public function testGetPrefixUtf8()
{
// UTF-8 string with non-ascii symbols (Russian alphabet)
$this->assertEquals(Zend_Search_Lucene_Index_Term::getPrefix('абвгдеёжзийклмнопрстуфхцчшщьыъэюя', 64), 'абвгдеёжзийклмнопрстуфхцчшщьыъэюя');
$this->assertEquals(Zend_Search_Lucene_Index_Term::getPrefix('абвгдеёжзийклмнопрстуфхцчшщьыъэюя', 33), 'абвгдеёжзийклмнопрстуфхцчшщьыъэюя');
$this->assertEquals(Zend_Search_Lucene_Index_Term::getPrefix('абвгдеёжзийклмнопрстуфхцчшщьыъэюя', 4), 'абвг');
$this->assertEquals(Zend_Search_Lucene_Index_Term::getPrefix('абвгдеёжзийклмнопрстуфхцчшщьыъэюя', 0), '');
}
/**
* Scans terms dictionary and returns next term
*
* @return Zend_Search_Lucene_Index_Term|null
*/
public function nextTerm()
{
if ($this->_tisFile === null || $this->_termCount == 0) {
$this->_lastTerm = null;
$this->_lastTermInfo = null;
$this->_lastTermPositions = null;
$this->_docMap = null;
// may be necessary for "empty" segment
$this->_tisFile = null;
$this->_frqFile = null;
$this->_prxFile = null;
return null;
}
$termPrefixLength = $this->_tisFile->readVInt();
$termSuffix = $this->_tisFile->readString();
$termFieldNum = $this->_tisFile->readVInt();
$termValue = Zend_Search_Lucene_Index_Term::getPrefix($this->_lastTerm->text, $termPrefixLength) . $termSuffix;
$this->_lastTerm = new Zend_Search_Lucene_Index_Term($termValue, $this->_fields[$termFieldNum]->name);
$docFreq = $this->_tisFile->readVInt();
$freqPointer = $this->_lastTermInfo->freqPointer + $this->_tisFile->readVInt();
$proxPointer = $this->_lastTermInfo->proxPointer + $this->_tisFile->readVInt();
if ($docFreq >= $this->_skipInterval) {
$skipOffset = $this->_tisFile->readVInt();
} else {
$skipOffset = 0;
}
$this->_lastTermInfo = new Zend_Search_Lucene_Index_TermInfo($docFreq, $freqPointer, $proxPointer, $skipOffset);
if ($this->_termsScanMode == self::SM_FULL_INFO || $this->_termsScanMode == self::SM_MERGE_INFO) {
$this->_lastTermPositions = array();
$this->_frqFile->seek($this->_lastTermInfo->freqPointer + $this->_frqFileOffset, SEEK_SET);
$freqs = array(); $docId = 0;
for( $count = 0; $count < $this->_lastTermInfo->docFreq; $count++ ) {
$docDelta = $this->_frqFile->readVInt();
if( $docDelta % 2 == 1 ) {
$docId += ($docDelta-1)/2;
$freqs[ $docId ] = 1;
} else {
$docId += $docDelta/2;
$freqs[ $docId ] = $this->_frqFile->readVInt();
}
}
$this->_prxFile->seek($this->_lastTermInfo->proxPointer + $this->_prxFileOffset, SEEK_SET);
foreach ($freqs as $docId => $freq) {
$termPosition = 0; $positions = array();
for ($count = 0; $count < $freq; $count++ ) {
$termPosition += $this->_prxFile->readVInt();
$positions[] = $termPosition;
}
if (isset($this->_docMap[$docId])) {
$this->_lastTermPositions[$this->_docMap[$docId]] = $positions;
}
}
}
$this->_termCount--;
if ($this->_termCount == 0) {
$this->_tisFile = null;
$this->_frqFile = null;
$this->_prxFile = null;
}
return $this->_lastTerm;
}
/**
* Query specific matches highlighting
*
* @param Zend_Search_Lucene_Search_Highlighter_Interface $highlighter Highlighter object (also contains doc for highlighting)
*/
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter)
{
$words = array();
//$1 'Zend/Search/Lucene/Index/Term.php';
$prefix = Zend_Search_Lucene_Index_Term::getPrefix($this->_term->text, $this->_prefixLength);
$prefixByteLength = strlen($prefix);
$prefixUtf8Length = Zend_Search_Lucene_Index_Term::getLength($prefix);
$termLength = Zend_Search_Lucene_Index_Term::getLength($this->_term->text);
$termRest = substr($this->_term->text, $prefixByteLength);
// we calculate length of the rest in bytes since levenshtein() is not UTF-8 compatible
$termRestLength = strlen($termRest);
$scaleFactor = 1 / (1 - $this->_minimumSimilarity);
$docBody = $highlighter->getDocument()->getFieldUtf8Value('body');
//$1 'Zend/Search/Lucene/Analysis/Analyzer.php';
$tokens = Zend_Search_Lucene_Analysis_Analyzer::getDefault()->tokenize($docBody, 'UTF-8');
foreach ($tokens as $token) {
$termText = $token->getTermText();
if (substr($termText, 0, $prefixByteLength) == $prefix) {
// Calculate similarity
$target = substr($termText, $prefixByteLength);
$maxDistance = isset($this->_maxDistances[strlen($target)]) ? $this->_maxDistances[strlen($target)] : $this->_calculateMaxDistance($prefixUtf8Length, $termRestLength, strlen($target));
if ($termRestLength == 0) {
// we don't have anything to compare. That means if we just add
// the letters for current term we get the new word
$similarity = $prefixUtf8Length == 0 ? 0 : 1 - strlen($target) / $prefixUtf8Length;
} else {
if (strlen($target) == 0) {
$similarity = $prefixUtf8Length == 0 ? 0 : 1 - $termRestLength / $prefixUtf8Length;
} else {
if ($maxDistance < abs($termRestLength - strlen($target))) {
//just adding the characters of term to target or vice-versa results in too many edits
//for example "pre" length is 3 and "prefixes" length is 8. We can see that
//given this optimal circumstance, the edit distance cannot be less than 5.
