1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* Class Document |
4
|
|
|
* |
5
|
|
|
* @filesource Document.php |
6
|
|
|
* @created 05.05.2017 |
7
|
|
|
* @package chillerlan\PrototypeDOM |
8
|
|
|
* @author Smiley <[email protected]> |
9
|
|
|
* @copyright 2017 Smiley |
10
|
|
|
* @license MIT |
11
|
|
|
*/ |
12
|
|
|
|
13
|
|
|
namespace chillerlan\PrototypeDOM; |
14
|
|
|
|
15
|
|
|
use chillerlan\PrototypeDOM\Node\{ |
16
|
|
|
Attr, CdataSection, CharacterData, Comment, DocumentFragment, DocumentType, Element, Entity, |
17
|
|
|
EntityReference, Node, Notation, ProcessingInstruction, PrototypeHTMLElement, PrototypeNode, Text, |
18
|
|
|
}; |
19
|
|
|
use DOMDocument, DOMException, DOMNode, DOMNodeList, DOMXPath; |
20
|
|
|
use Symfony\Component\CssSelector\CssSelectorConverter; |
21
|
|
|
|
22
|
|
|
use function count, in_array, is_file, is_iterable, is_readable, is_string; |
23
|
|
|
|
24
|
|
|
use const LIBXML_COMPACT, LIBXML_HTML_NODEFDTD, LIBXML_HTML_NOIMPLIED, LIBXML_NOERROR, LIBXML_NONET, XML_ELEMENT_NODE; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* |
28
|
|
|
*/ |
29
|
|
|
class Document extends DOMDocument{ |
30
|
|
|
|
31
|
|
|
protected const NODE_CLASSES = [ |
32
|
|
|
'DOMAttr' => Attr::class, |
33
|
|
|
'DOMCdataSection' => CdataSection::class, |
34
|
|
|
'DOMCharacterData' => CharacterData::class, |
35
|
|
|
'DOMComment' => Comment::class, |
36
|
|
|
'DOMDocumentFragment' => DocumentFragment::class, |
37
|
|
|
'DOMDocumentType' => DocumentType::class, |
38
|
|
|
'DOMElement' => Element::class, |
39
|
|
|
'DOMEntity' => Entity::class, |
40
|
|
|
'DOMEntityReference' => EntityReference::class, |
41
|
|
|
'DOMNode' => Node::class, |
42
|
|
|
'DOMNotation' => Notation::class, |
43
|
|
|
'DOMProcessingInstruction' => ProcessingInstruction::class, |
44
|
42 |
|
'DOMText' => Text::class, |
45
|
42 |
|
]; |
46
|
|
|
|
47
|
42 |
|
protected const LOAD_OPTIONS = LIBXML_COMPACT | LIBXML_NONET | LIBXML_HTML_NODEFDTD | LIBXML_HTML_NOIMPLIED | LIBXML_NOERROR; |
48
|
42 |
|
|
49
|
|
|
protected CssSelectorConverter $cssSelectorConverter; |
50
|
|
|
|
51
|
42 |
|
/** |
52
|
2 |
|
* Document constructor. |
53
|
|
|
* |
54
|
|
|
* @param string|\DOMNodeList|\chillerlan\PrototypeDOM\NodeList $content |
55
|
42 |
|
* @param bool|null $xml |
56
|
|
|
* @param string|null $version |
57
|
|
|
* @param string|null $encoding |
58
|
|
|
*/ |
59
|
|
|
public function __construct($content = null, bool $xml = null, string $version = null, string $encoding = null){ |
60
|
|
|
parent::__construct($version ?? '1.0', $encoding ?? 'UTF-8'); |
61
|
|
|
|
62
|
1 |
|
foreach($this::NODE_CLASSES as $baseClass => $extendedClass){ |
63
|
1 |
|
$this->registerNodeClass($baseClass, $extendedClass); |
64
|
|
|
} |
65
|
|
|
|
66
|
1 |
|
if($content !== null){ |
67
|
1 |
|
$this->loadDocument($content, $xml); |
68
|
|
|
} |
69
|
1 |
|
|
70
|
1 |
|
