| author | Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com> |
| Thu, 21 Mar 2013 19:50:53 +0100 | |
| changeset 807 | 877f952ae2bd |
| parent 207 | 621fa6caec0c |
| child 1230 | 68c69c656a2c |
| 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 Search |
|
807
877f952ae2bd
update Zend and twitter oauth end point
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
207
diff
changeset
|
18 |
* @copyright Copyright (c) 2005-2012 Zend Technologies USA Inc. (http://www.zend.com) |
|
0
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 |
|
807
877f952ae2bd
update Zend and twitter oauth end point
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
207
diff
changeset
|
20 |
* @version $Id: MultiTerm.php 24593 2012-01-05 20:35:02Z matthew $ |
|
0
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
21 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
22 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
23 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
24 |
/** Zend_Search_Lucene_Search_Query */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
25 |
require_once 'Zend/Search/Lucene/Search/Query.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
26 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
27 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
28 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
29 |
* @category Zend |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
30 |
* @package Zend_Search_Lucene |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
31 |
* @subpackage Search |
|
807
877f952ae2bd
update Zend and twitter oauth end point
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
207
diff
changeset
|
32 |
* @copyright Copyright (c) 2005-2012 Zend Technologies USA Inc. (http://www.zend.com) |
|
0
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
33 |
* @license http://framework.zend.com/license/new-bsd New BSD License |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
34 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
35 |
class Zend_Search_Lucene_Search_Query_MultiTerm extends Zend_Search_Lucene_Search_Query |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
36 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
37 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
38 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
39 |
* Terms to find. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
40 |
* Array of Zend_Search_Lucene_Index_Term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
41 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
42 |
* @var array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
43 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
44 |
private $_terms = array(); |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
47 |
* Term signs. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
48 |
* If true then term is required. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
49 |
* If false then term is prohibited. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
50 |
* If null then term is neither prohibited, nor required |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
51 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
52 |
* If array is null then all terms are required |
|
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 |
* @var array |
|
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 |
private $_signs; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
57 |
|
|
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 |
* Result vector. |
|
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 |
* @var array |
|
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 |
private $_resVector = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
64 |
|
|
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 |
* Terms positions vectors. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
67 |
* Array of Arrays: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
68 |
* term1Id => (docId => freq, ...) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
69 |
* term2Id => (docId => freq, ...) |
|
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 |
* @var array |
|
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 |
private $_termsFreqs = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
74 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
75 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
76 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
77 |
* A score factor based on the fraction of all query terms |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
78 |
* that a document contains. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
79 |
* float for conjunction queries |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
80 |
* array of float for non conjunction queries |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
81 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
82 |
* @var mixed |
|
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 |
private $_coord = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
85 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
86 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
87 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
88 |
* Terms weights |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
89 |
* array of Zend_Search_Lucene_Search_Weight |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
90 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
91 |
* @var array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
92 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
93 |
private $_weights = array(); |
|
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 |
|
|
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 |
* Class constructor. Create a new multi-term query object. |
|
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 |
* if $signs array is omitted then all terms are required |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
100 |
* it differs from addTerm() behavior, but should never be used |
|
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 |
* @param array $terms Array of Zend_Search_Lucene_Index_Term objects |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
103 |
* @param array $signs Array of signs. Sign is boolean|null. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
104 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
105 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
106 |
public function __construct($terms = null, $signs = null) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
107 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
108 |
if (is_array($terms)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
109 |
require_once 'Zend/Search/Lucene.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
110 |
if (count($terms) > Zend_Search_Lucene::getTermsPerQueryLimit()) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
111 |
throw new Zend_Search_Lucene_Exception('Terms per query limit is reached.'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
112 |
} |
|
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 |
$this->_terms = $terms; |
|
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 |
$this->_signs = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
117 |
// Check if all terms are required |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
118 |
if (is_array($signs)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
119 |
foreach ($signs as $sign ) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
120 |
if ($sign !== true) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
121 |
$this->_signs = $signs; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
122 |
break; |
|
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 |
} |
|
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 |
} |
|
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 |
|
|
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 |
* Add a $term (Zend_Search_Lucene_Index_Term) to this query. |
|
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 |
* The sign is specified as: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
134 |
* TRUE - term is required |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
135 |
* FALSE - term is prohibited |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
136 |
* NULL - term is neither prohibited, nor required |
|
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 |
* @param Zend_Search_Lucene_Index_Term $term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
139 |
* @param boolean|null $sign |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
140 |
* @return void |
|
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 |
public function addTerm(Zend_Search_Lucene_Index_Term $term, $sign = null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
143 |
if ($sign !== true || $this->_signs !== null) { // Skip, if all terms are required |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
144 |
if ($this->_signs === null) { // Check, If all previous terms are required |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
145 |
$this->_signs = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
146 |
