1
|
|
|
<?php declare(strict_types=1); |
2
|
|
|
namespace PHPHtmlParser; |
3
|
|
|
|
4
|
|
|
use PHPHtmlParser\Dom\AbstractNode; |
5
|
|
|
use PHPHtmlParser\Dom\Collection; |
6
|
|
|
use PHPHtmlParser\Dom\HtmlNode; |
7
|
|
|
use PHPHtmlParser\Dom\TextNode; |
8
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
9
|
|
|
use PHPHtmlParser\Exceptions\CircularException; |
10
|
|
|
use PHPHtmlParser\Exceptions\CurlException; |
11
|
|
|
use PHPHtmlParser\Exceptions\NotLoadedException; |
12
|
|
|
use PHPHtmlParser\Exceptions\ParentNotFoundException; |
13
|
|
|
use PHPHtmlParser\Exceptions\StrictException; |
14
|
|
|
use PHPHtmlParser\Exceptions\UnknownChildTypeException; |
15
|
|
|
use PHPHtmlParser\Exceptions\LogicalException; |
16
|
|
|
use stringEncode\Encode; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* Class Dom |
20
|
|
|
* |
21
|
|
|
* @package PHPHtmlParser |
22
|
|
|
*/ |
23
|
|
|
class Dom |
24
|
|
|
{ |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* The charset we would like the output to be in. |
28
|
|
|
* |
29
|
|
|
* @var string |
30
|
|
|
*/ |
31
|
|
|
protected $defaultCharset = 'UTF-8'; |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* Contains the root node of this dom tree. |
35
|
|
|
* |
36
|
|
|
* @var HtmlNode |
37
|
|
|
*/ |
38
|
|
|
public $root; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* The raw version of the document string. |
42
|
|
|
* |
43
|
|
|
* @var string |
44
|
|
|
*/ |
45
|
|
|
protected $raw; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* The document string. |
49
|
|
|
* |
50
|
|
|
* @var Content |
51
|
|
|
*/ |
52
|
|
|
protected $content = null; |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* The original file size of the document. |
56
|
|
|
* |
57
|
|
|
* @var int |
58
|
|
|
*/ |
59
|
|
|
protected $rawSize; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* The size of the document after it is cleaned. |
63
|
|
|
* |
64
|
|
|
* @var int |
65
|
|
|
*/ |
66
|
|
|
protected $size; |
67
|
|
|
|
68
|
|
|
/** |
69
|
|
|
* A global options array to be used by all load calls. |
70
|
|
|
* |
71
|
|
|
* @var array |
72
|
|
|
*/ |
73
|
|
|
protected $globalOptions = []; |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* A persistent option object to be used for all options in the |
77
|
|
|
* parsing of the file. |
78
|
|
|
* |
79
|
|
|
* @var Options |
80
|
|
|
*/ |
81
|
|
|
protected $options; |
82
|
|
|
|
83
|
|
|
/** |
84
|
|
|
* A list of tags which will always be self closing |
85
|
|
|
* |
86
|
|
|
* @var array |
87
|
|
|
*/ |
88
|
|
|
protected $selfClosing = [ |
89
|
|
|
'area', |
90
|
|
|
'base', |
91
|
|
|
'basefont', |
92
|
|
|
'br', |
93
|
|
|
'col', |
94
|
|
|
'embed', |
95
|
|
|
'hr', |
96
|
|
|
'img', |
97
|
|
|
'input', |
98
|
|
|
'keygen', |
99
|
|
|
'link', |
100
|
|
|
'meta', |
101
|
|
|
'param', |
102
|
|
|
'source', |
103
|
|
|
'spacer', |
104
|
|
|
'track', |
105
|
|
|
'wbr' |
106
|
|
|
]; |
107
|
|
|
|
108
|
|
|
/** |
109
|
|
|
* A list of tags where there should be no /> at the end (html5 style) |
110
|
|
|
* |
111
|
|
|
* @var array |
112
|
|
|
*/ |
113
|
|
|
protected $noSlash = []; |
114
|
|
|
|
115
|
|
|
/** |
116
|
|
|
* Returns the inner html of the root node. |
117
|
|
|
* |
118
|
|
|
* @return string |
119
|
|
|
* @throws ChildNotFoundException |
120
|
|
|
* @throws UnknownChildTypeException |
121
|
|
|
*/ |
122
|
24 |
|
public function __toString(): string |
123
|
|
|
{ |
124
|
24 |
|
return $this->root->innerHtml(); |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
/** |
128
|
|
|
* A simple wrapper around the root node. |
129
|
|
|
* |
130
|
|
|
* @param string $name |
131
|
|
|
* @return mixed |
132
|
|
|
*/ |
133
|
18 |
|
public function __get($name) |
134
|
|
|
{ |
135
|
18 |
|
