Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
76.21% |
583 / 765 |
|
56.52% |
26 / 46 |
CRAP | |
0.00% |
0 / 1 |
| ZestInst | |
76.21% |
583 / 765 |
|
56.52% |
26 / 46 |
1053.28 | |
0.00% |
0 / 1 |
| sort | |
50.00% |
3 / 6 |
|
0.00% |
0 / 1 |
6.00 | |||
| next | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
3 | |||
| prev | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
3 | |||
| child | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
4 | |||
| lastChild | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
4 | |||
| parentIsElement | |
83.33% |
5 / 6 |
|
0.00% |
0 / 1 |
4.07 | |||
| nodeIsDocument | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| unichr | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| unquote | |
60.00% |
12 / 20 |
|
0.00% |
0 / 1 |
12.10 | |||
| decodeid | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| encodeid | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
5 | |||
| makeInside | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| reSource | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| replace | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
1 | |||
| truncateUrl | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| xpathQuote | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
2 | |||
| getElementsById | |
66.67% |
16 / 24 |
|
0.00% |
0 / 1 |
17.33 | |||
| docFragHelper | |
87.50% |
7 / 8 |
|
0.00% |
0 / 1 |
4.03 | |||
| getElementsByTagName | |
87.50% |
21 / 24 |
|
0.00% |
0 / 1 |
9.16 | |||
| getElementsByClassName | |
84.21% |
16 / 19 |
|
0.00% |
0 / 1 |
4.06 | |||
| parseNth | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
4 | |||
| nth | |
90.48% |
19 / 21 |
|
0.00% |
0 / 1 |
10.09 | |||
| addSelector0 | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| addSelector1 | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| initSelectors | |
70.53% |
146 / 207 |
|
0.00% |
0 / 1 |
69.17 | |||
| selectorsAttr | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
4 | |||
| addOperator | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| initOperators | |
67.35% |
33 / 49 |
|
0.00% |
0 / 1 |
17.01 | |||
| addCombinator | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| initCombinators | |
94.87% |
37 / 39 |
|
0.00% |
0 / 1 |
12.02 | |||
| makeRef | |
0.00% |
0 / 25 |
|
0.00% |
0 / 1 |
90 | |||
| initRules | |
100.00% |
28 / 28 |
|
100.00% |
1 / 1 |
1 | |||
| compile | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| doCompile | |
68.75% |
44 / 64 |
|
0.00% |
0 / 1 |
21.87 | |||
| tokQname | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
4 | |||
| tok | |
90.48% |
19 / 21 |
|
0.00% |
0 / 1 |
10.09 | |||
| makeSimple | |
100.00% |
9 / 9 |
|
100.00% |
1 / 1 |
4 | |||
| makeTest | |
100.00% |
11 / 11 |
|
100.00% |
1 / 1 |
4 | |||
| makeSubject | |
0.00% |
0 / 17 |
|
0.00% |
0 / 1 |
20 | |||
| compileGroup | |
69.23% |
9 / 13 |
|
0.00% |
0 / 1 |
5.73 | |||
| findInternal | |
100.00% |
21 / 21 |
|
100.00% |
1 / 1 |
8 | |||
| find | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
9 | |||
| matches | |
90.91% |
10 / 11 |
|
0.00% |
0 / 1 |
2.00 | |||
| newBadSelectorException | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| isStandardsMode | |
80.00% |
4 / 5 |
|
0.00% |
0 / 1 |
3.07 | |||
| __construct | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
6 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace Wikimedia\Zest; |
| 4 | |
| 5 | use DOMDocument; |
| 6 | use DOMDocumentFragment; |
| 7 | use DOMElement; |
| 8 | use DOMNode; |
| 9 | use InvalidArgumentException; |
| 10 | use Throwable; |
| 11 | |
| 12 | /** |
| 13 | * Zest.php (https://github.com/wikimedia/zest.php) |
| 14 | * Copyright (c) 2019, C. Scott Ananian. (MIT licensed) |
| 15 | * PHP port based on: |
| 16 | * |
| 17 | * Zest (https://github.com/chjj/zest) |
| 18 | * A css selector engine. |
| 19 | * Copyright (c) 2011-2012, Christopher Jeffrey. (MIT Licensed) |
| 20 | * Domino version based on Zest v0.1.3 with bugfixes applied. |
| 21 | */ |
| 22 | |
| 23 | class ZestInst { |
| 24 | |
| 25 | /** @var ZestFunc[] */ |
| 26 | private $compileCache = []; |
| 27 | |
| 28 | /** |
| 29 | * Helpers |
| 30 | */ |
| 31 | |
| 32 | /** |
| 33 | * Sort query results in document order. |
| 34 | * @param array &$results |
| 35 | * @param bool $isStandardsMode |
| 36 | */ |
| 37 | private static function sort( &$results, bool $isStandardsMode ): void { |
| 38 | if ( count( $results ) < 2 ) { |
| 39 | return; |
| 40 | } |
| 41 | if ( $isStandardsMode ) { |
| 42 | // DOM spec-compliant version: |
| 43 | usort( $results, static function ( $a, $b ) { |
| 44 | return ( $a->compareDocumentPosition( $b ) & 2 ) ? 1 : -1; |
| 45 | } ); |
| 46 | } |
| 47 | // PHP's dom extension returns true for method_exists on |
| 48 | // compareDocumentPosition, but when called it throws a |
| 49 | // "Not yet implemented" exception |
| 50 | |
| 51 | // If compareDocumentPosition isn't implemented, skip the sort. |
| 52 | // Our results are generally added as a result of an in-order |
| 53 | // traversal of the tree, so absent any funny business with complex |
| 54 | // selectors, our natural order should be more-or-less sorted. |
| 55 | } |
| 56 | |
| 57 | /** |
| 58 | * @param DOMNode $el |
| 59 | * @return ?DOMNode |
| 60 | */ |
| 61 | private static function next( $el ) { |
| 62 | while ( ( $el = $el->nextSibling ) && $el->nodeType !== 1 ) { |
| 63 | // no op |
| 64 | } |
| 65 | return $el; |
| 66 | } |
| 67 | |
| 68 | /** |
| 69 | * @param DOMNode $el |
| 70 | * @return ?DOMNode |
| 71 | */ |
| 72 | private static function prev( $el ) { |
| 73 | while ( ( $el = $el->previousSibling ) && $el->nodeType !== 1 ) { |
| 74 | // no op |
| 75 | } |
| 76 | return $el; |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * @param DOMNode $el |
| 81 | * @return ?DOMNode |
| 82 | */ |
| 83 | private static function child( $el ) { |
| 84 | if ( $el = $el->firstChild ) { |
| 85 | while ( $el->nodeType !== 1 && ( $el = $el->nextSibling ) ) { |
| 86 | // no op |
| 87 | } |
| 88 | } |
| 89 | return $el; |
| 90 | } |
| 91 | |
| 92 | /** |
| 93 | * @param DOMNode $el |
| 94 | * @return ?DOMNode |
| 95 | */ |
| 96 | private static function lastChild( $el ) { |
| 97 | if ( $el = $el->lastChild ) { |
| 98 | while ( $el->nodeType !== 1 && ( $el = $el->previousSibling ) ) { |
| 99 | // no op |
| 100 | } |
| 101 | } |
| 102 | return $el; |
| 103 | } |
| 104 | |
| 105 | /** |
| 106 | * @param DOMNode $n |
| 107 | * @return bool |
| 108 | */ |
| 109 | private static function parentIsElement( $n ): bool { |
| 110 | $parent = $n->parentNode; |
| 111 | if ( !$parent ) { |
| 112 | return false; |
| 113 | } |
| 114 | // The root `html` element (node type 9) can be a first- or |
| 115 | // last-child, too, which means that the document (or document |
| 116 | // fragment) counts as an "element". |
| 117 | return $parent->nodeType === 1 /* Element */ || |
| 118 | self::nodeIsDocument( $parent ) /* Document */ || |
| 119 | $parent->nodeType === 11; /* DocumentFragment */ |
| 120 | } |
| 121 | |
| 122 | /** |
| 123 | * @param DOMNode $n |
| 124 | * @return bool |
| 125 | */ |
| 126 | private static function nodeIsDocument( $n ): bool { |
| 127 | $nodeType = $n->nodeType; |
| 128 | return $nodeType === 9 /* Document */ || |
| 129 | // In PHP, if you load a document with |
| 130 | // DOMDocument::loadHTML, your root DOMDocument will have node |
| 131 | // type 13 (!) which is PHP's bespoke "XML_HTML_DOCUMENT_NODE" |
| 132 | // and Not A Real Thing. But we'll recognize it anyway... |
| 133 | $nodeType === 13; /* HTMLDocument */ |
| 134 | } |
| 135 | |
| 136 | private static function unichr( int $codepoint ): string { |
| 137 | if ( extension_loaded( 'intl' ) ) { |
| 138 | return \IntlChar::chr( $codepoint ); |
| 139 | } else { |
| 140 | return mb_chr( $codepoint, "utf-8" ); |
| 141 | } |
| 142 | } |
| 143 | |
| 144 | private static function unquote( string $str ): string { |
| 145 | if ( !$str ) { |
| 146 | return $str; |
| 147 | } |
| 148 | self::initRules(); |
| 149 | $ch = $str[ 0 ]; |
| 150 | if ( $ch === '"' || $ch === "'" ) { |
| 151 | if ( substr( $str, -1 ) === $ch ) { |
| 152 | $str = substr( $str, 1, -1 ); |
| 153 | } else { |
| 154 | // bad string. |
| 155 | $str = substr( $str, 1 ); |
| 156 | } |
| 157 | return preg_replace_callback( self::$rules->str_escape, function ( array $matches ) { |
| 158 | $s = $matches[0]; |
| 159 | if ( !preg_match( '/^\\\(?:([0-9A-Fa-f]+)|([\r\n\f]+))/', $s, $m ) ) { |
