tt-help-cli-ycl 1.3.22 → 1.3.24

package/src/scraper/modules/comment-extractor.js

@@ -1,74 +1,74 @@
-import { delay, getDelayConfig, closeCommentPanel } from "./page-helpers.js";
-import { scrollAndCollect } from "./scroll-collector.js";
-import { waitAndGetCaptcha } from "./captcha-handler.js";
-
-async function openCommentPanel(page) {
-  const tabs = page.locator('[class*="tabbar-item"]');
-  const commentTab = tabs.filter({ hasText: "评论" }).first();
-  await commentTab.click();
-
-  // 等待短暂时间让页面渲染
-  await new Promise(r => setTimeout(r, 2000));
-
-  // 检测验证码
-  const captchaResult = await waitAndGetCaptcha(page, {
-    waitMs: 180000,
-    pollInterval: 5000,
-    log: console.error,
-  });
-
-  await page
-    .waitForSelector('[class*="CommentListContainer"]', { timeout: 5000 })
-    .catch(() => {});
-  await page
-    .waitForFunction(
-      () => {
-        const list = document.querySelector('[class*="CommentListContainer"]');
-        return list && list.children.length > 0;
-      },
-      { timeout: 10000 },
-    )
-    .catch(() => {});
-
-  return { captchaDetected: !!captchaResult.detected };
-}
-
-async function extractCommentAuthors(page, maxComments = 10) {
-  const panelResult = await openCommentPanel(page);
-
-  const config = getDelayConfig();
-  const allAuthors = await scrollAndCollect(page, {
-    container: '[class*="CommentMain"]',
-    findScrollable: true,
-    collectFn: (container) => {
-      const list = document.querySelector('[class*="CommentListContainer"]');
-      if (!list) return { items: [] };
-      const authors = [];
-      Array.from(list.children).forEach((wrapper) => {
-        const link = wrapper.querySelector(
-          '[class*="UsernameContentWrapper"] a',
-        );
-        if (link) {
-          const href = link.href || link.getAttribute("href");
-          const m = href && href.match(/@([^/]+)/);
-          if (m) authors.push("@" + m[1]);
-        }
-      });
-      return { items: authors };
-    },
-    uniqueKey: (a) => a,
-    maxItems: maxComments,
-    delayRange: [Math.round(config.commentMax * 0.3), config.commentMax],
-    staleThreshold: 2,
-  });
-
-  await closeCommentPanel(page);
-  await delay(Math.round(config.commentMax * 0.3), config.commentMax);
-
-  return {
-    authors: allAuthors.slice(0, maxComments),
-    captchaDetected: panelResult.captchaDetected,
-  };
-}
-
-export { extractCommentAuthors };
+import { delay, getDelayConfig, closeCommentPanel } from "./page-helpers.js";
+import { scrollAndCollect } from "./scroll-collector.js";
+import { waitAndGetCaptcha } from "./captcha-handler.js";
+
+async function openCommentPanel(page) {
+  const tabs = page.locator('[class*="tabbar-item"]');
+  const commentTab = tabs.filter({ hasText: "评论" }).first();
+  await commentTab.click();
+
+  // 等待短暂时间让页面渲染
+  await new Promise(r => setTimeout(r, 2000));
+
+  // 检测验证码
+  const captchaResult = await waitAndGetCaptcha(page, {
+    waitMs: 180000,
+    pollInterval: 5000,
+    log: console.error,
+  });
+
+  await page
+    .waitForSelector('[class*="CommentListContainer"]', { timeout: 5000 })
+    .catch(() => {});
+  await page
+    .waitForFunction(
+      () => {
+        const list = document.querySelector('[class*="CommentListContainer"]');
+        return list && list.children.length > 0;
+      },
+      { timeout: 10000 },
+    )
+    .catch(() => {});
+
+  return { captchaDetected: !!captchaResult.detected };
+}
+
+async function extractCommentAuthors(page, maxComments = 10) {
+  const panelResult = await openCommentPanel(page);
