| 1 |
<?php |
| 2 |
|
| 3 |
/* |
| 4 |
* This file is part of the Symfony package. |
| 5 |
* |
| 6 |
* (c) Fabien Potencier <fabien@symfony.com> |
| 7 |
* |
| 8 |
* For the full copyright and license information, please view the LICENSE |
| 9 |
* file that was distributed with this source code. |
| 10 |
*/ |
| 11 |
|
| 12 |
namespace Symfony\Component\CssSelector\Parser; |
| 13 |
|
| 14 |
use Symfony\Component\CssSelector\Exception\InternalErrorException; |
| 15 |
use Symfony\Component\CssSelector\Exception\SyntaxErrorException; |
| 16 |
use Symfony\Component\CssSelector\Node; |
| 17 |
use Symfony\Component\CssSelector\Parser\Tokenizer\Tokenizer; |
| 18 |
|
| 19 |
/** |
| 20 |
* CSS selector parser. |
| 21 |
* |
| 22 |
* This component is a port of the Python cssselect library, |
| 23 |
* which is copyright Ian Bicking, @see https://github.com/scrapy/cssselect. |
| 24 |
* |
| 25 |
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com> |
| 26 |
* |
| 27 |
* @internal |
| 28 |
*/ |
| 29 |
class Parser implements ParserInterface |
| 30 |
{ |
| 31 |
private const HAS_NESTING_LIMIT = 16; |
| 32 |
private const NESTING_LIMIT = 16; |
| 33 |
|
| 34 |
private Tokenizer $tokenizer; |
| 35 |
private int $hasNestingDepth = 0; |
| 36 |
private int $nestingDepth = 0; |
| 37 |
|
| 38 |
public function __construct(?Tokenizer $tokenizer = null) |
| 39 |
{ |
| 40 |
$this->tokenizer = $tokenizer ?? new Tokenizer(); |
| 41 |
} |
| 42 |
|
| 43 |
public function parse(string $source): array |
| 44 |
{ |
| 45 |
$reader = new Reader($source); |
| 46 |
$stream = $this->tokenizer->tokenize($reader); |
| 47 |
|
| 48 |
return $this->parseSelectorList($stream); |
| 49 |
} |
| 50 |
|
| 51 |
/** |
| 52 |
* Parses the arguments for ":nth-child()" and friends. |
| 53 |
* |
| 54 |
* @param Token[] $tokens |
| 55 |
* |
| 56 |
* @throws SyntaxErrorException |
| 57 |
*/ |
| 58 |
public static function parseSeries(array $tokens): array |
| 59 |
{ |
| 60 |
foreach ($tokens as $token) { |
| 61 |
if ($token->isString()) { |
| 62 |
throw SyntaxErrorException::stringAsFunctionArgument(); |
| 63 |
} |
| 64 |
} |
| 65 |
|
| 66 |
$joined = trim(implode('', array_map(static fn (Token $token) => $token->getValue(), $tokens))); |
| 67 |
|
| 68 |
$int = static function ($string) { |
| 69 |
if (!is_numeric($string)) { |
| 70 |
throw SyntaxErrorException::stringAsFunctionArgument(); |
| 71 |
} |
| 72 |
|
| 73 |
return (int) $string; |
| 74 |
}; |
| 75 |
|
| 76 |
switch (true) { |
| 77 |
case 'odd' === $joined: |
| 78 |
return [2, 1]; |
| 79 |
case 'even' === $joined: |
| 80 |
return [2, 0]; |
| 81 |
case 'n' === $joined: |
| 82 |
return [1, 0]; |
| 83 |
case !str_contains($joined, 'n'): |
| 84 |
return [0, $int($joined)]; |
| 85 |
} |
| 86 |
|
| 87 |
$split = explode('n', $joined); |
| 88 |
$first = $split[0] ?? null; |
| 89 |
|
| 90 |
return [ |
| 91 |
$first ? ('-' === $first || '+' === $first ? $int($first.'1') : $int($first)) : 1, |
| 92 |
isset($split[1]) && $split[1] ? $int($split[1]) : 0, |
| 93 |
]; |
| 94 |
} |
| 95 |
|
| 96 |
private function parseSelectorList(TokenStream $stream, bool $isArgument = false): array |
| 97 |
{ |
| 98 |
$stream->skipWhitespace(); |
| 99 |
$selectors = []; |
| 100 |
|
| 101 |
while (true) { |
| 102 |
