search404.module in Search 404 5

File

search404.module

<?php

/* @file
 * The search404 module files, does all the searching
 * when a 404 occurs
 */

/**
 * Implementation of hook_menu().
 */
function search404_menu($may_cache) {
  $items = array();
  if ($may_cache) {
    $items[] = array(
      'path' => 'search404',
      'title' => t('Page not found'),
      'access' => TRUE,
      'callback' => 'search404_page',
      'type' => MENU_LOCAL_TASK,
    );
    $items[] = array(
      'path' => 'admin/settings/search404',
      'title' => t('Search 404 settings'),
      'description' => t('Administer search 404.'),
      'callback' => 'drupal_get_form',
      'callback arguments' => array(
        'search404_settings',
      ),
      'access' => user_access('administer site configuration'),
      'type' => MENU_NORMAL_ITEM,
    );
  }
  return $items;
}

/**
 * Replacement for search_get_keys
 * WARNING: This function can potentially return dangerous
 *              potential SQL inject/XSS
 * data. Return must be sanatized before use.
 */
function search404_get_keys() {

  // Abort query on certain extensions, e.g: gif jpg jpeg png
  $extensions = preg_split('/\\s+/', variable_get('search404_ignore_query', 'gif jpg jpeg bmp png'));
  $extensions = trim(implode('|', $extensions));
  if (!empty($extensions) && preg_match("/\\.({$extensions})\$/", $_REQUEST['destination'])) {
    return FALSE;
  }
  $keys = $_REQUEST['destination'];
  $misc_var = variable_get('search404_regex', '');
  if (!empty($misc_var)) {
    $keys = preg_grep($misc_var, $keys);
    $keys = $keys[0];
  }

  // Ingore certain extensions from query
  $extensions = preg_split('/\\s+/', variable_get('search404_ignore_extensions', 'htm html php'));
  $extensions = trim(implode('|', $extensions));
  if (!empty($extensions)) {
    $keys = preg_replace("/\\.({$extensions})\$/", '', $keys);
  }
  $keys = preg_split('/[' . PREG_CLASS_SEARCH_EXCLUDE . ']+/u', $keys);

  // Ignore certain words
  $keys = array_diff($keys, explode(' ', variable_get('search404_ignore', 'and or the')));
  foreach ($keys as $a => $b) {
    $keys[$a] = check_plain($b);
  }
  $modifier = variable_get('search404_use_or', FALSE) ? ' OR ' : ' ';
  $keys = trim(implode($modifier, $keys));
  return $keys;
}

/**
 * Detect search from search engine (BETA)
 * WARNING: This function can potentially return dangerous
 *              potential SQL inject/XSS
 * data. Return must be sanatized before use.
 */
function search404_search_engine_query() {
  $engines = array(
    'altavista' => 'q',
    'aol' => 'query',
    'google' => 'q',
    'live' => 'q',
    'lycos' => 'query',
    'msn' => 'q',
    'yahoo' => 'p',
  );
  $parsed_url = parse_url($_SERVER['HTTP_REFERER']);
  drupal_set_message('Parsed url :' . $parsed_url);
  $remote_host = $parsed_url['host'];
  drupal_set_message('Remote host :' . $remote_host);
  $query_string = $parsed_url['query'];
  drupal_set_message('Query string :' . $query_string);
  parse_str($query_string, $query);
  if (!$parsed_url === FALSE && !empty($remote_host) && !empty($query_string) && count($query)) {
    foreach ($engines as $host => $key) {
      if (strpos($remote_host, $host) !== FALSE && array_key_exists($key, $query)) {
        return trim($query[$key]);
      }
    }
  }
  return FALSE;
}

/**
 * Main search function.
 * Started with: http://drupal.org/node/12668
 * Updated to be more similar to search_view
 * Beware of messy code
 */
function search404_page() {
  $output = '<p>' . t('The page you requested was not found.') . '</p>';
  if (module_exists('search') && user_access('search content')) {
    $keys = "";
    if (variable_get('search404_use_search_engine', FALSE)) {
      $keys = search404_search_engine_query();
    }
    if (!$keys) {
      $keys = search404_get_keys();
    }
    if ($keys) {

      // TODO: watchdog?
      $results = module_invoke('node', 'search', 'search', $keys);
      if (isset($results) && is_array($results) && count($results) == 1 && variable_get('search404_jump', FALSE)) {

        // First, check to see if there is exactly 1 result
        drupal_set_message(t('The page you requested does not exist. A search for %keys resulted in this page.', array(
          '%keys' => check_plain($keys),
        )), 'status');

        // overwrite $_REQUEST['destination'] because it is set by drupal_not_found()
        $_REQUEST['destination'] = 'node/' . $results[0]['node']->nid;
        drupal_goto();
      }
      elseif (isset($results) && is_array($results) && count($results) > 1 && variable_get('search404_first', FALSE)) {
        drupal_set_message(t('The page you requested does not exist. A search for %keys resulted in this page.', array(
          '%keys' => check_plain($keys),
        )), 'status');

