channel-worker 2.5.47 → 2.5.49

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "channel-worker",
-  "version": "2.5.47",
+  "version": "2.5.49",
   "description": "Channel Manager worker daemon — runs on remote machines to execute video pipeline jobs",
   "main": "lib/daemon.js",
   "bin": {

package/scripts/nurture_facebook.js

@@ -21,6 +21,11 @@
 
 const { assertAccountUsable } = require('./lib/fb-guard');
 
+// Every markup shape Facebook has used for "one post in the feed". Used by the
+// secondary helpers (ad check, open-post) via closest()/querySelectorAll;
+// scanFeed picks the single best-yielding one per render instead.
+const UNIT_SEL = "div[role='feed'] > div, div[aria-posinset], [data-pagelet^='FeedUnit'], [role='article']";
+
 // ─── helpers ────────────────────────────────────────────────────────────────
 function randInt(min, max) { return Math.floor(min + Math.random() * (max - min + 1)); }
 function chance(p) { return Math.random() < p; }
@@ -107,6 +112,26 @@ async function dismissDialogs(page, log) {
 async function scanFeed(page, { skipKeys = [], wantLike = false } = {}) {
   return page.evaluate(({ skip, wantLike }) => {
     const skipSet = new Set(skip);
+
+    // WHICH element is "a post"? Not a fixed selector — Facebook's feed markup
+    // moves. Measured on the live account 2026-08-11: the whole document held
+    // exactly ONE [role='article'] while the feed scrolled fine, so a hardcoded
+    // selector counted 1 post per 10 minutes. Try the known shapes and keep
+    // whichever yields the most post-sized blocks on THIS render.
+    const CANDIDATES = [
+      "div[role='feed'] > div",
+      'div[aria-posinset]',
+      "[data-pagelet^='FeedUnit']",
+      "[role='article']",
+    ];
+    let sel = null, nodes = [];
+    for (const s of CANDIDATES) {
+      const found = [...document.querySelectorAll(s)].filter((e) => e.getBoundingClientRect().height > 80);
+      if (found.length > nodes.length) { nodes = found; sel = s; }
+    }
+    // Drop blocks nested inside another block of the same kind (comments,
+    // embedded shares) — only outermost units are posts.
+    const units = nodes.filter((n) => !nodes.some((o) => o !== n && o.contains(n)));
     const keyOf = (art) => {
       for (const a of art.querySelectorAll('a[href]')) {
         const h = a.getAttribute('href') || '';
@@ -120,9 +145,7 @@ async function scanFeed(page, { skipKeys = [], wantLike = false } = {}) {
     const mid = window.innerHeight / 2;
     const posts = [];
     let best = null, bestDist = Infinity, bestKey = '';
-    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;
+    for (const art of units) {
       const r = art.getBoundingClientRect();
       if (r.bottom < 0 || r.top > window.innerHeight || r.height < 80) continue;
       const text = art.innerText || '';
@@ -140,7 +163,6 @@ async function scanFeed(page, { skipKeys = [], wantLike = false } = {}) {
         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');
@@ -148,10 +170,25 @@ async function scanFeed(page, { skipKeys = [], wantLike = false } = {}) {
         break;
       }
     }
+    // Which element is actually scrolling? Facebook has shipped both a
+    // window-scrolled feed and one inside its own overflow container; if we
+    // measure the wrong one, "the page didn't move" is a false alarm.
+    let scroller = 'window', scrollTop = Math.round(window.scrollY || 0);
+    if (!scrollTop) {
+      for (const el of document.querySelectorAll('div')) {
+        if (el.scrollTop > 50 && el.scrollHeight > el.clientHeight + 200) {
+          scroller = 'div.' + String(el.className || '').split(' ')[0];
+          scrollTop = Math.round(el.scrollTop);
+          break;
+        }
+      }
+    }
     return {
-      posts, likeKey,
+      posts, likeKey, scroller, scrollTop,
       scrollY: Math.round(window.scrollY || document.documentElement.scrollTop || 0),
-      articlesInDom: document.querySelectorAll("[role='article']").length,
+      articlesInDom: units.length,
+      selector: sel || '(none)',
+      sample: posts.slice(0, 3).map((p) => p.key.slice(0, 46)),
     };
   }, { skip: skipKeys, wantLike }).catch(() => ({ posts: [], likeKey: null, scrollY: 0, articlesInDom: 0 }));
 }
@@ -184,12 +221,12 @@ async function clickMarkedLike(page, log) {
 // Watch a video that's playing in the feed, counting only seconds where
 // playback actually ADVANCES (a frozen/buffering player must not bank time).
 async function watchFeedVideo(page, minSec, maxSec, log) {
-  const has = await page.evaluate(() => {
+  const has = await page.evaluate((UNIT) => {
     const vids = [...document.querySelectorAll('video')];
     for (const v of vids) {
       const r = v.getBoundingClientRect();
       if (r.height < 100 || r.bottom < 0 || r.top > window.innerHeight) continue;
-      const art = v.closest("[role='article']");
+      const art = v.closest(UNIT);
       // 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;
@@ -199,7 +236,7 @@ async function watchFeedVideo(page, minSec, maxSec, log) {
       return true;
     }
     return false;
-  }).catch(() => false);
+  }, UNIT_SEL).catch(() => false);
   if (!has) return false;
 
