94 lines
3.4 KiB
PHP
94 lines
3.4 KiB
PHP
<?php
|
|
|
|
/**
|
|
* Processor for removing stopwords from index and search terms.
|
|
*/
|
|
class SearchApiStopWords extends SearchApiAbstractProcessor {
|
|
|
|
public function configurationForm() {
|
|
$form = parent::configurationForm();
|
|
|
|
$form += array(
|
|
'help' => array(
|
|
'#markup' => '<p>' . t('Provide a stopwords file or enter the words in this form. If you do both, both will be used. Read about !stopwords.', array('!stopwords' => l(t('stop words'), "http://en.wikipedia.org/wiki/Stop_words"))) . '</p>',
|
|
),
|
|
'file' => array(
|
|
'#type' => 'textfield',
|
|
'#title' => t('Stopwords file URI'),
|
|
'#title' => t('Enter the URI of your stopwords.txt file'),
|
|
'#description' => t('This must be a stream-type description like <code>public://stopwords/stopwords.txt</code> or <code>http://example.com/stopwords.txt</code> or <code>private://stopwords.txt</code>.'),
|
|
),
|
|
'stopwords' => array(
|
|
'#type' => 'textarea',
|
|
'#title' => t('Stopwords'),
|
|
'#description' => t('Enter a space and/or linebreak separated list of stopwords that will be removed from content before it is indexed and from search terms before searching.'),
|
|
'#default_value' => t("but\ndid\nthe this that those\netc"),
|
|
),
|
|
);
|
|
|
|
if (!empty($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);
|
|
|
|
$stopwords = trim($values['stopwords']);
|
|
$uri = $values['file'];
|
|
if (empty($stopwords) && empty($uri)) {
|
|
$el = $form['file'];
|
|
form_error($el, $el['#title'] . ': ' . t('At stopwords file or words are required.'));
|
|
}
|
|
if (!empty($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)) {
|
|
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(' ', $words);
|
|
}
|
|
|
|
public function postprocessSearchResults(array &$response, SearchApiQuery $query) {
|
|
if (isset($this->ignored)) {
|
|
if (isset($response['ignored'])) {
|
|
$response['ignored'] = array_merge($response['ignored'], $this->ignored);
|
|
}
|
|
else $response['ignored'] = $this->ignored;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* @return
|
|
* An array whose keys are the stopwords set in either the file or the text
|
|
* field.
|
|
*/
|
|
protected function getStopWords() {
|
|
if (isset($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;
|
|
}
|
|
} |