web/lib/Zend/Search/Lucene/Index/Writer.php
author Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
Tue, 15 Mar 2011 16:32:22 +0100
changeset 68 ecaf28ffe26e
parent 0 web/Zend/Search/Lucene/Index/Writer.php@4eba9c11703f
parent 64 web/Zend/Search/Lucene/Index/Writer.php@162c1de6545a
child 207 621fa6caec0c
permissions -rw-r--r--
Merge with cc1eea280cdb9d27ecdc9a2898de7a2b9835cde7
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     1
<?php
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     2
/**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     3
 * Zend Framework
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     4
 *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     5
 * LICENSE
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     6
 *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     7
 * This source file is subject to the new BSD license that is bundled
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     8
 * with this package in the file LICENSE.txt.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
     9
 * It is also available through the world-wide-web at this URL:
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    10
 * http://framework.zend.com/license/new-bsd
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    11
 * If you did not receive a copy of the license and are unable to
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    12
 * obtain it through the world-wide-web, please send an email
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    13
 * to license@zend.com so we can send you a copy immediately.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    14
 *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    15
 * @category   Zend
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    16
 * @package    Zend_Search_Lucene
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    17
 * @subpackage Index
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    18
 * @copyright  Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    19
 * @license    http://framework.zend.com/license/new-bsd     New BSD License
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    20
 * @version    $Id: Writer.php 20096 2010-01-06 02:05:09Z bkarwin $
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    21
 */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    22
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    23
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    24
/** Zend_Search_Lucene_LockManager */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    25
require_once 'Zend/Search/Lucene/LockManager.php';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    26
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    27
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    28
/**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    29
 * @category   Zend
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    30
 * @package    Zend_Search_Lucene
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    31
 * @subpackage Index
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    32
 * @copyright  Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    33
 * @license    http://framework.zend.com/license/new-bsd     New BSD License
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    34
 */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    35
class Zend_Search_Lucene_Index_Writer
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    36
{
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    37
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    38
     * @todo Implement Analyzer substitution
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    39
     * @todo Implement Zend_Search_Lucene_Storage_DirectoryRAM and Zend_Search_Lucene_Storage_FileRAM to use it for
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    40
     *       temporary index files
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    41
     * @todo Directory lock processing
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    42
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    43
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    44
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    45
     * Number of documents required before the buffered in-memory
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    46
     * documents are written into a new Segment
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    47
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    48
     * Default value is 10
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    49
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    50
     * @var integer
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    51
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    52
    public $maxBufferedDocs = 10;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    53
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    54
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    55
     * Largest number of documents ever merged by addDocument().
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    56
     * Small values (e.g., less than 10,000) are best for interactive indexing,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    57
     * as this limits the length of pauses while indexing to a few seconds.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    58
     * Larger values are best for batched indexing and speedier searches.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    59
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    60
     * Default value is PHP_INT_MAX
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    61
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    62
     * @var integer
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    63
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    64
    public $maxMergeDocs = PHP_INT_MAX;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    65
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    66
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    67
     * Determines how often segment indices are merged by addDocument().
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    68
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    69
     * With smaller values, less RAM is used while indexing,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    70
     * and searches on unoptimized indices are faster,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    71
     * but indexing speed is slower.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    72
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    73
     * With larger values, more RAM is used during indexing,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    74
     * and while searches on unoptimized indices are slower,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    75
     * indexing is faster.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    76
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    77
     * Thus larger values (> 10) are best for batch index creation,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    78
     * and smaller values (< 10) for indices that are interactively maintained.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    79
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    80
     * Default value is 10
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    81
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    82
     * @var integer
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    83
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    84
    public $mergeFactor = 10;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    85
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    86
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    87
     * File system adapter.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    88
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    89
     * @var Zend_Search_Lucene_Storage_Directory
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    90
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    91
    private $_directory = null;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    92
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    93
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    94
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    95
     * Changes counter.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    96
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    97
     * @var integer
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    98
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
    99
    private $_versionUpdate = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   100
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   101
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   102
     * List of the segments, created by index writer
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   103
     * Array of Zend_Search_Lucene_Index_SegmentInfo objects
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   104
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   105
     * @var array
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   106
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   107
    private $_newSegments = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   108
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   109
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   110
     * List of segments to be deleted on commit
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   111
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   112
     * @var array
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   113
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   114
    private $_segmentsToDelete = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   115
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   116
