1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace voku\helper; |
6
|
|
|
|
7
|
|
|
/** @noinspection PhpHierarchyChecksInspection */ |
8
|
|
|
class SimpleHtmlDom extends AbstractSimpleHtmlDom implements \IteratorAggregate, SimpleHtmlDomInterface |
9
|
|
|
{ |
10
|
|
|
/** |
11
|
|
|
* @var \DOMElement|\DOMNode |
12
|
|
|
*/ |
13
|
|
|
protected $node; |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* @param \DOMElement|\DOMNode $node |
17
|
|
|
*/ |
18
|
107 |
|
public function __construct(\DOMNode $node) |
19
|
|
|
{ |
20
|
107 |
|
$this->node = $node; |
21
|
107 |
|
} |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* @param string $name |
25
|
|
|
* @param array $arguments |
26
|
|
|
* |
27
|
|
|
* @throws \BadMethodCallException |
28
|
|
|
* |
29
|
|
|
* @return SimpleHtmlDomInterface|string|null |
30
|
|
|
*/ |
31
|
9 |
View Code Duplication |
public function __call($name, $arguments) |
|
|
|
|
32
|
|
|
{ |
33
|
9 |
|
$name = \strtolower($name); |
34
|
|
|
|
35
|
9 |
|
if (isset(self::$functionAliases[$name])) { |
36
|
9 |
|
return \call_user_func_array([$this, self::$functionAliases[$name]], $arguments); |
37
|
|
|
} |
38
|
|
|
|
39
|
|
|
throw new \BadMethodCallException('Method does not exist'); |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* Returns children of node. |
44
|
|
|
* |
45
|
|
|
* @param int $idx |
46
|
|
|
* |
47
|
|
|
* @return SimpleHtmlDomInterface|SimpleHtmlDomInterface[]|SimpleHtmlDomNodeInterface|null |
48
|
|
|
*/ |
49
|
2 |
|
public function childNodes(int $idx = -1) |
50
|
|
|
{ |
51
|
2 |
|
$nodeList = $this->getIterator(); |
52
|
|
|
|
53
|
2 |
|
if ($idx === -1) { |
54
|
2 |
|
return $nodeList; |
55
|
|
|
} |
56
|
|
|
|
57
|
2 |
|
return $nodeList[$idx] ?? null; |
58
|
|
|
} |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* Find list of nodes with a CSS selector. |
62
|
|
|
* |
63
|
|
|
* @param string $selector |
64
|
|
|
* @param int|null $idx |
65
|
|
|
* |
66
|
|
|
* @return SimpleHtmlDomInterface|SimpleHtmlDomInterface[]|SimpleHtmlDomNodeInterface |
67
|
|
|
*/ |
68
|
26 |
|
public function find(string $selector, $idx = null) |
69
|
|
|
{ |
70
|
26 |
|
return $this->getHtmlDomParser()->find($selector, $idx); |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
/** |
74
|
|
|
* Find one node with a CSS selector. |
75
|
|
|
* |
76
|
|
|
* @param string $selector |
77
|
|
|
* |
78
|
|
|
* @return SimpleHtmlDomInterface |
79
|
|
|
*/ |
80
|
1 |
|
public function findOne(string $selector): SimpleHtmlDomInterface |
81
|
|
|
{ |
82
|
1 |
|
return $this->find($selector, 0); |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* Find nodes with a CSS selector. |
87
|
|
|
* |
88
|
|
|
* @param string $selector |
89
|
|
|
* |
90
|
|
|
* @return SimpleHtmlDomInterface[]|SimpleHtmlDomNodeInterface |
91
|
|
|
*/ |
92
|
|
|
public function findMulti(string $selector): SimpleHtmlDomNodeInterface |
93
|
|
|
{ |
94
|
|
|
return $this->find($selector, null); |
95
|
|
|
} |
96
|
|
|
|
97
|
|
|
/** |
98
|
|
|
* Returns the first child of node. |
99
|
|
|
* |
100
|
|
|
* @return SimpleHtmlDomInterface|null |
101
|
|
|
*/ |
102
|
4 |
|
public function firstChild() |
103
|
|
|
{ |
104
|
|
|
/** @var \DOMNode|null $node */ |
105
|
4 |
|
$node = $this->node->firstChild; |
106
|
|
|
|
107
|
4 |
|
if ($node === null) { |
108
|
1 |
|
return null; |
109
|
|
|
} |
110
|
|
|
|
111
|
4 |
|
return new static($node); |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
/** |
115
|
|
|
* Returns an array of attributes. |
116
|
|
|
* |
117
|
|
|
* @return array|null |
118
|
|
|
*/ |
119
|
2 |
|
public function getAllAttributes() |
120
|
|
|
