Compare commits
3 Commits
28d52be2c3
...
main
| Author | SHA1 | Date | |
|---|---|---|---|
| 52d64287ae | |||
| 5882540cf8 | |||
| 8cf101dee5 |
1
.gitignore
vendored
1
.gitignore
vendored
@@ -11,3 +11,4 @@ src/*.png
|
|||||||
debug/
|
debug/
|
||||||
|
|
||||||
.DS_Store
|
.DS_Store
|
||||||
|
|
||||||
|
|||||||
21
AGENTS.md
Normal file
21
AGENTS.md
Normal file
@@ -0,0 +1,21 @@
|
|||||||
|
# Engineering & Scraping Guidelines
|
||||||
|
|
||||||
|
## 1. Git Commit & Repository Hygiene
|
||||||
|
- **Conventional Commits:** Write clear, concise commit messages prefixed with standard types (e.g., `feat:`, `fix:`, `refactor:`, `docs:`).
|
||||||
|
- **Exclude Transient Artifacts:** Never stage or commit generated debug files (e.g., `src/*.html`, `src/*.png`, large logs) to version control. Always maintain `.gitignore` to keep git packs lean and prevent remote unpacker errors on Gitea/GitHub.
|
||||||
|
|
||||||
|
## 2. Web Scraping & DOM Inspection Strategy
|
||||||
|
- **Rendered DOM Dumping for Debugging:** When debugging extraction failures on dynamic or anti-bot protected pages:
|
||||||
|
- Capture the post-render HTML via `await page.content()` and save it to `src/hotel-debug.html`.
|
||||||
|
- Capture a synchronized screenshot (`src/hotel-debug.png`).
|
||||||
|
- Inspect the raw rendered DOM to verify if the page was served as standard SSR, an A/B test SPA variant, or intercepted by a challenge (e.g. AWS WAF).
|
||||||
|
- **Multi-Tier Extraction Hierarchy:** Always implement fallbacks in this order:
|
||||||
|
1. **Deep JSON-LD Parsing:** Traverse all `<script type="application/ld+json">` tags, including `@graph` arrays and varied accommodation schemas (`Hotel`, `Resort`, `Apartment`, `Hostel`, `LodgingBusiness`, etc.).
|
||||||
|
2. **DOM Selectors:** Fallback to standard test IDs, classes, and element hierarchy (`data-testid`, header titles, score components).
|
||||||
|
3. **Meta Tags:** Fallback to OpenGraph and Twitter meta tags (`og:title`, `og:image`, `og:description`).
|
||||||
|
4. **URL Slug Extraction:** As an ultimate safety net, derive clean human-readable names from the URL path if the HTML body was obstructed.
|
||||||
|
- **Anti-Bot & WAF Handling:**
|
||||||
|
- Sanitize input URLs by stripping transient tokens (`chal_t`, `force_referer`, session IDs).
|
||||||
|
- Emulate full browser properties (`window.chrome`, `navigator.plugins`, `navigator.languages`).
|
||||||
|
- Actively detect WAF challenge interstitial pages and wait for client-side JavaScript token computation and automatic redirection before evaluating the DOM.
|
||||||
|
|
||||||
362
src/index.js
362
src/index.js
@@ -23,7 +23,8 @@ const WESTERN_COUNTRIES = [
|
|||||||
'United Kingdom', 'UK', 'Niederlande', 'Netherlands', 'Frankreich', 'France',
|
'United Kingdom', 'UK', 'Niederlande', 'Netherlands', 'Frankreich', 'France',
|
||||||
'Spanien', 'Spain', 'Italien', 'Italy', 'Kanada', 'Canada', 'Australien', 'Australia',
|
'Spanien', 'Spain', 'Italien', 'Italy', 'Kanada', 'Canada', 'Australien', 'Australia',
|
||||||
'Belgien', 'Belgium', 'Schweden', 'Sweden', 'Norwegen', 'Norway', 'Dänemark', 'Denmark',
|
'Belgien', 'Belgium', 'Schweden', 'Sweden', 'Norwegen', 'Norway', 'Dänemark', 'Denmark',
|
||||||
'Finnland', 'Finland', 'Irland', 'Ireland', 'Luxemburg', 'Luxembourg', 'Indien', 'India'
|
'Finnland', 'Finland', 'Irland', 'Ireland', 'Luxemburg', 'Luxembourg',
|
||||||
|
'Neuseeland', 'New Zealand'
|
||||||
];
|
];
|
||||||
|
|
||||||
// ==========================================
|
// ==========================================
|
||||||
@@ -139,12 +140,32 @@ function cleanBookingUrl(rawUrl) {
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// ==========================================
|
||||||
|
// HILFSFUNKTION: SICHERE EVALUATION (RESISTENT GEGEN REDIRECTS)
|
||||||
|
// ==========================================
|
||||||
|
async function safeEvaluate(page, fn, fallback = null, arg = undefined) {
|
||||||
|
try {
|
||||||
|
return arg !== undefined ? await page.evaluate(fn, arg) : await page.evaluate(fn);
|
||||||
|
} catch (err) {
|
||||||
|
if (err.message && (err.message.includes('Execution context was destroyed') || err.message.includes('navigating') || err.message.includes('Target closed'))) {
