tt-help-cli-ycl 1.4.136 → 1.4.139

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.4.136",
+  "version": "1.4.139",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/cli/comments.js

@@ -5,10 +5,11 @@
 //   手动模式: tt-help comments <URL> [N] [--save]            单视频评论提取
 // ────────────────────────────────────────
 
-import { writeFileSync, unlinkSync } from "fs";
-import os from "os";
-import path from "path";
-import { execSync } from "child_process";
+import {
+  getBrowserMemory,
+  formatMemoryWithBrowser,
+} from "../lib/browser-memory.js";
+import { reportConsecutiveTaskErrors } from "../lib/client-error-report.js";
 
 import { fetchUserCommentsAPI } from "../lib/api-interceptor-comment.js";
 import {
@@ -52,9 +53,9 @@ const CFG = {
   interval: 10,
   /** 并行数 */
   parallel: 1,
-  /** 连续 403 后切换端口 */
+  /** 连续 403 后重启浏览器 */
   consecutive403Blocked: 3,
-  /** 连续验证码后切换端口 */
+  /** 连续验证码后重启浏览器 */
   maxConsecutiveCaptcha: 3,
   /** 内存回收 — 触发回收的 RSS(MB) 阈值 */
   recycleRssMb: 2000,
@@ -66,11 +67,11 @@ const CFG = {
   recycleCooldownMs: 180000,
   /** 内存回收 — 每 N 个任务强制回收（0=禁用） */
   periodicRecycleEvery: 0,
-  /** 内存回收 — 回收时是否切换端口 */
-  switchPortOnRecycle: true,
+  /** 内存回收 — 回收时是否切换端口（默认关闭：只重启浏览器不换端口） */
+  switchPortOnRecycle: false,
   /** 连续任务处理错误阈值（达到后上报服务端） */
   consecutiveTaskErrorsMax: 20,
-  /** 单端口最大运行时间(ms)，超时自动切换端口（默认 48 分钟） */
+  /** 单端口最大运行时间(ms)，超时自动重启浏览器（默认 48 分钟） */
   maxSessionRunTimeMs: 48 * 60 * 1000,
 };
 
@@ -103,7 +104,7 @@ function loadEnvConfig() {
     int("COMMENTS_PERIODIC_RECYCLE_EVERY", CFG.periodicRecycleEvery),
   );
   CFG.switchPortOnRecycle =
-    String(process.env.COMMENTS_SWITCH_PORT_ON_RECYCLE || "1") !== "0";
+    String(process.env.COMMENTS_SWITCH_PORT_ON_RECYCLE || "0") !== "0";
   CFG.consecutiveTaskErrorsMax = Math.max(
     1,
     int("COMMENTS_CONSECUTIVE_TASK_ERRORS_MAX", CFG.consecutiveTaskErrorsMax),
@@ -114,95 +115,6 @@ function loadEnvConfig() {
   );
 }
 
