custom.module in Spam 5.3

File

filters/custom/custom.module
View source
<?php

/**
 * Custom spam filter module
 * Copyright(c) 2007-2008
 *  Jeremy Andrews <jeremy@tag1consulting.com>.
 *
 * Allows manual definition of words and regular expressions to detect spam
 * content.
 */
define(SPAM_CUSTOM_STYLE_PLAIN, 0);
define(SPAM_CUSTOM_STYLE_REGEX, 1);
define(SPAM_CUSTOM_STATUS_NOTSPAM, -2);
define(SPAM_CUSTOM_STATUS_PROBABLYNOT, -1);
define(SPAM_CUSTOM_STATUS_DISABLED, 0);
define(SPAM_CUSTOM_STATUS_PROBABLY, 1);
define(SPAM_CUSTOM_STATUS_SPAM, 2);
define(SPAM_CUSTOM_SCAN_CONTENT, 0x1);
define(SPAM_CUSTOM_SCAN_REFERRER, 0x4);
define(SPAM_CUSTOM_SCAN_USERAGENT, 0x8);

// TODO: support actions

//define(SPAM_CUSTOM_ACTION_DELETE, 0x1);

//define(SPAM_CUSTOM_ACTION_MAIL, 0x2);

/**
 * Spam API Hook
 */
function custom_spamapi($op, $type = NULL, $content = array(), $fields = array(), $extra = NULL) {
  switch ($op) {
    case 'filter':
      if (!module_invoke('spam', 'filter_enabled', 'custom', $type, $content, $fields, $extra)) {
        return;
      }
      return custom_spam_filter($content, $type, $fields, $extra);
    case 'filter_module':
      return 'custom';
    case 'filter_info':
      return array(
        'name' => t('Custom filter'),
        'module' => t('custom'),
        'description' => t('Custom spam filters.'),
        'help' => t('The custom spam filter module allows you to manually define custom spam filter rules.'),
      );
    case 'filter_install':
      return array(
        'status' => SPAM_FILTER_ENABLED,
        'gain' => 250,
        'weight' => -4,
      );
  }
}

/**
 * Drupal _menu() hook.
 */
function custom_menu($may_cache) {
  $items = array();
  if ($may_cache) {
    $items[] = array(
      'path' => 'admin/settings/spam/filters/custom',
      'title' => t('Custom'),
      'callback' => 'drupal_get_form',
      'callback arguments' => array(
        'custom_admin_settings',
      ),
      'description' => t('Configure the custom spam filter module.'),
      'type' => MENU_LOCAL_TASK,
    );
    $items[] = array(
      'path' => 'admin/settings/spam/filters/custom/list',
      'title' => t('List'),
      'callback' => 'drupal_get_form',
      'callback arguments' => array(
        'custom_admin_settings',
      ),
      'description' => t('Configure the custom spam filter module.'),
      'type' => MENU_DEFAULT_LOCAL_TASK,
    );
    $items[] = array(
      'path' => 'admin/settings/spam/filters/custom/create',
      'title' => t('Create'),
      'callback' => 'drupal_get_form',
      'callback arguments' => array(
        'custom_admin_filter',
      ),
      'description' => t('Create a custom spam filter.'),
      'type' => MENU_LOCAL_TASK,
    );
  }
  else {
    $cid = arg(5);
    $items[] = array(
      'path' => "admin/settings/spam/filters/custom/{$cid}/edit",
      'title' => t('Create'),
      'callback' => 'drupal_get_form',
      'callback arguments' => array(
        'custom_admin_filter',
        $cid,
      ),
      'description' => t('Edit a custom spam filter.'),
      'type' => MENU_LOCAL_TASK,
    );
  }
  return $items;
}

