953cbb7 · typeable/pandoc

git.iko.soy/typeable/pandoc

Moved more into T.P.Readers.LaTeX.Lang.

commit953cbb7f6ac42bde750546ef88e57841656ff8b9 authorJohn MacFarlane <jgm@berkeley.edu> dateWed 3 Mar 2021, 11:08 (6 years ago) parent2cbe949

3 files changed+97−82

compare with master

unifiedsplit

src/Text/Pandoc/Readers/LaTeX.hs+6−78 src/Text/Pandoc/Readers/LaTeX/Inline.hs+62−3 src/Text/Pandoc/Readers/LaTeX/Lang.hs+29−1

src/Text/Pandoc/Readers/LaTeX.hs+6−78

@@ -34,7 +34,7 @@ import qualified Data.Set as Set

3434import Data.Text (Text)

3535import qualified Data.Text as T

3636import System.FilePath (addExtension, replaceExtension, takeExtension)

37−import Text.Pandoc.BCP47 (Lang (..), renderLang)

37+import Text.Pandoc.BCP47 (renderLang)

3838import Text.Pandoc.Builder as B

3939import Text.Pandoc.Class.PandocPure (PandocPure)

4040import Text.Pandoc.Class.PandocMonad (PandocMonad (..), getResourcePath,

@@ -58,11 +58,12 @@ import Text.Pandoc.Readers.LaTeX.Math (dollarsMath, inlineEnvironments,

5858 theoremEnvironment)

5959import Text.Pandoc.Readers.LaTeX.Table (tableEnvironments)

6060import Text.Pandoc.Readers.LaTeX.Macro (macroDef)

61−import Text.Pandoc.Readers.LaTeX.Lang (polyglossiaLangToBCP47,

61+import Text.Pandoc.Readers.LaTeX.Lang (inlineLanguageCommands,

6262 babelLangToBCP47, setDefaultLanguage)

6363import Text.Pandoc.Readers.LaTeX.SIunitx (siunitxCommands)

6464import Text.Pandoc.Readers.LaTeX.Inline (acronymCommands, refCommands,

6565 nameCommands, charCommands,

66+ biblatexInlineCommands,

6667 verbCommands, rawInlineOr,

6768 listingsLanguage)

6869import Text.Pandoc.Shared

@@ -350,8 +351,7 @@ unescapeURL = T.concat . go . T.splitOn "\\"

350351

351352inlineCommands :: PandocMonad m => M.Map Text (LP m Inlines)

352353inlineCommands = M.unions

353− [ inlineLanguageCommands

354− , accentCommands tok

354+ [ accentCommands tok

355355 , citationCommands inline

356356 , siunitxCommands tok

357357 , acronymCommands

@@ -359,6 +359,8 @@ inlineCommands = M.unions

359359 , nameCommands

360360 , verbCommands

361361 , charCommands

362+ , inlineLanguageCommands tok

363+ , biblatexInlineCommands tok

362364 , rest ]

363365 where

364366 rest = M.fromList

@@ -373,12 +375,6 @@ inlineCommands = M.unions

373375 , ("texttt", ttfamily)

374376 , ("sout", extractSpaces strikeout <$> tok)

375377 , ("alert", skipopts >> spanWith ("",["alert"],[]) <$> tok) -- beamer

376− , ("lq", return (str "‘"))

377− , ("rq", return (str "’"))

378− , ("textquoteleft", return (str "‘"))

379− , ("textquoteright", return (str "’"))

380− , ("textquotedblleft", return (str "“"))

381− , ("textquotedblright", return (str "”"))

382378 , ("textsuperscript", extractSpaces superscript <$> tok)

383379 , ("textsubscript", extractSpaces subscript <$> tok)

384380 , ("textbf", extractSpaces strong <$> tok)

@@ -447,11 +443,6 @@ inlineCommands = M.unions

447443 , ("toggletrue", braced >>= setToggle True)

448444 , ("togglefalse", braced >>= setToggle False)

449445 , ("iftoggle", try $ ifToggle >> inline)

