tt-help-cli-ycl 1.4.52 → 1.4.54

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.4.52",
+  "version": "1.4.54",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/cli/attach.js

@@ -57,11 +57,13 @@ export async function handleAttach(options) {
     attachCountries,
     customProxy,
     attachPoolSize,
+    attachWebserver,
     showHelp,
   } = options;
+  const webserverUrl = attachWebserver || null;
   const effectiveProxy = customProxy || configuredProxy;
 
-  if (effectiveProxy) {
+  if (effectiveProxy && !webserverUrl) {
     attachLog(`[Attach] 使用代理: ${effectiveProxy}`);
   }
   let shuttingDown = false;
@@ -87,12 +89,18 @@ export async function handleAttach(options) {
     attachLog(
       "  --size <N>            浏览器页面池大小（默认: 3，不建议超过 3）",
     );
+    attachLog(
+      "  --webserver <URL>     webserver 模式：通过 HTTP 调用集中式 webserver，不启动本地浏览器",
+    );
     attachLog("");
     attachLog("说明:");
     attachLog(
       "  后台轮询服务端 /api/user-update-tasks 接口，自动抓取 TikTok 用户信息",
     );
     attachLog("  抓取完成后通过 POST /api/user-info-batch 批量回传结果");
+    attachLog(
+      "  默认启动本地浏览器（CDP 模式），传 --webserver 则走 HTTP 接口，无需浏览器",
+    );
     attachLog("  浏览器崩溃时自动重启，支持长时间无人值守运行");
     attachLog(
       "  --size 控制单个浏览器内并发的 page 数，增大可提升吞吐但可能触发 TikTok 限流",
@@ -126,10 +134,32 @@ export async function handleAttach(options) {
 
   const { apiGet, apiPost } = createApiClient({ log: false });
 
-  const scraper = new TikTokScraper({
-    poolSize: attachPoolSize || 3,
-    proxyServer: effectiveProxy || null,
-  });
+  // ── webserver 模式：不启动本地浏览器，通过 HTTP 调用集中式 webserver ──
+  let scraper = null;
+  let getUserInfo;
+
+  if (webserverUrl) {
+    attachLog(`[Attach] webserver 模式: ${webserverUrl}`);
+    getUserInfo = async (uniqueId) => {
+      const res = await fetch(`${webserverUrl}/api/tiktok/user`, {
+        method: "POST",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify({ uniqueId }),
+        signal: AbortSignal.timeout(30000),
+      });
+      const data = await res.json();
+      if (!res.ok || data.error) return null;
+      return data;
+    };
+  } else {
+    scraper = new TikTokScraper({
+      poolSize: attachPoolSize || 3,
+      proxyServer: effectiveProxy || null,
+    });
+    await scraper.init();
+    attachLog("[Attach] 浏览器初始化完成，开始循环接收任务...");
+    getUserInfo = (uniqueId) => scraper.getUserInfo(uniqueId);
+  }
   const shutdown = async (signal) => {
     if (shuttingDown) return;
     shuttingDown = true;
@@ -152,9 +182,6 @@ export async function handleAttach(options) {
   process.on("SIGTERM", onSigterm);
 
   try {
-    await scraper.init();
-    attachLog("[Attach] 浏览器初始化完成，开始循环接收任务...");
-
     let loopCount = 0;
     let browserRestartCount = 0;
     let taskBatchCount = 0;
@@ -162,8 +189,8 @@ export async function handleAttach(options) {
     while (!shuttingDown) {
       loopCount++;
 
-      // 检查浏览器是否存活，不存活则重启
-      if (!scraper.isAlive()) {
+      // 检查浏览器是否存活，不存活则重启（仅本地浏览器模式）
+      if (scraper && !scraper.isAlive()) {
         attachLog(
           `[Attach] 浏览器已关闭，正在重启 (${++browserRestartCount})...`,
         );
@@ -174,7 +201,7 @@ export async function handleAttach(options) {
       // 收到退出信号且当前无任务正在处理，则退出
       if (shuttingDown) {
         attachLog("[Attach] 当前任务已完成，正在关闭浏览器...");
-        await scraper.close().catch(() => {});
+        if (scraper) await scraper.close().catch(() => {});
         if (forceExitTimer) {
           clearTimeout(forceExitTimer);
           forceExitTimer = null;
@@ -192,12 +219,14 @@ export async function handleAttach(options) {
       );
 
