1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace PHPHtmlParser\Dom\Node; |
6
|
|
|
|
7
|
|
|
use PHPHtmlParser\Contracts\Selector\SelectorInterface; |
8
|
|
|
use PHPHtmlParser\Dom\Tag; |
9
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
10
|
|
|
use PHPHtmlParser\Exceptions\CircularException; |
11
|
|
|
use PHPHtmlParser\Exceptions\ParentNotFoundException; |
12
|
|
|
use PHPHtmlParser\Exceptions\Tag\AttributeNotFoundException; |
13
|
|
|
use PHPHtmlParser\Finder; |
14
|
|
|
use PHPHtmlParser\Selector\Selector; |
15
|
|
|
use stringEncode\Encode; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* Dom node object. |
19
|
|
|
* |
20
|
|
|
* @property-read string $outerhtml |
21
|
|
|
* @property-read string $innerhtml |
22
|
|
|
* @property-read string $innerText |
23
|
|
|
* @property-read string $text |
24
|
|
|
* @property-read Tag $tag |
25
|
|
|
* @property-read InnerNode $parent |
26
|
|
|
*/ |
27
|
|
|
abstract class AbstractNode |
28
|
|
|
{ |
29
|
|
|
/** |
30
|
|
|
* Contains the tag name/type. |
31
|
|
|
* |
32
|
|
|
* @var ?Tag |
33
|
|
|
*/ |
34
|
|
|
protected $tag; |
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; |
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
|
|
|
* @var int |
77
|
|
|
*/ |
78
|
|
|
private static $count = 0; |
79
|
|
|
|
80
|
|
|
/** |
81
|
|
|
* Creates a unique id for this node. |
82
|
|
|
*/ |
83
|
564 |
|
public function __construct() |
84
|
|
|
{ |
85
|
564 |
|
$this->id = self::$count; |
86
|
564 |
|
++self::$count; |
87
|
564 |
|
} |
88
|
|
|
|
89
|
|
|
/** |
90
|
|
|
* Attempts to clear out any object references. |
91
|
|
|
*/ |
92
|
134 |
|
public function __destruct() |
93
|
|
|
{ |
94
|
134 |
|
$this->tag = null; |
|
|
|
|
95
|
134 |
|
$this->parent = null; |
|
|
|
|
96
|
134 |
|
$this->attr = []; |
97
|
134 |
|
$this->children = []; |
98
|
134 |
|
} |
99
|
|
|
|
100
|
|
|
/** |
101
|
|
|
* Magic get method for attributes and certain methods. |
102
|
|
|
* |
103
|
|
|
* @return mixed |
104
|
|
|
*/ |
105
|
330 |
|
public function __get(string $key) |
106
|
|
|
{ |
107
|
|
|
// check attribute first |
108
|
330 |
|
if ($this->getAttribute($key) !== null) { |
109
|
18 |
|
return $this->getAttribute($key); |
110
|
|
|
} |
111
|
327 |
|
switch (\strtolower($key)) { |
112
|
327 |
|
case 'outerhtml': |
113
|
42 |
|
return $this->outerHtml(); |
114
|
318 |
|
case 'innerhtml': |
115
|
81 |
|
return $this->innerHtml(); |
116
|
315 |
|
case 'innertext': |
117
|
6 |
|
return $this->innerText(); |
|
|
|
|
118
|
312 |
|
case 'text': |
119
|
51 |
|
return $this->text(); |
120
|
297 |
|
case 'tag': |
121
|
297 |
|
return $this->getTag(); |
122
|
3 |
|
case 'parent': |
123
|
|
|
return $this->getParent(); |
124
|
|
|
} |
125
|
3 |
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* Simply calls the outer text method. |
129
|
|
|
* |
130
|
|
|
* @return string |
131
|
|
|
*/ |
132
|
9 |
|
public function __toString() |
133
|
|
|
{ |
134
|
9 |
|
return $this->outerHtml(); |
135
|
|
|
} |
136
|
|
|
|
137
|
|
|
/** |
138
|
|
|
* @param bool $htmlSpecialCharsDecode |
139
|
|
|
*/ |
140
|
294 |
|
public function setHtmlSpecialCharsDecode($htmlSpecialCharsDecode = false): void |
141
|
|
|
{ |
142
|
294 |
|
$this->htmlSpecialCharsDecode = $htmlSpecialCharsDecode; |
143
|
294 |
|
} |
144
|
|
|
|
145
|
|
|
/** |
146
|
|
|
* Returns the id of this object. |
147
|
|
|
*/ |
148
|
495 |
|
public function id(): int |
149
|
|
|
{ |
150
|
495 |
|
return $this->id; |
151
|
|
