1
|
|
|
<?php declare(strict_types=1); |
2
|
|
|
namespace PHPHtmlParser\Dom; |
3
|
|
|
|
4
|
|
|
use PHPHtmlParser\Exceptions\CircularException; |
5
|
|
|
use PHPHtmlParser\Exceptions\ParentNotFoundException; |
6
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
7
|
|
|
use PHPHtmlParser\Selector\Selector; |
8
|
|
|
use PHPHtmlParser\Selector\Parser as SelectorParser; |
9
|
|
|
use stringEncode\Encode; |
10
|
|
|
use PHPHtmlParser\Finder; |
11
|
|
|
|
12
|
|
|
/** |
13
|
|
|
* Dom node object. |
14
|
|
|
* @property string $outerhtml |
15
|
|
|
* @property string $innerhtml |
16
|
|
|
* @property string $text |
17
|
|
|
* @property int $prev |
18
|
|
|
* @property int $next |
19
|
|
|
* @property Tag $tag |
20
|
|
|
* @property InnerNode $parent |
21
|
|
|
*/ |
22
|
|
|
abstract class AbstractNode |
23
|
|
|
{ |
24
|
|
|
/** |
25
|
|
|
* @var int |
26
|
|
|
*/ |
27
|
|
|
private static $count = 0; |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* Contains the tag name/type |
31
|
|
|
* |
32
|
|
|
* @var ?Tag |
33
|
|
|
*/ |
34
|
|
|
protected $tag = null; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Contains a list of attributes on this tag. |
38
|
|
|
* |
39
|
|
|
* @var array |
40
|
|
|
*/ |
41
|
|
|
protected $attr = []; |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* Contains the parent Node. |
45
|
|
|
* |
46
|
|
|
* @var ?InnerNode |
47
|
|
|
*/ |
48
|
|
|
protected $parent = null; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* The unique id of the class. Given by PHP. |
52
|
|
|
* |
53
|
|
|
* @var int |
54
|
|
|
*/ |
55
|
|
|
protected $id; |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* The encoding class used to encode strings. |
59
|
|
|
* |
60
|
|
|
* @var mixed |
61
|
|
|
*/ |
62
|
|
|
protected $encode; |
63
|
|
|
|
64
|
|
|
/** |
65
|
|
|
* An array of all the children. |
66
|
|
|
* |
67
|
|
|
* @var array |
68
|
|
|
*/ |
69
|
|
|
protected $children = []; |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* @var bool |
73
|
|
|
*/ |
74
|
|
|
protected $htmlSpecialCharsDecode = false; |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* Creates a unique id for this node. |
78
|
|
|
*/ |
79
|
510 |
|
public function __construct() |
80
|
|
|
{ |
81
|
510 |
|
$this->id = self::$count; |
82
|
510 |
|
self::$count++; |
83
|
510 |
|
} |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* Magic get method for attributes and certain methods. |
87
|
|
|
* |
88
|
|
|
* @param string $key |
89
|
|
|
* @return mixed |
90
|
|
|
*/ |
91
|
174 |
|
public function __get(string $key) |
92
|
|
|
{ |
93
|
|
|
// check attribute first |
94
|
174 |
|
if ( ! is_null($this->getAttribute($key))) { |
|
|
|
|
95
|
21 |
|
return $this->getAttribute($key); |
96
|
|
|
} |
97
|
156 |
|
switch (strtolower($key)) { |
98
|
156 |
|
case 'outerhtml': |
99
|
36 |
|
return $this->outerHtml(); |
100
|
120 |
|
case 'innerhtml': |
101
|
66 |
|
return $this->innerHtml(); |
102
|
57 |
|
case 'text': |
103
|
51 |
|
return $this->text(); |
104
|
6 |
|
case 'tag': |
105
|
3 |
|
return $this->getTag(); |
106
|
3 |
|
case 'parent': |
107
|
