auditfiles.usednotreferenced.inc in Audit Files 7.3

Generates report showing files in file_usage, but not referenced by content.

File

auditfiles.usednotreferenced.inc

<?php

/**
 * @file
 * Generates report showing files in file_usage, but not referenced by content.
 */

/**
 * The following are functions for displaying the list of files on the page.
 */

/**
 * Generates the report.
 *
 * This cannot be sorted, because a result set that is too large will time out.
 *
 * @param array $form
 *   The form definition.
 * @param array $form_state
 *   The current state of the form.
 *
 * @return array
 *   The form definition.
 */
function auditfiles_used_not_referenced_form(array $form, array &$form_state) {

  // Check to see if the confirmation form needs to be displayed instead of the
  // normal form.
  if (isset($form_state['storage']['confirm'])) {
    return _auditfiles_used_not_referenced_confirm_operation($form, $form_state);
  }

  // Get the records to display.
  // Check to see if there is saved data, and if so, use that.
  $rows = variable_get('auditfiles_used_not_referenced_files_to_display', array());
  if (empty($rows)) {

    // The data is not saved and the batch operation has not been run, so get
    // the data using the default options.
    $file_ids = _auditfiles_used_not_referenced_get_file_list();
    if (!empty($file_ids)) {
      foreach ($file_ids as $file_id) {
        $rows[$file_id] = _auditfiles_used_not_referenced_get_file_data($file_id);
      }
    }

    // Save the data for persistent use.
    variable_set('auditfiles_used_not_referenced_files_to_display', $rows);
  }

  // Set up the pager.
  if (!empty($rows)) {
    $items_per_page = variable_get('auditfiles_report_options_items_per_page', 50);
    if (!empty($items_per_page)) {
      $current_page = pager_default_initialize(count($rows), $items_per_page);

      // Break the total data set into page sized chunks.
      $pages = array_chunk($rows, $items_per_page, TRUE);
    }
  }

  // Define the form.
  // Setup the record count and related messages.
  $maximum_records = variable_get('auditfiles_report_options_maximum_records', 250);
  if (!empty($rows)) {
    if ($maximum_records > 0) {
      $file_count_message = 'Found at least @count files in the file_usage table that are not referenced in content.';
    }
    else {
      $file_count_message = 'Found @count files in the file_usage table that are not referenced in content.';
    }
    $form_count = format_plural(count($rows), 'Found 1 file in the file_usage table that is not referenced in content.', $file_count_message);
  }
  else {
    $form_count = t('Found no files in the file_usage table that are not referenced in content.');
  }

  // Add the button to reset the record selection.
  $form['reset_records'] = array(
    '#type' => 'submit',
    '#value' => t('Reset file list'),
    '#suffix' => '<div>' . t("Use this button to reset this report's variables and load the page anew.") . '</div>',
  );

  // Add the button to batch process the list of results.
  if ($maximum_records > 0) {
    $form['batch_process'] = array(
      '#type' => 'submit',
      '#value' => t('Load all files'),
      '#suffix' => '<div>' . t('Use this button to load the number of records specified with the "Batch size" administrative configuration setting.') . '</div>',
    );
  }

  // Create the form table.
  $form['files'] = array(
    '#type' => 'tableselect',
    '#header' => _auditfiles_used_not_referenced_get_header(),
    '#empty' => t('No items found.'),
    '#prefix' => '<div><em>' . $form_count . '</em></div>',
  );

  // Add the data.
  if (!empty($rows) && !empty($pages)) {
    $form['files']['#options'] = $pages[$current_page];
  }
  elseif (!empty($rows)) {
    $form['files']['#options'] = $rows;
  }
  else {
    $form['files']['#options'] = array();
  }

  // Add any action buttons.
  if (!empty($rows)) {
    $form['actions'] = array(
      '#type' => 'actions',
    );
    $form['actions']['submit'] = array(
      '#type' => 'submit',
      '#value' => t('Delete selected items from the file_usage table'),
    );

