feeds.module

<?php
// $Id$

/**
 * @file
 * Feeds - basic API functions and hook implementations.
 */

// Common request time, use as point of reference and to avoid calls to time().
define('FEEDS_REQUEST_TIME', time());
// Do not schedule a feed for refresh.
define('FEEDS_SCHEDULE_NEVER', -1);
// Never expire feed items.
define('FEEDS_EXPIRE_NEVER', -1);
// An object that is not persistent. Compare EXPORT_IN_DATABASE, EXPORT_IN_CODE.
define('FEEDS_EXPORT_NONE', 0x0);
// The Drupal Queue FeedsScheduler may use for scheduling importing or expiry.
define('FEEDS_SCHEDULER_QUEUE', 'feeds_queue');
// Status of batched operations.
define('FEEDS_BATCH_COMPLETE', 1);
define('FEEDS_BATCH_ACTIVE', 0);

/**
 * @defgroup hooks Hook and callback implementations
 * @{
 */

/**
 * Implementation of hook_cron().
 */
function feeds_cron() {
  feeds_scheduler()->cron();
}

/**
 * Implementation of hook_cron_queue_info().
 * Invoked by drupal_queue module if present.
 */
function feeds_cron_queue_info() {
  $queues = array();
  $queues[FEEDS_SCHEDULER_QUEUE] = array(
    'worker callback' => 'feeds_scheduler_work',
    'time' => variable_get('feeds_worker_time', 15),
  );
  return $queues;
}

/**
 * Implementation of hook_perm().
 */
function feeds_perm() {
  $perms = array('administer feeds');
  foreach (feeds_importer_load_all() as $importer) {
    $perms[] = 'import '. $importer->id .' feeds';
    $perms[] = 'clear '. $importer->id .' feeds';
  }
  return $perms;
}

/**
 * Implementation of hook_forms().
 */
function feeds_forms() {
  // Declare form callbacks for all known classes derived from FeedsConfigurable.
  $forms = array();
  $forms['FeedsImporter_feeds_config_form']['callback'] = 'feeds_config_form';
  $plugins = feeds_get_plugins();
  foreach ($plugins as $plugin) {
    // See feeds_get_config_form().
    $forms[$plugin['handler']['class'] .'_feeds_config_form']['callback'] = 'feeds_config_form';
  }
  return $forms;
}

/**
 * Implementation of hook_menu().
 */
function feeds_menu() {
  // Register a callback for all feed configurations that are not attached to a content type.
  $items = array();
  foreach (feeds_importer_load_all() as $importer) {
    if (empty($importer->config['content_type'])) {
      $items['import/'. $importer->id] = array(
        'title' => $importer->config['name'],
        'page callback' => 'drupal_get_form',
        'page arguments' => array('feeds_import_form', 1),
        'access callback' => 'feeds_access',
        'access arguments' => array('import', $importer->id),
        'file' => 'feeds.pages.inc',
      );
      $items['import/'. $importer->id .'/import'] = array(
        'title' => 'Import',
        'type' => MENU_DEFAULT_LOCAL_TASK,
        'weight' => -10,
      );
      $items['import/'. $importer->id .'/delete-items'] = array(
        'title' => 'Delete items',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('feeds_delete_tab_form', 1),
        'access callback' => 'feeds_access',
        'access arguments' => array('clear', $importer->id),
        'file' => 'feeds.pages.inc',
        'type' => MENU_LOCAL_TASK,
      );
    }
    else {
      $items['node/%node/import'] = array(
        'title' => 'Import',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('feeds_import_tab_form', 1),
        'access callback' => 'feeds_access',
        'access arguments' => array('import', 1),
        'file' => 'feeds.pages.inc',
        'type' => MENU_LOCAL_TASK,
        'weight' => 10,
      );
      $items['node/%node/delete-items'] = array(
        'title' => 'Delete items',
        'page callback' => 'drupal_get_form',
        'page arguments' => array('feeds_delete_tab_form', NULL, 1),
        'access callback' => 'feeds_access',
        'access arguments' => array('clear', 1),
        'file' => 'feeds.pages.inc',
        'type' => MENU_LOCAL_TASK,
        'weight' => 11,
      );
    }
    $items += $importer->fetcher->menuItem();
  }
  if (count($items)) {
    $items['import'] = array(
      'title' => 'Import',
      'page callback' => 'feeds_page',
      'access callback' => 'feeds_page_access',
      'file' => 'feeds.pages.inc',
    );
  }
  return $items;
}

