tt-help-cli-ycl 1.0.8 → 1.1.0

package/src/scraper/core.cjs

@@ -1,214 +1,214 @@
-const {
-  closeCommentPanel,
-  delay,
-  ensureBrowserReady,
-  ensureTikTokPage,
-  setDelayConfig,
-  getDelayConfig,
-  retryWithBackoff,
-} = require("./modules/page-helpers.cjs");
-const { extractCommentAuthors } = require("./modules/comment-extractor.cjs");
-const { extractGuessVideos } = require("./modules/guess-extractor.cjs");
-
-async function scrapeSingleVideo(page, maxComments, maxGuess, log, location = 'ES') {
-  const config = getDelayConfig();
-
-  await page
-    .waitForSelector('[class*="VideoMeta"]', { timeout: 10000 })
-    .catch(() => {});
-  await delay(Math.round(config.commentMax * 0.3), config.commentMax);
-
-  const userData = await page.evaluate(() => {
-    const result = {};
-
-    const m = window.location.href.match(/\/@([^\/]+)\/video/);
-    if (m) result.uniqueId = m[1];
-
-    const authorEls = document.querySelectorAll('[class*="Author"]');
-    for (const el of authorEls) {
-      const text = (el.textContent || "").trim();
-      if (text && !text.includes("TikTok") && !text.includes("Share")) {
-        result.nickname = text;
-        break;
-      }
-    }
-
-    const html = document.documentElement.outerHTML;
-    const locMatch = html.match(/"locationCreated":"([^"]*)/);
-    if (locMatch) result.locationCreated = locMatch[1];
-
-    return result;
-  });
-
-  const videoAuthor = userData.uniqueId ? "@" + userData.uniqueId : null;
-  if (!videoAuthor) {
-    throw new Error("无法获取视频作者");
-  }
-
-  let guessVideos = [];
-  let commentUsers = [];
-
-  if (userData.locationCreated === location) {
-    guessVideos = await extractGuessVideos(page, maxGuess);
-
-    commentUsers = await extractCommentAuthors(page, maxComments);
-    await closeCommentPanel(page);
-    await delay(Math.round(config.commentMax * 0.3), config.commentMax);
-  }
-
-  const uniqueUsers = [...new Set(commentUsers)];
-
-  return {
-    videoAuthor,
-    uniqueId: userData.uniqueId,
-    nickname: userData.nickname,
-    locationCreated: userData.locationCreated,
-    commentUsers: uniqueUsers,
-    guessVideos,
-  };
-}
-
-async function runScrape(options) {
-  const {
-    videoUrl,
-    maxVideos = 20,
-    maxComments = 999,
-    maxGuess = 10,
-    preset = null,
-    switchMax = null,
-    commentMax = null,
-    log = console.error,
-    browser: externalBrowser = null,
-    page: externalPage = null,
-  } = options;
-
-  if (preset) {
-    setDelayConfig(preset);
-  } else if (switchMax || commentMax) {
-    setDelayConfig({
-      switchMax: switchMax || 5000,
-      commentMax: commentMax || 3000,
-    });
-  }
-
-  const config = getDelayConfig();
-
-  let browser, page;
-  const isExternal = !!(externalBrowser && externalPage);
-
-  if (!isExternal) {
-    log(`视频地址: ${videoUrl}`);
-    log(
-      `视频数: ${maxVideos}, 评论数: ${maxComments}, 猜你喜欢: ${maxGuess}, 切换延迟: ${config.switchMax}ms, 评论延迟: ${config.commentMax}ms`,
-    );
-  }
-
-  if (isExternal) {
-    browser = externalBrowser;
-    page = externalPage;
-  } else {
-    browser = await ensureBrowserReady();
-    try {
-      page = await ensureTikTokPage(browser, videoUrl);
-    } catch (e) {
-      await browser.close().catch(() => {});
-      throw e;
-    }
-  }
-
-  await retryWithBackoff(() => page.goto(videoUrl, { waitUntil: "load", timeout: 30000 }), { log });
-  await delay(Math.round(config.switchMax * 0.5), config.switchMax);
-  await closeCommentPanel(page);
-  await delay(Math.round(config.commentMax * 0.5), config.commentMax);
-
-  const allResults = [];
-  const videoAuthors = new Set();
