HTML reader: treat Plain as Para when needed.

For example, in

    Just a few glitches remaining.
    <ul><li> In this situation, one loses the list.
    </ul>
    And in this, the preformatting.
   <pre>Preformatted text not starting with its own blank line.
   </pre>

Thansk to Dirk Laurie for noticing the issue.

src/Text/Pandoc/Readers/HTML.hs

@@ -75,7 +75,7 @@ parseHeader tags = (Meta{docTitle = tit'', docAuthors = [], docDate = []}, rest)
                                                 t ~== TagOpen "body" []) tags
 
 parseBody :: TagParser [Block]
-parseBody = liftM concat $ manyTill block eof
+parseBody = liftM (fixPlains False . concat) $ manyTill block eof
 
 block :: TagParser [Block]
 block = choice
@@ -107,7 +107,7 @@ pBulletList = try $ do
   -- treat it as a list item, though it's not valid xhtml...
   skipMany nonItem
   items <- manyTill (pInTags "li" block >>~ skipMany nonItem) (pCloses "ul")
-  return [BulletList $ map fixPlains items]
+  return [BulletList $ map (fixPlains True) items]
 
 pOrderedList :: TagParser [Block]
 pOrderedList = try $ do
@@ -138,7 +138,7 @@ pOrderedList = try $ do
   -- treat it as a list item, though it's not valid xhtml...
   skipMany nonItem
   items <- manyTill (pInTags "li" block >>~ skipMany nonItem) (pCloses "ol")
-  return [OrderedList (start, style, DefaultDelim) $ map fixPlains items]
+  return [OrderedList (start, style, DefaultDelim) $ map (fixPlains True) items]
 
 pDefinitionList :: TagParser [Block]
 pDefinitionList = try $ do
@@ -154,16 +154,19 @@ pDefListItem = try $ do
   defs  <- many1 (try $ skipMany nonItem >> pInTags "dd" block)
   skipMany nonItem
   let term = intercalate [LineBreak] terms
-  return (term, map fixPlains defs)
+  return (term, map (fixPlains True) defs)
 
-fixPlains :: [Block] -> [Block]
-fixPlains bs = if any isParaish bs
-                  then map plainToPara bs
-                  else bs
+fixPlains :: Bool -> [Block] -> [Block]
+fixPlains inList bs = if any isParaish bs
+                         then map plainToPara bs
+                         else bs
   where isParaish (Para _) = True
         isParaish (CodeBlock _ _) = True
         isParaish (Header _ _) = True
         isParaish (BlockQuote _) = True
+        isParaish (BulletList _) = not inList
+        isParaish (OrderedList _ _) = not inList
+        isParaish (DefinitionList _) = not inList
         isParaish _        = False
         plainToPara (Plain xs) = Para xs
         plainToPara x = x
@@ -231,7 +234,7 @@ pCell celltype = try $ do
 pBlockQuote :: TagParser [Block]
 pBlockQuote = do
   contents <- pInTags "blockquote" block
-  return [BlockQuote contents]
+  return [BlockQuote $ fixPlains False contents]
 
 pPlain :: TagParser [Block]
 pPlain = do

tests/html-reader.html

@@ -298,40 +298,6 @@ These should not be escaped:  \$ \\ \> \[ \{
   <dd>Low-voiced stringed instrument.</dd>
 </dl>
 <hr />
-<h1>HTML Blocks</h1>
-<p>Simple block on one line:</p>
-foo<p>And nested without indentation:</p>
-foobar<p>Interpreted markdown in a table:</p>
-This is <em>emphasized</em>. And this is <strong>strong</strong><p>Here's a simple block:</p>
-foo<p>This should be a code block, though:</p>
-<pre><code>&lt;div>
-    foo
-&lt;/div>
-</code></pre>
-<p>As should this:</p>
-<pre><code>&lt;div>foo&lt;/div>
-</code></pre>
-<p>Now, nested:</p>
-foo<p>This should just be an HTML comment:</p>
-<p>Multiline:</p>
-<p>Code block:</p>
-<pre><code>&lt;!-- Comment -->
-</code></pre>
-<p>Just plain comment, with trailing spaces on the line:</p>
-<p>Code:</p>
-<pre><code>&lt;hr />
-</code></pre>
-<p>Hr's:</p>
-<hr />
-<hr />
-<hr />
-<hr />
-<hr />
-<hr />
-<hr />
-<hr />
-<hr />
-<hr />
 <h1>Inline Markup</h1>
 <p>This is <em>emphasized</em>, and so <em>is this</em>.</p>
 <p>This is <strong>strong</strong>, and so <strong>is this</strong>.</p>

