app/lib/podcast/feed.mjs

Functions

Imports

It imports app/lib/publications/entities.mjs. Imported by app/components/home-podcast.tsx, app/lib/podcast/podcast.server.ts, workers/csp.mjs.

Source

1// Pure, so node:test and the CSP builder can import it. Every feed URL is checked against a host
2// allowlist; the audio list is what workers/csp.mjs writes into `media-src`, so the two cannot disagree.
3
4import { decodeEntities } from "../publications/entities.mjs";
5
6export const PODCAST_FEED_URL = "https://germomics.com/feed/podcast/";
7
8export const PODCAST_SITE_URL = "https://germomics.com/";
9
10// `op3.dev` is the feed's analytics prefix and 302s to `media.germomics.com`; CSP checks the redirect
11// target too, so both are listed.
12export const PODCAST_AUDIO_HOSTS = ["op3.dev", "media.germomics.com"];
13
14const EPISODE_PAGE_HOSTS = ["germomics.com", "www.germomics.com"];
15
16/**
17 * @typedef {{
18 *   guid: string,
19 *   title: string,
20 *   link: string,
21 *   publishedAt: string,
22 *   durationSeconds: number | null,
23 *   description: string,
24 *   audioUrl: string,
25 *   audioType: string,
26 *   season: number | null,
27 *   episode: number | null,
28 * }} PodcastEpisode
29 */
30
31/**
32 * The shared decoder, plus ` `: podcast hosts write it in show notes, and every whitespace run
33 * here collapses to one space anyway. Replaced before decoding, so ` ` still reads literally.
34 *
35 * @param {string} text
36 */
37function decodeFeedEntities(text) {
38  return decodeEntities(text.replace(/ /g, " "));
39}
40
41/**
42 * @param {string} xml @param {string} tag
43 */
44function textOf(xml, tag) {
45  const escaped = tag.replace(":", "\\:");
46  const match = xml.match(new RegExp(`<${escaped}(?:\\s[^>]*)?>([\\s\\S]*?)</${escaped}>`));
47  if (!match) return "";
48  const raw = (match[1] ?? "").replace(/<!\[CDATA\[([\s\S]*?)\]\]>/g, "$1");
49  return decodeFeedEntities(raw.replace(/<[^>]*>/g, " ")).replace(/\s+/g, " ").trim();
50}
51
52/** @param {string} xml @param {string} tag @param {string} attr */
53function attrOf(xml, tag, attr) {
54  const element = xml.match(new RegExp(`<${tag.replace(":", "\\:")}\\s[^>]*>`));
55  if (!element) return "";
56  const value = element[0].match(new RegExp(`\\s${attr}="([^"]*)"`));
57  return value ? decodeFeedEntities(value[1] ?? "") : "";
58}
59
60/**
61 * @param {string} value @param {readonly string[]} hosts
62 */
63function allowedHttpsUrl(value, hosts) {
64  try {
65    const url = new URL(value);
66    if (url.protocol !== "https:" || url.username || url.password || url.port) return null;
67    return hosts.includes(url.hostname) ? url.href : null;
68  } catch {
69    return null;
70  }
71}
72
73/**
74 * @param {string} value
75 */
76export function parseDuration(value) {
77  if (!/^\d+(:\d{1,2}){0,2}$/.test(value)) return null;
78  return value.split(":").reduce((total, part) => total * 60 + Number(part), 0);
79}
80
81/** @param {string} value */
82function positiveInt(value) {
83  return /^\d+$/.test(value) && Number(value) > 0 ? Number(value) : null;
84}
85
86/**
87 * @param {string} xml
88 * @returns {PodcastEpisode[]}
89 */
90export function parsePodcastFeed(xml) {
91  /** @type {PodcastEpisode[]} */
92  const episodes = [];
93  for (const [, item = ""] of xml.matchAll(/<item>([\s\S]*?)<\/item>/g)) {
94    const title = textOf(item, "title");
95    const guid = textOf(item, "guid");
96    const link = allowedHttpsUrl(textOf(item, "link"), EPISODE_PAGE_HOSTS);
97    const audioUrl = allowedHttpsUrl(attrOf(item, "enclosure", "url"), PODCAST_AUDIO_HOSTS);
98    const audioType = attrOf(item, "enclosure", "type");
99    const published = new Date(textOf(item, "pubDate"));
100    if (!title || !guid || !link || !audioUrl || !audioType.startsWith("audio/")) continue;
101    if (Number.isNaN(published.getTime())) continue;
102    episodes.push({
103      guid,
104      title,
105      link,
106      publishedAt: published.toISOString(),
107      durationSeconds: parseDuration(textOf(item, "itunes:duration")),
108      description: textOf(item, "description"),
109      audioUrl,
110      audioType,
111      season: positiveInt(textOf(item, "itunes:season")),
112      episode: positiveInt(textOf(item, "itunes:episode")),
113    });
114  }
115  return episodes.sort((a, b) => b.publishedAt.localeCompare(a.publishedAt));
116}