if ($isArgument && $stream->getPeek()->isDelimiter([')'])) { |
| 103 |
break; |
| 104 |
} |
| 105 |
|
| 106 |
$selectors[] = $this->parserSelectorNode($stream, $isArgument); |
| 107 |
|
| 108 |
if ($stream->getPeek()->isDelimiter([','])) { |
| 109 |
$stream->getNext(); |
| 110 |
$stream->skipWhitespace(); |
| 111 |
} else { |
| 112 |
break; |
| 113 |
} |
| 114 |
} |
| 115 |
|
| 116 |
return $selectors; |
| 117 |
} |
| 118 |
|
| 119 |
private function parserSelectorNode(TokenStream $stream, bool $isArgument = false, bool $insideRelativeSelector = false): Node\SelectorNode |
| 120 |
{ |
| 121 |
[$result, $pseudoElement] = $this->parseSimpleSelector($stream, false, $isArgument, $insideRelativeSelector); |
| 122 |
|
| 123 |
while (true) { |
| 124 |
$stream->skipWhitespace(); |
| 125 |
$peek = $stream->getPeek(); |
| 126 |
|
| 127 |
if ( |
| 128 |
$peek->isFileEnd() |
| 129 |
|| $peek->isDelimiter([',']) |
| 130 |
|| ($isArgument && $peek->isDelimiter([')'])) |
| 131 |
) { |
| 132 |
break; |
| 133 |
} |
| 134 |
|
| 135 |
if (null !== $pseudoElement) { |
| 136 |
throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector'); |
| 137 |
} |
| 138 |
|
| 139 |
if ($peek->isDelimiter(['+', '>', '~'])) { |
| 140 |
$combinator = $stream->getNext()->getValue(); |
| 141 |
$stream->skipWhitespace(); |
| 142 |
} else { |
| 143 |
$combinator = ' '; |
| 144 |
} |
| 145 |
|
| 146 |
[$nextSelector, $pseudoElement] = $this->parseSimpleSelector($stream, false, $isArgument, $insideRelativeSelector); |
| 147 |
$result = new Node\CombinedSelectorNode($result, $combinator, $nextSelector); |
| 148 |
} |
| 149 |
|
| 150 |
return new Node\SelectorNode($result, $pseudoElement); |
| 151 |
} |
| 152 |
|
| 153 |
/** |
| 154 |
* @return list<array{0: string, 1: Node\SelectorNode}> |
| 155 |
* |
| 156 |
* @throws SyntaxErrorException |
| 157 |
* @throws InternalErrorException |
| 158 |
*/ |
| 159 |
private function parseRelativeSelector(TokenStream $stream): array |
| 160 |
{ |
| 161 |
if ($this->hasNestingDepth >= self::HAS_NESTING_LIMIT) { |
| 162 |
throw SyntaxErrorException::nestedHas(); |
| 163 |
} |
| 164 |
|
| 165 |
++$this->hasNestingDepth; |
| 166 |
|
| 167 |
try { |
| 168 |
$arguments = []; |
| 169 |
while (true) { |
| 170 |
$stream->skipWhitespace(); |
| 171 |
$peek = $stream->getPeek(); |
| 172 |
|
| 173 |
if ($peek->isDelimiter(['+', '>', '~'])) { |
| 174 |
$combinator = $stream->getNext()->getValue(); |
| 175 |
$stream->skipWhitespace(); |
| 176 |
$peek = $stream->getPeek(); |
| 177 |
} else { |
| 178 |
$combinator = ' '; |
| 179 |
} |
| 180 |
|
| 181 |
if ($peek->isString() || $peek->isNumber()) { |
| 182 |
throw SyntaxErrorException::unexpectedToken('an argument', $stream->getNext()); |
| 183 |
} |
| 184 |
|
| 185 |
$selector = $this->parserSelectorNode($stream, true, true); |
| 186 |
|
| 187 |
if (null !== $pseudoElement = $selector->getPseudoElement()) { |
| 188 |
throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'inside :has()'); |
| 189 |
} |
| 190 |
|
| 191 |
$arguments[] = [$combinator, $selector]; |
| 192 |
|
| 193 |
if ($stream->getPeek()->isDelimiter([','])) { |
| 194 |
$stream->getNext(); |
| 195 |
continue; |
| 196 |
} |
| 197 |
|
| 198 |
break; |
| 199 |
} |
| 200 |
|
| 201 |
$next = $stream->getNext(); |
| 202 |
if (!$next->isDelimiter([')'])) { |
