radio-explorer/server/sources/iconScraper.js

// Best-effort icon resolver for radio stations.
// Order:
//   1. Radio-Browser favicon by exact-ish name (only if station.source !== 'radiobrowser', else
//      we already have it).
//   2. Scrape <link rel="icon">, <link rel="apple-touch-icon">, <meta property="og:image">
//      from the homepage HTML.
//   3. HEAD-probe /favicon.ico at the homepage origin.
// Returns the best absolute URL found, or null.

const UA = 'OnlineRadioExplorer/0.1 (+icon-scraper)';
const FETCH_TIMEOUT_MS = 8000;
const MAX_HTML_BYTES = 256 * 1024;
const RB_BASE = 'https://de1.api.radio-browser.info';

function withTimeout(ms) {
  const ctl = new AbortController();
  const t = setTimeout(() => ctl.abort(), ms);
  return { signal: ctl.signal, done: () => clearTimeout(t) };
}

async function fetchText(url) {
  const t = withTimeout(FETCH_TIMEOUT_MS);
  try {
    const res = await fetch(url, {
      headers: { 'User-Agent': UA, 'Accept': 'text/html,application/xhtml+xml' },
      redirect: 'follow',
      signal: t.signal
    });
    if (!res.ok) return null;
    const reader = res.body?.getReader();
    if (!reader) return null;
    let received = 0;
    const chunks = [];
    while (true) {
      const { done, value } = await reader.read();
      if (done) break;
      received += value.length;
      chunks.push(value);
      if (received >= MAX_HTML_BYTES) { try { await reader.cancel(); } catch {} break; }
    }
    return Buffer.concat(chunks.map((c) => Buffer.from(c))).toString('utf8');
  } catch {
    return null;
  } finally { t.done(); }
}

async function head(url) {
  const t = withTimeout(FETCH_TIMEOUT_MS);
  try {
    const res = await fetch(url, { method: 'HEAD', headers: { 'User-Agent': UA }, signal: t.signal, redirect: 'follow' });
    return res.ok;
  } catch { return false; } finally { t.done(); }
}

function abs(base, href) {
  if (!href) return null;
  try { return new URL(href, base).toString(); } catch { return null; }
}

// Extract candidate icon URLs from raw HTML. Returns array of { href, size } sorted best-first.
function parseIconCandidates(html, baseUrl) {
  const out = [];
  // <link rel="...icon..." href="..." sizes="...">
  const linkRe = /<link\b([^>]*?)\/?>/gi;
  let m;
  while ((m = linkRe.exec(html))) {
    const attrs = m[1];
    const rel = (/\brel\s*=\s*["']([^"']+)["']/i.exec(attrs) || [])[1] || '';
    if (!/icon/i.test(rel)) continue;
    const href = (/\bhref\s*=\s*["']([^"']+)["']/i.exec(attrs) || [])[1];
    if (!href) continue;
    const sizes = (/\bsizes\s*=\s*["']([^"']+)["']/i.exec(attrs) || [])[1] || '';
    const sz = parseInt((/(\d+)x\d+/.exec(sizes) || [])[1] || '0', 10);
    const apple = /apple-touch-icon/i.test(rel) ? 64 : 0; // bias: apple-touch-icons usually larger PNGs
    const u = abs(baseUrl, href);
    if (u) out.push({ href: u, score: sz + apple });
  }
  // <meta property="og:image" content="...">
  const metaRe = /<meta\b([^>]*?)\/?>/gi;
  while ((m = metaRe.exec(html))) {
    const attrs = m[1];
    const prop = (/\b(?:property|name)\s*=\s*["']([^"']+)["']/i.exec(attrs) || [])[1] || '';
    if (!/^og:image|^twitter:image/i.test(prop)) continue;
    const content = (/\bcontent\s*=\s*["']([^"']+)["']/i.exec(attrs) || [])[1];
    const u = abs(baseUrl, content);
    if (u) out.push({ href: u, score: 200 }); // og:image preferred
  }
  out.sort((a, b) => b.score - a.score);
  // de-dupe preserving order
  const seen = new Set();
  return out.filter((c) => (seen.has(c.href) ? false : (seen.add(c.href), true)));
}

async function fromRadioBrowserByName(name) {
  if (!name) return null;
  try {
    const url = `${RB_BASE}/json/stations/search?name=${encodeURIComponent(name)}&limit=3&hidebroken=true&order=clickcount&reverse=true`;
    const t = withTimeout(FETCH_TIMEOUT_MS);
    const res = await fetch(url, { headers: { 'User-Agent': UA }, signal: t.signal });
    t.done();
    if (!res.ok) return null;
    const list = await res.json();
    const target = name.toLowerCase().trim();
    const exact = list.find((s) => (s.name || '').toLowerCase().trim() === target);
    const pick = exact || list[0];
    if (pick?.favicon) return pick.favicon;
  } catch {}
  return null;
}

async function fromHomepage(homepage) {
  if (!homepage) return null;
  let base;
  try { base = new URL(homepage); } catch { return null; }
  const html = await fetchText(base.toString());
  if (html) {
    const cands = parseIconCandidates(html, base.toString());
    for (const c of cands) {
      if (await head(c.href)) return c.href;
    }
  }
  // last resort: /favicon.ico
  const ico = `${base.origin}/favicon.ico`;
  if (await head(ico)) return ico;
  return null;
}

/**
 * Try to find an icon URL for a station.
 * @param {{ name?: string, homepage?: string|null, source?: string }} station
 * @returns {Promise<string|null>}
 */
export async function scrapeIcon(station) {
  if (!station) return null;
  // For non-RB stations, RB often still has an entry → cheap win.
  if (station.source !== 'radiobrowser') {
    const rb = await fromRadioBrowserByName(station.name);
    if (rb) return rb;
  }
  return fromHomepage(station.homepage);
}