channel-worker 2.5.52 → 2.5.54

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "channel-worker",
-  "version": "2.5.52",
+  "version": "2.5.54",
   "description": "Channel Manager worker daemon — runs on remote machines to execute video pipeline jobs",
   "main": "lib/daemon.js",
   "bin": {

package/scripts/upload_facebook.js

@@ -1255,24 +1255,58 @@ async function runOnce({ page, payload, log }) {
             await page.locator(editBtn.selector).click({ timeout: 5000 });
             await page.waitForTimeout(2500);
 
-            // Wait for the thumb-edit modal to mount. Header text =
-            // "Chỉnh sửa hình thu nhỏ".
+            // Wait for the thumb editor to mount. FB serves (A/B, seen
+            // 2026-08-19) TWO variants and any given account can get either:
+            //   modal  — a separate [role='dialog'] titled "Chỉnh sửa hình
+            //            thu nhỏ" (the original form).
+            //   nested — the editor renders INSIDE the reel-composer dialog
+            //            ("Cài đặt thước phim"): no own dialog, and the
+            //            header sits too deep in innerText for a 300-char
+            //            signature check. Recognize it by its upload button
+            //            (aria "Tải hình thu nhỏ tùy chỉnh lên…"), which only
+            //            exists while the editor is open, and tag the nearest
+            //            container that also holds Lưu + Hủy so the scoped
+            //            selectors below keep working for both variants.
+            // Check modal FIRST so old-UI accounts behave exactly as before.
             await page.evaluate(() => document.querySelectorAll("[__fbpw_thumb_edit__]").forEach((el) => el.removeAttribute('__fbpw_thumb_edit__'))).catch(() => {});
-            const thumbDialogReady = await page.evaluate(() => {
+            const detectThumbEditor = () => page.evaluate(() => {
               document.querySelectorAll('[__fbpw_thumb_dialog__]').forEach((el) => el.removeAttribute('__fbpw_thumb_dialog__'));
+              const vis = (el) => {
+                const r = el.getBoundingClientRect();
+                if (r.width < 8 || r.height < 8) return false;
+                const cs = getComputedStyle(el);
+                return cs.visibility !== 'hidden' && cs.display !== 'none' && cs.opacity !== '0';
+              };
               for (const dlg of document.querySelectorAll("[role='dialog']")) {
-                const r = dlg.getBoundingClientRect();
-                if (r.width < 8 || r.height < 8) continue;
-                const cs = getComputedStyle(dlg);
-                if (cs.visibility === 'hidden' || cs.display === 'none' || cs.opacity === '0') continue;
+                if (!vis(dlg)) continue;
                 const sig = `${dlg.getAttribute('aria-label') || ''}\n${(dlg.innerText || '').slice(0, 300)}`;
                 if (!/Chỉnh sửa hình thu nhỏ|Edit thumbnail/i.test(sig)) continue;
                 dlg.setAttribute('__fbpw_thumb_dialog__', '1');
-                return true;
+                return 'modal';
+              }
+              const uploadBtn = [...document.querySelectorAll('[aria-label]')].find((el) => {
+                const al = el.getAttribute('aria-label') || '';
+                return /Tải hình thu nhỏ|thumbnail/i.test(al) && /Tải|upload/i.test(al) && vis(el);
+              });
+              if (uploadBtn) {
+                const hasBtn = (root, re) => [...root.querySelectorAll("[role='button'], button")]
+                  .some((b) => re.test((b.innerText || '').trim()) && vis(b));
+                let box = uploadBtn.parentElement;
+                while (box && box !== document.body && !(hasBtn(box, /^(Lưu|Save)$/) && hasBtn(box, /^(Hủy|Cancel)$/))) box = box.parentElement;
+                if (box === document.body) box = null;
+                (box || uploadBtn.closest("[role='dialog']") || document.body).setAttribute('__fbpw_thumb_dialog__', '1');
+                return 'nested';
               }
-              return false;
-            }).catch(() => false);
-            if (!thumbDialogReady) throw new Error('thumbnail editor dialog did not mount');
+              return null;
+            }).catch(() => null);
+            let thumbEditorVia = await detectThumbEditor();
+            const mountDeadline = Date.now() + 12_000;
+            while (!thumbEditorVia && Date.now() < mountDeadline) {
+              await page.waitForTimeout(2000);
+              thumbEditorVia = await detectThumbEditor();
+            }
+            if (!thumbEditorVia) throw new Error('thumbnail editor dialog did not mount');
+            log('info', `[fb-pw] page-wall thumb — editor mounted (variant=${thumbEditorVia})`);
 
