Incorrect parsing un-indented HTML fragment

[wjordan]

I came across an issue parsing an un-indented HTML fragment, where the parser seems to be incorrectly treating part of the HTML as a block of paragraph-text (wrapped in a <p> tag and quote-escaped). Here's a minimal-repro test case where the issue should be self-explanatory:

$ ruby -rredcarpet -e "puts Redcarpet::Markdown.new(Redcarpet::Render::HTML).render(ARGF.read)" <<MD
<div>
<div>
</div>
</div>
<script>
  x = "hello";
</script>
MD
<div>
<div>
</div>

<p></div>
<script>
  x = &quot;hello&quot;;
</script></p>

Compare against:

• commonmarker:

$ ruby -rcommonmarker -e "puts CommonMarker.render_html(ARGF.read)" <<MD
<div>
<div>
</div>
</div>
<script>
  x = "hello";
</script>
MD
<div>
<div>
</div>
</div>
<script>
  x = "hello";
</script>

• kramdown:

$ ruby -rkramdown -e "puts Kramdown::Document.new(ARGF.read).to_html" <<MD
<div>
<div>
</div>
</div>
<script>
  x = "hello";
</script>
MD
<div>
<div>
</div>
</div>
<script>
  x = "hello";
</script>

• bluecloth:

$ ruby -rbluecloth -e "puts BlueCloth.new(ARGF.read).text" <<MD
<div>
<div>
</div>
</div>
<script>
  x = "hello";
</script>
MD
<div>
<div>
</div>
</div>
<script>
  x = "hello";
</script>