tt-help-cli-ycl 1.3.34 → 1.3.35

package/src/scraper/explore-core.js

@@ -1,167 +1,177 @@
-import { ensureBrowserReady, delay } 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,
-} from "../lib/constants.js";
-
-async function processExplore(page, username, options, log) {
-  const {
-    maxVideos = 16,
-    enableFollow = true,
-    loggedIn = false, // 由外部传入登录状态，避免每次调用 isLoggedIn(page)
-    maxFollowing = 50,
-    maxFollowers = 50,
-    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 videoList = await collectVideos(page, username, maxVideos, log);
-
-    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 videoArray = videoList ? [...videoList.values()] : [];
-    result.collectedVideos = videoArray.length;
-
-    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;
-      result.noVideo = true;
-      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}`;
-
-      locationCreated = await extractVideoLocation(firstVideoUrl);
-    }
-
-    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) {
-        await delay(100, 1000);
-        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 isSeller = result.userInfo?.ttSeller === true;
-            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] ${e.message}`);
-
-    // 被封会抛出 "被封: username" 异常
-    if (e.message.startsWith("被封:")) {
-      result.processed = false;
-      result.noVideo = false;
-      log(`  @${username} 认定为被封`);
-    }
-  }
-
-  return result;
-}
-
-export { processExplore };
+import { ensureBrowserReady, delay } 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,
+} from "../lib/constants.js";
+
+async function processExplore(page, username, options, log) {
+  const {
+    maxVideos = 16,
+    enableFollow = true,
+    loggedIn = false, // 由外部传入登录状态，避免每次调用 isLoggedIn(page)
+    maxFollowing = 50,
+    maxFollowers = 50,
+    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 videoList = await collectVideos(page, username, maxVideos, log);
+
+    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 videoArray = videoList ? [...videoList.values()] : [];
+    result.collectedVideos = videoArray.length;
+
+    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;
+      result.noVideo = true;
+      log(`  @${username} 没有视频，标记已处理`);
+      return result;
+    }
+
+    // 从最多 5 个视频并发获取 locationCreated，取众数
+    const SAMPLE_SIZE = 5;
+    let locationCreated = null;
+    const sampleVideos = videoArray.slice(0, SAMPLE_SIZE);
+    if (sampleVideos.length > 0) {
+      const sampleUrls = sampleVideos.map(v =>
+        v.href.startsWith("http") ? v.href : `https://www.tiktok.com${v.href}`
+      );
+      const locations = await Promise.all(sampleUrls.map(url => extractVideoLocation(url)));
+      log(`  国家采样(${locations.length}个): [${locations.filter(Boolean).join(", ") || "无数据"}]`);
+      const freq = {};
+      for (const loc of locations) {
+        if (loc) {
+          const key = loc.toUpperCase();
+          freq[key] = (freq[key] || 0) + 1;
+        }
+      }
+      const entries = Object.entries(freq).sort((a, b) => b[1] - a[1]);
+      locationCreated = entries.length > 0 ? entries[0][0] : null;
+    }
+
+    result.locationCreated = locationCreated || null;
+    log(`  国家: ${result.locationCreated || "未知"} (众数)`);
+
+    // 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) {
+        await delay(100, 1000);
+        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 isSeller = result.userInfo?.ttSeller === true;
+            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] ${e.message}`);
+
+    // 被封会抛出 "被封: username" 异常
+    if (e.message.startsWith("被封:")) {
+      result.processed = false;
+      result.noVideo = false;
+      log(`  @${username} 认定为被封`);
+    }
+  }
+
+  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 };
+}