Parsing: Fixed uri -- escape unicode URLs.

Otherwise Network.URI.parseURI fails on e.g. Chinese
URLs.  Changed an incorrect test in markdown-reader-more.

src/Text/Pandoc/Parsing.hs

@@ -153,7 +153,7 @@ import Text.Parsec
 import Text.Parsec.Pos (newPos)
 import Data.Char ( toLower, toUpper, ord, isAscii, isAlphaNum, isDigit, isPunctuation )
 import Data.List ( intercalate, transpose )
-import Network.URI ( parseURI, URI (..), isAllowedInURI )
+import Network.URI ( parseURI, URI (..), isAllowedInURI, isUnescapedInURI, escapeURIString )
 import Text.Pandoc.Shared
 import qualified Data.Map as M
 import Text.TeXMath.Macros (applyMacros, Macro, parseMacroDefinitions)
@@ -380,7 +380,7 @@ uri = try $ do
   str <- liftM concat $ many1 $ inParens <|> count 1 (innerPunct <|> uriChar)
   str' <- option str $ char '/' >> return (str ++ "/")
   -- now see if they amount to an absolute URI
-  case parseURI (escapeURI str') of
+  case parseURI (escapeURIString isUnescapedInURI str') of
        Just uri' -> if uriScheme uri' `elem` protocols
                        then return (str', show uri')
                        else fail "not a URI"

tests/markdown-reader-more.native

@@ -29,7 +29,7 @@
 ,Para [Str "`hi"]
 ,Para [Str "there`"]
 ,Header 2 ("multilingual-urls",[],[]) [Str "Multilingual",Space,Str "URLs"]
-,Para [RawInline "html" "<http://\27979.com?\27979=\27979>"]
+,Para [Link [Str "http://\27979.com?\27979=\27979"] ("http://%E6%B5%8B.com?%E6%B5%8B=%E6%B5%8B","")]
 ,Para [Link [Str "foo"] ("/bar/\27979?x=\27979","title")]
 ,Para [Link [Str "\27979@foo.\27979.baz"] ("mailto:\27979@foo.\27979.baz","")]
 ,Header 2 ("numbered-examples",[],[]) [Str "Numbered",Space,Str "examples"]