url.module in Spam 5.3

File

filters/url/url.module

<?php

/**
 * URL filter plug in for the spam module.
 * Copyright(c) 2007-2008
 *  Jeremy Andrews <[e-mail]>.  All rights reserved.
 *
 */
function url_spamapi($op, $type = NULL, $content = array(), $fields = array(), $extra = NULL) {

  // Don't both with this hook unless the filter is actually enabled.
  switch ($op) {
    case 'filter':
      if (!module_invoke('spam', 'filter_enabled', 'url', $type, $content, $fields, $extra)) {
        return;
      }
      return url_spam_filter($content, $type, $fields, $extra);
    case 'filter_module':
      return 'url';
    case 'filter_info':
      return array(
        'name' => t('URL filter'),
        'module' => t('url'),
        'description' => t('A spam url filter.'),
        'help' => t('The url filter blocks posts containing spam-URLs, automatically learned with the bayesian filter module.'),
      );
      break;
    case 'filter_install':
      return array(
        'status' => SPAM_FILTER_ENABLED,
        'gain' => 250,
        'weight' => -6,
      );
    case 'mark_as_spam':
    case 'mark_as_not_spam':
      if (!module_invoke('spam', 'filter_enabled', 'url', $type, $content, $fields, $extra)) {
        return;
      }
      spam_log(SPAM_DEBUG, 'url_spamapi', t('@op', array(
        '@op' => $op,
      )), $type, $extra['id']);
      $fields = spam_invoke_module($type, 'filter_fields', $extra['content']);
      $urls = _url_extract($extra['content'], $type, $fields, $extra);
      url_update($urls, $op == 'mark_as_spam' ? TRUE : FALSE, $type, $extra['id']);
      break;
  }
}
function _url_extract($content, $type, $fields, $extra = array()) {
  static $urls = array();
  $id = spam_invoke_module($type, 'content_id', $content, $extra);
  if (is_object($content)) {
    $content = (array) $content;
  }
  if (!isset($urls["{$type}-{$id}"])) {
    $string = '';
    foreach ($fields['main'] as $field) {
      $string .= $content["{$field}"] . ' ';
    }
    if (is_array($fields['other'])) {
      foreach ($fields['other'] as $field) {
        $string .= $content["{$field}"] . ' ';
      }
    }

    // TODO: Improve this matching.  We don't actually extract mailto: urls.
    $URI = "(http://|https://|ftp://|mailto:)";

    // Find all urls in content.
    preg_match_all("!(<p>|[ \n\r\t\\(]*)({$URI}([a-zA-Z0-9@:%_~#?&=.,/;-]*[a-zA-Z0-9@:%_~#&=/;-]))([.,?]?)(?=(</p>|[ \n\r\t\\)]*))!i", $string, $matches);
    foreach ($matches[2] as $url) {
      $url = preg_replace("'{$URI}'", '', $url);

      // get full domain (ie www.sample.com)
      preg_match("/^()?([^\\/\"\\']+)/i", $url, $domain);

      // get root domain (ie sample.com)
      preg_match("/[^\\.\\/]+\\.[^\\.\\/]+\$/", $domain[2], $root);
      $url = htmlspecialchars(strtolower($root[0]));
      _url_count($url);
      $u[] = $url;
    }
    $urls["{$type}-{$id}"] = $u;
  }
  return $urls["{$type}-{$id}"];
}

/**
 * Search for known spam urls in content.
 */
function url_spam_filter($content, $type, $fields, $extra = array(), $filter_test = FALSE) {
  $action = array();
  $id = spam_invoke_module($type, 'content_id', $content, $extra);
  $spam = FALSE;
  $urls = _url_extract($content, $type, $fields, $extra);
  if (is_array($urls) && !empty($urls)) {
    $count = _url_count();
    $limit = variable_get('url_limit_total', 10);
    if ($limit > -1 && $count['total'] > $limit) {
      spam_log(SPAM_VERBOSE, 'url_spam_filter', t('total urls(@total) > url_limit_total(@limit)', array(
        '@total' => $count['total'],
        '@limit' => variable_get('url_limit_total', 10),
      )), $type, $id);
      $action['url'][] = array(
        'limit' => 'total',
        'total' => $count['total'],
      );
      $action['total'] = 99;
      return $action;
    }
    $limit = variable_get('url_limit_repeat', 5);
    if ($limit > -1) {

      // Sort urls from most repeated to least repeated.
      asort($count);

      // skip count['total']
      array_pop($count);
      $max = array_pop($count);
      if ($max > $limit) {
        spam_log(SPAM_VERBOSE, 'url_spam_filter', t('repeated urls(@total) > url_limit_repeat(@limit)', array(
          '@total' => $max,
          '@limit' => variable_get('url_limit_repeat', 5),
        )), $type, $id);
        $action['url'][] = array(
          'limit' => 'repeat',
          'total' => $max,
        );
        $action['total'] = 99;
      }
    }
    foreach ($urls as $url) {
      $p = db_fetch_object(db_query("SELECT probability FROM {bayesian_tokens} WHERE class = 'url' AND token = '%s'", $url));
      $action['url'][] = array(
        'url' => $url,
        'probability' => $p->probability,
      );
      if ($p->probability >= variable_get('spam_threshold', SPAM_DEFAULT_THRESHOLD)) {
        spam_log(SPAM_VERBOSE, 'url_spam_filter', t('found spam url(@url) probability(@probability)', array(
          '@url' => $url,
          '@probability' => $p->probability,
        )), $type, $id);
        $spam = TRUE;
        break;
      }
      spam_log(SPAM_DEBUG, 'url_spam_filter', t('not spam url(@url) probability(@probability)', array(
        '@url' => $url,
        '@probability' => $p->probability,
      )), $type, $id);
    }
  }
  if ($spam) {
    $action['total'] = 99;
  }
  else {
    $action['total'] = 0;
  }
  return $action;
}

/**
 * Update url probabilities in database.
 */
function url_update($urls, $yes, $type, $id) {
  module_invoke('bayesian', 'tokens_update', 'url', $urls, $yes, $type, $id);
}

/**
 * Keep track of the total number of URLs found in the current content.
 *
 * @param $url  A URL to be added to a static array.
 * @return      Array of URLs showing how many times each URL is present, and
 *              the total number of arrays.
 */
function _url_count($url = NULL) {

  // build up an array of all URLs seen in current content
  static $urls = array();
  if ($url != NULL) {
    $urls["{$url}"]++;
    $urls['total']++;
  }
  return $urls;
}

Functions

Name	Description
url_spamapi	URL filter plug in for the spam module. Copyright(c) 2007-2008 Jeremy Andrews <[e-mail]>. All rights reserved.
url_spam_filter	Search for known spam urls in content.
url_update	Update url probabilities in database.
_url_count	Keep track of the total number of URLs found in the current content.
_url_extract