channel-worker 2.5.45 → 2.5.47

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "channel-worker",
-  "version": "2.5.45",
+  "version": "2.5.47",
   "description": "Channel Manager worker daemon — runs on remote machines to execute video pipeline jobs",
   "main": "lib/daemon.js",
   "bin": {

package/scripts/nurture_facebook.js

@@ -25,6 +25,44 @@ const { assertAccountUsable } = require('./lib/fb-guard');
 function randInt(min, max) { return Math.floor(min + Math.random() * (max - min + 1)); }
 function chance(p) { return Math.random() < p; }
 
+// Park the cursor over the middle of the feed column. MUST be done before any
+// wheel event: mouse.wheel scrolls whatever is under the pointer, and the
+// pointer starts at (0,0) — FB's fixed top-nav — so every wheel tick landed on
+// a non-scrolling element. Measured on the first live session: 3 posts in 9.6
+// minutes, six "feed idle" bailouts, because the page never actually moved.
+async function centerMouse(page) {
+  const vp = await page.evaluate(() => ({ w: window.innerWidth, h: window.innerHeight })).catch(() => null);
+  if (!vp) return;
+  const jitter = (n) => n + randInt(-40, 40);
+  await page.mouse.move(jitter(Math.floor(vp.w / 2)), jitter(Math.floor(vp.h / 2))).catch(() => {});
+}
+
+// Scroll by `dy` and VERIFY the page moved. Falls back through three rungs:
+// wheel → re-center the cursor and wheel again → window.scrollBy. Returns the
+// distance actually scrolled.
+async function humanScroll(page, dy) {
+  const readY = () => page.evaluate(() => window.scrollY || document.documentElement.scrollTop || 0).catch(() => 0);
+  const before = await readY();
+  await page.mouse.wheel(0, dy).catch(() => {});
+  await page.waitForTimeout(randInt(350, 700));
+  let after = await readY();
+  if (Math.abs(after - before) < 5) {
+    await centerMouse(page);
+    await page.mouse.wheel(0, dy).catch(() => {});
+    await page.waitForTimeout(randInt(350, 700));
+    after = await readY();
+  }
+  if (Math.abs(after - before) < 5) {
+    // Last resort — a programmatic scroll. Not a trusted event, but scrolling
+    // isn't gated on trust the way clicks are, and a feed that never moves is
+    // worth less than a feed that moves synthetically.
+    await page.evaluate((d) => window.scrollBy(0, d), dy).catch(() => {});
+    await page.waitForTimeout(randInt(300, 600));
+    after = await readY();
+  }
+  return Math.abs(after - before);
+}
+
 // Dismiss FB cookie / cross-sell / "not now" popups. Best-effort, never throws.
 // Never touches the composer or a reel viewer.
 async function dismissDialogs(page, log) {
@@ -55,60 +93,73 @@ async function dismissDialogs(page, log) {
   }
 }
 
-// Count feed posts that entered the viewport since the last call, marking them
-// so they're counted once. Sponsored posts are marked too (so they're never
-// liked or watched) but do NOT count as "seen" — an ad isn't reading.
-async function markPostsInView(page) {
-  return page.evaluate(() => {
-    let fresh = 0;
-    for (const art of document.querySelectorAll("[role='article']")) {
-      // Comments render as NESTED articles — only top-level posts count.
-      if (art.parentElement && art.parentElement.closest("[role='article']")) continue;
-      if (art.hasAttribute('data-nur-seen')) continue;
-      const r = art.getBoundingClientRect();
-      const inView = r.bottom > 0 && r.top < window.innerHeight && r.height > 80;
-      if (!inView) continue;
-      art.setAttribute('data-nur-seen', '1');
-      const head = (art.innerText || '').slice(0, 400);
-      if (/Được tài trợ|Sponsored|Tài trợ/i.test(head)) { art.setAttribute('data-nur-ad', '1'); continue; }
-      fresh++;
-    }
-    return fresh;
-  }).catch(() => 0);
-}
+// Scan the posts currently in the viewport.
+//
+// Identity is the POST, never the DOM node. Facebook virtualises the feed: it
+// recycles the same <div role="article"> element for a different post as you
+// scroll, so a "seen" flag written onto the node marks every future post that
+// lands in it as already-seen. First live run: 1 post counted in 4.5 minutes of
+// real scrolling. The key is the permalink id (falling back to a text prefix),
+// which travels with the content instead of the container.
+//
+// Also marks a like candidate in the same pass — one evaluate keeps the key
+// derivation in exactly one place instead of drifting across two copies.
+async function scanFeed(page, { skipKeys = [], wantLike = false } = {}) {
+  return page.evaluate(({ skip, wantLike }) => {
+    const skipSet = new Set(skip);
+    const keyOf = (art) => {
+      for (const a of art.querySelectorAll('a[href]')) {
+        const h = a.getAttribute('href') || '';
+        const m = h.match(/\/posts\/[\w.]+|story_fbid=\d+|\/reel\/\d+|\/videos\/\d+|\/permalink\/\d+|\/photo\/?\?fbid=\d+/);
+        if (m) return m[0];
+      }
+      // No permalink (some suggested/aggregated cards) — fall back to content.
+      return 'txt:' + (art.innerText || '').replace(/\s+/g, ' ').slice(0, 90);
+    };
 
