<?php

/**
 * @file
 * Anti-spam module that uses data from www.stopforumspam.com to protect the user registration form against known spammers and spambots.
 */

define('SPAMBOT_ACTION_NONE', 0);
define('SPAMBOT_ACTION_BLOCK', 1);
define('SPAMBOT_ACTION_DELETE', 2);
define('SPAMBOT_DEFAULT_CRITERIA_EMAIL', 1);
define('SPAMBOT_DEFAULT_CRITERIA_USERNAME', 0);
define('SPAMBOT_DEFAULT_CRITERIA_IP', 20);
define('SPAMBOT_DEFAULT_DELAY', 0);
define('SPAMBOT_DEFAULT_CRON_USER_LIMIT', 0);
define('SPAMBOT_DEFAULT_BLOCKED_MESSAGE', 'Your email address or username or IP address is blacklisted.');
define('SPAMBOT_MAX_EVIDENCE_LENGTH', 1024);

/**
 * Implements hook_menu().
 */
function spambot_menu() {
  $items['admin/config/system/spambot'] = array(
    'title' => 'Spambot',
    'description' => 'Configure the spambot module',
    'page callback' => 'drupal_get_form',
    'page arguments' => array('spambot_settings_form'),
    'access arguments' => array('administer site configuration'),
    'file' => 'spambot.admin.inc',
  );

  $items['user/%user/spambot'] = array(
    'title' => 'Spam',
    'page callback' => 'spambot_user_spam',
    'page arguments' => array(1),
    'access arguments' => array('administer users'),
    'type' => MENU_LOCAL_TASK,
    'file' => 'spambot.pages.inc',
  );

  return $items;
}

/**
 * Implements hook_permission().
 */
function spambot_permission() {
  return array(
    'protected from spambot scans' => array(
      'title' => t('Protected from spambot scans'),
      'description' => t('Roles with this access permission would not be checked for spammer'),
    ),
  );
}

/**
 * Implements hook_admin_paths().
 */
function spambot_admin_paths() {
  $paths = array(
    'user/*/spambot' => TRUE,
  );

  return $paths;
}

/**
 * Implements hook_form_FORM_ID_alter().
 */
function spambot_form_user_register_form_alter(&$form, &$form_state) {
  if (variable_get('spambot_user_register_protect', TRUE) && !user_access('protected from spambot scans')) {
    $form['#validate'][] = 'spambot_user_register_form_validate';
  }
}

/**
 * Implements hook_form_FORM_ID_alter().
 * Adds an Operations link to the user administration form with a link to user/$uid/spambot
 */
function spambot_form_user_admin_account_alter(&$form, &$form_state, $form_id){
  foreach ($form['accounts']['#options'] as $uid => $user_options) {
    // Change $form['accounts']['#options'][$uid]['operations']['data'] into a multi-item render array so we can append to it
    $form['accounts']['#options'][$uid]['operations']['data'] = array(
      'edit' => $form['accounts']['#options'][$uid]['operations']['data']
    );
    $form['accounts']['#options'][$uid]['operations']['data']['spam'] = array(
      '#type' => 'link',
      '#title' => t('spam'),
      '#href' => "user/$uid/spambot",
      // Ugly hack to insert a space
      '#prefix' => ' ',
    );
  }
}

/**
 * Implements hook_node_insert().
 */
function spambot_node_insert($node) {
  db_insert('node_spambot')
    ->fields(array(
      'nid' => $node->nid,
      'uid' => $node->uid,
      'hostname' => ip_address(),
    ))
    ->execute();
}

/**
 * Implements hook_node_delete().
 */
function spambot_node_delete($node) {
  db_delete('node_spambot')
    ->condition('nid', $node->nid)
    ->execute();
}

/**
 * Implements hook_cron().
 */
function spambot_cron() {
  if ($limit = variable_get('spambot_cron_user_limit', SPAMBOT_DEFAULT_CRON_USER_LIMIT)) {
    $last_uid = variable_get('spambot_last_checked_uid', 0);
    if ($last_uid < 1) {
      // Skip scanning the anonymous user.
      $last_uid = 1;
    }

