FeedsImporter.inc

<?php

/**
 * @file
 * FeedsImporter class and related.
 */

/**
 * A FeedsImporter object describes how an external source should be fetched,
 * parsed and processed. Feeds can manage an arbitrary amount of importers.
 *
 * A FeedsImporter holds a pointer to a FeedsFetcher, a FeedsParser and a
 * FeedsProcessor plugin. It further contains the configuration for itself and
 * each of the three plugins.
 *
 * Its most important responsibilities are configuration management, interfacing
 * with the job scheduler and expiring of all items produced by this
 * importer.
 *
 * When a FeedsImporter is instantiated, it loads its configuration. Then it
 * instantiates one fetcher, one parser and one processor plugin depending on
 * the configuration information. After instantiating them, it sets them to
 * the configuration information it holds for them.
 */
class FeedsImporter extends FeedsConfigurable {

  // Every feed has a fetcher, a parser and a processor.
  // These variable names match the possible return values of
  // FeedsPlugin::typeOf().
  protected $fetcher, $parser, $processor;

  // This array defines the variable names of the plugins above.
  protected $plugin_types = array('fetcher', 'parser', 'processor');

  /**
   * Instantiate class variables, initialize and configure
   * plugins.
   */
  protected function __construct($id) {
    parent::__construct($id);

    // Try to load information from database.
    $this->load();

    // Instantiate fetcher, parser and processor, set their configuration if
    // stored info is available.
    foreach ($this->plugin_types as $type) {
      $plugin = feeds_plugin($this->config[$type]['plugin_key'], $this->id);

      if (isset($this->config[$type]['config'])) {
        $plugin->setConfig($this->config[$type]['config']);
      }
      $this->$type = $plugin;
    }
  }

  /**
   * Report how many items *should* be created on one page load by this
   * importer.
   *
   * Note:
   *
   * It depends on whether parser implements batching if this limit is actually
   * respected. Further, if no limit is reported it doesn't mean that the
   * number of items that can be created on one page load is actually without
   * limit.
   *
   * @return
   *   A positive number defining the number of items that can be created on
   *   one page load. 0 if this number is unlimited.
   */
  public function getLimit() {
    return $this->processor->getLimit();
  }

  /**
   * Save configuration.
   */
  public function save() {
    $save = new stdClass();
    $save->id = $this->id;
    $save->config = $this->getConfig();

    if ($config = db_query("SELECT config FROM {feeds_importer} WHERE id = :id", array(':id' => $this->id))->fetchField()) {
      drupal_write_record('feeds_importer', $save, 'id');
      // Only rebuild menu if content_type has changed. Don't worry about
      // rebuilding menus when creating a new importer since it will default
      // to the standalone page.
      $config = unserialize($config);
      if ($config['content_type'] != $save->config['content_type']) {
        variable_set('menu_rebuild_needed', TRUE);
      }
    }
    else {
      drupal_write_record('feeds_importer', $save);
    }
  }

  /**
   * Load configuration and unpack.
   */
  public function load() {
    ctools_include('export');
    if ($config = ctools_export_load_object('feeds_importer', 'conditions', array('id' => $this->id))) {
      $config = array_shift($config);
      $this->export_type = $config->export_type;
      $this->disabled = isset($config->disabled) ? $config->disabled : FALSE;
      $this->config = $config->config;
      return TRUE;
    }
    return FALSE;
  }

  /**
   * Deletes configuration.
   *
   * Removes configuration information from database, does not delete
   * configuration itself.
   */
  public function delete() {
    db_delete('feeds_importer')
      ->condition('id', $this->id)
      ->execute();

    feeds_reschedule($this->id);
  }

  /**
   * Set plugin.
   *
   * @param $plugin_key
   *   A fetcher, parser or processor plugin.
   *
   * @todo Error handling, handle setting to the same plugin.
   */
  public function setPlugin($plugin_key) {
    // $plugin_type can be either 'fetcher', 'parser' or 'processor'
    if ($plugin_type = FeedsPlugin::typeOf($plugin_key)) {
      if ($plugin = feeds_plugin($plugin_key, $this->id)) {
        // Unset existing plugin, switch to new plugin.
        unset($this->$plugin_type);
        $this->$plugin_type = $plugin;
        // Set configuration information, blow away any previous information on
        // this spot.
        $this->config[$plugin_type] = array('plugin_key' => $plugin_key);
      }
    }
  }

  /**
   * Copy a FeedsImporter configuration into this importer.
   *
   * @param FeedsConfigurable $configurable
   *   The feeds importer object to copy from.
   */
   public function copy(FeedsConfigurable $configurable) {
     parent::copy($configurable);

     if ($configurable instanceof FeedsImporter) {
       // Instantiate new fetcher, parser and processor and initialize their
       // configurations.
       foreach ($this->plugin_types as $plugin_type) {
         $this->setPlugin($configurable->config[$plugin_type]['plugin_key']);
         $this->$plugin_type->setConfig($configurable->config[$plugin_type]['config']);
       }
     }
   }