| 160 | return substr( $s, 1 ); |
| 161 | } |
| 162 | if ( $m[ 2 ] ) { |
| 163 | return ''; /* escaped newlines are ignored in strings. */ |
| 164 | } |
| 165 | $cp = intval( $m[ 1 ], 16 ); |
| 166 | return self::unichr( $cp ); |
| 167 | }, $str ); |
| 168 | } elseif ( preg_match( self::$rules->ident, $str ) ) { |
| 169 | return self::decodeid( $str ); |
| 170 | } else { |
| 171 | // NUMBER, PERCENTAGE, DIMENSION, etc |
| 172 | return $str; |
| 173 | } |
| 174 | } |
| 175 | |
| 176 | private static function decodeid( string $str ): string { |
| 177 | return preg_replace_callback( self::$rules->escape, function ( array $matches ) { |
| 178 | $s = $matches[0]; |
| 179 | if ( !preg_match( '/^\\\([0-9A-Fa-f]+)/', $s, $m ) ) { |
| 180 | return $s[ 1 ]; |
| 181 | } |
| 182 | $cp = intval( $m[ 1 ], 16 ); |
| 183 | return self::unichr( $cp ); |
| 184 | }, $str ); |
| 185 | } |
| 186 | |
| 187 | /** |
| 188 | * Escape an identifier for CSS. |
| 189 | * This is equivalent to CSS.escape |
| 190 | * (https://drafts.csswg.org/cssom/#the-css.escape()-method) |
| 191 | * and is the opposite of self::decodeid(). |
| 192 | */ |
| 193 | private static function encodeid( string $str ): string { |
| 194 | return preg_replace_callback( '/(\\x00)|([\\x01-\\x1F\\x7F])|(^[0-9])|(^-[0-9])|(^-$)|([^-A-Za-z0-9_\\x{80}-\\x{10FFFF}])/u', static function ( array $matches ) { |
| 195 | if ( isset( $matches[1] ) ) { |
| 196 | return "\u{FFFD}"; |
| 197 | } elseif ( isset( $matches[2] ) || isset( $matches[3] ) ) { |
| 198 | $cp = mb_ord( $matches[0], "UTF-8" ); |
| 199 | return '\\' . dechex( $cp ) . ' '; |
| 200 | } elseif ( isset( $matches[4] ) ) { |
| 201 | $cp = mb_ord( $matches[0][1], "UTF-8" ); |
| 202 | return '-\\' . dechex( $cp ) . ' '; |
| 203 | } else { |
| 204 | return '\\' . $matches[0]; |
| 205 | } |
| 206 | }, $str, -1, $ignore, PREG_UNMATCHED_AS_NULL ); |
| 207 | } |
| 208 | |
| 209 | private static function makeInside( string $start, string $end ): string { |
| 210 | $regex = preg_replace( |
| 211 | '/>/', $end, preg_replace( |
| 212 | '/</', $start, self::reSource( self::$rules->inside ) |
| 213 | ) |
| 214 | ); |
| 215 | return '/' . $regex . '/Su'; |
| 216 | } |
| 217 | |
| 218 | private static function reSource( string $regex ): string { |
| 219 | // strip delimiter and flags from regular expression |
| 220 | return preg_replace( '/(^\/)|(\/[a-z]*$)/Diu', '', $regex ); |
| 221 | } |
| 222 | |
| 223 | private static function replace( string $regex, string $name, string $val ): string { |
| 224 | $regex = self::reSource( $regex ); |
| 225 | $regex = str_replace( $name, self::reSource( $val ), $regex ); |
| 226 | return '/' . $regex . '/Su'; |
| 227 | } |
| 228 | |
| 229 | private static function truncateUrl( string $url, int $num ): string { |
| 230 | $url = preg_replace( '/^(?:\w+:\/\/|\/+)/', '', $url ); |
| 231 | $url = preg_replace( '/(?:\/+|\/*#.*?)$/', '', $url ); |
| 232 | return implode( '/', explode( '/', $url, $num ) ); |
| 233 | } |
| 234 | |
| 235 | private static function xpathQuote( string $s ): string { |
| 236 | // Ugly-but-functional escape mechanism for xpath query |
| 237 | $parts = explode( "'", $s ); |
| 238 | $parts = array_map( static function ( string $ss ) { |
| 239 | return "'$ss'"; |
| 240 | }, $parts ); |
| 241 | if ( count( $parts ) === 1 ) { |
| 242 | return $parts[0]; |
| 243 | } else { |
| 244 | return 'concat(' . implode( ',"\'",', $parts ) . ')'; |
| 245 | } |
| 246 | } |
| 247 | |
| 248 | /** |
| 249 | * Get descendants by ID. |
| 250 | * |
| 251 | * The PHP DOM doesn't provide this method for DOMElement, and the |
| 252 | * implementation in DOMDocument is broken. |
| 253 | * |
| 254 | * Further, the web spec only provides for returning a single element |
| 255 | * here. This function can support returning *all* of the matches for |
| 256 | * a given ID, if the underlying DOM implementation supports this. |
| 257 | * |
| 258 | * This is an *exclusive* query; that is, $context should never be included |
| 259 | * among the results. |
| 260 | * |
| 261 | * Although a `getElementsById` key can be passed in the options array |
| 262 | * to override the default implementation, for efficiency it is recommended |
| 263 | * that clients subclass ZestInst and override this entire method if |
| 264 | * they can provide an efficient id index. |
| 265 | * |
| 266 | * @param DOMDocument|DOMDocumentFragment|DOMElement $context |
| 267 | * The scoping root for the search |
| 268 | * @param string $id |
| 269 | * @param array $opts Additional match-context options (optional) |
| 270 | * @return array<DOMElement> A list of the elements with the given ID. When there are more |
| 271 | * than one, this method might return all of them or only the first one. |
| 272 | */ |
| 273 | public function getElementsById( $context, string $id, array $opts = [] ): array { |
| 274 | if ( is_callable( $opts['getElementsById'] ?? null ) ) { |
| 275 | // Third-party DOM implementation might provide a way to |
| 276 | // get multiple results for a given ID. |
| 277 | // Note that this must work for DocumentFragment and Element |
| 278 | // as well! |
| 279 | $func = $opts['getElementsById']; |
| 280 | return $func( $context, $id ); |
| 281 | } |
| 282 | // Neither PHP nor the web standards provide an DOMElement-scoped |
| 283 | // version of getElementById, so we can't call this directly on |
| 284 | // $context -- but that's okay because (1) IDs should be unique, and |
| 285 | // (2) we verify the scope of the returned element below |
| 286 | // anyway (to work around bugs with deleted-but-not-gc'ed |
| 287 | // nodes). |
| 288 | $doc = self::nodeIsDocument( $context ) ? |
| 289 | $context : $context->ownerDocument; |
| 290 | $r = $doc->getElementById( $id ); |
| 291 | // Note that $r could be null here because the |
| 292 | // DOMDocument hasn't had an "id attribute" set, even if the id |
| 293 | // exists in the document. See: |
| 294 | // http://php.net/manual/en/domdocument.getelementbyid.php |
| 295 | if ( $r !== null ) { |
| 296 | // Verify that this node is actually connected to the |
| 297 | // document (or to the context), since the element |
| 298 | // isn't removed from the index immediately when it |
| 299 | // is deleted. (Also PHP's call is not scoped.) |
| 300 | // (Note that scoped getElementsById is *exclusive* of $context, |
| 301 | // so we start this search at r's parent node.) |
| 302 | for ( $parent = $r->parentNode; $parent; $parent = $parent->parentNode ) { |
| 303 | if ( $parent === $context ) { |
| 304 | return [ $r ]; |
| 305 | } |
| 306 | } |
| 307 | // It's possible a deleted-but-still-indexed element was |
| 308 | // shadowing a later-added element, so we can't return |
| 309 | // null here directly; fallback to a full search. |
| 310 | } |
| 311 | if ( $this->isStandardsMode( $context, $opts ) ) { |
| 312 | // The workaround below only works (and is only necessary!) |
| 313 | // when this is a PHP-provided \DOMDocument. For 3rd-party |
| 314 | // DOM implementations, we assume that getElementById() was |
| 315 | // reliable. |
| 316 | // @phan-suppress-next-line PhanUndeclaredProperty |
| 317 | if ( $context->isConnected || $id === '' ) { |
| 318 | return []; |
| 319 | } |
| 320 | // For disconnected Elements and DocumentFragments, we need |
| 321 | // to do this the hard/slow way |
| 322 | $r = []; |
| 323 | foreach ( $this->getElementsByTagName( $context, '*', $opts ) as $el ) { |
| 324 | // Work around Phan 8.1 / 7.4 issues by telling Phan what |
| 325 | // the expected type is. Zest could be used with newer DOM |
| 326 | // implementations that could return null for missing attributes. |
| 327 | // See https://developer.mozilla.org/en-US/docs/Web/API/Element/getAttribute#non-existing_attributes |
| 328 | // But, Phan doesn't know that and only sees the native PHP |
| 329 | // implementation that returns '' for missing attributes. |
| 330 | $elId = $el->getAttribute( 'id' ); |
| 331 | '@phan-var ?string $elId'; |
| 332 | if ( $id === ( $elId ?? '' ) ) { |
| 333 | $r[] = $el; |
| 334 | } |
| 335 | } |
| 336 | return $r; |
| 337 | } |
| 338 | // Do an xpath search, which is still a full traversal of the tree |
| 339 | // (sigh) but 25% faster than traversing it wholly in PHP. |
| 340 | $xpath = new \DOMXPath( $doc ); |
| 341 | $query = './/*[@id=' . self::xpathQuote( $id ) . ']'; |
| 342 | if ( $context->nodeType === 11 ) { |
| 343 | // ugh, PHP dom extension workaround: nodes which are direct |