+
+  const config = getDelayConfig();
+  const allAuthors = await scrollAndCollect(page, {
+    container: '[class*="CommentMain"]',
+    findScrollable: true,
+    collectFn: (container) => {
+      const list = document.querySelector('[class*="CommentListContainer"]');
+      if (!list) return { items: [] };
+      const authors = [];
+      Array.from(list.children).forEach((wrapper) => {
+        const link = wrapper.querySelector(
+          '[class*="UsernameContentWrapper"] a',
+        );
+        if (link) {
+          const href = link.href || link.getAttribute("href");
+          const m = href && href.match(/@([^/]+)/);
+          if (m) authors.push("@" + m[1]);
+        }
+      });
+      return { items: authors };
+    },
+    uniqueKey: (a) => a,
+    maxItems: maxComments,
+    delayRange: [Math.round(config.commentMax * 0.3), config.commentMax],
+    staleThreshold: 2,
+  });
+
+  await closeCommentPanel(page);
+  await delay(Math.round(config.commentMax * 0.3), config.commentMax);
+
+  return {
+    authors: allAuthors.slice(0, maxComments),
+    captchaDetected: panelResult.captchaDetected,
+  };
+}
+
+export { extractCommentAuthors };

package/src/scraper/modules/follow-extractor.js

@@ -1,118 +1,118 @@
-import { delay, getDelayConfig } from "./page-helpers.js";
-import { scrollAndCollect } from "./scroll-collector.js";
-
-const FILTER_WORDS = ["主页", "已关注", "粉丝", "推荐"];
-
-async function waitForListContent(page, minChildren = 1, timeout = 15000) {
-  await page
-    .waitForFunction(
-      (min) => {
-        const container = document.querySelector(
-          "[class*=DivUserListContainer]",
-        );
-        return container && container.children.length >= min;
-      },
-      minChildren,
-      { timeout },
-    )
-    .catch(() => {});
-}
-
-async function openFollowModal(page) {
-  const el = await page.$("[data-e2e=following]");
-  if (!el) {
-    throw new Error(
-      "未找到 [data-e2e=following] 元素，请确认当前页面为用户主页",
-    );
-  }
-  await el.evaluate((el) => el.parentElement.click());
-  await page.waitForSelector("[class*=DivUserListContainer]", { timeout: 15000 });
-  await waitForListContent(page, 1, 3000);
-}
-
-async function switchToFollowersTab(page) {
-  await page.evaluate(() => {
-    const tabs = document.querySelectorAll("[class*=DivTabItem]");
-    for (const tab of tabs) {
-      if (tab.textContent?.includes("粉丝")) {
-        tab.click();
-        return;
-      }
-    }
-    throw new Error("未找到粉丝 Tab");
-  });
-  await page.waitForSelector("[class*=DivUserListContainer]", { timeout: 15000 });
-  await waitForListContent(page, 1, 3000);
-}
-
-async function closeFollowModal(page) {
-  await page.evaluate(() => {
-    const closeBtn = document.querySelector("[data-e2e=follow-popup-close]");
-    if (closeBtn) closeBtn.click();
-  });
-  await page.waitForTimeout(500);
-}
-
-function createUserCollectFn() {
-  return (container) => {
-    const FILTER_WORDS = ["主页", "已关注", "粉丝", "推荐"];
-    const modal = document.querySelector("[class*=eyhy6180]");
-    const root = modal || document;
-    const users = [];
-    const seen = new Set();
-    const links = root.querySelectorAll('a[href*="/@"]');
-    for (const link of links) {
-      const match = link.href.match(/@([^/?]+)/);
-      if (!match) continue;
-      const handle = "@" + decodeURIComponent(match[1]);
-      const text = (link.textContent || "").trim();
-      if (text.length <= 2) continue;
-      if (FILTER_WORDS.includes(text)) continue;
-      if (seen.has(handle)) continue;