/**
 * Menu loader callback.
 */
function feeds_importer_load($id) {
  return feeds_importer($id);
}

/**
 * Implementation of hook_theme().
 */
function feeds_theme() {
  return array(
    'feeds_info' => array(
      'file' => 'feeds.pages.inc',
    ),
  );
}

/**
 * Menu access callback.
 *
 * @param $action
 *   One of 'import' or 'clear'.
 * @param $param
 *   Node object or FeedsImporter id.
 */
function feeds_access($action, $param) {
  if (is_string($param)) {
    $importer_id = $param;
  }
  elseif ($param->type) {
    $importer_id = feeds_get_importer_id($param->type);
  }

  // Check for permissions if feed id is present, otherwise return FALSE.
  if ($importer_id) {
    if (user_access('administer feeds') || user_access($action .' '. $importer_id .' feeds')) {
      return TRUE;
    }
  }
  return FALSE;
}

/**
 * Menu access callback.
 *
 * @todo Create cached function that returns available configurations
 * and use it for enumerating feed configuration ids.
 */
function feeds_page_access() {
  if (user_access('administer feeds')) {
    return TRUE;
  }
  foreach (feeds_importer_load_all() as $importer) {
    if (user_access('import '. $importer->id .' feed')) {
      return TRUE;
    }
  }
  return FALSE;
}

/**
 * Implementation of hook_views_api().
 */
function feeds_views_api() {
  return array(
    'api' => '2.0',
    'path' => drupal_get_path('module', 'feeds') .'/views',
  );
}

/**
 * Implementation of hook_ctools_plugin_api().
 */
function feeds_ctools_plugin_api($owner, $api) {
  if ($owner == 'feeds' && $api == 'plugins') {
    return array('version' => 1);
  }
}

/**
 * Implementation of hook_ctools_plugin_plugins().
 *
 * Psuedo hook defintion plugin system options and defaults.
 */
function feeds_ctools_plugin_plugins() {
  return array(
    'cache' => TRUE,
    'use hooks' => TRUE,
  );
}

/**
 * Implementation of hook_feeds_plugins().
 */
function feeds_feeds_plugins() {
  module_load_include('inc', 'feeds', 'feeds.plugins');
  return _feeds_feeds_plugins();
}

/**
 * Implementation of hook_nodeapi().
 */
function feeds_nodeapi(&$node, $op, $form) {
  static $last_title;

  // Break out node processor related nodeapi functionality.
  _feeds_nodeapi_node_processor($node, $op);

  if ($importer_id = feeds_get_importer_id($node->type)) {
    switch ($op) {
      case 'validate':
        // On validation stage we are working with a FeedsSource object that is
        // not tied to a nid - when creating a new node there is $node->nid at
        // this stage.
        $source = feeds_source($importer_id);

        // If node title is empty, try to retrieve title from feed.
        if (trim($node->title) == '') {
          try {
            $source->addConfig($node->feeds);
            // @todo Too many indirections. Clean up.
            $batch = $source->importer->fetcher->fetch($source);
            $source->importer->parser->parse($batch, $source);
            // Keep the title in a static cache and populate $node->title on
            // 'presave' as node module looses any changes to $node after
            // 'validate'.
            if (!$last_title = $batch->getTitle()) {
              throw new Exception();
            }
          }
          catch (Exception $e) {
            drupal_set_message($e->getMessage(), 'error');
            form_set_error('title', t('Could not retrieve title from feed.'), 'error');
          }
        }

