1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Bavix\AdvancedHtmlDom; |
4
|
|
|
|
5
|
|
|
class AdvancedHtmlBase |
6
|
|
|
{ |
7
|
|
|
|
8
|
|
|
/** |
9
|
|
|
* @var |
10
|
|
|
*/ |
11
|
|
|
public $doc; |
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* @var |
15
|
|
|
*/ |
16
|
|
|
public $dom; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* @var |
20
|
|
|
*/ |
21
|
|
|
public $node; |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* @var bool |
25
|
|
|
*/ |
26
|
|
|
public $is_text = false; |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* @see https://github.com/monkeysuffrage/advanced_html_dom/issues/19 |
30
|
|
|
*/ |
31
|
4 |
|
public function __destruct() |
32
|
|
|
{ |
33
|
4 |
|
$this->doc = $this->dom = $this->node = null; |
34
|
4 |
|
unset($this->doc, $this->dom, $this->node); |
35
|
4 |
|
Cleanup::all(); |
36
|
4 |
|
} |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* @return string |
40
|
|
|
*/ |
41
|
|
|
public function __toString() |
42
|
|
|
{ |
43
|
|
|
return (string)$this->html(); |
44
|
|
|
} |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* @return mixed |
48
|
|
|
*/ |
49
|
3 |
|
public function html() |
50
|
|
|
{ |
51
|
3 |
|
return $this->doc->dom->saveXML($this->node); |
52
|
|
|
} |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* @return $this |
56
|
|
|
*/ |
57
|
|
|
public function remove() |
58
|
|
|
{ |
59
|
|
|
$this->node->parentNode->removeChild($this->node); |
60
|
|
|
|
61
|
|
|
return $this; |
62
|
|
|
} |
63
|
|
|
|
64
|
|
|
/** |
65
|
|
|
* @return Str |
66
|
|
|
*/ |
67
|
|
|
public function str() |
68
|
|
|
{ |
69
|
|
|
return new Str($this->text); |
|
|
|
|
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* @param $re |
74
|
|
|
* |
75
|
|
|
* @return bool |
76
|
|
|
*/ |
77
|
|
|
public function match($re) |
78
|
|
|
{ |
79
|
|
|
$str = new Str($this->text); |
|
|
|
|
80
|
|
|
|
81
|
|
|
return $str->match($re); |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* @param $re |
86
|
|
|
* |
87
|
|
|
* @return mixed |
88
|
|
|
*/ |
89
|
|
|
public function scan($re) |
90
|
|
|
{ |
91
|
|
|
$str = new Str($this->text); |
|
|
|
|
92
|
|
|
|
93
|
|
|
return $str->scan($re); |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
/** |
97
|
|
|
* @param $str |
98
|
|
|
* |
99
|
|
|
* @return string |
100
|
|
|
*/ |
101
|
|
|
public function clean($str) |
102
|
|
|
{ |
103
|
|
|
return $this->trim(\preg_replace('/\s+/', ' ', $str)); |
104
|
|
|
} |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* @param $str |
108
|
|
|
* |
109
|
|
|
* @return string |
110
|
|
|
*/ |
111
|
|
|
public function trim($str) |
112
|
|
|
{ |
113
|
|
|
return \trim($str); |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
/** |
117
|
|
|
* @param $key |
118
|
|
|
* @param $args |
119
|
|
|
* |
120
|
|
|
* @return AHTMLNode[]|AHTMLNodeList|null |
121
|
|
|
*/ |
122
|
6 |
|
public function __call($key, $args) |
123
|
|
|
{ |
124
|
6 |
|
$key = \strtolower(\str_replace('_', '', $key)); |
125
|
|
|
switch ($key) { |
126
|
6 |
|
