phrase.php

来自「PHP 知识管理系统（基于树结构的知识管理系统）, 英文原版的PHP源码。」· PHP 代码 · 共 543 行 · 第 1/2 页

     *
     * @param Zend_Search_Lucene $reader
     * @return Zend_Search_Lucene_Search_Weight
     */
    public function createWeight($reader)
    {
        $this->_weight = new Zend_Search_Lucene_Search_Weight_Phrase($this, $reader);
        return $this->_weight;
    }


    /**
     * Score calculator for exact phrase queries (terms sequence is fixed)
     *
     * @param integer $docId
     * @return float
     */
    public function _exactPhraseFreq($docId)
    {
        $freq = 0;

        // Term Id with lowest cardinality
        $lowCardTermId = null;

        // Calculate $lowCardTermId
        foreach ($this->_terms as $termId => $term) {
            if ($lowCardTermId === null ||
                count($this->_termsPositions[$termId][$docId]) <
                count($this->_termsPositions[$lowCardTermId][$docId]) ) {
                    $lowCardTermId = $termId;
                }
        }

        // Walk through positions of the term with lowest cardinality
        foreach ($this->_termsPositions[$lowCardTermId][$docId] as $lowCardPos) {
            // We expect phrase to be found
            $freq++;

            // Walk through other terms
            foreach ($this->_terms as $termId => $term) {
                if ($termId != $lowCardTermId) {
                    $expectedPosition = $lowCardPos +
                                            ($this->_offsets[$termId] -
                                             $this->_offsets[$lowCardTermId]);

                    if (!in_array($expectedPosition, $this->_termsPositions[$termId][$docId])) {
                        $freq--;  // Phrase wasn't found.
                        break;
                    }
                }
            }
        }

        return $freq;
    }

    /**
     * Score calculator for sloppy phrase queries (terms sequence is fixed)
     *
     * @param integer $docId
     * @param Zend_Search_Lucene $reader
     * @return float
     */
    public function _sloppyPhraseFreq($docId, Zend_Search_Lucene $reader)
    {
        $freq = 0;

        $phraseQueue = array();
        $phraseQueue[0] = array(); // empty phrase
        $lastTerm = null;

        // Walk through the terms to create phrases.
        foreach ($this->_terms as $termId => $term) {
            $queueSize = count($phraseQueue);
            $firstPass = true;

            // Walk through the term positions.
            // Each term position produces a set of phrases.
            foreach ($this->_termsPositions[$termId][$docId] as $termPosition ) {
                if ($firstPass) {
                    for ($count = 0; $count < $queueSize; $count++) {
                        $phraseQueue[$count][$termId] = $termPosition;
                    }
                } else {
                    for ($count = 0; $count < $queueSize; $count++) {
                        if ($lastTerm !== null &&
                            abs( $termPosition - $phraseQueue[$count][$lastTerm] -
                                 ($this->_offsets[$termId] - $this->_offsets[$lastTerm])) > $this->_slop) {
                            continue;
                        }

                        $newPhraseId = count($phraseQueue);
                        $phraseQueue[$newPhraseId]          = $phraseQueue[$count];
                        $phraseQueue[$newPhraseId][$termId] = $termPosition;
                    }

                }

                $firstPass = false;
            }
            $lastTerm = $termId;
        }


        foreach ($phraseQueue as $phrasePos) {
            $minDistance = null;

            for ($shift = -$this->_slop; $shift <= $this->_slop; $shift++) {
                $distance = 0;
                $start = reset($phrasePos) - reset($this->_offsets) + $shift;

                foreach ($this->_terms as $termId => $term) {
                    $distance += abs($phrasePos[$termId] - $this->_offsets[$termId] - $start);

                    if($distance > $this->_slop) {
                        break;
                    }
                }

                if ($minDistance === null || $distance < $minDistance) {
                    $minDistance = $distance;
                }
            }

            if ($minDistance <= $this->_slop) {
                $freq += $reader->getSimilarity()->sloppyFreq($minDistance);
            }
        }

        return $freq;
    }

    /**
     * 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);
    }

    /**
     * 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 $reader
     * @return float
     */
    public function score($docId, $reader)
    {
        if (isset($this->_resVector[$docId])) {
            if ($this->_slop == 0) {
                $freq = $this->_exactPhraseFreq($docId);
            } else {
                $freq = $this->_sloppyPhraseFreq($docId, $reader);
            }

            if ($freq != 0) {
                $tf = $reader->getSimilarity()->tf($freq);
                $weight = $this->_weight->getValue();
                $norm = $reader->norm($docId, reset($this->_terms)->field);

                return $tf * $weight * $norm * $this->getBoost();
            }

            // Included in result, but culculated freq is zero
            return 0;
        } else {
            return 0;
        }
    }

    /**
     * Return query terms
     *
     * @return array
     */
    public function getQueryTerms()
    {
        return $this->_terms;
    }

    /**
     * Highlight query terms
     *
     * @param integer &$colorIndex
     * @param Zend_Search_Lucene_Document_Html $doc
     */
    public function highlightMatchesDOM(Zend_Search_Lucene_Document_Html $doc, &$colorIndex)
    {
        $words = array();
        foreach ($this->_terms as $term) {
            $words[] = $term->text;
        }

        $doc->highlight($words, $this->_getHighlightColor($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 = '';

        if (isset($this->_terms[0]) && $this->_terms[0]->field !== null) {
            $query .= $this->_terms[0]->field . ':';
        }

        $query .= '"';

        foreach ($this->_terms as $id => $term) {
            if ($id != 0) {
                $query .= ' ';
            }
            $query .= $term->text;
        }

        $query .= '"';

        if ($this->_slop != 0) {
            $query .= '~' . $this->_slop;
        }

        return $query;
    }
}