|
} |
152
|
|
|
|
153
|
|
|
/** |
154
|
|
|
* Returns the parent of node. |
155
|
|
|
* |
156
|
|
|
* @return InnerNode |
157
|
|
|
*/ |
158
|
264 |
|
public function getParent(): ?InnerNode |
159
|
|
|
{ |
160
|
264 |
|
return $this->parent; |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
/** |
164
|
|
|
* Sets the parent node. |
165
|
|
|
* |
166
|
|
|
* @throws ChildNotFoundException |
167
|
|
|
* @throws CircularException |
168
|
|
|
*/ |
169
|
489 |
|
public function setParent(InnerNode $parent): AbstractNode |
170
|
|
|
{ |
171
|
|
|
// remove from old parent |
172
|
489 |
|
if ($this->parent !== null) { |
173
|
33 |
|
if ($this->parent->id() == $parent->id()) { |
174
|
|
|
// already the parent |
175
|
27 |
|
return $this; |
176
|
|
|
} |
177
|
|
|
|
178
|
9 |
|
$this->parent->removeChild($this->id); |
179
|
|
|
} |
180
|
|
|
|
181
|
489 |
|
$this->parent = $parent; |
|
|
|
|
182
|
|
|
|
183
|
|
|
// assign child to parent |
184
|
489 |
|
$this->parent->addChild($this); |
185
|
|
|
|
186
|
489 |
|
return $this; |
187
|
|
|
} |
188
|
|
|
|
189
|
|
|
/** |
190
|
|
|
* Removes this node and all its children from the |
191
|
|
|
* DOM tree. |
192
|
|
|
* |
193
|
|
|
* @return void |
194
|
|
|
*/ |
195
|
3 |
|
public function delete() |
196
|
|
|
{ |
197
|
3 |
|
if ($this->parent !== null) { |
198
|
3 |
|
$this->parent->removeChild($this->id); |
199
|
|
|
} |
200
|
3 |
|
$this->parent->clear(); |
201
|
3 |
|
$this->clear(); |
202
|
3 |
|
} |
203
|
|
|
|
204
|
|
|
/** |
205
|
|
|
* Sets the encoding class to this node. |
206
|
|
|
* |
207
|
|
|
* @return void |
208
|
|
|
*/ |
209
|
249 |
|
public function propagateEncoding(Encode $encode) |
210
|
|
|
{ |
211
|
249 |
|
$this->encode = $encode; |
212
|
249 |
|
$this->tag->setEncoding($encode); |
213
|
249 |
|
} |
214
|
|
|
|
215
|
|
|
/** |
216
|
|
|
* Checks if the given node id is an ancestor of |
217
|
|
|
* the current node. |
218
|
|
|
*/ |
219
|
492 |
|
public function isAncestor(int $id): bool |
220
|
|
|
{ |
221
|
492 |
|
if ($this->getAncestor($id) !== null) { |
222
|
6 |
|
return true; |
223
|
|
|
} |
224
|
|
|
|
225
|
492 |
|
return false; |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
/** |
229
|
|
|
* Attempts to get an ancestor node by the given id. |
230
|
|
|
* |
231
|
|
|
* @return AbstractNode|null |
232
|
|
|
*/ |
233
|
495 |
|
public function getAncestor(int $id) |
234
|
|
|
{ |
235
|
495 |
|
if ($this->parent !== null) { |
236
|
333 |
|
if ($this->parent->id() == $id) { |
237
|
12 |
|
return $this->parent; |
238
|
|
|
} |
239
|
|
|
|
240
|
324 |
|
return $this->parent->getAncestor($id); |
241
|
|
|
} |
242
|
495 |
|
} |
243
|
|
|
|
244
|
|
|
/** |
245
|
|
|
* Checks if the current node has a next sibling. |
246
|
|
|
*/ |
247
|
|
|
public function hasNextSibling(): bool |
248
|
|
|
{ |
249
|
|
|
try { |
250
|
|
|
$this->nextSibling(); |
251
|
|
|
|
252
|
|
|
// sibling found, return true; |
253
|
|
|
return true; |
254
|
|
|
} catch (ParentNotFoundException $e) { |
255
|
|
|
// no parent, no next sibling |
256
|
|
|
unset($e); |
257
|
|
|
|
258
|
|
|
return false; |
259
|
|
|
} catch (ChildNotFoundException $e) { |
260
|
|
|
// no sibling found |
261
|
|
|
unset($e); |
262
|
|
|
|
263
|
|
|
return false; |
264
|
|
|
} |
265
|
|
|
} |
266
|
|
|
|
267
|
|
|
/** |
268
|
|
|
* Attempts to get the next sibling. |
269
|
|
|
* |
270
|
|
|
* @throws ChildNotFoundException |
271
|
|
|
* @throws ParentNotFoundException |
272
|
|
|
*/ |
273
|
30 |
