channel-worker 1.6.0 → 1.6.1

package/lib/stats-syncer.js

@@ -1,8 +1,9 @@
 const WebSocket = require('ws');
 
 /**
- * Sync YouTube Studio stats via Nstbrowser CDP.
- * Flow: Launch profile → connect CDP → navigate to Studio → scrape stats → close
+ * Sync YouTube stats via Nstbrowser CDP.
+ * Flow: Launch profile → connect CDP → get channel ID from Studio →
+ *       navigate to channel About page → read #additional-info-container table
  */
 class StatsSyncer {
   constructor(nstManager, apiClient) {
@@ -92,221 +93,114 @@ class StatsSyncer {
           await s('Page.enable');
           await s('Runtime.enable');
 
-          let stats = { subscribers: 0, total_views: 0, video_count: 0 };
-
-          // Step 1: Go to YouTube Studio → auto-redirects to correct channel
+          // Step 1: Go to YouTube Studio to get channel ID
           console.log('[stats] Navigating to YouTube Studio...');
           await s('Page.navigate', { url: 'https://studio.youtube.com' });
           await this.waitForLoad(s);
-          await this.sleep(5000);
+          await this.sleep(4000);
 
-          // Get channel ID from URL
           const urlResult = await s('Runtime.evaluate', { expression: 'window.location.href', returnByValue: true });
           const studioUrl = urlResult?.result?.value || '';
           console.log(`[stats] Studio URL: ${studioUrl}`);
 
-          const channelIdMatch = studioUrl.match(/channel\/(UC[\w-]+)/);
-          const channelId = channelIdMatch ? channelIdMatch[1] : null;
-
-          if (!channelId) {
-            // Try ytcfg
+          // Extract channel ID
+          let cid = null;
+          const cidMatch = studioUrl.match(/channel\/(UC[\w-]+)/);
+          if (cidMatch) {
+            cid = cidMatch[1];
+          } else {
             const cfgResult = await s('Runtime.evaluate', {
               expression: '(window.ytcfg && window.ytcfg.get) ? window.ytcfg.get("CHANNEL_ID") : ""',
               returnByValue: true,
             });
-            console.log(`[stats] ytcfg CHANNEL_ID: ${cfgResult?.result?.value}`);
+            cid = cfgResult?.result?.value || null;
           }
 
-          const cid = channelId || (await s('Runtime.evaluate', {
-            expression: '(window.ytcfg && window.ytcfg.get) ? window.ytcfg.get("CHANNEL_ID") : ""',
-            returnByValue: true,
-          }))?.result?.value;
-
-          // Step 2: Navigate to Studio Analytics page (English URL works regardless of UI language)
-          if (cid) {
-            const analyticsUrl = `https://studio.youtube.com/channel/${cid}/analytics/tab-overview/period-default`;
-            console.log(`[stats] Navigating to analytics: ${analyticsUrl}`);
-            await s('Page.navigate', { url: analyticsUrl });
-            await this.waitForLoad(s);
-            await this.sleep(6000);
-
-            // Scrape analytics page — look for numbers regardless of language
-            const analyticsResult = await s('Runtime.evaluate', {
-              expression: `
-                (function() {
-                  var result = { subscribers: 0, total_views: 0, video_count: 0, debug_text: '' };
-                  var text = document.body ? document.body.innerText : '';
-                  result.debug_text = text.substring(0, 800);
-
-                  // YouTube Studio Analytics shows metric cards with labels and numbers
-                  // The layout is: Label (in any language) followed by a number
-                  // We look for large numbers near subscriber/view keywords in ANY language
-
-                  // Strategy: find all metric-like elements by looking for number patterns
-                  // Studio analytics typically shows: Views, Watch time, Subscribers
-                  // in cards with the number prominently displayed
-
-                  // Try to find subscriber-related numbers
-                  // Multi-language patterns: subscribers, người đăng ký, abonnés, Abonnenten, etc.
-                  var subKeywords = /subscri|đăng ký|abonn|подписч|구독/i;
-                  var viewKeywords = /views?|lượt xem|vues?|aufrufe|просмотр|조회/i;
-
-                  var lines = text.split('\\n').map(function(l) { return l.trim(); }).filter(function(l) { return l.length > 0; });
-
-                  for (var i = 0; i < lines.length; i++) {
-                    var line = lines[i];
-                    var nextLine = lines[i + 1] || '';
-                    var prevLine = lines[i - 1] || '';
-
-                    // Check if this line or adjacent lines contain keywords
-                    if (subKeywords.test(line) || subKeywords.test(prevLine)) {
-                      // Look for a number in this line or nearby
-                      var numInLine = extractNumber(line);
-                      var numInNext = extractNumber(nextLine);
-                      if (numInLine > 0 && result.subscribers === 0) result.subscribers = numInLine;
-                      else if (numInNext > 0 && result.subscribers === 0) result.subscribers = numInNext;
-                    }
-
-                    if (viewKeywords.test(line) || viewKeywords.test(prevLine)) {
-                      var numInLine2 = extractNumber(line);
-                      var numInNext2 = extractNumber(nextLine);
-                      if (numInLine2 > 0 && result.total_views === 0) result.total_views = numInLine2;
-                      else if (numInNext2 > 0 && result.total_views === 0) result.total_views = numInNext2;
-                    }
-                  }
-
-                  // Also try: Current subscribers count shown in Studio sidebar
-                  // Pattern: "Current subscribers\\n123" or "Người đăng ký hiện tại\\n123"
-                  var currentSubMatch = text.match(/(?:current subscribers|người đăng ký hiện tại|subscriber|đăng ký)[\\s\\n]*([\\d,\\.]+[KMB]?)/i);
-                  if (currentSubMatch && result.subscribers === 0) {
-                    result.subscribers = parseHumanNum(currentSubMatch[1]);
-                  }
-
-                  function extractNumber(str) {
-                    if (!str) return 0;
-                    // Match standalone numbers like "1,234" or "5.2K" or "123"
-                    var m = str.match(/^([\\d,\\.]+[KMB]?)$/i) || str.match(/\\b([\\d,\\.]+[KMB]?)\\b/);
-                    if (m) return parseHumanNum(m[1]);
-                    return 0;
-                  }
-
-                  function parseHumanNum(str) {
-                    if (!str) return 0;
-                    str = str.replace(/,/g, '').trim();
-                    var m = str.match(/([\\d\\.]+)\\s*([KMB])/i);
-                    if (m) {
-                      var n = parseFloat(m[1]);
-                      var s = m[2].toUpperCase();
-                      if (s === 'K') return Math.round(n * 1000);
-                      if (s === 'M') return Math.round(n * 1000000);
-                      if (s === 'B') return Math.round(n * 1000000000);
-                    }
-                    return parseInt(str, 10) || 0;
-                  }
-
-                  return JSON.stringify(result);
-                })()
-              `,
-              returnByValue: true,
-            });
-
-            try {
-              var parsed = JSON.parse(analyticsResult?.result?.value || '{}');
-              console.log('[stats] Analytics debug text (first 300):', (parsed.debug_text || '').substring(0, 300));
-              if (parsed.subscribers > 0) stats.subscribers = parsed.subscribers;
-              if (parsed.total_views > 0) stats.total_views = parsed.total_views;
-            } catch {}
+          if (!cid) {
+            throw new Error('Could not find channel ID from YouTube Studio');
           }
+          console.log(`[stats] Channel ID: ${cid}`);
 
