tt-help-cli-ycl 1.3.1 → 1.3.2

package/README.md

@@ -1,17 +1,17 @@
-# tt-help
-
-TikTok user & video data scraper — extract ttSeller, verified, locationCreated from HTML source.
-
-## Install
-
-```bash
-npm i -g tt-help
-```
-
-## Usage
-
-```bash
-tt-help [options] <urls...>
-```
-
-示例、代理配置、和更多用法请查看仓库或 `src/main.mjs` 中的帮助文本。
+# tt-help
+
+TikTok user & video data scraper — extract ttSeller, verified, locationCreated from HTML source.
+
+## Install
+
+```bash
+npm i -g tt-help
+```
+
+## Usage
+
+```bash
+tt-help [options] <urls...>
+```
+
+示例、代理配置、和更多用法请查看仓库或 `src/main.mjs` 中的帮助文本。

package/cli.js

@@ -1,9 +1,9 @@
-#!/usr/bin/env node
-import { fileURLToPath } from 'url';
-import { dirname, resolve } from 'path';
-
-const __filename = fileURLToPath(import.meta.url);
-const __dirname = dirname(__filename);
-
-const mainPath = resolve(__dirname, 'src', 'main.mjs');
-await import(`file://${mainPath}`);
+#!/usr/bin/env node
+import { fileURLToPath } from 'url';
+import { dirname, resolve } from 'path';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+
+const mainPath = resolve(__dirname, 'src', 'main.mjs');
+await import(`file://${mainPath}`);

package/package.json

@@ -1,44 +1,44 @@
-{
-  "name": "tt-help-cli-ycl",
-  "version": "1.3.1",
-  "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
-  "type": "module",
-  "bin": {
-    "tt-help": "cli.js"
-  },
-  "main": "src/main.mjs",
-  "files": [
-    "cli.js",
-    "src/"
-  ],
-  "scripts": {
-    "start": "node src/main.mjs"
-  },
-  "keywords": [
-    "tiktok",
-    "scraper",
-    "cli",
-    "seller",
-    "ttSeller"
-  ],
-  "author": "jsjhycl",
-  "license": "ISC",
-  "engines": {
-    "node": ">=18"
-  },
-  "publishConfig": {
-    "access": "public"
-  },
-  "repository": {
-    "type": "git",
-    "url": "git+https://github.com/jsjhycl/tt-help-cli.git"
-  },
-  "bugs": {
-    "url": "https://github.com/jsjhycl/tt-help-cli/issues"
-  },
-  "homepage": "https://github.com/jsjhycl/tt-help-cli#readme",
-  "dependencies": {
-    "playwright": "^1.59.1",
-    "undici": "^8.1.0"
-  }
-}
+{
+  "name": "tt-help-cli-ycl",
+  "version": "1.3.2",
+  "description": "TikTok user & video data scraper - extract ttSeller, verified, locationCreated from HTML source",
+  "type": "module",
+  "bin": {
+    "tt-help": "cli.js"
+  },
+  "main": "src/main.mjs",
+  "files": [
+    "cli.js",
+    "src/"
+  ],
+  "scripts": {
+    "start": "node src/main.mjs"
+  },
+  "keywords": [
+    "tiktok",
+    "scraper",
+    "cli",
+    "seller",
+    "ttSeller"
+  ],
+  "author": "jsjhycl",
+  "license": "ISC",
+  "engines": {
+    "node": ">=18"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/jsjhycl/tt-help-cli.git"
+  },
+  "bugs": {
+    "url": "https://github.com/jsjhycl/tt-help-cli/issues"
+  },
+  "homepage": "https://github.com/jsjhycl/tt-help-cli#readme",
+  "dependencies": {
+    "playwright": "^1.59.1",
+    "undici": "^8.1.0"
+  }
+}

package/src/cli/auto.js

@@ -45,7 +45,7 @@ export async function handleAuto(options) {
     console.error(`种子用户: ${added} 个新增, ${skipped} 个已存在`);
   }
 