| 203 |
throw SyntaxErrorException::unexpectedToken('")"', $next); |
| 204 |
} |
| 205 |
|
| 206 |
return $arguments; |
| 207 |
} finally { |
| 208 |
--$this->hasNestingDepth; |
| 209 |
} |
| 210 |
} |
| 211 |
|
| 212 |
/** |
| 213 |
* Parses next simple node (hash, class, pseudo, negation). |
| 214 |
* |
| 215 |
* @throws SyntaxErrorException |
| 216 |
* @throws InternalErrorException |
| 217 |
*/ |
| 218 |
private function parseSimpleSelector(TokenStream $stream, bool $insideNegation = false, bool $isArgument = false, bool $insideRelativeSelector = false): array |
| 219 |
{ |
| 220 |
$stream->skipWhitespace(); |
| 221 |
|
| 222 |
$selectorStart = \count($stream->getUsed()); |
| 223 |
$result = $this->parseElementNode($stream); |
| 224 |
$pseudoElement = null; |
| 225 |
|
| 226 |
while (true) { |
| 227 |
$peek = $stream->getPeek(); |
| 228 |
if ($peek->isWhitespace() |
| 229 |
|| $peek->isFileEnd() |
| 230 |
|| $peek->isDelimiter([',', '+', '>', '~']) |
| 231 |
|| ($isArgument && $peek->isDelimiter([')'])) |
| 232 |
) { |
| 233 |
break; |
| 234 |
} |
| 235 |
|
| 236 |
if (null !== $pseudoElement) { |
| 237 |
throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector'); |
| 238 |
} |
| 239 |
|
| 240 |
if ($peek->isHash()) { |
| 241 |
$result = new Node\HashNode($result, $stream->getNext()->getValue()); |
| 242 |
} elseif ($peek->isDelimiter(['.'])) { |
| 243 |
$stream->getNext(); |
| 244 |
$result = new Node\ClassNode($result, $stream->getNextIdentifier()); |
| 245 |
} elseif ($peek->isDelimiter(['['])) { |
| 246 |
$stream->getNext(); |
| 247 |
$result = $this->parseAttributeNode($result, $stream); |
| 248 |
} elseif ($peek->isDelimiter([':'])) { |
| 249 |
$stream->getNext(); |
| 250 |
|
| 251 |
if ($stream->getPeek()->isDelimiter([':'])) { |
| 252 |
$stream->getNext(); |
| 253 |
$pseudoElement = $stream->getNextIdentifier(); |
| 254 |
|
| 255 |
continue; |
| 256 |
} |
| 257 |
|
| 258 |
$identifier = $stream->getNextIdentifier(); |
| 259 |
if (\in_array(strtolower($identifier), ['first-line', 'first-letter', 'before', 'after'], true)) { |
| 260 |
// Special case: CSS 2.1 pseudo-elements can have a single ':'. |
| 261 |
// Any new pseudo-element must have two. |
| 262 |
$pseudoElement = $identifier; |
| 263 |
|
| 264 |
continue; |
| 265 |
} |
| 266 |
|
| 267 |
if (!$stream->getPeek()->isDelimiter(['('])) { |
| 268 |
$result = new Node\PseudoNode($result, $identifier); |
| 269 |
if ('Pseudo[Element[*]:scope]' === $result->__toString()) { |
| 270 |
$used = \count($stream->getUsed()); |
| 271 |
$prevSeparators = [',']; |
| 272 |
if ($insideRelativeSelector) { |
| 273 |
$prevSeparators = [',', '(', '>', '+', '~']; |
| 274 |
} |
| 275 |
if (!(2 === $used |
| 276 |
|| 3 === $used && $stream->getUsed()[0]->isWhiteSpace() |
| 277 |
|| $used >= 3 && $stream->getUsed()[$used - 3]->isDelimiter($prevSeparators) |
| 278 |
|| $used >= 4 |
| 279 |
&& $stream->getUsed()[$used - 3]->isWhiteSpace() |
| 280 |
&& $stream->getUsed()[$used - 4]->isDelimiter($prevSeparators) |
| 281 |
)) { |
| 282 |
throw SyntaxErrorException::notAtTheStartOfASelector('scope'); |
| 283 |
} |
| 284 |
} |
| 285 |
continue; |
| 286 |
} |
| 287 |
|
| 288 |
$stream->getNext(); |
| 289 |
$stream->skipWhitespace(); |
| 290 |
|
| 291 |