| 344 | // children of the DocumentFragment are not matched unless we |
| 345 | // use a ./ query in addition to the .// query. |
| 346 | $query = "./" . substr( $query, 3 ) . "|$query"; |
| 347 | } |
| 348 | return iterator_to_array( $xpath->query( $query, $context ) ); |
| 349 | } |
| 350 | |
| 351 | private function docFragHelper( $docFrag, string $sel, array $opts, callable $collectFunc ) { |
| 352 | $result = []; |
| 353 | for ( $n = $docFrag->firstChild; $n; $n = $n->nextSibling ) { |
| 354 | if ( $n->nodeType !== 1 ) { |
| 355 | continue; // Not an element |
| 356 | } |
| 357 | // See if $n itself should be included |
| 358 | if ( $this->matches( $n, $sel, $opts ) ) { |
| 359 | $result[] = $n; |
| 360 | } |
| 361 | // Now include all of $n's children |
| 362 | array_splice( $result, count( $result ), 0, $collectFunc( $n ) ); |
| 363 | } |
| 364 | return $result; |
| 365 | } |
| 366 | |
| 367 | /** |
| 368 | * Get descendants by tag name. |
| 369 | * The PHP DOM doesn't provide this method for DOMElement, and the |
| 370 | * implementation in DOMDocument has performance issues. |
| 371 | * |
| 372 | * This is an *exclusive* query; that is, $context should never be included |
| 373 | * among the results. |
| 374 | * |
| 375 | * Clients can subclass and override this to provide a more efficient |
| 376 | * implementation if one is available. |
| 377 | * |
| 378 | * @param DOMDocument|DOMDocumentFragment|DOMElement $context |
| 379 | * @param string $tagName |
| 380 | * @param array $opts Additional match-context options (optional) |
| 381 | * @return array<DOMElement> |
| 382 | */ |
| 383 | public function getElementsByTagName( $context, string $tagName, array $opts = [] ) { |
| 384 | if ( $context->nodeType === 11 /* DocumentFragment */ ) { |
| 385 | // DOM standards don't define getElementsByTagName on |
| 386 | // DocumentFragment, and XPath supports it but has a bug which |
| 387 | // omits root elements. So emulate in both these cases. |
| 388 | $selector = $tagName === '*' ? '*' : self::encodeid( $tagName ); |
| 389 | return $this->docFragHelper( |
| 390 | $context, $selector, $opts, |
| 391 | function ( $el ) use ( $tagName, $opts ): array { |
| 392 | return $this->getElementsByTagName( $el, $tagName, $opts ); |
| 393 | } |
| 394 | ); |
| 395 | } |
| 396 | if ( $this->isStandardsMode( $context, $opts ) ) { |
| 397 | // For third-party DOM implementations, just use native func. |
| 398 | return iterator_to_array( |
| 399 | $context->getElementsByTagName( $tagName ) |
| 400 | ); |
| 401 | } |
| 402 | // This *should* just be a call to PHP's `getElementByTagName` |
| 403 | // function *BUT* PHP's implementation is 100x slower than using |
| 404 | // XPath to get the same results (!) |
| 405 | |
| 406 | // XXX this assumes default PHP DOM implementation, which |
| 407 | // reports lowercase tag names in DOMNode->tagName (even though |
| 408 | // the DOM spec says it should report uppercase) |
| 409 | $tagName = strtolower( $tagName ); |
| 410 | |
| 411 | $doc = self::nodeIsDocument( $context ) ? |
| 412 | $context : $context->ownerDocument; |
| 413 | $xpath = new \DOMXPath( $doc ); |
| 414 | $ns = $doc->documentElement === null ? 'force use of local-name' : |
| 415 | $doc->documentElement->namespaceURI; |
| 416 | if ( $tagName === '*' ) { |
| 417 | $query = ".//*"; |
| 418 | } elseif ( $ns || !preg_match( '/^[_a-z][-.0-9_a-z]*$/S', $tagName ) ) { |
| 419 | $query = './/*[local-name()=' . self::xpathQuote( $tagName ) . ']'; |
| 420 | } else { |
| 421 | $query = ".//$tagName"; |
| 422 | } |
| 423 | return iterator_to_array( $xpath->query( $query, $context ) ); |
| 424 | } |
| 425 | |
| 426 | /** |
| 427 | * Clients can subclass and override this to provide a more efficient |
| 428 | * implementation if one is available. |
| 429 | * |
| 430 | * This is an *exclusive* query; that is, $context should never be included |
| 431 | * among the results. |
| 432 | * |
| 433 | * @param DOMDocument|DOMDocumentFragment|DOMElement $context |
| 434 | * @param string $className |
| 435 | * @param array $opts |
| 436 | * @return array<DOMElement> |
| 437 | */ |
| 438 | protected function getElementsByClassName( $context, string $className, $opts ) { |
| 439 | if ( $context->nodeType === 11 /* DocumentFragment */ ) { |
| 440 | // DOM standards don't define getElementsByClassName on |
| 441 | // DocumentFragment, and XPath supports it but has a bug which |
| 442 | // omits root elements. So emulate in both these cases. |
| 443 | return $this->docFragHelper( |
| 444 | $context, |
| 445 | // NOTE this only works when $className is a single class, |
| 446 | // but that's the only way we invoke it. |
| 447 | "." . self::encodeid( $className ), |
| 448 | $opts, |
| 449 | function ( $el ) use ( $className, $opts ): array { |
| 450 | return $this->getElementsByClassName( $el, $className, $opts ); |
| 451 | } |
| 452 | ); |
| 453 | } |
| 454 | if ( $this->isStandardsMode( $context, $opts ) ) { |
| 455 | // For third-party DOM implementations, just use native func. |
| 456 | return iterator_to_array( |
| 457 | // @phan-suppress-next-line PhanUndeclaredMethod |
| 458 | $context->getElementsByClassName( $className ) |
| 459 | ); |
| 460 | } |
| 461 | |
| 462 | // PHP doesn't have an implementation of this method; use XPath |
| 463 | // to quickly get results. (It would be faster still if there was an |
| 464 | // actual index, but this will be about 25% faster than doing the |
| 465 | // tree traversal all in PHP.) |
| 466 | $doc = self::nodeIsDocument( $context ) ? |
| 467 | $context : $context->ownerDocument; |
| 468 | $xpath = new \DOMXPath( $doc ); |
| 469 | $quotedClassName = self::xpathQuote( " $className " ); |
| 470 | $query = ".//*[contains(concat(' ', normalize-space(@class), ' '), $quotedClassName)]"; |
| 471 | return iterator_to_array( $xpath->query( $query, $context ) ); |
| 472 | } |
| 473 | |
| 474 | /** |
| 475 | * Handle `nth` Selectors |
| 476 | */ |
| 477 | private static function parseNth( string $param ): object { |
| 478 | $param = preg_replace( '/\s+/', '', $param ); |
| 479 | |
| 480 | if ( $param === 'even' ) { |
| 481 | $param = '2n+0'; |
| 482 | } elseif ( $param === 'odd' ) { |
| 483 | $param = '2n+1'; |
| 484 | } elseif ( strpos( $param, 'n' ) === false ) { |
| 485 | $param = '0n' . $param; |
| 486 | } |
| 487 | |
| 488 | preg_match( '/^([+-])?(\d+)?n([+-])?(\d+)?$/', $param, $cap, PREG_UNMATCHED_AS_NULL ); |
| 489 | |
| 490 | $group = intval( ( $cap[1] ?? '' ) . ( $cap[2] ?? '1' ), 10 ); |
| 491 | $offset = intval( ( $cap[3] ?? '' ) . ( $cap[4] ?? '0' ), 10 ); |
| 492 | return (object)[ |
| 493 | 'group' => $group, |
| 494 | 'offset' => $offset, |
| 495 | ]; |
| 496 | } |
| 497 | |
| 498 | /** |
| 499 | * @param string $param |
| 500 | * @param callable(DOMNode,DOMNode,array):bool $test |
| 501 | * @param bool $last |
| 502 | * @return callable(DOMNode,array):bool |
| 503 | */ |
| 504 | private static function nth( string $param, callable $test, bool $last ): callable { |
| 505 | $param = self::parseNth( $param ); |
| 506 | $group = $param->group; |
| 507 | $offset = $param->offset; |
| 508 | $find = ( !$last ) ? [ self::class, 'child' ] : [ self::class, 'lastChild' ]; |
| 509 | $advance = ( !$last ) ? [ self::class, 'next' ] : [ self::class, 'prev' ]; |
| 510 | return function ( $el, array $opts ) use ( $find, $test, $offset, $group, $advance ): bool { |
| 511 | if ( !self::parentIsElement( $el ) ) { |
| 512 | return false; |
| 513 | } |
| 514 | |
| 515 | $rel = call_user_func( $find, $el->parentNode ); |
| 516 | $pos = 0; |
| 517 | |
| 518 | while ( $rel ) { |
| 519 | if ( call_user_func( $test, $rel, $el, $opts ) ) { |
| 520 | $pos++; |
| 521 | } |
| 522 | if ( $rel === $el ) { |
| 523 | $pos -= $offset; |
| 524 | return ( $group && $pos ) |
| 525 | ? ( $pos % $group ) === 0 && ( ( $pos < 0 ) === ( $group < 0 ) ) |
| 526 | : !$pos; |
| 527 | } |
| 528 | $rel = call_user_func( $advance, $rel ); |
| 529 | } |
| 530 | return false; |
| 531 | }; |
| 532 | } |
| 533 | |
| 534 | /** |
| 535 | * Simple Selectors which take no arguments. |
| 536 | * @var array<string,(callable(DOMNode,array):bool)> |
| 537 | */ |
| 538 | private $selectors0; |
| 539 | |
| 540 | /** |
| 541 | * Simple Selectors which take one argument. |
| 542 | * @var array<string,(callable(string,ZestInst):(callable(DOMNode,array):bool))> |
| 543 | */ |
| 544 | private $selectors1; |
| 545 | |
| 546 | /** |
| 547 | * Add a custom selector that takes no parameters. |
| 548 | * @param string $key Name of the selector |
