diff --git a/COPYRIGHT b/COPYRIGHT --- a/COPYRIGHT +++ b/COPYRIGHT @@ -1,5 +1,5 @@ Pandoc -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This code is released under the [GPL], version 2 or later: @@ -32,25 +32,25 @@ licenses. ---------------------------------------------------------------------- -Text/Pandoc/Writers/Texinfo.hs -Copyright (C) 2008 John MacFarlane and Peter Wang +src/Text/Pandoc/Writers/Texinfo.hs +Copyright (C) 2008-2010 John MacFarlane and Peter Wang Released under the GPL. ---------------------------------------------------------------------- -Text/Pandoc/Writers/OpenDocument.hs -Copyright (C) 2008 Andrea Rossato +src/Text/Pandoc/Writers/OpenDocument.hs +Copyright (C) 2008-2010 Andrea Rossato and John MacFarlane Released under the GPL. ---------------------------------------------------------------------- -Text/Pandoc/Biblio.hs +src/Text/Pandoc/Biblio.hs Copyright (C) 2008 Andrea Rossato Released under the GPL. ---------------------------------------------------------------------- -LaTeXMathML.js +data/LaTeXMathML.js Adapted by Jeff Knisely and Douglas Woodall from ASCIIMathML.js v. 1.4.7 Copyright (C) 2005 Peter Jipsen @@ -58,18 +58,20 @@ Released under the GPL. ---------------------------------------------------------------------- -S5 slides.js and css files +data/MathMLInHTML.js +Copyright (C) 2004 Peter Jipsen http://www.chapman.edu/~jipsen + +Released under the GNU General Public License version 2 or later. +See the GNU General Public License +(at http://www.gnu.org/copyleft/gpl.html) for more details. + +---------------------------------------------------------------------- +s5/default +S5 slides.js and css files by Eric A. Meyer BuildFlags -> PackageDescription -> LocalBuildInfo -> IO () -makeManPages _ flags _ _ = mapM_ (makeManPage (fromFlag $ buildVerbosity flags)) manpages +makeManPages _ flags _ buildInfo = + mapM_ (makeManPage pandocPath (fromFlag $ buildVerbosity flags)) manpages + where pandocPath = (buildDir buildInfo) "pandoc" "pandoc" manpages :: [FilePath] manpages = ["pandoc.1", "markdown2pdf.1"] @@ -57,21 +59,20 @@ manDir = "man" "man1" -- | Build a man page from markdown source in man/man1. -makeManPage :: Verbosity -> FilePath -> IO () -makeManPage verbosity manpage = do - let pandoc = "dist" "build" "pandoc" "pandoc" +makeManPage :: FilePath -> Verbosity -> FilePath -> IO () +makeManPage pandoc verbosity manpage = do let page = manDir manpage - let source = manDir manpage <.> "md" + let source = page <.> "md" modifiedDeps <- modifiedDependencies page [source] unless (null modifiedDeps) $ do ec <- runProcess pandoc ["-s", "-S", "-r", "markdown", "-w", "man", "--template=templates/man.template", "-o", page, source] Nothing Nothing Nothing Nothing (Just stderr) >>= waitForProcess case ec of - ExitSuccess -> unless (verbosity == silent) $ - putStrLn $ "Created " ++ manDir manpage - _ -> do putStrLn $ "Error creating " ++ manDir manpage - exitWith ec + ExitSuccess -> unless (verbosity == silent) $ + putStrLn $ "Created " ++ page + ExitFailure n -> putStrLn ("Error creating " ++ page ++ + ". Exit code = " ++ show n) >> exitWith ec installScripts :: PackageDescription -> LocalBuildInfo -> Verbosity -> CopyDest -> IO () diff --git a/changelog b/changelog --- a/changelog +++ b/changelog @@ -1,3 +1,26 @@ +pandoc (1.5.1) + + [ John MacFarlane ] + + * Fixed treatment of unicode characters in URIs. + + Shared now exports escapeURI and unescapeURI. These handle + UTF8 encoding and decoding as well as URI escaping/unescaping. + + Shared: uri and emailAddress now return a pair of the original + parsed text and the escaped URI (in the latter