case 'innertext': |
127
|
1 |
|
return ($this->is_text || !$this->children->length) ? $this->text() : $this->find('./text()|./*')->outertext; |
|
|
|
|
128
|
6 |
|
case 'plaintext': |
129
|
|
|
return $this->text(); |
130
|
6 |
|
case 'outertext': |
131
|
6 |
|
case 'html': |
132
|
6 |
|
case 'save': |
133
|
|
|
return $this->html(); |
134
|
6 |
|
case 'innerhtml': |
135
|
|
|
$ret = ''; |
136
|
|
|
foreach ($this->node->childNodes as $child) { |
137
|
|
|
$ret .= $this->doc->dom->saveHTML($child); |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
return $ret; |
141
|
|
|
|
142
|
6 |
|
case 'tag': |
143
|
1 |
|
return $this->node->nodeName; |
144
|
6 |
|
case 'next': |
145
|
|
|
return $this->at('./following-sibling::*[1]|./following-sibling::text()[1]|./following-sibling::comment()[1]'); |
|
|
|
|
146
|
|
|
|
147
|
6 |
|
case 'index': |
148
|
|
|
return $this->search('./preceding-sibling::*')->length + 1; |
|
|
|
|
149
|
|
|
|
150
|
|
|
/* |
151
|
|
|
DOMNode::insertBefore — Adds a new child |
152
|
|
|
*/ |
153
|
|
|
|
154
|
|
|
// simple-html-dom junk methods |
155
|
6 |
|
case 'clear': |
156
|
|
|
return; |
157
|
|
|
|
158
|
|
|
// search functions |
159
|
6 |
|
case 'at': |
160
|
3 |
|
case 'getelementbytagname': |
161
|
6 |
|
return $this->find($args[0], 0); |
162
|
|
|
|
163
|
3 |
|
case 'search': |
164
|
3 |
|
case 'getelementsbytagname': |
165
|
1 |
|
return isset($args[1]) ? $this->find($args[0], $args[1]) : $this->find($args[0]); |
166
|
|
|
|
167
|
3 |
|
case 'getelementbyid': |
168
|
|
|
return $this->find('#' . $args[0], 0); |
169
|
3 |
|
case 'getelementsbyid': |
170
|
|
|
return isset($args[1]) ? $this->find('#' . $args[0], $args[1]) : $this->find('#' . $args[0]); |
171
|
|
|
|
172
|
|
|
// attributes |
173
|
3 |
|
case 'hasattribute': |
174
|
|
|
return !$this->is_text && $this->node->hasAttribute($args[0]); |
175
|
|
|
|
176
|
3 |
|
case 'getattribute': |
177
|
|
|
$arg = $args[0]; |
178
|
|
|
|
179
|
|
|
return $this->$arg; |
180
|
3 |
|
case 'setattribute': |
181
|
|
|
$arg0 = $args[0]; |
182
|
|
|
$arg1 = $args[1]; |
183
|
|
|
|
184
|
|
|
return $this->$arg0 = $arg1; |
185
|
3 |
|
case 'removeattribute': |
186
|
|
|
$arg = $args[0]; |
187
|
|
|
|
188
|
|
|
return $this->$arg = null; |
189
|
|
|
|
190
|
|
|
// wrap |
191
|
3 |
|
case 'wrap': |
192
|
|
|
return $this->replace('<' . $args[0] . '>' . $this . '</' . $args[0] . '>'); |
|
|
|
|
193
|
3 |
|
case 'unwrap': |
194
|
|
|
return $this->parent->replace($this); |
|
|
|
|
195
|
|
|
|
196
|
3 |
|
case 'str': |
197
|
|
|
return new Str($this->text); |
|
|
|
|
198
|
|
|
|
199
|
|
|
// heirarchy |
200
|
3 |
|
case 'firstchild': |
201
|
|
|
return $this->at('> *'); |
202
|
3 |
|
case 'lastchild': |
203
|
|
|
return $this->at('> *:last'); |
204
|
3 |
|
case 'nextsibling': |
205
|
|
|
return $this->at('+ *'); |
206
|
3 |
|
case 'prevsibling': |
207
|
|
|
return $this->at('./preceding-sibling::*[1]'); |
208
|
3 |
|
case 'parent': |
209
|
|
|
return $this->at('./..'); |
210
|
3 |
|
case 'children': |
211
|
3 |
|
case 'childnodes': |
212
|
1 |
|
