islandora_log_404_cleaner.module

<?php

/**
 * Implements hook_cron().
 */
function islandora_log_404_cleaner_cron() {
  islandora_log_404_cleaner_remove_entries();
}

function islandora_log_404_cleaner_remove_entries() {
  $database = \Drupal::database();
  // Query covers both _format=json and _format=jsonld.
  $entries_result = $database->query("SELECT wid, location FROM {watchdog} WHERE type = 'page not found' AND location LIKE '%_format=json%'");
  $stale_not_founds = [];
  foreach ($entries_result as $row) {
    $client = \Drupal::httpClient();
    try {
      $result = $client->head($row->location);
      if ($result->getStatusCode() == 200 && str_contains($row->location, '/media/')) {
        $stale_not_founds[] = $row->wid;
      }
    }
    catch (ClientException $e) {
      \Drupal::logger('islandora_log_404_cleaner')->error('Caught exception when pinging media: ' . $e->getMessage() . "\n");
    }
  }

  // Now that we have a list of media that do in fact exist, delete the entries from the watchdog table.
  if (count($stale_not_founds) > 0) {
    foreach ($stale_not_founds as $wid) {
      $delete_query = \Drupal::database()->delete('watchdog');
      $delete_query->condition('wid', $wid);
      $delete_query->execute();
    }
  }
}