{ |
121
|
2 |
|
if ($this->node->hasAttributes()) { |
122
|
2 |
|
$attributes = []; |
123
|
2 |
|
foreach ($this->node->attributes as $attr) { |
124
|
2 |
|
$attributes[$attr->name] = HtmlDomParser::putReplacedBackToPreserveHtmlEntities($attr->value); |
125
|
|
|
} |
126
|
|
|
|
127
|
2 |
|
return $attributes; |
128
|
|
|
} |
129
|
|
|
|
130
|
1 |
|
return null; |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
/** |
134
|
|
|
* Return attribute value. |
135
|
|
|
* |
136
|
|
|
* @param string $name |
137
|
|
|
* |
138
|
|
|
* @return string |
139
|
|
|
*/ |
140
|
14 |
|
public function getAttribute(string $name): string |
141
|
|
|
{ |
142
|
14 |
|
if ($this->node instanceof \DOMElement) { |
143
|
14 |
|
return HtmlDomParser::putReplacedBackToPreserveHtmlEntities( |
144
|
14 |
|
$this->node->getAttribute($name) |
145
|
|
|
); |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
return ''; |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
/** |
152
|
|
|
* Return element by #id. |
153
|
|
|
* |
154
|
|
|
* @param string $id |
155
|
|
|
* |
156
|
|
|
* @return SimpleHtmlDomInterface |
157
|
|
|
*/ |
158
|
1 |
|
public function getElementById(string $id): SimpleHtmlDomInterface |
159
|
|
|
{ |
160
|
1 |
|
return $this->findOne("#${id}"); |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
/** |
164
|
|
|
* Returns elements by #id. |
165
|
|
|
* |
166
|
|
|
* @param string $id |
167
|
|
|
* @param int|null $idx |
168
|
|
|
* |
169
|
|
|
* @return SimpleHtmlDomInterface|SimpleHtmlDomInterface[]|SimpleHtmlDomNodeInterface |
170
|
|
|
*/ |
171
|
|
|
public function getElementsById(string $id, $idx = null) |
172
|
|
|
{ |
173
|
|
|
return $this->find("#${id}", $idx); |
174
|
|
|
} |
175
|
|
|
|
176
|
|
|
/** |
177
|
|
|
* Return elements by .class. |
178
|
|
|
* |
179
|
|
|
* @param string $class |
180
|
|
|
* |
181
|
|
|
* @return SimpleHtmlDomInterface[]|SimpleHtmlDomNodeInterface |
182
|
|
|
*/ |
183
|
|
|
public function getElementByClass(string $class): SimpleHtmlDomNodeInterface |
184
|
|
|
{ |
185
|
|
|
return $this->findMulti(".${class}"); |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
/** |
189
|
|
|
* Return element by tag name. |
190
|
|
|
* |
191
|
|
|
* @param string $name |
192
|
|
|
* |
193
|
|
|
* @return SimpleHtmlDomInterface |
194
|
|
|
*/ |
195
|
|
|
public function getElementByTagName(string $name): SimpleHtmlDomInterface |
196
|
|
|
{ |
197
|
1 |
|
if ($this->node instanceof \DOMElement) { |
198
|
1 |
|
$node = $this->node->getElementsByTagName($name)->item(0); |
199
|
|
|
} else { |
200
|
|
|
$node = null; |
201
|
|
|
} |
202
|
|
|
|
203
|
1 |
|
if ($node === null) { |
204
|
|
|
return new SimpleHtmlDomBlank(); |
205
|
|
|
} |
206
|
|
|
|
207
|
1 |
|
return new static($node); |
208
|
|
|
} |
209
|
|
|
|
210
|
|
|
/** |
211
|
|
|
* Returns elements by tag name. |
212
|
|
|
* |
213
|
|
|
* @param string $name |
214
|
|
|
* @param int|null $idx |
215
|
|
|
* |
216
|
|
|
* @return SimpleHtmlDomInterface|SimpleHtmlDomInterface[]|SimpleHtmlDomNodeInterface |
217
|
|
|
*/ |
218
|
|
View Code Duplication |
public function getElementsByTagName(string $name, $idx = null) |
|
|
|
|
219
|
|
|
{ |
220
|
1 |
|
if ($this->node instanceof \DOMElement) { |
221
|
1 |
|
$nodesList = $this->node->getElementsByTagName($name); |
222
|
|
|
} else { |
223
|
|
|
$nodesList = []; |
224
|
|
|
} |
225
|
|
|
|
226
|
1 |
|
$elements = new SimpleHtmlDomNode(); |
227
|
|
|
|
228
|
1 |
|
foreach ($nodesList as $node) { |
229
|
1 |
|
$elements[] = new static($node); |
230
|
|
|
} |
231
|
|
|
|
232
|
|
|
// return all elements |
233
|
1 |
|
if ($idx === null) { |
234
|
1 |
|