-// ────────────────────────────────────────
-// Section 2: 内存监控工具
-// ────────────────────────────────────────
-
-function formatMemoryUsage(mem = process.memoryUsage()) {
-  const mb = (n) => (n / 1024 / 1024).toFixed(0);
-  return `rss:${mb(mem.rss)}MB heap:${mb(mem.heapUsed)}MB ext:${mb(mem.external)}MB ab:${mb(mem.arrayBuffers)}MB`;
-}
-
-/**
- * 获取 Edge 浏览器进程的内存占用（跨平台：macOS/Linux/Windows）
- */
-function getBrowserMemory(port = null) {
-  try {
-    const platform = os.platform();
-    let cmd;
-    if (platform === "darwin") {
-      if (port) {
-        cmd = `ps aux | grep "[M]icrosoft Edge" | grep "remote-debugging-port=${port}" | awk '{sum += $6} END {printf "%d", sum/1024}'`;
-      } else {
-        cmd =
-          'ps aux | grep "[M]icrosoft Edge" | awk \'{sum += $6} END {printf "%d", sum/1024}\'';
-      }
-      const result = execSync(cmd, { timeout: 5000 }).toString().trim();
-      const mb = parseInt(result, 10);
-      return mb > 0 ? mb : null;
-    } else if (platform === "linux") {
-      if (port) {
-        cmd = `ps aux | grep "[m]sedge" | grep "remote-debugging-port=${port}" | awk '{sum += $6} END {printf "%d", sum/1024}'`;
-      } else {
-        cmd =
-          'ps aux | grep "[m]sedge" | awk \'{sum += $6} END {printf "%d", sum/1024}\'';
-      }
-      const result = execSync(cmd, { timeout: 5000 }).toString().trim();
-      const mb = parseInt(result, 10);
-      return mb > 0 ? mb : null;
-    } else if (platform === "win32") {
-      const ps1Path = path.join(os.tmpdir(), `edge-mem-${Date.now()}.ps1`);
-      try {
-        let ps1Content;
-        if (port) {
-          ps1Content =
-            `$procs = Get-CimInstance Win32_Process -Filter "Name LIKE 'msedge%'" -ErrorAction SilentlyContinue; ` +
-            `$sum = 0; ` +
-            `foreach ($p in $procs) { ` +
-            `  if ($p.CommandLine -and $p.CommandLine -like "*--remote-debugging-port=${port}*") { ` +
-            `    $sum += $p.WorkingSetSize ` +
-            `  } ` +
-            `}; ` +
-            `Write-Output $sum`;
-        } else {
-          // 无端口场景：Get-Process（.NET API，不依赖 WMI）统计全部 msedge 工作集，
-          // 与原 Get-CimInstance 全量求和行为等价；WMI 损坏/禁用的机器也能返回
-          ps1Content =
-            `$procs = Get-Process msedge -ErrorAction SilentlyContinue; ` +
-            `$sum = 0; ` +
-            `foreach ($p in $procs) { $sum += $p.WorkingSet64 }; ` +
-            `Write-Output $sum`;
-        }
-        writeFileSync(ps1Path, ps1Content);
-        cmd = `powershell.exe -NoProfile -ExecutionPolicy Bypass -File "${ps1Path}"`;
-        const result = execSync(cmd, { timeout: 5000 }).toString().trim();
-        try {
-          unlinkSync(ps1Path);
-        } catch {}
-        const bytes = parseInt(result, 10);
-        const mb = bytes > 0 ? Math.round(bytes / 1048576) : 0;
-        return mb > 0 ? mb : null;
-      } catch {
-        try {
-          unlinkSync(ps1Path);
-        } catch {}
-        return null;
-      }
-    }
-    return null;
-  } catch {
-    return null;
-  }
-}
-
-function formatMemoryWithBrowser(port = null) {
-  const nodeMem = formatMemoryUsage();
-  const edgeMB = getBrowserMemory(port);
-  if (!edgeMB) return nodeMem;
-  const portLabel = port ? `:${port}` : "";
-  return `${nodeMem} edge${portLabel}:${edgeMB}MB`;
-}
-
 // ────────────────────────────────────────
 // Section 3: 页面辅助
 // ────────────────────────────────────────
@@ -384,7 +296,7 @@ class CdpSessionManager {
     if (now - this._lastMemCheck < 30000) return;
     this._lastMemCheck = now;
 
-    // 定时轮换：单端口运行超时自动切换
+    // 定时轮换：单端口运行超时自动重启浏览器
     const sessionElapsed = now - this.sessionStartTime;
     if (sessionElapsed >= CFG.maxSessionRunTimeMs) {
       const mins = Math.round(sessionElapsed / 60000);
@@ -474,10 +386,14 @@ async function prepareVideoPage(page, href) {
   await page.keyboard.press("Escape");
   await new Promise((r) => setTimeout(r, 1000));
 
-  // 如果评论面板已打开，先关闭
+  // 如果右侧面板已打开（评论/猜你喜欢/商品 tab 面板），先安全关闭
+  // 同时识别旧版 RightPanelContainer 与新版 VideoDetailContainer
   const panelOpen = await safeEvaluate(
     page,
-    () => !!document.querySelector('[class*="RightPanelContainer"]'),
+    () =>
+      !!document.querySelector(
+        '[class*="RightPanelContainer"], [class*="VideoDetailContainer"]',
+      ),
   );
   if (panelOpen) {
     await closeCommentPanel(page);
@@ -849,7 +765,7 @@ async function runAutoMode(options) {
 
                 if (consecutiveCaptchaCount >= CFG.maxConsecutiveCaptcha) {
                   console.error(
-                    `  ♻️ 连续${CFG.maxConsecutiveCaptcha}次验证码 → 切换端口`,
+                    `  ♻️ 连续${CFG.maxConsecutiveCaptcha}次验证码 → 重启浏览器`,
                   );
                   await session.recycleSession("连续验证码", {
                     switchPort: true,
@@ -899,7 +815,7 @@ async function runAutoMode(options) {
               const remainingMins =
                 remainingMs > 0 ? Math.round(remainingMs / 60000) : 0;
               console.error(
-                `── #${totalTasks()} ── ${formatMemoryWithBrowser(session.port)}  处理=${processedCount} 跳过=${skippedCount} 错误=${errorCount}  距切换≈${remainingMins}min`,
+                `── #${totalTasks()} ── ${formatMemoryWithBrowser(session.port)}  处理=${processedCount} 跳过=${skippedCount} 错误=${errorCount}  距轮换≈${remainingMins}min`,
               );
             }
           } catch (err) {
@@ -918,7 +834,7 @@ async function runAutoMode(options) {
                 `        403 (${consecutive403}/${CFG.consecutive403Blocked})`,
               );
               if (consecutive403 >= CFG.consecutive403Blocked) {
-                console.error("  🚫 连续403 → 等2min切换端口");
+                console.error("  🚫 连续403 → 等2min重启浏览器");
                 await new Promise((r) => setTimeout(r, 120000));
                 await session.recycleSession("连续 403 封禁");
                 consecutive403 = 0;
@@ -931,19 +847,18 @@ async function runAutoMode(options) {
               console.error(`        ${err.message}`);
             }
 
