Added parsing for vins

index.js

@@ -36,8 +36,10 @@ app.post('/convertGalleryToHar', async (req, res) => {
   try {
     const payloads = await run(url, processor);
     res.status(200).json({
+      vin: payloads.vin,
+      mileage: payloads.mileage,
       log: {
-        entries: payloads
+        entries: payloads.payloads
       }
       // payloads
     })
@@ -65,7 +67,12 @@ async function run(url, processor) {
   await page.setUserAgent("Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/44.0.2403.157 Safari/537.36");
   console.log('Loading page...');
   await page.goto(url, { timeout: 60000 });
-  console.log('page loaded')
+  console.log('page loaded');
+  console.log('attempting to parse fields');
+  const vin = await processor.parseVIN(page);
+  console.log('parsed VIN:', vin);
+  const mileage = await processor.parseMileage(page);
+  console.log('parsed Mileage:', mileage);
   const galleryUrls = await processor.execute(page);
 
 
@@ -90,7 +97,11 @@ async function run(url, processor) {
   })))
   console.log('URLS done downloading')
   await browser.close();
-  return payloads;
+  return {
+    vin,
+    payloads,
+    mileage
+  };
 }
 
 // run();

processors/barrettjackson.processor.js

@@ -1,9 +1,8 @@
 //sample url = https://BarrettJacksonCDN.azureedge.net/staging/carlist/items/Fullsize/Cars/263634/263634_Rear_3-4_Web.jpg
 //sample urlt= https://BarrettJacksonCDN.azureedge.net/staging/carlist/items/Thumbnails/Cars/263634/263634_Misc_6ca0ca58-2b21-482d-ba01-4078083b5520_T.jpg
 
+const genericVinParserFactory = require("./generic-vin-parser");
 
-const _ = require('lodash');
-const path = require('path');
 module.exports = {
     baseUrl: 'www.barrett-jackson.com',
     execute: async function (page) {
@@ -21,6 +20,12 @@ module.exports = {
         const sources = sourcesFromThumbnails;
         
        return sources;
+    },
+    parseVIN: genericVinParserFactory({
+        vinElementSelector: `span#VIN`
+    }),
+    parseMileage: async function (page) {
+        return null;
     }
 }
 function convertThumbnailUrlToFullSize(url) {

processors/bringatrailer.processor.js

@@ -1,5 +1,6 @@
-const genericBootstrapFactory = require("./generic-bootstrap")
+const genericBootstrapFactory = require("./generic-bootstrap");
-
+const genericVinParserFactory = require("./generic-vin-parser");
+const _ = require('lodash')
 module.exports = {
     execute: genericBootstrapFactory({
         pageLoadIndicator: '.gallery',
@@ -7,5 +8,11 @@ module.exports = {
         carouselTrigger: '.gallery > a:nth-child(1)'
 
     }),
-    baseUrl: 'bringatrailer.com'
+    baseUrl: 'bringatrailer.com',
+    parseVIN: genericVinParserFactory({
+        vinElementSelector: `div.item ul li`
+    }),
+    parseMileage: async function (page) {
+        return null;
+    }
 }

processors/classiccars.processor.js

@@ -1,4 +1,7 @@
-const genericBootstrapFactory = require("./generic-bootstrap")
+const genericBootstrapFactory = require("./generic-bootstrap");
+const genericVinParserFactory = require("./generic-vin-parser");
+
+const _ = require('lodash');
 
 module.exports = {
   execute: genericBootstrapFactory({
@@ -7,5 +10,12 @@ module.exports = {
     vinSelector: 'li.p-vin > span:nth-child(2)',
     carouselTrigger: 'div.swiper-slide-active > div > img.u-photo'
   }),
-    baseUrl :'classiccars.com'
+  baseUrl: 'classiccars.com',
+  parseVIN: genericVinParserFactory({
+    vinElementSelector: `li.border-btm.p-vin span.w40.d-inline-block.b.fs-14.gray`
+  }),
+  parseMileage: async function (page) {
+    return null;
+  }
+  //li.border-btm.p-vin span.w40.d-inline-block.b.fs-14.gray
 }