foreach ($this->_terms as $prevTerm) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
147 |
$this->_signs[] = true; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
148 |
} |
|
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 |
$this->_signs[] = $sign; |
|
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 |
$this->_terms[] = $term; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
154 |
} |
|
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 |
|
|
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 |
* Re-write query into primitive queries in the context of specified index |
|
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 |
* @param Zend_Search_Lucene_Interface $index |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
161 |
* @return Zend_Search_Lucene_Search_Query |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
162 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
163 |
public function rewrite(Zend_Search_Lucene_Interface $index) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
164 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
165 |
if (count($this->_terms) == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
166 |
require_once 'Zend/Search/Lucene/Search/Query/Empty.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
167 |
return new Zend_Search_Lucene_Search_Query_Empty(); |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
170 |
// Check, that all fields are qualified |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
171 |
$allQualified = true; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
172 |
foreach ($this->_terms as $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
173 |
if ($term->field === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
174 |
$allQualified = false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
175 |
break; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
176 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
177 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
178 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
179 |
if ($allQualified) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
180 |
return $this; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
181 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
182 |
/** transform multiterm query to boolean and apply rewrite() method to subqueries. */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
183 |
require_once 'Zend/Search/Lucene/Search/Query/Boolean.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
184 |
$query = new Zend_Search_Lucene_Search_Query_Boolean(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
185 |
$query->setBoost($this->getBoost()); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
186 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
187 |
require_once 'Zend/Search/Lucene/Search/Query/Term.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
188 |
foreach ($this->_terms as $termId => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
189 |
$subquery = new Zend_Search_Lucene_Search_Query_Term($term); |
|
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 |
$query->addSubquery($subquery->rewrite($index), |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
192 |
($this->_signs === null)? true : $this->_signs[$termId]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
193 |
} |
|
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 |
return $query; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
196 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
197 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
198 |
|
|
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 |
* Optimize query in the context of specified index |
|
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 |
* @param Zend_Search_Lucene_Interface $index |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
203 |
* @return Zend_Search_Lucene_Search_Query |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
204 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
205 |
public function optimize(Zend_Search_Lucene_Interface $index) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
206 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
207 |
$terms = $this->_terms; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
208 |
$signs = $this->_signs; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
209 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
210 |
foreach ($terms as $id => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
211 |
if (!$index->hasTerm($term)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
212 |
if ($signs === null || $signs[$id] === true) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
213 |
// Term is required |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
214 |
require_once 'Zend/Search/Lucene/Search/Query/Empty.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
215 |
return new Zend_Search_Lucene_Search_Query_Empty(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
216 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
217 |
// Term is optional or prohibited |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
218 |
// Remove it from terms and signs list |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
219 |
unset($terms[$id]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
220 |
unset($signs[$id]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
221 |
} |
|
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 |
} |
|
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 |
// Check if all presented terms are prohibited |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
226 |
$allProhibited = true; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
227 |
if ($signs === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
228 |
$allProhibited = false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
229 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
230 |
foreach ($signs as $sign) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
231 |
if ($sign !== false) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
232 |
$allProhibited = false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
233 |
break; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
234 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
235 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
236 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
237 |
if ($allProhibited) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
238 |
require_once 'Zend/Search/Lucene/Search/Query/Empty.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
239 |
return new Zend_Search_Lucene_Search_Query_Empty(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
240 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
241 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
242 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
243 |
* @todo make an optimization for repeated terms |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
244 |
* (they may have different signs) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
245 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
246 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
247 |
if (count($terms) == 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
248 |
// It's already checked, that it's not a prohibited term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
249 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
250 |
// It's one term query with one required or optional element |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
251 |
require_once 'Zend/Search/Lucene/Search/Query/Term.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
252 |
$optimizedQuery = new Zend_Search_Lucene_Search_Query_Term(reset($terms)); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
253 |
$optimizedQuery->setBoost($this->getBoost()); |
|
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 |
return $optimizedQuery; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
256 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
257 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
258 |
if (count($terms) == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
259 |
require_once 'Zend/Search/Lucene/Search/Query/Empty.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
260 |
return new Zend_Search_Lucene_Search_Query_Empty(); |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
263 |
$optimizedQuery = new Zend_Search_Lucene_Search_Query_MultiTerm($terms, $signs); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
264 |
$optimizedQuery->setBoost($this->getBoost()); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
265 |
return $optimizedQuery; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
266 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
267 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
268 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
269 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
270 |
* Returns query term |
|
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 |
* @return array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
273 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
274 |
public function getTerms() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