-  const { ensureBrowserReady, processUser } = await import('../lib/auto-browser.mjs');
+  const { ensureBrowserReady, processUser } = await import('../scraper/auto-core.mjs');
   const browser = await ensureBrowserReady();
 
   try {

package/src/cli/explore.js

@@ -42,7 +42,7 @@ export async function handleExplore(options) {
   console.error(`关注/粉丝: ${exploreEnableFollow ? '启用' : '禁用'}`);
   if (exploreMaxUsers > 0) console.error(`上限: ${exploreMaxUsers} 个用户`);
 
-  const { ensureBrowserReady, processExplore } = await import('../lib/auto-browser.mjs');
+  const { ensureBrowserReady, processExplore } = await import('../scraper/explore-core.mjs');
   const browser = await ensureBrowserReady();
 
   try {

package/src/cli/progress.js

@@ -1,111 +1,111 @@
-import { writeFileSync } from 'fs';
-import { formatOutput } from '../lib/output.js';
-import { deduplicate } from '../lib/output.js';
-import { applyFilter, formatFilterDescription } from '../lib/filter.js';
-import { calculateConcurrency, createMultiProgressBars, renderMultiProgressBars, clearProgressBars } from '../lib/io.js';
-import { randomDelay } from '../lib/delay.js';
-
-export async function processUrlsWithProgress({
-  urls,
-  proxyUrl,
-  outputFile,
-  outputFormat,
-  filter,
-  processFn,
-  label = '数据',
-}) {
-  const allResults = [];
-  const errors = [];
-
-  if (urls.length === 0) {
-    console.error('\n未获取到数据');
-    if (outputFile) writeFileSync(outputFile, '[]', 'utf-8');
-    return;
-  }
-
-  const concurrency = calculateConcurrency(urls.length);
-  const bars = createMultiProgressBars(concurrency);
-
-  const slots = Array.from({ length: concurrency }, () => []);
-  urls.forEach((url, i) => slots[i % concurrency].push(url));
-
-  bars.forEach((bar, i) => {
-    bar.total = slots[i].length;
-    bar.status = slots[i].length > 0 ? 'running' : 'done';
-  });
-
-  renderMultiProgressBars(bars);
-
-  const workers = slots.map(async (slotUrls, slotIndex) => {
-    for (const url of slotUrls) {
-      bars[slotIndex].url = url;
-      renderMultiProgressBars(bars);
-
-      await randomDelay();
-
-      try {
-        const results = await processFn(url, proxyUrl);
-        allResults.push(...results);
-        bars[slotIndex].current++;
-        bars[slotIndex].status = 'running';
-      } catch (err) {
-        errors.push({ url, message: err.message });
-        bars[slotIndex].current++;
-        bars[slotIndex].status = 'error';
-      }
-
-      renderMultiProgressBars(bars);
-    }
-    bars[slotIndex].status = bars[slotIndex].current === bars[slotIndex].total ? 'done' : 'error';
-    renderMultiProgressBars(bars);
-  });
-
-  await Promise.all(workers);
-  clearProgressBars();
-
-  const uniqueResults = deduplicate(allResults);
-  const filteredResults = applyFilter(uniqueResults, filter);
-
-  if (errors.length > 0) {
-    const firstMsg = errors[0].message;
-    const isProxyError = ['不可用', '连接被拒绝', '连接中断', '超时', '无法解析']
-      .some(kw => firstMsg.includes(kw));
-
-    if (filteredResults.length === 0) {
-      if (isProxyError) {
-        console.error(`  所有请求失败，请检查代理: ${proxyUrl}\n`);
-      } else {
-        const show = errors.slice(0, 5);
-        for (const e of show) console.error(`  ✗ ${e.url}: ${e.message}\n`);
-        if (errors.length > 5) console.error(`  ... 还有 ${errors.length - 5} 个失败\n`);
-      }
-      console.error('未获取到数据');
-      if (outputFile) writeFileSync(outputFile, '[]', 'utf-8');
-      return;
-    } else {
-      if (isProxyError) {
-        console.error(`  ${errors.length} 个请求失败，请检查代理: ${proxyUrl}\n`);
-      } else {
-        console.error(`  ${errors.length} 个失败:`);
-        const show = errors.slice(0, 5);
-        for (const e of show) console.error(`    ✗ ${e.url}: ${e.message}`);
-        if (errors.length > 5) console.error(`    ... 还有 ${errors.length - 5} 个`);
-      }
-    }
-  }
-
-  const output = formatOutput(filteredResults, outputFormat);
-
-  if (outputFile) {
