writer.php

zend的加强包 zend的加强包

        $this->commit();
    }

    /**
     * Update segments file by adding current segment to a list
     *
     * @throws Zend_Search_Lucene_Exception
     */
    private function _updateSegments()
    {
        // Get an exclusive index lock
        Zend_Search_Lucene_LockManager::obtainWriteLock($this->_directory);

        $generation = Zend_Search_Lucene::getActualGeneration($this->_directory);
        $segmentsFile   = $this->_directory->getFileObject(Zend_Search_Lucene::getSegmentFileName($generation), false);
        $newSegmentFile = $this->_directory->createFile(Zend_Search_Lucene::getSegmentFileName(++$generation), false);

        try {
            $genFile = $this->_directory->getFileObject('segments.gen', false);
        } catch (Zend_Search_Lucene_Exception $e) {
            if (strpos($e->getMessage(), 'is not readable') !== false) {
                $genFile = $this->_directory->createFile('segments.gen');
            } else {
                throw $e;
            }
        }
        
        $genFile->writeInt((int)0xFFFFFFFE);
        // Write generation (first copy)
        $genFile->writeLong($generation);

        try {
            // Write format marker
            $newSegmentFile->writeInt((int)0xFFFFFFFD);
    
            // Skip format identifier
            $segmentsFile->seek(4, SEEK_CUR);
            // $version = $segmentsFile->readLong() + $this->_versionUpdate;
            // Process version on 32-bit platforms
            $versionHigh = $segmentsFile->readInt();
            $versionLow  = $segmentsFile->readInt();
            $version = $versionHigh * ((double)0xFFFFFFFF + 1) +
                       (($versionLow < 0)? (double)0xFFFFFFFF - (-1 - $versionLow) : $versionLow);
            $version += $this->_versionUpdate;
            $this->_versionUpdate = 0;
            $newSegmentFile->writeInt((int)($version/((double)0xFFFFFFFF + 1)));
            $newSegmentFile->writeInt((int)($version & 0xFFFFFFFF));
    
            // Write segment name counter
            $newSegmentFile->writeInt($segmentsFile->readInt());
    
            // Get number of segments offset
            $numOfSegmentsOffset = $newSegmentFile->tell();
            // Write dummy data (segment counter)
            $newSegmentFile->writeInt(0);
    
            // Read number of segemnts
            $segmentsCount = $segmentsFile->readInt();

            $segments = array();
            for ($count = 0; $count < $segmentsCount; $count++) {
                $segName = $segmentsFile->readString();
                $segSize = $segmentsFile->readInt();
    
                if ($generation == 1 /* retrieved generation is 0 */) {
                    // pre-2.1 index format
                    $delGenHigh = 0;
                    $delGenLow  = 0;
                    $hasSingleNormFile = false;
                    $numField = (int)0xFFFFFFFF;
                    $isCompound = 1;
                } else {
                    //$delGen          = $segmentsFile->readLong();
                    $delGenHigh        = $segmentsFile->readInt();
                    $delGenLow         = $segmentsFile->readInt();
                    $hasSingleNormFile = $segmentsFile->readByte();
                    $numField          = $segmentsFile->readInt();
    
                    $normGens = array();
                    if ($numField != (int)0xFFFFFFFF) {
                        for ($count1 = 0; $count1 < $numField; $count1++) {
                            $normGens[] = $segmentsFile->readLong();
                        }
                    }
                    $isCompound        = $segmentsFile->readByte();
                }
    
                if (!in_array($segName, $this->_segmentsToDelete)) {
                    // Load segment if necessary
                    if (!isset($this->_segmentInfos[$segName])) {
                        $delGen = $delGenHigh * ((double)0xFFFFFFFF + 1) +
                                     (($delGenLow < 0)? (double)0xFFFFFFFF - (-1 - $delGenLow) : $delGenLow);
                        $this->_segmentInfos[$segName] = 
                                    new Zend_Search_Lucene_Index_SegmentInfo($this->_directory,
                                                                             $segName,
                                                                             $segSize,
                                                                             $delGen,
                                                                             $hasSingleNormFile,
                                                                             $isCompound);
                    } else {
                        // Retrieve actual detetions file generation number
                        $delGen = $this->_segmentInfos[$segName]->getDelGen();
                        
                        if ($delGen >= 0) {
                            $delGenHigh = (int)($delGen/((double)0xFFFFFFFF + 1));
                            $delGenLow  =(int)($delGen & 0xFFFFFFFF);
                        } else {
                            $delGenHigh = $delGenLow = (int)0xFFFFFFFF;
                        }
                    }
                    
                    $newSegmentFile->writeString($segName);
                    $newSegmentFile->writeInt($segSize);
                    $newSegmentFile->writeInt($delGenHigh);
                    $newSegmentFile->writeInt($delGenLow);
                    $newSegmentFile->writeByte($hasSingleNormFile);
                    $newSegmentFile->writeInt($numField);
                    if ($numField != (int)0xFFFFFFFF) {
                        foreach ($normGens as $normGen) {
                            $newSegmentFile->writeLong($normGen);
                        }
                    }
                    $newSegmentFile->writeByte($isCompound);
    