if (\count($elements) === 0) { |
235
|
|
|
return new SimpleHtmlDomNodeBlank(); |
236
|
|
|
} |
237
|
|
|
|
238
|
1 |
|
return $elements; |
239
|
|
|
} |
240
|
|
|
|
241
|
|
|
// handle negative values |
242
|
|
|
if ($idx < 0) { |
243
|
|
|
$idx = \count($elements) + $idx; |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
// return one element |
247
|
|
|
return $elements[$idx] ?? new SimpleHtmlDomBlank(); |
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* Create a new "HtmlDomParser"-object from the current context. |
252
|
|
|
* |
253
|
|
|
* @return HtmlDomParser |
254
|
|
|
*/ |
255
|
|
|
public function getHtmlDomParser(): HtmlDomParser |
256
|
|
|
{ |
257
|
72 |
|
return new HtmlDomParser($this); |
258
|
|
|
} |
259
|
|
|
|
260
|
|
|
/** |
261
|
|
|
* Retrieve an external iterator. |
262
|
|
|
* |
263
|
|
|
* @see http://php.net/manual/en/iteratoraggregate.getiterator.php |
264
|
|
|
* |
265
|
|
|
* @return SimpleHtmlDomNode |
266
|
|
|
* <p> |
267
|
|
|
* An instance of an object implementing <b>Iterator</b> or |
268
|
|
|
* <b>Traversable</b> |
269
|
|
|
* </p> |
270
|
|
|
*/ |
271
|
|
|
public function getIterator(): SimpleHtmlDomNodeInterface |
272
|
|
|
{ |
273
|
3 |
|
$elements = new SimpleHtmlDomNode(); |
274
|
3 |
|
if ($this->node->hasChildNodes()) { |
275
|
3 |
|
foreach ($this->node->childNodes as $node) { |
276
|
3 |
|
$elements[] = new static($node); |
277
|
|
|
} |
278
|
|
|
} |
279
|
|
|
|
280
|
3 |
|
return $elements; |
281
|
|
|
} |
282
|
|
|
|
283
|
|
|
/** |
284
|
|
|
* @return \DOMNode |
285
|
|
|
*/ |
286
|
|
|
public function getNode(): \DOMNode |
287
|
|
|
{ |
288
|
73 |
|
return $this->node; |
289
|
|
|
} |
290
|
|
|
|
291
|
|
|
/** |
292
|
|
|
* Determine if an attribute exists on the element. |
293
|
|
|
* |
294
|
|
|
* @param string $name |
295
|
|
|
* |
296
|
|
|
* @return bool |
297
|
|
|
*/ |
298
|
|
|
public function hasAttribute(string $name): bool |
299
|
|
|
{ |
300
|
2 |
|
if (!$this->node instanceof \DOMElement) { |
301
|
|
|
return false; |
302
|
|
|
} |
303
|
|
|
|
304
|
2 |
|
return $this->node->hasAttribute($name); |
305
|
|
|
} |
306
|
|
|
|
307
|
|
|
/** |
308
|
|
|
* Get dom node's outer html. |
309
|
|
|
* |
310
|
|
|
* @param bool $multiDecodeNewHtmlEntity |
311
|
|
|
* |
312
|
|
|
* @return string |
313
|
|
|
*/ |
314
|
|
|
public function html(bool $multiDecodeNewHtmlEntity = false): string |
315
|
|
|
{ |
316
|
27 |
|
return $this->getHtmlDomParser()->html($multiDecodeNewHtmlEntity); |
317
|
|
|
} |
318
|
|
|
|
319
|
|
|
/** |
320
|
|
|
* Get dom node's inner html. |
321
|
|
|
* |
322
|
|
|
* @param bool $multiDecodeNewHtmlEntity |
323
|
|
|
* |
324
|
|
|
* @return string |
325
|
|
|
*/ |
326
|
|
|
public function innerHtml(bool $multiDecodeNewHtmlEntity = false): string |
327
|
|
|
{ |
328
|
12 |
|
return $this->getHtmlDomParser()->innerHtml($multiDecodeNewHtmlEntity); |
329
|
|
|
} |
330
|
|
|
|
331
|
|
|
/** |
332
|
|
|
* Returns the last child of node. |
333
|
|
|
* |
334
|
|
|
* @return SimpleHtmlDomInterface|null |
335
|
|
|
*/ |
336
|
|
|
public function lastChild() |
337
|
|
|
{ |
338
|
|
|
/** @var \DOMNode|null $node */ |
339
|
4 |
|
$node = $this->node->lastChild; |
340
|
|
|
|
341
|
4 |
|
if ($node === null) { |
342
|
1 |
|
return null; |
343
|
|
|
} |
344
|
|
|
|
345
|
4 |
|
return new static($node); |
346
|
|
|
} |
347
|
|
|
|
348
|
|
|
/** |
349
|
|
|
* Returns the next sibling of node. |
350
|
|
|
* |
351
|
|
|
* @return SimpleHtmlDomInterface|null |
352
|
|
|
*/ |
353
|
|
|
public function nextSibling() |
354
|
|
|
{ |
355
|
|
|
/** @var \DOMNode|null $node */ |
356
|
1 |
|
$node = $this->node->nextSibling; |
