1
|
|
|
<?php |
2
|
|
|
namespace PHPHtmlParser\Selector; |
3
|
|
|
|
4
|
|
|
use PHPHtmlParser\Dom\AbstractNode; |
5
|
|
|
use PHPHtmlParser\Dom\Collection; |
6
|
|
|
use PHPHtmlParser\Dom\InnerNode; |
7
|
|
|
use PHPHtmlParser\Dom\LeafNode; |
8
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* Class Selector |
12
|
|
|
* |
13
|
|
|
* @package PHPHtmlParser |
14
|
|
|
*/ |
15
|
|
|
class Selector |
16
|
|
|
{ |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* @var array |
20
|
|
|
*/ |
21
|
|
|
protected $selectors = []; |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* Constructs with the selector string |
25
|
|
|
* |
26
|
|
|
* @param string $selector |
27
|
|
|
*/ |
28
|
255 |
|
public function __construct(string $selector, ParserInterface $parser) |
29
|
|
|
{ |
30
|
255 |
|
$this->selectors = $parser->parseSelectorString($selector); |
31
|
255 |
|
} |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* Returns the selectors that where found in __construct |
35
|
|
|
* |
36
|
|
|
* @return array |
37
|
|
|
*/ |
38
|
12 |
|
public function getSelectors() |
39
|
|
|
{ |
40
|
12 |
|
return $this->selectors; |
41
|
|
|
} |
42
|
|
|
|
43
|
|
|
/** |
44
|
|
|
* Attempts to find the selectors starting from the given |
45
|
|
|
* node object. |
46
|
|
|
* |
47
|
|
|
* @param AbstractNode $node |
48
|
|
|
* @return Collection |
49
|
|
|
*/ |
50
|
243 |
|
public function find(AbstractNode $node): Collection |
51
|
|
|
{ |
52
|
243 |
|
$results = new Collection; |
53
|
243 |
|
foreach ($this->selectors as $selector) { |
54
|
243 |
|
$nodes = [$node]; |
55
|
243 |
|
if (count($selector) == 0) { |
56
|
|
|
continue; |
57
|
|
|
} |
58
|
|
|
|
59
|
243 |
|
$options = []; |
60
|
243 |
|
foreach ($selector as $rule) { |
61
|
243 |
|
if ($rule['alterNext']) { |
62
|
3 |
|
$options[] = $this->alterNext($rule); |
63
|
3 |
|
continue; |
64
|
|
|
} |
65
|
243 |
|
$nodes = $this->seek($nodes, $rule, $options); |
66
|
|
|
// clear the options |
67
|
243 |
|
$options = []; |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
// this is the final set of nodes |
71
|
243 |
|
foreach ($nodes as $result) { |
72
|
228 |
|
$results[] = $result; |
73
|
|
|
} |
74
|
|
|
} |
75
|
|
|
|
76
|
243 |
|
return $results; |
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
|
80
|
|
|
/** |
81
|
|
|
* Attempts to find all children that match the rule |
82
|
|
|
* given. |
83
|
|
|
* |
84
|
|
|
* @param array $nodes |
85
|
|
|
* @param array $rule |
86
|
|
|
* @param array $options |
87
|
|
|
* @return array |
88
|
|
|
* @recursive |
89
|
|
|
*/ |
90
|
243 |
|
protected function seek(array $nodes, array $rule, array $options): array |
91
|
|
|
{ |
92
|
|
|
// XPath index |
93
|
243 |
|
if (array_key_exists('tag', $rule) && |
94
|
243 |
|
array_key_exists('key', $rule) && |
95
|
243 |
|
is_numeric($rule['key']) |
96
|
|
|
) { |
97
|
3 |
|
$count = 0; |
98
|
|
|
/** @var AbstractNode $node */ |
99
|
3 |
|
foreach ($nodes as $node) { |
100
|
3 |
|
if ($rule['tag'] == '*' || |
101
|
3 |
|
$rule['tag'] == $node->getTag()->name() |
102
|
|
|
) { |
103
|
3 |
|
++$count; |
104
|
3 |
|
if ($count == $rule['key']) { |
105
|
|
|
// found the node we wanted |
106
|
3 |
|
return [$node]; |
107
|
|
|
} |
108
|
|
|
} |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
return []; |
112
|
|
|
} |
113
|
|
|
|
114
|
240 |
|
$options = $this->flattenOptions($options); |
115
|
|
|
|
116
|
240 |
|
$return = []; |
117
|
|
|
/** @var InnerNode $node */ |
118
|
240 |
|
foreach ($nodes as $node) { |
119
|
|
|
// check if we are a leaf |
120
|
240 |
|