tests/html-reader.native

@@ -183,39 +183,6 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
  ,([Str "Cello",LineBreak,Str "Violoncello"],
    [[Plain [Str "Low",Str "-",Str "voiced",Space,Str "stringed",Space,Str "instrument",Str "."]]])]
 ,HorizontalRule
-,Header 1 [Str "HTML",Space,Str "Blocks"]
-,Para [Str "Simple",Space,Str "block",Space,Str "on",Space,Str "one",Space,Str "line:"]
-,Plain [Str "foo"]
-,Para [Str "And",Space,Str "nested",Space,Str "without",Space,Str "indentation:"]
-,Plain [Str "foobar"]
-,Para [Str "Interpreted",Space,Str "markdown",Space,Str "in",Space,Str "a",Space,Str "table:"]
-,Plain [Str "This",Space,Str "is",Space,Emph [Str "emphasized"],Str ".",Space,Str "And",Space,Str "this",Space,Str "is",Space,Strong [Str "strong"]]
-,Para [Str "Here",Str "'",Str "s",Space,Str "a",Space,Str "simple",Space,Str "block:"]
-,Plain [Str "foo"]
-,Para [Str "This",Space,Str "should",Space,Str "be",Space,Str "a",Space,Str "code",Space,Str "block,",Space,Str "though:"]
-,CodeBlock ("",[],[]) "<div>\n    foo\n</div>"
-,Para [Str "As",Space,Str "should",Space,Str "this:"]
-,CodeBlock ("",[],[]) "<div>foo</div>"
-,Para [Str "Now,",Space,Str "nested:"]
-,Plain [Str "foo"]
-,Para [Str "This",Space,Str "should",Space,Str "just",Space,Str "be",Space,Str "an",Space,Str "HTML",Space,Str "comment:"]
-,Para [Str "Multiline:"]
-,Para [Str "Code",Space,Str "block:"]
-,CodeBlock ("",[],[]) "<!-- Comment -->"
-,Para [Str "Just",Space,Str "plain",Space,Str "comment,",Space,Str "with",Space,Str "trailing",Space,Str "spaces",Space,Str "on",Space,Str "the",Space,Str "line:"]
-,Para [Str "Code:"]
-,CodeBlock ("",[],[]) "<hr />"
-,Para [Str "Hr",Str "'",Str "s:"]
-,HorizontalRule
-,HorizontalRule
-,HorizontalRule
-,HorizontalRule
-,HorizontalRule
-,HorizontalRule
-,HorizontalRule
-,HorizontalRule
-,HorizontalRule
-,HorizontalRule
 ,Header 1 [Str "Inline",Space,Str "Markup"]
 ,Para [Str "This",Space,Str "is",Space,Emph [Str "emphasized"],Str ",",Space,Str "and",Space,Str "so",Space,Emph [Str "is",Space,Str "this"],Str "."]
 ,Para [Str "This",Space,Str "is",Space,Strong [Str "strong"],Str ",",Space,Str "and",Space,Str "so",Space,Strong [Str "is",Space,Str "this"],Str "."]
@@ -293,7 +260,7 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
 ,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","title preceded by a tab"),Str "."]
 ,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","title with \"quotes\" in it")]
 ,Para [Link [Str "URL",Space,Str "and",Space,Str "title"] ("/url/","title with single quotes")]
-,Plain [Str "Email",Space,Str "link",Space,Str "(nobody",Space,Str "[at]",Space,Str "nowhere",Str ".",Str "net)"]
+,Para [Str "Email",Space,Str "link",Space,Str "(nobody",Space,Str "[at]",Space,Str "nowhere",Str ".",Str "net)"]
 ,Para [Link [Str "Empty"] ("",""),Str "."]
 ,Header 2 [Str "Reference"]
 ,Para [Str "Foo",Space,Link [Str "bar"] ("/url/",""),Str "."]
@@ -319,7 +286,7 @@ Pandoc (Meta {docTitle = [Str "Pandoc",Space,Str "Test",Space,Str "Suite"], docA
  [[Plain [Str "In",Space,Str "a",Space,Str "list?"]]
  ,[Plain [Link [Str "http://example",Str ".",Str "com/"] ("http://example.com/","")]]
  ,[Plain [Str "It",Space,Str "should",Str "."]]]
-,Plain [Str "An",Space,Str "e",Str "-",Str "mail",Space,Str "address:",Space,Str "nobody",Space,Str "[at]",Space,Str "nowhere",Str ".",Str "net"]
+,Para [Str "An",Space,Str "e",Str "-",Str "mail",Space,Str "address:",Space,Str "nobody",Space,Str "[at]",Space,Str "nowhere",Str ".",Str "net"]
 ,BlockQuote
  [Para [Str "Blockquoted:",Space,Link [Str "http://example",Str ".",Str "com/"] ("http://example.com/","")]]
 ,Para [Str "Auto",Str "-",Str "links",Space,Str "should",Space,Str "not",Space,Str "occur",Space,Str "here:",Space,Code ("",[],[]) "<http://example.com/>"]