357
|
|
|
|
358
|
1 |
|
if ($node === null) { |
359
|
1 |
|
return null; |
360
|
|
|
} |
361
|
|
|
|
362
|
1 |
|
return new static($node); |
363
|
|
|
} |
364
|
|
|
|
365
|
|
|
/** |
366
|
|
|
* Returns the parent of node. |
367
|
|
|
* |
368
|
|
|
* @return SimpleHtmlDomInterface |
369
|
|
|
*/ |
370
|
|
|
public function parentNode(): SimpleHtmlDomInterface |
371
|
|
|
{ |
372
|
1 |
|
return new static($this->node->parentNode); |
|
|
|
|
373
|
|
|
} |
374
|
|
|
|
375
|
|
|
/** |
376
|
|
|
* Nodes can get partially destroyed in which they're still an |
377
|
|
|
* actual DOM node (such as \DOMElement) but almost their entire |
378
|
|
|
* body is gone, including the `nodeType` attribute. |
379
|
|
|
* |
380
|
|
|
* @return bool true if node has been destroyed |
381
|
|
|
*/ |
382
|
|
|
public function isRemoved(): bool |
383
|
|
|
{ |
384
|
|
|
return !isset($this->node->nodeType); |
385
|
|
|
} |
386
|
|
|
|
387
|
|
|
/** |
388
|
|
|
* Returns the previous sibling of node. |
389
|
|
|
* |
390
|
|
|
* @return SimpleHtmlDomInterface|null |
391
|
|
|
*/ |
392
|
|
|
public function previousSibling() |
393
|
|
|
{ |
394
|
|
|
/** @var \DOMNode|null $node */ |
395
|
1 |
|
$node = $this->node->previousSibling; |
396
|
|
|
|
397
|
1 |
|
if ($node === null) { |
398
|
1 |
|
return null; |
399
|
|
|
} |
400
|
|
|
|
401
|
1 |
|
return new static($node); |
402
|
|
|
} |
403
|
|
|
|
404
|
|
|
/** |
405
|
|
|
* Replace child node. |
406
|
|
|
* |
407
|
|
|
* @param string $string |
408
|
|
|
* |
409
|
|
|
* @return SimpleHtmlDomInterface |
410
|
|
|
*/ |
411
|
|
|
protected function replaceChildWithString(string $string): SimpleHtmlDomInterface |
412
|
|
|
{ |
413
|
7 |
|
if (!empty($string)) { |
414
|
6 |
|
$newDocument = new HtmlDomParser($string); |
415
|
|
|
|
416
|
6 |
|
$tmpDomString = $this->normalizeStringForComparision($newDocument); |
417
|
6 |
|
$tmpStr = $this->normalizeStringForComparision($string); |
418
|
6 |
|
if ($tmpDomString !== $tmpStr) { |
419
|
|
|
throw new \RuntimeException( |
420
|
|
|
'Not valid HTML fragment!' . "\n" . |
421
|
|
|
$tmpDomString . "\n" . |
422
|
|
|
$tmpStr |
423
|
|
|
); |
424
|
|
|
} |
425
|
|
|
} |
426
|
|
|
|
427
|
7 |
|
if (\count($this->node->childNodes) > 0) { |
428
|
7 |
|
foreach ($this->node->childNodes as $node) { |
429
|
7 |
|
$this->node->removeChild($node); |
430
|
|
|
} |
431
|
|
|
} |
432
|
|
|
|
433
|
7 |
|
if (!empty($newDocument)) { |
434
|
6 |
|
$newDocument = $this->cleanHtmlWrapper($newDocument); |
435
|
6 |
|
$ownerDocument = $this->node->ownerDocument; |
436
|
|
|
if ( |
437
|
6 |
|
$ownerDocument !== null |
438
|
|
|
&& |
439
|
6 |
|
$newDocument->getDocument()->documentElement !== null |
440
|
|
|
) { |
441
|
6 |
|
$newNode = $ownerDocument->importNode($newDocument->getDocument()->documentElement, true); |
442
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
443
|
6 |
|
$this->node->appendChild($newNode); |
444
|
|
|
} |
445
|
|
|
} |
446
|
|
|
|
447
|
7 |
|
return $this; |
448
|
|
|
} |
449
|
|
|
|
450
|
|
|
/** |
451
|
|
|
* Replace this node with text |
452
|
|
|
* |
453
|
|
|
* @param string $string |
454
|
|
|
* |
455
|
|
|
* @return SimpleHtmlDomInterface |
456
|
|
|
*/ |
457
|
|
|
protected function replaceTextWithString($string): SimpleHtmlDomInterface |
458
|
|
|
{ |
459
|
1 |
|
if (empty($string)) { |
460
|
1 |
|
$this->node->parentNode->removeChild($this->node); |
461
|
|
|
|
462
|
1 |
|
return $this; |
463
|
|
|
} |
464
|
|
|
|
465
|
1 |
|
$ownerDocument = $this->node->ownerDocument; |
466
|
1 |
|
if ($ownerDocument !== null) { |
467
|
1 |
|
$newElement = $ownerDocument->createTextNode($string); |
468
|
1 |
