1864bb0 · typeable/pandoc

git.iko.soy/typeable/pandoc

Data files changes.

* Added `embed_data_files` flag. (not yet used) * Shared no longer exports `findDataFile`. * `readDataFile` now returns a strict bytestring. * Shared now exports `readDataFileUTF8` which returns a string like the old `readDataFile`. * Rewrote modules to use new data file functions and to avoid using functions from Paths_pandoc directly.

commit1864bb0994cc70b99ea8207b3a22438c343348c0 authorJohn MacFarlane <fiddlosopher@gmail.com> dateSat 29 Dec 2012, 17:44 (14 years ago) parent32c5a8e

9 files changed+64−67

compare with master

unifiedsplit

benchmark/benchmark-pandoc.hs+3−3 pandoc.cabal+7 pandoc.hs+13−9 src/Text/Pandoc/SelfContained.hs+3−13 src/Text/Pandoc/Shared.hs+27−14 src/Text/Pandoc/Templates.hs+2−2 src/Text/Pandoc/Writers/Docx.hs+2−11 src/Text/Pandoc/Writers/EPUB.hs+4−3 src/Text/Pandoc/Writers/ODT.hs+3−12

benchmark/benchmark-pandoc.hs+3−3

@@ -1,5 +1,5 @@

11import Text.Pandoc

2−import Text.Pandoc.Shared (readDataFile, normalize)

2+import Text.Pandoc.Shared (readDataFileUTF8, normalize)

33import Criterion.Main

44import Criterion.Config

55import Text.JSON.Generic

@@ -35,8 +35,8 @@ main :: IO ()

3535main = do

3636 args <- getArgs

3737 (conf,_) <- parseArgs defaultConfig{ cfgSamples = Last $ Just 20 } defaultOptions args

38− inp <- readDataFile (Just ".") "README"

39− inp2 <- readDataFile (Just ".") "tests/testsuite.txt"

38+ inp <- readDataFileUTF8 (Just ".") "README"

39+ inp2 <- readDataFileUTF8 (Just ".") "tests/testsuite.txt"

4040 let opts = def{ readerSmart = True }

4141 let doc = readMarkdown opts $ inp ++ unlines (drop 3 $ lines inp2)

4242 let readerBs = map (readerBench doc) readers

pandoc.cabal+7

@@ -187,6 +187,9 @@ Source-repository head

187187Flag blaze_html_0_5

188188 Description: Use blaze-html 0.5 and blaze-markup 0.5

189189 Default: True

190+Flag embed_data_files

191+ Description: Embed data files in binary for relocatable executable.

192+ Default: False

190193

191194Library

192195 Build-Depends: base >= 4.2 && <5,

@@ -224,6 +227,10 @@ Library

224227 else

225228 build-depends:

226229 blaze-html >= 0.4.3.0 && < 0.5

230+ if flag(embed_data_files)

231+ build-depends: file-embed >= 0.0.4 && < 0.1,

232+ template-haskell >= 2.4 && < 2.9

233+ cpp-options: -DEMBED_DATA_FILES

227234 if impl(ghc >= 7.0.1)

228235 Ghc-Options: -O2 -rtsopts -Wall -fno-warn-unused-do-bind -dno-debug-output

229236 else

pandoc.hs+13−9

@@ -33,7 +33,7 @@ module Main where

3333import Text.Pandoc

3434import Text.Pandoc.PDF (tex2pdf)

3535import Text.Pandoc.Readers.LaTeX (handleIncludes)

36−import Text.Pandoc.Shared ( tabFilter, readDataFile, safeRead,

36+import Text.Pandoc.Shared ( tabFilter, readDataFileUTF8, safeRead,

3737 headerShift, normalize, err, warn )

3838import Text.Pandoc.XML ( toEntities, fromEntities )

3939import Text.Pandoc.SelfContained ( makeSelfContained )

@@ -889,24 +889,27 @@ main = do

889889 E.catch (UTF8.readFile tp')

890890 (\e -> if isDoesNotExistError e

891891 then E.catch

892− (readDataFile datadir $

893− "templates" </> tp')

892+ (readDataFileUTF8 datadir

893+ ("templates" </> tp'))

