xinningsu /
html-query
| 1 | <?php |
||
| 2 | |||
| 3 | namespace Sulao\HtmlQuery; |
||
| 4 | |||
| 5 | use DOMDocument; |
||
| 6 | use DOMNode; |
||
| 7 | use DOMNodeList; |
||
| 8 | use DOMXPath; |
||
| 9 | use Symfony\Component\CssSelector\CssSelectorConverter; |
||
| 10 | use Traversable; |
||
| 11 | |||
| 12 | /** |
||
| 13 | * Class Helper |
||
| 14 | * |
||
| 15 | * @package Sulao\HtmlQuery |
||
| 16 | */ |
||
| 17 | class Helper |
||
| 18 | { |
||
| 19 | /** |
||
| 20 | * Convert a css selector to xpath |
||
| 21 | * |
||
| 22 | * @param string $selector |
||
| 23 | * @param string $prefix |
||
| 24 | * |
||
| 25 | * @return string |
||
| 26 | */ |
||
| 27 | 80 | public static function toXpath( |
|
| 28 | string $selector, |
||
| 29 | string $prefix = 'descendant::' |
||
| 30 | ): string { |
||
| 31 | 80 | static $converter; |
|
| 32 | 80 | $converter = $converter ?: new CssSelectorConverter(); |
|
| 33 | |||
| 34 | 80 | return $converter->toXPath($selector, $prefix); |
|
| 35 | } |
||
| 36 | |||
| 37 | /** |
||
| 38 | * Strict Array Unique |
||
| 39 | * |
||
| 40 | * @param array|Traversable $arr |
||
| 41 | * |
||
| 42 | * @return array |
||
| 43 | */ |
||
| 44 | 83 | public static function strictArrayUnique($arr): array |
|
| 45 | { |
||
| 46 | 83 | $uniqueArr = []; |
|
| 47 | 83 | foreach ($arr as $value) { |
|
| 48 | 81 | if (!in_array($value, $uniqueArr, true)) { |
|
| 49 | 81 | $uniqueArr[] = $value; |
|
| 50 | } |
||
| 51 | } |
||
| 52 | |||
| 53 | 83 | return $uniqueArr; |
|
| 54 | } |
||
| 55 | |||
| 56 | /** |
||
| 57 | * Strict array intersect |
||
| 58 | * |
||
| 59 | * @param array $arr1 |
||
| 60 | * @param array $arr2 |
||
| 61 | * |
||
| 62 | * @return array |
||
| 63 | */ |
||
| 64 | 5 | public static function strictArrayIntersect(array $arr1, array $arr2): array |
|
| 65 | { |
||
| 66 | 5 | $arr = array_filter($arr1, function ($val1) use ($arr2) { |
|
| 67 | 5 | return in_array($val1, $arr2, true); |
|
| 68 | 5 | }); |
|
| 69 | |||
| 70 | 5 | return array_values($arr); |
|
| 71 | } |
||
| 72 | |||
| 73 | /** |
||
| 74 | * Strict array diff |
||
| 75 | * |
||
| 76 | * @param array $arr1 |
||
| 77 | * @param array $arr2 |
||
| 78 | * |
||
| 79 | * @return array |
||
| 80 | */ |
||
| 81 | 3 | public static function strictArrayDiff(array $arr1, array $arr2): array |
|
| 82 | { |
||
| 83 | 3 | $arr = array_filter($arr1, function ($val1) use ($arr2) { |
|
| 84 | 3 | return !in_array($val1, $arr2, true); |
|
| 85 | 3 | }); |
|
| 86 | |||
| 87 | 3 | return array_values($arr); |
|
| 88 | } |
||
| 89 | |||
| 90 | /** |
||
| 91 | * Case insensitive search |
||
| 92 | * |
||
| 93 | * @param string $needle |
||
| 94 | * @param string[] $haystack |
||
| 95 | * |
||
| 96 | * @return array |
||
| 97 | */ |
||
| 98 | 2 | public static function caseInsensitiveSearch( |
|
| 99 | string $needle, |
||
| 100 | array $haystack |
||
| 101 | ): array { |
||
| 102 | 2 | $needle = strtolower($needle); |
|
| 103 | 2 | $match = array_filter($haystack, function ($value) use ($needle) { |
|
| 104 | 2 | return $needle === strtolower($value); |
|
| 105 | 2 | }); |
|
| 106 | |||
| 107 | 2 | return array_values($match); |
|
| 108 | } |
||
| 109 | |||
| 110 | /** |
||
| 111 | * Split the class attr value to a class array |
||
| 112 | * |
||
| 113 | * @param string $className |
||
| 114 | * |
||
| 115 | * @return array |
||
| 116 | */ |
||
| 117 | 6 | public static function splitClass(string $className): array |
|
| 118 | { |
||
| 119 | 6 | return preg_split('/\s+/', trim($className)) ?: []; |
|
| 120 | } |
||
| 121 | |||
| 122 | /** |
||
| 123 | * Split style to css array |
||
| 124 | * |
||
| 125 | * @param string $style |
||
| 126 | * |
||
| 127 | * @return array |
||
| 128 | */ |
||
| 129 | 2 | public static function splitCss(string $style): array |