|
$newNode = $ownerDocument->importNode($newElement, true); |
469
|
1 |
|
$this->node->parentNode->replaceChild($newNode, $this->node); |
470
|
1 |
|
$this->node = $newNode; |
471
|
|
|
} |
472
|
|
|
|
473
|
1 |
|
return $this; |
474
|
|
|
} |
475
|
|
|
|
476
|
|
|
/** |
477
|
|
|
* Replace this node. |
478
|
|
|
* |
479
|
|
|
* @param string $string |
480
|
|
|
* |
481
|
|
|
* @return SimpleHtmlDomInterface |
482
|
|
|
*/ |
483
|
|
|
protected function replaceNodeWithString(string $string): SimpleHtmlDomInterface |
484
|
|
|
{ |
485
|
4 |
|
if (empty($string)) { |
486
|
2 |
|
$this->node->parentNode->removeChild($this->node); |
487
|
|
|
|
488
|
2 |
|
return $this; |
489
|
|
|
} |
490
|
|
|
|
491
|
3 |
|
$newDocument = new HtmlDomParser($string); |
492
|
|
|
|
493
|
3 |
|
$tmpDomOuterTextString = $this->normalizeStringForComparision($newDocument); |
494
|
3 |
|
$tmpStr = $this->normalizeStringForComparision($string); |
495
|
3 |
|
if ($tmpDomOuterTextString !== $tmpStr) { |
496
|
|
|
throw new \RuntimeException( |
497
|
|
|
'Not valid HTML fragment!' . "\n" |
498
|
|
|
. $tmpDomOuterTextString . "\n" . |
499
|
|
|
$tmpStr |
500
|
|
|
); |
501
|
|
|
} |
502
|
|
|
|
503
|
3 |
|
$newDocument = $this->cleanHtmlWrapper($newDocument, true); |
504
|
3 |
|
$ownerDocument = $this->node->ownerDocument; |
505
|
|
|
if ( |
506
|
3 |
|
$ownerDocument === null |
507
|
|
|
|| |
508
|
3 |
|
$newDocument->getDocument()->documentElement === null |
509
|
|
|
) { |
510
|
|
|
return $this; |
511
|
|
|
} |
512
|
|
|
|
513
|
3 |
|
$newNode = $ownerDocument->importNode($newDocument->getDocument()->documentElement, true); |
514
|
|
|
|
515
|
3 |
|
$this->node->parentNode->replaceChild($newNode, $this->node); |
516
|
3 |
|
$this->node = $newNode; |
517
|
|
|
|
518
|
|
|
// Remove head element, preserving child nodes. (again) |
519
|
|
View Code Duplication |
if ( |
|
|
|
|
520
|
3 |
|
$this->node->parentNode instanceof \DOMElement |
521
|
|
|
&& |
522
|
3 |
|
$newDocument->getIsDOMDocumentCreatedWithoutHeadWrapper() |
523
|
|
|
) { |
524
|
2 |
|
$html = $this->node->parentNode->getElementsByTagName('head')[0]; |
525
|
2 |
|
if ($this->node->parentNode->ownerDocument !== null) { |
526
|
2 |
|
$fragment = $this->node->parentNode->ownerDocument->createDocumentFragment(); |
527
|
2 |
|
if ($html !== null) { |
528
|
|
|
/** @var \DOMNode $html */ |
529
|
1 |
|
while ($html->childNodes->length > 0) { |
530
|
1 |
|
$tmpNode = $html->childNodes->item(0); |
531
|
1 |
|
if ($tmpNode !== null) { |
532
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
533
|
1 |
|
$fragment->appendChild($tmpNode); |
534
|
|
|
} |
535
|
|
|
} |
536
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
537
|
1 |
|
$html->parentNode->replaceChild($fragment, $html); |
538
|
|
|
} |
539
|
|
|
} |
540
|
|
|
} |
541
|
|
|
|
542
|
3 |
|
return $this; |
543
|
|
|
} |
544
|
|
|
|
545
|
|
|
/** |
546
|
|
|
* Normalize the given input for comparision. |
547
|
|
|
* |
548
|
|
|
* @param HtmlDomParser|string $input |
549
|
|
|
* |
550
|
|
|
* @return string |
551
|
|
|
*/ |
552
|
|
|
private function normalizeStringForComparision($input): string |
553
|
|
|
{ |
554
|
9 |
|
if ($input instanceof HtmlDomParser) { |
555
|
9 |
|
$string = $input->outerText(); |
556
|
|
|
|
557
|
9 |
|
if ($input->getIsDOMDocumentCreatedWithoutHeadWrapper()) { |
558
|
|
|
/** @noinspection HtmlRequiredTitleElement */ |
559
|
9 |
|
$string = \str_replace(['<head>', '</head>'], '', $string); |
560
|
|
|
} |
561
|
|
|
} else { |
562
|
9 |
|
$string = (string) $input; |
563
|
|
|
} |
564
|
|
|
|
565
|
|
|
return |
566
|
9 |
|
\urlencode( |
567
|
9 |
|
\urldecode( |
568
|
