1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace PHPHtmlParser\Selector; |
6
|
|
|
|
7
|
|
|
use PHPHtmlParser\Dom\AbstractNode; |
8
|
|
|
use PHPHtmlParser\Dom\Collection; |
9
|
|
|
use PHPHtmlParser\Dom\InnerNode; |
10
|
|
|
use PHPHtmlParser\Dom\LeafNode; |
11
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* Class Selector |
15
|
|
|
* |
16
|
|
|
* @package PHPHtmlParser |
17
|
|
|
*/ |
18
|
|
|
class Selector |
19
|
|
|
{ |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* @var array |
23
|
|
|
*/ |
24
|
|
|
protected $selectors = []; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* @var bool |
28
|
|
|
*/ |
29
|
|
|
private $depthFirst = false; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Constructs with the selector string |
33
|
|
|
* @param string $selector |
34
|
|
|
* @param ParserInterface $parser |
35
|
|
|
*/ |
36
|
315 |
|
public function __construct(string $selector, ParserInterface $parser) |
37
|
|
|
{ |
38
|
315 |
|
$this->selectors = $parser->parseSelectorString($selector); |
39
|
315 |
|
} |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* Returns the selectors that where found in __construct |
43
|
|
|
* @return array |
44
|
|
|
*/ |
45
|
12 |
|
public function getSelectors() |
46
|
|
|
{ |
47
|
12 |
|
return $this->selectors; |
48
|
|
|
} |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* @param bool $status |
52
|
|
|
* @return void |
53
|
|
|
*/ |
54
|
252 |
|
public function setDepthFirstFind(bool $status): void |
55
|
|
|
{ |
56
|
252 |
|
$this->depthFirst = $status; |
57
|
252 |
|
} |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* Attempts to find the selectors starting from the given |
61
|
|
|
* node object. |
62
|
|
|
* @param AbstractNode $node |
63
|
|
|
* @return Collection |
64
|
|
|
* @throws ChildNotFoundException |
65
|
|
|
*/ |
66
|
138 |
|
public function find(AbstractNode $node): Collection |
67
|
|
|
{ |
68
|
138 |
|
$collection = new Collection(); |
69
|
138 |
|
foreach($this->findGenerator($node) as $item) { |
70
|
129 |
|
$collection[] = $item; |
71
|
|
|
} |
72
|
138 |
|
return $collection; |
73
|
|
|
} |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* Attempts to find the selectors starting from the given |
77
|
|
|
* node object. |
78
|
|
|
* @param AbstractNode $node |
79
|
|
|
* @return \Generator |
80
|
|
|
* @throws ChildNotFoundException |
81
|
|
|
*/ |
82
|
303 |
|
public function findGenerator(AbstractNode $node): \Generator |
83
|
|
|
{ |
84
|
303 |
|
foreach ($this->selectors as $selector) { |
85
|
303 |
|
$nodes = [$node]; |
86
|
303 |
|
if (count($selector) == 0) { |
87
|
|
|
continue; |
88
|
|
|
} |
89
|
|
|
|
90
|
303 |
|
$options = []; |
91
|
303 |
|
for ($i=0; $i<count($selector); $i++) { |
|
|
|
|
92
|
303 |
|
$rule = $selector[$i]; |
93
|
303 |
|
$isLast = ($i == count($selector) - 1); |
94
|
|
|
|
95
|
303 |
|
if ($rule['alterNext']) { |
96
|
3 |
|
$options[] = $this->alterNext($rule); |
97
|
3 |
|
continue; |
98
|
|
|
} |
99
|
303 |
|
$next = []; |
