tt-help-cli-ycl 1.3.80 → 1.3.82

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tt-help-cli-ycl",
-  "version": "1.3.80",
+  "version": "1.3.82",
   "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
   "type": "module",
   "bin": {

package/src/cli/refresh.js

@@ -549,6 +549,7 @@ export async function handleRefresh(options) {
 
       const payload = {
         userInfo: refreshUserInfo,
+        topRecentVideo: result.topRecentVideo || null,
         discoveredFollowing: (result.discoveredFollowing || []).map((f) => ({
           handle: Array.isArray(f) ? f[0] : f,
           displayName: Array.isArray(f) ? f[1] : null,

package/src/cli/watch.js

@@ -5,19 +5,40 @@ import { startWatchServer, openBrowser } from "../watch/server.js";
 
 export async function handleWatch(options) {
   const dataAnchor = options.dataAnchor || options.outputFile;
-  const { watchPort } = options;
+  const { watchPort, llmRefill, llmRefillMinScore, llmRefillSampleSize } =
+    options;
 
   if (!dataAnchor) {
-    console.error("用法: tt-help watch -o <db路径> [-p 端口]");
+    console.error("用法: tt-help watch -o <db路径> [-p 端口] [--llm-refill]");
     console.error("      tt-help watch -o data/result.db");
     console.error("      tt-help watch -o data/result.db -p 8080");
+    console.error(
+      "      tt-help watch -o data/result.db --llm-refill --llm-refill-min 60",
+    );
     process.exit(1);
   }
 
   mkdirSync(path.dirname(path.resolve(dataAnchor)), { recursive: true });
 
-  const store = createStore(dataAnchor);
-  const { server, port } = await startWatchServer(dataAnchor, watchPort, store);
+  const storeOptions = {};
+  if (llmRefill) {
+    storeOptions.refillLlm = {
+      llmScore: true,
+      llmMinScore: llmRefillMinScore ?? 60,
+      llmSampleSize: llmRefillSampleSize ?? 100,
+    };
+    console.error(
+      `[watch] LLM 自动打分已启用: 最低分 ${storeOptions.refillLlm.llmMinScore}, 采样 ${storeOptions.refillLlm.llmSampleSize}`,
+    );
+  }
+
+  const store = createStore(dataAnchor, storeOptions);
+  const { server, port } = await startWatchServer(
+    dataAnchor,
+    watchPort,
+    store,
+    storeOptions,
+  );
   openBrowser(port);
 
   process.once("SIGINT", () => {

package/src/lib/api-interceptor.js

@@ -18,7 +18,12 @@ async function processAPIResponse(
   for (const item of firstPageItems) {
     if (items.length >= maxVideos) break;
     const href = `https://www.tiktok.com/@${username}/video/${item.id}`;
-    items.push({ id: item.id, href, createTime: item.createTime || null });
+    items.push({
+      id: item.id,
+      href,
+      createTime: item.createTime || null,
+      playCount: item.stats?.playCount || 0,
+    });
   }
 
   let cursor = data.cursor;
@@ -32,7 +37,7 @@ async function processAPIResponse(
 
     try {
       const pageData = await (() => {
-        // 重试包装：处理页面导航导致的执行上下文销毁
+        // 重试包装：处理页面导航导致的执行上下文销毁和 CDP 断连
         const tryEval = async (retries = 3) => {
           for (let i = 0; i < retries; i++) {
             try {
@@ -42,8 +47,11 @@ async function processAPIResponse(
               }, newUrl);
             } catch (e) {
               if (
-                e.message.includes("Execution context was destroyed") &&
-                i < retries - 1
+                e.message &&
+                (e.message.includes("Execution context was destroyed") ||
+                  e.message.includes("Target closed") ||
+                  e.message.includes("Connection closed") ||
+                  e.message.includes("Protocol error"))
               ) {
                 await delay(500 * (i + 1), 500 * (i + 1));
               } else {
@@ -63,6 +71,7 @@ async function processAPIResponse(
             id: item.id,
             href,
             createTime: item.createTime || null,
+            playCount: item.stats?.playCount || 0,
           });
         }
       }
@@ -89,6 +98,14 @@ async function processAPIResponse(
 async function fetchUserVideosAPI(page, username, maxVideos, log) {
   log(`  [API拦截] 获取 @${username} 视频 ...`);
 
+  // CDP 健康检查：确保 page 可用
+  try {
+    await page.evaluate(() => 1);
+  } catch (e) {
+    log(`  [API拦截] CDP 连接异常: ${e.message}`);
+    throw new Error(`CDP 连接异常: ${e.message}`);
+  }
+
   let apiRequestUrl = null;
   let sawApiRequest = false;
 

package/src/lib/args.js

@@ -347,6 +347,9 @@ function parseInfoArgs(args) {
 function parseWatchArgs(args) {
   let dataAnchor = "./result.db";
   let watchPort = 3001;
+  let llmRefill = true;
+  let llmRefillMinScore = null;
+  let llmRefillSampleSize = null;
 
   for (let i = 0; i < args.length; i++) {
     const arg = args[i];
@@ -354,6 +357,14 @@ function parseWatchArgs(args) {
       dataAnchor = args[++i];
     } else if (arg === "-p") {
       watchPort = parseInt(args[++i]) || 3001;
+    } else if (arg === "--llm-refill") {
+      llmRefill = true;
+    } else if (arg === "--no-llm-refill") {
+      llmRefill = false;
+    } else if (arg === "--llm-refill-min") {
+      llmRefillMinScore = parseInt(args[++i]) || 60;
+    } else if (arg === "--llm-refill-sample") {
+      llmRefillSampleSize = parseInt(args[++i]) || 100;
     }
   }
 