        // Invoke source
        // Node module magically moved $form['feeds'] to $node->feeds :P
        $source->configFormValidate($node->feeds);
        break;
      case 'presave':
        if (!empty($last_title)) {
          $node->title = $last_title;
        }
        $last_title = NULL;
        break;
      case 'insert':
      case 'update':
        // Add configuration to feed source and save.
        $source = feeds_source($importer_id, $node->nid);
        $source->addConfig($node->feeds);
        $source->save();

        // Refresh feed if import on create is selected and suppress_import is
        // not set.
        if ($op == 'insert' && feeds_importer($importer_id)->config['import_on_create'] && !isset($node->feeds['suppress_import'])) {
          feeds_batch_set(t('Importing'), 'import', $importer_id, $node->nid);
        }
        // Add import to scheduler.
        feeds_scheduler()->add($importer_id, 'import', $node->nid);
        // Add expiry to schedule, in case this is the first feed of this
        // configuration.
        feeds_scheduler()->add($importer_id, 'expire');
        break;
      case 'delete':
        // Remove feed from scheduler and delete source.
        feeds_scheduler()->remove($importer_id, 'import', $node->nid);
        feeds_source($importer_id, $node->nid)->delete();
        break;
    }
  }
}

/**
 * Break out FeedsNodeProcessor specific nodeapi operations.
 */
function _feeds_nodeapi_node_processor($node, $op) {
  switch ($op) {
    case 'load':
      if ($result = db_fetch_object(db_query('SELECT imported, guid, url, feed_nid FROM {feeds_node_item} WHERE nid = %d', $node->nid))) {
        $node->feeds_node_item = $result;
      }
      break;
    case 'insert':
      if (isset($node->feeds_node_item)) {
        $node->feeds_node_item->nid = $node->nid;
        drupal_write_record('feeds_node_item', $node->feeds_node_item);
      }
      break;
    case 'update':
      if (isset($node->feeds_node_item)) {
        $node->feeds_node_item->nid = $node->nid;
        drupal_write_record('feeds_node_item', $node->feeds_node_item, 'nid');
      }
      break;
    case 'delete':
      if (isset($node->feeds_node_item)) {
        db_query('DELETE FROM {feeds_node_item} WHERE nid = %d', $node->nid);
      }
      break;
  }
}

/**
 * Implementation of hook_form_alter().
 */
function feeds_form_alter(&$form, $form_state, $form_id) {
  if ($form['#id'] == 'node-form') {
    if ($importer_id = feeds_get_importer_id($form['type']['#value'])) {
      // Set title to not required, try to retrieve it from feed.
      $form['title']['#required'] = FALSE;

      // Build form.
      $source = feeds_source($importer_id, empty($form['nid']['#value']) ? 0 : $form['nid']['#value']);
      $form['feeds'] = array(
        '#type' => 'fieldset',
        '#title' => t('Feed'),
        '#tree' => TRUE,
      );
      $form['feeds'] += $source->configForm($form_state);
      $form['#feed_id'] = $importer_id;
    }
  }
}

/**
 * Refreshes a feed identified by $feed_info.
 *
 * Used as a worker callback for drupal_queue.
 *
 * @param $feed_info
 *   Array where the key 'id' is the id of a FeedsImporter object and the key
 *   'feed_nid' is the node id of feed node.
 */
function feeds_scheduler_work($feed_info) {
  feeds_scheduler()->work($feed_info);
}

/**
 * @}
 */

/**
 * @defgroup batch Batch functions
 */

/**
 * Batch helper.
 *
 * @param $title
 *   Title to show to user when executing batch.
 * @param $method
 *   Method to execute on importer; one of 'import', 'clear' or 'expire'.
 * @param $importer_id
 *   Identifier of a FeedsImporter object.
 * @param $feed_nid
 *   If importer is attached to content type, feed node id identifying the
 *   source to be imported.
 */
function feeds_batch_set($title, $method, $importer_id, $feed_nid = 0) {
  $batch = array(
    'title' => $title,
    'operations' => array(
      array('feeds_batch', array($method, $importer_id, $feed_nid)),
    ),
    'progress_message' => '',
  );
  batch_set($batch);
}

