tt-help-cli-ycl 1.4.29 → 1.4.31

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.4.29",
+  "version": "1.4.31",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/cli/comments.js

@@ -736,15 +736,13 @@ async function runAutoMode(options) {
             try {
               const modals = await detectModals(session.page);
               const loginModal = modals.find(
-                (m) =>
-                  m.type === "login-modal" || m.type === "modal-container",
+                (m) => m.type === "login-modal" || m.type === "modal-container",
               );
               if (loginModal) {
                 console.error(`        🔔 检测到登录弹窗，正在关闭...`);
-                const { closed, failed } = await closeAllModals(
-                  session.page,
-                  { waitMs: 300 },
-                );
+                const { closed, failed } = await closeAllModals(session.page, {
+                  waitMs: 300,
+                });
                 console.error(
                   `        🔔 弹窗关闭: closed=${JSON.stringify(closed)} failed=${JSON.stringify(failed)}`,
                 );

package/src/cli/search.js

@@ -71,6 +71,7 @@ export async function handleSearch(parsed) {
     serverUrl,
     basePort = 3333,
     portCount = 10,
+    parallel = 1,
     proxy: cliProxy,
   } = searchOpts || {};
 
@@ -196,7 +197,9 @@ export async function handleSearch(parsed) {
       try {
         return await p.evaluate(async () => {
           try {
-            const res = await fetch("https://www.tiktok.com/passport/web/account/info/");
+            const res = await fetch(
+              "https://www.tiktok.com/passport/web/account/info/",
+            );
             const data = await res.json();
             return data?.data?.unique_id || null;
           } catch {}
@@ -410,6 +413,7 @@ export async function handleSearch(parsed) {
         let searchError = null;
         try {
           ids = await collectSearchAuthors(browser, tag, {
+            parallel,
             log,
             exclude: myUsername ? [myUsername] : [],
           });
@@ -432,6 +436,7 @@ export async function handleSearch(parsed) {
           // 重试搜索
           try {
             ids = await collectSearchAuthors(browser, tag, {
+              parallel,
               log,
               exclude: myUsername ? [myUsername] : [],
             });
@@ -457,6 +462,7 @@ export async function handleSearch(parsed) {
             await new Promise((r) => setTimeout(r, 5000));
             try {
               ids = await collectSearchAuthors(browser, tag, {
+                parallel,
                 log,
                 exclude: myUsername ? [myUsername] : [],
               });

package/src/lib/args.js

@@ -560,6 +560,7 @@ function parseSearchArgs(args) {
   let serverUrl = null;
   let basePort = 3333;
   let portCount = 10;
+  let parallel = 1;
   let proxy = null;
 
   for (let i = 0; i < args.length; i++) {
@@ -570,6 +571,8 @@ function parseSearchArgs(args) {
       basePort = parseInt(args[++i]) || 3333;
     } else if (arg === "--port-count") {
       portCount = parseInt(args[++i]) || 10;
+    } else if (arg === "-p" || arg === "--parallel") {
+      parallel = parseInt(args[++i]) || 1;
     } else if (arg === "--proxy") {
       proxy = args[++i];
     }
@@ -577,7 +580,7 @@ function parseSearchArgs(args) {
 
   return {
     subcommand: "search",
-    search: { serverUrl, basePort, portCount, proxy },
+    search: { serverUrl, basePort, portCount, parallel, proxy },
     urls: [],
     outputFormat: "json",
     exploreCount: 0,

package/src/lib/constants.js

@@ -316,19 +316,21 @@ const HELP_TEXT = [
   "",
   "  search [选项]",
   "    基于 productive 标签搜索 TikTok 作者（score>70 的标签作为搜索词）",
-  "    5 个搜索 Tab 并行采集（综合/视频/用户/直播/照片），去重后提交到 job_base",
+  "    4 个搜索 Tab 采集（综合/视频/用户/照片），去重后提交到 job_base",
   "    选项:",
   "      -s, --server <URL>        服务端地址（默认 http://127.0.0.1:3000）",
   "      --base-port <端口号>      起始 CDP 端口，默认 3333",
   "      --port-count <数量>       端口数量（账户数），默认 10",
+  "      -p, --parallel <N>        并行搜索数，默认 1（串行）",
   "      --proxy <地址>            浏览器代理",
   "    行为:",
   "      - 从 tags 表取 score>70 的 productive 标签",
   "      - 去掉 # 号后作为搜索关键词",
-  "      - 并行搜索 5 个 tab，去重后提交到 job_base（来源=search）",
+  "      - 搜索 4 个 tab，去重后提交到 job_base（来源=search）",
   "      - 猜测国家取标签 countries 数组的第一个",
   "    示例: tt-help search",
   "          tt-help search -s http://127.0.0.1:3001",
+  "          tt-help search -p 2 -s http://127.0.0.1:3001",
   "          tt-help search --base-port 3333 -s http://127.0.0.1:3001",
   "          tt-help search --proxy http://127.0.0.1:7890",
   "",

package/src/lib/search-collector.js

@@ -14,7 +14,7 @@ const TAB_URLS = {
   照片: (q) => `https://www.tiktok.com/search/photo?q=${encodeURIComponent(q)}`,
 };
 