-    writeFileSync(outputFile, output, 'utf-8');
-    console.log(`\n结果已写入: ${outputFile}`);
-  } else {
-    process.stdout.write(output + '\n');
-  }
-
-  if (filter) {
-    console.log(`\n共 ${uniqueResults.length} 个${label}，过滤后 ${filteredResults.length} 个（过滤条件: ${formatFilterDescription(filter)}）`);
-  } else {
-    console.log(`\n共 ${filteredResults.length} 个${label}`);
-  }
-}
+import { writeFileSync } from 'fs';
+import { formatOutput } from '../lib/output.js';
+import { deduplicate } from '../lib/output.js';
+import { applyFilter, formatFilterDescription } from '../lib/filter.js';
+import { calculateConcurrency, createMultiProgressBars, renderMultiProgressBars, clearProgressBars } from '../lib/io.js';
+import { randomDelay } from '../lib/delay.js';
+
+export async function processUrlsWithProgress({
+  urls,
+  proxyUrl,
+  outputFile,
+  outputFormat,
+  filter,
+  processFn,
+  label = '数据',
+}) {
+  const allResults = [];
+  const errors = [];
+
+  if (urls.length === 0) {
+    console.error('\n未获取到数据');
+    if (outputFile) writeFileSync(outputFile, '[]', 'utf-8');
+    return;
+  }
+
+  const concurrency = calculateConcurrency(urls.length);
+  const bars = createMultiProgressBars(concurrency);
+
+  const slots = Array.from({ length: concurrency }, () => []);
+  urls.forEach((url, i) => slots[i % concurrency].push(url));
+
+  bars.forEach((bar, i) => {
+    bar.total = slots[i].length;
+    bar.status = slots[i].length > 0 ? 'running' : 'done';
+  });
+
+  renderMultiProgressBars(bars);
+
+  const workers = slots.map(async (slotUrls, slotIndex) => {
+    for (const url of slotUrls) {
+      bars[slotIndex].url = url;
+      renderMultiProgressBars(bars);
+
+      await randomDelay();
+
+      try {
+        const results = await processFn(url, proxyUrl);
+        allResults.push(...results);
+        bars[slotIndex].current++;
+        bars[slotIndex].status = 'running';
+      } catch (err) {
+        errors.push({ url, message: err.message });
+        bars[slotIndex].current++;
+        bars[slotIndex].status = 'error';
+      }
+
+      renderMultiProgressBars(bars);
+    }
+    bars[slotIndex].status = bars[slotIndex].current === bars[slotIndex].total ? 'done' : 'error';
+    renderMultiProgressBars(bars);
+  });
+
+  await Promise.all(workers);
+  clearProgressBars();
+
+  const uniqueResults = deduplicate(allResults);
+  const filteredResults = applyFilter(uniqueResults, filter);
+
+  if (errors.length > 0) {
+    const firstMsg = errors[0].message;
+    const isProxyError = ['不可用', '连接被拒绝', '连接中断', '超时', '无法解析']
+      .some(kw => firstMsg.includes(kw));
+
+    if (filteredResults.length === 0) {
+      if (isProxyError) {
+        console.error(`  所有请求失败，请检查代理: ${proxyUrl}\n`);
+      } else {
+        const show = errors.slice(0, 5);
+        for (const e of show) console.error(`  ✗ ${e.url}: ${e.message}\n`);
+        if (errors.length > 5) console.error(`  ... 还有 ${errors.length - 5} 个失败\n`);
+      }
+      console.error('未获取到数据');
+      if (outputFile) writeFileSync(outputFile, '[]', 'utf-8');
+      return;
+    } else {
+      if (isProxyError) {
+        console.error(`  ${errors.length} 个请求失败，请检查代理: ${proxyUrl}\n`);
+      } else {
+        console.error(`  ${errors.length} 个失败:`);
+        const show = errors.slice(0, 5);
+        for (const e of show) console.error(`    ✗ ${e.url}: ${e.message}`);
+        if (errors.length > 5) console.error(`    ... 还有 ${errors.length - 5} 个`);
+      }
+    }
+  }
+
+  const output = formatOutput(filteredResults, outputFormat);
+
+  if (outputFile) {
+    writeFileSync(outputFile, output, 'utf-8');
+    console.log(`\n结果已写入: ${outputFile}`);
+  } else {
+    process.stdout.write(output + '\n');
+  }
+
+  if (filter) {
+    console.log(`\n共 ${uniqueResults.length} 个${label}，过滤后 ${filteredResults.length} 个（过滤条件: ${formatFilterDescription(filter)}）`);
+  } else {
+    console.log(`\n共 ${filteredResults.length} 个${label}`);
+  }
+}

