|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
/* For licensing terms, see /license.txt */ |
|
6
|
|
|
|
|
7
|
|
|
namespace Chamilo\CoreBundle\Search\Xapian; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* Service responsible for indexing documents into the Xapian database. |
|
11
|
|
|
*/ |
|
12
|
|
|
final class XapianIndexService |
|
13
|
|
|
{ |
|
14
|
|
|
private const DEFAULT_LANGUAGE = 'english'; |
|
15
|
|
|
|
|
16
|
|
|
public function __construct( |
|
17
|
|
|
private readonly SearchIndexPathResolver $indexPathResolver, |
|
18
|
|
|
) { |
|
19
|
|
|
} |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* Indexes a simple demo document so we can verify that search works end-to-end. |
|
23
|
|
|
* |
|
24
|
|
|
* @return int The Xapian internal document id |
|
25
|
|
|
* |
|
26
|
|
|
* @throws \RuntimeException When indexing fails |
|
27
|
|
|
*/ |
|
28
|
|
|
public function indexDemoDocument(): int |
|
29
|
|
|
{ |
|
30
|
|
|
$now = new \DateTimeImmutable('now'); |
|
31
|
|
|
|
|
32
|
|
|
$fields = [ |
|
33
|
|
|
'title' => 'Demo test document', |
|
34
|
|
|
'content' => 'This is a test document indexed from XapianIndexService in Chamilo 2.', |
|
35
|
|
|
'created_at' => $now->format(\DATE_ATOM), |
|
36
|
|
|
]; |
|
37
|
|
|
|
|
38
|
|
|
$terms = [ |
|
39
|
|
|
'XTdemo', |
|
40
|
|
|
'XTchamilo', |
|
41
|
|
|
]; |
|
42
|
|
|
|
|
43
|
|
|
return $this->indexDocument($fields, $terms); |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Indexes a generic document. |
|
48
|
|
|
* |
|
49
|
|
|
* @param array<string,mixed> $fields Arbitrary data to store and index as free-text |
|
50
|
|
|
* @param string[] $terms Optional list of additional terms to add to the document |
|
51
|
|
|
* @param string|null $language Language used for stemming (defaults to english) |
|
52
|
|
|
* |
|
53
|
|
|
* @return int The Xapian internal document id |
|
54
|
|
|
* |
|
55
|
|
|
* @throws \RuntimeException When Xapian fails during indexing |
|
56
|
|
|
*/ |
|
57
|
|
|
public function indexDocument( |
|
58
|
|
|
array $fields, |
|
59
|
|
|
array $terms = [], |
|
60
|
|
|
?string $language = null, |
|
61
|
|
|
): int { |
|
62
|
|
|
if (!\class_exists(\XapianWritableDatabase::class)) { |
|
63
|
|
|
throw new \RuntimeException('Xapian PHP extension is not loaded.'); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
$db = $this->openWritableDatabase(); |
|
67
|
|
|
|
|
68
|
|
|
$doc = new \XapianDocument(); |
|
69
|
|
|
$termGen = new \XapianTermGenerator(); |
|
70
|
|
|
|
|
71
|
|
|
$lang = $language ?: self::DEFAULT_LANGUAGE; |
|
72
|
|
|
$stemmer = new \XapianStem($lang); |
|
73
|
|
|
|
|
74
|
|
|
$termGen->set_stemmer($stemmer); |
|
75
|
|
|
$termGen->set_document($doc); |
|
76
|
|
|
|
|
77
|
|
|
// Index all field values as free-text (title, content, etc.) |
|
78
|
|
|
foreach ($fields as $value) { |
|
79
|
|
|
if ($value === null) { |
|
80
|
|
|
continue; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
if (!\is_string($value)) { |
|
84
|
|
|
$value = (string) $value; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
$termGen->index_text($value, 1); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
// Add explicit terms if provided |
|
91
|
|
|
foreach ($terms as $term) { |
|
92
|
|
|
$term = (string) $term; |
|
93
|
|
|
if ($term === '') { |
|
94
|
|
|
continue; |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
$doc->add_term($term, 1); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
// Store fields as serialized payload (compatible with the search service decode) |
|
101
|
|
|
$doc->set_data(\serialize($fields)); |
|
102
|
|
|
|
|
103
|
|
|
try { |
|
104
|
|
|
$docId = $db->add_document($doc); |
|
105
|
|
|
$db->flush(); |
|
106
|
|
|
|
|
107
|
|
|
error_log('[Xapian] XapianIndexService::indexDocument: document added with docId=' |
|
108
|
|
|
.var_export($docId, true) |
|
109
|
|
|
); |
|
110
|
|
|
|
|
111
|
|
|
return $docId; |
|
112
|
|
|
} catch (\Throwable $e) { |
|
113
|
|
|
throw new \RuntimeException( |
|
114
|
|
|
\sprintf('Failed to index document in Xapian: %s', $e->getMessage()), |
|
115
|
|
|
0, |
|
116
|
|
|
$e |
|
117
|
|
|
); |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
/** |
|
122
|
|
|
* Deletes a document from the Xapian index using its internal document id. |
|
123
|
|
|
* |
|
124
|
|
|
* @throws \RuntimeException When Xapian fails during deletion |
|
125
|
|
|
*/ |
|
126
|
|
|
public function deleteDocument(int $docId): void |
|
127
|
|
|
{ |
|
128
|
|
|
if (!\class_exists(\XapianWritableDatabase::class)) { |
|
129
|
|
|
throw new \RuntimeException('Xapian PHP extension is not loaded.'); |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
$db = $this->openWritableDatabase(); |
|
133
|
|
|
|
|
134
|
|
|
try { |
|
135
|
|
|
error_log('[Xapian] XapianIndexService::deleteDocument: deleting docId=' |
|
136
|
|
|
.var_export($docId, true) |
|
137
|
|
|
); |
|
138
|
|
|
|
|
139
|
|
|
$db->delete_document($docId); |
|
140
|
|
|
$db->flush(); |
|
141
|
|
|
} catch (\Throwable $e) { |
|
142
|
|
|
throw new \RuntimeException( |
|
143
|
|
|
\sprintf('Failed to delete document in Xapian: %s', $e->getMessage()), |
|
144
|
|
|
0, |
|
145
|
|
|
$e |
|
146
|
|
|
); |
|
147
|
|
|
} |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
/** |
|
151
|
|
|
* Opens the writable Xapian database using DB_CREATE_OR_OPEN. |
|
152
|
|
|
*/ |
|
153
|
|
|
private function openWritableDatabase(): \XapianWritableDatabase |
|
154
|
|
|
{ |
|
155
|
|
|
$indexDir = $this->indexPathResolver->getIndexDir(); |
|
156
|
|
|
|
|
157
|
|
|
return new \XapianWritableDatabase($indexDir, \Xapian::DB_CREATE_OR_OPEN); |
|
158
|
|
|
} |
|
159
|
|
|
} |
|
160
|
|
|
|
chamilo /
chamilo-lms