$nl = $this->search('./*'); |
213
|
|
|
|
214
|
1 |
|
return isset($args[0]) ? $nl[$args[0]] : $nl; |
215
|
2 |
|
case 'child': // including text/comment nodes |
216
|
|
|
$nl = $this->search('./*|./text()|./comment()'); |
217
|
|
|
|
218
|
|
|
return isset($args[0]) ? $nl[$args[0]] : $nl; |
219
|
|
|
|
220
|
|
|
} |
221
|
|
|
|
222
|
|
|
// $doc->spans[x] |
223
|
2 |
|
if (\preg_match(TAGS_REGEX, $key, $m)) { |
224
|
|
|
return $this->find($m[1]); |
225
|
|
|
} |
226
|
2 |
|
if (\preg_match(TAG_REGEX, $key, $m)) { |
227
|
|
|
return $this->find($m[1], 0); |
228
|
|
|
} |
229
|
|
|
|
230
|
2 |
|
if (\preg_match('/(clean|trim|str)(.*)/', $key, $m) && isset($m[2])) { |
231
|
|
|
list($arg0, $arg1, $arg2) = $m; |
232
|
|
|
|
233
|
|
|
return $this->$arg1($this->$arg2); |
234
|
|
|
} |
235
|
|
|
|
236
|
2 |
|
if (\in_array($key, ['dom', 'node', 'doc'])) { |
237
|
|
|
return null; |
238
|
|
|
} |
239
|
|
|
|
240
|
2 |
|
if (!\preg_match(ATTRIBUTE_REGEX, $key, $m)) { |
241
|
|
|
\trigger_error('Unknown method or property: ' . $key, E_USER_WARNING); |
242
|
|
|
} |
243
|
2 |
|
if (!$this->node || $this->is_text) { |
244
|
|
|
return null; |
245
|
|
|
} |
246
|
|
|
|
247
|
2 |
|
return $this->node->getAttribute($key); |
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* @return mixed |
252
|
|
|
*/ |
253
|
2 |
|
public function text() |
254
|
|
|
{ |
255
|
2 |
|
return $this->node->nodeValue; |
256
|
|
|
} |
257
|
|
|
|
258
|
|
|
// magic methods |
259
|
|
|
|
260
|
|
|
/** |
261
|
|
|
* @param $css |
262
|
|
|
* @param null $index |
|
|
|
|
263
|
|
|
* |
264
|
|
|
* @return array|AHTMLNode|AHTMLNodeList |
265
|
|
|
*/ |
266
|
6 |
|
public function find($css, $index = null) |
267
|
|
|
{ |
268
|
6 |
|
$xpath = CSS::xpath_for($css); |
269
|
|
|
|
270
|
6 |
|
if (!isset($this->doc, $this->doc->xpath)) { |
271
|
|
|
return null; |
272
|
|
|
} |
273
|
|
|
|
274
|
6 |
|
if (null === $index) { |
|
|
|
|
275
|
5 |
|
return new AHTMLNodeList($this->doc->xpath->query($xpath, $this->node), $this->doc); |
276
|
|
|
} |
277
|
|
|
|
278
|
6 |
|
$nl = $this->doc->xpath->query($xpath, $this->node); |
279
|
6 |
|
if ($index < 0) { |
280
|
|
|
$index = $nl->length + $index; |
281
|
|
|
} |
282
|
6 |
|
$node = $nl->item($index); |
283
|
|
|
|
284
|
6 |
|
return $node ? new AHTMLNode($node, $this->doc) : null; |
285
|
|
|
} |
286
|
|
|
|
287
|
|
|
/** |
288
|
|
|
* @param $key |
289
|
|
|
* |
290
|
|
|
* @return mixed |
291
|
|
|
*/ |
292
|
4 |
|
public function __get($key) |
293
|
|
|
{ |
294
|
4 |
|
return $this->$key(); |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
/** |
298
|
|
|
* @param $name |
299
|
|
|
* @param $value |
300
|
|
|
*/ |
301
|
|
|
public function __set($name, $value) |
302
|
|
|
{ |
303
|
|
|
throw new \InvalidArgumentException(__METHOD__); |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
/** |
307
|
|
|
* @param $name |
308
|
|
|
* |
309
|
|
|
* @return bool |
310
|
|
|
*/ |
311
|
|
|
public function __isset($name) |
312
|
|
|
{ |
313
|
|
|
return true; |
314
|
|
|
} |
315
|
|
|
} |
316
|
|
|
|