dictionary where
-  output dict =
+  output aDictionary =
     "<<" `mappend` keyValues `mappend` ">>"
     where
-      keyValues = join " " $ outputKeyVal <$> Map.toList dict
+      keyValues = join " " $ outputKeyVal <$> Map.toList aDictionary
       outputKeyVal :: (Name, DirectObject) -> OBuilder
-      outputKeyVal (key, val) = Output.concat [output key, " ", output val]
+      outputKeyVal (key, val) = mconcat [output key, " ", output val]
 
-dictionary :: Parser u Dictionary
+dictionary :: MonadParser m => m Dictionary
 dictionary =
   string "<<" *> blank *> keyValPairs <* string ">>" <?> "dictionary"
   where
@@ -170,7 +179,7 @@ dictionary =
 --
 -- Null
 --
-nullObject :: Parser u ()
+nullObject :: MonadParser m => m ()
 nullObject = string "null" *> return () <?> "null object"
 
 --
@@ -181,7 +190,7 @@ data IndirectObjCoordinates = IndirectObjCoordinates {
     , versionNumber :: Int
   } deriving Show
 
-reference :: Parser u IndirectObjCoordinates
+reference :: MonadParser m => m IndirectObjCoordinates
 reference = IndirectObjCoordinates
   <$> (fmap ObjectId integer) <*> integer <* char 'R' <?> "reference to an object"
 
@@ -204,23 +213,25 @@ instance Output DirectObject where
   output (NumberObject n) = output n
   output (StringObject s) = output s
   output (NameObject n) = output n
-  output (Array a) = Output.concat ["[", join " " a, "]"]
+  output (Array a) = mconcat ["[", join " " a, "]"]
   output (Dictionary d) = output d
   output (Null) = "null"
   output (Reference (IndirectObjCoordinates {objectId, versionNumber})) =
     Output.string (printf "%d %d R" (getObjectId objectId) versionNumber)
 
-directObject :: Parser u DirectObject
-directObject =
-    Boolean <$> boolean
-  <|> Reference <$> reference {- defined before Number because Number is a prefix of it -}
-  <|> NumberObject <$> number
-  <|> StringObject <$> stringObject
-  <|> NameObject <$> name
-  <|> Array <$> array
-  <|> Dictionary <$> dictionary
-  <|> const Null <$> nullObject
-  <?> "direct object"
+directObject :: MonadParser m => m DirectObject
+directObject = (peek >>= dispatch) <?> "direct object"
+  where
+    dispatch 't' = Boolean <$> boolean
+    dispatch 'f' = Boolean <$> boolean
+    dispatch '(' = StringObject <$> stringObject
+    dispatch '<' = StringObject <$> stringObject <|> Dictionary <$> dictionary
+    dispatch '/' = NameObject <$> name
+    dispatch '[' = Array <$> array
+    dispatch 'n' = nullObject *> return Null
+    dispatch _ =
+        Reference <$> reference {- defined before Number because Number is a prefix of it -}
+      <|> NumberObject <$> number
 
 --
 -- Object
@@ -235,7 +246,7 @@ data Object =
 
 instance Output Object where
   output (Direct d) = output d
-  output (Stream {header, streamContent}) = Output.concat [
+  output (Stream {header, streamContent}) = mconcat [
         output header, newLine
       , Output.line "stream"
       , byteString streamContent
@@ -250,7 +261,7 @@ data Occurrence = Comment String | Indirect IndirectObjCoordinates deriving Show
 outputOccurrence :: IndexedObjects -> Occurrence -> OBuilder
 outputOccurrence _ (Comment c) = Output.line c
 outputOccurrence objects (Indirect (IndirectObjCoordinates {objectId, versionNumber})) =
-  saveOffset (Object objectId) >> Output.concat [
+  saveOffset (Object objectId) >> mconcat [
         Output.line (printf "%d %d obj" (getObjectId objectId) versionNumber)
       , output (objects ! objectId), newLine
       , Output.line "endobj"
@@ -383,7 +394,7 @@ outputBody (occurrences, objects) =
 instance Output Content where
   output (Content {occurrences, objects, docStructure}) =
     fmap (updateXrefs xRef) <$> getOffsets (outputBody (occurrences, objects))
-    >>= \(body, (xref, startXRef)) -> Output.concat [
+    >>= \(body, (xref, startXRef)) -> mconcat [
           body
         , Output.line "xref"
         , output xref

