| author | Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com> |
| Mon, 19 Dec 2011 00:56:02 +0100 | |
| changeset 428 | 8b51d23b758f |
| parent 68 | ecaf28ffe26e |
| child 207 | 621fa6caec0c |
| permissions | -rw-r--r-- |
|
0
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
1 |
<?php |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
2 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
3 |
* Zend Framework |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
4 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
5 |
* LICENSE |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
6 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
7 |
* This source file is subject to the new BSD license that is bundled |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
8 |
* with this package in the file LICENSE.txt. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
9 |
* It is also available through the world-wide-web at this URL: |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
10 |
* http://framework.zend.com/license/new-bsd |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
11 |
* If you did not receive a copy of the license and are unable to |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
12 |
* obtain it through the world-wide-web, please send an email |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
13 |
* to license@zend.com so we can send you a copy immediately. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
14 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
15 |
* @category Zend |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
16 |
* @package Zend_Search_Lucene |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
17 |
* @subpackage Index |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
18 |
* @copyright Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
19 |
* @license http://framework.zend.com/license/new-bsd New BSD License |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
20 |
* @version $Id: DocumentWriter.php 20096 2010-01-06 02:05:09Z bkarwin $ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
21 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
22 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
23 |
/** Zend_Search_Lucene_Index_SegmentWriter */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
24 |
require_once 'Zend/Search/Lucene/Index/SegmentWriter.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
25 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
26 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
27 |
* @category Zend |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
28 |
* @package Zend_Search_Lucene |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
29 |
* @subpackage Index |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
30 |
* @copyright Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
31 |
* @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
|
32 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
33 |
class Zend_Search_Lucene_Index_SegmentWriter_DocumentWriter extends Zend_Search_Lucene_Index_SegmentWriter |
|
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 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
36 |
* Term Dictionary |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
37 |
* Array of the Zend_Search_Lucene_Index_Term objects |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
38 |
* Corresponding Zend_Search_Lucene_Index_TermInfo object stored in the $_termDictionaryInfos |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
39 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
40 |
* @var array |
|
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 |
protected $_termDictionary; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
43 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
44 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
45 |
* Documents, which contain the term |
|
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 |
* @var array |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
48 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
49 |
protected $_termDocs; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
50 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
51 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
52 |
* Object constructor. |
|
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 |
* @param Zend_Search_Lucene_Storage_Directory $directory |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
55 |
* @param string $name |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
56 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
57 |
public function __construct(Zend_Search_Lucene_Storage_Directory $directory, $name) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
58 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
59 |
parent::__construct($directory, $name); |
|
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 |
$this->_termDocs = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
62 |
$this->_termDictionary = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
63 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
64 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
65 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
66 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
67 |
* Adds a document to this segment. |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
68 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
69 |
* @param Zend_Search_Lucene_Document $document |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
70 |
* @throws Zend_Search_Lucene_Exception |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
71 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
72 |
public function addDocument(Zend_Search_Lucene_Document $document) |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
73 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
74 |
/** Zend_Search_Lucene_Search_Similarity */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
75 |
require_once 'Zend/Search/Lucene/Search/Similarity.php'; |
|
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 |
$storedFields = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
78 |
$docNorms = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
79 |
$similarity = Zend_Search_Lucene_Search_Similarity::getDefault(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
80 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
81 |
foreach ($document->getFieldNames() as $fieldName) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
82 |
$field = $document->getField($fieldName); |
|
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 |
if ($field->storeTermVector) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
85 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
86 |
* @todo term vector storing support |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
87 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
88 |
require_once 'Zend/Search/Lucene/Exception.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
89 |
throw new Zend_Search_Lucene_Exception('Store term vector functionality is not supported yet.'); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
90 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
91 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
92 |
if ($field->isIndexed) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
93 |
if ($field->isTokenized) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
94 |
/** Zend_Search_Lucene_Analysis_Analyzer */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
95 |
require_once 'Zend/Search/Lucene/Analysis/Analyzer.php'; |
|
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 |
$analyzer = Zend_Search_Lucene_Analysis_Analyzer::getDefault(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
98 |
$analyzer->setInput($field->value, $field->encoding); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
99 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
100 |
$position = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
101 |
$tokenCounter = 0; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
102 |
while (($token = $analyzer->nextToken()) !== null) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
103 |
$tokenCounter++; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
104 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
105 |
$term = new Zend_Search_Lucene_Index_Term($token->getTermText(), $field->name); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
106 |
$termKey = $term->key(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
107 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
108 |
if (!isset($this->_termDictionary[$termKey])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
109 |
// New term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
110 |
$this->_termDictionary[$termKey] = $term; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
111 |
$this->_termDocs[$termKey] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
112 |
$this->_termDocs[$termKey][$this->_docCount] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
113 |
} else if (!isset($this->_termDocs[$termKey][$this->_docCount])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
114 |
// Existing term, but new term entry |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
