Syndication of note appends a duplicate of a URL contained in the post content

[sentience]

Describe the bug

This post (as shown in the Indieweb web UI), when syndicated to Mastodon, was posted with the URL present in the post (https://kevinyank.com/) also appended to it:

{
  "html": "<p>Plenty of work still to do (especially on the home page), but my site redesign was already looking better than what was public, so I’ve merged my branch. <a href=\"https://kevinyank.com/\">https://kevinyank.com/</a> is now showing my work-in-progress redesign. #indieweb</p>",
  "text": "Plenty of work still to do (especially on the home page), but my site redesign was already looking better than what was public, so I’ve merged my branch. [https://kevinyank.com/](https://kevinyank.com/) is now showing my work-in-progress redesign. #indieweb"
}

Here's the note in my content store.

Here's the resulting Mastodon post.

Expected behaviour

I would have expected the note to be posted to Mastodon as entered into Indiekit, without the URL added to the end.

Environment

• Indiekit version: v1.0.0-beta.1
• Browser: Safari
• Server/host: Private Synology server
• Content store URL: https://github.com/sentience/kevinyank.com
• Publication URL: https://kevinyank.com/notes/kne3f/

[paulrobertlloyd]

This is expected behaviour, sort of! After converting Markdown to HTML, the last href (if present) is appended to the syndicated note. So if you were to post the following:

This is [my new website](https://kevinyank.com/).

this would be syndicated as:

This is my new website. https://kevinyank.com/

Your example is the one case where this approach falls apart! 🙈

I’ve noticed that Micro.blog is able to maintain linked text when it syndicates notes to Mastodon; I quickly experimented with sending HTML, but that spat out raw HTML… clearly there’s a way to send linked text over Mastodon’s API, I just need to research what is the correct means of doing so. I think that’s the solution to the above issue.

[sentience]

For the record, I ran into this issue because I submitted a note with a URL inline, and Indiekit converted it into the Markdown link that you see above. Maybe it shouldn't do that…?

[paulrobertlloyd]

Oh, right. I probably have markdown-it's linkify option on - disabling that would prevent this behaviour (leaving the syndication target to do its own linkify-ing).

[sentience]

In my head it could make sense for Indiekit to linkify my Markdown when converting it to HTML, but I don't expect it to modify the Markdown I give it. Unless my mental model is broken…?

[paulrobertlloyd]

This might be a side-effect of the change made in b64b7d4 to fix #563, released in Beta 1.

There’s a bit of to-and-fro between text and HTML representations, and Indiekit needing to account for one or both values being sent (content.text and/or content.html), or one a content value that could conceivably contain either text or HTML.

There are some tests here (that could possibly be expanded on) that account for this that may help shine a light on the different scenarios being handled:

indiekit/packages/endpoint-micropub/tests/unit/jf2.js

Line 133 in b64b7d4

test("Gets text and HTML values from `content` property", (t) => {

The conversion between text and html values happens here:

indiekit/packages/endpoint-micropub/lib/jf2.js

Line 133 in b68b87f

export const getContentProperty = (properties) => {

[paulrobertlloyd]

Tested syndicating this same update to my private test accounts, given the fix in #573. Here’s what this note looks like:

Mastodon

Twitter