tt-help-cli-ycl 1.4.10 → 1.4.11

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.4.10",
+  "version": "1.4.11",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/cli/comments.js

@@ -5,7 +5,9 @@
 //   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 { fetchUserCommentsAPI } from "../lib/api-interceptor-comment.js";
@@ -65,6 +67,8 @@ const CFG = {
   periodicRecycleEvery: 0,
   /** 内存回收 — 回收时是否切换端口 */
   switchPortOnRecycle: true,
+  /** 连续任务处理错误阈值（达到后上报服务端） */
+  consecutiveTaskErrorsMax: 20,
 };
 
 /** 从环境变量覆写配置 */
@@ -97,6 +101,10 @@ function loadEnvConfig() {
   );
   CFG.switchPortOnRecycle =
     String(process.env.COMMENTS_SWITCH_PORT_ON_RECYCLE || "1") !== "0";
+  CFG.consecutiveTaskErrorsMax = Math.max(
+    1,
+    int("COMMENTS_CONSECUTIVE_TASK_ERRORS_MAX", CFG.consecutiveTaskErrorsMax),
+  );
 }
 
 // ────────────────────────────────────────
@@ -108,32 +116,82 @@ function formatMemoryUsage(mem = process.memoryUsage()) {
   return `rss:${mb(mem.rss)}MB heap:${mb(mem.heapUsed)}MB ext:${mb(mem.external)}MB ab:${mb(mem.arrayBuffers)}MB`;
 }
 
-function getBrowserMemory(port) {
+/**
+ * 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();
-    if (platform !== "darwin" && platform !== "linux") return null;
-    const procName = platform === "darwin" ? "Microsoft Edge" : "msedge";
-    const c = procName[0];
-    const rest = procName.slice(1);
     let cmd;
-    if (port) {
-      cmd = `ps aux | grep "[${c}]${rest}" | grep "remote-debugging-port=${port}" | awk '{sum += $6} END {printf "%d", sum/1024}'`;
-    } else {
-      cmd = `ps aux | grep "[${c}]${rest}" | awk '{sum += $6} END {printf "%d", sum/1024}'`;
+    if (platform === "darwin") {
+      if (port) {
+        cmd = `ps aux | grep "[M]icrosoft Edge" | grep "remote-debugging-port=${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 $sum`;
+        } else {
+          ps1Content =
+            `$procs = Get-CimInstance Win32_Process -Filter "Name LIKE 'msedge%'" -ErrorAction SilentlyContinue; ` +
+            `$sum = 0; ` +
+            `foreach ($p in $procs) { $sum += $p.WorkingSetSize }; ` +
+            `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;
+      }
     }
-    const mb = parseInt(execSync(cmd, { timeout: 5000 }).toString().trim(), 10);
-    return mb > 0 ? mb : null;
+    return null;
   } catch {
     return null;
   }
 }
 
-function formatMemoryWithBrowser(port) {
+function formatMemoryWithBrowser(port = null) {
   const nodeMem = formatMemoryUsage();
   const edgeMB = getBrowserMemory(port);
   if (!edgeMB) return nodeMem;
-  const label = port ? ` edge:${port}:${edgeMB}MB` : ` edge:${edgeMB}MB`;
-  return nodeMem + label;
+  const portLabel = port ? `:${port}` : "";
+  return `${nodeMem} edge${portLabel}:${edgeMB}MB`;
 }
 
 // ────────────────────────────────────────
@@ -481,7 +539,8 @@ async function runAutoMode(options) {
     browserRestartCount = 0;
   let consecutiveErrors = 0,
     consecutive403 = 0,
-    consecutiveCaptchaCount = 0;
+    consecutiveCaptchaCount = 0,
+    consecutiveTaskErrors = 0;
 
   // ── CDP 会话 ──
   const session = new CdpSessionManager({
@@ -720,6 +779,7 @@ async function runAutoMode(options) {
             await apiPut(`${serverUrl}/api/comment-task/${id}`);
             processedCount++;
             consecutive403 = 0;
+            consecutiveTaskErrors = 0;
 
             // 每 10 个任务输出内存汇总
             if (totalTasks() % 10 === 0) {
@@ -729,6 +789,7 @@ async function runAutoMode(options) {
             }
           } catch (err) {
             errorCount++;
+            consecutiveTaskErrors++;
             const isBrowserClosed = isBrowserClosedError(err);
             const is403 =
               err.message &&
@@ -754,6 +815,22 @@ async function runAutoMode(options) {
             } else {
               console.error(`        ${err.message}`);
             }
+
+            // 连续任务错误达到阈值，汇报服务端
+            if (consecutiveTaskErrors >= CFG.consecutiveTaskErrorsMax) {
+              console.error(
+                `  ⚠️ 连续任务错误 ${consecutiveTaskErrors}/${CFG.consecutiveTaskErrorsMax} → 汇报服务端`,
+              );
+              apiPost(`${serverUrl}/api/error-report`, {
+                userId,
+                username: "",
+                errorType: "consecutive_task_errors",
+                errorMessage: `连续 ${consecutiveTaskErrors} 个任务处理失败 (port=${session.port})`,
+                stage: "comments",
+                errorStack: "",
+              }).catch(() => {});
+              consecutiveTaskErrors = 0;
+            }
           }
 
           // 任务间内存检查
@@ -761,6 +838,7 @@ async function runAutoMode(options) {
         }
       } catch (err) {
         errorCount++;
+        consecutiveTaskErrors++;
         if (isBrowserClosedError(err)) {
           browserRestartCount++;
           console.error(`  ⚠ 浏览器断开 (#${browserRestartCount}) → 重启`);
@@ -768,6 +846,22 @@ async function runAutoMode(options) {
         } else {
           console.error(`  × ${err.message}`);
         }
+
+        // 连续轮间错误达到阈值，汇报服务端
+        if (consecutiveTaskErrors >= CFG.consecutiveTaskErrorsMax) {
+          console.error(
+            `  ⚠️ 连续轮间错误 ${consecutiveTaskErrors}/${CFG.consecutiveTaskErrorsMax} → 汇报服务端`,
+          );
+          apiPost(`${serverUrl}/api/error-report`, {
+            userId,
+            username: "",
+            errorType: "consecutive_task_errors",
+            errorMessage: `连续 ${consecutiveTaskErrors} 个轮次处理失败 (port=${session.port})`,
+            stage: "comments",
+            errorStack: "",
+          }).catch(() => {});
+          consecutiveTaskErrors = 0;
+        }
       }
 
       // 轮间内存检查 + 等待
@@ -794,7 +888,9 @@ async function runManualMode(url, options) {
     save: shouldSave,
     serverUrl = defaultServer,
     basePort = 9222,
+    proxy: proxyOption,
   } = options;
+  const effectiveProxy = proxyOption || configuredProxy;
 
   let browser;
   try {
@@ -930,5 +1026,6 @@ export async function handleComments(options) {
     save: commentsSave,
     serverUrl: commentsServer || defaultServer,
     basePort: commentsBasePort,
+    proxy: commentsProxy,
   });
 }