ScraperControl/scripts/test-url-discovery.ts

import { discoverMassScheduleUrl } from '../src/scrapers/url-discovery';

const TEST_SITES = [
  'https://www.saintpatrickscathedral.org',
  'https://www.holynamecathedral.org',
  'https://www.olacathedral.org',
];

const CONFIDENCE_ICONS: Record<string, string> = {
  high: '🟢',
  medium: '🟡',
  low: '🔴',
};

const METHOD_DESCRIPTIONS: Record<string, string> = {
  pattern: 'Found via URL pattern matching',
  link: 'Found via link crawling',
  homepage: 'Fell back to homepage',
};

async function testSingleUrl(url: string) {
  console.log('\n' + '='.repeat(70));
  console.log('NEARESTMASS URL DISCOVERY TEST');
  console.log('='.repeat(70));
  console.log(`\nURL: ${url}`);
  console.log(`Time: ${new Date().toISOString()}`);
  console.log('\n' + '-'.repeat(70));

  console.log('\n[1/2] Discovering mass schedule URL...');
  const startTime = Date.now();
  const result = await discoverMassScheduleUrl(url);
  const elapsed = Date.now() - startTime;
  console.log(`      ✓ Discovery completed in ${elapsed}ms`);

  console.log('\n[2/2] Results:');
  console.log(`      Discovered URL: ${result.url}`);
  console.log(`      Method: ${result.method} (${METHOD_DESCRIPTIONS[result.method]})`);
  console.log(`      Confidence: ${CONFIDENCE_ICONS[result.confidence]} ${result.confidence}`);

  console.log('\n' + '='.repeat(70));
  console.log('SUMMARY');
  console.log('='.repeat(70));
  console.log(`Input:      ${url}`);
  console.log(`Output:     ${result.url}`);
  console.log(`Method:     ${result.method}`);
  console.log(`Confidence: ${result.confidence}`);
  console.log(`Time:       ${elapsed}ms`);
  console.log('='.repeat(70) + '\n');
}

async function testMultipleSites() {
  console.log('\n' + '='.repeat(70));
  console.log('NEARESTMASS URL DISCOVERY TEST (BATCH)');
  console.log('='.repeat(70));
  console.log(`\nTesting ${TEST_SITES.length} sites...`);
  console.log(`Time: ${new Date().toISOString()}`);

  const results: Array<{
    site: string;
    url: string;
    method: string;
    confidence: string;
    elapsed: number;
  }> = [];

  for (let i = 0; i < TEST_SITES.length; i++) {
    const site = TEST_SITES[i];
    console.log('\n' + '-'.repeat(70));
    console.log(`[${i + 1}/${TEST_SITES.length}] Testing: ${site}`);
    console.log('-'.repeat(70));

    const startTime = Date.now();
    const result = await discoverMassScheduleUrl(site);
    const elapsed = Date.now() - startTime;

    console.log(`\n      Discovered URL: ${result.url}`);
    console.log(`      Method: ${result.method} (${METHOD_DESCRIPTIONS[result.method]})`);
    console.log(`      Confidence: ${CONFIDENCE_ICONS[result.confidence]} ${result.confidence}`);
    console.log(`      Time: ${elapsed}ms`);

    results.push({
      site,
      url: result.url,
      method: result.method,
      confidence: result.confidence,
      elapsed,
    });

    // Rate limiting between sites
    if (i < TEST_SITES.length - 1) {
      console.log('\n      Waiting 2s before next site...');
      await new Promise((r) => setTimeout(r, 2000));
    }
  }

  // Summary table
  console.log('\n' + '='.repeat(70));
  console.log('SUMMARY');
  console.log('='.repeat(70));

  const highCount = results.filter((r) => r.confidence === 'high').length;
  const mediumCount = results.filter((r) => r.confidence === 'medium').length;
  const lowCount = results.filter((r) => r.confidence === 'low').length;
  const totalTime = results.reduce((sum, r) => sum + r.elapsed, 0);

  console.log(`\nSites tested:  ${results.length}`);
  console.log(`High conf:     ${highCount} 🟢`);
  console.log(`Medium conf:   ${mediumCount} 🟡`);
  console.log(`Low conf:      ${lowCount} 🔴`);
  console.log(`Total time:    ${totalTime}ms`);

  console.log('\n' + '-'.repeat(70));
  console.log('RESULTS BY SITE');
  console.log('-'.repeat(70));

  for (const r of results) {
    console.log(`\n${r.site}`);
    console.log(`  → ${r.url}`);
    console.log(`    ${CONFIDENCE_ICONS[r.confidence]} ${r.confidence} via ${r.method}`);
  }

  console.log('\n' + '='.repeat(70) + '\n');
}

async function main() {
  const testUrl = process.argv[2];

  if (testUrl) {
    await testSingleUrl(testUrl);
  } else {
    await testMultipleSites();
  }
}

main().catch(console.error);