/**
 * Adminsitrative interface for configuring custom spam filter rules.
 */
function custom_admin_settings() {
  $form = array();
  $form['options'] = array(
    '#type' => 'fieldset',
    '#title' => t('Options'),
    '#prefix' => '<div class="container-inline">',
    '#suffix' => '</div>',
  );
  $options = array();
  foreach (module_invoke_all('spam_custom_operations') as $operation => $op) {
    $options[$operation] = $op['label'];
  }
  $form['options']['operation'] = array(
    '#type' => 'select',
    '#options' => $options,
    '#default_value' => 'scan',
  );
  $form['options']['submit'] = array(
    '#type' => 'submit',
    '#value' => t('Execute'),
  );
  $rows = array();
  $result = pager_query('SELECT * FROM {spam_custom} ORDER BY weight ASC');
  while ($custom = db_fetch_object($result)) {
    $all[$custom->cid] = '';

    // The filter text.
    $form['filter'][$custom->cid] = array(
      '#value' => $custom->filter,
    );

    // What style of filter.
    if ($custom->style == SPAM_CUSTOM_STYLE_PLAIN) {
      $form['style'][$custom->cid] = array(
        '#value' => t('Plain text'),
      );
    }
    else {
      if ($custom->style == SPAM_CUSTOM_STYLE_REGEX) {
        $form['style'][$custom->cid] = array(
          '#value' => t('Regular expression'),
        );
      }
    }

    // What to scan.
    $scan = array();
    if ($custom->scan & SPAM_CUSTOM_SCAN_CONTENT) {
      $scan[] = t('Content');
    }
    if ($custom->scan & SPAM_CUSTOM_SCAN_REFERRER) {
      $scan[] = t('Referrer');
    }
    if ($custom->scan & SPAM_CUSTOM_SCAN_USERAGENT) {
      $scan[] = t('User agent');
    }
    $form['scan'][$custom->cid] = array(
      '#value' => implode(', ', $scan),
    );

    // What status to apply.
    switch ($custom->status) {
      case SPAM_CUSTOM_STATUS_NOTSPAM:
        $status = t('Mark as not spam');
        break;
      case SPAM_CUSTOM_STATUS_PROBABLYNOT:
        $status = t('Mark as probably not spam');
        break;
      case SPAM_CUSTOM_STATUS_DISABLED:
        $status = t('Disabled');
        break;
      case SPAM_CUSTOM_STATUS_PROBABLY:
        $status = t('Mark as probably spam');
        break;
      case SPAM_CUSTOM_STATUS_SPAM:
        $status = t('Mark as spam');
        break;
      default:
        $status = t('Unknown');
        break;
    }
    $form['status'][$custom->cid] = array(
      '#value' => $status,
    );

    // How many times this filter has been matched.
    $form['matches'][$custom->cid] = array(
      '#value' => $custom->matches,
    );

    // The last time this filter was matched.
    $last = $custom->last ? t('@time ago', array(
      '@time' => format_interval(time() - $custom->last),
    )) : t('Never');
    $form['last'][$custom->cid] = array(
      '#value' => $last,
    );

    // Link to edit the filter.
    $form['edit'][$custom->cid] = array(
      '#value' => l(t('edit'), "admin/settings/spam/filters/custom/{$custom->cid}/edit"),
    );
    $rows[] = $row;
  }
  $form['create2'] = array(
    '#value' => '[' . l(t('create custom filter'), 'admin/settings/spam/filters/custom/create') . ']',
    '#prefix' => '<div class="container-inline">',
    '#suffix' => '</div>',
  );
  $form['custom'] = array(
    '#type' => 'checkboxes',
    '#options' => $all,
  );
  $form['pager'] = array(
    '#value' => theme('pager', NULL, 50, 0),
  );
  return $form;
}

