Less/greater than signs in headings appear as &lt/&gt in TOC

[p-ouellette]

That's an mkdocs site with the readthedocs theme. It's using git master HEAD of mkdocs and markdown. The issue was caused by 74c0166. I haven't tested this using just markdown so it might be related to mkdocs.

[facelessuser]

What version of Markdown are you using? I'm not seeing this in the current version.

[waylan]

I just tested the readthedocs theme with the latest on master and do not see the issue.

Note that in the picture you can see the Markdown text, the TOC and the main body. All appear correct.

[p-ouellette]

Reproduced using just markdown 3.2.1:

echo -e "[TOC]\n# `<test>`" | python -m markdown -x toc

<div class="toc">
<ul>
<li><a href="#lttestgt">&amp;lt;test&amp;gt;</a></li>
</ul>
</div>
<h1 id="lttestgt"><code>&lt;test&gt;</code></h1>

With markdown 3.2 the above command outputs:

<div class="toc">
<ul>
<li><a href="#lttestgt">&lt;test&gt;</a></li>
</ul>
</div>
<h1 id="lttestgt"><code>&lt;test&gt;</code></h1>

[waylan]

Ah, I missed that that was a code span. It would appear that code spans were not being considered in this comment. We do need the smart escape after all.

_{Sent with GitHawk}

[facelessuser]

Okay, I guess my results were different as I'm using pymdownx.inlinehilite. It uses the htmlstash instead of AtomicString. I guess there is a difference in how those get handled...

[facelessuser]

I guess htmlstash must work while AtomicString does not due to the fact that the htmlstash placeholders don't get cleaned up until after the toc is generated. I imagine it just duplicates the placeholder, and the placeholders get replaced with the content post toc.

So yeah, we generally need a solution when handling AtomicStrings. I assume we can just detect AtomicString types and not escape them? I don't think we specifically need a smart escape.

[facelessuser]

I'll take a stab at it.

[facelessuser]

Yeah, I think all we really need is something like this:

def get_name(el):
    """Get title name."""

    text = []
    for c in el.itertext():
        if isinstance(c, AtomicString):
            text.append(html.unescape(c))
        else:
            text.append(c)
    name = ''.join(text).strip()
    return name

This prevents us from getting weird ids such as lttestgt Which we previously got from `<text>`. And it returns a string that we can handle sanely. It just unescapes Atomic strings with html.unescape which uses html5 rules for entities (py34+).