a2c4261 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

HTML reader: allow enabling `raw_tex` extension.

a2c4261John MacFarlane8 years ago

MANUAL.txt+2−1

compare with master

unifiedsplit

@@ -1917,7 +1917,8 @@ This extension can be enabled/disabled for the following formats

1917(in addition to `markdown`):

1918

1919input formats

1920−: `latex`, `org`, `textile`

1920+: `latex`, `org`, `textile`, `html` (environments, `\ref`, and

1921+ `\eqref` only)

1921

1922

1922output formats

1923output formats

1923: `textile`, `commonmark`

1924: `textile`, `commonmark`

src/Text/Pandoc/Readers/HTML.hs+28−3

@@ -68,11 +68,13 @@ import qualified Text.Pandoc.Builder as B

68import Text.Pandoc.Class (PandocMonad (..))

69import Text.Pandoc.CSS (foldOrElse, pickStyleAttrProps)

70import Text.Pandoc.Definition

71+import Text.Pandoc.Readers.LaTeX (rawLaTeXInline)

72+import Text.Pandoc.Readers.LaTeX.Types (Macro)

71import Text.Pandoc.Error

73import Text.Pandoc.Error

72import Text.Pandoc.Logging

74import Text.Pandoc.Logging

73import Text.Pandoc.Options (

75import Text.Pandoc.Options (

74 Extension (Ext_epub_html_exts, Ext_empty_paragraphs, Ext_native_divs,

76 Extension (Ext_epub_html_exts, Ext_empty_paragraphs, Ext_native_divs,

75− Ext_native_spans, Ext_raw_html, Ext_line_blocks),

77+ Ext_native_spans, Ext_raw_html, Ext_line_blocks, Ext_raw_tex),

76 ReaderOptions (readerExtensions, readerStripComments),

78 ReaderOptions (readerExtensions, readerStripComments),

77 extensionEnabled)

79 extensionEnabled)

78import Text.Pandoc.Parsing hiding ((<|>))

80import Text.Pandoc.Parsing hiding ((<|>))

@@ -102,7 +104,8 @@ readHtml opts inp = do

102 (m:_) -> messageString m

104 (m:_) -> messageString m

103 result <- flip runReaderT def $

105 result <- flip runReaderT def $

104 runParserT parseDoc

106 runParserT parseDoc

105− (HTMLState def{ stateOptions = opts } [] Nothing Set.empty M.empty [])

107+ (HTMLState def{ stateOptions = opts }

108+ [] Nothing Set.empty M.empty [] M.empty)

106 "source" tags

109 "source" tags

107 case result of

110 case result of

108 Right doc -> return doc

111 Right doc -> return doc

@@ -124,7 +127,8 @@ data HTMLState =

124 baseHref :: Maybe URI,

127 baseHref :: Maybe URI,

125 identifiers :: Set.Set String,

128 identifiers :: Set.Set String,

126 headerMap :: M.Map Inlines String,

129 headerMap :: M.Map Inlines String,

127− logMessages :: [LogMessage]

130+ logMessages :: [LogMessage],

131+ macros :: M.Map Text Macro

128 }

132 }

129

133

130data HTMLLocal = HTMLLocal { quoteContext :: QuoteContext

134data HTMLLocal = HTMLLocal { quoteContext :: QuoteContext

@@ -907,9 +911,25 @@ pTagContents =

907 <|> pStr

911 <|> pStr

908 <|> pSpace

912 <|> pSpace

909 <|> smartPunctuation pTagContents

913 <|> smartPunctuation pTagContents

914+ <|> pRawTeX

910 <|> pSymbol

915 <|> pSymbol

911 <|> pBad

916 <|> pBad

912

917

918+pRawTeX :: PandocMonad m => InlinesParser m Inlines

919+pRawTeX = do

920+ lookAhead $ try $ do

921+ char '\\'

922+ choice $ map (try . string) ["begin", "eqref", "ref"]

923+ guardEnabled Ext_raw_tex

924+ inp <- getInput

925+ st <- getState

926+ res <- lift $ runParserT (withRaw rawLaTeXInline) st "chunk" (T.unpack inp)

927+ case res of

928+ Left _ -> mzero

929+ Right (contents, raw) -> do

930+ _ <- count (length raw) anyChar

931+ return $ B.rawInline "tex" contents

932+

913pStr :: PandocMonad m => InlinesParser m Inlines

933pStr :: PandocMonad m => InlinesParser m Inlines

914pStr = do

934pStr = do

915 result <- many1 $ satisfy $ \c ->

935 result <- many1 $ satisfy $ \c ->

@@ -923,6 +943,7 @@ isSpecial '\'' = True

923isSpecial '.' = True

943isSpecial '.' = True

924isSpecial '-' = True

944isSpecial '-' = True

925isSpecial '$' = True

945isSpecial '$' = True

946+isSpecial '\\' = True

926isSpecial '\8216' = True

947isSpecial '\8216' = True

927isSpecial '\8217' = True

948isSpecial '\8217' = True

928isSpecial '\8220' = True

949isSpecial '\8220' = True

@@ -1249,6 +1270,10 @@ isSpace _ = False

1249

1270

1250-- Instances

1271-- Instances

1251

1272

1273+instance HasMacros HTMLState where

1274+ extractMacros = macros

1275+ updateMacros f st = st{ macros = f $ macros st }

1276+

1252instance HasIdentifierList HTMLState where

1277instance HasIdentifierList HTMLState where

1253 extractIdentifierList = identifiers

1278 extractIdentifierList = identifiers

1254 updateIdentifierList f s = s{ identifiers = f (identifiers s) }

1279 updateIdentifierList f s = s{ identifiers = f (identifiers s) }

test/command/1126.mdnew file+29

@@ -0,0 +1,29 @@

1+```

2+% pandoc -f html -t latex

3+\begin{eqnarray}

4+A&=&B,\\

5+C&=&D

6+\end{eqnarray}

7+^D

8+\textbackslash{}begin\{eqnarray\}

9+A\&=\&B,\textbackslash{}\textbackslash{} C\&=\&D

10+\textbackslash{}end\{eqnarray\}

11+```

12+

13+```

14+% pandoc -f html+raw_tex -t latex

15+<p>See \eqref{myeq}.</p>

16+\begin{eqnarray}

17+A&=&B,\\

18+C&=&D

19+\\label{myeq}

20+\end{eqnarray}

21+^D

22+See \eqref{myeq}.

23+

24+\begin{eqnarray}

25+A&=&B,\\

26+C&=&D

27+\\label{myeq}

28+\end{eqnarray}

29+```