feat: login-tools.js — per-source login check & auto-login

- check <source>: open checkUrl, assert loggedInSelector (exit 3 when
  logged out); login <source>: recipe-driven multi-step form flow with
  $USER/$PASS substitution from NH_LOGIN_* env, tab-relocation hardening
  against concurrent harvests, post-login verification
- recipes live in logins/<source>.json; ships digitimes-feed.json
- honors CDP_PORT_OVERRIDE / CDP_PROFILE_OVERRIDE (per-source instances)

skills/news-harvester/logins/digitimes-feed.json

+{
+  "checkUrl": "https://www.digitimes.com/",
+  "loggedInSelector": "a[href*=logout], a[href*=signout]",
+  "steps": [
+    {
+      "url": "https://www.digitimes.com/register/signin.php",
+      "fill": { "input[name=signInEMail]": "$USER" },
+      "submit": "form button[type=submit], form input[type=submit]"
+    },
+    {
+      "marker": "input[type=password]",
+      "fill": { "input[type=password]": "$PASS" },
+      "submit": "form button[type=submit], form input[type=submit]",
+      "waitMs": 8000
+    }
+  ]
+}

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

+#!/usr/bin/env node
+/**
+ * login-tools.js — 采集前登录预检/自动登录(按站点配方驱动)
+ *
+ * 用法:
+ *   node scripts/login-tools.js check <source>   检查登录态;已登录 exit 0,未登录 exit 3
+ *   node scripts/login-tools.js login  <source>  自动登录;凭据来自环境变量
+ *                                                NH_LOGIN_USER / NH_LOGIN_PASS
+ *
+ * 配方:logins/<source>.json
+ *   {
+ *     "checkUrl": "https://example.com/",
+ *     "loggedInSelector": "a[href*=logout]",
+ *     "steps": [
+ *       {"url": "https://example.com/signin", "fill": {"#email": "$USER"},
+ *        "submit": "form button[type=submit]"},
+ *       {"fill": {"input[type=password]": "$PASS"},
+ *        "submit": "form button[type=submit]", "waitMs": 8000}
+ *     ]
+ *   }
+ *   fill 值中的 $USER / $PASS 会被凭据替换;多步登录按序执行。
+ *
+ * 浏览器实例:尊重 CDP_PORT_OVERRIDE / CDP_PROFILE_OVERRIDE(与 ensure-chromium 一致),
+ * 未设置时用 config.chromium.cdpPort(默认 9222)。
+ * 退出码:0 成功;1 环境/配方错误;2 浏览器不可用;3 未登录;4 无配方;5 登录失败。
+ */
+
+const fs = require('fs');
+const http = require('http');
+const path = require('path');
+const WebSocket = require(path.join(__dirname, '..', 'node_modules', 'ws'));
+
+const ROOT = path.join(__dirname, '..');
+const LOGIN_USER = process.env.NH_LOGIN_USER || '';
+const LOGIN_PASS = process.env.NH_LOGIN_PASS || '';
+
+function cdpPort() {
+  if (process.env.CDP_PORT_OVERRIDE) return parseInt(process.env.CDP_PORT_OVERRIDE, 10);
+  try {
+    const cfg = JSON.parse(fs.readFileSync(path.join(ROOT, 'config.json'), 'utf8'));
+    if (cfg.chromium && cfg.chromium.cdpPort) return parseInt(cfg.chromium.cdpPort, 10);
+  } catch (e) { /* default */ }
+  return 9222;
+}
+
+function loadRecipe(source) {
+  const p = path.join(ROOT, 'logins', `${source}.json`);
+  if (!fs.existsSync(p)) return null;
+  return JSON.parse(fs.readFileSync(p, 'utf8'));
+}
+
+const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
+
+async function api(pathname, method = 'GET') {
+  const res = await fetch(`http://127.0.0.1:${cdpPort()}${pathname}`, { method });
+  return res.json();
+}
+
+async function findTab(pred) {
+  const tabs = await api('/json');
+  return tabs.find((t) => t.type === 'page' && pred(t));
+}
+
+function connect(tab) {
+  return new Promise((resolve, reject) => {
+    const ws = new WebSocket(tab.webSocketDebuggerUrl, { perMessageDeflate: false });
+    let id = 0;
+    const pending = new Map();
+    ws.on('message', (raw) => {
+      const m = JSON.parse(raw);
+      if (m.id && pending.has(m.id)) { pending.get(m.id)(m); pending.delete(m.id); }
+    });
+    ws.on('error', reject);
+    ws.on('open', () => {
+      const send = (method, params = {}) => new Promise((res, rej) => {
+        const i = ++id;
+        const timer = setTimeout(() => rej(new Error('CDP 调用超时: ' + method)), 20000);
+        pending.set(i, (m) => { clearTimeout(timer); res(m); });
+        ws.send(JSON.stringify({ id: i, method, params }));
+      });
+      resolve({ ws, send });
+    });
+  });
+}
+
+async function evalIn(tab, expression) {
+  const { ws, send } = await connect(tab);
+  try {
+    await send('Runtime.enable');