-// Like ONE organic post near the centre of the viewport. Returns true on a
-// confirmed like. Deliberately picky: skips ads, skips already-liked posts, and
-// never grabs a COMMENT's like button (nested article).
-async function likeVisiblePost(page, log) {
-  const found = await page.evaluate(() => {
     const mid = window.innerHeight / 2;
-    let best = null, bestDist = Infinity;
+    const posts = [];
+    let best = null, bestDist = Infinity, bestKey = '';
     for (const art of document.querySelectorAll("[role='article']")) {
-      if (art.parentElement && art.parentElement.closest("[role='article']")) continue; // comment
-      if (art.hasAttribute('data-nur-ad') || art.hasAttribute('data-nur-liked')) continue;
+      // Comments render as NESTED articles — only top-level posts count.
+      if (art.parentElement && art.parentElement.closest("[role='article']")) continue;
       const r = art.getBoundingClientRect();
-      if (r.bottom < 60 || r.top > window.innerHeight - 60 || r.height < 120) continue;
+      if (r.bottom < 0 || r.top > window.innerHeight || r.height < 80) continue;
+      const text = art.innerText || '';
+      const isAd = /Được tài trợ|Sponsored|Tài trợ/i.test(text.slice(0, 400));
+      const key = keyOf(art);
+      posts.push({ key, isAd });
+      if (!wantLike || isAd || skipSet.has(key)) continue;
       const dist = Math.abs((r.top + r.bottom) / 2 - mid);
-      if (dist < bestDist) { best = art; bestDist = dist; }
+      if (dist < bestDist) { best = art; bestDist = dist; bestKey = key; }
     }
-    if (!best) return null;
-    if (/Được tài trợ|Sponsored/i.test((best.innerText || '').slice(0, 400))) { best.setAttribute('data-nur-ad', '1'); return null; }
-    for (const b of best.querySelectorAll("[role='button']")) {
-      const aria = (b.getAttribute('aria-label') || '').trim();
-      if (!/^(thích|like)$/i.test(aria)) continue;
-      if (b.getAttribute('aria-pressed') === 'true') continue;   // already liked
-      if (b.closest("[role='article']") !== best) continue;      // belongs to a comment
-      const r = b.getBoundingClientRect();
-      if (r.width < 8 || r.height < 8 || b.offsetParent === null) continue;
-      best.setAttribute('data-nur-liked', 'pending');
-      b.setAttribute('__nur_like__', '1');
-      return { aria };
+
+    let likeKey = null;
+    if (best) {
+      for (const b of best.querySelectorAll("[role='button']")) {
+        const aria = (b.getAttribute('aria-label') || '').trim();
+        if (!/^(thích|like)$/i.test(aria)) continue;
+        if (b.getAttribute('aria-pressed') === 'true') continue;   // already liked
+        if (b.closest("[role='article']") !== best) continue;       // a comment's button
+        const br = b.getBoundingClientRect();
+        if (br.width < 8 || br.height < 8 || b.offsetParent === null) continue;
+        b.setAttribute('__nur_like__', '1');
+        likeKey = bestKey;
+        break;
+      }
     }
-    return null;
-  }).catch(() => null);
-  if (!found) return false;
+    return {
+      posts, likeKey,
+      scrollY: Math.round(window.scrollY || document.documentElement.scrollTop || 0),
+      articlesInDom: document.querySelectorAll("[role='article']").length,
+    };
+  }, { skip: skipKeys, wantLike }).catch(() => ({ posts: [], likeKey: null, scrollY: 0, articlesInDom: 0 }));
+}
 