$this->cssSelectorConverter = new CssSelectorConverter; |
71
|
|
|
} |
72
|
|
|
|
73
|
1 |
|
/********* |
74
|
1 |
|
* magic * |
75
|
|
|
*********/ |
76
|
1 |
|
|
77
|
|
|
/** |
78
|
|
|
* @return string|null |
79
|
|
|
*/ |
80
|
1 |
|
public function getTitle():?string{ |
81
|
|
|
return $this->select(['head > title'])->item(0)->nodeValue ?? null; |
82
|
1 |
|
} |
83
|
|
|
|
84
|
|
|
/** |
85
|
|
|
* @throws \DOMException |
86
|
|
|
*/ |
87
|
|
|
public function setTitle(string $title):void{ |
88
|
|
|
$currentTitle = $this->select(['head > title'])->item(0); |
89
|
|
|
|
90
|
2 |
|
if($currentTitle instanceof Element){ |
91
|
|
|
$currentTitle->update($title); |
92
|
|
|
|
93
|
2 |
|
return; |
94
|
1 |
|
} |
95
|
1 |
|
|
96
|
|
|
$head = $this->select(['head'])->item(0); |
97
|
|
|
$currentTitle = $this->newElement('title')->update($title); |
98
|
|
|
|
99
|
|
|
if(!$head){ |
100
|
1 |
|
$html = $this->select(['html'])->first(); |
101
|
1 |
|
|
102
|
|
|
if(!$html instanceof PrototypeHTMLElement){ |
103
|
1 |
|
throw new DOMException('html header missing'); |
104
|
1 |
|
} |
105
|
1 |
|
|
106
|
|
|
$head = $this->newElement('head'); |
107
|
1 |
|
$html->insert_top($head); |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
$head->insert($currentTitle); |
|
|
|
|
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* @param string|\DOMNodeList|\chillerlan\PrototypeDOM\NodeList $content |
115
|
|
|
* @param bool|null $xml |
116
|
8 |
|
* |
117
|
|
|
* @return \chillerlan\PrototypeDOM\Document |
118
|
|
|
* @throws \DOMException |
119
|
8 |
|
*/ |
120
|
8 |
|
public function loadDocument($content, bool $xml = null):Document{ |
121
|
7 |
|
|
122
|
7 |
|
if($content instanceof NodeList){ |
123
|
7 |
|
return $this->insertNodeList($content); |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
if($content instanceof DOMNodeList){ |
127
|
|
|
return $this->insertNodeList(new NodeList($content)); |
128
|
|
|
} |
129
|
|
|
|
130
|
|
|
if(!is_string($content)){ |
|
|
|
|
131
|
|
|
throw new DOMException('invalid document content'); |
132
|
|
|
} |
133
|
|
|
|
134
|
|
|
if(is_file($content) && is_readable($content)){ |
135
|
8 |
|
return $this->loadDocumentFile($content, $xml); |
136
|
8 |
|
} |
137
|
8 |
|
|
138
|
|
|
return $this->loadDocumentString($content, $xml); |
139
|
8 |
|
} |
140
|
|
|
|
141
|
|
|
/** |
142
|
|
|
* @throws \DOMException |
143
|
|
|
*/ |
144
|
|
|
public function loadDocumentFile(string $file, bool $xml = null, int $options = null):Document{ |
145
|
|
|
$options = $options ?? $this::LOAD_OPTIONS; |
146
|
|
|
|
147
|
3 |
|
$result = $xml === true |
148
|
3 |
|
? $this->load($file, $options) |
149
|
|
|
: $this->loadHTMLFile($file, $options); |
150
|
3 |
|
|
151
|
3 |
|
if($result === false){ |
|
|
|
|
152
|
|
|
throw new DOMException('failed to load document from file: '.$file); // @codeCoverageIgnore |
153
|
|
|
} |
154
|
3 |
|
|