//which is 8-3 or more precisesly abs(3-8).
//if our maximum edit distance is 4, then we can discard this word
//without looking at it.
$similarity = 0;
} else {
$similarity = 1 - levenshtein($termRest, $target) / ($prefixUtf8Length + min($termRestLength, strlen($target)));
}
}
}
if ($similarity > $this->_minimumSimilarity) {
$words[] = $termText;
}
}
}
$highlighter->highlight($words);
}
/**
* 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)
{
$this->_matches = array();
$this->_scores = array();
$this->_termKeys = array();
if ($this->_term->field === null) {
// Search through all fields
$fields = $index->getFieldNames(true);
} else {
$fields = array($this->_term->field);
}
$prefix = Zend_Search_Lucene_Index_Term::getPrefix($this->_term->text, $this->_prefixLength);
$prefixByteLength = strlen($prefix);
$prefixUtf8Length = Zend_Search_Lucene_Index_Term::getLength($prefix);
$termLength = Zend_Search_Lucene_Index_Term::getLength($this->_term->text);
$termRest = substr($this->_term->text, $prefixByteLength);
// we calculate length of the rest in bytes since levenshtein() is not UTF-8 compatible
$termRestLength = strlen($termRest);
$scaleFactor = 1 / (1 - $this->_minimumSimilarity);
foreach ($fields as $field) {
$index->resetTermsStream();
if ($prefix != '') {
$index->skipTo(new Zend_Search_Lucene_Index_Term($prefix, $field));
while ($index->currentTerm() !== null && $index->currentTerm()->field == $field && substr($index->currentTerm()->text, 0, $prefixByteLength) == $prefix) {
// Calculate similarity
$target = substr($index->currentTerm()->text, $prefixByteLength);
$maxDistance = isset($this->_maxDistances[strlen($target)]) ? $this->_maxDistances[strlen($target)] : $this->_calculateMaxDistance($prefixUtf8Length, $termRestLength, strlen($target));
if ($termRestLength == 0) {
// we don't have anything to compare. That means if we just add
// the letters for current term we get the new word
$similarity = $prefixUtf8Length == 0 ? 0 : 1 - strlen($target) / $prefixUtf8Length;
} else {
if (strlen($target) == 0) {
$similarity = $prefixUtf8Length == 0 ? 0 : 1 - $termRestLength / $prefixUtf8Length;
} else {
if ($maxDistance < abs($termRestLength - strlen($target))) {
//just adding the characters of term to target or vice-versa results in too many edits
//for example "pre" length is 3 and "prefixes" length is 8. We can see that
//given this optimal circumstance, the edit distance cannot be less than 5.
//which is 8-3 or more precisesly abs(3-8).
//if our maximum edit distance is 4, then we can discard this word
//without looking at it.
$similarity = 0;
} else {
$similarity = 1 - levenshtein($termRest, $target) / ($prefixUtf8Length + min($termRestLength, strlen($target)));
}
}
}
if ($similarity > $this->_minimumSimilarity) {
$this->_matches[] = $index->currentTerm();
$this->_termKeys[] = $index->currentTerm()->key();
$this->_scores[] = ($similarity - $this->_minimumSimilarity) * $scaleFactor;
}
$index->nextTerm();
}
} else {
$index->skipTo(new Zend_Search_Lucene_Index_Term('', $field));
while ($index->currentTerm() !== null && $index->currentTerm()->field == $field) {
// Calculate similarity
$target = $index->currentTerm()->text;
$maxDistance = isset($this->_maxDistances[strlen($target)]) ? $this->_maxDistances[strlen($target)] : $this->_calculateMaxDistance(0, $termRestLength, strlen($target));
if ($maxDistance < abs($termRestLength - strlen($target))) {
//just adding the characters of term to target or vice-versa results in too many edits
//for example "pre" length is 3 and "prefixes" length is 8. We can see that
//given this optimal circumstance, the edit distance cannot be less than 5.
//which is 8-3 or more precisesly abs(3-8).
//if our maximum edit distance is 4, then we can discard this word
//without looking at it.
$similarity = 0;
} else {
$similarity = 1 - levenshtein($termRest, $target) / min($termRestLength, strlen($target));
}
if ($similarity > $this->_minimumSimilarity) {
$this->_matches[] = $index->currentTerm();
$this->_termKeys[] = $index->currentTerm()->key();
$this->_scores[] = ($similarity - $this->_minimumSimilarity) * $scaleFactor;
}
$index->nextTerm();
}
}
$index->closeTermsStream();
}
if (count($this->_matches) == 0) {
return new Zend_Search_Lucene_Search_Query_Empty();
} else {
if (count($this->_matches) == 1) {
return new Zend_Search_Lucene_Search_Query_Term(reset($this->_matches));
} else {
$rewrittenQuery = new Zend_Search_Lucene_Search_Query_Boolean();
array_multisort($this->_scores, SORT_DESC, SORT_NUMERIC, $this->_termKeys, SORT_ASC, SORT_STRING, $this->_matches);
$termCount = 0;
foreach ($this->_matches as $id => $matchedTerm) {
$subquery = new Zend_Search_Lucene_Search_Query_Term($matchedTerm);
$subquery->setBoost($this->_scores[$id]);
$rewrittenQuery->addSubquery($subquery);
$termCount++;
if ($termCount >= self::MAX_CLAUSE_COUNT) {
break;
}
}
return $rewrittenQuery;
}
}
}