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   117
     * Current segment to add documents
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   118
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   119
     * @var Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   120
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   121
    private $_currentSegment = null;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   122
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   123
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   124
     * Array of Zend_Search_Lucene_Index_SegmentInfo objects for this index.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   125
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   126
     * It's a reference to the corresponding Zend_Search_Lucene::$_segmentInfos array
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   127
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   128
     * @var array Zend_Search_Lucene_Index_SegmentInfo
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   129
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   130
    private $_segmentInfos;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   131
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   132
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   133
     * Index target format version
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   134
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   135
     * @var integer
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   136
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   137
    private $_targetFormatVersion;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   138
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   139
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   140
     * List of indexfiles extensions
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   141
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   142
     * @var array
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   143
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   144
    private static $_indexExtensions = array('.cfs' => '.cfs',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   145
                                             '.cfx' => '.cfx',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   146
                                             '.fnm' => '.fnm',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   147
                                             '.fdx' => '.fdx',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   148
                                             '.fdt' => '.fdt',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   149
                                             '.tis' => '.tis',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   150
                                             '.tii' => '.tii',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   151
                                             '.frq' => '.frq',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   152
                                             '.prx' => '.prx',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   153
                                             '.tvx' => '.tvx',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   154
                                             '.tvd' => '.tvd',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   155
                                             '.tvf' => '.tvf',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   156
                                             '.del' => '.del',
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   157
                                             '.sti' => '.sti' );
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   158
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   159
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   160
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   161
     * Create empty index
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   162
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   163
     * @param Zend_Search_Lucene_Storage_Directory $directory
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   164
     * @param integer $generation
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   165
     * @param integer $nameCount
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   166
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   167
    public static function createIndex(Zend_Search_Lucene_Storage_Directory $directory, $generation, $nameCount)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   168
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   169
        if ($generation == 0) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   170
            // Create index in pre-2.1 mode
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   171
            foreach ($directory->fileList() as $file) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   172
                if ($file == 'deletable' ||
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   173
                    $file == 'segments'  ||
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   174
                    isset(self::$_indexExtensions[ substr($file, strlen($file)-4)]) ||
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   175
                    preg_match('/\.f\d+$/i', $file) /* matches <segment_name>.f<decimal_nmber> file names */) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   176
                        $directory->deleteFile($file);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   177
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   178
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   179
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   180
            $segmentsFile = $directory->createFile('segments');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   181
            $segmentsFile->writeInt((int)0xFFFFFFFF);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   182
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   183
            // write version (initialized by current time)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   184
            $segmentsFile->writeLong(round(microtime(true)));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   185
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   186
            // write name counter
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   187
            $segmentsFile->writeInt($nameCount);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   188
            // write segment counter
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   189
            $segmentsFile->writeInt(0);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   190
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   191
            $deletableFile = $directory->createFile('deletable');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   192
            // write counter
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   193
            $deletableFile->writeInt(0);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   194
        } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   195
            $genFile = $directory->createFile('segments.gen');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   196
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   197
            $genFile->writeInt((int)0xFFFFFFFE);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   198
            // Write generation two times
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   199
            $genFile->writeLong($generation);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   200
            $genFile->writeLong($generation);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   201
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   202
            $segmentsFile = $directory->createFile(Zend_Search_Lucene::getSegmentFileName($generation));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   203
            $segmentsFile->writeInt((int)0xFFFFFFFD);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   204
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   205
            // write version (initialized by current time)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   206
            $segmentsFile->writeLong(round(microtime(true)));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   207
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   208
            // write name counter
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   209
            $segmentsFile->writeInt($nameCount);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   210
            // write segment counter
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   211
            $segmentsFile->writeInt(0);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   212
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   213
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   214
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   215
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   216
     * Open the index for writing
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   217
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   218
     * @param Zend_Search_Lucene_Storage_Directory $directory
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   219
     * @param array $segmentInfos
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   220
     * @param integer $targetFormatVersion
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   221
     * @param Zend_Search_Lucene_Storage_File $cleanUpLock
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   222
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   223
    public function __construct(Zend_Search_Lucene_Storage_Directory $directory, &$segmentInfos, $targetFormatVersion)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   224
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   225
