pingback.module

<?php

function pingback_perm() {
  return array('administer pingbacks');
}


function pingback_menu($may_cache) {
  $items = array();
  if ($may_cache) {
    $items[] = array(
      'path' => 'admin/settings/pingback',
      'title' => t('Pingback'),
      'description' => t('Configure pingbacks.'),
      'callback' => 'drupal_get_form',
      'callback arguments' => array('pingback_settings_form'),
      'access' => user_access('administer pingbacks'),
    );
  }
  else {
    //I developed this module with Drupal 6 pattern/convention, so page callbacks are on their own files
    //unconditionally load those files in Drupal 5
    require_once drupal_get_path('module', 'pingback') . '/pingback.admin.inc';
    //set the server autodiscovery
    //paths like node/10/edit are NOT pingback-enabled
    if (arg(0) == 'node' && is_numeric(arg(1)) && arg(3) == NULL) {
      drupal_set_header('X-Pingback: ' .$GLOBALS['base_url'] . '/xmlrpc.php');
      //drupal_set_html_head();
    }
  }
  return $items;
}

function pingback_form_alter($form_id, &$form) {
  global $user;
  if ($form_id == 'node_type_form' && isset($form['identity']['type'])) {
    $type = $form['#node_type']->type;
    $form['workflow']['pingback'] = array(
      '#type' => 'radios',
      '#title' => t('Pingbacks'),
      '#options' => array(1 => t('Enabled'), 0 => t('Disabled')),
      '#default_value' => _pingback_valid_for_node_type($type),
      '#description' => t('Enable pingbacks for this node type.')
    );
  }
  else if (isset($form['type']) && $form['type']['#value'] .'_node_form' == $form_id) {
    $node = $form['#node'];
    if (_pingback_valid_for_node_type($node->type)) {
      // if there are any past successful pingbacks from this posting, add them to the node editing page.
      $past_successes_listing = array();
      $q = db_query("SELECT url FROM {pingback_sent} WHERE nid = %d", $node->nid);
      while ($pb = db_fetch_object($q)) {
        $past_successes_listing[] = $pb->url;
      }
      // add listing of successfully pingbacked URLs
      if (count($past_successes_listing)) {
        $form['pingback'] = array(
          '#type' => 'fieldset',
          '#title' => t('Pingbacks'),
          '#collapsible' => TRUE
        );
        $form['pingback'][] = array(
          '#type' => 'markup',
          '#value' => theme('item_list', $past_successes_listing, t('Successfully pingbacked URLs')),
        );
        //t('These URLs have been successfuly pinged by this post.')
      }
    }
  }
  //hide pingback input format if desired for anon users
  else if (
    $form_id == 'comment_form'
    && (!$user->uid)
    && variable_get('pingback_hide_format_for_anon', 0)
    //&& (isset($GLOBALS['pingback_bypass_format_hiding']) ? !$GLOBALS['pingback_bypass_format_hiding'] : TRUE)
  ) {
    //dpm($form);
    $alternate_formats = array();
    foreach ($form['comment_filter']['format'] as $k => $v) {
      //dpm($k);
      if (!element_property($k) && isset($v['#return_value'])) {
        if ($v['#return_value'] == variable_get('pingback_input_format', FILTER_FORMAT_DEFAULT)) {
          unset($form['comment_filter']['format'][$k]);
        }
        else {
          // Make a list of alternate formats for the comment form.
          $alternate_formats[] = $k;
        }
      }
    }
    if (count($alternate_formats) == 1) {
      // There is only one available format. Remove fieldset and go back to hidden form field.
      $new_form[$alternate_formats[0]] = array(
        '#type' => 'value',
        '#value' => $alternate_formats[0],
        '#parents' => $form['comment_filter']['format'][$alternate_formats[0]]['#parents'],
      );
      $new_form['format']['guidelines'] = array(
        '#title' => t('Formatting guidelines'),
        '#value' => $form['comment_filter']['format'][$alternate_formats[0]]['#description'],
      );
      $form['comment_filter']['format'] = $new_form;
    }
  }
}


/**
 * Menu callback: lists pingbacks.  TODO: ability to delete them!
 */ 
 //pingback_list_pingbacks() a better name?
function pingback_list_page() {
  //$result = db_query("SELECT ");
  return 'TODO';
}


function pingback_nodeapi(&$node, $op, $teaser = NULL, $page = NULL) {
  
  if (_pingback_valid_for_node_type($node->type)) {
    switch ($op) {
      case 'insert':
      case 'update':
          if (variable_get('pingback_mode', 'off') == 'submit') {
            global $pingback_nid;
            $pingback_nid = $node->nid;
          }
          else { //mode == 'cron'
            //queue this nid in variable pingback_nid_queue, but take care for not queuing existing nids
            $q = variable_get('pingback_nid_queue', array());
            if (!in_array($node->nid, $q)) {
              $q[] = $node->nid;
              variable_set('pingback_nid_queue', $q);
            }
          }
        break;
        