return $this->root->$name; |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
/** |
139
|
|
|
* Attempts to load the dom from any resource, string, file, or URL. |
140
|
|
|
* @param string $str |
141
|
|
|
* @param array $options |
142
|
|
|
* @return Dom |
143
|
|
|
* @throws ChildNotFoundException |
144
|
|
|
* @throws CircularException |
145
|
|
|
* @throws CurlException |
146
|
|
|
* @throws StrictException |
147
|
|
|
*/ |
148
|
183 |
|
public function load(string $str, array $options = []): Dom |
149
|
|
|
{ |
150
|
183 |
|
AbstractNode::resetCount(); |
151
|
|
|
// check if it's a file |
152
|
183 |
|
if (strpos($str, "\n") === false && is_file($str)) { |
153
|
6 |
|
return $this->loadFromFile($str, $options); |
154
|
|
|
} |
155
|
|
|
// check if it's a url |
156
|
177 |
|
if (preg_match("/^https?:\/\//i", $str)) { |
157
|
|
|
return $this->loadFromUrl($str, $options); |
158
|
|
|
} |
159
|
|
|
|
160
|
177 |
|
return $this->loadStr($str, $options); |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
/** |
164
|
|
|
* Loads the dom from a document file/url |
165
|
|
|
* @param string $file |
166
|
|
|
* @param array $options |
167
|
|
|
* @return Dom |
168
|
|
|
* @throws ChildNotFoundException |
169
|
|
|
* @throws CircularException |
170
|
|
|
* @throws StrictException |
171
|
|
|
* @throws LogicalException |
172
|
|
|
*/ |
173
|
51 |
|
public function loadFromFile(string $file, array $options = []): Dom |
174
|
|
|
{ |
175
|
51 |
|
$content = file_get_contents($file); |
176
|
51 |
|
if ($content === false) { |
177
|
|
|
throw new LogicalException('file_get_contents failed and returned false when trying to read "'.$file.'".'); |
178
|
|
|
} |
179
|
51 |
|
return $this->loadStr($content, $options); |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
/** |
183
|
|
|
* Use a curl interface implementation to attempt to load |
184
|
|
|
* the content from a url. |
185
|
|
|
* @param string $url |
186
|
|
|
* @param array $options |
187
|
|
|
* @param CurlInterface|null $curl |
188
|
|
|
* @return Dom |
189
|
|
|
* @throws ChildNotFoundException |
190
|
|
|
* @throws CircularException |
191
|
|
|
* @throws CurlException |
192
|
|
|
* @throws StrictException |
193
|
|
|
*/ |
194
|
6 |
|
public function loadFromUrl(string $url, array $options = [], CurlInterface $curl = null): Dom |
195
|
|
|
{ |
196
|
6 |
|
if (is_null($curl)) { |
197
|
|
|
// use the default curl interface |
198
|
|
|
$curl = new Curl; |
199
|
|
|
} |
200
|
6 |
|
$content = $curl->get($url, $options); |
201
|
|
|
|
202
|
6 |
|
return $this->loadStr($content, $options); |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
/** |
206
|
|
|
* Parsers the html of the given string. Used for load(), loadFromFile(), |
207
|
|
|
* and loadFromUrl(). |
208
|
|
|
* @param string $str |
209
|
|
|
* @param array $option |
210
|
|
|
* @return Dom |
211
|
|
|
* @throws ChildNotFoundException |
212
|
|
|
* @throws CircularException |
213
|
|
|
* @throws StrictException |
214
|
|
|
*/ |
215
|
246 |
|
public function loadStr(string $str, array $option = []): Dom |
216
|
|
|
{ |
217
|
246 |
|
$this->options = new Options; |
218
|
246 |
|
$this->options->setOptions($this->globalOptions) |
219
|
246 |
|
->setOptions($option); |
220
|
|
|
|
221
|
246 |
|
$this->rawSize = strlen($str); |
222
|
246 |
|
$this->raw = $str; |
223
|
|
|
|
224
|
246 |
|
$html = $this->clean($str); |
225
|
|
|
|
226
|
246 |
|
$this->size = strlen($str); |
227
|
246 |
|
$this->content = new Content($html); |
228
|
|
|
|
229
|
246 |
|
$this->parse(); |
230
|
240 |
|
$this->detectCharset(); |
231
|
|
|
|
232
|
240 |
|
return $this; |
233
|
|
|
} |
234
|
|
|
|
235
|
|
|
/** |
236
|
|
|
* Sets a global options array to be used by all load calls. |
237
|
|
|
* |
238
|
|
|