        $this->_directory           = $directory;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   226
        $this->_segmentInfos        = &$segmentInfos;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   227
        $this->_targetFormatVersion = $targetFormatVersion;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   228
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   229
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   230
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   231
     * Adds a document to this index.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   232
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   233
     * @param Zend_Search_Lucene_Document $document
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   234
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   235
    public function addDocument(Zend_Search_Lucene_Document $document)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   236
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   237
        /** Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   238
        require_once 'Zend/Search/Lucene/Index/SegmentWriter/DocumentWriter.php';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   239
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   240
        if ($this->_currentSegment === null) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   241
            $this->_currentSegment =
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   242
                new Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter($this->_directory, $this->_newSegmentName());
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   243
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   244
        $this->_currentSegment->addDocument($document);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   245
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   246
        if ($this->_currentSegment->count() >= $this->maxBufferedDocs) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   247
            $this->commit();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   248
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   249
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   250
        $this->_maybeMergeSegments();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   251
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   252
        $this->_versionUpdate++;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   253
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   254
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   255
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   256
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   257
     * Check if we have anything to merge
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   258
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   259
     * @return boolean
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   260
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   261
    private function _hasAnythingToMerge()
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   262
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   263
        $segmentSizes = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   264
        foreach ($this->_segmentInfos as $segName => $segmentInfo) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   265
            $segmentSizes[$segName] = $segmentInfo->count();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   266
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   267
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   268
        $mergePool   = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   269
        $poolSize    = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   270
        $sizeToMerge = $this->maxBufferedDocs;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   271
        asort($segmentSizes, SORT_NUMERIC);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   272
        foreach ($segmentSizes as $segName => $size) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   273
            // Check, if segment comes into a new merging block
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   274
            while ($size >= $sizeToMerge) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   275
                // Merge previous block if it's large enough
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   276
                if ($poolSize >= $sizeToMerge) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   277
                    return true;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   278
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   279
                $mergePool   = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   280
                $poolSize    = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   281
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   282
                $sizeToMerge *= $this->mergeFactor;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   283
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   284
                if ($sizeToMerge > $this->maxMergeDocs) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   285
                    return false;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   286
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   287
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   288
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   289
            $mergePool[] = $this->_segmentInfos[$segName];
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   290
            $poolSize += $size;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   291
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   292
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   293
        if ($poolSize >= $sizeToMerge) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   294
            return true;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   295
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   296
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   297
        return false;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   298
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   299
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   300
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   301
     * Merge segments if necessary
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   302
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   303
    private function _maybeMergeSegments()
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   304
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   305
        if (Zend_Search_Lucene_LockManager::obtainOptimizationLock($this->_directory) === false) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   306
            return;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   307
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   308
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   309
        if (!$this->_hasAnythingToMerge()) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   310
            Zend_Search_Lucene_LockManager::releaseOptimizationLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   311
            return;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   312
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   313
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   314
        // Update segments list to be sure all segments are not merged yet by another process
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   315
        //
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   316
        // Segment merging functionality is concentrated in this class and surrounded
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   317
        // by optimization lock obtaining/releasing.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   318
        // _updateSegments() refreshes segments list from the latest index generation.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   319
        // So only new segments can be added to the index while we are merging some already existing
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   320
        // segments.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   321
        // Newly added segments will be also included into the index by the _updateSegments() call
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   322
        // either by another process or by the current process with the commit() call at the end of _mergeSegments() method.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   323
        // That's guaranteed by the serialisation of _updateSegments() execution using exclusive locks.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   324
        $this->_updateSegments();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   325
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   326
        // Perform standard auto-optimization procedure
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   327
        $segmentSizes = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   328
        foreach ($this->_segmentInfos as $segName => $segmentInfo) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   329
            $segmentSizes[$segName] = $segmentInfo->count();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   330
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   331
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   332
        $mergePool   = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   333
        $poolSize    = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   334
        $sizeToMerge = $this->maxBufferedDocs;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   335
        asort($segmentSizes, SORT_NUMERIC);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   336
        foreach ($segmentSizes as $segName => $size) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   337
            // Check, if segment comes into a new merging block
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   338
