Mediawiki reader: improve table parsing.

This fixes detection of table attributes and also
handles `!` characters in cells.

Closes #4508.

1 files changed, 18 insertions, 5 deletions

diff --git a/src/Text/Pandoc/Readers/MediaWiki.hs b/src/Text/Pandoc/Readers/MediaWiki.hs
index a07f66e7a..498f80216 100644
--- a/src/Text/Pandoc/Readers/MediaWiki.hs
+++ b/src/Text/Pandoc/Readers/MediaWiki.hs
@@ -232,7 +232,8 @@ para = do
 table :: PandocMonad m => MWParser m Blocks
 table = do
   tableStart
-  styles <- option [] parseAttrs
+  styles <- option [] $
+               parseAttrs <* skipMany spaceChar <* optional (char '|')
   skipMany spaceChar
   optional blanklines
   let tableWidth = case lookup "width" styles of
@@ -283,17 +284,29 @@ rowsep = try $ guardColumnOne *> skipSpaces *> sym "|-" <*
 
 cellsep :: PandocMonad m => MWParser m ()
 cellsep = try $ do
+  col <- sourceColumn <$> getPosition
   skipSpaces
-  (char '|' *> notFollowedBy (oneOf "-}+") *> optional (char '|'))
-    <|> (char '!' *> optional (char '!'))
+  let pipeSep = do
+        char '|'
+        notFollowedBy (oneOf "-}+")
+        if col == 1
+           then optional (char '|')
+           else void (char '|')
+  let exclSep = do
+        char '!'
+        if col == 1
+           then optional (char '!')
+           else void (char '!')
+  pipeSep <|> exclSep
 
 tableCaption :: PandocMonad m => MWParser m Inlines
 tableCaption = try $ do
   guardColumnOne
   skipSpaces
   sym "|+"
-  optional (try $ parseAttr *> skipSpaces *> char '|' *> skipSpaces)
-  (trimInlines . mconcat) <$> many (notFollowedBy (cellsep <|> rowsep) *> inline)
+  optional (try $ parseAttr *> skipSpaces *> char '|' *> blanklines)
+  (trimInlines . mconcat) <$>
+    many (notFollowedBy (cellsep <|> rowsep) *> inline)
 
 tableRow :: PandocMonad m => MWParser m [((Alignment, Double), Blocks)]
 tableRow = try $ skipMany htmlComment *> many tableCell