| 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\SyntaxErrorException; |
| 15 |
use Symfony\Component\CssSelector\Node; |
| 16 |
use Symfony\Component\CssSelector\Parser\Tokenizer\Tokenizer; |
| 17 |
|
| 18 |
/** |
| 19 |
* CSS selector parser. |
| 20 |
* |
| 21 |
* This component is a port of the Python cssselect library, |
| 22 |
* which is copyright Ian Bicking, @see https://github.com/scrapy/cssselect. |
| 23 |
* |
| 24 |
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com> |
| 25 |
* |
| 26 |
* @internal |
| 27 |
*/ |
| 28 |
class Parser implements ParserInterface |
| 29 |
{ |
| 30 |
private Tokenizer $tokenizer; |
| 31 |
|
| 32 |
public function __construct(?Tokenizer $tokenizer = null) |
| 33 |
{ |
| 34 |
$this->tokenizer = $tokenizer ?? new Tokenizer(); |
| 35 |
} |
| 36 |
|
| 37 |
public function parse(string $source): array |
| 38 |
{ |
| 39 |
$reader = new Reader($source); |
| 40 |
$stream = $this->tokenizer->tokenize($reader); |
| 41 |
|
| 42 |
return $this->parseSelectorList($stream); |
| 43 |
} |
| 44 |
|
| 45 |
/** |
| 46 |
* Parses the arguments for ":nth-child()" and friends. |
| 47 |
* |
| 48 |
* @param Token[] $tokens |
| 49 |
* |
| 50 |
* @throws SyntaxErrorException |
| 51 |
*/ |
| 52 |
public static function parseSeries(array $tokens): array |
| 53 |
{ |
| 54 |
foreach ($tokens as $token) { |
| 55 |
if ($token->isString()) { |
| 56 |
throw SyntaxErrorException::stringAsFunctionArgument(); |
| 57 |
} |
| 58 |
} |
| 59 |
|
| 60 |
$joined = trim(implode('', array_map(fn (Token $token) => $token->getValue(), $tokens))); |
| 61 |
|
| 62 |
$int = function ($string) { |
| 63 |
if (!is_numeric($string)) { |
| 64 |
throw SyntaxErrorException::stringAsFunctionArgument(); |
| 65 |
} |
| 66 |
|
| 67 |
return (int) $string; |
| 68 |
}; |
| 69 |
|
| 70 |
switch (true) { |
| 71 |
case 'odd' === $joined: |
| 72 |
return [2, 1]; |
| 73 |
case 'even' === $joined: |
| 74 |
return [2, 0]; |
| 75 |
case 'n' === $joined: |
| 76 |
return [1, 0]; |
| 77 |
case !str_contains($joined, 'n'): |
| 78 |
return [0, $int($joined)]; |
| 79 |
} |
| 80 |
|
| 81 |
$split = explode('n', $joined); |
| 82 |
$first = $split[0] ?? null; |
| 83 |
|
| 84 |
return [ |
| 85 |
$first ? ('-' === $first || '+' === $first ? $int($first.'1') : $int($first)) : 1, |
| 86 |
isset($split[1]) && $split[1] ? $int($split[1]) : 0, |
| 87 |
]; |
| 88 |
} |
| 89 |
|
| 90 |
private function parseSelectorList(TokenStream $stream): array |
| 91 |
{ |
| 92 |
$stream->skipWhitespace(); |
| 93 |
$selectors = []; |
| 94 |
|
| 95 |
while (true) { |
| 96 |
$selectors[] = $this->parserSelectorNode($stream); |
| 97 |
|
| 98 |
if ($stream->getPeek()->isDelimiter([','])) { |
| 99 |
$stream->getNext(); |
| 100 |
$stream->skipWhitespace(); |
| 101 |
} else { |
| 102 |
break; |
| 103 |
} |
| 104 |
} |
| 105 |
|
| 106 |
return $selectors; |
| 107 |
} |
| 108 |
|
| 109 |
private function parserSelectorNode(TokenStream $stream): Node\SelectorNode |
