cannaiq/backend/archive/enrich-azdhs-from-google-maps.ts

import { firefox } from 'playwright';
import { pool } from './src/db/migrate';
import { getRandomProxy } from './src/utils/proxyManager';

interface DispensaryEnrichment {
  id: number;
  azdhs_name: string;
  address: string;
  city: string;
  state: string;
  zip: string;
  dba_name?: string;
  website?: string;
  google_phone?: string;
  google_rating?: number;
  google_review_count?: number;
  confidence: 'high' | 'medium' | 'low';
  notes?: string;
}

async function enrichFromGoogleMaps() {
  console.log('🦊 Enriching AZDHS dispensaries from Google Maps using Firefox\n');

  // Get a proxy
  const proxy = await getRandomProxy();
  if (!proxy) {
    console.log('❌ No proxies available');
    await pool.end();
    return;
  }

  console.log(`🔌 Using proxy: ${proxy.server}\n`);

  const browser = await firefox.launch({
    headless: true,
    firefoxUserPrefs: {
      'geo.enabled': true,
      'geo.provider.use_corelocation': true,
      'geo.prompt.testing': true,
      'geo.prompt.testing.allow': true,
    }
  });

  const contextOptions: any = {
    viewport: { width: 1920, height: 1080 },
    userAgent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:121.0) Gecko/20100101 Firefox/121.0',
    geolocation: { latitude: 33.4484, longitude: -112.0740 }, // Phoenix, AZ
    permissions: ['geolocation'],
    proxy: {
      server: proxy.server,
      username: proxy.username,
      password: proxy.password
    }
  };

  const context = await browser.newContext(contextOptions);
  const page = await context.newPage();

  try {
    // Get all dispensaries that don't have website yet
    const result = await pool.query(`
      SELECT id, slug, name, address, city, state, zip, phone, website, dba_name
      FROM dispensaries
      WHERE website IS NULL OR website = ''
      ORDER BY id
      LIMIT 50
    `);

    const dispensaries = result.rows;
    console.log(`📋 Found ${dispensaries.length} dispensaries to enrich\n`);

    let changesCreated = 0;
    let failed = 0;
    let skipped = 0;

    for (const disp of dispensaries) {
      console.log(`\n🔍 Processing: ${disp.name}`);
      console.log(`   Address: ${disp.address}, ${disp.city}, ${disp.state} ${disp.zip}`);

      try {
        // Search Google Maps with dispensary name + address for better results
        const searchQuery = `${disp.name} ${disp.address}, ${disp.city}, ${disp.state} ${disp.zip}`;
        const encodedQuery = encodeURIComponent(searchQuery);
        const url = `https://www.google.com/maps/search/${encodedQuery}`;

        console.log(`   📍 Searching Maps: ${searchQuery}`);
        await page.goto(url, {
          waitUntil: 'domcontentloaded',
          timeout: 30000
        });

        // Wait for results
        await page.waitForTimeout(3000);

        // Extract business data from the first result
        const businessData = await page.evaluate(() => {
          const data: any = {};

          // Try to find the place name from the side panel
          const nameSelectors = [
            'h1[class*="fontHeadline"]',
            'h1.DUwDvf',
            '[data-item-id*="name"] h1'
          ];

          for (const selector of nameSelectors) {
            const el = document.querySelector(selector);
            if (el?.textContent) {
              data.name = el.textContent.trim();
              break;
            }
          }

          // Try to find website
          const websiteSelectors = [
            'a[data-item-id="authority"]',
            'a[data-tooltip="Open website"]',
            'a[aria-label*="Website"]'
          ];

          for (const selector of websiteSelectors) {
            const el = document.querySelector(selector) as HTMLAnchorElement;
            if (el?.href && !el.href.includes('google.com')) {
              data.website = el.href;
              break;
            }
          }

          // Try to find phone
          const phoneSelectors = [
            'button[data-item-id*="phone"]',
            'button[aria-label*="Phone"]',
            '[data-tooltip*="Copy phone number"]'
          ];

          for (const selector of phoneSelectors) {
            const el = document.querySelector(selector);
            if (el?.textContent) {
              const phoneMatch = el.textContent.match(/\(?\d{3}\)?[-.\s]?\d{3}[-.\s]?\d{4}/);
              if (phoneMatch) {
                data.phone = phoneMatch[0];
                break;
              }
            }
          }

          // Try to find rating
          const ratingEl = document.querySelector('[role="img"][aria-label*="stars"]');
          if (ratingEl) {
            const label = ratingEl.getAttribute('aria-label');
            const match = label?.match(/(\d+\.?\d*)\s*stars?/);
            if (match) {
              data.rating = parseFloat(match[1]);
            }
          }