| 549 | * @param callable(DOMNode,array):bool $func |
| 550 | * The selector match function |
| 551 | */ |
| 552 | public function addSelector0( string $key, callable $func ) { |
| 553 | $this->selectors0[$key] = $func; |
| 554 | } |
| 555 | |
| 556 | /** |
| 557 | * Add a custom selector that takes 1 parameter, which is passed as a |
| 558 | * string. |
| 559 | * @param string $key Name of the selector |
| 560 | * @param callable(string,ZestInst):(callable(DOMNode,array):bool)|callable(string):(callable(DOMNode,array):bool) $func |
| 561 | * The selector match function |
| 562 | */ |
| 563 | public function addSelector1( string $key, callable $func ) { |
| 564 | $this->selectors1[$key] = $func; |
| 565 | } |
| 566 | |
| 567 | private function initSelectors() { |
| 568 | // Careful: this method is only called once, on the singleton |
| 569 | // ZestInst, which all child ZestInst instances inherit their |
| 570 | // default selector lists from. But as a result $this is |
| 571 | // always the singleton; be sure to use the $self argument (for |
| 572 | // selector1) or $opts['this'] (for selector0) to access the |
| 573 | // dynamically-bound $this. |
| 574 | |
| 575 | $this->addSelector0( '*', static function ( $el, $opts ): bool { |
| 576 | return true; |
| 577 | } ); |
| 578 | $this->addSelector1( 'type', static function ( string $type ): callable { |
| 579 | $type = strtolower( $type ); |
| 580 | return static function ( $el, $opts ) use ( $type ): bool { |
| 581 | return strtolower( $el->nodeName ) === $type; |
| 582 | }; |
| 583 | } ); |
| 584 | $this->addSelector1( 'typeNoNS', static function ( string $type ): callable { |
| 585 | $type = strtolower( $type ); |
| 586 | return static function ( $el, $opts ) use ( $type ): bool { |
| 587 | return ( $el->namespaceURI ?? '' ) === '' && |
| 588 | strtolower( $el->nodeName ) === $type; |
| 589 | }; |
| 590 | } ); |
| 591 | $this->addSelector0( ':first-child', function ( $el, $opts ): bool { |
| 592 | return !self::prev( $el ) && self::parentIsElement( $el ); |
| 593 | } ); |
| 594 | $this->addSelector0( ':last-child', function ( $el, $opts ): bool { |
| 595 | return !self::next( $el ) && self::parentIsElement( $el ); |
| 596 | } ); |
| 597 | $this->addSelector0( ':only-child', function ( $el, $opts ): bool { |
| 598 | return !self::prev( $el ) && !self::next( $el ) |
| 599 | && self::parentIsElement( $el ); |
| 600 | } ); |
| 601 | $this->addSelector1( ':nth-child', function ( string $param ): callable { |
| 602 | return self::nth( $param, static function ( $rel, $el, $opts ): bool { |
| 603 | return true; |
| 604 | }, false /* last */ ); |
| 605 | } ); |
| 606 | $this->addSelector1( ':nth-last-child', function ( string $param ): callable { |
| 607 | return self::nth( $param, static function ( $rel, $el, $opts ): bool { |
| 608 | return true; |
| 609 | }, true /* last */ ); |
| 610 | } ); |
| 611 | $this->addSelector0( ':root', static function ( $el, $opts ): bool { |
| 612 | return $el->ownerDocument->documentElement === $el; |
| 613 | } ); |
| 614 | $this->addSelector0( ':empty', static function ( $el, $opts ): bool { |
| 615 | return !$el->firstChild; |
| 616 | } ); |
| 617 | $this->addSelector1( ':not', static function ( string $sel, ZestInst $self ) { |
| 618 | $test = $self->compileGroup( $sel ); |
| 619 | return static function ( $el, $opts ) use ( $test ): bool { |
| 620 | return !call_user_func( $test, $el, $opts ); |
| 621 | }; |
| 622 | } ); |
| 623 | $this->addSelector0( ':first-of-type', function ( $el, $opts ): bool { |
| 624 | if ( !self::parentIsElement( $el ) ) { |
| 625 | return false; |
| 626 | } |
| 627 | $type = $el->nodeName; |
| 628 | while ( $el = self::prev( $el ) ) { |
| 629 | if ( $el->nodeName === $type ) { |
| 630 | return false; |
| 631 | } |
| 632 | } |
| 633 | return true; |
| 634 | } ); |
| 635 | $this->addSelector0( ':last-of-type', function ( $el, $opts ): bool { |
| 636 | if ( !self::parentIsElement( $el ) ) { |
| 637 | return false; |
| 638 | } |
| 639 | $type = $el->nodeName; |
| 640 | while ( $el = self::next( $el ) ) { |
| 641 | if ( $el->nodeName === $type ) { |
| 642 | return false; |
| 643 | } |
| 644 | } |
| 645 | return true; |
| 646 | } ); |
| 647 | $this->addSelector0( ':only-of-type', static function ( $el, $opts ): bool { |
| 648 | $self = $opts['this']; |
| 649 | return $self->selectors0[ ':first-of-type' ]( $el, $opts ) && |
| 650 | $self->selectors0[ ':last-of-type' ]( $el, $opts ); |
| 651 | } ); |
| 652 | $makeNthOfType = static function ( bool $last ) { |
| 653 | return function ( string $param ) use ( $last ): callable { |
| 654 | return self::nth( $param, static function ( $rel, $el, $opts ): bool { |
| 655 | return $rel->nodeName === $el->nodeName; |
| 656 | }, $last ); |
| 657 | }; |
| 658 | }; |
| 659 | $this->addSelector1( ':nth-of-type', $makeNthOfType( false ) ); |
| 660 | $this->addSelector1( ':nth-last-of-type', $makeNthOfType( true ) ); |
| 661 | /** @suppress PhanUndeclaredProperty not defined in PHP DOM */ |
| 662 | $this->addSelector0( ':checked', static function ( $el, $opts ): bool { |
| 663 | '@phan-var DOMElement $el'; |
| 664 | $self = $opts['this']; |
| 665 | if ( $self->isStandardsMode( $el, $opts ) ) { |
| 666 | // These properties don't exist in the PHP DOM, and in fact |
| 667 | // they are supposed to reflect the *dynamic* state of the |
| 668 | // widget, not the 'default' state (which is given by the |
| 669 | // attribute value) |
| 670 | if ( isset( $el->checked ) || isset( $el->selected ) ) { |
| 671 | return ( isset( $el->checked ) && $el->checked ) || |
| 672 | ( isset( $el->selected ) && $el->selected ); |
| 673 | } |
| 674 | } |
| 675 | return $el->hasAttribute( 'checked' ) || $el->hasAttribute( 'selected' ); |
| 676 | } ); |
| 677 | $this->addSelector0( ':indeterminate', static function ( $el, $opts ): bool { |
| 678 | $self = $opts['this']; |
| 679 | return !$self->selectors0[ ':checked' ]( $el, $opts ); |
| 680 | } ); |
| 681 | /** @suppress PhanUndeclaredProperty not defined in PHP DOM */ |
| 682 | $this->addSelector0( ':enabled', static function ( $el, $opts ): bool { |
| 683 | '@phan-var DOMElement $el'; |
| 684 | $self = $opts['this']; |
| 685 | if ( $self->isStandardsMode( $el, $opts ) && isset( $el->type ) ) { |
| 686 | $type = $el->type; // this does case normalization in spec |
| 687 | } else { |
| 688 | $type = $el->getAttribute( 'type' ); |
| 689 | } |
| 690 | return !$el->hasAttribute( 'disabled' ) && $type !== 'hidden'; |
| 691 | } ); |
| 692 | $this->addSelector0( ':disabled', static function ( $el, $opts ): bool { |
| 693 | '@phan-var DOMElement $el'; |
| 694 | return $el->hasAttribute( 'disabled' ); |
| 695 | } ); |
| 696 | /* |
| 697 | $this->addSelector0( ':target', function ( $el ) use ( &$window ) { |
| 698 | return $el->id === $window->location->hash->substring( 1 ); |
| 699 | }); |
| 700 | $this->addSelector0( ':focus', function ( $el ) { |
| 701 | return $el === $el->ownerDocument->activeElement; |
| 702 | }); |
| 703 | */ |
| 704 | $this->addSelector1( ':is', static function ( string $sel, ZestInst $self ): callable { |
| 705 | return $self->compileGroup( $sel ); |
| 706 | } ); |
| 707 | // :matches is an older name for :is; see |
| 708 | // https://github.com/w3c/csswg-drafts/issues/3258 |
| 709 | $this->addSelector1( ':matches', static function ( string $sel, ZestInst $self ): callable { |
| 710 | return $self->selectors1[ ':is' ]( $sel, $self ); |
| 711 | } ); |
| 712 | $makeNthMatch = static function ( bool $last ) { |
| 713 | return function ( string $param, ZestInst $self ) use ( $last ): callable { |
| 714 | $args = preg_split( '/\s*,\s*/', $param ); |
| 715 | $arg = array_shift( $args ); |
| 716 | $test = $self->compileGroup( implode( ',', $args ) ); |
| 717 | |
| 718 | return self::nth( $arg, static function ( $rel, $el, $opts ) use ( $test ): bool { |
| 719 | return call_user_func( $test, $el, $opts ); |
| 720 | }, $last ); |
| 721 | }; |
| 722 | }; |
| 723 | $this->addSelector1( ':nth-match', $makeNthMatch( false ) ); |
| 724 | $this->addSelector1( ':nth-last-match', $makeNthMatch( true ) ); |
| 725 | /* |
| 726 | $this->addSelector0( ':links-here', function ( $el ) use ( &$window ) { |
| 727 | return $el . '' === $window->location . ''; |
| 728 | }); |
| 729 | */ |
| 730 | $this->addSelector1( ':lang', static function ( string $param ): callable { |
| 731 | return static function ( $el, $opts ) use ( $param ): bool { |
| 732 | while ( $el ) { |
