lucene.php

很棒的在线教学系统

<?php
/**
 * Zend Framework
 *
 * LICENSE
 *
 * This source file is subject to the new BSD license that is bundled
 * with this package in the file LICENSE.txt.
 * It is also available through the world-wide-web at this URL:
 * http://framework.zend.com/license/new-bsd
 * If you did not receive a copy of the license and are unable to
 * obtain it through the world-wide-web, please send an email
 * to [e-mail] so we can send you a copy immediately.
 *
 * @category   Zend
 * @package    Zend_Search_Lucene
 * @copyright  Copyright (c) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
 * @license    http://framework.zend.com/license/new-bsd     New BSD License
 */

/** Zend_Search_Lucene_Exception */
require_once "Zend/Search/Lucene/Exception.php";

/** Zend_Search_Lucene_Document */
require_once "Zend/Search/Lucene/Document.php";

/** Zend_Search_Lucene_Document_Html */
require_once "Zend/Search/Lucene/Document/Html.php";

/** Zend_Search_Lucene_Storage_Directory_Filesystem */
require_once "Zend/Search/Lucene/Storage/Directory/Filesystem.php";

/** Zend_Search_Lucene_Storage_File_Memory */
require_once "Zend/Search/Lucene/Storage/File/Memory.php";

/** Zend_Search_Lucene_Index_Term */
require_once "Zend/Search/Lucene/Index/Term.php";

/** Zend_Search_Lucene_Index_TermInfo */
require_once "Zend/Search/Lucene/Index/TermInfo.php";

/** Zend_Search_Lucene_Index_SegmentInfo */
require_once "Zend/Search/Lucene/Index/SegmentInfo.php";

/** Zend_Search_Lucene_Index_FieldInfo */
require_once "Zend/Search/Lucene/Index/FieldInfo.php";

/** Zend_Search_Lucene_Index_Writer */
require_once "Zend/Search/Lucene/Index/Writer.php";

/** Zend_Search_Lucene_Search_QueryParser */
require_once "Zend/Search/Lucene/Search/QueryParser.php";

/** Zend_Search_Lucene_Search_QueryHit */
require_once "Zend/Search/Lucene/Search/QueryHit.php";

/** Zend_Search_Lucene_Search_Similarity */
require_once "Zend/Search/Lucene/Search/Similarity.php";

/** Zend_Search_Lucene_Index_SegmentInfoPriorityQueue */
require_once "Zend/Search/Lucene/Index/SegmentInfoPriorityQueue.php";

/** Zend_Search_Lucene_LockManager */
if (@$CFG->block_search_softlock){
    require_once "Zend/Search/Lucene/SoftLockManager.php";
} else {
    require_once "Zend/Search/Lucene/LockManager.php";
}



/** Zend_Search_Lucene_Interface */
require_once "Zend/Search/Lucene/Interface.php";

/** Zend_Search_Lucene_Proxy */
require_once "Zend/Search/Lucene/Proxy.php";


/**
 * @category   Zend
 * @package    Zend_Search_Lucene
 * @copyright  Copyright (c) 2005-2008 Zend Technologies USA Inc. (http://www.zend.com)
 * @license    http://framework.zend.com/license/new-bsd     New BSD License
 */
class Zend_Search_Lucene implements Zend_Search_Lucene_Interface
{
    /**
     * Default field name for search
     *
     * Null means search through all fields
     *
     * @var string
     */
    private static $_defaultSearchField = null;

    /**
     * Result set limit
     *
     * 0 means no limit
     *
     * @var integer
     */
    private static $_resultSetLimit = 0;

    /**
     * File system adapter.
     *
     * @var Zend_Search_Lucene_Storage_Directory
     */
    private $_directory = null;

    /**
     * File system adapter closing option
     *
     * @var boolean
     */
    private $_closeDirOnExit = true;

    /**
     * Writer for this index, not instantiated unless required.
     *
     * @var Zend_Search_Lucene_Index_Writer
     */
    private $_writer = null;

    /**
     * Array of Zend_Search_Lucene_Index_SegmentInfo objects for this index.
     *
     * @var array Zend_Search_Lucene_Index_SegmentInfo
     */
    private $_segmentInfos = array();

    /**
     * Number of documents in this index.
     *
     * @var integer
     */
    private $_docCount = 0;

    /**
     * Flag for index changes
     *
     * @var boolean
     */
    private $_hasChanges = false;


    /**
     * Signal, that index is already closed, changes are fixed and resources are cleaned up
     *
     * @var boolean
     */
    private $_closed = false;

    /**
     * Number of references to the index object
     *
     * @var integer
     */
    private $_refCount = 0;

    /**
     * Current segment generation
     *
     * @var integer
     */
    private $_generation;