package/src/cli/scrape.js

@@ -1,47 +1,47 @@
-import { writeFileSync } from 'fs';
-
-export async function handleScrape(options) {
-  const { scrapeUrl, scrapePreset, scrapeMaxVideos, scrapeMaxComments, scrapeMaxGuess, scrapeSwitchDelay, scrapeCommentDelay, outputFile } = options;
-
-  if (!scrapeUrl) {
-    console.error('用法: tt-help scrape <视频URL> [preset] [最大视频数] [最大评论数] [-o 输出路径]');
-    console.error('预设: fast, normal, slow, stealth');
-    console.error('选项: -o, --output <路径>     输出到文件（默认输出到 stdout）');
-    console.error('      --switch-delay <ms>     视频切换延迟（毫秒）');
-    console.error('      --comment-delay <ms>    评论滚动延迟（毫秒）');
-    process.exit(1);
-  }
-
-  const { runScrape } = await import('../lib/scrape-browser.mjs');
-
-  let browser;
-  try {
-    const { output, browser: b } = await runScrape({
-      videoUrl: scrapeUrl,
-      maxVideos: scrapeMaxVideos,
-      maxComments: scrapeMaxComments,
-      maxGuess: scrapeMaxGuess,
-      preset: scrapePreset,
-      switchMax: scrapeSwitchDelay,
-      commentMax: scrapeCommentDelay,
-      log: console.error,
-    });
-    browser = b;
-
-    const json = JSON.stringify(output, null, 2);
-    if (outputFile) {
-      writeFileSync(outputFile, json, 'utf-8');
-      console.error(`结果已写入: ${outputFile}`);
-    } else {
-      process.stdout.write(json + '\n');
-    }
-
-    const stats = output.stats;
-    console.error(`\n共 ${stats.totalVideos} 个视频, ${stats.uniqueVideoAuthors} 个视频作者, ${stats.uniqueCommentAuthors} 个评论作者, ${stats.uniqueGuessAuthors} 个猜你喜欢作者`);
-  } catch (err) {
-    console.error(`浏览器抓取失败: ${err.message}`);
-    process.exit(1);
-  } finally {
-    if (browser) await browser.close().catch(() => {});
-  }
-}
+import { writeFileSync } from 'fs';
+
+export async function handleScrape(options) {
+  const { scrapeUrl, scrapePreset, scrapeMaxVideos, scrapeMaxComments, scrapeMaxGuess, scrapeSwitchDelay, scrapeCommentDelay, outputFile } = options;
+
+  if (!scrapeUrl) {
+    console.error('用法: tt-help scrape <视频URL> [preset] [最大视频数] [最大评论数] [-o 输出路径]');
+    console.error('预设: fast, normal, slow, stealth');
+    console.error('选项: -o, --output <路径>     输出到文件（默认输出到 stdout）');
+    console.error('      --switch-delay <ms>     视频切换延迟（毫秒）');
+    console.error('      --comment-delay <ms>    评论滚动延迟（毫秒）');
+    process.exit(1);
+  }
+
+  const { runScrape } = await import('../scraper/core.mjs');
+
+  let browser;
+  try {
+    const { output, browser: b } = await runScrape({
+      videoUrl: scrapeUrl,
+      maxVideos: scrapeMaxVideos,
+      maxComments: scrapeMaxComments,
+      maxGuess: scrapeMaxGuess,
+      preset: scrapePreset,
+      switchMax: scrapeSwitchDelay,
+      commentMax: scrapeCommentDelay,
+      log: console.error,
+    });
+    browser = b;
+
+    const json = JSON.stringify(output, null, 2);
+    if (outputFile) {
+      writeFileSync(outputFile, json, 'utf-8');
+      console.error(`结果已写入: ${outputFile}`);
+    } else {
+      process.stdout.write(json + '\n');
+    }
+
+    const stats = output.stats;
+    console.error(`\n共 ${stats.totalVideos} 个视频, ${stats.uniqueVideoAuthors} 个视频作者, ${stats.uniqueCommentAuthors} 个评论作者, ${stats.uniqueGuessAuthors} 个猜你喜欢作者`);
+  } catch (err) {
+    console.error(`浏览器抓取失败: ${err.message}`);
+    process.exit(1);
+  } finally {
+    if (browser) await browser.close().catch(() => {});
+  }
+}

