1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Sulao\HtmlQuery; |
4
|
|
|
|
5
|
|
|
use DOMDocument, DOMNode, DOMNodeList, DOMXPath; |
6
|
|
|
use Symfony\Component\CssSelector\CssSelectorConverter; |
7
|
|
|
use Traversable; |
8
|
|
|
|
9
|
|
|
/** |
10
|
|
|
* Class Helper |
11
|
|
|
* |
12
|
|
|
* @package Sulao\HtmlQuery |
13
|
|
|
*/ |
14
|
|
|
class Helper |
15
|
|
|
{ |
16
|
|
|
/** |
17
|
|
|
* Convert a css selector to xpath |
18
|
|
|
* |
19
|
|
|
* @param string $selector |
20
|
|
|
* @param string $prefix |
21
|
|
|
* |
22
|
|
|
* @return string |
23
|
|
|
*/ |
24
|
80 |
|
public static function toXpath( |
25
|
|
|
string $selector, |
26
|
|
|
string $prefix = 'descendant::' |
27
|
|
|
): string { |
28
|
80 |
|
static $converter; |
29
|
80 |
|
$converter = $converter ?: new CssSelectorConverter(); |
30
|
|
|
|
31
|
80 |
|
return $converter->toXPath($selector, $prefix); |
32
|
|
|
} |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* Strict Array Unique |
36
|
|
|
* |
37
|
|
|
* @param array|Traversable $arr |
38
|
|
|
* |
39
|
|
|
* @return array |
40
|
|
|
*/ |
41
|
83 |
|
public static function strictArrayUnique($arr): array |
42
|
|
|
{ |
43
|
83 |
|
$uniqueArr = []; |
44
|
83 |
|
foreach ($arr as $value) { |
45
|
81 |
|
if (!in_array($value, $uniqueArr, true)) { |
46
|
81 |
|
$uniqueArr[] = $value; |
47
|
|
|
} |
48
|
|
|
} |
49
|
|
|
|
50
|
83 |
|
return $uniqueArr; |
51
|
|
|
} |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* Strict array intersect |
55
|
|
|
* |
56
|
|
|
* @param array $arr1 |
57
|
|
|
* @param array $arr2 |
58
|
|
|
* |
59
|
|
|
* @return array |
60
|
|
|
*/ |
61
|
5 |
|
public static function strictArrayIntersect(array $arr1, array $arr2): array |
62
|
|
|
{ |
63
|
|
|
$arr = array_filter($arr1, function ($val1) use ($arr2) { |
64
|
5 |
|
return in_array($val1, $arr2, true); |
65
|
5 |
|
}); |
66
|
|
|
|
67
|
5 |
|
return array_values($arr); |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* Strict array diff |
72
|
|
|
* |
73
|
|
|
* @param array $arr1 |
74
|
|
|
* @param array $arr2 |
75
|
|
|
* |
76
|
|
|
* @return array |
77
|
|
|
*/ |
78
|
3 |
|
public static function strictArrayDiff(array $arr1, array $arr2): array |
79
|
|
|
{ |
80
|
|
|
$arr = array_filter($arr1, function ($val1) use ($arr2) { |
81
|
3 |
|
return !in_array($val1, $arr2, true); |
82
|
3 |
|
}); |
83
|
|
|
|
84
|
3 |
|
return array_values($arr); |
85
|
|
|
} |
86
|
|
|
|
87
|
|
|
/** |
88
|
|
|
* Case insensitive search |
89
|
|
|
* |
90
|
|
|
* @param string $needle |
91
|
|
|
* @param string[] $haystack |
92
|
|
|
* |
93
|
|
|
* @return array |
94
|
|
|
*/ |
95
|
2 |
|
public static function caseInsensitiveSearch( |
96
|
|
|
string $needle, |
97
|
|
|
array $haystack |
98
|
|
|
): array { |
99
|
2 |
|
$needle = strtolower($needle); |
100
|
|
|
$match = array_filter($haystack, function ($value) use ($needle) { |
101
|
2 |
|
return $needle === strtolower($value); |
102
|
2 |
|
}); |
103
|
|
|
|
104
|
2 |
|
return array_values($match); |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
/** |
108
|
|
|
* Split the class attr value to a class array |
109
|
|
|
* |
110
|
|
|
* @param string $className |
111
|
|
|
* |
112
|
|
|
* @return array |
113
|
|
|
*/ |
114
|
6 |
|
public static function splitClass(string $className): array |
115
|
|
|
{ |
116
|
6 |
|
return preg_split('/\s+/', trim($className)) ?: []; |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
/** |
120
|
|
|
* Split style to css array |
121
|
|
|
* |
122
|
|
|
* @param string $style |
123
|
|
|
* |
124
|
|
|
* @return array |
125
|
|
|
*/ |
126
|
2 |
|
public static function splitCss(string $style): array |
