| 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/SegmentInfo.php@4eba9c11703f |
| parent 64 | web/Zend/Search/Lucene/Index/SegmentInfo.php@162c1de6545a |
| child 207 | 621fa6caec0c |
| permissions | -rw-r--r-- |
|
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: SegmentInfo.php 22987 2010-09-21 10:39:53Z alexander $ |
|
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 |
/** Zend_Search_Lucene_Index_TermsStream_Interface */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
24 |
require_once 'Zend/Search/Lucene/Index/TermsStream/Interface.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
25 |
|
|
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 |
/** Zend_Search_Lucene_Search_Similarity */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
28 |
require_once 'Zend/Search/Lucene/Search/Similarity.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
29 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
30 |
/** Zend_Search_Lucene_Index_FieldInfo */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
31 |
require_once 'Zend/Search/Lucene/Index/FieldInfo.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
32 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
33 |
/** Zend_Search_Lucene_Index_Term */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
34 |
require_once 'Zend/Search/Lucene/Index/Term.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
35 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
36 |
/** Zend_Search_Lucene_Index_TermInfo */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
37 |
require_once 'Zend/Search/Lucene/Index/TermInfo.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
38 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
39 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
40 |
* @category Zend |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
41 |
* @package Zend_Search_Lucene |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
42 |
* @subpackage Index |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
43 |
* @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
|
44 |
* @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
|
45 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
46 |
class Zend_Search_Lucene_Index_SegmentInfo implements Zend_Search_Lucene_Index_TermsStream_Interface |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
49 |
* "Full scan vs fetch" boundary. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
50 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
51 |
* If filter selectivity is less than this value, then full scan is performed |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
52 |
* (since term entries fetching has some additional overhead). |
|
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 |
const FULL_SCAN_VS_FETCH_BOUNDARY = 5; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
55 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
56 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
57 |
* Number of docs in a segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
58 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
59 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
60 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
61 |
private $_docCount; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
62 |
|
|
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 |
* Segment name |
|
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 |
* @var string |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
67 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
68 |
private $_name; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
69 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
70 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
71 |
* Term Dictionary Index |
|
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 |
* Array of arrays (Zend_Search_Lucene_Index_Term objects are represented as arrays because |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
74 |
* of performance considerations) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
75 |
* [0] -> $termValue |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
76 |
* [1] -> $termFieldNum |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
77 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
78 |
* Corresponding Zend_Search_Lucene_Index_TermInfo object stored in the $_termDictionaryInfos |
|
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 |
* @var array |
|
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 |
private $_termDictionary; |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
85 |
* Term Dictionary Index TermInfos |
|
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 |
* Array of arrays (Zend_Search_Lucene_Index_TermInfo objects are represented as arrays because |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
88 |
* of performance considerations) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
89 |
* [0] -> $docFreq |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
90 |
* [1] -> $freqPointer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
91 |
* [2] -> $proxPointer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
92 |
* [3] -> $skipOffset |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
93 |
* [4] -> $indexPointer |
|
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 |
* @var array |
|
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 |
private $_termDictionaryInfos; |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
100 |
* Segment fields. Array of Zend_Search_Lucene_Index_FieldInfo objects for this segment |
|
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 |
* @var array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
103 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
104 |
private $_fields; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
105 |
|
|
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 |
* Field positions in a dictionary. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
108 |
* (Term dictionary contains filelds ordered by names) |
|
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 |
* @var array |
|
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 |
private $_fieldsDicPositions; |
|
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 |
|
|
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 |
* Associative array where the key is the file name and the value is data offset |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
117 |
* in a compound segment file (.csf). |
|
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 array |
|
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 $_segFiles; |
|
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 |
* Associative array where the key is the file name and the value is file size (.csf). |
|
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 |
* @var 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 |
private $_segFileSizes; |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
131 |
* Delete file generation number |
|
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 |
* -2 means autodetect latest delete generation |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
134 |
* -1 means 'there is no delete file' |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
135 |
* 0 means pre-2.1 format delete file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
136 |
* X specifies used delete file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
137 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
138 |
* @var integer |
|
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 |
private $_delGen; |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
143 |
* Segment has single norms file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
144 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
145 |
* If true then one .nrm file is used for all fields |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
146 |
* Otherwise .fN files are used |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
147 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
148 |
* @var boolean |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
149 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
150 |
private $_hasSingleNormFile; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
151 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
152 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
153 |
* Use compound segment file (*.cfs) to collect all other segment files |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
154 |
* (excluding .del files) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
155 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
156 |
* @var boolean |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
157 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
158 |
private $_isCompound; |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
162 |
* File system adapter. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
163 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
164 |
* @var Zend_Search_Lucene_Storage_Directory_Filesystem |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
165 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
166 |
private $_directory; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
167 |
|
|
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 |
* Normalization factors. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
170 |
* An array fieldName => normVector |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
171 |
* normVector is a binary string. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
172 |
* Each byte corresponds to an indexed document in a segment and |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
173 |
* encodes normalization factor (float value, encoded by |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
174 |
* Zend_Search_Lucene_Search_Similarity::encodeNorm()) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
175 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
176 |
* @var array |
|
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 |
private $_norms = array(); |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
181 |
* List of deleted documents. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
182 |
* bitset if bitset extension is loaded or array otherwise. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
183 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
184 |
* @var mixed |
|
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 |
private $_deleted = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
187 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
188 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
189 |
* $this->_deleted update flag |
|
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 |
* @var boolean |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
192 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
193 |
private $_deletedDirty = false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
194 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
195 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
196 |
* True if segment uses shared doc store |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
197 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
198 |
* @var boolean |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
199 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
200 |
private $_usesSharedDocStore; |
|
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 |
/* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
203 |
* Shared doc store options. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
204 |
* It's an assotiative array with the following items: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
205 |
* - 'offset' => $docStoreOffset The starting document in the shared doc store files where this segment's documents begin |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
206 |
* - 'segment' => $docStoreSegment The name of the segment that has the shared doc store files. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
207 |
* - 'isCompound' => $docStoreIsCompoundFile True, if compound file format is used for the shared doc store files (.cfx file). |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
208 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
209 |
private $_sharedDocStoreOptions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
210 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
211 |
|
|
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 |
* Zend_Search_Lucene_Index_SegmentInfo constructor |
|
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 |
* @param Zend_Search_Lucene_Storage_Directory $directory |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
216 |
* @param string $name |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
217 |
* @param integer $docCount |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
218 |
* @param integer $delGen |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
219 |
* @param array|null $docStoreOptions |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
220 |
* @param boolean $hasSingleNormFile |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
221 |
* @param boolean $isCompound |
|
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, $name, $docCount, $delGen = 0, $docStoreOptions = null, $hasSingleNormFile = false, $isCompound = null) |
|
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->_name = $name; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
227 |
$this->_docCount = $docCount; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
228 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
229 |
if ($docStoreOptions !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
230 |
$this->_usesSharedDocStore = true; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
231 |
$this->_sharedDocStoreOptions = $docStoreOptions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
232 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
233 |
if ($docStoreOptions['isCompound']) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
234 |
$cfxFile = $this->_directory->getFileObject($docStoreOptions['segment'] . '.cfx'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
235 |
$cfxFilesCount = $cfxFile->readVInt(); |
|
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 |
$cfxFiles = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
238 |
$cfxFileSizes = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
239 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
240 |
for ($count = 0; $count < $cfxFilesCount; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
241 |
$dataOffset = $cfxFile->readLong(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
242 |
if ($count != 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
243 |
$cfxFileSizes[$fileName] = $dataOffset - end($cfxFiles); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
244 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
245 |
$fileName = $cfxFile->readString(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
246 |
$cfxFiles[$fileName] = $dataOffset; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
247 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
248 |
if ($count != 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
249 |
$cfxFileSizes[$fileName] = $this->_directory->fileLength($docStoreOptions['segment'] . '.cfx') - $dataOffset; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
250 |
} |
|
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->_sharedDocStoreOptions['files'] = $cfxFiles; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
253 |
$this->_sharedDocStoreOptions['fileSizes'] = $cfxFileSizes; |
|
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 |
$this->_hasSingleNormFile = $hasSingleNormFile; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
258 |
$this->_delGen = $delGen; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
259 |
$this->_termDictionary = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
260 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
261 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
262 |
if ($isCompound !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
263 |
$this->_isCompound = $isCompound; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
264 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
265 |
// It's a pre-2.1 segment or isCompound is set to 'unknown' |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
266 |
// Detect if segment uses compound file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
267 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
268 |
try { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
269 |
// Try to open compound file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
270 |
$this->_directory->getFileObject($name . '.cfs'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
271 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
272 |
// Compound file is found |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
273 |
$this->_isCompound = true; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
274 |
} catch (Zend_Search_Lucene_Exception $e) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
275 |
if (strpos($e->getMessage(), 'is not readable') !== false) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
276 |
// Compound file is not found or is not readable |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
277 |
$this->_isCompound = false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
278 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
279 |
throw new Zend_Search_Lucene_Exception($e->getMessage(), $e->getCode(), $e); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
280 |
} |
|
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 |
} |
|
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 |
$this->_segFiles = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
285 |
if ($this->_isCompound) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
286 |
$cfsFile = $this->_directory->getFileObject($name . '.cfs'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
287 |
$segFilesCount = $cfsFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
288 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
289 |
for ($count = 0; $count < $segFilesCount; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
290 |
$dataOffset = $cfsFile->readLong(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
291 |
if ($count != 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
292 |
$this->_segFileSizes[$fileName] = $dataOffset - end($this->_segFiles); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
293 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
294 |
$fileName = $cfsFile->readString(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
295 |
$this->_segFiles[$fileName] = $dataOffset; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
296 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
297 |
if ($count != 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
298 |
$this->_segFileSizes[$fileName] = $this->_directory->fileLength($name . '.cfs') - $dataOffset; |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
302 |
$fnmFile = $this->openCompoundFile('.fnm'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
303 |
$fieldsCount = $fnmFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
304 |
$fieldNames = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
305 |
$fieldNums = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
306 |
$this->_fields = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
307 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
308 |
for ($count=0; $count < $fieldsCount; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
309 |
$fieldName = $fnmFile->readString(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
310 |
$fieldBits = $fnmFile->readByte(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
311 |
$this->_fields[$count] = new Zend_Search_Lucene_Index_FieldInfo($fieldName, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
