thibault/plankton
1
0
Fork 0
Code Issues Pull requests 1 Projects Releases Packages 1 Wiki Activity Actions 1

feat(api): read rss results from indexers #1

Merged
thibault merged 1 commit from feat/indexer-rss into 0.2 2026-10-08 18:30:24 +02:00
Conversation 0 Commits 1 Files changed 11 +608 -50
thibault commented 2026-10-08 18:03:11 +02:00
Owner
Copy link

What

PR D of the indexer engine: the rss results type.

  • src/indexers/rss.rs: reads a feed with quick-xml's plain Reader (no serde, no namespace resolution, so names stay as written: ex:seeders). Each <item> (RSS 2.0 / 1.0) or <entry> (Atom) becomes a small JSON value: an element is its trimmed text (CDATA, predefined entities and character references decoded), or, when it has attributes or children, an object of its children by qualified name, @attr keys and its own text under #text. Repeated elements: the first one wins. The root must be rss, feed or RDF; malformed XML, unclosed elements or another document (an HTML page) is an error.
  • extract::rss runs the existing field reading on each item, so sizes, counts, dates (RFC 2822 pubDate), infohash from the field or the magnet, magnet built from infohash + trackers, and the "left out" count all work unchanged. lookup learns the @attr segment, and a path leading to an element with attributes reads its #text.
  • The engine fetches rss with the same limits, cache and filters; the "rss is not supported yet" skip is gone, so /api/indexers/check reports rss indexers like json ones.
  • .torrent links stay plain links: nothing downloads them, an item needs an infohash or a magnet holding one.

Tests

  • rss.rs: namespaced elements, attributes, CDATA, entities, repeated elements, Atom entries, malformed / unclosed / HTML / empty answers.
  • Fake indexer: /rss (namespaced infohash, CDATA title, 1.4 GiB, RFC 2822 date, .torrent enclosure via enclosure.@url, an item with only a base32 magnet in <link>, an item with neither), /rss-broken (malformed), /rss-named (the fake catalog's releases as a feed). An rss indexer pointed at an HTML page now fails instead of being skipped.
  • API: an anime search through an rss indexer keeps the batch and the season pack and drops the single episode, as from JSON.

124 tests pass; fmt and clippy clean.

Docs

docs/INDEXERS.md: status line, a complete fictional RSS feed and definition, an "RSS results" section (paths, attributes, decoding, first wins, UTF-8, Atom), new troubleshooting errors. README mentions RSS; ROADMAP phase 7 gets an rss step ✅.

New dependency

quick-xml = "0.41": already in Cargo.lock through librqbit (librqbit-upnp), same version, so nothing new is downloaded or built. A small, fast, widely used XML pull parser; the plain reader keeps the code short and gives prefixed names as written, which definitions use.

🤖 Generated with Claude Code

## What PR D of the indexer engine: the `rss` results type. - `src/indexers/rss.rs`: reads a feed with quick-xml's plain `Reader` (no serde, no namespace resolution, so names stay as written: `ex:seeders`). Each `<item>` (RSS 2.0 / 1.0) or `<entry>` (Atom) becomes a small JSON value: an element is its trimmed text (CDATA, predefined entities and character references decoded), or, when it has attributes or children, an object of its children by qualified name, `@attr` keys and its own text under `#text`. Repeated elements: the first one wins. The root must be `rss`, `feed` or `RDF`; malformed XML, unclosed elements or another document (an HTML page) is an error. - `extract::rss` runs the existing field reading on each item, so sizes, counts, dates (RFC 2822 `pubDate`), infohash from the field or the magnet, magnet built from infohash + trackers, and the "left out" count all work unchanged. `lookup` learns the `@attr` segment, and a path leading to an element with attributes reads its `#text`. - The engine fetches rss with the same limits, cache and filters; the "rss is not supported yet" skip is gone, so `/api/indexers/check` reports rss indexers like json ones. - `.torrent` links stay plain links: nothing downloads them, an item needs an infohash or a magnet holding one. ## Tests - rss.rs: namespaced elements, attributes, CDATA, entities, repeated elements, Atom entries, malformed / unclosed / HTML / empty answers. - Fake indexer: `/rss` (namespaced infohash, CDATA title, `1.4 GiB`, RFC 2822 date, `.torrent` enclosure via `enclosure.@url`, an item with only a base32 magnet in `<link>`, an item with neither), `/rss-broken` (malformed), `/rss-named` (the fake catalog's releases as a feed). An rss indexer pointed at an HTML page now fails instead of being skipped. - API: an anime search through an rss indexer keeps the batch and the season pack and drops the single episode, as from JSON. 124 tests pass; fmt and clippy clean. ## Docs docs/INDEXERS.md: status line, a complete fictional RSS feed and definition, an "RSS results" section (paths, attributes, decoding, first wins, UTF-8, Atom), new troubleshooting errors. README mentions RSS; ROADMAP phase 7 gets an `rss` step ✅. ## New dependency `quick-xml = "0.41"`: already in Cargo.lock through librqbit (librqbit-upnp), same version, so nothing new is downloaded or built. A small, fast, widely used XML pull parser; the plain reader keeps the code short and gives prefixed names as written, which definitions use. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
feat(api): read rss results from indexers
Some checks failed
CI / test (pull_request) Has been cancelled
Details
CI / web (pull_request) Has been cancelled
Details
0bd815ff04 
Indexers whose results type is rss are now searched and checked like json
ones. Each <item> (or Atom <entry>) of the feed becomes a small JSON value,
read with the same paths, format inference, magnet building and filters.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01947SGTYxD1CJLsk2PcULRA
ci: trigger the checks again
Some checks failed
CI / test (pull_request) Has been cancelled
Details
CI / web (pull_request) Has been cancelled
Details
5d7533f920 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
thibault force-pushed feat/indexer-rss from 5d7533f920
Some checks failed
CI / test (pull_request) Has been cancelled
Details
CI / web (pull_request) Has been cancelled
Details
to 0bd815ff04
Some checks failed
CI / test (pull_request) Has been cancelled
Details
CI / web (pull_request) Has been cancelled
Details
2026-10-08 18:19:25 +02:00 Compare
thibault force-pushed feat/indexer-rss from 0bd815ff04
Some checks failed
CI / test (pull_request) Has been cancelled
Details
CI / web (pull_request) Has been cancelled
Details
to 46b732d124
All checks were successful
CI / web (pull_request) Successful in 7s
Details
CI / test (pull_request) Successful in 25s
Details
2026-10-08 18:19:48 +02:00 Compare
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
thibault/plankton!1
No description provided.