tt-help-cli-ycl 1.3.11 → 1.3.13

package/src/cli/config.js

@@ -0,0 +1,116 @@
+import { HELP_TEXT, configPath, saveBrowser, saveUserId, getConfigText } from '../lib/constants.js';
+import { readFileSync, writeFileSync, existsSync } from 'fs';
+import { fileURLToPath } from 'url';
+import { dirname, join } from 'path';
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+const pkgPath = join(__dirname, '..', '..', 'package.json');
+const { version } = JSON.parse(readFileSync(pkgPath, 'utf-8'));
+
+function showConfig(urls, outputFile) {
+  const configLines = getConfigText();
+  if (urls.length > 0) {
+    configLines.push(`\n  URL数量: ${urls.length}`);
+  }
+  if (outputFile) {
+    configLines.push(`  输出文件: ${outputFile}`);
+  }
+  console.error(configLines.join('\n'));
+}
+
+function showUsage() {
+  console.error(HELP_TEXT.join('\n'));
+  process.exit(0);
+}
+
+function handleConfig(action, key, value) {
+  switch (action) {
+    case 'show': {
+      const configLines = getConfigText();
+      console.error(configLines.join('\n'));
+      break;
+    }
+
+    case 'set': {
+      if (!key) {
+        console.error('用法: tt-help config set <key> <value>');
+        console.error('  可用 key: proxy, server, browser, userId');
+        return;
+      }
+
+      switch (key) {
+        case 'proxy':
+          if (!value) {
+            console.error('请提供 proxy 的值');
+            console.error('用法: tt-help config set proxy <代理地址>');
+            return;
+          }
+          saveProxy(value);
+          console.error(`代理已更新: ${value}`);
+          break;
+
+        case 'server':
+          if (!value) {
+            console.error('请提供 server 的值');
+            return;
+          }
+          saveServer(value);
+          console.error(`服务器已更新: ${value}`);
+          break;
+
+        case 'browser':
+          if (!value) {
+            console.error('请提供 browser 的值');
+            console.error('用法: tt-help config set browser <浏览器路径>');
+            console.error('  或: tt-help config set-browser <浏览器路径>');
+            return;
+          }
+          saveBrowser(value);
+          console.error(`浏览器路径已更新: ${value}`);
+          break;
+
+        case 'userId':
+          if (!value) {
+            console.error('请提供 userId 的值');
+            return;
+          }
+          saveUserId(value);
+          console.error(`用户号已更新: ${value}`);
+          break;
+
+        default:
+          console.error(`未知配置项: ${key}`);
+          console.error('  可用 key: proxy, server, browser, userId');
+      }
+      break;
+    }
+
+    case 'reset': {
+      if (existsSync(configPath)) {
+        writeFileSync(configPath, '{}', 'utf-8');
+        console.error('配置已重置为默认');
+      } else {
+        console.error('配置文件不存在或已是默认状态');
+      }
+      break;
+    }
+
+    default:
+      console.error(`未知配置命令: ${action}`);
+      console.error('用法: tt-help config [show|set|reset]');
+  }
+}
+
+function saveProxy(newProxy) {
+  const cfg = existsSync(configPath) ? JSON.parse(readFileSync(configPath, 'utf-8')) : {};
+  cfg.proxy = newProxy;
+  writeFileSync(configPath, JSON.stringify(cfg, null, 2), 'utf-8');
+}
+
+function saveServer(newServer) {
+  const cfg = existsSync(configPath) ? JSON.parse(readFileSync(configPath, 'utf-8')) : {};
+  cfg.server = newServer;
+  writeFileSync(configPath, JSON.stringify(cfg, null, 2), 'utf-8');
+}
+
