<?php
/**
 * @file file_checker.module
 * We try to harmonzie the db table "files" with the actual filesystem representation.
 * There we flag missing physical files with a new file status FILE_STATUS_MISSING.
 */

// Additional file status flag, complementing ~FILE_STATUS_PERMANENT and FILE_STATUS_PERMANENT
define('FILE_STATUS_MISSING', 2);

include_once('file_checker.pages.inc');
include_once('file_checker.views.inc');

/**
 * Verify all files in the files table and check their representation on the file system.
 * Therefore we create a batch job and separate the whole set into separate junks.
 * @param $process
 *   boolean whether to process the batch immediately, not needed of being set as
 *   on form submission a batch will be triggered automatically
 */
function file_checker_process_verification($process = TRUE, $batch_size = NULL) {
  // calculate number of files
  $total = file_checker_get_count();
  // split processing into junks
  $batch_size = isset($batch_size) ? $batch_size : variable_get('file_checker_process_batch_size', 10000);
  $operations = array();
  for ($i = 0; $i < $total; $i += $batch_size) {
    // the last op will be smaller, don't exceed total size
    $count = ($total > ($i + $batch_size)) ? $batch_size : ($total - $i);
    $operations[] = array('file_checker_verify_range', array($i, $count, $total));
  }
  // Set batch processor
  $batch = array(
    'operations' => $operations,
    'finished' => 'file_checker_verify_finish', // called after operations have been finished
    'title' => t('Identify and flag missing physical files'),
    'init_message' => t('File Checker is starting to process %total files', array('%total' => $total)),
    'progress_message' => NULL, // suppress here, message to be set in operation function
    'error_message' => t('File Checker has encountered an error.'),
    'progressive' => FALSE,
  );
  batch_set($batch); // no need to call
  if ($process) {
    batch_process();
  }
}

/**
 * This batch operation callback checks a range of files.
 * @param $from
 *   file offset
 * @param $count
 *   number of files to check
 * @param $total
 *   total number of files
 * @param $context
 *   batch callback context information
 */
function file_checker_verify_range($from, $count, $total, &$context) {
  //$result = db_query("SELECT * FROM {file_usage} WHERE count>0");
  $result = db_select('file_usage', 'f')
    ->fields('f')
    ->condition('count', 0, '>')
    ->range($from, $count)
    ->execute();

  foreach ($result as $data) {
    // Temporary files (those marked with status = 0) older than the maximum temp age
    // were, or will be, removed during a cron run.
    $files = db_select('file_managed', 'f')
      ->fields('f')
      ->condition('fid', $data->fid)
      //->condition('filesize', 0, '>')
      ->condition('status', 0, '>')
      ->execute();
    foreach ($files as $file) {
      file_checker_verify($file);
      break; // only need to check one, more would be redundant
    }
  }
  // correct status message here, as we process in larger junks
  $context['message'] = t('Processed @current out of @total.', array('@current' => ($from + $count), '@total' => $total));
}

/**
 * Check if a file contained in the files table physically exists on the file system.
 * If a file cannot be accessed, we flag the file as missed, i. e. FILE_STATUS_MISSING.
 * @param $file
 *   file object
 */
function file_checker_verify($file) {
  // determine file path
  //$path = file_create_path($file->filepath); // determine file path
  $path = file_stream_wrapper_uri_normalize($file->uri);
  // check if file can be accessed
  if (!file_exists($path) || !is_readable($path)) {
    $file->status = FILE_STATUS_MISSING;
    drupal_write_record('file_managed', $file, 'fid');
  }
  // if files are falsely marked as missing, revert the status back
  elseif (($file->status & FILE_STATUS_MISSING) == FILE_STATUS_MISSING) {
    $file->status = FILE_STATUS_PERMANENT;
    drupal_write_record('file_managed', $file, 'fid');
  }
}

/**
 * Implements hook_cron()
 *
 * Process the file system checking on cron if demanded. This should only be used on small sites if at all.
 */
function file_checker_cron() {
  if (variable_get('file_checker_process_on_cron', FALSE)) {
    file_checker_process_verification(TRUE);
  }
}

/**
 * Batch callback finisher operation
 * @param $success
 *   boolean about operation success
 * @param $results
 *   potential ouput
 * @param $operations
 *   operations processed
 */
function file_checker_verify_finish($success, $results, $operations) {
  if ($success) {
    variable_set('file_checker_last_run', REQUEST_TIME);
    drupal_set_message(t('File verification has been completed.'));
    watchdog('file_checker', t('File verification has been completed.'));
  }
  else {
    // A fatal error occurred during batch processing
    $error_operation = reset($operations);
    $operation = array_shift($error_operation);
    $arguments = array_shift($error_operation);
    $arguments_as_string = implode(', ', $arguments);
    watchdog('file_checker', "Error on file verification when calling operation '%s'('%s')", array($operation, $arguments_as_string), WATCHDOG_ERROR);
    drupal_set_message(t('An error occurred and has been recorded in the system log.'), 'error');
  }
}

/**
 * Implements hook_menu().
 *
 * @return
 *   array of menu items
 */
function file_checker_menu() {
  $items['admin/config/system/file_checker'] = array(
    'title' => 'File Checker',
    'description' => t('Show missing physical files.'),
    'page callback' => 'drupal_get_form',
    'page arguments' => array('file_checker_admin_settings'),
    'access callback' => 'file_checker_access',
    'type' => MENU_NORMAL_ITEM,
    'file' => 'file_checker.pages.inc',
  );

  $items['admin/reports/file_checker'] = array(
    'title' => 'File list',
    'description' => t('Show physical files, filter them by status, e. g. missing.'),
    'page callback' => 'file_checker_overview',
    'access callback' => 'file_checker_access',
    'type' => MENU_NORMAL_ITEM,
    'file' => 'file_checker.pages.inc',
  );
  return $items;
}

function file_checker_access() {
  return user_access('access file checker') ;
}

/**
 * Implements hook_permission().
 * @return multitype:number
 */
function file_checker_permission() {
  return array(
    'access file checker' => array(
      'title' => t('Access file checker'),
      'description' => t('Allow users to access the file checker'),
    ),
  );
}

/**
 * Implements hook_views_api().
 * @return multitype:number
 */
function file_checker_views_api() {
  return array(
    'api' => 2,
  );
}

/**
 * Total count of files.
 * @return
 *   SQL statement string
 */
function file_checker_get_count() {
  $total = db_query('SELECT count(*) FROM {file_usage} WHERE count>0')->fetchField();
  return $total;
}

/**
 * Get a list of (relevant) db columuns in the files table. We could also call drupal_get_schema('files'),
 * but this is more efficient and we have the columns in the order we want them to have.
 *
 * @return
 *  array of file table columns
 */
function file_checker_get_db_table_file_cols() {
  return array(
    'fid' => array('type' => 'serial'),
    'timestamp' => array('type' => 'int'),
    'filename' => array('type' => 'varchar'),
    'uri' => array('type' => 'varchar'),
    'filemime' => array('type' => 'varchar'),
    'filesize' => array('type' => 'int'),
    'uid' => array('type' => 'int'),
    'status' => array('type' => 'int'),
    'type' => array('type' => 'varchar'),
  );
}