30 lines
1019 B
JavaScript
30 lines
1019 B
JavaScript
// 检查百家号页面状态：登录态、页面文字、可用的数据入口
|
|
import { connect, attachPage, evaluate, saveJson } from './lib.mjs';
|
|
|
|
const cdp = await connect();
|
|
const hit = await attachPage(cdp, 'baijiahao.baidu.com');
|
|
if (!hit) { console.log('NO_BAIJIAHAO_TAB'); process.exit(1); }
|
|
|
|
const { sessionId } = hit;
|
|
await cdp.send('Runtime.enable', {}, sessionId).catch(() => {});
|
|
|
|
const info = await evaluate(cdp, sessionId, `(() => {
|
|
const t = document.body ? document.body.innerText : '';
|
|
return {
|
|
url: location.href,
|
|
title: document.title,
|
|
textLen: t.length,
|
|
textSample: t.slice(0, 3000),
|
|
links: Array.from(document.querySelectorAll('a')).slice(0, 60).map(a => ({
|
|
text: (a.innerText || '').trim().slice(0, 40),
|
|
href: a.href
|
|
})).filter(x => x.href)
|
|
};
|
|
})()`);
|
|
|
|
saveJson('out/page-info.json', info);
|
|
console.log('URL=' + info.url.replace(/[^\x20-\x7E]/g, '?'));
|
|
console.log('TEXTLEN=' + info.textLen);
|
|
console.log('SAVED out/page-info.json');
|
|
process.exit(0);
|