|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Xparse\RecursivePagination; |
|
4
|
|
|
|
|
5
|
|
|
|
|
6
|
|
|
use Xparse\ElementFinder\ElementFinder; |
|
7
|
|
|
use Xparse\Parser\Parser; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* |
|
11
|
|
|
*/ |
|
12
|
|
|
class RecursivePagination { |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* @var array |
|
16
|
|
|
*/ |
|
17
|
|
|
protected $queue = []; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* @var Parser |
|
21
|
|
|
*/ |
|
22
|
|
|
protected $parser; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var array |
|
26
|
|
|
*/ |
|
27
|
|
|
protected $elementSelector = []; |
|
28
|
|
|
|
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @param Parser $parser |
|
32
|
|
|
* @param string|array $expression |
|
33
|
|
|
* @throws \Exception |
|
34
|
|
|
*/ |
|
35
|
|
|
public function __construct(Parser $parser, $expression) { |
|
36
|
|
|
$this->parser = $parser; |
|
37
|
|
|
|
|
38
|
|
|
$this->setExpression($expression); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* @param array|string $links |
|
44
|
|
|
* @param bool $state |
|
45
|
|
|
* @throws \InvalidArgumentException |
|
46
|
|
|
* @return $this |
|
47
|
|
|
*/ |
|
48
|
|
|
public function addToQueue($links, $state = false) { |
|
49
|
|
|
if (!is_string($links) and !is_array($links)) { |
|
50
|
|
|
throw new \InvalidArgumentException('Links should be an array or a string'); |
|
51
|
|
|
} |
|
52
|
|
|
foreach ((array) $links as $url) { |
|
53
|
|
|
if (!is_string($url)) { |
|
54
|
|
|
throw new \InvalidArgumentException('url should be a string'); |
|
55
|
|
|
} |
|
56
|
|
|
$this->queue[$url] = $state; |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
return $this; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* @return ElementFinder|null |
|
65
|
|
|
* @throws \Exception |
|
66
|
|
|
* @throws \InvalidArgumentException |
|
67
|
|
|
*/ |
|
68
|
|
|
public function getNextPage() { |
|
69
|
|
|
|
|
70
|
|
|
$link = array_search(false, $this->queue, true); |
|
71
|
|
|
if ($link === false) { |
|
72
|
|
|
return null; |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
$this->queue[$link] = true; |
|
76
|
|
|
$page = $this->parser->get($link); |
|
77
|
|
|
|
|
78
|
|
|
if ($page === null) { |
|
79
|
|
|
return null; |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
foreach ($this->elementSelector as $expression => $state) { |
|
83
|
|
|
$queueLinks = $page->value($expression)->all(); |
|
84
|
|
|
$countQueueLinks = count($queueLinks); |
|
85
|
|
|
if ($countQueueLinks > 0) { |
|
86
|
|
|
$queueLinks = array_combine($queueLinks, array_fill(0, $countQueueLinks, false)); |
|
87
|
|
|
$this->queue = array_merge($queueLinks, $this->queue); |
|
88
|
|
|
} |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
return $page; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
|
|
95
|
|
|
/** |
|
96
|
|
|
* @param string|array $expression |
|
97
|
|
|
* @throws \InvalidArgumentException |
|
98
|
|
|
*/ |
|
99
|
|
|
private function setExpression($expression) { |
|
100
|
|
|
|
|
101
|
|
|
if (!is_string($expression) and !is_array($expression)) { |
|
102
|
|
|
throw new \InvalidArgumentException('Invalid expression, should be array or string'); |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
$expression = array_filter((array) $expression); |
|
106
|
|
|
|
|
107
|
|
|
if (count($expression) === 0) { |
|
108
|
|
|
throw new \InvalidArgumentException('Expression might be not empty'); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
foreach ($expression as $path) { |
|
112
|
|
|
if (!is_string($path)) { |
|
113
|
|
|
throw new \InvalidArgumentException('Invalid expression, should be a string'); |
|
114
|
|
|
} |
|
115
|
|
|
$this->elementSelector[$path] = true; |
|
116
|
|
|
} |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
} |
xparse /
RecursivePagination
