Bug: Feed entries with <description> element don't have a description

[reynir]

RSS feeds with a <description> element are not populated with a description. See e.g.: https://github.com/ocaml/ocaml.org/blob/ff7b4c3457b895fe9be83f3df251709558900bfd/data/planet/robur.coop/miragevpn--tls-crypt-v2.md?plain=1#L3

https://blog.robur.coop/feed.xml

It seems River fetches the page and scrapes html meta headers instead of the nice <description></description> served in the feed. For atom feeds I believe the <summary></summary> element should be used.

[sabine]

Looks like it.

The upstream dependency river (https://github.com/kayceesrk/river) needs to be extended to properly handle extracting descriptions from the feed.

In river's post.ml, we need to add a new field description : string option that is populated from the <description>, or, respectively <summary> tags from the RSS2.0, respectively Atom feed.

Also, we should consider whether we still need to fall back to the metadata if these tags are not present.

[sabine]

This needs to be addressed upstream in the dependency river. Here's a corresponding issue: https://github.com/tarides/river/issues/12

[0xRamsi]

Not sure if this is the same bug or just related.

I use QuiteRSS, and I see entries and titles, but no content/description, and further, the link is just the link to another feed. E.g. https://tarides.com/feed.xml or https://alan.petitepomme.net/cwn/cwn.rss instead of getting to the actual article. So all I can do, is subscribe to the other feed.

[reynir]

I think it is different (but maybe related). This issue is about how the rss/atom feeds are scraped and turned into markdown documents with metadata. I don't know how the markdown files are turned into the rss feed and how that might affect what you observe.

[0xRamsi]

I looked into the feed. I believe it's (mostly) an issue with QuiteRSS. I open an issue there: https://github.com/QuiteRSS/quiterss/issues/1603