925815b · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Split off T.P.Readers.LaTeX.Accent.

925815bJohn MacFarlane6 years ago

pandoc.cabal+1

compare with master

unifiedsplit

@@ -630,6 +630,7 @@ library

630 Text.Pandoc.Readers.LaTeX.Parsing,

631 Text.Pandoc.Readers.LaTeX.Lang,

632 Text.Pandoc.Readers.LaTeX.SIunitx,

633+ Text.Pandoc.Readers.LaTeX.Accent,

633 Text.Pandoc.Readers.Odt.Base,

634 Text.Pandoc.Readers.Odt.Base,

634 Text.Pandoc.Readers.Odt.Namespaces,

635 Text.Pandoc.Readers.Odt.Namespaces,

635 Text.Pandoc.Readers.Odt.StyleReader,

636 Text.Pandoc.Readers.Odt.StyleReader,

src/Text/Pandoc/Readers/LaTeX.hs+8−60

@@ -57,6 +57,7 @@ import Text.Pandoc.Parsing hiding (blankline, many, mathDisplay, mathInline,

57import Text.Pandoc.Readers.LaTeX.Types (ExpansionPoint (..), Macro (..),

58 ArgSpec (..), Tok (..), TokType (..))

59import Text.Pandoc.Readers.LaTeX.Parsing

60+import Text.Pandoc.Readers.LaTeX.Accent (accentCommands)

60import Text.Pandoc.Readers.LaTeX.Lang (polyglossiaLangToBCP47,

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

61 babelLangToBCP47)

62 babelLangToBCP47)

62import Text.Pandoc.Readers.LaTeX.SIunitx

63import Text.Pandoc.Readers.LaTeX.SIunitx

@@ -64,7 +65,6 @@ import Text.Pandoc.Shared

64import qualified Text.Pandoc.Translations as Translations

65import qualified Text.Pandoc.Translations as Translations

65import Text.Pandoc.Walk

66import Text.Pandoc.Walk

66import qualified Text.Pandoc.Builder as B

67import qualified Text.Pandoc.Builder as B

67−import qualified Data.Text.Normalize as Normalize

68import Safe

70-- for debugging:

@@ -247,9 +247,6 @@ doxspace =

247 startsWithLetter _ = False

248

249

250−lit :: Text -> LP m Inlines

251−lit = pure . str

252−

253removeDoubleQuotes :: Text -> Text

250removeDoubleQuotes :: Text -> Text

254removeDoubleQuotes t =

251removeDoubleQuotes t =

255 Data.Maybe.fromMaybe t $ T.stripPrefix "\"" t >>= T.stripSuffix "\""

252 Data.Maybe.fromMaybe t $ T.stripPrefix "\"" t >>= T.stripSuffix "\""

@@ -296,6 +293,9 @@ quoted' f starter ender = do

296 cs -> cs)

293 cs -> cs)

297 else lit startchs

294 else lit startchs

298

295

296+lit :: Text -> LP m Inlines

297+lit = pure . str

298+

299enquote :: PandocMonad m => Bool -> Maybe Text -> LP m Inlines

300enquote starred mblang = do

301 skipopts

@@ -631,7 +631,10 @@ inlineEnvironments = M.fromList [

631 ]

632

633inlineCommands :: PandocMonad m => M.Map Text (LP m Inlines)

634+inlineCommands =

634−inlineCommands = M.union inlineLanguageCommands $ M.fromList

635+ M.union inlineLanguageCommands $

636+ M.union (accentCommands tok) $

637+ M.fromList

635 [ ("emph", extractSpaces emph <$> tok)

638 [ ("emph", extractSpaces emph <$> tok)

636 , ("textit", extractSpaces emph <$> tok)

639 , ("textit", extractSpaces emph <$> tok)

637 , ("textsl", extractSpaces emph <$> tok)

640 , ("textsl", extractSpaces emph <$> tok)

