d339b29 · typeable/pandoc

git.iko.soy/typeable/pandoc

Added skeleton of basic docbook reader.

commitd339b29967878f64d5fe45d03d214476e9d88f7e authorJohn MacFarlane <fiddlosopher@gmail.com> dateSat 14 Apr 2012, 16:44 (15 years ago) parente37c452

4 files changed+40−1

compare with master

unifiedsplit

pandoc.cabal+1 src/Text/Pandoc.hs+2 src/Text/Pandoc/Pretty.hs+1−1 src/Text/Pandoc/Readers/DocBook.hs+36

pandoc.cabal+1

@@ -246,6 +246,7 @@ Library

246246 Text.Pandoc.Readers.LaTeX,

247247 Text.Pandoc.Readers.Markdown,

248248 Text.Pandoc.Readers.RST,

249+ Text.Pandoc.Readers.DocBook,

249250 Text.Pandoc.Readers.TeXMath,

250251 Text.Pandoc.Readers.Textile,

251252 Text.Pandoc.Readers.Native,

src/Text/Pandoc.hs+2

@@ -119,6 +119,7 @@ import Text.Pandoc.Definition

119119import Text.Pandoc.Generic

120120import Text.Pandoc.Readers.Markdown

121121import Text.Pandoc.Readers.RST

122+import Text.Pandoc.Readers.DocBook

122123import Text.Pandoc.Readers.LaTeX

123124import Text.Pandoc.Readers.HTML

124125import Text.Pandoc.Readers.Textile

@@ -162,6 +163,7 @@ readers = [("native" , \_ -> readNative)

162163 ,("rst" , readRST)

163164 ,("rst+lhs" , \st ->

164165 readRST st{ stateLiterateHaskell = True})

166+ ,("docbook" , readDocBook)

165167 ,("textile" , readTextile) -- TODO : textile+lhs

166168 ,("html" , readHtml)

167169 ,("latex" , readLaTeX)

src/Text/Pandoc/Pretty.hs+1−1

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

1−{-# LANGUAGE GeneralizedNewtypeDeriving #-}

1+{-# LANGUAGE GeneralizedNewtypeDeriving, DatatypeContexts #-}

22{-

src/Text/Pandoc/Readers/DocBook.hsnew file+36

@@ -0,0 +1,36 @@

1+module Text.Pandoc.Readers.DocBook ( readDocBook ) where

2+import Text.Pandoc.Parsing (ParserState(..), defaultParserState)

3+import Text.Pandoc.Definition

4+import Text.Pandoc.Builder

5+import Text.XML.Light

6+import Data.Monoid

7+import Data.Char (isSpace)

9+readDocBook :: ParserState -> String -> Pandoc

10+readDocBook st inp = Pandoc (Meta [] [] []) $ toList blocks

11+ where blocks = mconcat $ map (parseBlock st) $ parseXML inp

12+

13+parseBlock :: ParserState -> Content -> Blocks

14+parseBlock st (Text (CData _ s _)) = if all isSpace s

15+ then mempty

16+ else plain $ text s

17+parseBlock st (Elem e) =

18+ case qName (elName e) of

19+ "para" -> para $ trimInlines $ mconcat

20+ $ map (parseInline st) $ elContent e

21+ _ -> mconcat $ map (parseBlock st) $ elContent e

22+parseBlock st (CRef _) = mempty

23+

24+parseInline :: ParserState -> Content -> Inlines

25+parseInline st (Text (CData _ s _)) = text s

26+parseInline st (Elem e) =

27+ case qName (elName e) of

28+ "emphasis" -> case lookupAttrBy (\attr -> qName attr == "role")

29+ (elAttribs e) of

30+ Just "strong" -> strong innerInlines

31+ _ -> emph innerInlines

32+ _ -> innerInlines

33+ where innerInlines = trimInlines . mconcat . map (parseInline st)

34+ $ elContent e

35+parseInline st (CRef _) = mempty

36+