dc69255 · typeable/pandoc

src/Text/Pandoc/Entities.hs−18

compare with master

unifiedsplit

@@ -34,7 +34,6 @@ module Text.Pandoc.Entities (

34 encodeEntities,

35 decodeEntities,

36 escapeSGMLChar,

37− stringToSGML,

38 characterEntity

37 characterEntity

39 ) where

38 ) where

40import Data.Char ( chr, ord )

39import Data.Char ( chr, ord )

@@ -115,23 +114,6 @@ decodeEntities str =

115 Left err -> error $ "\nError: " ++ show err

114 Left err -> error $ "\nError: " ++ show err

116 Right result -> result

115 Right result -> result

117

116

118−-- | Escape string for SGML, preserving entity references.

119−stringToSGML :: String -> String

120−stringToSGML str =

121− let regular = do

122− str <- many1 (satisfy (not . needsEscaping))

123− return str

124− special = do

125− notFollowedBy characterEntity

126− c <- anyChar

127− return $ escapeSGMLChar c

128− entity = do

129− ent <- manyTill anyChar (char ';')

130− return (ent ++ ";") in

131− case parse (many (regular <|> special <|> entity)) str str of

132− Left err -> error $ "\nError: " ++ show err

133− Right result -> concat result

134−

135entityTable :: [(String, Char)]

117entityTable :: [(String, Char)]

136entityTable = [

118entityTable = [

137 (""", chr 34),

119 (""", chr 34),

src/Text/Pandoc/Readers/HTML.hs+3−1

@@ -445,7 +445,9 @@ extractAttribute name [] = Nothing

445extractAttribute name ((attrName, contents):rest) =

446 let name' = map toLower name

447 attrName' = map toLower attrName in

448+ if (attrName' == name')

449+ then Just (decodeEntities contents)

448− if (attrName' == name') then Just contents else extractAttribute name rest

450+ else extractAttribute name rest

449

451

450link = try (do

452link = try (do

451 (tag, attributes) <- htmlTag "a"

453 (tag, attributes) <- htmlTag "a"

src/Text/Pandoc/Readers/Markdown.hs+4−4

@@ -42,7 +42,7 @@ import Text.Pandoc.Readers.HTML ( rawHtmlBlock,

42 anyHtmlTag, anyHtmlEndTag,

43 htmlEndTag, extractTagType,

44 htmlBlockElement )

45−import Text.Pandoc.Entities ( characterEntity )

45+import Text.Pandoc.Entities ( characterEntity, decodeEntities )

46import Text.ParserCombinators.Parsec

47

48-- | Read markdown from an input string and return a Pandoc document.

@@ -144,14 +144,14 @@ authorsLine = try (do

144 skipSpaces

145 authors <- sepEndBy (many1 (noneOf ",;\n")) (oneOf ",;")

146 newline

147− return (map removeLeadingTrailingSpace authors))

147+ return (map (decodeEntities . removeLeadingTrailingSpace) authors))

148

149dateLine = try (do

150 char '%'

151 skipSpaces

152 date <- many (noneOf "\n")

153 newline

154− return (removeTrailingSpace date))

154+ return (decodeEntities $ removeTrailingSpace date))

155

156titleBlock = try (do

157 failIfStrict

@@ -894,7 +894,7 @@ titleWith startChar endChar = try (do

894 char endChar

895 skipSpaces

896 notFollowedBy (noneOf ")\n")))

897− return tit)

897+ return $ decodeEntities tit)

898

899title = choice [ titleWith '(' ')',

900 titleWith '"' '"',

src/Text/Pandoc/Shared.hs+3−3

@@ -72,7 +72,7 @@ module Text.Pandoc.Shared (

72 ) where

73import Text.Pandoc.Definition

74import Text.ParserCombinators.Parsec as Parsec

75−import Text.Pandoc.Entities ( decodeEntities, encodeEntities, stringToSGML )

75+import Text.Pandoc.Entities ( decodeEntities, encodeEntities )

76import Text.PrettyPrint.HughesPJ as PP ( text, char, (<>),

77 ($$), nest, Doc, isEmpty )

78import Data.Char ( toLower, ord )

@@ -524,8 +524,8 @@ replaceRefLinksInline keytable other = other

524-- | Return a text object with a string of formatted SGML attributes.

525attributeList :: [(String, String)] -> Doc

526attributeList = text . concatMap

527− (\(a, b) -> " " ++ stringToSGML a ++ "=\"" ++

527+ (\(a, b) -> " " ++ encodeEntities a ++ "=\"" ++

528− stringToSGML b ++ "\"")

528+ encodeEntities b ++ "\"")

529

530-- | Put the supplied contents between start and end tags of tagType,

531-- with specified attributes and (if specified) indentation.

src/Text/Pandoc/Writers/Docbook.hs+7−7

@@ -32,7 +32,7 @@ module Text.Pandoc.Writers.Docbook (

32 ) where

33import Text.Pandoc.Definition

34import Text.Pandoc.Shared

35−import Text.Pandoc.Entities ( encodeEntities, stringToSGML )

35+import Text.Pandoc.Entities ( encodeEntities )

36import Data.Char ( toLower, ord )

37import Data.List ( isPrefixOf, partition, drop )

38import Text.PrettyPrint.HughesPJ hiding ( Str )

