|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare( strict_types = 1 ); |
|
4
|
|
|
|
|
5
|
|
|
namespace Wikibase\JsonDumpReader; |
|
6
|
|
|
|
|
7
|
|
|
use Deserializers\Deserializer; |
|
8
|
|
|
use Iterator; |
|
9
|
|
|
use Wikibase\JsonDumpReader\Iterator\EntityDumpIterator; |
|
10
|
|
|
use Wikibase\JsonDumpReader\Iterator\ObjectDumpIterator; |
|
11
|
|
|
use Wikibase\JsonDumpReader\Reader\Bz2DumpReader; |
|
12
|
|
|
use Wikibase\JsonDumpReader\Reader\ExtractedDumpReader; |
|
13
|
|
|
use Wikibase\JsonDumpReader\Reader\GzDumpReader; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Package public |
|
17
|
|
|
* @since 1.0.0 |
|
18
|
|
|
* |
|
19
|
|
|
* @licence GNU GPL v2+ |
|
20
|
|
|
* @author Jeroen De Dauw < [email protected] > |
|
21
|
|
|
*/ |
|
22
|
|
|
class JsonDumpFactory { |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* Creates a DumpReader that can read lines from a bz2 compressed JSON dump. |
|
26
|
|
|
* @since 1.1.0 |
|
27
|
|
|
* |
|
28
|
|
|
* @param string $dumpFilePath |
|
29
|
|
|
* @param int $initialPosition since 1.3.0 |
|
30
|
|
|
* |
|
31
|
|
|
* @return SeekableDumpReader |
|
32
|
|
|
*/ |
|
33
|
3 |
|
public function newGzDumpReader( string $dumpFilePath, int $initialPosition = 0 ): SeekableDumpReader { |
|
34
|
3 |
|
return new GzDumpReader( $dumpFilePath, $initialPosition ); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* Creates a DumpReader that can read lines from an uncompressed JSON dump. |
|
39
|
|
|
* @since 1.0.0 |
|
40
|
|
|
* |
|
41
|
|
|
* @param string $dumpFilePath |
|
42
|
|
|
* @param int $initialPosition |
|
43
|
|
|
* |
|
44
|
|
|
* @return SeekableDumpReader |
|
45
|
|
|
*/ |
|
46
|
1 |
|
public function newExtractedDumpReader( string $dumpFilePath, int $initialPosition = 0 ): SeekableDumpReader { |
|
47
|
1 |
|
return new ExtractedDumpReader( $dumpFilePath, $initialPosition ); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* Creates a DumpReader that can read lines from a bz2 compressed JSON dump. |
|
52
|
|
|
* @since 1.0.0 |
|
53
|
|
|
* |
|
54
|
|
|
* @param string $dumpFilePath |
|
55
|
|
|
* |
|
56
|
|
|
* @return DumpReader |
|
57
|
|
|
*/ |
|
58
|
2 |
|
public function newBz2DumpReader( string $dumpFilePath ): DumpReader { |
|
59
|
2 |
|
return new Bz2DumpReader( $dumpFilePath ); |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* Creates an Iterator over each JSON serialized Entity in the dump. |
|
64
|
|
|
* @since 1.0.0 |
|
65
|
|
|
* |
|
66
|
|
|
* @param DumpReader $dumpReader |
|
67
|
|
|
* @param callable $onError Gets called with a single string parameter on error |
|
68
|
|
|
* |
|
69
|
|
|
* @return Iterator string[] |
|
70
|
|
|
*/ |
|
71
|
16 |
|
public function newStringDumpIterator( DumpReader $dumpReader, callable $onError = null ): Iterator { |
|
72
|
|
|
$iterator = new \RewindableGenerator( function() use ( $dumpReader, $onError ) { |
|
73
|
16 |
|
while ( true ) { |
|
74
|
|
|
try { |
|
75
|
16 |
|
$line = $dumpReader->nextJsonLine(); |
|
76
|
|
|
|
|
77
|
16 |
|
if ( $line === null ) { |
|
78
|
13 |
|
return; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
15 |
|
yield $line; |
|
82
|
|
|
} |
|
83
|
|
|
catch ( DumpReadingException $ex ) { |
|
84
|
|
|
if ( $onError !== null ) { |
|
85
|
|
|
call_user_func( $onError, $ex->getMessage() ); |
|
86
|
|
|
} |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
16 |
|
} ); |
|
90
|
|
|
|
|
91
|
|
|
$iterator->onRewind( function() use ( $dumpReader ) { |
|
92
|
16 |
|
$dumpReader->rewind(); |
|
93
|
16 |
|
} ); |
|
94
|
|
|
|
|
95
|
16 |
|
return $iterator; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Creates an Iterator over each Entity in the dump as PHP array/object in the JSON format. |
|
100
|
|
|
* This is essentially a json_decode map of the string dump iterator. |
|
101
|
|
|
* @since 1.0.0 |
|
102
|
|
|
* |
|
103
|
|
|
* @param DumpReader $dumpReader |
|
104
|
|
|
* @param callable $onError Gets called with a single string parameter on error |
|
105
|
|
|
* |
|
106
|
|
|
* @return Iterator array[] |
|
107
|
|
|
*/ |
|
108
|
14 |
|
public function newObjectDumpIterator( DumpReader $dumpReader, callable $onError = null ): Iterator { |
|
109
|
14 |
|
$iterator = new ObjectDumpIterator( |
|
110
|
14 |
|
$this->newStringDumpIterator( $dumpReader, $onError ) |
|
111
|
|
|
); |
|
112
|
|
|
|
|
113
|
14 |
|
$iterator->onError( $onError ); |
|
114
|
|
|
|
|
115
|
14 |
|
return $iterator; |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
/** |
|
119
|
|
|
* Creates an Iterator over each Entity in the dump, fully deserialized as EntityDocument. |
|
120
|
|
|
* @since 1.0.0 |
|
121
|
|
|
* |
|
122
|
|
|
* @param DumpReader $dumpReader |
|
123
|
|
|
* @param Deserializer $entityDeserializer |
|
124
|
|
|
* @param callable $onError Gets called with a single string parameter on error |
|
125
|
|
|
* |
|
126
|
|
|
* @return Iterator EntityDocument[] |
|
127
|
|
|
*/ |
|
128
|
8 |
|
public function newEntityDumpIterator( DumpReader $dumpReader, Deserializer $entityDeserializer, |
|
129
|
|
|
callable $onError = null ): Iterator { |
|
130
|
|
|
|
|
131
|
8 |
|
$iterator = new EntityDumpIterator( |
|
132
|
8 |
|
$this->newObjectDumpIterator( $dumpReader, $onError ), |
|
133
|
|
|
$entityDeserializer |
|
134
|
|
|
); |
|
135
|
|
|
|
|
136
|
8 |
|
$iterator->onError( $onError ); |
|
137
|
|
|
|
|
138
|
8 |
|
return $iterator; |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
} |
JeroenDeDauw /
JsonDumpReader