127
|
|
|
{ |
128
|
2 |
|
$arr = explode(';', $style); |
129
|
2 |
|
$arr = array_map('trim', $arr); |
130
|
|
|
|
131
|
2 |
|
$css = []; |
132
|
2 |
|
foreach ($arr as $value) { |
133
|
2 |
|
$row = explode(':', $value, 2); |
134
|
2 |
|
if (count($row) !== 2) { |
135
|
1 |
|
continue; |
136
|
|
|
} |
137
|
|
|
|
138
|
2 |
|
$css[trim($row[0])] = trim($row[1]); |
139
|
|
|
} |
140
|
|
|
|
141
|
2 |
|
return $css; |
142
|
|
|
} |
143
|
|
|
|
144
|
|
|
/** |
145
|
|
|
* Implode css array to style string |
146
|
|
|
* |
147
|
|
|
* @param array $css |
148
|
|
|
* |
149
|
|
|
* @return string |
150
|
|
|
*/ |
151
|
2 |
|
public static function implodeCss(array $css): string |
152
|
|
|
{ |
153
|
2 |
|
$arr = []; |
154
|
2 |
|
foreach ($css as $key => $value) { |
155
|
2 |
|
$arr[] = $key . ': ' . $value; |
156
|
|
|
} |
157
|
|
|
|
158
|
2 |
|
$style = $arr ? implode('; ', $arr) . ';' : ''; |
159
|
|
|
|
160
|
2 |
|
return $style; |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
/** |
164
|
|
|
* Determine whether the string is raw html, |
165
|
|
|
* otherwise consider it as a css selector |
166
|
|
|
* |
167
|
|
|
* @param string $html |
168
|
|
|
* |
169
|
|
|
* @return bool |
170
|
|
|
*/ |
171
|
37 |
|
public static function isRawHtml(string $html): bool |
172
|
|
|
{ |
173
|
37 |
|
if ($html[0] === '<' && $html[-1] === '>' && strlen($html) >= 3) { |
174
|
8 |
|
return true; |
175
|
|
|
} |
176
|
|
|
|
177
|
35 |
|
return (bool) preg_match('/^\s*(<[\w\W]+>)[^>]*$/', $html); |
178
|
|
|
} |
179
|
|
|
|
180
|
|
|
/** |
181
|
|
|
* Determine whether the selector is a id selector |
182
|
|
|
* |
183
|
|
|
* @param string $selector |
184
|
|
|
* @param string|null $id |
185
|
|
|
* |
186
|
|
|
* @return bool |
187
|
|
|
*/ |
188
|
79 |
|
public static function isIdSelector( |
189
|
|
|
string $selector, |
190
|
|
|
string &$id = null |
191
|
|
|
): bool { |
192
|
79 |
|
if (preg_match('/^#([\w-]+)$/', $selector, $match)) { |
193
|
5 |
|
$id = $match[1]; |
194
|
|
|
|
195
|
5 |
|
return true; |
196
|
|
|
} |
197
|
|
|
|
198
|
77 |
|
return false; |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
/** |
202
|
|
|
* Query xpath to an array of DOMNode |
203
|
|
|
* |
204
|
|
|
* @param string $xpath |
205
|
|
|
* @param DOMDocument $doc |
206
|
|
|
* @param DOMNode|null $node |
207
|
|
|
* |
208
|
|
|
* @return DOMNode[] |
209
|
|
|
*/ |
210
|
79 |
|
public static function xpathQuery( |
211
|
|
|
string $xpath, |
212
|
|
|
DOMDocument $doc, |
213
|
|
|
?DOMNode $node = null |
214
|
|
|
): array { |
215
|
79 |
|
$docXpath = new DOMXpath($doc); |
216
|
79 |
|
$nodeList = $docXpath->query($xpath, $node); |
217
|
|
|
|
218
|
79 |
|
if (!($nodeList instanceof DOMNodeList)) { |
219
|
1 |
|
return []; |
220
|
|
|
} |
221
|
|
|
|
222
|
79 |
|
return iterator_to_array($nodeList); |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
/** |
226
|
|
|
* Get the node with the relationship of current node. |
227
|
|
|
* |
228
|
|
|
* @param DOMNode $node |
229
|
|
|
* @param string $relation |
230
|
|
|
* |
231
|
|
|
* @return DOMNode|null |
232
|
|
|
*/ |
233
|
4 |
|
public static function getRelationNode(DOMNode $node, string $relation) |
234
|
|
|
{ |
235
|
|
|
/** @var DOMNode $node */ |
236
|
4 |
|
while (($node = $node->$relation) |
237
|
4 |
|
&& $node instanceof DOMNode |
238
|
4 |
|
&& $node->nodeType !== XML_DOCUMENT_NODE |
239
|
|
|
) { |
240
|
4 |
|
if ($node->nodeType !== XML_ELEMENT_NODE) { |
241
|
3 |
|
continue; |
242
|
|
|
} |
243
|
|
|
|
244
|
4 |
|
return $node; |
245
|
|
|
} |
246
|
|
|
|
247
|
3 |
|
return null; |
248
|
|
|
} |
249
|
|
|
} |
250
|
|
|
|