tt-help-cli-ycl 1.3.22 → 1.3.24

package/src/cli/watch.js

@@ -1,31 +1,31 @@
-import { existsSync } from 'fs';
-import { createStore } from '../watch/data-store.js';
-import { startWatchServer, openBrowser } from '../watch/server.js';
-
-export async function handleWatch(options) {
-  const { outputFile, watchPort } = options;
-
-  if (!outputFile) {
-    console.error('用法: tt-help watch -o <数据文件> [-p 端口]');
-    console.error('示例: tt-help watch -o data.json');
-    console.error('      tt-help watch -o data.json -p 8080');
-    process.exit(1);
-  }
-
-  if (!existsSync(outputFile)) {
-    console.error(`文件不存�? ${outputFile}`);
-    process.exit(1);
-  }
-
-  const store = createStore(outputFile);
-  const { server, port } = await startWatchServer(outputFile, watchPort, store);
-  openBrowser(port);
-
-  process.once('SIGINT', () => {
-    store.stopBackup();
-    server.close();
-    process.exit(0);
-  });
-
-  console.error('�?Ctrl+C 停止监控服务');
-}
+import { existsSync } from 'fs';
+import { createStore } from '../watch/data-store.js';
+import { startWatchServer, openBrowser } from '../watch/server.js';
+
+export async function handleWatch(options) {
+  const { outputFile, watchPort } = options;
+
+  if (!outputFile) {
+    console.error('用法: tt-help watch -o <数据文件> [-p 端口]');
+    console.error('示例: tt-help watch -o data.json');
+    console.error('      tt-help watch -o data.json -p 8080');
+    process.exit(1);
+  }
+
+  if (!existsSync(outputFile)) {
+    console.error(`文件不存�? ${outputFile}`);
+    process.exit(1);
+  }
+
+  const store = createStore(outputFile);
+  const { server, port } = await startWatchServer(outputFile, watchPort, store);
+  openBrowser(port);
+
+  process.once('SIGINT', () => {
+    store.stopBackup();
+    server.close();
+    process.exit(0);
+  });
+
+  console.error('�?Ctrl+C 停止监控服务');
+}

package/src/lib/api-interceptor.js

@@ -1,129 +1,191 @@
-import { delay } from './delay.js';
-import { retryWithBackoff } from './retry.js';
-import { assertPageUrl } from './browser/page.js';
-
-/**
- * 通过拦截 TikTok 内部 API 获取用户视频列表
- * 比 DOM 滚动解析快 5-10 倍
- *
- * @param {import('playwright').Page} page - Playwright page (CDP 连接)
- * @param {string} username - TikTok 用户名
- * @param {number} maxVideos - 最大视频数
- * @param {Function} log - 日志函数
- * @returns {Map<string, {id: string, href: string}>} 与 collectVideos 返回格式一致
- */
-async function fetchUserVideosAPI(page, username, maxVideos, log) {
-  const url = `https://www.tiktok.com/@${username}`;
-  const items = [];
-
-  // 1. 注册 response 拦截器 + request URL 捕获
-  let apiResolve = null;
-  const apiPromise = new Promise(r => { apiResolve = r; });
-
-  let apiRequestUrl = null;
-
-  const responseHandler = async (response) => {
-    if (response.url().includes('/api/post/item_list/')) {
-      try {
-        apiResolve(await response.json());
-      } catch (e) {
-        apiResolve(null);
-      }
-    }
-  };
-
-  const requestHandler = (request) => {
-    if (request.url().includes('/api/post/item_list/') && !apiRequestUrl) {
-      apiRequestUrl = request.url();
-    }
-  };
-
-  page.on('response', responseHandler);
-  page.on('request', requestHandler);
-
-  try {
-    // 2. 导航并等待 API 响应
-    log('  [API拦截] 导航到用户页，等待 /api/post/item_list/ ...');
-    const t0 = Date.now();
-
-    await retryWithBackoff(async () => {
-      await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
-      await assertPageUrl(page, `@${username}`);
-    }, { maxRetries: 3, baseDelay: 3000, log });
-
-    const data = await Promise.race([
-      apiPromise,
-      new Promise(r => setTimeout(() => r(null), 8000)),
-    ]);
-
-    const elapsed = Date.now() - t0;
-
-    if (!data || !data.itemList) {
-      log(`  [API拦截] ${elapsed}ms 后未拿到 API 数据`);