|
||||||
|
await page.waitForLoadState('domcontentloaded').catch(() => {});
|
||||||
|
await page.waitForTimeout(2500);
|
||||||
|
try {
|
||||||
|
return arg !== undefined ? await page.evaluate(fn, arg) : await page.evaluate(fn);
|
||||||
|
} catch (innerErr) {
|
||||||
|
return fallback;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
return fallback;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
// ==========================================
|
// ==========================================
|
||||||
// BULLMQ WORKER: SCRAPING & REVIEWS
|
// BULLMQ WORKER: SCRAPING & REVIEWS
|
||||||
// ==========================================
|
// ==========================================
|
||||||
const worker = new Worker('scrape-jobs', async (job) => {
|
const worker = new Worker('scrape-jobs', async (job) => {
|
||||||
const { url, webhookUrl, appWebhookUrl, hotelId, debug } = job.data;
|
const { url, webhookUrl, appWebhookUrl, hotelId, debug, mode } = job.data;
|
||||||
const isReviewScan = Boolean(webhookUrl && webhookUrl.startsWith('http'));
|
const isReviewScan = Boolean((webhookUrl && webhookUrl.startsWith('http')) || mode === 'reviews' || mode === 'full');
|
||||||
const isDebugMode = Boolean(debug || DEBUG_HTML);
|
const isDebugMode = Boolean(debug || DEBUG_HTML);
|
||||||
|
|
||||||
console.log(`\n==================================================`);
|
console.log(`\n==================================================`);
|
||||||
@@ -167,29 +188,21 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
args: [
|
args: [
|
||||||
'--no-sandbox',
|
'--no-sandbox',
|
||||||
'--disable-setuid-sandbox',
|
'--disable-setuid-sandbox',
|
||||||
|
'--disable-dev-shm-usage',
|
||||||
'--disable-blink-features=AutomationControlled',
|
'--disable-blink-features=AutomationControlled',
|
||||||
'--disable-infobars'
|
'--window-size=1920,1080'
|
||||||
]
|
]
|
||||||
});
|
});
|
||||||
|
|
||||||
const context = await browser.newContext({
|
const context = await browser.newContext({
|
||||||
userAgent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
|
userAgent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36',
|
||||||
viewport: { width: 1366, height: 900 },
|
viewport: { width: 1920, height: 1080 },
|
||||||
locale: 'de-DE',
|
locale: 'de-DE',
|
||||||
timezoneId: 'Europe/Berlin',
|
timezoneId: 'Europe/Berlin'
|
||||||
extraHTTPHeaders: {
|
|
||||||
'Accept-Language': 'de-DE,de;q=0.9,en-US;q=0.8,en;q=0.7',
|
|
||||||
'Upgrade-Insecure-Requests': '1',
|
|
||||||
'Sec-Ch-Ua': '"Not/A)Brand";v="8", "Chromium";v="126", "Google Chrome";v="126"',
|
|
||||||
'Sec-Ch-Ua-Mobile': '?0',
|
|
||||||
'Sec-Ch-Ua-Platform': '"Windows"'
|
|
||||||
}
|
|
||||||
});
|
});
|
||||||
|
|
||||||
const page = await context.newPage();
|
|
||||||
|
|
||||||
// Verhindert Erkennung über navigator.webdriver und emuliert Standard-Browser-Eigenschaften
|
// Verhindert Erkennung über navigator.webdriver und emuliert Standard-Browser-Eigenschaften
|
||||||
await page.addInitScript(() => {
|
await context.addInitScript(() => {
|
||||||
Object.defineProperty(navigator, 'webdriver', { get: () => undefined });
|
Object.defineProperty(navigator, 'webdriver', { get: () => undefined });
|
||||||
window.chrome = {
|
window.chrome = {
|
||||||
runtime: {},
|
runtime: {},
|
||||||
@@ -205,18 +218,20 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
});
|
});
|
||||||
});
|
});
|
||||||
|
|
||||||
|
const page = await context.newPage();
|
||||||
|
|
||||||
let targetUrl = cleanBookingUrl(url);
|
let targetUrl = cleanBookingUrl(url);
|
||||||
if (isReviewScan && !targetUrl.includes('#tab-reviews')) {
|
|
||||||
targetUrl = targetUrl.replace('#tab-main', '') + '#tab-reviews';
|
|
||||||
}
|
|
||||||
|
|
||||||
console.log(`[Job ${job.id}] 🌐 Bereinigte Ziel-URL: ${targetUrl}`);
|
console.log(`[Job ${job.id}] 🌐 Bereinigte Ziel-URL: ${targetUrl}`);
|
||||||
|
|
||||||
await page.goto(targetUrl, { waitUntil: 'domcontentloaded', timeout: 45000 });
|
await page.goto(targetUrl, { waitUntil: 'domcontentloaded', timeout: 45000 });
|
||||||
|
await page.waitForLoadState('domcontentloaded').catch(() => {});
|
||||||
await page.waitForTimeout(2000);
|