       if (!tasks || tasks.length === 0) {
-        const heap = getHeapUsage();
-        if (heap.ratio >= HEAP_RESTART_RATIO) {
-          await recycleScraper(
-            scraper,
-            `空闲时堆占用 ${(heap.ratio * 100).toFixed(1)}%`,
-          );
+        if (scraper) {
+          const heap = getHeapUsage();
+          if (heap.ratio >= HEAP_RESTART_RATIO) {
+            await recycleScraper(
+              scraper,
+              `空闲时堆占用 ${(heap.ratio * 100).toFixed(1)}%`,
+            );
+          }
         }
         if (loopCount === 1) {
           attachLog(`[Attach] 当前无待更新任务，${attachInterval}秒后重试...`);
@@ -223,7 +252,7 @@ export async function handleAttach(options) {
             : "";
           attachLog(`  → 获取 @${uniqueId} 的用户信息...${countryTag}`);
           try {
-            const info = await scraper.getUserInfo(uniqueId);
+            const info = await getUserInfo(uniqueId);
             return { uniqueId, info, error: null };
           } catch (err) {
             return { uniqueId, info: null, error: err };
@@ -319,38 +348,40 @@ export async function handleAttach(options) {
 
       attachLog(`  本批结果: ${successCount} 成功, ${failCount} 失败\n`);
 
-      // CDN 限流比例超过 30% 时，冷却 + 重启浏览器
-      const cdnRatio = cdnBlockedCount / tasks.length;
-      if (cdnRatio > 0.3) {
-        const coolDownSeconds = cdnRatio > 0.8 ? 120 : 60;
-        attachLog(
-          `  [Attach] CDN限流比例 ${(cdnRatio * 100).toFixed(0)}% (${cdnBlockedCount}/${tasks.length})，冷却 ${coolDownSeconds} 秒后重启浏览器...`,
-        );
-        await new Promise((r) => setTimeout(r, coolDownSeconds * 1000));
-        await recycleScraper(
-          scraper,
-          `CDN限流比例过高 (${cdnBlockedCount}/${tasks.length})`,
-        );
-        browserRestartCount++;
-        taskBatchCount = 0;
-      }
+      // CDN 限流比例超过 30% 时，冷却 + 重启浏览器（仅本地浏览器模式）
+      if (scraper) {
+        const cdnRatio = cdnBlockedCount / tasks.length;
+        if (cdnRatio > 0.3) {
+          const coolDownSeconds = cdnRatio > 0.8 ? 120 : 60;
+          attachLog(
+            `  [Attach] CDN限流比例 ${(cdnRatio * 100).toFixed(0)}% (${cdnBlockedCount}/${tasks.length})，冷却 ${coolDownSeconds} 秒后重启浏览器...`,
+          );
+          await new Promise((r) => setTimeout(r, coolDownSeconds * 1000));
+          await recycleScraper(
+            scraper,
+            `CDN限流比例过高 (${cdnBlockedCount}/${tasks.length})`,
+          );
+          browserRestartCount++;
+          taskBatchCount = 0;
+        }
 
-      const heap = getHeapUsage();
-      if (heap.ratio >= HEAP_RESTART_RATIO) {
-        await recycleScraper(
-          scraper,
-          `本批后堆占用 ${(heap.ratio * 100).toFixed(1)}%`,
-        );
-        browserRestartCount++;
-        taskBatchCount = 0;
-      } else if (taskBatchCount >= MAX_TASK_BATCHES_BEFORE_RESTART) {
-        await recycleScraper(scraper, `已连续处理 ${taskBatchCount} 批任务`);
-        browserRestartCount++;
-        taskBatchCount = 0;
-      }
+        const heap = getHeapUsage();
+        if (heap.ratio >= HEAP_RESTART_RATIO) {
+          await recycleScraper(
+            scraper,
+            `本批后堆占用 ${(heap.ratio * 100).toFixed(1)}%`,
+          );
+          browserRestartCount++;
+          taskBatchCount = 0;
+        } else if (taskBatchCount >= MAX_TASK_BATCHES_BEFORE_RESTART) {
+          await recycleScraper(scraper, `已连续处理 ${taskBatchCount} 批任务`);
+          browserRestartCount++;
+          taskBatchCount = 0;
+        }
 