+// Click the like button scanFeed already marked (`__nur_like__`). Returns true
+// on a confirmed like. The picking — skip ads, skip already-liked, never a
+// comment's button — happened in scanFeed; this only performs the click.
+async function clickMarkedLike(page, log) {
   const btn = page.locator("[__nur_like__='1']").first();
   let ok = false;
   try {
@@ -120,17 +171,14 @@ async function likeVisiblePost(page, log) {
     // and React's handler ignores it (same lesson as the YouTube ad-skip).
     await btn.click({ timeout: 4000 });
     await page.waitForTimeout(randInt(900, 1800));
-    ok = await btn.evaluate((el) => el.getAttribute('aria-pressed') === 'true').catch(() => true);
-    if (ok === null || ok === undefined) ok = true;
+    const pressed = await btn.evaluate((el) => el.getAttribute('aria-pressed')).catch(() => null);
+    ok = pressed === null ? true : pressed === 'true';  // button gone/re-rendered → treat as done
   } catch (e) {
     log('info', `[nurture-fb] like failed: ${String(e.message || e).slice(0, 80)}`);
   }
-  await page.evaluate((liked) => {
-    document.querySelectorAll('[__nur_like__]').forEach(e => e.removeAttribute('__nur_like__'));
-    document.querySelectorAll("[data-nur-liked='pending']").forEach(e => e.setAttribute('data-nur-liked', liked ? '1' : 'fail'));
-  }, !!ok).catch(() => {});
+  await page.evaluate(() => document.querySelectorAll('[__nur_like__]').forEach(e => e.removeAttribute('__nur_like__'))).catch(() => {});
   if (ok) log('info', '[nurture-fb] liked a post');
-  return !!ok;
+  return ok;
 }
 
 // Watch a video that's playing in the feed, counting only seconds where
@@ -142,7 +190,9 @@ async function watchFeedVideo(page, minSec, maxSec, log) {
       const r = v.getBoundingClientRect();
       if (r.height < 100 || r.bottom < 0 || r.top > window.innerHeight) continue;
       const art = v.closest("[role='article']");
-      if (art && art.hasAttribute('data-nur-ad')) continue;      // never watch ads
+      // Ads are re-detected from text — a node flag would be stale the moment
+      // FB recycles the article for a different post.
+      if (art && /Được tài trợ|Sponsored/i.test((art.innerText || '').slice(0, 400))) continue;
       if (v.hasAttribute('data-nur-watched')) continue;
       v.setAttribute('data-nur-watched', '1');
       if (v.paused) { try { v.play(); } catch {} }
@@ -183,7 +233,9 @@ async function openPostAndRead(page, log) {
     let best = null, bestDist = Infinity;
     for (const art of document.querySelectorAll("[role='article']")) {
       if (art.parentElement && art.parentElement.closest("[role='article']")) continue;
-      if (art.hasAttribute('data-nur-ad') || art.hasAttribute('data-nur-opened')) continue;
+      // Node-level flags can't be trusted (FB recycles articles), so ads are
+      // re-checked from the text right here.
+      if (/Được tài trợ|Sponsored/i.test((art.innerText || '').slice(0, 400))) continue;
       const r = art.getBoundingClientRect();
       if (r.bottom < 60 || r.top > window.innerHeight - 60) continue;
       const dist = Math.abs((r.top + r.bottom) / 2 - mid);
@@ -193,7 +245,6 @@ async function openPostAndRead(page, log) {
     for (const a of best.querySelectorAll("a[href*='/posts/'], a[href*='/permalink/'], a[href*='story_fbid'], a[href*='/photo']")) {
       const r = a.getBoundingClientRect();
       if (r.width < 8 || r.height < 8 || a.offsetParent === null) continue;
-      best.setAttribute('data-nur-opened', '1');
       a.setAttribute('__nur_open__', '1');
       return true;
     }
@@ -211,8 +262,9 @@ async function openPostAndRead(page, log) {
   await page.waitForTimeout(randInt(2500, 4000));
 
   // Read comments: a couple of small scrolls with human pauses.
+  await centerMouse(page);
   for (let i = 0; i < randInt(2, 4); i++) {
-    await page.mouse.wheel(0, randInt(200, 600)).catch(() => {});
+    await humanScroll(page, randInt(200, 600));
     await page.waitForTimeout(randInt(1500, 4000));
   }
   log('info', '[nurture-fb] opened a post and read comments');
@@ -241,7 +293,7 @@ async function followVisiblePage(page, log) {
       if (r.width < 8 || r.height < 8 || b.offsetParent === null) continue;
       if (r.bottom < 0 || r.top > window.innerHeight) continue;
       const art = b.closest("[role='article']");
-      if (art && art.hasAttribute('data-nur-ad')) continue;   // not an ad's CTA
+      if (art && /Được tài trợ|Sponsored/i.test((art.innerText || '').slice(0, 400))) continue; // not an ad's CTA
       b.setAttribute('__nur_follow__', '1');
       return label;
     }
@@ -288,14 +340,28 @@ async function run({ page, payload, log }) {
   await assertAccountUsable(page, log);
   await dismissDialogs(page, log);
 
