|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Wikibase\EntityStore\Internal; |
|
4
|
|
|
|
|
5
|
|
|
use Deserializers\Deserializer; |
|
6
|
|
|
use Deserializers\Exceptions\DeserializationException; |
|
7
|
|
|
use Iterator; |
|
8
|
|
|
use Psr\Log\LoggerInterface; |
|
9
|
|
|
use Wikibase\DataModel\Entity\EntityDocument; |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* Internal class |
|
13
|
|
|
|
|
14
|
|
|
* @licence GPLv2+ |
|
15
|
|
|
* @author Thomas Pellissier Tanon |
|
16
|
|
|
*/ |
|
17
|
|
|
class IncrementalXmlDumpReader implements Iterator { |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* @var resource |
|
21
|
|
|
*/ |
|
22
|
|
|
private $fileStream; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var Deserializer |
|
26
|
|
|
*/ |
|
27
|
|
|
private $entityDeserializer; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @var EntityDocument|null |
|
31
|
|
|
*/ |
|
32
|
|
|
private $currentEntity = null; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* @var LoggerInterface |
|
36
|
|
|
*/ |
|
37
|
|
|
private $logger; |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @param string $fileName |
|
41
|
|
|
* @param Deserializer $entityDeserializer |
|
42
|
|
|
* @param LoggerInterface $logger |
|
43
|
|
|
*/ |
|
44
|
2 |
|
public function __construct( $fileName, Deserializer $entityDeserializer, LoggerInterface $logger ) { |
|
45
|
2 |
|
$this->fileStream = fopen( $fileName, 'r' ); |
|
46
|
2 |
|
$this->entityDeserializer = $entityDeserializer; |
|
47
|
2 |
|
$this->logger = $logger; |
|
48
|
2 |
|
} |
|
49
|
|
|
|
|
50
|
2 |
|
public function __destruct() { |
|
51
|
2 |
|
fclose( $this->fileStream ); |
|
52
|
2 |
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* @see Iterator::current |
|
56
|
|
|
*/ |
|
57
|
2 |
|
public function current() { |
|
58
|
2 |
|
return $this->currentEntity; |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* @see Iterator::next |
|
63
|
|
|
*/ |
|
64
|
2 |
|
public function next() { |
|
65
|
2 |
|
$this->currentEntity = null; |
|
66
|
|
|
|
|
67
|
2 |
|
$isEntity = false; |
|
68
|
2 |
|
$text = ''; |
|
69
|
2 |
|
while( true ) { |
|
70
|
2 |
|
$line = fgets( $this->fileStream ); |
|
71
|
|
|
|
|
72
|
2 |
|
if( $line === false ) { |
|
73
|
2 |
|
return; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
2 |
|
if( preg_match( '/<model>wikibase-[a-z]+<\/model>/', $line ) ) { |
|
77
|
2 |
|
$isEntity = true; |
|
78
|
2 |
|
} elseif( preg_match( '/<text.*>(.*)<\/text>/', $line, $m ) ) { |
|
79
|
2 |
|
$text = $m[1]; |
|
80
|
2 |
|
} elseif( preg_match( '/<\/page>/', $line, $m ) && $isEntity ) { |
|
81
|
|
|
|
|
82
|
2 |
|
$json = json_decode( html_entity_decode( $text ), true ); |
|
83
|
2 |
|
if( $json === null ) { |
|
84
|
2 |
|
$this->logger->error( 'The serialization is not a valid JSON document.' ); |
|
85
|
2 |
|
continue; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
2 |
|
if( array_key_exists( 'redirect', $json ) ) { |
|
89
|
2 |
|
$this->logger->info( 'Entity redirection not supported.' ); |
|
90
|
2 |
|
continue; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
try { |
|
94
|
2 |
|
$this->currentEntity = $this->entityDeserializer->deserialize( $json ); |
|
95
|
2 |
|
return; |
|
96
|
|
|
} catch( DeserializationException $e ) { |
|
97
|
|
|
$id = array_key_exists( 'id', $json ) ? $json['id'] : ''; |
|
98
|
|
|
$this->logger->error( 'Deserialization of entity ' . $id . ' failed: ' . $e->getMessage() ); |
|
99
|
|
|
} |
|
100
|
|
|
} |
|
101
|
2 |
|
} |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
|
|
/** |
|
105
|
|
|
* @see Iterator::key |
|
106
|
|
|
*/ |
|
107
|
|
|
public function key() { |
|
108
|
|
|
if ( $this->currentEntity === null ) { |
|
109
|
|
|
return null; |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
return $this->currentEntity->getId()->getSerialization(); |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
/** |
|
116
|
|
|
* @see Iterator::valid |
|
117
|
|
|
*/ |
|
118
|
2 |
|
public function valid() { |
|
119
|
2 |
|
return $this->currentEntity !== null; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
/** |
|
123
|
|
|
* @see Iterator::rewind |
|
124
|
|
|
*/ |
|
125
|
2 |
|
public function rewind() { |
|
126
|
2 |
|
fseek( $this->fileStream, 0 ); |
|
127
|
2 |
|
$this->next(); |
|
128
|
2 |
|
} |
|
129
|
|
|
} |
|
130
|
|
|
|
ProjetPP /
WikibaseEntityStore