if ($node instanceof LeafNode || |
121
|
240 |
|
! $node->hasChildren() |
122
|
|
|
) { |
123
|
12 |
|
continue; |
124
|
|
|
} |
125
|
|
|
|
126
|
240 |
|
$children = []; |
127
|
240 |
|
$child = $node->firstChild(); |
128
|
240 |
|
while ( ! is_null($child)) { |
129
|
|
|
// wild card, grab all |
130
|
240 |
|
if ($rule['tag'] == '*' && is_null($rule['key'])) { |
131
|
12 |
|
$return[] = $child; |
132
|
12 |
|
$child = $this->getNextChild($node, $child); |
133
|
12 |
|
continue; |
134
|
|
|
} |
135
|
|
|
|
136
|
240 |
|
$pass = $this->checkTag($rule, $child); |
137
|
240 |
|
if ($pass && ! is_null($rule['key'])) { |
138
|
84 |
|
$pass = $this->checkKey($rule, $child); |
139
|
|
|
} |
140
|
240 |
|
if ($pass && ! is_null($rule['key']) && |
141
|
240 |
|
! is_null($rule['value']) && $rule['value'] != '*' |
142
|
|
|
) { |
143
|
81 |
|
$pass = $this->checkComparison($rule, $child); |
144
|
|
|
} |
145
|
|
|
|
146
|
240 |
|
if ($pass) { |
147
|
|
|
// it passed all checks |
148
|
195 |
|
$return[] = $child; |
149
|
|
|
} else { |
150
|
|
|
// this child failed to be matched |
151
|
225 |
|
if ($child instanceof InnerNode && |
152
|
225 |
|
$child->hasChildren() |
153
|
|
|
) { |
154
|
|
|
// we still want to check its children |
155
|
213 |
|
$children[] = $child; |
156
|
|
|
} |
157
|
|
|
} |
158
|
|
|
|
159
|
240 |
|
$child = $this->getNextChild($node, $child); |
160
|
|
|
} |
161
|
|
|
|
162
|
240 |
|
if (( ! isset($options['checkGrandChildren']) || |
163
|
240 |
|
$options['checkGrandChildren']) |
164
|
240 |
|
&& count($children) > 0 |
165
|
|
|
) { |
166
|
|
|
// we have children that failed but are not leaves. |
167
|
210 |
|
$matches = $this->seek($children, $rule, $options); |
168
|
210 |
|
foreach ($matches as $match) { |
169
|
196 |
|
$return[] = $match; |
170
|
|
|
} |
171
|
|
|
} |
172
|
|
|
} |
173
|
|
|
|
174
|
240 |
|
return $return; |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
/** |
178
|
|
|
* Attempts to match the given arguments with the given operator. |
179
|
|
|
* |
180
|
|
|
* @param string $operator |
181
|
|
|
* @param string $pattern |
182
|
|
|
* @param string $value |
183
|
|
|
* @return bool |
184
|
|
|
*/ |
185
|
81 |
|
protected function match(string $operator, string $pattern, string $value): bool |
186
|
|
|
{ |
187
|
81 |
|
$value = strtolower($value); |
188
|
81 |
|
$pattern = strtolower($pattern); |
189
|
27 |
|
switch ($operator) { |
190
|
81 |
|
case '=': |
191
|
81 |
|
return $value === $pattern; |
192
|
|
|
case '!=': |
193
|
|
|
return $value !== $pattern; |
194
|
|
|
case '^=': |
195
|
|
|
return preg_match('/^'.preg_quote($pattern, '/').'/', $value) == 1; |
196
|
|
|
case '$=': |
197
|
|
|
return preg_match('/'.preg_quote($pattern, '/').'$/', $value) == 1; |
198
|
|
|
case '*=': |
199
|
|
|
if ($pattern[0] == '/') { |
200
|
|
|
return preg_match($pattern, $value) == 1; |
201
|
|
|
} |
202
|
|
|
|
203
|
|
|
return preg_match("/".$pattern."/i", $value) == 1; |
204
|
|
|
} |
205
|
|
|
|
206
|
|
|
return false; |
207
|
|
|
} |
208
|
|
|
|
209
|
|
|
/** |
210
|
|
|
* Attempts to figure out what the alteration will be for |
211
|
|
|
* the next element. |
212
|
|
|
* |
213
|
|
|
* @param array $rule |
214
|
|
|
* @return array |
215
|
|
|
*/ |
216
|
3 |
|
protected function alterNext(array $rule): array |
217
|
|
|
{ |
218
|
3 |
|
$options = []; |
219
|
3 |
|
if ($rule['tag'] == '>') { |
220
|
3 |
|
$options['checkGrandChildren'] = false; |
221
|
|
|
} |
222
|
|
|
|
223
|
3 |
|
return $options; |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
/** |
227
|
|
|
* Flattens the option array. |
228
|
