@web-auto/webauto 0.1.18 → 0.1.19

package/apps/webauto/entry/smart-reply-cli.mjs

@@ -0,0 +1,267 @@
+#!/usr/bin/env node
+/**
+ * smart-reply-cli.mjs
+ * 
+ * 智能回复 CLI 工具
+ * 
+ * 流程：
+ * 1. 从已爬取的评论数据中分析高频关键词
+ * 2. 展示给用户选择
+ * 3. 用户指定回复中心意思
+ * 4. dryrun 模式验证
+ * 5. 真实回复（可选）
+ */
+
+import fs from 'fs';
+import path from 'path';
+import os from 'os';
+import { createInterface } from 'readline';
+import { spawn } from 'child_process';
+
+// 统一路径解析（与项目一致）
+function resolveWebautoRoot() {
+  const explicit = String(process.env.WEBAUTO_HOME || '').trim();
+  if (explicit) return explicit;
+  
+  const legacy = String(process.env.WEBAUTO_ROOT || process.env.WEBAUTO_PORTABLE_ROOT || '').trim();
+  if (legacy) {
+    const base = path.basename(legacy).toLowerCase();
+    return (base === '.webauto' || base === 'webauto') ? legacy : path.join(legacy, '.webauto');
+  }
+  
+  if (process.platform === 'win32') {
+    const dExists = fs.existsSync('D:\\');
+    return dExists ? 'D:\\webauto' : path.join(os.homedir(), '.webauto');
+  }
+  return path.join(os.homedir(), '.webauto');
+}
+
+const WEBAUTO_ROOT = resolveWebautoRoot();
+const DOWNLOAD_ROOT = path.join(WEBAUTO_ROOT, 'download');
+
+// 简单的中文分词
+function tokenize(text) {
+  return text
+    .replace(/[，。！？、；：""''（）【】《》\s]+/g, ' ')
+    .split(/\s+/)
+    .filter(t => t.length >= 2 && t.length <= 10);
+}
+
+// 统计词频
+function countKeywords(comments) {
+  const wordFreq = new Map();
+  
+  for (const comment of comments) {
+    const tokens = tokenize(comment.content || comment.text || '');
+    for (const token of tokens) {
+      wordFreq.set(token, (wordFreq.get(token) || 0) + 1);
+    }
+  }
+  
+  return Array.from(wordFreq.entries())
+    .sort((a, b) => b[1] - a[1])
+    .map(([word, count]) => ({ word, count }));
+}
+
+// 加载评论数据
+function loadComments(keyword) {
+  const comments = [];
+  
+  const searchDirs = [DOWNLOAD_ROOT];
+  
+  for (const searchDir of searchDirs) {
+    if (!fs.existsSync(searchDir)) continue;
+    
+    const findJsonlFiles = (dir) => {
+      const files = [];
+      try {
+        const entries = fs.readdirSync(dir, { withFileTypes: true });
+        for (const entry of entries) {
+          if (entry.name === 'node_modules') continue;
+          const full = path.join(dir, entry.name);
+          if (entry.isDirectory()) {
+            files.push(...findJsonlFiles(full));
+          } else if (entry.name === 'comments.jsonl') {
+            files.push(full);
+          }
+        }
+      } catch {}
+      return files;
+    };
+    
+    const files = findJsonlFiles(searchDir);
+    
+    for (const file of files) {
+      if (keyword && !file.includes(keyword)) continue;
+      
+      try {
+        const lines = fs.readFileSync(file, 'utf-8').split('\n').filter(Boolean);
+        for (const line of lines) {
+          try {
+            const comment = JSON.parse(line);
+            comments.push({ ...comment, file });
+          } catch {}
+        }
+      } catch {}
+    }
+  }
+  
+  return comments;
+}
+
+// CLI 交互
+async function main() {
+  const rl = createInterface({
+    input: process.stdin,
+    output: process.stdout,
+  });
+  
+  const question = (prompt) => new Promise((resolve) => {
+    rl.question(prompt, resolve);
+  });
+  
+  console.log('========================================');
+  console.log('  小红书智能回复工具');
+  console.log('========================================\n');
+  console.log(`数据目录: ${WEBAUTO_ROOT}\n`);
+  
+  // 1. 输入关键词
+  const keyword = await question('请输入已爬取的关键词目录（如 deepseek）: ');
+  
+  console.log('\n正在加载评论数据...');
+  const comments = loadComments(keyword.trim());
+  
+  if (comments.length === 0) {
