tt-help-cli-ycl 1.3.12 → 1.3.14

package/src/lib/retry.js

@@ -1,45 +1,45 @@
-import { delay } from './delay.js';
-
-const RETRYABLE_PATTERNS = [
-  'interrupted',
-  'Navigation.*interrupted',
-  'net::',
-  'ECONN',
-  'ETIMEDOUT',
-  'ENOTFOUND',
-  'EAI_AGAIN',
-  'ESOCKETRESET',
-  'connection.*refused',
-  'connection.*reset',
-  'failed.*navigate',
-  'target.*closed',
-  'crash',
-  '代理错误',
-];
-
-export function isRetryableError(error) {
-  if (!error) return false;
-  const msg = error.message || error.toString() || '';
-  return RETRYABLE_PATTERNS.some(p => new RegExp(p, 'i').test(msg));
-}
-
-export async function retryWithBackoff(fn, { maxRetries = 3, baseDelay = 3000, log } = {}) {
-  let lastError;
-  for (let attempt = 0; attempt <= maxRetries; attempt++) {
-    try {
-      return await fn();
-    } catch (error) {
-      lastError = error;
-      if (attempt >= maxRetries || !isRetryableError(error)) {
-        throw error;
-      }
-      const jitter = Math.random() * 2000;
-      const waitTime = baseDelay * Math.pow(2, attempt) + jitter;
-      if (log) {
-        log(`  [重试] ${attempt + 1}/${maxRetries}，${Math.round(waitTime / 1000)}s 后重试...`);
-      }
-      await delay(Math.round(waitTime), Math.round(waitTime));
-    }
-  }
-  throw lastError;
-}
+import { delay } from './delay.js';
+
+const RETRYABLE_PATTERNS = [
+  'interrupted',
+  'Navigation.*interrupted',
+  'net::',
+  'ECONN',
+  'ETIMEDOUT',
+  'ENOTFOUND',
+  'EAI_AGAIN',
+  'ESOCKETRESET',
+  'connection.*refused',
+  'connection.*reset',
+  'failed.*navigate',
+  'target.*closed',
+  'crash',
+  '代理错误',
+];
+
+export function isRetryableError(error) {
+  if (!error) return false;
+  const msg = error.message || error.toString() || '';
+  return RETRYABLE_PATTERNS.some(p => new RegExp(p, 'i').test(msg));
+}
+
+export async function retryWithBackoff(fn, { maxRetries = 3, baseDelay = 3000, log } = {}) {
+  let lastError;
+  for (let attempt = 0; attempt <= maxRetries; attempt++) {
+    try {
+      return await fn();
+    } catch (error) {
+      lastError = error;
+      if (attempt >= maxRetries || !isRetryableError(error)) {
+        throw error;
+      }
+      const jitter = Math.random() * 2000;
+      const waitTime = baseDelay * Math.pow(2, attempt) + jitter;
+      if (log) {
+        log(`  [重试] ${attempt + 1}/${maxRetries}，${Math.round(waitTime / 1000)}s 后重试...`);
+      }
+      await delay(Math.round(waitTime), Math.round(waitTime));
+    }
+  }
+  throw lastError;
+}