| 733 | if ( $el->nodeType === 1 /* Element */ ) { |
| 734 | '@phan-var DOMElement $el'; |
| 735 | // PHP DOM doesn't have 'lang' property |
| 736 | $lang = $el->getAttribute( 'lang' ); |
| 737 | if ( $lang ) { |
| 738 | return strpos( $lang, $param ) === 0; |
| 739 | } |
| 740 | } |
| 741 | $el = $el->parentNode; |
| 742 | } |
| 743 | return false; |
| 744 | }; |
| 745 | } ); |
| 746 | $this->addSelector1( ':dir', static function ( string $param ): callable { |
| 747 | return static function ( $el, $opts ) use ( $param ): bool { |
| 748 | while ( $el ) { |
| 749 | if ( $el->nodeType === 1 /* Element */ ) { |
| 750 | '@phan-var DOMElement $el'; |
| 751 | $dir = $el->getAttribute( 'dir' ); |
| 752 | if ( $dir ) { |
| 753 | return $dir === $param; |
| 754 | } |
| 755 | } |
| 756 | $el = $el->parentNode; |
| 757 | } |
| 758 | return false; |
| 759 | }; |
| 760 | } ); |
| 761 | $this->addSelector0( ':scope', static function ( $el, $opts ): bool { |
| 762 | $self = $opts['this']; |
| 763 | $scope = $opts['scope'] ?? null; |
| 764 | if ( $scope !== null && $scope->nodeType === 1 ) { |
| 765 | return $el === $scope; |
| 766 | } |
| 767 | // If the scoping root is missing or not an element, then :scope |
| 768 | // should be a synonym for :root |
| 769 | return $self->selectors0[ ':root' ]( $el, $opts ); |
| 770 | } ); |
| 771 | /* |
| 772 | $this->addSelector0( ':any-link', function ( $el ):bool { |
| 773 | return gettype( $el->href ) === 'string'; |
| 774 | }); |
| 775 | $this->addSelector( ':local-link', function ( $el ) use ( &$window ) { |
| 776 | if ( $el->nodeName ) { |
| 777 | return $el->href && $el->host === $window->location->host; |
| 778 | } |
| 779 | // XXX this is really selector1 not selector0 |
| 780 | $param = +$el + 1; |
| 781 | return function ( $el ) use ( &$window, $param ) { |
| 782 | if ( !$el->href ) { return; } |
| 783 | |
| 784 | $url = $window->location . ''; |
| 785 | $href = $el . ''; |
| 786 | |
| 787 | return self::truncateUrl( $url, $param ) === self::truncateUrl( $href, $param ); |
| 788 | }; |
| 789 | }); |
| 790 | $this->addSelector0( ':default', function ( $el ):bool { |
| 791 | return !!$el->defaultSelected; |
| 792 | }); |
| 793 | $this->addSelector0( ':valid', function ( $el ):bool { |
| 794 | return $el->willValidate || ( $el->validity && $el->validity->valid ); |
| 795 | }); |
| 796 | */ |
| 797 | $this->addSelector0( ':invalid', static function ( $el, $opts ): bool { |
| 798 | $self = $opts['this']; |
| 799 | return !$self->selectors0[ ':valid' ]( $el, $opts ); |
| 800 | } ); |
| 801 | /* |
| 802 | $this->addSelector0( ':in-range', function ( $el ):bool { |
| 803 | return $el->value > $el->min && $el->value <= $el->max; |
| 804 | }); |
| 805 | */ |
| 806 | $this->addSelector0( ':out-of-range', static function ( $el, $opts ): bool { |
| 807 | $self = $opts['this']; |
| 808 | return !$self->selectors0[ ':in-range' ]( $el, $opts ); |
| 809 | } ); |
| 810 | $this->addSelector0( ':required', static function ( $el, $opts ): bool { |
| 811 | '@phan-var DOMElement $el'; |
| 812 | return $el->hasAttribute( 'required' ); |
| 813 | } ); |
| 814 | $this->addSelector0( ':optional', static function ( $el, $opts ): bool { |
| 815 | $self = $opts['this']; |
| 816 | return !$self->selectors0[ ':required' ]( $el, $opts ); |
| 817 | } ); |
| 818 | $this->addSelector0( ':read-only', static function ( $el, $opts ): bool { |
| 819 | '@phan-var DOMElement $el'; |
| 820 | if ( $el->hasAttribute( 'readOnly' ) ) { |
| 821 | return true; |
| 822 | } |
| 823 | |
| 824 | $attr = $el->getAttribute( 'contenteditable' ); |
| 825 | $name = strtolower( $el->nodeName ); |
| 826 | |
| 827 | $name = $name !== 'input' && $name !== 'textarea'; |
| 828 | |
| 829 | return ( $name || $el->hasAttribute( 'disabled' ) ) && $attr == null; |
| 830 | } ); |
| 831 | $this->addSelector0( ':read-write', static function ( $el, $opts ): bool { |
| 832 | $self = $opts['this']; |
| 833 | return !$self->selectors0[ ':read-only' ]( $el, $opts ); |
| 834 | } ); |
| 835 | foreach ( [ |
| 836 | ':hover', |
| 837 | ':active', |
| 838 | ':link', |
| 839 | ':visited', |
| 840 | ':column', |
| 841 | ':nth-column', |
| 842 | ':nth-last-column', |
| 843 | ':current', |
| 844 | ':past', |
| 845 | ':future', |
| 846 | ] as $selector ) { |
| 847 | $this->addSelector0( |
| 848 | $selector, |
| 849 | /** |
| 850 | * @param DOMNode $el |
| 851 | * @param array $opts |
| 852 | * @return never |
| 853 | */ |
| 854 | static function ( $el, $opts ) use ( $selector ): bool { |
| 855 | $self = $opts['this']; |
| 856 | throw $self->newBadSelectorException( $selector . ' is not supported.' ); |
| 857 | } |
| 858 | ); |
| 859 | } |
| 860 | // Non-standard, for compatibility purposes. |
| 861 | $this->addSelector1( ':contains', static function ( string $param ): callable { |
| 862 | return static function ( $el ) use ( $param ): bool { |
| 863 | $text = $el->textContent; |
| 864 | return strpos( $text, $param ) !== false; |
| 865 | }; |
| 866 | } ); |
| 867 | $this->addSelector1( ':has', static function ( string $param ): callable { |
| 868 | return static function ( $el, array $opts ) use ( $param ): bool { |
| 869 | '@phan-var DOMElement $el'; |
| 870 | $self = $opts['this']; |
| 871 | return count( $self->find( $param, $el, $opts ) ) > 0; |
| 872 | }; |
| 873 | } ); |
| 874 | // Potentially add more pseudo selectors for |
| 875 | // compatibility with sizzle and most other |
| 876 | // selector engines (?). |
| 877 | } |
| 878 | |
| 879 | /** @return callable(DOMNode,array):bool */ |
| 880 | private function selectorsAttr( string $key, string $op, string $val, bool $i ): callable { |
| 881 | $op = $this->operators[ $op ]; |
| 882 | return static function ( $el, $opts ) use ( $key, $i, $op, $val ): bool { |
| 883 | /* XXX: the below all assumes a more complete PHP DOM than we have |
| 884 | switch ( $key ) { |
| 885 | #case 'for': |
| 886 | # $attr = $el->htmlFor; // Not supported in PHP DOM |
| 887 | # break; |
| 888 | case 'class': |
| 889 | // PHP DOM doesn't support $el->className |
| 890 | // className is '' when non-existent |
| 891 | // getAttribute('class') is null |
| 892 | if ($el->hasAttributes() && $el->hasAttribute( 'class' ) ) { |
| 893 | $attr = $el->getAttribute( 'class' ); |
| 894 | } else { |
| 895 | $attr = null; |
| 896 | } |
| 897 | break; |
| 898 | case 'href': |
| 899 | case 'src': |
| 900 | $attr = $el->getAttribute( $key, 2 ); |
| 901 | break; |
| 902 | case 'title': |
| 903 | // getAttribute('title') can be '' when non-existent sometimes? |
| 904 | if ($el->hasAttribute('title')) { |
| 905 | $attr = $el->getAttribute( 'title' ); |
| 906 | } else { |
| 907 | $attr = null; |
| 908 | } |
| 909 | break; |
| 910 | // careful with attributes with special getter functions |
| 911 | case 'id': |
| 912 | case 'lang': |
| 913 | case 'dir': |
| 914 | case 'accessKey': |
| 915 | case 'hidden': |
| 916 | case 'tabIndex': |
| 917 | case 'style': |
| 918 | if ( $el->getAttribute ) { |
| 919 | $attr = $el->getAttribute( $key ); |
| 920 | break; |
| 921 | } |
| 922 | // falls through |
| 923 | default: |
| 924 | if ( $el->hasAttribute && !$el->hasAttribute( $key ) ) { |
| 925 | break; |
| 926 | } |
| 927 | $attr = ( $el[ $key ] != null ) ? |
| 928 | $el[ $key ] : |
| 929 | $el->getAttribute && $el->getAttribute( $key ); |
| 930 | break; |
| 931 | } |
| 932 | */ |
| 933 | // This is our simple PHP DOM version |
| 934 | '@phan-var DOMElement $el'; |
| 935 | if ( $el->hasAttributes() && $el->hasAttribute( $key ) ) { |
| 936 | $attr = $el->getAttribute( $key ); |
| 937 | } else { |
| 938 | return false; |
| 939 | } |
| 940 | // End simple PHP DOM version |
| 941 | if ( $i ) { |
| 942 | $attr = strtolower( $attr ); |
| 943 | $val = strtolower( $val ); |
| 944 | } |
| 945 | return call_user_func( $op, $attr, $val ); |
| 946 | }; |
| 947 | } |
| 948 | |
| 949 | /** |
| 950 | * Attribute Operators |
| 951 | * @var array<string,(callable(string,string):bool)> |
| 952 | */ |
| 953 | private $operators; |
| 954 | |
| 955 | /** |
| 956 | * Add a custom operator |
| 957 | * @param string $key Name of the operator |
| 958 | * @param callable(string,string):bool $func |
| 959 | * The operator match function |
| 960 | */ |
| 961 | public function addOperator( string $key, callable $func ) { |
| 962 | $this->operators[$key] = $func; |
| 963 | } |
| 964 | |
| 965 | private function initOperators() { |
| 966 | // Be careful: $this always points to the singleton ZestInst |
