get vin data in scrapes

Author: quicklywilliam

scrapers/carmax/scrape.js

@@ -95,10 +95,27 @@ class CarMaxScraper extends BaseScraper {
       // Wait a bit for content to render
       await new Promise(resolve => setTimeout(resolve, 1000));
 
-      // Get page HTML and parse
+      // Get page HTML
       const html = await this.page.content();
+
+      // Extract VIN data from JavaScript array: const cars = [{"stockNumber":...,"vin":"..."}...]
+      const vinMap = new Map();
+      const carsMatch = html.match(/const cars = (\[.*?\]);/s);
+      if (carsMatch) {
+        try {
+          const carsData = JSON.parse(carsMatch[1]);
+          for (const car of carsData) {
+            if (car.stockNumber && car.vin) {
+              vinMap.set(car.stockNumber.toString(), car.vin);
+            }
+          }
+        } catch (e) {
+          // Failed to parse, continue without VINs
+        }
+      }
+
       const $ = cheerio.load(html);
-      const pageListings = parseListings($, query.make, query.model);
+      const pageListings = parseListings($, query.make, query.model, vinMap);
 
       // Deduplicate - only add listings we haven't seen before
       for (const listing of pageListings) {
@@ -162,7 +179,7 @@ function buildSearchUrl(make, model) {
   return `https://www.carmax.com/cars/${makeSlug}/${modelSlug}`;
 }
 
-function parseListings($, make, model) {
+function parseListings($, make, model, vinMap = new Map()) {
   const listings = [];
 
   // CarMax car tiles
@@ -177,6 +194,9 @@ function parseListings($, make, model) {
         return; // Skip listings without valid IDs
       }
 
+      // Get VIN from map
+      const vin = vinMap.get(stockId) || null;
+
       // Extract price
       const priceText = $card.find('.scct--price-miles-info--price').text().trim();
       const price = parseInt(priceText.replace(/[$,*]/g, ''));
@@ -229,6 +249,7 @@ function parseListings($, make, model) {
 
       listings.push({
         id: stockId,
+        vin,
         make,
         model,
         year,

scrapers/carvana/scrape.js

@@ -97,13 +97,55 @@ class CarvanaScraper extends BaseScraper {
 
       await new Promise(resolve => setTimeout(resolve, 1000));
 
-      // Get page HTML and parse
-      const html = await this.page.content();
-      const $ = cheerio.load(html);
-
-      const pageListings = parseListings($, query.make, query.model);
-
-      allListings.push(...pageListings);
+      // Extract vehicle data from __NEXT_DATA__ script tag
+      const pageListings = await this.page.evaluate((make, model) => {
+        const script = document.getElementById('__NEXT_DATA__');
+        if (!script) return [];
+
+        try {
+          const data = JSON.parse(script.textContent);
+          const vehiclesArray = data?.props?.pageProps?.forProviders?.forInventoryContext?.inventoryData?.inventory?.vehicles || [];
+
+          return vehiclesArray.map(v => ({
+            vin: v.vin,
+            year: v.year,
+            make: v.make,
+            model: v.model,
+            trim: v.trim,
+            price: v.price?.total,
+            mileage: v.mileage,
+            vehicleId: v.id
+          }));
+        } catch (e) {
+          return [];
+        }
+      }, query.make, query.model);
+
+      // Filter to only matching make/model and format
+      const formattedListings = pageListings
+        .filter(v => {
+          if (!v.vin) return false;
+          const makeLower = v.make?.toLowerCase() || '';
+          const modelLower = v.model?.toLowerCase() || '';
+          const queryMakeLower = query.make.toLowerCase();
+          const queryModelLower = query.model.toLowerCase();
+          return makeLower.includes(queryMakeLower) && modelLower.includes(queryModelLower);
+        })
+        .map(v => ({
+          id: `carvana-${v.vehicleId}`,
+          vin: v.vin,
+          make: query.make,
+          model: query.model,
+          year: v.year,
+          trim: v.trim,
+          price: v.price || null,
+          mileage: v.mileage || null,
+          location: 'Carvana',
+          url: `https://www.carvana.com/vehicle/${v.vehicleId}`,
+          listing_date: new Date().toISOString().split('T')[0]
+        }));
+
+      allListings.push(...formattedListings);
 
       // Stop if we've reached the target count
       if (allListings.length >= targetCount) {

scrapers/plattauto/scrape.js

@@ -161,9 +161,10 @@ function parseListings($, make, model) {
       const urlPath = linkElement.attr('href');
       const url = urlPath ? (urlPath.startsWith('http') ? urlPath : `https://www.plattauto.com${urlPath}`) : null;
 
-      // Extract stock number or VIN for ID - required for tracking
+      // Extract stock number and VIN
       const stockText = $card.find('.dws-vehicle-field-stock-number').text().trim().replace(/\s+/g, ' ');
-      const vinText = $card.find('.dws-vehicle-field-vin').text().trim().replace(/\s+/g, ' ');
+      const vinText = $card.find('.dws-vehicle-field-vin').text().trim().replace(/\s+/g, ' ').replace(/^VIN\s+/i, '');
+      const vin = vinText || null;
       const id = stockText || vinText;
       if (!id) {
         console.error(`    ⚠ Warning: Could not extract stock number or VIN from listing`);
@@ -178,6 +179,7 @@ function parseListings($, make, model) {
       seenIds.add(id);
       listings.push({
         id,
+        vin,
         make,
         model,
         year,