* @param array $options |
239
|
|
|
* @return Dom |
240
|
|
|
* @chainable |
241
|
|
|
*/ |
242
|
54 |
|
public function setOptions(array $options): Dom |
243
|
|
|
{ |
244
|
54 |
|
$this->globalOptions = $options; |
245
|
|
|
|
246
|
54 |
|
return $this; |
247
|
|
|
} |
248
|
|
|
|
249
|
|
|
/** |
250
|
|
|
* Find elements by css selector on the root node. |
251
|
|
|
* @param string $selector |
252
|
|
|
* @param int|null $nth |
253
|
|
|
* @return mixed|Collection|null |
254
|
|
|
* @throws ChildNotFoundException |
255
|
|
|
* @throws NotLoadedException |
256
|
|
|
*/ |
257
|
180 |
|
public function find(string $selector, int $nth = null) |
258
|
|
|
{ |
259
|
180 |
|
$this->isLoaded(); |
260
|
|
|
|
261
|
177 |
|
$depthFirstSearch = $this->options->get('depthFirstSearch'); |
262
|
177 |
|
if (is_bool($depthFirstSearch)) { |
263
|
177 |
|
$result = $this->root->find($selector, $nth, $depthFirstSearch); |
264
|
|
|
} else { |
265
|
|
|
$result = $this->root->find($selector, $nth); |
266
|
|
|
} |
267
|
|
|
|
268
|
177 |
|
return $result; |
269
|
|
|
} |
270
|
|
|
|
271
|
|
|
/** |
272
|
|
|
* Find element by Id on the root node |
273
|
|
|
* @param int $id |
274
|
|
|
* @return bool|AbstractNode |
275
|
|
|
* @throws ChildNotFoundException |
276
|
|
|
* @throws NotLoadedException |
277
|
|
|
* @throws ParentNotFoundException |
278
|
|
|
*/ |
279
|
9 |
|
public function findById(int $id) |
280
|
|
|
{ |
281
|
9 |
|
$this->isLoaded(); |
282
|
|
|
|
283
|
9 |
|
return $this->root->findById($id); |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
/** |
287
|
|
|
* Adds the tag (or tags in an array) to the list of tags that will always |
288
|
|
|
* be self closing. |
289
|
|
|
* |
290
|
|
|
* @param string|array $tag |
291
|
|
|
* @return Dom |
292
|
|
|
* @chainable |
293
|
|
|
*/ |
294
|
6 |
|
public function addSelfClosingTag($tag): Dom |
295
|
|
|
{ |
296
|
6 |
|
if ( ! is_array($tag)) { |
297
|
3 |
|
$tag = [$tag]; |
298
|
|
|
} |
299
|
6 |
|
foreach ($tag as $value) { |
300
|
6 |
|
$this->selfClosing[] = $value; |
301
|
|
|
} |
302
|
|
|
|
303
|
6 |
|
return $this; |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
/** |
307
|
|
|
* Removes the tag (or tags in an array) from the list of tags that will |
308
|
|
|
* always be self closing. |
309
|
|
|
* |
310
|
|
|
* @param string|array $tag |
311
|
|
|
* @return Dom |
312
|
|
|
* @chainable |
313
|
|
|
*/ |
314
|
3 |
|
public function removeSelfClosingTag($tag): Dom |
315
|
|
|
{ |
316
|
3 |
|
if ( ! is_array($tag)) { |
317
|
3 |
|
$tag = [$tag]; |
318
|
|
|
} |
319
|
3 |
|
$this->selfClosing = array_diff($this->selfClosing, $tag); |
320
|
|
|
|
321
|
3 |
|
return $this; |
322
|
|
|
} |
323
|
|
|
|
324
|
|
|
/** |
325
|
|
|
* Sets the list of self closing tags to empty. |
326
|
|
|
* |
327
|
|
|
* @return Dom |
328
|
|
|
* @chainable |
329
|
|
|
*/ |
330
|
3 |
|
public function clearSelfClosingTags(): Dom |
331
|
|
|
{ |
332
|
3 |
|
$this->selfClosing = []; |
333
|
|
|
|
334
|
3 |
|
return $this; |
335
|
|
|
} |
336
|
|
|
|
337
|
|
|
|
338
|
|
|
/** |
339
|
|
|
* Adds a tag to the list of self closing tags that should not have a trailing slash |
340
|
|
|
* |
341
|
|
|
* @param $tag |
342
|
|
|
* @return Dom |
343
|
|
|
* @chainable |
344
|
|
|
*/ |
345
|
3 |
|
public function addNoSlashTag($tag): Dom |
346
|
|
|
{ |
347
|
3 |
|
if ( ! is_array($tag)) { |
348
|
3 |
|
$tag = [$tag]; |
349
|
|
|
} |
350
|
3 |
|
foreach ($tag as $value) { |
351
|
3 |
|
$this->noSlash[] = $value; |
352
|
|
|
} |
353
|
|
|
|
354
|
3 |
|
return $this; |
355
|
|
|
} |
356
|
|
|
|
357
|
|
|
/** |
358
|
|
|
* Removes a tag from the list of no-slash tags. |
359
|
|
|
* |
360
|
|
|
* @param $tag |
361
|
|
|
* @return Dom |
362
|
|
|
* @chainable |
363
|
|
|
*/ |
364
|
|
|
