pdf-toolbox-content 0.0.3.3 → 0.0.4.0

raw patch · 5 files changed

+51/−29 lines, 5 filesdep ~pdf-toolbox-corePVP: major bump suggested

API removals or changes: PVP suggests a major version bump

Dependency ranges changed: pdf-toolbox-core

API changes (from Hackage documentation)

Files

• changelog.md +4/−0
• lib/Pdf/Toolbox/Content/FontInfo.hs +1/−1
• lib/Pdf/Toolbox/Content/Parser.hs +22/−4
• lib/Pdf/Toolbox/Content/Processor.hs +22/−22
• pdf-toolbox-content.cabal +2/−2

@@ -1,3 +1,7 @@+0.0.4.0++* switch to errors-2.0+* workaround for parse error on inline image in content stream  0.0.3.3 

@@ -121,7 +121,7 @@     n <- intValue x     ws <- forM arr $ \w -> fromObject w >>= realValue     go res {cidFontWidthsChars = Map.fromList (zip [n ..] ws) `mappend` cidFontWidthsChars res} xs-  go _ _ = left $ UnexpectedError "Can't parse CIDFont width"+  go _ _ = throwE $ UnexpectedError "Can't parse CIDFont width"  -- | Get glyph width by glyph code cidFontGetWidth :: CIDFontWidths -> Int -> Maybe Double

@@ -9,6 +9,7 @@ where  import Data.Int+import qualified Data.ByteString.Char8 as ByteString8 import Data.Attoparsec.ByteString.Char8 (Parser) import qualified Data.Attoparsec.ByteString.Char8 as Parser import Data.IORef@@ -48,11 +49,11 @@         `catch` (\e -> return $ Left $ UnexpectedError $ show (e :: Streams.ParseException))       case e of         Right expr -> return expr-        Left er -> left er+        Left er -> throwE er     case expr of       Nothing -> case args of                    [] -> return Nothing-                   _ -> left $ UnexpectedError $ "Args without op: " ++ show args+                   _ -> throwE $ UnexpectedError $ "Args without op: " ++ show args       Just (Obj o) -> go (o : args)       Just (Op o) -> return $ Just (o, reverse args) @@ -74,7 +75,7 @@         case ss of           [] -> return Nothing           (h:t) -> do-            reader <- runEitherT $ mkReader h t+            reader <- runExceptT $ mkReader h t             case reader of               Left e -> liftIO $ ioError $ userError $ show e               Right r -> do@@ -87,7 +88,24 @@   = (skipSpace >> Parser.endOfInput >> return Nothing)   <|> do     skipSpace-    fmap Just $ fmap Obj parseObject <|> fmap (Op . toOp) (Parser.takeWhile1 isRegularChar)+    fmap Just $ fmap Obj parseObject+      <|> fmap (Op . toOp) (Parser.takeWhile1 isRegularChar)+      -- See Note Inline image+      <|> fmap (Op . UnknownOp . ByteString8.pack . return) Parser.anyChar++{- Note Inline image++There is at least one case that doesn't fit the way we represent content+stream operators: inline images, see Pdf1.7:8.9.7++Inline image looks like "BI" operator, a number of key/value pairs, "ID"+operator, image data and "EI" operator. I have no idea how to handle that case.+There is no data length, so the "EI" string is the only indicator for end+of image data. What if the data contains "EI"?++For now lets skip all unknown bytes as unknown operator. That seems to work,+but it is not reliable.+-}  -- Treat comments as spaces skipSpace :: Parser ()