            while ($size >= $sizeToMerge) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   339
                // Merge previous block if it's large enough
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   340
                if ($poolSize >= $sizeToMerge) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   341
                    $this->_mergeSegments($mergePool);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   342
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   343
                $mergePool   = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   344
                $poolSize    = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   345
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   346
                $sizeToMerge *= $this->mergeFactor;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   347
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   348
                if ($sizeToMerge > $this->maxMergeDocs) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   349
                    Zend_Search_Lucene_LockManager::releaseOptimizationLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   350
                    return;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   351
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   352
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   353
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   354
            $mergePool[] = $this->_segmentInfos[$segName];
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   355
            $poolSize += $size;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   356
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   357
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   358
        if ($poolSize >= $sizeToMerge) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   359
            $this->_mergeSegments($mergePool);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   360
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   361
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   362
        Zend_Search_Lucene_LockManager::releaseOptimizationLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   363
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   364
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   365
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   366
     * Merge specified segments
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   367
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   368
     * $segments is an array of SegmentInfo objects
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   369
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   370
     * @param array $segments
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   371
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   372
    private function _mergeSegments($segments)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   373
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   374
        $newName = $this->_newSegmentName();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   375
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   376
        /** Zend_Search_Lucene_Index_SegmentMerger */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   377
        require_once 'Zend/Search/Lucene/Index/SegmentMerger.php';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   378
        $merger = new Zend_Search_Lucene_Index_SegmentMerger($this->_directory,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   379
                                                             $newName);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   380
        foreach ($segments as $segmentInfo) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   381
            $merger->addSource($segmentInfo);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   382
            $this->_segmentsToDelete[$segmentInfo->getName()] = $segmentInfo->getName();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   383
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   384
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   385
        $newSegment = $merger->merge();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   386
        if ($newSegment !== null) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   387
            $this->_newSegments[$newSegment->getName()] = $newSegment;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   388
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   389
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   390
        $this->commit();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   391
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   392
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   393
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   394
     * Update segments file by adding current segment to a list
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   395
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   396
     * @throws Zend_Search_Lucene_Exception
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   397
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   398
    private function _updateSegments()
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   399
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   400
        // Get an exclusive index lock
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   401
        Zend_Search_Lucene_LockManager::obtainWriteLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   402
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   403
        // Write down changes for the segments
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   404
        foreach ($this->_segmentInfos as $segInfo) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   405
            $segInfo->writeChanges();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   406
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   407
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   408
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   409
        $generation = Zend_Search_Lucene::getActualGeneration($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   410
        $segmentsFile   = $this->_directory->getFileObject(Zend_Search_Lucene::getSegmentFileName($generation), false);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   411
        $newSegmentFile = $this->_directory->createFile(Zend_Search_Lucene::getSegmentFileName(++$generation), false);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   412
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   413
        try {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   414
            $genFile = $this->_directory->getFileObject('segments.gen', false);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   415
        } catch (Zend_Search_Lucene_Exception $e) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   416
            if (strpos($e->getMessage(), 'is not readable') !== false) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   417
                $genFile = $this->_directory->createFile('segments.gen');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   418
            } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   419
                throw new Zend_Search_Lucene_Exception($e->getMessage(), $e->getCode(), $e);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   420
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   421
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   422
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   423
        $genFile->writeInt((int)0xFFFFFFFE);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   424
        // Write generation (first copy)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   425
        $genFile->writeLong($generation);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   426
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   427
        try {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   428
            // Write format marker
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   429
            if ($this->_targetFormatVersion == Zend_Search_Lucene::FORMAT_2_1) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   430
                $newSegmentFile->writeInt((int)0xFFFFFFFD);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   431
            } else if ($this->_targetFormatVersion == Zend_Search_Lucene::FORMAT_2_3) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   432
                $newSegmentFile->writeInt((int)0xFFFFFFFC);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   433
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   434
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   435
            // Read src file format identifier
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   436
            $format = $segmentsFile->readInt();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   437
            if ($format == (int)0xFFFFFFFF) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   438
                $srcFormat = Zend_Search_Lucene::FORMAT_PRE_2_1;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   439
            } else if ($format == (int)0xFFFFFFFD) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   440
                $srcFormat = Zend_Search_Lucene::FORMAT_2_1;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   441
            } else if ($format == (int)0xFFFFFFFC) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   442
                $srcFormat = Zend_Search_Lucene::FORMAT_2_3;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   443
            } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   444
                throw new Zend_Search_Lucene_Exception('Unsupported segments file format');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   445
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   446
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   447
            $version = $segmentsFile->readLong() + $this->_versionUpdate;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   448
