clawdbot-workspace/reonomy-scraper-working.js

#!/usr/bin/env node

/**
 * Reonomy Lead Scraper - Working JSON Fallback Version
 *
 * Extracts property and owner leads from Reonomy dashboard/search
 * and saves to JSON (no Google Sheets dependency).
 */

const puppeteer = require('puppeteer');
const fs = require('fs');
const path = require('path');

// Configuration
const REONOMY_EMAIL = process.env.REONOMY_EMAIL || 'henry@realestateenhanced.com';
const REONOMY_PASSWORD = process.env.REONOMY_PASSWORD || '9082166532';
const SEARCH_LOCATION = process.env.REONOMY_LOCATION || 'New York, NY';
const HEADLESS = process.env.HEADLESS === 'true';

// Output file
const OUTPUT_FILE = path.join(__dirname, 'reonomy-leads.json');
const LOG_FILE = path.join(__dirname, 'reonomy-scraper.log');

function log(message) {
  const timestamp = new Date().toISOString();
  const logMessage = `[${timestamp}] ${message}\n`;
  console.log(message);
  fs.appendFileSync(LOG_FILE, logMessage);
}

function sleep(ms) {
  return new Promise(resolve => setTimeout(resolve, ms));
}

/**
 * Save leads to JSON file
 */
function saveLeads(leads) {
  const data = {
    scrapeDate: new Date().toISOString(),
    leadCount: leads.length,
    location: SEARCH_LOCATION,
    leads: leads
  };

  try {
    fs.writeFileSync(OUTPUT_FILE, JSON.stringify(data, null, 2));
    log(`💾 Saved ${leads.length} leads to ${OUTPUT_FILE}`);
    return OUTPUT_FILE;
  } catch (error) {
    log(`❌ Error saving to JSON: ${error.message}`);
    return null;
  }
}

/**
 * Extract properties from page
 */
async function extractProperties(page) {
  log('🔍 Extracting property data...');

  const properties = await page.evaluate(() => {
    const results = [];

    const propertyLinks = Array.from(document.querySelectorAll('a[href*="/property/"]'));

    propertyLinks.forEach(link => {
      const text = (link.innerText || link.textContent || '').trim();

      const addressMatch = text.match(/^(\d+.+),\s*([A-Za-z\s]+),\s*([A-Z]{2})\s*(\d{5})/);

      if (addressMatch) {
        results.push({
          fullText: text,
          address: addressMatch[1].trim(),
          city: addressMatch[2].trim(),
          state: addressMatch[3].trim(),
          zip: addressMatch[4].trim(),
          url: link.href,
          remainingText: text.substring(addressMatch[0].length).trim()
        });
      }
    });

    return results;
  });

  const scrapeDate = new Date().toISOString().split('T')[0];
  const leads = [];

  for (const prop of properties) {
    const sqFtMatch = prop.remainingText.match(/(\d+\.?\d*)\s*k?\s*SF/i);
    const sqFt = sqFtMatch ? sqFtMatch[0] : '';
    const propertyType = prop.remainingText.replace(sqFt, '').trim() || '';

    const lead = {
      scrapeDate,
      ownerName: '',
      propertyAddress: prop.address,
      city: prop.city,
      state: prop.state,
      zip: prop.zip,
      propertyType,
      squareFootage: sqFt,
      ownerLocation: '',
      propertyCount: '',
      propertyUrl: prop.url,
      ownerUrl: '',
      email: '',
      phone: ''
    };

    leads.push(lead);
  }

  log(`✅ Extracted ${leads.length} properties`);
  return leads;
}

/**
 * Extract owners from page
 */
async function extractOwners(page) {
  log('🔍 Extracting owner data...');

  const owners = await page.evaluate(() => {
    const results = [];

    const ownerLinks = Array.from(document.querySelectorAll('a[href*="/person/"]'));

    ownerLinks.forEach(link => {
      const text = (link.innerText || link.textContent || '').trim();

      const lines = text.split('\n').map(l => l.trim()).filter(l => l);

      if (lines.length >= 2) {
        const ownerName = lines[0];
        const location = lines.find(l => l.includes(',')) || '';
        const propertyCountMatch = text.match(/(\d+)\s*propert/i);
        const propertyCount = propertyCountMatch ? propertyCountMatch[1] : '';

        results.push({
          ownerName,
          location,
          propertyCount,
          url: link.href,
          fullText: text
        });
      }
    });

    return results;
  });

  const scrapeDate = new Date().toISOString().split('T')[0];
  const leads = [];

  for (const owner of owners) {
    let city = '';
    let state = '';
    let ownerLocation = owner.location;

    if (ownerLocation.includes(',')) {
      const parts = ownerLocation.split(',').map(p => p.trim());