-      seen.add(handle);
-      users.push({ handle, displayName: text });
-    }
-    return { items: users };
-  };
-}
-
-async function extractUsersFromModal(page, maxUsers) {
-  const config = getDelayConfig();
-  const minDelay = Math.max(300, Math.round(config.commentMax * 0.3));
-  const maxDelay = Math.max(800, config.commentMax);
-
-  const allUsers = await scrollAndCollect(page, {
-    container: "[class*=DivUserListContainer]",
-    findScrollable: false,
-    collectFn: createUserCollectFn(),
-    uniqueKey: (u) => u.handle,
-    maxItems: maxUsers,
-    delayRange: [minDelay, maxDelay],
-    staleThreshold: 2,
-  });
-
-  return allUsers.slice(0, maxUsers);
-}
-
-async function extractFollowAndFollowers(page, options = {}) {
-  const { maxFollowing = 999, maxFollowers = 999, log = () => {} } = options;
-
-  await openFollowModal(page);
-
-  const following = await extractUsersFromModal(page, maxFollowing);
-  log(`  已关注: ${following.length}`);
-
-  await switchToFollowersTab(page);
-
-  const followers = await extractUsersFromModal(page, maxFollowers);
-  log(`  粉丝: ${followers.length}`);
-
-  await closeFollowModal(page);
-
-  return {
-    following: following.map((u) => [u.handle, u.displayName]),
-    followers: followers.map((u) => [u.handle, u.displayName]),
-  };
-}
-
-export { extractFollowAndFollowers };
+import { delay, getDelayConfig } from "./page-helpers.js";
+import { scrollAndCollect } from "./scroll-collector.js";
+
+const FILTER_WORDS = ["主页", "已关注", "粉丝", "推荐"];
+
+async function waitForListContent(page, minChildren = 1, timeout = 15000) {
+  await page
+    .waitForFunction(
+      (min) => {
+        const container = document.querySelector(
+          "[class*=DivUserListContainer]",
+        );
+        return container && container.children.length >= min;
+      },
+      minChildren,
+      { timeout },
+    )
+    .catch(() => {});
+}
+
+async function openFollowModal(page) {
+  const el = await page.$("[data-e2e=following]");
+  if (!el) {
+    throw new Error(
+      "未找到 [data-e2e=following] 元素，请确认当前页面为用户主页",
+    );
+  }
+  await el.evaluate((el) => el.parentElement.click());
+  await page.waitForSelector("[class*=DivUserListContainer]", { timeout: 30000 });
+  await waitForListContent(page, 1, 5000);
+}
+
+async function switchToFollowersTab(page) {
+  await page.evaluate(() => {
+    const tabs = document.querySelectorAll("[class*=DivTabItem]");
+    for (const tab of tabs) {
+      if (tab.textContent?.includes("粉丝")) {
+        tab.click();
+        return;
+      }
+    }
+    throw new Error("未找到粉丝 Tab");
+  });
+  await page.waitForSelector("[class*=DivUserListContainer]", { timeout: 30000 });
+  await waitForListContent(page, 1, 5000);
+}
+
+async function closeFollowModal(page) {
+  await page.evaluate(() => {
+    const closeBtn = document.querySelector("[data-e2e=follow-popup-close]");
+    if (closeBtn) closeBtn.click();
+  });
+  await page.waitForTimeout(500);
+}
+
+function createUserCollectFn() {
+  return (container) => {
+    const FILTER_WORDS = ["主页", "已关注", "粉丝", "推荐"];
+    const modal = document.querySelector("[class*=eyhy6180]");
+    const root = modal || document;
+    const users = [];
+    const seen = new Set();
+    const links = root.querySelectorAll('a[href*="/@"]');
+    for (const link of links) {
+      const match = link.href.match(/@([^/?]+)/);
+      if (!match) continue;
+      const handle = "@" + decodeURIComponent(match[1]);
+      const text = (link.textContent || "").trim();
+      if (text.length <= 2) continue;