/**
 * Batch callback.
 *
 * @param $method
 *   Method to execute on importer; one of 'import' or 'clear'.
 * @param $importer_id
 *   Identifier of a FeedsImporter object.
 * @param $feed_nid
 *   If importer is attached to content type, feed node id identifying the
 *   source to be imported.
 * @param $context
 *   Batch context.
 */
function feeds_batch($method, $importer_id, $feed_nid = 0, &$context) {
  $context['finished'] = 1;
  try {
    $context['finished'] = feeds_source($importer_id, $feed_nid)->$method();
  }
  catch (Exception $e) {
    drupal_set_message($e->getMessage(), 'error');
  }
}

/**
 * @}
 */

/**
 * @defgroup utility Utility functions
 * @{
 */

/**
 * Load all importers.
 *
 * @param $load_disabled
 *   Pass TRUE to load all importers, enabled or disabled, pass FALSE to only
 *   retrieve enabled importers.
 *
 * @return
 *   An array of all feed configurations available.
 */
function feeds_importer_load_all($load_disabled = FALSE) {
  $feeds = array();
  // This function can get called very early in install process through
  // menu_router_rebuild(). Do not try to include CTools if not available.
  if (function_exists('ctools_include')) {
    ctools_include('export');
    $configs = ctools_export_load_object('feeds_importer', 'all');
    foreach ($configs as $config) {
      if (!empty($config->id) && ($load_disabled || empty($config->disabled))) {
        $feeds[$config->id] = feeds_importer($config->id);
      }
    }
  }
  return $feeds;
}

/**
 * Return whether a importer is enabled or not.
 *
 * Use only for importer configurations that ship with feeds. This function
 * assumes that importers are initially off and then enabled manually on
 * admin/build/feeds.
 *
 * @return
 *   TRUE if the importer is enabled, FALSE if not.
 */
function feeds_importer_enabled($id) {
  // 'default_feeds_importer' variable is used internally by CTools to determine
  // the enabled/disabled state of a configuration. This variable may not be
  // completely populated.
  $disabled = variable_get('default_feeds_importer', FALSE);
  if ($disabled === FALSE) {
    $disabled = array();
    foreach (feeds_importer_load_all() as $importer) {
      $disabled[$importer->id] = $disabled->disabled;
    }
    variable_set('default_feeds_importer', $disabled);
  }
  // If disabled is not present, assume that importer configuration is disabled.
  if (!isset($disabled[$id])) {
    return FALSE;
  }
  return !$disabled[$id];
}

/**
 * Get a an enabled importer configuration by content type.
 *
 * @todo speed this up by DB caching the result.
 *
 * @param $content_type
 *   A node type string.
 *
 * @return
 *   A FeedsImporter id if there is an importer for the given content type,
 *   FALSE otherwise.
 */
function feeds_get_importer_id($content_type) {
  static $ids = array();
  if (!isset($ids[$content_type])) {
    $ids[$content_type] = FALSE;
    foreach (feeds_importer_load_all() as $importer) {
      if ((!$importer->disabled) && $importer->config['content_type'] == $content_type) {
        $ids[$content_type] = $importer->id;
        break;
      }
    }
  }
  return $ids[$content_type];
}

/**
 * Export a FeedsImporter configuration to code.
 */
function feeds_export($importer_id, $indent = '') {
  ctools_include('export');
  $result = ctools_export_load_object('feeds_importer', 'names', array('id' => $importer_id));
  if (isset($result[$importer_id])) {
    return ctools_export_object('feeds_importer', $result[$importer_id], $indent);
  }
}