@@ -100,13 +100,13 @@ processOp :: Monad m => Operator -> Processor -> PdfE m Processor  processOp (Op_q, []) p = return p {prStateStack = prState p : prStateStack p}-processOp (Op_q, args) _ = left $ UnexpectedError $ "Op_q: wrong number of arguments: " ++ show args+processOp (Op_q, args) _ = throwE $ UnexpectedError $ "Op_q: wrong number of arguments: " ++ show args  processOp (Op_Q, []) p =   case prStateStack p of-    [] -> left $ UnexpectedError "Op_Q: state is empty"+    [] -> throwE $ UnexpectedError "Op_Q: state is empty"     (x:xs) -> return p {prState = x, prStateStack = xs}-processOp (Op_Q, args) _ = left $ UnexpectedError $ "Op_Q: wrong number of arguments: " ++ show args+processOp (Op_Q, args) _ = throwE $ UnexpectedError $ "Op_Q: wrong number of arguments: " ++ show args  processOp (Op_BT, []) p = do   ensureInTextObject False p@@ -116,7 +116,7 @@     gsTextMatrix = identity,     gsTextLineMatrix = identity     }}-processOp (Op_BT, args) _ = left $ UnexpectedError $ "Op_BT: wrong number of arguments: " ++ show args+processOp (Op_BT, args) _ = throwE $ UnexpectedError $ "Op_BT: wrong number of arguments: " ++ show args  processOp (Op_ET, []) p = do   ensureInTextObject True p@@ -124,7 +124,7 @@   return p {prState = gstate {     gsInText = False     }}-processOp (Op_ET, args) _ = left $ UnexpectedError $ "Op_ET: wrong number of arguments: " ++ show args+processOp (Op_ET, args) _ = throwE $ UnexpectedError $ "Op_ET: wrong number of arguments: " ++ show args  processOp (Op_Td, [txo, tyo]) p = do   ensureInTextObject True p@@ -136,13 +136,13 @@     gsTextMatrix = tm,     gsTextLineMatrix = tm     }}-processOp (Op_Td, args) _ = left $ UnexpectedError $ "Op_Td: wrong number of arguments: " ++ show args+processOp (Op_Td, args) _ = throwE $ UnexpectedError $ "Op_Td: wrong number of arguments: " ++ show args  processOp (Op_TD, [txo, tyo]) p = do   l <- fromObject tyo >>= realValue   p' <- processOp (Op_TL, [ONumber $ NumReal $ negate l]) p   processOp (Op_Td, [txo, tyo]) p'-processOp (Op_TD, args) _ = left $ UnexpectedError $ "Op_TD: wrong number of arguments: " ++ show args+processOp (Op_TD, args) _ = throwE $ UnexpectedError $ "Op_TD: wrong number of arguments: " ++ show args  processOp (Op_Tm, [a', b', c', d', e', f']) p = do   ensureInTextObject True p@@ -158,14 +158,14 @@     gsTextMatrix = tm,     gsTextLineMatrix = tm     }}-processOp (Op_Tm, args) _ = left $ UnexpectedError $ "Op_Tm: wrong number of arguments: " ++ show args+processOp (Op_Tm, args) _ = throwE $ UnexpectedError $ "Op_Tm: wrong number of arguments: " ++ show args  processOp (Op_T_star, []) p = do   ensureInTextObject True p   let gstate = prState p       l = gsTextLeading gstate   processOp (Op_TD, map (ONumber . NumReal) [0, negate l]) p-processOp (Op_T_star, args) _ = left $ UnexpectedError $ "Op_T_star: wrong number of arguments: " ++ show args+processOp (Op_T_star, args) _ = throwE $ UnexpectedError $ "Op_T_star: wrong number of arguments: " ++ show args  processOp (Op_TL, [lo]) p = do   l <- fromObject lo >>= realValue@@ -173,7 +173,7 @@   return p {prState = gstate {     gsTextLeading = l     }}-processOp (Op_TL, args) _ = left $ UnexpectedError $ "Op_TL: wrong number of arguments: " ++ show args+processOp (Op_TL, args) _ = throwE $ UnexpectedError $ "Op_TL: wrong number of arguments: " ++ show args  processOp (Op_cm, [a', b', c', d', e', f']) p = do   a <- fromObject a' >>= realValue@@ -187,7 +187,7 @@   return p {prState = gstate {     gsCurrentTransformMatrix = ctm     }}-processOp (Op_cm, args) _ = left $ UnexpectedError $ "Op_cm: wrong number of arguments: " ++ show args+processOp (Op_cm, args) _ = throwE $ UnexpectedError $ "Op_cm: wrong number of arguments: " ++ show args  processOp (Op_Tf, [fontO, szO]) p = do   font <- fromObject fontO@@ -197,17 +197,17 @@     gsFont = Just font,     gsFontSize = Just sz     }}-processOp (Op_Tf, args) _ = left $ UnexpectedError $ "Op_Tf: wrong number of agruments: " ++ show args+processOp (Op_Tf, args) _ = throwE $ UnexpectedError $ "Op_Tf: wrong number of agruments: " ++ show args  processOp (Op_Tj, [OStr str]) p = do   let gstate = prState p   fontName <-     case gsFont gstate of-      Nothing -> left $ UnexpectedError "Op_Tj: font not set"+      Nothing -> throwE $ UnexpectedError "Op_Tj: font not set"       Just fn -> return fn   fontSize <-     case gsFontSize gstate of-      Nothing -> left $ UnexpectedError "Op_Tj: font size not set"+      Nothing -> throwE $ UnexpectedError "Op_Tj: font size not set"       Just fs -> return fs   let (tm, glyphs) = positionGlyghs fontSize (gsCurrentTransformMatrix gstate)                        (gsTextMatrix gstate) (gsTextCharSpacing gstate) (gsTextWordSpacing gstate) $@@ -218,17 +218,17 @@       gsTextMatrix = tm       }     }-processOp (Op_Tj, args) _ = left $ UnexpectedError $ "Op_Tj: wrong number of agruments:" ++ show args+processOp (Op_Tj, args) _ = throwE $ UnexpectedError $ "Op_Tj: wrong number of agruments:" ++ show args  processOp (Op_TJ, [OArray (Array array)]) p = do   let gstate = prState p   fontName <-     case gsFont gstate of-      Nothing -> left $ UnexpectedError "Op_Tj: font not set"+      Nothing -> throwE $ UnexpectedError "Op_Tj: font not set"       Just fn -> return fn   fontSize <-     case gsFontSize gstate of-      Nothing -> left $ UnexpectedError "Op_Tj: font size not set"+      Nothing -> throwE $ UnexpectedError "Op_Tj: font size not set"       Just fs -> return fs   let (textMatrix, glyphs) = loop (gsTextMatrix gstate) [] array         where@@ -246,7 +246,7 @@       gsTextMatrix = textMatrix       }     }-processOp (Op_TJ, args) _ = left $ UnexpectedError $ "Op_TJ: wrong number of agruments:" ++ show args+processOp (Op_TJ, args) _ = throwE $ UnexpectedError $ "Op_TJ: wrong number of agruments:" ++ show args  processOp (Op_Tc, [o]) p = do   spacing <- fromObject o >>= realValue@@ -256,7 +256,7 @@       gsTextCharSpacing = spacing       }     }-processOp (Op_Tc, args) _ = left $ UnexpectedError $ "Op_Tc: wrong number of agruments:" ++ show args+processOp (Op_Tc, args) _ = throwE $ UnexpectedError $ "Op_Tc: wrong number of agruments:" ++ show args  processOp (Op_Tw, [o]) p = do   spacing <- fromObject o >>= realValue@@ -266,18 +266,18 @@       gsTextWordSpacing = spacing       }     }-processOp (Op_Tw, args) _ = left $ UnexpectedError $ "Op_Tw: wrong number of agruments:" ++ show args+processOp (Op_Tw, args) _ = throwE $ UnexpectedError $ "Op_Tw: wrong number of agruments:" ++ show args  processOp (Op_apostrophe, [o]) p = do   p' <- processOp (Op_T_star, []) p   processOp (Op_Tj, [o]) p'-processOp (Op_apostrophe, args) _ = left $ UnexpectedError $ "Op_apostrophe: wrong number of agruments:" ++ show args+processOp (Op_apostrophe, args) _ = throwE $ UnexpectedError $ "Op_apostrophe: wrong number of agruments:" ++ show args  processOp _ p = return p  ensureInTextObject :: Monad m => Bool -> Processor -> PdfE m () ensureInTextObject inText p =-  unless (inText == gsInText (prState p)) $ left $+  unless (inText == gsInText (prState p)) $ throwE $     UnexpectedError $ "ensureInTextObject: expected: " ++ show inText ++ ", found: " ++ show (gsInText $ prState p)  positionGlyghs :: Double -> Transform Double -> Transform Double -> Double -> Double -> [(Glyph, Double)] -> (Transform Double, [Glyph])

@@ -1,5 +1,5 @@ name:                pdf-toolbox-content-version:             0.0.3.3+version:             0.0.4.0 synopsis:            A collection of tools for processing PDF files license:             BSD3 license-file:        LICENSE@@ -41,4 +41,4 @@                        base16-bytestring,                        text,                        io-streams,-                       pdf-toolbox-core >=0.0.2.0 && <0.0.4+                       pdf-toolbox-core ==0.0.4.*