Art-gallery/scripts/backfill-updated-at.js

/**
 * Backfill catalog updated_at from linked image file mtimes where possible.
 *
 * Dev:  npm run dev:backfill-updated-at
 * Prod: npm run harmonize:backfill-updated-at
 */
const fs = require('fs');
const path = require('path');
const pg = require('pg');
const {
  assertDevDatabase,
  assertProdDatabase,
  loadDevPgConfig,
  loadProdPgConfig,
  rootDir,
} = require('./db-env');
const { printCliResult } = require('./lib/cli-result');

const { Client } = pg;

function parseArgs(argv) {
  return { prod: argv.includes('--prod') };
}

function resolveImageDir(prod) {
  if (prod) {
    const envPath = path.join(rootDir, 'infra', 'docker', '.env.prod');
    if (fs.existsSync(envPath)) {
      for (const line of fs.readFileSync(envPath, 'utf8').split(/\r?\n/)) {
        const trimmed = line.trim();
        if (trimmed.startsWith('IMAGE_DIR=')) {
          const val = trimmed.slice('IMAGE_DIR='.length).trim().replace(/^["']|["']$/g, '');
          if (val) return path.resolve(val);
        }
      }
    }
    return path.resolve('\\\\192.168.10.122\\Gallery\\data\\images');
  }
  require('dotenv').config({ path: path.join(rootDir, '.env') });
  return path.resolve(process.env.IMAGE_DIR || path.join(rootDir, 'data', 'images'));
}

function fileMtimeMs(imageDir, relPath) {
  if (!relPath || typeof relPath !== 'string') return null;
  const abs = path.join(imageDir, relPath.replace(/^\//, ''));
  try {
    if (!fs.existsSync(abs)) return null;
    return fs.statSync(abs).mtimeMs;
  } catch {
    return null;
  }
}

function maxMtime(imageDir, paths) {
  let max = null;
  for (const rel of paths) {
    const ms = fileMtimeMs(imageDir, rel);
    if (ms != null && (max == null || ms > max)) max = ms;
  }
  return max;
}

async function main() {
  const { prod } = parseArgs(process.argv.slice(2));
  const config = prod ? loadProdPgConfig() : loadDevPgConfig();
  if (prod) assertProdDatabase(config.database);
  else assertDevDatabase(config.database);

  const imageDir = resolveImageDir(prod);
  const client = new Client(config);
  await client.connect();

  const fallback = new Date();
  let updated = 0;

  const { rows: artists } = await client.query(
    'SELECT id, portrait_path, portrait_thumb_path FROM artists',
  );
  for (const row of artists) {
    const ms = maxMtime(imageDir, [row.portrait_path, row.portrait_thumb_path]);
    const ts = ms != null ? new Date(ms) : fallback;
    await client.query('UPDATE artists SET updated_at = $1 WHERE id = $2', [ts, row.id]);
    updated += 1;
  }

  const { rows: paintings } = await client.query(
    'SELECT id, image_path, thumbnail_path FROM paintings',
  );
  for (const row of paintings) {
    const ms = maxMtime(imageDir, [row.image_path, row.thumbnail_path]);
    const ts = ms != null ? new Date(ms) : fallback;
    await client.query('UPDATE paintings SET updated_at = $1 WHERE id = $2', [ts, row.id]);
    updated += 1;
  }

  const childUpdates = [
    ['artist_periods', 'artist_id', 'artists'],
    ['painting_influences', 'painting_id', 'paintings'],
    ['painting_influence_sources', 'painting_id', 'paintings'],
    ['painting_annotations', 'painting_id', 'paintings'],
  ];

  for (const [child, fk, parent] of childUpdates) {
    const res = await client.query(
      `UPDATE ${child} c
       SET updated_at = GREATEST(c.updated_at, p.updated_at)
       FROM ${parent} p
       WHERE c.${fk} = p.id`,
    );
    updated += res.rowCount || 0;
  }

  for (const table of ['historical_eras', 'art_movements']) {
    const res = await client.query(
      `UPDATE ${table} SET updated_at = $1 WHERE updated_at IS NOT NULL`,
      [fallback],
    );
    updated += res.rowCount || 0;
  }

  await client.query(
    `UPDATE art_movements m
     SET updated_at = GREATEST(m.updated_at, e.updated_at)
     FROM historical_eras e
     WHERE m.era_id = e.id`,
  );

  await client.query(
    `UPDATE artists a
     SET updated_at = GREATEST(a.updated_at, m.updated_at)
     FROM art_movements m
     WHERE a.movement_id = m.id`,
  );

  await client.end();

  printCliResult({
    script: 'backfill-updated-at',
    ok: true,
    summary: `Backfilled updated_at on ${config.database} (${updated} row touches).`,
    details: [`Image dir: ${imageDir}`],
  });
}

main().catch((err) => {
  printCliResult({
    script: 'backfill-updated-at',
    ok: false,
    summary: err.message,
  });
});