             // Click "Tải lên" button — opens OS file picker. Use filechooser
             // race to inject the file path directly. Scope every action to the
@@ -1355,6 +1389,11 @@ async function runOnce({ page, payload, log }) {
                 const RETRY_AFTER_MS = 12_000;
                 let lastRetryAt = Date.now();
                 while (Date.now() < closeDeadline) {
+                  // "Editor still open" must cover both variants: the modal
+                  // form (dialog whose text leads with the header) and the
+                  // nested form (no own dialog — its upload button, aria
+                  // "Tải hình thu nhỏ tùy chỉnh lên…", only exists while the
+                  // editor is showing).
                   const still = await page.evaluate(() => {
                     const dlgs = document.querySelectorAll("[role='dialog']");
                     for (const dlg of dlgs) {
@@ -1364,6 +1403,15 @@ async function runOnce({ page, payload, log }) {
                         if (r.width > 8 && r.height > 8) return true;
                       }
                     }
+                    for (const el of document.querySelectorAll('[aria-label]')) {
+                      const al = el.getAttribute('aria-label') || '';
+                      if (!(/Tải hình thu nhỏ|thumbnail/i.test(al) && /Tải|upload/i.test(al))) continue;
+                      const r = el.getBoundingClientRect();
+                      if (r.width < 8 || r.height < 8) continue;
+                      const cs = getComputedStyle(el);
+                      if (cs.visibility === 'hidden' || cs.display === 'none' || cs.opacity === '0') continue;
+                      return true;
+                    }
                     return false;
                   }).catch(() => false);
                   if (!still) { modalClosed = true; break; }
@@ -2184,6 +2232,30 @@ async function runOnce({ page, payload, log }) {
     //       - Page wall display order isn't strictly chronological
     //       - "Quảng bá thước phim" appears on ALL Page reels, not just new
     //       - Title match alone can't distinguish same-title duplicates
+    // Fresh-tile scrape, dùng ở lượt đầu VÀ lượt retry (a.2b) — cùng một luật:
+    // chỉ nhận tile có timestamp tươi ("Vừa xong"/"X phút"), tuyệt đối không
+    // nhận tile cũ.
+    const scrapeFreshReelTile = () => page.evaluate(() => {
+      const FRESH_RE = /vừa xong|vài giây|^\s*\d{1,2}\s*giây|^\s*[1-5]\s*phút\b|\b[1-5]\s*phút trước|just now|few seconds ago|\b[1-5] min(ute)?s? ago/i;
+      const anchors = document.querySelectorAll("a[href*='/reel/']");
+      for (const a of anchors) {
+        const href = a.getAttribute('href') || '';
+        const m = href.match(/\/reel\/(\d{8,18})/);
+        if (!m) continue;
+        const r = a.getBoundingClientRect();
+        if (r.width < 8 || r.height < 8) continue;
+        let ctx = a;
+        for (let depth = 0; depth < 5 && ctx; depth++) {
+          const raw = (ctx.innerText || ctx.textContent || '').slice(0, 500);
+          if (FRESH_RE.test(raw)) {
+            return { href, id: m[1], depth, timestamp: (raw.match(FRESH_RE) || [''])[0] };
+          }
+          ctx = ctx.parentElement;
+        }
+      }
+      return null;
+    }).catch(() => null);
+    let reelsTabUrl = null; // giữ lại cho lượt retry (a.2b)
     try {
       // Find a link to the page's reels tab. Multi-strategy:
       //   1. Any existing href containing "sk=reels_tab"
@@ -2224,35 +2296,14 @@ async function runOnce({ page, payload, log }) {
         return null;
       }).catch(() => null);
       if (reelsTabHref) {
+        reelsTabUrl = reelsTabHref;
         log('info', `[fb-pw] navigating to page reels tab: ${reelsTabHref.slice(0, 100)}`);
         await page.goto(reelsTabHref, { waitUntil: 'domcontentloaded', timeout: 30_000 }).catch(() => {});
         await page.waitForTimeout(5000);
-        const fresh = await page.evaluate(() => {
-          const norm = (s) => (s || '').toLowerCase().normalize('NFD').replace(/\p{Diacritic}/gu, '');
-          const FRESH_RE = /vừa xong|vài giây|^\s*\d{1,2}\s*giây|^\s*[1-5]\s*phút\b|\b[1-5]\s*phút trước|just now|few seconds ago|\b[1-5] min(ute)?s? ago/i;
-          // Reel tiles on the reels tab — each is an <a href="/reel/<id>/">
-          // wrapping a thumbnail + meta. Iterate in DOM order (newest first
-          // on reels tab). For each tile, check if its subtree has a fresh
-          // timestamp.
-          const anchors = document.querySelectorAll("a[href*='/reel/']");
-          for (const a of anchors) {
-            const href = a.getAttribute('href') || '';
-            const m = href.match(/\/reel\/(\d{8,18})/);
-            if (!m) continue;
-            const r = a.getBoundingClientRect();
-            if (r.width < 8 || r.height < 8) continue;
-            // Look within the anchor's subtree + closest meaningful ancestor.
-            let ctx = a;
-            for (let depth = 0; depth < 5 && ctx; depth++) {
-              const raw = (ctx.innerText || ctx.textContent || '').slice(0, 500);
-              if (FRESH_RE.test(raw)) {
-                return { href, id: m[1], depth, timestamp: (raw.match(FRESH_RE) || [''])[0] };
-              }
-              ctx = ctx.parentElement;
-            }
-          }
-          return null;
-        }).catch(() => null);
+        // Reel tiles on the reels tab — each is an <a href="/reel/<id>/">
+        // wrapping a thumbnail + meta, DOM order = newest first. Only a tile
+        // with a FRESH timestamp counts (see scrapeFreshReelTile).
+        const fresh = await scrapeFreshReelTile();
         if (fresh) {
           const full = fresh.href.startsWith('http') ? fresh.href : `https://www.facebook.com${fresh.href}`;
           postUrl = full;
@@ -2374,24 +2425,52 @@ async function runOnce({ page, payload, log }) {
       }
     }
 
