annotate vendor/symfony/css-selector/Parser/Parser.php @ 19:fa3358dc1485 tip

Add ndrum files
author Chris Cannam
date Wed, 28 Aug 2019 13:14:47 +0100
parents 129ea1e6d783
children
rev   line source
Chris@0 1 <?php
Chris@0 2
Chris@0 3 /*
Chris@0 4 * This file is part of the Symfony package.
Chris@0 5 *
Chris@0 6 * (c) Fabien Potencier <fabien@symfony.com>
Chris@0 7 *
Chris@0 8 * For the full copyright and license information, please view the LICENSE
Chris@0 9 * file that was distributed with this source code.
Chris@0 10 */
Chris@0 11
Chris@0 12 namespace Symfony\Component\CssSelector\Parser;
Chris@0 13
Chris@0 14 use Symfony\Component\CssSelector\Exception\SyntaxErrorException;
Chris@0 15 use Symfony\Component\CssSelector\Node;
Chris@0 16 use Symfony\Component\CssSelector\Parser\Tokenizer\Tokenizer;
Chris@0 17
Chris@0 18 /**
Chris@0 19 * CSS selector parser.
Chris@0 20 *
Chris@0 21 * This component is a port of the Python cssselect library,
Chris@0 22 * which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
Chris@0 23 *
Chris@0 24 * @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
Chris@0 25 *
Chris@0 26 * @internal
Chris@0 27 */
Chris@0 28 class Parser implements ParserInterface
Chris@0 29 {
Chris@0 30 private $tokenizer;
Chris@0 31
Chris@0 32 public function __construct(Tokenizer $tokenizer = null)
Chris@0 33 {
Chris@0 34 $this->tokenizer = $tokenizer ?: new Tokenizer();
Chris@0 35 }
Chris@0 36
Chris@0 37 /**
Chris@0 38 * {@inheritdoc}
Chris@0 39 */
Chris@0 40 public function parse($source)
Chris@0 41 {
Chris@0 42 $reader = new Reader($source);
Chris@0 43 $stream = $this->tokenizer->tokenize($reader);
Chris@0 44
Chris@0 45 return $this->parseSelectorList($stream);
Chris@0 46 }
Chris@0 47
Chris@0 48 /**
Chris@0 49 * Parses the arguments for ":nth-child()" and friends.
Chris@0 50 *
Chris@0 51 * @param Token[] $tokens
Chris@0 52 *
Chris@0 53 * @return array
Chris@0 54 *
Chris@0 55 * @throws SyntaxErrorException
Chris@0 56 */
Chris@0 57 public static function parseSeries(array $tokens)
Chris@0 58 {
Chris@0 59 foreach ($tokens as $token) {
Chris@0 60 if ($token->isString()) {
Chris@0 61 throw SyntaxErrorException::stringAsFunctionArgument();
Chris@0 62 }
Chris@0 63 }
Chris@0 64
Chris@0 65 $joined = trim(implode('', array_map(function (Token $token) {
Chris@0 66 return $token->getValue();
Chris@0 67 }, $tokens)));
Chris@0 68
Chris@0 69 $int = function ($string) {
Chris@0 70 if (!is_numeric($string)) {
Chris@0 71 throw SyntaxErrorException::stringAsFunctionArgument();
Chris@0 72 }
Chris@0 73
Chris@0 74 return (int) $string;
Chris@0 75 };
Chris@0 76
Chris@0 77 switch (true) {
Chris@0 78 case 'odd' === $joined:
Chris@17 79 return [2, 1];
Chris@0 80 case 'even' === $joined:
Chris@17 81 return [2, 0];
Chris@0 82 case 'n' === $joined:
Chris@17 83 return [1, 0];
Chris@0 84 case false === strpos($joined, 'n'):
Chris@17 85 return [0, $int($joined)];
Chris@0 86 }
Chris@0 87
Chris@0 88 $split = explode('n', $joined);
Chris@0 89 $first = isset($split[0]) ? $split[0] : null;
Chris@0 90
Chris@17 91 return [
Chris@0 92 $first ? ('-' === $first || '+' === $first ? $int($first.'1') : $int($first)) : 1,
Chris@0 93 isset($split[1]) && $split[1] ? $int($split[1]) : 0,
Chris@17 94 ];
Chris@0 95 }
Chris@0 96
Chris@0 97 /**
Chris@0 98 * Parses selector nodes.
