8014809 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

LaTeX reader: Allow accents with combining characters.

8014809John MacFarlane13 years ago

src/Text/Pandoc/Readers/LaTeX.hs+185−185

compare with master

unifiedsplit

@@ -543,196 +543,196 @@ doLHSverb = codeWith ("",["haskell"],[]) <$> manyTill (satisfy (/='\n')) (char '

543lit :: String -> LP Inlines

544lit = pure . str

545

546−accent :: (Char -> Char) -> Inlines -> LP Inlines

546+accent :: (Char -> String) -> Inlines -> LP Inlines

547accent f ils =

548 case toList ils of

549− (Str (x:xs) : ys) -> return $ fromList $ (Str (f x : xs) : ys)

549+ (Str (x:xs) : ys) -> return $ fromList $ (Str (f x ++ xs) : ys)

550 [] -> mzero

551 _ -> return ils

552

553−grave :: Char -> Char

553+grave :: Char -> String

554−grave 'A' = 'À'

554+grave 'A' = "À"

555−grave 'E' = 'È'

555+grave 'E' = "È"

556−grave 'I' = 'Ì'

556+grave 'I' = "Ì"

557−grave 'O' = 'Ò'

557+grave 'O' = "Ò"

558−grave 'U' = 'Ù'

558+grave 'U' = "Ù"

559−grave 'a' = 'à'

559+grave 'a' = "à"

560−grave 'e' = 'è'

560+grave 'e' = "è"

561−grave 'i' = 'ì'

561+grave 'i' = "ì"

562−grave 'o' = 'ò'

562+grave 'o' = "ò"

563−grave 'u' = 'ù'

563+grave 'u' = "ù"

564−grave c = c

564+grave c = [c]

565−

565+

566−acute :: Char -> Char

566+acute :: Char -> String

567−acute 'A' = 'Á'

567+acute 'A' = "Á"

568−acute 'E' = 'É'

568+acute 'E' = "É"

569−acute 'I' = 'Í'

569+acute 'I' = "Í"

570−acute 'O' = 'Ó'

570+acute 'O' = "Ó"

571−acute 'U' = 'Ú'

571+acute 'U' = "Ú"

572−acute 'Y' = 'Ý'

572+acute 'Y' = "Ý"

573−acute 'a' = 'á'

573+acute 'a' = "á"

574−acute 'e' = 'é'

574+acute 'e' = "é"

575−acute 'i' = 'í'

575+acute 'i' = "í"

576−acute 'o' = 'ó'

576+acute 'o' = "ó"

577−acute 'u' = 'ú'

577+acute 'u' = "ú"

578−acute 'y' = 'ý'

578+acute 'y' = "ý"

579−acute 'C' = 'Ć'

579+acute 'C' = "Ć"

580−acute 'c' = 'ć'

580+acute 'c' = "ć"

581−acute 'L' = 'Ĺ'

581+acute 'L' = "Ĺ"

582−acute 'l' = 'ĺ'

582+acute 'l' = "ĺ"

583−acute 'N' = 'Ń'

583+acute 'N' = "Ń"

584−acute 'n' = 'ń'

584+acute 'n' = "ń"

585−acute 'R' = 'Ŕ'

585+acute 'R' = "Ŕ"

586−acute 'r' = 'ŕ'

586+acute 'r' = "ŕ"

587−acute 'S' = 'Ś'

587+acute 'S' = "Ś"

588−acute 's' = 'ś'

588+acute 's' = "ś"

589−acute 'Z' = 'Ź'

589+acute 'Z' = "Ź"

590−acute 'z' = 'ź'

590+acute 'z' = "ź"

591−acute c = c

591+acute c = [c]

592−

592+

593−circ :: Char -> Char

593+circ :: Char -> String

594−circ 'A' = 'Â'

594+circ 'A' = "Â"

595−circ 'E' = 'Ê'

595+circ 'E' = "Ê"

596−circ 'I' = 'Î'

596+circ 'I' = "Î"

597−circ 'O' = 'Ô'

597+circ 'O' = "Ô"

598−circ 'U' = 'Û'

598+circ 'U' = "Û"

599−circ 'a' = 'â'

599+circ 'a' = "â"

600−circ 'e' = 'ê'

600+circ 'e' = "ê"

601−circ 'i' = 'î'

601+circ 'i' = "î"

602−circ 'o' = 'ô'

602+circ 'o' = "ô"

603−circ 'u' = 'û'

603+circ 'u' = "û"

604−circ 'C' = 'Ĉ'

604+circ 'C' = "Ĉ"

605−circ 'c' = 'ĉ'

605+circ 'c' = "ĉ"

606−circ 'G' = 'Ĝ'

606+circ 'G' = "Ĝ"

607−circ 'g' = 'ĝ'

607+circ 'g' = "ĝ"

608−circ 'H' = 'Ĥ'

608+circ 'H' = "Ĥ"

609−circ 'h' = 'ĥ'

