|
1
|
|
|
<?php declare(strict_types=1); |
|
2
|
|
|
|
|
3
|
|
|
namespace DOMWrap; |
|
4
|
|
|
|
|
5
|
|
|
use DOMWrap\Traits\{ |
|
6
|
|
|
CommonTrait, |
|
7
|
|
|
TraversalTrait, |
|
8
|
|
|
ManipulationTrait |
|
9
|
|
|
}; |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* Document Node |
|
13
|
|
|
* |
|
14
|
|
|
* @package DOMWrap |
|
15
|
|
|
* @license http://opensource.org/licenses/BSD-3-Clause BSD 3 Clause |
|
16
|
|
|
*/ |
|
17
|
|
|
class Document extends \DOMDocument |
|
18
|
|
|
{ |
|
19
|
|
|
use CommonTrait; |
|
20
|
|
|
use TraversalTrait; |
|
21
|
|
|
use ManipulationTrait; |
|
22
|
|
|
|
|
23
|
140 |
|
/** @var int */ |
|
24
|
140 |
|
protected $libxmlOptions = 0; |
|
25
|
|
|
|
|
26
|
140 |
|
public function __construct(string $version = '1.0', string $encoding = 'UTF-8') { |
|
27
|
140 |
|
parent::__construct($version, $encoding); |
|
28
|
140 |
|
|
|
29
|
140 |
|
$this->registerNodeClass('DOMText', 'DOMWrap\\Text'); |
|
30
|
140 |
|
$this->registerNodeClass('DOMElement', 'DOMWrap\\Element'); |
|
31
|
140 |
|
$this->registerNodeClass('DOMComment', 'DOMWrap\\Comment'); |
|
32
|
|
|
$this->registerNodeClass('DOMDocument', 'DOMWrap\\Document'); |
|
33
|
|
|
$this->registerNodeClass('DOMDocumentType', 'DOMWrap\\DocumentType'); |
|
34
|
|
|
$this->registerNodeClass('DOMProcessingInstruction', 'DOMWrap\\ProcessingInstruction'); |
|
35
|
|
|
} |
|
36
|
138 |
|
|
|
37
|
138 |
|
/** |
|
38
|
|
|
* Set libxml options. |
|
39
|
|
|
* |
|
40
|
|
|
* Multiple values must use bitwise OR. |
|
41
|
|
|
* eg: LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD |
|
42
|
|
|
* |
|
43
|
134 |
|
* @link http://php.net/manual/en/libxml.constants.php |
|
44
|
134 |
|
* |
|
45
|
|
|
* @param int $libxmlOptions |
|
46
|
|
|
*/ |
|
47
|
|
|
public function setLibxmlOptions(int $libxmlOptions): void { |
|
48
|
|
|
$this->libxmlOptions = $libxmlOptions; |
|
49
|
|
|
} |
|
50
|
1 |
|
|
|
51
|
1 |
|
/** |
|
52
|
|
|
* {@inheritdoc} |
|
53
|
|
|
*/ |
|
54
|
|
|
public function document(): ?\DOMDocument { |
|
55
|
1 |
|
return $this; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* {@inheritdoc} |
|
60
|
|
|
*/ |
|
61
|
1 |
|
public function collection(): NodeList { |
|
62
|
1 |
|
return $this->newNodeList([$this]); |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* {@inheritdoc} |
|
67
|
|
|
*/ |
|
68
|
1 |
|
public function result(NodeList $nodeList) { |
|
69
|
1 |
|
if ($nodeList->count()) { |
|
70
|
|
|
return $nodeList->first(); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
return null; |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* {@inheritdoc} |
|
78
|
|
|
*/ |
|
79
|
|
|
public function parent() { |
|
80
|
|
|
return null; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
1 |
|
* {@inheritdoc} |
|
85
|
1 |
|
*/ |
|
86
|
|
|
public function parents() { |
|
87
|
|
|
return $this->newNodeList(); |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
2 |
|
* {@inheritdoc} |
|
92
|
2 |
|
*/ |
|
93
|
|
|
public function replaceWith($newNode): self { |
|
94
|
|
|
$this->replaceChild($newNode, $this); |
|
95
|
|
|
|
|
96
|
|
|
return $this; |
|
97
|
|
|
} |
|
98
|
140 |
|
|
|
99
|
140 |
|
/** |
|
100
|
1 |
|
* {@inheritdoc} |
|
101
|
|
|
*/ |
|
102
|
|
|
public function _clone() { |
|
103
|
140 |
|
return null; |
|
104
|
140 |
|
} |
|
105
|
|
|
|
|
106
|
140 |
|
/** |
|
107
|
140 |
|
* {@inheritdoc} |
|
108
|
|
|
*/ |
|
109
|
140 |
|
public function getHtml(): string { |
|
110
|
1 |
|
return $this->getOuterHtml(); |
|
111
|
1 |
|
} |
|
112
|
1 |
|
|
|
113
|
|
|
/** |
|
114
|
|
|
* {@inheritdoc} |
|
115
|
|
|
*/ |
|
116
|
140 |
|