| 967 | $this->addOperator( '-', static function ( string $attr, string $val ): bool { |
| 968 | return true; |
| 969 | } ); |
| 970 | $this->addOperator( '=', static function ( string $attr, string $val ): bool { |
| 971 | return $attr === $val; |
| 972 | } ); |
| 973 | $this->addOperator( '*=', static function ( string $attr, string $val ): bool { |
| 974 | return strpos( $attr, $val ) !== false; |
| 975 | } ); |
| 976 | $this->addOperator( '~=', static function ( string $attr, string $val ): bool { |
| 977 | // https://drafts.csswg.org/selectors-4/#attribute-representation |
| 978 | // If "val" contains whitespace, it will never represent |
| 979 | // anything (since the words are separated by spaces) |
| 980 | if ( strcspn( $val, " \t\r\n\f" ) !== strlen( $val ) ) { |
| 981 | return false; |
| 982 | } |
| 983 | // Also if "val" is the empty string, it will never |
| 984 | // represent anything. |
| 985 | if ( strlen( $val ) === 0 ) { |
| 986 | return false; |
| 987 | } |
| 988 | $attrLen = strlen( $attr ); |
| 989 | $valLen = strlen( $val ); |
| 990 | for ( $s = 0; $s < $attrLen; $s = $i + 1 ) { |
| 991 | $i = strpos( $attr, $val, $s ); |
| 992 | if ( $i === false ) { |
| 993 | return false; |
| 994 | } |
| 995 | $j = $i + $valLen; |
| 996 | $f = ( $i === 0 ) ? ' ' : $attr[ $i - 1 ]; |
| 997 | $l = ( $j >= $attrLen ) ? ' ' : $attr[ $j ]; |
| 998 | $f = strtr( $f, "\t\r\n\f", " " ); |
| 999 | $l = strtr( $l, "\t\r\n\f", " " ); |
| 1000 | if ( $f === ' ' && $l === ' ' ) { |
| 1001 | return true; |
| 1002 | } |
| 1003 | } |
| 1004 | return false; |
| 1005 | } ); |
| 1006 | $this->addOperator( '|=', static function ( string $attr, string $val ): bool { |
| 1007 | $i = strpos( $attr, $val ); |
| 1008 | if ( $i !== 0 ) { |
| 1009 | return false; |
| 1010 | } |
| 1011 | $j = $i + strlen( $val ); |
| 1012 | if ( $j >= strlen( $attr ) ) { |
| 1013 | return true; |
| 1014 | } |
| 1015 | $l = $attr[ $j ]; |
| 1016 | return $l === '-'; |
| 1017 | } ); |
| 1018 | $this->addOperator( '^=', static function ( string $attr, string $val ): bool { |
| 1019 | return strpos( $attr, $val ) === 0; |
| 1020 | } ); |
| 1021 | $this->addOperator( '$=', static function ( string $attr, string $val ): bool { |
| 1022 | $i = strrpos( $attr, $val ); |
| 1023 | return $i !== false && $i + strlen( $val ) === strlen( $attr ); |
| 1024 | } ); |
| 1025 | // non-standard |
| 1026 | $this->addOperator( '!=', static function ( string $attr, string $val ): bool { |
| 1027 | return $attr !== $val; |
| 1028 | } ); |
| 1029 | } |
| 1030 | |
| 1031 | /** |
| 1032 | * Combinator Logic |
| 1033 | * @var array<string,(callable(callable(DOMNode,array):bool):(callable(DOMNode,array):(?DOMNode)))> |
| 1034 | */ |
| 1035 | private $combinators; |
| 1036 | |
| 1037 | /** |
| 1038 | * Add a custom combinator |
| 1039 | * @param string $key Name of the combinator |
| 1040 | * @param callable(callable(DOMNode,array):bool):(callable(DOMNode,array):(?DOMNode)) $func |
| 1041 | * The combinator match function |
| 1042 | */ |
| 1043 | public function addCombinator( string $key, callable $func ) { |
| 1044 | $this->combinators[$key] = $func; |
| 1045 | } |
| 1046 | |
| 1047 | private function initCombinators() { |
| 1048 | // Be careful: $this always points to the singleton ZestInst |
| 1049 | $this->addCombinator( ' ', static function ( callable $test ): callable { |
| 1050 | return static function ( $el, $opts ) use ( $test ) { |
| 1051 | while ( $el = $el->parentNode ) { |
| 1052 | if ( $el->nodeType === 1 && call_user_func( $test, $el, $opts ) ) { |
| 1053 | return $el; |
| 1054 | } |
| 1055 | } |
| 1056 | return null; |
| 1057 | }; |
| 1058 | } ); |
| 1059 | $this->addCombinator( '>', static function ( callable $test ): callable { |
| 1060 | return static function ( $el, $opts ) use ( $test ) { |
| 1061 | if ( $el = $el->parentNode ) { |
| 1062 | if ( $el->nodeType === 1 && call_user_func( $test, $el, $opts ) ) { |
| 1063 | return $el; |
| 1064 | } |
| 1065 | } |
| 1066 | return null; |
| 1067 | }; |
| 1068 | } ); |
| 1069 | $this->addCombinator( '+', function ( callable $test ): callable { |
| 1070 | return function ( $el, $opts ) use ( $test ) { |
| 1071 | if ( $el = self::prev( $el ) ) { |
| 1072 | if ( call_user_func( $test, $el, $opts ) ) { |
| 1073 | return $el; |
| 1074 | } |
| 1075 | } |
| 1076 | return null; |
| 1077 | }; |
| 1078 | } ); |
| 1079 | $this->addCombinator( '~', function ( callable $test ): callable { |
| 1080 | return function ( $el, $opts ) use ( $test ) { |
| 1081 | while ( $el = self::prev( $el ) ) { |
| 1082 | if ( call_user_func( $test, $el, $opts ) ) { |
| 1083 | return $el; |
| 1084 | } |
| 1085 | } |
| 1086 | return null; |
| 1087 | }; |
| 1088 | } ); |
| 1089 | $this->addCombinator( 'noop', static function ( callable $test ): callable { |
| 1090 | return static function ( $el, $opts ) use ( $test ) { |
| 1091 | if ( call_user_func( $test, $el, $opts ) ) { |
| 1092 | return $el; |
| 1093 | } |
| 1094 | return null; |
| 1095 | }; |
| 1096 | } ); |
| 1097 | } |
| 1098 | |
| 1099 | /** |
| 1100 | * @param callable(DOMNode,array):bool $test |
| 1101 | * @param string $name |
| 1102 | * @return ZestFunc |
| 1103 | */ |
| 1104 | private function makeRef( callable $test, string $name ): ZestFunc { |
| 1105 | $node = null; |
| 1106 | $ref = new ZestFunc( function ( $el, $opts ) use ( &$node, &$ref ): bool { |
| 1107 | $doc = $el->ownerDocument; |
| 1108 | $nodes = $this->getElementsByTagName( $doc, '*', $opts ); |
| 1109 | $i = count( $nodes ); |
| 1110 | |
| 1111 | while ( $i-- ) { |
| 1112 | $node = $nodes[$i]; |
| 1113 | if ( call_user_func( $ref->test->func, $el, $opts ) ) { |
| 1114 | $node = null; |
| 1115 | return true; |
| 1116 | } |
| 1117 | } |
| 1118 | |
| 1119 | $node = null; |
| 1120 | return false; |
| 1121 | } ); |
| 1122 | |
| 1123 | $ref->combinator = static function ( $el, $opts ) use ( &$node, $name, $test ) { |
| 1124 | if ( !$node || $node->nodeType !== 1 /* Element */ ) { |
| 1125 | return null; |
| 1126 | } |
| 1127 | |
| 1128 | $attr = $node->getAttribute( $name ) ?? ''; |
| 1129 | if ( $attr !== '' && $attr[ 0 ] === '#' ) { |
| 1130 | $attr = substr( $attr, 1 ); |
| 1131 | } |
| 1132 | |
| 1133 | $id = $node->getAttribute( 'id' ) ?? ''; |
| 1134 | if ( $attr === $id && call_user_func( $test, $node, $opts ) ) { |
| 1135 | return $node; |
| 1136 | } |
| 1137 | return null; |
| 1138 | }; |
| 1139 | |
| 1140 | return $ref; |
| 1141 | } |
| 1142 | |
| 1143 | /** |
| 1144 | * Grammar |
| 1145 | */ |
| 1146 | |
| 1147 | /** @var \stdClass */ |
| 1148 | private static $rules; |
| 1149 | |
| 1150 | public static function initRules() { |
| 1151 | self::$rules = (object)[ |
| 1152 | 'escape' => '/\\\(?:[^0-9A-Fa-f\r\n]|[0-9A-Fa-f]{1,6}[\r\n\t ]?)/', |
| 1153 | 'str_escape' => '/(escape)|\\\(\n|\r\n?|\f)/', |
| 1154 | 'nonascii' => '/[\x{00A0}-\x{FFFF}]/', |
| 1155 | 'cssid' => '/(?:(?!-?[0-9])(?:escape|nonascii|[-_a-zA-Z0-9])+)/', |
| 1156 | 'qname' => '/^ *((?:\*?\|)?cssid|\*)/', |
| 1157 | 'simple' => '/^(?:([.#]cssid)|pseudo|attr)/', |
| 1158 | 'ref' => '/^ *\/(cssid)\/ */', |
| 1159 | 'combinator' => '/^(?: +([^ \w*.#\\\]) +|( )+|([^ \w*.#\\\]))(?! *$)/', |
| 1160 | 'attr' => '/^\[(cssid)(?:([^\w]?=)(inside))?\]/', |
| 1161 | 'pseudo' => '/^(:cssid)(?:\((inside)\))?/', |
| 1162 | 'inside' => "/(?:\"(?:\\\\\"|[^\"])*\"|'(?:\\\\'|[^'])*'|<[^\"'>]*>|\\\\[\"'>]|[^\"'>])*/", |
| 1163 | 'ident' => '/^(cssid)$/', |
| 1164 | ]; |
| 1165 | self::$rules->cssid = self::replace( self::$rules->cssid, 'nonascii', self::$rules->nonascii ); |
| 1166 | self::$rules->cssid = self::replace( self::$rules->cssid, 'escape', self::$rules->escape ); |
| 1167 | self::$rules->qname = self::replace( self::$rules->qname, 'cssid', self::$rules->cssid ); |
| 1168 | self::$rules->simple = self::replace( self::$rules->simple, 'cssid', self::$rules->cssid ); |
| 1169 | self::$rules->ref = self::replace( self::$rules->ref, 'cssid', self::$rules->cssid ); |
| 1170 | self::$rules->attr = self::replace( self::$rules->attr, 'cssid', self::$rules->cssid ); |
| 1171 | self::$rules->pseudo = self::replace( self::$rules->pseudo, 'cssid', self::$rules->cssid ); |
| 1172 | self::$rules->inside = self::replace( self::$rules->inside, "[^\"'>]*", self::$rules->inside ); |
| 1173 | self::$rules->attr = self::replace( self::$rules->attr, 'inside', self::makeInside( '\[', '\]' ) ); |
| 1174 | self::$rules->pseudo = self::replace( self::$rules->pseudo, 'inside', self::makeInside( '\(', '\)' ) ); |