            $this->_versionUpdate = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   449
            $newSegmentFile->writeLong($version);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   450
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   451
            // Write segment name counter
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   452
            $newSegmentFile->writeInt($segmentsFile->readInt());
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   453
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   454
            // Get number of segments offset
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   455
            $numOfSegmentsOffset = $newSegmentFile->tell();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   456
            // Write dummy data (segment counter)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   457
            $newSegmentFile->writeInt(0);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   458
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   459
            // Read number of segemnts
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   460
            $segmentsCount = $segmentsFile->readInt();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   461
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   462
            $segments = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   463
            for ($count = 0; $count < $segmentsCount; $count++) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   464
                $segName = $segmentsFile->readString();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   465
                $segSize = $segmentsFile->readInt();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   466
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   467
                if ($srcFormat == Zend_Search_Lucene::FORMAT_PRE_2_1) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   468
                    // pre-2.1 index format
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   469
                    $delGen            = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   470
                    $hasSingleNormFile = false;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   471
                    $numField          = (int)0xFFFFFFFF;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   472
                    $isCompoundByte    = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   473
                    $docStoreOptions   = null;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   474
                } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   475
                    $delGen = $segmentsFile->readLong();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   476
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   477
                    if ($srcFormat == Zend_Search_Lucene::FORMAT_2_3) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   478
                        $docStoreOffset = $segmentsFile->readInt();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   479
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   480
                        if ($docStoreOffset != (int)0xFFFFFFFF) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   481
                            $docStoreSegment        = $segmentsFile->readString();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   482
                            $docStoreIsCompoundFile = $segmentsFile->readByte();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   483
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   484
                            $docStoreOptions = array('offset'     => $docStoreOffset,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   485
                                                     'segment'    => $docStoreSegment,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   486
                                                     'isCompound' => ($docStoreIsCompoundFile == 1));
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   487
                        } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   488
                            $docStoreOptions = null;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   489
                        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   490
                    } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   491
                        $docStoreOptions = null;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   492
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   493
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   494
                    $hasSingleNormFile = $segmentsFile->readByte();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   495
                    $numField          = $segmentsFile->readInt();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   496
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   497
                    $normGens = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   498
                    if ($numField != (int)0xFFFFFFFF) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   499
                        for ($count1 = 0; $count1 < $numField; $count1++) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   500
                            $normGens[] = $segmentsFile->readLong();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   501
                        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   502
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   503
                    $isCompoundByte    = $segmentsFile->readByte();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   504
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   505
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   506
                if (!in_array($segName, $this->_segmentsToDelete)) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   507
                    // Load segment if necessary
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   508
                    if (!isset($this->_segmentInfos[$segName])) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   509
                        if ($isCompoundByte == 0xFF) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   510
                            // The segment is not a compound file
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   511
                            $isCompound = false;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   512
                        } else if ($isCompoundByte == 0x00) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   513
                            // The status is unknown
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   514
                            $isCompound = null;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   515
                        } else if ($isCompoundByte == 0x01) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   516
                            // The segment is a compound file
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   517
                            $isCompound = true;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   518
                        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   519
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   520
                        /** Zend_Search_Lucene_Index_SegmentInfo */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   521
                        require_once 'Zend/Search/Lucene/Index/SegmentInfo.php';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   522
                        $this->_segmentInfos[$segName] =
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   523
                                    new Zend_Search_Lucene_Index_SegmentInfo($this->_directory,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   524
                                                                             $segName,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   525
                                                                             $segSize,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   526
                                                                             $delGen,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   527
                                                                             $docStoreOptions,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   528
                                                                             $hasSingleNormFile,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   529
                                                                             $isCompound);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   530
                    } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   531
                        // Retrieve actual deletions file generation number
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   532
                        $delGen = $this->_segmentInfos[$segName]->getDelGen();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   533
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   534
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   535
                    $newSegmentFile->writeString($segName);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   536
                    $newSegmentFile->writeInt($segSize);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   537
                    $newSegmentFile->writeLong($delGen);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   538
                    if ($this->_targetFormatVersion == Zend_Search_Lucene::FORMAT_2_3) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   539
                        if ($docStoreOptions !== null) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   540
                            $newSegmentFile->writeInt($docStoreOffset);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   541
                            $newSegmentFile->writeString($docStoreSegment);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   542