    $uids = db_select('users')
      ->fields('users', array('uid'))
      ->condition('uid', $last_uid, '>')
      ->orderBy('uid')
      ->range(0, $limit)
      ->execute()
      ->fetchCol();

    if ($uids) {
      $action = variable_get('spambot_spam_account_action', SPAMBOT_ACTION_NONE);
      $accounts = user_load_multiple($uids);

      foreach ($accounts as $account) {
        if ($account->status || variable_get('spambot_check_blocked_accounts', FALSE)) {
          $result = spambot_account_is_spammer($account);

          if ($result > 0) {
            $link = l(t('spammer'), 'user/' . $account->uid);
            switch (user_access('protected from spambot scans', $account) ? SPAMBOT_ACTION_NONE : $action) {
              case SPAMBOT_ACTION_BLOCK:
                if ($account->status) {
                  // Block spammer's account.
                  $account->status = 0;
                  user_save($account);
                  watchdog('spambot', 'Blocked spam account: @name &lt;@email&gt; (uid @uid)', array(
                    '@name' => $account->name,
                    '@email' => $account->mail,
                    '@uid' => $account->uid,
                  ), WATCHDOG_NOTICE, $link);
                }
                else {
                  // Don't block an already blocked account.
                  watchdog('spambot', 'Spam account already blocked: @name &lt;@email&gt; (uid @uid)', array(
                    '@name' => $account->name,
                    '@email' => $account->mail,
                    '@uid' => $account->uid,
                  ), WATCHDOG_NOTICE, $link);
                }
                break;

              case SPAMBOT_ACTION_DELETE:
                user_delete($account->uid);
                watchdog('spambot', 'Deleted spam account: @name &lt;@email&gt; (uid @uid)', array(
                  '@name' => $account->name,
                  '@email' => $account->mail,
                  '@uid' => $account->uid,
                ), WATCHDOG_NOTICE, $link);
                break;

              default:
                watchdog('spambot', 'Found spam account: @name &lt;@email&gt; (uid @uid)', array(
                  '@name' => $account->name,
                  '@email' => $account->mail,
                  '@uid' => $account->uid,
                ), WATCHDOG_NOTICE, $link);
                break;
            }

            // Mark this uid as successfully checked.
            variable_set('spambot_last_checked_uid', $account->uid);
          }
          else if ($result == 0) {
            // Mark this uid as successfully checked.
            variable_set('spambot_last_checked_uid', $account->uid);
          }
          else if ($result < 0) {
            // Error contacting service, so pause processing.
            break;
          }
        }
      }
    }
  }
}

/**
 * Validate callback for user_register form.
 */
function spambot_user_register_form_validate(&$form, &$form_state) {
  $values = $form_state['values'];
  $form_errors = form_get_errors();

  $email_threshold = variable_get('spambot_criteria_email', SPAMBOT_DEFAULT_CRITERIA_EMAIL);
  $username_threshold = variable_get('spambot_criteria_username', SPAMBOT_DEFAULT_CRITERIA_USERNAME);
  $ip_threshold = variable_get('spambot_criteria_ip', SPAMBOT_DEFAULT_CRITERIA_IP);

  // Build request parameters according to the criteria to use.
  $request = array();
  if (!empty($values['mail']) && $email_threshold > 0 && !spambot_check_whitelist('email', $values['mail'])) {
    $request['email'] = $values['mail'];
  }

  if (!empty($values['name']) && $username_threshold > 0 && !spambot_check_whitelist('username', $values['name'])) {
    $request['username'] = $values['name'];
  }

  $ip = ip_address();
  if ($ip_threshold > 0 && $ip != '127.0.0.1' && !spambot_check_whitelist('ip', $ip)) {
    $request['ip'] = $ip;
  }

  // Only do a remote API request if there is anything to check.
  if ($request && !$form_errors) {
    $data = array();
    if (spambot_sfs_request($request, $data)) {
      $substitutions = array(
        '@email' => $values['mail'],
        '%email' => $values['mail'],
        '@username' => $values['name'],
        '%username' => $values['name'],
        '@ip' => $ip,
        '%ip' => $ip,
      );

      $reasons = array();
      if ($email_threshold > 0 && !empty($data['email']['appears']) && $data['email']['frequency'] >= $email_threshold) {
        form_set_error('mail', t(variable_get('spambot_blocked_message_email', SPAMBOT_DEFAULT_BLOCKED_MESSAGE), $substitutions));
        $reasons[] = t('email=@value', array('@value' => $request['email']));
      }
      if ($username_threshold > 0 && !empty($data['username']['appears']) && $data['username']['frequency'] >= $username_threshold) {
        form_set_error('name', t(variable_get('spambot_blocked_message_username', SPAMBOT_DEFAULT_BLOCKED_MESSAGE), $substitutions));
        $reasons[] = t('username=@value', array('@value' => $request['username']));
      }
      if ($ip_threshold > 0 && !empty($data['ip']['appears']) && $data['ip']['frequency'] >= $ip_threshold) {
        form_set_error('', t(variable_get('spambot_blocked_message_ip', SPAMBOT_DEFAULT_BLOCKED_MESSAGE), $substitutions));
        $reasons[] = t('ip=@value', array('@value' => $request['ip']));
      }

      if ($reasons) {
        if (variable_get('spambot_log_blocked_registration', TRUE)) {
          watchdog('spambot', 'Blocked registration: @reasons', array('@reasons' => implode(',', $reasons)));