312 |
$fieldBits & 0x01 /* field is indexed */, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
313 |
$count, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
314 |
$fieldBits & 0x02 /* termvectors are stored */, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
315 |
$fieldBits & 0x10 /* norms are omitted */, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
316 |
$fieldBits & 0x20 /* payloads are stored */); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
317 |
if ($fieldBits & 0x10) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
318 |
// norms are omitted for the indexed field |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
319 |
$this->_norms[$count] = str_repeat(chr(Zend_Search_Lucene_Search_Similarity::encodeNorm(1.0)), $docCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
320 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
321 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
322 |
$fieldNums[$count] = $count; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
323 |
$fieldNames[$count] = $fieldName; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
324 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
325 |
array_multisort($fieldNames, SORT_ASC, SORT_REGULAR, $fieldNums); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
326 |
$this->_fieldsDicPositions = array_flip($fieldNums); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
327 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
328 |
if ($this->_delGen == -2) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
329 |
// SegmentInfo constructor is invoked from index writer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
330 |
// Autodetect current delete file generation number |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
331 |
$this->_delGen = $this->_detectLatestDelGen(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
332 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
333 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
334 |
// Load deletions |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
335 |
$this->_deleted = $this->_loadDelFile(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
336 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
337 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
338 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
339 |
* Load detetions file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
340 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
341 |
* Returns bitset or an array depending on bitset extension availability |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
342 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
343 |
* @return mixed |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
344 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
345 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
346 |
private function _loadDelFile() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
347 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
348 |
if ($this->_delGen == -1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
349 |
// There is no delete file for this segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
350 |
return null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
351 |
} else if ($this->_delGen == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
352 |
// It's a segment with pre-2.1 format delete file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
353 |
// Try to load deletions file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
354 |
return $this->_loadPre21DelFile(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
355 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
356 |
// It's 2.1+ format deleteions file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
357 |
return $this->_load21DelFile(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
358 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
359 |
} |
|
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 |
* Load pre-2.1 detetions file |
|
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 |
* Returns bitset or an array depending on bitset extension availability |
|
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 |
* @return mixed |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
367 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
368 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
369 |
private function _loadPre21DelFile() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
370 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
371 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
372 |
try { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
373 |
// '.del' files always stored in a separate file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
374 |
// Segment compound is not used |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
375 |
$delFile = $this->_directory->getFileObject($this->_name . '.del'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
376 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
377 |
$byteCount = $delFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
378 |
$byteCount = ceil($byteCount/8); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
379 |
$bitCount = $delFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
380 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
381 |
if ($bitCount == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
382 |
$delBytes = ''; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
383 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
384 |
$delBytes = $delFile->readBytes($byteCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
385 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
386 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
387 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
388 |
return $delBytes; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
389 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
390 |
$deletions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
391 |
for ($count = 0; $count < $byteCount; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
392 |
$byte = ord($delBytes[$count]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
393 |
for ($bit = 0; $bit < 8; $bit++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
394 |
if ($byte & (1<<$bit)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
395 |
$deletions[$count*8 + $bit] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
396 |
} |
|
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 |
} |
|
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 |
return $deletions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
401 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
402 |
} catch(Zend_Search_Lucene_Exception $e) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
403 |
if (strpos($e->getMessage(), 'is not readable') === false) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
404 |
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
|
405 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
406 |
// There is no deletion file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
407 |
$this->_delGen = -1; |
|
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 |
return null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
410 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
411 |
} |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
414 |
* Load 2.1+ format detetions file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
415 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
416 |
* Returns bitset or an array depending on bitset extension availability |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
417 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
418 |
* @return mixed |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
419 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
420 |
private function _load21DelFile() |
|
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 |
$delFile = $this->_directory->getFileObject($this->_name . '_' . base_convert($this->_delGen, 10, 36) . '.del'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
423 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
424 |
$format = $delFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
425 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
426 |
if ($format == (int)0xFFFFFFFF) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
427 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
428 |
$deletions = bitset_empty(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
429 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
430 |
$deletions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
431 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
432 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
433 |
$byteCount = $delFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
434 |
$bitCount = $delFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
435 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
436 |
$delFileSize = $this->_directory->fileLength($this->_name . '_' . base_convert($this->_delGen, 10, 36) . '.del'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
437 |
$byteNum = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
438 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
439 |
do { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
440 |
$dgap = $delFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
441 |
$nonZeroByte = $delFile->readByte(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
442 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
443 |
$byteNum += $dgap; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
444 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
445 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
446 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
447 |
for ($bit = 0; $bit < 8; $bit++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
448 |
if ($nonZeroByte & (1<<$bit)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
449 |
bitset_incl($deletions, $byteNum*8 + $bit); |
|
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 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
452 |
return $deletions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
453 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
454 |
for ($bit = 0; $bit < 8; $bit++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
455 |
if ($nonZeroByte & (1<<$bit)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
456 |
$deletions[$byteNum*8 + $bit] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
457 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
458 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
459 |
return (count($deletions) > 0) ? $deletions : null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
460 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
461 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
462 |
} while ($delFile->tell() < $delFileSize); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
463 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
464 |
// $format is actually byte count |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
465 |
$byteCount = ceil($format/8); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
466 |
$bitCount = $delFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
467 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
468 |
if ($bitCount == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
469 |
$delBytes = ''; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
470 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
471 |
$delBytes = $delFile->readBytes($byteCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
472 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
473 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
474 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
475 |
return $delBytes; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
476 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
477 |
$deletions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
478 |
for ($count = 0; $count < $byteCount; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
479 |
$byte = ord($delBytes[$count]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
480 |
for ($bit = 0; $bit < 8; $bit++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
481 |
if ($byte & (1<<$bit)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
482 |
$deletions[$count*8 + $bit] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
483 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
484 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
485 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
486 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
487 |
return (count($deletions) > 0) ? $deletions : null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
488 |
} |
|
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 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
491 |
|
|
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 |
* Opens index file stoted within compound index file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
494 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
495 |
* @param string $extension |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
496 |
* @param boolean $shareHandler |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
497 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
498 |
* @return Zend_Search_Lucene_Storage_File |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
499 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
500 |
public function openCompoundFile($extension, $shareHandler = true) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
501 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
502 |
if (($extension == '.fdx' || $extension == '.fdt') && $this->_usesSharedDocStore) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
503 |
$fdxFName = $this->_sharedDocStoreOptions['segment'] . '.fdx'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
504 |
$fdtFName = $this->_sharedDocStoreOptions['segment'] . '.fdt'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
505 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
506 |
if (!$this->_sharedDocStoreOptions['isCompound']) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
507 |
$fdxFile = $this->_directory->getFileObject($fdxFName, $shareHandler); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
508 |
$fdxFile->seek($this->_sharedDocStoreOptions['offset']*8, SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
509 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
510 |
if ($extension == '.fdx') { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
511 |
// '.fdx' file is requested |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
512 |
return $fdxFile; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
513 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
514 |
// '.fdt' file is requested |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
515 |
$fdtStartOffset = $fdxFile->readLong(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
516 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
517 |
$fdtFile = $this->_directory->getFileObject($fdtFName, $shareHandler); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
518 |
$fdtFile->seek($fdtStartOffset, SEEK_CUR); |
|
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 |
return $fdtFile; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
521 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
522 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
523 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
524 |
if( !isset($this->_sharedDocStoreOptions['files'][$fdxFName]) ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
525 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
526 |
throw new Zend_Search_Lucene_Exception('Shared doc storage segment compound file doesn\'t contain ' |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
527 |
. $fdxFName . ' file.' ); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
528 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
529 |
if( !isset($this->_sharedDocStoreOptions['files'][$fdtFName]) ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
530 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
531 |
throw new Zend_Search_Lucene_Exception('Shared doc storage segment compound file doesn\'t contain ' |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
532 |
. $fdtFName . ' file.' ); |
|
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 |
// Open shared docstore segment file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
536 |
$cfxFile = $this->_directory->getFileObject($this->_sharedDocStoreOptions['segment'] . '.cfx', $shareHandler); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
537 |
// Seek to the start of '.fdx' file within compound file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
538 |
$cfxFile->seek($this->_sharedDocStoreOptions['files'][$fdxFName]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
539 |
// Seek to the start of current segment documents section |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
540 |
$cfxFile->seek($this->_sharedDocStoreOptions['offset']*8, SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
541 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
542 |
if ($extension == '.fdx') { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
543 |
// '.fdx' file is requested |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
544 |
return $cfxFile; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
545 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
546 |
// '.fdt' file is requested |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
547 |
$fdtStartOffset = $cfxFile->readLong(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
548 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
549 |
// Seek to the start of '.fdt' file within compound file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
550 |
$cfxFile->seek($this->_sharedDocStoreOptions['files'][$fdtFName]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
551 |
// Seek to the start of current segment documents section |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
552 |
$cfxFile->seek($fdtStartOffset, SEEK_CUR); |
|
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 |
return $fdtFile; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
555 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
556 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
557 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
558 |
$filename = $this->_name . $extension; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
559 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
560 |
if (!$this->_isCompound) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
561 |
return $this->_directory->getFileObject($filename, $shareHandler); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
562 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
563 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
564 |
if( !isset($this->_segFiles[$filename]) ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
565 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
566 |
throw new Zend_Search_Lucene_Exception('Segment compound file doesn\'t contain ' |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
567 |
. $filename . ' file.' ); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
568 |
} |
|
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 |
$file = $this->_directory->getFileObject($this->_name . '.cfs', $shareHandler); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
571 |
$file->seek($this->_segFiles[$filename]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
572 |
return $file; |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
575 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
576 |
* Get compound file length |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