9 |
|
\trim( |
569
|
9 |
|
\str_replace( |
570
|
|
|
[ |
571
|
9 |
|
' ', |
572
|
|
|
"\n", |
573
|
|
|
"\r", |
574
|
|
|
'/>', |
575
|
|
|
], |
576
|
|
|
[ |
577
|
9 |
|
'', |
578
|
|
|
'', |
579
|
|
|
'', |
580
|
|
|
'>', |
581
|
|
|
], |
582
|
9 |
|
\strtolower($string) |
583
|
|
|
) |
584
|
|
|
) |
585
|
|
|
) |
586
|
|
|
); |
587
|
|
|
} |
588
|
|
|
|
589
|
|
|
/** |
590
|
|
|
* @param HtmlDomParser $newDocument |
591
|
|
|
* @param bool $removeExtraHeadTag |
592
|
|
|
* |
593
|
|
|
* @return HtmlDomParser |
594
|
|
|
*/ |
595
|
|
|
protected function cleanHtmlWrapper(HtmlDomParser $newDocument, $removeExtraHeadTag = false): HtmlDomParser |
596
|
|
|
{ |
597
|
|
|
if ( |
598
|
9 |
|
$newDocument->getIsDOMDocumentCreatedWithoutHtml() |
599
|
|
|
|| |
600
|
9 |
|
$newDocument->getIsDOMDocumentCreatedWithoutHtmlWrapper() |
601
|
|
|
) { |
602
|
|
|
|
603
|
|
|
// Remove doc-type node. |
604
|
9 |
|
if ($newDocument->getDocument()->doctype !== null) { |
605
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
606
|
|
|
$newDocument->getDocument()->doctype->parentNode->removeChild($newDocument->getDocument()->doctype); |
607
|
|
|
} |
608
|
|
|
|
609
|
|
|
// Remove html element, preserving child nodes. |
610
|
9 |
|
$html = $newDocument->getDocument()->getElementsByTagName('html')->item(0); |
611
|
9 |
|
$fragment = $newDocument->getDocument()->createDocumentFragment(); |
612
|
9 |
|
if ($html !== null) { |
613
|
6 |
|
while ($html->childNodes->length > 0) { |
614
|
6 |
|
$tmpNode = $html->childNodes->item(0); |
615
|
6 |
|
if ($tmpNode !== null) { |
616
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
617
|
6 |
|
$fragment->appendChild($tmpNode); |
618
|
|
|
} |
619
|
|
|
} |
620
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
621
|
6 |
|
$html->parentNode->replaceChild($fragment, $html); |
622
|
|
|
} |
623
|
|
|
|
624
|
|
|
// Remove body element, preserving child nodes. |
625
|
9 |
|
$body = $newDocument->getDocument()->getElementsByTagName('body')->item(0); |
626
|
9 |
|
$fragment = $newDocument->getDocument()->createDocumentFragment(); |
627
|
9 |
|
if ($body instanceof \DOMElement) { |
628
|
4 |
|
while ($body->childNodes->length > 0) { |
629
|
4 |
|
$tmpNode = $body->childNodes->item(0); |
630
|
4 |
|
if ($tmpNode !== null) { |
631
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
632
|
4 |
|
$fragment->appendChild($tmpNode); |
633
|
|
|
} |
634
|
|
|
} |
635
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
636
|
4 |
|
$body->parentNode->replaceChild($fragment, $body); |
637
|
|
|
|
638
|
|
|
// At this point DOMDocument still added a "<p>"-wrapper around our string, |
639
|
|
|
// so we replace it with "<simpleHtmlDomP>" and delete this at the ending ... |
640
|
4 |
|
$item = $newDocument->getDocument()->getElementsByTagName('p')->item(0); |
641
|
4 |
|
if ($item !== null) { |
642
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
643
|
4 |
|
$this->changeElementName($item, 'simpleHtmlDomP'); |
644
|
|
|
} |
645
|
|
|
} |
646
|
|
|
} |
647
|
|
|
|
648
|
|
|
// Remove head element, preserving child nodes. |
649
|
|
View Code Duplication |
if ( |
|
|
|
|
650
|
9 |
|
$removeExtraHeadTag |
651
|
|
|
&& |
652
|
9 |
|
$this->node->parentNode instanceof \DOMElement |
653
|
|
|
&& |
654
|
9 |
|
$newDocument->getIsDOMDocumentCreatedWithoutHeadWrapper() |
655
|
|
|
) { |
656
|
2 |
|
$html = $this->node->parentNode->getElementsByTagName('head')[0]; |
657
|
2 |
|
if ($this->node->parentNode->ownerDocument !== null) { |
658
|
2 |
|
$fragment = $this->node->parentNode->ownerDocument->createDocumentFragment(); |
659
|
2 |
|
if ($html !== null) { |
660
|
|
|