                    $segments[$segName] = $segSize;
                }
            }
            $segmentsFile->close();
    
            $segmentsCount = count($segments) + count($this->_newSegments);
    
            foreach ($this->_newSegments as $segName => $segmentInfo) {
                $newSegmentFile->writeString($segName);
                $newSegmentFile->writeInt($segmentInfo->count());
    
                // delete file generation: -1 (there is no delete file yet)
                $newSegmentFile->writeInt((int)0xFFFFFFFF);$newSegmentFile->writeInt((int)0xFFFFFFFF);
                // HasSingleNormFile
                $newSegmentFile->writeByte($segmentInfo->hasSingleNormFile());
                // NumField
                $newSegmentFile->writeInt((int)0xFFFFFFFF);
                // IsCompoundFile
                $newSegmentFile->writeByte($segmentInfo->isCompound());
    
                $segments[$segmentInfo->getName()] = $segmentInfo->count();
                $this->_segmentInfos[$segName] = $segmentInfo;
            }
            $this->_newSegments = array();
    
            $newSegmentFile->seek($numOfSegmentsOffset);
            $newSegmentFile->writeInt($segmentsCount);  // Update segments count
            $newSegmentFile->close();
        } catch (Exception $e) {
            /** Restore previous index generation */
            $generation--;
            $genFile->seek(4, SEEK_SET);
            // Write generation number twice
            $genFile->writeLong($generation); $genFile->writeLong($generation);

            // Release index write lock
            Zend_Search_Lucene_LockManager::releaseWriteLock($this->_directory);
            
            // Throw the exception
            throw $e;
        }

        // Write generation (second copy)
        $genFile->writeLong($generation);

        
        // Check if another update process is not running now
        // If yes, skip clean-up procedure
        if (Zend_Search_Lucene_LockManager::escalateReadLock($this->_directory)) {
            /**
             * Clean-up directory
             */
            $filesToDelete = array();
            $filesTypes    = array();
            $filesNumbers  = array();
            
            // list of .del files of currently used segments
            // each segment can have several generations of .del files
            // only last should not be deleted
            $delFiles = array();
            
            foreach ($this->_directory->fileList() as $file) {
                if ($file == 'deletable') {
                    // 'deletable' file
                    $filesToDelete[] = $file;
                    $filesTypes[]    = 0; // delete this file first, since it's not used starting from Lucene v2.1
                    $filesNumbers[]  = 0;
                } else if ($file == 'segments') {
                    // 'segments' file
    
                    $filesToDelete[] = $file;
                    $filesTypes[]    = 1; // second file to be deleted "zero" version of segments file (Lucene pre-2.1)
                    $filesNumbers[]  = 0;
                } else if (preg_match('/^segments_[a-zA-Z0-9]+$/i', $file)) {
                    // 'segments_xxx' file
                    // Check if it's not a just created generation file
                    if ($file != Zend_Search_Lucene::getSegmentFileName($generation)) {
                        $filesToDelete[] = $file;
                        $filesTypes[]    = 2; // first group of files for deletions
                        $filesNumbers[]  = (int)base_convert(substr($file, 9), 36, 10); // ordered by segment generation numbers 
                    }
                } else if (preg_match('/(^_([a-zA-Z0-9]+))\.f\d+$/i', $file, $matches)) {
                    // one of per segment files ('<segment_name>.f<decimal_number>')
                    // Check if it's not one of the segments in the current segments set
                    if (!isset($segments[$matches[1]])) {
                        $filesToDelete[] = $file;
                        $filesTypes[]    = 3; // second group of files for deletions
                        $filesNumbers[]  = (int)base_convert($matches[2], 36, 10); // order by segment number 
                    }
                } else if (preg_match('/(^_([a-zA-Z0-9]+))(_([a-zA-Z0-9]+))\.del$/i', $file, $matches)) {
                    // one of per segment files ('<segment_name>_<del_generation>.del' where <segment_name> is '_<segment_number>')
                    // Check if it's not one of the segments in the current segments set
                    if (!isset($segments[$matches[1]])) {
                        $filesToDelete[] = $file;
                        $filesTypes[]    = 3; // second group of files for deletions
                        $filesNumbers[]  = (int)base_convert($matches[2], 36, 10); // order by segment number 
                    } else {
                        $segmentNumber = (int)base_convert($matches[2], 36, 10);
                        $delGeneration = (int)base_convert($matches[4], 36, 10);
                        if (!isset($delFiles[$segmentNumber])) {
                            $delFiles[$segmentNumber] = array();
                        }
                        $delFiles[$segmentNumber][$delGeneration] = $file;
                    }
                } else if (isset(self::$_indexExtensions[substr($file, strlen($file)-4)])) {
                    // one of per segment files ('<segment_name>.<ext>')
                    $segmentName = substr($file, 0, strlen($file) - 4);
                    // Check if it's not one of the segments in the current segments set
                    if (!isset($segments[$segmentName])  &&
                        ($this->_currentSegment === null  ||  $this->_currentSegment->getName() != $segmentName)) {
                        $filesToDelete[] = $file;
                        $filesTypes[]    = 3; // second group of files for deletions
                        $filesNumbers[]  = (int)base_convert(substr($file, 1 /* skip '_' */, strlen($file)-5), 36, 10); // order by segment number 
                    }
                }
            }

