jeremy.herve.bzh / skypress
A calm place to write long-form, and publish it to the open social web. skypress.blog/
0
Fork 0

Configure Feed

Select the types of activity you want to include in your feed.

skypress / docs / specs / sp11-rss-feeds.md
at 72366dcee8fd0ea8de4e979a656dfa9a500f0bdb 3.3 kB View raw View code
Jeremy Herve Add full-content RSS feeds, one per publication
29731bb4

SP11 — Full-content RSS feeds (one per publication)#

  • Status: Built
  • Depends on: SP4 (public renderer), SP10 (publication system)
  • Decisions: 0011 (full-content feeds, hand-rolled builder, route, discovery)

Goal#

Give every publication a machine-readable, full-content RSS feed so readers can subscribe. The feed carries the whole sanitised article HTML, not just a summary.

Route#

skypress.blog/@handle/{slug}/rss.xml   → RSS 2.0 feed for that publication

A read-through server endpoint (prerender = false) in src/pages/[author]/[slug]/rss.xml.ts, mirroring the publication page's resolution and error behaviour:

  1. Validate @handle + slug; 404 otherwise.
  2. resolveAuthor(handle) → { did, pdsUrl }; 404 if unresolvable.
  3. resolveReaderPublication(pdsUrl, did, slug); 404 if unknown.
  4. listRecords(site.standard.document, 100) — one SSRF-guarded PDS fetch; no per-article round-trip (records already include content.blocks + textContent).
  5. buildPublicationFeedXml(...) → XML.
  6. Respond application/rss+xml; charset=utf-8, cache-control: public, max-age=300.

Feed contents#

  • Channel: publication name (title), home URL (link), description, a self-referencing atom:link, and lastBuildDate from the newest item.
  • Items (newest 20, by publishedAt, newest-first; undated documents skipped):
    • title, link + guid = canonical article URL (/@handle/{slug}/{rkey}),
    • pubDate (RFC-822, via Date.toUTCString()),
    • description = stored description or a 200-char textContent fallback,
    • content:encoded = full sanitised article HTML in a CDATA section.

Pipeline reuse (no drift, guardrails by reuse)#

The item body uses the exact reader pipeline: resolveBlobImageUrls → renderBlocks (src/lib/blocks/render.ts) → sanitizeArticleHtml (src/lib/reader/sanitize.ts). So:

  • HTML is sanitised before output (AGENTS.md #6b) — same code as the article page;
  • the route's PDS fetch is SSRF-guarded via listRecords/safeFetch (AGENTS.md #6a);
  • reading-page fidelity and feed fidelity can never diverge.

Modules#

  • src/lib/feed/rss.ts — pure, dependency-free RSS 2.0 builder. Owns XML + CDATA escaping.
  • src/lib/feed/publication-feed.ts — pure transform: filter to this publication, sort, cap, render each item, assemble channel. Network-free, fully unit-testable.
  • src/pages/[author]/[slug]/rss.xml.ts — the endpoint (fetch + delegate).

Discovery#

  • Autodiscovery <link rel="alternate" type="application/rss+xml" title="{name} — RSS"> in the publication page and every article page <head>.
  • A visible "RSS" link in the publication hero.

Tests (TDD; escaping is test-locked)#

  • rss.test.ts — document shape, namespaces, channel/item fields, XML escaping, the ]]> CDATA-split, RFC-822 dates, ordering, empty channel.
  • publication-feed.test.ts — ownership filter, newest-first sort, 20-item cap, undated skip, full HTML in content:encoded, canonical links + self feed URL, description fallback, empty publication.

Out of scope#

  • Atom/JSON Feed (RSS 2.0 only for v1).
  • Per-author or cross-publication feeds (no clear channel identity — Decision 0011).
  • Pagination / <lastBuildDate>-based conditional GETs.