/** * Calculate result vector for non Conjunction query * (like '+something -another') * * @param Zend_Search_Lucene $reader */ private function _calculateNonConjunctionResult($reader) { if (extension_loaded('bitset')) { $required = null; $neither = bitset_empty(); $prohibited = bitset_empty(); foreach ($this->_terms as $termId => $term) { $termDocs = bitset_from_array($reader->termDocs($term)); if ($this->_signs[$termId] === true) { // required if ($required !== null) { $required = bitset_intersection($required, $termDocs); } else { $required = $termDocs; } } elseif ($this->_signs[$termId] === false) { // prohibited $prohibited = bitset_union($prohibited, $termDocs); } else { // neither required, nor prohibited $neither = bitset_union($neither, $termDocs); } $this->_termsPositions[$termId] = $reader->termPositions($term); } if ($required === null) { $required = $neither; } $this->_resVector = bitset_intersection($required, bitset_invert($prohibited, $reader->count())); } else { $required = null; $neither = array(); $prohibited = array(); foreach ($this->_terms as $termId => $term) { $termDocs = array_flip($reader->termDocs($term)); if ($this->_signs[$termId] === true) { // required if ($required !== null) { // substitute for bitset_intersection foreach ($required as $key => $value) { if (!isset($termDocs[$key])) { unset($required[$key]); } } } else { $required = $termDocs; } } elseif ($this->_signs[$termId] === false) { // prohibited // substitute for bitset_union foreach ($termDocs as $key => $value) { $prohibited[$key] = $value; } } else { // neither required, nor prohibited // substitute for bitset_union foreach ($termDocs as $key => $value) { $neither[$key] = $value; } } $this->_termsPositions[$termId] = $reader->termPositions($term); } if ($required === null) { $required = $neither; } foreach ($required as $key => $value) { if (isset($prohibited[$key])) { unset($required[$key]); } } $this->_resVector = $required; } }
/** * Score specified document * * @param integer $docId * @param Zend_Search_Lucene $reader * @return float */ public function score($docId, $reader) { if ($this->_docVector === null) { if (extension_loaded('bitset')) { $this->_docVector = bitset_from_array($reader->termDocs($this->_term)); } else { $this->_docVector = array_flip($reader->termDocs($this->_term)); } $this->_termPositions = $reader->termPositions($this->_term); $this->_initWeight($reader); } $match = extension_loaded('bitset') ? bitset_in($this->_docVector, $docId) : isset($this->_docVector[$docId]); if ($this->_sign && $match) { return $reader->getSimilarity()->tf(count($this->_termPositions[$docId])) * $this->_weight->getValue() * $reader->norm($docId, $this->_term->field); } else { return 0; } }
/** * Calculate result vector * * @param Zend_Search_Lucene $reader */ private function _calculateResult($reader) { if (extension_loaded('bitset')) { foreach ($this->_terms as $termId => $term) { if ($this->_resVector === null) { $this->_resVector = bitset_from_array($reader->termDocs($term)); } else { $this->_resVector = bitset_intersection($this->_resVector, bitset_from_array($reader->termDocs($term))); } $this->_termsPositions[$termId] = $reader->termPositions($term); } } else { foreach ($this->_terms as $termId => $term) { if ($this->_resVector === null) { $this->_resVector = array_flip($reader->termDocs($term)); } else { $termDocs = array_flip($reader->termDocs($term)); foreach ($this->_resVector as $key => $value) { if (!isset($termDocs[$key])) { unset($this->_resVector[$key]); } } } $this->_termsPositions[$termId] = $reader->termPositions($term); } } }
/** * Execute query in context of index reader * It also initializes necessary internal structures * * @param Zend_Search_Lucene $reader */ public function execute($reader) { $this->_docVector = array_flip($reader->termDocs($this->_term)); $this->_termPositions = $reader->termPositions($this->_term); // Initialize weight if it's not done yet $this->_initWeight($reader); }