public function removeNoSlashTag($tag): Dom |
365
|
|
|
{ |
366
|
|
|
if ( ! is_array($tag)) { |
367
|
|
|
$tag = [$tag]; |
368
|
|
|
} |
369
|
|
|
$this->noSlash = array_diff($this->noSlash, $tag); |
370
|
|
|
|
371
|
|
|
return $this; |
372
|
|
|
} |
373
|
|
|
|
374
|
|
|
/** |
375
|
|
|
* Empties the list of no-slash tags. |
376
|
|
|
* |
377
|
|
|
* @return Dom |
378
|
|
|
* @chainable |
379
|
|
|
*/ |
380
|
|
|
public function clearNoSlashTags(): Dom |
381
|
|
|
{ |
382
|
|
|
$this->noSlash = []; |
383
|
|
|
|
384
|
|
|
return $this; |
385
|
|
|
} |
386
|
|
|
|
387
|
|
|
/** |
388
|
|
|
* Simple wrapper function that returns the first child. |
389
|
|
|
* @return AbstractNode |
390
|
|
|
* @throws ChildNotFoundException |
391
|
|
|
* @throws NotLoadedException |
392
|
|
|
*/ |
393
|
3 |
|
public function firstChild(): AbstractNode |
394
|
|
|
{ |
395
|
3 |
|
$this->isLoaded(); |
396
|
|
|
|
397
|
3 |
|
return $this->root->firstChild(); |
398
|
|
|
} |
399
|
|
|
|
400
|
|
|
/** |
401
|
|
|
* Simple wrapper function that returns the last child. |
402
|
|
|
* @return AbstractNode |
403
|
|
|
* @throws ChildNotFoundException |
404
|
|
|
* @throws NotLoadedException |
405
|
|
|
*/ |
406
|
3 |
|
public function lastChild(): AbstractNode |
407
|
|
|
{ |
408
|
3 |
|
$this->isLoaded(); |
409
|
|
|
|
410
|
3 |
|
return $this->root->lastChild(); |
411
|
|
|
} |
412
|
|
|
|
413
|
|
|
/** |
414
|
|
|
* Simple wrapper function that returns count of child elements |
415
|
|
|
* |
416
|
|
|
* @return int |
417
|
|
|
* @throws NotLoadedException |
418
|
|
|
*/ |
419
|
3 |
|
public function countChildren(): int |
420
|
|
|
{ |
421
|
3 |
|
$this->isLoaded(); |
422
|
|
|
|
423
|
3 |
|
return $this->root->countChildren(); |
424
|
|
|
} |
425
|
|
|
|
426
|
|
|
/** |
427
|
|
|
* Get array of children |
428
|
|
|
* |
429
|
|
|
* @return array |
430
|
|
|
* @throws NotLoadedException |
431
|
|
|
*/ |
432
|
3 |
|
public function getChildren(): array |
433
|
|
|
{ |
434
|
3 |
|
$this->isLoaded(); |
435
|
|
|
|
436
|
3 |
|
return $this->root->getChildren(); |
437
|
|
|
} |
438
|
|
|
|
439
|
|
|
/** |
440
|
|
|
* Check if node have children nodes |
441
|
|
|
* |
442
|
|
|
* @return bool |
443
|
|
|
* @throws NotLoadedException |
444
|
|
|
*/ |
445
|
3 |
|
public function hasChildren(): bool |
446
|
|
|
{ |
447
|
3 |
|
$this->isLoaded(); |
448
|
|
|
|
449
|
3 |
|
return $this->root->hasChildren(); |
450
|
|
|
} |
451
|
|
|
|
452
|
|
|
/** |
453
|
|
|
* Simple wrapper function that returns an element by the |
454
|
|
|
* id. |
455
|
|
|
* @param $id |
456
|
|
|
* @return mixed|Collection|null |
457
|
|
|
* @throws ChildNotFoundException |
458
|
|
|
* @throws NotLoadedException |
459
|
|
|
*/ |
460
|
12 |
|
public function getElementById($id) |
461
|
|
|
{ |
462
|
12 |
|
$this->isLoaded(); |
463
|
|
|
|
464
|
12 |
|
return $this->find('#'.$id, 0); |
465
|
|
|
} |
466
|
|
|
|
467
|
|
|
/** |
468
|
|
|
* Simple wrapper function that returns all elements by |
469
|
|
|
* tag name. |
470
|
|
|
* @param string $name |
471
|
|
|
* @return mixed|Collection|null |
472
|
|
|
* @throws ChildNotFoundException |
473
|
|
|
* @throws NotLoadedException |
474
|
|
|
*/ |
475
|
15 |
|
public function getElementsByTag(string $name) |
476
|
|
|
{ |
477
|
15 |
|
$this->isLoaded(); |
478
|
|
|
|
479
|
15 |
|
return $this->find($name); |
480
|
|
|
} |
481
|
|
|
|
482
|
|
|
/** |
483
|
|
|
* Simple wrapper function that returns all elements by |
484
|
|
|
* class name. |
485
|
|
|
* @param string $class |
486
|
|
|
* @return mixed|Collection|null |
487
|
|
|
* @throws ChildNotFoundException |
488
|
|
|
* @throws NotLoadedException |
489
|
|
|
*/ |
490
|
3 |
|
public function getElementsByClass(string $class) |
491
|
|
|
{ |
492
|
3 |
|
$this->isLoaded(); |
