1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace PHPHtmlParser\Selector; |
6
|
|
|
|
7
|
|
|
use PHPHtmlParser\Contracts\Selector\SeekerInterface; |
8
|
|
|
use PHPHtmlParser\Dom\Node\AbstractNode; |
9
|
|
|
use PHPHtmlParser\Dom\Node\InnerNode; |
10
|
|
|
use PHPHtmlParser\Dom\Node\LeafNode; |
11
|
|
|
use PHPHtmlParser\DTO\Selector\RuleDTO; |
12
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
13
|
|
|
|
14
|
|
|
class Seeker implements SeekerInterface |
15
|
|
|
{ |
16
|
|
|
/** |
17
|
|
|
* Attempts to find all children that match the rule |
18
|
|
|
* given. |
19
|
|
|
* |
20
|
|
|
* @var InnerNode[] |
21
|
|
|
* |
22
|
|
|
* @throws ChildNotFoundException |
23
|
|
|
*/ |
24
|
342 |
|
public function seek(array $nodes, RuleDTO $rule, array $options): array |
25
|
|
|
{ |
26
|
|
|
// XPath index |
27
|
342 |
|
if ($rule->getTag() !== null && \is_numeric($rule->getKey())) { |
28
|
3 |
|
$count = 0; |
29
|
3 |
|
foreach ($nodes as $node) { |
30
|
3 |
|
if ($rule->getTag() == '*' |
31
|
3 |
|
|| $rule->getTag() == $node->getTag() |
32
|
3 |
|
->name() |
33
|
|
|
) { |
34
|
3 |
|
++$count; |
35
|
3 |
|
if ($count == $rule->getKey()) { |
36
|
|
|
// found the node we wanted |
37
|
3 |
|
return [$node]; |
38
|
|
|
} |
39
|
|
|
} |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
return []; |
43
|
|
|
} |
44
|
|
|
|
45
|
339 |
|
$options = $this->flattenOptions($options); |
46
|
|
|
|
47
|
339 |
|
$return = []; |
48
|
339 |
|
foreach ($nodes as $node) { |
49
|
|
|
// check if we are a leaf |
50
|
336 |
|
if ($node instanceof LeafNode || !$node->hasChildren() |
51
|
|
|
) { |
52
|
12 |
|
continue; |
53
|
|
|
} |
54
|
|
|
|
55
|
336 |
|
$children = []; |
56
|
336 |
|
$child = $node->firstChild(); |
57
|
336 |
|
while (!\is_null($child)) { |
58
|
|
|
// wild card, grab all |
59
|
336 |
|
if ($rule->getTag() == '*' && \is_null($rule->getKey())) { |
60
|
15 |
|
$return[] = $child; |
61
|
15 |
|
$child = $this->getNextChild($node, $child); |
62
|
15 |
|
continue; |
63
|
|
|
} |
64
|
|
|
|
65
|
336 |
|
$pass = $this->checkTag($rule, $child); |
66
|
336 |
|
if ($pass && $rule->getKey() !== null) { |
67
|
123 |
|
$pass = $this->checkKey($rule, $child); |
68
|
|
|
} |
69
|
336 |
|
if ($pass && |
70
|
336 |
|
$rule->getKey() !== null && |
71
|
336 |
|
$rule->getValue() !== null && |
72
|
336 |
|
$rule->getValue() != '*' |
73
|
|
|
) { |
74
|
114 |
|
$pass = $this->checkComparison($rule, $child); |
75
|
|
|
} |
76
|
|
|
|
77
|
336 |
|
if ($pass) { |
78
|
|
|
// it passed all checks |
79
|
267 |
|
$return[] = $child; |
80
|
|
|
} |
81
|
|
|
// this child failed to be matched |
82
|
336 |
|
if ($child instanceof InnerNode && $child->hasChildren() |
83
|
|
|
) { |
84
|
285 |
|
if (!isset($options['checkGrandChildren']) |
85
|
285 |
|
|| $options['checkGrandChildren'] |
86
|
|
|
) { |
87
|
|
|
// we have a child that failed but are not leaves. |
88
|
285 |
|
$matches = $this->seek([$child], $rule, $options); |
89
|
285 |
|
foreach ($matches as $match) { |
90
|
147 |
|
$return[] = $match; |
91
|
|
|
} |
92
|
|
|
} |
93
|
|
|
} |
94
|
|
|
|
95
|
336 |
|
$child = $this->getNextChild($node, $child); |
96
|
|
|
} |
97
|
|
|
|
98
|
336 |
|
if ((!isset($options['checkGrandChildren']) |
99
|
336 |
|
|| $options['checkGrandChildren']) |
100
|
336 |
|
&& \count($children) > 0 |
101
|
|
|
) { |
102
|
|
|
// we have children that failed but are not leaves. |
103
|
|
|