609+circ 'h' = "ĥ"

610−circ 'J' = 'Ĵ'

610+circ 'J' = "Ĵ"

611−circ 'j' = 'ĵ'

611+circ 'j' = "ĵ"

612−circ 'S' = 'Ŝ'

612+circ 'S' = "Ŝ"

613−circ 's' = 'ŝ'

613+circ 's' = "ŝ"

614−circ 'W' = 'Ŵ'

614+circ 'W' = "Ŵ"

615−circ 'w' = 'ŵ'

615+circ 'w' = "ŵ"

616−circ 'Y' = 'Ŷ'

616+circ 'Y' = "Ŷ"

617−circ 'y' = 'ŷ'

617+circ 'y' = "ŷ"

618−circ c = c

618+circ c = [c]

619−

619+

620−tilde :: Char -> Char

620+tilde :: Char -> String

621−tilde 'A' = 'Ã'

621+tilde 'A' = "Ã"

622−tilde 'a' = 'ã'

622+tilde 'a' = "ã"

623−tilde 'O' = 'Õ'

623+tilde 'O' = "Õ"

624−tilde 'o' = 'õ'

624+tilde 'o' = "õ"

625−tilde 'I' = 'Ĩ'

625+tilde 'I' = "Ĩ"

626−tilde 'i' = 'ĩ'

626+tilde 'i' = "ĩ"

627−tilde 'U' = 'Ũ'

627+tilde 'U' = "Ũ"

628−tilde 'u' = 'ũ'

628+tilde 'u' = "ũ"

629−tilde 'N' = 'Ñ'

629+tilde 'N' = "Ñ"

630−tilde 'n' = 'ñ'

630+tilde 'n' = "ñ"

631−tilde c = c

631+tilde c = [c]

632−

632+

633−umlaut :: Char -> Char

633+umlaut :: Char -> String

634−umlaut 'A' = 'Ä'

634+umlaut 'A' = "Ä"

635−umlaut 'E' = 'Ë'

635+umlaut 'E' = "Ë"

636−umlaut 'I' = 'Ï'

636+umlaut 'I' = "Ï"

637−umlaut 'O' = 'Ö'

637+umlaut 'O' = "Ö"

638−umlaut 'U' = 'Ü'

638+umlaut 'U' = "Ü"

639−umlaut 'a' = 'ä'

639+umlaut 'a' = "ä"

640−umlaut 'e' = 'ë'

640+umlaut 'e' = "ë"

641−umlaut 'i' = 'ï'

641+umlaut 'i' = "ï"

642−umlaut 'o' = 'ö'

642+umlaut 'o' = "ö"

643−umlaut 'u' = 'ü'

643+umlaut 'u' = "ü"

644−umlaut c = c

644+umlaut c = [c]

645−

645+

646−dot :: Char -> Char

646+dot :: Char -> String

647−dot 'C' = 'Ċ'

647+dot 'C' = "Ċ"

648−dot 'c' = 'ċ'

648+dot 'c' = "ċ"

649−dot 'E' = 'Ė'

649+dot 'E' = "Ė"

650−dot 'e' = 'ė'

650+dot 'e' = "ė"

651−dot 'G' = 'Ġ'

651+dot 'G' = "Ġ"

652−dot 'g' = 'ġ'

652+dot 'g' = "ġ"

653−dot 'I' = 'İ'

653+dot 'I' = "İ"

654−dot 'Z' = 'Ż'

654+dot 'Z' = "Ż"

655−dot 'z' = 'ż'

655+dot 'z' = "ż"

656−dot c = c

656+dot c = [c]

657−

657+

658−macron :: Char -> Char

658+macron :: Char -> String

659−macron 'A' = 'Ā'

659+macron 'A' = "Ā"

660−macron 'E' = 'Ē'

660+macron 'E' = "Ē"

661−macron 'I' = 'Ī'

661+macron 'I' = "Ī"

662−macron 'O' = 'Ō'

662+macron 'O' = "Ō"

663−macron 'U' = 'Ū'

663+macron 'U' = "Ū"

664−macron 'a' = 'ā'

664+macron 'a' = "ā"

665−macron 'e' = 'ē'

665+macron 'e' = "ē"

666−macron 'i' = 'ī'

666+macron 'i' = "ī"

667−macron 'o' = 'ō'

667+macron 'o' = "ō"

668−macron 'u' = 'ū'

668+macron 'u' = "ū"

669−macron c = c

669+macron c = [c]

670−

670+

671−cedilla :: Char -> Char

671+cedilla :: Char -> String

672−cedilla 'c' = 'ç'

672+cedilla 'c' = "ç"

673−cedilla 'C' = 'Ç'

673+cedilla 'C' = "Ç"

674−cedilla 's' = 'ş'

674+cedilla 's' = "ş"

675−cedilla 'S' = 'Ş'

675+cedilla 'S' = "Ş"

676−cedilla 't' = 'ţ'