+      if (FILTER_WORDS.includes(text)) continue;
+      if (seen.has(handle)) continue;
+      seen.add(handle);
+      users.push({ handle, displayName: text });
+    }
+    return { items: users };
+  };
+}
+
+async function extractUsersFromModal(page, maxUsers) {
+  const config = getDelayConfig();
+  const minDelay = Math.max(300, Math.round(config.commentMax * 0.3));
+  const maxDelay = Math.max(800, config.commentMax);
+
+  const allUsers = await scrollAndCollect(page, {
+    container: "[class*=DivUserListContainer]",
+    findScrollable: false,
+    collectFn: createUserCollectFn(),
+    uniqueKey: (u) => u.handle,
+    maxItems: maxUsers,
+    delayRange: [minDelay, maxDelay],
+    staleThreshold: 2,
+  });
+
+  return allUsers.slice(0, maxUsers);
+}
+
+async function extractFollowAndFollowers(page, options = {}) {
+  const { maxFollowing = 999, maxFollowers = 999, log = () => {} } = options;
+
+  await openFollowModal(page);
+
+  const following = await extractUsersFromModal(page, maxFollowing);
+  log(`  已关注: ${following.length}`);
+
+  await switchToFollowersTab(page);
+
+  const followers = await extractUsersFromModal(page, maxFollowers);
+  log(`  粉丝: ${followers.length}`);
+
+  await closeFollowModal(page);
+
+  return {
+    following: following.map((u) => [u.handle, u.displayName]),
+    followers: followers.map((u) => [u.handle, u.displayName]),
+  };
+}
+
+export { extractFollowAndFollowers };

package/src/scraper/modules/guess-extractor.js

@@ -1,51 +1,51 @@
-import { delay, getDelayConfig, closeCommentPanel } from './page-helpers.js';
-import { scrollAndCollect } from './scroll-collector.js';
-
-async function openGuessTab(page) {
-  const tabs = page.locator('[class*="tabbar-item"]');
-  const guessTab = tabs.filter({ hasText: /猜你喜欢/i }).first();
-  await guessTab.click();
-  const config = getDelayConfig();
-  await delay(Math.round(config.commentMax * 0.5), config.commentMax);
-  await page.waitForSelector('[class*="Related"]', { timeout: 5000 }).catch(() => {});
-}
-
-async function extractGuessVideos(page, maxVideos = 10) {
-  await openGuessTab(page);
-
-  const config = getDelayConfig();
-  const allVideos = await scrollAndCollect(page, {
-    container: '[class*="Related"]',
-    findScrollable: true,
-    collectFn: (container) => {
-      const items = [];
-      Array.from(container.querySelectorAll('[class*="DivItemContainer"]')).forEach(item => {
-        const link = item.querySelector('a[href*="/video/"]');
-        if (link) {
-          const href = link.href || link.getAttribute('href');
-          const m = href && href.match(/@([^/]+)\/video\/(\d+)/);
-          if (m) {
-            items.push({
-              author: '@' + m[1],
-              videoId: m[2],
-              url: href,
-              title: '',
-            });
-          }
-        }
-      });
-      return { items };
-    },
-    uniqueKey: (v) => v.videoId,
-    maxItems: maxVideos,
-    delayRange: [Math.round(config.commentMax * 0.3), config.commentMax],
-    staleThreshold: 3,
-  });
-
-  await closeCommentPanel(page);
-  await delay(Math.round(config.commentMax * 0.3), config.commentMax);
-
-  return allVideos.slice(0, maxVideos);
-}
-
-export { extractGuessVideos };
+import { delay, getDelayConfig, closeCommentPanel } from './page-helpers.js';
+import { scrollAndCollect } from './scroll-collector.js';
+
+async function openGuessTab(page) {
+  const tabs = page.locator('[class*="tabbar-item"]');
+  const guessTab = tabs.filter({ hasText: /猜你喜欢/i }).first();
+  await guessTab.click();