+    const r = await send('Runtime.evaluate', { expression, returnByValue: true });
+    if (r.result && r.result.exceptionDetails) {
+      throw new Error('页面执行异常: ' + (r.result.exceptionDetails.text || ''));
+    }
+    return r.result ? r.result.result && r.result.result.value : undefined;
+  } finally {
+    ws.close();
+  }
+}
+
+async function browserAlive() {
+  return new Promise((resolve) => {
+    const req = http.get(`http://127.0.0.1:${cdpPort()}/json/version`, (res) => {
+      res.resume();
+      resolve(true);
+    });
+    req.on('error', () => resolve(false));
+    req.setTimeout(1500, () => { req.destroy(); resolve(false); });
+  });
+}
+
+function substitute(value) {
+  return String(value).replace(/\$USER/g, LOGIN_USER).replace(/\$PASS/g, LOGIN_PASS);
+}
+
+async function main() {
+  const [command, source] = process.argv.slice(2);
+  if (!command || !source) {
+    console.error('用法: node scripts/login-tools.js check|login <source>');
+    process.exit(1);
+  }
+  const recipe = loadRecipe(source);
+  if (!recipe) {
+    console.error(`无登录配方: logins/${source}.json(该源无需登录或未配置)`);
+    process.exit(4);
+  }
+  if (command === 'login' && (!LOGIN_USER || !LOGIN_PASS)) {
+    console.error('缺少凭据: 需要环境变量 NH_LOGIN_USER / NH_LOGIN_PASS');
+    process.exit(1);
+  }
+  if (!(await browserAlive())) {
+    console.error(`浏览器实例不可用(端口 ${cdpPort()}),请先运行 ensure-chromium.js`);
+    process.exit(2);
+  }
+
+  // ── check:打开检查页,看登录态选择器 ──
+  if (command === 'check') {
+    await api(`/json/new?${encodeURIComponent(recipe.checkUrl)}`, 'PUT').catch(() => {});
+    await sleep(5000);
+    const tab = await findTab((t) => t.url.includes(new URL(recipe.checkUrl).host));
+    if (!tab) { console.error('检查页标签丢失(可能被并发采集关闭)'); process.exit(3); }
+    const logged = await evalIn(tab,
+      `!!document.querySelector(${JSON.stringify(recipe.loggedInSelector)})`);
+    console.log(logged ? `已登录: ${source}` : `未登录: ${source}`);
+    process.exit(logged ? 0 : 3);
+  }
+
+  // ── login:按配方多步执行 ──
+  if (command === 'login') {
+    for (const step of recipe.steps) {
+      if (step.url) {
+        await api(`/json/new?${encodeURIComponent(step.url)}`, 'PUT').catch(() => {});
+        await sleep(5000);
+      }
+      // 重新定位自己的标签(并发采集可能关掉闲置标签)
+      const marker = step.url ? new URL(step.url).pathname : (step.marker || 'input');
+      let tab = null;
+      for (let i = 0; i < 4 && !tab; i++) {
+        tab = await findTab((t) => (step.url ? t.url.includes(marker) : true));
+        if (!tab) await sleep(3000);
+      }
+      if (!tab) { console.error('登录标签丢失,中止'); process.exit(5); }
+      const fills = Object.entries(step.fill || {}).map(([sel, val]) =>
+        `(() => { const el = document.querySelector(${JSON.stringify(sel)}); if (!el) return 'MISSING:${sel}'; el.focus(); el.value = ${JSON.stringify(substitute(val))}; el.dispatchEvent(new Event('input', {bubbles:true})); el.dispatchEvent(new Event('change', {bubbles:true})); return 'OK'; })()`).join(',\n');
+      const result = await evalIn(tab, `JSON.stringify([${fills}])`);
+      const results = JSON.parse(result || '[]');
+      if (results.some((r) => String(r).startsWith('MISSING'))) {
+        console.error('表单字段未找到: ' + results.join('; '));
+        process.exit(5);
+      }
+      if (step.submit) {
+        await evalIn(tab, `(() => { const b = document.querySelector(${JSON.stringify(step.submit)}); const f = b ? b.closest('form') : null; if (b) b.click(); else if (f) f.requestSubmit(); return 'OK'; })()`);
+      }
+      await sleep(step.waitMs || 6000);
+    }
+    // 验证
+    await api(`/json/new?${encodeURIComponent(recipe.checkUrl)}`, 'PUT').catch(() => {});
+    await sleep(5000);
+    const tab = await findTab((t) => t.url.includes(new URL(recipe.checkUrl).host));
+    const logged = tab && await evalIn(tab,
+      `!!document.querySelector(${JSON.stringify(recipe.loggedInSelector)})`);
+    console.log(logged ? `✅ 自动登录成功: ${source}` : `❌ 登录后未检测到登录态: ${source}`);
+    process.exit(logged ? 0 : 5);
+  }
+
+  console.error('未知命令: ' + command);
+  process.exit(1);
+}
+
+main().catch((e) => { console.error('ERR', e.message); process.exit(1); });