| 110 |
{ |
| 111 |
[$result, $pseudoElement] = $this->parseSimpleSelector($stream); |
| 112 |
|
| 113 |
while (true) { |
| 114 |
$stream->skipWhitespace(); |
| 115 |
$peek = $stream->getPeek(); |
| 116 |
|
| 117 |
if ($peek->isFileEnd() || $peek->isDelimiter([','])) { |
| 118 |
break; |
| 119 |
} |
| 120 |
|
| 121 |
if (null !== $pseudoElement) { |
| 122 |
throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector'); |
| 123 |
} |
| 124 |
|
| 125 |
if ($peek->isDelimiter(['+', '>', '~'])) { |
| 126 |
$combinator = $stream->getNext()->getValue(); |
| 127 |
$stream->skipWhitespace(); |
| 128 |
} else { |
| 129 |
$combinator = ' '; |
| 130 |
} |
| 131 |
|
| 132 |
[$nextSelector, $pseudoElement] = $this->parseSimpleSelector($stream); |
| 133 |
$result = new Node\CombinedSelectorNode($result, $combinator, $nextSelector); |
| 134 |
} |
| 135 |
|
| 136 |
return new Node\SelectorNode($result, $pseudoElement); |
| 137 |
} |
| 138 |
|
| 139 |
/** |
| 140 |
* Parses next simple node (hash, class, pseudo, negation). |
| 141 |
* |
| 142 |
* @throws SyntaxErrorException |
| 143 |
*/ |
| 144 |
private function parseSimpleSelector(TokenStream $stream, bool $insideNegation = false): array |
| 145 |
{ |
| 146 |
$stream->skipWhitespace(); |
| 147 |
|
| 148 |
$selectorStart = \count($stream->getUsed()); |
| 149 |
$result = $this->parseElementNode($stream); |
| 150 |
$pseudoElement = null; |
| 151 |
|
| 152 |
while (true) { |
| 153 |
$peek = $stream->getPeek(); |
| 154 |
if ($peek->isWhitespace() |
| 155 |
|| $peek->isFileEnd() |
| 156 |
|| $peek->isDelimiter([',', '+', '>', '~']) |
| 157 |
|| ($insideNegation && $peek->isDelimiter([')'])) |
| 158 |
) { |
| 159 |
break; |
| 160 |
} |
| 161 |
|
| 162 |
if (null !== $pseudoElement) { |
| 163 |
throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector'); |
| 164 |
} |
| 165 |
|
| 166 |
if ($peek->isHash()) { |
| 167 |
$result = new Node\HashNode($result, $stream->getNext()->getValue()); |
| 168 |
} elseif ($peek->isDelimiter(['.'])) { |
| 169 |
$stream->getNext(); |
| 170 |
$result = new Node\ClassNode($result, $stream->getNextIdentifier()); |
| 171 |
} elseif ($peek->isDelimiter(['['])) { |
| 172 |
$stream->getNext(); |
| 173 |
$result = $this->parseAttributeNode($result, $stream); |
| 174 |
} elseif ($peek->isDelimiter([':'])) { |
| 175 |
$stream->getNext(); |
| 176 |
|
| 177 |
if ($stream->getPeek()->isDelimiter([':'])) { |
| 178 |
$stream->getNext(); |
| 179 |
$pseudoElement = $stream->getNextIdentifier(); |
| 180 |
|
| 181 |
continue; |
| 182 |
} |
| 183 |
|
| 184 |
$identifier = $stream->getNextIdentifier(); |
| 185 |
if (\in_array(strtolower($identifier), ['first-line', 'first-letter', 'before', 'after'])) { |
| 186 |
// Special case: CSS 2.1 pseudo-elements can have a single ':'. |
| 187 |
// Any new pseudo-element must have two. |
| 188 |
$pseudoElement = $identifier; |
| 189 |
|
| 190 |
continue; |