/**
 * Format the custom filter admin page.
 */
function theme_custom_admin_settings($form) {
  _custom_upgrade();
  $header = array(
    theme('table_select_header_cell'),
    t('Filter'),
    t('Style'),
    t('Scan'),
    t('Status'),
    t('Matches'),
    t('Last'),
    '',
  );
  $output = drupal_render($form['options']);
  $rows = array();
  if (isset($form['filter']) && is_array($form['filter'])) {
    foreach (element_children($form['filter']) as $key) {
      $row = array();
      $row[] = drupal_render($form['custom'][$key]);
      $row[] = drupal_render($form['filter'][$key]);
      $row[] = drupal_render($form['style'][$key]);
      $row[] = drupal_render($form['scan'][$key]);
      $row[] = drupal_render($form['status'][$key]);
      $row[] = drupal_render($form['matches'][$key]);
      $row[] = drupal_render($form['last'][$key]);
      $row[] = drupal_render($form['edit'][$key]);
      $rows[] = $row;
    }
    $output .= theme('table', $header, $rows);
    if ($form['pager']['#value']) {
      $output .= drupal_render($form['pager']);
    }
  }
  else {
    $output .= theme('table', $header, $rows);
    $output .= '<em>' . t('No custom filters created.') . '</em>';
  }
  $output .= drupal_render($form);
  return $output;
}

/**
 * Define callbacks for custom filter options.
 */
function custom_spam_custom_operations() {
  $operations = array(
    'disable' => array(
      'label' => t('Disable'),
      'callback' => 'custom_spam_filter_operations',
      'callback arguments' => array(
        'disable',
      ),
    ),
    'delete' => array(
      'label' => t('Delete'),
      'callback' => 'custom_spam_filter_operations',
      'callback arguments' => array(
        'delete',
      ),
    ),
  );
  return $operations;
}

/**
 * Create or edit a custom spam filter.
 */
function custom_admin_filter($cid = NULL) {
  if ($cid) {
    drupal_set_title('Edit');
    $custom = db_fetch_object(db_query('SELECT * FROM {spam_custom} WHERE cid = %d', $cid));
    if (!isset($custom->cid)) {
      drupal_set_message(t('Failed to load custom filter.'));
      drupal_goto('admin/settings/spam/filters/custom');
    }
  }
  else {
    drupal_set_title('Create');
  }
  $form = array();
  $form['filter'] = array(
    '#type' => 'textfield',
    '#title' => t('Filter'),
    '#description' => t('Enter a custom filter string. You can enter a word, a phrase, or a regular expression.'),
    '#default_value' => $custom->cid ? $custom->filter : '',
    '#required' => TRUE,
  );
  $form['style'] = array(
    '#type' => 'radios',
    '#title' => t('Filter type'),
    '#description' => t('For a custom filter to match exactly what you type, select <code>plain text</code>.  If you would like to define a regular expression, your filter must be formatted as a <a href="http://www.php.net/manual/en/ref.pcre.php">Perl-compatible regular expression</a>.'),
    '#options' => array(
      SPAM_CUSTOM_STYLE_PLAIN => t('Plain text'),
      SPAM_CUSTOM_STYLE_REGEX => t('Regular expression'),
    ),
    '#default_value' => $custom->cid ? $custom->style : SPAM_CUSTOM_STYLE_PLAIN,
    '#required' => TRUE,
  );
  $options = array(
    SPAM_CUSTOM_SCAN_CONTENT => 'Content',
    SPAM_CUSTOM_SCAN_REFERRER => t('Referrer'),
    SPAM_CUSTOM_SCAN_USERAGENT => t('User agent'),
  );
  $scan = array();
  if ($custom->scan & SPAM_CUSTOM_SCAN_CONTENT) {
    $scan[] = SPAM_CUSTOM_SCAN_CONTENT;
  }
  if ($custom->scan & SPAM_CUSTOM_SCAN_REFERRER) {
    $scan[] = SPAM_CUSTOM_SCAN_REFERRER;
  }
  if ($custom->scan & SPAM_CUSTOM_SCAN_USERAGENT) {
    $scan[] = SPAM_CUSTOM_SCAN_USERAGENT;
  }
  $form['scan'] = array(
    '#type' => 'checkboxes',
    '#title' => t('Scan'),
    '#description' => t('Specify where you\'d like to apply your custom filter.'),
    '#options' => $options,
    '#required' => TRUE,
    '#default_value' => !empty($scan) ? $scan : array(
      SPAM_CUSTOM_SCAN_CONTENT,
    ),
  );
  $options = array();
  $form['status'] = array(
    '#type' => 'radios',
    '#title' => t('Status'),
    '#description' => t('Select the status to apply when your custom filter matches site content.  Filters are tested in the order they are displayed above, thus if content matches a filter that says to mark it as spam, and another to mark it as not spam, the first to match will be the actual status applied.'),
    '#options' => array(
      SPAM_CUSTOM_STATUS_DISABLED => t('Disabled'),
      SPAM_CUSTOM_STATUS_SPAM => t('Mark as spam'),
      SPAM_CUSTOM_STATUS_PROBABLY => t('Mark as probably spam'),
      SPAM_CUSTOM_STATUS_PROBABLYNOT => t('Mark as probably not spam'),
      SPAM_CUSTOM_STATUS_NOTSPAM => t('Mark as not spam'),
    ),
    '#default_value' => $custom->cid ? $custom->status : SPAM_CUSTOM_STATUS_SPAM,
    '#required' => TRUE,
  );
  $form['weight'] = array(
    '#type' => 'weight',
    '#title' => t('Weight'),
    '#description' => t('Give your custom filter a weight.  "Lighter" filters with smaller weights will run before "heavier" filters with larger weights.'),
    '#default_value' => $custom->weight,
  );
  $form['submit'] = array(
    '#type' => 'submit',
    '#value' => $custom->cid ? t('Update filter') : t('Create filter'),
  );
  if ($custom->cid) {
    $form['cid'] = array(
      '#type' => 'hidden',
      '#value' => $custom->cid,
    );
  }
  return $form;
}

