登录保温:检测型配方login防御 + 站点保温访问keepalive.js

- login-tools.js: 无 steps 的检测型配方(reuters/wsj)在 login 命令下不再抛
  'recipe.steps is not iterable',改为提示走面板人工登录(exit 6)
- keepalive.js: 用站点共享实例访问一次首页刷新反爬(DataDome)cookie,
  维持常驻访客画像;实例离线则拉起并在结束后回收(有采集锁时不回收);
  检测撞墙写状态文件(NH_KEEPALIVE_STATUS),撞墙 exit 3

skills/news-harvester/scripts/keepalive.js

+#!/usr/bin/env node
+/**
+ * keepalive.js - 站点保温访问:用该站点的共享浏览器实例访问一次首页,
+ * 让反爬服务(DataDome 等)的 JS 在真实 profile 里刷新 cookie,维持"常驻访客"画像,
+ * 降低凌晨采集时突然批量请求触发人机验证的概率。
+ *
+ * 行为:
+ *   1. 实例不在线 → 以站点 profile 无头拉起(ensure-chromium.js),结束后回收;
+ *      已在线(如正在采集)→ 只开一个标签访问,绝不回收实例。
+ *   2. 打开首页等待 NH_KEEPALIVE_WAIT_MS(默认 30s),检测是否撞墙并记录状态。
+ *
+ * env:
+ *   CDP_PORT_OVERRIDE     必填,站点共享实例端口
+ *   CDP_PROFILE_OVERRIDE  站点 profile 目录(拉起实例用)
+ *   NH_KEEPALIVE_URL      访问地址,默认 https://www.reuters.com/
+ *   NH_KEEPALIVE_WAIT_MS  页面停留时长,默认 30000
+ *   NH_KEEPALIVE_STATUS   可选,状态 JSON 写入路径(面板/诊断用)
+ *
+ * 退出码:0 正常;2 实例不可用;3 检测到反爬墙(保温失败,需要人工过一次验证)
+ */
+const { spawn, spawnSync } = require('child_process');
+const http = require('http');
+const path = require('path');
+const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
+
+const PORT = parseInt(process.env.CDP_PORT_OVERRIDE || '', 10);
+const URL_TO_VISIT = process.env.NH_KEEPALIVE_URL || 'https://www.reuters.com/';
+const WAIT_MS = parseInt(process.env.NH_KEEPALIVE_WAIT_MS || '30000', 10);
+const STATUS_PATH = process.env.NH_KEEPALIVE_STATUS || '';
+if (!PORT) { console.error('keepalive: 缺少 CDP_PORT_OVERRIDE'); process.exit(2); }
+
+const cdp = (p, method = 'GET') => new Promise((res, rej) => {
+  const r = http.request({ host: '127.0.0.1', port: PORT, path: p, method }, (resp) => {
+    let d = ''; resp.on('data', (c) => (d += c));
+    resp.on('end', () => { try { res(JSON.parse(d)); } catch { res(d); } });
+  });
+  r.on('error', rej); r.end();
+});
+
+async function instanceAlive() {
+  try { await cdp('/json/version'); return true; } catch { return false; }
+}
+
+async function main() {
+  const startedByUs = !(await instanceAlive());
+  if (startedByUs) {
+    const ensure = path.join(__dirname, 'ensure-chromium.js');
+    const r = spawnSync('node', [ensure], { stdio: 'inherit', env: process.env });
+    if (r.status !== 0 || !(await instanceAlive())) {
+      console.error('keepalive: 浏览器实例拉起失败');
+      process.exit(2);
+    }
+  }
+
+  let tabId = null, verdict = { wall: false };
+  try {
+    const tab = await cdp('/json/new', 'PUT');
+    tabId = tab.id;
+    const ws = new WebSocket((await cdp('/json')).find((t) => t.id === tabId).webSocketDebuggerUrl);
+    await new Promise((r, j) => { ws.onopen = r; ws.onerror = j; });
+    let seq = 0; const pending = new Map();
+    const send = (method, params = {}) => new Promise((res) => {
+      const s = ++seq; pending.set(s, res); ws.send(JSON.stringify({ id: s, method, params }));
+    });
+    ws.onmessage = (m) => {
+      const j = JSON.parse(m.data);
