tt-help-cli-ycl 1.3.12 → 1.3.14

package/src/scraper/explore-core.js

@@ -1,171 +1,162 @@
-import {
-  delay,
-  ensureBrowserReady,
-  setDelayConfig,
-  closeCommentPanel,
-  retryWithBackoff,
-  detectPageError,
-  isLoggedIn,
-  assertPageUrl,
-} from './modules/page-helpers.js';
-import { detectCaptcha } from './modules/captcha-handler.js';
-export { ensureBrowserReady };
-import {
-  getUserInfo,
-  collectVideos,
-} from '../videos/core.js';
-import { scrapeSingleVideo } from './core.js';
-import { extractFollowAndFollowers } from './modules/follow-extractor.js';
-import { extractCommentAuthors } from './modules/comment-extractor.js';
-import { extractGuessVideos } from './modules/guess-extractor.js';
-
-async function processExplore(page, username, options, log) {
-  const {
-    maxComments = 0,
-    maxGuess = 0,
-    enableFollow = true,
-    maxFollowing = 5,
-    maxFollowers = 5,
-    location = 'ES',
-  } = options;
-
-  const result = {
-    userInfo: null,
-    discoveredVideoAuthors: [],
-    discoveredCommentAuthors: [],
-    discoveredGuessAuthors: [],
-    discoveredFollowing: [],
-    discoveredFollowers: [],
-    collectedVideos: 0,
-    processed: false,
-    hasFollowData: false,
-    keepFollow: false,
-    locationCreated: null,
-    noVideo: false,
-    error: null,
-  };
-
-  try {
-    log(`  访问 @${username} 主页...`);
-    const homeUrl = `https://www.tiktok.com/@${username}`;
-    await retryWithBackoff(async () => {
-      await page.goto(homeUrl, { waitUntil: 'domcontentloaded', timeout: 30000 });
-      assertPageUrl(page, `@${username}`);
-    }, { log });
-    await page.waitForSelector('[class*="DivVideoList"]', { timeout: 10000 }).catch(() => {});
-    await delay(1000, 2000);
-
-    log('  获取用户信息...');
-    const info = await getUserInfo(page);
-    if (info) {
-      result.userInfo = info;
-      log(`  用户: ${info.nickname || username} | 粉丝: ${info.followerCount || '-'} | 视频: ${info.videoCount || '-'}`);
-    }
-
-    const captcha = await detectCaptcha(page);
-    if (captcha && captcha.visible) {
-      log(`[验证码] @${username} 页面出现验证码`);
-      result.captchaDetected = true;
-    }
-
-    const videoList = await collectVideos(page, username, 1, log);
-    const videoArray = videoList ? [...videoList.values()] : [];
-    result.collectedVideos = videoArray.length;
-
-    if (videoArray.length <= 0) {
-      result.processed = true;
-      result.noVideo = true;
-      const pageError = await detectPageError(page);
-      if (pageError) {
-        result.restricted = true;
-        log(`  @${username} 页面受限（${pageError}），标记跳过`);
-      } else {
-        log(`  @${username} 没有视频，标记已处理`);
-      }
-      return result;
-    }
-
-    if (enableFollow) {
-      const loggedIn = await isLoggedIn(page);
-      if (!loggedIn) {
-        log('  [跳过] 获取关注/粉丝：未登录，请先登录 TikTok');
-        result.hasFollowData = false;
-        result.discoveredFollowing = [];
-        result.discoveredFollowers = [];
-      } else {
-        try {
-          log('  获取关注/粉丝...');
-          const { following, followers } = await extractFollowAndFollowers(
-            page, { maxFollowing, maxFollowers, log }
-          );
-          result.discoveredFollowing = following || [];
-          result.discoveredFollowers = followers || [];
-          result.hasFollowData = true;
-          log(`  关注: ${result.discoveredFollowing.length}, 粉丝: ${result.discoveredFollowers.length}`);
-        } catch (e) {
-          log(`  关注/粉丝提取失败: ${e.message}`);
-          result.hasFollowData = false;
-          result.discoveredFollowing = [];
-          result.discoveredFollowers = [];
-        }
-      }
-    }
-
-    const firstVideo = videoArray[0];