    
    /**
     * Create index
     *
     * @param mixed $directory
     * @return Zend_Search_Lucene_Interface
     */
    public static function create($directory)
    {
        return new Zend_Search_Lucene_Proxy(new Zend_Search_Lucene($directory, true));
    }

    /**
     * Open index
     *
     * @param mixed $directory
     * @return Zend_Search_Lucene_Interface
     */
    public static function open($directory)
    {
        return new Zend_Search_Lucene_Proxy(new Zend_Search_Lucene($directory, false));
    }

    /** Generation retrieving counter */
    const GENERATION_RETRIEVE_COUNT = 10;

    /** Pause between generation retrieving attempts in milliseconds */
    const GENERATION_RETRIEVE_PAUSE = 50;

    /**
     * Get current generation number
     *
     * Returns generation number
     * 0 means pre-2.1 index format
     * -1 means there are no segments files.
     *
     * @param Zend_Search_Lucene_Storage_Directory $directory
     * @return integer
     * @throws Zend_Search_Lucene_Exception
     */
    public static function getActualGeneration(Zend_Search_Lucene_Storage_Directory $directory)
    {
        /**
         * Zend_Search_Lucene uses segments.gen file to retrieve current generation number
         *
         * Apache Lucene index format documentation mentions this method only as a fallback method
         *
         * Nevertheless we use it according to the performance considerations
         *
         * @todo check if we can use some modification of Apache Lucene generation determination algorithm
         *       without performance problems
         */

        try {
            for ($count = 0; $count < self::GENERATION_RETRIEVE_COUNT; $count++) {
                // Try to get generation file
                $genFile = $directory->getFileObject('segments.gen', false);

                $format = $genFile->readInt();
                if ($format != (int)0xFFFFFFFE) {
                    throw new Zend_Search_Lucene_Exception('Wrong segments.gen file format');
                }

                $gen1 = $genFile->readLong();
                $gen2 = $genFile->readLong();

                if ($gen1 == $gen2) {
                    return $gen1;
                }

                usleep(self::GENERATION_RETRIEVE_PAUSE * 1000);
            }

            // All passes are failed
            throw new Zend_Search_Lucene_Exception('Index is under processing now');
        } catch (Zend_Search_Lucene_Exception $e) {
            if (strpos($e->getMessage(), 'is not readable') !== false) {
                try {
                    // Try to open old style segments file
                    $segmentsFile = $directory->getFileObject('segments', false);

                    // It's pre-2.1 index
                    return 0;
                } catch (Zend_Search_Lucene_Exception $e) {
                    if (strpos($e->getMessage(), 'is not readable') !== false) {
                        return -1;
                    } else {
                        throw $e;
                    }
                }
            } else {
                throw $e;
            }
        }

        return -1;
    }

    /**
     * Get segments file name
     *
     * @param integer $generation
     * @return string
     */
    public static function getSegmentFileName($generation)
    {
        if ($generation == 0) {
            return 'segments';
        }

        return 'segments_' . base_convert($generation, 10, 36);
    }

    /**
     * Read segments file for pre-2.1 Lucene index format
     */
    private function _readPre21SegmentsFile()
    {
        $segmentsFile = $this->_directory->getFileObject('segments');

        $format = $segmentsFile->readInt();

        if ($format != (int)0xFFFFFFFF) {
            throw new Zend_Search_Lucene_Exception('Wrong segments file format');
        }

        // read version
        // $segmentsFile->readLong();
        $segmentsFile->readInt(); $segmentsFile->readInt();

        // read segment name counter
        $segmentsFile->readInt();

        $segments = $segmentsFile->readInt();

        $this->_docCount = 0;

        // read segmentInfos
        for ($count = 0; $count < $segments; $count++) {
            $segName = $segmentsFile->readString();
            $segSize = $segmentsFile->readInt();
            $this->_docCount += $segSize;

            $this->_segmentInfos[$segName] =
                                new Zend_Search_Lucene_Index_SegmentInfo($this->_directory,
                                                                         $segName,
                                                                         $segSize);
        }
    }

    /**
     * Read segments file
     *
     * @throws Zend_Search_Lucene_Exception
     */
    private function _readSegmentsFile()
    {
        $segmentsFile = $this->_directory->getFileObject(self::getSegmentFileName($this->_generation));

        $format = $segmentsFile->readInt();

        if ($format != (int)0xFFFFFFFD) {
            throw new Zend_Search_Lucene_Exception('Wrong segments file format');
        }

        // read version
        // $segmentsFile->readLong();
        $segmentsFile->readInt(); $segmentsFile->readInt();

        // read segment name counter
        $segmentsFile->readInt();