+      if (pending.has(j.id)) { pending.get(j.id)(j.result); pending.delete(j.id); }
+    };
+    // /json/new 不导航:显式 Page.navigate(Chromium 行为差异兜底)
+    await send('Page.enable');
+    await send('Page.navigate', { url: URL_TO_VISIT });
+    await sleep(WAIT_MS);
+    const r = await send('Runtime.evaluate', {
+      returnByValue: true,
+      expression: `JSON.stringify({
+        url: location.href,
+        ddIframe: !!document.querySelector('iframe[title*=datadome i], iframe[src*=datadome]'),
+        ddWidget: !!document.querySelector('#ddv1-captcha, [class*=ddcaptcha]'),
+        captchaWord: /unusual traffic|verify you|滑动|验证/i.test(document.body.innerText.slice(0, 4000)),
+        title: document.title.slice(0, 80)
+      })`,
+    });
+    verdict = JSON.parse((r && r.result && r.result.value) || '{}');
+    verdict.wall = !!(verdict.ddIframe || verdict.ddWidget || verdict.captchaWord);
+    ws.close();
+  } finally {
+    if (tabId) await cdp(`/json/close/${tabId}`).catch(() => {});
+  }
+
+  const line = `${new Date().toISOString()} visit=${URL_TO_VISIT} wall=${verdict.wall} url=${verdict.url || '?'} title=${verdict.title || '?'}`;
+  if (STATUS_PATH) {
+    require('fs').writeFileSync(STATUS_PATH, JSON.stringify({
+      ts: new Date().toISOString(), wall: verdict.wall, url: verdict.url, title: verdict.title,
+    }, null, 2));
+  }
+
+  if (startedByUs) {
+    // 我们拉起的实例才回收;期间若有采集任务锁定了该端口,留给对方收尾
+    const lockDir = process.env.NH_LOCK_DIR;
+    let busy = false;
+    if (lockDir) {
+      try {
+        busy = require('fs').readdirSync(lockDir).some((f) => f.startsWith(`port-${PORT}.`));
+      } catch { /* 目录不存在视为空闲 */ }
+    }
+    if (!busy) {
+      const quit = spawnSync('node', [path.join(__dirname, 'instance-quit.js')],
+        { stdio: 'inherit', env: process.env });
+      if (quit.status !== 0) console.error('keepalive: 实例回收失败(不影响保温效果)');
+    } else {
+      console.log('keepalive: 采集任务已接管实例,跳过回收');
+    }
+  }
+  console.log(verdict.wall ? `keepalive: ⚠️ 撞墙,需要人工过一次验证 — ${line}` : `keepalive: ✅ ${line}`);
+  process.exit(verdict.wall ? 3 : 0);
+}
+
+main().catch((e) => { console.error(`keepalive: ${e.message}`); process.exit(2); });

skills/news-harvester/scripts/login-tools.js

@@ -352,6 +352,12 @@ async function main() {
 
   // ── login:按配方多步执行(discover 成功后落入此处实登验证) ──
   if (command === 'login' || command === 'discover') {
+    // 检测型配方(只有 loggedInSelector/loggedOutSelector,无 steps):
+    // 只能检测登录态,不支持自动登录——如 reuters/wsj(DataDome 之下自动登录也不可行),人工登录兜底
+    if (!Array.isArray(recipe.steps) || recipe.steps.length === 0) {
+      console.error(`检测型配方(无自动登录步骤): logins/${source}.json——请在面板"站点登录→人工登录"完成,该配方仅用于登录态检测`);
+      process.exit(6);
+    }
     let ownTabId = null;  // 全程复用同一个标签(提交导航不换 target id)
     if (!await findTab((t) => t.url.includes(new URL(recipe.checkUrl).host))) {
       const opened = await openTab(recipe.checkUrl);