    }
  }
}


function pingback_cron() {
  $q = variable_get('pingback_nid_queue', array());
  $limit = variable_get('pingback_check_per_cron', 30);
  $count = 0;
  //dpr($q);
  while (($nid = array_shift($q)) && ($count++ < $limit)) {
    pingback_send_by_nid($nid, FALSE);
    //dpr("Sent pingbacks in node $nid");
  }
  variable_set('pingback_nid_queue', $q);
}

function pingback_exit() {
  global $pingback_nid;
  if (isset($pingback_nid)) {
    //reset the node_load() cache
    node_load($pingback_nid, NULL, TRUE);
    pingback_send_by_nid($pingback_nid, variable_get('pingback_notify_successful_pings', 1));
  }
}


function pingback_xmlrpc() {
  return array(
    array(
      'pingback.ping',
      'pingback_receive',
      array('string', 'string', 'string'),
      t('Handles pingback pings.'),
    ),
  );
}


/**
 * XML-RPC callback: process pingback.ping() call.
 */
function pingback_receive($pagelinkedfrom, $pagelinkedto) {
  //return xmlrpc_server_error(0, 'abcdefgh');
  //big thanks to WordPress codebase, specifically file xmlrpc.php, method pingback_ping() for becoming the reference implementation and theft victim ;)
  //note: $pagelinkedto is a URL from our own site, $pagelinkedfrom is a foreign URL
  
  if (!variable_get('pingback_receive', 1)) return xmlrpc_server_error(33, t("The specified target URL cannot be used as a target. It either doesn't exist, or it is not a pingback-enabled resource."));
  
  //don't really understand this part, supposed to unescape ampersand entities?
  $pagelinkedfrom = str_replace('&amp;', '&', $pagelinkedfrom);
  $pagelinkedto   = preg_replace('#&([^amp\;])#is', '&amp;$1', $pagelinkedto);
  $error_code = -1;

  // Check if the page linked to is in our site
  $pos1 = strpos($pagelinkedto, str_replace(array('http://www.','http://','https://www.','https://'), '', $GLOBALS['base_url']));
  if( !$pos1 ) {
    return new xmlrpc_server_error(0, t('Is there no link to us?'));
  }
  
  // let's find which post is linked to
  $nid = _pingback_url_to_nid($pagelinkedto);


  //dpm("(PB) URL='$pagelinkedto' ID='$post_ID' Found='$way'");

  $node = $nid ? node_load($nid) : FALSE;
  //watchdog('debug', '--- ' . $nid . ' --- ' . print_r($node, TRUE));
  
  if (!$node || !_pingback_valid_for_node($node)) // node not found
    return xmlrpc_server_error(33, t("The specified target URL cannot be used as a target. It either doesn't exist, or it is not a pingback-enabled resource."));

  if ($nid == _pingback_url_to_nid($pagelinkedfrom))
    return xmlrpc_server_error(0, t('The source URL and the target URL cannot both point to the same resource.'));
  
  if (!$node->status)
    return xmlrpc_server_error(33, t("The specified target URL cannot be used as a target. It either doesn't exist, or it is not a pingback-enabled resource."));

  // Let's check that the remote site didn't already pingback this entry
  $result = db_query("SELECT cid FROM {comments} WHERE  nid = %d AND homepage = '%s' AND format = %d", $nid, $pagelinkedfrom, variable_get('pingback_input_format', FILTER_FORMAT_DEFAULT));

  if (db_num_rows($result)) // We already have a Pingback from this URL
      return xmlrpc_server_error(48, 'The pingback has already been registered.');

  // very stupid, but gives time to the 'from' server to publish !
  sleep(1);

  // Let's check the remote site
  $r = drupal_http_request($pagelinkedfrom);
  if ($r->error)
    return xmlrpc_server_error(16, 'The source URL does not exist.');
  //watchdog('debug', print_r($r, TRUE));
  $linea = $r->data;