          $hook_args = array(
            'request' => $request,
            'reasons' => $reasons,
          );
          module_invoke_all('spambot_registration_blocked', $hook_args);
        }

        // Slow them down if configured.
        if ($delay = variable_get('spambot_blacklisted_delay', SPAMBOT_DEFAULT_DELAY)) {
          sleep($delay);
        }
      }
    }
  }
}

/**
 * Invoke www.stopforumspam.com's api
 *
 * @param array $query
 *   A keyed array of url parameters ie. array('email' => 'blah@blah.com')
 * @param array $data
 *   An array that will be filled with the data from www.stopforumspam.com.
 *
 * @return bool
 *   TRUE on successful request (and $data will contain the data), FALSE if error
 *
 * $data should be an array of the following form:
 * Array
 * (
 *     [success] => 1
 *     [email] => Array
 *         (
 *             [lastseen] => 2010-01-10 08:41:26
 *             [frequency] => 2
 *             [appears] => 1
 *         )
 *
 *     [username] => Array
 *         (
 *             [frequency] => 0
 *             [appears] => 0
 *         )
 * )
 *
 */
function spambot_sfs_request($query, &$data) {
  // An empty request results in no match
  if (empty($query)) {
    return FALSE;
  }

  // Use php serialisation format
  $query['f'] = 'serial';

  $url = 'http://www.stopforumspam.com/api?' . http_build_query($query, '', '&');
  $result = drupal_http_request($url);
  if (!empty($result->code) && $result->code == 200 && empty($result->error) && !empty($result->data)) {
    $data = unserialize($result->data);
    if (!empty($data['success'])) {
      return TRUE;
    }
    else {
      watchdog('spambot', "Request unsuccessful: @url <pre>\n@dump</pre>", array('@url' => $url, '@dump' => print_r($data, TRUE)));
    }
  }
  else {
    watchdog('spambot', "Error contacting service: @url <pre>\n@dump</pre>", array('@url' => $url, '@dump' => print_r($result, TRUE)));
  }

  return FALSE;
}

/**
 * Checks an account to see if it's a spammer.
 * This one uses configurable automated criteria checking of email and username only
 *
 * @param object $account
 *
 * @return int
 *   positive if spammer, 0 if not spammer, negative if error
 */
function spambot_account_is_spammer($account) {
  $email_threshold = variable_get('spambot_criteria_email', SPAMBOT_DEFAULT_CRITERIA_EMAIL);
  $username_threshold = variable_get('spambot_criteria_username', SPAMBOT_DEFAULT_CRITERIA_USERNAME);
  $ip_threshold = variable_get('spambot_criteria_ip', SPAMBOT_DEFAULT_CRITERIA_IP);

  // Build request parameters according to the criteria to use
  $request = array();
  if (!empty($account->mail) && $email_threshold > 0 && !spambot_check_whitelist('email', $account->mail)) {
    $request['email'] = $account->mail;
  }

  if (!empty($account->name) && $username_threshold > 0 && !spambot_check_whitelist('username', $account->name)) {
    $request['username'] = $account->name;
  }

  // Only do a remote API request if there is anything to check
  if ($request) {
    $data = array();
    if (spambot_sfs_request($request, $data)) {
      if (($email_threshold > 0 && !empty($data['email']['appears']) && $data['email']['frequency'] >= $email_threshold)
        || ($username_threshold > 0 && !empty($data['username']['appears']) && $data['username']['frequency'] >= $username_threshold)) {

        return 1;
      }
    }
    else {
      // Return error
      return -1;
    }
  }

  // Now check IP's
  // If any IP matches the threshold, then flag as a spammer
  if ($ip_threshold > 0) {
    $ips = spambot_account_ip_addresses($account);
    foreach ($ips as $ip) {
      // Skip the loopback interface
      if ($ip == '127.0.0.1') {
        continue;
      }

      $request = array('ip' => $ip);
      $data = array();
      if (spambot_sfs_request($request, $data)) {
        if (!empty($data['ip']['appears']) && $data['ip']['frequency'] >= $ip_threshold) {
          return 1;
        }
      }
      else {
        // Abort on error
        return -1;
      }
    }
  }