676+cedilla 't' = "ţ"

677−cedilla 'T' = 'Ţ'

677+cedilla 'T' = "Ţ"

678−cedilla 'e' = 'ȩ'

678+cedilla 'e' = "ȩ"

679−cedilla 'E' = 'Ȩ'

679+cedilla 'E' = "Ȩ"

680−cedilla 'h' = 'ḩ'

680+cedilla 'h' = "ḩ"

681−cedilla 'H' = 'Ḩ'

681+cedilla 'H' = "Ḩ"

682−cedilla 'o' = 'o̧'

682+cedilla 'o' = "o̧"

683−cedilla 'O' = 'O̧'

683+cedilla 'O' = "O̧"

684−cedilla c = c

684+cedilla c = [c]

685−

685+

686−hacek :: Char -> Char

686+hacek :: Char -> String

687−hacek 'A' = 'Ǎ'

687+hacek 'A' = "Ǎ"

688−hacek 'a' = 'ǎ'

688+hacek 'a' = "ǎ"

689−hacek 'C' = 'Č'

689+hacek 'C' = "Č"

690−hacek 'c' = 'č'

690+hacek 'c' = "č"

691−hacek 'D' = 'Ď'

691+hacek 'D' = "Ď"

692−hacek 'd' = 'ď'

692+hacek 'd' = "ď"

693−hacek 'E' = 'Ě'

693+hacek 'E' = "Ě"

694−hacek 'e' = 'ě'

694+hacek 'e' = "ě"

695−hacek 'G' = 'Ǧ'

695+hacek 'G' = "Ǧ"

696−hacek 'g' = 'ǧ'

696+hacek 'g' = "ǧ"

697−hacek 'H' = 'Ȟ'

697+hacek 'H' = "Ȟ"

698−hacek 'h' = 'ȟ'

698+hacek 'h' = "ȟ"

699−hacek 'I' = 'Ǐ'

699+hacek 'I' = "Ǐ"

700−hacek 'i' = 'ǐ'

700+hacek 'i' = "ǐ"

701−hacek 'j' = 'ǰ'

701+hacek 'j' = "ǰ"

702−hacek 'K' = 'Ǩ'

702+hacek 'K' = "Ǩ"

703−hacek 'k' = 'ǩ'

703+hacek 'k' = "ǩ"

704−hacek 'L' = 'Ľ'

704+hacek 'L' = "Ľ"

705−hacek 'l' = 'ľ'

705+hacek 'l' = "ľ"

706−hacek 'N' = 'Ň'

706+hacek 'N' = "Ň"

707−hacek 'n' = 'ň'

707+hacek 'n' = "ň"

708−hacek 'O' = 'Ǒ'

708+hacek 'O' = "Ǒ"

709−hacek 'o' = 'ǒ'

709+hacek 'o' = "ǒ"

710−hacek 'R' = 'Ř'

710+hacek 'R' = "Ř"

711−hacek 'r' = 'ř'

711+hacek 'r' = "ř"

712−hacek 'S' = 'Š'

712+hacek 'S' = "Š"

713−hacek 's' = 'š'

713+hacek 's' = "š"

714−hacek 'T' = 'Ť'

714+hacek 'T' = "Ť"

715−hacek 't' = 'ť'

715+hacek 't' = "ť"

716−hacek 'U' = 'Ǔ'

716+hacek 'U' = "Ǔ"

717−hacek 'u' = 'ǔ'

717+hacek 'u' = "ǔ"

718−hacek 'Z' = 'Ž'

718+hacek 'Z' = "Ž"

719−hacek 'z' = 'ž'

719+hacek 'z' = "ž"

720−hacek c = c

720+hacek c = [c]

721−

721+

722−breve :: Char -> Char

722+breve :: Char -> String

723−breve 'A' = 'Ă'

723+breve 'A' = "Ă"

724−breve 'a' = 'ă'

724+breve 'a' = "ă"

725−breve 'E' = 'Ĕ'

725+breve 'E' = "Ĕ"

726−breve 'e' = 'ĕ'

726+breve 'e' = "ĕ"

727−breve 'G' = 'Ğ'

727+breve 'G' = "Ğ"

728−breve 'g' = 'ğ'

728+breve 'g' = "ğ"

729−breve 'I' = 'Ĭ'

729+breve 'I' = "Ĭ"

730−breve 'i' = 'ĭ'

730+breve 'i' = "ĭ"

731−breve 'O' = 'Ŏ'

731+breve 'O' = "Ŏ"

732−breve 'o' = 'ŏ'

732+breve 'o' = "ŏ"

733−breve 'U' = 'Ŭ'

733+breve 'U' = "Ŭ"

734−breve 'u' = 'ŭ'

734+breve 'u' = "ŭ"

735−breve c = c

735+breve c = [c]

736

737tok :: LP Inlines

738tok = try $ grouped inline <|> inlineCommand <|> str <$> (count 1 $ inlineChar)