Chris@0 99 *
Chris@0 100 * @return array
Chris@0 101 */
Chris@0 102 private function parseSelectorList(TokenStream $stream)
Chris@0 103 {
Chris@0 104 $stream->skipWhitespace();
Chris@17 105 $selectors = [];
Chris@0 106
Chris@0 107 while (true) {
Chris@0 108 $selectors[] = $this->parserSelectorNode($stream);
Chris@0 109
Chris@17 110 if ($stream->getPeek()->isDelimiter([','])) {
Chris@0 111 $stream->getNext();
Chris@0 112 $stream->skipWhitespace();
Chris@0 113 } else {
Chris@0 114 break;
Chris@0 115 }
Chris@0 116 }
Chris@0 117
Chris@0 118 return $selectors;
Chris@0 119 }
Chris@0 120
Chris@0 121 /**
Chris@0 122 * Parses next selector or combined node.
Chris@0 123 *
Chris@0 124 * @return Node\SelectorNode
Chris@0 125 *
Chris@0 126 * @throws SyntaxErrorException
Chris@0 127 */
Chris@0 128 private function parserSelectorNode(TokenStream $stream)
Chris@0 129 {
Chris@0 130 list($result, $pseudoElement) = $this->parseSimpleSelector($stream);
Chris@0 131
Chris@0 132 while (true) {
Chris@0 133 $stream->skipWhitespace();
Chris@0 134 $peek = $stream->getPeek();
Chris@0 135
Chris@17 136 if ($peek->isFileEnd() || $peek->isDelimiter([','])) {
Chris@0 137 break;
Chris@0 138 }
Chris@0 139
Chris@0 140 if (null !== $pseudoElement) {
Chris@0 141 throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector');
Chris@0 142 }
Chris@0 143
Chris@17 144 if ($peek->isDelimiter(['+', '>', '~'])) {
Chris@0 145 $combinator = $stream->getNext()->getValue();
Chris@0 146 $stream->skipWhitespace();
Chris@0 147 } else {
Chris@0 148 $combinator = ' ';
Chris@0 149 }
Chris@0 150
Chris@0 151 list($nextSelector, $pseudoElement) = $this->parseSimpleSelector($stream);
Chris@0 152 $result = new Node\CombinedSelectorNode($result, $combinator, $nextSelector);
Chris@0 153 }
Chris@0 154
Chris@0 155 return new Node\SelectorNode($result, $pseudoElement);
Chris@0 156 }
Chris@0 157
Chris@0 158 /**
Chris@0 159 * Parses next simple node (hash, class, pseudo, negation).
Chris@0 160 *
Chris@0 161 * @param TokenStream $stream
Chris@0 162 * @param bool $insideNegation
Chris@0 163 *
Chris@0 164 * @return array
Chris@0 165 *
Chris@0 166 * @throws SyntaxErrorException
Chris@0 167 */
Chris@0 168 private function parseSimpleSelector(TokenStream $stream, $insideNegation = false)
Chris@0 169 {
Chris@0 170 $stream->skipWhitespace();
Chris@0 171
Chris@17 172 $selectorStart = \count($stream->getUsed());
Chris@0 173 $result = $this->parseElementNode($stream);
Chris@0 174 $pseudoElement = null;
Chris@0 175
Chris@0 176 while (true) {
Chris@0 177 $peek = $stream->getPeek();
Chris@0 178 if ($peek->isWhitespace()
Chris@0 179 || $peek->isFileEnd()
Chris@17 180 || $peek->isDelimiter([',', '+', '>', '~'])
Chris@17 181 || ($insideNegation && $peek->isDelimiter([')']))
Chris@0 182 ) {
Chris@0 183 break;
Chris@0 184 }
Chris@0 185
Chris@0 186 if (null !== $pseudoElement) {
Chris@0 187 throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector');
Chris@0 188 }
Chris@0 189
Chris@0 190 if ($peek->isHash()) {
Chris@0 191 $result = new Node\HashNode($result, $stream->getNext()->getValue());
Chris@17 192 } elseif ($peek->isDelimiter(['.'])) {
Chris@0 193 $stream->getNext();
Chris@0 194 $result = new Node\ClassNode($result, $stream->getNextIdentifier());
Chris@17 195 } elseif ($peek->isDelimiter(['['])) {
Chris@0 196 $stream->getNext();
Chris@0 197 $result = $this->parseAttributeNode($result, $stream);
Chris@17 198 } elseif ($peek->isDelimiter([':'])) {
Chris@0 199 $stream->getNext();
Chris@0 200
Chris@17 201 if ($stream->getPeek()->isDelimiter([':'])) {
Chris@0 202 $stream->getNext();
Chris@0 203 $pseudoElement = $stream->getNextIdentifier();
Chris@0 204
Chris@0 205 continue;
Chris@0 206 }
Chris@0 207
Chris@0 208 $identifier = $stream->getNextIdentifier();
Chris@17 209 if (\in_array(strtolower($identifier), ['first-line', 'first-letter', 'before', 'after'])) {
Chris@0 210 // Special case: CSS 2.1 pseudo-elements can have a single ':'.
