Drupal 9 - unused / unmanaged unsupervised file cleanup script.

file_cleanup.php

<?php

/**
 * @file
 * File cleanup script, to be executed after the migration.
 *
 * Invocation:
 * ddev drush scr web/modules/custom/server_migrate/file_cleanup.php .
 */

$start = time();

// Let's gather all the unused files.
// Unused here means that it is not referenced in a file reference field
// and not used by a media. This covers all the use-cases.
// Beware that `file_usage` might not be bulletproof.
$file_ids = \Drupal::database()->query("
SELECT
  fm.fid
FROM
  {file_managed} fm
LEFT JOIN
  {file_usage} fu
ON fm.fid=fu.fid
WHERE
  fu.fid IS NULL")->fetchCol();

$file_storage = \Drupal::entityTypeManager()->getStorage('file');
$file_storage->delete($file_storage->loadMultiple($file_ids));
print "Deleted " . count($file_ids) . " unused files\n";

/**
 * Filesystem service.
 *
 * @var \Drupal\Core\File\FileSystemInterface $fs_service
 */
$fs_service = \Drupal::service('file_system');

$file_uris = \Drupal::database()->query("
SELECT
  uri
FROM
  {file_managed}
")->fetchCol();
$used_paths = [];
foreach ($file_uris as $file_uri) {
  $used_paths[] = $fs_service->realpath($file_uri);
}

$directory = new \RecursiveDirectoryIterator($fs_service->realpath("public://"));
$iterator = new \RecursiveIteratorIterator($directory);
$delete_count = 0;

/**
 * File info.
 *
 * @var \SplFileInfo $info
 */
foreach ($iterator as $info) {
  $absolute_path = $info->getRealPath();
  if (in_array($absolute_path, $used_paths)) {
    continue;
  }
  if (is_dir($absolute_path)) {
    continue;
  }
  if (unlink($absolute_path)) {
    $delete_count++;
  }
}
print "Deleted $delete_count unmanaged files\n";
print "Execution time: " . (time() - $start) . " seconds\n";