   const budgetMs = randInt(minSec, maxSec) * 1000;
@@ -228,11 +265,11 @@ async function watchFeedVideo(page, minSec, maxSec, log) {
 // Phase 2+. Best-effort: if the click doesn't navigate, nothing is lost.
 async function openPostAndRead(page, log) {
   const urlBefore = page.url();
-  const marked = await page.evaluate(() => {
+  const marked = await page.evaluate((UNIT) => {
     const mid = window.innerHeight / 2;
     let best = null, bestDist = Infinity;
-    for (const art of document.querySelectorAll("[role='article']")) {
-      if (art.parentElement && art.parentElement.closest("[role='article']")) continue;
+    for (const art of document.querySelectorAll(UNIT)) {
+      if (art.parentElement && art.parentElement.closest(UNIT)) 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;
@@ -249,7 +286,7 @@ async function openPostAndRead(page, log) {
       return true;
     }
     return false;
-  }).catch(() => false);
+  }, UNIT_SEL).catch(() => false);
   if (!marked) return false;
 
   try {
@@ -283,7 +320,7 @@ async function openPostAndRead(page, log) {
 // Follow a page/creator surfaced by the feed itself (phase 3, ≤ config max).
 // Only clicks a Follow/Like-Page button that's already on screen — never hunts.
 async function followVisiblePage(page, log) {
-  const marked = await page.evaluate(() => {
+  const marked = await page.evaluate((UNIT) => {
     for (const b of document.querySelectorAll("[role='button'], a[role='button']")) {
       const t = (b.innerText || '').trim();
       const aria = (b.getAttribute('aria-label') || '').trim();
@@ -292,13 +329,13 @@ async function followVisiblePage(page, log) {
       const r = b.getBoundingClientRect();
       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']");
+      const art = b.closest(UNIT);
       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;
     }
     return null;
-  }).catch(() => null);
+  }, UNIT_SEL).catch(() => null);
   if (!marked) return false;
   let ok = false;
   try {
@@ -362,6 +399,11 @@ async function run({ page, payload, log }) {
       seenKeys.add(p.key);
       if (!p.isAd) fresh++;      // scrolling past an ad isn't reading
     }
+    // debug:true (probe runs only) — dumps what the scan actually saw so a
+    // wrong selector / duplicate key shows up as data instead of a guess.
+    if (payload.debug && ticks % 3 === 0) {
+      log('info', `[nurture-fb][dbg] t${ticks} scroller=${scan.scroller}@${scan.scrollTop} winY=${scan.scrollY} sel=${scan.selector} units=${scan.articlesInDom} fresh=${fresh} keys=${JSON.stringify(scan.sample)}`);
+    }
     postsSeen += fresh;
     postsSinceLike += fresh;
     // Feed that stops yielding posts for ~15 ticks = end of feed, a wall, or a
@@ -419,7 +461,7 @@ async function run({ page, payload, log }) {
     // moment activity trips a threshold — better to stop than to keep poking.
     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=${lastScan.articlesInDom}`);
+      log('info', `[nurture-fb] …${Math.round((Date.now() - t0) / 1000)}s: posts=${postsSeen} likes=${likes}/${likesTarget} videos=${videos} scrollY=${lastScan.scrollY} units=${lastScan.articlesInDom} sel=${lastScan.selector}`);
     }
 
     // Page refuses to move at all → this is not a nurture session, it's ten