894894 (\e' -> let _ = (e' :: E.SomeException)

895895 in throwIO e')

896896 else throwIO e)

897897

898898 variables' <- case mathMethod of

899899 LaTeXMathML Nothing -> do

900− s <- readDataFile datadir $ "data" </> "LaTeXMathML.js"

900+ s <- readDataFileUTF8 datadir

901+ ("data" </> "LaTeXMathML.js")

901902 return $ ("mathml-script", s) : variables

902903 MathML Nothing -> do

903− s <- readDataFile datadir $ "data"</>"MathMLinHTML.js"

904+ s <- readDataFileUTF8 datadir

905+ ("data"</>"MathMLinHTML.js")

904906 return $ ("mathml-script", s) : variables

905907 _ -> return variables

906908

907909 variables'' <- if "dzslides" `isPrefixOf` writerName'

908910 then do

909− dztempl <- readDataFile datadir $ "dzslides" </> "template.html"

911+ dztempl <- readDataFileUTF8 datadir

912+ ("dzslides" </> "template.html")

910913 let dzcore = unlines $ dropWhile (not . isPrefixOf "<!-- {{{{ dzslides core")

911914 $ lines dztempl

912915 return $ ("dzslides-core", dzcore) : variables'

@@ -927,15 +930,16 @@ main = do

927930 then do

928931 csl <- CSL.parseCSL =<<

929932 case mbCsl of

930− Nothing -> readDataFile datadir "default.csl"

933+ Nothing -> readDataFileUTF8 datadir

934+ "default.csl"

931935 Just cslfile -> do

932936 exists <- doesFileExist cslfile

933937 if exists

934938 then UTF8.readFile cslfile

935939 else do

936940 csldir <- getAppUserDataDirectory "csl"

937− readDataFile (Just csldir)

938− (replaceExtension cslfile "csl")

941+ readDataFileUTF8 (Just csldir)

942+ (replaceExtension cslfile "csl")

939943 abbrevs <- maybe (return []) CSL.readJsonAbbrevFile cslabbrevs

940944 return $ Just csl { CSL.styleAbbrevs = abbrevs }

941945 else return Nothing

src/Text/Pandoc/SelfContained.hs+3−13

@@ -41,7 +41,7 @@ import System.FilePath (takeExtension, dropExtension, takeDirectory, (</>))

4141import Data.Char (toLower, isAscii, isAlphaNum)

4242import Codec.Compression.GZip as Gzip

4343import qualified Data.ByteString.Lazy as L

44−import Text.Pandoc.Shared (findDataFile, renderTags')

44+import Text.Pandoc.Shared (readDataFile, renderTags')

4545import Text.Pandoc.MIME (getMimeType)

4646import System.Directory (doesFileExist)

4747import Text.Pandoc.UTF8 (toString, fromString)

@@ -55,18 +55,8 @@ getItem userdata f =

5555 ".gz" -> getMimeType $ dropExtension f

5656 x -> getMimeType x

5757 exists <- doesFileExist f

58− if exists

59− then do

60− cont <- B.readFile f

61− return (cont, mime)

62− else do

63− res <- findDataFile userdata f

64− exists' <- doesFileExist res

65− if exists'

66− then do

67− cont <- B.readFile res

68− return (cont, mime)

69− else error $ "Could not find `" ++ f ++ "'"

58+ cont <- if exists then B.readFile f else readDataFile userdata f

59+ return (cont, mime)

7060

7161-- TODO - have this return mime type too - then it can work for google

7262-- chart API, e.g.

src/Text/Pandoc/Shared.hs+27−14

@@ -1,4 +1,4 @@

1−{-# LANGUAGE DeriveDataTypeable #-}

1+{-# LANGUAGE DeriveDataTypeable, CPP #-}

22{-

@@ -64,8 +64,8 @@ module Text.Pandoc.Shared (

6464 renderTags',

6565 -- * File handling

6666 inDirectory,

67− findDataFile,

6867 readDataFile,

68+ readDataFileUTF8,

6969 -- * Error handling

7070 err,

7171 warn,

@@ -89,13 +89,18 @@ import System.FilePath ( (</>) )

8989import Data.Generics (Typeable, Data)

