← back to Handbag Authentication
crawler/govdeals-scraper.js
128 lines
const puppeteer = require('puppeteer');
/**
* GovDeals.com Scraper
* Government surplus auctions - police seizures, unclaimed property, etc.
*/
class GovDealsScraper {
constructor() {
this.baseUrl = 'https://www.govdeals.com';
}
async scrapeSearch(brand, page = 1) {
try {
const browser = await puppeteer.launch({
headless: 'new',
args: ['--no-sandbox', '--disable-setuid-sandbox']
});
const browserPage = await browser.newPage();
await browserPage.setUserAgent('Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36');
const searchUrl = `${this.baseUrl}/index.cfm?fa=Main.AdvSearchResultsNew&kWord=${encodeURIComponent(brand + ' handbag')}&searchType=3&sorting=1&page=${page}`;
console.log(`Scraping GovDeals: ${searchUrl}`);
await browserPage.goto(searchUrl, { waitUntil: 'networkidle2', timeout: 30000 });
await browserPage.waitForSelector('.listing-item, .item-row', { timeout: 10000 }).catch(() => {});
const listings = await browserPage.evaluate(() => {
const items = [];
const itemElements = document.querySelectorAll('.listing-item, .item-row, [class*="listing"]');
itemElements.forEach((item) => {
try {
const titleEl = item.querySelector('.listing-title, .title, h3, a');
const title = titleEl?.textContent?.trim() || '';
// Price extraction
let price = 0;
const priceEl = item.querySelector('.current-bid, .price, [class*="bid"], [class*="price"]');
if (priceEl) {
const priceText = priceEl.textContent.trim();
const priceMatch = priceText.match(/\$?([\d,]+\.?\d*)/);
if (priceMatch) {
price = parseFloat(priceMatch[1].replace(/,/g, ''));
}
}
const imgEl = item.querySelector('img');
const image = imgEl?.src || '';
const linkEl = item.querySelector('a');
const url = linkEl?.href || '';
if (url && !url.startsWith('http')) {
url = 'https://www.govdeals.com' + url;
}
const idMatch = url.match(/itemid=(\d+)/);
const id = idMatch?.[1] || url;
const timeEl = item.querySelector('.time-left, .ends, [class*="time"]');
const endTime = timeEl?.textContent?.trim() || '';
// Only include if it mentions designer brands
const isDesignerBag = title.match(/Louis Vuitton|Gucci|Chanel|Hermes|Prada|Dior|Fendi|Balenciaga|Celine|Bottega|Designer|Luxury/i);
if (title && price > 0 && isDesignerBag) {
items.push({
external_id: `govdeals_${id}`,
title,
price_jpy: Math.round(price * 150), // Convert USD to JPY estimate
price_usd: price,
image_url: image,
thumbnail_url: image,
product_url: url,
listing_type: 'auction',
auction_end_date: endTime,
condition: 'Seized/Surplus'
});
}
} catch (e) {
console.error('Error parsing GovDeals item:', e);
}
});
return items;
});
await browser.close();
return listings;
} catch (error) {
console.error(`Error scraping GovDeals page ${page}:`, error.message);
return [];
}
}
async scrapeAllBrands() {
const allListings = [];
const brands = ['Louis Vuitton', 'Gucci', 'Chanel', 'Hermes', 'Prada'];
for (const brand of brands) {
console.log(`\nScraping GovDeals for ${brand}...`);
for (let page = 1; page <= 3; page++) {
const listings = await this.scrapeSearch(brand, page);
if (listings.length === 0) break;
listings.forEach(listing => {
listing.source = 'govdeals';
// Try to extract brand from title
const brandMatch = listing.title.match(/Louis Vuitton|Gucci|Chanel|Hermes|Prada|Dior|Fendi/i);
listing.brand = brandMatch ? brandMatch[0] : brand;
});
allListings.push(...listings);
console.log(`Found ${listings.length} items on page ${page}`);
await new Promise(resolve => setTimeout(resolve, 3000));
}
}
return allListings;
}
}
module.exports = GovDealsScraper;