await page.waitForTimeout(2000);
|
||||||
|
|
||||||
// Prüfen, ob AWS WAF Challenge aktiv ist
|
// Prüfen, ob AWS WAF Challenge aktiv ist (sicher gegenüber gleichzeitiger Navigation)
|
||||||
const isWafChallenge = await page.evaluate(() => {
|
let isWafChallenge = false;
|
||||||
|
try {
|
||||||
|
isWafChallenge = await page.evaluate(() => {
|
||||||
return Boolean(
|
return Boolean(
|
||||||
document.querySelector('#challenge-container') ||
|
document.querySelector('#challenge-container') ||
|
||||||
document.querySelector('script[src*="challenge.js"]') ||
|
document.querySelector('script[src*="challenge.js"]') ||
|
||||||
@@ -224,6 +239,21 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
(document.title === '' && (document.body ? document.body.innerText.includes('verify that you') : false))
|
(document.title === '' && (document.body ? document.body.innerText.includes('verify that you') : false))
|
||||||
);
|
);
|
||||||
});
|
});
|
||||||
|
} catch (navErr) {
|
||||||
|
await page.waitForLoadState('domcontentloaded').catch(() => {});
|
||||||
|
await page.waitForTimeout(2500);
|
||||||
|
try {
|
||||||
|
isWafChallenge = await page.evaluate(() => {
|
||||||
|
return Boolean(
|
||||||
|
document.querySelector('#challenge-container') ||
|
||||||
|
document.querySelector('script[src*="challenge.js"]') ||
|
||||||
|
window.AwsWafIntegration
|
||||||
|
);
|
||||||
|
});
|
||||||
|
} catch (e) {
|
||||||
|
isWafChallenge = false;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
if (isWafChallenge) {
|
if (isWafChallenge) {
|
||||||
console.log(`[Job ${job.id}] 🛡️ AWS WAF Challenge erkannt. Warte auf automatische Token-Weiterleitung...`);
|
console.log(`[Job ${job.id}] 🛡️ AWS WAF Challenge erkannt. Warte auf automatische Token-Weiterleitung...`);
|
||||||
@@ -234,12 +264,14 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
return !hasChallenge || Boolean(hasContent);
|
return !hasChallenge || Boolean(hasContent);
|
||||||
}, { timeout: 25000 });
|
}, { timeout: 25000 });
|
||||||
console.log(`[Job ${job.id}] 🔓 Weiterleitung nach AWS WAF erfolgt!`);
|
console.log(`[Job ${job.id}] 🔓 Weiterleitung nach AWS WAF erfolgt!`);
|
||||||
|
await page.waitForLoadState('domcontentloaded').catch(() => {});
|
||||||
await page.waitForTimeout(2000);
|
await page.waitForTimeout(2000);
|
||||||
} catch (wafErr) {
|
} catch (wafErr) {
|
||||||
console.log(`[Job ${job.id}] ⚠️ AWS WAF Timeout. Versuche direkten Reload der Basis-URL...`);
|
console.log(`[Job ${job.id}] ⚠️ AWS WAF Timeout. Versuche direkten Reload der Basis-URL...`);
|
||||||
try {
|
try {
|
||||||
const parsed = new URL(targetUrl);
|
const parsed = new URL(targetUrl);
|
||||||
await page.goto(parsed.origin + parsed.pathname, { waitUntil: 'domcontentloaded', timeout: 30000 });
|
await page.goto(parsed.origin + parsed.pathname, { waitUntil: 'domcontentloaded', timeout: 30000 });
|
||||||
|
await page.waitForLoadState('domcontentloaded').catch(() => {});
|
||||||
await page.waitForTimeout(3000);
|
await page.waitForTimeout(3000);
|
||||||
} catch (reloadErr) {}
|
} catch (reloadErr) {}
|
||||||
}
|
}
|
||||||
@@ -252,21 +284,13 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
|
|
||||||
// Cookie-Banner & störende Overlays für alle Scans sofort schließen
|
// Cookie-Banner & störende Overlays für alle Scans sofort schließen
|
||||||
try {
|
try {
|
||||||
const dismissSelectors = [
|
const dismissBtns = await page.$$(
|
||||||
'#onetrust-accept-btn-handler',
|
'button[aria-label*="schließen"], button[aria-label*="Schließen"], button[aria-label*="Close"], button[aria-label*="Dismiss"], #onetrust-accept-btn-handler, button:has-text("Akzeptieren"), button:has-text("Alle akzeptieren"), button[data-testid="selection-modal-close-btn"]'
|
||||||
'button#onetrust-reject-all-handler',
|
);
|
||||||
'button[data-testid="selection-modal-close-btn"]',
|
for (const btn of dismissBtns) {
|
||||||
'button[aria-label="Schließen"]',
|
|
||||||
'button[aria-label="Close"]',
|
|
||||||
'button.modal-mask-closeBtn'
|
|
||||||
];
|
|
||||||
for (const btnSel of dismissSelectors) {
|
|
||||||
const btn = await page.$(btnSel);
|
|
||||||
if (btn) {
|
|
||||||
await btn.click().catch(() => {});
|
await btn.click().catch(() => {});