/** @var \DOMNode $html */ |
661
|
|
|
while ($html->childNodes->length > 0) { |
662
|
|
|
$tmpNode = $html->childNodes->item(0); |
663
|
|
|
if ($tmpNode !== null) { |
664
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
665
|
|
|
$fragment->appendChild($tmpNode); |
666
|
|
|
} |
667
|
|
|
} |
668
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
669
|
|
|
$html->parentNode->replaceChild($fragment, $html); |
670
|
|
|
} |
671
|
|
|
} |
672
|
|
|
} |
673
|
|
|
|
674
|
9 |
|
return $newDocument; |
675
|
|
|
} |
676
|
|
|
|
677
|
|
|
/** |
678
|
|
|
* Change the name of a tag in a "DOMNode". |
679
|
|
|
* |
680
|
|
|
* @param \DOMNode $node |
681
|
|
|
* @param string $name |
682
|
|
|
* |
683
|
|
|
* @return \DOMElement|false |
684
|
|
|
* <p>DOMElement a new instance of class DOMElement or false |
685
|
|
|
* if an error occured.</p> |
686
|
|
|
*/ |
687
|
|
|
protected function changeElementName(\DOMNode $node, string $name) |
688
|
|
|
{ |
689
|
4 |
|
$ownerDocument = $node->ownerDocument; |
690
|
4 |
|
if ($ownerDocument) { |
691
|
4 |
|
$newNode = $ownerDocument->createElement($name); |
692
|
|
|
} else { |
693
|
|
|
return false; |
694
|
|
|
} |
695
|
|
|
|
696
|
4 |
|
foreach ($node->childNodes as $child) { |
697
|
4 |
|
$child = $ownerDocument->importNode($child, true); |
698
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
699
|
4 |
|
$newNode->appendChild($child); |
700
|
|
|
} |
701
|
|
|
|
702
|
4 |
|
foreach ($node->attributes as $attrName => $attrNode) { |
703
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
704
|
|
|
$newNode->setAttribute($attrName, $attrNode); |
705
|
|
|
} |
706
|
|
|
|
707
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
708
|
4 |
|
$newNode->ownerDocument->replaceChild($newNode, $node); |
709
|
|
|
|
710
|
4 |
|
return $newNode; |
711
|
|
|
} |
712
|
|
|
|
713
|
|
|
/** |
714
|
|
|
* Set attribute value. |
715
|
|
|
* |
716
|
|
|
* @param string $name <p>The name of the html-attribute.</p> |
717
|
|
|
* @param string|null $value <p>Set to NULL or empty string, to remove the attribute.</p> |
718
|
|
|
* @param bool $strict </p> |
719
|
|
|
* $value must be NULL, to remove the attribute, |
720
|
|
|
* so that you can set an empty string as attribute-value e.g. autofocus="" |
721
|
|
|
* </p> |
722
|
|
|
* |
723
|
|
|
* @return SimpleHtmlDomInterface |
724
|
|
|
*/ |
725
|
|
|
public function setAttribute(string $name, $value = null, bool $strict = false): SimpleHtmlDomInterface |
726
|
|
|
{ |
727
|
|
|
if ( |
728
|
10 |
|
($strict && $value === null) |
729
|
|
|
|| |
730
|
10 |
|
(!$strict && empty($value)) |
731
|
|
|
) { |
732
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
733
|
2 |
|
$this->removeAttribute($name); |
734
|
10 |
|
} elseif (\method_exists($this->node, 'setAttribute')) { |
735
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
736
|
10 |
|
$this->node->setAttribute($name, $value); |
|
|
|
|
737
|
|
|
} |
738
|
|
|
|
739
|
10 |
|
return $this; |
|
|
|
|
740
|
|
|
} |
741
|
|
|
|
742
|
|
|
/** |
743
|
|
|
* @param string|string[]|null $value <p> |
744
|
|
|
* null === get the current input value |
745
|
|
|
* text === set a new input value |
746
|
|
|
* </p> |
747
|
|
|
* |
748
|
|
|
* @return string|string[]|null |
749
|
|
|
*/ |
750
|
|
|
public function val($value = null) |
751
|
|
|
{ |
752
|
1 |
|
if ($value === null) { |
753
|
|
|
if ( |
754
|
1 |
|
$this->tag === 'input' |
|
|
|
|
755
|
|
|
&& |
756
|
|
|
( |
757
|
1 |
|
$this->getAttribute('type') === 'text' |
758
|
|
|
|| |
759
|
1 |
|
!$this->hasAttribute('type') |
760
|
|
|
) |
761
|
|
|
) { |
762
|
1 |
|
return $this->getAttribute('value'); |