+export { handleConfig, showConfig, showUsage, version };

package/src/cli/explore-default.js

@@ -0,0 +1,83 @@
+import { parseFilter, applyFilter, formatFilterDescription } from '../lib/filter.js';
+import { writeFileSync } from 'fs';
+import { processUrlsWithProgress } from './progress.js';
+import { cleanError } from './utils.js';
+
+async function runExploreDefault(exploreCount, urls, proxyUrl, outputFile, outputFormat, pipeMode, filter) {
+  const allResults = [];
+
+  if (exploreCount > 0) {
+    try {
+      const { fetchExplore } = await import('../lib/explore-fetch.js');
+      const exploreResults = await fetchExplore(exploreCount);
+
+      console.log(`  获取到 ${exploreResults.length} 个视频\n`);
+      if (pipeMode) {
+        const videoUrls = exploreResults.map(r => r.url).filter(Boolean);
+        if (videoUrls.length > 0) {
+          await runScrapeDefault(videoUrls, proxyUrl, outputFile, outputFormat, filter);
+          return;
+        }
+      }
+      allResults.push(...exploreResults);
+    } catch (err) {
+      console.error(`  Explore 获取失败: ${cleanError(err.message)}\n`);
+      console.error(`  请确保代理 ${proxyUrl} 正常运行\n`);
+    }
+  }
+
+  if (urls.length > 0) {
+    const { processUrl } = await import('../lib/scrape.js');
+    await processUrlsWithProgress({
+      urls,
+      proxyUrl,
+      outputFile,
+      outputFormat,
+      filter,
+      processFn: (url, px) => processUrl(url, px),
+      label: '数据',
+      log: console.log,
+    });
+    return;
+  }
+
+  const { deduplicate, formatOutput } = await import('../lib/output.js');
+  const uniqueResults = deduplicate(allResults);
+  const filteredResults = applyFilter(uniqueResults, filter);
+
+  if (filteredResults.length === 0) {
+    console.log('\n未获取到数据');
+    if (outputFile) writeFileSync(outputFile, '[]', 'utf-8');
+    return;
+  }
+
+  const output = formatOutput(filteredResults, outputFormat);
+  if (outputFile) {
+    writeFileSync(outputFile, output, 'utf-8');
+    console.log(`\n结果已写入: ${outputFile}`);
+  } else {
+    console.log(output);
+  }
+
+  if (filter) {
+    console.log(`\n共 ${uniqueResults.length} 个数据，过滤后 ${filteredResults.length} 个（过滤条件: ${formatFilterDescription(filter)}）`);
+  } else {
+    console.log(`\n共 ${filteredResults.length} 个数据`);
+  }
+}
+
+async function runScrapeDefault(urls, proxyUrl, outputFile, outputFormat, filter) {
+  const { processUrl } = await import('../lib/scrape.js');
+  await processUrlsWithProgress({
+    urls,
+    proxyUrl,
+    outputFile,
+    outputFormat,
+    filter,
+    processFn: (url, px) => processUrl(url, px),
+    label: '用户的数据',
+    log: console.log,
+  });
+}
+
+export { runExploreDefault, runScrapeDefault };

package/src/cli/explore.js

@@ -1,181 +1,227 @@
-import { getOrCreatePage } from '../lib/browser/page.js';
-import { delay, getDelayConfig, setDelayConfig } from '../scraper/modules/page-helpers.mjs';
-import { userId as configuredUserId, saveUserId } from '../lib/constants.js';
-import { getMacOrUuid } from '../lib/mac-or-uuid.js';
-
-const MAX_RETRY_WAIT = 5 * 60 * 1000;
-
-async function withRetry(label, fn) {
-  let backoff = 1000;
-  while (true) {
-    try {
-      return await fn();
-    } catch (err) {
-      console.error(`[连接] ${label} 失败: ${err.message}，${backoff / 1000}秒后重试...`);
-      await new Promise(r => setTimeout(r, backoff));