@@ -703,48 +706,6 @@ inlineCommands = M.union inlineLanguageCommands $ M.fromList

703 , ("MakeTextLowercase", makeLowercase <$> tok)

706 , ("MakeTextLowercase", makeLowercase <$> tok)

704 , ("lowercase", makeLowercase <$> tok)

707 , ("lowercase", makeLowercase <$> tok)

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

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

706− , ("aa", lit "å")

707− , ("AA", lit "Å")

708− , ("ss", lit "ß")

709− , ("o", lit "ø")

710− , ("O", lit "Ø")

711− , ("L", lit "Ł")

712− , ("l", lit "ł")

713− , ("ae", lit "æ")

714− , ("AE", lit "Æ")

715− , ("oe", lit "œ")

716− , ("OE", lit "Œ")

717− , ("pounds", lit "£")

718− , ("euro", lit "€")

719− , ("copyright", lit "©")

720− , ("textasciicircum", lit "^")

721− , ("textasciitilde", lit "~")

722− , ("H", accent '\779' Nothing) -- hungarumlaut

723− , ("`", accent '\768' (Just '`')) -- grave

724− , ("'", accent '\769' (Just '\'')) -- acute

725− , ("^", accent '\770' (Just '^')) -- circ

726− , ("~", accent '\771' (Just '~')) -- tilde

727− , ("\"", accent '\776' Nothing) -- umlaut

728− , (".", accent '\775' Nothing) -- dot

729− , ("=", accent '\772' Nothing) -- macron

730− , ("|", accent '\781' Nothing) -- vertical line above

731− , ("b", accent '\817' Nothing) -- macron below

732− , ("c", accent '\807' Nothing) -- cedilla

733− , ("G", accent '\783' Nothing) -- doublegrave

734− , ("h", accent '\777' Nothing) -- hookabove

735− , ("d", accent '\803' Nothing) -- dotbelow

736− , ("f", accent '\785' Nothing) -- inverted breve

737− , ("r", accent '\778' Nothing) -- ringabove

738− , ("t", accent '\865' Nothing) -- double inverted breve

739− , ("U", accent '\782' Nothing) -- double vertical line above

740− , ("v", accent '\780' Nothing) -- hacek

741− , ("u", accent '\774' Nothing) -- breve

742− , ("k", accent '\808' Nothing) -- ogonek

743− , ("textogonekcentered", accent '\808' Nothing) -- ogonek

744− , ("i", lit "ı") -- dotless i

745− , ("j", lit "ȷ") -- dotless j

746− , ("newtie", accent '\785' Nothing) -- inverted breve

747− , ("textcircled", accent '\8413' Nothing) -- combining circle

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

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

749 guard $ not inTableCell

710 guard $ not inTableCell

750 optional opt

711 optional opt

@@ -960,19 +921,6 @@ inlineCommands = M.union inlineLanguageCommands $ M.fromList

960 , ("hyphen", pure (str "-"))

921 , ("hyphen", pure (str "-"))

961 ]

922 ]

962

923

963−accent :: PandocMonad m => Char -> Maybe Char -> LP m Inlines

964−accent combiningAccent fallBack = try $ do

965− ils <- tok

966− case toList ils of

967− (Str (T.uncons -> Just (x, xs)) : ys) -> return $ fromList $

968− -- try to normalize to the combined character:

969− Str (Normalize.normalize Normalize.NFC

970− (T.pack [x, combiningAccent]) <> xs) : ys

971− [Space] -> return $ str $ T.singleton $ fromMaybe combiningAccent fallBack

972− [] -> return $ str $ T.singleton $ fromMaybe combiningAccent fallBack

973− _ -> return ils

974−

975−

976lettrine :: PandocMonad m => LP m Inlines

924lettrine :: PandocMonad m => LP m Inlines

977lettrine = do

925lettrine = do

978 optional opt

926 optional opt

src/Text/Pandoc/Readers/LaTeX/Accent.hsnew file+78

@@ -0,0 +1,78 @@

1+{-# LANGUAGE ViewPatterns #-}

2+{-# LANGUAGE OverloadedStrings #-}

3+module Text.Pandoc.Readers.LaTeX.Accent

4+ ( accentCommands )