+  const config = getDelayConfig();
+  await delay(Math.round(config.commentMax * 0.5), config.commentMax);
+  await page.waitForSelector('[class*="Related"]', { timeout: 5000 }).catch(() => {});
+}
+
+async function extractGuessVideos(page, maxVideos = 10) {
+  await openGuessTab(page);
+
+  const config = getDelayConfig();
+  const allVideos = await scrollAndCollect(page, {
+    container: '[class*="Related"]',
+    findScrollable: true,
+    collectFn: (container) => {
+      const items = [];
+      Array.from(container.querySelectorAll('[class*="DivItemContainer"]')).forEach(item => {
+        const link = item.querySelector('a[href*="/video/"]');
+        if (link) {
+          const href = link.href || link.getAttribute('href');
+          const m = href && href.match(/@([^/]+)\/video\/(\d+)/);
+          if (m) {
+            items.push({
+              author: '@' + m[1],
+              videoId: m[2],
+              url: href,
+              title: '',
+            });
+          }
+        }
+      });
+      return { items };
+    },
+    uniqueKey: (v) => v.videoId,
+    maxItems: maxVideos,
+    delayRange: [Math.round(config.commentMax * 0.3), config.commentMax],
+    staleThreshold: 3,
+  });
+
+  await closeCommentPanel(page);
+  await delay(Math.round(config.commentMax * 0.3), config.commentMax);
+
+  return allVideos.slice(0, maxVideos);
+}
+
+export { extractGuessVideos };

package/src/scraper/modules/page-helpers.js

@@ -1,48 +1,48 @@
-import {
-  delay,
-  getDelayConfig,
-  setDelayConfig,
-  listDelayPresets,
-  DELAY_PRESETS,
-} from '../../lib/delay.js';
-import { ensureBrowserReady } from '../../lib/browser/cdp.js';
-import {
-  ensureTikTokPage,
-  closeCommentPanel,
-  findTikTokPage,
-  getOrCreatePage,
-  isLoggedIn,
-  assertPageUrl,
-} from '../../lib/browser/page.js';
-import { retryWithBackoff, isRetryableError } from '../../lib/retry.js';
-import {
-  extractUserSection,
-  parseUserSection,
-  extractLocationCreated,
-  USER_SECTION_SIZE,
-} from '../../lib/parser.js';
-import { detectPageError } from './page-error-detector.js';
-
-export {
-  delay,
-  setDelayConfig,
-  getDelayConfig,
-  listDelayPresets,
-  DELAY_PRESETS,
-  ensureBrowserReady,
-  ensureTikTokPage,
-  closeCommentPanel,
-  findTikTokPage,
-  getOrCreatePage,
-  isLoggedIn,
-  assertPageUrl,
-  retryWithBackoff,
-  isRetryableError,
-  extractUserSection,
-  parseUserSection,
-  extractLocationCreated,
-  USER_SECTION_SIZE,
-  detectPageError,
-};
-
-export const CDP_PORT = 9222;
+import {
+  delay,
+  getDelayConfig,
+  setDelayConfig,
+  listDelayPresets,
+  DELAY_PRESETS,
+} from '../../lib/delay.js';
+import { ensureBrowserReady } from '../../lib/browser/cdp.js';
+import {
+  ensureTikTokPage,
+  closeCommentPanel,
+  findTikTokPage,
+  getOrCreatePage,
+  isLoggedIn,
+  assertPageUrl,
+} from '../../lib/browser/page.js';
+import { retryWithBackoff, isRetryableError } from '../../lib/retry.js';
+import {
+  extractUserSection,
+  parseUserSection,
+  extractLocationCreated,
+  USER_SECTION_SIZE,
+} from '../../lib/parser.js';
+import { detectPageError } from './page-error-detector.js';
+
+export {
+  delay,
+  setDelayConfig,
+  getDelayConfig,
+  listDelayPresets,
+  DELAY_PRESETS,
+  ensureBrowserReady,
+  ensureTikTokPage,
+  closeCommentPanel,
+  findTikTokPage,
+  getOrCreatePage,
+  isLoggedIn,
+  assertPageUrl,
+  retryWithBackoff,
+  isRetryableError,
+  extractUserSection,
+  parseUserSection,
+  extractLocationCreated,
+  USER_SECTION_SIZE,
+  detectPageError,
+};
+
+export const CDP_PORT = 9222;