src/PDF/Object/Navigation.hs

@@ -0,0 +1,82 @@
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE ConstraintKinds #-}
+{-# LANGUAGE FlexibleContexts #-}
+module PDF.Object.Navigation (
+      (//)
+    , dictionaryById
+    , getDictionary
+    , getField
+    , follow
+    , objectById
+    , openStream
+    , origin
+  ) where
+
+import Codec.Compression.Zlib (decompress)
+import Control.Monad.Reader (MonadReader(..))
+import Control.Monad.Fail (MonadFail(..))
+import Data.ByteString (ByteString)
+import qualified Data.ByteString.Lazy as Lazy (fromStrict, toStrict)
+import Data.Map ((!))
+import qualified Data.Map as Map (lookup)
+import PDF.Object (
+      Content(..), Dictionary, DirectObject(..), IndirectObjCoordinates(..)
+    , Name(..), Object(..), Structure(..)
+  )
+import PDF.Output (ObjectId)
+import Prelude hiding (fail)
+import Text.Printf (printf)
+
+type PDFContent m = (MonadReader Content m, MonadFail m)
+
+castDictionary :: MonadFail m => Object -> m Dictionary
+castDictionary (Direct (Dictionary aDict)) = return aDict
+castDictionary obj = expected "dictionary : " obj
+
+castObjectId :: MonadFail m => DirectObject -> m ObjectId
+castObjectId (Reference (IndirectObjCoordinates {objectId})) = return objectId
+castObjectId directObject = expected "reference" directObject
+
+dictionaryById :: PDFContent m => ObjectId -> m Dictionary
+dictionaryById objectId = objectById objectId >>= castDictionary
+
+expected :: (MonadFail m, Show a) => String -> a -> m b
+expected name = fail . printf "Not a %s: %s" name . show
+
+getField :: MonadFail m => String -> Dictionary -> m DirectObject
+getField key aDictionary =
+  maybe (fail errorMessage) return (Map.lookup (Name key) aDictionary)
+  where
+    errorMessage =
+      printf "Key %s not found in dictionary %s" key (show aDictionary)
+
+follow :: PDFContent m => DirectObject -> m Object
+follow directObject = castObjectId directObject >>= objectById
+
+objectById :: PDFContent m => ObjectId -> m Object
+objectById objectId = do
+  content <- ask
+  return (objects content ! objectId)
+
+getDictionary :: PDFContent m => DirectObject -> m Dictionary
+getDictionary (Dictionary aDictionary) = return aDictionary
+getDictionary (Reference (IndirectObjCoordinates {objectId})) =
+  objectById objectId >>= castDictionary
+getDictionary aDirectObject =
+  expected "resource (dictionary or reference)" aDirectObject
+
+(//) :: PDFContent m => Dictionary -> [String] -> m DirectObject
+(//) aDict [] = return $ Dictionary aDict
+(//) aDict [key] = getField key aDict
+(//) aDict (key:keys) = getField key aDict >>= getDictionary >>= (// keys)
+
+origin :: PDFContent m => m Dictionary
+origin = trailer . docStructure <$> ask
+
+openStream :: MonadFail m => Object -> m ByteString
+openStream (Stream {header, streamContent}) = return $
+  case Map.lookup (Name "Filter") header of
+    Just (NameObject (Name "FlateDecode")) ->
+      Lazy.toStrict . decompress . Lazy.fromStrict $ streamContent
+    _ -> streamContent
+openStream obj = expected "stream" obj

src/PDF/Output.hs

@@ -12,7 +12,6 @@ module PDF.Output (
     , Resource(..)
     , byteString
     , char
-    , concat
     , getOffsets
     , join
     , line
@@ -32,7 +31,6 @@ import qualified Data.Map as Map (singleton)
 import Data.String (IsString(..))
 import Control.Monad.RWS (RWS, runRWS, ask, get, listen, modify, tell)
 import qualified PDF.EOL as EOL (Style(..))
-import Prelude hiding (concat)
 
 newtype ObjectId = ObjectId {getObjectId :: Int} deriving (Enum, Eq, Ord, Show)
 newtype Offset = Offset {getOffset :: Int} deriving (Show)
@@ -66,9 +64,6 @@ getOffsets (OContext builder) =
 append :: OBuilder -> OBuilder -> OBuilder
 append (OContext a) (OContext b) = OContext (mappend <$> a <*> b)
 