+    console.log('未找到评论数据，请先运行爬取任务');
+    rl.close();
+    return;
+  }
+  
+  console.log(`找到 ${comments.length} 条评论\n`);
+  
+  // 2. 分析高频词
+  console.log('正在分析评论关键词...\n');
+  const keywords = countKeywords(comments).slice(0, 20);
+  
+  console.log('高频关键词 TOP 20:');
+  console.log('-'.repeat(40));
+  keywords.forEach((item, i) => {
+    console.log(`${(i + 1).toString().padStart(2)}. ${item.word.padEnd(15)} (${item.count}次)`);
+  });
+  console.log('-'.repeat(40));
+  
+  // 3. 选择关键词
+  const selectedKeywords = await question('\n请选择要匹配的关键词（多个用逗号分隔，如: 牛逼,厉害）: ');
+  const matchKeywords = selectedKeywords.split(/[,，]/).map(k => k.trim()).filter(Boolean);
+  
+  if (matchKeywords.length === 0) {
+    console.log('未选择关键词');
+    rl.close();
+    return;
+  }
+  
+  // 4. 筛选命中评论
+  const matchedComments = comments.filter(c => {
+    const text = (c.content || c.text || '').toLowerCase();
+    return matchKeywords.some(k => text.includes(k.toLowerCase()));
+  });
+  
+  console.log(`\n命中评论数: ${matchedComments.length}`);
+  
+  if (matchedComments.length === 0) {
+    console.log('没有命中任何评论');
+    rl.close();
+    return;
+  }
+  
+  // 5. 展示部分命中评论
+  console.log('\n命中评论预览 (前5条):');
+  console.log('-'.repeat(60));
+  matchedComments.slice(0, 5).forEach((c, i) => {
+    const text = (c.content || c.text || '').slice(0, 50);
+    console.log(`${i + 1}. [${c.userName || '匿名'}] ${text}...`);
+  });
+  console.log('-'.repeat(60));
+  
+  // 6. 指定回复中心意思
+  const replyIntent = await question('\n请输入回复的中心意思（如: 感谢认可，告诉对方具体信息）: ');
+  
+  if (!replyIntent.trim()) {
+    console.log('未输入回复意图');
+    rl.close();
+    return;
+  }
+  
+  // 7. 选择模式
+  const mode = await question('\n选择模式: (1) dryrun预览 (2) 真实回复: ');
+  const dryRun = mode.trim() === '1' || mode.trim().toLowerCase() === 'dryrun';
+  
+  // 8. 确认
+  console.log('\n' + '='.repeat(50));
+  console.log('配置确认:');
+  console.log(`  关键词目录: ${keyword}`);
+  console.log(`  匹配关键词: ${matchKeywords.join(', ')}`);
+  console.log(`  命中评论数: ${matchedComments.length}`);
+  console.log(`  回复意图: ${replyIntent}`);
+  console.log(`  模式: ${dryRun ? 'dryrun（预览）' : '真实回复'}`);
+  console.log('='.repeat(50));
+  
+  const confirm = await question('\n确认执行？(y/n): ');
+  if (confirm.toLowerCase() !== 'y') {
+    console.log('已取消');
+    rl.close();
+    return;
+  }
+  
+  // 9. 输出配置文件
+  const config = {
+    keyword: keyword.trim(),
+    matchKeywords,
+    replyIntent: replyIntent.trim(),
+    dryRun,
+    comments: matchedComments.slice(0, dryRun ? 3 : undefined),
+    timestamp: new Date().toISOString(),
+  };
+  
+  const configPath = path.join(WEBAUTO_ROOT, 'smart-reply-config.json');
+  fs.writeFileSync(configPath, JSON.stringify(config, null, 2));
+  
+  console.log(`\n配置已保存到: ${configPath}`);
+  console.log('\n后续步骤:');
+  console.log('1. 确保 Unified API 运行中 (http://127.0.0.1:7701)');
+  console.log('2. 确保浏览器 session 已打开小红书页面');
+  console.log('3. 运行 SmartReplyBlock 进行实际回复');
+  
+  // 10. 测试 AI 生成（在关闭 rl 之前）
+  const testGenerate = await question('\n是否立即测试 AI 生成回复？(y/n): ');
+  if (testGenerate.toLowerCase() === 'y') {
+    console.log('\n正在生成测试回复...\n');
+    
+    const testComment = matchedComments[0];
+    const prompt = `你是一个小红书评论回复助手。请根据以下信息生成一条回复。
+
+## 帖子评论
+${testComment.content || testComment.text || ''}
+
+## 回复要求
+- 回复的中心意思：${replyIntent}
+- 回复风格：友好、自然、口语化
+- 字数限制：100字以内
+- 不要使用表情符号开头
+- 可以适当使用 1-2 个表情符号
+
+请直接输出回复内容，不要有任何解释或说明。`;
+
+    // 在关闭 rl 之前启动子进程
+    const child = spawn('iflow', ['-p', prompt], { stdio: 'inherit' });
+    
+    // 关闭 readline
+    rl.close();
+    
+    await new Promise(resolve => child.on('close', resolve));
+  } else {
+    rl.close();
+  }
+}
+
+main().catch(console.error);

