Properly escape URIs in all readers.

4 files changed, 37 insertions, 44 deletions

diff --git a/src/Text/Pandoc/Readers/HTML.hs b/src/Text/Pandoc/Readers/HTML.hs
index e68592b95..5c188e3d9 100644
--- a/src/Text/Pandoc/Readers/HTML.hs
+++ b/src/Text/Pandoc/Readers/HTML.hs
@@ -182,7 +182,7 @@ unsanitaryURI u =
              "ldaps:", "magnet:", "mms:", "msnim:", "notes:", "rsync:",
              "secondlife:", "skype:", "ssh:", "sftp:", "smb:", "sms:",
              "snews:", "webcal:", "ymsgr:"]
-  in  case parseURIReference (stringToURI u) of
+  in  case parseURIReference (escapeURI u) of
            Just p  -> (map toLower $ uriScheme p) `notElem` safeURISchemes
            Nothing -> True
 
@@ -746,7 +746,7 @@ link = try $ do
            Nothing  -> fail "no href"
   let title = fromMaybe "" $ extractAttribute "title" attributes
   lab <- inlinesTilEnd "a"
-  return $ Link (normalizeSpaces lab) (url, title)
+  return $ Link (normalizeSpaces lab) (escapeURI url, title)
 
 image :: GenParser Char ParserState Inline
 image = try $ do
@@ -756,5 +756,5 @@ image = try $ do
            Nothing  -> fail "no src"
   let title = fromMaybe "" $ extractAttribute "title" attributes
   let alt = fromMaybe "" (extractAttribute "alt" attributes)
-  return $ Image [Str alt] (url, title)
+  return $ Image [Str alt] (escapeURI url, title)
 
diff --git a/src/Text/Pandoc/Readers/LaTeX.hs b/src/Text/Pandoc/Readers/LaTeX.hs
index a2df27c07..36940fab0 100644
--- a/src/Text/Pandoc/Readers/LaTeX.hs
+++ b/src/Text/Pandoc/Readers/LaTeX.hs
@@ -738,7 +738,7 @@ url :: GenParser Char ParserState Inline
 url = try $ do
   string "\\url"
   url' <- charsInBalanced '{' '}'
-  return $ Link [Code url'] (url', "")
+  return $ Link [Code url'] (escapeURI url', "")
 
 link :: GenParser Char ParserState Inline
 link = try $ do
@@ -746,17 +746,16 @@ link = try $ do
   url' <- manyTill anyChar (char '}')
   char '{'
   label' <- manyTill inline (char '}') 
-  return $ Link (normalizeSpaces label') (url', "")
+  return $ Link (normalizeSpaces label') (escapeURI url', "")
 
 image :: GenParser Char ParserState Inline
 image = try $ do
   ("includegraphics", _, args) <- command
   let args' = filter isArg args -- filter out options
-  let src = if null args' then
-              ("", "")
-            else
-              (stripFirstAndLast (head args'), "")
-  return $ Image [Str "image"] src
+  let (src,tit) = case args' of
+                       []    -> ("", "")
+                       (x:_) -> (stripFirstAndLast x, "")
+  return $ Image [Str "image"] (escapeURI src, tit)
 
 footnote :: GenParser Char ParserState Inline
 footnote = try $ do
diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs
index fafd526e6..13edd0586 100644
--- a/src/Text/Pandoc/Readers/Markdown.hs
+++ b/src/Text/Pandoc/Readers/Markdown.hs
@@ -31,7 +31,7 @@ module Text.Pandoc.Readers.Markdown (
                                      readMarkdown 
                                     ) where
 
-import Data.List ( transpose, isPrefixOf, isSuffixOf, sortBy, findIndex, intercalate )
+import Data.List ( transpose, isSuffixOf, sortBy, findIndex, intercalate )
 import Data.Ord ( comparing )
 import Data.Char ( isAlphaNum )
 import Data.Maybe
@@ -73,10 +73,6 @@ specialChars = "\\[]*_~`<>$!^-.&'\"\8216\8217\8220\8221;"
 -- auxiliary functions
 --
 
--- | Replace spaces with %20
-uriEscapeSpaces :: String -> String
-uriEscapeSpaces = substitute " " "%20"
-
 indentSpaces :: GenParser Char ParserState [Char]
 indentSpaces = try $ do
   state <- getState
@@ -206,7 +202,7 @@ referenceKey = try $ do
   tit <- option "" referenceTitle
   blanklines
   endPos <- getPosition
-  let newkey = (lab, (uriEscapeSpaces $ removeTrailingSpace src,  tit))
+  let newkey = (lab, (escapeURI $ removeTrailingSpace src,  tit))
   st <- getState
   let oldkeys = stateKeys st
   updateState $ \s -> s { stateKeys = newkey : oldkeys }
@@ -1194,7 +1190,7 @@ source' = do
   tit <- option "" linkTitle
   skipSpaces
   eof
-  return (uriEscapeSpaces $ removeTrailingSpace src, tit)
+  return (escapeURI $ removeTrailingSpace src, tit)
 
 linkTitle :: GenParser Char st String
 linkTitle = try $ do 
@@ -1208,11 +1204,11 @@ linkTitle = try $ do
 link :: GenParser Char ParserState Inline
 link = try $ do
   lab <- reference
-  src <- source <|> referenceLink lab
+  (src, tit) <- source <|> referenceLink lab
   sanitize <- getState >>= return . stateSanitizeHTML
-  if sanitize && unsanitaryURI (fst src)
+  if sanitize && unsanitaryURI src
      then fail "Unsanitary URI"
-     else return $ Link lab src
+     else return $ Link lab (src, tit)
 