/**
 * Be sure that the custom filter is valid.
 */
function custom_admin_filter_validate($form_id, $form_values) {
  if ($form_values['style'] == SPAM_CUSTOM_STYLE_REGEX) {
    if (preg_match($form_values['filter'], 'test') === FALSE) {
      form_set_error('filter', t('Failed to validate your filter\'s regular expression.  It must be properly formatted as a <a href="http://www.php.net/manual/en/ref.pcre.php">Perl-compatible regular expression</a>.  Review the above error for details on the specific problem with your expression.'));
    }
  }
  if (isset($form_values['cid'])) {

    // update
    $cid = db_result(db_query("SELECT cid FROM {spam_custom} WHERE filter = '%s' AND cid != %d", $form_values['filter'], $form_values['cid']));
    if ($cid) {
      form_set_error($cid, t('Custom filter %filter already exists', array(
        '%filter' => $form_values['filter'],
      )));
    }
  }
  else {

    // create
    $cid = db_result(db_query("SELECT cid FROM {spam_custom} WHERE filter = '%s'", $form_values['filter']));
    if ($cid) {
      form_set_error($cid, t('Custom filter %filter already exists', array(
        '%filter' => $form_values['filter'],
      )));
    }
  }
}

/**
 * Create/update custom filer.
 */
function custom_admin_filter_submit($form_id, $form_values) {
  $scan = 0;
  if (is_array($form_values['scan'])) {
    foreach ($form_values['scan'] as $s) {
      $scan += $s;
    }
  }
  if (isset($form_values['cid'])) {
    db_query("UPDATE {spam_custom} SET filter = '%s', style = %d, status = %d, scan = %d, weight = %d WHERE cid = %d", $form_values['filter'], $form_values['style'], $form_values['status'], $scan, $form_values['weight'], $form_values['cid']);
    drupal_set_message(t('Custom filter %filter updated.', array(
      '%filter' => $form_values['filter'],
    )));
  }
  else {
    db_query("INSERT INTO {spam_custom} (filter, style, status, scan, weight) VALUES ('%s', %d, %d, %d, %d)", $form_values['filter'], $form_values['style'], $form_values['status'], $scan, $form_values['weight']);
    drupal_set_message(t('Custom filter %filter created.', array(
      '%filter' => $form_values['filter'],
    )));
  }
  drupal_goto('admin/settings/spam/filters/custom');
}

