c87d522 · typeable/pandoc

git.iko.soy/typeable/pandoc

Properly escape URIs in all readers.

commitc87d52223ad97ae7f21c69b09ec7b1467bd35670 authorJohn MacFarlane <jgm@berkeley.edu> dateTue 23 Mar 2010, 15:07 (17 years ago) parentc3384bb

4 files changed+37−44

compare with master

unifiedsplit

src/Text/Pandoc/Readers/HTML.hs+3−3 src/Text/Pandoc/Readers/LaTeX.hs+6−7 src/Text/Pandoc/Readers/Markdown.hs+9−16 src/Text/Pandoc/Readers/RST.hs+19−18

src/Text/Pandoc/Readers/HTML.hs+3−3

@@ -182,7 +182,7 @@ unsanitaryURI u =

182182 "ldaps:", "magnet:", "mms:", "msnim:", "notes:", "rsync:",

183183 "secondlife:", "skype:", "ssh:", "sftp:", "smb:", "sms:",

184184 "snews:", "webcal:", "ymsgr:"]

185− in case parseURIReference (stringToURI u) of

185+ in case parseURIReference (escapeURI u) of

186186 Just p -> (map toLower $ uriScheme p) `notElem` safeURISchemes

187187 Nothing -> True

188188

@@ -746,7 +746,7 @@ link = try $ do

746746 Nothing -> fail "no href"

747747 let title = fromMaybe "" $ extractAttribute "title" attributes

748748 lab <- inlinesTilEnd "a"

749− return $ Link (normalizeSpaces lab) (url, title)

749+ return $ Link (normalizeSpaces lab) (escapeURI url, title)

750750

751751image :: GenParser Char ParserState Inline

752752image = try $ do

@@ -756,5 +756,5 @@ image = try $ do

756756 Nothing -> fail "no src"

757757 let title = fromMaybe "" $ extractAttribute "title" attributes

758758 let alt = fromMaybe "" (extractAttribute "alt" attributes)

759− return $ Image [Str alt] (url, title)

759+ return $ Image [Str alt] (escapeURI url, title)

760760

src/Text/Pandoc/Readers/LaTeX.hs+6−7

@@ -738,7 +738,7 @@ url :: GenParser Char ParserState Inline

738738url = try $ do

739739 string "\\url"

740740 url' <- charsInBalanced '{' '}'

741− return $ Link [Code url'] (url', "")

741+ return $ Link [Code url'] (escapeURI url', "")

742742

743743link :: GenParser Char ParserState Inline

744744link = try $ do

@@ -746,17 +746,16 @@ link = try $ do

746746 url' <- manyTill anyChar (char '}')

747747 char '{'

748748 label' <- manyTill inline (char '}')

749− return $ Link (normalizeSpaces label') (url', "")

749+ return $ Link (normalizeSpaces label') (escapeURI url', "")

750750

751751image :: GenParser Char ParserState Inline

752752image = try $ do

753753 ("includegraphics", _, args) <- command

754754 let args' = filter isArg args -- filter out options

755− let src = if null args' then

756− ("", "")

757− else

758− (stripFirstAndLast (head args'), "")

759− return $ Image [Str "image"] src

755+ let (src,tit) = case args' of

756+ [] -> ("", "")

757+ (x:_) -> (stripFirstAndLast x, "")

758+ return $ Image [Str "image"] (escapeURI src, tit)

760759

761760footnote :: GenParser Char ParserState Inline

762761footnote = try $ do

src/Text/Pandoc/Readers/Markdown.hs+9−16

@@ -31,7 +31,7 @@ module Text.Pandoc.Readers.Markdown (

3131 readMarkdown

3232 ) where

3333

34−import Data.List ( transpose, isPrefixOf, isSuffixOf, sortBy, findIndex, intercalate )

34+import Data.List ( transpose, isSuffixOf, sortBy, findIndex, intercalate )

3535import Data.Ord ( comparing )

3636import Data.Char ( isAlphaNum )

3737import Data.Maybe

@@ -73,10 +73,6 @@ specialChars = "\\[]*_~`<>$!^-.&'\"\8216\8217\8220\8221;"

7373-- auxiliary functions

7474--

7575

76−-- | Replace spaces with %20

77−uriEscapeSpaces :: String -> String

78−uriEscapeSpaces = substitute " " "%20"

79−

8076indentSpaces :: GenParser Char ParserState [Char]

8177indentSpaces = try $ do

8278 state <- getState

@@ -206,7 +202,7 @@ referenceKey = try $ do

206202 tit <- option "" referenceTitle

207203 blanklines

208204 endPos <- getPosition

209− let newkey = (lab, (uriEscapeSpaces $ removeTrailingSpace src, tit))

205+ let newkey = (lab, (escapeURI $ removeTrailingSpace src, tit))

210206 st <- getState

211207 let oldkeys = stateKeys st

212208 updateState $ \s -> s { stateKeys = newkey : oldkeys }

@@ -1194,7 +1190,7 @@ source' = do

11941190 tit <- option "" linkTitle

11951191 skipSpaces

11961192 eof

1197− return (uriEscapeSpaces $ removeTrailingSpace src, tit)

1193+ return (escapeURI $ removeTrailingSpace src, tit)

11981194

11991195linkTitle :: GenParser Char st String

12001196linkTitle = try $ do

@@ -1208,11 +1204,11 @@ linkTitle = try $ do

12081204link :: GenParser Char ParserState Inline

12091205link = try $ do

12101206 lab <- reference

1211− src <- source <|> referenceLink lab

1207+ (src, tit) <- source <|> referenceLink lab

12121208 sanitize <- getState >>= return . stateSanitizeHTML

1213− if sanitize && unsanitaryURI (fst src)