-          // Step 3: Get video count + subscriber from channel page via ytInitialData
-          var channelPageUrl = cid
-            ? 'https://www.youtube.com/channel/' + cid
-            : 'https://www.youtube.com/@me';
-
-          console.log('[stats] Navigating to channel page:', channelPageUrl);
-          await s('Page.navigate', { url: channelPageUrl });
+          // Step 2: Navigate to channel About page
+          const aboutUrl = `https://www.youtube.com/channel/${cid}/about`;
+          console.log(`[stats] Navigating to: ${aboutUrl}`);
+          await s('Page.navigate', { url: aboutUrl });
           await this.waitForLoad(s);
           await this.sleep(4000);
 
-          var ytResult = await s('Runtime.evaluate', {
+          // Step 3: Read #additional-info-container table
+          const result = await s('Runtime.evaluate', {
             expression: `
               (function() {
-                var result = { subscribers: 0, total_views: 0, video_count: 0 };
-                try {
-                  var data = window.ytInitialData;
-                  if (!data) return JSON.stringify({ error: 'no ytInitialData' });
-
-                  // Try pageHeaderRenderer (new layout)
-                  var pageHeader = data.header && data.header.pageHeaderRenderer;
-                  if (pageHeader) {
-                    var content = pageHeader.content && pageHeader.content.pageHeaderViewModel;
-                    if (content && content.metadata && content.metadata.contentMetadataViewModel) {
-                      var rows = content.metadata.contentMetadataViewModel.metadataRows || [];
-                      for (var r = 0; r < rows.length; r++) {
-                        var parts = rows[r].metadataParts || [];
-                        for (var p = 0; p < parts.length; p++) {
-                          var t = (parts[p].text && parts[p].text.content) || '';
-                          // subscriber text like "123 subscribers" or "0 người đăng ký"
-                          if (/subscri|đăng ký/i.test(t)) {
-                            var m = t.match(/([\\d,\\.]+[KMB]?)/);
-                            if (m) result.subscribers = parseN(m[1]);
-                          }
-                          // video count like "1 video" or "15 video"
-                          if (/video/i.test(t)) {
-                            var m2 = t.match(/([\\d,]+)/);
-                            if (m2) result.video_count = parseInt(m2[1].replace(/,/g, ''), 10);
-                          }
-                        }
-                      }
-                    }
-                  }
+                var stats = { subscribers: -1, total_views: -1, video_count: -1, debug: '' };
 
