veodl 1.9.0 → 1.10.0

package/src/source-discovery.js

@@ -5,8 +5,44 @@ import path from 'node:path';
 import { setTimeout as delay } from 'node:timers/promises';
 import { cleanText, validateUrl } from './utils.js';
 
-const MEDIA_URL = /\.(?:m3u8|mpd|mp4|webm|mov)(?:[?#]|$)/i;
-const MEDIA_TYPE = /(?:mpegurl|dash\+xml|video\/(?:mp4|webm|quicktime))/i;
+const MEDIA_URL = /\.(?:m3u8|mpd|mp4|webm|mov)(?:[?#]|$)/i;
+const MEDIA_TYPE = /(?:mpegurl|dash\+xml|video\/(?:mp4|webm|quicktime))/i;
+const PLAY_CONTROL = String.raw`(() => {
+  const visited = new WeakSet();
+  const state = globalThis[Symbol.for('veo.playControls')] ||= { visited, clicks: 0 };
+  const labelMatches = node => {
+    const label = [node.getAttribute('aria-label'), node.getAttribute('title'), node.getAttribute('data-testid'),
+      node.getAttribute('value'), node.textContent].find(value => value && value.trim())?.trim() || '';
+    return /^(?:play|watch|start|replay|abspielen|ansehen|wiedergabe)(?:\s+(?:video|movie|film|now))?$/i.test(label)
+      || /(?:^|[-_\s])(?:big[-_])?play(?:[-_\s]|$)/i.test(node.className?.baseVal || node.className || '');
+  };
+  const documents = [document];
+  for (let index = 0; index < documents.length && index < 16; index++) {
+    for (const frame of documents[index].querySelectorAll('iframe')) {
+      try { if (frame.contentDocument && !documents.includes(frame.contentDocument)) documents.push(frame.contentDocument); } catch {}
+    }
+  }
+  for (const doc of documents) {
+    const controls = doc.querySelectorAll('button,[role="button"],[aria-label],[title],[data-testid],input[type="button"],input[type="image"],[class*="play"],[class*="Play"]');
+    for (const node of controls) {
+      if (state.clicks >= 8) return false;
+      if (state.visited.has(node) || node.disabled || node.getAttribute('aria-disabled') === 'true' || !labelMatches(node)) continue;
+      const style = doc.defaultView.getComputedStyle(node);
+      if (style.display === 'none' || style.visibility === 'hidden' || style.pointerEvents === 'none' || !node.getClientRects().length) continue;
+      state.visited.add(node);
+      state.clicks++;
+      node.click();
+      return true;
+    }
+    for (const video of doc.querySelectorAll('video')) {
+      if (state.visited.has(video) || !video.paused) continue;
+      state.visited.add(video);
+      video.play().catch(() => {});
+      return true;
+    }
+  }
+  return false;
+})()`;
 