package/src/lib/scrape.js

@@ -1,40 +1,89 @@
-import { extractUserSection, parseUserSection, extractLocationCreated } from './parser.js';
-import { fetchHtml, isProfileUrl } from './fetcher.js';
-import { toProfileUrl, isVideoUrl, extractUniqueId } from './url.js';
-
-export async function extractUserData(profileUrl, proxyUrl) {
-  const profileHtml = await fetchHtml(profileUrl, proxyUrl);
-  const section = extractUserSection(profileHtml);
-  if (!section) throw new Error('无法解析用户信息');
-  const data = parseUserSection(section);
-  data.locationCreated = extractLocationCreated(profileHtml);
-  return data;
-}
-
-export async function extractVideoLocation(videoUrl, proxyUrl) {
-  const videoHtml = await fetchHtml(videoUrl, proxyUrl);
-  return extractLocationCreated(videoHtml);
-}
-
-export async function processUrl(url, proxyUrl) {
-  if (isProfileUrl(url)) {
-    const profileUrl = toProfileUrl(url);
-    const profileData = await extractUserData(profileUrl, proxyUrl);
-    return [profileData];
-  }
-
-  if (isVideoUrl(url)) {
-    const profileHandle = extractUniqueId(url);
-    if (!profileHandle) throw new Error(`无法从视频URL提取用户主页: ${url}`);
-
-    const profileUrl = toProfileUrl(profileHandle);
-    const [profileData, locationCreated] = await Promise.all([
-      extractUserData(profileUrl, proxyUrl),
-      extractVideoLocation(url, proxyUrl),
-    ]);
-
-    return [{ ...profileData, locationCreated }];
-  }
-
-  return [];
-}
+import { TikTokScraper } from './tiktok-scraper.mjs';
+import { isProfileUrl, isVideoUrl, extractUniqueId, normalizeUsername } from './url.js';
+
+// Lazy singleton for TikTokScraper
+let scraperInstance = null;
+let scraperInitPromise = null;
+
+async function getScraper() {
+  if (scraperInstance) return scraperInstance;
+  if (scraperInitPromise) return scraperInitPromise;
+  scraperInitPromise = (async () => {
+    const scraper = new TikTokScraper();
+    await scraper.init();
+    scraperInstance = scraper;
+    scraperInitPromise = null;
+    return scraper;
+  })();
+  return scraperInitPromise;
+}
+
+export async function closeScraper() {
+  if (scraperInstance) {
+    await scraperInstance.close();
+    scraperInstance = null;
+  }
+}
+
+// Map parseUserInfo output to legacy parser.js format
+function mapUserInfo(user) {
+  if (!user) return null;
+  return {
+    uniqueId: user.uniqueId,
+    uid: user.id,
+    secUid: user.secUid,
+    nickname: user.nickname,
+    signature: user.bio,
+    ttSeller: user.ttSeller,
+    verified: user.verified,
+    followerCount: user.followerCount,
+    followingCount: user.followingCount,
+    heartCount: user.heartCount,
+    videoCount: user.videoCount,
+    diggCount: user.diggCount,
+    avatarLarger: user.avatar,
+    locationCreated: user.locationCreated,
+  };
+}
+
+// Map parseVideoInfo output to legacy format
+function mapVideoLocation(video) {
+  if (!video) return null;
+  return video.locationCreated;
+}
+
+export async function extractUserData(url) {
+  const scraper = await getScraper();
+  const uniqueId = extractUniqueId(url);
+  if (!uniqueId) throw new Error(`无法从URL提取用户名: ${url}`);
+  const user = await scraper.getUserInfo(normalizeUsername(uniqueId));
+  if (!user) throw new Error('无法解析用户信息');
+  return mapUserInfo(user);
+}
+
+export async function extractVideoLocation(videoUrl) {
+  const scraper = await getScraper();
+  const video = await scraper.getVideoInfo(videoUrl);
+  return mapVideoLocation(video);
+}
+
+export async function processUrl(url) {
+  if (isProfileUrl(url)) {
+    const profileData = await extractUserData(url);
+    return [profileData];
+  }
+
+  if (isVideoUrl(url)) {
+    const profileHandle = extractUniqueId(url);
+    if (!profileHandle) throw new Error(`无法从视频URL提取用户主页: ${url}`);
+
+    const [profileData, locationCreated] = await Promise.all([
+      extractUserData(url),
+      extractVideoLocation(url),
+    ]);
+
+    return [{ ...profileData, locationCreated }];
+  }
+
+  return [];
+}