        $segments = $segmentsFile->readInt();

        $this->_docCount = 0;

        // read segmentInfos
        for ($count = 0; $count < $segments; $count++) {
            $segName = $segmentsFile->readString();
            $segSize = $segmentsFile->readInt();

            // 2.1+ specific properties
            //$delGen          = $segmentsFile->readLong();
            $delGenHigh        = $segmentsFile->readInt();
            $delGenLow         = $segmentsFile->readInt();
            if ($delGenHigh == (int)0xFFFFFFFF  && $delGenLow == (int)0xFFFFFFFF) {
                $delGen = -1; // There are no deletes
            } else {
                $delGen = ($delGenHigh << 32) | $delGenLow;
            }

            $hasSingleNormFile = $segmentsFile->readByte();
            $numField          = $segmentsFile->readInt();

            $normGens = array();
            if ($numField != (int)0xFFFFFFFF) {
                for ($count1 = 0; $count1 < $numField; $count1++) {
                    $normGens[] = $segmentsFile->readLong();
                }

                throw new Zend_Search_Lucene_Exception('Separate norm files are not supported. Optimize index to use it with Zend_Search_Lucene.');
            }

            $isCompound        = $segmentsFile->readByte();


            $this->_docCount += $segSize;

            $this->_segmentInfos[$segName] =
                                new Zend_Search_Lucene_Index_SegmentInfo($this->_directory,
                                                                         $segName,
                                                                         $segSize,
                                                                         $delGen,
                                                                         $hasSingleNormFile,
                                                                         $isCompound);
        }
    }

    /**
     * Opens the index.
     *
     * IndexReader constructor needs Directory as a parameter. It should be
     * a string with a path to the index folder or a Directory object.
     *
     * @param mixed $directory
     * @throws Zend_Search_Lucene_Exception
     */
    public function __construct($directory = null, $create = false)
    {
        if ($directory === null) {
            throw new Zend_Search_Exception('No index directory specified');
        }

        if ($directory instanceof Zend_Search_Lucene_Storage_Directory_Filesystem) {
            $this->_directory      = $directory;
            $this->_closeDirOnExit = false;
        } else {
            $this->_directory      = new Zend_Search_Lucene_Storage_Directory_Filesystem($directory);
            $this->_closeDirOnExit = true;
        }

        $this->_segmentInfos = array();

        // Mark index as "under processing" to prevent other processes from premature index cleaning
        Zend_Search_Lucene_LockManager::obtainReadLock($this->_directory);
        
        // Escalate read lock to prevent current generation index files to be deleted while opening process is not done 
        Zend_Search_Lucene_LockManager::escalateReadLock($this->_directory);
        
        
        $this->_generation = self::getActualGeneration($this->_directory);
        
        if ($create) {
        	try {
        		Zend_Search_Lucene_LockManager::obtainWriteLock($this->_directory);
        	} catch (Zend_Search_Lucene_Exception $e) {
        		if (strpos($e->getMessage(), 'Can\'t obtain exclusive index lock') === false) {
        			throw $e;
        		} else {
        			throw new Zend_Search_Lucene_Exception('Can\'t create index. It\'s under processing now');
        		}
        	}

            if ($this->_generation == -1) {
                // Directory doesn't contain existing index, start from 1
                $this->_generation = 1;
                $nameCounter = 0;
            } else {
                // Directory contains existing index
                $segmentsFile = $this->_directory->getFileObject(self::getSegmentFileName($this->_generation));
                $segmentsFile->seek(12); // 12 = 4 (int, file format marker) + 8 (long, index version)

                $nameCounter = $segmentsFile->readInt();
                $this->_generation++;
            }

            Zend_Search_Lucene_Index_Writer::createIndex($this->_directory, $this->_generation, $nameCounter);

            Zend_Search_Lucene_LockManager::releaseWriteLock($this->_directory);
        }

        if ($this->_generation == -1) {
            throw new Zend_Search_Lucene_Exception('Index doesn\'t exists in the specified directory.');
        } else if ($this->_generation == 0) {
            $this->_readPre21SegmentsFile();
        } else {
            $this->_readSegmentsFile();
        }
        
        // De-escalate read lock to prevent current generation index files to be deleted while opening process is not done 
        Zend_Search_Lucene_LockManager::deEscalateReadLock($this->_directory);
    }

    /**
     * Close current index and free resources
     */
    private function _close()
    {
        if ($this->_closed) {
            // index is already closed and resources are cleaned up
            return;
        }

        $this->commit();

        // Release "under processing" flag
        Zend_Search_Lucene_LockManager::releaseReadLock($this->_directory);
                
        if ($this->_closeDirOnExit) {
            $this->_directory->close();