493
|
|
|
|
494
|
3 |
|
return $this->find('.'.$class); |
495
|
|
|
} |
496
|
|
|
|
497
|
|
|
/** |
498
|
|
|
* Checks if the load methods have been called. |
499
|
|
|
* |
500
|
|
|
* @throws NotLoadedException |
501
|
|
|
*/ |
502
|
204 |
|
protected function isLoaded(): void |
503
|
|
|
{ |
504
|
204 |
|
if (is_null($this->content)) { |
505
|
3 |
|
throw new NotLoadedException('Content is not loaded!'); |
506
|
|
|
} |
507
|
201 |
|
} |
508
|
|
|
|
509
|
|
|
/** |
510
|
|
|
* Cleans the html of any none-html information. |
511
|
|
|
* |
512
|
|
|
* @param string $str |
513
|
|
|
* @return string |
514
|
|
|
*/ |
515
|
246 |
|
protected function clean(string $str): string |
516
|
|
|
{ |
517
|
246 |
|
if ($this->options->get('cleanupInput') != true) { |
518
|
|
|
// skip entire cleanup step |
519
|
6 |
|
return $str; |
520
|
|
|
} |
521
|
|
|
|
522
|
240 |
|
$is_gzip = 0 === mb_strpos($str, "\x1f" . "\x8b" . "\x08", 0, "US-ASCII"); |
523
|
240 |
|
if ($is_gzip) { |
524
|
|
|
$str = gzdecode($str); |
525
|
|
|
if ($str === false) { |
526
|
|
|
throw new LogicalException('gzdecode returned false. Error when trying to decode the string.'); |
527
|
|
|
} |
528
|
|
|
} |
529
|
|
|
|
530
|
|
|
// remove white space before closing tags |
531
|
240 |
|
$str = mb_eregi_replace("'\s+>", "'>", $str); |
532
|
240 |
|
if ($str === false) { |
533
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to clean single quotes.'); |
534
|
|
|
} |
535
|
240 |
|
$str = mb_eregi_replace('"\s+>', '">', $str); |
536
|
240 |
|
if ($str === false) { |
537
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to clean double quotes.'); |
538
|
|
|
} |
539
|
|
|
|
540
|
|
|
// clean out the \n\r |
541
|
240 |
|
$replace = ' '; |
542
|
240 |
|
if ($this->options->get('preserveLineBreaks')) { |
543
|
3 |
|
$replace = ' '; |
544
|
|
|
} |
545
|
240 |
|
$str = str_replace(["\r\n", "\r", "\n"], $replace, $str); |
546
|
240 |
|
if ($str === false) { |
547
|
|
|
throw new LogicalException('str_replace returned false instead of a string. Error when attempting to clean input string.'); |
548
|
|
|
} |
549
|
|
|
|
550
|
|
|
// strip the doctype |
551
|
240 |
|
$str = mb_eregi_replace("<!doctype(.*?)>", '', $str); |
552
|
240 |
|
if ($str === false) { |
553
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to strip the doctype.'); |
554
|
|
|
} |
555
|
|
|
|
556
|
|
|
// strip out comments |
557
|
240 |
|
$str = mb_eregi_replace("<!--(.*?)-->", '', $str); |
558
|
240 |
|
if ($str === false) { |
559
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to strip comments.'); |
560
|
|
|
} |
561
|
|
|
|
562
|
|
|
// strip out cdata |
563
|
240 |
|
$str = mb_eregi_replace("<!\[CDATA\[(.*?)\]\]>", '', $str); |
564
|
240 |
|
if ($str === false) { |
565
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to strip out cdata.'); |
566
|
|
|
} |
567
|
|
|
|
568
|
|
|
// strip out <script> tags |
569
|
240 |
|
if ($this->options->get('removeScripts')) { |
570
|
237 |
|
$str = mb_eregi_replace("<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>", '', $str); |
571
|
237 |
|
if ($str === false) { |
572
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to remove scripts 1.'); |
573
|
|
|
} |
574
|
237 |
|
$str = mb_eregi_replace("<\s*script\s*>(.*?)<\s*/\s*script\s*>", '', $str); |
575
|
237 |
|
if ($str === false) { |
576
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to remove scripts 2.'); |
577
|
|
|
} |
578
|
|
|
} |
579
|
|
|
|
580
|
|
|
// strip out <style> tags |
581
|
240 |
|
if ($this->options->get('removeStyles')) { |
582
|
237 |
|
$str = mb_eregi_replace("<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>", '', $str); |
583
|
237 |
|
if ($str === false) { |
584
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to strip out style tags 1.'); |