-      if (backoff < MAX_RETRY_WAIT) backoff *= 2;
-    }
-  }
-}
-
-async function apiPost(url, body) {
-  return withRetry(`POST ${url}`, async () => {
-    const res = await fetch(url, {
-      method: 'POST',
-      headers: { 'Content-Type': 'application/json' },
-      body: JSON.stringify(body),
-    });
-    return res.json();
-  });
-}
-
-async function apiGet(url) {
-  return withRetry(`GET ${url}`, async () => {
-    const res = await fetch(url);
-    return res.json();
-  });
-}
-
-export async function handleExplore(options) {
-  const {
-    exploreUsernames, explorePreset, exploreMaxComments, exploreMaxGuess,
-    exploreEnableFollow, exploreMaxFollowing, exploreMaxFollowers,
-    exploreLocation, exploreMaxUsers, serverUrl,
-  } = options;
-
-  let userId = configuredUserId;
-  if (!userId) {
-    userId = await getMacOrUuid();
-    saveUserId(userId);
-    console.error(`[初始化] 未检测到本地用户编号，已生成并使用: ${userId}`);
-  }
-
-  setDelayConfig(explorePreset);
-
-  await apiGet(`${serverUrl}/api/stats`);
-
-  if (exploreUsernames && exploreUsernames.length > 0) {
-    const { added, skipped } = await apiPost(`${serverUrl}/api/users`, { usernames: exploreUsernames });
-    console.error(`种子用户: ${added} 个新增, ${skipped} 个已存在`);
-  }
-
-  console.error(`\n国家筛选: ${exploreLocation}`);
-  console.error(`评论: ${exploreMaxComments}, 猜你喜欢: ${exploreMaxGuess}`);
-  console.error(`关注/粉丝: ${exploreEnableFollow ? '启用' : '禁用'}`);
-  console.error(`服务器: ${serverUrl}（断开会自动重连）`);
-  if (exploreMaxUsers > 0) console.error(`上限: ${exploreMaxUsers} 个用户`);
-
-  const { ensureBrowserReady, processExplore } = await import('../scraper/explore-core.mjs');
-  const browser = await ensureBrowserReady();
-
-  const page = await getOrCreatePage(browser);
-
-  let processedCount = 0;
-  let errorCount = 0;
-  let consecutiveNetworkErrors = 0;
-
-  while (true) {
-    const job = await apiGet(`${serverUrl}/api/job?userId=${encodeURIComponent(userId)}`);
-    if (!job.hasJob) break;
-
-    const username = job.user.uniqueId;
-    processedCount++;
-
-    if (consecutiveNetworkErrors > 0) {
-      const waitTime = consecutiveNetworkErrors <= 2
-        ? 0
-        : consecutiveNetworkErrors <= 5
-          ? 30000
-          : 300000;
-      if (waitTime > 0) {
-        console.error(`  [网络] 连续 ${consecutiveNetworkErrors} 次网络异常，等待 ${waitTime / 1000}s 后重试...`);
-        await new Promise(r => setTimeout(r, waitTime));
-      }
-    }
-
-    console.error(`\n[${processedCount}] 探索 @${username}...`);
-
-    const { switchMax } = getDelayConfig();
-    await delay(switchMax, switchMax * 3);
-
-    const result = await processExplore(page, username, {
-      maxComments: exploreMaxComments,
-      maxGuess: exploreMaxGuess,
-      enableFollow: exploreEnableFollow,
-      maxFollowing: exploreMaxFollowing,
-      maxFollowers: exploreMaxFollowers,
-      location: exploreLocation,
-      browser,
-    }, console.error);
-
-    if (result.restricted) {
-      consecutiveNetworkErrors = 0;
-      await apiPost(`${serverUrl}/api/job/${username}`, { restricted: true, userInfo: result.userInfo || {} });
-      if (exploreMaxUsers > 0 && processedCount >= exploreMaxUsers) {
-        console.error(`\n已达上限 ${exploreMaxUsers} 个用户，停止处理`);
-        break;
-      }
-      continue;
-    }
-
-    if (result.error) {
-      consecutiveNetworkErrors++;
-      errorCount++;