-    const videoUrl = firstVideo.href.startsWith('http')
-      ? firstVideo.href
-      : `https://www.tiktok.com${firstVideo.href}`;
-
-    log(`  进入第一个视频: ${videoUrl}`);
-    await retryWithBackoff(() => page.goto(videoUrl, { waitUntil: 'domcontentloaded', timeout: 30000 }), { log });
-    assertPageUrl(page, videoUrl.split('/video/')[0]);
-    await delay(1500, 2500);
-
-    const videoData = await scrapeSingleVideo(page, 0, 0, log, 'NEVER_MATCH');
-    result.locationCreated = videoData.locationCreated || null;
-    log(`  视频作者: ${videoData.videoAuthor} | 国家: ${result.locationCreated || '未知'}`);
-
-    const isTargetLocation = result.locationCreated === location;
-
-    if (isTargetLocation) {
-      result.keepFollow = true;
-      log(`  国家匹配 (${location})，获取评论和猜你喜欢...`);
-
-      if (maxComments > 0) {
-        const commentResult = await extractCommentAuthors(page, maxComments);
-        result.discoveredCommentAuthors = commentResult || [];
-        await closeCommentPanel(page);
-        await delay(500, 1000);
-        log(`  评论用户: ${result.discoveredCommentAuthors.length}`);
-      }
-
-      if (maxGuess > 0) {
-        const guessResult = await extractGuessVideos(page, maxGuess);
-        result.discoveredGuessAuthors = (guessResult || []).map(v => v.author).filter(Boolean);
-        await closeCommentPanel(page);
-        await delay(500, 1000);
-        log(`  猜你喜欢作者: ${result.discoveredGuessAuthors.length}`);
-      }
-
-      result.discoveredVideoAuthors = [{
-        uniqueId: videoData.uniqueId,
-        nickname: videoData.nickname,
-        locationCreated: videoData.locationCreated,
-      }];
-    } else {
-      result.keepFollow = false;
-      log(`  国家不匹配 (${result.locationCreated} !== ${location})，跳过评论/猜你喜欢，丢弃关注/粉丝`);
-      result.discoveredFollowing = [];
-      result.discoveredFollowers = [];
-    }
-
-    result.processed = true;
-  } catch (e) {
-    result.error = e.message;
-    log(`  [错误] ${e.message}`);
-  }
-
-  return result;
-}
-
-export { processExplore };
+import {
+  delay,
+  ensureBrowserReady,
+  retryWithBackoff,
+  detectPageError,
+  isLoggedIn,
+  assertPageUrl,
+} from './modules/page-helpers.js';
+import { detectCaptcha } from './modules/captcha-handler.js';
+export { ensureBrowserReady };
+import { getUserInfo, collectVideos } from '../videos/core.js';
+import { extractFollowAndFollowers } from './modules/follow-extractor.js';
+import { extractVideoLocation } from '../lib/scrape.js';
+import { maxFollowing as globalMaxFollowing, maxFollowers as globalMaxFollowers, maxVideos as globalMaxVideos } from '../lib/constants.js';
+
+async function processExplore(page, username, options, log) {
+  const {
+    maxVideos = 1,
+    enableFollow = true,
+    maxFollowing = 5,
+    maxFollowers = 5,
+    location = 'PL,NL,BE,DE,FR,IT,ES,IE',
+  } = options;
+
+  const result = {
+    userInfo: null,
+    discoveredVideoAuthors: [],
+    discoveredCommentAuthors: [],
+    discoveredGuessAuthors: [],
+    discoveredFollowing: [],
+    discoveredFollowers: [],
+    collectedVideos: 0,
+    processed: false,
+    hasFollowData: false,
+    keepFollow: false,
+    locationCreated: null,
+    noVideo: false,
+    error: null,
+  };
+
+  try {
+    log(`  访问 @${username} 主页...`);
+    const homeUrl = `https://www.tiktok.com/@${username}`;
+    await retryWithBackoff(async () => {
+      await page.goto(homeUrl, { waitUntil: 'domcontentloaded', timeout: 30000 });
+      assertPageUrl(page, `@${username}`);
+    }, { log });
+    await page.waitForSelector('[class*="DivVideoList"]', { timeout: 10000 }).catch(() => {});
+    await delay(1000, 2000);
+
+    log('  获取用户信息...');
+    const info = await getUserInfo(page);
+    if (info) {
+      result.userInfo = info;