$matches = $this->seek($children, $rule, $options); |
104
|
|
|
foreach ($matches as $match) { |
105
|
|
|
$return[] = $match; |
106
|
|
|
} |
107
|
|
|
} |
108
|
|
|
} |
109
|
|
|
|
110
|
339 |
|
return $return; |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* Checks comparison condition from rules against node. |
115
|
|
|
*/ |
116
|
114 |
|
private function checkComparison(RuleDTO $rule, AbstractNode $node): bool |
117
|
|
|
{ |
118
|
114 |
|
if ($rule->getKey() == 'plaintext') { |
119
|
|
|
// plaintext search |
120
|
|
|
$nodeValue = $node->text(); |
121
|
|
|
$result = $this->checkNodeValue($nodeValue, $rule, $node); |
122
|
|
|
} else { |
123
|
|
|
// normal search |
124
|
114 |
|
if (!\is_array($rule->getKey())) { |
125
|
111 |
|
$nodeValue = $node->getAttribute($rule->getKey()); |
126
|
111 |
|
$result = $this->checkNodeValue($nodeValue, $rule, $node); |
127
|
|
|
} else { |
128
|
3 |
|
$result = true; |
129
|
3 |
|
foreach ($rule->getKey() as $index => $key) { |
130
|
3 |
|
$nodeValue = $node->getAttribute($key); |
131
|
3 |
|
$result = $result && |
132
|
3 |
|
$this->checkNodeValue($nodeValue, $rule, $node, $index); |
133
|
|
|
} |
134
|
|
|
} |
135
|
|
|
} |
136
|
|
|
|
137
|
114 |
|
return $result; |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
/** |
141
|
|
|
* Flattens the option array. |
142
|
|
|
* |
143
|
|
|
* @return array |
144
|
|
|
*/ |
145
|
339 |
|
private function flattenOptions(array $optionsArray) |
146
|
|
|
{ |
147
|
339 |
|
$options = []; |
148
|
339 |
|
foreach ($optionsArray as $optionArray) { |
149
|
3 |
|
foreach ($optionArray as $key => $option) { |
150
|
3 |
|
$options[$key] = $option; |
151
|
|
|
} |
152
|
|
|
} |
153
|
|
|
|
154
|
339 |
|
return $options; |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
/** |
158
|
|
|
* Returns the next child or null if no more children. |
159
|
|
|
* |
160
|
|
|
* @return AbstractNode|null |
161
|
|
|
*/ |
162
|
336 |
|
private function getNextChild( |
163
|
|
|
AbstractNode $node, |
164
|
|
|
AbstractNode $currentChild |
165
|
|
|
) { |
166
|
|
|
try { |
167
|
336 |
|
$child = null; |
168
|
336 |
|
if ($node instanceof InnerNode) { |
169
|
|
|
// get next child |
170
|
336 |
|
$child = $node->nextChild($currentChild->id()); |
171
|
|
|
} |
172
|
336 |
|
} catch (ChildNotFoundException $e) { |
173
|
|
|
// no more children |
174
|
336 |
|
unset($e); |
175
|
336 |
|
$child = null; |
176
|
|
|
} |
177
|
|
|
|
178
|
336 |
|
return $child; |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
/** |
182
|
|
|
* Checks tag condition from rules against node. |
183
|
|
|
*/ |
184
|
336 |
|
private function checkTag(RuleDTO $rule, AbstractNode $node): bool |
185
|
|
|
{ |
186
|
336 |
|
if (!empty($rule->getTag()) && $rule->getTag() != $node->getTag()->name() |
187
|
336 |
|
&& $rule->getTag() != '*' |
188
|
|
|
) { |
189
|
321 |
|
return false; |
190
|
|
|
} |
191
|
|
|
|
192
|
267 |
|
return true; |
193
|
|
|
} |
194
|
|
|
|
195
|
|
|
/** |
196
|
|
|
* Checks key condition from rules against node. |
197
|
|
|
*/ |
198
|
123 |
|
private function checkKey(RuleDTO $rule, AbstractNode $node): bool |
199
|
|
|
{ |
200
|
123 |
|
if (!\is_array($rule->getKey())) { |
201
|
120 |
|
if ($rule->isNoKey()) { |
202
|
|
|
if ($node->getAttribute($rule->getKey()) !== null) { |
203
|
|
|
return false; |
204
|
|
|
} |
205
|
|
|
} else { |
206
|
120 |
|
if ($rule->getKey() != 'plaintext' |
207
|
120 |
|
