Original post discovery: support for rel=alternate type=application/mf2+json

[cybeardjm]

Hi,

Been playing with the various IndieWeb and syndication features for WordPress today.

As it seems the old “Microformats2” plugin is not maintained anymore (doesn’t seem to work nice with the Twenty Twenty-Three theme anyway), I’m testing the MF2 Feeds plugin instead.

From the description: "Provides a Microformats2 JSON “Feed” for every WordPress URL, and helps to allow other sites to get pre-parsed Microformats-JSON even if the theme does not support Microformats2."

It generates these alternate links for posts:

<link rel="alternate" type="application/mf2+json" href="https://domain/sample-post/feed/mf2/" />
<link rel="alternate" type="application/jf2+json" href="https://domain/sample-post/feed/jf2/" />

But, they're not supported for discovery (so far, it seems) as alternate means to check for microformats in a post.
=> when I test a preview to publish on Mastodon, I get errors such as "Mastodon doesn't support type(s) h-cite, or no content was found." or "No h-entry found" (on a https://indiewebify.me/validate-h-entry/ test)

[snarfed]

Hmm, good idea! Right now we only support separate feeds with rel=feed, and only HTML: https://brid.gy/about#link . For other types like this, we probably wouldn't look for these URL paths specifically, and I doubt we'd add JF2 support, but we could maybe ingest MF2 via rel=alternate type=application/mf2+json.

Not a priority for me personally right now, but this would be a pretty reasonable first project if you want to contribute. Dev docs at https://bridgy.readthedocs.io/ , code for this would go somewhere around:

bridgy/original_post_discovery.py

Lines 322 to 348 in 6d5b0cb

	candidates = (author_mf2['rels'].get('feed', []) +
	[a.get('url') for a in author_mf2.get('alternates', [])
	if a.get('type') == MF2_HTML_MIME_TYPE])
	for feed_url in candidates:
	# check that it's html, not too big, etc
	feed_url, _, feed_ok = util.get_webmention_target(feed_url)
	if feed_url == author_url:
	logger.debug('author url is the feed url, ignoring')
	elif not feed_ok:
	logger.debug("skipping feed since it's not HTML or otherwise bad")
	else:
	feed_urls.add(feed_url)
	for feed_url in feed_urls:
	try:
	logger.debug(f"fetching author's rel-feed {feed_url}")
	feed_mf2 = util.fetch_mf2(feed_url)
	if not feed_mf2:
	logger.debug('nothing found')
	continue
	feeditems = _merge_hfeeds(feeditems, _find_feed_items(feed_mf2))
	domain = util.domain_from_link(feed_url)
	if source.updates is not None and domain not in source.domains:
	domains = source.updates.setdefault('domains', source.domains)
	if domain not in domains:
	logger.info(f'rel-feed found new domain {domain}! adding to source')
	domains.append(domain)

[cybeardjm]

Thx @snarfed - unfortunately, am no dev, just an "advanced user", so wouldn't even try to mess with the code...
Ok, will wait till it works some day, will stick with ActivityPub 100% in the mean time...

[snarfed]

Np, we can keep it open!

[dshanske]

I have another option we use in our Webmention plugin using the WordPress rest api... but someone would still need to implement.

[cybeardjm]

OK, been playing with various plugins all day long, including Webmention, from various Fediverse apps...
Thx

[benharri]

This would be very handy as I haven't figured out a good way to get bridgy to read my wordpress feed. Might have a go at adding mf2+json support