-            // 连续任务错误达到阈值，汇报服务端
+            // 连续任务错误达到阈值，汇报服务端 + 触发 daemon fallback 换节点
             if (consecutiveTaskErrors >= CFG.consecutiveTaskErrorsMax) {
               console.error(
-                `  ⚠️ 连续任务错误 ${consecutiveTaskErrors}/${CFG.consecutiveTaskErrorsMax} → 汇报服务端`,
+                `  ⚠️ 连续任务错误 ${consecutiveTaskErrors}/${CFG.consecutiveTaskErrorsMax} → 汇报服务端 + 触发 fallback`,
               );
-              apiPost(`${serverUrl}/api/error-report`, {
+              await reportConsecutiveTaskErrors({
+                post: apiPost,
+                baseUrl: serverUrl,
                 userId,
-                username: "",
-                errorType: "consecutive_task_errors",
-                errorMessage: `连续 ${consecutiveTaskErrors} 个任务处理失败 (port=${session.port})`,
+                message: `连续 ${consecutiveTaskErrors} 个任务处理失败 (port=${session.port})`,
                 stage: "comments",
-                errorStack: "",
-              }).catch(() => {});
+              });
               consecutiveTaskErrors = 0;
             }
           }
@@ -962,19 +877,18 @@ async function runAutoMode(options) {
           console.error(`  × ${err.message}`);
         }
 
-        // 连续轮间错误达到阈值，汇报服务端
+        // 连续轮间错误达到阈值，汇报服务端 + 触发 daemon fallback 换节点
         if (consecutiveTaskErrors >= CFG.consecutiveTaskErrorsMax) {
           console.error(
-            `  ⚠️ 连续轮间错误 ${consecutiveTaskErrors}/${CFG.consecutiveTaskErrorsMax} → 汇报服务端`,
+            `  ⚠️ 连续轮间错误 ${consecutiveTaskErrors}/${CFG.consecutiveTaskErrorsMax} → 汇报服务端 + 触发 fallback`,
           );
-          apiPost(`${serverUrl}/api/error-report`, {
+          await reportConsecutiveTaskErrors({
+            post: apiPost,
+            baseUrl: serverUrl,
             userId,
-            username: "",
-            errorType: "consecutive_task_errors",
-            errorMessage: `连续 ${consecutiveTaskErrors} 个轮次处理失败 (port=${session.port})`,
+            message: `连续 ${consecutiveTaskErrors} 个轮次处理失败 (port=${session.port})`,
             stage: "comments",
-            errorStack: "",
-          }).catch(() => {});
+          });
           consecutiveTaskErrors = 0;
         }
       }

package/src/cli/explore.js

@@ -557,6 +557,30 @@ export async function handleExplore(options) {
             }
           }
 
+          // 兜底：上报错误前再识别一次验证码
+          // 网络错误（如连续请求失败）常伴随验证码/限流页，而 explore-core 的 detectCaptcha
+          // 可能在验证码延迟渲染前已执行，或已识别到( captchaDetected=true )但后续流程又抛了异常出现 error。
+          // 这里在把错误上报为 network/other 前对当前页面再检测一次，识别到则转走验证码分支。
+          if (result.error) {
+            try {
+              const lateCaptcha = await detectCaptcha(page);
+              if (lateCaptcha && lateCaptcha.visible) {
+                console.error(`  [验证码] 兜底识别到验证码 @${username}`);
+                result.captchaDetected = true;
+                result.captchaStage = result.captchaStage || "process";
+                result.captchaMessage =
+                  result.captchaMessage || "上报错误前识别到验证码";
+                result.error = null;
+                result.errorStack = "";
+                result.retryable = false;
+                consecutiveNetworkErrors = 0;
+                errorCount--;
+              }
+            } catch (e) {
+              // 页面可能已销毁（浏览器崩溃等），忽略兜底检测异常
+            }
+          }
+
           if (result.error) {
             // 上报错误（重试后仍有错误才上报）
             await apiPost(`${serverUrl}/api/job/${username}`, {