|
|
* |
229
|
|
|
* @param array $optionsArray |
230
|
|
|
* @return array |
231
|
|
|
*/ |
232
|
240 |
|
protected function flattenOptions(array $optionsArray) |
233
|
|
|
{ |
234
|
240 |
|
$options = []; |
235
|
240 |
|
foreach ($optionsArray as $optionArray) { |
236
|
3 |
|
foreach ($optionArray as $key => $option) { |
237
|
3 |
|
$options[$key] = $option; |
238
|
|
|
} |
239
|
|
|
} |
240
|
|
|
|
241
|
240 |
|
return $options; |
242
|
|
|
} |
243
|
|
|
|
244
|
|
|
/** |
245
|
|
|
* Returns the next child or null if no more children. |
246
|
|
|
* |
247
|
|
|
* @param AbstractNode $node |
248
|
|
|
* @param AbstractNode $currentChild |
249
|
|
|
* @return AbstractNode|null |
250
|
|
|
*/ |
251
|
240 |
|
protected function getNextChild(AbstractNode $node, AbstractNode $currentChild) |
252
|
|
|
{ |
253
|
|
|
try { |
254
|
|
|
// get next child |
255
|
240 |
|
$child = $node->nextChild($currentChild->id()); |
|
|
|
|
256
|
240 |
|
} catch (ChildNotFoundException $e) { |
257
|
|
|
// no more children |
258
|
240 |
|
$child = null; |
259
|
|
|
} |
260
|
|
|
|
261
|
240 |
|
return $child; |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
/** |
265
|
|
|
* Checks tag condition from rules against node. |
266
|
|
|
* |
267
|
|
|
* @param array $rule |
268
|
|
|
* @param AbstractNode $node |
269
|
|
|
* @return bool |
270
|
|
|
*/ |
271
|
240 |
|
protected function checkTag(array $rule, AbstractNode $node): bool |
272
|
|
|
{ |
273
|
240 |
|
if ( ! empty($rule['tag']) && $rule['tag'] != $node->getTag()->name() && |
274
|
240 |
|
$rule['tag'] != '*' |
275
|
|
|
) { |
276
|
216 |
|
return false; |
277
|
|
|
} |
278
|
|
|
|
279
|
195 |
|
return true; |
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
/** |
283
|
|
|
* Checks key condition from rules against node. |
284
|
|
|
* |
285
|
|
|
* @param array $rule |
286
|
|
|
* @param AbstractNode $node |
287
|
|
|
* @return bool |
288
|
|
|
*/ |
289
|
84 |
|
protected function checkKey(array $rule, AbstractNode $node): bool |
290
|
|
|
{ |
291
|
84 |
|
if ($rule['noKey']) { |
292
|
|
|
if ( ! is_null($node->getAttribute($rule['key']))) { |
293
|
|
|
return false; |
294
|
|
|
} |
295
|
|
|
} else { |
296
|
84 |
|
if ($rule['key'] != 'plaintext' && !$node->hasAttribute($rule['key'])) { |
297
|
81 |
|
return false; |
298
|
|
|
} |
299
|
|
|
} |
300
|
|
|
|
301
|
84 |
|
return true; |
302
|
|
|
} |
303
|
|
|
|
304
|
|
|
/** |
305
|
|
|
* Checks comparison condition from rules against node. |
306
|
|
|
* |
307
|
|
|
* @param array $rule |
308
|
|
|
* @param AbstractNode $node |
309
|
|
|
* @return bool |
310
|
|
|
*/ |
311
|
81 |
|
public function checkComparison(array $rule, AbstractNode $node): bool |
312
|
|
|
{ |
313
|
81 |
|
if ($rule['key'] == 'plaintext') { |
314
|
|
|
// plaintext search |
315
|
|
|
$nodeValue = $node->text(); |
316
|
|
|
} else { |
317
|
|
|
// normal search |
318
|
81 |
|
$nodeValue = $node->getAttribute($rule['key']); |
319
|
|
|
} |
320
|
|
|
|
321
|
81 |
|
$check = $this->match($rule['operator'], $rule['value'], $nodeValue); |
322
|
|
|
|
323
|
|
|
// handle multiple classes |
324
|
81 |
|
if ( ! $check && $rule['key'] == 'class') { |
325
|
36 |
|
$nodeClasses = explode(' ', $node->getAttribute('class')); |
326
|
36 |
|
foreach ($nodeClasses as $class) { |
327
|
36 |
|
if ( ! empty($class)) { |
328
|
36 |
|
$check = $this->match($rule['operator'], $rule['value'], $class); |
329
|
|
|
} |
330
|
36 |
|
if ($check) { |
331
|
31 |
|
break; |
332
|
|
|
} |
333
|
|
|
} |
334
|
|
|
} |
335
|
|
|
|
336
|
81 |
|
return $check; |
337
|
|
|
} |
338
|
|
|
} |
339
|
|
|
|