577 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
578 |
* @param string $extension |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
579 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
580 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
581 |
public function compoundFileLength($extension) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
582 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
583 |
if (($extension == '.fdx' || $extension == '.fdt') && $this->_usesSharedDocStore) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
584 |
$filename = $this->_sharedDocStoreOptions['segment'] . $extension; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
585 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
586 |
if (!$this->_sharedDocStoreOptions['isCompound']) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
587 |
return $this->_directory->fileLength($filename); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
588 |
} |
|
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 |
if( !isset($this->_sharedDocStoreOptions['fileSizes'][$filename]) ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
591 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
592 |
throw new Zend_Search_Lucene_Exception('Shared doc store compound file doesn\'t contain ' |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
593 |
. $filename . ' file.' ); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
594 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
595 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
596 |
return $this->_sharedDocStoreOptions['fileSizes'][$filename]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
597 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
598 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
599 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
600 |
$filename = $this->_name . $extension; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
601 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
602 |
// Try to get common file first |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
603 |
if ($this->_directory->fileExists($filename)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
604 |
return $this->_directory->fileLength($filename); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
605 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
606 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
607 |
if( !isset($this->_segFileSizes[$filename]) ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
608 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
609 |
throw new Zend_Search_Lucene_Exception('Index compound file doesn\'t contain ' |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
610 |
. $filename . ' file.' ); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
611 |
} |
|
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 |
return $this->_segFileSizes[$filename]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
614 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
615 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
616 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
617 |
* Returns field index or -1 if field is not found |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
618 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
619 |
* @param string $fieldName |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
620 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
621 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
622 |
public function getFieldNum($fieldName) |
|
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 |
foreach( $this->_fields as $field ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
625 |
if( $field->name == $fieldName ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
626 |
return $field->number; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
627 |
} |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
630 |
return -1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
631 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
632 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
633 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
634 |
* Returns field info for specified field |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
635 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
636 |
* @param integer $fieldNum |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
637 |
* @return Zend_Search_Lucene_Index_FieldInfo |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
638 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
639 |
public function getField($fieldNum) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
640 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
641 |
return $this->_fields[$fieldNum]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
642 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
643 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
644 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
645 |
* Returns array of fields. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
646 |
* if $indexed parameter is true, then returns only indexed fields. |
|
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 |
* @param boolean $indexed |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
649 |
* @return array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
650 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
651 |
public function getFields($indexed = false) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
652 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
653 |
$result = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
654 |
foreach( $this->_fields as $field ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
655 |
if( (!$indexed) || $field->isIndexed ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
656 |
$result[ $field->name ] = $field->name; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
657 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
658 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
659 |
return $result; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
660 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
661 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
662 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
663 |
* Returns array of FieldInfo objects. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
664 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
665 |
* @return array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
666 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
667 |
public function getFieldInfos() |
|
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 |
return $this->_fields; |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
672 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
673 |
* Returns actual deletions file generation number. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
674 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
675 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
676 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
677 |
public function getDelGen() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
678 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
679 |
return $this->_delGen; |
|
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 |
* Returns the total number of documents in this segment (including deleted documents). |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
684 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
685 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
686 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
687 |
public function count() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
688 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
689 |
return $this->_docCount; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
690 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
691 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
692 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
693 |
* Returns number of deleted documents. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
694 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
695 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
696 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
697 |
private function _deletedCount() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
698 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
699 |
if ($this->_deleted === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
700 |
return 0; |
|
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 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
704 |
return count(bitset_to_array($this->_deleted)); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
705 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
706 |
return count($this->_deleted); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
707 |
} |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
710 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
711 |
* Returns the total number of non-deleted documents in this segment. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
712 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
713 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
714 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
715 |
public function numDocs() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
716 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
717 |
if ($this->hasDeletions()) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
718 |
return $this->_docCount - $this->_deletedCount(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
719 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
720 |
return $this->_docCount; |
|
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 |
* Get field position in a fields dictionary |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
726 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
727 |
* @param integer $fieldNum |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
728 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
729 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
730 |
private function _getFieldPosition($fieldNum) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
731 |
// Treat values which are not in a translation table as a 'direct value' |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
732 |
return isset($this->_fieldsDicPositions[$fieldNum]) ? |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
733 |
$this->_fieldsDicPositions[$fieldNum] : $fieldNum; |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
737 |
* Return segment name |
|
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 |
* @return string |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
740 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
741 |
public function getName() |
|
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 |
return $this->_name; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
744 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
745 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
746 |
|
|
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 |
* TermInfo cache |
|
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 |
* Size is 1024. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
751 |
* Numbers are used instead of class constants because of performance considerations |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
752 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
753 |
* @var array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
754 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
755 |
private $_termInfoCache = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
756 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
757 |
private function _cleanUpTermInfoCache() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
758 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
759 |
// Clean 256 term infos |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
760 |
foreach ($this->_termInfoCache as $key => $termInfo) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
761 |
unset($this->_termInfoCache[$key]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
762 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
763 |
// leave 768 last used term infos |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
764 |
if (count($this->_termInfoCache) == 768) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
765 |
break; |
|
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 |
|
|
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 |
* Load terms dictionary index |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
772 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
773 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
774 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
775 |
private function _loadDictionaryIndex() |
|
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 |
// Check, if index is already serialized |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
778 |
if ($this->_directory->fileExists($this->_name . '.sti')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
779 |
// Load serialized dictionary index data |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
780 |
$stiFile = $this->_directory->getFileObject($this->_name . '.sti'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
781 |
$stiFileData = $stiFile->readBytes($this->_directory->fileLength($this->_name . '.sti')); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
782 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
783 |
// Load dictionary index data |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
784 |
if (($unserializedData = @unserialize($stiFileData)) !== false) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
785 |
list($this->_termDictionary, $this->_termDictionaryInfos) = $unserializedData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
786 |
return; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
787 |
} |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
790 |
// Load data from .tii file and generate .sti file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
791 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
792 |
// Prefetch dictionary index data |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
793 |
$tiiFile = $this->openCompoundFile('.tii'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
794 |
$tiiFileData = $tiiFile->readBytes($this->compoundFileLength('.tii')); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
795 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
796 |
/** Zend_Search_Lucene_Index_DictionaryLoader */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
797 |
require_once 'Zend/Search/Lucene/Index/DictionaryLoader.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
798 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
799 |
// Load dictionary index data |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
800 |
list($this->_termDictionary, $this->_termDictionaryInfos) = |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
801 |
Zend_Search_Lucene_Index_DictionaryLoader::load($tiiFileData); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
802 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
803 |
$stiFileData = serialize(array($this->_termDictionary, $this->_termDictionaryInfos)); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
804 |
$stiFile = $this->_directory->createFile($this->_name . '.sti'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
805 |
$stiFile->writeBytes($stiFileData); |
|
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 |
|
|
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 |
* Scans terms dictionary and returns term info |
|
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 |
* @param Zend_Search_Lucene_Index_Term $term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
812 |
* @return Zend_Search_Lucene_Index_TermInfo |
|
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 |
public function getTermInfo(Zend_Search_Lucene_Index_Term $term) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
815 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
816 |
$termKey = $term->key(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
817 |
if (isset($this->_termInfoCache[$termKey])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
818 |
$termInfo = $this->_termInfoCache[$termKey]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
819 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
820 |
// Move termInfo to the end of cache |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
821 |
unset($this->_termInfoCache[$termKey]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
822 |
$this->_termInfoCache[$termKey] = $termInfo; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
823 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
824 |
return $termInfo; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
825 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
826 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
827 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
828 |
if ($this->_termDictionary === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
829 |
$this->_loadDictionaryIndex(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
830 |
} |
|
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 |
$searchField = $this->getFieldNum($term->field); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
833 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
834 |
if ($searchField == -1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
835 |
return null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
836 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
837 |
$searchDicField = $this->_getFieldPosition($searchField); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
838 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
839 |
// search for appropriate value in dictionary |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
840 |
$lowIndex = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
841 |
$highIndex = count($this->_termDictionary)-1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
842 |
while ($highIndex >= $lowIndex) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
843 |
// $mid = ($highIndex - $lowIndex)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
844 |
$mid = ($highIndex + $lowIndex) >> 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
845 |
$midTerm = $this->_termDictionary[$mid]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
846 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
847 |
$fieldNum = $this->_getFieldPosition($midTerm[0] /* field */); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
848 |
$delta = $searchDicField - $fieldNum; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
849 |
if ($delta == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
850 |
$delta = strcmp($term->text, $midTerm[1] /* text */); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
851 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
852 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
853 |
if ($delta < 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
854 |
$highIndex = $mid-1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
855 |
} elseif ($delta > 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
856 |
$lowIndex = $mid+1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
857 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
858 |
// return $this->_termDictionaryInfos[$mid]; // We got it! |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
859 |
$a = $this->_termDictionaryInfos[$mid]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
860 |
$termInfo = new Zend_Search_Lucene_Index_TermInfo($a[0], $a[1], $a[2], $a[3], $a[4]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
861 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
862 |
// Put loaded termInfo into cache |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