-      if (needRestart) {
-        attachLog("[Attach] 检测到浏览器异常，将在下一轮重启...");
+        if (needRestart) {
+          attachLog("[Attach] 检测到浏览器异常，将在下一轮重启...");
+        }
       }
 
       await new Promise((r) => setTimeout(r, 500));
@@ -367,7 +398,7 @@ export async function handleAttach(options) {
     process.removeListener("SIGINT", onSigint);
     process.removeListener("SIGTERM", onSigterm);
     // 只有在非信号退出时才关闭浏览器（信号退出已在循环内关闭）
-    if (!shuttingDown) {
+    if (!shuttingDown && scraper) {
       await scraper.close().catch(() => {});
     }
   }

package/src/lib/args.js

@@ -627,6 +627,7 @@ function parseAttachArgs(args) {
   let countries = [];
   let customProxy = null;
   let poolSize = 3;
+  let attachWebserver = null;
 
   for (let i = 0; i < args.length; i++) {
     const arg = args[i];
@@ -645,6 +646,8 @@ function parseAttachArgs(args) {
       customProxy = args[++i];
     } else if (arg === "--size") {
       poolSize = parseInt(args[++i], 10) || 3;
+    } else if (arg === "--webserver") {
+      attachWebserver = args[++i];
     }
   }
 
@@ -656,6 +659,7 @@ function parseAttachArgs(args) {
     attachCountries: countries,
     customProxy,
     attachPoolSize: poolSize,
+    attachWebserver,
     urls: [],
     outputFormat: "json",
     exploreCount: 0,

package/src/lib/constants.js

@@ -188,12 +188,13 @@ const HELP_TEXT = [
   "    示例: tt-help info https://www.tiktok.com/@nike",
   "          tt-help info https://www.tiktok.com/@nike/video/7234567890 --onlyvideo",
   "",
-  "  tt-help attach [-p 并行数] [-i 间隔秒数] [-s 服务端地址] [-c 国家列表]",
+  "  tt-help attach [-p 并行数] [-i 间隔秒数] [-s 服务端地址] [-c 国家列表] [--webserver URL]",
   "    后台轮询服务端任务接口，自动抓取 TikTok 用户信息",
   "    -p, --parallel <N>     并行抓取数（默认: 1）",
   "    -i, --interval <N>     无任务时轮询间隔，单位秒（默认: 10）",
   "    -s, --server <URL>     服务端地址（默认: http://127.0.0.1:3001）",
   "    -c, --countries <A,B>  猜测国家列表（逗号分隔，如 PL,DE,FR）",
+  "    --webserver <URL>      webserver 模式（不启动本地浏览器，走 HTTP 接口）",
   "    示例: tt-help attach -p 5 -i 10",
   "          tt-help attach -c PL,DE,FR -p 5",
   "",

package/src/webserver/server.mjs

@@ -222,14 +222,15 @@ async function handleTiktokProduct(req, res) {
     );
     return sendJSON(res, 200, result);
   } catch (e) {
-    // 先返回错误，不阻塞请求
     console.error(`[webserver] product FAIL | ${clientIP} | ${e.message}`);
-    sendJSON(res, 502, { error: e.message });
 
-    // 如果是 captcha，异步弹浏览器（不阻塞其他请求）
+    // 如果是 captcha，先触发上报+恢复（不阻塞其他请求）
+    // 必须在 sendJSON 之前调用，否则 sendJSON 抛异常会导致验证码通知丢失
     if (e.message?.includes("Blocked by TikTok")) {
       scheduleCaptchaRecovery(normalizedUrl, SCRAPER_CONFIG.proxyServer);
     }
+
+    sendJSON(res, 502, { error: e.message });
     return;
   }
 }
@@ -252,7 +253,7 @@ async function scheduleCaptchaRecovery(url, proxy) {
         headers: { "Content-Type": "application/json" },
         body: JSON.stringify({
           userId: `webserver-${process.env.HOSTNAME || os.hostname() || "unknown"}`,
-          errorType: "captcha_blocked",
+          errorType: "captcha",
           errorMessage:
             "TikTok Shop captcha detected, browser opened - manual intervention required",
           stage: "product_info",