if ('not' === strtolower($identifier)) { |
| 292 |
if ($insideNegation) { |
| 293 |
throw SyntaxErrorException::nestedNot(); |
| 294 |
} |
| 295 |
|
| 296 |
[$argument, $argumentPseudoElement] = $this->parseSimpleSelector($stream, true, true); |
| 297 |
$next = $stream->getNext(); |
| 298 |
|
| 299 |
if (null !== $argumentPseudoElement) { |
| 300 |
throw SyntaxErrorException::pseudoElementFound($argumentPseudoElement, 'inside :not()'); |
| 301 |
} |
| 302 |
|
| 303 |
if (!$next->isDelimiter([')'])) { |
| 304 |
throw SyntaxErrorException::unexpectedToken('")"', $next); |
| 305 |
} |
| 306 |
|
| 307 |
$result = new Node\NegationNode($result, $argument); |
| 308 |
} elseif ('is' === strtolower($identifier)) { |
| 309 |
$selectors = $this->parseNestedSelectorList($stream, 'is'); |
| 310 |
|
| 311 |
$next = $stream->getNext(); |
| 312 |
if (!$next->isDelimiter([')'])) { |
| 313 |
throw SyntaxErrorException::unexpectedToken('")"', $next); |
| 314 |
} |
| 315 |
|
| 316 |
$result = new Node\MatchingNode($result, $selectors); |
| 317 |
} elseif ('where' === strtolower($identifier)) { |
| 318 |
$selectors = $this->parseNestedSelectorList($stream, 'where'); |
| 319 |
|
| 320 |
$next = $stream->getNext(); |
| 321 |
if (!$next->isDelimiter([')'])) { |
| 322 |
throw SyntaxErrorException::unexpectedToken('")"', $next); |
| 323 |
} |
| 324 |
|
| 325 |
$result = new Node\SpecificityAdjustmentNode($result, $selectors); |
| 326 |
} elseif ('has' === strtolower($identifier)) { |
| 327 |
$result = new Node\RelationNode($result, $this->parseRelativeSelector($stream)); |
| 328 |
} else { |
| 329 |
$arguments = []; |
| 330 |
$next = null; |
| 331 |
|
| 332 |
while (true) { |
| 333 |
$stream->skipWhitespace(); |
| 334 |
$next = $stream->getNext(); |
| 335 |
|
| 336 |
if ($next->isIdentifier() |
| 337 |
|| $next->isString() |
| 338 |
|| $next->isNumber() |
| 339 |
|| $next->isDelimiter(['+', '-']) |
| 340 |
) { |
| 341 |
$arguments[] = $next; |
| 342 |
} elseif ($next->isDelimiter([')'])) { |
| 343 |
break; |
| 344 |
} else { |
| 345 |
throw SyntaxErrorException::unexpectedToken('an argument', $next); |
| 346 |
} |
| 347 |
} |
| 348 |
|
| 349 |
if (!$arguments) { |
| 350 |
throw SyntaxErrorException::unexpectedToken('at least one argument', $next); |
| 351 |
} |
| 352 |
|
| 353 |
$result = new Node\FunctionNode($result, $identifier, $arguments); |
| 354 |
} |
| 355 |
} else { |
| 356 |
throw SyntaxErrorException::unexpectedToken('selector', $peek); |
| 357 |
} |
| 358 |
} |
| 359 |
|
| 360 |
if (\count($stream->getUsed()) === $selectorStart) { |
| 361 |
throw SyntaxErrorException::unexpectedToken('selector', $stream->getPeek()); |
| 362 |
} |
| 363 |
|
| 364 |
return [$result, $pseudoElement]; |
| 365 |
} |
| 366 |
|
| 367 |
/** |
| 368 |
* @return Node\SelectorNode[] |
| 369 |
* |
| 370 |
* @throws SyntaxErrorException |
| 371 |
*/ |
| 372 |
private function parseNestedSelectorList(TokenStream $stream, string $identifier): array |
| 373 |
{ |
| 374 |
if ($this->nestingDepth >= self::NESTING_LIMIT) { |
| 375 |
throw new SyntaxErrorException(\sprintf('Got too deeply nested :%s().', $identifier)); |
| 376 |
} |
| 377 |
|
| 378 |
++$this->nestingDepth; |
| 379 |
|
| 380 |
try { |
| 381 |
return $this->parseSelectorList($stream, true); |
| 382 |
} finally { |
