Detect podcast RSS item enclosures (#50)

This commit is contained in:
RissRIce 2026-06-13 23:55:06 -06:00 committed by GitHub
parent be530ab687
commit aa3fdba277
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 29 additions and 2 deletions

View file

@ -29,6 +29,16 @@ describe("feed parsing", () => {
expect(result.sampleItems[0]).toMatchObject({ title: "Launch tiny products", url: "https://example.com/post" }); expect(result.sampleItems[0]).toMatchObject({ title: "Launch tiny products", url: "https://example.com/post" });
}); });
it("detects podcast RSS feeds from item enclosures", () => {
const result = parseFeedDocument(
"https://example.com/podcast.xml",
`<?xml version="1.0"?><rss version="2.0"><channel><title>Audio Show</title><link>https://example.com</link><item><title>Episode 1</title><enclosure url="https://example.com/episode-1.mp3" type="audio/mpeg" length="12345" /></item></channel></rss>`
);
expect(result.ok).toBe(true);
expect(result.kind).toBe("podcast");
});
it("parses Atom and JSON Feed documents", () => { it("parses Atom and JSON Feed documents", () => {
const atom = parseFeedDocument("https://example.com/atom.xml", `<feed><title>Atom Feed</title><entry><title>Entry</title><updated>2026-06-09T00:00:00Z</updated></entry></feed>`); const atom = parseFeedDocument("https://example.com/atom.xml", `<feed><title>Atom Feed</title><entry><title>Entry</title><updated>2026-06-09T00:00:00Z</updated></entry></feed>`);
const json = parseFeedDocument("https://example.com/feed.json", JSON.stringify({ version: "https://jsonfeed.org/version/1.1", title: "JSON Feed", items: [{ title: "Item", url: "https://example.com/item" }] }), "application/feed+json"); const json = parseFeedDocument("https://example.com/feed.json", JSON.stringify({ version: "https://jsonfeed.org/version/1.1", title: "JSON Feed", items: [{ title: "Item", url: "https://example.com/item" }] }), "application/feed+json");

View file

@ -96,7 +96,7 @@ function parseRss(feedUrl: string, rss: Record<string, unknown>): ValidationResu
title: title || "Untitled RSS Feed", title: title || "Untitled RSS Feed",
description: stringValue(channel.description), description: stringValue(channel.description),
homepageUrl: linkValue(channel.link), homepageUrl: linkValue(channel.link),
kind: detectRssKind(channel), kind: detectRssKind(channel, items),
language: stringValue(channel.language), language: stringValue(channel.language),
imageUrl: imageValue(channel.image), imageUrl: imageValue(channel.image),
lastPublishedAt: newestDate([stringValue(channel.lastBuildDate), stringValue(channel.pubDate), ...sampleItems.map((item) => item.publishedAt)]), lastPublishedAt: newestDate([stringValue(channel.lastBuildDate), stringValue(channel.pubDate), ...sampleItems.map((item) => item.publishedAt)]),
@ -132,13 +132,30 @@ function parseAtom(feedUrl: string, feed: Record<string, unknown>): ValidationRe
}; };
} }
function detectRssKind(channel: Record<string, unknown>): FeedKind { function detectRssKind(channel: Record<string, unknown>, items: Record<string, unknown>[]): FeedKind {
if (channel.itunes || channel["itunes:author"] || channel.enclosure) { if (channel.itunes || channel["itunes:author"] || channel.enclosure) {
return "podcast"; return "podcast";
} }
if (items.some(hasPodcastEnclosure)) {
return "podcast";
}
return "blog"; return "blog";
} }
function hasPodcastEnclosure(item: Record<string, unknown>) {
return asArray(item.enclosure)
.filter(isRecord)
.some((enclosure) => {
const type = stringValue(enclosure.type)?.toLowerCase();
const url = stringValue(enclosure.url);
return (
type?.startsWith("audio/") ||
type?.startsWith("video/") ||
/\.(mp3|m4a|mp4|m4v|ogg|oga|wav|aac|flac)(?:[?#].*)?$/i.test(url ?? "")
);
});
}
function imageValue(value: unknown) { function imageValue(value: unknown) {
if (isRecord(value)) { if (isRecord(value)) {
return stringValue(value.url); return stringValue(value.url);