450− -- biblatex misc

451− , ("RN", romanNumeralUpper)

452− , ("Rn", romanNumeralLower)

453− -- babel

454− , ("foreignlanguage", foreignlanguage)

455446 -- include

456447 , ("input", rawInlineOr "input" $ include "input")

457448 -- soul package

@@ -460,24 +451,6 @@ inlineCommands = M.unions

460451 , ("uline", underline <$> tok)

461452 -- plain tex stuff that should just be passed through as raw tex

462453 , ("ifdim", ifdim)

463− -- bibtex

464− , ("mkbibquote", spanWith nullAttr . doubleQuoted <$> tok)

465− , ("mkbibemph", spanWith nullAttr . emph <$> tok)

466− , ("mkbibitalic", spanWith nullAttr . emph <$> tok)

467− , ("mkbibbold", spanWith nullAttr . strong <$> tok)

468− , ("mkbibparens",

469− spanWith nullAttr . (\x -> str "(" <> x <> str ")") <$> tok)

470− , ("mkbibbrackets",

471− spanWith nullAttr . (\x -> str "[" <> x <> str "]") <$> tok)

472− , ("autocap", spanWith nullAttr <$> tok)

473− , ("textnormal", spanWith ("",["nodecor"],[]) <$> tok)

474− , ("bibstring",

475− (\x -> spanWith ("",[],[("bibstring",x)]) (str x)) . untokenize

476− <$> braced)

477− , ("adddot", pure (str "."))

478− , ("adddotspace", pure (spanWith nullAttr (str "." <> space)))

479− , ("addabbrvspace", pure space)

480− , ("hyphen", pure (str "-"))

481454 ]

482455

483456lettrine :: PandocMonad m => LP m Inlines

@@ -502,26 +475,6 @@ alterStr :: (Text -> Text) -> Inline -> Inline

502475alterStr f (Str xs) = Str (f xs)

503476alterStr _ x = x

504477

505−foreignlanguage :: PandocMonad m => LP m Inlines

506−foreignlanguage = do

507− babelLang <- untokenize <$> braced

508− case babelLangToBCP47 babelLang of

509− Just lang -> spanWith ("", [], [("lang", renderLang lang)]) <$> tok

510− _ -> tok

511−

512−inlineLanguageCommands :: PandocMonad m => M.Map Text (LP m Inlines)

513−inlineLanguageCommands = M.fromList $ mk <$> M.toList polyglossiaLangToBCP47

514− where

515− mk (polyglossia, bcp47Func) =

516− ("text" <> polyglossia, inlineLanguage bcp47Func)

517−

518−inlineLanguage :: PandocMonad m => (Text -> Lang) -> LP m Inlines

519−inlineLanguage bcp47Func = do

520− o <- option "" $ T.filter (\c -> c /= '[' && c /= ']')

521− <$> rawopt

522− let lang = renderLang $ bcp47Func o

523− extractSpaces (spanWith ("", [], [("lang", lang)])) <$> tok

524−

525478hyperlink :: PandocMonad m => LP m Inlines

526479hyperlink = try $ do

527480 src <- untokenize <$> braced

@@ -542,31 +495,6 @@ hypertargetInline = try $ do

542495 ils <- grouped inline

543496 return $ spanWith (ref, [], []) ils

544497

545−romanNumeralUpper :: (PandocMonad m) => LP m Inlines

546−romanNumeralUpper =

547− str . toRomanNumeral <$> romanNumeralArg

548−

549−romanNumeralLower :: (PandocMonad m) => LP m Inlines

550−romanNumeralLower =

551− str . T.toLower . toRomanNumeral <$> romanNumeralArg

552−

553−romanNumeralArg :: (PandocMonad m) => LP m Int

554−romanNumeralArg = spaces *> (parser <|> inBraces)

555− where

556− inBraces = do

557− symbol '{'

558− spaces

559− res <- parser

560− spaces

561− symbol '}'

562− return res

563− parser = do

564− s <- untokenize <$> many1 (satisfyTok isWordTok)