585
|
|
|
} |
586
|
237 |
|
$str = mb_eregi_replace("<\s*style\s*>(.*?)<\s*/\s*style\s*>", '', $str); |
587
|
237 |
|
if ($str === false) { |
588
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to strip out style tags 2.'); |
589
|
|
|
} |
590
|
|
|
} |
591
|
|
|
|
592
|
|
|
// strip out server side scripts |
593
|
240 |
|
if ($this->options->get('serverSideScripts')) { |
594
|
|
|
$str = mb_eregi_replace("(<\?)(.*?)(\?>)", '', $str); |
595
|
|
|
if ($str === false) { |
596
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to strip out service side scripts.'); |
597
|
|
|
} |
598
|
|
|
} |
599
|
|
|
|
600
|
|
|
// strip smarty scripts |
601
|
240 |
|
if ($this->options->get('removeSmartyScripts')) { |
602
|
237 |
|
$str = mb_eregi_replace("(\{\w)(.*?)(\})", '', $str); |
603
|
237 |
|
if ($str === false) { |
604
|
|
|
throw new LogicalException('mb_eregi_replace returned false instead of a string. Error when attempting to remove smarty scripts.'); |
605
|
|
|
} |
606
|
|
|
} |
607
|
|
|
|
608
|
240 |
|
return $str; |
609
|
|
|
} |
610
|
|
|
|
611
|
|
|
/** |
612
|
|
|
* Attempts to parse the html in content. |
613
|
|
|
* |
614
|
|
|
* @return void |
615
|
|
|
* @throws ChildNotFoundException |
616
|
|
|
* @throws CircularException |
617
|
|
|
* @throws StrictException |
618
|
|
|
*/ |
619
|
246 |
|
protected function parse(): void |
620
|
|
|
{ |
621
|
|
|
// add the root node |
622
|
246 |
|
$this->root = new HtmlNode('root'); |
623
|
246 |
|
$this->root->setHtmlSpecialCharsDecode($this->options->htmlSpecialCharsDecode); |
624
|
246 |
|
$activeNode = $this->root; |
625
|
246 |
|
while ( ! is_null($activeNode)) { |
626
|
246 |
|
$str = $this->content->copyUntil('<'); |
627
|
246 |
|
if ($str == '') { |
628
|
246 |
|
$info = $this->parseTag(); |
629
|
246 |
|
if ( ! $info['status']) { |
630
|
|
|
// we are done here |
631
|
240 |
|
$activeNode = null; |
632
|
240 |
|
continue; |
633
|
|
|
} |
634
|
|
|
|
635
|
|
|
// check if it was a closing tag |
636
|
240 |
|
if ($info['closing']) { |
637
|
231 |
|
$foundOpeningTag = true; |
638
|
231 |
|
$originalNode = $activeNode; |
639
|
231 |
|
while ($activeNode->getTag()->name() != $info['tag']) { |
640
|
78 |
|
$activeNode = $activeNode->getParent(); |
641
|
78 |
|
if (is_null($activeNode)) { |
642
|
|
|
// we could not find opening tag |
643
|
36 |
|
$activeNode = $originalNode; |
644
|
36 |
|
$foundOpeningTag = false; |
645
|
36 |
|
break; |
646
|
|
|
} |
647
|
|
|
} |
648
|
231 |
|
if ($foundOpeningTag) { |
649
|
231 |
|
$activeNode = $activeNode->getParent(); |
650
|
|
|
} |
651
|
231 |
|
continue; |
652
|
|
|
} |
653
|
|
|
|
654
|
240 |
|
if ( ! isset($info['node'])) { |
655
|
12 |
|
continue; |
656
|
|
|
} |
657
|
|
|
|
658
|
|
|
/** @var AbstractNode $node */ |
659
|
240 |
|
$node = $info['node']; |
660
|
240 |
|
$activeNode->addChild($node); |
661
|
|
|
|
662
|
|
|
// check if node is self closing |
663
|
240 |
|
if ( ! $node->getTag()->isSelfClosing()) { |
664
|
240 |
|
$activeNode = $node; |
665
|
|
|
} |
666
|
225 |
|
} else if ($this->options->whitespaceTextNode || |
667
|
225 |
|
trim($str) != '' |
668
|
|
|
) { |
669
|
|
|
// we found text we care about |
670
|
222 |
|
$textNode = new TextNode($str, $this->options->removeDoubleSpace); |
671
|
222 |
|
$textNode->setHtmlSpecialCharsDecode($this->options->htmlSpecialCharsDecode); |
672
|
222 |
|
$activeNode->addChild($textNode); |
673
|
|
|
} |
674
|
|
|
} |
675
|
240 |
|
} |
676
|
|
|
|
677
|
|
|
/** |
678
|
|
|
* Attempt to parse a tag out of the content. |
679
|
|
|
* |
680
|
|
|
* @return array |
681
|
|
|
* @throws StrictException |
682
|
|
|
*/ |
683
|
246 |
|
protected function parseTag(): array |
684
|
|
|
{ |
685
|
|
|
$return = [ |
686
|
246 |
|
'status' => false, |
