Art-gallery/scripts/fetch-artist-images.js

/**
 * Download artist portrait images from Wikipedia (and fallbacks) into data/images/portraits/.
 * Run: npm run dev:fetch-artist-images
 * Flags: --force (re-fetch even when portrait_path is set), --limit=N
 */
require('dotenv').config();
const fs = require('fs');
const path = require('path');
const pool = require('../server/db');
const { saveArtistPortrait, findLocalPortraitPath } = require('./image-fetcher');

const FORCE = process.argv.includes('--force');
const LIMIT = parseInt(process.argv.find((a) => a.startsWith('--limit='))?.split('=')[1] || '0', 10);
const IMAGE_DIR = path.resolve(process.env.IMAGE_DIR || path.join(__dirname, '../data/images'));

function portraitFileExists(portraitPath) {
  return !!portraitPath && fs.existsSync(path.join(IMAGE_DIR, portraitPath));
}

function needsPortrait(artist, force) {
  if (force) return true;
  const local = findLocalPortraitPath(artist.name, IMAGE_DIR);
  if (local) return artist.portrait_path !== local;
  if (artist.portrait_path) return !portraitFileExists(artist.portrait_path);
  return true;
}

async function main() {
  const { rows: artists } = await pool.query(`
    SELECT id, name, wikipedia_title, portrait_path
    FROM artists
    ORDER BY name
  `);

  let targets = artists.filter((a) => needsPortrait(a, FORCE));
  if (LIMIT > 0) targets = targets.slice(0, LIMIT);

  console.log(
    `Artists total: ${artists.length}, to fetch: ${targets.length}${FORCE ? ' (force)' : ''}${LIMIT > 0 ? ` (limit ${LIMIT})` : ''}`
  );

  let updated = 0;
  let linked = 0;
  let failed = 0;

  for (const artist of targets) {
    try {
      const local = findLocalPortraitPath(artist.name, IMAGE_DIR);
      if (local) {
        await pool.query('UPDATE artists SET portrait_path = $1 WHERE id = $2', [local, artist.id]);
        console.log(`↺ ${artist.name} — ${local}`);
        linked += 1;
        continue;
      }

      const wikiTitle = artist.wikipedia_title || artist.name;
      const saved = await saveArtistPortrait(artist.name, wikiTitle, IMAGE_DIR);
      if (!saved.path) {
        console.warn(`✗ ${artist.name} — no portrait found`);
        failed += 1;
        continue;
      }

      await pool.query('UPDATE artists SET portrait_path = $1 WHERE id = $2', [saved.path, artist.id]);
      console.log(`✓ ${artist.name} — ${saved.path} (${saved.source || 'Wikipedia'})`);
      updated += 1;
    } catch (err) {
      console.error(`✗ ${artist.name} — ${err.message}`);
      failed += 1;
    }
  }

  console.log(`\nDone: ${updated} downloaded, ${linked} linked from disk, ${failed} failed`);
  await pool.end();
}

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