|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Bavix\AdvancedHtmlDom; |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* Class AHTMLNode |
|
7
|
|
|
* |
|
8
|
|
|
* @package Bavix\AdvancedHtmlDom |
|
9
|
|
|
* |
|
10
|
|
|
* @property-read string $clean_text |
|
11
|
|
|
*/ |
|
12
|
|
|
class AHTMLNode extends AdvancedHtmlBase implements \ArrayAccess |
|
13
|
|
|
{ |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* @var |
|
17
|
|
|
*/ |
|
18
|
|
|
private $_path; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* AHTMLNode constructor. |
|
22
|
|
|
* |
|
23
|
|
|
* @param $node |
|
24
|
|
|
* @param $doc |
|
25
|
|
|
*/ |
|
26
|
4 |
|
public function __construct($node, $doc) |
|
27
|
|
|
{ |
|
28
|
4 |
|
$this->node = $node; |
|
29
|
4 |
|
$this->_path = $node->getNodePath(); |
|
30
|
4 |
|
$this->doc = $doc; |
|
31
|
4 |
|
$this->is_text = $node->nodeName === '#text'; |
|
32
|
4 |
|
} |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* @inheritdoc |
|
36
|
|
|
*/ |
|
37
|
2 |
|
public function __destruct() |
|
38
|
|
|
{ |
|
39
|
2 |
|
$this->_path = null; |
|
40
|
2 |
|
unset($this->_path); |
|
41
|
2 |
|
parent::__destruct(); |
|
42
|
2 |
|
} |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* @param $html |
|
46
|
|
|
* |
|
47
|
|
|
* @return mixed |
|
48
|
|
|
*/ |
|
49
|
|
|
private function get_fragment($html) |
|
50
|
|
|
{ |
|
51
|
|
|
$dom = $this->doc->dom; |
|
52
|
|
|
$fragment = $dom->createDocumentFragment() or die('nope'); |
|
|
|
|
|
|
53
|
|
|
$fragment->appendXML($html); |
|
54
|
|
|
|
|
55
|
|
|
return $fragment; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @param $html |
|
60
|
|
|
* |
|
61
|
|
|
* @return AHTMLNode|null |
|
62
|
|
|
*/ |
|
63
|
|
|
public function replace($html) |
|
64
|
|
|
{ |
|
65
|
|
|
$node = empty($html) ? null : $this->before($html); |
|
66
|
|
|
$this->remove(); |
|
67
|
|
|
|
|
68
|
|
|
return $node; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* @param $html |
|
73
|
|
|
* |
|
74
|
|
|
* @return AHTMLNode |
|
75
|
|
|
*/ |
|
76
|
|
|
public function before($html) |
|
77
|
|
|
{ |
|
78
|
|
|
$fragment = $this->get_fragment($html); |
|
79
|
|
|
$this->node->parentNode->insertBefore($fragment, $this->node); |
|
80
|
|
|
|
|
81
|
|
|
return new AHTMLNode($this->node->previousSibling, $this->doc); |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* @param $html |
|
86
|
|
|
*/ |
|
87
|
|
|
public function after($html) |
|
88
|
|
|
{ |
|
89
|
|
|
$fragment = $this->get_fragment($html); |
|
90
|
|
|
if ($ref_node = $this->node->nextSibling) |
|
91
|
|
|
{ |
|
92
|
|
|
$this->node->parentNode->insertBefore($fragment, $ref_node); |
|
93
|
|
|
} else { |
|
94
|
|
|
$this->node->parentNode->appendChild($fragment); |
|
95
|
|
|
} |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* @param $str |
|
100
|
|
|
* |
|
101
|
|
|
* @return mixed |
|
102
|
|
|
*/ |
|
103
|
|
|
public function decamelize($str) |
|
104
|
|
|
{ |
|
105
|
|
|
$str = \preg_replace_callback( |
|
106
|
|
|
'/(^|[a-z])([A-Z])/', |
|
107
|
|
|
function($matches) { |
|
108
|
|
|
return |
|
109
|
|
|
\strtolower( |
|
110
|
|
|
\strlen($matches[1]) |
|
111
|
|
|
? $matches[1] . '_' . $matches[2] : $matches[2] |
|
112
|
|
|
); |
|
113
|
|
|
}, |
|
114
|
|
|
$str |
|
115
|
|
|
); |
|
116
|
|
|
|
|
117
|
|
|
return \preg_replace('/ /', '_', \strtolower($str)); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* @return array |
|
122
|
|
|
*/ |
|
123
|
|
|
public function attributes() |
|
124
|
|
|
{ |
|
125
|
|
|
$ret = array(); |
|
126
|
|
|
foreach ($this->node->attributes as $attr) { |
|
127
|
|
|
$ret[$attr->nodeName] = $attr->nodeValue; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
return $ret; |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* @param null $key |
|
|
|
|
|
|
135
|
|
|
* @param int $level |
|
136
|
|
|
* |
|
137
|
|
|
* @return array |
|
138
|
|
|
*/ |
|
139
|
|
|
public function flatten($key = null, $level = 1) |
|
140
|
|
|
{ |
|
141