    // Add the pager.
    $form['pager'] = array(
      '#markup' => theme('pager'),
    );
  }
  return $form;
}

/**
 * Submit handler for the auditfiles_used_not_referenced_form form.
 */
function auditfiles_used_not_referenced_form_submit(array $form, array &$form_state) {

  // Check if an operation was performed.
  if (!empty($form_state['values']['op'])) {

    // Check which operation was performed and start the batch process.
    if ($form_state['values']['op'] == t('Load all files')) {

      // Clear the variable, so subsequent pages will load the correct data.
      variable_del('auditfiles_used_not_referenced_files_to_display');

      // Prepare and set the batch.
      batch_set(_auditfiles_used_not_referenced_batch_display_create_batch());
    }
    elseif ($form_state['values']['op'] == t('Reset file list')) {

      // Reset all the variables for this report, so subsequent pages loads will
      // load and use the correct data.
      db_delete('variable')
        ->condition('name', 'auditfiles_used_not_referenced_%', 'LIKE')
        ->execute();
      cache_clear_all('variables', 'cache_bootstrap');
    }
    elseif ($form_state['values']['op'] == t('Delete selected items from the file_usage table') && !empty($form_state['values']['files'])) {
      foreach ($form_state['values']['files'] as $file_id) {
        if (!empty($file_id)) {

          // At least one file was selected, and the operation has not been
          // confirmed, so modify the data to display the confirmation form.
          $form_state['storage']['files'] = $form_state['values']['files'];
          $form_state['storage']['op'] = $form_state['values']['op'];
          $form_state['storage']['confirm'] = TRUE;
          $form_state['rebuild'] = TRUE;
          return TRUE;
        }
      }
      drupal_set_message(t('No items were selected to operate on.'));
    }
    elseif ($form_state['values']['op'] == t('Yes')) {
      if ($form_state['values']['operation'] == 'delete') {

        // Prepare and set the batch.
        batch_set(_auditfiles_used_not_referenced_batch_delete_create_batch($form_state['values']['changelist']));
      }
    }
  }
}

/**
 * The following are functions that are common for all batches in this file.
 */

/**
 * Adds vaules to a batch definition that are common to all batches in the file.
 *
 * @return array
 *   The beginning of the batch definition.
 */
function _auditfiles_used_not_referenced_batch_set_common_values() {
  return array(
    'error_message' => t('One or more errors were encountered processing the files.'),
    'file' => drupal_get_path('module', 'auditfiles') . '/auditfiles.usednotreferenced.inc',
    'finished' => '_auditfiles_used_not_referenced_batch_finish_batch',
    'progress_message' => t('Completed @current of @total operations.'),
  );
}

/**
 * The function that is called when the batch is complete.
 */
function _auditfiles_used_not_referenced_batch_finish_batch($success, $results, $operations) {
  if ($success) {
    if (!empty($results['files_to_display'])) {

      // Save the gathered data for display.
      variable_set('auditfiles_used_not_referenced_files_to_display', $results['files_to_display']);
    }
  }
  else {

    // An error occurred.
    // $operations contains the operations that remained unprocessed.
    $error_operation = reset($operations);
    drupal_set_message(t('An error occurred while processing @operation with arguments : @args', array(
      '@operation' => $error_operation[0],
      '@args' => print_r($error_operation[0], TRUE),
    )));
  }
}

/**
 * The following are functions for preparing the batch for displaying the files.
 */

/**
 * Prepares the definition for the page display batch.
 *
 * @return array
 *   The batch definition.
 */
function _auditfiles_used_not_referenced_batch_display_create_batch() {
  $batch = _auditfiles_used_not_referenced_batch_set_common_values();
  $batch['title'] = t('Loading file audit data');
  $batch['operations'] = _auditfiles_used_not_referenced_batch_display_get_operations();
  return $batch;
}