|
||||||
await page.waitForTimeout(300);
|
await page.waitForTimeout(300);
|
||||||
}
|
}
|
||||||
}
|
|
||||||
} catch (e) {}
|
} catch (e) {}
|
||||||
|
|
||||||
// HTML-Dump & Screenshot bei Debug-Modus sichern
|
// HTML-Dump & Screenshot bei Debug-Modus sichern
|
||||||
@@ -290,7 +314,7 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
} catch (e) {}
|
} catch (e) {}
|
||||||
|
|
||||||
// 2. STAMMDATEN EXTRAHIEREN (JSON-LD + DOM + META FALLBACKS)
|
// 2. STAMMDATEN EXTRAHIEREN (JSON-LD + DOM + META FALLBACKS)
|
||||||
hotelDetails = await page.evaluate(() => {
|
hotelDetails = await safeEvaluate(page, () => {
|
||||||
let name = null;
|
let name = null;
|
||||||
let rating = null;
|
let rating = null;
|
||||||
let reviewCount = null;
|
let reviewCount = null;
|
||||||
@@ -450,7 +474,7 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
}
|
}
|
||||||
|
|
||||||
return { name, rating, reviewCount, address, image };
|
return { name, rating, reviewCount, address, image };
|
||||||
});
|
}, { name: null, rating: null, reviewCount: null, address: null, image: null });
|
||||||
|
|
||||||
// Fallback: Name aus URL-Slug ableiten wenn im HTML nichts gefunden wurde
|
// Fallback: Name aus URL-Slug ableiten wenn im HTML nichts gefunden wurde
|
||||||
if (!hotelDetails.name) {
|
if (!hotelDetails.name) {
|
||||||
@@ -458,7 +482,7 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
}
|
}
|
||||||
|
|
||||||
// 3. DOM-DATEN (Nächte, Personen, Zimmergröße)
|
// 3. DOM-DATEN (Nächte, Personen, Zimmergröße)
|
||||||
const extraDom = await page.evaluate(() => {
|
const extraDom = await safeEvaluate(page, () => {
|
||||||
try {
|
try {
|
||||||
const bodyText = document.body ? (document.body.innerText || '') : '';
|
const bodyText = document.body ? (document.body.innerText || '') : '';
|
||||||
const nMatch = bodyText.match(/(\d+)\s*Nächte?/i);
|
const nMatch = bodyText.match(/(\d+)\s*Nächte?/i);
|
||||||
@@ -472,14 +496,14 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
} catch (e) {
|
} catch (e) {
|
||||||
return { dNights: 1, dPersons: 2, roomSize: null };
|
return { dNights: 1, dPersons: 2, roomSize: null };
|
||||||
}
|
}
|
||||||
});
|
}, { dNights: 1, dPersons: 2, roomSize: null });
|
||||||
|
|
||||||
domData.dNights = extraDom.dNights;
|
domData.dNights = extraDom.dNights;
|
||||||
domData.dPersons = extraDom.dPersons;
|
domData.dPersons = extraDom.dPersons;
|
||||||
roomSizeM2 = extraDom.roomSize;
|
roomSizeM2 = extraDom.roomSize;
|
||||||
|
|
||||||
// PREIS AUSLESEN
|
// PREIS AUSLESEN
|
||||||
scrapedPrice = await page.evaluate(() => {
|
scrapedPrice = await safeEvaluate(page, () => {
|
||||||
try {
|
try {
|
||||||
const selectors = [
|
const selectors = [
|
||||||
'[data-testid="price-and-discounted-price"]',
|
'[data-testid="price-and-discounted-price"]',
|
||||||
@@ -502,10 +526,10 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
}
|
}
|
||||||
} catch (e) {}
|
} catch (e) {}
|
||||||
return null;
|
return null;
|
||||||
});
|
}, null);
|
||||||
|
|
||||||
// GENIUS BADGE PRÜFUNG
|
// GENIUS BADGE PRÜFUNG
|
||||||
hasGenius = await page.evaluate(() => {
|
hasGenius = await safeEvaluate(page, () => {
|
||||||
try {
|
try {
|
||||||
const geniusSelectors = [
|
const geniusSelectors = [
|
||||||
'.m-badge__genius',
|
'.m-badge__genius',
|
||||||
@@ -519,7 +543,7 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
} catch (e) {
|
} catch (e) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
});
|
}, false);
|
||||||
|
|
||||||
console.log(`[Job ${job.id}] 📋 Extrahierte Stammdaten:`, {
|
console.log(`[Job ${job.id}] 📋 Extrahierte Stammdaten:`, {
|
||||||
name: hotelDetails.name,
|
name: hotelDetails.name,
|
||||||
@@ -542,78 +566,29 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
if (isReviewScan) {
|
if (isReviewScan) {
|
||||||
console.log(`[Job ${job.id}] 🚀 Starte gezielte Review-Extraction für "${hotelDetails.name || 'Hotel'}"...`);
|
console.log(`[Job ${job.id}] 🚀 Starte gezielte Review-Extraction für "${hotelDetails.name || 'Hotel'}"...`);
|
||||||
|
|
||||||
// A. Cookie-Banner und störende Overlays schließen
|
// B. Herunterscrollen um Lazy-Components zu hydrieren und Review-Dialog zu öffnen
|
||||||
try {
|