/** 
 * Perform bulk operations on the filters.
 */
function custom_admin_settings_submit($form_id, $form_values) {
  if (is_array($form_values['custom'])) {
    foreach ($form_values['custom'] as $cid => $selected) {
      if ($selected) {
        $process[] = $cid;
      }
    }
  }
  if (!empty($process)) {
    foreach (module_invoke_all('spam_custom_operations') as $operation => $op) {
      $options[$operation] = $op;
    }
    $operation = $form_values['operation'];
    if (isset($options[$operation])) {
      $function = $options[$operation]['callback'];
      $arguments = $options[$operation]['callback arguments'];
      foreach ($process as $cid) {
        call_user_func_array($function, array_merge($arguments, array(
          $cid,
        )));
      }
    }
  }
}

/**
 * Perform custom operations.
 * TODO: Confirmation would be nice.
 */
function custom_spam_filter_operations($op, $cid) {
  $filter = db_fetch_object(db_query('SELECT cid, status, filter FROM {spam_custom} WHERE cid = %d', $cid));
  switch ($op) {
    case 'delete':
      if ($filter->cid) {
        db_query('DELETE FROM {spam_custom} WHERE cid = %d', $cid);
        drupal_set_message(t('Deleted custom filter %filter.', array(
          '%filter' => $filter->filter,
        )));
      }
      break;
    case 'disable':
      if ($filter->cid && $filter->status != SPAM_CUSTOM_STATUS_DISABLED) {
        db_query('UPDATE {spam_custom} SET status = %d WHERE cid = %d', SPAM_CUSTOM_STATUS_DISABLED, $cid);
        drupal_set_message(t('Disabled custom filter %filter.', array(
          '%filter' => $filter->filter,
        )));
      }
      break;
  }
}

/**
 * Apply enabled custom filter rules against content.
 */
function custom_spam_filter($content, $type, $fields, $extra = array(), $filter_test = FALSE) {
  $probably = $probably_not = 0;
  $id = spam_invoke_module($type, 'content_id', $content, $extra);
  $result = db_query('SELECT cid, filter, style, status, scan, action FROM {spam_custom} WHERE status != %d ORDER BY weight ASC', SPAM_CUSTOM_STATUS_DISABLED);
  while ($custom = db_fetch_object($result)) {
    $scan = '';
    if ($custom->scan & SPAM_CUSTOM_SCAN_CONTENT) {

      // scan content
      if (is_object($content)) {
        $content = (array) $content;
      }
      $scan .= spam_get_text($content, $type, $fields, $extra);
      spam_log(SPAM_DEBUG, 'custom_spam_filter', t('scanning content with %filter.', array(
        '%filter' => $custom->filter,
      )), $type, $id);
    }
    if ($custom->scan & SPAM_CUSTOM_SCAN_REFERRER) {

      // scan referrer
      // TODO: Determine if this is a live scan.  If not, don't scan referrer.
      $scan .= $_SERVER['HTTP_REFERER'];
      spam_log(SPAM_DEBUG, 'custom_spam_filter', t('scanning referrer with %filter.', array(
        '%filter' => $custom->filter,
      )), $type, $id);
    }
    if ($custom->scan & SPAM_CUSTOM_SCAN_USERAGENT) {

      // scan user agent
      // TODO: Determine if this is a live scan.  If not, don't scan user agent.
      $scan .= $_SERVER['HTTP_USER_AGENT'];
      spam_log(SPAM_DEBUG, 'custom_spam_filter', t('scanning user agent with %filter.', array(
        '%filter' => $custom->filter,
      )), $type, $id);
    }
    switch ($custom->style) {
      case SPAM_CUSTOM_STYLE_PLAIN:
        $match = preg_match_all("/{$custom->filter}/", $scan, $matches);
        break;
      case SPAM_CUSTOM_STYLE_REGEX:
        $match = preg_match_all($custom->filter, $scan, $matches);
        break;
    }
    if ($match) {

      // Record that we've had one or more matches.
      db_query('UPDATE {spam_custom} SET matches = matches + %d, last = %d WHERE cid = %d', $match, time(), $custom->cid);
      spam_log(SPAM_VERBOSE, 'custom_spam_filter', t('matched with %filter.', array(
        '%filter' => $custom->filter,
      )), $type, $id);
      $action['custom'][] = array(
        'filter' => $custom->filter,
        'status' => $custom->status,
        'style' => $custom->style,
        'scan' => $custom->scan,
        'extra' => $custom->extra,
      );
      switch ($custom->status) {
        case SPAM_CUSTOM_STATUS_SPAM:
          spam_log(SPAM_VERBOSE, 'custom_spam_filter', t('content is spam.'), $type, $id);