case, with + the mailto: prefix). + + HTML reader: unsanitaryURI has been modified to allow unicode + high characters in a URI. + + Readers: All link and image URIs are now escaped using + escapeURI. + + Markdown and RST writers: unescapeURI is used so that URIs + in these formats are human-readable. + + * Setup.hs: Don't assume that the build directory is "dist". + Instead, get it from localBuildInfo. + + * OpenDocument writer: Use a Map for stTextStyleAttr. + This avoids duplicates (and invalid xml). Resolves Issue #222. + pandoc (1.5.0.1) [ John MacFarlane ] diff --git a/man/man1/pandoc.1.md b/man/man1/pandoc.1.md --- a/man/man1/pandoc.1.md +++ b/man/man1/pandoc.1.md @@ -351,8 +351,6 @@ # SEE ALSO -`hsmarkdown`(1), -`html2markdown`(1), `markdown2pdf` (1). The *README* file distributed with Pandoc contains full documentation. diff --git a/pandoc.cabal b/pandoc.cabal --- a/pandoc.cabal +++ b/pandoc.cabal @@ -1,5 +1,5 @@ Name: pandoc -Version: 1.5.0.1 +Version: 1.5.1 Cabal-Version: >= 1.2 Build-Type: Custom License: GPL diff --git a/src/Text/Pandoc.hs b/src/Text/Pandoc.hs --- a/src/Text/Pandoc.hs +++ b/src/Text/Pandoc.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/CharacterReferences.hs b/src/Text/Pandoc/CharacterReferences.hs --- a/src/Text/Pandoc/CharacterReferences.hs +++ b/src/Text/Pandoc/CharacterReferences.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.CharacterReferences - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Definition.hs b/src/Text/Pandoc/Definition.hs --- a/src/Text/Pandoc/Definition.hs +++ b/src/Text/Pandoc/Definition.hs @@ -1,6 +1,6 @@ {-# OPTIONS_GHC -fglasgow-exts #-} -- for deriving Typeable {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -19,7 +19,7 @@ {- | Module : Text.Pandoc.Definition - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/ODT.hs b/src/Text/Pandoc/ODT.hs --- a/src/Text/Pandoc/ODT.hs +++ b/src/Text/Pandoc/ODT.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2008 John MacFarlane +Copyright (C) 2008-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.ODT - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2008-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Readers/HTML.hs b/src/Text/Pandoc/Readers/HTML.hs --- a/src/Text/Pandoc/Readers/HTML.hs +++ b/src/Text/Pandoc/Readers/HTML.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Readers.HTML - Copyright : Copyright (C) 2006-8 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane @@ -182,7 +182,7 @@ "ldaps:", "magnet:", "mms:", "msnim:", "notes:", "rsync:", "secondlife:", "skype:", "ssh:", "sftp:", "smb:", "sms:", "snews:", "webcal:", "ymsgr:"] - in case parseURIReference u of + in case parseURIReference (escapeURI u) of Just p -> (map toLower $ uriScheme p) `notElem` safeURISchemes Nothing -> True @@ -746,7 +746,7 @@ Nothing -> fail "no href" let title = fromMaybe "" $ extractAttribute "title" attributes lab <- inlinesTilEnd "a" - return $ Link (normalizeSpaces lab) (url, title) + return $ Link (normalizeSpaces lab) (escapeURI url, title) image :: GenParser Char ParserState Inline image = try $ do @@ -756,5 +756,5 @@ Nothing -> fail "no src" let title = fromMaybe "" $ extractAttribute "title" attributes let alt = fromMaybe "" (extractAttribute "alt" attributes) - return $ Image [Str alt] (url, title) + return $ Image [Str alt] (escapeURI url, title) diff --git a/src/Text/Pandoc/Readers/LaTeX.hs b/src/Text/Pandoc/Readers/LaTeX.hs --- a/src/Text/Pandoc/Readers/LaTeX.hs +++ b/src/Text/Pandoc/Readers/LaTeX.