Art-gallery/scripts/seed-wikipedia.js

/**
 * Seed historical eras, art movements, artists, periods, and flagship paintings.
 * Run: npm run seed
 * Flags: --force (re-seed), --fetch-images (download portraits and paintings from Wikipedia)
 */
require('dotenv').config();
const path = require('path');
const pool = require('../server/db');
const { ERAS, MOVEMENTS, buildArtists } = require('./seed-catalog-data');
const { savePaintingImages, saveImageForItem } = require('./image-fetcher');

const FORCE = process.argv.includes('--force');
const FETCH_IMAGES = process.argv.includes('--fetch-images');
const IMAGE_DIR = path.resolve(process.env.IMAGE_DIR || path.join(__dirname, '../data/images'));

async function tableCount(table) {
  const { rows } = await pool.query(`SELECT COUNT(*)::int AS n FROM ${table}`);
  return rows[0].n;
}

async function clearCatalog() {
  await pool.query(`
    TRUNCATE TABLE
      painting_annotations,
      painting_influence_sources,
      painting_influences,
      paintings,
      artist_periods,
      artists,
      art_movements,
      historical_eras
    RESTART IDENTITY CASCADE
  `);
}

async function insertEras() {
  const eraIds = new Map();
  for (const era of ERAS) {
    const { rows } = await pool.query(
      `INSERT INTO historical_eras (name, start_year, end_year, start_definite, end_definite, description, sort_order)
       VALUES ($1, $2, $3, $4, $5, $6, $7)
       RETURNING id`,
      [era.name, era.start_year, era.end_year, era.start_definite, era.end_definite, era.description, era.sort_order]
    );
    eraIds.set(era.name, rows[0].id);
  }
  return eraIds;
}

async function insertMovements(eraIds) {
  const movementIds = new Map();
  for (const mov of MOVEMENTS) {
    const eraId = eraIds.get(mov.era) ?? null;
    const { rows } = await pool.query(
      `INSERT INTO art_movements (name, start_year, end_year, era_id, description, color)
       VALUES ($1, $2, $3, $4, $5, $6)
       RETURNING id`,
      [mov.name, mov.start_year, mov.end_year, eraId, mov.description, mov.color]
    );
    movementIds.set(mov.name, rows[0].id);
  }
  return movementIds;
}

function periodName(artist) {
  if (artist.birth_year != null && artist.death_year != null) {
    return `${artist.birth_year}–${artist.death_year}`;
  }
  return 'Main period';
}

async function insertArtists(movementIds, artists) {
  let inserted = 0;
  let paintings = 0;
  let skippedPaintings = 0;
  let portraitsFetched = 0;
  let imagesFetched = 0;

  for (const artist of artists) {
    const movementId = movementIds.get(artist.movement);
    if (!movementId) {
      console.warn(`  Unknown movement "${artist.movement}" for ${artist.name} — skipped`);
      continue;
    }

    const artistRes = await pool.query(
      `INSERT INTO artists (name, birth_year, death_year, movement_id, wikipedia_title, century)
       VALUES ($1, $2, $3, $4, $5, $6)
       RETURNING id`,
      [artist.name, artist.birth_year, artist.death_year, movementId, artist.wikipedia_title, artist.century]
    );
    const artistId = artistRes.rows[0].id;
    inserted += 1;

    const periodRes = await pool.query(
      `INSERT INTO artist_periods (artist_id, name, start_year, end_year, sort_order)
       VALUES ($1, $2, $3, $4, 0)
       RETURNING id`,
      [artistId, periodName(artist), artist.birth_year, artist.death_year]
    );
    const periodId = periodRes.rows[0].id;

    if (FETCH_IMAGES) {
      try {
        const portraitBase = artist.name.replace(/[^a-zA-Z0-9_-]/g, '_');
        const saved = await saveImageForItem(artist.wikipedia_title, 'portraits', portraitBase, IMAGE_DIR);
        if (saved.path) {
          await pool.query('UPDATE artists SET portrait_path = $1 WHERE id = $2', [saved.path, artistId]);
          portraitsFetched += 1;
        }
      } catch (err) {
        console.warn(`  portrait: ${artist.name}: ${err.message}`);
      }
    }

    if (!artist.painting?.title) {
      skippedPaintings += 1;
      console.warn(`  no flagship painting for ${artist.name}`);
      continue;
    }

    const wikiTitle = artist.painting.wikipedia_title || artist.painting.title;
    const paintingRes = await pool.query(
      `INSERT INTO paintings (artist_id, period_id, title, year, wikipedia_title, sort_order)
       VALUES ($1, $2, $3, $4, $5, 0)
       RETURNING id`,
      [artistId, periodId, artist.painting.title, artist.painting.year ?? null, wikiTitle]
    );
    const paintingId = paintingRes.rows[0].id;
    paintings += 1;

    if (FETCH_IMAGES) {
      try {
        const base = `${artist.name}_${artist.painting.title}`.replace(/[^a-zA-Z0-9_-]/g, '_');
        const saved = await savePaintingImages(wikiTitle, base, IMAGE_DIR, {
          artistName: artist.name,
          paintingTitle: artist.painting.title,
        });
        if (saved.imagePath || saved.thumbnailPath) {
          await pool.query(
            'UPDATE paintings SET image_path = $1, thumbnail_path = $2 WHERE id = $3',
            [saved.imagePath, saved.thumbnailPath, paintingId]
          );
          imagesFetched += 1;
        }
      } catch (err) {
        console.warn(`  image: ${artist.name} — ${artist.painting.title}: ${err.message}`);
      }
    }

    console.log(`+ ${artist.name} (${artist.movement}) — ${artist.painting.title}`);
  }

  return { inserted, paintings, skippedPaintings, portraitsFetched, imagesFetched };
}

async function main() {
  const existing = await tableCount('artists');
  if (existing > 0 && !FORCE) {
    console.log(`Catalog already has ${existing} artists. Use --force to re-seed.`);
    return;
  }

  if (FORCE && existing > 0) {
    console.log('Clearing existing catalog…');
    await clearCatalog();
  }

  const artists = buildArtists();
  console.log(`Seeding ${ERAS.length} eras, ${MOVEMENTS.length} movements, ${artists.length} artists…`);

  const eraIds = await insertEras();
  const movementIds = await insertMovements(eraIds);
  const stats = await insertArtists(movementIds, artists);

  console.log('\nSeed complete.');
  console.log(`  Artists:   ${stats.inserted}`);
  console.log(`  Paintings: ${stats.paintings}${stats.skippedPaintings ? ` (${stats.skippedPaintings} without flagship)` : ''}`);
  if (FETCH_IMAGES) {
    console.log(`  Portraits downloaded: ${stats.portraitsFetched}`);
    console.log(`  Painting images:      ${stats.imagesFetched}`);
  } else {
    console.log('  Run with --fetch-images to download artwork files, or npm run fetch-images later.');
  }
  console.log('\nNext steps:');
  console.log('  npm run fetch-artist-bios');
  console.log('  npm run expand-catalog');
  console.log('  npm run update-influences');
}

main()
  .then(() => pool.end())
  .catch((err) => {
    console.error('Seed failed:', err.message);
    pool.end().finally(() => process.exit(1));
  });