def05d3 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

HTML reader: Parse LaTeX math if appropriate options are set.

def05d3John MacFarlane13 years ago

src/Text/Pandoc/Parsing.hs+35

compare with master

unifiedsplit

@@ -48,6 +48,8 @@ module Text.Pandoc.Parsing ( (>>~),

48 romanNumeral,

49 emailAddress,

50 uri,

51+ mathInline,

52+ mathDisplay,

51 withHorizDisplacement,

53 withHorizDisplacement,

52 withRaw,

54 withRaw,

53 escaped,

55 escaped,

@@ -455,6 +457,39 @@ uri = try $ do

455 let uri' = scheme ++ ":" ++ fromEntities str'

457 let uri' = scheme ++ ":" ++ fromEntities str'

456 return (uri', escapeURI uri')

458 return (uri', escapeURI uri')

457

459

460+mathInlineWith :: String -> String -> Parser [Char] st String

461+mathInlineWith op cl = try $ do

462+ string op

463+ notFollowedBy space

464+ words' <- many1Till (count 1 (noneOf "\n\\")

465+ <|> (char '\\' >> anyChar >>= \c -> return ['\\',c])

466+ <|> count 1 newline <* notFollowedBy' blankline

467+ *> return " ")

468+ (try $ string cl)

469+ notFollowedBy digit -- to prevent capture of $5

470+ return $ concat words'

471+

472+mathDisplayWith :: String -> String -> Parser [Char] st String

473+mathDisplayWith op cl = try $ do

474+ string op

475+ many1Till (noneOf "\n" <|> (newline >>~ notFollowedBy' blankline)) (try $ string cl)

476+

477+mathDisplay :: Parser [Char] ParserState String

478+mathDisplay =

479+ (guardEnabled Ext_tex_math_dollars >> mathDisplayWith "$$" "$$")

480+ <|> (guardEnabled Ext_tex_math_single_backslash >>

481+ mathDisplayWith "\\[" "\\]")

482+ <|> (guardEnabled Ext_tex_math_double_backslash >>

483+ mathDisplayWith "\\\\[" "\\\\]")

484+

485+mathInline :: Parser [Char] ParserState String

486+mathInline =

487+ (guardEnabled Ext_tex_math_dollars >> mathInlineWith "$" "$")

488+ <|> (guardEnabled Ext_tex_math_single_backslash >>

489+ mathInlineWith "\$" "\$")

490+ <|> (guardEnabled Ext_tex_math_double_backslash >>

491+ mathInlineWith "\\\$" "\\\$")

492+

458-- | Applies a parser, returns tuple of its results and its horizontal

493-- | Applies a parser, returns tuple of its results and its horizontal

459-- displacement (the difference between the source column at the end

494-- displacement (the difference between the source column at the end

460-- and the source column at the beginning). Vertical displacement

495-- and the source column at the beginning). Vertical displacement

src/Text/Pandoc/Readers/HTML.hs+8−1

@@ -467,7 +467,13 @@ pBlank = try $ do

467

468pTagContents :: Parser [Char] ParserState Inline

469pTagContents =

470+ Math InlineMath `fmap` mathInline

471+ <|> Math DisplayMath `fmap` mathDisplay

472+ <|> pStr

473+ <|> pSpace

470− pStr <|> pSpace <|> smartPunctuation pTagContents <|> pSymbol <|> pBad

474+ <|> smartPunctuation pTagContents

475+ <|> pSymbol

476+ <|> pBad

471

477

472pStr :: Parser [Char] ParserState Inline

478pStr :: Parser [Char] ParserState Inline

473pStr = do

479pStr = do

@@ -482,6 +488,7 @@ isSpecial '"' = True

482isSpecial '\'' = True

488isSpecial '\'' = True

483isSpecial '.' = True

489isSpecial '.' = True

484isSpecial '-' = True

490isSpecial '-' = True

491+isSpecial '$' = True

485isSpecial '\8216' = True

492isSpecial '\8216' = True

486isSpecial '\8217' = True

493isSpecial '\8217' = True

487isSpecial '\8220' = True

494isSpecial '\8220' = True

src/Text/Pandoc/Readers/LaTeX.hs+2−1

@@ -38,7 +38,8 @@ import Text.Pandoc.Definition

38import Text.Pandoc.Walk

39import Text.Pandoc.Shared

40import Text.Pandoc.Options

41−import Text.Pandoc.Parsing hiding ((<|>), many, optional, space)

41+import Text.Pandoc.Parsing hiding ((<|>), many, optional, space,

42+ mathDisplay, mathInline)

42import qualified Text.Pandoc.UTF8 as UTF8

43import qualified Text.Pandoc.UTF8 as UTF8

43import Data.Char ( chr, ord )

44import Data.Char ( chr, ord )

44import Control.Monad

45import Control.Monad

src/Text/Pandoc/Readers/Markdown.hs−33

@@ -1408,39 +1408,6 @@ math :: MarkdownParser (F Inlines)

1408math = (return . B.displayMath <$> (mathDisplay >>= applyMacros'))

1409 <|> (return . B.math <$> (mathInline >>= applyMacros'))

1410

1411−mathDisplay :: MarkdownParser String

1412−mathDisplay =

1413− (guardEnabled Ext_tex_math_dollars >> mathDisplayWith "$$" "$$")

1414− <|> (guardEnabled Ext_tex_math_single_backslash >>

1415− mathDisplayWith "\\[" "\\]")

1416− <|> (guardEnabled Ext_tex_math_double_backslash >>

1417− mathDisplayWith "\\\\[" "\\\\]")

1418−

1419−mathDisplayWith :: String -> String -> MarkdownParser String

1420−mathDisplayWith op cl = try $ do

1421− string op

1422− many1Till (noneOf "\n" <|> (newline >>~ notFollowedBy' blankline)) (try $ string cl)

1423−

1424−mathInline :: MarkdownParser String

1425−mathInline =

1426− (guardEnabled Ext_tex_math_dollars >> mathInlineWith "$" "$")

1427− <|> (guardEnabled Ext_tex_math_single_backslash >>

1428− mathInlineWith "\$" "\$")

1429− <|> (guardEnabled Ext_tex_math_double_backslash >>

1430− mathInlineWith "\\\$" "\\\$")

1431−

1432−mathInlineWith :: String -> String -> MarkdownParser String

1433−mathInlineWith op cl = try $ do

1434− string op

1435− notFollowedBy space

1436− words' <- many1Till (count 1 (noneOf "\n\\")

1437− <|> (char '\\' >> anyChar >>= \c -> return ['\\',c])

1438− <|> count 1 newline <* notFollowedBy' blankline

1439− *> return " ")

1440− (try $ string cl)

1441− notFollowedBy digit -- to prevent capture of $5

1442− return $ concat words'

1443−

1444-- Parses material enclosed in *s, **s, _s, or __s.

1411-- Parses material enclosed in *s, **s, _s, or __s.

1445-- Designed to avoid backtracking.

1412-- Designed to avoid backtracking.

1446enclosure :: Char

1413enclosure :: Char