Use cmark 0.5.

Closes #2605.

3 files changed, 37 insertions, 26 deletions

diff --git a/src/Text/Pandoc/Readers/CommonMark.hs b/src/Text/Pandoc/Readers/CommonMark.hs
index 71c7d05b2..d20d386e7 100644
--- a/src/Text/Pandoc/Readers/CommonMark.hs
+++ b/src/Text/Pandoc/Readers/CommonMark.hs
@@ -58,15 +58,19 @@ addBlocks = foldr addBlock []
 addBlock :: Node -> [Block] -> [Block]
 addBlock (Node _ PARAGRAPH nodes) =
   (Para (addInlines nodes) :)
-addBlock (Node _ HRULE _) =
+addBlock (Node _ THEMATIC_BREAK _) =
   (HorizontalRule :)
 addBlock (Node _ BLOCK_QUOTE nodes) =
   (BlockQuote (addBlocks nodes) :)
-addBlock (Node _ (HTML t) _) =
+addBlock (Node _ (HTML_BLOCK t) _) =
   (RawBlock (Format "html") (unpack t) :)
+-- Note:  the cmark parser will never generate CUSTOM_BLOCK,
+-- so we don't need to handle it:
+addBlock (Node _ (CUSTOM_BLOCK _onEnter _onExit) _nodes) =
+  id
 addBlock (Node _ (CODE_BLOCK info t) _) =
   (CodeBlock ("", take 1 (words (unpack info)), []) (unpack t) :)
-addBlock (Node _ (HEADER lev) nodes) =
+addBlock (Node _ (HEADING lev) nodes) =
   (Header lev ("",[],[]) (addInlines nodes) :)
 addBlock (Node _ (LIST listAttrs) nodes) =
   (constructor (map (setTightness . addBlocks . children) nodes) :)
@@ -104,8 +108,12 @@ addInline (Node _ (TEXT t) _) = (map toinl clumps ++)
         toinl xs         = Str xs
 addInline (Node _ LINEBREAK _) = (LineBreak :)
 addInline (Node _ SOFTBREAK _) = (SoftBreak :)
-addInline (Node _ (INLINE_HTML t) _) =
+addInline (Node _ (HTML_INLINE t) _) =
   (RawInline (Format "html") (unpack t) :)
+-- Note:  the cmark parser will never generate CUSTOM_BLOCK,
+-- so we don't need to handle it:
+addInline (Node _ (CUSTOM_INLINE _onEnter _onExit) _nodes) =
+  id
 addInline (Node _ (CODE t) _) =
   (Code ("",[],[]) (unpack t) :)
 addInline (Node _ EMPH nodes) =
diff --git a/src/Text/Pandoc/Shared.hs b/src/Text/Pandoc/Shared.hs
index 9d799fa52..aa07c81e1 100644
--- a/src/Text/Pandoc/Shared.hs
+++ b/src/Text/Pandoc/Shared.hs
@@ -909,9 +909,11 @@ fetchItem sourceURL s =
           case parseURI s' of  -- requires absolute URI
                -- We don't want to treat C:/ as a scheme:
                Just u' | length (uriScheme u') > 2 -> openURL (show u')
-               _ -> E.try readLocalFile -- get from local file system
-  where readLocalFile = do
-          cont <- BS.readFile fp
+               Just u' | uriScheme u' == "file:" ->
+                    E.try $ readLocalFile $ dropWhile (=='/') (uriPath u')
+               _ -> E.try $ readLocalFile fp -- get from local file system
+  where readLocalFile f = do
+          cont <- BS.readFile f
           return (cont, mime)
         dropFragmentAndQuery = takeWhile (\c -> c /= '?' && c /= '#')
         fp = unEscapeString $ dropFragmentAndQuery s
@@ -919,8 +921,9 @@ fetchItem sourceURL s =
                     ".gz" -> getMimeType $ dropExtension fp
                     ".svgz" -> getMimeType $ dropExtension fp ++ ".svg"
                     x     -> getMimeType x
-        ensureEscaped x@(_:':':'\\':_) = x -- likely windows path
-        ensureEscaped x = escapeURIString isAllowedInURI x
+        ensureEscaped = escapeURIString isAllowedInURI . map convertSlash
+        convertSlash '\\' = '/'
+        convertSlash x    = x
 