/**
 * Configures the operations for the batch process.
 *
 * @return array
 *   The operations to execute.
 */
function _auditfiles_used_not_referenced_batch_display_get_operations() {

  // Get and save the field data, so this is only done once for the entire batch
  // process.
  $file_fields = '';

  // Get a list of all fields on the site.
  $fields = field_info_fields();
  foreach ($fields as $field_name => $field) {

    // @todo
    // Consider adding a setting to allow the administrator the ability for
    // spcifying the field types.
    if (!empty($field['type']) && ($field['type'] == 'file' || $field['type'] == 'image')) {

      // Get the database table name for the field.
      $table = key($field['storage']['details']['sql'][FIELD_LOAD_CURRENT]);
      $field['table'] = $table;

      // Get the column name in the database table for the field.
      $column = $field['storage']['details']['sql'][FIELD_LOAD_CURRENT][$table]['fid'];
      $field['column'] = $column;
      $query = "SELECT t.{$column} FROM {{$table}} AS t INNER JOIN {file_usage} AS f ON f.fid = t.{$column}";
      $results = db_query($query)
        ->fetchAll();
      $field['results'] = $results;

      // Save the data for use in the operation.
      $file_fields[$field_name] = $field;
    }
  }
  $operations = array();
  $operations[] = array(
    '_auditfiles_used_not_referenced_batch_display_get_files',
    array(
      $file_fields,
    ),
  );
  $operations[] = array(
    '_auditfiles_used_not_referenced_batch_display_process_files',
    array(),
  );
  return $operations;
}

/**
 * The batch process operation for getting the files.
 *
 * @param array $file_fields
 *   The relevant fields to look for file information in.
 * @param array $context
 *   Used by the Batch API to keep track of and pass data from one operation to
 *   the next.
 */
function _auditfiles_used_not_referenced_batch_display_get_files(array $file_fields, array &$context) {
  if (empty($context['sandbox'])) {
    $context['sandbox'] = array();
    $context['sandbox']['progress'] = 0;
    $context['sandbox']['current_file'] = 0;
    $query = 'SELECT COUNT(DISTINCT fid) FROM {file_usage} fu';
    $record_count = db_query($query)
      ->fetchField();
    $batch_size = variable_get('auditfiles_report_options_batch_size', 1000);
    if ($batch_size > 0 && $record_count > $batch_size) {
      $context['sandbox']['max'] = $batch_size;
    }
    else {
      $context['sandbox']['max'] = $record_count;
    }
  }
  if (empty($context['results']['file_list'])) {
    $context['results']['file_list'] = array();
  }

  // Get the file data from the database.
  // This cannot be sorted any other way here, or the results are not complete.
  $query = 'SELECT DISTINCT fid
    FROM {file_usage}
    WHERE fid > :fid
    ORDER BY fid ASC
    LIMIT 20';
  $files = db_query($query, array(
    ':fid' => $context['sandbox']['current_file'],
  ))
    ->fetchCol();
  $files_in_fields = array();
  foreach ($file_fields as $field) {
    foreach ($field['results'] as $fid) {
      if (in_array($fid->{$field['column']}, $files)) {
        $files_in_fields[] = $fid->{$field['column']};
      }
    }
  }
  $file_list = array_diff($files, $files_in_fields);
  $context['results']['file_list'] = array_merge($context['results']['file_list'], $file_list);
  $file_id = end($files);

  // Update the progress information.
  $context['sandbox']['progress'] += 20;
  $context['sandbox']['current_file'] = $file_id;
  $context['message'] = t('Getting the list of files. Processed file @num1 of @num2. Last file ID processed: !file_id.', array(
    '@num1' => $context['sandbox']['progress'],
    '@num2' => $context['sandbox']['max'],
    '!file_id' => $file_id,
  ));
  if ($context['sandbox']['progress'] != $context['sandbox']['max']) {
    $context['finished'] = $context['sandbox']['progress'] >= $context['sandbox']['max'];
  }
}

