Nested Root Microformats Aren't Parsed Correctly

[zcdunn]

On my site, I use h-cite for contexts to remote urls (replies, reposts, etc). I have them marked up as described here, where the u-* property of the h-entry is on the same element as the root h-cite and not directly on the link itself. It seems like this library is not getting the url property of the nested h-cite right. It's also incorrectly parsing the category of the outer h-entry as the category of the nested h-cite

For this repost, you can see parse results below that show a repost-of property (at $.items[0].properties.repost-of[0]) with a nested h-cite that contains the correct url:

• pin13
• unmung

Parsed repost-of from pin13/unmung

{
    "type": [
        "h-cite"
    ],
    "properties": {
        "name": [
            "gilest.org: Make the indie web easier"
        ],
        "url": [
            "https://gilest.org/indie-easy.html"
        ]
    },
    "lang": "en-US",
    "value": "https://gilest.org/indie-easy.html"
}

Parsed repost-of from microformats2-elixir

%{
  "properties" => %{
    "category" => ["IndieWeb", "tech", "decentralization"],
    "name" => ["gilest.org: Make the indie web easier"],
    "url" => ["gilest.org: Make the indie web easier"]
  },
  "type" => ["h-cite"],
  "value" => "gilest.org: Make the indie web easier"
}

[ckruse]

Hi there,

thank you for the report and sorry that it took so long to answer. I'm a bit busy currently.

May I add this example to the repository as a test case?

[zcdunn]

Sure, you can add it to the repo. And there's no rush on this; thanks for taking the time to look at it.

[ckruse]

I think I found the problem… it looks like a bug in Floki, the HTML parser I use. The element name for this:

        <a
          href="https://gilest.org/indie-easy.html"
          class="u-url entry__link"
          itemprop="url"
        >

is an a\n instead of just a. I have to investigate further…

[ckruse]

Floki was innocent 🫣 it was my bug.

I fixed it and like to release a new version. Do you mind to check if it works for you now?

[zcdunn]

🎉 Thank you! That solved the url parsing.

It's still parsing the outer h-entry's category property as part of the inner h-cite. Should I open a separate issue for that?

[ckruse]

Sigh.

Both bugs are caused by a workaround for a bug in MochiWeb. It doesn't deal very well with whitespaces, so I have to replace them with their entities. And I do it within tags, too, which changes the parse tree of the document.

I have to rewrite whitespace handling…

[ckruse]

Can you check again? I completely overhauled whitespace handling, and your test case (and the old test cases) now work for me.

[zcdunn]

Yep, it's working for me now. Thank you!

[ckruse]

Cool :-) I just released v 1.0.1