1715d7c · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Org reader: Support more inline/display math variants

1715d7cAlbert Krewinkel13 years ago

src/Text/Pandoc/Readers/Org.hs+26−2

compare with master

unifiedsplit

@@ -512,6 +512,7 @@ inline =

512 , underline

513 , code

514 , math

515+ , displayMath

515 , verbatim

516 , verbatim

516 , subscript

517 , subscript

517 , superscript

518 , superscript

@@ -607,7 +608,15 @@ verbatim :: OrgParser Inlines

607verbatim = B.rawInline "" <$> verbatimBetween '~'

608verbatim = B.rawInline "" <$> verbatimBetween '~'

608

609

609math :: OrgParser Inlines

610math :: OrgParser Inlines

611+math = B.math <$> choice [ math1CharBetween '$'

610−math = B.math <$> mathStringBetween '$'

612+ , mathStringBetween '$'

613+ , rawMathBetween "\$" "\$"

614+ ]

615+

616+displayMath :: OrgParser Inlines

617+displayMath = B.displayMath <$> choice [ rawMathBetween "\\[" "\\]"

618+ , rawMathBetween "$$" "$$"

619+ ]

611

620

612subscript :: OrgParser Inlines

621subscript :: OrgParser Inlines

613subscript = B.subscript <$> (try $ char '_' *> maybeGroupedByBraces)

622subscript = B.subscript <$> (try $ char '_' *> maybeGroupedByBraces)

@@ -655,6 +664,21 @@ mathStringBetween c = try $ do

655 final <- mathEnd c

664 final <- mathEnd c

656 return $ body ++ [final]

665 return $ body ++ [final]

657

666

667+-- | Parse a single character between @c@ using math rules

668+math1CharBetween :: Char

669+ -> OrgParser String

670+math1CharBetween c = try $ do

671+ char c

672+ res <- noneOf $ c:mathForbiddenBorderChars

673+ char c

674+ eof <|> lookAhead (oneOf mathPostChars) *> return ()

675+ return [res]

676+

677+rawMathBetween :: String

678+ -> String

679+ -> OrgParser String

680+rawMathBetween s e = try $ string s *> manyTill anyChar (try $ string e)

681+

658-- | Parses the start (opening character) of emphasis

682-- | Parses the start (opening character) of emphasis

659emphasisStart :: Char -> OrgParser Char

683emphasisStart :: Char -> OrgParser Char

660emphasisStart c = try $ do

684emphasisStart c = try $ do

@@ -747,7 +771,7 @@ emphasisAllowedNewlines = 1

747

771

748-- | Chars allowed after an inline ($...$) math statement

772-- | Chars allowed after an inline ($...$) math statement

749mathPostChars :: [Char]

773mathPostChars :: [Char]

750−mathPostChars = "\t\n \"',-.:;?"

774+mathPostChars = "\t\n \"'),-.:;?"

751

775

752-- | Chars not allowed at the (inner) border of math

776-- | Chars not allowed at the (inner) border of math

753mathForbiddenBorderChars :: [Char]

777mathForbiddenBorderChars :: [Char]

tests/Tests/Readers/Org.hs+30−6

@@ -54,14 +54,26 @@ tests =

54 "=Robot.rock()=" =?>

55 para (code "Robot.rock()")

57− , "Math" =:

58− "$E=mc^2$" =?>

59− para (math "E=mc^2")

60−

61 , "Verbatim" =:

57 , "Verbatim" =:

62 "~word for word~" =?>

58 "~word for word~" =?>

63 para (rawInline "" "word for word")

59 para (rawInline "" "word for word")

61+ , "Math $..$" =:

62+ "$E=mc^2$" =?>

63+ para (math "E=mc^2")

64+

65+ , "Math $$..$$" =:

66+ "$$E=mc^2$$" =?>

67+ para (displayMath "E=mc^2")

68+

69+ , "Math \\[..\\]" =:

70+ "\\[E=ℎν\\]" =?>

71+ para (displayMath "E=ℎν")

72+

73+ , "Math \$..\$" =:

74+ "\$σ_x σ_p ≥ \\frac{ℏ}{2}\$" =?>

75+ para (math "σ_x σ_p ≥ \\frac{ℏ}{2}")

76+

65 , "Symbol" =:

77 , "Symbol" =:

66 "A * symbol" =?>

78 "A * symbol" =?>

67 para (str "A" <> space <> str "*" <> space <> "symbol")

79 para (str "A" <> space <> str "*" <> space <> "symbol")

@@ -86,14 +98,19 @@ tests =

86 unlines [ "this+that+ +so+on"

98 unlines [ "this+that+ +so+on"

87 , "seven*eight* nine*"

99 , "seven*eight* nine*"

88 , "+not+funny+"

100 , "+not+funny+"

89− , "this == self"

90 ] =?>

101 ] =?>

91 para (spcSep [ "this+that+", "+so+on"

102 para (spcSep [ "this+that+", "+so+on"

92 , "seven*eight*", "nine*"

103 , "seven*eight*", "nine*"

93 , strikeout "not+funny"

104 , strikeout "not+funny"

94− , "this" <> space <> "==" <> space <> "self"

95 ])

105 ])

106

107+ , "No empty markup" =:

108+ -- FIXME: __ is erroneously parsed as subscript "_"

109+ -- "// ** __ ++ == ~~ $$" =?>

110+ -- para (spcSep [ "//", "**", "__", "++", "==", "~~", "$$" ])

111+ "// ** ++ == ~~ $$" =?>

112+ para (spcSep [ "//", "**", "++", "==", "~~", "$$" ])

113+

97 , "Adherence to Org's rules for markup borders" =:

114 , "Adherence to Org's rules for markup borders" =:

98 "/t/& a/ / ./r/ (*l*) /e/! /b/." =?>

115 "/t/& a/ / ./r/ (*l*) /e/! /b/." =?>

99 para (spcSep [ emph $ "t/&" <> space <> "a"

116 para (spcSep [ emph $ "t/&" <> space <> "a"

@@ -109,6 +126,13 @@ tests =

109 para ((math "a\nb\nc") <> space <>

126 para ((math "a\nb\nc") <> space <>

110 spcSep [ "$d", "e", "f", "g$" ])

127 spcSep [ "$d", "e", "f", "g$" ])

111

128

129+ , "Single-character math" =:

130+ "$a$ $b$! $c$?" =?>

131+ para (spcSep [ math "a"

132+ , "$b$!"

133+ , (math "c") <> "?"

134+ ])

135+

112 , "Markup may not span more than two lines" =:

136 , "Markup may not span more than two lines" =:

113 unlines [ "/this *is +totally", "nice+ not*", "emph/" ] =?>

137 unlines [ "/this *is +totally", "nice+ not*", "emph/" ] =?>

114 para (spcSep [ "/this"

138 para (spcSep [ "/this"