/**
* 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;
}
/**
* {@inheritdoc}
*/
public function register(Application $app)
{
Analyzer::setDefault(new CaseInsensitive());
QueryParser::setDefaultEncoding('UTF-8');
$app['zendsearch.indices_path'] = array();
$app['zendsearch.indices.initializer'] = $app->protect(function () use($app) {
static $initialized = false;
if ($initialized) {
return;
}
$initialized = true;
$indices = array();
foreach ($app['zendsearch.indices_path'] as $name => $index) {
$indices[$name] = file_exists($index) ? Lucene::open($index) : Lucene::create($index);
}
$app['zendsearch.indices_collection'] = $indices;
});
$app['zendsearch.indices'] = $app->share(function ($app) {
$app['zendsearch.indices.initializer']();
return $app['zendsearch.indices_collection'];
});
$app['zendsearch.multisearcher'] = $app->share(function ($app) {
$app['zendsearch.indices.initializer']();
$multi = new MultiSearcher();
foreach ($app['zendsearch.indices'] as $index) {
$multi->addIndex($index);
}
return $multi;
});
$app['zendsearch'] = $app->share(function ($app) {
return $app['zendsearch.multisearcher'];
});
}
public function __construct(array $filterClasses, array $stopWordFiles, FilterFactory $stopwordsFilterFactory)
{
QueryParser::setDefaultEncoding('utf-8');
$this->filterClasses = $filterClasses;
$this->stopWordFiles = $stopWordFiles;
$this->stopwordsFilterFactory = $stopwordsFilterFactory;
}
public function init()
{
QueryParser::setDefaultEncoding('UTF-8');
if ($this->caseSensitivity) {
Analyzer::setDefault($this->parseNumeric ? new Utf8Num() : new Utf8());
} else {
Analyzer::setDefault($this->parseNumeric ? new CaseInsensitiveNum() : new CaseInsensitive());
}
$this->indexDirectory = FileHelper::normalizePath(Yii::getAlias($this->indexDirectory));
$this->luceneIndex = $this->getLuceneIndex($this->indexDirectory);
}
public function __construct(FrontendController $frontendController, $moduleName)
{
parent::__construct($frontendController, $moduleName);
$this->controllerRoutes = array('/' => array('GET' => 'showSearchResults'), '/rebuild-index' => array('GET' => 'rebuildIndex'));
Analyzer::setDefault(new CaseInsensitive());
QueryParser::setDefaultEncoding('UTF-8');
$this->registerService('index', 'generateIndex');
$cmsPage = $frontendController->getCmsPage();
if ($cmsPage !== null) {
$cmsPage->setLastModified(date('Y-m-d H:i:s'));
}
}
/**
* 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' => $query]);
}
protected function getIndex()
{
if ($this->index != null) {
return $this->index;
}
\ZendSearch\Lucene\Search\QueryParser::setDefaultEncoding('utf-8');
\ZendSearch\Lucene\Analysis\Analyzer\Analyzer::setDefault(new \ZendSearch\Lucene\Analysis\Analyzer\Common\Utf8Num\CaseInsensitive());
\ZendSearch\Lucene\Search\QueryParser::setDefaultOperator(\ZendSearch\Lucene\Search\QueryParser::B_AND);
try {
$index = \ZendSearch\Lucene\Lucene::open($this->getIndexPath());
} catch (\ZendSearch\Lucene\Exception\RuntimeException $ex) {
$index = \ZendSearch\Lucene\Lucene::create($this->getIndexPath());
}
$this->index = $index;
return $index;
}
/**
* Gets the index mapped by the given lucene identifier.
*
* @param string $identifier The lucene identifier.
*
* @return \ZendSearch\Lucene\Index The lucene index.
*/
public function getIndex($identifier)
{
$config = $this->getConfig($identifier);
$path = $config['path'];
if (!$this->checkPath($path)) {
$this->indexes[$identifier] = Lucene::create($path);
} else {
$this->indexes[$identifier] = Lucene::open($path);
}
Analyzer::setDefault(new $config['analyzer']());
$this->indexes[$identifier]->setMaxBufferedDocs($config['max_buffered_docs']);
$this->indexes[$identifier]->setMaxMergeDocs($config['max_merge_docs']);
$this->indexes[$identifier]->setMergeFactor($config['merge_factor']);
ZfFilesystem::setDefaultFilePermissions($config['permissions']);
if ($config['auto_optimized']) {
$this->indexes[$identifier]->optimize();
}
QueryParser::setDefaultEncoding($config['query_parser_encoding']);
return $this->indexes[$identifier];
}
/**
* Create a new zend instance.
*
* @param string $name
* @param string $driver
*
* @return void
*/
public function __construct($name, $driver)
{
parent::__construct($name, $driver);
\ZendSearch\Lucene\Search\QueryParser::setDefaultEncoding('UTF-8');
}
private function getIndex() : SearchIndexInterface
{
$path = $this->getIndexPath();
if (!$this->checkIndexPath($path)) {
$index = Lucene::create($path);
} else {
$index = Lucene::open($path);
}
Analyzer::setDefault(new CaseInsensitive());
LuceneFilesystem::setDefaultFilePermissions(0775);
QueryParser::setDefaultEncoding('UTF-8');
$index->setMaxBufferedDocs($this->options['max_buffered_docs']);
$index->setMaxMergeDocs($this->options['max_merge_docs']);
$index->setMergeFactor($this->options['merge_factor']);
$index->optimize();
return $index;
}
/**
*
* Инициализация Zend Lucene
*
* @param string[]|string $modelClasses
* @param string $indexLocation
* @throws SearchCanNotConnectException
*/
public function __construct($modelClasses, $indexLocation)
{
if (!is_array($modelClasses)) {
$modelClasses = array($modelClasses);
}
$this->modelClasses = $this->filterModelClasses($modelClasses);
$this->defaultAnalyzer = $this->getDefaultAnalyzer();
$this->analyzerForHighlighter = $this->getAnalyzerForHighlighter();
Analyzer::setDefault($this->defaultAnalyzer);
QueryParser::setDefaultEncoding('utf-8');
// сделаем ограничение количества записей результата поиска
// Lucene::setResultSetLimit(10000);
// открываем/создаём новый индекс
if (file_exists($indexLocation = $indexLocation)) {
try {
$this->connection = Lucene::open($indexLocation);
} catch (\Exception $ex) {
$this->connection = Lucene::create($indexLocation);
}
} else {
$this->connection = Lucene::create($indexLocation);
}
if (!$this->connection) {
throw new SearchCanNotConnectException($indexLocation);
}
}