+      log(`  用户: ${info.nickname || username} | 粉丝: ${info.followerCount || '-'} | 视频: ${info.videoCount || '-'}`);
+    }
+
+    const captcha = await detectCaptcha(page);
+    if (captcha && captcha.visible) {
+      log(`[验证码] @${username} 页面出现验证码`);
+      result.captchaDetected = true;
+      result.captchaStage = result.captchaStage || 'video-page';
+      result.captchaMessage = result.captchaMessage || '视频页出现验证码';
+    }
+
+    const isSeller = result.userInfo?.ttSeller === true;
+    const effectiveMaxVideos = isSeller ? globalMaxVideos : maxVideos;
+    if (isSeller) log(`  商家用户，视频采集数: ${effectiveMaxVideos}`);
+    const videoList = await collectVideos(page, username, effectiveMaxVideos, log);
+    const videoArray = videoList ? [...videoList.values()] : [];
+    result.collectedVideos = videoArray.length;
+
+    if (videoArray.length <= 0) {
+      result.processed = true;
+      result.noVideo = true;
+      const pageError = await detectPageError(page);
+      if (pageError) {
+        result.restricted = true;
+        log(`  @${username} 页面受限（${pageError}），标记跳过`);
+      } else {
+        log(`  @${username} 没有视频，标记已处理`);
+      }
+      return result;
+    }
+
+    // 从第一个视频获取 locationCreated
+    let locationCreated = null;
+    if (videoArray.length > 0) {
+      const firstVideo = videoArray[0];
+      const firstVideoUrl = firstVideo.href.startsWith('http')
+        ? firstVideo.href
+        : `https://www.tiktok.com${firstVideo.href}`;
+
+      try {
+        locationCreated = await extractVideoLocation(firstVideoUrl);
+      } catch (e) {
+        log(`  获取视频国家失败: ${e.message}`);
+      }
+    }
+
+    result.locationCreated = locationCreated || null;
+    log(`  国家: ${result.locationCreated || '未知'}`);
+
+    // 国家筛选
+    const locationList = (location || 'ES').split(',').map(s => s.trim().toUpperCase());
+    const isTargetLocation = locationList.includes(result.locationCreated?.toUpperCase?.() || result.locationCreated);
+
+    if (isTargetLocation) {
+      result.keepFollow = true;
+      log(`  国家匹配，获取关注/粉丝...`);
+
+      // 提取关注/粉丝
+      if (enableFollow) {
+        const loggedIn = await isLoggedIn(page);
+        if (!loggedIn) {
+          log('  [跳过] 获取关注/粉丝：未登录，请先登录 TikTok');
+          result.hasFollowData = false;
+          result.discoveredFollowing = [];
+          result.discoveredFollowers = [];
+        } else {
+          try {
+            const effectiveMaxFollowing = isSeller ? globalMaxFollowing : maxFollowing;
+            const effectiveMaxFollowers = isSeller ? globalMaxFollowers : maxFollowers;
+            if (isSeller) log(`  商家用户，关注采集: ${effectiveMaxFollowing}, 粉丝采集: ${effectiveMaxFollowers}`);
+            const { following, followers } = await extractFollowAndFollowers(
+              page, { maxFollowing: effectiveMaxFollowing, maxFollowers: effectiveMaxFollowers, log }
+            );
+            result.discoveredFollowing = following || [];
+            result.discoveredFollowers = followers || [];
+            result.hasFollowData = true;
+            log(`  关注: ${result.discoveredFollowing.length}, 粉丝: ${result.discoveredFollowers.length}`);
+          } catch (e) {
+            log(`  关注/粉丝提取失败: ${e.message}`);
+            result.hasFollowData = false;
+            result.discoveredFollowing = [];
+            result.discoveredFollowers = [];
+          }
+        }
+      }
+
+      // 携带视频列表供登记
+      result.videoList = videoArray;
+    } else {
+      // 国家不匹配
+      result.keepFollow = false;
+      result.discoveredFollowing = [];
+      result.discoveredFollowers = [];
+      result.hasFollowData = false;
+      log(`  国家不匹配，跳过`);
+    }
+
+    result.processed = true;
+  } catch (e) {