-  const commentUsers = new Set();
-  const allCommentAuthorsList = [];
-  const allGuessAuthors = new Set();
-  const allGuessVideos = [];
-
-  for (let i = 0; i < maxVideos; i++) {
-    await delay(Math.round(config.commentMax * 0.3), config.commentMax);
-
-    let result;
-    try {
-      result = await scrapeSingleVideo(page, maxComments, maxGuess, log);
-    } catch (e) {
-      log(`[${i + 1}/${maxVideos}] 跳过: ${e.message}`);
-      if (i < maxVideos - 1) {
-        await page.evaluate(() => {
-          const container = document.querySelector(
-            '[class*="ColumnListContainer"]',
-          );
-          if (container) container.scrollTop += 700;
-          else window.scrollBy(0, 700);
-        });
-        await delay(Math.round(config.switchMax * 0.5), config.switchMax);
-      }
-      continue;
-    }
-    allResults.push(result);
-    videoAuthors.add(result.videoAuthor);
-    result.commentUsers.forEach((u) => commentUsers.add(u));
-    allCommentAuthorsList.push(...result.commentUsers);
-    if (result.guessVideos) {
-      allGuessVideos.push(...result.guessVideos);
-      result.guessVideos.forEach((v) => {
-        if (v.author) allGuessAuthors.add(v.author);
-      });
-    }
-
-    if ((i + 1) % 5 === 0 || i === 0) {
-      log(
-        `[${i + 1}/${maxVideos}] ${result.videoAuthor} | 昵称: ${result.nickname || "-"} | 评论用户: ${result.commentUsers.length} | 猜你喜欢: ${result.guessVideos ? result.guessVideos.length : 0}`,
-      );
-    }
-
-    if (i < maxVideos - 1) {
-      await page.evaluate(() => {
-        const container = document.querySelector(
-          '[class*="ColumnListContainer"]',
-        );
-        if (container) container.scrollTop += 700;
-      });
-      await delay(2000, config.switchMax);
-    }
-  }
-
-  log(
-    `\n结果: 视频作者 ${videoAuthors.size} | 评论用户 ${commentUsers.size} | 总评论 ${allCommentAuthorsList.length} | 猜你喜欢作者 ${allGuessAuthors.size} | 总猜中视频 ${allGuessVideos.length}`,
-  );
-
-  const videoDetails = {};
-  for (const r of allResults) {
-    const key = r.videoAuthor;
-    if (!videoDetails[key]) {
-      videoDetails[key] = {
-        videoAuthor: r.videoAuthor,
-        uniqueId: r.uniqueId,
-        nickname: r.nickname,
-        locationCreated: r.locationCreated,
-      };
-    }
-  }
-
-  const output = {
-    videoDetails: Object.values(videoDetails),
-    commentUsers: [...commentUsers].sort(),
-    allCommentAuthorsList,
-    guessVideos: allGuessVideos,
-    guessAuthors: [...allGuessAuthors].sort(),
-    stats: {
-      totalVideos: allResults.length,
-      uniqueVideoAuthors: videoAuthors.size,
-      uniqueCommentAuthors: commentUsers.size,
-      uniqueGuessAuthors: allGuessAuthors.size,
-      totalGuessVideos: allGuessVideos.length,
-    },
-  };
-
-  return { output, browser, isExternal };
-}
-
+const {
+  closeCommentPanel,
+  delay,
+  ensureBrowserReady,
+  ensureTikTokPage,
+  setDelayConfig,
+  getDelayConfig,
+  retryWithBackoff,
+} = require("./modules/page-helpers.cjs");
+const { extractCommentAuthors } = require("./modules/comment-extractor.cjs");
+const { extractGuessVideos } = require("./modules/guess-extractor.cjs");
+
+async function scrapeSingleVideo(page, maxComments, maxGuess, log, location = 'ES') {
+  const config = getDelayConfig();
+
+  await page
+    .waitForSelector('[class*="VideoMeta"]', { timeout: 10000 })
+    .catch(() => {});
+  await delay(Math.round(config.commentMax * 0.3), config.commentMax);
+
+  const userData = await page.evaluate(() => {
+    const result = {};
+
+    const m = window.location.href.match(/\/@([^\/]+)\/video/);