          // Try to find review count
          const reviewEl = document.querySelector('[aria-label*="reviews"]');
          if (reviewEl) {
            const label = reviewEl.getAttribute('aria-label');
            const match = label?.match(/([\d,]+)\s*reviews?/);
            if (match) {
              data.reviewCount = parseInt(match[1].replace(/,/g, ''));
            }
          }

          return data;
        });

        console.log(`   Found data:`, businessData);

        // Determine confidence level
        let confidence: 'high' | 'medium' | 'low' = 'low';
        if (businessData.name && businessData.website && businessData.phone) {
          confidence = 'high';
        } else if (businessData.name && (businessData.website || businessData.phone)) {
          confidence = 'medium';
        }

        // Track if any changes were made for this dispensary
        let changesMadeForDispensary = 0;

        // Create change records for each field that has new data
        if (businessData.name && businessData.name !== disp.dba_name) {
          await pool.query(`
            INSERT INTO dispensary_changes (
              dispensary_id, field_name, old_value, new_value,
              confidence_score, source, change_notes
            ) VALUES ($1, $2, $3, $4, $5, $6, $7)
          `, [
            disp.id,
            'dba_name',
            disp.dba_name || null,
            businessData.name,
            confidence,
            'google_maps',
            `Found via Google Maps search for "${disp.name}"`
          ]);
          console.log(`   📝 Created change record for DBA name`);
          changesMadeForDispensary++;
        }

        if (businessData.website && businessData.website !== disp.website) {
          await pool.query(`
            INSERT INTO dispensary_changes (
              dispensary_id, field_name, old_value, new_value,
              confidence_score, source, change_notes, requires_recrawl
            ) VALUES ($1, $2, $3, $4, $5, $6, $7, $8)
          `, [
            disp.id,
            'website',
            disp.website || null,
            businessData.website,
            confidence,
            'google_maps',
            `Found via Google Maps search for "${disp.name}"`,
            true
          ]);
          console.log(`   📝 Created change record for website (requires recrawl)`);
          changesMadeForDispensary++;
        }

        if (businessData.phone && businessData.phone !== disp.phone) {
          await pool.query(`
            INSERT INTO dispensary_changes (
              dispensary_id, field_name, old_value, new_value,
              confidence_score, source, change_notes
            ) VALUES ($1, $2, $3, $4, $5, $6, $7)
          `, [
            disp.id,
            'phone',
            disp.phone || null,
            businessData.phone,
            confidence,
            'google_maps',
            `Found via Google Maps search for "${disp.name}"`
          ]);
          console.log(`   📝 Created change record for phone`);
          changesMadeForDispensary++;
        }

        if (businessData.rating) {
          await pool.query(`
            INSERT INTO dispensary_changes (
              dispensary_id, field_name, old_value, new_value,
              confidence_score, source, change_notes
            ) VALUES ($1, $2, $3, $4, $5, $6, $7)
          `, [
            disp.id,
            'google_rating',
            null,
            businessData.rating.toString(),
            confidence,
            'google_maps',
            `Google rating from Maps search`
          ]);
          console.log(`   📝 Created change record for Google rating`);
          changesMadeForDispensary++;
        }

        if (businessData.reviewCount) {
          await pool.query(`
            INSERT INTO dispensary_changes (
              dispensary_id, field_name, old_value, new_value,
              confidence_score, source, change_notes
            ) VALUES ($1, $2, $3, $4, $5, $6, $7)
          `, [
            disp.id,
            'google_review_count',
            null,
            businessData.reviewCount.toString(),
            confidence,
            'google_maps',
            `Google review count from Maps search`
          ]);
          console.log(`   📝 Created change record for Google review count`);
          changesMadeForDispensary++;
        }

        if (changesMadeForDispensary > 0) {
          console.log(`   ✅ Created ${changesMadeForDispensary} change record(s) for review (${confidence} confidence)`);
          changesCreated += changesMadeForDispensary;
        } else {
          console.log(`   ⏭️  No new data found`);
          skipped++;
        }

      } catch (error) {
        console.log(`   ❌ Error: ${error}`);
        failed++;
      }

      // Rate limiting - wait between requests
      await page.waitForTimeout(3000 + Math.random() * 2000);
    }

    console.log('\n' + '='.repeat(80));
    console.log(`\n📊 Summary:`);
    console.log(`   📝 Change records created: ${changesCreated}`);
    console.log(`   ⏭️  Skipped (no new data): ${skipped}`);
    console.log(`   ❌ Failed: ${failed}`);
    console.log(`\n💡 Visit the Change Approval page to review and approve these changes.`);

  } finally {
    await browser.close();
    await pool.end();
  }
}

async function main() {
  try {
    await enrichFromGoogleMaps();
  } catch (error) {
    console.error('Fatal error:', error);
    process.exit(1);
  }
}

main();