Dicken/dickendock
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

fix: KöPi - klickbar, Bilder via Proxy, Datum angezeigt

Browse Source
This commit is contained in:
Dicken
2026-07-08 23:47:05 +02:00
parent 1326958098
commit c544e14171
2 changed files with 51 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

48
backend/src/tools/koepi/routes.js
View File

@@ -98,23 +98,28 @@ async function scrapeOffersWithPuppeteer() {
const productImg = imgs[0]?.src || null;
const publisherImg = imgs[1]?.src || null;
// Text parsen: "neuKönig Pilsener bei Trinkgut"Aktuelle Angebote", Seite 6Mo. 6.7. - Sa. 11.7.2026..."
const storeMatch = text.match(/bei\s+(.+?)\s*[""„]/);
const titleMatch = text.match(/[""„](.+?)[""„]/);
const pageMatch = text.match(/Seite\s+(\d+)/);
const dateMatch = text.match(/(\w+\.\s+\d+\.\d+\.)\s*-\s*(\w+\.\s+\d+\.\d+\.\d+)/);
const validMatch = text.match(/(Noch \d+ Tage? gültig|Gültig bis \S+)/);
const isNew = text.startsWith('neu');
// Text parsen
const storeMatch = text.match(/bei\s+([^""„]+?)\s*[""„]/);
const titleMatch = text.match(/[""„]([^""„]+?)[""„]/);
const pageMatch = text.match(/Seite\s+(\d+)/);
const validMatch = text.match(/(Noch \d+ Tag[e]? g.ltig|G.ltig bis [\d\.]+)/);
const dateMatch = text.match(/(\d{1,2}\.\d{1,2}\.(?:\d{4})?)\s*-\s*(\w+\.?\s+\d{1,2}\.\d{1,2}\.\d{4})/);
const isNew = text.startsWith('neu');
// URL aus Link holen
const link = b.querySelector('a');
const url = link?.href || null;
return {
type: 'brochure',
store: storeMatch?.[1]?.trim() || null,
title: titleMatch?.[1]?.trim() || null,
page: pageMatch?.[1] ? parseInt(pageMatch[1]) : null,
dateRange: dateMatch ? `${dateMatch[1]} - ${dateMatch[2]}` : null,
validity: validMatch?.[1] || null,
type: 'brochure',
store: storeMatch?.[1]?.trim() || null,
title: titleMatch?.[1]?.trim() || null,
page: pageMatch?.[1] ? parseInt(pageMatch[1]) : null,
dateRange: dateMatch ? `${dateMatch[1]} - ${dateMatch[2]}` : null,
validity: validMatch?.[1] || null,
isNew,
productImage: productImg,
url,
productImage: productImg,
publisherLogo: publisherImg,
};
});
@@ -256,4 +261,19 @@ router.post('/clear-cache', authenticate, (req, res) => {
res.json({ ok: true });
});
// ── Bild-Proxy (umgeht CSP) ──────────────────────────────────────────────────
router.get('/img', authenticate, (req, res) => {
const url = req.query.url;
if (!url || !url.startsWith('https://')) return res.status(400).end();
const parsed = new URL(url);
const allowed = ['content-media.bonial.biz','publisher-media.bonial.biz','web-assets.kaufda.de'];
if (!allowed.some(h => parsed.hostname === h)) return res.status(403).end();
const reqImg = https.get(url, { headers: { 'User-Agent': 'Mozilla/5.0' } }, (imgRes) => {
res.setHeader('Content-Type', imgRes.headers['content-type'] || 'image/jpeg');
res.setHeader('Cache-Control', 'public, max-age=3600');
imgRes.pipe(res);
});
reqImg.on('error', () => res.status(502).end());
});
module.exports = router;