Markdown reader: properly escape URIs.

+ Resolves Issue #220.
+ Added escapeURI function to Markdown reader. This escapes
  links in a way that makes sense for markdown.  If they've
  used URI escapes like %20 in their link, these will be preserved.
  But if they've used a special character or space without escaping
  it, it will be escaped. This should make sense in most cases.
+ Previously pandoc collapsed adjacent spaces and replaced these
  sequences of spaces with + characters.  That isn't correct for
  a URI path (+ is to be used only in the query part).  We've also
  removed the space-collapsing behavior.

git-svn-id: https://pandoc.googlecode.com/svn/trunk@1847 788f1e2b-df1e-0410-8736-df70ead52e1b

1 files changed, 10 insertions, 2 deletions

diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs
index e3052386a..7131af2b7 100644
--- a/src/Text/Pandoc/Readers/Markdown.hs
+++ b/src/Text/Pandoc/Readers/Markdown.hs
@@ -45,6 +45,7 @@ import Text.Pandoc.Readers.HTML ( rawHtmlBlock, anyHtmlBlockTag,
 import Text.Pandoc.CharacterReferences ( decodeCharacterReferences )
 import Text.ParserCombinators.Parsec
 import Control.Monad (when, liftM, unless)
+import Network.URI ( unEscapeString, escapeURIString, isUnescapedInURI )
 
 -- | Read markdown from an input string and return a Pandoc document.
 readMarkdown :: ParserState -- ^ Parser state, including options for parser
@@ -73,6 +74,13 @@ specialChars = "\\[]*_~`<>$!^-.&'\"\8216\8217\8220\8221;"
 -- auxiliary functions
 --
 
+-- | Escape a URI in a markdown-appropriate way.  First,
+-- we unescape the string, since we don't want to screw things
+-- up if they've entered a properly encoded URI.  Then, we
+-- escape the result.
+escapeURI :: String -> String
+escapeURI = escapeURIString isUnescapedInURI . unEscapeString
+
 indentSpaces :: GenParser Char ParserState [Char]
 indentSpaces = try $ do
   state <- getState
@@ -194,7 +202,7 @@ referenceKey = try $ do
   tit <- option "" referenceTitle
   blanklines
   endPos <- getPosition
-  let newkey = (lab, (intercalate "+" $ words $ removeTrailingSpace src,  tit))
+  let newkey = (lab, (escapeURI $ removeTrailingSpace src,  tit))
   st <- getState
   let oldkeys = stateKeys st
   updateState $ \s -> s { stateKeys = newkey : oldkeys }
@@ -1173,7 +1181,7 @@ source' = do
   tit <- option "" linkTitle
   skipSpaces
   eof
-  return (intercalate "+" $ words $ removeTrailingSpace src, tit)
+  return (escapeURI $ removeTrailingSpace src, tit)
 
 linkTitle :: GenParser Char st String
 linkTitle = try $ do