/**
* Reset terms stream
*
* $startId - id for the fist document
* $compact - remove deleted documents
*
* Returns start document id for the next segment
*
* @param integer $startId
* @param integer $mode
* @throws \Zend\Search\Lucene\Exception\InvalidArgumentException
* @throws \Zend\Search\Lucene\Exception\InvalidFileFormatException
* @return integer
*/
public function resetTermsStream()
{
/**
* SegmentInfo->resetTermsStream() method actually takes two optional parameters:
* $startId (default value is 0)
* $mode (default value is self::SM_TERMS_ONLY)
*/
$argList = func_get_args();
if (count($argList) > 2) {
throw new InvalidArgumentException('Wrong number of arguments');
} else {
if (count($argList) == 2) {
$startId = $argList[0];
$mode = $argList[1];
} else {
if (count($argList) == 1) {
$startId = $argList[0];
$mode = self::SM_TERMS_ONLY;
} else {
$startId = 0;
$mode = self::SM_TERMS_ONLY;
}
}
}
if ($this->_tisFile !== null) {
$this->_tisFile = null;
}
$this->_tisFile = $this->openCompoundFile('.tis', false);
$this->_tisFileOffset = $this->_tisFile->tell();
$tiVersion = $this->_tisFile->readInt();
if ($tiVersion != (int) 0xfffffffe && $tiVersion != (int) 0xfffffffd) {
throw new InvalidFileFormatException('Wrong TermInfoFile file format');
}
$this->_termCount = $this->_termNum = $this->_tisFile->readLong();
// Read terms count
$this->_indexInterval = $this->_tisFile->readInt();
// Read Index interval
$this->_skipInterval = $this->_tisFile->readInt();
// Read skip interval
if ($tiVersion == (int) 0xfffffffd) {
$maxSkipLevels = $this->_tisFile->readInt();
}
if ($this->_frqFile !== null) {
$this->_frqFile = null;
}
if ($this->_prxFile !== null) {
$this->_prxFile = null;
}
$this->_docMap = array();
$this->_lastTerm = new Term('', -1);
$this->_lastTermInfo = new TermInfo(0, 0, 0, 0);
$this->_lastTermPositions = null;
$this->_termsScanMode = $mode;
switch ($mode) {
case self::SM_TERMS_ONLY:
// Do nothing
break;
case self::SM_FULL_INFO:
// break intentionally omitted
// break intentionally omitted
case self::SM_MERGE_INFO:
$this->_frqFile = $this->openCompoundFile('.frq', false);
$this->_frqFileOffset = $this->_frqFile->tell();
$this->_prxFile = $this->openCompoundFile('.prx', false);
$this->_prxFileOffset = $this->_prxFile->tell();
for ($count = 0; $count < $this->_docCount; $count++) {
if (!$this->isDeleted($count)) {
$this->_docMap[$count] = $startId + ($mode == self::SM_MERGE_INFO ? count($this->_docMap) : $count);
}
}
break;
default:
throw new InvalidArgumentException('Wrong terms scaning mode specified.');
break;
}
// Calculate next segment start id (since $this->_docMap structure may be cleaned by $this->nextTerm() call)
$nextSegmentStartId = $startId + ($mode == self::SM_MERGE_INFO ? count($this->_docMap) : $this->_docCount);
$this->nextTerm();
return $nextSegmentStartId;
}
/**
* Add term
*
* Term positions is an array( docId => array(pos1, pos2, pos3, ...), ... )
*
* @param \Zend\Search\Lucene\Index\Term $termEntry
* @param array $termDocs
*/
public function addTerm($termEntry, $termDocs)
{
$freqPointer = $this->_frqFile->tell();
$proxPointer = $this->_prxFile->tell();
$prevDoc = 0;
foreach ($termDocs as $docId => $termPositions) {
$docDelta = ($docId - $prevDoc) * 2;
$prevDoc = $docId;
if (count($termPositions) > 1) {
$this->_frqFile->writeVInt($docDelta);
$this->_frqFile->writeVInt(count($termPositions));
} else {
$this->_frqFile->writeVInt($docDelta + 1);
}
$prevPosition = 0;
foreach ($termPositions as $position) {
$this->_prxFile->writeVInt($position - $prevPosition);
$prevPosition = $position;
}
}
if (count($termDocs) >= self::$skipInterval) {
/**
* @todo Write Skip Data to a freq file.
* It's not used now, but make index more optimal
*/
$skipOffset = $this->_frqFile->tell() - $freqPointer;
} else {
$skipOffset = 0;
}
$term = new Index\Term($termEntry->text, $this->_fields[$termEntry->field]->number);
$termInfo = new Index\TermInfo(count($termDocs), $freqPointer, $proxPointer, $skipOffset);
$this->_dumpTermDictEntry($this->_tisFile, $this->_prevTerm, $term, $this->_prevTermInfo, $termInfo);
if (($this->_termCount + 1) % self::$indexInterval == 0) {
$this->_dumpTermDictEntry($this->_tiiFile, $this->_prevIndexTerm, $term, $this->_prevIndexTermInfo, $termInfo);
$indexPosition = $this->_tisFile->tell();
$this->_tiiFile->writeVInt($indexPosition - $this->_lastIndexPosition);
$this->_lastIndexPosition = $indexPosition;
}
$this->_termCount++;
}