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Readers.LaTeX - Copyright : Copyright (C) 2006-8 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane @@ -738,7 +738,7 @@ url = try $ do string "\\url" url' <- charsInBalanced '{' '}' - return $ Link [Code url'] (url', "") + return $ Link [Code url'] (escapeURI url', "") link :: GenParser Char ParserState Inline link = try $ do @@ -746,17 +746,16 @@ url' <- manyTill anyChar (char '}') char '{' label' <- manyTill inline (char '}') - return $ Link (normalizeSpaces label') (url', "") + return $ Link (normalizeSpaces label') (escapeURI url', "") image :: GenParser Char ParserState Inline image = try $ do ("includegraphics", _, args) <- command let args' = filter isArg args -- filter out options - let src = if null args' then - ("", "") - else - (stripFirstAndLast (head args'), "") - return $ Image [Str "image"] src + let (src,tit) = case args' of + [] -> ("", "") + (x:_) -> (stripFirstAndLast x, "") + return $ Image [Str "image"] (escapeURI src, tit) footnote :: GenParser Char ParserState Inline footnote = try $ do diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs --- a/src/Text/Pandoc/Readers/Markdown.hs +++ b/src/Text/Pandoc/Readers/Markdown.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Readers.Markdown - Copyright : Copyright (C) 2006-8 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane @@ -31,7 +31,7 @@ readMarkdown ) where -import Data.List ( transpose, isPrefixOf, isSuffixOf, sortBy, findIndex, intercalate ) +import Data.List ( transpose, isSuffixOf, sortBy, findIndex, intercalate ) import Data.Ord ( comparing ) import Data.Char ( isAlphaNum ) import Data.Maybe @@ -73,10 +73,6 @@ -- auxiliary functions -- --- | Replace spaces with %20 -uriEscapeSpaces :: String -> String -uriEscapeSpaces = substitute " " "%20" - indentSpaces :: GenParser Char ParserState [Char] indentSpaces = try $ do state <- getState @@ -206,7 +202,7 @@ tit <- option "" referenceTitle blanklines endPos <- getPosition - let newkey = (lab, (uriEscapeSpaces $ removeTrailingSpace src, tit)) + let newkey = (lab, (escapeURI $ removeTrailingSpace src, tit)) st <- getState let oldkeys = stateKeys st updateState $ \s -> s { stateKeys = newkey : oldkeys } @@ -1194,7 +1190,7 @@ tit <- option "" linkTitle skipSpaces eof - return (uriEscapeSpaces $ removeTrailingSpace src, tit) + return (escapeURI $ removeTrailingSpace src, tit) linkTitle :: GenParser Char st String linkTitle = try $ do @@ -1208,11 +1204,11 @@ link :: GenParser Char ParserState Inline link = try $ do lab <- reference - src <- source <|> referenceLink lab + (src, tit) <- source <|> referenceLink lab sanitize <- getState >>= return . stateSanitizeHTML - if sanitize && unsanitaryURI (fst src) + if sanitize && unsanitaryURI src then fail "Unsanitary URI" - else return $ Link lab src + else return $ Link lab (src, tit) -- a link like [this][ref] or [this][] or [this] referenceLink :: [Inline] @@ -1229,18 +1225,15 @@ autoLink :: GenParser Char ParserState Inline autoLink = try $ do char '<' - src <- uri <|> (emailAddress >>= (return . ("mailto:" ++))) + (orig, src) <- uri <|> emailAddress char '>' - let src' = if "mailto:" `isPrefixOf` src - then drop 7 src - else src st <- getState let sanitize = stateSanitizeHTML st if sanitize && unsanitaryURI src then fail "Unsanitary URI" else return $ if stateStrict st - then Link [Str src'] (src, "") - else Link [Code src'] (src, "") + then Link [Str orig] (src, "") + else Link [Code orig] (src, "") image :: GenParser Char ParserState Inline image = try $ do diff --git a/src/Text/Pandoc/Readers/RST.hs b/src/Text/Pandoc/Readers/RST.hs --- a/src/Text/Pandoc/Readers/RST.hs +++ b/src/Text/Pandoc/Readers/RST.