|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Bavix\AdvancedHtmlDom; |
|
4
|
|
|
|
|
5
|
|
|
use Bavix\AdvancedHtmlDom\CacheSystem\InterfaceCache; |
|
6
|
|
|
|
|
7
|
|
|
class AdvancedHtmlDom extends AdvancedHtmlBase |
|
8
|
|
|
{ |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* @var |
|
12
|
|
|
*/ |
|
13
|
|
|
public $xpath; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* @var |
|
17
|
|
|
*/ |
|
18
|
|
|
public $root; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* @var InterfaceCache |
|
22
|
|
|
*/ |
|
23
|
|
|
protected $cache; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* AdvancedHtmlDom constructor. |
|
27
|
|
|
* |
|
28
|
|
|
* @param null $html |
|
|
|
|
|
|
29
|
|
|
* @param bool $is_xml |
|
30
|
|
|
*/ |
|
31
|
4 |
|
public function __construct($html = null, $is_xml = false) |
|
32
|
|
|
{ |
|
33
|
4 |
|
$this->doc = $this; |
|
34
|
4 |
|
if ($html) |
|
35
|
|
|
{ |
|
36
|
1 |
|
$this->load($html, $is_xml); |
|
37
|
|
|
} |
|
38
|
4 |
|
} |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* @inheritdoc |
|
42
|
|
|
*/ |
|
43
|
1 |
|
public function __destruct() |
|
44
|
|
|
{ |
|
45
|
1 |
|
$this->xpath = $this->root = null; |
|
46
|
1 |
|
unset($this->xpath, $this->root); |
|
47
|
1 |
|
parent::__destruct(); |
|
48
|
1 |
|
} |
|
49
|
|
|
|
|
50
|
|
|
/** |
|
51
|
|
|
* @param InterfaceCache $cache |
|
52
|
|
|
*/ |
|
53
|
|
|
public function setCache(InterfaceCache $cache) |
|
54
|
|
|
{ |
|
55
|
|
|
$this->cache = $cache; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @param string $url |
|
60
|
|
|
* |
|
61
|
|
|
* @return mixed |
|
62
|
|
|
*/ |
|
63
|
3 |
|
public function cache($url) |
|
64
|
|
|
{ |
|
65
|
3 |
|
if (!$this->cache) |
|
66
|
|
|
{ |
|
67
|
3 |
|
return \file_get_contents($url); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
return $this->cache->get($url); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* @param $html |
|
75
|
|
|
* @param bool $is_xml |
|
76
|
|
|
*/ |
|
77
|
4 |
|
public function load($html, $is_xml = false) |
|
78
|
|
|
{ |
|
79
|
4 |
|
$this->dom = new \DOMDocument(); |
|
80
|
4 |
|
if ($is_xml) |
|
81
|
|
|
{ |
|
82
|
|
|
$html = \preg_replace('/xmlns=".*?"/ ', '', $html); |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
4 |
|
$this->dom->loadHTML($html, LIBXML_NOERROR); |
|
86
|
4 |
|
$this->xpath = new \DOMXPath($this->dom); |
|
87
|
|
|
//$this->root = new AHTMLNode($this->dom->documentElement, $this->doc); |
|
|
|
|
|
|
88
|
4 |
|
$this->root = $this->at('body'); |
|
|
|
|
|
|
89
|
4 |
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @param $file |
|
93
|
|
|
* @param bool $is_xml |
|
94
|
|
|
* |
|
95
|
|
|
* @deprecated loadFile |
|
96
|
|
|
*/ |
|
97
|
|
|
public function load_file($file, $is_xml = false) |
|
98
|
|
|
{ |
|
99
|
|
|
$this->loadFile($file, $is_xml); |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
/** |
|
103
|
|
|
* @param string $file |
|
104
|
|
|
* @param bool $is_xml |
|
105
|
|
|
* |
|
106
|
|
|
* @return $this |
|
107
|
|
|
*/ |
|
108
|
3 |
|
public function loadFile($file, $is_xml = false) |
|
109
|
|
|
{ |
|
110
|
3 |
|
$this->load($this->cache($file), $is_xml); |
|
111
|
|
|
|
|
112
|
3 |
|
return $this; |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
// special cases |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* @return mixed |
|
119
|
|
|
*/ |
|
120
|
|
|
public function text() |
|
121
|
|
|
{ |
|
122
|
|
|
return $this->root->text; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
/** |
|
126
|
|
|
* @return mixed |
|
127
|
|
|
*/ |
|
128
|
|
|
public function title() |
|
129
|
|
|
{ |
|
130
|
|
|
return $this->at('title')->text(); |
|
|
|
|
|
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
} |
|
134
|
|
|
|
bavix /
advanced-html-dom
Loading history...