  // Work around bug in strip_tags():
  $linea = str_replace('<!DOC', '<DOC', $linea);
  $linea = preg_replace( '/[\s\r\n\t]+/', ' ', $linea ); // normalize spaces
  $linea = preg_replace( "/ <(h1|h2|h3|h4|h5|h6|p|th|td|li|dt|dd|pre|caption|input|textarea|button|body)[^>]*>/", "\n\n", $linea );

  preg_match('|<title>([^<]*?)</title>|is', $linea, $matchtitle);
  $title = check_plain($matchtitle[1]);
  if ( empty( $title ) )
      return xmlrpc_server_error(32, 'We cannot find a title on that page.');

  $linea = strip_tags( $linea, '<a>' ); // just keep the tag we need

  $p = explode( "\n\n", $linea );

  $preg_target = preg_quote($pagelinkedto);

  foreach ( $p as $para ) {
    if ( strpos($para, $pagelinkedto) !== false ) { // it exists, but is it a link?
      preg_match("|<a[^>]+?".$preg_target."[^>]*>([^>]+?)</a>|", $para, $context);

      // If the URL isn't in a link context, keep looking
      if (empty($context)) continue;

      // We're going to use this fake tag to mark the context in a bit
      // the marker is needed in case the link text appears more than once in the paragraph
      //I edited <wpcontext></wpcontext> to <dpcontext></dpcontext> so it becomes more Drupal-ish!
      $excerpt = preg_replace('|\</?dpcontext\>|', '', $para);

      // prevent really long link text
      if ( strlen($context[1]) > 100 )
        $context[1] = substr($context[1], 0, 100) . '...';
      
      
      $marker = '<dpcontext>'.$context[1].'</dpcontext>';    // set up our marker
      $excerpt = str_replace($context[0], $marker, $excerpt); // swap out the link for our marker
      $excerpt = strip_tags($excerpt, '<dpcontext>');        // strip all tags but our context marker
      $excerpt = trim($excerpt);
      $preg_marker = preg_quote($marker);
      $excerpt = preg_replace("|.*?\s(.{0,100}$preg_marker.{0,100})\s.*|s", '$1', $excerpt);
      $excerpt = strip_tags($excerpt); // YES, again, to remove the marker wrapper
      
      break;
    }
  }

  if (empty($context)) // Link to target not found
      return xmlrpc_server_error(17, t('The source URL does not contain a link to the target URL, and so cannot be used as a source.'));
  
  //??? can someone explain about this?
  $pagelinkedfrom = preg_replace('#&([^amp\;])#is', '&amp;$1', $pagelinkedfrom);

  //$context = '[...] ' . wp_specialchars( $excerpt ) . ' [...]';
  //TODO: a custom filter for $excerpt
  $edit = array(
    'nid' => $nid,
    'subject' => t('Pingback'),
    'comment' => '[...] ' . $excerpt . ' [...]',
    'hostname' => $_SERVER['REMOTE_ADDR'],
    'format' => variable_get('pingback_input_format', FILTER_FORMAT_DEFAULT),
    'name' => $title,
    'homepage' => $pagelinkedfrom,
  );
  comment_save($edit);
  /*
  //bypass the hiding in pingback_form_alter() because we want to use the input format
  $GLOBALS['pingback_bypass_format_hiding'] = TRUE;
  drupal_execute('comment_form', $edit, array());
  $GLOBALS['pingback_bypass_format_hiding'] = FALSE;
  watchdog('debug', print_r(form_get_errors(), TRUE));
  */
  $message = t('Pingback from @source to @target registered! Keep the web talking! :-)', array('@source' => $pagelinkedfrom, '@target' => $pagelinkedto));
  //comment.module already logs new comments
  //watchdog('pingback', $message);
  return $message;
}





/*--- theme_pingback_* ---*/

function theme_pingback($pb, $links = 0) {
  return theme('comment', $pb, $links);
}





/*--- APIs ---*/

/**
 * Discover a pingback server with pingback autodiscovery schemes.
 * @param $target the absolute URL to search for its server. This should have passed check_url() first.
 */
function pingback_discover($target) {
  $server = '';
  //#1: send a HEAD to check for X-Pingback header
  $r = drupal_http_request($target, array(), 'HEAD');
  //dpm($r);
  if (!$r->error) {
    if (is_array($r->headers) && isset($r->headers['X-Pingback'])) {
      $server = $r->headers['X-Pingback'];
    }
    else {
      //#2: search for <link rel="pingback" href="(server)" /> tags
      $get = drupal_http_request($target);
      if (!$get->error) {
        //dpm($get->data);
        //this regexp is the one provided in the spec
        if (preg_match('#<link rel="pingback" href="([^"]+)" ?/?>#', $get->data, $matches)) {
          $server = $matches[1];
        }
        