|
public function nextSibling(): AbstractNode |
274
|
|
|
{ |
275
|
30 |
|
if ($this->parent === null) { |
276
|
3 |
|
throw new ParentNotFoundException('Parent is not set for this node.'); |
277
|
|
|
} |
278
|
|
|
|
279
|
27 |
|
return $this->parent->nextChild($this->id); |
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
/** |
283
|
|
|
* Attempts to get the previous sibling. |
284
|
|
|
* |
285
|
|
|
* @throws ChildNotFoundException |
286
|
|
|
* @throws ParentNotFoundException |
287
|
|
|
*/ |
288
|
9 |
|
public function previousSibling(): AbstractNode |
289
|
|
|
{ |
290
|
9 |
|
if ($this->parent === null) { |
291
|
3 |
|
throw new ParentNotFoundException('Parent is not set for this node.'); |
292
|
|
|
} |
293
|
|
|
|
294
|
6 |
|
return $this->parent->previousChild($this->id); |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
/** |
298
|
|
|
* Gets the tag object of this node. |
299
|
|
|
*/ |
300
|
363 |
|
public function getTag(): Tag |
301
|
|
|
{ |
302
|
363 |
|
return $this->tag; |
303
|
|
|
} |
304
|
|
|
|
305
|
|
|
/** |
306
|
|
|
* Replaces the tag for this node. |
307
|
|
|
* |
308
|
|
|
* @param string|Tag $tag |
309
|
|
|
*/ |
310
|
3 |
|
public function setTag($tag): AbstractNode |
311
|
|
|
{ |
312
|
3 |
|
if (\is_string($tag)) { |
313
|
3 |
|
$tag = new Tag($tag); |
314
|
|
|
} |
315
|
|
|
|
316
|
3 |
|
$this->tag = $tag; |
|
|
|
|
317
|
|
|
|
318
|
|
|
// clear any cache |
319
|
3 |
|
$this->clear(); |
320
|
|
|
|
321
|
3 |
|
return $this; |
322
|
|
|
} |
323
|
|
|
|
324
|
|
|
/** |
325
|
|
|
* A wrapper method that simply calls the getAttribute method |
326
|
|
|
* on the tag of this node. |
327
|
|
|
*/ |
328
|
6 |
|
public function getAttributes(): array |
329
|
|
|
{ |
330
|
6 |
|
$attributes = $this->tag->getAttributes(); |
331
|
6 |
|
foreach ($attributes as $name => $attributeDTO) { |
332
|
3 |
|
$attributes[$name] = $attributeDTO->getValue(); |
333
|
|
|
} |
334
|
|
|
|
335
|
6 |
|
return $attributes; |
336
|
|
|
} |
337
|
|
|
|
338
|
|
|
/** |
339
|
|
|
* A wrapper method that simply calls the getAttribute method |
340
|
|
|
* on the tag of this node. |
341
|
|
|
*/ |
342
|
351 |
|
public function getAttribute(string $key): ?string |
343
|
|
|
{ |
344
|
|
|
try { |
345
|
351 |
|
$attributeDTO = $this->tag->getAttribute($key); |
346
|
330 |
|
} catch (AttributeNotFoundException $e) { |
347
|
|
|
// no attribute with this key exists, returning null. |
348
|
330 |
|
unset($e); |
349
|
|
|
|
350
|
330 |
|
return null; |
351
|
|
|
} |
352
|
|
|
|
353
|
147 |
|
return $attributeDTO->getValue(); |
354
|
|
|
} |
355
|
|
|
|
356
|
|
|
/** |
357
|
|
|
* A wrapper method that simply calls the hasAttribute method |
358
|
|
|
* on the tag of this node. |
359
|
|
|
*/ |
360
|
123 |
|
public function hasAttribute(string $key): bool |
361
|
|
|
{ |
362
|
123 |
|
return $this->tag->hasAttribute($key); |
363
|
|
|
} |
364
|
|
|
|
365
|
|
|
/** |
366
|
|
|
* A wrapper method that simply calls the setAttribute method |
367
|
|
|
* on the tag of this node. |
368
|
|
|
*/ |
369
|
24 |
|
public function setAttribute(string $key, ?string $value, bool $doubleQuote = true): AbstractNode |
370
|
|
|
{ |
371
|
24 |
|
$this->tag->setAttribute($key, $value, $doubleQuote); |
372
|
|
|
|
373
|
|
|
//clear any cache |
374
|
24 |
|
$this->clear(); |
375
|
|
|
|
376
|
24 |
|
return $this; |
377
|
|
|
} |
378
|
|
|
|
379
|
|
|
/** |
380
|
|
|
* A wrapper method that simply calls the removeAttribute method |
381
|