|
|
return $this->getParent(); |
108
|
|
|
} |
109
|
|
|
|
110
|
3 |
|
return null; |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* Attempts to clear out any object references. |
115
|
|
|
*/ |
116
|
126 |
|
public function __destruct() |
117
|
|
|
{ |
118
|
126 |
|
$this->tag = null; |
119
|
126 |
|
$this->parent = null; |
120
|
126 |
|
$this->attr = []; |
121
|
126 |
|
$this->children = []; |
122
|
126 |
|
} |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* Simply calls the outer text method. |
126
|
|
|
* |
127
|
|
|
* @return string |
128
|
|
|
*/ |
129
|
12 |
|
public function __toString() |
130
|
|
|
{ |
131
|
12 |
|
return $this->outerHtml(); |
132
|
|
|
} |
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* @param bool $htmlSpecialCharsDecode |
136
|
|
|
* @return void |
137
|
|
|
*/ |
138
|
246 |
|
public function setHtmlSpecialCharsDecode($htmlSpecialCharsDecode = false): void |
139
|
|
|
{ |
140
|
246 |
|
$this->htmlSpecialCharsDecode = $htmlSpecialCharsDecode; |
141
|
246 |
|
} |
142
|
|
|
|
143
|
|
|
|
144
|
|
|
/** |
145
|
|
|
* Reset node counter |
146
|
|
|
* |
147
|
|
|
* @return void |
148
|
|
|
*/ |
149
|
183 |
|
public static function resetCount() |
150
|
|
|
{ |
151
|
183 |
|
self::$count = 0; |
152
|
183 |
|
} |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* Returns the id of this object. |
156
|
|
|
* |
157
|
|
|
* @return int |
158
|
|
|
*/ |
159
|
444 |
|
public function id(): int |
160
|
|
|
{ |
161
|
444 |
|
return $this->id; |
162
|
|
|
} |
163
|
|
|
|
164
|
|
|
/** |
165
|
|
|
* Returns the parent of node. |
166
|
|
|
* |
167
|
|
|
* @return AbstractNode |
168
|
|
|
*/ |
169
|
240 |
|
public function getParent() |
170
|
|
|
{ |
171
|
240 |
|
return $this->parent; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
/** |
175
|
|
|
* Sets the parent node. |
176
|
|
|
* @param InnerNode $parent |
177
|
|
|
* @return AbstractNode |
178
|
|
|
* @throws ChildNotFoundException |
179
|
|
|
* @throws CircularException |
180
|
|
|
*/ |
181
|
438 |
|
public function setParent(InnerNode $parent): AbstractNode |
182
|
|
|
{ |
183
|
|
|
// remove from old parent |
184
|
438 |
|
if ( ! is_null($this->parent)) { |
185
|
30 |
|
if ($this->parent->id() == $parent->id()) { |
186
|
|
|
// already the parent |
187
|
27 |
|
return $this; |
188
|
|
|
} |
189
|
|
|
|
190
|
6 |
|
$this->parent->removeChild($this->id); |
191
|
|
|
} |
192
|
|
|
|
193
|
438 |
|
$this->parent = $parent; |
194
|
|
|
|
195
|
|
|
// assign child to parent |
196
|
438 |
|
$this->parent->addChild($this); |
197
|
|
|
|
198
|
438 |
|
return $this; |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
/** |
202
|
|
|
* Removes this node and all its children from the |
203
|
|
|
* DOM tree. |
204
|
|
|
* |
205
|
|
|
* @return void |
206
|
|
|
*/ |
207
|
3 |
|
public function delete() |
208
|
|
|
{ |
209
|
3 |
|
if ( ! is_null($this->parent)) { |
210
|
3 |
|
$this->parent->removeChild($this->id); |
211
|
|
|
} |
212
|
3 |
|
$this->parent->clear(); |
213
|
3 |
|
$this->clear(); |
214
|
3 |
|
} |
215
|
|
|
|
216
|
|
|
/** |
217
|
|
|
* Sets the encoding class to this node. |