275 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
276 |
return $this->_terms; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
277 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
278 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
279 |
|
|
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 |
* Return terms signs |
|
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 |
* @return array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
284 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
285 |
public function getSigns() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
286 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
287 |
return $this->_signs; |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
290 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
291 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
292 |
* Set weight for specified term |
|
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 |
* @param integer $num |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
295 |
* @param Zend_Search_Lucene_Search_Weight_Term $weight |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
296 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
297 |
public function setWeight($num, $weight) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
298 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
299 |
$this->_weights[$num] = $weight; |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
303 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
304 |
* Constructs an appropriate Weight implementation for this query. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
305 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
306 |
* @param Zend_Search_Lucene_Interface $reader |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
307 |
* @return Zend_Search_Lucene_Search_Weight |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
308 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
309 |
public function createWeight(Zend_Search_Lucene_Interface $reader) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
310 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
311 |
require_once 'Zend/Search/Lucene/Search/Weight/MultiTerm.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
312 |
$this->_weight = new Zend_Search_Lucene_Search_Weight_MultiTerm($this, $reader); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
313 |
return $this->_weight; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
314 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
315 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
316 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
317 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
318 |
* Calculate result vector for Conjunction query |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
319 |
* (like '+something +another') |
|
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 |
* @param Zend_Search_Lucene_Interface $reader |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
322 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
323 |
private function _calculateConjunctionResult(Zend_Search_Lucene_Interface $reader) |
|
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 |
$this->_resVector = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
326 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
327 |
if (count($this->_terms) == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
328 |
$this->_resVector = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
329 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
330 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
331 |
// Order terms by selectivity |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
332 |
$docFreqs = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
333 |
$ids = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
334 |
foreach ($this->_terms as $id => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
335 |
$docFreqs[] = $reader->docFreq($term); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
336 |
$ids[] = $id; // Used to keep original order for terms with the same selectivity and omit terms comparison |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
337 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
338 |
array_multisort($docFreqs, SORT_ASC, SORT_NUMERIC, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
339 |
$ids, SORT_ASC, SORT_NUMERIC, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
340 |
$this->_terms); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
341 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
342 |
require_once 'Zend/Search/Lucene/Index/DocsFilter.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
343 |
$docsFilter = new Zend_Search_Lucene_Index_DocsFilter(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
344 |
foreach ($this->_terms as $termId => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
345 |
$termDocs = $reader->termDocs($term, $docsFilter); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
346 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
347 |
// Treat last retrieved docs vector as a result set |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
348 |
// (filter collects data for other terms) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
349 |
$this->_resVector = array_flip($termDocs); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
350 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
351 |
foreach ($this->_terms as $termId => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
352 |
$this->_termsFreqs[$termId] = $reader->termFreqs($term, $docsFilter); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
353 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
354 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
355 |
// ksort($this->_resVector, SORT_NUMERIC); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
356 |
// Docs are returned ordered. Used algorithms doesn't change elements order. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
357 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
358 |
|
|
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 |
* Calculate result vector for non Conjunction query |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
362 |
* (like '+something -another') |
|
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 |
* @param Zend_Search_Lucene_Interface $reader |
|
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 |
private function _calculateNonConjunctionResult(Zend_Search_Lucene_Interface $reader) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
367 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
368 |
$requiredVectors = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
369 |
$requiredVectorsSizes = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
370 |
$requiredVectorsIds = array(); // is used to prevent arrays comparison |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
371 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
372 |
$optional = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
373 |
$prohibited = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
374 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
375 |
foreach ($this->_terms as $termId => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
376 |
$termDocs = array_flip($reader->termDocs($term)); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
377 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
378 |
if ($this->_signs[$termId] === true) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
379 |
// required |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
380 |
$requiredVectors[] = $termDocs; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
381 |
$requiredVectorsSizes[] = count($termDocs); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
382 |
$requiredVectorsIds[] = $termId; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
383 |
} elseif ($this->_signs[$termId] === false) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
384 |
// prohibited |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
385 |
// array union |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
386 |
$prohibited += $termDocs; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
387 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
388 |
// neither required, nor prohibited |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
389 |
// array union |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
390 |
$optional += $termDocs; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
391 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
392 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
393 |
$this->_termsFreqs[$termId] = $reader->termFreqs($term); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
394 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
395 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
396 |
// sort resvectors in order of subquery cardinality increasing |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
397 |
array_multisort($requiredVectorsSizes, SORT_ASC, SORT_NUMERIC, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
398 |
$requiredVectorsIds, SORT_ASC, SORT_NUMERIC, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