565− let (digits, rest) = T.span isDigit s

566− unless (T.null rest) $

567− Prelude.fail "Non-digits in argument to \\Rn or \\RN"

568− safeRead digits

569−

570498newToggle :: (Monoid a, PandocMonad m) => [Tok] -> LP m a

571499newToggle name = do

572500 updateState $ \st ->

src/Text/Pandoc/Readers/LaTeX/Inline.hs+62−3

@@ -13,6 +13,7 @@ module Text.Pandoc.Readers.LaTeX.Inline

1313 , verbCommands

1414 , charCommands

1515 , nameCommands

16+ , biblatexInlineCommands

1617 , refCommands

1718 , rawInlineOr

1819 , listingsLanguage

@@ -23,15 +24,17 @@ import qualified Data.Map as M

2324import Data.Text (Text)

2425import qualified Data.Text as T

2526import Text.Pandoc.Builder

27+import Text.Pandoc.Shared (toRomanNumeral, safeRead)

2628import Text.Pandoc.Readers.LaTeX.Types (Tok (..), TokType (..))

27−import Control.Applicative (optional)

28−import Control.Monad (guard, mzero, mplus)

29+import Control.Applicative (optional, (<|>))

30+import Control.Monad (guard, mzero, mplus, unless)

2931import Text.Pandoc.Class.PandocMonad (PandocMonad (..), translateTerm)

3032import Text.Pandoc.Readers.LaTeX.Parsing

3133import Text.Pandoc.Extensions (extensionEnabled, Extension(..))

3234import Text.Pandoc.Parsing (getOption, updateState, getState, notFollowedBy,

3335 manyTill, getInput, setInput, incSourceColumn,

34− option)

36+ option, many1)

37+import Data.Char (isDigit)

3538import Text.Pandoc.Highlighting (fromListingsLanguage,)

3639import Data.Maybe (maybeToList)

3740import Text.Pandoc.Options (ReaderOptions(..))

@@ -127,6 +130,31 @@ nlToSpace :: Char -> Char

127130nlToSpace '\n' = ' '

128131nlToSpace x = x

129132

133+romanNumeralUpper :: (PandocMonad m) => LP m Inlines

134+romanNumeralUpper =

135+ str . toRomanNumeral <$> romanNumeralArg

136+

137+romanNumeralLower :: (PandocMonad m) => LP m Inlines

138+romanNumeralLower =

139+ str . T.toLower . toRomanNumeral <$> romanNumeralArg

140+

141+romanNumeralArg :: (PandocMonad m) => LP m Int

142+romanNumeralArg = spaces *> (parser <|> inBraces)

143+ where

144+ inBraces = do

145+ symbol '{'

146+ spaces

147+ res <- parser

148+ spaces

149+ symbol '}'

150+ return res

151+ parser = do

152+ s <- untokenize <$> many1 (satisfyTok isWordTok)

153+ let (digits, rest) = T.span isDigit s

154+ unless (T.null rest) $

155+ Prelude.fail "Non-digits in argument to \\Rn or \\RN"

156+ safeRead digits

157+

130158

131159

132160verbCommands :: PandocMonad m => M.Map Text (LP m Inlines)

@@ -157,6 +185,12 @@ charCommands = M.fromList

157185 , ("{", lit "{")

158186 , ("}", lit "}")

159187 , ("qed", lit "\a0\x25FB")

188+ , ("lq", return (str "‘"))

189+ , ("rq", return (str "’"))

190+ , ("textquoteleft", return (str "‘"))

191+ , ("textquoteright", return (str "’"))

192+ , ("textquotedblleft", return (str "“"))

193+ , ("textquotedblright", return (str "”"))

160194 , ("/", pure mempty) -- italic correction

161195 , ("\\", linebreak <$ (do inTableCell <- sInTableCell <$> getState

162196 guard $ not inTableCell

@@ -185,6 +219,31 @@ charCommands = M.fromList

185219 , ("hyp", lit "-")

186220 ]