218
|
|
|
* |
219
|
|
|
* @param Encode $encode |
220
|
|
|
* @return void |
221
|
|
|
*/ |
222
|
222 |
|
public function propagateEncoding(Encode $encode) |
223
|
|
|
{ |
224
|
222 |
|
$this->encode = $encode; |
225
|
222 |
|
$this->tag->setEncoding($encode); |
226
|
222 |
|
} |
227
|
|
|
|
228
|
|
|
/** |
229
|
|
|
* Checks if the given node id is an ancestor of |
230
|
|
|
* the current node. |
231
|
|
|
* |
232
|
|
|
* @param int $id |
233
|
|
|
* @return bool |
234
|
|
|
*/ |
235
|
441 |
|
public function isAncestor(int $id): Bool |
236
|
|
|
{ |
237
|
441 |
|
if ( ! is_null($this->getAncestor($id))) { |
238
|
6 |
|
return true; |
239
|
|
|
} |
240
|
|
|
|
241
|
441 |
|
return false; |
242
|
|
|
} |
243
|
|
|
|
244
|
|
|
/** |
245
|
|
|
* Attempts to get an ancestor node by the given id. |
246
|
|
|
* |
247
|
|
|
* @param int $id |
248
|
|
|
* @return null|AbstractNode |
249
|
|
|
*/ |
250
|
444 |
|
public function getAncestor(int $id) |
251
|
|
|
{ |
252
|
444 |
|
if ( ! is_null($this->parent)) { |
253
|
309 |
|
if ($this->parent->id() == $id) { |
254
|
12 |
|
return $this->parent; |
255
|
|
|
} |
256
|
|
|
|
257
|
300 |
|
return $this->parent->getAncestor($id); |
258
|
|
|
} |
259
|
|
|
|
260
|
444 |
|
return null; |
261
|
|
|
} |
262
|
|
|
|
263
|
|
|
/** |
264
|
|
|
* Checks if the current node has a next sibling. |
265
|
|
|
* |
266
|
|
|
* @return bool |
267
|
|
|
*/ |
268
|
9 |
|
public function hasNextSibling(): bool |
269
|
|
|
{ |
270
|
|
|
try |
271
|
|
|
{ |
272
|
9 |
|
$this->nextSibling(); |
273
|
|
|
|
274
|
|
|
// sibling found, return true; |
275
|
9 |
|
return true; |
276
|
|
|
} |
277
|
9 |
|
catch (ParentNotFoundException $e) |
278
|
|
|
{ |
279
|
|
|
// no parent, no next sibling |
280
|
|
|
unset($e); |
281
|
|
|
return false; |
282
|
|
|
} |
283
|
9 |
|
catch (ChildNotFoundException $e) |
284
|
|
|
{ |
285
|
|
|
// no sibling found |
286
|
9 |
|
unset($e); |
287
|
9 |
|
return false; |
288
|
|
|
} |
289
|
|
|
} |
290
|
|
|
|
291
|
|
|
/** |
292
|
|
|
* Attempts to get the next sibling. |
293
|
|
|
* @return AbstractNode |
294
|
|
|
* @throws ChildNotFoundException |
295
|
|
|
* @throws ParentNotFoundException |
296
|
|
|
*/ |
297
|
39 |
|
public function nextSibling(): AbstractNode |
298
|
|
|
{ |
299
|
39 |
|
if (is_null($this->parent)) { |
300
|
3 |
|
throw new ParentNotFoundException('Parent is not set for this node.'); |
301
|
|
|
} |
302
|
|
|
|
303
|
36 |
|
return $this->parent->nextChild($this->id); |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
/** |
307
|
|
|
* Attempts to get the previous sibling. |
308
|
|
|
* @return AbstractNode |
309
|
|
|
* @throws ChildNotFoundException |
310
|
|
|
* @throws ParentNotFoundException |
311
|
|
|
*/ |
312
|
9 |
|
public function previousSibling(): AbstractNode |
313
|
|
|
{ |
314
|
9 |
|
if (is_null($this->parent)) { |
315
|
3 |
|
throw new ParentNotFoundException('Parent is not set for this node.'); |
316
|
|
|
} |
317
|
|
|
|
318
|
6 |
|
return $this->parent->previousChild($this->id); |
319
|
|
|
} |
320
|
|
|
|
321
|
|
|
/** |
322
|
|
|