      if (parts.length >= 2 && /^[A-Z]{2}$/.test(parts[parts.length - 1])) {
        state = parts[parts.length - 1];
        const cityWithPrefix = parts[parts.length - 2];
        const cityMatch = cityWithPrefix.match(/(\b[A-Z][a-z]+(?:\s+[A-Z][a-z]+)*)$/);
        city = cityMatch ? cityMatch[1] : '';
      } else if (parts.length === 2) {
        city = parts[0];
        state = parts[1];
      }
    }

    const lead = {
      scrapeDate,
      ownerName: owner.ownerName,
      propertyAddress: '',
      city,
      state,
      zip: '',
      propertyType: '',
      squareFootage: '',
      ownerLocation: owner.location,
      propertyCount: owner.propertyCount,
      propertyUrl: '',
      ownerUrl: owner.url,
      email: '',
      phone: ''
    };

    leads.push(lead);
  }

  log(`✅ Extracted ${leads.length} owners`);
  return leads;
}

/**
 * Main scraper
 */
async function scrapeLeads() {
  log('🚀 Starting Reonomy Lead Scraper (JSON Fallback Mode)...\n');

  const browser = await puppeteer.launch({
    headless: HEADLESS ? 'new' : false,
    args: ['--no-sandbox', '--disable-setuid-sandbox', '--window-size=1920,1080']
  });

  const page = await browser.newPage();
  await page.setViewport({ width: 1920, height: 1080 });

  try {
    // Login
    log('\n📍 Step 1: Logging into Reonomy...');
    await page.goto('https://app.reonomy.com/#!/account', {
      waitUntil: 'domcontentloaded',
      timeout: 60000
    });

    await sleep(2000);

    await page.type('input[type="email"]', REONOMY_EMAIL, { delay: 100 });
    await page.type('input[type="password"]', REONOMY_PASSWORD, { delay: 100 });

    await page.click('button[type="submit"]');
    log('⏳ Logging in...');

    await sleep(8000);

    const url = page.url();
    if (url.includes('login') || url.includes('auth')) {
      throw new Error('Login failed. Please check credentials.');
    }

    log('✅ Successfully logged in!');

    // Navigate to search
    log('\n📍 Step 2: Navigating to search...');
    await page.goto('https://app.reonomy.com/#!/search', {
      waitUntil: 'networkidle2',
      timeout: 60000
    });

    await sleep(3000);
    log('✅ On search page');

    // Search
    log(`\n📍 Step 3: Searching for: ${SEARCH_LOCATION}...`);

    const searchInput = await page.waitForSelector('input[placeholder*="address"], input[placeholder*="Search"]', {
      timeout: 10000
    }).catch(() => {
      return page.waitForSelector('input[type="text"]', { timeout: 5000 });
    });

    if (searchInput) {
      await searchInput.click({ clickCount: 3 });
      await searchInput.type(SEARCH_LOCATION, { delay: 100 });
      await sleep(1000);
      await page.keyboard.press('Enter');
      log('⏳ Searching...');
      await sleep(5000);
    }

    // Extract leads
    log('\n📍 Step 4: Extracting lead data...');
    const allLeads = [];

    const properties = await extractProperties(page);
    allLeads.push(...properties);

    const owners = await extractOwners(page);
    allLeads.push(...owners);

    log(`\n✅ Total leads extracted: ${allLeads.length}`);

    if (allLeads.length === 0) {
      log('\n⚠️  No leads found. Taking screenshot for debugging...');
      await page.screenshot({ path: '/tmp/reonomy-no-leads.png', fullPage: true });
      log('📸 Screenshot saved: /tmp/reonomy-no-leads.png');
    } else {
      // Save to JSON
      log('\n📍 Step 5: Saving leads to JSON file...');
      saveLeads(allLeads);
    }

    log('\n✅ Scraping complete!');
    log(`💾 Leads saved to: ${OUTPUT_FILE}`);
    log(`📝 Log file: ${LOG_FILE}`);

    return { leadCount: allLeads.length, outputFile: OUTPUT_FILE };

  } catch (error) {
    log(`\n❌ Error: ${error.message}`);
    log(error.stack);

    try {
      await page.screenshot({ path: '/tmp/reonomy-error.png', fullPage: true });
      log('📸 Error screenshot saved: /tmp/reonomy-error.png');
    } catch (e) {}

    throw error;

  } finally {
    await browser.close();
    log('\n🔚 Browser closed');
  }
}

// Run
scrapeLeads()
  .then(result => {
    log(`\n🎉 Success! ${result.leadCount} leads scraped.`);
    console.log(`\n💾 View your leads at: ${result.outputFile}`);
    process.exit(0);
  })
  .catch(error => {
    log(`\n💥 Scraper failed: ${error.message}`);
    process.exit(1);
  });