| 383 |
--$this->nestingDepth; |
| 384 |
} |
| 385 |
} |
| 386 |
|
| 387 |
private function parseElementNode(TokenStream $stream): Node\ElementNode |
| 388 |
{ |
| 389 |
$peek = $stream->getPeek(); |
| 390 |
|
| 391 |
if ($peek->isIdentifier() || $peek->isDelimiter(['*'])) { |
| 392 |
if ($peek->isIdentifier()) { |
| 393 |
$namespace = $stream->getNext()->getValue(); |
| 394 |
} else { |
| 395 |
$stream->getNext(); |
| 396 |
$namespace = null; |
| 397 |
} |
| 398 |
|
| 399 |
if ($stream->getPeek()->isDelimiter(['|'])) { |
| 400 |
$stream->getNext(); |
| 401 |
$element = $stream->getNextIdentifierOrStar(); |
| 402 |
} else { |
| 403 |
$element = $namespace; |
| 404 |
$namespace = null; |
| 405 |
} |
| 406 |
} else { |
| 407 |
$element = $namespace = null; |
| 408 |
} |
| 409 |
|
| 410 |
return new Node\ElementNode($namespace, $element); |
| 411 |
} |
| 412 |
|
| 413 |
private function parseAttributeNode(Node\NodeInterface $selector, TokenStream $stream): Node\AttributeNode |
| 414 |
{ |
| 415 |
$stream->skipWhitespace(); |
| 416 |
$attribute = $stream->getNextIdentifierOrStar(); |
| 417 |
|
| 418 |
if (null === $attribute && !$stream->getPeek()->isDelimiter(['|'])) { |
| 419 |
throw SyntaxErrorException::unexpectedToken('"|"', $stream->getPeek()); |
| 420 |
} |
| 421 |
|
| 422 |
if ($stream->getPeek()->isDelimiter(['|'])) { |
| 423 |
$stream->getNext(); |
| 424 |
|
| 425 |
if ($stream->getPeek()->isDelimiter(['='])) { |
| 426 |
$namespace = null; |
| 427 |
$stream->getNext(); |
| 428 |
$operator = '|='; |
| 429 |
} else { |
| 430 |
$namespace = $attribute; |
| 431 |
$attribute = $stream->getNextIdentifier(); |
| 432 |
$operator = null; |
| 433 |
} |
| 434 |
} else { |
| 435 |
$namespace = $operator = null; |
| 436 |
} |
| 437 |
|
| 438 |
if (null === $operator) { |
| 439 |
$stream->skipWhitespace(); |
| 440 |
$next = $stream->getNext(); |
| 441 |
|
| 442 |
if ($next->isDelimiter([']'])) { |
| 443 |
return new Node\AttributeNode($selector, $namespace, $attribute, 'exists', null); |
| 444 |
} elseif ($next->isDelimiter(['='])) { |
| 445 |
$operator = '='; |
| 446 |
} elseif ($next->isDelimiter(['^', '$', '*', '~', '|', '!']) |
| 447 |
&& $stream->getPeek()->isDelimiter(['=']) |
| 448 |
) { |
| 449 |
$operator = $next->getValue().'='; |
| 450 |
$stream->getNext(); |
| 451 |
} else { |
| 452 |
throw SyntaxErrorException::unexpectedToken('operator', $next); |
| 453 |
} |
| 454 |
} |
| 455 |
|
| 456 |
$stream->skipWhitespace(); |
| 457 |
$value = $stream->getNext(); |
| 458 |
|
| 459 |
if ($value->isNumber()) { |
| 460 |
// if the value is a number, it's casted into a string |
| 461 |
$value = new Token(Token::TYPE_STRING, (string) $value->getValue(), $value->getPosition()); |
| 462 |
} |
| 463 |
|
| 464 |
if (!($value->isIdentifier() || $value->isString())) { |
| 465 |
throw SyntaxErrorException::unexpectedToken('string or identifier', $value); |
| 466 |
} |
| 467 |
|
| 468 |
$stream->skipWhitespace(); |
| 469 |
$next = $stream->getNext(); |
| 470 |
|
| 471 |
if (!$next->isDelimiter([']'])) { |
| 472 |
throw SyntaxErrorException::unexpectedToken('"]"', $next); |
| 473 |
} |
| 474 |
|
| 475 |
return new Node\AttributeNode($selector, $namespace, $attribute, $operator, $value->getValue()); |
| 476 |
} |
| 477 |
} |
| 478 |
|
