/**
* 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;
}
public function newInstance($path)
{
if (!is_file($path)) {
throw new \InvalidArgumentException("File '{$path}' with stop words doesn't exist.");
}
$stopWordsFilter = new StopWordsFilter();
$stopWordsFilter->loadFromFile($path);
return $stopWordsFilter;
}
/**
* @return CaseInsensitive
*/
private function getDefaultAnalyzer()
{
// используем анализатор для кодировки UTF-8 нечувствительный к регистру
$analyzer = new CaseInsensitive();
// добавляем к анализатору фильтры стоп-слов и морфологический фильтр
$stopWordsFilter = new StopWords();
$stopWordsFilter->loadFromFile(__DIR__ . '/phpmorphy/stop-words/stop-words-russian.txt');
$analyzer->addFilter($stopWordsFilter);
$stopWordsFilter = new StopWords();
$stopWordsFilter->loadFromFile(__DIR__ . '/phpmorphy/stop-words/stop-words-english4.txt');
$analyzer->addFilter($stopWordsFilter);
$analyzer->addFilter(new MorphyFilter());
return $analyzer;
}
