class SearchApiStopWords in Search API 7

Processor for removing stopwords from index and search terms.

Hierarchy

• class \SearchApiAbstractProcessor implements SearchApiProcessorInterface
  • class \SearchApiStopWords

1 string reference to 'SearchApiStopWords'

search_api_search_api_processor_info in ./search_api.module

Implements hook_search_api_processor_info().

File

includes/processor_stopwords.inc, line 11

Contains SearchApiStopWords.

class SearchApiStopWords extends SearchApiAbstractProcessor {

  /**
   * Holds all words ignored for the last query.
   *
   * @var array
   */
  protected $ignored = array();
  public function configurationForm() {
    $form = parent::configurationForm();
    $form += array(
      'help' => array(
        '#markup' => '<p>' . t('Provide a stopwords file or enter the words in this form. If you do both, both will be used. Read about !stopwords.', array(
          '!stopwords' => l(t('stop words'), "http://en.wikipedia.org/wiki/Stop_words"),
        )) . '</p>',
      ),
      'file' => array(
        '#type' => 'textfield',
        '#title' => t('Stopwords file'),
        '#description' => t('This must be a stream-type description like <code>public://stopwords/stopwords.txt</code> or <code>http://example.com/stopwords.txt</code> or <code>private://stopwords.txt</code>.'),
      ),
      'stopwords' => array(
        '#type' => 'textarea',
        '#title' => t('Stopwords'),
        '#description' => t('Enter a space and/or linebreak separated list of stopwords that will be removed from content before it is indexed and from search terms before searching.'),
        '#default_value' => t("but\ndid\nthe this that those\netc"),
      ),
    );
    if (!empty($this->options)) {
      $form['file']['#default_value'] = $this->options['file'];
      $form['stopwords']['#default_value'] = $this->options['stopwords'];
    }
    return $form;
  }
  public function configurationFormValidate(array $form, array &$values, array &$form_state) {
    parent::configurationFormValidate($form, $values, $form_state);
    $uri = $values['file'];
    if (!empty($uri) && !@file_get_contents($uri)) {
      $el = $form['file'];
      form_error($el, t('Stopwords file') . ': ' . t('The file %uri is not readable or does not exist.', array(
        '%uri' => $uri,
      )));
    }
  }
  public function process(&$value) {
    $stopwords = $this
      ->getStopWords();
    if (empty($stopwords) || !is_string($value)) {
      return;
    }
    $words = preg_split('/\\s+/', $value);
    foreach ($words as $sub_key => $sub_value) {
      if (isset($stopwords[$sub_value])) {
        unset($words[$sub_key]);
        $this->ignored[] = $sub_value;
      }
    }
    $value = implode(' ', $words);
  }
  public function preprocessSearchQuery(SearchApiQuery $query) {
    $this->ignored = array();
    parent::preprocessSearchQuery($query);
  }
  public function postprocessSearchResults(array &$response, SearchApiQuery $query) {
    if ($this->ignored) {
      if (isset($response['ignored'])) {
        $response['ignored'] = array_merge($response['ignored'], $this->ignored);
      }
      else {
        $response['ignored'] = $this->ignored;
      }
    }
  }

  /**
   * Retrieves the processor's configured stopwords.
   *
   * @return array
   *   An array whose keys are the stopwords set in either the file or the text
   *   field.
   */
  protected function getStopWords() {
    if (isset($this->stopwords)) {
      return $this->stopwords;
    }
    $file_words = $form_words = array();
    if (!empty($this->options['file']) && ($stopwords_file = file_get_contents($this->options['file']))) {
      $file_words = preg_split('/\\s+/', $stopwords_file);
    }
    if (!empty($this->options['stopwords'])) {
      $form_words = preg_split('/\\s+/', $this->options['stopwords']);
    }
    $this->stopwords = array_flip(array_merge($file_words, $form_words));
    return $this->stopwords;
  }

}

Members

Name	Modifiers	Type	Description	Overrides
SearchApiAbstractProcessor::$index	protected	property
SearchApiAbstractProcessor::$options	protected	property
SearchApiAbstractProcessor::configurationFormSubmit	public	function	Submit callback for the form returned by configurationForm(). Overrides SearchApiProcessorInterface::configurationFormSubmit
SearchApiAbstractProcessor::implodeTokens	protected	function	Internal helper function for imploding tokens into a single string.
SearchApiAbstractProcessor::normalizeTokens	protected	function	Internal helper function for normalizing tokens.
SearchApiAbstractProcessor::preprocessIndexItems	public	function	Calls processField() for all appropriate fields. Overrides SearchApiProcessorInterface::preprocessIndexItems
SearchApiAbstractProcessor::processField	protected	function	Method for preprocessing field data.
SearchApiAbstractProcessor::processFieldValue	protected	function	Called for processing a single text element in a field. The default implementation just calls process().	2
SearchApiAbstractProcessor::processFilters	protected	function	Method for preprocessing query filters.
SearchApiAbstractProcessor::processFilterValue	protected	function	Called for processing a single filter value. The default implementation just calls process().
SearchApiAbstractProcessor::processKey	protected	function	Called for processing a single search keyword. The default implementation just calls process().
SearchApiAbstractProcessor::processKeys	protected	function	Method for preprocessing search keys.
SearchApiAbstractProcessor::supportsIndex	public	function	Check whether this processor is applicable for a certain index. Overrides SearchApiProcessorInterface::supportsIndex
SearchApiAbstractProcessor::testField	protected	function	Determines whether to process data from the given field.
SearchApiAbstractProcessor::testType	protected	function	Determines whether fields of the given type should normally be processed.
SearchApiAbstractProcessor::__construct	public	function	Constructor, saving its arguments into properties. Overrides SearchApiProcessorInterface::__construct	2
SearchApiStopWords::$ignored	protected	property	Holds all words ignored for the last query.
SearchApiStopWords::configurationForm	public	function	Display a form for configuring this processor. Since forcing users to specify options for disabled processors makes no sense, none of the form elements should have the '#required' attribute set. Overrides SearchApiAbstractProcessor::configurationForm
SearchApiStopWords::configurationFormValidate	public	function	Validation callback for the form returned by configurationForm(). Overrides SearchApiAbstractProcessor::configurationFormValidate
SearchApiStopWords::getStopWords	protected	function	Retrieves the processor's configured stopwords.
SearchApiStopWords::postprocessSearchResults	public	function	Does nothing. Overrides SearchApiAbstractProcessor::postprocessSearchResults
SearchApiStopWords::preprocessSearchQuery	public	function	Calls processKeys() for the keys and processFilters() for the filters. Overrides SearchApiAbstractProcessor::preprocessSearchQuery
SearchApiStopWords::process	public	function	Function that is ultimately called for all text by the standard implementation, and does nothing by default. Overrides SearchApiAbstractProcessor::process