            $maxGenNumber = 0;
            // process .del files of currently used segments
            foreach ($delFiles as $segmentNumber => $segmentDelFiles) {
                ksort($delFiles[$segmentNumber], SORT_NUMERIC);
                array_pop($delFiles[$segmentNumber]); // remove last delete file generation from candidates for deleting
                
                end($delFiles[$segmentNumber]);
                $lastGenNumber = key($delFiles[$segmentNumber]);
                if ($lastGenNumber > $maxGenNumber) {
                    $maxGenNumber = $lastGenNumber; 
                }
            }
            foreach ($delFiles as $segmentNumber => $segmentDelFiles) {
                foreach ($segmentDelFiles as $delGeneration => $file) {
                        $filesToDelete[] = $file;
                        $filesTypes[]    = 4; // third group of files for deletions
                        $filesNumbers[]  = $segmentNumber*$maxGenNumber + $delGeneration; // order by <segment_number>,<del_generation> pair 
                }
            }
            
            // Reorder files for deleting
            array_multisort($filesTypes,    SORT_ASC, SORT_NUMERIC,
                            $filesNumbers,  SORT_ASC, SORT_NUMERIC,
                            $filesToDelete, SORT_ASC, SORT_STRING);
            
            foreach ($filesToDelete as $file) {
                try {
                    $this->_directory->deleteFile($file);
                } catch (Zend_Search_Lucene_Exception $e) {
                    if (strpos($e->getMessage(), 'Can\'t delete file') === false) {
                        // That's not "file is under processing or already deleted" exception
                        // Pass it through
                        throw $e;
                    }
                }
            }
            
            // Return read lock into the previous state
            Zend_Search_Lucene_LockManager::deEscalateReadLock($this->_directory);
        } else {
            // Only release resources if another index reader is running now
            foreach ($this->_segmentsToDelete as $segName) {
                foreach (self::$_indexExtensions as $ext) {
                    $this->_directory->purgeFile($segName . $ext);
                }
            }
        }

        // Clean-up _segmentsToDelete container
        $this->_segmentsToDelete = array();
        

        // Release index write lock
        Zend_Search_Lucene_LockManager::releaseWriteLock($this->_directory);

        // Remove unused segments from segments list
        foreach ($this->_segmentInfos as $segName => $segmentInfo) {
            if (!isset($segments[$segName])) {
                unset($this->_segmentInfos[$segName]);
            }
        }
    }

    /**
     * Commit current changes
     */
    public function commit()
    {
        if ($this->_currentSegment !== null) {
            $newSegment = $this->_currentSegment->close();
            if ($newSegment !== null) {
                $this->_newSegments[$newSegment->getName()] = $newSegment;
            }
            $this->_currentSegment = null;
        }

        $this->_updateSegments();
    }


    /**
     * Merges the provided indexes into this index.
     *
     * @param array $readers
     * @return void
     */
    public function addIndexes($readers)
    {
        /**
         * @todo implementation
         */
    }

    /**
     * Merges all segments together into new one
     * 
     * Returns true on success and false if another optimization or auto-optimization process 
     * is running now 
     *
     * @return boolean
     */
    public function optimize()
    {
        if (Zend_Search_Lucene_LockManager::obtainOptimizationLock($this->_directory) === false) {
            return false;
        }

        $this->_mergeSegments($this->_segmentInfos);
        
        Zend_Search_Lucene_LockManager::releaseOptimizationLock($this->_directory);
        
        return true;
    }

    /**
     * Get name for new segment
     *
     * @return string
     */
    private function _newSegmentName()
    {
        Zend_Search_Lucene_LockManager::obtainWriteLock($this->_directory);
        
        $generation = Zend_Search_Lucene::getActualGeneration($this->_directory);
        $segmentsFile = $this->_directory->getFileObject(Zend_Search_Lucene::getSegmentFileName($generation), false);

        $segmentsFile->seek(12); // 12 = 4 (int, file format marker) + 8 (long, index version)
        $segmentNameCounter = $segmentsFile->readInt();

        $segmentsFile->seek(12); // 12 = 4 (int, file format marker) + 8 (long, index version)
        $segmentsFile->writeInt($segmentNameCounter + 1);

        // Flash output to guarantee that wrong value will not be loaded between unlock and
        // return (which calls $segmentsFile destructor)
        $segmentsFile->flush();

        Zend_Search_Lucene_LockManager::releaseWriteLock($this->_directory);
        
        return '_' . base_convert($segmentNameCounter, 10, 36);
    }

}