                            $newSegmentFile->writeByte($docStoreIsCompoundFile);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   543
                        } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   544
                            // Set DocStoreOffset to -1
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   545
                            $newSegmentFile->writeInt((int)0xFFFFFFFF);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   546
                        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   547
                    } else if ($docStoreOptions !== null) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   548
                        // Release index write lock
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   549
                        Zend_Search_Lucene_LockManager::releaseWriteLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   550
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   551
                        throw new Zend_Search_Lucene_Exception('Index conversion to lower format version is not supported.');
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   552
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   553
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   554
                    $newSegmentFile->writeByte($hasSingleNormFile);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   555
                    $newSegmentFile->writeInt($numField);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   556
                    if ($numField != (int)0xFFFFFFFF) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   557
                        foreach ($normGens as $normGen) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   558
                            $newSegmentFile->writeLong($normGen);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   559
                        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   560
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   561
                    $newSegmentFile->writeByte($isCompoundByte);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   562
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   563
                    $segments[$segName] = $segSize;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   564
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   565
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   566
            $segmentsFile->close();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   567
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   568
            $segmentsCount = count($segments) + count($this->_newSegments);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   569
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   570
            foreach ($this->_newSegments as $segName => $segmentInfo) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   571
                $newSegmentFile->writeString($segName);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   572
                $newSegmentFile->writeInt($segmentInfo->count());
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   573
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   574
                // delete file generation: -1 (there is no delete file yet)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   575
                $newSegmentFile->writeInt((int)0xFFFFFFFF);$newSegmentFile->writeInt((int)0xFFFFFFFF);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   576
                if ($this->_targetFormatVersion == Zend_Search_Lucene::FORMAT_2_3) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   577
                    // docStoreOffset: -1 (segment doesn't use shared doc store)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   578
                    $newSegmentFile->writeInt((int)0xFFFFFFFF);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   579
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   580
                // HasSingleNormFile
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   581
                $newSegmentFile->writeByte($segmentInfo->hasSingleNormFile());
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   582
                // NumField
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   583
                $newSegmentFile->writeInt((int)0xFFFFFFFF);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   584
                // IsCompoundFile
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   585
                $newSegmentFile->writeByte($segmentInfo->isCompound() ? 1 : -1);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   586
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   587
                $segments[$segmentInfo->getName()] = $segmentInfo->count();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   588
                $this->_segmentInfos[$segName] = $segmentInfo;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   589
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   590
            $this->_newSegments = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   591
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   592
            $newSegmentFile->seek($numOfSegmentsOffset);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   593
            $newSegmentFile->writeInt($segmentsCount);  // Update segments count
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   594
            $newSegmentFile->close();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   595
        } catch (Exception $e) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   596
            /** Restore previous index generation */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   597
            $generation--;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   598
            $genFile->seek(4, SEEK_SET);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   599
            // Write generation number twice
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   600
            $genFile->writeLong($generation); $genFile->writeLong($generation);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   601
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   602
            // Release index write lock
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   603
            Zend_Search_Lucene_LockManager::releaseWriteLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   604
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   605
            // Throw the exception
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   606
            require_once 'Zend/Search/Lucene/Exception.php';
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   607
            throw new Zend_Search_Lucene_Exception($e->getMessage(), $e->getCode(), $e);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   608
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   609
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   610
        // Write generation (second copy)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   611
        $genFile->writeLong($generation);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   612
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   613
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   614
        // Check if another update or read process is not running now
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   615
        // If yes, skip clean-up procedure
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   616
        if (Zend_Search_Lucene_LockManager::escalateReadLock($this->_directory)) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   617
            /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   618
             * Clean-up directory
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   619
             */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   620
            $filesToDelete = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   621
            $filesTypes    = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   622
            $filesNumbers  = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   623
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   624
            // list of .del files of currently used segments
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   625
            // each segment can have several generations of .del files
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   626
            // only last should not be deleted
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   627
            $delFiles = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   628
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   629
            foreach ($this->_directory->fileList() as $file) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   630
                if ($file == 'deletable') {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   631
                    // 'deletable' file
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   632
                    $filesToDelete[] = $file;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   633
                    $filesTypes[]    = 0; // delete this file first, since it's not used starting from Lucene v2.1
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   634
                    $filesNumbers[]  = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   635
                } else if ($file == 'segments') {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   636
                    // 'segments' file
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   637
                    $filesToDelete[] = $file;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   638
                    $filesTypes[]    = 1; // second file to be deleted "zero" version of segments file (Lucene pre-2.1)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   639
                    $filesNumbers[]  = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   640
                } else if (preg_match('/^segments_[a-zA-Z0-9]+$/i', $file)) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   641
                    // 'segments_xxx' file