+    // (a.2b) RETRY reels_tab sau khi chờ — reel mới thường cần 1-3 phút xử lý
+    //       phía FB rồi mới hiện lên tab, nên lượt scrape đầu (chạy ngay sau
+    //       khi bấm Đăng) hay trượt và mọi thứ rơi xuống last-resort bốc nhầm
+    //       reel cũ (4 lần gần nhất 2026-08-19 đều thế). Một lượt chờ 75s +
+    //       re-scrape cứu được phần lớn ca đó, vẫn giữ luật fresh-timestamp
+    //       nên không thể nhận nhầm tile cũ.
+    if (!postUrl && reelsTabUrl) {
+      log('info', '[fb-pw] no post URL yet — waiting 75s for FB to surface the new reel on reels_tab, then re-scraping…');
+      await page.waitForTimeout(75_000);
+      await page.goto(reelsTabUrl, { waitUntil: 'domcontentloaded', timeout: 30_000 }).catch(() => {});
+      await page.waitForTimeout(5000);
+      const fresh2 = await scrapeFreshReelTile();
+      if (fresh2) {
+        postUrl = fresh2.href.startsWith('http') ? fresh2.href : `https://www.facebook.com${fresh2.href}`;
+        log('info', `[fb-pw] post URL from reels_tab RETRY (id=${fresh2.id}, timestamp="${fresh2.timestamp}"): ${postUrl}`);
+      } else {
+        log('warn', '[fb-pw] reels_tab retry: still no fresh tile');
+      }
+    }
+
     // (a.3) LAST-RESORT: first reel tile on the profile reels tab. UNRELIABLE —
     //       the just-published reel may not be on the tab yet (still
     //       processing), so the first tile can be a STALE older reel. Only used
     //       when every authoritative source above (fresh-timestamp tile,
-    //       title-match, inline CTA, network capture) returned nothing.
+    //       title-match, inline CTA, network capture, timed retry) returned
+    //       nothing. Provably-stale filter: mọi reel ID đã bắt được TRƯỚC khi
+    //       bấm Đăng là reel có sẵn trên page — reel mới không thể là chúng.
+    //       Thà post_url rỗng còn hơn ghi sai (2026-08-19: nhánh này ghi reel
+    //       02/08 cho bài vừa đăng → link trong DB trỏ nhầm bài cũ).
     if (!postUrl) {
-      const tileHref = await page.evaluate(() => {
+      const preIds = capturedReelIds.slice(0, capturedReelIdsSnapshotLen);
+      const tileHref = await page.evaluate((staleIds) => {
+        const staleSet = new Set(staleIds);
         const anchors = document.querySelectorAll("a[role='link']");
         for (const a of anchors) {
           const aria = (a.getAttribute('aria-label') || '').toLowerCase();
           if (!/bản xem trước ô thước phim|reel tile preview|reel preview/.test(aria)) continue;
           const href = a.getAttribute('href') || '';
           const m = href.match(/\/reel\/(\d{8,20})/);
-          if (m) return { href, aria: aria.slice(0, 60) };
+          if (m) return { href, aria: aria.slice(0, 60), id: m[1], provablyStale: staleSet.has(m[1]) };
         }
         return null;
-      }).catch(() => null);
-      if (tileHref) {
+      }, preIds).catch(() => null);
+      if (tileHref && tileHref.provablyStale) {
+        log('warn', `[fb-pw] LAST-RESORT tile ${tileHref.id} was already captured PRE-publish — provably an OLD reel; leaving post_url empty instead of recording a wrong one`);
+      } else if (tileHref) {
         postUrl = tileHref.href.startsWith('http') ? tileHref.href : `https://www.facebook.com${tileHref.href}`;
         log('warn', `[fb-pw] post URL from FIRST reel tile on reels_tab (LAST-RESORT, may be STALE): ${postUrl}`);
       }