|
1
|
|
|
<?php |
|
2
|
|
|
declare(strict_types=1); |
|
3
|
|
|
|
|
4
|
|
|
namespace Rarst\Hugo\wprss2hugo; |
|
5
|
|
|
|
|
6
|
|
|
use Prewk\XmlStringStreamer; |
|
7
|
|
|
use Prewk\XmlStringStreamer\Parser\StringWalker; |
|
8
|
|
|
use Prewk\XmlStringStreamer\Stream\File; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Incremental wrapper for the XML export file. |
|
12
|
|
|
*/ |
|
13
|
|
|
class Export implements \Iterator |
|
14
|
|
|
{ |
|
15
|
|
|
/** @var XmlStringStreamer */ |
|
16
|
|
|
private $streamer; |
|
17
|
|
|
|
|
18
|
|
|
/** @var string */ |
|
19
|
|
|
private $currentNode; |
|
20
|
|
|
|
|
21
|
|
|
/** @var int */ |
|
22
|
|
|
private $size; |
|
23
|
|
|
|
|
24
|
|
|
/** @var int */ |
|
25
|
|
|
private $bytes = 0; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Set up for a given export file path. |
|
29
|
|
|
*/ |
|
30
|
|
|
public function __construct(string $file) |
|
31
|
|
|
{ |
|
32
|
|
|
$this->size = filesize($file); |
|
33
|
|
|
$stream = new File($file, 16384, function (string $chunk, int $readBytes) { |
|
34
|
|
|
$this->bytes = $readBytes; |
|
35
|
|
|
}); |
|
36
|
|
|
$parser = new StringWalker([ |
|
37
|
|
|
'captureDepth' => 3, |
|
38
|
|
|
'expectGT' => true, |
|
39
|
|
|
]); |
|
40
|
|
|
$this->streamer = new XmlStringStreamer($parser, $stream); |
|
41
|
|
|
|
|
42
|
|
|
$this->next(); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* File read progress in percent. |
|
47
|
|
|
*/ |
|
48
|
|
|
public function progress(): int |
|
49
|
|
|
{ |
|
50
|
|
|
return (int)round($this->bytes / $this->size * 100); |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
/** |
|
54
|
|
|
* Current XML node. |
|
55
|
|
|
*/ |
|
56
|
|
|
public function current() |
|
57
|
|
|
{ |
|
58
|
|
|
return $this->transform($this->currentNode); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* Convert XML string to an element object. |
|
63
|
|
|
*/ |
|
64
|
|
|
private function transform(string $xml): \SimpleXMLElement |
|
65
|
|
|
{ |
|
66
|
|
|
return simplexml_load_string($this->removeNamespaces($xml)); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* Remove namespaces from partial XML string for simplicity of processing. |
|
71
|
|
|
*/ |
|
72
|
|
|
private function removeNamespaces(string $xml): string |
|
73
|
|
|
{ |
|
74
|
|
|
$xml = strtr($xml, [ |
|
75
|
|
|
'<content:encoded>' => '<content>', |
|
76
|
|
|
'</content:encoded>' => '</content>', |
|
77
|
|
|
'<excerpt:encoded>' => '<excerpt>', |
|
78
|
|
|
'</excerpt:encoded>' => '</excerpt>', |
|
79
|
|
|
]); |
|
80
|
|
|
|
|
81
|
|
|
$xml = preg_replace('~<([/]?)[[:alpha:]_]+?:([[:alpha:]_]+?)([/]?)>~', '<$1$2$3>', $xml); |
|
82
|
|
|
|
|
83
|
|
|
return $xml; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
public function next() |
|
87
|
|
|
{ |
|
88
|
|
|
$this->currentNode = (string)$this->streamer->getNode(); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* Iterator key (not applicable). |
|
93
|
|
|
*/ |
|
94
|
|
|
public function key() |
|
95
|
|
|
{ |
|
96
|
|
|
return null; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* If got valid node iteration. |
|
101
|
|
|
*/ |
|
102
|
|
|
public function valid() |
|
103
|
|
|
{ |
|
104
|
|
|
return '' !== $this->currentNode; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
public function rewind() |
|
108
|
|
|
{ |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
Rarst /
wprss2hugo