+  await centerMouse(page);   // wheel events need the cursor over the feed column
+
   let postsSeen = 0, likes = 0, videos = 0, pagesFollowed = 0, postsOpened = 0;
   let postsSinceLike = likeGap;   // allow the first like once enough posts scroll by
-  let ticks = 0, emptyTicks = 0;
+  let ticks = 0, emptyTicks = 0, stuckScrolls = 0;
   const deadline = t0 + sessionSec * 1000;
 
+  const seenKeys = new Set();    // post ids already counted (survives DOM recycling)
+  const likedKeys = new Set();   // post ids already attempted — never twice
+  let lastScan = { scrollY: 0, articlesInDom: 0 };
+
   while (Date.now() < deadline) {
     ticks++;
-    const fresh = await markPostsInView(page);
+    const wantLike = likes < likesTarget && postsSinceLike >= likeGap;
+    const scan = await scanFeed(page, { skipKeys: [...likedKeys], wantLike });
+    lastScan = scan;
+    let fresh = 0;
+    for (const p of scan.posts) {
+      if (seenKeys.has(p.key)) continue;
+      seenKeys.add(p.key);
+      if (!p.isAd) fresh++;      // scrolling past an ad isn't reading
+    }
     postsSeen += fresh;
     postsSinceLike += fresh;
     // Feed that stops yielding posts for ~15 ticks = end of feed, a wall, or a
@@ -318,9 +384,11 @@ async function run({ page, payload, log }) {
     }
 
     // Like — phase 2+ only, never two posts in a row (enforced by likeGap).
-    if (likes < likesTarget && postsSinceLike >= likeGap) {
-      if (await likeVisiblePost(page, log)) { likes++; postsSinceLike = 0; }
-      else postsSinceLike = Math.max(0, postsSinceLike - 1); // retry a bit later
+    // scanFeed already marked the button; a post is only ever attempted once.
+    if (wantLike && scan.likeKey) {
+      likedKeys.add(scan.likeKey);
+      if (await clickMarkedLike(page, log)) { likes++; postsSinceLike = 0; }
+      else postsSinceLike = Math.max(0, postsSinceLike - 1); // try another post later
     }
 
     // Open a post to read comments — phase 2+.
@@ -339,16 +407,26 @@ async function run({ page, payload, log }) {
     if (roll < 0.15) {
       await page.waitForTimeout(randInt(6000, 20000));            // stop and read
     } else if (roll < 0.23) {
-      await page.mouse.wheel(0, -randInt(150, 400)).catch(() => {});
+      await humanScroll(page, -randInt(150, 400));                // re-read what just passed
       await page.waitForTimeout(randInt(1200, 3000));
     } else {
-      await page.mouse.wheel(0, randInt(250, 900)).catch(() => {});
+      const moved = await humanScroll(page, randInt(250, 900));
+      if (moved < 5) stuckScrolls++; else stuckScrolls = 0;
       await page.waitForTimeout(randInt(800, 4000));
     }
 
     // Mid-session health check (~ every 12 ticks). FB throws a checkpoint the
     // moment activity trips a threshold — better to stop than to keep poking.
-    if (ticks % 12 === 0) await assertAccountUsable(page, log);
+    if (ticks % 12 === 0) {
+      await assertAccountUsable(page, log);
+      log('info', `[nurture-fb] …${Math.round((Date.now() - t0) / 1000)}s: posts=${postsSeen} likes=${likes}/${likesTarget} videos=${videos} scrollY=${lastScan.scrollY} articles=${lastScan.articlesInDom}`);
+    }
+
+    // Page refuses to move at all → this is not a nurture session, it's ten
+    // minutes of pretending. Fail loudly instead of banking a fake session.
+    if (stuckScrolls >= 20) {
+      throw new Error(`nurture-fb: trang không cuộn được sau ${stuckScrolls} lần thử (posts=${postsSeen}) — layout FB đổi hoặc feed không tải`);
+    }
   }
 
   const durationSec = Math.round((Date.now() - t0) / 1000);