-concat :: [OBuilder] -> OBuilder
-concat = foldl mappend mempty
-
 #if MIN_VERSION_base(4,11,0)
 instance Semigroup OBuilder where
   (<>) = append
@@ -94,7 +89,7 @@ instance Output Bool where
   output True = string "true"
 
 instance Output a => Output [a] where
-  output = concat . fmap output
+  output = mconcat . fmap output
 
 join :: Output a => String -> [a] -> OBuilder
 join _ [] = mempty
@@ -116,7 +111,7 @@ char :: Char -> OBuilder
 char c = lift char8 c <* offset (+1)
 
 string :: String -> OBuilder
-string s = lift string8 s <* offset (+ toEnum (length s))
+string s = lift string8 s <* offset (+ length s)
 
 line :: String -> OBuilder
 line l = string l `mappend` newLine

src/PDF/Pages.hs

@@ -0,0 +1,133 @@
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE GeneralizedNewtypeDeriving #-}
+module PDF.Pages (
+      Page(..)
+    , get
+    , getAll
+  ) where
+
+import Control.Applicative (Alternative(..), (<|>))
+import Control.Monad (MonadPlus(..), foldM)
+import Control.Monad.Fail (MonadFail(..))
+import Control.Monad.RWS (RWST(..), evalRWST, modify)
+import qualified Control.Monad.RWS as RWS (get)
+import Data.Map (Map)
+import qualified Data.Map as Map (empty, fromList, insert, lookup, toList)
+import Data.Text (Text)
+import PDF.CMap (cMap)
+import qualified PDF.Content as Content (parse)
+import PDF.Content.Text (renderText)
+import PDF.Encoding (encoding)
+import PDF.Font (Font, FontSet)
+import PDF.Object (
+      Content(..), Dictionary, DirectObject(..), IndirectObjCoordinates(..)
+    , Name(..)
+  ,)
+import PDF.Object.Navigation (
+    (//), dictionaryById, getDictionary, getField, follow, openStream, origin
+  )
+import PDF.Output (ObjectId(..))
+import Prelude hiding (fail)
+import Text.Printf (printf)
+
+type CachedFonts = Map ObjectId Font
+newtype Error a = Error {
+    runError :: Either String a
+  } deriving (Alternative, Functor, Applicative, Monad, MonadPlus)
+type T = RWST Content () CachedFonts Error
+data Page = Page {
+      contents :: [Text]
+    , source :: ObjectId
+  }
+
+instance MonadFail Error where
+  fail = Error . Left
+
+getFontDictionary :: Dictionary -> T Dictionary
+getFontDictionary pageDict =
+    ((pageDict // ["Resources", "Font"]) >>= getDictionary)
+  <|> return Map.empty
+
+cache :: (ObjectId -> T Font) -> ObjectId -> T Font
+cache loader objectId =
+  (maybe load return . Map.lookup objectId) =<< RWS.get
+  where
+    load = do
+      value <- loader objectId
+      modify $ Map.insert objectId value
+      return value
+
+loadFont :: ObjectId -> T Font
+loadFont objectId = dictionaryById objectId >>= tryMappings
+  where
+    tryMappings dictionary =
+        loadCMap dictionary
+      <|> (getField "Encoding" dictionary >>= loadEncoding)
+      <|> (fail $ unknownFormat (show objectId) (show dictionary))
+    unknownFormat = printf "Unknown font format for object #%s : %s"
+    loadCMap :: Dictionary -> T Font
+    loadCMap dictionary =
+      getField "ToUnicode" dictionary >>= follow >>= openStream >>= cMap
+    loadEncoding :: DirectObject -> T Font
+    loadEncoding (NameObject (Name name)) = encoding name
+    loadEncoding directObject =
+      fail $ printf "Encoding must be a name, not that : %s" $ show directObject
+
+loadFonts :: Dictionary -> T FontSet
+loadFonts = foldM addFont Map.empty . Map.toList
+  where
+    addFont :: FontSet -> (Name, DirectObject) -> T FontSet
+    addFont output (name, Reference (IndirectObjCoordinates {objectId})) =
+      flip (Map.insert name) output <$> cache loadFont objectId
+    addFont output _ = return output
+
+several :: DirectObject -> [DirectObject]
+several (Array l) =  l
+several directObject = [directObject]
+
+pagesList :: T [ObjectId]
+pagesList = do
+  pages <- (origin :: T Dictionary) >>= (// ["Root", "Pages"]) >>= getDictionary
+  case Map.lookup (Name "Kids") pages of
+    Just (Array kids) -> return $ getReferences kids
+    _ -> return []
+
+getReferences :: [DirectObject] -> [ObjectId]
+getReferences objects = do
+  object <- objects
+  case object of
+    Reference (IndirectObjCoordinates {objectId}) -> [objectId]
+    _ -> []
+
+extractText :: Dictionary -> T [Text]
+extractText pageDict = do
+  fonts <- loadFonts =<< getFontDictionary pageDict
+  objects <- several <$> getField "Contents" pageDict
+  concat <$> mapM (loadContent fonts) objects
+  where
+    loadContent :: FontSet -> DirectObject -> T [Text]
+    loadContent fonts directObject =
+        follow directObject
+      >>= openStream
+      >>= (either fail return . Content.parse)
+      >>= renderText fonts
+
+loadPage :: ObjectId -> T Page
+loadPage source = do
+  contents <- extractText =<< dictionaryById source
+  return $ Page {contents, source}
+
+getAll :: Content -> Either String (Map Int Page)
+getAll content = runError $ fst <$> evalRWST getPages content Map.empty
+  where
+    numbered = Map.fromList . zip [1..]
+    getPages = numbered <$> (mapM loadPage =<< pagesList)
+
+get :: Content -> Int -> Either String Page
+get content pageNumber
+  | pageNumber < 1 = Left "Pages start at 1"
+  | otherwise = runError $ fst <$> evalRWST getPage content Map.empty
+  where
+    firstPage [] = fail "Page is out of bounds"
+    firstPage (p:_) = loadPage p
+    getPage = drop (pageNumber - 1) <$> pagesList >>= firstPage