+    if (m) result.uniqueId = m[1];
+
+    const authorEls = document.querySelectorAll('[class*="Author"]');
+    for (const el of authorEls) {
+      const text = (el.textContent || "").trim();
+      if (text && !text.includes("TikTok") && !text.includes("Share")) {
+        result.nickname = text;
+        break;
+      }
+    }
+
+    const html = document.documentElement.outerHTML;
+    const locMatch = html.match(/"locationCreated":"([^"]*)/);
+    if (locMatch) result.locationCreated = locMatch[1];
+
+    return result;
+  });
+
+  const videoAuthor = userData.uniqueId ? "@" + userData.uniqueId : null;
+  if (!videoAuthor) {
+    throw new Error("无法获取视频作者");
+  }
+
+  let guessVideos = [];
+  let commentUsers = [];
+
+  if (userData.locationCreated === location) {
+    guessVideos = await extractGuessVideos(page, maxGuess);
+
+    commentUsers = await extractCommentAuthors(page, maxComments);
+    await closeCommentPanel(page);
+    await delay(Math.round(config.commentMax * 0.3), config.commentMax);
+  }
+
+  const uniqueUsers = [...new Set(commentUsers)];
+
+  return {
+    videoAuthor,
+    uniqueId: userData.uniqueId,
+    nickname: userData.nickname,
+    locationCreated: userData.locationCreated,
+    commentUsers: uniqueUsers,
+    guessVideos,
+  };
+}
+
+async function runScrape(options) {
+  const {
+    videoUrl,
+    maxVideos = 20,
+    maxComments = 999,
+    maxGuess = 10,
+    preset = null,
+    switchMax = null,
+    commentMax = null,
+    log = console.error,
+    browser: externalBrowser = null,
+    page: externalPage = null,
+  } = options;
+
+  if (preset) {
+    setDelayConfig(preset);
+  } else if (switchMax || commentMax) {
+    setDelayConfig({
+      switchMax: switchMax || 5000,
+      commentMax: commentMax || 3000,
+    });
+  }
+
+  const config = getDelayConfig();
+
+  let browser, page;
+  const isExternal = !!(externalBrowser && externalPage);
+
+  if (!isExternal) {
+    log(`视频地址: ${videoUrl}`);
+    log(
+      `视频数: ${maxVideos}, 评论数: ${maxComments}, 猜你喜欢: ${maxGuess}, 切换延迟: ${config.switchMax}ms, 评论延迟: ${config.commentMax}ms`,
+    );
+  }
+
+  if (isExternal) {
+    browser = externalBrowser;
+    page = externalPage;
+  } else {
+    browser = await ensureBrowserReady();
+    try {
+      page = await ensureTikTokPage(browser, videoUrl);
+    } catch (e) {
+      await browser.close().catch(() => {});
+      throw e;
+    }
+  }
+
+  await retryWithBackoff(() => page.goto(videoUrl, { waitUntil: "load", timeout: 30000 }), { log });
+  await delay(Math.round(config.switchMax * 0.5), config.switchMax);
+  await closeCommentPanel(page);
+  await delay(Math.round(config.commentMax * 0.5), config.commentMax);
+
+  const allResults = [];
+  const videoAuthors = new Set();
+  const commentUsers = new Set();
+  const allCommentAuthorsList = [];
+  const allGuessAuthors = new Set();
+  const allGuessVideos = [];
+
+  for (let i = 0; i < maxVideos; i++) {
+    await delay(Math.round(config.commentMax * 0.3), config.commentMax);
+
+    let result;
+    try {
+      result = await scrapeSingleVideo(page, maxComments, maxGuess, log);
+    } catch (e) {
+      log(`[${i + 1}/${maxVideos}] 跳过: ${e.message}`);
+      if (i < maxVideos - 1) {
+        await page.evaluate(() => {
+          const container = document.querySelector(
+            '[class*="ColumnListContainer"]',
+          );
+          if (container) container.scrollTop += 700;
+          else window.scrollBy(0, 700);
+        });
+        await delay(Math.round(config.switchMax * 0.5), config.switchMax);
+      }
+      continue;
+    }
