pdf-toolbox-document 0.0.3.0 → 0.0.4.0

raw patch · 4 files changed

+40/−8 lines, 4 filesdep ~pdf-toolbox-corePVP ok

version bump matches the API change (PVP)

Dependency ranges changed: pdf-toolbox-core

API changes (from Hackage documentation)

Files

• changelog.md +6/−0
• lib/Pdf/Toolbox/Document/Page.hs +28/−4
• lib/Pdf/Toolbox/Document/Pdf.hs +2/−1
• pdf-toolbox-document.cabal +4/−3

@@ -0,0 +1,6 @@++0.0.4.0++* extracting text: try to insert spaces and newlines+* fix attoparsec module deprication warnings+* fix AMP warnings

@@ -13,10 +13,12 @@ ) where +import Data.Monoid import Data.Functor import qualified Data.Traversable as Traversable import Data.Text (Text)-import qualified Data.Text as Text+import qualified Data.Text.Lazy as TextL+import qualified Data.Text.Lazy.Builder as TextB import qualified Data.Map as Map import Control.Monad @@ -94,8 +96,8 @@  -- | Extract text from the page ----- Right now it doesn't even try to insert additional spaces or newlines,--- and returns text as it is embeded. But someday it will.+-- It tries to add spaces between chars if they don't present+-- as actual characters in content stream. pageExtractText :: (MonadPdf m, MonadIO m) => Page -> PdfE m Text pageExtractText page = do   -- load fonts and create glyph decoder@@ -125,8 +127,30 @@   let loop p = do         next <- readNextOperator is         case next of-          Nothing -> return $ Text.concat $ mapMaybe glyphText $ concat $ prGlyphs p           Just op -> processOp op p >>= loop+          Nothing -> return $ glyphsToText (prGlyphs p)   loop $ mkProcessor {     prGlyphDecoder = glyphDecoder     }++-- | Convert glyphs to text, trying to add spaces and newlines+--+-- It takes list of spans. Each span is a list of glyphs that are outputed in one shot.+-- So we don't need to add space inside span, only between them.+glyphsToText :: [[Glyph]] -> Text+glyphsToText = TextL.toStrict . TextB.toLazyText . snd . foldl step ((Vector 0 0, False), mempty)+  where+  step acc [] = acc+  step ((Vector lx2 ly2, wasSpace), res) sp =+    let Vector x1 y1 = glyphTopLeft (head sp)+        Vector x2 _ = glyphBottomRight (last sp)+        Vector _ y2 = glyphTopLeft (last sp)+        space =+          if abs (ly2 - y1) < 1.8+            then  if wasSpace || abs (lx2 - x1) < 1.8+                    then mempty+                    else TextB.singleton ' '+            else TextB.singleton '\n'+        txt = TextB.fromLazyText $ TextL.fromChunks $ mapMaybe glyphText sp+        endWithSpace = glyphText (last sp) == Just " "+    in ((Vector x2 y2, endWithSpace), mconcat [res, space, txt])

@@ -27,6 +27,7 @@ import qualified Data.ByteString as BS import Data.Map (Map) import qualified Data.Map as Map+import Control.Applicative import Control.Monad import Control.Monad.Trans.Class import Control.Monad.Trans.State@@ -51,7 +52,7 @@  -- | Basic implementation of pdf monad newtype Pdf' m a = Pdf' (StateT PdfState m a)-  deriving (Monad, Functor, MonadIO, MonadTrans)+  deriving (Monad, Functor, Applicative, MonadIO, MonadTrans)  -- | Convenient type alias type Pdf m = PdfE (Pdf' m)

@@ -1,15 +1,16 @@ name:                pdf-toolbox-document-version:             0.0.3.0+version:             0.0.4.0 synopsis:            A collection of tools for processing PDF files. license:             BSD3 license-file:        LICENSE author:              Yuras Shumovich maintainer:          Yuras Shumovich <shumovichy@gmail.com>-copyright:           Copyright (c) Yuras Shumovich 2012-2013+copyright:           Copyright (c) Yuras Shumovich 2012-2014 category:            PDF build-type:          Simple cabal-version:       >=1.8 homepage:            https://github.com/Yuras/pdf-toolbox+extra-source-files:  changelog.md description:   Mid level tools for processing PDF files.   .@@ -42,5 +43,5 @@                        io-streams,                        cipher-rc4,                        cryptohash,-                       pdf-toolbox-core ==0.0.2.*,+                       pdf-toolbox-core >=0.0.2 && <0.0.4,                        pdf-toolbox-content ==0.0.3.*