fix: wayback-save fire-and-forget + CDX polling statt Body-Wait
This commit is contained in:
@@ -259,7 +259,7 @@ router.get('/wayback', authenticate, async (req, res) => {
|
||||
});
|
||||
|
||||
// ── POST /api/tools/paywallkiller/wayback-save ────────────────────────────
|
||||
// Erstellt einen neuen Wayback-Snapshot und wartet auf das Ergebnis
|
||||
// Schießt Save-Request ab (fire-and-forget), pollt dann CDX auf Ergebnis
|
||||
router.post('/wayback-save', authenticate, async (req, res) => {
|
||||
const { url } = req.body;
|
||||
if (!url || typeof url !== 'string') return res.status(400).json({ error: 'url fehlt' });
|
||||
@@ -267,33 +267,67 @@ router.post('/wayback-save', authenticate, async (req, res) => {
|
||||
let targetUrl = url.trim();
|
||||
if (!targetUrl.startsWith('http')) targetUrl = 'https://' + targetUrl;
|
||||
|
||||
// Save API aufrufen: GET https://web.archive.org/save/URL
|
||||
// Antwort: 302 Redirect auf https://web.archive.org/web/TIMESTAMP/URL
|
||||
try {
|
||||
const saveUrl = `https://web.archive.org/save/${targetUrl}`;
|
||||
const saveResult = await fetchPage(saveUrl); // fetchPage folgt Redirects
|
||||
const https = require('https');
|
||||
|
||||
// Nach Redirect landet man auf web.archive.org/web/TIMESTAMP/URL
|
||||
const final = saveResult.finalUrl || '';
|
||||
const match = final.match(/web\.archive\.org\/web\/(\d{14})\//);
|
||||
if (match) {
|
||||
return res.json({ success: true, waybackUrl: final, timestamp: match[1] });
|
||||
// ── Hilfsfunktion: CDX prüfen ──────────────────────────────────────────
|
||||
async function checkCdx() {
|
||||
const cdxUrl = `https://web.archive.org/cdx/search/cdx?url=${encodeURIComponent(targetUrl)}&output=json&limit=1&fl=timestamp&filter=statuscode:200&fastLatest=true`;
|
||||
try {
|
||||
const r = await fetchPage(cdxUrl);
|
||||
if (r.statusCode === 200 && r.body) {
|
||||
const data = JSON.parse(r.body);
|
||||
if (Array.isArray(data) && data.length >= 2 && data[1]?.[0]) {
|
||||
return `https://web.archive.org/web/${data[1][0]}/${targetUrl}`;
|
||||
}
|
||||
}
|
||||
} catch {}
|
||||
return null;
|
||||
}
|
||||
|
||||
// ── Save-Request fire-and-forget (kein Warten auf Body) ───────────────
|
||||
function fireSaveRequest() {
|
||||
return new Promise((resolve) => {
|
||||
const saveUrl = new URL(`https://web.archive.org/save/${targetUrl}`);
|
||||
const req = https.get({
|
||||
hostname: saveUrl.hostname,
|
||||
path: saveUrl.pathname + saveUrl.search,
|
||||
headers: {
|
||||
'User-Agent': 'Mozilla/5.0 (compatible; WaybackSave/1.0)',
|
||||
'Accept': 'text/html',
|
||||
},
|
||||
timeout: 8000,
|
||||
}, (res) => {
|
||||
// Nur Headers lesen, Body sofort verwerfen
|
||||
const location = res.headers.location || '';
|
||||
res.resume(); // Body ignorieren
|
||||
resolve(location);
|
||||
});
|
||||
req.on('error', () => resolve(''));
|
||||
req.on('timeout', () => { req.destroy(); resolve(''); });
|
||||
});
|
||||
}
|
||||
|
||||
try {
|
||||
// Save abschießen
|
||||
const location = await fireSaveRequest();
|
||||
|
||||
// Direkt aus Location-Header auslesen wenn vorhanden
|
||||
const directMatch = location.match(/\/web\/(\d{14})\//);
|
||||
if (directMatch) {
|
||||
const waybackUrl = location.startsWith('http') ? location : `https://web.archive.org${location}`;
|
||||
return res.json({ success: true, waybackUrl, timestamp: directMatch[1] });
|
||||
}
|
||||
|
||||
// Fallback: CDX nochmal prüfen (Save kann asynchron sein)
|
||||
await new Promise(r => setTimeout(r, 3000));
|
||||
const cdxUrl = `https://web.archive.org/cdx/search/cdx?url=${encodeURIComponent(targetUrl)}&output=json&limit=1&fl=timestamp&filter=statuscode:200&fastLatest=true`;
|
||||
const cdxResult = await fetchPage(cdxUrl);
|
||||
if (cdxResult.statusCode === 200) {
|
||||
let data;
|
||||
try { data = JSON.parse(cdxResult.body); } catch { data = null; }
|
||||
if (Array.isArray(data) && data.length >= 2 && data[1]?.[0]) {
|
||||
const timestamp = data[1][0];
|
||||
return res.json({ success: true, waybackUrl: `https://web.archive.org/web/${timestamp}/${targetUrl}`, timestamp });
|
||||
// Polling: bis zu 6x alle 5s CDX abfragen (max 30s warten)
|
||||
for (let i = 0; i < 6; i++) {
|
||||
await new Promise(r => setTimeout(r, 5000));
|
||||
const found = await checkCdx();
|
||||
if (found) {
|
||||
return res.json({ success: true, waybackUrl: found });
|
||||
}
|
||||
}
|
||||
|
||||
return res.status(502).json({ error: 'Snapshot konnte nicht erstellt werden. Die Seite blockiert möglicherweise die Wayback Machine.' });
|
||||
return res.status(502).json({ error: 'Archivierung läuft noch oder wurde blockiert. Versuche es in einer Minute erneut.' });
|
||||
} catch (err) {
|
||||
return res.status(500).json({ error: 'Save fehlgeschlagen: ' + err.message });
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user