package/src/lib/tiktok-scraper.mjs

@@ -0,0 +1,176 @@
+import { chromium } from 'playwright';
+import { detectBrowser } from './browser/launch.js';
+import { parseUserInfo, parseVideoInfo } from './parse-ssr.mjs';
+
+const DEFAULT_POOL_SIZE = 3;
+const DEFAULT_WAF_TTL = 120000;
+const DEFAULT_WARM_URL = 'https://www.tiktok.com/@nike';
+
+function delay(ms) {
+  return new Promise(r => setTimeout(r, ms));
+}
+
+class PageSlot {
+  constructor(page) {
+    this.page = page;
+    this.lock = new PromiseQueue();
+  }
+}
+
+class PromiseQueue {
+  constructor() {
+    this._queue = [];
+    this._processing = false;
+  }
+  async run(task) {
+    return new Promise((resolve, reject) => {
+      this._queue.push({ task, resolve, reject });
+      this._process();
+    });
+  }
+  async _process() {
+    if (this._processing) return;
+    this._processing = true;
+    while (this._queue.length > 0) {
+      const { task, resolve, reject } = this._queue.shift();
+      try {
+        const result = await task();
+        resolve(result);
+      } catch (e) {
+        reject(e);
+      }
+    }
+    this._processing = false;
+  }
+}
+
+export class TikTokScraper {
+  constructor({ poolSize = DEFAULT_POOL_SIZE, wafTtl = DEFAULT_WAF_TTL, warmUrl = DEFAULT_WARM_URL } = {}) {
+    this.poolSize = poolSize;
+    this.wafTtl = wafTtl;
+    this.warmUrl = warmUrl;
+    this.browser = null;
+    this.context = null;
+    this.slots = [];
+    this.slotIdx = 0;
+    this.lastWarmTime = 0;
+    this.warmPromise = null;
+  }
+
+  async init() {
+    const executablePath = detectBrowser();
+    if (!executablePath) {
+      throw new Error('未找到本地浏览器（Chrome/Edge），请先安装浏览器或执行 npx playwright install');
+    }
+    this.browser = await chromium.launch({
+      headless: true,
+      executablePath,
+      args: ['--no-sandbox', '--disable-setuid-sandbox', '--disable-dev-shm-usage'],
+    });
+    this.context = await this.browser.newContext();
+    for (let i = 0; i < this.poolSize; i++) {
+      this.slots.push(new PageSlot(await this.context.newPage()));
+    }
+    await this.warmWaf();
+  }
+
+  async close() {
+    if (this.browser) {
+      await this.browser.close();
+      this.browser = null;
+      this.context = null;
+      this.slots = [];
+    }
+  }
+
+  async restart() {
+    await this.close();
+    await this.init();
+  }
+
+  isAlive() {
+    try {
+      return !!(this.browser && this.browser.isConnected());
+    } catch {
+      return false;
+    }
+  }
+
+  async warmWaf() {
+    if (this.warmPromise) return this.warmPromise;
+    this.warmPromise = (async () => {
+      const page = this.slots[0].page;
+      await page.goto(this.warmUrl, { waitUntil: 'domcontentloaded', timeout: 15000 });
+      await delay(1500);
+      this.lastWarmTime = Date.now();
+      this.warmPromise = null;
+    })();
+    return this.warmPromise;
+  }
+
+  _needWarm() {
+    return Date.now() - this.lastWarmTime > this.wafTtl;
+  }
+
+  _pickSlot() {
+    const slot = this.slots[this.slotIdx % this.poolSize];
+    this.slotIdx++;
+    return slot;
+  }
+
+  async _ensurePage(slot) {
+    try {
+      if (!slot.page.isClosed()) return slot.page;
+    } catch {}
+    slot.page = await this.context.newPage();
+    return slot.page;
+  }
+
+  async _fetchViewSource(url, slot) {
+    const page = await this._ensurePage(slot);
+
+    await page.goto('view-source:' + url, {
+      waitUntil: 'domcontentloaded',
+      timeout: 15000,
+    });
+
+    return await page.evaluate(() => {
+      const rows = document.querySelectorAll('tr');
+      let content = '';
+      rows.forEach(r => {
+        const lc = r.querySelector('.line-content');
+        if (lc) content += lc.textContent + '\n';
+      });
+      return content;
+    });
+  }
+
+  async getUserInfo(uniqueId) {
+    if (this._needWarm()) await this.warmWaf();
+    const slot = this._pickSlot();
+    return slot.lock.run(async () => {
+      const rawHtml = await this._fetchViewSource(
+        `https://www.tiktok.com/@${uniqueId}`,
+        slot
+      );
+      return parseUserInfo(rawHtml);
+    });
+  }
+
+  async getVideoInfo(videoUrl) {
+    if (this._needWarm()) await this.warmWaf();
+    const slot = this._pickSlot();
+    return slot.lock.run(async () => {
+      const rawHtml = await this._fetchViewSource(videoUrl, slot);
+      return parseVideoInfo(rawHtml);
+    });
+  }
+
+  async getUserAndVideo(videoUrl) {
+    if (this._needWarm()) await this.warmWaf();
+    const video = await this.getVideoInfo(videoUrl);
+    if (!video) return null;
+    const user = await this.getUserInfo(video.author.uniqueId);
+    return { user, video };
+  }
+}