9090import qualified Control.Monad.State as S

9191import Control.Monad (msum)

92−import Paths_pandoc (getDataFileName)

9392import Text.Pandoc.Pretty (charWidth)

9493import System.Locale (defaultTimeLocale)

9594import Data.Time

9695import System.IO (stderr)

9796import Text.HTML.TagSoup (renderTagsOptions, RenderOptions(..), Tag(..),

9897 renderOptions)

98+import qualified Data.ByteString as B

99+#ifdef EMBED_DATA_FILES

100+import Data.FileEmbed

101+#else

102+import Paths_pandoc (getDataFileName)

103+#endif

99104

100105--

101106-- List processing

@@ -499,20 +504,28 @@ inDirectory path action = do

499504 setCurrentDirectory oldDir

500505 return result

501506

502−-- | Get file path for data file, either from specified user data directory,

503−-- or, if not found there, from Cabal data directory.

504−findDataFile :: Maybe FilePath -> FilePath -> IO FilePath

505−findDataFile Nothing f = getDataFileName f

506−findDataFile (Just u) f = do

507− ex <- doesFileExist (u </> f)

508− if ex

509− then return (u </> f)

510− else getDataFileName f

507+readDefaultDataFile :: FilePath -> IO B.ByteString

508+readDefaultDataFile fname =

509+#ifdef EMBED_DATA_FILES

510+ TODO

511+#else

512+ getDataFileName fname >>= B.readFile

513+#endif

511514

512515-- | Read file from specified user data directory or, if not found there, from

513516-- Cabal data directory.

514−readDataFile :: Maybe FilePath -> FilePath -> IO String

515−readDataFile userDir fname = findDataFile userDir fname >>= UTF8.readFile

517+readDataFile :: Maybe FilePath -> FilePath -> IO B.ByteString

518+readDataFile Nothing fname = readDefaultDataFile fname

519+readDataFile (Just userDir) fname = do

520+ exists <- doesFileExist (userDir </> fname)

521+ if exists

522+ then B.readFile (userDir </> fname)

523+ else readDefaultDataFile fname

524+

525+-- | Same as 'readDataFile' but returns a String instead of a ByteString.

526+readDataFileUTF8 :: Maybe FilePath -> FilePath -> IO String

527+readDataFileUTF8 userDir fname =

528+ UTF8.toString `fmap` readDataFile userDir fname

516529

517530--

518531-- Error reporting

src/Text/Pandoc/Templates.hs+2−2

@@ -80,7 +80,7 @@ import Text.Blaze (preEscapedString, Html)

8080#endif

8181import Text.Pandoc.UTF8 (fromStringLazy)

8282import Data.ByteString.Lazy (ByteString)

83−import Text.Pandoc.Shared (readDataFile)

83+import Text.Pandoc.Shared (readDataFileUTF8)

8484import qualified Control.Exception.Extensible as E (try, IOException)

8585

8686-- | Get default template for the specified writer.

@@ -98,7 +98,7 @@ getDefaultTemplate user writer = do

9898 "multimarkdown" -> getDefaultTemplate user "markdown"

9999 "markdown_github" -> getDefaultTemplate user "markdown"

100100 _ -> let fname = "templates" </> "default" <.> format

101− in E.try $ readDataFile user fname

101+ in E.try $ readDataFileUTF8 user fname

102102

103103data TemplateState = TemplateState Int [(String,String)]

104104

src/Text/Pandoc/Writers/Docx.hs+2−11

@@ -29,14 +29,12 @@ Conversion of 'Pandoc' documents to docx.

2929-}

3030module Text.Pandoc.Writers.Docx ( writeDocx ) where

3131import Data.List ( intercalate )

32−import System.FilePath ( (</>) )

3332import qualified Data.ByteString.Lazy as B

3433import qualified Data.Map as M

3534import qualified Text.Pandoc.UTF8 as UTF8

3635import System.IO ( stderr )

3736import Codec.Archive.Zip

3837import Data.Time.Clock.POSIX

39−import Paths_pandoc ( getDataFileName )