687
|
|
|
'closing' => false, |
688
|
|
|
'node' => null, |
689
|
|
|
]; |
690
|
246 |
|
if ($this->content->char() != '<') { |
691
|
|
|
// we are not at the beginning of a tag |
692
|
237 |
|
return $return; |
693
|
|
|
} |
694
|
|
|
|
695
|
|
|
// check if this is a closing tag |
696
|
240 |
|
if ($this->content->fastForward(1)->char() == '/') { |
697
|
|
|
// end tag |
698
|
231 |
|
$tag = $this->content->fastForward(1) |
699
|
231 |
|
->copyByToken('slash', true); |
700
|
|
|
// move to end of tag |
701
|
231 |
|
$this->content->copyUntil('>'); |
702
|
231 |
|
$this->content->fastForward(1); |
703
|
|
|
|
704
|
|
|
// check if this closing tag counts |
705
|
231 |
|
$tag = strtolower($tag); |
706
|
231 |
|
if (in_array($tag, $this->selfClosing, true)) { |
707
|
12 |
|
$return['status'] = true; |
708
|
|
|
|
709
|
12 |
|
return $return; |
710
|
|
|
} else { |
711
|
231 |
|
$return['status'] = true; |
712
|
231 |
|
$return['closing'] = true; |
713
|
231 |
|
$return['tag'] = strtolower($tag); |
714
|
|
|
} |
715
|
|
|
|
716
|
231 |
|
return $return; |
717
|
|
|
} |
718
|
|
|
|
719
|
240 |
|
$tag = strtolower($this->content->copyByToken('slash', true)); |
720
|
240 |
|
if (trim($tag) == '') |
721
|
|
|
{ |
722
|
|
|
// no tag found, invalid < found |
723
|
3 |
|
return $return; |
724
|
|
|
} |
725
|
240 |
|
$node = new HtmlNode($tag); |
726
|
240 |
|
$node->setHtmlSpecialCharsDecode($this->options->htmlSpecialCharsDecode); |
727
|
|
|
|
728
|
|
|
// attributes |
729
|
240 |
|
while ($this->content->char() != '>' && |
730
|
240 |
|
$this->content->char() != '/') { |
731
|
231 |
|
$space = $this->content->skipByToken('blank', true); |
732
|
231 |
|
if (empty($space)) { |
733
|
6 |
|
$this->content->fastForward(1); |
734
|
6 |
|
continue; |
735
|
|
|
} |
736
|
|
|
|
737
|
231 |
|
$name = $this->content->copyByToken('equal', true); |
738
|
231 |
|
if ($name == '/') { |
739
|
|
|
break; |
740
|
|
|
} |
741
|
|
|
|
742
|
231 |
|
if (empty($name)) { |
743
|
123 |
|
$this->content->skipByToken('blank'); |
744
|
123 |
|
continue; |
745
|
|
|
} |
746
|
|
|
|
747
|
228 |
|
$this->content->skipByToken('blank'); |
748
|
228 |
|
if ($this->content->char() == '=') { |
749
|
228 |
|
$attr = []; |
750
|
228 |
|
$this->content->fastForward(1) |
751
|
228 |
|
->skipByToken('blank'); |
752
|
228 |
|
switch ($this->content->char()) { |
753
|
228 |
|
case '"': |
754
|
213 |
|
$attr['doubleQuote'] = true; |
755
|
213 |
|
$this->content->fastForward(1); |
756
|
213 |
|
$string = $this->content->copyUntil('"', true); |
757
|
|
|
do { |
758
|
213 |
|
$moreString = $this->content->copyUntilUnless('"', '=>'); |
759
|
213 |
|
$string .= $moreString; |
760
|
213 |
|
} while ( ! empty($moreString)); |
761
|
213 |
|
$attr['value'] = $string; |
762
|
213 |
|
$this->content->fastForward(1); |
763
|
213 |
|
$node->getTag()->$name = $attr; |
764
|
213 |
|
break; |
765
|
21 |
|
case "'": |
766
|
18 |
|
$attr['doubleQuote'] = false; |
767
|
18 |
|
$this->content->fastForward(1); |
768
|
18 |
|
$string = $this->content->copyUntil("'", true); |
769
|
|
|
do { |
770
|
18 |
|
$moreString = $this->content->copyUntilUnless("'", '=>'); |
771
|
18 |
|
$string .= $moreString; |
772
|
18 |
|
} while ( ! empty($moreString)); |
773
|
18 |
|
$attr['value'] = $string; |
774
|
18 |
|
$this->content->fastForward(1); |
775
|
18 |
|
$node->getTag()->$name = $attr; |
776
|
18 |
|
break; |
777
|
|
|
default: |
778
|
3 |
|
$attr['doubleQuote'] = true; |
779
|
3 |
|
$attr['value'] = $this->content->copyByToken('attr', true); |
780
|
3 |
|
$node->getTag()->$name = $attr; |
781
|
228 |
|
break; |
782
|
|
|
} |
783
|
|
|
} else { |
784
|
|
|
// no value attribute |
785
|
72 |
|
if ($this->options->strict) { |
786
|
|
|
// can't have this in strict html |
787