src/PDF/Parser.hs

@@ -1,56 +1,75 @@
+{-# LANGUAGE FlexibleInstances #-}
+{-# LANGUAGE ConstraintKinds #-}
+{-# LANGUAGE UndecidableInstances #-}
 module PDF.Parser (
-      Parser
+      MonadParser(..)
+    , Parser
     , (<?>)
-    , block
-    , char
-    , choice
-    , count
-    , decNumber
-    , hexNumber
-    , many
     , octDigit
     , on
-    , oneOf
-    , option
     , runParser
-    , sepBy
-    , string
-    , takeAll
-    , takeAll1
+    , evalParser
   ) where
 
-import Control.Applicative ((<|>), empty)
+import Control.Applicative (Alternative, (<|>))
+import Control.Monad (MonadPlus)
+import Control.Monad.Fail (MonadFail(..))
 import Control.Monad.State (StateT(..), evalStateT)
-import Control.Monad.Trans (lift)
+import Control.Monad.Trans (MonadTrans(..))
 import qualified Data.Attoparsec.ByteString.Char8 as Atto (
-    Parser, char, parseOnly, satisfy, string, take, takeWhile, takeWhile1
+      Parser, char, endOfInput, parseOnly, peekChar', satisfy, string, take
+    , takeWhile, takeWhile1
   )
 import Data.ByteString (ByteString)
+import Data.ByteString.Char8.Util (B16Int(..))
 import Data.Char (toLower)
 import Data.Set (Set)
 import qualified Data.Set as Set (fromList, member, unions)
+import Prelude hiding (fail)
+
+type MonadDeps m = (MonadFail m, MonadPlus m)
+
+class MonadDeps m => MonadParser m where
+  block :: Int -> m ByteString
+  char :: Char -> m Char
+  decNumber :: m ByteString
+  endOfInput :: m ()
+  hexNumber :: m B16Int
+  oneOf :: String -> m Char
+  peek :: m Char
+  string :: ByteString -> m ByteString
+  takeAll :: (Char -> Bool) -> m ByteString
+  takeAll1 :: (Char -> Bool) -> m ByteString
+
+instance MonadParser Atto.Parser where
+  block = Atto.take
+  char = Atto.char
+  endOfInput = Atto.endOfInput
+  decNumber = Atto.takeWhile1 (`Set.member` digits)
+  hexNumber = B16Int <$> Atto.takeWhile1 (`Set.member` hexDigits)
+  oneOf charSet = Atto.satisfy (`elem` charSet)
+  peek = Atto.peekChar'
+  string s = Atto.string s <?> show s
+  takeAll = Atto.takeWhile
+  takeAll1 = Atto.takeWhile1
+
+instance (MonadParser m, MonadTrans t, MonadDeps (t m)) => MonadParser (t m) where
+  block = lift . block
+  char = lift . char
+  endOfInput = lift $ endOfInput
+  decNumber = lift $ decNumber
+  hexNumber = lift $ hexNumber
+  oneOf = lift . oneOf
+  peek = lift $ peek
+  string = lift . string
+  takeAll = lift . takeAll
+  takeAll1 = lift . takeAll1
 