-      return null;
-    }
-
-    // 3. 提取首页视频
-    const firstPageItems = data.itemList || [];
-    for (const item of firstPageItems) {
-      if (items.length >= maxVideos) break;
-      const href = `https://www.tiktok.com/@${username}/video/${item.id}`;
-      items.push({ id: item.id, href });
-    }
-
-    log(`  [API拦截] ${elapsed}ms 获取首页 ${firstPageItems.length} 条视频`);
-
-    // 4. 翻页获取后续视频
-    let cursor = data.cursor;
-    let hasMore = data.hasMore;
-
-    while (hasMore && cursor && items.length < maxVideos) {
-      if (!apiRequestUrl) {
-        log('  [API拦截] 未捕获到 API 请求 URL，无法翻页');
-        break;
-      }
-
-      const newUrl = apiRequestUrl.replace(/cursor=\d+/, `cursor=${cursor}`);
-
-      try {
-        const pageData = await page.evaluate(async (u) => {
-          const res = await fetch(u);
-          return await res.json();
-        }, newUrl);
-
-        if (pageData && pageData.itemList) {
-          for (const item of pageData.itemList) {
-            if (items.length >= maxVideos) break;
-            const href = `https://www.tiktok.com/@${username}/video/${item.id}`;
-            items.push({ id: item.id, href });
-          }
-          log(`  [API拦截] 翻页 cursor=${cursor}，获取 ${pageData.itemList.length} 条，累计 ${items.length}`);
-        }
-
-        cursor = pageData.cursor;
-        hasMore = pageData.hasMore;
-      } catch (e) {
-        log(`  [API拦截] 翻页失败: ${e.message}`);
-        break;
-      }
-
-      await delay(300, 600);
-    }
-
-    log(`  [API拦截] 总计获取 ${items.length} 条视频`);
-
-    // 转成 Map 返回，与 collectVideos 一致
-    const videoMap = new Map();
-    for (const v of items) {
-      if (!videoMap.has(v.id)) videoMap.set(v.id, v);
-    }
-
-    return videoMap;
-  } finally {
-    // 5. 必须清理拦截器，防止累积
-    page.off('response', responseHandler);
-    page.off('request', requestHandler);
-  }
-}
-
-export { fetchUserVideosAPI };
+import { delay } from './delay.js';
+import { retryWithBackoff } from './retry.js';
+import { assertPageUrl } from './browser/page.js';
+
+/**
+ * 处理 API 响应数据：提取首页视频 + 翻页
+ */
+async function processAPIResponse(data, username, maxVideos, items, page, apiRequestUrl, log) {
+  // 提取首页视频
+  const firstPageItems = data.itemList || [];
+  for (const item of firstPageItems) {
+    if (items.length >= maxVideos) break;
+    const href = `https://www.tiktok.com/@${username}/video/${item.id}`;
+    items.push({ id: item.id, href });
+  }
+
+  log(`  [API拦截] 获取首页 ${firstPageItems.length} 条视频`);
+
+  // 翻页获取后续视频
+  let cursor = data.cursor;
+  let hasMore = data.hasMore;
+
+  while (hasMore && cursor && items.length < maxVideos) {
+    const reqUrl = apiRequestUrl || data.apiRequestUrl;
+    if (!reqUrl) {
+      log('  [API拦截] 未捕获到 API 请求 URL，无法翻页');
+      break;
+    }
+
+    const newUrl = reqUrl.replace(/cursor=\d+/, `cursor=${cursor}`);
+
+    try {
+      const pageData = await page.evaluate(async (u) => {
+        const res = await fetch(u);
+        return await res.json();
+      }, newUrl);
+
+      if (pageData && pageData.itemList) {
+        for (const item of pageData.itemList) {
+          if (items.length >= maxVideos) break;
+          const href = `https://www.tiktok.com/@${username}/video/${item.id}`;
+          items.push({ id: item.id, href });
+        }
+        log(`  [API拦截] 翻页 cursor=${cursor}，获取 ${pageData.itemList.length} 条，累计 ${items.length}`);
+      }
+
+      cursor = pageData.cursor;
+      hasMore = pageData.hasMore;
+    } catch (e) {
+      log(`  [API拦截] 翻页失败: ${e.message}`);
+      break;
+    }
+
+    await delay(300, 600);
+  }
+
+  log(`  [API拦截] 总计获取 ${items.length} 条视频`);
+
+  // 转成 Map 返回，与 collectVideos 一致
+  const videoMap = new Map();
+  for (const v of items) {
+    if (!videoMap.has(v.id)) videoMap.set(v.id, v);