  /**
   * Get configuration of this feed.
   */
  public function getConfig() {
    foreach (array('fetcher', 'parser', 'processor') as $type) {
      $this->config[$type]['config'] = $this->$type->getConfig();
    }
    return parent::getConfig();
  }

  /**
   * Return defaults for feed configuration.
   */
  public function configDefaults() {
    return array(
      'name' => '',
      'description' => '',
      'fetcher' => array(
        'plugin_key' => 'FeedsHTTPFetcher',
      ),
      'parser' => array(
        'plugin_key' => 'FeedsSyndicationParser',
      ),
      'processor' => array(
        'plugin_key' => 'FeedsNodeProcessor',
      ),
      'content_type' => '',
      'update' => 0,
      'import_period' => 1800, // Refresh every 30 minutes by default.
      'expire_period' => 3600, // Expire every hour by default, this is a hidden setting.
      'import_on_create' => TRUE, // Import on submission.
      'process_in_background' => FALSE,
    );
  }

  /**
   * Override parent::configForm().
   */
  public function configForm(&$form_state) {
    $config = $this->getConfig();
    $form = array();
    $form['name'] = array(
      '#type' => 'textfield',
      '#title' => t('Name'),
      '#description' => t('A human readable name of this importer.'),
      '#default_value' => $config['name'],
      '#required' => TRUE,
    );
    $form['description'] = array(
      '#type' => 'textfield',
      '#title' => t('Description'),
      '#description' => t('A description of this importer.'),
      '#default_value' => $config['description'],
    );
    $node_types = node_type_get_names();
    array_walk($node_types, 'check_plain');
    $form['content_type'] = array(
      '#type' => 'select',
      '#title' => t('Attach to content type'),
      '#description' => t('If "Use standalone form" is selected a source is imported by using a form under !import_form.
                           If a content type is selected a source is imported by creating a node of that content type.',
                           array('!import_form' => l(url('import', array('absolute' => TRUE)), 'import', array('attributes' => array('target' => '_new'))))),
      '#options' => array('' => t('Use standalone form')) + $node_types,
      '#default_value' => $config['content_type'],
    );
    $cron_required =  ' ' . l(t('Requires cron to be configured.'), 'http://drupal.org/cron', array('attributes' => array('target' => '_new')));
    $period = drupal_map_assoc(array(900, 1800, 3600, 10800, 21600, 43200, 86400, 259200, 604800, 2419200), 'format_interval');
    foreach ($period as &$p) {
      $p = t('Every !p', array('!p' => $p));
    }
    $period = array(
      FEEDS_SCHEDULE_NEVER => t('Off'),
      0 => t('As often as possible'),
    ) + $period;
    $form['import_period'] = array(
      '#type' => 'select',
      '#title' => t('Periodic import'),
      '#options' => $period,
      '#description' => t('Choose how often a source should be imported periodically.') . $cron_required,
      '#default_value' => $config['import_period'],
    );
    $form['import_on_create'] = array(
      '#type' => 'checkbox',
      '#title' => t('Import on submission'),
      '#description' => t('Check if import should be started at the moment a standalone form or node form is submitted.'),
      '#default_value' => $config['import_on_create'],
    );
    $form['process_in_background'] = array(
      '#type' => 'checkbox',
      '#title' => t('Process in background'),
      '#description' => t('For very large imports. If checked, import and delete tasks started from the web UI will be handled by a cron task in the background rather than by the browser. This does not affect periodic imports, they are handled by a cron task in any case.') . $cron_required,
      '#default_value' => $config['process_in_background'],
    );
    return $form;
  }

  /**
   * Reschedule if import period changes.
   */
  public function configFormSubmit(&$values) {
    if ($this->config['import_period'] != $values['import_period']) {
      feeds_reschedule($this->id);
    }
    parent::configFormSubmit($values);
  }

  /**
   * Implements FeedsConfigurable::dependencies().
   */
  public function dependencies() {
    $dependencies = parent::dependencies();
    foreach ($this->plugin_types as $plugin_type) {
      $dependencies = array_merge($dependencies, $this->$plugin_type->dependencies());
    }
    return $dependencies;
  }
}

/**
 * Helper, see FeedsDataProcessor class.
 */
function feeds_format_expire($timestamp) {
  if ($timestamp == FEEDS_EXPIRE_NEVER) {
    return t('Never');
  }
  return t('after !time', array('!time' => format_interval($timestamp)));
}