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   642
                    // Check if it's not a just created generation file
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   643
                    if ($file != Zend_Search_Lucene::getSegmentFileName($generation)) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   644
                        $filesToDelete[] = $file;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   645
                        $filesTypes[]    = 2; // first group of files for deletions
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   646
                        $filesNumbers[]  = (int)base_convert(substr($file, 9), 36, 10); // ordered by segment generation numbers
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   647
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   648
                } else if (preg_match('/(^_([a-zA-Z0-9]+))\.f\d+$/i', $file, $matches)) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   649
                    // one of per segment files ('<segment_name>.f<decimal_number>')
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   650
                    // Check if it's not one of the segments in the current segments set
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   651
                    if (!isset($segments[$matches[1]])) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   652
                        $filesToDelete[] = $file;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   653
                        $filesTypes[]    = 3; // second group of files for deletions
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   654
                        $filesNumbers[]  = (int)base_convert($matches[2], 36, 10); // order by segment number
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   655
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   656
                } else if (preg_match('/(^_([a-zA-Z0-9]+))(_([a-zA-Z0-9]+))\.del$/i', $file, $matches)) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   657
                    // one of per segment files ('<segment_name>_<del_generation>.del' where <segment_name> is '_<segment_number>')
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   658
                    // Check if it's not one of the segments in the current segments set
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   659
                    if (!isset($segments[$matches[1]])) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   660
                        $filesToDelete[] = $file;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   661
                        $filesTypes[]    = 3; // second group of files for deletions
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   662
                        $filesNumbers[]  = (int)base_convert($matches[2], 36, 10); // order by segment number
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   663
                    } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   664
                        $segmentNumber = (int)base_convert($matches[2], 36, 10);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   665
                        $delGeneration = (int)base_convert($matches[4], 36, 10);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   666
                        if (!isset($delFiles[$segmentNumber])) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   667
                            $delFiles[$segmentNumber] = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   668
                        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   669
                        $delFiles[$segmentNumber][$delGeneration] = $file;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   670
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   671
                } else if (isset(self::$_indexExtensions[substr($file, strlen($file)-4)])) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   672
                    // one of per segment files ('<segment_name>.<ext>')
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   673
                    $segmentName = substr($file, 0, strlen($file) - 4);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   674
                    // Check if it's not one of the segments in the current segments set
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   675
                    if (!isset($segments[$segmentName])  &&
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   676
                        ($this->_currentSegment === null  ||  $this->_currentSegment->getName() != $segmentName)) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   677
                        $filesToDelete[] = $file;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   678
                        $filesTypes[]    = 3; // second group of files for deletions
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   679
                        $filesNumbers[]  = (int)base_convert(substr($file, 1 /* skip '_' */, strlen($file)-5), 36, 10); // order by segment number
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   680
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   681
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   682
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   683
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   684
            $maxGenNumber = 0;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   685
            // process .del files of currently used segments
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   686
            foreach ($delFiles as $segmentNumber => $segmentDelFiles) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   687
                ksort($delFiles[$segmentNumber], SORT_NUMERIC);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   688
                array_pop($delFiles[$segmentNumber]); // remove last delete file generation from candidates for deleting
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   689
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   690
                end($delFiles[$segmentNumber]);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   691
                $lastGenNumber = key($delFiles[$segmentNumber]);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   692
                if ($lastGenNumber > $maxGenNumber) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   693
                    $maxGenNumber = $lastGenNumber;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   694
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   695
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   696
            foreach ($delFiles as $segmentNumber => $segmentDelFiles) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   697
                foreach ($segmentDelFiles as $delGeneration => $file) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   698
                        $filesToDelete[] = $file;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   699
                        $filesTypes[]    = 4; // third group of files for deletions
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   700
                        $filesNumbers[]  = $segmentNumber*$maxGenNumber + $delGeneration; // order by <segment_number>,<del_generation> pair
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   701
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   702
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   703
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   704
            // Reorder files for deleting
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   705
            array_multisort($filesTypes,    SORT_ASC, SORT_NUMERIC,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   706
                            $filesNumbers,  SORT_ASC, SORT_NUMERIC,
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   707
                            $filesToDelete, SORT_ASC, SORT_STRING);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   708
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   709
            foreach ($filesToDelete as $file) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   710
                try {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   711
                    /** Skip shared docstore segments deleting */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   712
                    /** @todo Process '.cfx' files to check if them are already unused */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   713
                    if (substr($file, strlen($file)-4) != '.cfx') {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   714
                        $this->_directory->deleteFile($file);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   715
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   716
                } catch (Zend_Search_Lucene_Exception $e) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   717
                    if (strpos($e->getMessage(), 'Can\'t delete file') === false) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   718
                        // That's not "file is under processing or already deleted" exception
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   719
                        // Pass it through
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   720
                        throw new Zend_Search_Lucene_Exception($e->getMessage(), $e->getCode(), $e);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   721
                    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   722
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   723
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   724
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   725
            // Return read lock into the previous state
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   726
            Zend_Search_Lucene_LockManager::deEscalateReadLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   727
        } else {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   728
            // Only release resources if another index reader is running now
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   729
