public function testDefaultSearchField()
{
$currentDefaultSearchField = Zend_Search_Lucene::getDefaultSearchField();
$this->assertEquals($currentDefaultSearchField, null);
Zend_Search_Lucene::setDefaultSearchField('anotherField');
$this->assertEquals(Zend_Search_Lucene::getDefaultSearchField(), 'anotherField');
Zend_Search_Lucene::setDefaultSearchField($currentDefaultSearchField);
}
/**
* 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');
}
/**
* 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 $index)
{
// Allow to use wildcards within phrases
// They are either removed by text analyzer or used as a part of keyword for keyword fields
//
// if (strpos($this->_phrase, '?') !== false || strpos($this->_phrase, '*') !== false) {
// require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
// throw new Zend_Search_Lucene_Search_QueryParserException('Wildcards are only allowed in a single terms.');
// }
// Split query into subqueries if field name is not specified
if ($this->_field === null) {
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php';
$query = new Zend_Search_Lucene_Search_Query_Boolean();
$query->setBoost($this->getBoost());
require_once 'Zend/Search/Lucene.php';
if (Zend_Search_Lucene::getDefaultSearchField() === null) {
$searchFields = $index->getFieldNames(true);
} else {
$searchFields = array(Zend_Search_Lucene::getDefaultSearchField());
}
foreach ($searchFields as $fieldName) {
$subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Phrase($this->_phrase, $this->_phraseEncoding, $fieldName);
$subquery->setSlop($this->getSlop());
$query->addSubquery($subquery->rewrite($index));
}
$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->_phrase, $this->_field);
if ($index->hasTerm($term)) {
require_once 'Zend/Search/Lucene/Search/Query/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
// tokenize phrase using current analyzer and process it as a phrase query
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($this->_phrase, $this->_phraseEncoding);
if (count($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/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
//It's non-trivial phrase query
$position = -1;
require_once 'Zend/Search/Lucene/Search/Query/Phrase.php';
$query = new Zend_Search_Lucene_Search_Query_Phrase();
require_once 'Zend/Search/Lucene/Index/Term.php';
foreach ($tokens as $token) {
$position += $token->getPositionIncrement();
$term = new Zend_Search_Lucene_Index_Term($token->getTermText(), $this->_field);
$query->addTerm($term, $position);
$query->setSlop($this->getSlop());
}
$this->_matches = $query->getQueryTerms();
return $query;
}
/**
* Object constructor
*/
public function __construct($text, $field = null)
{
$this->field = $field === null ? Zend_Search_Lucene::getDefaultSearchField() : $field;
$this->text = $text;
}
/**
* 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/MultiTerm.php';
$query = new Zend_Search_Lucene_Search_Query_MultiTerm();
$query->setBoost($this->getBoost());
$hasInsignificantSubqueries = false;
require_once 'Zend/Search/Lucene.php';
if (Zend_Search_Lucene::getDefaultSearchField() === null) {
$searchFields = $index->getFieldNames(true);
} else {
$searchFields = array(Zend_Search_Lucene::getDefaultSearchField());
}
require_once 'Zend/Search/Lucene/Search/Query/Preprocessing/Term.php';
foreach ($searchFields as $fieldName) {
$subquery = new Zend_Search_Lucene_Search_Query_Preprocessing_Term($this->_word, $this->_encoding, $fieldName);
$rewrittenSubquery = $subquery->rewrite($index);
foreach ($rewrittenSubquery->getQueryTerms() as $term) {
$query->addTerm($term);
}
if ($rewrittenSubquery instanceof Zend_Search_Lucene_Search_Query_Insignificant) {
$hasInsignificantSubqueries = true;
}
}
if (count($query->getTerms()) == 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();
}
}
$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/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
// -------------------------------------
// 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) {
$word = iconv($this->_encoding, 'UTF-8', $this->_word);
$wildcardsPattern = '/[*?]/u';
$subPatternsEncoding = 'UTF-8';
} else {
$word = $this->_word;
$wildcardsPattern = '/[*?]/';
$subPatternsEncoding = $this->_encoding;
}
$subPatterns = preg_split($wildcardsPattern, $word, -1, PREG_SPLIT_OFFSET_CAPTURE);
if (count($subPatterns) > 1) {
// Wildcard query is recognized
$pattern = '';
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php';
foreach ($subPatterns as $id => $subPattern) {
// Append corresponding wildcard character to the pattern before each sub-pattern (except first)
if ($id != 0) {
$pattern .= $word[$subPattern[1] - 1];
}
// Check if each subputtern is a single word in terms of current analyzer
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($subPattern[0], $subPatternsEncoding);
if (count($tokens) > 1) {
require_once 'Zend/Search/Lucene/Search/QueryParserException.php';
throw new Zend_Search_Lucene_Search_QueryParserException('Wildcard search is supported only for non-multiple word terms');
}
foreach ($tokens as $token) {
$pattern .= $token->getTermText();
}
}
require_once 'Zend/Search/Lucene/Index/Term.php';
$term = new Zend_Search_Lucene_Index_Term($pattern, $this->_field);
require_once 'Zend/Search/Lucene/Search/Query/Wildcard.php';
$query = new Zend_Search_Lucene_Search_Query_Wildcard($term);
$query->setBoost($this->getBoost());
// Get rewritten query. Important! It also fills terms matching container.
$rewrittenQuery = $query->rewrite($index);
$this->_matches = $query->getQueryTerms();
return $rewrittenQuery;
}
// -------------------------------------
// Recognize 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/Term.php';
$query = new Zend_Search_Lucene_Search_Query_Term($term);
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
//It's not insignificant or one term query
require_once 'Zend/Search/Lucene/Search/Query/MultiTerm.php';
$query = new Zend_Search_Lucene_Search_Query_MultiTerm();
/**
* @todo Process $token->getPositionIncrement() to support stemming, synonyms and other
* analizer design features
*/
require_once 'Zend/Search/Lucene/Index/Term.php';
foreach ($tokens as $token) {
$term = new Zend_Search_Lucene_Index_Term($token->getTermText(), $this->_field);
$query->addTerm($term, true);
// all subterms are required
}
$query->setBoost($this->getBoost());
$this->_matches = $query->getQueryTerms();
return $query;
}
/**
* Get default search field.
*
* Null means, that search is performed through all fields by default
*
* @return string
*/
public static function getDefaultSearchField()
{
return Zend_Search_Lucene::getDefaultSearchField();
}
public function testDefaultSearchField()
{
$index = Zend_Search_Lucene::open(dirname(__FILE__) . '/_index23Sample/_files');
$storedDefaultSearchField = Zend_Search_Lucene::getDefaultSearchField();
Zend_Search_Lucene::setDefaultSearchField('path');
$hits = $index->find('contributing');
$this->assertEquals(count($hits), 5);
$expectedResultset = array(array(8, 0.847922, 'IndexSource/contributing.html'), array(0, 0.678337, 'IndexSource/contributing.documentation.html'), array(1, 0.678337, 'IndexSource/contributing.wishlist.html'), array(2, 0.678337, 'IndexSource/contributing.patches.html'), array(7, 0.678337, 'IndexSource/contributing.bugs.html'));
foreach ($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]);
}
Zend_Search_Lucene::setDefaultSearchField($storedDefaultSearchField);
}