155
|
|
|
return $this; |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
/** |
159
|
|
|
* @throws \DOMException |
160
|
|
|
*/ |
161
|
|
|
public function loadDocumentString(string $documentSource, bool $xml = null, int $options = null):Document{ |
162
|
|
|
$options = $options ?? $this::LOAD_OPTIONS; |
163
|
|
|
|
164
|
|
|
$result = $xml === true |
165
|
|
|
? $this->loadXML($documentSource, $options) |
166
|
|
|
: $this->loadHTML($documentSource, $options); |
167
|
|
|
|
168
|
41 |
|
if($result === false){ |
|
|
|
|
169
|
41 |
|
throw new DOMException('failed to load document from string'); // @codeCoverageIgnore |
170
|
|
|
} |
171
|
|
|
|
172
|
|
|
return $this; |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
/** |
176
|
|
|
* @throws \DOMException |
177
|
|
|
*/ |
178
|
41 |
|
public function toNodeList($content):NodeList{ |
179
|
41 |
|
|
180
|
|
|
if($content instanceof NodeList){ |
181
|
|
|
return $content; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
if($content instanceof DOMNode || $content instanceof PrototypeNode){ |
185
|
|
|
return new NodeList([$content]); |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
if($content instanceof DOMNodeList || is_iterable($content)){ |
189
|
41 |
|
return new NodeList($content); |
190
|
41 |
|
} |
191
|
|
|
|
192
|
|
|
if(is_string($content)){ |
193
|
|
|
$document = new self; |
194
|
|
|
$document->loadHTML('<html lang="en"><body id="-import-content">'.$content.'</body></html>'); |
195
|
|
|
|
196
|
|
|
return $document->toNodeList($document->getElementById('-import-content')->childNodes); |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
throw new DOMException('invalid content'); |
200
|
1 |
|
} |
201
|
1 |
|
|
202
|
|
|
/*********** |
203
|
1 |
|
* generic * |
204
|
|
|
***********/ |
205
|
1 |
|
|
206
|
1 |
|
/** |
207
|
|
|
* |
208
|
|
|
*/ |
209
|
|
|
public function selector2xpath(string $selector, string $axis = null):string{ |
210
|
|
|
return $this->cssSelectorConverter->toXPath($selector, $axis ?? '//'); |
211
|
1 |
|
} |
212
|
|
|
|
213
|
|
|
/** |
214
|
|
|
* |
215
|
|
|
*/ |
216
|
|
|
public function query(string $xpath, DOMNode $contextNode = null):?NodeList{ |
217
|
|
|
$q = (new DOMXPath($this))->query($xpath, $contextNode); |
218
|
|
|
|
219
|
2 |
|
return $q !== false ? new NodeList($q) : null; |
220
|
|
|
} |
221
|
|
|
|
222
|
2 |
|
/** |
223
|
2 |
|
* |
224
|
|
|
*/ |
225
|
|
|
public function querySelectorAll(string $selector, DOMNode $contextNode = null, string $axis = null):?NodeList{ |
226
|
2 |
|
return $this->query($this->cssSelectorConverter->toXPath($selector, $axis ?? 'descendant-or-self::'), $contextNode); |
227
|
|
|
} |
228
|
|
|
|
229
|
|
|
/** |
230
|
|
|
* |
231
|
|
|
*/ |
232
|
|
|
public function removeElementsBySelector(array $selectors, DOMNode $contextNode = null, string $axis = null):Document{ |
233
|
|
|