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Readers.RST - Copyright : Copyright (C) 2006-8 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane @@ -556,7 +556,7 @@ contents <- many1 (try (many spaceChar >> newline >> many1 spaceChar >> noneOf " \t\n") <|> noneOf "\n") blanklines - return contents + return $ escapeURI $ removeLeadingTrailingSpace $ contents imageKey :: GenParser Char ParserState ([Inline], (String, [Char])) imageKey = try $ do @@ -565,13 +565,13 @@ skipSpaces string "image::" src <- targetURI - return (normalizeSpaces ref, (removeLeadingTrailingSpace src, "")) + return (normalizeSpaces ref, (src, "")) anonymousKey :: GenParser Char st ([Inline], (String, [Char])) anonymousKey = try $ do oneOfStrings [".. __:", "__"] src <- targetURI - return ([Str "_"], (removeLeadingTrailingSpace src, "")) + return ([Str "_"], (src, "")) regularKey :: GenParser Char ParserState ([Inline], (String, [Char])) regularKey = try $ do @@ -579,7 +579,7 @@ ref <- referenceName char ':' src <- targetURI - return (normalizeSpaces ref, (removeLeadingTrailingSpace src, "")) + return (normalizeSpaces ref, (src, "")) -- -- tables @@ -883,7 +883,8 @@ src <- manyTill (noneOf ">\n") (char '>') skipSpaces string "`_" - return $ Link (normalizeSpaces label') (removeLeadingTrailingSpace src, "") + return $ Link (normalizeSpaces label') + (escapeURI $ removeLeadingTrailingSpace src, "") referenceLink :: GenParser Char ParserState Inline referenceLink = try $ do @@ -891,25 +892,25 @@ key <- option label' (do{char '_'; return [Str "_"]}) -- anonymous link state <- getState let keyTable = stateKeys state - src <- case lookupKeySrc keyTable key of - Nothing -> fail "no corresponding key" - Just target -> return target + (src,tit) <- case lookupKeySrc keyTable key of + Nothing -> fail "no corresponding key" + Just target -> return target -- if anonymous link, remove first anon key so it won't be used again let keyTable' = if (key == [Str "_"]) -- anonymous link? - then delete ([Str "_"], src) keyTable -- remove first anon key + then delete ([Str "_"], (src,tit)) keyTable -- remove first anon key else keyTable setState $ state { stateKeys = keyTable' } - return $ Link (normalizeSpaces label') src + return $ Link (normalizeSpaces label') (src, tit) autoURI :: GenParser Char ParserState Inline autoURI = do - src <- uri - return $ Link [Str src] (src, "") + (orig, src) <- uri + return $ Link [Str orig] (src, "") autoEmail :: GenParser Char ParserState Inline autoEmail = do - src <- emailAddress - return $ Link [Str src] ("mailto:" ++ src, "") + (orig, src) <- emailAddress + return $ Link [Str orig] (src, "") autoLink :: GenParser Char ParserState Inline autoLink = autoURI <|> autoEmail @@ -921,7 +922,7 @@ ref <- manyTill inline (char '|') state <- getState let keyTable = stateKeys state - src <- case lookupKeySrc keyTable ref of - Nothing -> fail "no corresponding key" - Just target -> return target - return $ Image (normalizeSpaces ref) src + (src,tit) <- case lookupKeySrc keyTable ref of + Nothing -> fail "no corresponding key" + Just target -> return target + return $ Image (normalizeSpaces ref) (src, tit) diff --git a/src/Text/Pandoc/Readers/TeXMath.hs b/src/Text/Pandoc/Readers/TeXMath.hs --- a/src/Text/Pandoc/Readers/TeXMath.hs +++ b/src/Text/Pandoc/Readers/TeXMath.