tt-help-cli-ycl 1.3.45 → 1.3.47

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

@@ -1,250 +1,250 @@
-import { delay, getDelayConfig } from "./page-helpers.js";
-import { scrollAndCollect } from "./scroll-collector.js";
-import { extractUniqueId, toProfileUrl } from "../../lib/url.js";
-
-const FILTER_WORDS = ["主页", "已关注", "粉丝", "推荐"];
-
-const FOLLOW_TRIGGER_SELECTORS = [
-  "[data-e2e=following]",
-  'a[href$="/following"]',
-  'a[href*="/following"]',
-  '[data-e2e*="following"]',
-];
-
-async function waitForFollowTrigger(page, timeout = 15000) {
-  await page
-    .waitForFunction(
-      (selectors) => {
-        for (const selector of selectors) {
-          if (document.querySelector(selector)) return true;
-        }
-
-        const textMatchers = [/^关注$/, /^Following$/i, /^已关注$/];
-        const nodes = document.querySelectorAll("a,button,div,span");
-        for (const node of nodes) {
-          const text = (node.textContent || "").trim();
-          if (textMatchers.some((matcher) => matcher.test(text))) return true;
-        }
-
-        return false;
-      },
-      FOLLOW_TRIGGER_SELECTORS,
-      { timeout },
-    )
-    .catch(() => {});
-}
-
-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(() => {});
-}
-
-function getProfileUrlForRetry(page) {
-  const currentUrl = page.url();
-  const uniqueId = extractUniqueId(currentUrl);
-  if (uniqueId) return toProfileUrl(uniqueId);
-  return currentUrl;
-}
-
-async function openFollowModal(page, log = () => {}) {
-  const tryOpen = async () =>
-    page.evaluate((selectors) => {
-      const clickTarget = (node) => {
-        if (!node) return false;
-        const clickable =
-          node.closest('a,button,[role="button"]') ||
-          node.parentElement ||
-          node;
-        clickable.click();
-        return true;
-      };
-
-      for (const selector of selectors) {
-        const node = document.querySelector(selector);
-        if (clickTarget(node)) return selector;
-      }
-
-      const textMatchers = [/^关注$/, /^Following$/i, /^已关注$/];
-      const nodes = Array.from(document.querySelectorAll("a,button,div,span"));
-      for (const node of nodes) {
-        const text = (node.textContent || "").trim();
-        if (!text) continue;
-        if (textMatchers.some((matcher) => matcher.test(text))) {
-          if (clickTarget(node)) return `text:${text}`;
-        }
-      }
-
-      return null;
-    }, FOLLOW_TRIGGER_SELECTORS);
-
-  let opened = null;
-  for (let attempt = 1; attempt <= 3; attempt++) {
-    await waitForFollowTrigger(page, attempt === 1 ? 15000 : 8000);
-    opened = await tryOpen();
-    if (opened) {
-      if (attempt >= 2) {
-        log(`  [关注弹窗] 第 ${attempt} 次已点击关注入口: ${opened}`);
-      }
-      break;
-    }
-
-    if (attempt >= 2) {
-      log(`  [关注弹窗] 第 ${attempt} 次仍未找到可点击的关注入口`);
-    }
-
-    await page
-      .evaluate(() => {
-        window.scrollTo({ top: 0, behavior: "instant" });
-      })
-      .catch(() => {});
-    await delay(800, 1500);
-  }
-
-  if (!opened) {
-    log("  [关注弹窗] 重试后仍未找到关注入口");
-    throw new Error(
-      "未找到关注入口元素，请确认当前页面为用户主页或页面结构已变化",
-    );
-  }
-
-  // 等待用户列表容器，超时说明可能被弹窗遮挡
-  let containerReady = false;
-  for (let attempt = 1; attempt <= 2; attempt++) {
-    try {
-      await page.waitForSelector("[class*=DivUserListContainer]", {
-        timeout: 30000,
-      });
-      containerReady = true;
-      if (attempt >= 2) {
-        log(`  [关注弹窗] 第 ${attempt} 次等待后，关注列表容器已出现`);
-      }
-      break;
-    } catch (e) {
-      if (attempt === 1) {
-        // 第一次超时，重新 goto 用户主页后重试
-        const retryUrl = getProfileUrlForRetry(page);
-        log(
-          `  [关注弹窗] 第 1 次等待列表超时，准备 goto 主页重试: ${retryUrl} | ${e.message}`,
-        );
-        await page.goto(retryUrl, { waitUntil: "domcontentloaded" });
-        await delay(2000, 3000);
-        log("  [关注弹窗] 主页跳转完成，重新点击关注入口");
-        // 重新点击 follow 入口
-        opened = await tryOpen();
-        if (!opened) {
-          log("  [关注弹窗] goto 主页后仍未找到关注入口");
-          throw new Error(
-            "goto 主页后仍未找到关注入口元素，请确认当前页面为用户主页",
-          );
-        }
-        log(`  [关注弹窗] goto 主页后已重新点击关注入口: ${opened}`);
-      } else {
-        log(`  [关注弹窗] 第 2 次等待列表仍超时，结束本次提取: ${e.message}`);
-        throw e;
-      }
-    }
-  }
-
-  if (containerReady) {
-    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, log);
-
-  const following = await extractUsersFromModal(page, maxFollowing);
-  log(`  已关注: ${following.length}`);
-
-  await switchToFollowersTab(page);
-
-  const followers = await extractUsersFromModal(page, maxFollowers);
-  log(`  粉丝: ${followers.length}`);
-
-  await closeFollowModal(page);
-
-  return {
-    following: following.map((u) => [u.handle, u.displayName]),