863 |
$this->_termInfoCache[$termKey] = $termInfo; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
864 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
865 |
return $termInfo; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
866 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
867 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
868 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
869 |
if ($highIndex == -1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
870 |
// Term is out of the dictionary range |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
871 |
return null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
872 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
873 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
874 |
$prevPosition = $highIndex; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
875 |
$prevTerm = $this->_termDictionary[$prevPosition]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
876 |
$prevTermInfo = $this->_termDictionaryInfos[$prevPosition]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
877 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
878 |
$tisFile = $this->openCompoundFile('.tis'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
879 |
$tiVersion = $tisFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
880 |
if ($tiVersion != (int)0xFFFFFFFE /* pre-2.1 format */ && |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
881 |
$tiVersion != (int)0xFFFFFFFD /* 2.1+ format */) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
882 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
883 |
throw new Zend_Search_Lucene_Exception('Wrong TermInfoFile file format'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
884 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
885 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
886 |
$termCount = $tisFile->readLong(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
887 |
$indexInterval = $tisFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
888 |
$skipInterval = $tisFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
889 |
if ($tiVersion == (int)0xFFFFFFFD /* 2.1+ format */) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
890 |
$maxSkipLevels = $tisFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
891 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
892 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
893 |
$tisFile->seek($prevTermInfo[4] /* indexPointer */ - (($tiVersion == (int)0xFFFFFFFD)? 24 : 20) /* header size*/, SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
894 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
895 |
$termValue = $prevTerm[1] /* text */; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
896 |
$termFieldNum = $prevTerm[0] /* field */; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
897 |
$freqPointer = $prevTermInfo[1] /* freqPointer */; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
898 |
$proxPointer = $prevTermInfo[2] /* proxPointer */; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
899 |
for ($count = $prevPosition*$indexInterval + 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
900 |
$count <= $termCount && |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
901 |
( $this->_getFieldPosition($termFieldNum) < $searchDicField || |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
902 |
($this->_getFieldPosition($termFieldNum) == $searchDicField && |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
903 |
strcmp($termValue, $term->text) < 0) ); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
904 |
$count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
905 |
$termPrefixLength = $tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
906 |
$termSuffix = $tisFile->readString(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
907 |
$termFieldNum = $tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
908 |
$termValue = Zend_Search_Lucene_Index_Term::getPrefix($termValue, $termPrefixLength) . $termSuffix; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
909 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
910 |
$docFreq = $tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
911 |
$freqPointer += $tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
912 |
$proxPointer += $tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
913 |
if( $docFreq >= $skipInterval ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
914 |
$skipOffset = $tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
915 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
916 |
$skipOffset = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
917 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
918 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
919 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
920 |
if ($termFieldNum == $searchField && $termValue == $term->text) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
921 |
$termInfo = new Zend_Search_Lucene_Index_TermInfo($docFreq, $freqPointer, $proxPointer, $skipOffset); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
922 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
923 |
$termInfo = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
924 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
925 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
926 |
// Put loaded termInfo into cache |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
927 |
$this->_termInfoCache[$termKey] = $termInfo; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
928 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
929 |
if (count($this->_termInfoCache) == 1024) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
930 |
$this->_cleanUpTermInfoCache(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
931 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
932 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
933 |
return $termInfo; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
934 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
935 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
936 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
937 |
* Returns IDs of all the documents containing term. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
938 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
939 |
* @param Zend_Search_Lucene_Index_Term $term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
940 |
* @param integer $shift |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
941 |
* @param Zend_Search_Lucene_Index_DocsFilter|null $docsFilter |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
942 |
* @return array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
943 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
944 |
public function termDocs(Zend_Search_Lucene_Index_Term $term, $shift = 0, $docsFilter = null) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
945 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
946 |
$termInfo = $this->getTermInfo($term); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
947 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
948 |
if (!$termInfo instanceof Zend_Search_Lucene_Index_TermInfo) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
949 |
if ($docsFilter !== null && $docsFilter instanceof Zend_Search_Lucene_Index_DocsFilter) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
950 |
$docsFilter->segmentFilters[$this->_name] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
951 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
952 |
return array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
953 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
954 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
955 |
$frqFile = $this->openCompoundFile('.frq'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
956 |
$frqFile->seek($termInfo->freqPointer,SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
957 |
$docId = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
958 |
$result = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
959 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
960 |
if ($docsFilter !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
961 |
if (!$docsFilter instanceof Zend_Search_Lucene_Index_DocsFilter) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
962 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
963 |
throw new Zend_Search_Lucene_Exception('Documents filter must be an instance of Zend_Search_Lucene_Index_DocsFilter or null.'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
964 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
965 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
966 |
if (isset($docsFilter->segmentFilters[$this->_name])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
967 |
// Filter already has some data for the current segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
968 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
969 |
// Make short name for the filter (which doesn't need additional dereferencing) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
970 |
$filter = &$docsFilter->segmentFilters[$this->_name]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
971 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
972 |
// Check if filter is not empty |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
973 |
if (count($filter) == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
974 |
return array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
975 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
976 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
977 |
if ($this->_docCount/count($filter) < self::FULL_SCAN_VS_FETCH_BOUNDARY) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
978 |
// Perform fetching |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
979 |
// --------------------------------------------------------------- |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
980 |
$updatedFilterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
981 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
982 |
for( $count=0; $count < $termInfo->docFreq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
983 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
984 |
if( $docDelta % 2 == 1 ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
985 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
986 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
987 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
988 |
// read freq |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
989 |
$frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
990 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
991 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
992 |
if (isset($filter[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
993 |
$result[] = $shift + $docId; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
994 |
$updatedFilterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
995 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
996 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
997 |
$docsFilter->segmentFilters[$this->_name] = $updatedFilterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
998 |
// --------------------------------------------------------------- |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
999 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1000 |
// Perform full scan |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1001 |
$updatedFilterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1002 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1003 |
for( $count=0; $count < $termInfo->docFreq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1004 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1005 |
if( $docDelta % 2 == 1 ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1006 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1007 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1008 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1009 |
// read freq |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1010 |
$frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1011 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1012 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1013 |
if (isset($filter[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1014 |
$result[] = $shift + $docId; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1015 |
$updatedFilterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1016 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1017 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1018 |
$docsFilter->segmentFilters[$this->_name] = $updatedFilterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1019 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1020 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1021 |
// Filter is present, but doesn't has data for the current segment yet |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1022 |
$filterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1023 |
for( $count=0; $count < $termInfo->docFreq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1024 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1025 |
if( $docDelta % 2 == 1 ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1026 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1027 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1028 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1029 |
// read freq |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1030 |
$frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1031 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1032 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1033 |
$result[] = $shift + $docId; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1034 |
$filterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1035 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1036 |
$docsFilter->segmentFilters[$this->_name] = $filterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1037 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1038 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1039 |
for( $count=0; $count < $termInfo->docFreq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1040 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1041 |
if( $docDelta % 2 == 1 ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1042 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1043 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1044 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1045 |
// read freq |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1046 |
$frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1047 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1048 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1049 |
$result[] = $shift + $docId; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1050 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1051 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1052 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1053 |
return $result; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1054 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1055 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1056 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1057 |
* Returns term freqs array. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1058 |
* Result array structure: array(docId => freq, ...) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1059 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1060 |
* @param Zend_Search_Lucene_Index_Term $term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1061 |
* @param integer $shift |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1062 |
* @param Zend_Search_Lucene_Index_DocsFilter|null $docsFilter |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1063 |
* @return Zend_Search_Lucene_Index_TermInfo |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1064 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1065 |
public function termFreqs(Zend_Search_Lucene_Index_Term $term, $shift = 0, $docsFilter = null) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1066 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1067 |
$termInfo = $this->getTermInfo($term); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1068 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1069 |
if (!$termInfo instanceof Zend_Search_Lucene_Index_TermInfo) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1070 |
if ($docsFilter !== null && $docsFilter instanceof Zend_Search_Lucene_Index_DocsFilter) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1071 |
$docsFilter->segmentFilters[$this->_name] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1072 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1073 |
return array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1074 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1075 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1076 |
$frqFile = $this->openCompoundFile('.frq'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1077 |
$frqFile->seek($termInfo->freqPointer,SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1078 |
$result = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1079 |
$docId = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1080 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1081 |
$result = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1082 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1083 |
if ($docsFilter !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1084 |
if (!$docsFilter instanceof Zend_Search_Lucene_Index_DocsFilter) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1085 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1086 |
throw new Zend_Search_Lucene_Exception('Documents filter must be an instance of Zend_Search_Lucene_Index_DocsFilter or null.'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1087 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1088 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1089 |
if (isset($docsFilter->segmentFilters[$this->_name])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1090 |
// Filter already has some data for the current segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1091 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1092 |
// Make short name for the filter (which doesn't need additional dereferencing) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1093 |
$filter = &$docsFilter->segmentFilters[$this->_name]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1094 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1095 |
// Check if filter is not empty |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1096 |
if (count($filter) == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1097 |
return array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1098 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1099 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1100 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1101 |
if ($this->_docCount/count($filter) < self::FULL_SCAN_VS_FETCH_BOUNDARY) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1102 |
// Perform fetching |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1103 |
// --------------------------------------------------------------- |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1104 |
$updatedFilterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1105 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1106 |
for ($count = 0; $count < $termInfo->docFreq; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1107 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1108 |
if ($docDelta % 2 == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1109 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1110 |
if (isset($filter[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1111 |
$result[$shift + $docId] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1112 |
$updatedFilterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1113 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1114 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1115 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1116 |
$freq = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1117 |
if (isset($filter[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1118 |
$result[$shift + $docId] = $freq; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1119 |
$updatedFilterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1120 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1121 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1122 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1123 |
$docsFilter->segmentFilters[$this->_name] = $updatedFilterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1124 |
// --------------------------------------------------------------- |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1125 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1126 |
// Perform full scan |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1127 |
$updatedFilterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1128 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1129 |
for ($count = 0; $count < $termInfo->docFreq; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1130 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1131 |
if ($docDelta % 2 == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1132 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1133 |
if (isset($filter[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1134 |
$result[$shift + $docId] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1135 |
$updatedFilterData[$docId] = 1; // 1 is just some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1136 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1137 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1138 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1139 |
$freq = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1140 |
if (isset($filter[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1141 |
$result[$shift + $docId] = $freq; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1142 |
$updatedFilterData[$docId] = 1; // 1 is just some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1143 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1144 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1145 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1146 |
$docsFilter->segmentFilters[$this->_name] = $updatedFilterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1147 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1148 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1149 |
// Filter doesn't has data for current segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1150 |
$filterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1151 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1152 |
for ($count = 0; $count < $termInfo->docFreq; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1153 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1154 |
if ($docDelta % 2 == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1155 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1156 |
$result[$shift + $docId] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1157 |
$filterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1158 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1159 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1160 |
$result[$shift + $docId] = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1161 |
$filterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1162 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1163 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1164 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1165 |
$docsFilter->segmentFilters[$this->_name] = $filterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1166 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1167 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1168 |
for ($count = 0; $count < $termInfo->docFreq; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1169 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1170 |
if ($docDelta % 2 == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1171 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1172 |
$result[$shift + $docId] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1173 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1174 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1175 |
$result[$shift + $docId] = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1176 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1177 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1178 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1179 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1180 |
return $result; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1181 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1182 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1183 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1184 |
* Returns term positions array. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1185 |
* Result array structure: array(docId => array(pos1, pos2, ...), ...) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1186 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1187 |
* @param Zend_Search_Lucene_Index_Term $term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1188 |
* @param integer $shift |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1189 |
* @param Zend_Search_Lucene_Index_DocsFilter|null $docsFilter |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1190 |
* @return Zend_Search_Lucene_Index_TermInfo |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1191 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1192 |
public function termPositions(Zend_Search_Lucene_Index_Term $term, $shift = 0, $docsFilter = null) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1193 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1194 |
$termInfo = $this->getTermInfo($term); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1195 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1196 |
if (!$termInfo instanceof Zend_Search_Lucene_Index_TermInfo) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1197 |
if ($docsFilter !== null && $docsFilter instanceof Zend_Search_Lucene_Index_DocsFilter) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1198 |
$docsFilter->segmentFilters[$this->_name] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1199 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1200 |
return array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1201 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1202 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1203 |
$frqFile = $this->openCompoundFile('.frq'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1204 |
$frqFile->seek($termInfo->freqPointer,SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1205 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1206 |
$docId = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1207 |
$freqs = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1208 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1209 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1210 |
if ($docsFilter !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1211 |
if (!$docsFilter instanceof Zend_Search_Lucene_Index_DocsFilter) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1212 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1213 |
throw new Zend_Search_Lucene_Exception('Documents filter must be an instance of Zend_Search_Lucene_Index_DocsFilter or null.'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1214 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1215 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1216 |
if (isset($docsFilter->segmentFilters[$this->_name])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1217 |
// Filter already has some data for the current segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1218 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1219 |
// Make short name for the filter (which doesn't need additional dereferencing) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1220 |
$filter = &$docsFilter->segmentFilters[$this->_name]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1221 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1222 |
// Check if filter is not empty |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1223 |
if (count($filter) == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1224 |
return array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1225 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1226 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1227 |
if ($this->_docCount/count($filter) < self::FULL_SCAN_VS_FETCH_BOUNDARY) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1228 |
// Perform fetching |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1229 |
// --------------------------------------------------------------- |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1230 |
for ($count = 0; $count < $termInfo->docFreq; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1231 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1232 |
if ($docDelta % 2 == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1233 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1234 |
$freqs[$docId] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1235 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1236 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1237 |
$freqs[$docId] = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1238 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1239 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1240 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1241 |
$updatedFilterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1242 |
$result = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1243 |
$prxFile = $this->openCompoundFile('.prx'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1244 |
$prxFile->seek($termInfo->proxPointer, SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1245 |
foreach ($freqs as $docId => $freq) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1246 |
$termPosition = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1247 |
$positions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1248 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1249 |
// we have to read .prx file to get right position for next doc |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1250 |
// even filter doesn't match current document |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1251 |
for ($count = 0; $count < $freq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1252 |
$termPosition += $prxFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1253 |
$positions[] = $termPosition; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1254 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1255 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1256 |
// Include into updated filter and into result only if doc is matched by filter |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1257 |
if (isset($filter[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1258 |
$updatedFilterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1259 |
$result[$shift + $docId] = $positions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1260 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1261 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1262 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1263 |
$docsFilter->segmentFilters[$this->_name] = $updatedFilterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1264 |
// --------------------------------------------------------------- |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1265 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1266 |
// Perform full scan |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1267 |
for ($count = 0; $count < $termInfo->docFreq; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1268 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1269 |
if ($docDelta % 2 == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1270 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1271 |
$freqs[$docId] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1272 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1273 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1274 |
$freqs[$docId] = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1275 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1276 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1277 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1278 |
$updatedFilterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1279 |
$result = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1280 |
$prxFile = $this->openCompoundFile('.prx'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1281 |
$prxFile->seek($termInfo->proxPointer, SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1282 |
foreach ($freqs as $docId => $freq) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1283 |
$termPosition = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1284 |
$positions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1285 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1286 |
// we have to read .prx file to get right position for next doc |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1287 |
// even filter doesn't match current document |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1288 |
for ($count = 0; $count < $freq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1289 |
$termPosition += $prxFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1290 |
$positions[] = $termPosition; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1291 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1292 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1293 |
// Include into updated filter and into result only if doc is matched by filter |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1294 |
if (isset($filter[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1295 |
$updatedFilterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1296 |
$result[$shift + $docId] = $positions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1297 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1298 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1299 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1300 |
$docsFilter->segmentFilters[$this->_name] = $updatedFilterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1301 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1302 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1303 |
// Filter doesn't has data for current segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1304 |
for ($count = 0; $count < $termInfo->docFreq; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1305 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1306 |
if ($docDelta % 2 == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1307 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1308 |
$freqs[$docId] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1309 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1310 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1311 |
$freqs[$docId] = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1312 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1313 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1314 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1315 |
$filterData = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1316 |
$result = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1317 |
$prxFile = $this->openCompoundFile('.prx'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1318 |
$prxFile->seek($termInfo->proxPointer, SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1319 |
foreach ($freqs as $docId => $freq) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1320 |
$filterData[$docId] = 1; // 1 is just a some constant value, so we don't need additional var dereference here |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1321 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1322 |
$termPosition = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1323 |
$positions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1324 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1325 |
for ($count = 0; $count < $freq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1326 |
$termPosition += $prxFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1327 |
$positions[] = $termPosition; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1328 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1329 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1330 |
$result[$shift + $docId] = $positions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1331 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1332 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1333 |
$docsFilter->segmentFilters[$this->_name] = $filterData; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1334 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1335 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1336 |
for ($count = 0; $count < $termInfo->docFreq; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1337 |
$docDelta = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1338 |
if ($docDelta % 2 == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1339 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1340 |
$freqs[$docId] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1341 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1342 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1343 |
$freqs[$docId] = $frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1344 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1345 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1346 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1347 |
$result = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1348 |
$prxFile = $this->openCompoundFile('.prx'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1349 |
$prxFile->seek($termInfo->proxPointer, SEEK_CUR); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1350 |
foreach ($freqs as $docId => $freq) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1351 |
$termPosition = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1352 |
$positions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1353 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1354 |
for ($count = 0; $count < $freq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1355 |
$termPosition += $prxFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1356 |
$positions[] = $termPosition; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1357 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1358 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1359 |
$result[$shift + $docId] = $positions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1360 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1361 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1362 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1363 |
return $result; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1364 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1365 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1366 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1367 |
* Load normalizatin factors from an index file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1368 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1369 |
* @param integer $fieldNum |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1370 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1371 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1372 |
private function _loadNorm($fieldNum) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1373 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1374 |