1209+ if sanitize && unsanitaryURI src

12141210 then fail "Unsanitary URI"

1215− else return $ Link lab src

1211+ else return $ Link lab (src, tit)

12161212

12171213-- a link like [this][ref] or [this][] or [this]

12181214referenceLink :: [Inline]

@@ -1229,18 +1225,15 @@ referenceLink lab = do

12291225autoLink :: GenParser Char ParserState Inline

12301226autoLink = try $ do

12311227 char '<'

1232− src <- uri <|> (emailAddress >>= (return . ("mailto:" ++)))

1228+ (orig, src) <- uri <|> emailAddress

12331229 char '>'

1234− let src' = if "mailto:" `isPrefixOf` src

1235− then drop 7 src

1236− else src

12371230 st <- getState

12381231 let sanitize = stateSanitizeHTML st

12391232 if sanitize && unsanitaryURI src

12401233 then fail "Unsanitary URI"

12411234 else return $ if stateStrict st

1242− then Link [Str src'] (src, "")

1243− else Link [Code src'] (src, "")

1235+ then Link [Str orig] (src, "")

1236+ else Link [Code orig] (src, "")

12441237

12451238image :: GenParser Char ParserState Inline

12461239image = try $ do

src/Text/Pandoc/Readers/RST.hs+19−18

@@ -556,7 +556,7 @@ targetURI = do

556556 contents <- many1 (try (many spaceChar >> newline >>

557557 many1 spaceChar >> noneOf " \t\n") <|> noneOf "\n")

558558 blanklines

559− return contents

559+ return $ escapeURI $ removeLeadingTrailingSpace $ contents

560560

561561imageKey :: GenParser Char ParserState ([Inline], (String, [Char]))

562562imageKey = try $ do

@@ -565,13 +565,13 @@ imageKey = try $ do

565565 skipSpaces

566566 string "image::"

567567 src <- targetURI

568− return (normalizeSpaces ref, (removeLeadingTrailingSpace src, ""))

568+ return (normalizeSpaces ref, (src, ""))

569569

570570anonymousKey :: GenParser Char st ([Inline], (String, [Char]))

571571anonymousKey = try $ do

572572 oneOfStrings [".. __:", "__"]

573573 src <- targetURI

574− return ([Str "_"], (removeLeadingTrailingSpace src, ""))

574+ return ([Str "_"], (src, ""))

575575

576576regularKey :: GenParser Char ParserState ([Inline], (String, [Char]))

577577regularKey = try $ do

@@ -579,7 +579,7 @@ regularKey = try $ do

579579 ref <- referenceName

580580 char ':'

581581 src <- targetURI

582− return (normalizeSpaces ref, (removeLeadingTrailingSpace src, ""))

582+ return (normalizeSpaces ref, (src, ""))

583583

584584--

585585-- tables

@@ -883,7 +883,8 @@ explicitLink = try $ do

883883 src <- manyTill (noneOf ">\n") (char '>')

884884 skipSpaces

885885 string "`_"

886− return $ Link (normalizeSpaces label') (removeLeadingTrailingSpace src, "")

886+ return $ Link (normalizeSpaces label')

887+ (escapeURI $ removeLeadingTrailingSpace src, "")

887888

888889referenceLink :: GenParser Char ParserState Inline

889890referenceLink = try $ do

@@ -891,25 +892,25 @@ referenceLink = try $ do

891892 key <- option label' (do{char '_'; return [Str "_"]}) -- anonymous link

892893 state <- getState

893894 let keyTable = stateKeys state

894− src <- case lookupKeySrc keyTable key of

895− Nothing -> fail "no corresponding key"

896− Just target -> return target

895+ (src,tit) <- case lookupKeySrc keyTable key of

896+ Nothing -> fail "no corresponding key"

897+ Just target -> return target

897898 -- if anonymous link, remove first anon key so it won't be used again

898899 let keyTable' = if (key == [Str "_"]) -- anonymous link?

899− then delete ([Str "_"], src) keyTable -- remove first anon key

900+ then delete ([Str "_"], (src,tit)) keyTable -- remove first anon key

900901 else keyTable

901902 setState $ state { stateKeys = keyTable' }

902− return $ Link (normalizeSpaces label') src

903+ return $ Link (normalizeSpaces label') (src, tit)

903904

904905autoURI :: GenParser Char ParserState Inline

905906autoURI = do

906− src <- uri

907− return $ Link [Str src] (src, "")

907+ (orig, src) <- uri

908+ return $ Link [Str orig] (src, "")

908909

909910autoEmail :: GenParser Char ParserState Inline

910911autoEmail = do

911− src <- emailAddress

912− return $ Link [Str src] ("mailto:" ++ src, "")

912+ (orig, src) <- emailAddress

913+ return $ Link [Str orig] (src, "")

913914

914915autoLink :: GenParser Char ParserState Inline

915916autoLink = autoURI <|> autoEmail

@@ -921,7 +922,7 @@ image = try $ do

921922 ref <- manyTill inline (char '|')

922923 state <- getState

923924 let keyTable = stateKeys state

924− src <- case lookupKeySrc keyTable ref of

925− Nothing -> fail "no corresponding key"

926− Just target -> return target

927− return $ Image (normalizeSpaces ref) src

925+ (src,tit) <- case lookupKeySrc keyTable ref of

926+ Nothing -> fail "no corresponding key"

927+ Just target -> return target

928+ return $ Image (normalizeSpaces ref) (src, tit)