Art-gallery/scripts/find-duplicate-paintings-near.js

require('dotenv').config();
const pool = require('../server/db');

async function main() {
  const sameWiki = await pool.query(`
    SELECT a.name AS artist, p.wikipedia_title, COUNT(*)::int AS cnt,
           array_agg(p.id ORDER BY p.id) AS ids,
           array_agg(p.title ORDER BY p.id) AS titles,
           array_agg(COALESCE(p.year::text, '?') ORDER BY p.id) AS years
    FROM paintings p
    JOIN artists a ON a.id = p.artist_id
    WHERE p.wikipedia_title IS NOT NULL AND p.wikipedia_title <> ''
    GROUP BY a.id, a.name, p.wikipedia_title
    HAVING COUNT(*) > 1
    ORDER BY cnt DESC, a.name
    LIMIT 30
  `);

  const sameImage = await pool.query(`
    SELECT a.name AS artist, p.image_path, COUNT(*)::int AS cnt,
           array_agg(p.id ORDER BY p.id) AS ids,
           array_agg(p.title ORDER BY p.id) AS titles
    FROM paintings p
    JOIN artists a ON a.id = p.artist_id
    WHERE p.image_path IS NOT NULL AND p.image_path <> ''
    GROUP BY a.id, a.name, p.image_path
    HAVING COUNT(*) > 1
    ORDER BY cnt DESC, a.name
    LIMIT 20
  `);

  console.log('=== Same Wikipedia title (different painting rows) ===');
  console.log(`Groups: ${sameWiki.rows.length} shown (top 30)`);
  for (const row of sameWiki.rows) {
    console.log(`\n[${row.cnt}x] ${row.artist} — wiki: "${row.wikipedia_title}"`);
    row.titles.forEach((t, i) => console.log(`  id ${row.ids[i]}: "${t}" (${row.years[i]})`));
  }

  console.log('\n=== Same image file (different painting rows) ===');
  for (const row of sameImage.rows) {
    console.log(`\n[${row.cnt}x] ${row.artist} — ${row.image_path}`);
    row.titles.forEach((t, i) => console.log(`  id ${row.ids[i]}: "${t}"`));
  }

  await pool.end();
}

main().catch((err) => {
  console.error(err);
  process.exit(1);
});