processors/davidsclassiccars.processor.js

@@ -1,6 +1,9 @@
+const _ = require('lodash');
+const genericVinParserFactory = require("./generic-vin-parser");
+
 module.exports = {
     baseUrl: 'davidsclassiccars.com',
-    execute: async function(page) {
+    execute: async function (page) {
         const pageLoadIndicator = '.bx-viewport';
         await page.waitForSelector(pageLoadIndicator);
         const imageSelector = '.carimage > img';
@@ -13,5 +16,12 @@ module.exports = {
         }));
         return sources;
 
+    },
+    parseVIN: genericVinParserFactory({
+        vinElementSelector: `table.car-features tbody tr td.options-full span`
+    }),
+    parseMileage: async function (page) {
+        return null;
     }
+   
 }

processors/ebay.processor.js

@@ -1,5 +1,5 @@
-const _ = require('lodash');
+
-const path = require('path');
+const genericVinParserFactory = require("./generic-vin-parser");
 module.exports = {
     baseUrl: 'ebay.com',
     execute: async function (page) {
@@ -77,6 +77,12 @@ module.exports = {
             
             return sources;
         }
+    },
+    parseVIN: genericVinParserFactory({
+        vinElementSelector: `div.ux-layout-section__row div.ux-labels-values__values div.ux-labels-values__values-content div span.ux-textspans`
+    }),
+    parseMileage: async function (page) {
+        return null;
     }
 }
 function convertThumbnailUrlToFullSize(url) {

processors/generic-vin-parser.js

@@ -0,0 +1,12 @@
+const _ = require('lodash');
+module.exports = function(config) {
+    return async function (page) {
+        const vinSelector = config.vinElementSelector;
+        const vinRegex = /(?<vin>A\d\w\d{3}\w\d{6})/i;
+        const possibleVinElements = await page.$$(vinSelector);
+        const evaluatedVinElements = await Promise.all(possibleVinElements.map(async element => await page.evaluate(el => el.textContent, element)));
+        const correctElement = _.find(evaluatedVinElements, element => vinRegex.test(element));
+    
+        return correctElement ? vinRegex.exec(correctElement).groups.vin : null;
+    }
+}

processors/hemmings.processor.js

@@ -1,4 +1,6 @@
-const genericBootstrapFactory = require("./generic-bootstrap")
+const genericBootstrapFactory = require("./generic-bootstrap");
+const genericVinParserFactory = require("./generic-vin-parser");
+
 
 module.exports = {
     execute: genericBootstrapFactory({
@@ -7,6 +9,12 @@ module.exports = {
         vinSelector: '.leading-loose > li:nth-child(2) > span:nth-child(2)',
         carouselTrigger: '.bg-center'
       }),
-    baseUrl :'hemmings.com'
+    baseUrl :'hemmings.com',
+    parseVIN: genericVinParserFactory({
+      vinElementSelector: `div.py-4.px-4.flex.flex-col.lg:flex-row.flex-wrap div.w-full.lg:w-1/2.my-2 div.flex.flex-col div.text-lg.lg:text-xl.whitespace-normal`
+    }),
+    parseMileage: async function (page) {
+      return null;
+    }
 }
 

processors/topclassiccarsforsale.processor.js

@@ -1,3 +1,4 @@
+const genericVinParserFactory = require("./generic-vin-parser");
 module.exports = {
     baseUrl: 'topclassiccarsforsale.com',
     async execute(page) {
@@ -11,5 +12,11 @@ module.exports = {
             return { url: this.baseUrl + src };
         }));
         return sources;
+    },
+    parseVIN: genericVinParserFactory({
+        vinElementSelector: `ul.fullinfo li`
+    }),
+    parseMileage: async function (page) {
+        return null;
     }
 }