3dab6c5 · typeable/pandoc

40import Data.Maybe ( fromMaybe )

41import Data.Char ( chr, toUpper )

42import Data.List ( isPrefixOf, isSuffixOf )

43−import Control.Monad ( when, liftM )

43+import Control.Monad ( when )

44

45-- | Parse LaTeX from string and return 'Pandoc' document.

46readLaTeX :: ParserState -- ^ Parser state, including options for parser

119

120-- | Process LaTeX preamble, extracting metadata.

121processLaTeXPreamble :: GenParser Char ParserState ()

122−processLaTeXPreamble =

122+processLaTeXPreamble = do

123+ try $ string "\\documentclass"

123− skipMany $ choice [ bibliographic, unknownCommand,

124+ skipMany $ bibliographic <|> macro <|> commentBlock <|> skipChar

124− commentBlock, skipToken ]

125

126-- | Parse LaTeX and return 'Pandoc'.

127parseLaTeX :: GenParser Char ParserState Pandoc

128parseLaTeX = do

129+ skipMany $ spaces >> comment

129 spaces

130 spaces

130− blocks <- try (processLaTeXPreamble >> spaces >> environment "document")

131+ blocks <- try (processLaTeXPreamble >> environment "document")

131− <|> (many block >>~ (spaces >> eof))

132+ <|> (many block >>~ (spaces >> eof))

132 state <- getState

133 state <- getState

133 let blocks' = filter (/= Null) blocks

134 let blocks' = filter (/= Null) blocks

134 let title' = stateTitle state

135 let title' = stateTitle state

157 , unknownEnvironment

158 , unknownEnvironment

158 , ignore

159 , ignore

159 , unknownCommand

160 , unknownCommand

160− , skipToken

161 ] <?> "block"

162

163--

407 else anyEnvironment -- otherwise just the contents

408 return result

409

410+group :: GenParser Char ParserState Inline

411+group = do

412+ res <- bracketedText '{' '}'

413+ return $ TeX $ "{" ++ res ++ "}"

414+

410-- \ignore{} is used conventionally in literate haskell for definitions

415-- \ignore{} is used conventionally in literate haskell for definitions

411-- that are to be processed by the compiler but not printed.

416-- that are to be processed by the compiler but not printed.

412ignore :: GenParser Char ParserState Block

417ignore :: GenParser Char ParserState Block

437commandsToIgnore :: [String]

442commandsToIgnore :: [String]

438commandsToIgnore = ["special","pdfannot","pdfstringdef","bibliography"]

443commandsToIgnore = ["special","pdfannot","pdfstringdef","bibliography"]

439

444

440−skipToken :: GenParser Char ParserState Block

445+skipChar :: GenParser Char ParserState Block

446+skipChar = do

447+ satisfy (/='\\') <|>

448+ (notFollowedBy' (lookAhead $ string "\\begin{document}") >> anyChar)

449+ spaces

441−skipToken = satisfy (/='\\') >> spaces >> return Null

450+ return Null

442

451

443commentBlock :: GenParser Char st Block

452commentBlock :: GenParser Char st Block

444commentBlock = comment >> return Null

453commentBlock = comment >> return Null

474 , accentedChar

483 , accentedChar

475 , nonbreakingSpace

484 , nonbreakingSpace

476 , cite

485 , cite

486+ , group

477 , specialChar

487 , specialChar

478 , rawLaTeXInline'

488 , rawLaTeXInline'

479 , escapedChar

489 , escapedChar

906rawLaTeXInline' = do

916rawLaTeXInline' = do

907 notFollowedBy' $ oneOfStrings ["\\begin", "\\end", "\\item", "\\ignore",

917 notFollowedBy' $ oneOfStrings ["\\begin", "\\end", "\\item", "\\ignore",

908 "\\section"]

918 "\\section"]

909− rawLaTeXInline <|> liftM TeX (bracketedText '{' '}')

919+ rawLaTeXInline

910

920

911-- | Parse any LaTeX command and return it in a raw TeX inline element.

921-- | Parse any LaTeX command and return it in a raw TeX inline element.

912rawLaTeXInline :: GenParser Char ParserState Inline

922rawLaTeXInline :: GenParser Char ParserState Inline