Fix parsing bug affected indented code after raw HTML.

Closes #6009, #5360.

src/Text/Pandoc/Readers/Markdown.hs

@@ -1094,18 +1094,20 @@ rawHtmlBlocks = do
   -- inline will not be parsed as inline tags
   oldInHtmlBlock <- stateInHtmlBlock <$> getState
   updateState $ \st -> st{ stateInHtmlBlock = Just tagtype }
-  let closer = htmlTag (\x -> x ~== TagClose tagtype)
-  let block' = try $
-               do notFollowedBy' closer
-                  gobbleAtMostSpaces indentlevel
-                  block
+  let closer = htmlTag (~== TagClose tagtype)
+  let block' = try $ do
+                 gobbleAtMostSpaces indentlevel
+                 notFollowedBy' closer
+                 block
   contents <- mconcat <$> many block'
   result <-
-    (closer >>= \(_, rawcloser) -> return (
-                return (B.rawBlock "html" $ stripMarkdownAttribute raw) <>
+    try
+    (do gobbleAtMostSpaces indentlevel
+        (_, rawcloser) <- closer
+        return (return (B.rawBlock "html" $ stripMarkdownAttribute raw) <>
                 contents <>
                 return (B.rawBlock "html" rawcloser)))
-      <|> return (return (B.rawBlock "html" raw) <> contents)
+      <|> (return (return (B.rawBlock "html" raw) <> contents))
   updateState $ \st -> st{ stateInHtmlBlock = oldInHtmlBlock }
   return result
 

test/command/5360.md

@@ -0,0 +1,19 @@
+```
+% pandoc -t native
+::: {.foo}
+<table>
+ <tr>
+  <td>hi</td>
+ </tr>
+</table>
+:::
+^D
+[Div ("",["foo"],[])
+ [RawBlock (Format "html") "<table>"
+ ,RawBlock (Format "html") "<tr>"
+ ,RawBlock (Format "html") "<td>"
+ ,Plain [Str "hi"]
+ ,RawBlock (Format "html") "</td>"
+ ,RawBlock (Format "html") "</tr>"
+ ,RawBlock (Format "html") "</table>"]]
+```

test/command/6009.md

@@ -0,0 +1,18 @@
+```
+% pandoc -t native
+  <tr>
+   <td>
+   </td>
+  </tr>
+
+x
+
+    y
+^D
+[RawBlock (Format "html") "<tr>"
+,RawBlock (Format "html") "<td>"
+,RawBlock (Format "html") "</td>"
+,RawBlock (Format "html") "</tr>"
+,Para [Str "x"]
+,CodeBlock ("",[],[]) "y"]
+```