package/src/lib/url.js

@@ -1,52 +1,52 @@
-const BASE_URL = 'https://www.tiktok.com';
-
-export function extractUniqueId(url) {
-  const m = url.match(/\/@([^/]+)/);
-  return m ? m[1] : null;
-}
-
-export function extractVideoId(url) {
-  const m = url.match(/\/video\/(\d+)/);
-  return m ? m[1] : null;
-}
-
-export function normalizeUsername(input) {
-  return (input || '').replace(/^@/, '');
-}
-
-export function toProfileUrl(handle) {
-  const clean = normalizeUsername(handle);
-  return `${BASE_URL}/@${clean}`;
-}
-
-export function toVideoUrl(handle, videoId) {
-  const clean = normalizeUsername(handle);
-  return `${BASE_URL}/@${clean}/video/${videoId}`;
-}
-
-export function ensureAbsoluteUrl(href) {
-  if (href.startsWith('http')) return href;
-  return `${BASE_URL}${href}`;
-}
-
-export function isProfileUrl(url) {
-  return /\/@[\w-]+(?:$|[?#])/.test(url);
-}
-
-export function isVideoUrl(url) {
-  return /\/video\/\d+/.test(url);
-}
-
-export function extractDisplayPath(url) {
-  try {
-    const parts = new URL(url).pathname.split('/').filter(Boolean);
-    return parts.slice(-2).join('/');
-  } catch {
-    return url;
-  }
-}
-
-export function extractAuthorFromVideoUrl(url) {
-  const m = url.match(/@([^/]+)\/video/);
-  return m ? '@' + m[1] : null;
-}
+const BASE_URL = 'https://www.tiktok.com';
+
+export function extractUniqueId(url) {
+  const m = url.match(/\/@([\w.-]+)/);
+  return m ? m[1] : null;
+}
+
+export function extractVideoId(url) {
+  const m = url.match(/\/video\/(\d+)/);
+  return m ? m[1] : null;
+}
+
+export function normalizeUsername(input) {
+  return (input || '').replace(/^@/, '');
+}
+
+export function toProfileUrl(handle) {
+  const clean = normalizeUsername(handle);
+  return `${BASE_URL}/@${clean}`;
+}
+
+export function toVideoUrl(handle, videoId) {
+  const clean = normalizeUsername(handle);
+  return `${BASE_URL}/@${clean}/video/${videoId}`;
+}
+
+export function ensureAbsoluteUrl(href) {
+  if (href.startsWith('http')) return href;
+  return `${BASE_URL}${href}`;
+}
+
+export function isProfileUrl(url) {
+  return /\/@[\w.-]+(?:$|[?#])/.test(url);
+}
+
+export function isVideoUrl(url) {
+  return /\/video\/\d+/.test(url);
+}
+
+export function extractDisplayPath(url) {
+  try {
+    const parts = new URL(url).pathname.split('/').filter(Boolean);
+    return parts.slice(-2).join('/');
+  } catch {
+    return url;
+  }
+}
+
+export function extractAuthorFromVideoUrl(url) {
+  const m = url.match(/@([^/]+)\/video/);
+  return m ? '@' + m[1] : null;
+}