| 1175 | self::$rules->simple = self::replace( self::$rules->simple, 'pseudo', self::$rules->pseudo ); |
| 1176 | self::$rules->simple = self::replace( self::$rules->simple, 'attr', self::$rules->attr ); |
| 1177 | self::$rules->ident = self::replace( self::$rules->ident, 'cssid', self::$rules->cssid ); |
| 1178 | self::$rules->str_escape = self::replace( self::$rules->str_escape, 'escape', self::$rules->escape ); |
| 1179 | } |
| 1180 | |
| 1181 | /** |
| 1182 | * Compiling |
| 1183 | */ |
| 1184 | |
| 1185 | private function compile( string $sel ): ZestFunc { |
| 1186 | if ( !isset( $this->compileCache[$sel] ) ) { |
| 1187 | $this->compileCache[$sel] = $this->doCompile( $sel ); |
| 1188 | } |
| 1189 | return $this->compileCache[$sel]; |
| 1190 | } |
| 1191 | |
| 1192 | private function doCompile( string $sel ): ZestFunc { |
| 1193 | $sel = preg_replace( '/^\s+|\s+$/', '', $sel ); |
| 1194 | $test = null; |
| 1195 | $filter = []; |
| 1196 | $buff = []; |
| 1197 | $subject = null; |
| 1198 | $qname = null; |
| 1199 | $cap = null; |
| 1200 | $op = null; |
| 1201 | $ref = null; |
| 1202 | |
| 1203 | while ( $sel ) { |
| 1204 | if ( preg_match( self::$rules->qname, $sel, $cap ) ) { |
| 1205 | $sel = substr( $sel, strlen( $cap[0] ) ); |
| 1206 | $qname = self::decodeid( $cap[ 1 ] ); |
| 1207 | $buff[] = $this->tokQname( $qname ); |
| 1208 | // strip off *| or | prefix |
| 1209 | if ( substr( $qname, 0, 1 ) === '|' ) { |
| 1210 | $qname = substr( $qname, 1 ); |
| 1211 | } elseif ( substr( $qname, 0, 2 ) === '*|' ) { |
| 1212 | $qname = substr( $qname, 2 ); |
| 1213 | } |
| 1214 | } elseif ( preg_match( self::$rules->simple, $sel, $cap, PREG_UNMATCHED_AS_NULL ) ) { |
| 1215 | $sel = substr( $sel, strlen( $cap[0] ) ); |
| 1216 | $qname = '*'; |
| 1217 | $buff[] = $this->tokQname( $qname ); |
| 1218 | $buff[] = $this->tok( $cap ); |
| 1219 | } else { |
| 1220 | throw $this->newBadSelectorException( 'Invalid selector.' ); |
| 1221 | } |
| 1222 | |
| 1223 | while ( preg_match( self::$rules->simple, $sel, $cap, PREG_UNMATCHED_AS_NULL ) ) { |
| 1224 | $sel = substr( $sel, strlen( $cap[0] ) ); |
| 1225 | $buff[] = $this->tok( $cap ); |
| 1226 | } |
| 1227 | |
| 1228 | if ( $sel && $sel[ 0 ] === '!' ) { |
| 1229 | $sel = substr( $sel, 1 ); |
| 1230 | $subject = $this->makeSubject(); |
| 1231 | $subject->qname = $qname; |
| 1232 | $buff[] = $subject->simple; |
| 1233 | } |
| 1234 | |
| 1235 | if ( preg_match( self::$rules->ref, $sel, $cap ) ) { |
| 1236 | $sel = substr( $sel, strlen( $cap[0] ) ); |
| 1237 | $ref = $this->makeRef( self::makeSimple( $buff ), self::decodeid( $cap[ 1 ] ) ); |
| 1238 | $filter[] = $ref->combinator; |
| 1239 | $buff = []; |
| 1240 | continue; |
| 1241 | } |
| 1242 | |
| 1243 | if ( preg_match( self::$rules->combinator, $sel, $cap, PREG_UNMATCHED_AS_NULL ) ) { |
| 1244 | $sel = substr( $sel, strlen( $cap[0] ) ); |
| 1245 | $op = $cap[ 1 ] ?? $cap[ 2 ] ?? $cap[ 3 ]; |
| 1246 | if ( $op === ',' ) { |
| 1247 | $filter[] = $this->combinators['noop']( self::makeSimple( $buff ) ); |
| 1248 | break; |
| 1249 | } |
| 1250 | } else { |
| 1251 | $op = 'noop'; |
| 1252 | } |
| 1253 | |
| 1254 | if ( !isset( $this->combinators[ $op ] ) ) { |
| 1255 | throw $this->newBadSelectorException( 'Bad combinator: ' . $op ); |
| 1256 | } |
| 1257 | $filter[] = $this->combinators[ $op ]( self::makeSimple( $buff ) ); |
| 1258 | $buff = []; |
| 1259 | } |
| 1260 | |
| 1261 | $test = self::makeTest( $filter ); |
| 1262 | $test->qname = $qname; |
| 1263 | $test->sel = $sel; |
| 1264 | |
| 1265 | if ( $subject ) { |
| 1266 | $subject->lname = $test->qname; |
| 1267 | |
| 1268 | $subject->test = $test; |
| 1269 | // @phan-suppress-next-line PhanPluginDuplicateExpressionAssignment |
| 1270 | $subject->qname = $subject->qname; |
| 1271 | $subject->sel = $test->sel; |
| 1272 | $test = $subject; |
| 1273 | } |
| 1274 | |
| 1275 | if ( $ref ) { |
| 1276 | $ref->test = $test; |
| 1277 | $ref->qname = $test->qname; |
| 1278 | $ref->sel = $test->sel; |
| 1279 | $test = $ref; |
| 1280 | } |
| 1281 | |
| 1282 | return $test; |
| 1283 | } |
| 1284 | |
| 1285 | /** @return callable(DOMNode,array):bool */ |
| 1286 | private function tokQname( string $cap ): callable { |
| 1287 | // qname |
| 1288 | if ( $cap === '*' ) { |
| 1289 | return $this->selectors0['*']; |
| 1290 | } elseif ( substr( $cap, 0, 1 ) === '|' ) { |
| 1291 | // no namespace |
| 1292 | return $this->selectors1['typeNoNS']( substr( $cap, 1 ), $this ); |
| 1293 | } elseif ( substr( $cap, 0, 2 ) === '*|' ) { |
| 1294 | // any namespace including no namespace |
| 1295 | return $this->selectors1['type']( substr( $cap, 2 ), $this ); |
| 1296 | } else { |
| 1297 | return $this->selectors1['type']( $cap, $this ); |
| 1298 | } |
| 1299 | } |
| 1300 | |
| 1301 | /** @return callable(DOMNode,array):bool */ |
| 1302 | private function tok( array $cap ): callable { |
| 1303 | // class/id |
| 1304 | if ( $cap[ 1 ] ) { |
| 1305 | return $cap[ 1 ][ 0 ] === '.' |
| 1306 | // XXX unescape here? or in attr? |
| 1307 | ? $this->selectorsAttr( 'class', '~=', self::decodeid( substr( $cap[ 1 ], 1 ) ), false ) : |
| 1308 | $this->selectorsAttr( 'id', '=', self::decodeid( substr( $cap[ 1 ], 1 ) ), false ); |
| 1309 | } |
| 1310 | |
| 1311 | // pseudo-name |
| 1312 | // inside-pseudo |
| 1313 | if ( $cap[ 2 ] ) { |
| 1314 | $id = self::decodeid( $cap[ 2 ] ); |
| 1315 | if ( isset( $cap[3] ) && $cap[ 3 ] ) { |
| 1316 | if ( !isset( $this->selectors1[ $id ] ) ) { |
| 1317 | throw $this->newBadSelectorException( "Unknown Selector: $id" ); |
| 1318 | } |
| 1319 | return $this->selectors1[ $id ]( self::unquote( $cap[ 3 ] ), $this ); |
| 1320 | } else { |
| 1321 | if ( !isset( $this->selectors0[ $id ] ) ) { |
| 1322 | throw $this->newBadSelectorException( "Unknown Selector: $id" ); |
| 1323 | } |
| 1324 | return $this->selectors0[ $id ]; |
| 1325 | } |
| 1326 | } |
| 1327 | |
| 1328 | // attr name |
| 1329 | // attr op |
| 1330 | // attr value |
| 1331 | if ( $cap[ 4 ] ) { |
| 1332 | $value = $cap[ 6 ] ?? ''; |
| 1333 | $i = preg_match( "/[\"'\\s]\\s*I\$/i", $value ); |
| 1334 | if ( $i ) { |
| 1335 | $value = preg_replace( '/\s*I$/i', '', $value, 1 ); |
| 1336 | } |
| 1337 | return $this->selectorsAttr( self::decodeid( $cap[ 4 ] ), $cap[ 5 ] ?? '-', self::unquote( $value ), (bool)$i ); |
| 1338 | } |
| 1339 | |
| 1340 | throw $this->newBadSelectorException( 'Unknown Selector.' ); |
| 1341 | } |
| 1342 | |
| 1343 | /** |
| 1344 | * Returns true if all $func return true |
| 1345 | * @param array<callable(DOMNode,array):bool> $func |
| 1346 | * @return callable(DOMNode,array):bool |
| 1347 | */ |
| 1348 | private static function makeSimple( array $func ): callable { |
| 1349 | $l = count( $func ); |
| 1350 | |
| 1351 | // Potentially make sure |
| 1352 | // `el` is truthy. |
| 1353 | if ( $l < 2 ) { |
| 1354 | return $func[ 0 ]; |
| 1355 | } |
| 1356 | |
| 1357 | return static function ( $el, $opts ) use ( $l, $func ): bool { |
| 1358 | for ( $i = 0; $i < $l; $i++ ) { |
| 1359 | if ( !call_user_func( $func[ $i ], $el, $opts ) ) { |
| 1360 | return false; |
| 1361 | } |
| 1362 | } |
| 1363 | return true; |
| 1364 | }; |
| 1365 | } |
| 1366 | |
| 1367 | /** |
| 1368 | * Returns the element that all $func return |
| 1369 | * @param array<callable(DOMNode,array):(?DOMNode)> $func |
| 1370 | * @return ZestFunc |
| 1371 | */ |
| 1372 | private static function makeTest( array $func ): ZestFunc { |
| 1373 | if ( count( $func ) < 2 ) { |
| 1374 | return new ZestFunc( static function ( $el, $opts ) use ( $func ): bool { |
| 1375 | return (bool)call_user_func( $func[ 0 ], $el, $opts ); |
| 1376 | } ); |
| 1377 | } |
| 1378 | return new ZestFunc( static function ( $el, $opts ) use ( $func ): bool { |
| 1379 | $i = count( $func ); |
| 1380 | while ( $i-- ) { |
| 1381 | if ( !( $el = call_user_func( $func[ $i ], $el, $opts ) ) ) { |
| 1382 | return false; |
| 1383 | } |
| 1384 | } |
| 1385 | return true; |
| 1386 | } ); |
| 1387 | } |
| 1388 | |
| 1389 | /** |
| 1390 | * Return a skeleton ZestFunc for the caller to fill in. |
| 1391 | * @return ZestFunc |
| 1392 | */ |
| 1393 | private function makeSubject(): ZestFunc { |
| 1394 | $target = null; |
| 1395 | |
| 1396 | $subject = new ZestFunc( function ( $el, $opts ) use ( &$subject, &$target ): bool { |
| 1397 | $node = $el->ownerDocument; |
| 1398 | $scope = $this->getElementsByTagName( $node, $subject->lname, $opts ); |
| 1399 | $i = count( $scope ); |
| 1400 | |
| 1401 | while ( $i-- ) { |
| 1402 | if ( call_user_func( $subject->test->func, $scope[$i], $opts ) && $target === $el ) { |