187221

222+biblatexInlineCommands :: PandocMonad m

223+ => LP m Inlines -> M.Map Text (LP m Inlines)

224+biblatexInlineCommands tok = M.fromList

225+ -- biblatex misc

226+ [ ("RN", romanNumeralUpper)

227+ , ("Rn", romanNumeralLower)

228+ , ("mkbibquote", spanWith nullAttr . doubleQuoted <$> tok)

229+ , ("mkbibemph", spanWith nullAttr . emph <$> tok)

230+ , ("mkbibitalic", spanWith nullAttr . emph <$> tok)

231+ , ("mkbibbold", spanWith nullAttr . strong <$> tok)

232+ , ("mkbibparens",

233+ spanWith nullAttr . (\x -> str "(" <> x <> str ")") <$> tok)

234+ , ("mkbibbrackets",

235+ spanWith nullAttr . (\x -> str "[" <> x <> str "]") <$> tok)

236+ , ("autocap", spanWith nullAttr <$> tok)

237+ , ("textnormal", spanWith ("",["nodecor"],[]) <$> tok)

238+ , ("bibstring",

239+ (\x -> spanWith ("",[],[("bibstring",x)]) (str x)) . untokenize

240+ <$> braced)

241+ , ("adddot", pure (str "."))

242+ , ("adddotspace", pure (spanWith nullAttr (str "." <> space)))

243+ , ("addabbrvspace", pure space)

244+ , ("hyphen", pure (str "-"))

245+ ]

246+

188247nameCommands :: PandocMonad m => M.Map Text (LP m Inlines)

189248nameCommands = M.fromList

190249 [ ("figurename", doTerm Translations.Figure)

src/Text/Pandoc/Readers/LaTeX/Lang.hs+29−1

@@ -15,15 +15,43 @@ module Text.Pandoc.Readers.LaTeX.Lang

1515 ( setDefaultLanguage

1616 , polyglossiaLangToBCP47

1717 , babelLangToBCP47

18+ , inlineLanguageCommands

1819 )

1920where

2021import qualified Data.Map as M

22+import Data.Text (Text)

2123import qualified Data.Text as T

24+import Text.Pandoc.Shared (extractSpaces)

2225import Text.Pandoc.BCP47 (Lang(..), renderLang)

2326import Text.Pandoc.Class (PandocMonad(..), setTranslations)

2427import Text.Pandoc.Readers.LaTeX.Parsing

2528import Text.Pandoc.Parsing (updateState, option)

26−import Text.Pandoc.Builder (Blocks, setMeta, str)

29+import Text.Pandoc.Builder (Blocks, Inlines, setMeta, str, spanWith)

30+

31+foreignlanguage :: PandocMonad m => LP m Inlines -> LP m Inlines

32+foreignlanguage tok = do

33+ babelLang <- untokenize <$> braced

34+ case babelLangToBCP47 babelLang of

35+ Just lang -> spanWith ("", [], [("lang", renderLang lang)]) <$> tok

36+ _ -> tok

37+

38+inlineLanguageCommands :: PandocMonad m

39+ => LP m Inlines -> M.Map Text (LP m Inlines)

40+inlineLanguageCommands tok =

41+ M.fromList $

42+ ("foreignlanguage", foreignlanguage tok) :

43+ (mk <$> M.toList polyglossiaLangToBCP47)

44+ where

45+ mk (polyglossia, bcp47Func) =

46+ ("text" <> polyglossia, inlineLanguage tok bcp47Func)

47+

48+inlineLanguage :: PandocMonad m

49+ => LP m Inlines -> (Text -> Lang) -> LP m Inlines

50+inlineLanguage tok bcp47Func = do

51+ o <- option "" $ T.filter (\c -> c /= '[' && c /= ']')

52+ <$> rawopt

53+ let lang = renderLang $ bcp47Func o

54+ extractSpaces (spanWith ("", [], [("lang", lang)])) <$> tok

2755

2856setDefaultLanguage :: PandocMonad m => LP m Blocks

2957setDefaultLanguage = do