  // Return no match
  return 0;
}

/**
 * Retrieves a list of IP addresses for an account
 *
 * @param object $account
 *   Account to retrieve IP addresses for
 *
 * @return array
 *   An array of IP addresses, or an empty array if none found
 */
function spambot_account_ip_addresses($account) {
  $hostnames = array();

  // Retrieve IPs from node_spambot table.
  $items = db_select('node_spambot')
    ->distinct()
    ->fields('node_spambot', array('hostname'))
    ->condition('uid', $account->uid, '=')
    ->execute()
    ->fetchCol();
  $hostnames = array_merge($hostnames, $items);

  // Retrieve IPs from any sessions which may still exist.
  $items = db_select('sessions')
    ->distinct()
    ->fields('sessions', array('hostname'))
    ->condition('uid', $account->uid, '=')
    ->execute()
    ->fetchCol();
  $hostnames = array_merge($hostnames, $items);

  // Retrieve IPs from comments.
  if (module_exists('comment')) {
    $items = db_select('comment')
      ->distinct()
      ->fields('comment', array('hostname'))
      ->condition('uid', $account->uid, '=')
      ->execute()
      ->fetchCol();
    $hostnames = array_merge($hostnames, $items);
  }

  // Retrieve IPs from statistics.
  if (module_exists('statistics')) {
    $items = db_select('accesslog')
      ->distinct()
      ->fields('accesslog', array('hostname'))
      ->condition('uid', $account->uid, '=')
      ->execute()
      ->fetchCol();
    $hostnames = array_merge($hostnames, $items);
  }

  // Retrieve IPs from user stats.
  if (module_exists('user_stats')) {
    $items = db_select('user_stats_ips')
      ->distinct()
      ->fields('user_stats_ips', array('ip_address'))
      ->condition('uid', $account->uid, '=')
      ->execute()
      ->fetchCol();
    $hostnames = array_merge($hostnames, $items);
  }

  $hostnames = array_unique($hostnames);
  return $hostnames;
}

/**
 * Reports an account as a spammer. Requires ip address and evidence of a single incident
 *
 * @param object $account
 *   Account to report
 * @param string $ip
 *   IP address to report
 * @param string $evidence
 *   Evidence to report
 *
 * @return bool
 *   TRUE if successful, FALSE if error
 */
function spambot_report_account($account, $ip, $evidence) {
  $success = FALSE;

  if ($key = variable_get('spambot_sfs_api_key', FALSE)) {
    $query['api_key'] = $key;
    $query['email'] = $account->mail;
    $query['username'] = $account->name;
    $query['ip_addr'] = $ip;
    $query['evidence'] = truncate_utf8($evidence, SPAMBOT_MAX_EVIDENCE_LENGTH);

    $url = 'http://www.stopforumspam.com/add.php';
    $options = array(
      'headers' => array('Content-type' => 'application/x-www-form-urlencoded'),
      'method' => 'POST',
      'data' => http_build_query($query, '', '&'),
    );
    $result = drupal_http_request($url, $options);

    if (!empty($result->code) && $result->code == 200 && !empty($result->data) && stripos($result->data, 'data submitted successfully') !== FALSE) {
      $success = TRUE;
    }
    elseif (stripos($result->data, 'duplicate') !== FALSE) {
      // www.stopforumspam.com can return a 503 code with data = '<p>recent duplicate entry</p>'
      // which we will treat as successful.
      $success = TRUE;
    }
    else {
      watchdog('spambot', "Error reporting account: @url <pre>\n@dump</pre>", array('@url' => $url, '@dump' => print_r($result, TRUE)));
    }
  }

  return $success;
}

/**
 * Check if current data $type is whitelisted.
 *
 * @param string $type
 *  Type can be one of these three values:
 *  'ip', 'email' or 'username'.
 * @param string $value
 *  Value to be checked.
 *
 * @return bool
 */
function spambot_check_whitelist($type, $value) {
  switch ($type) {
    case 'ip':
      $whitelist_ips = variable_get('spambot_whitelist_ip', '');
      $result = strpos($whitelist_ips, $value) !== FALSE;
      break;

    case 'email':
      $whitelist_usernames = variable_get('spambot_whitelist_email', '');
      $result = strpos($whitelist_usernames, $value) !== FALSE;
      break;

    case 'username':
      $whitelist_emails = variable_get('spambot_whitelist_username', '');
      $result = strpos($whitelist_emails, $value) !== FALSE;
      break;

    default:
      $result = FALSE;
      break;
  }

  return $result;
}