/**
 * The batch process operation for formatting the files for display on the page.
 *
 * @param array $context
 *   Used by the Batch API to keep track of and pass data from one operation to
 *   the next.
 */
function _auditfiles_used_not_referenced_batch_display_process_files(array &$context) {
  if (empty($context['sandbox'])) {
    $context['sandbox'] = array();
    if (empty($context['results']['file_list'])) {
      $context['sandbox']['progress'] = 1;
      $context['sandbox']['max'] = 1;
    }
    else {
      $context['sandbox']['progress'] = 0;
      $context['sandbox']['max'] = count($context['results']['file_list']);
    }
  }
  if (empty($context['results']['files_to_display'])) {
    $context['results']['files_to_display'] = array();
  }
  if (!empty($context['results']['file_list'])) {
    $file_list = array_slice($context['results']['file_list'], 0, 20, TRUE);
    $query = 'SELECT fu.fid, fu.module, fu.type, fu.id, fu.count
      FROM {file_usage} fu
      INNER JOIN {file_managed} fm ON fm.fid = fu.fid
      WHERE fu.fid IN (:file_list)';
    $files_list = db_query($query, array(
      ':file_list' => $file_list,
    ))
      ->fetchAll();
    foreach ($files_list as $file) {
      $context['results']['files_to_display'][$file->fid] = _auditfiles_used_not_referenced_get_file_data($file->fid);
      unset($context['results']['file_list'][$file->fid]);

      // Update the progress information.
      $context['sandbox']['progress']++;
      $context['message'] = t('Processing the file list. Processed file @num1 of @num2. Last file ID processed: !file_id.', array(
        '@num1' => $context['sandbox']['progress'],
        '@num2' => $context['sandbox']['max'],
        '!file_id' => $file->fid,
      ));
    }
  }
  if ($context['sandbox']['progress'] != $context['sandbox']['max']) {
    $context['finished'] = $context['sandbox']['progress'] >= $context['sandbox']['max'];
  }
}

/**
 * The following are functions for the batch delete operation.
 */

/**
 * Presents a confimation form to verify the user wants to complete the action.
 *
 * @param array $form
 *   The form definition.
 * @param array $form_state
 *   The current state of the form.
 *
 * @return array
 *   A form array for a confirmation form.
 */
function _auditfiles_used_not_referenced_confirm_operation(array $form, array &$form_state) {
  $values = $form_state['values'];
  $form['changelist'] = array(
    '#prefix' => '<ul>',
    '#suffix' => '</ul>',
    '#tree' => TRUE,
  );

  // Prepare the list of items to present to the user.
  if (!empty($values['files'])) {
    foreach ($values['files'] as $file_id) {
      if (!empty($file_id)) {
        $file = file_load($file_id);
        if (!empty($file)) {
          $form['changelist'][$file_id] = array(
            '#type' => 'hidden',
            '#value' => $file_id,
            '#prefix' => '<li><strong>' . $file->filename . '</strong> ' . t('will be deleted from the file_usage table.'),
            '#suffix' => "</li>\n",
          );
        }
      }
      else {

        // Unsetting the unprocessed files prevents confirm_submit from dealing
        // with them.
        unset($form_state['values']['files'][$file_id]);
      }
    }
  }
  $form['operation'] = array(
    '#type' => 'hidden',
    '#value' => 'delete',
  );

  // Tell the submit handler to process the confirmation.
  $form['process'] = array(
    '#type' => 'hidden',
    '#value' => 'TRUE',
  );

  // Go back to the main form, when done with this one.
  $form['destination'] = array(
    '#type' => 'hidden',
    '#value' => 'admin/reports/auditfiles/usednotreferenced',
  );
  return confirm_form($form, t('Delete these items from the file_usage table?'), 'admin/reports/auditfiles/usednotreferenced', '<strong>' . t('This action cannot be undone.') . '</strong>', t('Yes'), t('No'));
}