$nodes = $this->select($selectors, $contextNode, $axis ?? 'descendant-or-self::'); |
234
|
|
|
|
235
|
|
|
if(count($nodes) > 0){ |
236
|
|
|
|
237
|
|
|
foreach($nodes as $node){ |
238
|
|
|
$node->removeNode(); |
|
|
|
|
239
|
|
|
} |
240
|
|
|
|
241
|
8 |
|
} |
242
|
8 |
|
|
243
|
2 |
|
return $this; |
244
|
8 |
|
} |
245
|
|
|
|
246
|
|
|
/** |
247
|
|
|
* |
248
|
|
|
*/ |
249
|
|
|
public function insertNodeList(NodeList $nodeList):Document{ |
250
|
|
|
|
251
|
|
|
foreach($nodeList as $node){ |
252
|
|
|
$this->appendChild($this->importNode($node->cloneNode(true), true)); |
253
|
|
|
} |
254
|
|
|
|
255
|
|
|
return $this; |
256
|
|
|
} |
257
|
41 |
|
|
258
|
|
|
/** |
259
|
41 |
|
* |
260
|
41 |
|
*/ |
261
|
|
|
public function getElementsByClassName(string $className):NodeList{ |
262
|
|
|
return $this->select(['.'.$className]); |
263
|
41 |
|
} |
264
|
2 |
|
|
265
|
|
|
/************* |
266
|
|
|
* prototype * |
267
|
41 |
|
*************/ |
268
|
|
|
|
269
|
41 |
|
/** |
270
|
|
|
* @link http://api.prototypejs.org/dom/Element/inspect/ |
271
|
41 |
|
*/ |
272
|
|
|
public function inspect(DOMNode $context = null, bool $xml = null):string{ |
273
|
41 |
|
|
274
|
41 |
|
if($xml === true){ |
275
|
|
|
return $this->saveXML($context); |
276
|
|
|
} |
277
|
|
|
|
278
|
|
|
return $this->saveHTML($context); |
279
|
|
|
} |
280
|
|
|
|
281
|
41 |
|
/** |
282
|
|
|
* @link http://api.prototypejs.org/dom/Element/select/ |
283
|
|
|
* @see https://secure.php.net/manual/dom.constants.php |
284
|
|
|
*/ |
285
|
|
|
public function select(array $selectors = null, DOMNode $contextNode = null, string $axis = null, int $nodeType = null):NodeList{ |
286
|
|
|
$nodeType = $nodeType ?? XML_ELEMENT_NODE; |
287
|
|
|
$elements = new NodeList; |
288
|
|
|
|
289
|
|
|
foreach($selectors ?? ['*'] as $selector){ |
290
|
|
|
|
291
|
|
|
if(!is_string($selector)){ |
292
|
|
|
continue; |
293
|
|
|
} |
294
|
3 |
|
|
295
|
3 |
|
foreach($this->querySelectorAll($selector, $contextNode, $axis ?? 'descendant-or-self::') as $element){ |
296
|
|
|
|
297
|
3 |
|
if($element->nodeType === $nodeType){ |
298
|
|
|
$elements[] = $element; |
299
|
3 |
|
} |
300
|
|
|
|
301
|
3 |
|
} |
302
|
3 |
|
|
303
|
|
|
} |
304
|
|
|
|
305
|
3 |
|
return $elements; |
306
|
2 |
|
} |
307
|
|
|
|
308
|
|
|
/** |
309
|
|
|
* @see http://api.prototypejs.org/dom/Element/recursivelyCollect/ |
310
|
|
|
* @see https://secure.php.net/manual/dom.constants.php |
311
|
|
|
*/ |
312
|
|
|
public function recursivelyCollect(DOMNode $element, string $property, int $maxLength = null, int $nodeType = null):NodeList{ |
313
|
3 |
|
$nodeType = $nodeType ?? XML_ELEMENT_NODE; |
314
|
|
|
$maxLength = $maxLength ?? -1; |
315
|
|
|
$nodes = new NodeList; |
316
|
|
|
|
317
|
|
|
if(in_array($property, ['parentNode', 'previousSibling', 'nextSibling'])){ |
318
|
|
|
|
319
|
|
|
while($element = $element->{$property}){ |
320
|
|