 type Parser s = StateT s Atto.Parser
 
-(<?>) :: Parser s a -> String -> Parser s a
+(<?>) :: (Alternative m, MonadFail m) => m a -> String -> m a
 (<?>) parser debugMessage = parser <|> fail debugMessage
 
-block :: Int -> Parser s ByteString
-block = lift . Atto.take
-
-char :: Char -> Parser s Char
-char = lift . Atto.char
-
-choice :: [Parser s a] -> Parser s a
-choice = foldr (<|>) empty
-
-count :: Int -> Parser s a -> Parser s [a]
-count 0 _ = return []
-count n p = (:) <$> p <*> count (n-1) p
-
-decNumber :: Parser s ByteString
-decNumber = lift $ Atto.takeWhile1 (`Set.member` digits)
-
 digits :: Set Char
 digits = Set.fromList ['0'..'9']
 
@@ -59,13 +78,7 @@ hexDigits = Set.unions [digits, Set.fromList af, Set.fromList $ toLower <$> af]
   where
     af = ['A'..'F']
 
-hexNumber :: Parser s ByteString
-hexNumber = lift $ Atto.takeWhile1 (`Set.member` hexDigits)
-
-many :: Parser s a -> Parser s [a]
-many parser = (:) <$> parser <*> many parser <|> return []
-
-octDigit :: Parser s Char
+octDigit :: MonadParser m => m Char
 octDigit = oneOf ['0'..'7']
 
 on :: Parser s a -> ByteString -> Parser s (Either String a)
@@ -74,25 +87,8 @@ on (StateT parserF) input = StateT $ \state ->
     Left errorMsg -> return (Left errorMsg, state)
     Right (result, newState) -> return (Right result, newState)
 
-oneOf :: String -> Parser s Char
-oneOf charSet = lift $ Atto.satisfy (`elem` charSet)
+runParser :: Parser s a -> s -> ByteString -> Either String (a, s)
+runParser parser initState = Atto.parseOnly (runStateT parser initState)
 
-option :: a -> Parser s a -> Parser s a
-option defaultValue p = p <|> pure defaultValue
-
-runParser :: Parser s a -> s -> ByteString -> Either String a
-runParser parser initState =
-  Atto.parseOnly (evalStateT parser initState)
-
-sepBy :: Parser s a -> Parser s b -> Parser s [a]
-sepBy parser separator =
-  option [] $ (:) <$> parser <*> many (separator *> parser)
-
-string :: ByteString -> Parser s ByteString
-string = lift . Atto.string
-
-takeAll :: (Char -> Bool) -> Parser s ByteString
-takeAll = lift . Atto.takeWhile
-
-takeAll1 :: (Char -> Bool) -> Parser s ByteString
-takeAll1 = lift . Atto.takeWhile1
+evalParser :: Parser s a -> s -> ByteString -> Either String a
+evalParser parser initState = Atto.parseOnly (evalStateT parser initState)