/**
 * Creates the batch for deleting files from the file_usage table.
 *
 * @param array $fileids
 *   The list of file IDs to be processed.
 *
 * @return array
 *   The definition of the batch.
 */
function _auditfiles_used_not_referenced_batch_delete_create_batch(array $fileids) {
  $batch = _auditfiles_used_not_referenced_batch_set_common_values();
  $batch['title'] = t('Deleting files from the file_usage table');
  $operations = array();

  // Remove all the empty values from the array.
  $file_ids = array();
  foreach ($fileids as $file_id) {
    if ($file_id != 0) {
      $file_ids[] = $file_id;
    }
  }

  // Fill in the $operations variable.
  foreach ($file_ids as $file_id) {
    $operations[] = array(
      '_auditfiles_used_not_referenced_batch_delete_process_batch',
      array(
        $file_id,
      ),
    );
  }
  $batch['operations'] = $operations;
  return $batch;
}

/**
 * The batch process for deleting the file.
 *
 * @param int $file_id
 *   The ID of a file to delete.
 * @param array $context
 *   Used by the Batch API to keep track of and pass data from one operation to
 *   the next.
 */
function _auditfiles_used_not_referenced_batch_delete_process_batch($file_id, array &$context) {

  // Process the current file.
  _auditfiles_used_not_referenced_batch_delete_process_file($file_id);

  // The contents of 'results' are available as $results in the 'finished'
  // function.
  $context['results'][] = $file_id;

  // Set a progress message.
  $context['message'] = t('Processed file ID %file_id.', array(
    '%file_id' => $file_id,
  ));
}

/**
 * Deletes the specified file from the file_usage table.
 *
 * @param int $file_id
 *   The ID of the file to delete from the database.
 */
function _auditfiles_used_not_referenced_batch_delete_process_file($file_id) {
  $num_rows = db_delete('file_usage')
    ->condition('fid', $file_id)
    ->execute();
  if (empty($num_rows)) {
    drupal_set_message(t('There was a problem deleting the record with file ID %fid from the file_usage table. Check the logs for more information.', array(
      '%fid' => $file_id,
    )), 'warning');
  }
  else {

    // Remove the deleted files from the list of files to display.
    $rows = variable_get('auditfiles_used_not_referenced_files_to_display', array());
    unset($rows[$file_id]);
    variable_set('auditfiles_used_not_referenced_files_to_display', $rows);
  }
}

/**
 * The following are functions for retrieving and processing the file data.
 */

/**
 * Retrieves the file IDs to operate on.
 *
 * @return array
 *   The file IDs.
 */
function _auditfiles_used_not_referenced_get_file_list() {

  // Get a list of files in the file_usage database table.
  $query = 'SELECT DISTINCT fid FROM {file_usage} fu';

  // If record limition has been configured, only use those records within that
  // specification.
  $maximum_records = variable_get('auditfiles_report_options_maximum_records', 250);
  if ($maximum_records > 0) {
    $query .= ' LIMIT ' . $maximum_records;
  }
  $files_in_file_usage = db_query($query)
    ->fetchCol();
  $files_in_fields = array();

  // Get a list of all fields on the site.
  $fields = field_info_fields();
  foreach ($fields as $field) {

    // @todo
    // Consider adding a setting to allow the administrator the ability for
    // spcifying the field types.
    if (!empty($field['type']) && ($field['type'] == 'file' || $field['type'] == 'image')) {

      // Get the database table name for the field.
      $table = key($field['storage']['details']['sql'][FIELD_LOAD_CURRENT]);

      // Get the column name in the database table for the field.
      $column = $field['storage']['details']['sql'][FIELD_LOAD_CURRENT][$table]['fid'];
      $query = "SELECT t.{$column} FROM {{$table}} AS t INNER JOIN {file_usage} AS f ON f.fid = t.{$column}";
      $result = db_query($query);
      foreach ($result as $fid) {
        if (in_array($fid->{$column}, $files_in_file_usage)) {
          $files_in_fields[] = $fid->{$column};
        }
      }
    }
  }
  return array_diff($files_in_file_usage, $files_in_fields);
}