        // overwrite $_REQUEST['destination'] because it is set by drupal_not_found()
        $_REQUEST['destination'] = 'node/' . $results[0]['node']->nid;
        drupal_goto();
      }
      else {
        drupal_set_message(t('The page you requested does not exist. For your convenience, a search was performed using the query %keys.', array(
          '%keys' => check_plain($keys),
        )), 'error');
        if (isset($results) && is_array($results) && count($results) > 0) {
          drupal_add_css(drupal_get_path('module', 'search') . '/search.css', 'module', 'all', FALSE);

          // EVIL HAXX!
          $oldgetq = $_GET['q'];
          $olddestination = $_REQUEST['destination'];
          unset($_REQUEST['destination']);
          $_GET['q'] = "search/node/{$keys}";
          $results = theme('search_page', $results, 'node');
          $_GET['q'] = $oldgetq;
          $_REQUEST['destination'] = $olddestination;

          // END OF EVIL HAXX!
        }
        else {
          $results = search_help('search#noresults');
        }
        $results = theme('box', t('Search results'), $results);
      }
    }

    // Construct the search form.
    $output .= drupal_get_form('search_form', NULL, $keys, 'node');
    $output .= $results;
  }
  return $output;
}

/**
 * Settings form
 */
function search404_settings() {
  $form['search404_jump'] = array(
    '#type' => 'checkbox',
    '#title' => t('Jump directly to the search result when there is only one result.'),
    '#default_value' => variable_get('search404_jump', FALSE),
  );
  $form['search404_first'] = array(
    '#type' => 'checkbox',
    '#title' => t('Jump directly to the first search result when there are multiple results.'),
    '#default_value' => variable_get('search404_first', FALSE),
  );
  $form['advanced'] = array(
    '#type' => 'fieldset',
    '#title' => t('Advanced settings'),
    '#description' => t("WARNING. Some of these settings can mess up stuff, don't touch unless you know what you are doing."),
    '#collapsible' => TRUE,
    '#collapsed' => TRUE,
  );
  $form['advanced']['search404_use_or'] = array(
    '#type' => 'checkbox',
    '#title' => t('Use OR between keywords when searching.'),
    '#default_value' => variable_get('search404_use_or', FALSE),
  );
  $form['advanced']['search404_use_search_engine'] = array(
    '#type' => 'checkbox',
    '#title' => t('Use auto-detection of keywords from search engine referer. BETA! Not for production sites, use at your own risk.'),
    '#default_value' => variable_get('search404_use_search_engine', FALSE),
  );
  $form['advanced']['search404_ignore'] = array(
    '#type' => 'textfield',
    '#title' => t('Words to ignore'),
    '#description' => t('These words will be ignored from query. Separate words with a space, e.g.: "and or the".'),
    '#default_value' => variable_get('search404_ignore', 'and or the'),
  );
  $form['advanced']['search404_ignore_extensions'] = array(
    '#type' => 'textfield',
    '#title' => t('Extensions to ignore'),
    '#description' => t('These extensions will be ignored from query. Separate extensions with a space, e.g.: "htm html php". Do not include leading dot.'),
    '#default_value' => variable_get('search404_ignore_extensions', 'htm html php'),
  );
  $form['advanced']['search404_ignore_query'] = array(
    '#type' => 'textfield',
    '#title' => t('Extensions to abort search'),
    '#description' => t('A search will not be performed for a query ending in the following extensions. Separate extensions with a space, e.g.: "gif jpg jpeg bmp png". Do not include leading dot.'),
    '#default_value' => variable_get('search404_ignore_query', 'gif jpg jpeg bmp png'),
  );
  $form['advanced']['search404_regex'] = array(
    '#type' => 'textfield',
    '#title' => t('PCRE REGEX'),
    '#description' => t('This regex will applied to all queries. It uses the code:<p>%code</p>Look directly at the source code to understand underlying syntax. See also <a href="http://php.net/pcre">PCRE pages in the PHP Manual</a>.', array(
      '%code' => "\$keys = preg_grep(variable_get('search404_regex'), \$keys);\n\$keys = \$keys[0];",
      '%function' => 'search404_get_keys()',
    )),
    '#default_value' => variable_get('search404_regex', ''),
  );
  return system_settings_form($form);
}

Functions

Name	Description
search404_get_keys	Replacement for search_get_keys WARNING: This function can potentially return dangerous potential SQL inject/XSS data. Return must be sanatized before use.
search404_menu	Implementation of hook_menu().
search404_page	Main search function. Started with: http://drupal.org/node/12668 Updated to be more similar to search_view Beware of messy code
search404_search_engine_query	Detect search from search engine (BETA) WARNING: This function can potentially return dangerous potential SQL inject/XSS data. Return must be sanatized before use.
search404_settings	Settings form