/**
* Get index
* @return \ZendSearch\Lucene\Index
*/
private function index()
{
if (!isset(self::$index)) {
$analyzer = new CaseInsensitive();
if ($this->config()->exists('zend_search', 'stop_words')) {
$stop_word_filter = new StopWords();
$words = $this->getRealPath($this->config()->get('zend_search', 'stop_words'));
if ($words !== false) {
$stop_word_filter->loadFromFile($words);
} else {
throw new \InvalidArgumentException('Path not found');
}
$analyzer->addFilter($stop_word_filter);
}
if ($this->config()->exists('zend_search', 'morphy_dicts')) {
$morphy_dicts = $this->getRealPath($this->config()->get('zend_search', 'morphy_dicts'));
if ($morphy_dicts !== false) {
$analyzer->addFilter(new Morphy($morphy_dicts, $this->config()->getCharset()));
} else {
throw new \InvalidArgumentException('Path not found');
}
}
Analyzer::setDefault($analyzer);
Lucene::setResultSetLimit($this->limit);
QueryParser::setDefaultEncoding($this->config()->getCharset());
$index = $this->config() - get('zend_search', 'index');
$path = $this->getRealPath($index);
self::$index = $path ? Lucene::open($path) : Lucene::create($index);
}
return self::$index;
}
/**
* Lists all Post models.
* @return mixed
*/
public function actionIndex()
{
$searchModel = new PostSearch();
$dataProvider = $searchModel->search(Yii::$app->request->post());
//setlocale(LC_ALL, 'en_US.UTF-8');
setlocale(LC_CTYPE, 'ru_RU.UTF-8');
//Lucene\Lucene::setDefaultSearchField('contents');
Lucene\Search\QueryParser::setDefaultEncoding('UTF-8');
Lucene\Analysis\Analyzer\Analyzer::setDefault(new Lucene\Analysis\Analyzer\Common\Utf8\CaseInsensitive());
Lucene\Lucene::setResultSetLimit(10);
// create blog posts index located in /data/posts_index ,make sure the folder is writable
$index = Lucene\Lucene::create('data/posts_index');
$posts = Post::find()->all();
//var_dump($posts);die();
// iterate through posts and build the index
foreach ($posts as $p) {
$doc = new Lucene\Document();
$doc->addField(Lucene\Document\Field::UnIndexed('entry_id', $p->id));
$doc->addField(Lucene\Document\Field::Keyword('title', $p->title));
$doc->addField(Lucene\Document\Field::text('contents', $p->content));
$index->addDocument($doc);
}
// commit the index
$index->commit();
//Lucene\Analysis\Analyzer\Analyzer::setDefault(new Lucene\Analysis\Analyzer\Common\Utf8\CaseInsensitive());
// explode the search query to individual words
$words = explode(' ', urldecode(Yii::$app->getRequest()->getQueryParam('q')));
// start a search query and add a term for each word to it
$query = new Lucene\Search\Query\MultiTerm();
foreach ($words as $w) {
$query->addTerm(new Lucene\Index\Term($w));
}
// open and query the index
$index = Lucene\Lucene::open('data/posts_index');
$results = $index->find($query);
// the search results
//var_dump($results);
return $this->render('index', ['searchModel' => $searchModel, 'dataProvider' => $dataProvider, 'search' => $results, 'query' => $query]);
}
public function testLimitingResult()
{
$index = Lucene\Lucene::open(__DIR__ . '/_index23Sample/_files');
$storedResultSetLimit = Lucene\Lucene::getResultSetLimit();
Lucene\Lucene::setResultSetLimit(3);
$hits = $index->find('"reporting bugs"', 'path');
$this->assertEquals(count($hits), 3);
$expectedResultset = array(array(7, 0.212395, 'IndexSource/contributing.bugs.html'), array(0, 0.247795, 'IndexSource/contributing.documentation.html'), array(2, 0.176996, 'IndexSource/contributing.patches.html'));
foreach ($hits as $resId => $hit) {
$this->assertEquals($hit->id, $expectedResultset[$resId][0]);
$this->assertTrue(abs($hit->score - $expectedResultset[$resId][1]) < 1.0E-6);
$this->assertEquals($hit->path, $expectedResultset[$resId][2]);
}
Lucene\Lucene::setResultSetLimit($storedResultSetLimit);
}
/**
* Search page for the term in the index.
* @param string $term
* @return array ('results' => array \ZendSearch\Lucene\Search\QueryHit, 'query' => string)
*/
public function find($term)
{
Wildcard::setMinPrefixLength($this->minPrefixLength);
Lucene::setResultSetLimit($this->resultsLimit);
return ['results' => $this->luceneIndex->find($term), 'query' => $term];
}
/**
* Search page for the term in the index.
* @param string $term
* @param array $fields (string => string)
* @return array ('results' => \ZendSearch\Lucene\Search\QueryHit[], 'query' => string)
*/
public function find($term, $fields = [])
{
Wildcard::setMinPrefixLength($this->minPrefixLength);
Lucene::setResultSetLimit($this->resultsLimit);
if (empty($fields)) {
return ['results' => $this->luceneIndex->find($term), 'query' => $term];
}
$fieldTerms[] = new IndexTerm($term);
foreach ($fields as $field => $fieldText) {
$fieldTerms[] = new IndexTerm($fieldText, $field);
}
return ['results' => $this->luceneIndex->find(new MultiTerm($fieldTerms)), 'query' => $term];
}
/**
* Search page for the term in the index.
*
* @param $term
* @param array $fields
* @return array|\ZendSearch\Lucene\Search\QueryHit
*/
public function search($term, $fields = [])
{
Wildcard::setMinPrefixLength($this->minPrefixLength);
Lucene::setResultSetLimit($this->resultsLimit);
$query = new MultiTerm();
if (count($fields)) {
foreach ($fields as $field => $value) {
$query->addTerm(new IndexTerm($value, $field), true);
}
}
$subTerm = explode(' ', $term);
foreach ($subTerm as $value) {
$query->addTerm(new IndexTerm($value), true);
}
return $this->luceneIndex->find($query);
}