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2007 John MacFarlane +Copyright (C) 2007-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Readers.TeXMath - Copyright : Copyright (C) 2007 John MacFarlane + Copyright : Copyright (C) 2007-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Shared.hs b/src/Text/Pandoc/Shared.hs --- a/src/Text/Pandoc/Shared.hs +++ b/src/Text/Pandoc/Shared.hs @@ -1,6 +1,6 @@ {-# LANGUAGE DeriveDataTypeable #-} {- -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -19,7 +19,7 @@ {- | Module : Text.Pandoc.Shared - Copyright : Copyright (C) 2006-8 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane @@ -28,7 +28,7 @@ Utility functions and definitions used by the various Pandoc modules. -} -module Text.Pandoc.Shared ( +module Text.Pandoc.Shared ( -- * List processing splitBy, splitByIndices, @@ -43,6 +43,8 @@ stripFirstAndLast, camelCaseToHyphenated, toRomanNumeral, + escapeURI, + unescapeURI, wrapped, wrapIfNeeded, wrappedTeX, @@ -114,10 +116,11 @@ import Text.PrettyPrint.HughesPJ ( Doc, fsep, ($$), (<>), empty, isEmpty, text, nest ) import qualified Text.PrettyPrint.HughesPJ as PP import Text.Pandoc.CharacterReferences ( characterReference ) -import Data.Char ( toLower, toUpper, ord, isLower, isUpper, isAlpha, +import Data.Char ( toLower, toUpper, ord, isLower, isUpper, isAlpha, isAscii, isPunctuation ) import Data.List ( find, isPrefixOf, intercalate ) -import Network.URI ( parseURI, URI (..), isAllowedInURI ) +import Network.URI ( parseURI, URI (..), isAllowedInURI, escapeURIString, unEscapeString ) +import Codec.Binary.UTF8.String ( encodeString, decodeString ) import System.Directory import System.FilePath ( () ) -- Note: ghc >= 6.12 (base >=4.2) supports unicode through iconv @@ -131,6 +134,7 @@ import qualified Control.Monad.State as S import Control.Monad (join) import Paths_pandoc (getDataFileName) + -- -- List processing -- @@ -228,6 +232,17 @@ _ | x >= 1 -> "I" ++ toRomanNumeral (x - 1) _ -> "" +-- | Escape unicode characters in a URI. Characters that are +-- already valid in a URI, including % and ?, are left alone. +escapeURI :: String -> String +escapeURI = escapeURIString isAllowedInURI . encodeString + +-- | Unescape unicode and some special characters in a URI, but +-- without introducing spaces. +unescapeURI :: String -> String +unescapeURI = escapeURIString (\c -> isAllowedInURI c || not (isAscii c)) . + decodeString . unEscapeString + -- | Wrap inlines to line length. wrapped :: Monad m => ([Inline] -> m Doc) -> [Inline] -> m Doc wrapped listWriter sect = (mapM listWriter $ splitBy Space sect) >>= @@ -479,25 +494,30 @@ dom <- many1 $ try (char '.' >> many1 domainChar ) return $ intercalate "." (first:dom) --- | Parses an email address; returns string. -emailAddress :: GenParser Char st [Char] +-- | Parses an email address; returns original and corresponding +-- escaped mailto: URI. +emailAddress :: GenParser Char st (String, String) emailAddress = try $ do firstLetter <- alphaNum restAddr <- many emailChar let addr = firstLetter:restAddr char '@' dom <- domain - return $ addr ++ '@':dom + let full = addr ++ '@':dom + return (full, escapeURI $ "mailto:" ++ full) --- | Parses a URI. -uri :: GenParser Char st String +-- | Parses a URI. Returns pair of original and URI-escaped version. +uri :: GenParser Char st (String, String) uri = try $ do - str <- many1 $ satisfy isAllowedInURI - case parseURI str of - Just uri' -> if uriScheme uri' `elem` [ "http:", "https:", "ftp:", - "file:", "mailto:", - "news:", "telnet:" ] - then return $ show uri' + let protocols = [ "http:", "https:", "ftp:", "file:", "mailto:", + "news:", "telnet:" ] + lookAhead $ oneOfStrings protocols + -- scan non-ascii characters and ascii characters allowed in a URI + str <- many1 $ satisfy (\c -> not (isAscii c) || isAllowedInURI c) + -- now see if they amount to an absolute URI + case parseURI (escapeURI str) of + Just uri' -> if uriScheme uri' `elem` protocols + then return (str, show uri') else fail "not a URI" Nothing -> fail "not a URI" diff --git a/src/Text/Pandoc/Templates.hs b/src/Text/Pandoc/Templates.hs --- a/src/Text/Pandoc/Templates.hs +++ b/src/Text/Pandoc/Templates.hs @@ -1,6 +1,6 @@ {-# LANGUAGE TypeSynonymInstances #-} {- -Copyright (C) 2009 John MacFarlane +Copyright (C) 2009-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -19,7 +19,7 @@ {- | Module : Text.Pandoc.Templates - Copyright : Copyright (C) 2009 John MacFarlane + Copyright : Copyright (C) 2009-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/ConTeXt.hs b/src/Text/Pandoc/Writers/ConTeXt.hs --- a/src/Text/Pandoc/Writers/ConTeXt.hs +++ b/src/Text/Pandoc/Writers/ConTeXt.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2007-8 John MacFarlane +Copyright (C) 2007-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.ConTeXt - Copyright : Copyright (C) 2007-8 John MacFarlane + Copyright : Copyright (C) 2007-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/Docbook.hs b/src/Text/Pandoc/Writers/Docbook.hs --- a/src/Text/Pandoc/Writers/Docbook.hs +++ b/src/Text/Pandoc/Writers/Docbook.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.Docbook - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/HTML.hs b/src/Text/Pandoc/Writers/HTML.hs --- a/src/Text/Pandoc/Writers/HTML.hs +++ b/src/Text/Pandoc/Writers/HTML.hs @@ -1,6 +1,6 @@ {-# OPTIONS_GHC -fno-warn-deprecations #-} {- -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -19,7 +19,7 @@ {- | Module : Text.Pandoc.Writers.HTML - Copyright : Copyright (C) 2006-8 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/LaTeX.hs b/src/Text/Pandoc/Writers/LaTeX.hs --- a/src/Text/Pandoc/Writers/LaTeX.hs +++ b/src/Text/Pandoc/Writers/LaTeX.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.LaTeX - Copyright : Copyright (C) 2006-8 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/Man.hs b/src/Text/Pandoc/Writers/Man.hs --- a/src/Text/Pandoc/Writers/Man.hs +++ b/src/Text/Pandoc/Writers/Man.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2007 John MacFarlane +Copyright (C) 2007-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.Man - Copyright : Copyright (C) 2007 John MacFarlane + Copyright : Copyright (C) 2007-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/Markdown.hs b/src/Text/Pandoc/Writers/Markdown.hs --- a/src/Text/Pandoc/Writers/Markdown.hs +++ b/src/Text/Pandoc/Writers/Markdown.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.Markdown - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane @@ -400,9 +400,10 @@ inlineToMarkdown _ (LineBreak) = return $ text " \n" inlineToMarkdown _ Space = return $ char ' ' inlineToMarkdown opts (Cite _ cits) = inlineListToMarkdown opts cits -inlineToMarkdown opts (Link txt (src, tit)) = do +inlineToMarkdown opts (Link txt (src', tit)) = do linktext <- inlineListToMarkdown opts txt let linktitle = if null tit then empty else text $ " \"" ++ tit ++ "\"" + let src = unescapeURI src' let srcSuffix = if isPrefixOf "mailto:" src then drop 7 src else src let useRefLinks = writerReferenceLinks opts let useAuto = null tit && txt == [Code srcSuffix] @@ -423,7 +424,7 @@ (alternate == [Str source]) -- to prevent autolinks then [Str "image"] else alternate - linkPart <- inlineToMarkdown opts (Link txt (source, tit)) + linkPart <- inlineToMarkdown opts (Link txt (unescapeURI source, tit)) return $ char '!' <> linkPart inlineToMarkdown _ (Note contents) = do modify (\st -> st{ stNotes = contents : stNotes st }) diff --git a/src/Text/Pandoc/Writers/MediaWiki.hs b/src/Text/Pandoc/Writers/MediaWiki.hs --- a/src/Text/Pandoc/Writers/MediaWiki.hs +++ b/src/Text/Pandoc/Writers/MediaWiki.