src/PDF/Text.hs

@@ -0,0 +1,319 @@
+{-# LANGUAGE NamedFieldPuns #-}
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE ConstraintKinds #-}
+{-# LANGUAGE FlexibleContexts #-}
+module PDF.Text {-(
+    pageContents
+  )-} where
+
+import Control.Applicative ((<|>))
+import Control.Monad (foldM)
+import Control.Monad.Fail (MonadFail)
+import Control.Monad.Reader (ReaderT, runReaderT, asks)
+import Control.Monad.State (MonadState, evalStateT, get, modify, put)
+import Data.Attoparsec.ByteString.Char8 (choice, sepBy)
+import Data.ByteString (ByteString)
+import Data.ByteString.Char8 (pack, unpack)
+import Data.Char (toLower)
+import Data.Map ((!), (!?), Map)
+import qualified Data.Map as Map (fromList)
+import Data.Text (Text)
+import PDF.Font (Font, FontSet, emptyFont)
+import PDF.Object (
+      DirectObject(..), StringObject(..)
+    , array, blank, name, regular, stringObject, toByteString
+  )
+import PDF.Parser (MonadParser(..), (<?>), Parser, evalParser)
+
+data StateOperator =
+  C_m | W_ | J | J_ | M | D_ | R_i | I_ | G_s -- general graphic state
+  deriving (Bounded, Enum, Show)
+data PathOperator =
+    M_ | L_ | C_ | V_ | Y_ | H_ | R_e -- path construction
+  | S | S_ | F_ | F | Fstar | B | Bstar | B_ | B_star | N_ -- path painting
+  | W | Wstar -- clipping path
+  deriving (Bounded, Enum, Show)
+data ColorOperator =
+    CS | C_s | SC | SCN | S_c | S_cn | G | G_ | RG | R_g | K | K_
+  deriving (Bounded, Enum, Show)
+data TextOperator =
+    Td | TD | Tm | Tstar -- text positioning
+  | TJ | Tj | Quote | DQuote -- text showing
+  | Tc | Tw | Tz | TL | Tf | Tr | Ts -- text state
+  deriving (Bounded, Enum, Show)
+data Argument = Raw ByteString | Typed DirectObject deriving Show
+type Call a = (a, [Argument])
+
+type Operator a = (Bounded a, Enum a, Show a)
+
+code :: Operator a => a -> ByteString
+code = pack . expand . show
+  where
+    expand "" = ""
+    expand (c:'_':s) = toLower c : expand s
+    expand ('s':'t':'a':'r':s) = '*' : expand s
+    expand ('Q':'u':'o':'t':'e':s) = '\'' : expand s
+    expand ('D':'Q':'u':'o':'t':'e':s) = '"' : expand s
+    expand (c:s) = c : expand s
+
+{-
+instance Show StateOperator where
+  show Cm = "cm"
+  show W_ = "w"
+  show J = "J"
+  show J_ = "j"
+  show M = "M"
+  show D = "d"
+  show Ri = "ri"
+  show I = "i"
+  show Gs = "gs"
+
+instance Show PathOperator where
+  show M_ = "m"
+  show L_ = "l"
+  show C_ = "c"
+  show V_ = "v"
+  show Y_ = "y"
+  show H_
+      ("m", (M_, \l -> case l of [Raw _, Raw _] -> True ; _ -> False))
+    , ("l", (L_, \l -> case l of [Raw _, Raw _] -> True ; _ -> False))
+    , ("c", (L_, \l -> case l of [Raw _, Raw _, Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False))
+    , ("v", (L_, \l -> case l of [Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False))
+    , ("y", (L_, \l -> case l of [Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False))
+    , ("h", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("re", (L_, \l -> case l of [Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False))
+    , ("S", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("s", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("f", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("F", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("F*", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("B", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("B*", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("b", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("b*", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("n", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("W", (L_, \l -> case l of [] -> True ; _ -> False))
+    , ("W*", (L_, \l -> case l of [] -> True ; _ -> False))
+
+instance Show ColorOperator where
+
+instance Show TextOperator where
+  show Td = "Td"
+  show TD = "TD"
+  show Tm = "Tm"
+  show Tstar = "T*"
+  show TJ = "TJ"
+  show Tj = "Tj"
+  show Quote = "'"
+  show DQuote = "\""
+  show Tc = "Tc"
+  show Tw = "Tw"
+  show Tz = "Tz"
+  show TL = "TL"
+  show Tf = "Tf"
+  show Tr = "Tr"
+  show Ts = "Ts"
+-}
+
+stateOperator :: OperatorTable StateOperator
+stateOperator = Map.fromList $ (\(op, checker) -> (code op, (op, checker))) <$> [