5+where

7+import Text.Pandoc.Class

8+import Text.Pandoc.Readers.LaTeX.Parsing

9+import Text.Pandoc.Builder as B

10+import qualified Data.Map as M

11+import Data.Text (Text)

12+import Data.Maybe (fromMaybe)

13+import Text.Pandoc.Parsing

14+import qualified Data.Text as T

15+import qualified Data.Text.Normalize as Normalize

16+

17+accentCommands :: PandocMonad m => LP m Inlines -> M.Map Text (LP m Inlines)

18+accentCommands tok =

19+ let accent = accentWith tok

20+ lit = pure . str

21+ in M.fromList

22+ [ ("aa", lit "å")

23+ , ("AA", lit "Å")

24+ , ("ss", lit "ß")

25+ , ("o", lit "ø")

26+ , ("O", lit "Ø")

27+ , ("L", lit "Ł")

28+ , ("l", lit "ł")

29+ , ("ae", lit "æ")

30+ , ("AE", lit "Æ")

31+ , ("oe", lit "œ")

32+ , ("OE", lit "Œ")

33+ , ("pounds", lit "£")

34+ , ("euro", lit "€")

36+ , ("textasciicircum", lit "^")

37+ , ("textasciitilde", lit "~")

38+ , ("H", accent '\779' Nothing) -- hungarumlaut

39+ , ("`", accent '\768' (Just '`')) -- grave

40+ , ("'", accent '\769' (Just '\'')) -- acute

41+ , ("^", accent '\770' (Just '^')) -- circ

42+ , ("~", accent '\771' (Just '~')) -- tilde

43+ , ("\"", accent '\776' Nothing) -- umlaut

44+ , (".", accent '\775' Nothing) -- dot

45+ , ("=", accent '\772' Nothing) -- macron

46+ , ("|", accent '\781' Nothing) -- vertical line above

47+ , ("b", accent '\817' Nothing) -- macron below

48+ , ("c", accent '\807' Nothing) -- cedilla

49+ , ("G", accent '\783' Nothing) -- doublegrave

50+ , ("h", accent '\777' Nothing) -- hookabove

51+ , ("d", accent '\803' Nothing) -- dotbelow

52+ , ("f", accent '\785' Nothing) -- inverted breve

53+ , ("r", accent '\778' Nothing) -- ringabove

54+ , ("t", accent '\865' Nothing) -- double inverted breve

55+ , ("U", accent '\782' Nothing) -- double vertical line above

56+ , ("v", accent '\780' Nothing) -- hacek

57+ , ("u", accent '\774' Nothing) -- breve

58+ , ("k", accent '\808' Nothing) -- ogonek

59+ , ("textogonekcentered", accent '\808' Nothing) -- ogonek

60+ , ("i", lit "ı") -- dotless i

61+ , ("j", lit "ȷ") -- dotless j

62+ , ("newtie", accent '\785' Nothing) -- inverted breve

63+ , ("textcircled", accent '\8413' Nothing) -- combining circle

64+ ]

65+

66+accentWith :: PandocMonad m

67+ => LP m Inlines -> Char -> Maybe Char -> LP m Inlines

68+accentWith tok combiningAccent fallBack = try $ do

69+ ils <- tok

70+ case toList ils of

71+ (Str (T.uncons -> Just (x, xs)) : ys) -> return $ fromList $

72+ -- try to normalize to the combined character:

73+ Str (Normalize.normalize Normalize.NFC

74+ (T.pack [x, combiningAccent]) <> xs) : ys

75+ [Space] -> return $ str $ T.singleton $ fromMaybe combiningAccent fallBack

76+ [] -> return $ str $ T.singleton $ fromMaybe combiningAccent fallBack

77+ _ -> return ils

78+