package/src/main.js

@@ -1,28 +1,22 @@
 import { parseArgs } from './lib/args.js';
 import { proxy, HELP_TEXT, getConfigText } from './lib/constants.js';
-import { parseFilter } from './lib/filter.js';
-import { handleScrape } from './cli/scrape.js';
-import { handleVideos } from './cli/videos.js';
-import { handleAuto } from './cli/auto.js';
+import { handleInfo } from './cli/info.js';
 import { handleExplore } from './cli/explore.js';
+import { handleAttach } from './cli/attach.js';
 import { handleWatch } from './cli/watch.js';
 import { handleConfig, showConfig, showUsage, version } from './cli/config.js';
-import { runExploreDefault, runScrapeDefault } from './cli/explore-default.js';
 
 async function main() {
   const parsed = parseArgs();
 
   switch (parsed.subcommand) {
-    case 'scrape': return handleScrape(parsed);
-    case 'videos': return handleVideos(parsed);
-    case 'auto':   return handleAuto(parsed);
-    case 'explore':return handleExplore(parsed);
-    case 'watch':  return handleWatch(parsed);
+    case 'explore': return handleExplore(parsed);
+    case 'info':    return handleInfo(parsed);
+    case 'attach':  return handleAttach(parsed);
+    case 'watch':   return handleWatch(parsed);
   }
 
-  const { urls, outputFile, outputFormat, exploreCount, showConfig: showCfg, showHelp, showVersion, customProxy, configAction, configKey, configValue, pipeMode, filterStr } = parsed;
-  const proxyUrl = customProxy || proxy;
-  const filter = parseFilter(filterStr);
+  const { urls, outputFile, outputFormat, exploreCount, showConfig: showCfg, showHelp, showVersion, customProxy, configAction, configKey, configValue } = parsed;
 
   if (showVersion) {
     console.log(version);
@@ -33,11 +27,13 @@ async function main() {
   if (showCfg) return showConfig(urls, outputFile);
   if (urls.length === 0 && exploreCount === 0) return showUsage();
 
+  // 默认行为：URL 走 info，--explore 走 explore
   if (exploreCount > 0) {
-    await runExploreDefault(exploreCount, urls, proxyUrl, outputFile, outputFormat, pipeMode, filter);
-  } else {
-    await runScrapeDefault(urls, proxyUrl, outputFile, outputFormat, filter);
+    return handleExplore({ ...parsed, subcommand: 'explore' });
   }
+
+  // 有 URL 默认走 info
+  return handleInfo(parsed);
 }
 
 main().catch(err => {

package/src/results/user-videos-bar.lar.lar.moeta.json

@@ -0,0 +1,37 @@
+{
+  "user": {
+    "uniqueId": "bar.lar.lar.moeta",
+    "secUid": "MS4wLjABAAAA3cgKTWvKfga0JAWeakAzx3zQ-aFAC8RuQvxD4HQFraKKsc_TbOIyMo3_ofVlXofV",
+    "nickname": "Bar Lar Lar Moetain",
+    "ttSeller": false,
+    "verified": false,
+    "followerCount": 24000,
+    "videoCount": 749,
+    "followingCount": 4293,
+    "heartCount": 254300,
+    "signature": ""
+  },
+  "totalVideos": 5,
+  "videos": [
+    {
+      "id": "7638231799084158228",
+      "url": "https://www.tiktok.com/@bar.lar.lar.moeta/video/7638231799084158228"
+    },
+    {
+      "id": "7638162444698914068",
+      "url": "https://www.tiktok.com/@bar.lar.lar.moeta/video/7638162444698914068"
+    },
+    {
+      "id": "7638116251767819541",
+      "url": "https://www.tiktok.com/@bar.lar.lar.moeta/video/7638116251767819541"
+    },
+    {
+      "id": "7638069637321690388",
+      "url": "https://www.tiktok.com/@bar.lar.lar.moeta/video/7638069637321690388"
+    },
+    {
+      "id": "7637927171025112341",
+      "url": "https://www.tiktok.com/@bar.lar.lar.moeta/video/7637927171025112341"
+    }
+  ]
+}