-      await apiPost(`${serverUrl}/api/job/${username}`, { error: result.error });
-      const errorType = consecutiveNetworkErrors > 1 ? 'network' : 'other';
-      await withRetry('report error', () =>
-        apiPost(`${serverUrl}/api/error-report`, {
-          userId,
-          username,
-          errorType,
-          errorMessage: result.error,
-        })
-      ).catch(() => {});
-      if (exploreMaxUsers > 0 && processedCount >= exploreMaxUsers) {
-        console.error(`\n已达上限 ${exploreMaxUsers} 个用户，停止处理`);
-        break;
-      }
-      continue;
-    }
-
-    if (result.captchaDetected) {
-      await withRetry('report captcha', () =>
-        apiPost(`${serverUrl}/api/error-report`, {
-          userId,
-          username,
-          errorType: 'captcha',
-          errorMessage: '页面出现验证码',
-        })
-      ).catch(() => {});
-    }
-
-    consecutiveNetworkErrors = 0;
-
-    const payload = {
-      userInfo: result.userInfo || {},
-      discoveredVideoAuthors: result.discoveredVideoAuthors || [],
-      discoveredCommentAuthors: result.discoveredCommentAuthors || [],
-      discoveredGuessAuthors: result.discoveredGuessAuthors || [],
-      discoveredFollowing: result.discoveredFollowing || [],
-      discoveredFollowers: result.discoveredFollowers || [],
-      processed: result.processed,
-      hasFollowData: result.hasFollowData,
-      keepFollow: result.keepFollow,
-      locationCreated: result.locationCreated,
-      noVideo: result.noVideo,
-    };
-    await apiPost(`${serverUrl}/api/job/${username}`, payload);
-    console.error('  已提交');
-
-    if (exploreMaxUsers > 0 && processedCount >= exploreMaxUsers) {
-      console.error(`\n已达上限 ${exploreMaxUsers} 个用户，停止处理`);
-      break;
-    }
-  }
-
-  const stats = await apiGet(`${serverUrl}/api/stats`);
-  console.error(`\n完成: ${processedCount} 个用户处理, ${errorCount} 个出错`);
-  console.error(`  总用户: ${stats.totalUsers}, 已完成: ${stats.processedUsers}, 待处理: ${stats.pendingUsers}, 错误: ${stats.errorUsers}`);
-
-  await browser.close().catch(() => {});
-}
+import { getOrCreatePage, isBrowserClosedError, relaunchBrowser } from '../lib/browser/page.js';
+import { delay, getDelayConfig, setDelayConfig } from '../scraper/modules/page-helpers.js';
+import { userId as configuredUserId, saveUserId } from '../lib/constants.js';
+import { getMacOrUuid } from '../lib/mac-or-uuid.js';
+import { ensureBrowserReady as ensureBrowserReadyCDP } from '../lib/browser/cdp.js';
+import path from 'path';
+import os from 'os';
+
+const MAX_RETRY_WAIT = 5 * 60 * 1000;
+
+async function withRetry(label, fn) {
+  let backoff = 1000;
+  while (true) {
+    try {
+      return await fn();
+    } catch (err) {
+      console.error(`[连接] ${label} 失败: ${err.message}，${backoff / 1000}秒后重试...`);
+      await new Promise(r => setTimeout(r, backoff));
+      if (backoff < MAX_RETRY_WAIT) backoff *= 2;
+    }
+  }
+}
+
+async function apiPost(url, body) {
+  return withRetry(`POST ${url}`, async () => {
+    const res = await fetch(url, {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify(body),
+    });
+    return res.json();
+  });
+}
+
+async function apiGet(url) {
+  return withRetry(`GET ${url}`, async () => {
+    const res = await fetch(url);
+    return res.json();
+  });
+}
+
+export async function handleExplore(options) {
+  const {