console.log(`[Job ${job.id}] 📜 Scrolle herunter, um verzögert geladene Review-Komponenten zu aktivieren...`);
|
||||||
const dismissSelectors = [
|
await page.evaluate(() => window.scrollTo(0, 4000));
|
||||||
'#onetrust-accept-btn-handler',
|
await page.waitForTimeout(2500);
|
||||||
'button#onetrust-reject-all-handler',
|
|
||||||
'button[data-testid="selection-modal-close-btn"]',
|
// C. Klick auf Review-Trigger ausführen
|
||||||
'button[aria-label="Schließen"]',
|
console.log(`[Job ${job.id}] 👉 Öffne Review-Dialog über 'Alle Bewertungen lesen' / 'Gästebewertungen'...`);
|
||||||
'button[aria-label="Close"]'
|
await page.evaluate(() => {
|
||||||
];
|
const btn = document.querySelector('button[data-testid="fr-read-all-reviews"]') ||
|
||||||
for (const btnSel of dismissSelectors) {
|
document.querySelector('button[data-testid="review-score-read-all-actionable"]') ||
|
||||||
const btn = await page.$(btnSel);
|
document.getElementById('reviews-tab-trigger') ||
|
||||||
|
document.querySelector('[data-testid="Property-Header-Nav-Tab-Trigger-reviews"]');
|
||||||
if (btn) {
|
if (btn) {
|
||||||
await btn.click().catch(() => {});
|
btn.scrollIntoView({ behavior: 'instant', block: 'center' });
|
||||||
await page.waitForTimeout(400);
|
btn.click();
|
||||||
}
|
btn.dispatchEvent(new MouseEvent('click', { bubbles: true, cancelable: true, view: window }));
|
||||||
}
|
|
||||||
} catch (e) {}
|
|
||||||
|
|
||||||
// B. 500px nach unten scrollen, damit Bookings Sticky-Header (#reviews-tab-trigger) überhaupt sichtbar & aktiv wird
|
|
||||||
await page.evaluate(() => window.scrollBy(0, 550));
|
|
||||||
await page.waitForTimeout(1500);
|
|
||||||
|
|
||||||
// C. Klick per echtem Maus-Event auf den Trigger ausführen
|
|
||||||
let clicked = false;
|
|
||||||
try {
|
|
||||||
// Liste möglicher Klick-Ziele (Header-Tab, Scorecard-Badge oder Text "Gästebewertungen")
|
|
||||||
const triggerLocators = [
|
|
||||||
page.locator('#reviews-tab-trigger').first(),
|
|
||||||
page.locator('[data-testid="Property-Header-Nav-Tab-Trigger-reviews"]').first(),
|
|
||||||
page.locator('a:has-text("Gästebewertungen")').first(),
|
|
||||||
page.locator('button:has-text("Gästebewertungen")').first(),
|
|
||||||
page.locator('[data-testid="review-score-right-component"]').first(),
|
|
||||||
page.locator('[data-testid="review-score-component"]').first()
|
|
||||||
];
|
|
||||||
|
|
||||||
for (const loc of triggerLocators) {
|
|
||||||
if (await loc.count() > 0 && await loc.isVisible()) {
|
|
||||||
console.log(`[Job ${job.id}] 👉 Sichtbaren Review-Trigger gefunden. Führe Klick aus...`);
|
|
||||||
|
|
||||||
// Reale Maus-Koordinaten ermitteln & klicken
|
|
||||||
const box = await loc.boundingBox();
|
|
||||||
if (box) {
|
|
||||||
await page.mouse.click(box.x + box.width / 2, box.y + box.height / 2);
|
|
||||||
clicked = true;
|
|
||||||
await page.waitForTimeout(3500);
|
|
||||||
break;
|
|
||||||
} else {
|
|
||||||
await loc.click({ force: true }).catch(() => {});
|
|
||||||
clicked = true;
|
|
||||||
await page.waitForTimeout(3500);
|
|
||||||
break;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
// Fallback: JS-Klick direkt auf das HTML-Element mit Text "Gästebewertungen"
|
|
||||||
if (!clicked) {
|
|
||||||
console.log(`[Job ${job.id}] ⚠️ Fallback: Klick per DOM-Suche nach 'Gästebewertungen'...`);
|
|
||||||
clicked = await page.evaluate(() => {
|
|
||||||
const el = Array.from(document.querySelectorAll('a, button, span, div'))
|
|
||||||
.find(e => e.innerText && e.innerText.includes('Gästebewertungen') && (e.id === 'reviews-tab-trigger' || e.closest('#reviews-tab-trigger') || e.getAttribute('data-testid')?.includes('reviews')));
|
|
||||||
if (el) {
|
|
||||||
const clickTarget = el.closest('a') || el.closest('button') || el;
|
|
||||||
clickTarget.click();
|
|
||||||
clickTarget.dispatchEvent(new MouseEvent('click', { bubbles: true, cancelable: true, view: window }));
|
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
return false;
|
return false;
|
||||||
}).catch(() => false);
|
}).catch(() => false);
|
||||||
|
|
||||||
if (clicked) await page.waitForTimeout(3500);
|
// 4 Sekunden warten bis das Dialog-Panel mit den 1.000+ Bewertungen vollständig geöffnet ist
|
||||||
}
|
await page.waitForTimeout(4000);
|
||||||
|
|
||||||
// Screenshot zur visuellen Kontrolle aktualisieren
|
// Screenshot zur visuellen Kontrolle aktualisieren
|
||||||
try {
|
try {
|
||||||
@@ -621,44 +596,13 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
console.log(`[Job ${job.id}] 📸 Screenshot aktualisiert unter: scraper-service/src/review-debug.png`);
|
console.log(`[Job ${job.id}] 📸 Screenshot aktualisiert unter: scraper-service/src/review-debug.png`);
|
||||||
} catch (e) {}
|
} catch (e) {}
|
||||||
|
|
||||||
} catch (e) {
|
// D. Sortierung auf "Neueste zuerst" (NEWEST_FIRST) umstellen
|
||||||
console.log(`[Job ${job.id}] ⚠️ Fehler beim Klick auf Review-Trigger:`, e.message);
|
|
||||||
}
|
|
||||||
|
|
||||||
// Warten bis Review-Karten im Sliding Panel oder DOM sichtbar sind
|
|
||||||
try {
|
|
||||||
await page.waitForSelector(
|
|
||||||
'[data-testid="review-card"], [data-testid="review-cards"], #hp-reviews-sliding, div[data-component="hotel/reviews/review-card"], li.review_item',
|
|
||||||
{ timeout: 8000 }
|
|
||||||
);
|
|
||||||
console.log(`[Job ${job.id}] 👁️ Review-Bereich erfolgreich gerendert.`);
|
|
||||||
} catch (e) {
|
|
||||||
console.log(`[Job ${job.id}] ⚠️ Wartezeit abgelaufen – starte Auslesen aus dem aktuellen DOM...`);
|
|
||||||
}
|
|
||||||
|
|
||||||
// C. Sortierung auf "Neueste zuerst" (NEWEST_FIRST) umstellen
|
|
||||||
try {
|
try {
|
||||||
const sorterLocator = page.locator('#reviewListSorters, [data-testid="reviews-sorter-component"], select[name="reviewListSorters"]').first();
|
const sorterLocator = page.locator('#reviewListSorters, [data-testid="reviews-sorter-component"], select[name="reviewListSorters"]').first();
|
||||||
if (await sorterLocator.count() > 0) {
|
if (await sorterLocator.count() > 0) {
|
||||||
console.log(`[Job ${job.id}] 🔄 Stelle Sortierung auf 'NEWEST_FIRST' (Neueste zuerst) um...`);
|
console.log(`[Job ${job.id}] 🔄 Stelle Sortierung auf 'NEWEST_FIRST' (Neueste zuerst) um...`);
|
||||||
await sorterLocator.selectOption('NEWEST_FIRST');
|
await sorterLocator.selectOption('NEWEST_FIRST');
|
||||||
await page.waitForTimeout(2500); // Warten bis Booking die Liste per AJAX neu lädt
|
await page.waitForTimeout(3000); // Warten bis Booking die Liste per AJAX neu lädt
|
||||||
} else {
|
|
||||||
// Fallback via JavaScript Event-Dispatch
|
|
||||||
const changed = await page.evaluate(() => {
|
|
||||||
const select = document.querySelector('#reviewListSorters, [data-testid="reviews-sorter-component"], select[name="reviewListSorters"]');
|
|
||||||
if (select) {
|
|
||||||
select.value = 'NEWEST_FIRST';
|
|
||||||
select.dispatchEvent(new Event('change', { bubbles: true }));
|
|
||||||
return true;
|
|
||||||
}
|
|
||||||
return false;
|
|
||||||
}).catch(() => false);
|
|
||||||
|
|
||||||
if (changed) {
|
|
||||||
console.log(`[Job ${job.id}] 🔄 Sortierung via JS-Event auf 'NEWEST_FIRST' umgestellt.`);
|
|
||||||
await page.waitForTimeout(2500);
|
|
||||||
}
|
|
||||||
}
|
}
|
||||||
} catch (e) {
|
} catch (e) {
|
||||||
console.log(`[Job ${job.id}] ⚠️ Sortierung konnte nicht angepasst werden:`, e.message);
|
console.log(`[Job ${job.id}] ⚠️ Sortierung konnte nicht angepasst werden:`, e.message);
|
||||||
@@ -671,123 +615,87 @@ const worker = new Worker('scrape-jobs', async (job) => {
|
|||||||
pageCount++;
|
pageCount++;
|
||||||
console.log(`[Job ${job.id}] 📄 Scanne Bewertungsseite ${pageCount} (Bisher gesammelt: ${extractedReviews.length})...`);
|
console.log(`[Job ${job.id}] 📄 Scanne Bewertungsseite ${pageCount} (Bisher gesammelt: ${extractedReviews.length})...`);
|
||||||
|
|
||||||
const evalResult = await page.evaluate(({ westernCountries }) => {
|
// Extrahiere alle Review-Karten aus dem Dialog / DOM
|
||||||
let cards = Array.from(document.querySelectorAll([
|
const pageReviews = await page.evaluate(() => {
|
||||||
'[data-testid="review-card"]',
|
const cards = Array.from(document.querySelectorAll(
|
||||||
'[data-testid="review-cards"] > li',
|
'#reviewCardsSection [data-testid="review-card"], div[role="dialog"] [data-testid="review-card"], [data-testid="review-card"]'
|
||||||
'[data-testid="review-cards"] > div',
|
));
|
||||||
'div[data-testid*="review-card"]',