Chris@0 211 // Any new pseudo-element must have two.
Chris@0 212 $pseudoElement = $identifier;
Chris@0 213
Chris@0 214 continue;
Chris@0 215 }
Chris@0 216
Chris@17 217 if (!$stream->getPeek()->isDelimiter(['('])) {
Chris@0 218 $result = new Node\PseudoNode($result, $identifier);
Chris@0 219
Chris@0 220 continue;
Chris@0 221 }
Chris@0 222
Chris@0 223 $stream->getNext();
Chris@0 224 $stream->skipWhitespace();
Chris@0 225
Chris@0 226 if ('not' === strtolower($identifier)) {
Chris@0 227 if ($insideNegation) {
Chris@0 228 throw SyntaxErrorException::nestedNot();
Chris@0 229 }
Chris@0 230
Chris@0 231 list($argument, $argumentPseudoElement) = $this->parseSimpleSelector($stream, true);
Chris@0 232 $next = $stream->getNext();
Chris@0 233
Chris@0 234 if (null !== $argumentPseudoElement) {
Chris@0 235 throw SyntaxErrorException::pseudoElementFound($argumentPseudoElement, 'inside ::not()');
Chris@0 236 }
Chris@0 237
Chris@17 238 if (!$next->isDelimiter([')'])) {
Chris@0 239 throw SyntaxErrorException::unexpectedToken('")"', $next);
Chris@0 240 }
Chris@0 241
Chris@0 242 $result = new Node\NegationNode($result, $argument);
Chris@0 243 } else {
Chris@17 244 $arguments = [];
Chris@0 245 $next = null;
Chris@0 246
Chris@0 247 while (true) {
Chris@0 248 $stream->skipWhitespace();
Chris@0 249 $next = $stream->getNext();
Chris@0 250
Chris@0 251 if ($next->isIdentifier()
Chris@0 252 || $next->isString()
Chris@0 253 || $next->isNumber()
Chris@17 254 || $next->isDelimiter(['+', '-'])
Chris@0 255 ) {
Chris@0 256 $arguments[] = $next;
Chris@17 257 } elseif ($next->isDelimiter([')'])) {
Chris@0 258 break;
Chris@0 259 } else {
Chris@0 260 throw SyntaxErrorException::unexpectedToken('an argument', $next);
Chris@0 261 }
Chris@0 262 }
Chris@0 263
Chris@0 264 if (empty($arguments)) {
Chris@0 265 throw SyntaxErrorException::unexpectedToken('at least one argument', $next);
Chris@0 266 }
Chris@0 267
Chris@0 268 $result = new Node\FunctionNode($result, $identifier, $arguments);
Chris@0 269 }
Chris@0 270 } else {
Chris@0 271 throw SyntaxErrorException::unexpectedToken('selector', $peek);
Chris@0 272 }
Chris@0 273 }
Chris@0 274
Chris@17 275 if (\count($stream->getUsed()) === $selectorStart) {
Chris@0 276 throw SyntaxErrorException::unexpectedToken('selector', $stream->getPeek());
Chris@0 277 }
Chris@0 278
Chris@17 279 return [$result, $pseudoElement];
Chris@0 280 }
Chris@0 281
Chris@0 282 /**
Chris@0 283 * Parses next element node.
