/** * Transform entry to a subquery * * @param string $encoding * @return Zend_Search_Lucene_Search_Query * @throws Zend_Search_Lucene_Search_QueryParserException */ public function getQuery($encoding) { if ($this->_fuzzyQuery) { /** Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy */ require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Fuzzy.php'; $query = new Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy($this->_term, $encoding, $this->_field !== null ? iconv($encoding, 'UTF-8', $this->_field) : null, $this->_similarity); $query->setBoost($this->_boost); return $query; } /** Zend_Search_Lucene_Search_Query_Preprocessing_Term */ require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Term.php'; $query = new Zend_Search_Lucene_Search_Query_Preprocessing_Term($this->_term, $encoding, $this->_field !== null ? iconv($encoding, 'UTF-8', $this->_field) : null); $query->setBoost($this->_boost); return $query; }
/** * Re-write query into primitive queries in the context of specified index * * @param Zend_Search_Lucene_Interface $index * @return Zend_Search_Lucene_Search_Query */ public function rewrite(Zend_Search_Lucene_Interface $index) { if ($this->_field === null) { require_once 'Zend/Search/Lucene/Search/Query/Boolean.php'; $query = new Zend_Search_Lucene_Search_Query_Boolean(); $hasInsignificantSubqueries = false; require_once 'Zend/Search/Lucene.php'; if (Zend_Search_Lucene::getDefaultSearchField() === null) { $searchFields = $index->getFieldNames(true); } else { $searchFields = array(Zend_Search_Lucene::getDefaultSearchField()); } require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Fuzzy.php'; foreach ($searchFields as $fieldName) { $subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy($this->_word, $this->_encoding, $fieldName, $this->_minimumSimilarity); $rewrittenSubquery = $subquery->rewrite($index); if (!($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant || $rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Empty)) { $query->addSubquery($rewrittenSubquery); } if ($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant) { $hasInsignificantSubqueries = true; } } $subqueries = $query->getSubqueries(); if (count($subqueries) == 0) { $this->_matches = array(); if ($hasInsignificantSubqueries) { require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php'; return new Zend_Search_Lucene_Search_Query_Insignificant(); } else { require_once 'Zend/Search/Lucene/Search/Query/Empty.php'; return new Zend_Search_Lucene_Search_Query_Empty(); } } if (count($subqueries) == 1) { $query = reset($subqueries); } $query->setBoost($this->getBoost()); $this->_matches = $query->getQueryTerms(); return $query; } // ------------------------------------- // Recognize exact term matching (it corresponds to Keyword fields stored in the index) // encoding is not used since we expect binary matching require_once 'Zend/Search/Lucene/Index/Term.php'; $term = new Zend_Search_Lucene_Index_Term($this->_word, $this->_field); if ($index->hasTerm($term)) { require_once 'Zend/Search/Lucene/Search/Query/Fuzzy.php'; $query = new Zend_Search_Lucene_Search_Query_Fuzzy($term, $this->_minimumSimilarity); $query->setBoost($this->getBoost()); // Get rewritten query. Important! It also fills terms matching container. $rewrittenQuery = $query->rewrite($index); $this->_matches = $query->getQueryTerms(); return $rewrittenQuery; } // ------------------------------------- // Recognize wildcard queries /** @todo check for PCRE unicode support may be performed through Zend_Environment in some future */ if (@preg_match('/\\pL/u', 'a') == 1) { $subPatterns = preg_split('/[*?]/u', iconv($this->_encoding, 'UTF-8', $this->_word)); } else { $subPatterns = preg_split('/[*?]/', $this->_word); } if (count($subPatterns) > 1) { require_once 'Zend/Search/Lucene/Search/QueryParserException.php'; throw new Zend_Search_Lucene_Search_QueryParserException('Fuzzy search doesn\'t support wildcards (except within Keyword fields).'); } // ------------------------------------- // Recognize one-term multi-term and "insignificant" queries require_once 'Zend/Search/Lucene/Analysis/Analyzer.php'; $tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_word, $this->_encoding); if (count($tokens) == 0) { $this->_matches = array(); require_once 'Zend/Search/Lucene/Search/Query/Insignificant.php'; return new Zend_Search_Lucene_Search_Query_Insignificant(); } if (count($tokens) == 1) { require_once 'Zend/Search/Lucene/Index/Term.php'; $term = new Zend_Search_Lucene_Index_Term($tokens[0]->getTermText(), $this->_field); require_once 'Zend/Search/Lucene/Search/Query/Fuzzy.php'; $query = new Zend_Search_Lucene_Search_Query_Fuzzy($term, $this->_minimumSimilarity); $query->setBoost($this->getBoost()); // Get rewritten query. Important! It also fills terms matching container. $rewrittenQuery = $query->rewrite($index); $this->_matches = $query->getQueryTerms(); return $rewrittenQuery; } // Word is tokenized into several tokens require_once 'Zend/Search/Lucene/Search/QueryParserException.php'; throw new Zend_Search_Lucene_Search_QueryParserException('Fuzzy search is supported only for non-multiple word terms'); }
/** * Transform entry to a subquery * * @param string $encoding * @return Zend_Search_Lucene_Search_Query * @throws Zend_Search_Lucene_Search_QueryParserException */ public function getQuery($encoding) { if ($this->_fuzzyQuery) { $query = new Zend_Search_Lucene_Search_Query_Preprocessing_Fuzzy($this->_term, $encoding, $this->_field !== null ? iconv($encoding, 'UTF-8', $this->_field) : null, $this->_similarity); $query->setBoost($this->_boost); return $query; } $query = new Zend_Search_Lucene_Search_Query_Preprocessing_Term($this->_term, $encoding, $this->_field !== null ? iconv($encoding, 'UTF-8', $this->_field) : null); $query->setBoost($this->_boost); return $query; }