|
|
|
|
|
142
|
|
|
$children = $this->children; |
|
|
|
|
|
|
143
|
|
|
$ret = array(); |
|
144
|
|
|
$tag = $this->tag; |
|
|
|
|
|
|
145
|
|
|
|
|
146
|
|
|
if ($this->at('./preceding-sibling::' . $this->tag) || $this->at('./following-sibling::' . $this->tag) || ($key = $this->tag . 's')) |
|
|
|
|
|
|
147
|
|
|
{ |
|
148
|
|
|
$count = $this->search('./preceding-sibling::' . $this->tag)->length + 1; |
|
|
|
|
|
|
149
|
|
|
$tag .= '_' . $count; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
if ($children->length == 0) |
|
153
|
|
|
{ |
|
154
|
|
|
$ret[$this->decamelize(\implode(' ', \array_filter(array($key, $tag))))] = $this->text; |
|
|
|
|
|
|
155
|
|
|
} else { |
|
156
|
|
|
$flatten = []; |
|
157
|
|
|
foreach ($children as $child) |
|
158
|
|
|
{ |
|
159
|
|
|
$flatten[] = $child->flatten(\implode(' ', \array_filter(array($key, $level <= 0 ? $tag : null))), $level - 1); |
|
160
|
|
|
// $ret = array_merge($ret, $child->flatten(implode(' ', array_filter(array($key, $level <= 0 ? $tag : null))), $level - 1)); |
|
|
|
|
|
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
|
|
$ret = \array_merge($ret, ...$flatten); |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
return $ret; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
/** |
|
170
|
|
|
* @param $name |
|
171
|
|
|
* @param $value |
|
172
|
|
|
*/ |
|
173
|
|
|
public function __set($name, $value) |
|
174
|
|
|
{ |
|
175
|
|
|
switch ($name) |
|
176
|
|
|
{ |
|
177
|
|
|
case 'text': |
|
178
|
|
|
case 'innertext': |
|
179
|
|
|
case 'innerText': |
|
180
|
|
|
case 'plaintext': |
|
181
|
|
|
$this->node->nodeValue = $value; |
|
182
|
|
|
|
|
183
|
|
|
return; |
|
184
|
|
|
case 'outertext': |
|
185
|
|
|
$this->replace($value); |
|
186
|
|
|
|
|
187
|
|
|
return; |
|
188
|
|
|
case 'tag': |
|
189
|
|
|
$el = $this->replace('<' . $value . '>' . $this->innerhtml . '</' . $value . '>'); |
|
|
|
|
|
|
190
|
|
|
foreach ($this->node->attributes as $_name => $att) |
|
191
|
|
|
{ |
|
192
|
|
|
$el->$_name = $att->nodeValue; |
|
193
|
|
|
} |
|
194
|
|
|
$this->node = $el->node; |
|
195
|
|
|
|
|
196
|
|
|
return; |
|
197
|
|
|
|
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
$this->offsetSet($name, $value); |
|
201
|
|
|
} |
|
202
|
|
|
|
|
203
|
|
|
/** |
|
204
|
|
|
* @param mixed $offset |
|
205
|
|
|
* |
|
206
|
|
|
* @return bool |
|
207
|
|
|
*/ |
|
208
|
|
|
public function offsetExists($offset) |
|
209
|
|
|
{ |
|
210
|
|
|
return true; |
|
211
|
|
|
} |
|
212
|
|
|
|
|
213
|
|
|
/** |
|
214
|
|
|
* @param mixed $offset |
|
215
|
|
|
* |
|
216
|
|
|
* @return mixed |
|
217
|
|
|
*/ |
|
218
|
|
|
public function offsetGet($offset) |
|
219
|
|
|
{ |
|
220
|
|
|
return $this->node->getAttribute($offset); |
|
221
|
|
|
} |
|
222
|
|
|
|
|
223
|
|
|
/** |
|
224
|
|
|
* @param mixed $key |
|
225
|
|
|
* @param mixed $value |
|
226
|
|
|
*/ |
|
227
|
|
|
public function offsetSet($key, $value) |
|
228
|
|
|
{ |
|
229
|
|
|
if (\in_array($key, ['_path','dom','doc','node'])) |
|
230
|
|
|
{ |
|
231
|
|
|
return; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
if ($value) |
|
235
|
|
|
{ |
|
236
|
|
|
$this->node->setAttribute($key, $value); |
|
237
|
|
|
return; |
|
238
|
|
|
} |
|
239
|
|
|
|
|
240
|
|
|
$this->node->removeAttribute($key); |
|
241
|
|
|
//trigger_error('offsetSet not implemented', E_USER_WARNING); |
|
|
|
|
|
|
242
|
|
|
} |
|
243
|
|
|
|
|
244
|
|
|
/** |
|
245
|
|
|
* @param mixed $offset |
|
246
|
|
|
*/ |
|
247
|
|
|
public function offsetUnset($offset) |
|
248
|
|
|
{ |
|
249
|
|
|
\trigger_error('offsetUnset not implemented', E_USER_WARNING); |
|
250
|
|
|
} |
|
251
|
|
|
|
|
252
|
|
|
/** |
|
253
|
|
|
* @return mixed |
|
254
|
|
|
*/ |
|
255
|
|
|
public function title() |
|
256
|
|
|
{ |
|
257
|
|
|
return $this->node->getAttribute('title'); |
|
258
|
|
|
} |
|
259
|
|
|
} |
|
260
|
|
|
|
In general, usage of exit should be done with care and only when running in a scripting context like a CLI script.