if ($this->_hasSingleNormFile) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1375 |
$normfFile = $this->openCompoundFile('.nrm'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1376 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1377 |
$header = $normfFile->readBytes(3); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1378 |
$headerFormatVersion = $normfFile->readByte(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1379 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1380 |
if ($header != 'NRM' || $headerFormatVersion != (int)0xFF) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1381 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1382 |
throw new Zend_Search_Lucene_Exception('Wrong norms file format.'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1383 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1384 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1385 |
foreach ($this->_fields as $fNum => $fieldInfo) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1386 |
if ($fieldInfo->isIndexed) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1387 |
$this->_norms[$fNum] = $normfFile->readBytes($this->_docCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1388 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1389 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1390 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1391 |
$fFile = $this->openCompoundFile('.f' . $fieldNum); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1392 |
$this->_norms[$fieldNum] = $fFile->readBytes($this->_docCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1393 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1394 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1395 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1396 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1397 |
* Returns normalization factor for specified documents |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1398 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1399 |
* @param integer $id |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1400 |
* @param string $fieldName |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1401 |
* @return float |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1402 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1403 |
public function norm($id, $fieldName) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1404 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1405 |
$fieldNum = $this->getFieldNum($fieldName); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1406 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1407 |
if ( !($this->_fields[$fieldNum]->isIndexed) ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1408 |
return null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1409 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1410 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1411 |
if (!isset($this->_norms[$fieldNum])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1412 |
$this->_loadNorm($fieldNum); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1413 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1414 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1415 |
return Zend_Search_Lucene_Search_Similarity::decodeNorm( ord($this->_norms[$fieldNum][$id]) ); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1416 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1417 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1418 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1419 |
* Returns norm vector, encoded in a byte string |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1420 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1421 |
* @param string $fieldName |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1422 |
* @return string |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1423 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1424 |
public function normVector($fieldName) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1425 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1426 |
$fieldNum = $this->getFieldNum($fieldName); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1427 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1428 |
if ($fieldNum == -1 || !($this->_fields[$fieldNum]->isIndexed)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1429 |
$similarity = Zend_Search_Lucene_Search_Similarity::getDefault(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1430 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1431 |
return str_repeat(chr($similarity->encodeNorm( $similarity->lengthNorm($fieldName, 0) )), |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1432 |
$this->_docCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1433 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1434 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1435 |
if (!isset($this->_norms[$fieldNum])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1436 |
$this->_loadNorm($fieldNum); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1437 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1438 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1439 |
return $this->_norms[$fieldNum]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1440 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1441 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1442 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1443 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1444 |
* Returns true if any documents have been deleted from this index segment. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1445 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1446 |
* @return boolean |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1447 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1448 |
public function hasDeletions() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1449 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1450 |
return $this->_deleted !== null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1451 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1452 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1453 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1454 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1455 |
* Returns true if segment has single norms file. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1456 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1457 |
* @return boolean |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1458 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1459 |
public function hasSingleNormFile() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1460 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1461 |
return $this->_hasSingleNormFile ? true : false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1462 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1463 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1464 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1465 |
* Returns true if segment is stored using compound segment file. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1466 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1467 |
* @return boolean |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1468 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1469 |
public function isCompound() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1470 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1471 |
return $this->_isCompound; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1472 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1473 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1474 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1475 |
* Deletes a document from the index segment. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1476 |
* $id is an internal document id |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1477 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1478 |
* @param integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1479 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1480 |
public function delete($id) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1481 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1482 |
$this->_deletedDirty = true; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1483 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1484 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1485 |
if ($this->_deleted === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1486 |
$this->_deleted = bitset_empty($id); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1487 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1488 |
bitset_incl($this->_deleted, $id); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1489 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1490 |
if ($this->_deleted === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1491 |
$this->_deleted = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1492 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1493 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1494 |
$this->_deleted[$id] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1495 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1496 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1497 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1498 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1499 |
* Checks, that document is deleted |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1500 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1501 |
* @param integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1502 |
* @return boolean |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1503 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1504 |
public function isDeleted($id) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1505 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1506 |
if ($this->_deleted === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1507 |
return false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1508 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1509 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1510 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1511 |
return bitset_in($this->_deleted, $id); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1512 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1513 |
return isset($this->_deleted[$id]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1514 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1515 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1516 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1517 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1518 |
* Detect latest delete generation |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1519 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1520 |
* Is actualy used from writeChanges() method or from the constructor if it's invoked from |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1521 |
* Index writer. In both cases index write lock is already obtained, so we shouldn't care |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1522 |
* about it |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1523 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1524 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1525 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1526 |
private function _detectLatestDelGen() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1527 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1528 |
$delFileList = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1529 |
foreach ($this->_directory->fileList() as $file) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1530 |
if ($file == $this->_name . '.del') { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1531 |
// Matches <segment_name>.del file name |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1532 |
$delFileList[] = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1533 |
} else if (preg_match('/^' . $this->_name . '_([a-zA-Z0-9]+)\.del$/i', $file, $matches)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1534 |
// Matches <segment_name>_NNN.del file names |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1535 |
$delFileList[] = (int)base_convert($matches[1], 36, 10); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1536 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1537 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1538 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1539 |
if (count($delFileList) == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1540 |
// There is no deletions file for current segment in the directory |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1541 |
// Set deletions file generation number to 1 |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1542 |
return -1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1543 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1544 |
// There are some deletions files for current segment in the directory |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1545 |
// Set deletions file generation number to the highest nuber |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1546 |
return max($delFileList); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1547 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1548 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1549 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1550 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1551 |
* Write changes if it's necessary. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1552 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1553 |
* This method must be invoked only from the Writer _updateSegments() method, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1554 |
* so index Write lock has to be already obtained. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1555 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1556 |
* @internal |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1557 |
* @throws Zend_Search_Lucene_Exceptions |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1558 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1559 |
public function writeChanges() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1560 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1561 |
// Get new generation number |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1562 |
$latestDelGen = $this->_detectLatestDelGen(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1563 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1564 |
if (!$this->_deletedDirty) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1565 |
// There was no deletions by current process |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1566 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1567 |
if ($latestDelGen == $this->_delGen) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1568 |
// Delete file hasn't been updated by any concurrent process |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1569 |
return; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1570 |
} else if ($latestDelGen > $this->_delGen) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1571 |
// Delete file has been updated by some concurrent process |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1572 |
// Reload deletions file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1573 |
$this->_delGen = $latestDelGen; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1574 |
$this->_deleted = $this->_loadDelFile(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1575 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1576 |
return; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1577 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1578 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1579 |
throw new Zend_Search_Lucene_Exception('Delete file processing workflow is corrupted for the segment \'' . $this->_name . '\'.'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1580 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1581 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1582 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1583 |
if ($latestDelGen > $this->_delGen) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1584 |
// Merge current deletions with latest deletions file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1585 |
$this->_delGen = $latestDelGen; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1586 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1587 |
$latestDelete = $this->_loadDelFile(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1588 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1589 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1590 |
$this->_deleted = bitset_union($this->_deleted, $latestDelete); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1591 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1592 |
$this->_deleted += $latestDelete; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1593 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1594 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1595 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1596 |
if (extension_loaded('bitset')) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1597 |
$delBytes = $this->_deleted; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1598 |
$bitCount = count(bitset_to_array($delBytes)); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1599 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1600 |
$byteCount = floor($this->_docCount/8)+1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1601 |
$delBytes = str_repeat(chr(0), $byteCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1602 |
for ($count = 0; $count < $byteCount; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1603 |
$byte = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1604 |
for ($bit = 0; $bit < 8; $bit++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1605 |
if (isset($this->_deleted[$count*8 + $bit])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1606 |
$byte |= (1<<$bit); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1607 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1608 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1609 |
$delBytes[$count] = chr($byte); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1610 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1611 |
$bitCount = count($this->_deleted); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1612 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1613 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1614 |
if ($this->_delGen == -1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1615 |
// Set delete file generation number to 1 |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1616 |
$this->_delGen = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1617 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1618 |
// Increase delete file generation number by 1 |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1619 |
$this->_delGen++; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1620 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1621 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1622 |
$delFile = $this->_directory->createFile($this->_name . '_' . base_convert($this->_delGen, 10, 36) . '.del'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1623 |
$delFile->writeInt($this->_docCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1624 |
$delFile->writeInt($bitCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1625 |
$delFile->writeBytes($delBytes); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1626 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1627 |
