FeedsImporter.inc in Feeds 7.2

FeedsImporter class and related.

File

includes/FeedsImporter.inc

<?php

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

/**
 * Class for a Feeds importer.
 *
 * 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().
   */

  /**
   * The selected fetcher for this importer.
   *
   * @var FeedsFetcher|null
   */
  protected $fetcher;

  /**
   * The selected parser for this importer.
   *
   * @var FeedsParser|null
   */
  protected $parser;

  /**
   * The selected processor for this importer.
   *
   * @var FeedsProcessor|null
   */
  protected $processor;

  /**
   * This array defines the variable names of the plugins above.
   *
   * @var string[]
   */
  protected $plugin_types = array(
    'fetcher',
    'parser',
    'processor',
  );

  /**
   * Instantiate class variables, initialize and configure plugins.
   *
   * @param string $id
   *   The importer ID.
   */
  protected function __construct($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;
    }
  }

  /**
   * Reports how many items *should* be created on one request 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 int
   *   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 string $plugin_key
   *   The name of a fetcher, parser or processor plugin.
   *
   * @todo Error handling, handle setting to the same plugin.
   */
  public function setPlugin($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();
  }

  /**
   * Validates the configuration.
   */
  public function validateConfig() {
    $errors = parent::validateConfig();
    $config = $this
      ->getConfig();

    // Check if "Attach to content type" setting is the same as content type on
    // node processor.
    if ($this->processor instanceof FeedsNodeProcessor) {
      $processor_config = $this->processor
        ->getConfig();
      if (!empty($config['content_type']) && $processor_config['bundle'] == $config['content_type']) {
        $message = t('The importer is attached to the same content type as the content type selected on the node processor. Unless you have a very advanced use case, these two should never be the same.');
        if (!module_exists('feeds_news')) {
          $message .= ' ' . t('Enable the Feeds News module for an example of how the "@content_type" setting can be used.', array(
            '@content_type' => t('Attach to content type'),
          ));
        }
        elseif ($this->id != 'feed') {
          $message .= ' ' . t('See the importer !importer_name (provided by the Feeds News module) for an example of how the "@content_type" setting can be used.', array(
            '!importer_name' => l(check_plain(feeds_importer('feed')->config['name']), 'admin/structure/feeds/feed'),
            '@content_type' => t('Attach to content type'),
          ));
        }
        $errors[] = $message;
      }
    }

    // Check for a combination of incompatible settings.
    if (!$config['import_on_create'] && empty($config['content_type'])) {
      if ($config['import_period'] == FEEDS_SCHEDULE_NEVER) {
        $errors[] = t('"@import_period" and "@import_on_create" are both turned off and the importer is not attached to a content type. Unless you have alternative methods of running imports for this importer, Feeds will not import anything for this importer.', array(
          '@import_period' => t('Periodic import'),
          '@import_on_create' => t('Import on submission'),
        ));
      }
      elseif ($config['process_in_background']) {
        $errors[] = t('Since "@import_on_create" is turned off and the importer is not attached to a content type, the "@process_in_background" setting may have no effect. When submitting the standalone form with the "@import_on_create" setting turned off, the feed is only scheduled for periodic import.', array(
          '@import_on_create' => t('Import on submission'),
          '@process_in_background' => t('Process in background'),
        ));
      }
    }

    // Validate errors of each plugin as well.
    $plugin_types = array(
      'fetcher' => t('Fetcher'),
      'parser' => t('Parser'),
      'processor' => t('Processor'),
    );
    foreach ($plugin_types as $type => $plugin_label) {

      // Check if plugin exists.
      $plugin = feeds_plugin($this->config[$type]['plugin_key'], $this->id);
      if ($plugin instanceof FeedsMissingPlugin) {
        $errors[] = t('The plugin %plugin is unavailable.', array(
          '%plugin' => $this->config[$type]['plugin_key'],
        ));
        continue;
      }
      $plugin_errors = $this->{$type}
        ->validateConfig();
      foreach ($plugin_errors as $key => $error) {
        $errors[] = t('@plugin: !error', array(
          '@plugin' => $plugin_label,
          '!error' => $error,
        ));
      }
    }
    return $errors;
  }

  /**
   * 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,
      // Refresh every 30 minutes by default.
      'import_period' => 1800,
      // Expire every hour by default, this is a hidden setting.
      'expire_period' => 3600,
      // Import on submission.
      'import_on_create' => TRUE,
      'process_in_background' => FALSE,
    ) + parent::configDefaults();
  }

  /**
   * Overrides parent::configForm().
   *
   * @param array $form_state
   *   The current state of the form.
   *
   * @return array
   *   The form definition.
   */
  public function configForm(&$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' => '<p>' . 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',
          ),
        )),
      )) . '</p>',
      '#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. If not checked and when using the standalone form, be sure to configure periodic import.'),
      '#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;
  }

  /**
   * Overrides parent::configFormSubmit().
   *
   * Reschedule if import period changes.
   *
   * @param array $values
   *   An array that contains the values entered by the user through configForm.
   */
  public function configFormSubmit(&$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;
  }

}

/**
 * Formats a time interval.
 *
 * @param int $timestamp
 *   The length of the interval in seconds.
 *
 * @return string
 *   A translated string representation of the interval.
 */
function feeds_format_expire($timestamp) {
  if ($timestamp == FEEDS_EXPIRE_NEVER) {
    return t('Never');
  }
  return t('after !time', array(
    '!time' => format_interval($timestamp),
  ));
}

Functions

Name	Description
feeds_format_expire	Formats a time interval.

Classes

Name	Description
FeedsImporter	Class for a Feeds importer.