* Gets the tag object of this node. |
323
|
|
|
* |
324
|
|
|
* @return Tag |
325
|
|
|
*/ |
326
|
315 |
|
public function getTag(): Tag |
327
|
|
|
{ |
328
|
315 |
|
return $this->tag; |
329
|
|
|
} |
330
|
|
|
|
331
|
|
|
/** |
332
|
|
|
* A wrapper method that simply calls the getAttribute method |
333
|
|
|
* on the tag of this node. |
334
|
|
|
* |
335
|
|
|
* @return array |
336
|
|
|
*/ |
337
|
6 |
|
public function getAttributes(): array |
338
|
|
|
{ |
339
|
6 |
|
$attributes = $this->tag->getAttributes(); |
340
|
6 |
|
foreach ($attributes as $name => $info) { |
341
|
3 |
|
$attributes[$name] = $info['value']; |
342
|
|
|
} |
343
|
|
|
|
344
|
6 |
|
return $attributes; |
345
|
|
|
} |
346
|
|
|
|
347
|
|
|
/** |
348
|
|
|
* A wrapper method that simply calls the getAttribute method |
349
|
|
|
* on the tag of this node. |
350
|
|
|
* |
351
|
|
|
* @param string $key |
352
|
|
|
* @return string|null |
353
|
|
|
*/ |
354
|
237 |
|
public function getAttribute(string $key): ?string |
355
|
|
|
{ |
356
|
237 |
|
$attribute = $this->tag->getAttribute($key); |
357
|
237 |
|
$attributeValue = $attribute['value']; |
358
|
|
|
|
359
|
237 |
|
return $attributeValue; |
360
|
|
|
} |
361
|
|
|
|
362
|
|
|
/** |
363
|
|
|
* A wrapper method that simply calls the hasAttribute method |
364
|
|
|
* on the tag of this node. |
365
|
|
|
* |
366
|
|
|
* @param string $key |
367
|
|
|
* @return bool |
368
|
|
|
*/ |
369
|
99 |
|
public function hasAttribute(string $key): bool |
370
|
|
|
{ |
371
|
99 |
|
return $this->tag->hasAttribute($key); |
372
|
|
|
} |
373
|
|
|
|
374
|
|
|
/** |
375
|
|
|
* A wrapper method that simply calls the setAttribute method |
376
|
|
|
* on the tag of this node. |
377
|
|
|
* |
378
|
|
|
* @param string $key |
379
|
|
|
* @param string|array $value |
380
|
|
|
* @return AbstractNode |
381
|
|
|
* @chainable |
382
|
|
|
*/ |
383
|
24 |
|
public function setAttribute(string $key, $value): AbstractNode |
384
|
|
|
{ |
385
|
24 |
|
$this->tag->setAttribute($key, $value); |
386
|
|
|
|
387
|
|
|
//clear any cache |
388
|
24 |
|
$this->clear(); |
389
|
|
|
|
390
|
24 |
|
return $this; |
391
|
|
|
} |
392
|
|
|
|
393
|
|
|
/** |
394
|
|
|
* A wrapper method that simply calls the removeAttribute method |
395
|
|
|
* on the tag of this node. |
396
|
|
|
* |
397
|
|
|
* @param string $key |
398
|
|
|
* @return void |
399
|
|
|
*/ |
400
|
3 |
|
public function removeAttribute(string $key): void |
401
|
|
|
{ |
402
|
3 |
|
$this->tag->removeAttribute($key); |
403
|
|
|
|
404
|
|
|
//clear any cache |
405
|
3 |
|
$this->clear(); |
406
|
3 |
|
} |
407
|
|
|
|
408
|
|
|
/** |
409
|
|
|
* A wrapper method that simply calls the removeAllAttributes |
410
|
|
|
* method on the tag of this node. |
411
|
|
|
* |
412
|
|
|
* @return void |
413
|
|
|
*/ |
414
|
3 |
|
public function removeAllAttributes(): void |
415
|
|
|
{ |
416
|
3 |
|
$this->tag->removeAllAttributes(); |
417
|
|
|
|
418
|
|
|
//clear any cache |
419
|
3 |
|
$this->clear(); |
420
|
3 |
|
} |
421
|
|
|
/** |
422
|
|
|
* Function to locate a specific ancestor tag in the path to the root. |
423
|
|
|
* |