| 191 |
} |
| 192 |
|
| 193 |
if (!$stream->getPeek()->isDelimiter(['('])) { |
| 194 |
$result = new Node\PseudoNode($result, $identifier); |
| 195 |
if ('Pseudo[Element[*]:scope]' === $result->__toString()) { |
| 196 |
$used = \count($stream->getUsed()); |
| 197 |
if (!(2 === $used |
| 198 |
|| 3 === $used && $stream->getUsed()[0]->isWhiteSpace() |
| 199 |
|| $used >= 3 && $stream->getUsed()[$used - 3]->isDelimiter([',']) |
| 200 |
|| $used >= 4 |
| 201 |
&& $stream->getUsed()[$used - 3]->isWhiteSpace() |
| 202 |
&& $stream->getUsed()[$used - 4]->isDelimiter([',']) |
| 203 |
)) { |
| 204 |
throw SyntaxErrorException::notAtTheStartOfASelector('scope'); |
| 205 |
} |
| 206 |
} |
| 207 |
continue; |
| 208 |
} |
| 209 |
|
| 210 |
$stream->getNext(); |
| 211 |
$stream->skipWhitespace(); |
| 212 |
|
| 213 |
if ('not' === strtolower($identifier)) { |
| 214 |
if ($insideNegation) { |
| 215 |
throw SyntaxErrorException::nestedNot(); |
| 216 |
} |
| 217 |
|
| 218 |
[$argument, $argumentPseudoElement] = $this->parseSimpleSelector($stream, true); |
| 219 |
$next = $stream->getNext(); |
| 220 |
|
| 221 |
if (null !== $argumentPseudoElement) { |
| 222 |
throw SyntaxErrorException::pseudoElementFound($argumentPseudoElement, 'inside ::not()'); |
| 223 |
} |
| 224 |
|
| 225 |
if (!$next->isDelimiter([')'])) { |
| 226 |
throw SyntaxErrorException::unexpectedToken('")"', $next); |
| 227 |
} |
| 228 |
|
| 229 |
$result = new Node\NegationNode($result, $argument); |
| 230 |
} else { |
| 231 |
$arguments = []; |
| 232 |
$next = null; |
| 233 |
|
| 234 |
while (true) { |
| 235 |
$stream->skipWhitespace(); |
| 236 |
$next = $stream->getNext(); |
| 237 |
|
| 238 |
if ($next->isIdentifier() |
| 239 |
|| $next->isString() |
| 240 |
|| $next->isNumber() |
| 241 |
|| $next->isDelimiter(['+', '-']) |
| 242 |
) { |
| 243 |
$arguments[] = $next; |
| 244 |
} elseif ($next->isDelimiter([')'])) { |
| 245 |
break; |
| 246 |
} else { |
| 247 |
throw SyntaxErrorException::unexpectedToken('an argument', $next); |
| 248 |
} |
| 249 |
} |
| 250 |
|
| 251 |
if (!$arguments) { |
| 252 |
throw SyntaxErrorException::unexpectedToken('at least one argument', $next); |
| 253 |
} |
| 254 |
|
| 255 |
$result = new Node\FunctionNode($result, $identifier, $arguments); |
| 256 |
} |
| 257 |
} else { |
| 258 |
throw SyntaxErrorException::unexpectedToken('selector', $peek); |
| 259 |
} |
| 260 |
} |
| 261 |
|
| 262 |
if (\count($stream->getUsed()) === $selectorStart) { |
| 263 |
throw SyntaxErrorException::unexpectedToken('selector', $stream->getPeek()); |
| 264 |
} |
| 265 |
|
| 266 |
return [$result, $pseudoElement]; |
| 267 |
} |
| 268 |
|
| 269 |
private function parseElementNode(TokenStream $stream): Node\ElementNode |
| 270 |
{ |
| 271 |
$peek = $stream->getPeek(); |
| 272 |
|
| 273 |
if ($peek->isIdentifier() || $peek->isDelimiter(['*'])) { |
| 274 |
if ($peek->isIdentifier()) { |
| 275 |
$namespace = $stream->getNext()->getValue(); |