763
|
|
|
} |
764
|
|
|
|
765
|
|
|
if ( |
766
|
1 |
|
$this->hasAttribute('checked') |
767
|
|
|
&& |
768
|
1 |
|
\in_array($this->getAttribute('type'), ['checkbox', 'radio'], true) |
769
|
|
|
) { |
770
|
1 |
|
return $this->getAttribute('value'); |
771
|
|
|
} |
772
|
|
|
|
773
|
1 |
|
if ($this->node->nodeName === 'select') { |
774
|
|
|
$valuesFromDom = []; |
775
|
|
|
$options = $this->getElementsByTagName('option'); |
776
|
|
|
if ($options instanceof SimpleHtmlDomNode) { |
777
|
|
|
foreach ($options as $option) { |
778
|
|
|
if ($this->hasAttribute('checked')) { |
779
|
|
|
/** @noinspection UnnecessaryCastingInspection */ |
780
|
|
|
$valuesFromDom[] = (string) $option->getAttribute('value'); |
781
|
|
|
} |
782
|
|
|
} |
783
|
|
|
} |
784
|
|
|
|
785
|
|
|
if (\count($valuesFromDom) === 0) { |
786
|
|
|
return null; |
787
|
|
|
} |
788
|
|
|
|
789
|
|
|
return $valuesFromDom; |
|
|
|
|
790
|
|
|
} |
791
|
|
|
|
792
|
1 |
|
if ($this->node->nodeName === 'textarea') { |
793
|
1 |
|
return $this->node->nodeValue; |
794
|
|
|
} |
795
|
|
|
} else { |
796
|
|
|
/** @noinspection NestedPositiveIfStatementsInspection */ |
797
|
1 |
|
if (\in_array($this->getAttribute('type'), ['checkbox', 'radio'], true)) { |
798
|
1 |
|
if ($value === $this->getAttribute('value')) { |
799
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
800
|
1 |
|
$this->setAttribute('checked', 'checked'); |
801
|
|
|
} else { |
802
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
803
|
1 |
|
$this->removeAttribute('checked'); |
804
|
|
|
} |
805
|
1 |
|
} elseif ($this->node instanceof \DOMElement && $this->node->nodeName === 'select') { |
806
|
|
|
foreach ($this->node->getElementsByTagName('option') as $option) { |
807
|
|
|
/** @var \DOMElement $option */ |
808
|
|
|
if ($value === $option->getAttribute('value')) { |
809
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
810
|
|
|
$option->setAttribute('selected', 'selected'); |
811
|
|
|
} else { |
812
|
|
|
$option->removeAttribute('selected'); |
813
|
|
|
} |
814
|
|
|
} |
815
|
1 |
|
} elseif ($this->node->nodeName === 'input' && \is_string($value)) { |
816
|
|
|
// Set value for input elements |
817
|
|
|
/** @noinspection UnusedFunctionResultInspection */ |
818
|
1 |
|
$this->setAttribute('value', $value); |
819
|
1 |
|
} elseif ($this->node->nodeName === 'textarea' && \is_string($value)) { |
820
|
1 |
|
$this->node->nodeValue = $value; |
821
|
|
|
} |
822
|
|
|
} |
823
|
|
|
|
824
|
1 |
|
return null; |
825
|
|
|
} |
826
|
|
|
|
827
|
|
|
/** |
828
|
|
|
* Remove attribute. |
829
|
|
|
* |
830
|
|
|
* @param string $name <p>The name of the html-attribute.</p> |
831
|
|
|
* |
832
|
|
|
* @return SimpleHtmlDomInterface |
833
|
|
|
*/ |
834
|
|
|
public function removeAttribute(string $name): SimpleHtmlDomInterface |
835
|
|
|
{ |
836
|
2 |
|
if (\method_exists($this->node, 'removeAttribute')) { |
837
|
2 |
|
$this->node->removeAttribute($name); |
|
|
|
|
838
|
|
|
} |
839
|
|
|
|
840
|
2 |
|
return $this; |
|
|
|
|
841
|
|
|
} |
842
|
|
|
|
843
|
|
|
/** |
844
|
|
|
* Get dom node's plain text. |
845
|
|
|
* |
846
|
|
|
* @return string |
847
|
|
|
*/ |
848
|
|
|
public function text(): string |
849
|
|
|
{ |
850
|
17 |
|
return $this->getHtmlDomParser()->fixHtmlOutput($this->node->textContent); |
851
|
|
|
} |
852
|
|
|
} |
853
|
|
|
|
Duplicated code is one of the most pungent code smells. If you need to duplicate the same code in three or more different places, we strongly encourage you to look into extracting the code into a single class or operation.
You can also find more detailed suggestions in the “Code” section of your repository.