424
|
|
|
* @param string $tag |
425
|
|
|
* @return AbstractNode |
426
|
|
|
* @throws ParentNotFoundException |
427
|
|
|
*/ |
428
|
6 |
|
public function ancestorByTag(string $tag): AbstractNode |
429
|
|
|
{ |
430
|
|
|
// Start by including ourselves in the comparison. |
431
|
6 |
|
$node = $this; |
432
|
|
|
|
433
|
6 |
|
while ( ! is_null($node)) { |
434
|
6 |
|
if ($node->tag->name() == $tag) { |
435
|
3 |
|
return $node; |
436
|
|
|
} |
437
|
|
|
|
438
|
3 |
|
$node = $node->getParent(); |
439
|
|
|
} |
440
|
|
|
|
441
|
3 |
|
throw new ParentNotFoundException('Could not find an ancestor with "'.$tag.'" tag'); |
442
|
|
|
} |
443
|
|
|
|
444
|
|
|
/** |
445
|
|
|
* Find elements by css selector |
446
|
|
|
* @param string $selector |
447
|
|
|
* @param int|null $nth |
448
|
|
|
* @param bool $depthFirst |
449
|
|
|
* @return mixed|Collection|null |
450
|
|
|
* @throws ChildNotFoundException |
451
|
|
|
*/ |
452
|
240 |
|
public function find(string $selector, int $nth = null, bool $depthFirst = false) |
453
|
|
|
{ |
454
|
240 |
|
$selector = new Selector($selector, new SelectorParser()); |
455
|
240 |
|
$selector->setDepthFirstFind($depthFirst); |
456
|
240 |
|
$nodes = $selector->find($this); |
457
|
|
|
|
458
|
240 |
|
if ( ! is_null($nth)) { |
459
|
|
|
// return nth-element or array |
460
|
240 |
|
if (isset($nodes[$nth])) { |
461
|
123 |
|
return $nodes[$nth]; |
462
|
|
|
} |
463
|
|
|
|
464
|
210 |
|
return null; |
465
|
|
|
} |
466
|
|
|
|
467
|
84 |
|
return $nodes; |
468
|
|
|
} |
469
|
|
|
|
470
|
|
|
/** |
471
|
|
|
* Find node by id |
472
|
|
|
* @param int $id |
473
|
|
|
* @return bool|AbstractNode |
474
|
|
|
* @throws ChildNotFoundException |
475
|
|
|
* @throws ParentNotFoundException |
476
|
|
|
*/ |
477
|
9 |
|
public function findById(int $id) |
478
|
|
|
{ |
479
|
9 |
|
$finder= new Finder($id); |
480
|
|
|
|
481
|
9 |
|
return $finder->find($this); |
482
|
|
|
} |
483
|
|
|
|
484
|
|
|
|
485
|
|
|
/** |
486
|
|
|
* Gets the inner html of this node. |
487
|
|
|
* |
488
|
|
|
* @return string |
489
|
|
|
*/ |
490
|
|
|
abstract public function innerHtml(): string; |
491
|
|
|
|
492
|
|
|
/** |
493
|
|
|
* Gets the html of this node, including it's own |
494
|
|
|
* tag. |
495
|
|
|
* |
496
|
|
|
* @return string |
497
|
|
|
*/ |
498
|
|
|
abstract public function outerHtml(): string; |
499
|
|
|
|
500
|
|
|
/** |
501
|
|
|
* Gets the text of this node (if there is any text). |
502
|
|
|
* |
503
|
|
|
* @return string |
504
|
|
|
*/ |
505
|
|
|
abstract public function text(): string; |
506
|
|
|
|
507
|
|
|
/** |
508
|
|
|
* Call this when something in the node tree has changed. Like a child has been added |
509
|
|
|
* or a parent has been changed. |
510
|
|
|
* |
511
|
|
|
* @return void |
512
|
|
|
*/ |
513
|
|
|
abstract protected function clear(): void; |
514
|
|
|
|
515
|
|
|
/** |
516
|
|
|
* Check is node type textNode |
517
|
|
|
* |
518
|
|
|
* @return boolean |
519
|
|
|
*/ |
520
|
9 |
|
public function isTextNode(): bool |
521
|
|
|
{ |
522
|
|
|
|
523
|
9 |
|
return false; |
524
|
|
|
} |
525
|
|
|
} |
526
|
|
|
|