100
|
|
|
|
101
|
303 |
|
foreach($this->seek($nodes, $rule, $options) as $foundNode) { |
102
|
246 |
|
if (!$isLast) { |
103
|
51 |
|
$next[] = $foundNode; |
104
|
|
|
} else { |
105
|
246 |
|
yield $foundNode; |
106
|
|
|
} |
107
|
|
|
} |
108
|
291 |
|
$nodes = $next; |
109
|
|
|
// clear the options |
110
|
291 |
|
$options = []; |
111
|
|
|
} |
112
|
|
|
} |
113
|
291 |
|
} |
114
|
|
|
|
115
|
|
|
/** |
116
|
|
|
* Attempts to find all children that match the rule |
117
|
|
|
* given. |
118
|
|
|
* @param array $nodes |
119
|
|
|
* @param array $rule |
120
|
|
|
* @param array $options |
121
|
|
|
* @return \Generator |
122
|
|
|
* @throws ChildNotFoundException |
123
|
|
|
*/ |
124
|
303 |
|
protected function seek(array $nodes, array $rule, array $options): \Generator |
125
|
|
|
{ |
126
|
|
|
// XPath index |
127
|
303 |
|
if (array_key_exists('tag', $rule) && array_key_exists('key', $rule) |
128
|
303 |
|
&& is_numeric($rule['key']) |
129
|
|
|
) { |
130
|
3 |
|
$count = 0; |
131
|
|
|
/** @var AbstractNode $node */ |
132
|
3 |
|
foreach ($nodes as $node) { |
133
|
3 |
|
if ($rule['tag'] == '*' |
134
|
3 |
|
|| $rule['tag'] == $node->getTag() |
135
|
3 |
|
->name() |
136
|
|
|
) { |
137
|
3 |
|
++$count; |
138
|
3 |
|
if ($count == $rule['key']) { |
139
|
|
|
// found the node we wanted |
140
|
3 |
|
yield $node; |
141
|
|
|
} |
142
|
|
|
} |
143
|
|
|
} |
144
|
|
|
|
145
|
3 |
|
return; |
146
|
|
|
} |
147
|
|
|
|
148
|
300 |
|
$options = $this->flattenOptions($options); |
149
|
|
|
|
150
|
|
|
/** @var InnerNode $node */ |
151
|
300 |
|
foreach ($nodes as $node) { |
152
|
|
|
// check if we are a leaf |
153
|
300 |
|
if ($node instanceof LeafNode || !$node->hasChildren() |
154
|
|
|
) { |
155
|
12 |
|
continue; |
156
|
|
|
} |
157
|
|
|
|
158
|
300 |
|
$children = []; |
159
|
300 |
|
$child = $node->firstChild(); |
160
|
300 |
|
while (!is_null($child)) { |
161
|
|
|
// wild card, grab all |
162
|
300 |
|
if ($rule['tag'] == '*' && is_null($rule['key'])) { |
163
|
15 |
|
yield $child; |
164
|
15 |
|
$child = $this->getNextChild($node, $child); |
165
|
15 |
|
continue; |
166
|
|
|
} |
167
|
|
|
|
168
|
300 |
|
$pass = $this->checkTag($rule, $child); |
169
|
300 |
|
if ($pass && !is_null($rule['key'])) { |
170
|
99 |
|
$pass = $this->checkKey($rule, $child); |
171
|
|
|
} |
172
|
300 |
|
if ($pass && !is_null($rule['key']) && !is_null($rule['value']) |
173
|
300 |
|
&& $rule['value'] != '*' |
174
|
|
|
) { |
175
|
96 |
|
$pass = $this->checkComparison($rule, $child); |
176
|
|
|
} |
177
|
|
|
|
178
|
300 |
|
if ($pass) { |
179
|
|
|
// it passed all checks |
180
|
243 |
|
yield $child; |
181
|
|
|
} else { |
182
|
|
|
// this child failed to be matched |
183
|
276 |
|
if ($child instanceof InnerNode && $child->hasChildren() |
184
|
|
|
) { |
185
|
249 |
|
if ($this->depthFirst) { |
186
|
231 |
|
if (!isset($options['checkGrandChildren']) |
187
|
231 |
|
|| $options['checkGrandChildren'] |
188
|
|
|
) { |
189
|
|
|
// we have a child that failed but are not leaves. |
