API generation should remove Matplotlib figure captions #732
Description
This RST results in this Sphinx-generated HTML:
Note the Fig 1 caption and how it has 3 links to download source files for the image:
This is how we render that RST in our platform after running the HTML through our pipeline:
The 3 URLs all are 404s. Instead, we should remove the caption entirely. The 3 URLs are not useful and we don't plan to support them—you can download the image directly. That would only leave the text Fig. 1, which we decided isn't a priority.
--
This impacts the Runtime 0.18 docs in #729, along with some historical docs if you run rg '_downloads' -l, e.g.:
--
Example HTML from Sphinx for a single figure:
<figure class="align-default" id="id1">
<img alt="../_images/fake_provider-1_00.png" class="plot-directive" src="../_images/fake_provider-1_00.png" />
<figcaption>
<p>
<span class="caption-number">Fig. 1 </span>
<span class="caption-text">
(
<a class="reference download internal" download="" href="../_downloads/a640acbc08577560dc62a3c02c6ca2ac/fake_provider-1_00.png">
<code class="xref download docutils literal notranslate"><span class="pre">png</span></code>
</a>
,
<a class="reference download internal" download="" href="../_downloads/98e08086a49350bea51e64248343d7ac/fake_provider-1_00.hires.png">
<code class="xref download docutils literal notranslate"><span class="pre">hires.png</span></code>
</a>
,
<a class="reference download internal" download="" href="../_downloads/684bf35d507376624fcead10d9aedaed/fake_provider-1_00.pdf">
<code class="xref download docutils literal notranslate"><span class="pre">pdf</span></code>
</a>
)
</span>
<a class="headerlink" href="#id1" title="Link to this image">¶</a>
</p>
</figcaption>
</figure>We want to simplify the HTML via a new helper function in processHtml.ts. It should look like this:
<figure class="align-default" id="id1">
<img alt="../_images/fake_provider-1_00.png" class="plot-directive" src="../_images/fake_provider-1_00.png">
</figure>The function needs to use https://cheerio.js.org to remove the <figcaption> but if it matches this selector: figcaption span.caption-text a.download.internal.reference. We don't want to remove any generic figcaption.
We should this in processHtml.test.ts, including that we don't remove generic figcaptions.
The final step is to run npm run regen-apis.