AI/reader
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/jina-ai/reader synced 2025-08-15 02:15:56 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: block rough requests

Browse Source
This commit is contained in:
Yanlong Wang 2024-05-16 20:22:26 +08:00
parent 8ec8c1e718
commit ed9e9f43cf
No known key found for this signature in database
GPG Key ID: C0A623C0BADF9F37
1 changed files with 23 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
backend/functions/src/services/puppeteer.ts
View File

@ -246,10 +246,33 @@ function giveSnapshot() {
return r;
}
`));
preparations.push(page.setRequestInterception(true));
await Promise.all(preparations);
await page.goto('about:blank', { waitUntil: 'domcontentloaded' });
page.on('request', (req) => {
const requestUrl = req.url();
if (!requestUrl.startsWith("http:") && !requestUrl.startsWith("https:") && requestUrl !== 'about:blank') {
return req.abort('blockedbyclient', 1000);
}
const parsedUrl = new URL(requestUrl);
if (
parsedUrl.hostname === 'localhost' ||
parsedUrl.hostname.startsWith('127.')
) {
return req.abort('blockedbyclient', 1000);
}
const continueArgs = req.continueRequestOverrides
? [req.continueRequestOverrides(), 0] as const
: [];
return req.continue(continueArgs[0], continueArgs[1]);
});
await page.evaluateOnNewDocument(`
let aftershot = undefined;
const handlePageLoad = () => {