When converting to mediawiki MW headings are prepended with a span

`<span id="HEADING-ID"></span>`, evidently to provide an anchor to

pandoc's original/automatic id so that internal links will still work. However

if there is a `<section>` element[^1] I frequently end up with an identical id

tag on the div resulting from the section element and the automatically

inserted span element, which seems like a bug to me:

``````html

<section id="head-2" class="level2">

<h2>Head 2</h2>

<section id="head-3" class="level3">

<h3>Head 3</h3>

<ol type="1">

<li><p>Li 1</p></li>

<li><p>Li 2</p>

<p>Li 2 para 2</p></li>

<li><p>Li 3</p></li>

</ol>

<p>Text</p>

<ol type="i">

<li>Li i</li>

<li>Li ii</li>

</ol>

</section>

</section>

``````

``````mediawiki

<div id="head-2" class="section level2">

<span id="head-2"></span>

== Head 2 ==

<div id="head-3" class="section level3">

<span id="head-3"></span>

=== Head 3 ===

<ol style="list-style-type: decimal;">

<li><p>Li 1</p></li>

<li><p>Li 2</p>

<p>Li 2 para 2</p></li>

<li><p>Li 3</p></li></ol>

Text

<ol style="list-style-type: lower-roman;">

<li>Li i</li>

<li>Li ii</li></ol>

</div>

</div>

``````

It seems that the only fix currently is to go through the MW source after

conversion and manually remove any offending spans.

I wish the mediawiki writer were smart enough to not insert the span above the

heading if its id would be identical to that of a parent section div. Would that

be possible, and is this enough of a bug to submit an issue?

In the actual use case I modify the HTML input with an HTML editing

script/library before conversion to mediawiki, and as a part of that I add

attributes to the section elements which I want to be there on the div in

the MW output, so removin

[^1]: In my case inserted by pandoc into the HTML now used as source at an

earlier run.