+      (C_m, \l -> case l of [Raw _, Raw _, Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (W_, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (J, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (J_, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (M, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (D_, \l -> case l of [Raw _, Raw _] -> True ; _ -> False)
+    , (R_i, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (I_, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (G_s, \l -> case l of [Typed (NameObject _)] -> True ; _ -> False)
+  ]
+
+pathOperator :: OperatorTable PathOperator
+pathOperator = Map.fromList $ (\(op, checker) -> (code op, (op, checker))) <$> [
+      (M_, \l -> case l of [Raw _, Raw _] -> True ; _ -> False)
+    , (L_, \l -> case l of [Raw _, Raw _] -> True ; _ -> False)
+    , (C_, \l -> case l of [Raw _, Raw _, Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (V_, \l -> case l of [Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (Y_, \l -> case l of [Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (H_, \l -> case l of [] -> True ; _ -> False)
+    , (R_e, \l -> case l of [Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (S, \l -> case l of [] -> True ; _ -> False)
+    , (S_, \l -> case l of [] -> True ; _ -> False)
+    , (F_, \l -> case l of [] -> True ; _ -> False)
+    , (F, \l -> case l of [] -> True ; _ -> False)
+    , (Fstar, \l -> case l of [] -> True ; _ -> False)
+    , (B, \l -> case l of [] -> True ; _ -> False)
+    , (Bstar, \l -> case l of [] -> True ; _ -> False)
+    , (B_, \l -> case l of [] -> True ; _ -> False)
+    , (B_star, \l -> case l of [] -> True ; _ -> False)
+    , (N_, \l -> case l of [] -> True ; _ -> False)
+    , (W, \l -> case l of [] -> True ; _ -> False)
+    , (Wstar, \l -> case l of [] -> True ; _ -> False)
+  ]
+
+colorOperator :: OperatorTable ColorOperator
+colorOperator = Map.fromList $ (\(op, checker) -> (code op, (op, checker))) <$> [
+      (CS, \l -> case l of [Typed (NameObject _)] -> True ; _ -> False)
+    , (C_s, \l -> case l of [Typed (NameObject _)] -> True ; _ -> False)
+    , (SC, \_ -> True)
+    , (SCN, \_ -> True)
+    , (S_c, \_ -> True)
+    , (S_cn, \_ -> True)
+    , (G, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (G_, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (RG, \l -> case l of [Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (R_g, \l -> case l of [Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (K, \l -> case l of [Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (K_, \l -> case l of [Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False)
+  ]
+
+{-
+stateOperator (Cm, [Raw _, Raw _, Raw _, Raw _, Raw _, Raw _]) = True
+stateOperator (W, [Raw _]) = True
+stateOperator (J, [Raw _]) = True
+stateOperator (J_, [Raw _]) = True
+stateOperator (M, [Raw _]) = True
+stateOperator (D, [Raw _, Raw _]) = True
+stateOperator (Ri, [Raw _]) = True
+stateOperator (I, [Raw _]) = True
+stateOperator (Gs, [Typed (NameObject _)]) = True
+stateOperator _ = False
+-}
+
+textOperator :: OperatorTable TextOperator
+textOperator = Map.fromList $ (\(op, checker) -> (code op, (op, checker))) <$> [
+      (Td, \l -> case l of [Raw _, Raw _] -> True ; _ -> False)
+    , (TD, \l -> case l of [Raw _, Raw _] -> True ; _ -> False)
+    , (Tm, \l -> case l of [Raw _, Raw _, Raw _, Raw _, Raw _, Raw _] -> True ; _ -> False)
+    , (Tstar, \l -> case l of [] -> True ; _ -> False)
+    , (TJ, \l -> case l of [Typed (Array _)] -> True ; _ -> False)
+    , (Tj, \l -> case l of [Typed (StringObject _)] -> True ; _ -> False)
+    , (Quote, \l -> case l of [Typed (StringObject _)] -> True ; _ -> False)
+    , (DQuote, \l -> case l of [Typed (StringObject _)] -> True ; _ -> False)
+    , (Tc, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (Tw, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (Tz, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (TL, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (Tf, \l -> case l of [Typed (NameObject _), Raw _] -> True ; _ -> False)
+    , (Tr, \l -> case l of [Raw _] -> True ; _ -> False)
+    , (Ts, \l -> case l of [Raw _] -> True ; _ -> False)
+  ]
+
+  {-