$this->_deletedDirty = false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1628 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1629 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1630 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1631 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1632 |
* Term Dictionary File object for stream like terms reading |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1633 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1634 |
* @var Zend_Search_Lucene_Storage_File |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1635 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1636 |
private $_tisFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1637 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1638 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1639 |
* Actual offset of the .tis file data |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1640 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1641 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1642 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1643 |
private $_tisFileOffset; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1644 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1645 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1646 |
* Frequencies File object for stream like terms reading |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1647 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1648 |
* @var Zend_Search_Lucene_Storage_File |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1649 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1650 |
private $_frqFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1651 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1652 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1653 |
* Actual offset of the .frq file data |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1654 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1655 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1656 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1657 |
private $_frqFileOffset; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1658 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1659 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1660 |
* Positions File object for stream like terms reading |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1661 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1662 |
* @var Zend_Search_Lucene_Storage_File |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1663 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1664 |
private $_prxFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1665 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1666 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1667 |
* Actual offset of the .prx file in the compound file |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1668 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1669 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1670 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1671 |
private $_prxFileOffset; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1672 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1673 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1674 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1675 |
* Actual number of terms in term stream |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1676 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1677 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1678 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1679 |
private $_termCount = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1680 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1681 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1682 |
* Overall number of terms in term stream |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1683 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1684 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1685 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1686 |
private $_termNum = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1687 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1688 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1689 |
* Segment index interval |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1690 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1691 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1692 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1693 |
private $_indexInterval; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1694 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1695 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1696 |
* Segment skip interval |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1697 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1698 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1699 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1700 |
private $_skipInterval; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1701 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1702 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1703 |
* Last TermInfo in a terms stream |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1704 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1705 |
* @var Zend_Search_Lucene_Index_TermInfo |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1706 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1707 |
private $_lastTermInfo = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1708 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1709 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1710 |
* Last Term in a terms stream |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1711 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1712 |
* @var Zend_Search_Lucene_Index_Term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1713 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1714 |
private $_lastTerm = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1715 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1716 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1717 |
* Map of the document IDs |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1718 |
* Used to get new docID after removing deleted documents. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1719 |
* It's not very effective from memory usage point of view, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1720 |
* but much more faster, then other methods |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1721 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1722 |
* @var array|null |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1723 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1724 |
private $_docMap = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1725 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1726 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1727 |
* An array of all term positions in the documents. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1728 |
* Array structure: array( docId => array( pos1, pos2, ...), ...) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1729 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1730 |
* Is set to null if term positions loading has to be skipped |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1731 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1732 |
* @var array|null |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1733 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1734 |
private $_lastTermPositions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1735 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1736 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1737 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1738 |
* Terms scan mode |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1739 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1740 |
* Values: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1741 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1742 |
* self::SM_TERMS_ONLY - terms are scanned, no additional info is retrieved |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1743 |
* self::SM_FULL_INFO - terms are scanned, frequency and position info is retrieved |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1744 |
* self::SM_MERGE_INFO - terms are scanned, frequency and position info is retrieved |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1745 |
* document numbers are compacted (shifted if segment has deleted documents) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1746 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1747 |
* @var integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1748 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1749 |
private $_termsScanMode; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1750 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1751 |
/** Scan modes */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1752 |
const SM_TERMS_ONLY = 0; // terms are scanned, no additional info is retrieved |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1753 |
const SM_FULL_INFO = 1; // terms are scanned, frequency and position info is retrieved |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1754 |
const SM_MERGE_INFO = 2; // terms are scanned, frequency and position info is retrieved |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1755 |
// document numbers are compacted (shifted if segment contains deleted documents) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1756 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1757 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1758 |
* Reset terms stream |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1759 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1760 |
* $startId - id for the fist document |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1761 |
* $compact - remove deleted documents |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1762 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1763 |
* Returns start document id for the next segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1764 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1765 |
* @param integer $startId |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1766 |
* @param integer $mode |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1767 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1768 |
* @return integer |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1769 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1770 |
public function resetTermsStream(/** $startId = 0, $mode = self::SM_TERMS_ONLY */) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1771 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1772 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1773 |
* SegmentInfo->resetTermsStream() method actually takes two optional parameters: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1774 |
* $startId (default value is 0) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1775 |
* $mode (default value is self::SM_TERMS_ONLY) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1776 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1777 |
$argList = func_get_args(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1778 |
if (count($argList) > 2) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1779 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1780 |
throw new Zend_Search_Lucene_Exception('Wrong number of arguments'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1781 |
} else if (count($argList) == 2) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1782 |
$startId = $argList[0]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1783 |
$mode = $argList[1]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1784 |
} else if (count($argList) == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1785 |
$startId = $argList[0]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1786 |
$mode = self::SM_TERMS_ONLY; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1787 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1788 |
$startId = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1789 |
$mode = self::SM_TERMS_ONLY; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1790 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1791 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1792 |
if ($this->_tisFile !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1793 |
$this->_tisFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1794 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1795 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1796 |
$this->_tisFile = $this->openCompoundFile('.tis', false); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1797 |
$this->_tisFileOffset = $this->_tisFile->tell(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1798 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1799 |
$tiVersion = $this->_tisFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1800 |
if ($tiVersion != (int)0xFFFFFFFE /* pre-2.1 format */ && |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1801 |
$tiVersion != (int)0xFFFFFFFD /* 2.1+ format */) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1802 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1803 |
throw new Zend_Search_Lucene_Exception('Wrong TermInfoFile file format'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1804 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1805 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1806 |
$this->_termCount = |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1807 |
$this->_termNum = $this->_tisFile->readLong(); // Read terms count |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1808 |
$this->_indexInterval = $this->_tisFile->readInt(); // Read Index interval |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1809 |
$this->_skipInterval = $this->_tisFile->readInt(); // Read skip interval |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1810 |
if ($tiVersion == (int)0xFFFFFFFD /* 2.1+ format */) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1811 |
$maxSkipLevels = $this->_tisFile->readInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1812 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1813 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1814 |
if ($this->_frqFile !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1815 |
$this->_frqFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1816 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1817 |
if ($this->_prxFile !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1818 |
$this->_prxFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1819 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1820 |
$this->_docMap = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1821 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1822 |
$this->_lastTerm = new Zend_Search_Lucene_Index_Term('', -1); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1823 |
$this->_lastTermInfo = new Zend_Search_Lucene_Index_TermInfo(0, 0, 0, 0); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1824 |
$this->_lastTermPositions = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1825 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1826 |
$this->_termsScanMode = $mode; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1827 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1828 |
switch ($mode) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1829 |
case self::SM_TERMS_ONLY: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1830 |
// Do nothing |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1831 |
break; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1832 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1833 |
case self::SM_FULL_INFO: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1834 |
// break intentionally omitted |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1835 |
case self::SM_MERGE_INFO: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1836 |
$this->_frqFile = $this->openCompoundFile('.frq', false); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1837 |
$this->_frqFileOffset = $this->_frqFile->tell(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1838 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1839 |
$this->_prxFile = $this->openCompoundFile('.prx', false); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1840 |
$this->_prxFileOffset = $this->_prxFile->tell(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1841 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1842 |
for ($count = 0; $count < $this->_docCount; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1843 |
if (!$this->isDeleted($count)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1844 |
$this->_docMap[$count] = $startId + (($mode == self::SM_MERGE_INFO) ? count($this->_docMap) : $count); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1845 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1846 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1847 |
break; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1848 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1849 |
default: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1850 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1851 |
throw new Zend_Search_Lucene_Exception('Wrong terms scaning mode specified.'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1852 |
break; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1853 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1854 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1855 |
// Calculate next segment start id (since $this->_docMap structure may be cleaned by $this->nextTerm() call) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1856 |
$nextSegmentStartId = $startId + (($mode == self::SM_MERGE_INFO) ? count($this->_docMap) : $this->_docCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1857 |
$this->nextTerm(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1858 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1859 |
return $nextSegmentStartId; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1860 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1861 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1862 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1863 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1864 |
* Skip terms stream up to the specified term preffix. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1865 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1866 |
* Prefix contains fully specified field info and portion of searched term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1867 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1868 |
* @param Zend_Search_Lucene_Index_Term $prefix |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1869 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1870 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1871 |
public function skipTo(Zend_Search_Lucene_Index_Term $prefix) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1872 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1873 |
if ($this->_termDictionary === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1874 |
$this->_loadDictionaryIndex(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1875 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1876 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1877 |
$searchField = $this->getFieldNum($prefix->field); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1878 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1879 |
if ($searchField == -1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1880 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1881 |
* Field is not presented in this segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1882 |
* Go to the end of dictionary |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1883 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1884 |
$this->_tisFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1885 |