-const DEFAULT_TABS = ["综合", "视频", "用户", "直播", "照片"];
+const DEFAULT_TABS = ["综合", "视频", "用户", "照片"];
 
 const SCROLL = {
   maxRounds: 60,
@@ -58,6 +58,17 @@ async function scrollPage(page) {
   });
 }
 
+async function waitForSearchResults(page, timeoutMs = 20000) {
+  try {
+    await page.waitForSelector('[class*="DivItemContainer"]', {
+      timeout: timeoutMs,
+    });
+    return true;
+  } catch {
+    return false;
+  }
+}
+
 async function processOneTab(page, tabName, keyword, allIds, excludeSet, log) {
   const urlFn = TAB_URLS[tabName];
   if (!urlFn) {
@@ -68,7 +79,7 @@ async function processOneTab(page, tabName, keyword, allIds, excludeSet, log) {
   const tag = `[${tabName}]`;
   const url = urlFn(keyword);
 
-  // 首屏正常加载
+  // 首屏加载
   log(`${tag} 加载...`);
   try {
     await page.goto(url, { waitUntil: "domcontentloaded", timeout: 30000 });
@@ -76,7 +87,12 @@ async function processOneTab(page, tabName, keyword, allIds, excludeSet, log) {
     log(`${tag} 导航失败: ${e.message}`);
     return;
   }
-  await delay(4000);
+
+  // 等待搜索结果实际加载（最长 20 秒）
+  const ready = await waitForSearchResults(page, 20000);
+  if (!ready) {
+    log(`${tag} 等待超时，继续采集`);
+  }
 
   // 首屏渲染完成后屏蔽图片/字体/媒体，加速后续滚动
   await page.route("**/*", (route) => {
@@ -143,43 +159,41 @@ async function processOneTab(page, tabName, keyword, allIds, excludeSet, log) {
  */
 export async function collectSearchAuthors(browser, keyword, opts = {}) {
   const tabs = opts.tabs || DEFAULT_TABS;
+  const parallel = Math.max(1, opts.parallel || 1);
   const log = opts.log || ((...args) => console.error(...args));
   const excludeSet = new Set(opts.exclude || []);
 
-  log(`[collect] "${keyword}"  Tab: [${tabs.join(", ")}]`);
+  log(`[collect] "${keyword}"  Tab: [${tabs.join(", ")}]  并行: ${parallel}`);
   if (excludeSet.size > 0)
     log(`[collect] 排除: [${[...excludeSet].join(", ")}]`);
 
   const context = browser.contexts()[0];
-  const existingPage = context.pages()[0];
+  const mainPage = context.pages()[0];
   const allIds = new Set();
 
-  // 并行：每个 tab 独立页面同时跑
-  const newPages = []; // 追踪所有新建页面，确保清理
-  const tasks = tabs.map(async (tabName, i) => {
-    let page = null;
-    try {
-      page = i === 0 ? existingPage : await context.newPage();
-      if (i > 0) newPages.push(page);
-      await processOneTab(page, tabName, keyword, allIds, excludeSet, log);
-    } catch (e) {
-      log(`[${tabName}] 异常: ${e.message}`);
-    } finally {
-      if (i > 0 && page) {
-        try {
-          await page.close();
-        } catch {}
+  // 分批并行：每批 parallel 个 tab，批内并行、批间串行
+  for (let i = 0; i < tabs.length; i += parallel) {
+    const batch = tabs.slice(i, i + parallel);
+    const batchPages = [];
+
+    const tasks = batch.map(async (tabName, j) => {
+      let page = null;
+      try {
+        page = j === 0 ? mainPage : await context.newPage();
+        if (j > 0) batchPages.push(page);
+        await processOneTab(page, tabName, keyword, allIds, excludeSet, log);
+      } catch (e) {
+        log(`[${tabName}] 异常: ${e.message}`);
+      } finally {
+        if (j > 0 && page) {
+          try {
+            await page.close();
+          } catch {}
+        }
       }
-    }
-  });
-
-  await Promise.all(tasks);
+    });
 
-  // 兜底：确保所有新建页面都已关闭
-  for (const p of newPages) {
-    try {
-      if (!p.isClosed()) await p.close();
-    } catch {}
+    await Promise.all(tasks);
   }
 
   return [...allIds].sort();