@@ -362,6 +373,9 @@ function parseWatchArgs(args) {
     outputFile: dataAnchor,
     dataAnchor,
     watchPort,
+    llmRefill,
+    llmRefillMinScore,
+    llmRefillSampleSize,
     urls: [],
     outputFormat: "json",
     exploreCount: 0,

package/src/lib/page-error-detector.js

@@ -54,23 +54,40 @@ const PATTERNS = {
   service_error: ["出错了", "很抱歉"],
 };
 
-export async function detectPageError(page) {
-  return page.evaluate((patterns) => {
-    const body = document.body;
-    if (!body) return null;
-    const bodyText = body.innerText;
-    const lower = bodyText.toLowerCase();
+export async function detectPageError(page, timeout = 10000) {
+  try {
+    return await page.evaluate(
+      (patterns) => {
+        const body = document.body;
+        if (!body) return null;
+        const bodyText = body.innerText;
+        const lower = bodyText.toLowerCase();
 
-    for (const [type, phrases] of Object.entries(patterns)) {
-      for (const phrase of phrases) {
-        if (lower.includes(phrase.toLowerCase())) {
-          return type;
+        for (const [type, phrases] of Object.entries(patterns)) {
+          for (const phrase of phrases) {
+            if (lower.includes(phrase.toLowerCase())) {
+              return type;
+            }
+          }
         }
-      }
-    }
 
-    return null;
-  }, PATTERNS);
+        return null;
+      },
+      PATTERNS,
+    );
+  } catch (e) {
+    // CDP 断连或超时：返回 null 而非永久挂起
+    if (
+      e.message &&
+      (e.message.includes("Timeout") ||
+        e.message.includes("Target closed") ||
+        e.message.includes("Connection closed") ||
+        e.message.includes("Protocol error"))
+    ) {
+      return null;
+    }
+    throw e;
+  }
 }
 
 /**

package/src/lib/scroll-collector.js

@@ -30,7 +30,7 @@ async function doCollect(
       }
 
       const fn = eval("(" + fnStr + ")");
-      return fn(el, args);
+      return args !== undefined ? fn(el, args) : fn(el);
     },
     { fn: fnStr, containerSelector: container, findScrollableFlag: findScrollable, args: extraArgs },
   );

package/src/scraper/explore-core.js

@@ -8,6 +8,7 @@ import {
   DEFAULT_TARGET_LOCATIONS_CSV,
   findFirstMatchingLocation,
   isLocationInList,
+  normalizeLocation,
   normalizeLocationList,
 } from "../lib/target-locations.js";
 import {
@@ -40,6 +41,7 @@ async function processExplore(page, username, options, log) {
     keepFollow: false,
     locationCreated: null,
     latestVideoTime: null,
+    topRecentVideo: null,
     noVideo: false,
     restricted: false,
     error: null,
@@ -84,6 +86,27 @@ async function processExplore(page, username, options, log) {
       if (result.userInfo) result.userInfo.latestVideoTime = latestCreateTime;
     }
 
+    // 找出 7 天内发布且播放量最大的视频
+    const SEVEN_DAYS_SECONDS = 7 * 24 * 60 * 60;
+    const nowSeconds = Math.floor(Date.now() / 1000);
+    const recentVideos = videoArray.filter(
+      (v) => v.createTime && nowSeconds - v.createTime <= SEVEN_DAYS_SECONDS,
+    );
+    if (recentVideos.length > 0) {
+      const topVideo = recentVideos.reduce((max, v) =>
+        v.playCount > max.playCount ? v : max,
+      );
+      result.topRecentVideo = {
+        id: topVideo.id,
+        href: topVideo.href,
+        playCount: topVideo.playCount,
+        createTime: topVideo.createTime,
+      };
+      log(
+        `  7天内最高播放视频: ${topVideo.playCount} 次播放 (${recentVideos.length} 个候选)`,
+      );
+    }
+
     if (videoArray.length <= 0) {
       // 视频为空：可能是页面受限或用户真的没有视频
       result.processed = true;
@@ -110,7 +133,10 @@ async function processExplore(page, username, options, log) {
       log(
         `  国家采样(${locations.length}个): [${locations.filter(Boolean).join(", ") || "无数据"}]`,
       );
-      const normalizedLocations = normalizeLocationList(locations, []);
+      // 直接标准化，不去重（保留重复值用于频率统计）
+      const normalizedLocations = locations
+        .map((loc) => normalizeLocation(loc))
+        .filter(Boolean);
 
       // 统计频率
       const freq = {};

package/src/videos/core.js

@@ -7,14 +7,18 @@ import {
 import { fetchUserVideosAPI } from "../lib/api-interceptor.js";
 
 async function getUserInfo(page) {
-  // 重试包装：处理页面导航导致的执行上下文销毁
+  // 重试包装：处理页面导航导致的执行上下文销毁和 CDP 断连
   const evaluateWithRetry = async (fn, retries = 3) => {
     for (let i = 0; i < retries; i++) {
       try {
         return await page.evaluate(fn);
       } catch (e) {
         if (
-          e.message.includes("Execution context was destroyed") &&
+          e.message &&
+          (e.message.includes("Execution context was destroyed") ||
+            e.message.includes("Target closed") ||
+            e.message.includes("Connection closed") ||
+            e.message.includes("Protocol error")) &&
           i < retries - 1
         ) {
           await new Promise((r) => setTimeout(r, 500 * (i + 1)));