+    exploreUsernames, explorePreset, exploreMaxComments, exploreMaxGuess,
+    exploreEnableFollow, exploreMaxFollowing, exploreMaxFollowers,
+    exploreLocation, exploreMaxUsers, serverUrl,
+    explorePort, exploreProfile, exploreUserId,
+  } = options;
+
+  let userId = exploreUserId || configuredUserId;
+  if (!userId) {
+    userId = await getMacOrUuid();
+    saveUserId(userId);
+    console.error(`[初始化] 未检测到本地用户编号，已生成并使用: ${userId}`);
+  }
+
+  setDelayConfig(explorePreset);
+
+  await apiGet(`${serverUrl}/api/stats`);
+
+  if (exploreUsernames && exploreUsernames.length > 0) {
+    const { added, skipped } = await apiPost(`${serverUrl}/api/users`, { usernames: exploreUsernames });
+    console.error(`种子用户: ${added} 个新增, ${skipped} 个已存在`);
+  }
+
+  console.error(`\n国家筛选: ${exploreLocation}`);
+  console.error(`评论: ${exploreMaxComments}, 猜你喜欢: ${exploreMaxGuess}`);
+  console.error(`关注/粉丝: ${exploreEnableFollow ? '启用' : '禁用'}`);
+  console.error(`服务器: ${serverUrl}（断开会自动重连）`);
+  if (exploreMaxUsers > 0) console.error(`上限: ${exploreMaxUsers} 个用户`);
+  console.error(`CDP 端口: ${explorePort || 9222}, 用户编号: ${userId}`);
+  if (exploreProfile) console.error(`浏览器配置: ${exploreProfile}`);
+
+  const cdpOptions = {};
+  if (explorePort) cdpOptions.port = explorePort;
+  if (exploreProfile) {
+    cdpOptions.userDataDir = path.join(os.homedir(), 'Library', 'Application Support', `Microsoft Edge For Testing_${exploreProfile}`);
+  }
+
+  let browser = await ensureBrowserReadyCDP(cdpOptions);
+  const { processExplore } = await import('../scraper/explore-core.js');
+
+  const page = await getOrCreatePage(browser);
+
+  let processedCount = 0;
+  let errorCount = 0;
+  let consecutiveNetworkErrors = 0;
+
+  while (true) {
+    const job = await apiGet(`${serverUrl}/api/job?userId=${encodeURIComponent(userId)}`);
+    if (!job.hasJob) break;
+
+    const username = job.user.uniqueId;
+    processedCount++;
+
+    if (consecutiveNetworkErrors > 0) {
+      const waitTime = consecutiveNetworkErrors <= 2
+        ? 0
+        : consecutiveNetworkErrors <= 5
+          ? 30000
+          : 300000;
+      if (waitTime > 0) {
+        console.error(`  [网络] 连续 ${consecutiveNetworkErrors} 次网络异常，等待 ${waitTime / 1000}s 后重试...`);
+        await new Promise(r => setTimeout(r, waitTime));
+      }
+    }
+
+    console.error(`\n[${processedCount}] 探索 @${username}...`);
+
+    const { switchMax } = getDelayConfig();
+    await delay(switchMax, switchMax * 3);
+
+    let result = await processExplore(page, username, {
+      maxComments: exploreMaxComments,
+      maxGuess: exploreMaxGuess,
+      enableFollow: exploreEnableFollow,
+      maxFollowing: exploreMaxFollowing,
+      maxFollowers: exploreMaxFollowers,
+      location: exploreLocation,
+      browser,
+    }, console.error);
+
+    // 浏览器关闭检测：processExplore 内部 catch 了异常，需要从 result.error 判断
+    if (result.error && isBrowserClosedError(new Error(result.error))) {
+      const newBrowser = await relaunchBrowser(cdpOptions, explorePort || 9222);
+      browser = newBrowser;
+      const newPage = await getOrCreatePage(browser);
+      Object.assign(page, newPage);
+      // 重试当前用户
+      result = await processExplore(page, username, {
+        maxComments: exploreMaxComments,
+        maxGuess: exploreMaxGuess,
+        enableFollow: exploreEnableFollow,
+        maxFollowing: exploreMaxFollowing,
