ScraperControl/scripts/save-schedules-to-db.ts

#!/usr/bin/env tsx
/**
 * Save mass schedules to database using scrapeChurch() service
 */

import { config } from 'dotenv';
config({ path: '.env.local' });
config({ path: '.env' });

import { scrapeChurch } from '../src/lib/scraper-service';
import { prisma } from '../src/lib/db';

const PRIORITY_COUNTRIES = ['FR', 'DE', 'ES', 'PL', 'BR'];
const CHURCHES_PER_COUNTRY = 5; // Start small to verify it works

interface ScrapeResult {
  churchId: string;
  churchName: string;
  country: string;
  success: boolean;
  schedulesCreated: number;
  error?: string;
}

async function saveSchedulesToDb() {
  console.log('Starting database save operation...\n');
  console.log(`Target: ${CHURCHES_PER_COUNTRY} churches per country`);
  console.log(`Countries: ${PRIORITY_COUNTRIES.join(', ')}\n`);

  const results: ScrapeResult[] = [];
  let totalChurches = 0;
  let totalSuccess = 0;
  let totalSchedules = 0;

  for (const country of PRIORITY_COUNTRIES) {
    console.log(`\n${'='.repeat(60)}`);
    console.log(`${country} - Finding churches to scrape...`);
    console.log('='.repeat(60));

    // Get churches with websites that haven't been scraped yet
    const churches = await prisma.church.findMany({
      where: {
        country,
        website: { not: null },
        source: 'osm',
        lastScrapedAt: null, // Only unscrapped churches
      },
      take: CHURCHES_PER_COUNTRY,
      orderBy: { createdAt: 'asc' },
    });

    console.log(`Found ${churches.length} churches to scrape\n`);

    for (let i = 0; i < churches.length; i++) {
      const church = churches[i];
      totalChurches++;

      process.stdout.write(`[${i + 1}/${churches.length}] ${church.name.substring(0, 40).padEnd(40)} `);

      try {
        // Use the scrapeChurch service which saves to database
        const result = await scrapeChurch(church.id);

        if (result.success) {
          totalSuccess++;
          totalSchedules += result.schedulesCreated;
          process.stdout.write(`✅ ${result.schedulesCreated} schedules saved\n`);

          results.push({
            churchId: church.id,
            churchName: church.name,
            country,
            success: true,
            schedulesCreated: result.schedulesCreated,
          });
        } else {
          process.stdout.write(`❌ ${result.error}\n`);

          results.push({
            churchId: church.id,
            churchName: church.name,
            country,
            success: false,
            schedulesCreated: 0,
            error: result.error,
          });
        }
      } catch (err: any) {
        process.stdout.write(`❌ ERROR: ${err.message}\n`);

        results.push({
          churchId: church.id,
          churchName: church.name,
          country,
          success: false,
          schedulesCreated: 0,
          error: err.message,
        });
      }
    }
  }

  // Final summary
  console.log('\n\n');
  console.log('═'.repeat(80));
  console.log('DATABASE SAVE SUMMARY');
  console.log('═'.repeat(80));
  console.log('');
  console.log(`Total churches processed: ${totalChurches}`);
  console.log(`Successful scrapes: ${totalSuccess} (${((totalSuccess / totalChurches) * 100).toFixed(1)}%)`);
  console.log(`Total schedules saved to database: ${totalSchedules}`);
  console.log('');

  // Verify database records
  console.log('Verifying database records...\n');

  const dbScheduleCount = await prisma.massSchedule.count();
  const dbChurchesWithSchedules = await prisma.church.count({
    where: {
      massSchedules: {
        some: {},
      },
    },
  });

  console.log(`✓ Total mass schedules in database: ${dbScheduleCount}`);
  console.log(`✓ Churches with schedules: ${dbChurchesWithSchedules}`);
  console.log('');

  // Show sample of saved schedules
  console.log('Sample of saved schedules:\n');

  const sampleChurches = await prisma.church.findMany({
    where: {
      massSchedules: {
        some: {},
      },
    },
    include: {
      massSchedules: {
        take: 3,
        orderBy: { dayOfWeek: 'asc' },
      },
    },
    take: 3,
  });

  const dayNames = ['Sun', 'Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat'];

  sampleChurches.forEach(church => {
    console.log(`${church.name} (${church.country}):`);
    church.massSchedules.forEach(schedule => {
      console.log(`  ${dayNames[schedule.dayOfWeek]} ${schedule.time} - ${schedule.language} ${schedule.massType || ''}`);
    });
    console.log('');
  });

  await prisma.$disconnect();
}

saveSchedulesToDb().catch(console.error);