+    allResults.push(result);
+    videoAuthors.add(result.videoAuthor);
+    result.commentUsers.forEach((u) => commentUsers.add(u));
+    allCommentAuthorsList.push(...result.commentUsers);
+    if (result.guessVideos) {
+      allGuessVideos.push(...result.guessVideos);
+      result.guessVideos.forEach((v) => {
+        if (v.author) allGuessAuthors.add(v.author);
+      });
+    }
+
+    if ((i + 1) % 5 === 0 || i === 0) {
+      log(
+        `[${i + 1}/${maxVideos}] ${result.videoAuthor} | 昵称: ${result.nickname || "-"} | 评论用户: ${result.commentUsers.length} | 猜你喜欢: ${result.guessVideos ? result.guessVideos.length : 0}`,
+      );
+    }
+
+    if (i < maxVideos - 1) {
+      await page.evaluate(() => {
+        const container = document.querySelector(
+          '[class*="ColumnListContainer"]',
+        );
+        if (container) container.scrollTop += 700;
+      });
+      await delay(2000, config.switchMax);
+    }
+  }
+
+  log(
+    `\n结果: 视频作者 ${videoAuthors.size} | 评论用户 ${commentUsers.size} | 总评论 ${allCommentAuthorsList.length} | 猜你喜欢作者 ${allGuessAuthors.size} | 总猜中视频 ${allGuessVideos.length}`,
+  );
+
+  const videoDetails = {};
+  for (const r of allResults) {
+    const key = r.videoAuthor;
+    if (!videoDetails[key]) {
+      videoDetails[key] = {
+        videoAuthor: r.videoAuthor,
+        uniqueId: r.uniqueId,
+        nickname: r.nickname,
+        locationCreated: r.locationCreated,
+      };
+    }
+  }
+
+  const output = {
+    videoDetails: Object.values(videoDetails),
+    commentUsers: [...commentUsers].sort(),
+    allCommentAuthorsList,
+    guessVideos: allGuessVideos,
+    guessAuthors: [...allGuessAuthors].sort(),
+    stats: {
+      totalVideos: allResults.length,
+      uniqueVideoAuthors: videoAuthors.size,
+      uniqueCommentAuthors: commentUsers.size,
+      uniqueGuessAuthors: allGuessAuthors.size,
+      totalGuessVideos: allGuessVideos.length,
+    },
+  };
+
+  return { output, browser, isExternal };
+}
+
 module.exports = { scrapeSingleVideo, runScrape };

package/src/scraper/index.cjs

@@ -1,97 +1,97 @@
-const { runScrape } = require("./core.cjs");
-
-async function main() {
-  const rawArgs = process.argv.slice(2);
-
-  let outputPath = null;
-  const args = [];
-  for (let i = 0; i < rawArgs.length; i++) {
-    if (rawArgs[i] === "-o" || rawArgs[i] === "--output") {
-      outputPath = rawArgs[++i];
-    } else {
-      args.push(rawArgs[i]);
-    }
-  }
-
-  const videoUrl = args[0];
-
-  let preset = null;
-  let maxVideos = 20;
-  let maxComments = 999;
-  let maxGuess = 10;
-  let switchMax = null;
-  let commentMax = null;
-
-  if (args[1]) {
-    if (["fast", "normal", "slow", "stealth"].includes(args[1].toLowerCase())) {
-      preset = args[1].toLowerCase();
-      maxVideos = parseInt(args[2]) || 20;
-      maxComments = parseInt(args[3]) || 999;
-      maxGuess = parseInt(args[4]) || 10;
-    } else {
-      maxVideos = parseInt(args[1]) || 20;
-      maxComments = parseInt(args[2]) || 999;
-      maxGuess = parseInt(args[3]) || 10;
-      switchMax = parseInt(args[4]) || null;
-      commentMax = parseInt(args[5]) || null;
-    }
-  }
-
-  if (!videoUrl) {
-    console.error("用法:");
-    console.error(
-      "  预设模式: node index.cjs <视频URL> <preset> [最大视频数] [最大评论数] [猜你喜欢数] [-o 输出路径]",
-    );
-    console.error(
-      "  手动模式: node index.cjs <视频URL> [最大视频数] [最大评论数] [猜你喜欢数] [切换延迟ms] [评论延迟ms] [-o 输出路径]",
-    );
-    console.error(
-      "预设: fast(1s/0.8s), normal(2.5s/1.5s), slow(5s/3s), stealth(8s/5s)",
-    );
-    console.error(
-      "选项: -o, --output <路径>  输出到文件; 不指定则输出到 stdout",
-    );