+textOperator (Td, [Raw _, Raw _]) = True
+textOperator (TD, [Raw _, Raw _]) = True
+textOperator (Tm, [Raw _, Raw _, Raw _, Raw _, Raw _, Raw _]) = True
+textOperator (Tstar, []) = True
+textOperator (TJ, [Typed (Array _)]) = True
+textOperator (Tj, [Typed (StringObject _)]) = True
+textOperator (Quote, [Typed (StringObject _)]) = True
+textOperator (DQuote, [Typed (StringObject _)]) = True
+textOperator (Tc, [Raw _]) = True
+textOperator (Tw, [Raw _]) = True
+textOperator (Tz, [Raw _]) = True
+textOperator (TL, [Raw _]) = True
+textOperator (Tf, [Typed (NameObject _), Raw _]) = True
+textOperator (Tr, [Raw _]) = True
+textOperator (Ts, [Raw _]) = True
+textOperator _ = False
+-}
+
+type ArgumentStackParser m = (MonadState [Argument] m, MonadParser m)
+--type Operator a = (Bounded a, Enum a, Show a)
+type OperatorTable a = Map ByteString (TypeChecker a)
+type TypeChecker a = (a, [Argument] -> Bool)
+
+parseShowable :: (Show a, MonadParser m) => a -> m a
+parseShowable textOp = string (pack $ show textOp) *> return textOp
+
+callChunk :: MonadParser m => OperatorTable a -> m (Either (TypeChecker a) Argument)
+callChunk table =
+    (Right <$> choice [stringArg, nameArg, arrayArg])
+  <|> operatorOrRawArg
+  <?> "call chunk"
+  where
+    operatorOrRawArg = do
+      chunk <- takeAll1 regular <* blank
+      case table !? chunk of
+        Nothing -> return . Right $ Raw chunk
+        Just typeChecker -> return $ Left typeChecker
+
+stackParser :: (ArgumentStackParser m, Operator a) => OperatorTable a -> m (Call a)
+stackParser table = either popCall push =<< (callChunk table)
+  where
+    push arg = modify (arg:) >> stackParser table
+    popCall (operator, predicate) = do
+      arguments <- reverse <$> get
+      let call = (operator, arguments)
+      if predicate arguments then return call else fail (unpack $ code operator)
+
+a :: (Operator a, MonadParser m) => OperatorTable a -> m (Call a)
+a table = evalStateT (stackParser table) []
+
+argument :: MonadParser m => m Argument
+argument = Raw <$> takeAll1 regular <* blank
+
+arrayArg :: MonadParser m => m Argument
+arrayArg = Typed . Array <$> array <* blank
+
+nameArg :: MonadParser m => m Argument
+nameArg = Typed . NameObject <$> name <* blank
+
+stringArg :: MonadParser m => m Argument
+stringArg = Typed . StringObject <$> stringObject <* blank
+
+type ParserWithFont = ReaderT FontSet (Parser Font)
+
+pageContents :: FontSet -> ByteString -> Either String [Text]
+pageContents fontSet input =
+  evalParser (runReaderT (page) fontSet) emptyFont input
+
+several :: MonadParser m => m [a] -> m [a]
+several p = concat <$> (p `sepBy` blank)
+
+page :: ParserWithFont [Text]
+page = several (graphicState <|> text) <* blank <* endOfInput <?> "Text page contents"
+
+graphicState :: ParserWithFont [Text]
+graphicState =
+  string "q" *> blank *> insideQ <* blank <* string "Q" <?> "Graphic state"
+  where
+    insideQ = several (command <|> graphicState <|> text)
+    ignore x = a x *> return []
+    command =
+      ignore stateOperator <|> ignore pathOperator <|> ignore colorOperator
+
+text :: ParserWithFont [Text]
+text =
+  string "BT" *> blank *> commands <* blank <* string "ET" <?> "Text operators"
+  where
+    commands = several (a textOperator >>= runOperator)
+
+runOperator :: Call TextOperator -> ParserWithFont [Text]
+runOperator (Tf, [Typed (NameObject fontName), _]) =
+  asks (! fontName) >>= put >> return []
+
+runOperator (Tstar, []) = return ["\n"]
+
+runOperator (TJ, [Typed (Array arrayObject)]) =
+  replicate 1 <$> foldM appendText "" arrayObject
+  where
+    appendText bs (StringObject outputString) =
+      mappend bs <$> decodeString outputString
+    appendText bs _ = return bs
+
+runOperator (Tj, [Typed (StringObject outputString)]) =
+  replicate 1 <$> decodeString outputString
+
+runOperator (Quote, [Typed (StringObject outputString)]) =
+  (\bs -> ["\n", bs]) <$> decodeString outputString
+
+runOperator (DQuote, [Typed (StringObject outputString)]) =
+  (\bs -> ["\n", bs]) <$> decodeString outputString
+
+runOperator _ = return []
+
+decodeString :: (MonadFail m, MonadState Font m) => StringObject -> m Text
+decodeString input = do
+  font <- get
+  either fail return . font $ toByteString input