&& !$node->hasAttribute($rule->getKey()) |
208
|
|
|
) { |
209
|
120 |
|
return false; |
210
|
|
|
} |
211
|
|
|
} |
212
|
|
|
} else { |
213
|
3 |
|
if ($rule->isNoKey()) { |
214
|
|
|
foreach ($rule->getKey() as $key) { |
215
|
|
|
if (!\is_null($node->getAttribute($key))) { |
216
|
|
|
return false; |
217
|
|
|
} |
218
|
|
|
} |
219
|
|
|
} else { |
220
|
3 |
|
foreach ($rule->getKey() as $key) { |
221
|
3 |
|
if ($key != 'plaintext' |
222
|
3 |
|
&& !$node->hasAttribute($key) |
223
|
|
|
) { |
224
|
|
|
return false; |
225
|
|
|
} |
226
|
|
|
} |
227
|
|
|
} |
228
|
|
|
} |
229
|
|
|
|
230
|
123 |
|
return true; |
231
|
|
|
} |
232
|
|
|
|
233
|
114 |
|
private function checkNodeValue( |
234
|
|
|
?string $nodeValue, |
235
|
|
|
RuleDTO $rule, |
236
|
|
|
AbstractNode $node, |
237
|
|
|
?int $index = null |
238
|
|
|
): bool { |
239
|
114 |
|
$check = false; |
240
|
|
|
if ( |
241
|
114 |
|
$rule->getValue() !== null && |
242
|
114 |
|
\is_string($rule->getValue()) && |
243
|
114 |
|
$nodeValue !== null |
244
|
|
|
) { |
245
|
69 |
|
$check = $this->match($rule->getOperator(), $rule->getValue(), $nodeValue); |
246
|
|
|
} |
247
|
|
|
|
248
|
|
|
// handle multiple classes |
249
|
114 |
|
$key = $rule->getKey(); |
250
|
|
|
if ( |
251
|
114 |
|
!$check && |
252
|
114 |
|
$key == 'class' && |
253
|
114 |
|
\is_array($rule->getValue()) |
254
|
|
|
) { |
255
|
54 |
|
$nodeClasses = \explode(' ', $node->getAttribute('class') ?? ''); |
256
|
54 |
|
foreach ($rule->getValue() as $value) { |
257
|
54 |
|
foreach ($nodeClasses as $class) { |
258
|
|
|
if ( |
259
|
54 |
|
!empty($class) && |
260
|
54 |
|
\is_string($rule->getOperator()) |
261
|
|
|
) { |
262
|
54 |
|
$check = $this->match($rule->getOperator(), $value, $class); |
263
|
|
|
} |
264
|
54 |
|
if ($check) { |
265
|
54 |
|
break; |
266
|
|
|
} |
267
|
|
|
} |
268
|
54 |
|
if (!$check) { |
269
|
45 |
|
break; |
270
|
|
|
} |
271
|
|
|
} |
272
|
|
|
} elseif ( |
273
|
72 |
|
!$check && |
274
|
72 |
|
\is_array($key) && |
275
|
72 |
|
!\is_null($nodeValue) && |
276
|
72 |
|
\is_string($rule->getOperator()) && |
277
|
72 |
|
\is_string($rule->getValue()[$index]) |
278
|
|
|
) { |
279
|
3 |
|
$check = $this->match($rule->getOperator(), $rule->getValue()[$index], $nodeValue); |
280
|
|
|
} |
281
|
|
|
|
282
|
114 |
|
return $check; |
283
|
|
|
} |
284
|
|
|
|
285
|
|
|
/** |
286
|
|
|
* Attempts to match the given arguments with the given operator. |
287
|
|
|
*/ |
288
|
114 |
|
private function match( |
289
|
|
|
string $operator, |
290
|
|
|
string $pattern, |
291
|
|
|
string $value |
292
|
|
|
): bool { |
293
|
114 |
|
$value = \strtolower($value); |
294
|
114 |
|
$pattern = \strtolower($pattern); |
295
|
114 |
|
switch ($operator) { |
296
|
114 |
|
case '=': |
297
|
102 |
|
return $value === $pattern; |
298
|
12 |
|
case '!=': |
299
|
3 |
|
return $value !== $pattern; |
300
|
9 |
|
case '^=': |
301
|
3 |
|
return \preg_match('/^' . \preg_quote($pattern, '/') . '/', |
302
|
3 |
|
$value) == 1; |
303
|
6 |
|
case '$=': |
304
|
3 |
|
return \preg_match('/' . \preg_quote($pattern, '/') . '$/', |
305
|
3 |
|
$value) == 1; |
306
|
3 |
|
case '*=': |
307
|
3 |
|
if ($pattern[0] == '/') { |
308
|
3 |
|
return \preg_match($pattern, $value) == 1; |
309
|
|
|
} |
310
|
|
|
|
311
|
|
|
return \preg_match('/' . $pattern . '/i', $value) == 1; |
312
|
|
|
default: |
313
|
|
|
return false; |
314
|
|
|
} |
315
|
|
|
} |
316
|
|
|
} |
317
|
|
|
|