package/src/cli/utils.js

@@ -1,18 +1,18 @@
-import { writeFileSync } from 'fs';
-
-export function cleanError(msg) {
-  return msg
-    .replace(/\x1b\[[0-9;]*m/g, '')
-    .replace(/\s*- navigating to.*/s, '')
-    .replace(/\s*Call log:/s, '')
-    .trim();
-}
-
-export function writeJson(data, outputFile) {
-  const json = JSON.stringify(data, null, 2);
-  if (outputFile) {
-    writeFileSync(outputFile, json, 'utf-8');
-  } else {
-    process.stdout.write(json + '\n');
-  }
-}
+import { writeFileSync } from 'fs';
+
+export function cleanError(msg) {
+  return msg
+    .replace(/\x1b\[[0-9;]*m/g, '')
+    .replace(/\s*- navigating to.*/s, '')
+    .replace(/\s*Call log:/s, '')
+    .trim();
+}
+
+export function writeJson(data, outputFile) {
+  const json = JSON.stringify(data, null, 2);
+  if (outputFile) {
+    writeFileSync(outputFile, json, 'utf-8');
+  } else {
+    process.stdout.write(json + '\n');
+  }
+}

package/src/cli/videos.js

@@ -1,41 +1,41 @@
-import { writeFileSync } from 'fs';
-
-export async function handleVideos(options) {
-  const { videosUsername, videosMax, outputFile } = options;
-
-  if (!videosUsername) {
-    console.error('用法: tt-help videos <用户名> [最大视频数] [-o 输出路径]');
-    console.error('示例: tt-help videos bar.lar.lar.moeta 1000');
-    console.error('      tt-help videos username 50 -o videos.json');
-    console.error('');
-    console.error('选项: -o, --output <路径>  输出到文件（默认输出到 stdout）');
-    process.exit(1);
-  }
-
-  const { runGetUserVideos } = await import('../lib/get-user-videos-browser.mjs');
-
-  let browser;
-  try {
-    const { output, browser: b } = await runGetUserVideos({
-      username: videosUsername,
-      maxVideos: videosMax,
-      log: console.error,
-    });
-    browser = b;
-
-    const json = JSON.stringify(output, null, 2);
-    if (outputFile) {
-      writeFileSync(outputFile, json, 'utf-8');
-      console.error(`结果已写入: ${outputFile}`);
-    } else {
-      process.stdout.write(json + '\n');
-    }
-
-    console.error(`\n共 ${output.videos.length} 个视频, 用户: @${videosUsername}`);
-  } catch (err) {
-    console.error(`获取用户视频失败: ${err.message}`);
-    process.exit(1);
-  } finally {
-    if (browser) await browser.close().catch(() => {});
-  }
-}
+import { writeFileSync } from 'fs';
+
+export async function handleVideos(options) {
+  const { videosUsername, videosMax, outputFile } = options;
+
+  if (!videosUsername) {
+    console.error('用法: tt-help videos <用户名> [最大视频数] [-o 输出路径]');
+    console.error('示例: tt-help videos bar.lar.lar.moeta 1000');