package/apps/webauto/entry/weibo-unified.mjs

@@ -2,20 +2,71 @@
 import minimist from 'minimist';
 import { runWorkflowById } from '../../../dist/modules/workflow/src/runner.js';
 import { pathToFileURL } from 'node:url';
+import { assertProfileUsable } from './lib/profile-policy.mjs';
+import { syncWeiboAccountByProfile } from './lib/account-detect.mjs';
+import { cleanupIncompleteProfiles } from './lib/account-store.mjs';
+import { ensureSessionInitialized } from './lib/session-init.mjs';
 
 const WEIBO_HOME_URL = 'https://www.weibo.com';
-const DEFAULT_PROFILE = 'profile-0';
+const DEFAULT_PROFILE = '';
+
+function parseBoolean(value, fallback = false) {
+  if (value === undefined || value === null || value === '') return fallback;
+  if (typeof value === 'boolean') return value;
+  const text = String(value).trim().toLowerCase();
+  if (['1', 'true', 'yes', 'on'].includes(text)) return true;
+  if (['0', 'false', 'no', 'off'].includes(text)) return false;
+  return fallback;
+}
+
+function parseIntFlag(value, fallback, min = 1) {
+  const num = Number(value);
+  if (!Number.isFinite(num)) return fallback;
+  return Math.max(min, Math.floor(num));
+}
+
+async function ensureWeiboLoginValid(profileId) {
+  const state = await syncWeiboAccountByProfile(profileId);
+  const valid = state?.valid === true && Boolean(String(state?.accountId || '').trim());
+  if (!valid) {
+    const reason = String(state?.reason || state?.status || 'invalid').trim() || 'invalid';
+    throw new Error(`weibo account invalid, login gate blocked: ${profileId} (${reason})`);
+  }
+  return state;
+}
+
+function resolveWeiboWorkflow(argv = {}) {
+  const explicitWorkflow = String(argv.workflow || '').trim();
+  if (explicitWorkflow) return explicitWorkflow;
+  const taskType = String(argv['task-type'] || argv.taskType || 'search').trim().toLowerCase();
+  if (!taskType || taskType === 'search') return 'weibo-search-v1';
+  throw new Error(`unsupported weibo task-type: ${taskType} (currently only search)`);
+}
 
 async function runCommand(argv) {
-  const profile = String(argv.profile || DEFAULT_PROFILE).trim();
+  cleanupIncompleteProfiles();
+  const profile = assertProfileUsable(String(argv.profile || DEFAULT_PROFILE).trim());
   if (!profile) {
     throw new Error('Profile ID is required. Use --profile <id>');
   }
+  const initResult = await ensureSessionInitialized(profile, {
+    url: WEIBO_HOME_URL,
+    rootDir: process.cwd(),
+    timeoutMs: 60000,
+  });
+  if (!initResult?.ok) {
+    throw new Error(`weibo session init failed: ${initResult?.error || 'unknown_error'}`);
+  }
+  await ensureWeiboLoginValid(profile);
 
-  const workflowId = String(argv.workflow || '').trim();
+  const workflowId = resolveWeiboWorkflow(argv);
   const keyword = String(argv.keyword || '').trim();
   const targetCount = Number(argv['max-notes'] || argv.target || 50);
   const maxComments = Number(argv['max-comments'] || 0); // 0 means no limit
+  const maxPages = Number(argv['max-pages'] || 10);
+  const collectComments = parseBoolean(argv['collect-comments'], maxComments > 0);
+  const tabCount = parseIntFlag(argv['tab-count'], 2, 1);
+  const tabOpenDelayMs = parseIntFlag(argv['tab-open-delay'], 800, 0);
 
   if (!keyword && workflowId === 'weibo-search-v1') {
     throw new Error('Keyword is required for search tasks. Use --keyword <text>');
@@ -29,13 +80,13 @@ async function runCommand(argv) {
     env,
     targetCount,
     maxComments,
+    maxPages,
+    collectComments,
+    tabCount,
+    tabOpenDelayMs,
     // Add other common parameters as needed
   };
 
-  if (!workflowId) {
-    throw new Error('Workflow ID is required. e.g., --workflow weibo-search-v1');
-  }
-
   console.log(`[Weibo Unified] Running workflow: ${workflowId} with profile: ${profile}`);
   const result = await runWorkflowById(workflowId, initialContext);
 
@@ -53,7 +104,6 @@ async function main() {
     boolean: ['help'],
     alias: { k: 'keyword', t: 'target', h: 'help' },
     default: {
-      profile: DEFAULT_PROFILE,
       env: 'debug',
       target: 50,
     },
@@ -67,10 +117,15 @@ Commands:
   search          Perform a Weibo search task.
 