115 |
$this->_termDocs[$termKey][$this->_docCount] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
116 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
117 |
$position += $token->getPositionIncrement(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
118 |
$this->_termDocs[$termKey][$this->_docCount][] = $position; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
119 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
120 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
121 |
if ($tokenCounter == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
122 |
// Field contains empty value. Treat it as non-indexed and non-tokenized |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
123 |
$field = clone($field); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
124 |
$field->isIndexed = $field->isTokenized = false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
125 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
126 |
$docNorms[$field->name] = chr($similarity->encodeNorm( $similarity->lengthNorm($field->name, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
127 |
$tokenCounter)* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
128 |
$document->boost* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
129 |
$field->boost )); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
130 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
131 |
} else if (($fieldUtf8Value = $field->getUtf8Value()) == '') { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
132 |
// Field contains empty value. Treat it as non-indexed and non-tokenized |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
133 |
$field = clone($field); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
134 |
$field->isIndexed = $field->isTokenized = false; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
135 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
136 |
$term = new Zend_Search_Lucene_Index_Term($fieldUtf8Value, $field->name); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
137 |
$termKey = $term->key(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
138 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
139 |
if (!isset($this->_termDictionary[$termKey])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
140 |
// New term |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
141 |
$this->_termDictionary[$termKey] = $term; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
142 |
$this->_termDocs[$termKey] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
143 |
$this->_termDocs[$termKey][$this->_docCount] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
144 |
} else if (!isset($this->_termDocs[$termKey][$this->_docCount])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
145 |
// Existing term, but new term entry |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
146 |
$this->_termDocs[$termKey][$this->_docCount] = array(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
147 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
148 |
$this->_termDocs[$termKey][$this->_docCount][] = 0; // position |
|
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 |
$docNorms[$field->name] = chr($similarity->encodeNorm( $similarity->lengthNorm($field->name, 1)* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
151 |
$document->boost* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
152 |
$field->boost )); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
153 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
154 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
155 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
156 |
if ($field->isStored) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
157 |
$storedFields[] = $field; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
158 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
159 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
160 |
$this->addField($field); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
161 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
162 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
163 |
foreach ($this->_fields as $fieldName => $field) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
164 |
if (!$field->isIndexed) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
165 |
continue; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
166 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
167 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
168 |
if (!isset($this->_norms[$fieldName])) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
169 |
$this->_norms[$fieldName] = str_repeat(chr($similarity->encodeNorm( $similarity->lengthNorm($fieldName, 0) )), |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
170 |
$this->_docCount); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
171 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
172 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
173 |
if (isset($docNorms[$fieldName])){ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
174 |
$this->_norms[$fieldName] .= $docNorms[$fieldName]; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
175 |
} else { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
176 |
$this->_norms[$fieldName] .= chr($similarity->encodeNorm( $similarity->lengthNorm($fieldName, 0) )); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
177 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
178 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
179 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
180 |
$this->addStoredFields($storedFields); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
181 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
182 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
183 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
184 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
185 |
* Dump Term Dictionary (.tis) and Term Dictionary Index (.tii) segment files |
|
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 |
protected function _dumpDictionary() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
188 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
189 |
ksort($this->_termDictionary, SORT_STRING); |
|
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 |
$this->initializeDictionaryFiles(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
192 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
193 |
foreach ($this->_termDictionary as $termId => $term) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
194 |
$this->addTerm($term, $this->_termDocs[$termId]); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
195 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
196 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
197 |
$this->closeDictionaryFiles(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
198 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
199 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
200 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
201 |
/** |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
202 |
* Close segment, write it to disk and return segment info |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
203 |
* |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
204 |
* @return Zend_Search_Lucene_Index_SegmentInfo |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
205 |
*/ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
206 |
public function close() |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
207 |
{ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
208 |
if ($this->_docCount == 0) { |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
209 |
return null; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
210 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
211 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
212 |
$this->_dumpFNM(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
213 |
$this->_dumpDictionary(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
214 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
215 |
$this->_generateCFS(); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
216 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
217 |
/** Zend_Search_Lucene_Index_SegmentInfo */ |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
218 |
require_once 'Zend/Search/Lucene/Index/SegmentInfo.php'; |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
219 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
220 |
return new Zend_Search_Lucene_Index_SegmentInfo($this->_directory, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
221 |
$this->_name, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
222 |
$this->_docCount, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
223 |
-1, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
224 |
null, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
225 |
true, |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
226 |
true); |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
227 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
228 |
|
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
229 |
} |
|
4eba9c11703f
first import
Yves-Marie Haussonne <1218002+ymph@users.noreply.github.com>
parents:
diff
changeset
|
230 |