4038import Text.Pandoc.Definition

4139import Text.Pandoc.Generic

4240import System.Directory

@@ -104,15 +102,8 @@ writeDocx opts doc@(Pandoc (Meta tit auths date) _) = do

104102 refArchive <- liftM toArchive $

105103 case writerReferenceDocx opts of

106104 Just f -> B.readFile f

107− Nothing -> do

108− let defaultDocx = getDataFileName "reference.docx" >>= B.readFile

109− case datadir of

110− Nothing -> defaultDocx

111− Just d -> do

112− exists <- doesFileExist (d </> "reference.docx")

113− if exists

114− then B.readFile (d </> "reference.docx")

115− else defaultDocx

105+ Nothing -> (B.fromChunks . (:[])) `fmap`

106+ readDataFile datadir "reference.docx"

116107

117108 (newContents, st) <- runStateT (writeOpenXML opts{writerWrapText = False} doc)

118109 defaultWriterState

src/Text/Pandoc/Writers/EPUB.hs+4−3

@@ -30,13 +30,13 @@ Conversion of 'Pandoc' documents to EPUB.

3030module Text.Pandoc.Writers.EPUB ( writeEPUB2, writeEPUB3 ) where

3131import Data.IORef

3232import Data.Maybe ( fromMaybe, isNothing )

33−import Data.List ( isPrefixOf, isInfixOf, intercalate )

33+import Data.List ( isInfixOf, intercalate )

3434import System.Environment ( getEnv )

3535import Text.Printf (printf)

3636import System.FilePath ( (</>), takeBaseName, takeExtension, takeFileName )

3737import qualified Data.ByteString.Lazy as B

3838import qualified Data.ByteString.Lazy.Char8 as B8

39−import Text.Pandoc.UTF8 ( fromStringLazy )

39+import Text.Pandoc.UTF8 ( fromStringLazy, toString )

4040import Codec.Archive.Zip

4141import Data.Time.Clock.POSIX

4242import Data.Time

@@ -321,7 +321,8 @@ writeEPUB version opts doc@(Pandoc meta _) = do

321321 -- stylesheet

322322 stylesheet <- case writerEpubStylesheet opts of

323323 Just s -> return s

324− Nothing -> readDataFile (writerUserDataDir opts) "epub.css"

324+ Nothing -> toString `fmap`

325+ readDataFile (writerUserDataDir opts) "epub.css"

325326 let stylesheetEntry = mkEntry "stylesheet.css" $ fromStringLazy stylesheet

326327

327328 -- construct archive

src/Text/Pandoc/Writers/ODT.hs+3−12

@@ -35,15 +35,13 @@ import qualified Data.ByteString.Lazy as B

3535import Text.Pandoc.UTF8 ( fromStringLazy )

3636import Codec.Archive.Zip

3737import Data.Time.Clock.POSIX

38−import Paths_pandoc ( getDataFileName )

3938import Text.Pandoc.Options ( WriterOptions(..) )

40−import Text.Pandoc.Shared ( stringify )

39+import Text.Pandoc.Shared ( stringify, readDataFile )

4140import Text.Pandoc.ImageSize ( readImageSize, sizeInPoints )

4241import Text.Pandoc.MIME ( getMimeType )

4342import Text.Pandoc.Definition

4443import Text.Pandoc.Generic

4544import Text.Pandoc.Writers.OpenDocument ( writeOpenDocument )

46−import System.Directory

4745import Control.Monad (liftM)

4846import Network.URI ( unEscapeString )

4947import Text.Pandoc.XML

@@ -59,15 +57,8 @@ writeODT opts doc@(Pandoc (Meta title _ _) _) = do

5957 refArchive <- liftM toArchive $

6058 case writerReferenceODT opts of

6159 Just f -> B.readFile f

62− Nothing -> do

63− let defaultODT = getDataFileName "reference.odt" >>= B.readFile

64− case datadir of

65− Nothing -> defaultODT

66− Just d -> do

67− exists <- doesFileExist (d </> "reference.odt")

68− if exists

69− then B.readFile (d </> "reference.odt")

70− else defaultODT

60+ Nothing -> (B.fromChunks . (:[])) `fmap`

61+ readDataFile datadir "reference.odt"

7162 -- handle pictures

7263 picEntriesRef <- newIORef ([] :: [Entry])

7364 let sourceDir = writerSourceDirectory opts