 Options:
-  --profile <id>      Camo profile ID to use (default: ${DEFAULT_PROFILE})
+  --profile <id>      Camo profile ID to use (required; must be pre-created)
+  --task-type <type>  当前仅支持 search
   --keyword <text>    Keyword to search for (required for search command)
   --target <n>        Target number of posts to collect (default: 50)
+  --max-pages <n>     Maximum search result pages to scan for links (default: 10)
   --max-comments <n>  Maximum comments to collect per post (default: 0, no limit)
+  --collect-comments <bool>  Whether to collect comments in content phase (default: auto by --max-comments)
+  --tab-count <n>     Tab count for round-robin detail collection (default: 2)
+  --tab-open-delay <ms> Delay after opening each extra tab (default: 800)
   --env <debug|prod>  Environment for data storage (default: debug)
   --help              Show this help message
 
@@ -104,11 +159,25 @@ if (isDirectExec) {
 }
 
 export async function runWeiboUnified(argv) {
-  const workflowId = String(argv.workflow || 'weibo-search-v1').trim();
+  cleanupIncompleteProfiles();
+  const workflowId = resolveWeiboWorkflow(argv);
   const keyword = String(argv.keyword || argv.k || '').trim();
-  const profile = String(argv.profile || DEFAULT_PROFILE).trim();
+  const profile = assertProfileUsable(String(argv.profile || DEFAULT_PROFILE).trim());
+  const initResult = await ensureSessionInitialized(profile, {
+    url: WEIBO_HOME_URL,
+    rootDir: process.cwd(),
+    timeoutMs: 60000,
+  });
+  if (!initResult?.ok) {
+    throw new Error(`weibo session init failed: ${initResult?.error || 'unknown_error'}`);
+  }
+  await ensureWeiboLoginValid(profile);
   const targetCount = Number(argv['max-notes'] || argv.target || argv['max-notes'] || 50);
   const maxComments = Number(argv['max-comments'] || 0);
+  const maxPages = Number(argv['max-pages'] || 10);
+  const collectComments = parseBoolean(argv['collect-comments'], maxComments > 0);
+  const tabCount = parseIntFlag(argv['tab-count'], 2, 1);
+  const tabOpenDelayMs = parseIntFlag(argv['tab-open-delay'], 800, 0);
   const env = String(argv.env || 'debug').trim();
 
   if (!keyword) {
@@ -121,6 +190,10 @@ export async function runWeiboUnified(argv) {
     env,
     targetCount,
     maxComments,
+    maxPages,
+    collectComments,
+    tabCount,
+    tabOpenDelayMs,
   };
 
   const result = await runWorkflowById(workflowId, initialContext);

package/apps/webauto/entry/xhs-orchestrate.mjs

@@ -4,7 +4,7 @@ import { runUnified } from './xhs-unified.mjs';
 
 async function main() {
   const argv = minimist(process.argv.slice(2));
-  const mode = String(argv.mode || 'phase1-phase2-unified').trim();
+  const mode = String(argv.mode || 'phase1-phase2-unified').trim().toLowerCase();
 
   if (mode === 'phase1-only') {
     console.log(JSON.stringify({ event: 'xhs.orchestrate.skip', mode, reason: 'phase1 is merged into runtime bootstrap' }));
@@ -13,6 +13,7 @@ async function main() {
 
   if (mode === 'phase1-phase2') {
     await runUnified(argv, {
+      stage: 'content',
       doComments: false,
       doLikes: false,
       doReply: false,
@@ -22,6 +23,47 @@ async function main() {
     return;
   }
 
+  if (mode === 'links-only') {
+    await runUnified(argv, {
+      stage: 'links',
+      doHomepage: false,
+      doImages: false,
+      doComments: false,
+      doLikes: false,
+      doReply: false,
+      doOcr: false,
+      persistComments: false,
+    });
+    return;
+  }
+
+  if (mode === 'content-only') {
+    await runUnified(argv, {
+      stage: 'content',
+      doLikes: false,
+      doReply: false,
+    });
+    return;
+  }
+
+  if (mode === 'like-only') {
+    await runUnified(argv, {
+      stage: 'like',
+      doLikes: true,
+      doReply: false,
+    });
+    return;
+  }
+
+  if (mode === 'reply-only') {
+    await runUnified(argv, {
+      stage: 'reply',
+      doLikes: false,
+      doReply: true,
+    });
+    return;
+  }
+
   if (mode === 'phase1-phase2-unified' || mode === 'unified-only') {
     await runUnified(argv);
     return;