190
|
231 |
|
$matches = $this->seek([$child], $rule, $options); |
191
|
231 |
|
foreach ($matches as $match) { |
192
|
131 |
|
yield $match; |
193
|
|
|
} |
194
|
|
|
} |
195
|
|
|
} else { |
196
|
|
|
// we still want to check its children |
197
|
72 |
|
$children[] = $child; |
198
|
|
|
} |
199
|
|
|
} |
200
|
|
|
} |
201
|
|
|
|
202
|
297 |
|
$child = $this->getNextChild($node, $child); |
203
|
|
|
} |
204
|
|
|
|
205
|
294 |
|
if ((!isset($options['checkGrandChildren']) |
206
|
294 |
|
|| $options['checkGrandChildren']) |
207
|
294 |
|
&& count($children) > 0 |
208
|
|
|
) { |
209
|
|
|
// we have children that failed but are not leaves. |
210
|
69 |
|
$matches = $this->seek($children, $rule, $options); |
211
|
69 |
|
foreach ($matches as $match) { |
212
|
138 |
|
yield $match; |
213
|
|
|
} |
214
|
|
|
} |
215
|
|
|
} |
216
|
294 |
|
} |
217
|
|
|
|
218
|
|
|
/** |
219
|
|
|
* Attempts to match the given arguments with the given operator. |
220
|
|
|
* @param string $operator |
221
|
|
|
* @param string $pattern |
222
|
|
|
* @param string $value |
223
|
|
|
* @return bool |
224
|
|
|
*/ |
225
|
96 |
|
protected function match( |
226
|
|
|
string $operator, |
227
|
|
|
string $pattern, |
228
|
|
|
string $value |
229
|
|
|
): bool { |
230
|
96 |
|
$value = strtolower($value); |
231
|
96 |
|
$pattern = strtolower($pattern); |
232
|
64 |
|
switch ($operator) { |
233
|
96 |
|
case '=': |
234
|
96 |
|
return $value === $pattern; |
235
|
|
|
case '!=': |
236
|
|
|
return $value !== $pattern; |
237
|
|
|
case '^=': |
238
|
|
|
return preg_match('/^' . preg_quote($pattern, '/') . '/', |
239
|
|
|
$value) == 1; |
240
|
|
|
case '$=': |
241
|
|
|
return preg_match('/' . preg_quote($pattern, '/') . '$/', |
242
|
|
|
$value) == 1; |
243
|
|
|
case '*=': |
244
|
|
|
if ($pattern[0] == '/') { |
245
|
|
|
return preg_match($pattern, $value) == 1; |
246
|
|
|
} |
247
|
|
|
|
248
|
|
|
return preg_match("/" . $pattern . "/i", $value) == 1; |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
return false; |
252
|
|
|
} |
253
|
|
|
|
254
|
|
|
/** |
255
|
|
|
* Attempts to figure out what the alteration will be for |
256
|
|
|
* the next element. |
257
|
|
|
* @param array $rule |
258
|
|
|
* @return array |
259
|
|
|
*/ |
260
|
3 |
|
protected function alterNext(array $rule): array |
261
|
|
|
{ |
262
|
3 |
|
$options = []; |
263
|
3 |
|
if ($rule['tag'] == '>') { |
264
|
3 |
|
$options['checkGrandChildren'] = false; |
265
|
|
|
} |
266
|
|
|
|
267
|
3 |
|
return $options; |
268
|
|
|
} |
269
|
|
|
|
270
|
|
|
/** |
271
|
|
|
* Flattens the option array. |
272
|
|
|
* @param array $optionsArray |
273
|
|
|
* @return array |
274
|
|
|
*/ |
275
|
300 |
|
protected function flattenOptions(array $optionsArray) |
276
|
|
|
{ |
277
|
300 |
|
$options = []; |
278
|
300 |
|
foreach ($optionsArray as $optionArray) { |
279
|
3 |
|
foreach ($optionArray as $key => $option) { |
280
|
3 |
|
$options[$key] = $option; |
281
|
|
|
} |
282
|
|
|
} |
283
|
|
|
|
284
|
300 |
|