|
const items = [];
|
||||||
'li[data-testid*="review-card"]',
|
const currentYear = new Date().getFullYear();
|
||||||
'li.review_list_new_item_block',
|
|
||||||
'div.c-review-block__item',
|
|
||||||
'div[itemprop="review"]',
|
|
||||||
'li.review_item'
|
|
||||||
].join(',')));
|
|
||||||
|
|
||||||
if (cards.length === 0) {
|
for (const card of cards) {
|
||||||
const container = document.querySelector('#hp-reviews-sliding, #reviewCardsSection, [data-testid="review-cards"], #blockdisplay4');
|
const titleEl = card.querySelector('[data-testid="review-title"], h4');
|
||||||
if (container) {
|
const posEl = card.querySelector('[data-testid="review-positive-text"]');
|
||||||
cards = Array.from(container.querySelectorAll('li, div')).filter(el => (el.innerText || '').length > 40);
|
const negEl = card.querySelector('[data-testid="review-negative-text"]');
|
||||||
}
|
const dateEl = card.querySelector('[data-testid="review-date"]');
|
||||||
}
|
const flagImg = card.querySelector('[data-testid="review-avatar"] .fff1944c52 img[alt], [data-testid="review-country"] img[alt], img.b8d1620349');
|
||||||
|
const countrySpan = card.querySelector('[data-testid="review-avatar"] .fff1944c52 span:not(.cec4c9cfef), [data-testid="review-country"]');
|
||||||
|
|
||||||
// Filterboxen und Scorecard-Zusammenfassungen explizit ausschließen
|
const title = titleEl ? titleEl.innerText.trim() : '';
|
||||||
cards = cards.filter(el => {
|
const positive = posEl ? posEl.innerText.trim().replace(/^„|“$/g, '').trim() : '';
|
||||||
const text = el.innerText || '';
|
const negative = negEl ? negEl.innerText.trim().replace(/^„|“$/g, '').trim() : '';
|
||||||
return text.length > 25 &&
|
const dateText = dateEl ? dateEl.innerText.trim() : 'Aktuell';
|
||||||
!text.includes('Wählen Sie Themen') &&
|
const country = (flagImg ? flagImg.getAttribute('alt') : (countrySpan ? countrySpan.innerText : '')).trim();
|
||||||
!text.includes('Kategorien:') &&
|
|
||||||
!el.classList.contains('review_score_breakdown');
|
|
||||||
});
|
|
||||||
|
|
||||||
const results = [];
|
|
||||||
const debugLogs = [];
|
|
||||||
const currentYear = 2026;
|
|
||||||
|
|
||||||
for (let idx = 0; idx < cards.length; idx++) {
|
|
||||||
const card = cards[idx];
|
|
||||||
const cardText = card.innerText || '';
|
const cardText = card.innerText || '';
|
||||||
if (!cardText.trim()) continue;
|
|
||||||
|
|
||||||
// 1. Herkunftsland des Gastes ermitteln
|
|
||||||
const countryEl = card.querySelector('[data-testid="review-country"], .bui-avatar-or-image__title, .user_country, .c-review__country');
|
|
||||||
const guestCountryText = countryEl ? countryEl.innerText : cardText;
|
|
||||||
|
|
||||||
const isWestern = westernCountries.some(c => guestCountryText.toLowerCase().includes(c.toLowerCase()));
|
|
||||||
if (!isWestern) {
|
|
||||||
const preview = guestCountryText.substring(0, 50).replace(/\n/g, ' ');
|
|
||||||
debugLogs.push(`Karte #${idx + 1}: ❌ Land nicht in Liste ("${preview}")`);
|
|
||||||
continue;
|
|
||||||
}
|
|
||||||
|
|
||||||
// 2. Datum / Jahr prüfen (letzte 2 Jahre)
|
|
||||||
const yearMatch = cardText.match(/\b(20\d{2})\b/);
|
const yearMatch = cardText.match(/\b(20\d{2})\b/);
|
||||||
if (yearMatch) {
|
if (yearMatch) {
|
||||||
const reviewYear = parseInt(yearMatch[1], 10);
|
const reviewYear = parseInt(yearMatch[1], 10);
|
||||||
if (currentYear - reviewYear > 2) {
|
if (currentYear - reviewYear > 1) {
|
||||||
debugLogs.push(`Karte #${idx + 1}: ❌ Zu alt (${reviewYear})`);
|
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
// 3. Texte extrahieren
|
|
||||||
const titleEl = card.querySelector('[data-testid="review-title"], .c-review__title, h4, .review_title');
|
|
||||||
const posEl = card.querySelector('[data-testid="review-positive-text"], .c-review__row--positive, .review_pos');
|
|
||||||
const negEl = card.querySelector('[data-testid="review-negative-text"], .c-review__row--negative, .review_neg');
|
|
||||||
const dateEl = card.querySelector('[data-testid="review-date"], .c-review__date, .review_item_date');
|
|
||||||
|
|
||||||
const title = titleEl ? titleEl.innerText.trim() : '';
|
|
||||||
const positive = posEl ? posEl.innerText.trim() : '';
|
|
||||||