            foreach ($this->_segmentsToDelete as $segName) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   730
                foreach (self::$_indexExtensions as $ext) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   731
                    $this->_directory->purgeFile($segName . $ext);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   732
                }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   733
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   734
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   735
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   736
        // Clean-up _segmentsToDelete container
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   737
        $this->_segmentsToDelete = array();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   738
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   739
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   740
        // Release index write lock
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   741
        Zend_Search_Lucene_LockManager::releaseWriteLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   742
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   743
        // Remove unused segments from segments list
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   744
        foreach ($this->_segmentInfos as $segName => $segmentInfo) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   745
            if (!isset($segments[$segName])) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   746
                unset($this->_segmentInfos[$segName]);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   747
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   748
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   749
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   750
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   751
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   752
     * Commit current changes
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   753
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   754
    public function commit()
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   755
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   756
        if ($this->_currentSegment !== null) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   757
            $newSegment = $this->_currentSegment->close();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   758
            if ($newSegment !== null) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   759
                $this->_newSegments[$newSegment->getName()] = $newSegment;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   760
            }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   761
            $this->_currentSegment = null;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   762
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   763
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   764
        $this->_updateSegments();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   765
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   766
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   767
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   768
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   769
     * Merges the provided indexes into this index.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   770
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   771
     * @param array $readers
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   772
     * @return void
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   773
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   774
    public function addIndexes($readers)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   775
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   776
        /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   777
         * @todo implementation
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   778
         */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   779
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   780
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   781
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   782
     * Merges all segments together into new one
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   783
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   784
     * Returns true on success and false if another optimization or auto-optimization process
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   785
     * is running now
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   786
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   787
     * @return boolean
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   788
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   789
    public function optimize()
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   790
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   791
        if (Zend_Search_Lucene_LockManager::obtainOptimizationLock($this->_directory) === false) {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   792
            return false;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   793
        }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   794
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   795
        // Update segments list to be sure all segments are not merged yet by another process
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   796
        //
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   797
        // Segment merging functionality is concentrated in this class and surrounded
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   798
        // by optimization lock obtaining/releasing.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   799
        // _updateSegments() refreshes segments list from the latest index generation.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   800
        // So only new segments can be added to the index while we are merging some already existing
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   801
        // segments.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   802
        // Newly added segments will be also included into the index by the _updateSegments() call
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   803
        // either by another process or by the current process with the commit() call at the end of _mergeSegments() method.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   804
        // That's guaranteed by the serialisation of _updateSegments() execution using exclusive locks.
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   805
        $this->_updateSegments();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   806
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   807
        $this->_mergeSegments($this->_segmentInfos);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   808
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   809
        Zend_Search_Lucene_LockManager::releaseOptimizationLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   810
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   811
        return true;
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   812
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   813
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   814
    /**
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   815
     * Get name for new segment
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   816
     *
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   817
     * @return string
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   818
     */
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   819
    private function _newSegmentName()
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   820
    {
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   821
        Zend_Search_Lucene_LockManager::obtainWriteLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   822
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   823
        $generation = Zend_Search_Lucene::getActualGeneration($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   824
        $segmentsFile = $this->_directory->getFileObject(Zend_Search_Lucene::getSegmentFileName($generation), false);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   825
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   826
        $segmentsFile->seek(12); // 12 = 4 (int, file format marker) + 8 (long, index version)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   827
        $segmentNameCounter = $segmentsFile->readInt();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   828
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   829
        $segmentsFile->seek(12); // 12 = 4 (int, file format marker) + 8 (long, index version)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   830
        $segmentsFile->writeInt($segmentNameCounter + 1);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   831
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   832
        // Flash output to guarantee that wrong value will not be loaded between unlock and
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   833
        // return (which calls $segmentsFile destructor)
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   834
        $segmentsFile->flush();
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   835
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   836
        Zend_Search_Lucene_LockManager::releaseWriteLock($this->_directory);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   837
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   838
        return '_' . base_convert($segmentNameCounter, 10, 36);
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   839
    }
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   840
4eba9c11703f first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff changeset
   841
}