 -- | Like 'fetchItem', but also looks for items in a 'MediaBag'.
 fetchItem' :: MediaBag -> Maybe String -> String
diff --git a/src/Text/Pandoc/Writers/CommonMark.hs b/src/Text/Pandoc/Writers/CommonMark.hs
index a786dfd24..262f491a8 100644
--- a/src/Text/Pandoc/Writers/CommonMark.hs
+++ b/src/Text/Pandoc/Writers/CommonMark.hs
@@ -97,8 +97,8 @@ blockToNodes (Para xs) = (node PARAGRAPH (inlinesToNodes xs) :)
 blockToNodes (CodeBlock (_,classes,_) xs) =
   (node (CODE_BLOCK (T.pack (unwords classes)) (T.pack xs)) [] :)
 blockToNodes (RawBlock fmt xs)
-  | fmt == Format "html" = (node (HTML (T.pack xs)) [] :)
-  | otherwise = id
+  | fmt == Format "html" = (node (HTML_BLOCK (T.pack xs)) [] :)
+  | otherwise = (node (CUSTOM_BLOCK (T.pack xs) (T.empty)) [] :)
 blockToNodes (BlockQuote bs) =
   (node BLOCK_QUOTE (blocksToNodes bs) :)
 blockToNodes (BulletList items) =
@@ -116,8 +116,8 @@ blockToNodes (OrderedList (start, _sty, delim) items) =
                                 _         -> PERIOD_DELIM,
                listTight = isTightList items,
                listStart = start }) (map (node ITEM . blocksToNodes) items) :)
-blockToNodes HorizontalRule = (node HRULE [] :)
-blockToNodes (Header lev _ ils) = (node (HEADER lev) (inlinesToNodes ils) :)
+blockToNodes HorizontalRule = (node THEMATIC_BREAK [] :)
+blockToNodes (Header lev _ ils) = (node (HEADING lev) (inlinesToNodes ils) :)
 blockToNodes (Div _ bs) = (blocksToNodes bs ++)
 blockToNodes (DefinitionList items) = blockToNodes (BulletList items')
   where items' = map dlToBullet items
@@ -128,7 +128,7 @@ blockToNodes (DefinitionList items) = blockToNodes (BulletList items')
         dlToBullet (term, xs) =
           Para term : concat xs
 blockToNodes t@(Table _ _ _ _ _) =
-  (node (HTML (T.pack $! writeHtmlString def $! Pandoc nullMeta [t])) [] :)
+  (node (HTML_BLOCK (T.pack $! writeHtmlString def $! Pandoc nullMeta [t])) [] :)
 blockToNodes Null = id
 
 inlinesToNodes :: [Inline] -> [Node]
@@ -142,25 +142,25 @@ inlineToNodes SoftBreak = (node SOFTBREAK [] :)
 inlineToNodes (Emph xs) = (node EMPH (inlinesToNodes xs) :)
 inlineToNodes (Strong xs) = (node STRONG (inlinesToNodes xs) :)
 inlineToNodes (Strikeout xs) =
-  ((node (INLINE_HTML (T.pack "<s>")) [] : inlinesToNodes xs ++
-   [node (INLINE_HTML (T.pack "</s>")) []]) ++ )
+  ((node (HTML_INLINE (T.pack "<s>")) [] : inlinesToNodes xs ++
+   [node (HTML_INLINE (T.pack "</s>")) []]) ++ )
 inlineToNodes (Superscript xs) =
-  ((node (INLINE_HTML (T.pack "<sup>")) [] : inlinesToNodes xs ++
-   [node (INLINE_HTML (T.pack "</sup>")) []]) ++ )
+  ((node (HTML_INLINE (T.pack "<sup>")) [] : inlinesToNodes xs ++
+   [node (HTML_INLINE (T.pack "</sup>")) []]) ++ )
 inlineToNodes (Subscript xs) =
-  ((node (INLINE_HTML (T.pack "<sub>")) [] : inlinesToNodes xs ++
-   [node (INLINE_HTML (T.pack "</sub>")) []]) ++ )
+  ((node (HTML_INLINE (T.pack "<sub>")) [] : inlinesToNodes xs ++
+   [node (HTML_INLINE (T.pack "</sub>")) []]) ++ )
 inlineToNodes (SmallCaps xs) =
-  ((node (INLINE_HTML (T.pack "<span style=\"font-variant:small-caps;\">")) []
+  ((node (HTML_INLINE (T.pack "<span style=\"font-variant:small-caps;\">")) []
     : inlinesToNodes xs ++
-    [node (INLINE_HTML (T.pack "</span>")) []]) ++ )
+    [node (HTML_INLINE (T.pack "</span>")) []]) ++ )
 inlineToNodes (Link _ ils (url,tit)) =
   (node (LINK (T.pack url) (T.pack tit)) (inlinesToNodes ils) :)
 inlineToNodes (Image _ ils (url,tit)) =
   (node (IMAGE (T.pack url) (T.pack tit)) (inlinesToNodes ils) :)
 inlineToNodes (RawInline fmt xs)
-  | fmt == Format "html" = (node (INLINE_HTML (T.pack xs)) [] :)
-  | otherwise = id
+  | fmt == Format "html" = (node (HTML_INLINE (T.pack xs)) [] :)
+  | otherwise = (node (CUSTOM_INLINE (T.pack xs) (T.empty)) [] :)
 inlineToNodes (Quoted qt ils) =
   ((node (TEXT start) [] : inlinesToNodes ils ++ [node (TEXT end) []]) ++)
   where (start, end) = case qt of
@@ -170,9 +170,9 @@ inlineToNodes (Code _ str) = (node (CODE (T.pack str)) [] :)
 inlineToNodes (Math mt str) =
   case mt of
     InlineMath  ->
-      (node (INLINE_HTML (T.pack ("\\(" ++ str ++ "\\)"))) [] :)
+      (node (HTML_INLINE (T.pack ("\\(" ++ str ++ "\\)"))) [] :)
     DisplayMath ->
-      (node (INLINE_HTML (T.pack ("\\[" ++ str ++ "\\]"))) [] :)
+      (node (HTML_INLINE (T.pack ("\\[" ++ str ++ "\\]"))) [] :)
 inlineToNodes (Span _ ils) = (inlinesToNodes ils ++)
 inlineToNodes (Cite _ ils) = (inlinesToNodes ils ++)
 inlineToNodes (Note _) = id -- should not occur