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2008 John MacFarlane +Copyright (C) 2008-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.MediaWiki - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2008-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/OpenDocument.hs b/src/Text/Pandoc/Writers/OpenDocument.hs --- a/src/Text/Pandoc/Writers/OpenDocument.hs +++ b/src/Text/Pandoc/Writers/OpenDocument.hs @@ -1,6 +1,7 @@ {-# LANGUAGE PatternGuards #-} {- -Copyright (C) 2008 Andrea Rossato +Copyright (C) 2008-2010 Andrea Rossato +and John MacFarlane. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -19,7 +20,7 @@ {- | Module : Text.Pandoc.Writers.OpenDocument - Copyright : Copyright (C) 2008 Andrea Rossato + Copyright : Copyright (C) 2008-2010 Andrea Rossato and John MacFarlane License : GNU GPL, version 2 or above Maintainer : Andrea Rossato @@ -40,6 +41,7 @@ import Control.Arrow ( (***), (>>>) ) import Control.Monad.State hiding ( when ) import Data.Char (chr) +import qualified Data.Map as Map -- | Auxiliary function to convert Plain block to Para. plainToPara :: Block -> Block @@ -56,7 +58,7 @@ , stParaStyles :: [Doc] , stListStyles :: [(Int, [Doc])] , stTextStyles :: [Doc] - , stTextStyleAttr :: [(TextStyle,[(String,String)])] + , stTextStyleAttr :: Map.Map TextStyle [(String,String)] , stIndentPara :: Int , stInDefinition :: Bool , stTight :: Bool @@ -69,7 +71,7 @@ , stParaStyles = [] , stListStyles = [] , stTextStyles = [] - , stTextStyleAttr = [] + , stTextStyleAttr = Map.empty , stIndentPara = 0 , stInDefinition = False , stTight = False @@ -91,11 +93,10 @@ addTextStyle i = modify $ \s -> s { stTextStyles = i : stTextStyles s } addTextStyleAttr :: (TextStyle, [(String,String)]) -> State WriterState () -addTextStyleAttr i = modify $ \s -> s { stTextStyleAttr = i : stTextStyleAttr s } +addTextStyleAttr (ts, xs) = modify $ \s -> s { stTextStyleAttr = Map.insert ts xs (stTextStyleAttr s) } -rmTextStyleAttr :: State WriterState () -rmTextStyleAttr = modify $ \s -> s { stTextStyleAttr = rmHead (stTextStyleAttr s) } - where rmHead l = if l /= [] then tail l else [] +rmTextStyleAttr :: TextStyle -> State WriterState () +rmTextStyleAttr ts = modify $ \s -> s { stTextStyleAttr = Map.delete ts (stTextStyleAttr s) } increaseIndent :: State WriterState () increaseIndent = modify $ \s -> s { stIndentPara = 1 + stIndentPara s } @@ -121,18 +122,18 @@ withTextStyle :: TextStyle -> State WriterState a -> State WriterState a withTextStyle s f = addTextStyleAttr (s,textStyleAttr s) >> - f >>= \r -> rmTextStyleAttr >> return r + f >>= \r -> rmTextStyleAttr s >> return r inTextStyle :: Doc -> State WriterState Doc inTextStyle d = do at <- gets stTextStyleAttr - if at == [] + if Map.null at then return d else do tn <- (+) 1 . length <$> gets stTextStyles addTextStyle $ inTags False "style:style" [("style:name" , "T" ++ show tn) ,("style:family", "text" )] - $ selfClosingTag "style:text-properties" (concatMap snd at) + $ selfClosingTag "style:text-properties" (concatMap snd $ Map.toList at) return $ inTags False "text:span" [("text:style-name","T" ++ show tn)] d inHeaderTags :: Int -> Doc -> Doc @@ -491,7 +492,7 @@ [ ("fo:text-align", x) , ("style:justify-single-word", "false")] -data TextStyle = Italic | Bold | Strike | Sub | Sup | SmallC deriving ( Eq ) +data TextStyle = Italic | Bold | Strike | Sub | Sup | SmallC deriving ( Eq,Ord ) textStyleAttr :: TextStyle -> [(String,String)] textStyleAttr s diff --git a/src/Text/Pandoc/Writers/RST.hs b/src/Text/Pandoc/Writers/RST.hs --- a/src/Text/Pandoc/Writers/RST.hs +++ b/src/Text/Pandoc/Writers/RST.