-    followers: followers.map((u) => [u.handle, u.displayName]),
-  };
-}
-
-export { extractFollowAndFollowers };
+import { delay, getDelayConfig } from "./page-helpers.js";
+import { scrollAndCollect } from "./scroll-collector.js";
+import { extractUniqueId, toProfileUrl } from "../../lib/url.js";
+
+const FILTER_WORDS = ["主页", "已关注", "粉丝", "推荐"];
+
+const FOLLOW_TRIGGER_SELECTORS = [
+  "[data-e2e=following]",
+  'a[href$="/following"]',
+  'a[href*="/following"]',
+  '[data-e2e*="following"]',
+];
+
+async function waitForFollowTrigger(page, timeout = 15000) {
+  await page
+    .waitForFunction(
+      (selectors) => {
+        for (const selector of selectors) {
+          if (document.querySelector(selector)) return true;
+        }
+
+        const textMatchers = [/^关注$/, /^Following$/i, /^已关注$/];
+        const nodes = document.querySelectorAll("a,button,div,span");
+        for (const node of nodes) {
+          const text = (node.textContent || "").trim();
+          if (textMatchers.some((matcher) => matcher.test(text))) return true;
+        }
+
+        return false;
+      },
+      FOLLOW_TRIGGER_SELECTORS,
+      { timeout },
+    )
+    .catch(() => {});
+}
+
+async function waitForListContent(page, minChildren = 1, timeout = 15000) {
+  await page
+    .waitForFunction(
+      (min) => {
+        const container = document.querySelector(
+          "[class*=DivUserListContainer]",
+        );
+        return container && container.children.length >= min;
+      },
+      minChildren,
+      { timeout },
+    )
+    .catch(() => {});
+}
+
+function getProfileUrlForRetry(page) {
+  const currentUrl = page.url();
+  const uniqueId = extractUniqueId(currentUrl);
+  if (uniqueId) return toProfileUrl(uniqueId);
+  return currentUrl;
+}
+
+async function openFollowModal(page, log = () => {}) {
+  const tryOpen = async () =>
+    page.evaluate((selectors) => {
+      const clickTarget = (node) => {
+        if (!node) return false;
+        const clickable =
+          node.closest('a,button,[role="button"]') ||
+          node.parentElement ||
+          node;
+        clickable.click();
+        return true;
+      };
+
+      for (const selector of selectors) {
+        const node = document.querySelector(selector);
+        if (clickTarget(node)) return selector;
+      }
+
+      const textMatchers = [/^关注$/, /^Following$/i, /^已关注$/];
+      const nodes = Array.from(document.querySelectorAll("a,button,div,span"));
+      for (const node of nodes) {
+        const text = (node.textContent || "").trim();
+        if (!text) continue;
+        if (textMatchers.some((matcher) => matcher.test(text))) {
+          if (clickTarget(node)) return `text:${text}`;
+        }
+      }
+
+      return null;
+    }, FOLLOW_TRIGGER_SELECTORS);
+
+  let opened = null;
+  for (let attempt = 1; attempt <= 3; attempt++) {
+    await waitForFollowTrigger(page, attempt === 1 ? 15000 : 8000);
+    opened = await tryOpen();
+    if (opened) {
+      if (attempt >= 2) {
+        log(`  [关注弹窗] 第 ${attempt} 次已点击关注入口: ${opened}`);
+      }
+      break;
+    }
+
+    if (attempt >= 2) {
+      log(`  [关注弹窗] 第 ${attempt} 次仍未找到可点击的关注入口`);
+    }
+
+    await page
+      .evaluate(() => {
+        window.scrollTo({ top: 0, behavior: "instant" });
+      })
+      .catch(() => {});
+    await delay(800, 1500);
+  }
+
+  if (!opened) {
+    log("  [关注弹窗] 重试后仍未找到关注入口");
+    throw new Error(
+      "未找到关注入口元素，请确认当前页面为用户主页或页面结构已变化",
+    );
+  }
+
+  // 等待用户列表容器，超时说明可能被弹窗遮挡
+  let containerReady = false;
+  for (let attempt = 1; attempt <= 2; attempt++) {
+    try {
+      await page.waitForSelector("[class*=DivUserListContainer]", {
+        timeout: 30000,
+      });
+      containerReady = true;
+      if (attempt >= 2) {
+        log(`  [关注弹窗] 第 ${attempt} 次等待后，关注列表容器已出现`);
+      }
+      break;
+    } catch (e) {
+      if (attempt === 1) {
+        // 第一次超时，重新 goto 用户主页后重试
+        const retryUrl = getProfileUrlForRetry(page);
+        log(
+          `  [关注弹窗] 第 1 次等待列表超时，准备 goto 主页重试: ${retryUrl} | ${e.message}`,
+        );
+        await page.goto(retryUrl, { waitUntil: "domcontentloaded" });
+        await delay(2000, 3000);
+        log("  [关注弹窗] 主页跳转完成，重新点击关注入口");
+        // 重新点击 follow 入口
+        opened = await tryOpen();
+        if (!opened) {
+          log("  [关注弹窗] goto 主页后仍未找到关注入口");
+          throw new Error(
+            "goto 主页后仍未找到关注入口元素，请确认当前页面为用户主页",
+          );
+        }
+        log(`  [关注弹窗] goto 主页后已重新点击关注入口: ${opened}`);
+      } else {
+        log(`  [关注弹窗] 第 2 次等待列表仍超时，结束本次提取: ${e.message}`);
+        throw e;
+      }
+    }
+  }
+
+  if (containerReady) {
+    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, log);
+
+  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 };