 -- a link like [this][ref] or [this][] or [this]
 referenceLink :: [Inline]
@@ -1229,18 +1225,15 @@ referenceLink lab = do
 autoLink :: GenParser Char ParserState Inline
 autoLink = try $ do
   char '<'
-  src <- uri <|> (emailAddress >>= (return . ("mailto:" ++)))
+  (orig, src) <- uri <|> emailAddress
   char '>'
-  let src' = if "mailto:" `isPrefixOf` src
-                then drop 7 src
-                else src 
   st <- getState
   let sanitize = stateSanitizeHTML st
   if sanitize && unsanitaryURI src
      then fail "Unsanitary URI"
      else return $ if stateStrict st
-                      then Link [Str src'] (src, "")
-                      else Link [Code src'] (src, "")
+                      then Link [Str orig] (src, "")
+                      else Link [Code orig] (src, "")
 
 image :: GenParser Char ParserState Inline
 image = try $ do
diff --git a/src/Text/Pandoc/Readers/RST.hs b/src/Text/Pandoc/Readers/RST.hs
index 40b02a25a..5e7ea512e 100644
--- a/src/Text/Pandoc/Readers/RST.hs
+++ b/src/Text/Pandoc/Readers/RST.hs
@@ -556,7 +556,7 @@ targetURI = do
   contents <- many1 (try (many spaceChar >> newline >> 
                           many1 spaceChar >> noneOf " \t\n") <|> noneOf "\n")
   blanklines
-  return contents
+  return $ escapeURI $ removeLeadingTrailingSpace $ contents
 
 imageKey :: GenParser Char ParserState ([Inline], (String, [Char]))
 imageKey = try $ do
@@ -565,13 +565,13 @@ imageKey = try $ do
   skipSpaces
   string "image::"
   src <- targetURI
-  return (normalizeSpaces ref, (removeLeadingTrailingSpace src, ""))
+  return (normalizeSpaces ref, (src, ""))
 
 anonymousKey :: GenParser Char st ([Inline], (String, [Char]))
 anonymousKey = try $ do
   oneOfStrings [".. __:", "__"]
   src <- targetURI
-  return ([Str "_"], (removeLeadingTrailingSpace src, ""))
+  return ([Str "_"], (src, ""))
 
 regularKey :: GenParser Char ParserState ([Inline], (String, [Char]))
 regularKey = try $ do
@@ -579,7 +579,7 @@ regularKey = try $ do
   ref <- referenceName
   char ':'
   src <- targetURI
-  return (normalizeSpaces ref, (removeLeadingTrailingSpace src, ""))
+  return (normalizeSpaces ref, (src, ""))
 
 --
 -- tables
@@ -883,7 +883,8 @@ explicitLink = try $ do
   src <- manyTill (noneOf ">\n") (char '>')
   skipSpaces
   string "`_"
-  return $ Link (normalizeSpaces label') (removeLeadingTrailingSpace src, "")
+  return $ Link (normalizeSpaces label')
+            (escapeURI $ removeLeadingTrailingSpace src, "")
 
 referenceLink :: GenParser Char ParserState Inline
 referenceLink = try $ do
@@ -891,25 +892,25 @@ referenceLink = try $ do
   key <- option label' (do{char '_'; return [Str "_"]}) -- anonymous link
   state <- getState
   let keyTable = stateKeys state
-  src <- case lookupKeySrc keyTable key of
-           Nothing     -> fail "no corresponding key"
-           Just target -> return target
+  (src,tit) <- case lookupKeySrc keyTable key of
+                    Nothing     -> fail "no corresponding key"
+                    Just target -> return target
   -- if anonymous link, remove first anon key so it won't be used again
   let keyTable' = if (key == [Str "_"]) -- anonymous link? 
-                    then delete ([Str "_"], src) keyTable -- remove first anon key 
+                    then delete ([Str "_"], (src,tit)) keyTable -- remove first anon key 
                     else keyTable                    
   setState $ state { stateKeys = keyTable' }
-  return $ Link (normalizeSpaces label') src 
+  return $ Link (normalizeSpaces label') (src, tit) 
 
 autoURI :: GenParser Char ParserState Inline
 autoURI = do
-  src <- uri
-  return $ Link [Str src] (src, "")
+  (orig, src) <- uri
+  return $ Link [Str orig] (src, "")
 
 autoEmail :: GenParser Char ParserState Inline
 autoEmail = do
-  src <- emailAddress
-  return $ Link [Str src] ("mailto:" ++ src, "")
+  (orig, src) <- emailAddress
+  return $ Link [Str orig] (src, "")
 
 autoLink :: GenParser Char ParserState Inline
 autoLink = autoURI <|> autoEmail
@@ -921,7 +922,7 @@ image = try $ do
   ref <- manyTill inline (char '|')
   state <- getState
   let keyTable = stateKeys state
-  src <- case lookupKeySrc keyTable ref of
-           Nothing     -> fail "no corresponding key"
-           Just target -> return target
-  return $ Image (normalizeSpaces ref) src
+  (src,tit) <- case lookupKeySrc keyTable ref of
+                     Nothing     -> fail "no corresponding key"
+                     Just target -> return target
+  return $ Image (normalizeSpaces ref) (src, tit)