|
3 |
|
$character = $this->content->getPosition(); |
788
|
3 |
|
throw new StrictException("Tag '$tag' has an attribute '$name' with out a value! (character #$character)"); |
789
|
|
|
} |
790
|
69 |
|
$node->getTag()->$name = [ |
791
|
|
|
'value' => null, |
792
|
|
|
'doubleQuote' => true, |
793
|
|
|
]; |
794
|
69 |
|
if ($this->content->char() != '>') { |
795
|
12 |
|
$this->content->rewind(1); |
796
|
|
|
} |
797
|
|
|
} |
798
|
|
|
} |
799
|
|
|
|
800
|
240 |
|
$this->content->skipByToken('blank'); |
801
|
240 |
|
$tag = strtolower($tag); |
802
|
240 |
|
if ($this->content->char() == '/') { |
803
|
|
|
// self closing tag |
804
|
120 |
|
$node->getTag()->selfClosing(); |
805
|
120 |
|
$this->content->fastForward(1); |
806
|
237 |
|
} elseif (in_array($tag, $this->selfClosing, true)) { |
807
|
|
|
|
808
|
|
|
// Should be a self closing tag, check if we are strict |
809
|
84 |
|
if ($this->options->strict) { |
810
|
3 |
|
$character = $this->content->getPosition(); |
811
|
3 |
|
throw new StrictException("Tag '$tag' is not self closing! (character #$character)"); |
812
|
|
|
} |
813
|
|
|
|
814
|
|
|
// We force self closing on this tag. |
815
|
81 |
|
$node->getTag()->selfClosing(); |
816
|
|
|
|
817
|
|
|
// Should this tag use a trailing slash? |
818
|
81 |
|
if(in_array($tag, $this->noSlash, true)) |
819
|
|
|
{ |
820
|
3 |
|
$node->getTag()->noTrailingSlash(); |
821
|
|
|
} |
822
|
|
|
|
823
|
|
|
} |
824
|
|
|
|
825
|
240 |
|
$this->content->fastForward(1); |
826
|
|
|
|
827
|
240 |
|
$return['status'] = true; |
828
|
240 |
|
$return['node'] = $node; |
829
|
|
|
|
830
|
240 |
|
return $return; |
831
|
|
|
} |
832
|
|
|
|
833
|
|
|
/** |
834
|
|
|
* Attempts to detect the charset that the html was sent in. |
835
|
|
|
* |
836
|
|
|
* @return bool |
837
|
|
|
* @throws ChildNotFoundException |
838
|
|
|
*/ |
839
|
240 |
|
protected function detectCharset(): bool |
840
|
|
|
{ |
841
|
|
|
// set the default |
842
|
240 |
|
$encode = new Encode; |
843
|
240 |
|
$encode->from($this->defaultCharset); |
844
|
240 |
|
$encode->to($this->defaultCharset); |
845
|
|
|
|
846
|
240 |
|
$enforceEncoding = $this->options->enforceEncoding; |
847
|
240 |
|
if ( ! is_null($enforceEncoding)) { |
848
|
|
|
// they want to enforce the given encoding |
849
|
|
|
$encode->from($enforceEncoding); |
850
|
|
|
$encode->to($enforceEncoding); |
851
|
|
|
|
852
|
|
|
return false; |
853
|
|
|
} |
854
|
|
|
|
855
|
|
|
/** @var AbstractNode $meta */ |
856
|
240 |
|
$meta = $this->root->find('meta[http-equiv=Content-Type]', 0); |
857
|
240 |
|
if (is_null($meta)) { |
858
|
|
|
// could not find meta tag |
859
|
210 |
|
$this->root->propagateEncoding($encode); |
860
|
|
|
|
861
|
210 |
|
return false; |
862
|
|
|
} |
863
|
30 |
|
$content = $meta->getAttribute('content'); |
|
|
|
|
864
|
30 |
|
if (is_null($content)) { |
|
|
|
|
865
|
|
|
// could not find content |
866
|
|
|
$this->root->propagateEncoding($encode); |
867
|
|
|
|
868
|
|
|
return false; |
869
|
|
|
} |
870
|
30 |
|
$matches = []; |
871
|
30 |
|
if (preg_match('/charset=(.+)/', $content, $matches)) { |
872
|
30 |
|
$encode->from(trim($matches[1])); |
873
|
30 |
|
$this->root->propagateEncoding($encode); |
874
|
|
|
|
875
|
30 |
|
return true; |
876
|
|
|
} |
877
|
|
|
|
878
|
|
|
// no charset found |
879
|
|
|
$this->root->propagateEncoding($encode); |
880
|
|
|
|
881
|
|
|
return false; |
882
|
|
|
} |
883
|
|
|
} |
884
|
|
|
|
This check looks for function or method calls that always return null and whose return value is assigned to a variable.
The method
getObject()
can return nothing but null, so it makes no sense to assign that value to a variable.The reason is most likely that a function or method is imcomplete or has been reduced for debug purposes.