|
| 130 | { |
||
| 131 | 2 | $arr = explode(';', $style); |
|
| 132 | 2 | $arr = array_map('trim', $arr); |
|
| 133 | |||
| 134 | 2 | $css = []; |
|
| 135 | 2 | foreach ($arr as $value) { |
|
| 136 | 2 | $row = explode(':', $value, 2); |
|
| 137 | 2 | if (count($row) !== 2) { |
|
| 138 | 1 | continue; |
|
| 139 | } |
||
| 140 | |||
| 141 | 2 | $css[trim($row[0])] = trim($row[1]); |
|
| 142 | } |
||
| 143 | |||
| 144 | 2 | return $css; |
|
| 145 | } |
||
| 146 | |||
| 147 | /** |
||
| 148 | * Implode css array to style string |
||
| 149 | * |
||
| 150 | * @param array $css |
||
| 151 | * |
||
| 152 | * @return string |
||
| 153 | */ |
||
| 154 | 2 | public static function implodeCss(array $css): string |
|
| 155 | { |
||
| 156 | 2 | $arr = []; |
|
| 157 | 2 | foreach ($css as $key => $value) { |
|
| 158 | 2 | $arr[] = $key . ': ' . $value; |
|
| 159 | } |
||
| 160 | |||
| 161 | 2 | $style = $arr ? implode('; ', $arr) . ';' : ''; |
|
| 162 | |||
| 163 | 2 | return $style; |
|
| 164 | } |
||
| 165 | |||
| 166 | /** |
||
| 167 | * Determine whether the string is raw html, |
||
| 168 | * otherwise consider it as a css selector |
||
| 169 | * |
||
| 170 | * @param string $html |
||
| 171 | * |
||
| 172 | * @return bool |
||
| 173 | */ |
||
| 174 | 37 | public static function isRawHtml(string $html): bool |
|
| 175 | { |
||
| 176 | 37 | if ($html[0] === '<' && $html[-1] === '>' && strlen($html) >= 3) { |
|
| 177 | 8 | return true; |
|
| 178 | } |
||
| 179 | |||
| 180 | 35 | return (bool) preg_match('/^\s*(<[\w\W]+>)[^>]*$/', $html); |
|
| 181 | } |
||
| 182 | |||
| 183 | /** |
||
| 184 | * Determine whether the selector is a id selector |
||
| 185 | * |
||
| 186 | * @param string $selector |
||
| 187 | * @param string|null $id |
||
| 188 | * |
||
| 189 | * @return bool |
||
| 190 | */ |
||
| 191 | 79 | public static function isIdSelector( |
|
| 192 | string $selector, |
||
| 193 | ?string &$id = null |
||
| 194 | ): bool { |
||
| 195 | 79 | if (preg_match('/^#([\w-]+)$/', $selector, $match)) { |
|
| 196 | 5 | $id = $match[1]; |
|
| 197 | |||
| 198 | 5 | return true; |
|
| 199 | } |
||
| 200 | |||
| 201 | 77 | return false; |
|
| 202 | } |
||
| 203 | |||
| 204 | /** |
||
| 205 | * Query xpath to an array of DOMNode |
||
| 206 | * |
||
| 207 | * @param string $xpath |
||
| 208 | * @param DOMDocument $doc |
||
| 209 | * @param DOMNode|null $node |
||
| 210 | * |
||
| 211 | * @return DOMNode[] |
||
| 212 | */ |
||
| 213 | 79 | public static function xpathQuery( |
|
| 214 | string $xpath, |
||
| 215 | DOMDocument $doc, |
||
| 216 | ?DOMNode $node = null |
||
| 217 | ): array { |
||
| 218 | 79 | $docXpath = new DOMXpath($doc); |
|
| 219 | 79 | $nodeList = $docXpath->query($xpath, $node); |
|
| 220 | |||
| 221 | 79 | if (!($nodeList instanceof DOMNodeList)) { |
|
|
0 ignored issues
–
show
introduced
by
Loading history...
|
|||
| 222 | 1 | return []; |
|
| 223 | } |
||
| 224 | |||
| 225 | 79 | return iterator_to_array($nodeList); |
|
| 226 | } |
||
| 227 | |||
| 228 | /** |
||
| 229 | * Get the node with the relationship of current node. |
||
| 230 | * |
||
| 231 | * @param DOMNode $node |
||
| 232 | * @param string $relation |
||
| 233 | * |
||
| 234 | * @return DOMNode|null |
||
| 235 | */ |
||
| 236 | 4 | public static function getRelationNode(DOMNode $node, string $relation) |
|
| 237 | { |
||
| 238 | /** @var DOMNode $node */ |
||
| 239 | while ( |
||
| 240 | 4 | ($node = $node->$relation) |
|
| 241 | 4 | && $node instanceof DOMNode |
|
| 242 | 4 | && $node->nodeType !== XML_DOCUMENT_NODE |
|
| 243 | ) { |
||
| 244 | 4 | if ($node->nodeType !== XML_ELEMENT_NODE) { |
|
| 245 | 3 | continue; |
|
| 246 | } |
||
| 247 | |||
| 248 | 4 | return $node; |
|
| 249 | } |
||
| 250 | |||
| 251 | 3 | return null; |
|
| 252 | } |
||
| 253 | } |
||
| 254 |