return $options; |
285
|
|
|
} |
286
|
|
|
|
287
|
|
|
/** |
288
|
|
|
* Returns the next child or null if no more children. |
289
|
|
|
* @param AbstractNode $node |
290
|
|
|
* @param AbstractNode $currentChild |
291
|
|
|
* @return AbstractNode|null |
292
|
|
|
*/ |
293
|
297 |
|
protected function getNextChild( |
294
|
|
|
AbstractNode $node, |
295
|
|
|
AbstractNode $currentChild |
296
|
|
|
) { |
297
|
|
|
try { |
298
|
297 |
|
$child = null; |
299
|
297 |
|
if ($node instanceof InnerNode) { |
300
|
|
|
// get next child |
301
|
297 |
|
$child = $node->nextChild($currentChild->id()); |
302
|
|
|
} |
303
|
294 |
|
} catch (ChildNotFoundException $e) { |
304
|
|
|
// no more children |
305
|
294 |
|
unset($e); |
306
|
294 |
|
$child = null; |
307
|
|
|
} |
308
|
|
|
|
309
|
297 |
|
return $child; |
310
|
|
|
} |
311
|
|
|
|
312
|
|
|
/** |
313
|
|
|
* Checks tag condition from rules against node. |
314
|
|
|
* @param array $rule |
315
|
|
|
* @param AbstractNode $node |
316
|
|
|
* @return bool |
317
|
|
|
*/ |
318
|
300 |
|
protected function checkTag(array $rule, AbstractNode $node): bool |
319
|
|
|
{ |
320
|
300 |
|
if (!empty($rule['tag']) && $rule['tag'] != $node->getTag()->name() |
321
|
300 |
|
&& $rule['tag'] != '*' |
322
|
|
|
) { |
323
|
264 |
|
return false; |
324
|
|
|
} |
325
|
|
|
|
326
|
243 |
|
return true; |
327
|
|
|
} |
328
|
|
|
|
329
|
|
|
/** |
330
|
|
|
* Checks key condition from rules against node. |
331
|
|
|
* @param array $rule |
332
|
|
|
* @param AbstractNode $node |
333
|
|
|
* @return bool |
334
|
|
|
*/ |
335
|
99 |
|
protected function checkKey(array $rule, AbstractNode $node): bool |
336
|
|
|
{ |
337
|
99 |
|
if (!is_array($rule['key'])) { |
338
|
96 |
|
if ($rule['noKey']) { |
339
|
|
|
if (!is_null($node->getAttribute($rule['key']))) { |
|
|
|
|
340
|
|
|
return false; |
341
|
|
|
} |
342
|
|
|
} else { |
343
|
96 |
|
if ($rule['key'] != 'plaintext' |
344
|
96 |
|
&& !$node->hasAttribute($rule['key']) |
345
|
|
|
) { |
346
|
96 |
|
return false; |
347
|
|
|
} |
348
|
|
|
} |
349
|
|
|
} else { |
350
|
3 |
|
if ($rule['noKey']) { |
351
|
|
|
foreach ($rule['key'] as $key) { |
352
|
|
|
if (!is_null($node->getAttribute($key))) { |
|
|
|
|
353
|
|
|
return false; |
354
|
|
|
} |
355
|
|
|
} |
356
|
|
|
} else { |
357
|
3 |
|
foreach ($rule['key'] as $key) { |
358
|
3 |
|
if ($key != 'plaintext' |
359
|
3 |
|
&& !$node->hasAttribute($key) |
360
|
|
|
) { |
361
|
1 |
|
return false; |
362
|
|
|
} |
363
|
|
|
} |
364
|
|
|
} |
365
|
|
|
} |
366
|
|
|
|
367
|
99 |
|
return true; |
368
|
|
|
} |
369
|
|
|
|
370
|
|
|
/** |
371
|
|
|
* Checks comparison condition from rules against node. |
372
|
|
|
* @param array $rule |
373
|
|
|
* @param AbstractNode $node |
374
|
|
|
* @return bool |
375
|
|
|
*/ |
376
|
96 |
|
public function checkComparison(array $rule, AbstractNode $node): bool |
377
|
|
|
{ |
378
|
96 |
|
if ($rule['key'] == 'plaintext') { |
379
|
|
|
// plaintext search |
380
|
|
|
$nodeValue = $node->text(); |
381
|
|
|