/** * Calculate result vector for non Conjunction query * (like '+something -another') * * @param Zend_Search_Lucene $reader */ private function _calculateNonConjunctionResult($reader) { $required = null; $optional = array(); $prohibited = array(); foreach ($this->_terms as $termId => $term) { $termDocs = array_flip($reader->termDocs($term)); if ($this->_signs[$termId] === true) { // required if ($required !== null) { // array intersection $required = array_intersect_key($required, $termDocs); } else { $required = $termDocs; } } elseif ($this->_signs[$termId] === false) { // prohibited // array union $prohibited += $termDocs; } else { // neither required, nor prohibited // array union $optional += $termDocs; } $this->_termsPositions[$termId] = $reader->termPositions($term); } if ($required !== null) { $this->_resVector = array_diff_key($required, $prohibited); } else { $this->_resVector = array_diff_key($optional, $prohibited); } ksort($this->_resVector, SORT_NUMERIC); }
/** * Execute query in context of index reader * It also initializes necessary internal structures * * @param Zend_Search_Lucene $reader */ public function execute($reader) { $this->_resVector = null; if (count($this->_terms) == 0) { $this->_resVector = array(); } foreach ($this->_terms as $termId => $term) { if ($this->_resVector === null) { $this->_resVector = array_flip($reader->termDocs($term)); } else { $this->_resVector = array_intersect_key($this->_resVector, array_flip($reader->termDocs($term))); } if (count($this->_resVector) == 0) { // Empty result set, we don't need to check other terms break; } $this->_termsPositions[$termId] = $reader->termPositions($term); } ksort($this->_resVector, SORT_NUMERIC); // Initialize weight if it's not done yet $this->_initWeight($reader); }
/** * This will refresh the search index for this topic (for all message content under this topic) * @param Zend_Search_Lucene $objIndex should be null if we are updating just one -- but for bulk index updates, you can pass in an already loaded index file * @return void */ public function RefreshSearchIndex($objIndex = null) { // Currently only implemented for Forum-based topic/message searches if ($this->TopicLink->TopicLinkTypeId != TopicLinkType::Forum) { return; } if (!$objIndex) { $objIndex = new Zend_Search_Lucene(__SEARCH_INDEXES__ . '/topics'); $blnIndexProvided = false; } else { $blnIndexProvided = true; } // Retrievew the Index Documents (if applicable) to delete them from the index $objSearchTerm = new Zend_Search_Lucene_Index_Term($this->Id, 'db_id'); foreach ($objIndex->termDocs($objSearchTerm) as $intDocId) { $objIndex->delete($intDocId); } // Create the Message Contents for this Topic $strContents = null; foreach ($this->GetMessageArray(QQ::OrderBy(QQN::Message()->ReplyNumber)) as $objMessage) { $strMessage = strip_tags(trim($objMessage->CompiledHtml)); $strMessage = html_entity_decode($strMessage, ENT_QUOTES, 'UTF-8'); $strContents .= $strMessage . "\r\n\r\n"; } // Create the Document $objDocument = new Zend_Search_Lucene_Document(); $objDocument->addField(Zend_Search_Lucene_Field::Keyword('db_id', $this->Id)); $objDocument->addField(Zend_Search_Lucene_Field::UnIndexed('topic_link_id', $this->TopicLinkId)); $objDocument->addField(Zend_Search_Lucene_Field::UnIndexed('topic_link_type_id', $this->TopicLink->TopicLinkTypeId)); $objDocument->addField(Zend_Search_Lucene_Field::UnIndexed('message_count', $this->MessageCount)); $objDocument->addField(Zend_Search_Lucene_Field::UnIndexed('last_post_date', $this->LastPostDate->Timestamp)); $objDocument->addField(Zend_Search_Lucene_Field::Text('title', $this->Name)); $objDocument->addField(Zend_Search_Lucene_Field::UnStored('contents', trim($strContents))); // Add Document to Index $objIndex->addDocument($objDocument); // Only call commit on the index if it was provided for us if (!$blnIndexProvided) { $objIndex->commit(); } }