boolean.php

Bug tracker, and reporter.

        if (count($prohibitedTerms) == 1) {
            // (boost factors are not significant for prohibited clauses)
            $subqueries[] = new Zend_Search_Lucene_Search_Query_Term(reset($prohibitedTerms));
            $signs[]      = false;

            // Clear prohibited terms list
            $prohibitedTerms = array();
        } else if (count($prohibitedTerms) > 1) {
            // prepare signs array
            $prohibitedSigns = array();
            foreach ($prohibitedTerms as $id => $term) {
                // all prohibited term are grouped as optional into multi-term query
                $prohibitedSigns[$id] = null;
            }

            // (boost factors are not significant for prohibited clauses)
            $subqueries[] = new Zend_Search_Lucene_Search_Query_MultiTerm($prohibitedTerms, $prohibitedSigns);
            // Clause sign is 'prohibited'
            $signs[]      = false;

            // Clear terms list
            $prohibitedTerms = array();
        }

        /** @todo Group terms with the same boost factors together */

        // Check, that all terms are processed
        // Replace candidate for optimized query
        if (count($terms) == 0  &&  count($prohibitedTerms) == 0) {
            $optimizedQuery = new Zend_Search_Lucene_Search_Query_Boolean($subqueries, $signs);
            $optimizedQuery->setBoost($this->getBoost());
        }

        return $optimizedQuery;
    }

    /**
     * Returns subqueries
     *
     * @return array
     */
    public function getSubqueries()
    {
        return $this->_subqueries;
    }


    /**
     * Return subqueries signs
     *
     * @return array
     */
    public function getSigns()
    {
        return $this->_signs;
    }


    /**
     * Constructs an appropriate Weight implementation for this query.
     *
     * @param Zend_Search_Lucene_Interface $reader
     * @return Zend_Search_Lucene_Search_Weight
     */
    public function createWeight(Zend_Search_Lucene_Interface $reader)
    {
        $this->_weight = new Zend_Search_Lucene_Search_Weight_Boolean($this, $reader);
        return $this->_weight;
    }


    /**
     * Calculate result vector for Conjunction query
     * (like '<subquery1> AND <subquery2> AND <subquery3>')
     */
    private function _calculateConjunctionResult()
    {
        $this->_resVector = null;

        if (count($this->_subqueries) == 0) {
            $this->_resVector = array();
        }

        $resVectors      = array();
        $resVectorsSizes = array();
        $resVectorsIds   = array(); // is used to prevent arrays comparison
        foreach ($this->_subqueries as $subqueryId => $subquery) {
            $resVectors[]      = $subquery->matchedDocs();
            $resVectorsSizes[] = count(end($resVectors));
            $resVectorsIds[]   = $subqueryId;
        }
        // sort resvectors in order of subquery cardinality increasing
        array_multisort($resVectorsSizes, SORT_ASC, SORT_NUMERIC,
                        $resVectorsIds,   SORT_ASC, SORT_NUMERIC,
                        $resVectors);
        
        foreach ($resVectors as $nextResVector) {
            if($this->_resVector === null) {
                $this->_resVector = $nextResVector;
            } else {
                //$this->_resVector = array_intersect_key($this->_resVector, $nextResVector);
                
                /**
                 * This code is used as workaround for array_intersect_key() slowness problem.
                 */
                $updatedVector = array();
                foreach ($this->_resVector as $id => $value) {
                    if (isset($nextResVector[$id])) {
                        $updatedVector[$id] = $value;
                    }
                }
                $this->_resVector = $updatedVector;
            }

            if (count($this->_resVector) == 0) {
                // Empty result set, we don't need to check other terms
                break;
            }
        }

        // ksort($this->_resVector, SORT_NUMERIC);
        // Used algorithm doesn't change elements order
    }


    /**
     * Calculate result vector for non Conjunction query
     * (like '<subquery1> AND <subquery2> AND NOT <subquery3> OR <subquery4>')
     */
    private function _calculateNonConjunctionResult()
    {
        $requiredVectors      = array();
        $requiredVectorsSizes = array();
        $requiredVectorsIds   = array(); // is used to prevent arrays comparison

        $optional = array();

        foreach ($this->_subqueries as $subqueryId => $subquery) {
            if ($this->_signs[$subqueryId] === true) {
                // required
                $requiredVectors[]      = $subquery->matchedDocs();
                $requiredVectorsSizes[] = count(end($requiredVectors));
                $requiredVectorsIds[]   = $subqueryId;
            } elseif ($this->_signs[$subqueryId] === false) {
                // prohibited
                // Do nothing. matchedDocs() may include non-matching id's
                // Calculating prohibited vector may take significant time, but do not affect the result
                // Skipped.  
            } else {
                // neither required, nor prohibited
                // array union
                $optional += $subquery->matchedDocs();
            }
        }

        // sort resvectors in order of subquery cardinality increasing
        array_multisort($requiredVectorsSizes, SORT_ASC, SORT_NUMERIC,
                        $requiredVectorsIds,   SORT_ASC, SORT_NUMERIC,
                        $requiredVectors);
        
        $required = null;
        foreach ($requiredVectors as $nextResVector) {
            if($required === null) {
                $required = $nextResVector;
            } else {
                //$required = array_intersect_key($required, $nextResVector);
                