+    console.error('      tt-help videos username 50 -o videos.json');
+    console.error('');
+    console.error('选项: -o, --output <路径>  输出到文件（默认输出到 stdout）');
+    process.exit(1);
+  }
+
+  const { runGetUserVideos } = await import('../videos/core.mjs');
+
+  let browser;
+  try {
+    const { output, browser: b } = await runGetUserVideos({
+      username: videosUsername,
+      maxVideos: videosMax,
+      log: console.error,
+    });
+    browser = b;
+
+    const json = JSON.stringify(output, null, 2);
+    if (outputFile) {
+      writeFileSync(outputFile, json, 'utf-8');
+      console.error(`结果已写入: ${outputFile}`);
+    } else {
+      process.stdout.write(json + '\n');
+    }
+
+    console.error(`\n共 ${output.videos.length} 个视频, 用户: @${videosUsername}`);
+  } catch (err) {
+    console.error(`获取用户视频失败: ${err.message}`);
+    process.exit(1);
+  } finally {
+    if (browser) await browser.close().catch(() => {});
+  }
+}

package/src/cli/watch.js

@@ -1,28 +1,28 @@
-import { writeFileSync, existsSync } from 'fs';
-import { startWatchServer, openBrowser } from '../watch/server.mjs';
-
-export async function handleWatch(options) {
-  const { outputFile, watchPort } = options;
-
-  if (!outputFile) {
-    console.error('用法: tt-help watch -o <数据文件> [-p 端口]');
-    console.error('示例: tt-help watch -o data.json');
-    console.error('      tt-help watch -o data.json -p 8080');
-    process.exit(1);
-  }
-
-  if (!existsSync(outputFile)) {
-    console.error(`文件不存在: ${outputFile}`);
-    process.exit(1);
-  }
-
-  const { server, port } = await startWatchServer(outputFile, watchPort);
-  openBrowser(port);
-
-  process.once('SIGINT', () => {
-    server.close();
-    process.exit(0);
-  });
-
-  console.error('按 Ctrl+C 停止监控服务');
-}
+import { writeFileSync, existsSync } from 'fs';
+import { startWatchServer, openBrowser } from '../watch/server.mjs';
+
+export async function handleWatch(options) {
+  const { outputFile, watchPort } = options;
+
+  if (!outputFile) {
+    console.error('用法: tt-help watch -o <数据文件> [-p 端口]');
+    console.error('示例: tt-help watch -o data.json');
+    console.error('      tt-help watch -o data.json -p 8080');
+    process.exit(1);
+  }
+
+  if (!existsSync(outputFile)) {
+    console.error(`文件不存在: ${outputFile}`);
+    process.exit(1);
+  }
+
+  const { server, port } = await startWatchServer(outputFile, watchPort);
+  openBrowser(port);
+
+  process.once('SIGINT', () => {
+    server.close();
+    process.exit(0);
+  });
+
+  console.error('按 Ctrl+C 停止监控服务');
+}