@@ -64,8 +64,8 @@ authorToDocbook name = inTagsIndented "author" $

64 then -- last name first

65 let (lastname, rest) = break (==',') name

66 firstname = removeLeadingSpace rest in

67− inTagsSimple "firstname" (text $ stringToSGML firstname) <>

67+ inTagsSimple "firstname" (text $ encodeEntities firstname) <>

68− inTagsSimple "surname" (text $ stringToSGML lastname)

68+ inTagsSimple "surname" (text $ encodeEntities lastname)

69 else -- last name last

70 let namewords = words name

71 lengthname = length namewords

@@ -73,8 +73,8 @@ authorToDocbook name = inTagsIndented "author" $

73 0 -> ("","")

74 1 -> ("", name)

75 n -> (joinWithSep " " (take (n-1) namewords), last namewords) in

76− inTagsSimple "firstname" (text $ stringToSGML firstname) $$

76+ inTagsSimple "firstname" (text $ encodeEntities firstname) $$

77− inTagsSimple "surname" (text $ stringToSGML lastname)

77+ inTagsSimple "surname" (text $ encodeEntities lastname)

78

79-- | Convert Pandoc document to string in Docbook format.

80writeDocbook :: WriterOptions -> Pandoc -> String

@@ -86,7 +86,7 @@ writeDocbook opts (Pandoc (Meta title authors date) blocks) =

86 then inTagsIndented "articleinfo" $

87 (inTagsSimple "title" (wrap opts title)) $$

88 (vcat (map authorToDocbook authors)) $$

89− (inTagsSimple "date" (text $ stringToSGML date))

89+ (inTagsSimple "date" (text $ encodeEntities date))

90 else empty

91 blocks' = replaceReferenceLinks blocks

92 (noteBlocks, blocks'') = partition isNoteBlock blocks'

@@ -227,7 +227,7 @@ inlineToDocbook opts (Image alt (Src src tit)) =

227 then empty

228 else inTagsIndented "objectinfo" $

229 inTagsIndented "title"

230− (text $ stringToSGML tit) in

230+ (text $ encodeEntities tit) in

231 inTagsIndented "inlinemediaobject" $

232 inTagsIndented "imageobject" $

233 titleDoc $$ selfClosingTag "imagedata" [("fileref", src)]

src/Text/Pandoc/Writers/HTML.hs+8−10

@@ -32,7 +32,7 @@ module Text.Pandoc.Writers.HTML (

32 ) where

33import Text.Pandoc.Definition

34import Text.Pandoc.Shared

35−import Text.Pandoc.Entities ( encodeEntities, stringToSGML )

35+import Text.Pandoc.Entities ( encodeEntities )

36import Text.Regex ( mkRegex, matchRegex )

37import Numeric ( showHex )

38import Data.Char ( ord, toLower )

@@ -127,11 +127,11 @@ htmlHeader opts (Meta title authors date) =

127 then empty

128 else selfClosingTag "meta" [("name", "author"),

129 ("content",

130− joinWithSep ", " (map stringToSGML authors))]

130+ joinWithSep ", " (map encodeEntities authors))]

131 datetext = if (date == "")

132 then empty

133 else selfClosingTag "meta" [("name", "date"),

134− ("content", stringToSGML date)] in

134+ ("content", encodeEntities date)] in

135 text (writerHeader opts) $$ authortext $$ datetext $$ titletext $$

136 text "</head>\n<body>"

137

@@ -248,20 +248,18 @@ inlineToHtml opts (TeX str) = text $ encodeEntities str

248inlineToHtml opts (HtmlInline str) = text str

249inlineToHtml opts (LineBreak) = selfClosingTag "br" []

250inlineToHtml opts Space = space

251−inlineToHtml opts (Link txt (Src src tit)) =

251+inlineToHtml opts (Link txt (Src src title)) =

252− let title = stringToSGML tit in

253 if (isPrefixOf "mailto:" src)

252 if (isPrefixOf "mailto:" src)

254 then obfuscateLink opts txt src

253 then obfuscateLink opts txt src

255− else inTags False "a" ([("href", encodeEntities src)] ++

254+ else inTags False "a" ([("href", src)] ++

256− if null tit then [] else [("title", title)])

255+ if null title then [] else [("title", title)])

257 (inlineListToHtml opts txt)

256 (inlineListToHtml opts txt)

258inlineToHtml opts (Link txt (Ref ref)) =

257inlineToHtml opts (Link txt (Ref ref)) =

259 char '[' <> (inlineListToHtml opts txt) <> text "][" <>

258 char '[' <> (inlineListToHtml opts txt) <> text "][" <>

260 (inlineListToHtml opts ref) <> char ']'

259 (inlineListToHtml opts ref) <> char ']'

261 -- this is what markdown does, for better or worse

260 -- this is what markdown does, for better or worse

262−inlineToHtml opts (Image alt (Src source tit)) =

261+inlineToHtml opts (Image alt (Src source title)) =

263− let title = stringToSGML tit

264− alternate = render $ inlineListToHtml opts alt in

262+ let alternate = render $ inlineListToHtml opts alt in

265 selfClosingTag "img" $ [("src", source)] ++

263 selfClosingTag "img" $ [("src", source)] ++

266 (if null alternate then [] else [("alt", alternate)]) ++

264 (if null alternate then [] else [("alt", alternate)]) ++

267 [("title", title)] -- note: null title is included, as in Markdown.pl

265 [("title", title)] -- note: null title is included, as in Markdown.pl