|
|
321
|
|
|
if($element->nodeType === $nodeType){ |
322
|
|
|
$nodes[] = $element; |
323
|
|
|
} |
324
|
|
|
|
325
|
4 |
|
if(count($nodes) === $maxLength){ |
326
|
|
|
break; |
327
|
4 |
|
} |
328
|
|
|
|
329
|
|
|
} |
330
|
4 |
|
|
331
|
|
|
} |
332
|
4 |
|
|
333
|
4 |
|
return $nodes; |
334
|
|
|
} |
335
|
|
|
|
336
|
4 |
|
/** |
337
|
|
|
* @see https://secure.php.net/manual/dom.constants.php |
338
|
|
|
*/ |
339
|
|
|
public function recursivelyFind(PrototypeNode $element, string $property = null, string $selector = null, int $index = null, int $nodeType = null):?DOMNode{ |
340
|
|
|
$nodeType = $nodeType ?? XML_ELEMENT_NODE; |
341
|
4 |
|
$index = $index ?? 0; |
342
|
|
|
|
343
|
|
|
if(in_array($property, ['parentNode', 'previousSibling', 'nextSibling'])){ |
344
|
|
|
|
345
|
|
|
while($element = $element->{$property}){ |
346
|
|
|
|
347
|
|
|
if($element->nodeType !== $nodeType || $selector !== null && !$element->match($selector) || --$index >= 0){ |
348
|
|
|
continue; |
349
|
|
|
} |
350
|
|
|
|
351
|
|
|
return $element; |
352
|
4 |
|
} |
353
|
|
|
|
354
|
4 |
|
} |
355
|
|
|
|
356
|
4 |
|
return null; |
357
|
4 |
|
} |
358
|
|
|
|
359
|
|
|
/** |
360
|
|
|
* @link http://api.prototypejs.org/dom/Element/match/ |
361
|
|
|
*/ |
362
|
4 |
|
public function match(DOMNode $element, string $selector):bool{ |
363
|
|
|
|
364
|
|
|
foreach($this->select([$selector]) as $match){ |
365
|
|
|
|
366
|
|
|
if($element->isSameNode($match)){ |
|
|
|
|
367
|
|
|
return true; |
368
|
|
|
} |
369
|
|
|
|
370
|
|
|
} |
371
|
|
|
|
372
|
|
|
return false; |
373
|
6 |
|
} |
374
|
|
|
|
375
|
6 |
|
/** |
376
|
|
|
* @link http://api.prototypejs.org/dom/Element/new/ |
377
|
6 |
|
*/ |
378
|
4 |
|
public function newElement(string $tag, array $attributes = null):PrototypeHTMLElement{ |
379
|
|
|
/** @var \chillerlan\PrototypeDOM\Node\PrototypeHTMLElement $element */ |
380
|
|
|
$element = $this->createElement($tag); |
381
|
6 |
|
|
382
|
|
|
if($attributes !== null){ |
383
|
|
|
$element->setAttributes($attributes); |
|
|
|
|
384
|
|
|
} |
385
|
|
|
|
386
|
|
|
return $element; |
387
|
|
|
} |
388
|
|
|
|
389
|
|
|
/** |
390
|
|
|
* @inheritDoc |
391
|
|
|
* |
392
|
|
|
* @return \chillerlan\PrototypeDOM\Node\PrototypeHTMLElement|\DOMNode|null |
393
|
|
|
* @throws \DOMException |
394
|
|
|
*/ |
395
|
|
|
public function getElementById($elementId):?DOMNode{ |
396
|
|
|
|
397
|
|
|
if(!is_string($elementId)){ |
398
|
|
|
throw new DOMException('invalid element id'); |
399
|
|
|
} |
400
|
|
|
|
401
|
|
|
return $this->select(['#'.$elementId])[0] ?? null; |
402
|
|
|
} |
403
|
|
|
|
404
|
|
|
} |
405
|
|
|
|
This check looks for calls to methods that do not seem to exist on a given type. It looks for the method on the type itself as well as in inherited classes or implemented interfaces.
This is most likely a typographical error or the method has been renamed.