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.RST - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane @@ -305,8 +305,9 @@ inlineToRST (Link [Code str] (src, _)) | src == str || src == "mailto:" ++ str = do let srcSuffix = if isPrefixOf "mailto:" src then drop 7 src else src - return $ text srcSuffix -inlineToRST (Link txt (src, tit)) = do + return $ text $ unescapeURI srcSuffix +inlineToRST (Link txt (src', tit)) = do + let src = unescapeURI src' useReferenceLinks <- get >>= (return . writerReferenceLinks . stOptions) linktext <- inlineListToRST $ normalizeSpaces txt if useReferenceLinks @@ -317,7 +318,8 @@ modify $ \st -> st { stLinks = refs' } return $ char '`' <> linktext <> text "`_" else return $ char '`' <> linktext <> text " <" <> text src <> text ">`_" -inlineToRST (Image alternate (source, tit)) = do +inlineToRST (Image alternate (source', tit)) = do + let source = unescapeURI source' pics <- get >>= (return . stImages) let labelsUsed = map fst pics let txt = if null alternate || alternate == [Str ""] || diff --git a/src/Text/Pandoc/Writers/RTF.hs b/src/Text/Pandoc/Writers/RTF.hs --- a/src/Text/Pandoc/Writers/RTF.hs +++ b/src/Text/Pandoc/Writers/RTF.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.RTF - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/S5.hs b/src/Text/Pandoc/Writers/S5.hs --- a/src/Text/Pandoc/Writers/S5.hs +++ b/src/Text/Pandoc/Writers/S5.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.S5 - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/Writers/Texinfo.hs b/src/Text/Pandoc/Writers/Texinfo.hs --- a/src/Text/Pandoc/Writers/Texinfo.hs +++ b/src/Text/Pandoc/Writers/Texinfo.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2008 John MacFarlane and Peter Wang +Copyright (C) 2008-2010 John MacFarlane and Peter Wang This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.Writers.Texinfo - Copyright : Copyright (C) 2008 John MacFarlane and Peter Wang + Copyright : Copyright (C) 2008-2010 John MacFarlane and Peter Wang License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/Text/Pandoc/XML.hs b/src/Text/Pandoc/XML.hs --- a/src/Text/Pandoc/XML.hs +++ b/src/Text/Pandoc/XML.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-7 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Text.Pandoc.XML - Copyright : Copyright (C) 2006-7 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/src/pandoc.hs b/src/pandoc.hs --- a/src/pandoc.hs +++ b/src/pandoc.hs @@ -1,5 +1,5 @@ {- -Copyright (C) 2006-8 John MacFarlane +Copyright (C) 2006-2010 John MacFarlane This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,7 +18,7 @@ {- | Module : Main - Copyright : Copyright (C) 2006-8 John MacFarlane + Copyright : Copyright (C) 2006-2010 John MacFarlane License : GNU GPL, version 2 or above Maintainer : John MacFarlane diff --git a/tests/markdown-reader-more.native b/tests/markdown-reader-more.native --- a/tests/markdown-reader-more.native +++ b/tests/markdown-reader-more.native @@ -27,5 +27,9 @@ , Para [Code "hi there"] , Para [Code "hi````there"] , Para [Str "`",Str "hi"] -, Para [Str "there",Str "`"] ] +, Para [Str "there",Str "`"] +, Header 2 [Str "Multilingual",Space,Str "URLs"] +, Para [Link [Code "http://\27979.com?\27979=\27979"] ("http://%E6%B5%8B.com?%E6%B5%8B=%E6%B5%8B","")] +, Para [Link [Str "foo"] ("/bar/%E6%B5%8B?x=%E6%B5%8B","title")] +, Para [Link [Code "\27979@foo.\27979.baz"] ("mailto:%E6%B5%8B@foo.%E6%B5%8B.baz","")] ] diff --git a/tests/markdown-reader-more.txt b/tests/markdown-reader-more.txt --- a/tests/markdown-reader-more.txt +++ b/tests/markdown-reader-more.txt @@ -83,3 +83,11 @@ there` +## Multilingual URLs + + + +[foo](/bar/测?x=测 "title") + +<测@foo.测.baz> +