$result = $this->checkNodeValue($nodeValue, $rule, $node); |
382
|
|
|
} else { |
383
|
|
|
// normal search |
384
|
96 |
|
if (!is_array($rule['key'])) { |
385
|
93 |
|
$nodeValue = $node->getAttribute($rule['key']); |
|
|
|
|
386
|
93 |
|
$result = $this->checkNodeValue($nodeValue, $rule, $node); |
387
|
|
|
} else { |
388
|
3 |
|
$result = true; |
389
|
3 |
|
foreach ($rule['key'] as $index => $key) { |
390
|
3 |
|
$nodeValue = $node->getAttribute($key); |
|
|
|
|
391
|
3 |
|
$result = $result && |
392
|
3 |
|
$this->checkNodeValue($nodeValue, $rule, $node, $index); |
393
|
|
|
} |
394
|
|
|
} |
395
|
|
|
} |
396
|
|
|
|
397
|
96 |
|
return $result; |
398
|
|
|
} |
399
|
|
|
|
400
|
|
|
/** |
401
|
|
|
* @param string|null $nodeValue |
402
|
|
|
* @param array $rule |
403
|
|
|
* @param AbstractNode $node |
404
|
|
|
* @param int|null $index |
405
|
|
|
* @return bool |
406
|
|
|
*/ |
407
|
96 |
|
private function checkNodeValue( |
408
|
|
|
?string $nodeValue, |
409
|
|
|
array $rule, |
410
|
|
|
AbstractNode $node, |
411
|
|
|
?int $index = null |
412
|
|
|
) : bool { |
413
|
96 |
|
$check = false; |
414
|
|
|
if ( |
415
|
96 |
|
array_key_exists('value', $rule) && !is_array($rule['value']) && |
416
|
96 |
|
!is_null($nodeValue) && |
417
|
96 |
|
array_key_exists('operator', $rule) && is_string($rule['operator']) && |
418
|
96 |
|
array_key_exists('value', $rule) && is_string($rule['value']) |
419
|
|
|
) { |
420
|
51 |
|
$check = $this->match($rule['operator'], $rule['value'], $nodeValue); |
421
|
|
|
} |
422
|
|
|
|
423
|
|
|
// handle multiple classes |
424
|
96 |
|
$key = $rule['key']; |
425
|
|
|
if ( |
426
|
96 |
|
!$check && |
427
|
96 |
|
$key == 'class' && |
428
|
96 |
|
array_key_exists('value', $rule) && is_array($rule['value']) |
429
|
|
|
) { |
430
|
51 |
|
$nodeClasses = explode(' ', $node->getAttribute('class') ?? ''); |
|
|
|
|
431
|
51 |
|
foreach ($rule['value'] as $value) { |
432
|
51 |
|
foreach ($nodeClasses as $class) { |
433
|
|
|
if ( |
434
|
51 |
|
!empty($class) && |
435
|
51 |
|
array_key_exists('operator', $rule) && is_string($rule['operator']) |
436
|
|
|
) { |
437
|
51 |
|
$check = $this->match($rule['operator'], $value, $class); |
438
|
|
|
} |
439
|
51 |
|
if ($check) { |
440
|
51 |
|
break; |
441
|
|
|
} |
442
|
|
|
} |
443
|
51 |
|
if (!$check) { |
444
|
43 |
|
break; |
445
|
|
|
} |
446
|
|
|
} |
447
|
|
|
} elseif ( |
448
|
54 |
|
!$check && |
449
|
54 |
|
is_array($key) && |
450
|
54 |
|
!is_null($nodeValue) && |
451
|
54 |
|
array_key_exists('operator', $rule) && is_string($rule['operator']) && |
452
|
54 |
|
array_key_exists('value', $rule) && is_string($rule['value'][$index]) |
453
|
|
|
) { |
454
|
3 |
|
$check = $this->match($rule['operator'], $rule['value'][$index], $nodeValue); |
455
|
|
|
} |
456
|
|
|
|
457
|
96 |
|
return $check; |
458
|
|
|
} |
459
|
|
|
} |
460
|
|
|
|
If the size of the collection does not change during the iteration, it is generally a good practice to compute it beforehand, and not on each iteration: