gerbeur/api/remote/providers/youtube.ts

/**
 * Previews YouTube videos, playlists and channels, for
 * `rich-content-service.ts`. Videos and playlists are read from YouTube's
 * oEmbed endpoint, channels from their page's OG tags.
 *
 * @module
 */

import type { RichContent } from "../../../shared/dump.ts";
import type { RichContentProvider } from "./provider.ts";
import { extractOgTag } from "../html.ts";
import { fetchWithTimeout, readTextCapped } from "../fetch.ts";

/**
 * Returns the id of the video a URL points at, on youtu.be, youtube.com or
 * www.youtube.com only; `m.` and `music.` links go to the generic provider.
 */
function extractVideoId(url: string): string | null {
  try {
    const u = new URL(url);
    if (u.hostname === "youtu.be") {
      return u.pathname.slice(1).split("/")[0] || null;
    }
    if (u.hostname === "youtube.com" || u.hostname === "www.youtube.com") {
      if (u.pathname === "/watch" || u.pathname.startsWith("/watch?")) {
        return u.searchParams.get("v");
      }
      if (
        u.pathname.startsWith("/embed/") || u.pathname.startsWith("/shorts/")
      ) {
        return u.pathname.split("/")[2] || null;
      }
    }
  } catch {
    // invalid URL
  }
  return null;
}

/**
 * Returns the `list` parameter of a URL, whatever its host: a link to another
 * site with a `list` parameter is taken for a YouTube playlist.
 */
function extractPlaylistId(url: string): string | null {
  try {
    return new URL(url).searchParams.get("list");
  } catch {
    return null;
  }
}

/**
 * Returns the start time of a YouTube URL in seconds, from `t` (`90`, `90s`,
 * `1h2m3s`) or `start`; null when there is none that parses.
 */
function extractStartSeconds(url: string): number | null {
  let raw: string | null;
  try {
    const u = new URL(url);
    raw = u.searchParams.get("t") ?? u.searchParams.get("start");
  } catch {
    return null;
  }
  if (!raw) return null;

  const plain = /^\d+$/.exec(raw);
  if (plain) return Number(raw);

  const hms = /^(?:(\d+)h)?(?:(\d+)m)?(?:(\d+)s)?$/i.exec(raw);
  if (!hms || (!hms[1] && !hms[2] && !hms[3])) return null;
  return Number(hms[1] ?? 0) * 3600 + Number(hms[2] ?? 0) * 60 +
    Number(hms[3] ?? 0);
}

/**
 * Returns the path of a youtube.com channel URL (`/channel/UC…`, `/@handle`,
 * `/c/name` or `/user/name`); null for any other URL.
 */
function extractChannelPath(url: string): string | null {
  try {
    const u = new URL(url);
    if (u.hostname === "youtube.com" || u.hostname === "www.youtube.com") {
      if (
        u.pathname.startsWith("/channel/") ||
        u.pathname.startsWith("/@") ||
        u.pathname.startsWith("/c/") ||
        u.pathname.startsWith("/user/")
      ) {
        return u.pathname;
      }
    }
  } catch {
    // invalid URL
  }
  return null;
}

/**
 * Fetches the title and thumbnail of `url` from YouTube's oEmbed endpoint;
 * empty when that fails, since the preview can do without them.
 */
async function fetchOEmbed(
  url: string,
): Promise<{ title?: string; thumbnailUrl?: string }> {
  try {
    const oembedUrl = `https://www.youtube.com/oembed?url=${
      encodeURIComponent(url)
    }&format=json`;
    const res = await fetchWithTimeout(oembedUrl);
    if (res.ok) {
      const data = await res.json() as {
        title?: string;
        thumbnail_url?: string;
      };
      return { title: data.title, thumbnailUrl: data.thumbnail_url };
    }
    await res.body?.cancel();
  } catch {
    // YouTube unreachable, or its answer not JSON.
  }
  return {};
}

/**
 * Comes first in the service's list. A video's embed keeps the link's playlist
 * and start time, a playlist embeds as a `videoseries`, and a channel gets no
 * embed. Its `fetch` never rejects: whatever YouTube doesn't answer is left
 * out.
 */
export const youtubeProvider: RichContentProvider = {
  name: "youtube",

  matches(url: string): boolean {
    return (
      extractVideoId(url) !== null ||
      extractPlaylistId(url) !== null ||
      extractChannelPath(url) !== null
    );
  },

  async fetch(url: string): Promise<RichContent> {
    const videoId = extractVideoId(url);
    const listId = extractPlaylistId(url);
    const channelPath = extractChannelPath(url);

    if (channelPath && !videoId) {
      // oEmbed doesn't take channel URLs.
      let title: string | undefined;
      let thumbnailUrl: string | undefined;
      try {
        const res = await fetchWithTimeout(url);
        if (res.ok) {
          const html = await readTextCapped(res);
          title = extractOgTag(html, "title");
          thumbnailUrl = extractOgTag(html, "image");
        } else {
          await res.body?.cancel();
        }
      } catch {
        // YouTube unreachable: the preview goes without.
      }
      return {
        type: "youtube",
        siteName: "YouTube",
        url,
        title,
        thumbnailUrl,
      };
    }

    if (listId && !videoId) {
      const { title, thumbnailUrl } = await fetchOEmbed(url);
      return {
        type: "youtube",
        siteName: "YouTube",
        url,
        title,
        thumbnailUrl,
        embedUrl:
          `https://www.youtube.com/embed/videoseries?list=${listId}&rel=0`,
      };
    }

    // Built, not taken from oEmbed, so it is there even when oEmbed fails.
    const thumbnailUrl = `https://img.youtube.com/vi/${videoId}/hqdefault.jpg`;
    const { title } = await fetchOEmbed(url);

    const embedParams = new URLSearchParams({ rel: "0" });
    if (listId) embedParams.set("list", listId);
    const start = extractStartSeconds(url);
    if (start) embedParams.set("start", String(start));

    return {
      type: "youtube",
      siteName: "YouTube",
      url,
      videoId: videoId!,
      title,
      thumbnailUrl,
      embedUrl: `https://www.youtube.com/embed/${videoId}?${embedParams}`,
    };
  },
};