+  }
+
+  return videoMap;
+}
+
+/**
+ * 通过拦截 TikTok 内部 API 获取用户视频列表
+ * 比 DOM 滚动解析快 5-10 倍
+ *
+ * @param {import('playwright').Page} page - Playwright page (CDP 连接)
+ * @param {string} username - TikTok 用户名
+ * @param {number} maxVideos - 最大视频数
+ * @param {Function} log - 日志函数
+ * @returns {Map<string, {id: string, href: string}>} 与 collectVideos 返回格式一致
+ * @throws {Error} API 超时或网络异常时抛出
+ */
+async function fetchUserVideosAPI(page, username, maxVideos, log) {
+  const url = `https://www.tiktok.com/@${username}`;
+  const items = [];
+
+  // 1. 注册 response 拦截器 + request URL 捕获
+  let apiResolve = null;
+  const apiPromise = new Promise(r => { apiResolve = r; });
+
+  let apiRequestUrl = null;
+
+  const responseHandler = async (response) => {
+    if (response.url().includes('/api/post/item_list/')) {
+      try {
+        apiResolve(await response.json());
+      } catch (e) {
+        apiResolve(null);
+      }
+    }
+  };
+
+  const requestHandler = (request) => {
+    if (request.url().includes('/api/post/item_list/') && !apiRequestUrl) {
+      apiRequestUrl = request.url();
+    }
+  };
+
+  page.on('response', responseHandler);
+  page.on('request', requestHandler);
+
+  try {
+    // 2. 导航并等待 API 响应
+    log('  [API拦截] 导航到用户页，等待 /api/post/item_list/ ...');
+    const t0 = Date.now();
+
+    await retryWithBackoff(async () => {
+      await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
+      await assertPageUrl(page, `@${username}`);
+    }, { maxRetries: 3, baseDelay: 3000, log });
+
+    const data = await Promise.race([
+      apiPromise,
+      new Promise(r => setTimeout(() => r(null), 8000)),
+    ]);
+
+    const elapsed = Date.now() - t0;
+
+    if (!data || !data.itemList) {
+      log(`  [API拦截] ${elapsed}ms 后未拿到 API 数据，重试...`);
+      // 清理后重试
+      page.off('response', responseHandler);
+      page.off('request', requestHandler);
+
+      // 重试：重新注册拦截器并导航
+      let apiResolve2 = null;
+      const apiPromise2 = new Promise(r => { apiResolve2 = r; });
+      let apiRequestUrl2 = null;
+
+      const responseHandler2 = async (response) => {
+        if (response.url().includes('/api/post/item_list/')) {
+          try {
+            const json = await response.json().catch(() => null);
+            apiResolve2(json);
+          } catch (e) {
+            apiResolve2(null);
+          }
+        }
+      };
+
+      const requestHandler2 = (request) => {
+        if (request.url().includes('/api/post/item_list/') && !apiRequestUrl2) {
+          apiRequestUrl2 = request.url();
+        }
+      };
+
+      page.on('response', responseHandler2);
+      page.on('request', requestHandler2);
+
+      try {
+        await retryWithBackoff(
+          async () => {
+            await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
+            assertPageUrl(page, url);
+          },
+          { retries: 1, log },
+        );
+
+        const data2 = await Promise.race([
+          apiPromise2,
+          new Promise(r => setTimeout(() => r(null), 8000)),
+        ]);
+        const elapsed2 = Date.now() - t0;
+
+        if (!data2 || !data2.itemList) {
+          throw new Error(`[API拦截] ${elapsed2}ms 后仍未拿到 API 数据（含重试），请检查网络`);
+        }
+
+        return await processAPIResponse(data2, username, maxVideos, items, page, apiRequestUrl2, log);
+      } finally {
+        page.off('response', responseHandler2);
+        page.off('request', requestHandler2);
+      }
+    }
+
+    // 3. 处理数据（提取首页 + 翻页）
+    return await processAPIResponse(data, username, maxVideos, items, page, apiRequestUrl, log);
+  } finally {
+    // 5. 必须清理拦截器，防止累积
+    page.off('response', responseHandler);
+    page.off('request', requestHandler);
+  }
+}
+
+export { fetchUserVideosAPI };