public function setHtml($html): self { |
|
117
|
|
|
if (!is_string($html) || trim($html) == '') { |
|
118
|
|
|
return $this; |
|
119
|
140 |
|
} |
|
120
|
|
|
|
|
121
|
140 |
|
$internalErrors = libxml_use_internal_errors(true); |
|
122
|
140 |
|
$disableEntities = libxml_disable_entity_loader(true); |
|
123
|
|
|
|
|
124
|
140 |
|
$html = $this->convertToUtf8($html); |
|
125
|
|
|
|
|
126
|
|
|
$this->loadHTML($html, $this->libxmlOptions); |
|
127
|
|
|
|
|
128
|
|
|
libxml_use_internal_errors($internalErrors); |
|
129
|
|
|
libxml_disable_entity_loader($disableEntities); |
|
130
|
|
|
|
|
131
|
|
|
return $this; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
/** |
|
135
|
|
|
* @param string $html |
|
136
|
|
|
* @param int $options |
|
137
|
|
|
* @return bool |
|
138
|
|
|
*/ |
|
139
|
|
|
public function loadHTML($html, $options = 0): bool { |
|
140
|
|
|
// Fix LibXML's crazy-ness RE root nodes |
|
141
|
|
|
// While importing HTML using the LIBXML_HTML_NOIMPLIED option LibXML insists |
|
142
|
|
|
// on having one root node. All subsequent nodes are appended to this first node. |
|
143
|
|
|
// To counter this we will create a fake element, allow LibXML to 'do its thing' |
|
144
|
|
|
// then undo it by taking the contents of the fake element, placing it back into |
|
145
|
|
|
// the root and then remove our fake element. |
|
146
|
|
|
if ($options & LIBXML_HTML_NOIMPLIED) { |
|
147
|
|
|
$html = '<domwrap></domwrap>' . $html; |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
$result = parent::loadHTML($html, $options); |
|
151
|
|
|
|
|
152
|
|
|
// Do our re-shuffling of nodes. |
|
153
|
|
|
if ($this->libxmlOptions & LIBXML_HTML_NOIMPLIED) { |
|
154
|
|
|
$this->children()->first()->contents()->each(function($node){ |
|
155
|
|
|
$this->append($node); |
|
156
|
|
|
}); |
|
157
|
|
|
|
|
158
|
|
|
$this->removeChild($this->children()->first()); |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
return $result; |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
private function getCharset(string $html): ?string { |
|
165
|
|
|
$charset = null; |
|
166
|
|
|
|
|
167
|
|
|
if (preg_match('@<meta.*?charset=["]?([^"\s]+)@im', $html, $matches)) { |
|
168
|
|
|
$charset = strtoupper($matches[1]); |
|
169
|
|
|
} |
|
170
|
|
|
|
|
171
|
|
|
return $charset; |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
|
|
private function convertToUtf8(string $html): string { |
|
175
|
|
|
if (mb_detect_encoding($html, mb_detect_order(), true) === 'UTF-8') { |
|
176
|
|
|
return $html; |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
$charset = $this->getCharset($html); |
|
180
|
|
|
|
|
181
|
|
|
if ($charset !== null) { |
|
182
|
|
|
$html = preg_replace('@(charset=["]?)([^"\s]+)([^"]*["]?)@im', '$1UTF-8$3', $html); |
|
183
|
|
|
$mbHasCharset = in_array($charset, array_map('strtoupper', mb_list_encodings())); |
|
184
|
|
|
|
|
185
|
|
|
if ($mbHasCharset) { |
|
186
|
|
|
$html = mb_convert_encoding($html, 'UTF-8', $charset); |
|
187
|
|
|
|
|
188
|
|
|
// Fallback to iconv if available. |
|
189
|
|
|
} elseif (extension_loaded('iconv')) { |
|
190
|
|
|
$htmlIconv = iconv($charset, 'UTF-8', $html); |
|
191
|
|
|
|
|
192
|
|
|
if ($htmlIconv !== false) { |
|
193
|
|
|
$html = $htmlIconv; |
|
194
|
|
|
} else { |
|
195
|
|
|
$charset = null; |
|
196
|
|
|
} |
|
197
|
|
|
} |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
if ($charset === null) { |
|
201
|
|
|
$html = mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'); |
|
202
|
|
|
} |
|
203
|
|
|
|
|
204
|
|
|
return $html; |
|
205
|
|
|
} |
|
206
|
|
|
} |
|
207
|
|
|
|
scotteh /
php-dom-wrapper
