|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Queryr\DumpReader\XmlReader; |
|
4
|
|
|
|
|
5
|
|
|
use Queryr\DumpReader\DumpReader; |
|
6
|
|
|
use Queryr\DumpReader\DumpReaderException; |
|
7
|
|
|
use Queryr\DumpReader\Page; |
|
8
|
|
|
use XMLReader; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* @licence GNU GPL v2+ |
|
12
|
|
|
* @author Jeroen De Dauw < [email protected] > |
|
13
|
|
|
*/ |
|
14
|
|
|
class DumpXmlReader extends DumpReader { |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* @var XMLReader |
|
18
|
|
|
*/ |
|
19
|
|
|
private $xmlReader; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* @var string |
|
23
|
|
|
*/ |
|
24
|
|
|
private $dumpFile; |
|
25
|
|
|
|
|
26
|
|
|
public function __construct( $dumpFile ) { |
|
27
|
|
|
$this->dumpFile = $dumpFile; |
|
28
|
|
|
|
|
29
|
|
|
$this->initReader(); |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
private function initReader() { |
|
33
|
|
|
$this->xmlReader = new XMLReader(); |
|
34
|
|
|
$this->xmlReader->open( $this->dumpFile ); |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
public function __destruct() { |
|
38
|
|
|
$this->closeReader(); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
private function closeReader() { |
|
42
|
|
|
$this->xmlReader->close(); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* @see DumpReader::rewind |
|
47
|
|
|
*/ |
|
48
|
|
|
public function rewind() { |
|
49
|
|
|
$this->closeReader(); |
|
50
|
|
|
$this->initReader(); |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* @see DumpReader::nextEntityPage |
|
55
|
|
|
* |
|
56
|
|
|
* @return Page|null |
|
57
|
|
|
* @throws DumpReaderException |
|
58
|
|
|
*/ |
|
59
|
|
|
public function nextEntityPage() { |
|
60
|
|
|
do { |
|
61
|
|
|
$page = $this->nextPage(); |
|
62
|
|
|
|
|
63
|
|
|
if ( $page === null ) { |
|
64
|
|
|
return null; |
|
65
|
|
|
} |
|
66
|
|
|
} while ( !$page->getRevision()->hasEntityModel() ); |
|
67
|
|
|
|
|
68
|
|
|
return $page; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* @return Page|null |
|
73
|
|
|
*/ |
|
74
|
|
|
private function nextPage() { |
|
75
|
|
|
while ( !$this->isPageNode() ) { |
|
76
|
|
|
$this->xmlReader->read(); |
|
77
|
|
|
|
|
78
|
|
|
if ( $this->xmlReader->nodeType === XMLReader::NONE ) { |
|
79
|
|
|
return null; |
|
80
|
|
|
} |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
$pageNode = new PageNode( $this->xmlReader->expand() ); |
|
84
|
|
|
|
|
85
|
|
|
$page = $pageNode->asPage(); |
|
86
|
|
|
$this->xmlReader->next(); |
|
87
|
|
|
return $page; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
private function isPageNode() { |
|
91
|
|
|
return $this->xmlReader->nodeType === XMLReader::ELEMENT && $this->xmlReader->name === 'page'; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
} |
|
95
|
|
|
|
JeroenDeDauw /
XmlDumpReader