+    result.error = e.message;
+    result.errorStack = e.stack || '';
+    log(`  [错误] ${e.message}`);
+  }
+
+  return result;
+}
+
+export { processExplore };

package/src/scraper/modules/captcha-handler.js

@@ -1,114 +1,114 @@
-export async function detectCaptcha(page) {
-  return page.evaluate(() => {
-    const container = document.querySelector('.captcha-verify-container');
-    if (!container) return null;
-
-    const r = container.getBoundingClientRect();
-    return {
-      exists: true,
-      visible: container.offsetParent !== null,
-      rect: {
-        x: Math.round(r.x),
-        y: Math.round(r.y),
-        w: Math.round(r.width),
-        h: Math.round(r.height),
-      },
-    };
-  });
-}
-
-export async function closeCaptcha(page) {
-  return page.evaluate(() => {
-    const closeBtn = document.getElementById('captcha_close_button');
-    if (!closeBtn) return { success: false, reason: 'close button not found' };
-
-    closeBtn.dispatchEvent(new MouseEvent('mousedown', { bubbles: true }));
-    closeBtn.dispatchEvent(new MouseEvent('mouseup', { bubbles: true }));
-    closeBtn.dispatchEvent(new MouseEvent('click', { bubbles: true }));
-
-    return { success: true };
-  });
-}
-
-export async function handleCaptcha(page, options = {}) {
-  const { waitMs = 2000 } = options;
-
-  const captcha = await detectCaptcha(page);
-  if (!captcha) return { detected: false, closed: false };
-
-  await new Promise(r => setTimeout(r, waitMs));
-
-  const result = await closeCaptcha(page);
-  if (!result.success) return { detected: true, closed: false, reason: result.reason };
-
-  await new Promise(r => setTimeout(r, 1000));
-
-  const stillThere = await detectCaptcha(page);
-  return { detected: true, closed: !stillThere };
-}
-
-export async function getIncognitoPage(browser, url, options = {}) {
-  const { waitMs = 3000 } = options;
-  const context = await browser.newContext();
-  const page = await context.newPage();
-  await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
-  await new Promise(r => setTimeout(r, waitMs));
-  return { page, context };
-}
-
-export async function waitAndGetCaptcha(page, options = {}) {
-  const { waitMs = 180000, pollInterval = 5000, log } = options;
-
-  const captcha = await detectCaptcha(page);
-  if (!captcha) return { detected: false, resolved: false, waited: 0 };
-
-  if (log) log('  检测到验证码，等待用户手动输入...');
-
-  const startTime = Date.now();
-  const deadline = startTime + waitMs;
-
-  while (Date.now() < deadline) {
-    await new Promise(r => setTimeout(r, pollInterval));
-    const remaining = await detectCaptcha(page);
-    if (!remaining) {
-      const waited = Math.round((Date.now() - startTime) / 1000);
-      if (log) log(`  验证码已解决（等待 ${waited}s）`);
-      return { detected: true, resolved: true, waited };
-    }
-  }
-
-  const waited = Math.round(waitMs / 1000);
-  if (log) log(`  验证码等待超时（${waited}s），继续执行`);
-
-  // 超时后尝试关闭验证码弹窗
-  await closeCaptcha(page);
-  await new Promise(r => setTimeout(r, 1000));
-
-  return { detected: true, resolved: false, waited };
-}
-
-export async function safeClickComment(page, options = {}) {
-  const { waitMs = 3000 } = options;
-
-  // 点击评论
-  await page.evaluate(() => {
-    const all = document.querySelectorAll('button');
-    for (const el of all) {
-      if (/^评论$/.test(el.textContent?.trim()) && el.offsetParent !== null && el.getBoundingClientRect().width > 0) {
-        el.click();
-        break;
-      }
-    }
-  });
-
-  await new Promise(r => setTimeout(r, waitMs));
-
-  // 检测并关闭验证码
-  const captcha = await detectCaptcha(page);
-  if (captcha) {
-    const result = await handleCaptcha(page);
-    return { clicked: true, captchaDetected: true, captchaClosed: result.closed };