      }
    }
  }
  if (!empty($server)) {
    return check_url($server);
  }
  else return '';
}

/**
 * Send pingbacks. Does nothing if the target does not have a pingback server.
 * @param $nid the source node ID.
 * @param $target the target absolute URL.
 * @param $source_is_absolute if this value is set to TRUE, $nid is interpreted as an absolute URL (which may originate not from the host site).
 * @return TRUE on success, FALSE otherwise. 
*/
function pingback_send($nid, $target, $source_is_absolute = FALSE) {
  if (!valid_url($target, TRUE)) return FALSE;
  if (!$source_is_absolute) {
    $source = url("node/$nid", NULL, NULL, TRUE);
    $result = db_query("SELECT nid FROM {pingback_sent} WHERE nid = %d AND url = '%s'", $nid, $target);
    if (db_num_rows($result) > 0) {
      //dpm('oops already sent');
      return FALSE;
    }
  }
  else {
    $source = $nid;
    if (!valid_url($source)) return FALSE;
  }
  //dpm($source);
  $retval = FALSE;

  //server autodiscovery
  $server = pingback_discover($target);
  //dpm($server);
  
  if (!empty($server)) {
    if(xmlrpc($server, 'pingback.ping', $source, $target)) {
      if (!$source_is_absolute) {
        db_query("INSERT INTO {pingback_sent} (nid, url, timestamp) VALUES (%d, '%s', %d)", $nid, $target, time());
      }
      watchdog('pingback', t('Pingback to %target from %source succeeded.', array('%source' => $source, '%target' => $target)));
      return TRUE;
    }
    else {
      watchdog('pingback', t('Pingback to %target from %source failed. Error @errno: @description', array('%source' => $source, '%target' => $target, '@errno' => xmlrpc_errno(), '@description' => xmlrpc_error_msg())), WATCHDOG_WARNING);
      return FALSE;
    }
  }
  return FALSE;
}

/**
 * Sends pingbacks in all URLs in specified node.
 */
function pingback_send_by_nid($nid, $message = TRUE) {
  global $base_root;
  
  $node = node_load($nid);
  $prepared = node_prepare($node);
  $urls = _pingback_extract_urls($prepared->body);
  if (isset($node->pingback_sent)) {
    //$urls = array_diff(_pingback_extract_urls($prepared->body), $node->pingback_sent);
  }
  $succesful = array();
  foreach ($urls as $url) {
    //dpm("Sending to " . check_plain($url));
    if (pingback_send($node->nid, $url)) {
      //dpm('success!');
      if ($message) $successful[] = "<a href=\"$url\">$url</a>";
    }
  }
  
  if ($message && count($successful)) {
    drupal_set_message(t('!urls pingbacked successfully.', array('!urls' => implode(', ', $successful))));
  }
  //drupal_set_message("URLs: " . implode(', ', $urls));
}


function pingback_comment_is_pingback($comment) {
  return $comment->format == variable_get('pingback_input_format', FILTER_FORMAT_DEFAULT);
}

/*--- private functions ---*/

function _pingback_valid_for_node_type($type) {
  return variable_get("pingback_$type", ($type == 'story' || $type == 'blog') ? 1 : 0);
}

function _pingback_valid_for_node($node) {
  return $node->comment == COMMENT_NODE_READ_WRITE;
}

function _pingback_extract_urls($text) {
  //regexp is stolen from trackback.module ;)
  preg_match_all("/(http|https):\/\/[a-zA-Z0-9@:%_~#?&=.,\/;-]*[a-zA-Z0-9@:%_~#&=\/;-]/", $text, $urls);
  return array_unique($urls[0]);
}

//maps any absolute url from this drupal site to nid if applicable.
//can also be used to check whether an absolute path is in the site and points to a node (e.g. node/1)
function _pingback_url_to_nid($url) {
  //first check if the url is really in our site, as well as getting the non-base-url part
  if (preg_match($a = '#^' . preg_quote($GLOBALS['base_url'], '#') . '/(.+)$#', $url, $matches)) {
    //dpm($matches[1]);
    //dpm(drupal_get_normal_path($matches[1]));
    if (!variable_get('clean_url', 0)) {
      // Clean URLs not enabled. Strip '?q=' from URL.
      $matches[1] = str_replace('?q=', '', $matches[1]);
    }
    if (preg_match($b = '#^node/([0-9]+)$#', drupal_get_normal_path($matches[1]), $matches2)) {
      return $matches2[1];
    } //else dpm($b);
  }

  //dpm($a);
  return FALSE;
}