+        maxFollowers: exploreMaxFollowers,
+        location: exploreLocation,
+        browser,
+      }, console.error);
+    }
+
+    if (result.restricted) {
+      consecutiveNetworkErrors = 0;
+      await apiPost(`${serverUrl}/api/job/${username}`, { restricted: true, userInfo: result.userInfo || {} });
+      if (exploreMaxUsers > 0 && processedCount >= exploreMaxUsers) {
+        console.error(`\n已达上限 ${exploreMaxUsers} 个用户，停止处理`);
+        break;
+      }
+      continue;
+    }
+
+    if (result.error) {
+      consecutiveNetworkErrors++;
+      errorCount++;
+      await apiPost(`${serverUrl}/api/job/${username}`, { error: result.error });
+      const errorType = consecutiveNetworkErrors > 1 ? 'network' : 'other';
+      await withRetry('report error', () =>
+        apiPost(`${serverUrl}/api/error-report`, {
+          userId,
+          username,
+          errorType,
+          errorMessage: result.error,
+          stage: 'process',
+          errorStack: result.errorStack || '',
+        })
+      ).catch(() => {});
+      if (exploreMaxUsers > 0 && processedCount >= exploreMaxUsers) {
+        console.error(`\n已达上限 ${exploreMaxUsers} 个用户，停止处理`);
+        break;
+      }
+      continue;
+    }
+
+    if (result.captchaDetected) {
+      await withRetry('report captcha', () =>
+        apiPost(`${serverUrl}/api/error-report`, {
+          userId,
+          username,
+          errorType: 'captcha',
+          errorMessage: result.captchaMessage || '页面出现验证码',
+          stage: result.captchaStage || 'video-page',
+          errorStack: '',
+        })
+      ).catch(() => {});
+    }
+
+    consecutiveNetworkErrors = 0;
+
+    const guessedLocation = result.locationCreated || null;
+
+    const payload = {
+      userInfo: result.userInfo || {},
+      discoveredVideoAuthors: (result.discoveredVideoAuthors || []).map(item =>
+        typeof item === 'object' ? { ...item, guessedLocation } : item
+      ),
+      discoveredCommentAuthors: (result.discoveredCommentAuthors || []).map(author => ({ author, guessedLocation })),
+      discoveredGuessAuthors: (result.discoveredGuessAuthors || []).map(author => ({ author, guessedLocation })),
+      discoveredFollowing: (result.discoveredFollowing || []).map(f => ({
+        handle: Array.isArray(f) ? f[0] : f,
+        displayName: Array.isArray(f) ? f[1] : null,
+        guessedLocation,
+      })),
+      discoveredFollowers: (result.discoveredFollowers || []).map(f => ({
+        handle: Array.isArray(f) ? f[0] : f,
+        displayName: Array.isArray(f) ? f[1] : null,
+        guessedLocation,
+      })),
+      processed: result.processed,
+      hasFollowData: result.hasFollowData,
+      keepFollow: result.keepFollow,
+      locationCreated: result.locationCreated,
+      noVideo: result.noVideo,
+    };
+    await apiPost(`${serverUrl}/api/job/${username}`, payload);
+    console.error('  已提交');
+
+    if (exploreMaxUsers > 0 && processedCount >= exploreMaxUsers) {
+      console.error(`\n已达上限 ${exploreMaxUsers} 个用户，停止处理`);
+      break;
+    }
+  }
+
+  const stats = await apiGet(`${serverUrl}/api/stats`);
+  console.error(`\n完成: ${processedCount} 个用户处理, ${errorCount} 个出错`);
+  console.error(`  总用户: ${stats.totalUsers}, 已完成: ${stats.processedUsers}, 待处理: ${stats.pendingUsers}, 错误: ${stats.errorUsers}`);
+
+  await browser.close().catch(() => {});
+}