/**
 * Log to a file like /mytmp/feeds_my_domain_org.log in temporary directory.
 */
function feeds_dbg($msg) {
  if (variable_get('feeds_debug', false)) {
    if (!is_string($msg)) {
      $msg = var_export($msg, true);
    }
    $filename = trim(str_replace('/', '_', $_SERVER['HTTP_HOST'] . base_path()), '_');
    $handle = fopen(file_directory_temp() ."/feeds_$filename.log", 'a');
    fwrite($handle, date('c') ."\t$msg\n");
    fclose($handle);
  }
}

/**
 * @}
 */

/**
 * @defgroup instantiators Instantiators
 * @{
 */

/**
 * Get an importer instance.
 *
 * @param $id
 *   The unique id of the importer object.
 *
 * @return
 *   A FeedsImporter object or an object of a class defined by the Drupal
 *   variable 'feeds_importer_class'. There is only one importer object
 *   per $id system-wide.
 */
function feeds_importer($id) {
  feeds_include('FeedsImporter');
  return FeedsConfigurable::instance(variable_get('feeds_importer_class', 'FeedsImporter'), $id);
}

/**
 * Get an instance of a source object.
 *
 * @param $importer_id
 *   A FeedsImporter id.
 * @param $feed_nid
 *   The node id of a feed node if the source is attached to a feed node.
 *
 * @return
 *   A FeedsSource object or an object of a class defiend by the Drupal
 *   variable 'source_class'.
 */
function feeds_source($importer_id, $feed_nid = 0) {
  feeds_include('FeedsImporter');
  return FeedsSource::instance($importer_id, $feed_nid);
}

/**
 * Get a scheduler instance.
 *
 * @return
 *   A FeedsScheduler object or an object of a class defined by the Drupal
 *   variable 'feeds_scheduler_class'.
 */
function feeds_scheduler() {
  feeds_include('FeedsImporter');
  feeds_include('FeedsScheduler');
  return FeedsScheduler::instance();
}

/**
 * @}
 */

/**
 * @defgroup plugins Plugin functions
 * @{
 *
 * @todo Encapsulate this in a FeedsPluginHandler class, move it to includes/
 * and only load it if we're manipulating plugins.
 */

/**
 * Get all available plugins. Does not list hidden plugins.
 *
 * @return
 *   An array where the keys are the plugin keys and the values
 *   are the plugin info arrays as defined in hook_feeds_plugins().
 */
function feeds_get_plugins() {
  ctools_include('plugins');
  $plugins = ctools_get_plugins('feeds', 'plugins');

  $result = array();
  foreach ($plugins as $key => $info) {
    if (!empty($info['hidden'])) {
      continue;
    }
    $result[$key] = $info;
  }

  // Sort plugins by name and return.
  uasort($result, 'feeds_plugin_compare');
  return $result;
}

/**
 * Sort callback for feeds_get_plugins().
 */
function feeds_plugin_compare($a, $b) {
  return strcasecmp($a['name'], $b['name']);
}

/**
 * Get all available plugins of a particular type.
 *
 * @param $type
 *   'fetcher', 'parser' or 'processor'
 */
function feeds_get_plugins_by_type($type) {
  $plugins = feeds_get_plugins();

  $result = array();
  foreach ($plugins as $key => $info) {
    if ($type == feeds_plugin_type($key)) {
      $result[$key] = $info;
    }
  }
  return $result;
}

/**
 * Get an instance of a class for a given plugin and id.
 *
 * @param $plugin
 *   A string that is the key of the plugin to load.
 * @param $id
 *   A string that is the id of the object.
 *
 * @return
 *   A FeedsPlugin object.
 *
 * @throws Exception
 *   If plugin can't be instantiated.
 */
function feeds_plugin_instance($plugin, $id) {
  feeds_include('FeedsImporter');
  ctools_include('plugins');
  if ($class = ctools_plugin_load_class('feeds', 'plugins', $plugin, 'handler')) {
    return FeedsConfigurable::instance($class, $id);
  }
  drupal_set_message(t('Missing Feeds plugin. Check whether all required libraries and modules are installed properly.'), 'warning');
  $class = ctools_plugin_load_class('feeds', 'plugins', 'FeedsMissingPlugin', 'handler');
  return FeedsConfigurable::instance($class, $id);
}