$this->_frqFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1886 |
$this->_prxFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1887 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1888 |
$this->_lastTerm = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1889 |
$this->_lastTermInfo = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1890 |
$this->_lastTermPositions = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1891 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1892 |
return; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1893 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1894 |
$searchDicField = $this->_getFieldPosition($searchField); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1895 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1896 |
// search for appropriate value in dictionary |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1897 |
$lowIndex = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1898 |
$highIndex = count($this->_termDictionary)-1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1899 |
while ($highIndex >= $lowIndex) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1900 |
// $mid = ($highIndex - $lowIndex)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1901 |
$mid = ($highIndex + $lowIndex) >> 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1902 |
$midTerm = $this->_termDictionary[$mid]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1903 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1904 |
$fieldNum = $this->_getFieldPosition($midTerm[0] /* field */); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1905 |
$delta = $searchDicField - $fieldNum; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1906 |
if ($delta == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1907 |
$delta = strcmp($prefix->text, $midTerm[1] /* text */); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1908 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1909 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1910 |
if ($delta < 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1911 |
$highIndex = $mid-1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1912 |
} elseif ($delta > 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1913 |
$lowIndex = $mid+1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1914 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1915 |
// We have reached term we are looking for |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1916 |
break; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1917 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1918 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1919 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1920 |
if ($highIndex == -1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1921 |
// Term is out of the dictionary range |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1922 |
$this->_tisFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1923 |
$this->_frqFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1924 |
$this->_prxFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1925 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1926 |
$this->_lastTerm = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1927 |
$this->_lastTermInfo = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1928 |
$this->_lastTermPositions = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1929 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1930 |
return; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1931 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1932 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1933 |
$prevPosition = $highIndex; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1934 |
$prevTerm = $this->_termDictionary[$prevPosition]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1935 |
$prevTermInfo = $this->_termDictionaryInfos[$prevPosition]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1936 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1937 |
if ($this->_tisFile === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1938 |
// The end of terms stream is reached and terms dictionary file is closed |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1939 |
// Perform mini-reset operation |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1940 |
$this->_tisFile = $this->openCompoundFile('.tis', false); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1941 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1942 |
if ($this->_termsScanMode == self::SM_FULL_INFO || $this->_termsScanMode == self::SM_MERGE_INFO) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1943 |
$this->_frqFile = $this->openCompoundFile('.frq', false); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1944 |
$this->_prxFile = $this->openCompoundFile('.prx', false); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1945 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1946 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1947 |
$this->_tisFile->seek($this->_tisFileOffset + $prevTermInfo[4], SEEK_SET); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1948 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1949 |
$this->_lastTerm = new Zend_Search_Lucene_Index_Term($prevTerm[1] /* text */, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1950 |
($prevTerm[0] == -1) ? '' : $this->_fields[$prevTerm[0] /* field */]->name); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1951 |
$this->_lastTermInfo = new Zend_Search_Lucene_Index_TermInfo($prevTermInfo[0] /* docFreq */, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1952 |
$prevTermInfo[1] /* freqPointer */, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1953 |
$prevTermInfo[2] /* proxPointer */, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1954 |
$prevTermInfo[3] /* skipOffset */); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1955 |
$this->_termCount = $this->_termNum - $prevPosition*$this->_indexInterval; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1956 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1957 |
if ($highIndex == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1958 |
// skip start entry |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1959 |
$this->nextTerm(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1960 |
} else if ($prefix->field == $this->_lastTerm->field && $prefix->text == $this->_lastTerm->text) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1961 |
// We got exact match in the dictionary index |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1962 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1963 |
if ($this->_termsScanMode == self::SM_FULL_INFO || $this->_termsScanMode == self::SM_MERGE_INFO) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1964 |
$this->_lastTermPositions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1965 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1966 |
$this->_frqFile->seek($this->_lastTermInfo->freqPointer + $this->_frqFileOffset, SEEK_SET); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1967 |
$freqs = array(); $docId = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1968 |
for( $count = 0; $count < $this->_lastTermInfo->docFreq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1969 |
$docDelta = $this->_frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1970 |
if( $docDelta % 2 == 1 ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1971 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1972 |
$freqs[ $docId ] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1973 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1974 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1975 |
$freqs[ $docId ] = $this->_frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1976 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1977 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1978 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1979 |
$this->_prxFile->seek($this->_lastTermInfo->proxPointer + $this->_prxFileOffset, SEEK_SET); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1980 |
foreach ($freqs as $docId => $freq) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1981 |
$termPosition = 0; $positions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1982 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1983 |
for ($count = 0; $count < $freq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1984 |
$termPosition += $this->_prxFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1985 |
$positions[] = $termPosition; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1986 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1987 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1988 |
if (isset($this->_docMap[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1989 |
$this->_lastTermPositions[$this->_docMap[$docId]] = $positions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1990 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1991 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1992 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1993 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1994 |
return; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1995 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1996 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1997 |
// Search term matching specified prefix |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1998 |
while ($this->_lastTerm !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1999 |
if ( strcmp($this->_lastTerm->field, $prefix->field) > 0 || |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2000 |
($prefix->field == $this->_lastTerm->field && strcmp($this->_lastTerm->text, $prefix->text) >= 0) ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2001 |
// Current term matches or greate than the pattern |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2002 |
return; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2003 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2004 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2005 |
$this->nextTerm(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2006 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2007 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2008 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2009 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2010 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2011 |
* Scans terms dictionary and returns next term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2012 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2013 |
* @return Zend_Search_Lucene_Index_Term|null |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2014 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2015 |
public function nextTerm() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2016 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2017 |
if ($this->_tisFile === null || $this->_termCount == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2018 |
$this->_lastTerm = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2019 |
$this->_lastTermInfo = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2020 |
$this->_lastTermPositions = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2021 |
$this->_docMap = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2022 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2023 |
// may be necessary for "empty" segment |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2024 |
$this->_tisFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2025 |
$this->_frqFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2026 |
$this->_prxFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2027 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2028 |
return null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2029 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2030 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2031 |
$termPrefixLength = $this->_tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2032 |
$termSuffix = $this->_tisFile->readString(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2033 |
$termFieldNum = $this->_tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2034 |
$termValue = Zend_Search_Lucene_Index_Term::getPrefix($this->_lastTerm->text, $termPrefixLength) . $termSuffix; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2035 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2036 |
$this->_lastTerm = new Zend_Search_Lucene_Index_Term($termValue, $this->_fields[$termFieldNum]->name); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2037 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2038 |
$docFreq = $this->_tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2039 |
$freqPointer = $this->_lastTermInfo->freqPointer + $this->_tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2040 |
$proxPointer = $this->_lastTermInfo->proxPointer + $this->_tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2041 |
if ($docFreq >= $this->_skipInterval) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2042 |
$skipOffset = $this->_tisFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2043 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2044 |
$skipOffset = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2045 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2046 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2047 |
$this->_lastTermInfo = new Zend_Search_Lucene_Index_TermInfo($docFreq, $freqPointer, $proxPointer, $skipOffset); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2048 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2049 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2050 |
if ($this->_termsScanMode == self::SM_FULL_INFO || $this->_termsScanMode == self::SM_MERGE_INFO) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2051 |
$this->_lastTermPositions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2052 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2053 |
$this->_frqFile->seek($this->_lastTermInfo->freqPointer + $this->_frqFileOffset, SEEK_SET); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2054 |
$freqs = array(); $docId = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2055 |
for( $count = 0; $count < $this->_lastTermInfo->docFreq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2056 |
$docDelta = $this->_frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2057 |
if( $docDelta % 2 == 1 ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2058 |
$docId += ($docDelta-1)/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2059 |
$freqs[ $docId ] = 1; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2060 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2061 |
$docId += $docDelta/2; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2062 |
$freqs[ $docId ] = $this->_frqFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2063 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2064 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2065 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2066 |
$this->_prxFile->seek($this->_lastTermInfo->proxPointer + $this->_prxFileOffset, SEEK_SET); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2067 |
foreach ($freqs as $docId => $freq) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2068 |
$termPosition = 0; $positions = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2069 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2070 |
for ($count = 0; $count < $freq; $count++ ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2071 |
$termPosition += $this->_prxFile->readVInt(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2072 |
$positions[] = $termPosition; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2073 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2074 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2075 |
if (isset($this->_docMap[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2076 |
$this->_lastTermPositions[$this->_docMap[$docId]] = $positions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2077 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2078 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2079 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2080 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2081 |
$this->_termCount--; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2082 |
if ($this->_termCount == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2083 |
$this->_tisFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2084 |
$this->_frqFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2085 |
$this->_prxFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2086 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2087 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2088 |
return $this->_lastTerm; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2089 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2090 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2091 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2092 |
* Close terms stream |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2093 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2094 |
* Should be used for resources clean up if stream is not read up to the end |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2095 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2096 |
public function closeTermsStream() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2097 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2098 |
$this->_tisFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2099 |
$this->_frqFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2100 |
$this->_prxFile = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2101 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2102 |
$this->_lastTerm = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2103 |
$this->_lastTermInfo = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2104 |
$this->_lastTermPositions = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2105 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2106 |
$this->_docMap = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2107 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2108 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2109 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2110 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2111 |
* Returns term in current position |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2112 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2113 |
* @return Zend_Search_Lucene_Index_Term|null |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2114 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2115 |
public function currentTerm() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2116 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2117 |
return $this->_lastTerm; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2118 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2119 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2120 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2121 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2122 |
* Returns an array of all term positions in the documents. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2123 |
* Return array structure: array( docId => array( pos1, pos2, ...), ...) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2124 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2125 |
* @return array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2126 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2127 |
public function currentTermPositions() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2128 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2129 |
return $this->_lastTermPositions; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2130 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2131 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2132 |