Chris@0 284 *
Chris@0 285 * @return Node\ElementNode
Chris@0 286 */
Chris@0 287 private function parseElementNode(TokenStream $stream)
Chris@0 288 {
Chris@0 289 $peek = $stream->getPeek();
Chris@0 290
Chris@17 291 if ($peek->isIdentifier() || $peek->isDelimiter(['*'])) {
Chris@0 292 if ($peek->isIdentifier()) {
Chris@0 293 $namespace = $stream->getNext()->getValue();
Chris@0 294 } else {
Chris@0 295 $stream->getNext();
Chris@0 296 $namespace = null;
Chris@0 297 }
Chris@0 298
Chris@17 299 if ($stream->getPeek()->isDelimiter(['|'])) {
Chris@0 300 $stream->getNext();
Chris@0 301 $element = $stream->getNextIdentifierOrStar();
Chris@0 302 } else {
Chris@0 303 $element = $namespace;
Chris@0 304 $namespace = null;
Chris@0 305 }
Chris@0 306 } else {
Chris@0 307 $element = $namespace = null;
Chris@0 308 }
Chris@0 309
Chris@0 310 return new Node\ElementNode($namespace, $element);
Chris@0 311 }
Chris@0 312
Chris@0 313 /**
Chris@0 314 * Parses next attribute node.
Chris@0 315 *
Chris@0 316 * @return Node\AttributeNode
Chris@0 317 *
Chris@0 318 * @throws SyntaxErrorException
Chris@0 319 */
Chris@0 320 private function parseAttributeNode(Node\NodeInterface $selector, TokenStream $stream)
Chris@0 321 {
Chris@0 322 $stream->skipWhitespace();
Chris@0 323 $attribute = $stream->getNextIdentifierOrStar();
Chris@0 324
Chris@17 325 if (null === $attribute && !$stream->getPeek()->isDelimiter(['|'])) {
Chris@0 326 throw SyntaxErrorException::unexpectedToken('"|"', $stream->getPeek());
Chris@0 327 }
Chris@0 328
Chris@17 329 if ($stream->getPeek()->isDelimiter(['|'])) {
Chris@0 330 $stream->getNext();
Chris@0 331
Chris@17 332 if ($stream->getPeek()->isDelimiter(['='])) {
Chris@0 333 $namespace = null;
Chris@0 334 $stream->getNext();
Chris@0 335 $operator = '|=';
Chris@0 336 } else {
Chris@0 337 $namespace = $attribute;
Chris@0 338 $attribute = $stream->getNextIdentifier();
Chris@0 339 $operator = null;
Chris@0 340 }
Chris@0 341 } else {
Chris@0 342 $namespace = $operator = null;
Chris@0 343 }
Chris@0 344
Chris@0 345 if (null === $operator) {
Chris@0 346 $stream->skipWhitespace();
Chris@0 347 $next = $stream->getNext();
Chris@0 348
Chris@17 349 if ($next->isDelimiter([']'])) {
Chris@0 350 return new Node\AttributeNode($selector, $namespace, $attribute, 'exists', null);
Chris@17 351 } elseif ($next->isDelimiter(['='])) {
Chris@0 352 $operator = '=';
Chris@17 353 } elseif ($next->isDelimiter(['^', '$', '*', '~', '|', '!'])
Chris@17 354 && $stream->getPeek()->isDelimiter(['='])
Chris@0 355 ) {
Chris@0 356 $operator = $next->getValue().'=';
Chris@0 357 $stream->getNext();
Chris@0 358 } else {
Chris@0 359 throw SyntaxErrorException::unexpectedToken('operator', $next);
Chris@0 360 }
Chris@0 361 }
Chris@0 362
Chris@0 363 $stream->skipWhitespace();
Chris@0 364 $value = $stream->getNext();
Chris@0 365
Chris@0 366 if ($value->isNumber()) {
Chris@0 367 // if the value is a number, it's casted into a string
Chris@0 368 $value = new Token(Token::TYPE_STRING, (string) $value->getValue(), $value->getPosition());
Chris@0 369 }
Chris@0 370
Chris@0 371 if (!($value->isIdentifier() || $value->isString())) {
Chris@0 372 throw SyntaxErrorException::unexpectedToken('string or identifier', $value);
Chris@0 373 }
Chris@0 374
Chris@0 375 $stream->skipWhitespace();
Chris@0 376 $next = $stream->getNext();
Chris@0 377
Chris@17 378 if (!$next->isDelimiter([']'])) {
Chris@0 379 throw SyntaxErrorException::unexpectedToken('"]"', $next);
Chris@0 380 }
Chris@0 381
Chris@0 382 return new Node\AttributeNode($selector, $namespace, $attribute, $operator, $value->getValue());
Chris@0 383 }
Chris@0 384 }