/**
 * Retrieves information about an individual file from the database.
 *
 * @param int $file_id
 *   The ID of the file to prepare for display.
 *
 * @return array
 *   The row for the table on the report, with the file's information formatted
 *   for display.
 */
function _auditfiles_used_not_referenced_get_file_data($file_id) {

  // Get the file's information from the file_managed table.
  $file_managed = db_query("SELECT * FROM {file_managed} fm WHERE fid = {$file_id}")
    ->fetchObject();
  if (empty($file_managed)) {

    // The file is not listed in the file_managed table. Display an error
    // message, instead of the file information.
    $row = array(
      'fid' => t('This file is not listed in the file_managed table. See the "!usednotmanaged" report.', array(
        '!usednotmanaged' => l(t('Used not managed'), 'admin/reports/auditfiles/usednotmanaged'),
      )),
      'uri' => '',
      'usage' => '',
    );
  }
  else {
    $usage = '<ul>';
    $results = db_query("SELECT * FROM {file_usage} WHERE fid = {$file_id}");
    foreach ($results as $file_usage) {

      // Create the usages list.
      $used_by = $file_usage->module;
      $type = $file_usage->type;
      $used_in = $file_usage->type == 'node' ? l($file_usage->id, 'node/' . $file_usage->id) : $file_usage->id;
      $times_used = $file_usage->count;
      $usage .= '<li>' . t('Used by module: %used_by, as object type: %type, in content ID: !used_in; Times used: %times_used', array(
        '%used_by' => $used_by,
        '%type' => $type,
        '!used_in' => $used_in,
        '%times_used' => $times_used,
      )) . '</li>';
    }
    $usage .= '</ul>';

    // Create the data for displaying in the table.
    $row = array(
      'fid' => $file_id,
      'uri' => $file_managed->uri,
      'usage' => $usage,
    );
  }
  return $row;
}

/**
 * The following are helper functions.
 */

/**
 * Returns the header to use for the display table.
 *
 * @return array
 *   The header to use.
 */
function _auditfiles_used_not_referenced_get_header() {
  return array(
    'fid' => array(
      'data' => t('File ID'),
    ),
    'uri' => array(
      'data' => t('File URI'),
    ),
    'usage' => array(
      'data' => t('Usages'),
    ),
  );
}

Functions

Name	Description
auditfiles_used_not_referenced_form	Generates the report.
auditfiles_used_not_referenced_form_submit	Submit handler for the auditfiles_used_not_referenced_form form.
_auditfiles_used_not_referenced_batch_delete_create_batch	Creates the batch for deleting files from the file_usage table.
_auditfiles_used_not_referenced_batch_delete_process_batch	The batch process for deleting the file.
_auditfiles_used_not_referenced_batch_delete_process_file	Deletes the specified file from the file_usage table.
_auditfiles_used_not_referenced_batch_display_create_batch	Prepares the definition for the page display batch.
_auditfiles_used_not_referenced_batch_display_get_files	The batch process operation for getting the files.
_auditfiles_used_not_referenced_batch_display_get_operations	Configures the operations for the batch process.
_auditfiles_used_not_referenced_batch_display_process_files	The batch process operation for formatting the files for display on the page.
_auditfiles_used_not_referenced_batch_finish_batch	The function that is called when the batch is complete.
_auditfiles_used_not_referenced_batch_set_common_values	Adds vaules to a batch definition that are common to all batches in the file.
_auditfiles_used_not_referenced_confirm_operation	Presents a confimation form to verify the user wants to complete the action.
_auditfiles_used_not_referenced_get_file_data	Retrieves information about an individual file from the database.
_auditfiles_used_not_referenced_get_file_list	Retrieves the file IDs to operate on.
_auditfiles_used_not_referenced_get_header	Returns the header to use for the display table.