| 1403 | $target = null; |
| 1404 | return true; |
| 1405 | } |
| 1406 | } |
| 1407 | |
| 1408 | $target = null; |
| 1409 | return false; |
| 1410 | } ); |
| 1411 | |
| 1412 | $subject->simple = static function ( $el, $opts ) use ( &$target ): bool { |
| 1413 | $target = $el; |
| 1414 | return true; |
| 1415 | }; |
| 1416 | |
| 1417 | return $subject; |
| 1418 | } |
| 1419 | |
| 1420 | /** |
| 1421 | * @return callable(DOMNode,array):bool |
| 1422 | */ |
| 1423 | private function compileGroup( string $sel ): callable { |
| 1424 | $test = $this->compile( $sel ); |
| 1425 | $tests = [ $test ]; |
| 1426 | |
| 1427 | while ( $test->sel ) { |
| 1428 | $test = $this->compile( $test->sel ); |
| 1429 | $tests[] = $test; |
| 1430 | } |
| 1431 | |
| 1432 | if ( count( $tests ) < 2 ) { |
| 1433 | return $test->func; |
| 1434 | } |
| 1435 | |
| 1436 | return static function ( $el, $opts ) use ( $tests ): bool { |
| 1437 | for ( $i = 0, $l = count( $tests ); $i < $l; $i++ ) { |
| 1438 | if ( call_user_func( $tests[ $i ]->func, $el, $opts ) ) { |
| 1439 | return true; |
| 1440 | } |
| 1441 | } |
| 1442 | return false; |
| 1443 | }; |
| 1444 | } |
| 1445 | |
| 1446 | /** |
| 1447 | * Selection |
| 1448 | */ |
| 1449 | |
| 1450 | // $node should be a DOMDocument, DOMDocumentFragment, or a DOMElement |
| 1451 | // These are "ParentNode" in the DOM spec. |
| 1452 | |
| 1453 | /** |
| 1454 | * @param string $sel |
| 1455 | * @param DOMDocument|DOMDocumentFragment|DOMElement $node |
| 1456 | * @param array $opts |
| 1457 | * @return DOMElement[] |
| 1458 | */ |
| 1459 | private function findInternal( string $sel, $node, $opts ): array { |
| 1460 | $results = []; |
| 1461 | $test = $this->compile( $sel ); |
| 1462 | $scope = $this->getElementsByTagName( $node, $test->qname, $opts ); |
| 1463 | $i = 0; |
| 1464 | $el = null; |
| 1465 | $needsSort = false; |
| 1466 | |
| 1467 | foreach ( $scope as $el ) { |
| 1468 | if ( call_user_func( $test->func, $el, $opts ) ) { |
| 1469 | $results[spl_object_id( $el )] = $el; |
| 1470 | } |
| 1471 | } |
| 1472 | |
| 1473 | if ( $test->sel ) { |
| 1474 | $needsSort = true; |
| 1475 | while ( $test->sel ) { |
| 1476 | $test = $this->compile( $test->sel ); |
| 1477 | $scope = $this->getElementsByTagName( $node, $test->qname, $opts ); |
| 1478 | foreach ( $scope as $el ) { |
| 1479 | if ( call_user_func( $test->func, $el, $opts ) ) { |
| 1480 | $results[spl_object_id( $el )] = $el; |
| 1481 | } |
| 1482 | } |
| 1483 | } |
| 1484 | } |
| 1485 | |
| 1486 | $results = array_values( $results ); |
| 1487 | if ( $needsSort ) { |
| 1488 | self::sort( $results, $this->isStandardsMode( $node, $opts ) ); |
| 1489 | } |
| 1490 | return $results; |
| 1491 | } |
| 1492 | |
| 1493 | /** |
| 1494 | * Find elements matching a CSS selector underneath $context. |
| 1495 | * |
| 1496 | * This search is exclusive; that is, `find(':scope', ...)` returns |
| 1497 | * no matches, although `:scope *` would return matches. |
| 1498 | * |
| 1499 | * @param string $sel The CSS selector string |
| 1500 | * @param DOMDocument|DOMDocumentFragment|DOMElement $context |
| 1501 | * The scoping root for the search |
| 1502 | * @param array $opts Additional match-context options (optional) |
| 1503 | * @return DOMElement[] Elements matching the CSS selector |
| 1504 | */ |
| 1505 | public function find( string $sel, $context, array $opts = [] ): array { |
| 1506 | $opts['this'] = $this; |
| 1507 | $opts['scope'] = $context; |
| 1508 | |
| 1509 | /* when context isn't a DocumentFragment and the selector is simple: */ |
| 1510 | if ( $context->nodeType !== 11 && strpos( $sel, ' ' ) === false ) { |
| 1511 | // https://www.w3.org/TR/CSS21/syndata.html#value-def-identifier |
| 1512 | // Valid identifiers starting with a hyphen or with escape |
| 1513 | // sequences will be handled correctly by the fall-through case. |
| 1514 | if ( $sel[ 0 ] === '#' && preg_match( '/^#[A-Za-z_](?:[-A-Za-z0-9_]|[^\0-\237])*$/Su', $sel ) ) { |
| 1515 | // Setting 'getElementsById' to `true` disables this |
| 1516 | // optimization and forces the hard/slow search in |
| 1517 | // order to guarantee that multiple elements will be |
| 1518 | // returned if there are multiple elements in the |
| 1519 | // $context with the given id. |
| 1520 | if ( ( $opts['getElementsById'] ?? null ) !== true ) { |
| 1521 | $id = substr( $sel, 1 ); |
| 1522 | return $this->getElementsById( $context, $id, $opts ); |
| 1523 | } |
| 1524 | } |
| 1525 | if ( $sel[ 0 ] === '.' && preg_match( '/^\.\w+$/', $sel ) ) { |
| 1526 | return $this->getElementsByClassName( $context, substr( $sel, 1 ), $opts ); |
| 1527 | } |
| 1528 | if ( preg_match( '/^\w+$/', $sel ) ) { |
| 1529 | return $this->getElementsByTagName( $context, $sel, $opts ); |
| 1530 | } |
| 1531 | } |
| 1532 | /* do things the hard/slow way */ |
| 1533 | return $this->findInternal( $sel, $context, $opts ); |
| 1534 | } |
| 1535 | |
| 1536 | /** |
| 1537 | * Determine whether an element matches the given selector. |
| 1538 | * |
| 1539 | * This test is inclusive; that is, `matches($el, ':scope')` |
| 1540 | * returns true. |
| 1541 | * |
| 1542 | * @param DOMNode $el The element to be tested |
| 1543 | * @param string $sel The CSS selector string |
| 1544 | * @param array $opts Additional match-context options (optional) |
| 1545 | * @return bool True iff the element matches the selector |
| 1546 | */ |
| 1547 | public function matches( $el, string $sel, array $opts = [] ): bool { |
| 1548 | $opts['this'] = $this; |
| 1549 | $opts['scope'] = $el; |
| 1550 | |
| 1551 | $test = new ZestFunc( static function ( $el, $opts ): bool { |
| 1552 | return true; |
| 1553 | } ); |
| 1554 | $test->sel = $sel; |
| 1555 | do { |
| 1556 | $test = $this->compile( $test->sel ); |
| 1557 | if ( call_user_func( $test->func, $el, $opts ) ) { |
| 1558 | return true; |
| 1559 | } |
| 1560 | } while ( $test->sel ); |
| 1561 | return false; |
| 1562 | } |
| 1563 | |
| 1564 | /** |
| 1565 | * Allow customization of the exception thrown for a bad selector. |
| 1566 | * @param string $msg Description of the failure |
| 1567 | * @return Throwable |
| 1568 | */ |
| 1569 | protected function newBadSelectorException( string $msg ): Throwable { |
| 1570 | return new InvalidArgumentException( $msg ); |
| 1571 | } |
| 1572 | |
| 1573 | /** |
| 1574 | * Allow subclasses to force Zest into "standards mode" (or not). |
| 1575 | * The default implementation looks for a 'standardsMode' key in the |
| 1576 | * option and if that is not present switches to standards mode if |
| 1577 | * the ownerDocument of the given node is not a \DOMDocument. |
| 1578 | * @param DOMNode $context a context node |
| 1579 | * @param array $opts The zest options array pased to ::find, ::matches, etc |
| 1580 | * @return bool True for standards mode, otherwise false. |
| 1581 | */ |
| 1582 | protected function isStandardsMode( $context, array $opts ): bool { |
| 1583 | // The $opts array can force a specific mode, if key is present |
| 1584 | if ( array_key_exists( 'standardsMode', $opts ) ) { |
| 1585 | return (bool)$opts['standardsMode']; |
| 1586 | } |
| 1587 | // Otherwise guess "not standard mode" if the node document is a |
| 1588 | // \DOMDocument, otherwise use standards mode. |
| 1589 | $doc = self::nodeIsDocument( $context ) ? |
| 1590 | $context : $context->ownerDocument; |
| 1591 | return !( $doc instanceof DOMDocument ); |
| 1592 | } |
| 1593 | |
| 1594 | /** @var ?ZestInst */ |
| 1595 | private static $singleton = null; |
| 1596 | |
| 1597 | /** |
| 1598 | * Create a new instance of Zest. Custom combinators and selectors |
| 1599 | * registered for each instance of Zest do not bleed |
| 1600 | * over into other instances. |
| 1601 | */ |
| 1602 | public function __construct() { |
| 1603 | $z = self::$singleton; |
| 1604 | $this->selectors0 = $z ? $z->selectors0 : []; |
| 1605 | $this->selectors1 = $z ? $z->selectors1 : []; |
| 1606 | $this->operators = $z ? $z->operators : []; |
| 1607 | $this->combinators = $z ? $z->combinators : []; |
| 1608 | if ( !$z ) { |
| 1609 | $this->initRules(); |
| 1610 | $this->initSelectors(); |
| 1611 | $this->initOperators(); |
| 1612 | $this->initCombinators(); |
| 1613 | self::$singleton = $this; |
| 1614 | // Now create another instance so that backing arrays are cloned |
| 1615 | // @phan-suppress-next-line PhanPossiblyInfiniteRecursionSameParams |
| 1616 | self::$singleton = new ZestInst; |
| 1617 | } |
| 1618 | } |
| 1619 | } |