399 |
$requiredVectors); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
400 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
401 |
$required = null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
402 |
foreach ($requiredVectors as $nextResVector) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
403 |
if($required === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
404 |
$required = $nextResVector; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
405 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
406 |
//$required = array_intersect_key($required, $nextResVector); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
407 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
408 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
409 |
* This code is used as workaround for array_intersect_key() slowness problem. |
|
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 |
$updatedVector = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
412 |
foreach ($required as $id => $value) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
413 |
if (isset($nextResVector[$id])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
414 |
$updatedVector[$id] = $value; |
|
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 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
417 |
$required = $updatedVector; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
418 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
419 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
420 |
if (count($required) == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
421 |
// Empty result set, we don't need to check other terms |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
422 |
break; |
|
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 |
} |
|
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 ($required !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
427 |
$this->_resVector = $required; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
428 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
429 |
$this->_resVector = $optional; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
430 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
431 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
432 |
if (count($prohibited) != 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
433 |
// $this->_resVector = array_diff_key($this->_resVector, $prohibited); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
434 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
435 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
436 |
* This code is used as workaround for array_diff_key() slowness problem. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
437 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
438 |
if (count($this->_resVector) < count($prohibited)) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
439 |
$updatedVector = $this->_resVector; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
440 |
foreach ($this->_resVector as $id => $value) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
441 |
if (isset($prohibited[$id])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
442 |
unset($updatedVector[$id]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
443 |
} |
|
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 |
$this->_resVector = $updatedVector; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
446 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
447 |
$updatedVector = $this->_resVector; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
448 |
foreach ($prohibited as $id => $value) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
449 |
unset($updatedVector[$id]); |
|
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 |
$this->_resVector = $updatedVector; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
452 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
453 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
454 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
455 |
ksort($this->_resVector, SORT_NUMERIC); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
456 |
} |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
460 |
* Score calculator for conjunction queries (all terms are required) |
|
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 |
* @param integer $docId |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
463 |
* @param Zend_Search_Lucene_Interface $reader |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
464 |
* @return float |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
465 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
466 |
public function _conjunctionScore($docId, Zend_Search_Lucene_Interface $reader) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
467 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
468 |
if ($this->_coord === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
469 |
$this->_coord = $reader->getSimilarity()->coord(count($this->_terms), |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
470 |
count($this->_terms) ); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
471 |
} |
|
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 |
$score = 0.0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
474 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
475 |
foreach ($this->_terms as $termId => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
476 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
477 |
* We don't need to check that term freq is not 0 |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
478 |
* Score calculation is performed only for matched docs |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
479 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
480 |
$score += $reader->getSimilarity()->tf($this->_termsFreqs[$termId][$docId]) * |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
481 |
$this->_weights[$termId]->getValue() * |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
482 |
$reader->norm($docId, $term->field); |
|
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 |
return $score * $this->_coord * $this->getBoost(); |
|
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 |
|
|
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 |
* Score calculator for non conjunction queries (not all terms are required) |
|
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 |
* @param integer $docId |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
493 |
* @param Zend_Search_Lucene_Interface $reader |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
494 |
* @return float |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
495 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
496 |
public function _nonConjunctionScore($docId, $reader) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
497 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
498 |
if ($this->_coord === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
499 |
$this->_coord = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
500 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
501 |
$maxCoord = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
502 |
foreach ($this->_signs as $sign) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
503 |
if ($sign !== false /* not prohibited */) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
504 |
$maxCoord++; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
505 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
506 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
507 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
508 |
for ($count = 0; $count <= $maxCoord; $count++) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
509 |
$this->_coord[$count] = $reader->getSimilarity()->coord($count, $maxCoord); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
510 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
511 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
512 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
513 |
$score = 0.0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
514 |
$matchedTerms = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
515 |
foreach ($this->_terms as $termId=>$term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
516 |
// Check if term is |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
517 |
if ($this->_signs[$termId] !== false && // not prohibited |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
518 |
isset($this->_termsFreqs[$termId][$docId]) // matched |
|
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 |
$matchedTerms++; |
|
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 |
* We don't need to check that term freq is not 0 |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
524 |
* Score calculation is performed only for matched docs |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
525 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
526 |
$score += |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
527 |
$reader->getSimilarity()->tf($this->_termsFreqs[$termId][$docId]) * |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
528 |
$this->_weights[$termId]->getValue() * |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
529 |
