gerbeur/api/remote/bandcamp-streams.ts

/**
 * Resolves a Bandcamp page to its streamable tracks when the native player
 * starts it, for `GET /api/bandcamp/tracks`. The stream URLs are signed and
 * expire 24 hours after the page is fetched, so unlike the embed URL they
 * can't be stored in the dump's preview. The browser plays them from bcbits
 * directly.
 *
 * @module
 */

import { APIException } from "../lib/api-exception.ts";
import { fetchWithTimeout, isValidHttpUrl, readTextCapped } from "./fetch.ts";
import { BANDCAMP_HOST_RE, parseTralbum } from "./bandcamp-tralbum.ts";
import type { Tralbum } from "../../shared/player.ts";

/**
 * Half the signatures' life, so a cached URL still has at least 12 hours left
 * when it is handed out.
 */
const TTL_MS = 12 * 60 * 60 * 1000;
/** The oldest write goes first. */
const MAX_ENTRIES = 200;

const cache = new Map<string, Tralbum>();

/**
 * Returns `url` without its query, fragment and trailing slashes, and with its
 * host in lowercase.
 */
function cacheKey(url: string): string {
  try {
    const u = new URL(url);
    return `${u.protocol}//${u.hostname.toLowerCase()}${
      u.pathname.replace(/\/+$/, "")
    }`;
  } catch {
    return url;
  }
}

function readCache(key: string): Tralbum | null {
  const hit = cache.get(key);
  if (!hit) return null;
  if (Date.now() - hit.resolvedAt > TTL_MS) {
    cache.delete(key);
    return null;
  }
  return hit;
}

function writeCache(key: string, value: Tralbum): void {
  // Map preserves insertion order, so the first key is the oldest write.
  if (cache.size >= MAX_ENTRIES) {
    const oldest = cache.keys().next().value;
    if (oldest !== undefined) cache.delete(oldest);
  }
  cache.set(key, value);
}

/**
 * Returns the tracklist of the Bandcamp page `pageUrl`, from the cache when it
 * was resolved in the last 12 hours. `force` skips the cache: the player sets
 * it when it re-resolves a queue whose URLs expired or are about to, which a
 * cached entry could hand back.
 *
 * @throws {APIException} 400 when `pageUrl` isn't a public bandcamp.com URL,
 * 502 when Bandcamp can't be reached or doesn't answer with a page, 404 when
 * the page has nothing streamable.
 */
export async function resolveBandcamp(
  pageUrl: string,
  { force = false }: { force?: boolean } = {},
): Promise<Tralbum> {
  if (!isValidHttpUrl(pageUrl)) {
    throw new APIException(400, "Invalid URL");
  }

  let hostname: string;
  try {
    hostname = new URL(pageUrl).hostname;
  } catch {
    throw new APIException(400, "Invalid URL");
  }
  if (!BANDCAMP_HOST_RE.test(hostname)) {
    throw new APIException(400, "Not a Bandcamp URL");
  }

  const key = cacheKey(pageUrl);
  if (!force) {
    const hit = readCache(key);
    if (hit) return hit;
  }

  let res: Response;
  try {
    res = await fetchWithTimeout(pageUrl, 8000);
  } catch {
    throw new APIException(502, "Could not reach Bandcamp");
  }
  // A body left unread keeps its connection open until GC, and this endpoint
  // is public: every early return cancels it.
  if (!res.ok) {
    await res.body?.cancel();
    throw new APIException(502, `Bandcamp returned ${res.status}`);
  }
  if (!(res.headers.get("content-type") ?? "").startsWith("text/html")) {
    await res.body?.cancel();
    throw new APIException(502, "Bandcamp did not return a page");
  }

  const parsed = parseTralbum(await readTextCapped(res), pageUrl);
  if (!parsed) {
    // No data, no audio or nothing streamable yet (a preorder). The client
    // plays the embed instead.
    throw new APIException(404, "No streamable tracks on this page");
  }

  writeCache(key, parsed);
  return parsed;
}