/**
 * Determines whether given plugin is derived from given base plugin.
 *
 * @todo Update variable names. Should be $plugin_key if it is a key string.
 *
 * @param $plugin
 *   String that identifies a Feeds plugin key.
 * @param $parent_plugin
 *   String that identifies a Feeds plugin key to be tested against.
 *
 * @return
 *   TRUE if $parent_plugin is directly *or indirectly* a parent of $plugin,
 *   FALSE otherwise.
 */
function feeds_plugin_child($plugin, $parent_plugin) {
  ctools_include('plugins');
  $plugins = ctools_get_plugins('feeds', 'plugins');
  $info = $plugins[$plugin];

  if (empty($info['handler']['parent'])) {
    return FALSE;
  }
  elseif ($info['handler']['parent'] == $parent_plugin) {
    return TRUE;
  }
  else {
    return feeds_plugin_child($info['handler']['parent'], $parent_plugin);
  }
}

/**
 * Determine the type of a plugin.
 *
 * @param $plugin_key
 *   String that is a Feeds plugin key.
 *
 * @return
 *   One of the following values:
 *   'fetcher' if the plugin is a fetcher
 *   'parser' if the plugin is a parser
 *   'processor' if the plugin is a processor
 *   FALSE otherwise.
 */
function feeds_plugin_type($plugin_key) {
  if (feeds_plugin_child($plugin_key, 'FeedsFetcher')) {
    return 'fetcher';
  }
  elseif (feeds_plugin_child($plugin_key, 'FeedsParser')) {
    return 'parser';
  }
  elseif (feeds_plugin_child($plugin_key, 'FeedsProcessor')) {
    return 'processor';
  }
  return FALSE;
}

/**
 * @}
 */

/**
 * @defgroup include Funtions for loading libraries
 * @{
 */

/**
 * Includes a feeds module include file.
 *
 * @param $file
 *   The filename without the .inc extension.
 * @param $directory
 *   The directory to include the file from. Do not include files from libraries
 *   directory. Use feeds_include_library() instead
 */
function feeds_include($file, $directory = 'includes') {
  static $included = array();
  if (!isset($included[$file])) {
    require './'. drupal_get_path('module', 'feeds') ."/$directory/$file.inc";
  }
  $included[$file] = TRUE;
}

/**
 * Include a library file.
 *
 * @param $file
 *   The filename to load from.
 * @param $library
 *   The name of the library. If libraries module is installed,
 *   feeds_include_library() will look for libraries with this name managed by
 *   libraries module.
 */
function feeds_include_library($file, $library) {
  static $included = array();
  if (!isset($included[$file])) {
    // Try first whether libraries module is present and load the file from
    // there. If this fails, require the library from the local path.
    if (module_exists('libraries') && file_exists(libraries_get_path($library) ."/$file")) {
      require libraries_get_path($library) ."/$file";
    }
    else {
      require './'. drupal_get_path('module', 'feeds') ."/libraries/$file";
    }
  }
  $included[$file] = TRUE;
}

/**
 * Checks whether a library is present.
 *
 * @param $file
 *   The filename to load from.
 * @param $library
 *   The name of the library. If libraries module is installed,
 *   feeds_library_exists() will look for libraries with this name managed by
 *   libraries module.
 */
function feeds_library_exists($file, $library) {
  if (module_exists('libraries') && file_exists(libraries_get_path($library) ."/$file")) {
    return TRUE;
  }
  elseif (file_exists(drupal_get_path('module', 'feeds') ."/libraries/$file")) {
    return TRUE;
  }
  return FALSE;
}

/**
 * @}
 */