|
|
* on the tag of this node. |
382
|
|
|
*/ |
383
|
3 |
|
public function removeAttribute(string $key): void |
384
|
|
|
{ |
385
|
3 |
|
$this->tag->removeAttribute($key); |
386
|
|
|
|
387
|
|
|
//clear any cache |
388
|
3 |
|
$this->clear(); |
389
|
3 |
|
} |
390
|
|
|
|
391
|
|
|
/** |
392
|
|
|
* A wrapper method that simply calls the removeAllAttributes |
393
|
|
|
* method on the tag of this node. |
394
|
|
|
*/ |
395
|
3 |
|
public function removeAllAttributes(): void |
396
|
|
|
{ |
397
|
3 |
|
$this->tag->removeAllAttributes(); |
398
|
|
|
|
399
|
|
|
//clear any cache |
400
|
3 |
|
$this->clear(); |
401
|
3 |
|
} |
402
|
|
|
|
403
|
|
|
/** |
404
|
|
|
* Function to locate a specific ancestor tag in the path to the root. |
405
|
|
|
* |
406
|
|
|
* @throws ParentNotFoundException |
407
|
|
|
*/ |
408
|
6 |
|
public function ancestorByTag(string $tag): AbstractNode |
409
|
|
|
{ |
410
|
|
|
// Start by including ourselves in the comparison. |
411
|
6 |
|
$node = $this; |
412
|
|
|
|
413
|
|
|
do { |
414
|
6 |
|
if ($node->tag->name() == $tag) { |
415
|
3 |
|
return $node; |
416
|
|
|
} |
417
|
|
|
|
418
|
3 |
|
$node = $node->getParent(); |
419
|
3 |
|
} while ($node !== null); |
420
|
|
|
|
421
|
3 |
|
throw new ParentNotFoundException('Could not find an ancestor with "' . $tag . '" tag'); |
422
|
|
|
} |
423
|
|
|
|
424
|
|
|
/** |
425
|
|
|
* Find elements by css selector. |
426
|
|
|
* |
427
|
|
|
* @throws ChildNotFoundException |
428
|
|
|
* |
429
|
|
|
* @return mixed|Collection|null |
430
|
|
|
*/ |
431
|
288 |
|
public function find(string $selectorString, ?int $nth = null, ?SelectorInterface $selector = null) |
432
|
|
|
{ |
433
|
288 |
|
if (\is_null($selector)) { |
434
|
288 |
|
$selector = new Selector($selectorString); |
435
|
|
|
} |
436
|
|
|
|
437
|
288 |
|
$nodes = $selector->find($this); |
438
|
|
|
|
439
|
288 |
|
if ($nth !== null) { |
440
|
|
|
// return nth-element or array |
441
|
288 |
|
if (isset($nodes[$nth])) { |
442
|
132 |
|
return $nodes[$nth]; |
443
|
|
|
} |
444
|
|
|
|
445
|
255 |
|
return; |
446
|
|
|
} |
447
|
|
|
|
448
|
111 |
|
return $nodes; |
449
|
|
|
} |
450
|
|
|
|
451
|
|
|
/** |
452
|
|
|
* Find node by id. |
453
|
|
|
* |
454
|
|
|
* @throws ChildNotFoundException |
455
|
|
|
* @throws ParentNotFoundException |
456
|
|
|
* |
457
|
|
|
* @return bool|AbstractNode |
458
|
|
|
*/ |
459
|
|
|
public function findById(int $id) |
460
|
|
|
{ |
461
|
|
|
$finder = new Finder($id); |
462
|
|
|
|
463
|
|
|
return $finder->find($this); |
464
|
|
|
} |
465
|
|
|
|
466
|
|
|
/** |
467
|
|
|
* Gets the inner html of this node. |
468
|
|
|
*/ |
469
|
|
|
abstract public function innerHtml(): string; |
470
|
|
|
|
471
|
|
|
/** |
472
|
|
|
* Gets the html of this node, including it's own |
473
|
|
|
* tag. |
474
|
|
|
*/ |
475
|
|
|
abstract public function outerHtml(): string; |
476
|
|
|
|
477
|
|
|
/** |
478
|
|
|
* Gets the text of this node (if there is any text). |
479
|
|
|
*/ |
480
|
|
|
abstract public function text(): string; |
481
|
|
|
|
482
|
|
|
/** |
483
|
|
|
* Check is node type textNode. |
484
|
|
|
*/ |
485
|
|
|
public function isTextNode(): bool |
486
|
|
|
{ |
487
|
|
|
return false; |
488
|
|
|
} |
489
|
|
|
|
490
|
|
|
/** |
491
|
|
|
* Call this when something in the node tree has changed. Like a child has been added |
492
|
|
|
* or a parent has been changed. |
493
|
|
|
*/ |
494
|
|
|
abstract protected function clear(): void; |
495
|
|
|
} |
496
|
|
|
|