-    process.exit(1);
-  }
-
-  let browser;
-  try {
-    const { output, browser: b, isExternal } = await runScrape({
-      videoUrl,
-      maxVideos,
-      maxComments,
-      maxGuess,
-      preset,
-      switchMax,
-      commentMax,
-      log: console.error,
-    });
-    if (!isExternal) {
-      browser = b;
-    }
-
-    const json = JSON.stringify(output, null, 2);
-    if (outputPath) {
-      const fs = require("fs");
-      const path = require("path");
-      const resultFile = path.isAbsolute(outputPath)
-        ? outputPath
-        : path.resolve(outputPath);
-      fs.mkdirSync(path.dirname(resultFile), { recursive: true });
-      fs.writeFileSync(resultFile, json);
-      console.error(`已保存到 ${resultFile}`);
-    } else {
-      process.stdout.write(json + "\n");
-    }
-  } catch (err) {
-    console.error(err.message);
-    process.exit(1);
-  } finally {
-    if (browser) await browser.close().catch(() => {});
-  }
-}
-
-main().catch((err) => {
-  console.error(err.message);
-  process.exit(1);
+const { runScrape } = require("./core.cjs");
+
+async function main() {
+  const rawArgs = process.argv.slice(2);
+
+  let outputPath = null;
+  const args = [];
+  for (let i = 0; i < rawArgs.length; i++) {
+    if (rawArgs[i] === "-o" || rawArgs[i] === "--output") {
+      outputPath = rawArgs[++i];
+    } else {
+      args.push(rawArgs[i]);
+    }
+  }
+
+  const videoUrl = args[0];
+
+  let preset = null;
+  let maxVideos = 20;
+  let maxComments = 999;
+  let maxGuess = 10;
+  let switchMax = null;
+  let commentMax = null;
+
+  if (args[1]) {
+    if (["fast", "normal", "slow", "stealth"].includes(args[1].toLowerCase())) {
+      preset = args[1].toLowerCase();
+      maxVideos = parseInt(args[2]) || 20;
+      maxComments = parseInt(args[3]) || 999;
+      maxGuess = parseInt(args[4]) || 10;
+    } else {
+      maxVideos = parseInt(args[1]) || 20;
+      maxComments = parseInt(args[2]) || 999;
+      maxGuess = parseInt(args[3]) || 10;
+      switchMax = parseInt(args[4]) || null;
+      commentMax = parseInt(args[5]) || null;
+    }
+  }
+
+  if (!videoUrl) {
+    console.error("用法:");
+    console.error(
+      "  预设模式: node index.cjs <视频URL> <preset> [最大视频数] [最大评论数] [猜你喜欢数] [-o 输出路径]",
+    );
+    console.error(
+      "  手动模式: node index.cjs <视频URL> [最大视频数] [最大评论数] [猜你喜欢数] [切换延迟ms] [评论延迟ms] [-o 输出路径]",
+    );
+    console.error(
+      "预设: fast(1s/0.8s), normal(2.5s/1.5s), slow(5s/3s), stealth(8s/5s)",
+    );
+    console.error(
+      "选项: -o, --output <路径>  输出到文件; 不指定则输出到 stdout",
+    );
+    process.exit(1);
+  }
+
+  let browser;
+  try {
+    const { output, browser: b, isExternal } = await runScrape({
+      videoUrl,
+      maxVideos,
+      maxComments,
+      maxGuess,
+      preset,
+      switchMax,
+      commentMax,
+      log: console.error,
+    });
+    if (!isExternal) {
+      browser = b;
+    }
+
+    const json = JSON.stringify(output, null, 2);
+    if (outputPath) {
+      const fs = require("fs");
+      const path = require("path");
+      const resultFile = path.isAbsolute(outputPath)
+        ? outputPath
+        : path.resolve(outputPath);
+      fs.mkdirSync(path.dirname(resultFile), { recursive: true });
+      fs.writeFileSync(resultFile, json);
+      console.error(`已保存到 ${resultFile}`);
+    } else {
+      process.stdout.write(json + "\n");
+    }
+  } catch (err) {
+    console.error(err.message);
+    process.exit(1);
+  } finally {
+    if (browser) await browser.close().catch(() => {});
+  }
+}
+
+main().catch((err) => {
+  console.error(err.message);
+  process.exit(1);
 });