Mediawiki reader: Parse table cell with attribs, to support rowspan, colspan (#8231)

2 files changed, 31 insertions, 12 deletions

diff --git a/src/Text/Pandoc/Readers/HTML.hs b/src/Text/Pandoc/Readers/HTML.hs
index 4fb6028e4..4e2be9382 100644
--- a/src/Text/Pandoc/Readers/HTML.hs
+++ b/src/Text/Pandoc/Readers/HTML.hs
@@ -21,6 +21,7 @@ module Text.Pandoc.Readers.HTML ( readHtml
                                 , isBlockTag
                                 , isTextTag
                                 , isCommentTag
+                                , toAttr
                                 ) where
 
 import Control.Applicative ((<|>))
diff --git a/src/Text/Pandoc/Readers/MediaWiki.hs b/src/Text/Pandoc/Readers/MediaWiki.hs
index 0fdfa8b84..8e86a0634 100644
--- a/src/Text/Pandoc/Readers/MediaWiki.hs
+++ b/src/Text/Pandoc/Readers/MediaWiki.hs
@@ -35,7 +35,7 @@ import Text.Pandoc.Definition
 import Text.Pandoc.Logging
 import Text.Pandoc.Options
 import Text.Pandoc.Parsing hiding (nested, tableCaption)
-import Text.Pandoc.Readers.HTML (htmlTag, isBlockTag, isCommentTag)
+import Text.Pandoc.Readers.HTML (htmlTag, isBlockTag, isCommentTag, toAttr)
 import Text.Pandoc.Shared (safeRead, stringify, stripTrailingNewlines,
                            trim, splitTextBy, tshow, formatCode)
 import Text.Pandoc.XML (fromEntities)
@@ -222,22 +222,21 @@ table = do
   optional rowsep
   hasheader <- option False $ True <$ lookAhead (skipSpaces *> char '!')
   (cellspecs',hdr) <- unzip <$> tableRow
-  let widths = map ((tableWidth *) . snd) cellspecs'
+  let widths = map (tableWidth *) cellspecs'
   let restwidth = tableWidth - sum widths
   let zerocols = length $ filter (==0.0) widths
   let defaultwidth = if zerocols == 0 || zerocols == length widths
                         then ColWidthDefault
                         else ColWidth $ restwidth / fromIntegral zerocols
   let widths' = map (\w -> if w > 0 then ColWidth w else defaultwidth) widths
-  let cellspecs = zip (map fst cellspecs') widths'
+  let cellspecs = zip (calculateAlignments hdr) widths'
   rows' <- many $ try $ rowsep *> (map snd <$> tableRow)
   optional blanklines
   tableEnd
-  let cols = length hdr
   let (headers,rows) = if hasheader
                           then (hdr, rows')
-                          else (replicate cols mempty, hdr:rows')
-  let toRow = Row nullAttr . map B.simpleCell
+                          else ([], hdr:rows')
+  let toRow = Row nullAttr
       toHeaderRow l = [toRow l | not (null l)]
   return $ B.table (B.simpleCaption $ B.plain caption)
                    cellspecs
@@ -245,6 +244,12 @@ table = do
                    [TableBody nullAttr 0 [] $ map toRow rows]
                    (TableFoot nullAttr [])
 
+calculateAlignments :: [Cell] -> [Alignment]
+calculateAlignments = map cellAligns
+  where
+    cellAligns :: Cell -> Alignment
+    cellAligns (Cell _ align _ _ _) = align
+
 parseAttrs :: PandocMonad m => MWParser m [(Text,Text)]
 parseAttrs = many1 parseAttr
 
@@ -252,7 +257,9 @@ parseAttr :: PandocMonad m => MWParser m (Text, Text)
 parseAttr = try $ do
   skipMany spaceChar
   k <- many1Char letter
+  skipMany spaceChar
   char '='
+  skipMany spaceChar
   v <- (char '"' >> many1TillChar (satisfy (/='\n')) (char '"'))
        <|> many1Char (satisfy $ \c -> not (isSpace c) && c /= '|')
   return (k,v)
@@ -289,6 +296,7 @@ cellsep = try $ do
 
 tableCaption :: PandocMonad m => MWParser m Inlines
 tableCaption = try $ do
+  optional rowsep
   guardColumnOne
   skipSpaces
   sym "|+"
@@ -296,14 +304,14 @@ tableCaption = try $ do
   trimInlines . mconcat <$>
     many (notFollowedBy (cellsep <|> rowsep) *> inline)
 
-tableRow :: PandocMonad m => MWParser m [((Alignment, Double), Blocks)]
+tableRow :: PandocMonad m => MWParser m [(Double, Cell)]
 tableRow = try $ skipMany htmlComment *> many tableCell
 
-tableCell :: PandocMonad m => MWParser m ((Alignment, Double), Blocks)
+tableCell :: PandocMonad m => MWParser m (Double, Cell)
 tableCell = try $ do
   cellsep
   skipMany spaceChar
-  attrs <- option [] $ try $ parseAttrs <* skipSpaces <* char '|' <*
+  attribs <- option [] $ try $ parseAttrs <* skipSpaces <* char '|' <*
                                  notFollowedBy (char '|')
   skipMany spaceChar
   pos' <- getPosition
@@ -311,15 +319,25 @@ tableCell = try $ do
                             ((snd <$> withRaw table) <|> countChar 1 anyChar))
   bs <- parseFromString (do setPosition pos'
                             mconcat <$> many block) ls
-  let align = case lookup "align" attrs of
+  let align = case lookup "align" attribs of
                     Just "left"   -> AlignLeft
                     Just "right"  -> AlignRight
                     Just "center" -> AlignCenter
                     _             -> AlignDefault
-  let width = case lookup "width" attrs of
+  let width = case lookup "width" attribs of
                     Just xs -> fromMaybe 0.0 $ parseWidth xs
                     Nothing -> 0.0
-  return ((align, width), bs)
+  let rowspan = RowSpan . fromMaybe 1 $
+                safeRead =<< lookup "rowspan" attribs
+  let colspan = ColSpan . fromMaybe 1 $
+                safeRead =<< lookup "colspan" attribs
+  let handledAttribs = ["align", "colspan", "rowspan"]
+      attribs' = foldr go [] attribs
+      go kv@(k, _) acc = case k of
+        -- drop attrib if it's already handled
+        _ | k `elem` handledAttribs -> acc
+        _ -> kv : acc
+  return (width, B.cellWith (toAttr attribs') align rowspan colspan bs)
 
 parseWidth :: Text -> Maybe Double
 parseWidth s =