airbnb-finder/test-scraper.ts

/**
 * Test script for Puppeteer-based Airbnb scraper
 * Run with: npx tsx test-scraper.ts
 */

import { scrapeAirbnbWithPuppeteer } from './src/lib/airbnb/puppeteer-scraper';

const TEST_URL = 'https://www.airbnb.com/rooms/52367822'; // Valid listing in Bad Bellingen, Germany

async function main() {
  console.log('========================================');
  console.log('Airbnb Puppeteer Scraper Test');
  console.log('========================================\n');
  
  console.log(`Testing URL: ${TEST_URL}\n`);
  console.log('Starting scraper (this may take 30-60 seconds)...\n');
  
  const startTime = Date.now();
  
  try {
    const result = await scrapeAirbnbWithPuppeteer(TEST_URL);
    
    const elapsed = ((Date.now() - startTime) / 1000).toFixed(1);
    console.log(`\n✅ Scraping completed in ${elapsed}s\n`);
    
    if (!result) {
      console.log('❌ Result is null - scraping may have failed');
      return;
    }
    
    console.log('========================================');
    console.log('EXTRACTED DATA');
    console.log('========================================\n');
    
    // Title
    console.log('📌 TITLE:');
    console.log(`   Value: ${result.title?.value || 'N/A'}`);
    console.log(`   Source: ${result.title?.source || 'N/A'}`);
    console.log(`   Confidence: ${result.title?.confidence || 'N/A'}\n`);
    
    // Price
    console.log('💰 PRICE:');
    console.log(`   Nightly: ${result.nightlyPrice?.value || 'N/A'} EUR`);
    console.log(`   Total: ${result.totalPrice?.value || 'N/A'} EUR`);
    console.log(`   Status: ${result.priceStatus || 'N/A'}\n`);
    
    // Location
    console.log('📍 LOCATION:');
    console.log(`   Text: ${result.locationText?.value || 'N/A'}`);
    console.log(`   Lat/Lng: ${result.latitude}, ${result.longitude}\n`);
    
    // Rating
    console.log('⭐ RATING:');
    console.log(`   Rating: ${result.rating?.value || 'N/A'}`);
    console.log(`   Reviews: ${result.reviewCount?.value || 'N/A'}\n`);
    
    // Capacity
    console.log('🏠 CAPACITY:');
    console.log(`   Guests: ${result.guestCount?.value || 'N/A'}`);
    console.log(`   Bedrooms: ${result.bedrooms?.value || 'N/A'}`);
    console.log(`   Beds: ${result.beds?.value || 'N/A'}`);
    console.log(`   Bathrooms: ${result.bathrooms?.value || 'N/A'}\n`);
    
    // Sleeping Options
    console.log('🛏️ SLEEPING OPTIONS:');
    if (result.sleepingOptions && result.sleepingOptions.length > 0) {
      result.sleepingOptions.forEach((opt, i) => {
        console.log(`   ${i + 1}. ${opt.quantity}x ${opt.bedType} (${opt.spotsPerUnit} spots, ${opt.quality})`);
      });
      console.log(`   Max sleeping places: ${result.maxSleepingPlaces}`);
      console.log(`   Suitable for 4: ${result.suitableFor4 ? '✅ Yes' : '❌ No'}`);
      console.log(`   Quality: ${result.sleepingDataQuality}`);
    } else {
      console.log('   No sleeping options extracted');
    }
    console.log('');
    
    // Host
    console.log('👤 HOST:');
    console.log(`   Name: ${result.hostName?.value || 'N/A'}\n`);
    
    // Images
    console.log('🖼️ IMAGES:');
    console.log(`   Count: ${result.images?.length || 0}`);
    if (result.images && result.images.length > 0) {
      console.log(`   First 3:`);
      result.images.slice(0, 3).forEach((img, i) => {
        console.log(`   ${i + 1}. ${img.substring(0, 80)}...`);
      });
    }
    console.log('');
    
    // Amenities
    console.log('✨ AMENITIES:');
    console.log(`   Count: ${result.amenities?.length || 0}`);
    if (result.amenities && result.amenities.length > 0) {
      console.log(`   First 10: ${result.amenities.slice(0, 10).join(', ')}`);
    }
    console.log('');
    
    // Description
    console.log('📝 DESCRIPTION:');
    const desc = result.description?.value || 'N/A';
    console.log(`   ${desc.substring(0, 200)}${desc.length > 200 ? '...' : ''}\n`);
    
    // External ID
    console.log('🔗 EXTERNAL ID:');
    console.log(`   ${result.externalId || 'N/A'}\n`);
    
    // Extraction Log
    console.log('📋 EXTRACTION LOG:');
    result.extractionLog?.forEach(log => {
      console.log(`   - ${log}`);
    });
    
    console.log('\n========================================');
    console.log('TEST COMPLETE');
    console.log('========================================');
    
  } catch (error) {
    const elapsed = ((Date.now() - startTime) / 1000).toFixed(1);
    console.log(`\n❌ Error after ${elapsed}s:`);
    console.error(error);
  }
}

main().catch(console.error);