 function abortable(promise, signal) {
   if (!signal) return promise;
@@ -149,19 +185,22 @@ export async function captureBrowserMedia(pageUrl, { signal, browserPaths = brow
       socket.addEventListener('open', resolve, { once: true });
       socket.addEventListener('error', reject, { once: true });
     }), signal);
-    const cdp = createCdp(socket);
-    const responses = new Map();
-    const bodyReads = [];
+    const cdp = createCdp(socket);
+    const responses = new Map();
+    const bodyReads = [];
+    const sessions = new Set();
     const collect = (value, mime = '') => {
       if (!MEDIA_URL.test(value) && !MEDIA_TYPE.test(mime)) return;
       try { urls.add(validateUrl(value)); } catch {}
     };
     cdp.on(message => {
-      if (message.method === 'Target.attachedToTarget') {
-        const attached = message.params.sessionId;
-        cdp.send('Network.enable', {}, attached).catch(() => {});
-        return;
-      }
+      if (message.method === 'Target.attachedToTarget') {
+        const attached = message.params.sessionId;
+        sessions.add(attached);
+        cdp.send('Network.enable', {}, attached).catch(() => {});
+        return;
+      }
+      if (message.method === 'Target.detachedFromTarget') sessions.delete(message.params.sessionId);
       if (message.method === 'Network.requestWillBeSent') collect(message.params.request?.url);
       if (message.method === 'Network.responseReceived') {
         const response = message.params.response;
@@ -182,19 +221,22 @@ export async function captureBrowserMedia(pageUrl, { signal, browserPaths = brow
       }
     });
     await abortable(cdp.send('Target.setAutoAttach', { autoAttach: true, waitForDebuggerOnStart: false, flatten: true }), signal);
-    const target = await abortable(cdp.send('Target.createTarget', { url: 'about:blank' }), signal);
-    const attached = await abortable(cdp.send('Target.attachToTarget', { targetId: target.targetId, flatten: true }), signal);
+    const target = await abortable(cdp.send('Target.createTarget', { url: 'about:blank' }), signal);
+    const attached = await abortable(cdp.send('Target.attachToTarget', { targetId: target.targetId, flatten: true }), signal);
+    sessions.add(attached.sessionId);
     await abortable(cdp.send('Network.enable', {}, attached.sessionId), signal);
     await abortable(cdp.send('Page.enable', {}, attached.sessionId), signal);
-    await abortable(cdp.send('Page.navigate', { url: pageUrl }, attached.sessionId), signal);
-    await delay(Math.min(observeMs / 2, 4000), undefined, { signal });
-    await abortable(cdp.send('Runtime.evaluate', { expression: `(() => {
-      for (const video of document.querySelectorAll('video')) video.play().catch(() => {});
-      const buttons = [...document.querySelectorAll('button,[role="button"]')];
-      const play = buttons.find(node => /^(play|watch|abspielen|ansehen|start)(?:\\s+(?:video|movie|film))?$/i.test((node.getAttribute('aria-label') || node.textContent || '').trim()));
-      if (play) play.click();
-    })()`, returnByValue: true }, attached.sessionId).catch(() => {}), signal);
-    await delay(Math.max(1000, observeMs / 2), undefined, { signal });
+    await abortable(cdp.send('Page.navigate', { url: pageUrl }, attached.sessionId), signal);
+    const observeUntil = Date.now() + observeMs;
+    while (Date.now() < observeUntil) {
+      signal?.throwIfAborted();
+      await Promise.allSettled([...sessions].map(sessionId => abortable(cdp.send('Runtime.evaluate', {
+        expression: PLAY_CONTROL, returnByValue: true,
+      }, sessionId), signal)));
+      if (!deepScan && urls.size > 0) break;
+      const remaining = observeUntil - Date.now();
+      if (remaining > 0) await delay(Math.min(300, remaining), undefined, { signal });
+    }
     await Promise.race([Promise.allSettled(bodyReads), delay(1000, undefined, { signal }).catch(() => {})]);
     return [...urls];
   } catch (error) {
@@ -208,14 +250,19 @@ export async function captureBrowserMedia(pageUrl, { signal, browserPaths = brow
 }
 
 /** Verify candidates with yt-dlp before offering them to the user. */
-export async function discoverSources(pageUrl, { signal, capture = captureBrowserMedia, inspect, limit = 30,
-  deepScan = false, timeoutMs = deepScan ? 120000 : 45000 } = {}) {
+export async function discoverSources(pageUrl, { signal, capture = captureBrowserMedia, inspect, limit = 30,
+  deepScan = false, timeoutMs = deepScan ? 120000 : 45000 } = {}) {
   signal?.throwIfAborted();
   const timeout = new AbortController();
   const timer = setTimeout(() => timeout.abort(new DOMException('Source scan timed out.', 'TimeoutError')), timeoutMs);
   const scanSignal = signal ? AbortSignal.any([signal, timeout.signal]) : timeout.signal;
-  const sources = [];
-  const seen = new Set();
+  const sources = [];
+  const seen = new Set();
+  const formatReferences = new Map();
+  const mediaKey = value => {
+    try { const address = new URL(value); return `${address.origin}${address.pathname}`; }
+    catch { return null; }
+  };
   try {
     const observeMs = deepScan ? Math.min(30000, Math.floor(timeoutMs * 0.6)) : Math.min(8000, Math.floor(timeoutMs * 0.5));
     let urls;
@@ -231,14 +278,19 @@ export async function discoverSources(pageUrl, { signal, capture = captureBrowse
           || !metadata.formats?.length && !metadata.url) continue;
         const address = new URL(url);
         const key = `${address.origin}${address.pathname}:${metadata.id || ''}`;
-        if (seen.has(key)) continue;
-        seen.add(key);
-        sources.push({ ...summarizeSource(metadata, url, sources.length + 1, pageUrl), url });
+        if (seen.has(key)) continue;
+        seen.add(key);
+        sources.push({ ...summarizeSource(metadata, url, sources.length + 1, pageUrl), url });
+        formatReferences.set(url, new Set((metadata.formats || []).map(format => mediaKey(format.url)).filter(Boolean)));
       } catch { /* A request alone is not proof of downloadable media. */ }
     }
     signal?.throwIfAborted();
-    Object.defineProperty(sources, 'timedOut', { value: timeout.signal.aborted });
-    return sources;
+    // A master playlist and its variant streams describe the same video.
+    const distinct = sources.filter(source => !sources.some(other => other !== source
+      && formatReferences.get(other.url)?.has(mediaKey(source.url))));
+    distinct.forEach((source, index) => { source.index = index + 1; });
+    Object.defineProperty(distinct, 'timedOut', { value: timeout.signal.aborted });
+    return distinct;
   } finally {
     clearTimeout(timer);
   }