                /**
                 * This code is used as workaround for array_intersect_key() slowness problem.
                 */
                $updatedVector = array();
                foreach ($required as $id => $value) {
                    if (isset($nextResVector[$id])) {
                        $updatedVector[$id] = $value;
                    }
                }
                $required = $updatedVector;
            }

            if (count($required) == 0) {
                // Empty result set, we don't need to check other terms
                break;
            }
        }
                
        
        if ($required !== null) {
            $this->_resVector = &$required;
        } else {
            $this->_resVector = &$optional;
        }

        ksort($this->_resVector, SORT_NUMERIC);
    }


    /**
     * Score calculator for conjunction queries (all subqueries are required)
     *
     * @param integer $docId
     * @param Zend_Search_Lucene_Interface $reader
     * @return float
     */
    public function _conjunctionScore($docId, Zend_Search_Lucene_Interface $reader)
    {
        if ($this->_coord === null) {
            $this->_coord = $reader->getSimilarity()->coord(count($this->_subqueries),
                                                            count($this->_subqueries) );
        }

        $score = 0;

        foreach ($this->_subqueries as $subquery) {
            $subscore = $subquery->score($docId, $reader);

            if ($subscore == 0) {
                return 0;
            }

            $score += $subquery->score($docId, $reader) * $this->_coord;
        }

        return $score * $this->_coord * $this->getBoost();
    }


    /**
     * Score calculator for non conjunction queries (not all subqueries are required)
     *
     * @param integer $docId
     * @param Zend_Search_Lucene_Interface $reader
     * @return float
     */
    public function _nonConjunctionScore($docId, Zend_Search_Lucene_Interface $reader)
    {
        if ($this->_coord === null) {
            $this->_coord = array();

            $maxCoord = 0;
            foreach ($this->_signs as $sign) {
                if ($sign !== false /* not prohibited */) {
                    $maxCoord++;
                }
            }

            for ($count = 0; $count <= $maxCoord; $count++) {
                $this->_coord[$count] = $reader->getSimilarity()->coord($count, $maxCoord);
            }
        }

        $score = 0;
        $matchedSubqueries = 0;
        foreach ($this->_subqueries as $subqueryId => $subquery) {
            $subscore = $subquery->score($docId, $reader);

            // Prohibited
            if ($this->_signs[$subqueryId] === false && $subscore != 0) {
                return 0;
            }

            // is required, but doen't match
            if ($this->_signs[$subqueryId] === true &&  $subscore == 0) {
                return 0;
            }

            if ($subscore != 0) {
                $matchedSubqueries++;
                $score += $subscore;
            }
        }

        return $score * $this->_coord[$matchedSubqueries] * $this->getBoost();
    }

    /**
     * Execute query in context of index reader
     * It also initializes necessary internal structures
     *
     * @param Zend_Search_Lucene_Interface $reader
     */
    public function execute(Zend_Search_Lucene_Interface $reader)
    {
        // Initialize weight if it's not done yet
        $this->_initWeight($reader);

        foreach ($this->_subqueries as $subquery) {
            $subquery->execute($reader);
        }

        if ($this->_signs === null) {
            $this->_calculateConjunctionResult();
        } else {
            $this->_calculateNonConjunctionResult();
        }
    }



    /**
     * Get document ids likely matching the query
     *
     * It's an array with document ids as keys (performance considerations)
     *
     * @return array
     */
    public function matchedDocs()
    {
        return $this->_resVector;
    }

    /**
     * Score specified document
     *
     * @param integer $docId
     * @param Zend_Search_Lucene_Interface $reader
     * @return float
     */
    public function score($docId, Zend_Search_Lucene_Interface $reader)
    {
        if (isset($this->_resVector[$docId])) {
            if ($this->_signs === null) {
                return $this->_conjunctionScore($docId, $reader);
            } else {
                return $this->_nonConjunctionScore($docId, $reader);
            }
        } else {
            return 0;
        }
    }

    /**
     * Return query terms
     *
     * @return array
     */
    public function getQueryTerms()
    {
        $terms = array();

        foreach ($this->_subqueries as $id => $subquery) {
            if ($this->_signs === null  ||  $this->_signs[$id] !== false) {
                $terms = array_merge($terms, $subquery->getQueryTerms());
            }
        }

        return $terms;
    }

    /**
     * Highlight query terms
     *
     * @param integer &$colorIndex
     * @param Zend_Search_Lucene_Document_Html $doc
     */
    public function highlightMatchesDOM(Zend_Search_Lucene_Document_Html $doc, &$colorIndex)
    {
        foreach ($this->_subqueries as $id => $subquery) {
            if ($this->_signs === null  ||  $this->_signs[$id] !== false) {
                $subquery->highlightMatchesDOM($doc, $colorIndex);
            }
        }
    }

    /**
     * Print a query
     *
     * @return string
     */
    public function __toString()
    {
        // It's used only for query visualisation, so we don't care about characters escaping

        $query = '';

        foreach ($this->_subqueries as $id => $subquery) {
            if ($id != 0) {
                $query .= ' ';
            }

            if ($this->_signs === null || $this->_signs[$id] === true) {
                $query .= '+';
            } else if ($this->_signs[$id] === false) {
                $query .= '-';
            }

            $query .= '(' . $subquery->__toString() . ')';

            if ($subquery->getBoost() != 1) {
                $query .= '^' . round($subquery->getBoost(), 4);
            }
        }

        return $query;
    }
}