-  }
-
-  return { clicked: true, captchaDetected: false, captchaClosed: false };
-}
+export async function detectCaptcha(page) {
+  return page.evaluate(() => {
+    const container = document.querySelector('.captcha-verify-container');
+    if (!container) return null;
+
+    const r = container.getBoundingClientRect();
+    return {
+      exists: true,
+      visible: container.offsetParent !== null,
+      rect: {
+        x: Math.round(r.x),
+        y: Math.round(r.y),
+        w: Math.round(r.width),
+        h: Math.round(r.height),
+      },
+    };
+  });
+}
+
+export async function closeCaptcha(page) {
+  return page.evaluate(() => {
+    const closeBtn = document.getElementById('captcha_close_button');
+    if (!closeBtn) return { success: false, reason: 'close button not found' };
+
+    closeBtn.dispatchEvent(new MouseEvent('mousedown', { bubbles: true }));
+    closeBtn.dispatchEvent(new MouseEvent('mouseup', { bubbles: true }));
+    closeBtn.dispatchEvent(new MouseEvent('click', { bubbles: true }));
+
+    return { success: true };
+  });
+}
+
+export async function handleCaptcha(page, options = {}) {
+  const { waitMs = 2000 } = options;
+
+  const captcha = await detectCaptcha(page);
+  if (!captcha) return { detected: false, closed: false };
+
+  await new Promise(r => setTimeout(r, waitMs));
+
+  const result = await closeCaptcha(page);
+  if (!result.success) return { detected: true, closed: false, reason: result.reason };
+
+  await new Promise(r => setTimeout(r, 1000));
+
+  const stillThere = await detectCaptcha(page);
+  return { detected: true, closed: !stillThere };
+}
+
+export async function getIncognitoPage(browser, url, options = {}) {
+  const { waitMs = 3000 } = options;
+  const context = await browser.newContext();
+  const page = await context.newPage();
+  await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 30000 });
+  await new Promise(r => setTimeout(r, waitMs));
+  return { page, context };
+}
+
+export async function waitAndGetCaptcha(page, options = {}) {
+  const { waitMs = 180000, pollInterval = 5000, log } = options;
+
+  const captcha = await detectCaptcha(page);
+  if (!captcha) return { detected: false, resolved: false, waited: 0 };
+
+  if (log) log('  检测到验证码，等待用户手动输入...');
+
+  const startTime = Date.now();
+  const deadline = startTime + waitMs;
+
+  while (Date.now() < deadline) {
+    await new Promise(r => setTimeout(r, pollInterval));
+    const remaining = await detectCaptcha(page);
+    if (!remaining) {
+      const waited = Math.round((Date.now() - startTime) / 1000);
+      if (log) log(`  验证码已解决（等待 ${waited}s）`);
+      return { detected: true, resolved: true, waited };
+    }
+  }
+
+  const waited = Math.round(waitMs / 1000);
+  if (log) log(`  验证码等待超时（${waited}s），继续执行`);
+
+  // 超时后尝试关闭验证码弹窗
+  await closeCaptcha(page);
+  await new Promise(r => setTimeout(r, 1000));
+
+  return { detected: true, resolved: false, waited };
+}
+
+export async function safeClickComment(page, options = {}) {
+  const { waitMs = 3000 } = options;
+
+  // 点击评论
+  await page.evaluate(() => {
+    const all = document.querySelectorAll('button');
+    for (const el of all) {
+      if (/^评论$/.test(el.textContent?.trim()) && el.offsetParent !== null && el.getBoundingClientRect().width > 0) {
+        el.click();
+        break;
+      }
+    }
+  });
+
+  await new Promise(r => setTimeout(r, waitMs));
+
+  // 检测并关闭验证码
+  const captcha = await detectCaptcha(page);
+  if (captcha) {
+    const result = await handleCaptcha(page);
+    return { clicked: true, captchaDetected: true, captchaClosed: result.closed };
+  }
+
+  return { clicked: true, captchaDetected: false, captchaClosed: false };
+}