const negative = negEl ? negEl.innerText.trim() : '';
|
|
||||||
const dateText = dateEl ? dateEl.innerText.trim() : (yearMatch ? yearMatch[0] : 'Aktuell');
|
|
||||||
|
|
||||||
let fallbackText = '';
|
let fallbackText = '';
|
||||||
if (!positive && !negative) {
|
if (!positive && !negative) {
|
||||||
fallbackText = cardText.replace(title, '').substring(0, 300).trim();
|
fallbackText = cardText.replace(title, '').substring(0, 300).trim();
|
||||||
}
|
}
|
||||||
|
|
||||||
if (title || positive || negative || fallbackText) {
|
if (title || positive || negative || fallbackText) {
|
||||||
results.push({ title, positive: positive || fallbackText, negative, date: dateText });
|
items.push({ title, positive: positive || fallbackText, negative, date: dateText, country });
|
||||||
debugLogs.push(`Karte #${idx + 1}: ✅ Akzeptiert ("${title || (positive ? positive.substring(0, 25) : 'Review')}")`);
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
return items;
|
||||||
|
}).catch(err => {
|
||||||
|
console.error(`[Job ${job.id}] Evaluate error on page ${pageCount}:`, err.message);
|
||||||
|
return [];
|
||||||
|
});
|
||||||
|
|
||||||
return {
|
console.log(`[Job ${job.id}] 🔍 Echte Gästekarten auf Seite ${pageCount} gefunden: ${pageReviews.length}`);
|
||||||
totalCards: cards.length,
|
|
||||||
results,
|
|
||||||
debugLogs: debugLogs.slice(0, 8)
|
|
||||||
};
|
|
||||||
}, { westernCountries: WESTERN_COUNTRIES });
|
|
||||||
|
|
||||||
console.log(`[Job ${job.id}] 🔍 Echte Gästekarten gefunden: ${evalResult.totalCards}`);
|
for (const rev of pageReviews) {
|
||||||
if (evalResult.debugLogs) {
|
if (rev.country) {
|
||||||
evalResult.debugLogs.forEach(log => console.log(`[Job ${job.id}] └─ ${log}`));
|
const isWestern = WESTERN_COUNTRIES.some(c => rev.country.toLowerCase().includes(c.toLowerCase()));
|
||||||
|
if (!isWestern) {
|
||||||
|
console.log(`[Job ${job.id}] └─ ❌ Land nicht in Liste: "${rev.country}"`);
|
||||||
|
continue;
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
// Deduplizierung
|
|
||||||
for (const rev of evalResult.results) {
|
|
||||||
const key = `${rev.title}-${rev.positive}-${rev.date}`;
|
const key = `${rev.title}-${rev.positive}-${rev.date}`;
|
||||||
if (!seenKeys.has(key) && extractedReviews.length < 80) {
|
if (!seenKeys.has(key) && extractedReviews.length < 80) {
|
||||||
seenKeys.add(key);
|
seenKeys.add(key);
|
||||||
extractedReviews.push(rev);
|
extractedReviews.push(rev);
|
||||||
|
console.log(`[Job ${job.id}] └─ ✅ Akzeptiert: "${rev.title || rev.positive.substring(0, 30)}" (${rev.country || 'International'})`);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
if (extractedReviews.length >= 80) break;
|
if (extractedReviews.length >= 80) break;
|
||||||
|
|
||||||
// Weiterblättern im Modal oder der Hauptseite
|
// Weiterblättern im Modal
|
||||||
try {
|
const nextResult = await page.evaluate(() => {
|
||||||
const nextBtn = await page.$(
|
const nextBtn = document.querySelector('div[role="dialog"] button[aria-label="Nächste Seite"], button[aria-label="Nächste Seite"], button[aria-label*="Nächste Seite"]');
|
||||||
'#hp-reviews-sliding button[aria-label*="Nächste"], #hp-reviews-sliding button[aria-label*="Next"], #reviewCardsSection button:has(svg), button[aria-label*="Nächste"], button[aria-label*="Next"], a.paging-next, li.bui-pagination__next-arrow a'
|
if (nextBtn && !nextBtn.disabled && !nextBtn.hasAttribute('disabled')) {
|
||||||
);
|
nextBtn.scrollIntoView({ behavior: 'instant', block: 'center' });
|
||||||
if (nextBtn) {
|
nextBtn.click();
|
||||||
const isDisabled = await nextBtn.getAttribute('disabled');
|
nextBtn.dispatchEvent(new MouseEvent('click', { bubbles: true, cancelable: true, view: window }));
|
||||||
if (isDisabled === null) {
|
return true;
|
||||||
await nextBtn.click({ force: true });
|
}
|
||||||
await page.waitForTimeout(2500);
|
return false;
|
||||||
} else break;
|
}).catch(() => false);
|
||||||
} else break;
|
|
||||||
} catch (e) {
|
if (nextResult) {
|
||||||
|
await page.waitForTimeout(3000);
|
||||||
|
} else {
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|||||||
Binary file not shown.
|
Before Width: | Height: | Size: 652 KiB |
Reference in New Issue
Block a user