          // no need to scan any more, we've found spam
          $action['total'] = 99;
          return $action;
        case SPAM_CUSTOM_STATUS_NOTSPAM:
          spam_log(SPAM_VERBOSE, 'custom_spam_filter', t('content is not spam.'), $type, $id);

          // no need to scan any more, we've found non-spam
          $action['total'] = 1;
          return $action;
        case SPAM_CUSTOM_STATUS_PROBABLYNOT:
          spam_log(SPAM_DEBUG, 'custom_spam_filter', t('content is probably not spam.'), $type, $id);

          // maintain internal counter that this is probably not spam
          $probably_not += $match;
          break;
        case SPAM_CUSTOM_STATUS_PROBABLY:
          spam_log(SPAM_DEBUG, 'custom_spam_filter', t('content is probably spam.'), $type, $id);

          // maintain internal counter that this is probably spam
          $probably += $match;
          break;
      }
    }
  }
  if ($probably && $probably_not) {
    if ($probably >= $probably_not) {
      $probably -= $probably_not;
      $probably_not = 0;
    }
    else {
      $probably_not -= $probably;
      $probably = 0;
    }
  }
  if ($probably) {
    spam_log(SPAM_VERBOSE, 'custom_spam_filter', t('matched adjusted total of !number probably spam rule(s).', array(
      '!number' => $probably,
    )), $type, $id);
    if ($probably >= variable_get('spam_custom_probably', 3)) {
      $action['total'] = 99;
    }
    else {
      $action['total'] = variable_get('spam_custom_probably_value', variable_get('spam_threshold', SPAM_DEFAULT_THRESHOLD));
    }
  }
  else {
    spam_log(SPAM_VERBOSE, 'custom_spam_filter', t('matched adjusted total of !number probably-not spam rule(s).', array(
      '!number' => $probably_not,
    )), $type, $id);
    if ($probably_not >= variable_get('spam_custom_probablynot', 3)) {
      $action['total'] = 1;
    }
    else {
      $action['total'] = variable_get('spam_custom_probablynot_value', 40);
    }
  }
  return $action;
}

/**
 * Upgrade from 5.x-1.x spam module.  This doesn't live in the custom.install 
 * file, as there was no custom.module in the 5.x-1.x release (it was part of 
 * the core spam module).  This doesn't live in spam.install because the
 * spam_custom table isn't created there.
 */
function _custom_upgrade() {
  if (variable_get('spam_custom_upgrade', 1) == 1) {
    require_once drupal_get_path('module', 'custom') . '/custom-upgrade.inc';
    custom_upgrade();
    variable_set('spam_custom_upgrade', 0);
    drupal_goto('admin/settings/spam/filters/custom');
  }
}
Functions

Name	Description
custom_admin_filter	Create or edit a custom spam filter.
custom_admin_filter_submit	Create/update custom filer.
custom_admin_filter_validate	Be sure that the custom filter is valid.
custom_admin_settings	Adminsitrative interface for configuring custom spam filter rules.
custom_admin_settings_submit	Perform bulk operations on the filters.
custom_menu	Drupal _menu() hook.
custom_spamapi	Spam API Hook
custom_spam_custom_operations	Define callbacks for custom filter options.
custom_spam_filter	Apply enabled custom filter rules against content.
custom_spam_filter_operations	Perform custom operations. TODO: Confirmation would be nice.
theme_custom_admin_settings	Format the custom filter admin page.
_custom_upgrade	Upgrade from 5.x-1.x spam module. This doesn't live in the custom.install file, as there was no custom.module in the 5.x-1.x release (it was part of the core spam module). This doesn't live in spam.install because the spam_custom table…
You are here

custom.module in Spam 5.3

File

Functions

API Navigation