-                  // Try c4TabbedHeaderRenderer (old layout)
-                  var c4Header = data.header && data.header.c4TabbedHeaderRenderer;
-                  if (c4Header) {
-                    var subText = c4Header.subscriberCountText && c4Header.subscriberCountText.simpleText || '';
-                    if (subText) {
-                      var m3 = subText.match(/([\\d,\\.]+[KMB]?)/);
-                      if (m3 && result.subscribers === 0) result.subscribers = parseN(m3[1]);
-                    }
-                    if (c4Header.videosCountText) {
-                      var vText = c4Header.videosCountText.runs ? c4Header.videosCountText.runs.map(function(r){return r.text}).join('') : (c4Header.videosCountText.simpleText || '');
-                      var m4 = vText.match(/([\\d,]+)/);
-                      if (m4) result.video_count = parseInt(m4[1].replace(/,/g, ''), 10);
-                    }
-                  }
+                // Find the table inside #additional-info-container
+                var container = document.querySelector('#additional-info-container');
+                if (!container) {
+                  stats.debug = 'no #additional-info-container found';
+                  // Fallback: try to find stats from page text
+                  var text = document.body ? document.body.innerText : '';
+                  stats.debug = text.substring(0, 500);
+                  return JSON.stringify(stats);
+                }
 
-                } catch (e) {
-                  return JSON.stringify({ error: e.message });
+                var table = container.querySelector('table');
+                if (!table) {
+                  stats.debug = 'no table in container, text: ' + container.innerText.substring(0, 300);
+                  return JSON.stringify(stats);
                 }
 
-                function parseN(str) {
-                  if (!str) return 0;
-                  str = str.replace(/,/g, '').trim();
-                  var m = str.match(/([\\d\\.]+)\\s*([KMB])/i);
-                  if (m) {
-                    var n = parseFloat(m[1]);
-                    var s = m[2].toUpperCase();
-                    if (s === 'K') return Math.round(n * 1000);
-                    if (s === 'M') return Math.round(n * 1000000);
-                    if (s === 'B') return Math.round(n * 1000000000);
+                // Each row has: icon cell + text cell (e.g. "1 subscriber", "3 videos", "743 views")
+                var rows = table.querySelectorAll('tr');
+                var texts = [];
+                rows.forEach(function(row) {
+                  var text = row.innerText.trim();
+                  texts.push(text);
+
+                  // Match patterns: "X subscriber(s)", "X video(s)", "X view(s)"
+                  // Also Vietnamese: "X người đăng ký", "X video", "X lượt xem"
+                  var num = 0;
+                  var numMatch = text.match(/([\\d,\\.]+)/);
+                  if (numMatch) {
+                    num = parseInt(numMatch[1].replace(/,/g, ''), 10);
+                  }
+
+                  // Check for K/M/B suffix
+                  var suffixMatch = text.match(/([\\d,\\.]+)\\s*([KMB])/i);
+                  if (suffixMatch) {
+                    var n = parseFloat(suffixMatch[1].replace(/,/g, ''));
+                    var suffix = suffixMatch[2].toUpperCase();
+                    if (suffix === 'K') num = Math.round(n * 1000);
+                    else if (suffix === 'M') num = Math.round(n * 1000000);
+                    else if (suffix === 'B') num = Math.round(n * 1000000000);
                   }
-                  return parseInt(str, 10) || 0;
-                }
 
-                return JSON.stringify(result);
+                  if (/subscriber|đăng ký/i.test(text)) {
+                    stats.subscribers = num;
+                  } else if (/\\bvideos?\\b/i.test(text)) {
+                    stats.video_count = num;
+                  } else if (/\\bviews?\\b|lượt xem/i.test(text)) {
+                    stats.total_views = num;
+                  }
+                });
+
+                stats.debug = texts.join(' | ');
+                return JSON.stringify(stats);
               })()
             `,
             returnByValue: true,
           });
 
+          var stats = { subscribers: 0, total_views: 0, video_count: 0 };
           try {
-            var ytData = JSON.parse(ytResult?.result?.value || '{}');
-            console.log('[stats] ytInitialData result:', JSON.stringify(ytData));
-            if (ytData.subscribers > 0 && stats.subscribers === 0) stats.subscribers = ytData.subscribers;
-            if (ytData.video_count > 0) stats.video_count = ytData.video_count;
-            if (ytData.total_views > 0 && stats.total_views === 0) stats.total_views = ytData.total_views;
-          } catch {}
+            var parsed = JSON.parse(result?.result?.value || '{}');
+            console.log(`[stats] About page data: ${parsed.debug}`);
+            if (parsed.subscribers >= 0) stats.subscribers = parsed.subscribers;
+            if (parsed.total_views >= 0) stats.total_views = parsed.total_views;
+            if (parsed.video_count >= 0) stats.video_count = parsed.video_count;
+          } catch (e) {
+            console.error(`[stats] Parse error: ${e.message}`);
+          }
 
           clearTimeout(timeout);
           ws.close();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "channel-worker",
-  "version": "1.6.0",
+  "version": "1.6.1",
   "description": "Channel Manager worker daemon — runs on remote machines to execute video pipeline jobs",
   "main": "lib/daemon.js",
   "bin": {