HTML to Atom: title element has encoded HTML

[gRegorLove]

Minor, but I noticed with this feed that one of my notes has encoded HTML getting into the <title>:

<title>I like this CSS image reset after watching Kevin Powell’s walkthrough. &lt;...</title>

The original post uses e-content and I'm guessing Granary is using the parsed html attribute for the title. A possible solution might be to use value attribute instead, for posts that don't have a name of course:

    "content": [
        {
            "html": "<p>I like this CSS image reset after watching <a href=\"https://www.youtube.com/watch?v=345V2MU3E_w\">Kevin Powell&#x2019;s walkthrough</a>.</p>\n\n<p>Also intrigued by the post he linked, &#x201C;<a class=\"h-cite\" href=\"https://csswizardry.com/2023/09/the-ultimate-lqip-lcp-technique/\">The Ultimate Low-Quality Image Placeholder Technique</a>.&#x201D;</p>",
            "value": "I like this CSS image reset after watching Kevin Powell\u2019s walkthrough.\nAlso intrigued by the post he linked, \u201cThe Ultimate Low-Quality Image Placeholder Technique.\u201d",
            "lang": "en"
        }
    ],

[gRegorLove]

Oh I just realized it's not the first youtube <a href> getting encoded, but some later HTML element. I'm not sure then. Let me know if something is off with my HTML that's causing it then.

[snarfed]

Huh! Thanks for the nudge. This is an odd one! It doesn't reproduce locally for me at all, with either the /stream/ feed or the specific post; both have <title>I like this CSS image reset after watching Kevin Powell’s walkthrough. </title>, no <.... I do see it on prod https://granary.io/ with both though. Hrm.

[snarfed]

Looked at this again, I'm now able to reproduce it locally. Output is still a bit different, I'm guessing that's because I'm using a different HTML parser locally vs in prod.

I suspect we're generating title from HTML content, then ellipsizing, and we end up with just the opening < of a tag, which we then entity-encode.

[snarfed]

Fixed!