$reader->norm($docId, $term->field); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
530 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
531 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
532 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
533 |
return $score * $this->_coord[$matchedTerms] * $this->getBoost(); |
|
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 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
536 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
537 |
* Execute query in context of index reader |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
538 |
* It also initializes necessary internal structures |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
539 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
540 |
* @param Zend_Search_Lucene_Interface $reader |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
541 |
* @param Zend_Search_Lucene_Index_DocsFilter|null $docsFilter |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
542 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
543 |
public function execute(Zend_Search_Lucene_Interface $reader, $docsFilter = null) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
544 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
545 |
if ($this->_signs === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
546 |
$this->_calculateConjunctionResult($reader); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
547 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
548 |
$this->_calculateNonConjunctionResult($reader); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
549 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
550 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
551 |
// Initialize weight if it's not done yet |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
552 |
$this->_initWeight($reader); |
|
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 |
|
|
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 |
* Get document ids likely matching the query |
|
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 |
* It's an array with document ids as keys (performance considerations) |
|
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 |
* @return array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
561 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
562 |
public function matchedDocs() |
|
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 |
return $this->_resVector; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
565 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
566 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
567 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
568 |
* Score specified document |
|
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 |
* @param integer $docId |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
571 |
* @param Zend_Search_Lucene_Interface $reader |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
572 |
* @return float |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
573 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
574 |
public function score($docId, Zend_Search_Lucene_Interface $reader) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
575 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
576 |
if (isset($this->_resVector[$docId])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
577 |
if ($this->_signs === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
578 |
return $this->_conjunctionScore($docId, $reader); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
579 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
580 |
return $this->_nonConjunctionScore($docId, $reader); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
581 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
582 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
583 |
return 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
584 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
585 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
586 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
587 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
588 |
* Return query terms |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
589 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
590 |
* @return array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
591 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
592 |
public function getQueryTerms() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
593 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
594 |
if ($this->_signs === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
595 |
return $this->_terms; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
596 |
} |
|
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 |
$terms = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
599 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
600 |
foreach ($this->_signs as $id => $sign) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
601 |
if ($sign !== false) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
602 |
$terms[] = $this->_terms[$id]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
603 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
604 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
605 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
606 |
return $terms; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
607 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
608 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
609 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
610 |
* Query specific matches highlighting |
|
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 |
* @param Zend_Search_Lucene_Search_Highlighter_Interface $highlighter Highlighter object (also contains doc for highlighting) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
613 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
614 |
protected function _highlightMatches(Zend_Search_Lucene_Search_Highlighter_Interface $highlighter) |
|
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 |
$words = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
617 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
618 |
if ($this->_signs === null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
619 |
foreach ($this->_terms as $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
620 |
$words[] = $term->text; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
621 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
622 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
623 |
foreach ($this->_signs as $id => $sign) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
624 |
if ($sign !== false) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
625 |
$words[] = $this->_terms[$id]->text; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
626 |
} |
|
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 |
$highlighter->highlight($words); |
|
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 |
* Print a query |
|
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 |
* @return string |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
637 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
638 |
public function __toString() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
639 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
640 |
// It's used only for query visualisation, so we don't care about characters escaping |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
641 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
642 |
$query = ''; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
643 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
644 |
foreach ($this->_terms as $id => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
645 |
if ($id != 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
646 |
$query .= ' '; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
647 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
648 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
649 |
if ($this->_signs === null || $this->_signs[$id] === true) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
650 |
$query .= '+'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
651 |
} else if ($this->_signs[$id] === false) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
652 |
$query .= '-'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
653 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
654 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
655 |
if ($term->field !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
656 |
$query .= $term->field . ':'; |
|
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 |
$query .= $term->text; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
659 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
660 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
661 |
if ($this->getBoost() != 1) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
662 |
$query = '(' . $query . ')^' . round($this->getBoost(), 4); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
663 |
} |
|
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 $query; |
|
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 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
668 |