| 276 |
} else { |
| 277 |
$stream->getNext(); |
| 278 |
$namespace = null; |
| 279 |
} |
| 280 |
|
| 281 |
if ($stream->getPeek()->isDelimiter(['|'])) { |
| 282 |
$stream->getNext(); |
| 283 |
$element = $stream->getNextIdentifierOrStar(); |
| 284 |
} else { |
| 285 |
$element = $namespace; |
| 286 |
$namespace = null; |
| 287 |
} |
| 288 |
} else { |
| 289 |
$element = $namespace = null; |
| 290 |
} |
| 291 |
|
| 292 |
return new Node\ElementNode($namespace, $element); |
| 293 |
} |
| 294 |
|
| 295 |
private function parseAttributeNode(Node\NodeInterface $selector, TokenStream $stream): Node\AttributeNode |
| 296 |
{ |
| 297 |
$stream->skipWhitespace(); |
| 298 |
$attribute = $stream->getNextIdentifierOrStar(); |
| 299 |
|
| 300 |
if (null === $attribute && !$stream->getPeek()->isDelimiter(['|'])) { |
| 301 |
throw SyntaxErrorException::unexpectedToken('"|"', $stream->getPeek()); |
| 302 |
} |
| 303 |
|
| 304 |
if ($stream->getPeek()->isDelimiter(['|'])) { |
| 305 |
$stream->getNext(); |
| 306 |
|
| 307 |
if ($stream->getPeek()->isDelimiter(['='])) { |
| 308 |
$namespace = null; |
| 309 |
$stream->getNext(); |
| 310 |
$operator = '|='; |
| 311 |
} else { |
| 312 |
$namespace = $attribute; |
| 313 |
$attribute = $stream->getNextIdentifier(); |
| 314 |
$operator = null; |
| 315 |
} |
| 316 |
} else { |
| 317 |
$namespace = $operator = null; |
| 318 |
} |
| 319 |
|
| 320 |
if (null === $operator) { |
| 321 |
$stream->skipWhitespace(); |
| 322 |
$next = $stream->getNext(); |
| 323 |
|
| 324 |
if ($next->isDelimiter([']'])) { |
| 325 |
return new Node\AttributeNode($selector, $namespace, $attribute, 'exists', null); |
| 326 |
} elseif ($next->isDelimiter(['='])) { |
| 327 |
$operator = '='; |
| 328 |
} elseif ($next->isDelimiter(['^', '$', '*', '~', '|', '!']) |
| 329 |
&& $stream->getPeek()->isDelimiter(['=']) |
| 330 |
) { |
| 331 |
$operator = $next->getValue().'='; |
| 332 |
$stream->getNext(); |
| 333 |
} else { |
| 334 |
throw SyntaxErrorException::unexpectedToken('operator', $next); |
| 335 |
} |
| 336 |
} |
| 337 |
|
| 338 |
$stream->skipWhitespace(); |
| 339 |
$value = $stream->getNext(); |
| 340 |
|
| 341 |
if ($value->isNumber()) { |
| 342 |
// if the value is a number, it's casted into a string |
| 343 |
$value = new Token(Token::TYPE_STRING, (string) $value->getValue(), $value->getPosition()); |
| 344 |
} |
| 345 |
|
| 346 |
if (!($value->isIdentifier() || $value->isString())) { |
| 347 |
throw SyntaxErrorException::unexpectedToken('string or identifier', $value); |
| 348 |
} |
| 349 |
|
| 350 |
$stream->skipWhitespace(); |
| 351 |
$next = $stream->getNext(); |
| 352 |
|
| 353 |
if (!$next->isDelimiter([']'])) { |
| 354 |
throw SyntaxErrorException::unexpectedToken('"]"', $next); |
| 355 |
} |
| 356 |
|
| 357 |
return new Node\AttributeNode($selector, $namespace, $attribute, $operator, $value->getValue()); |
| 358 |
} |
| 359 |
} |
| 360 |
|
