Chris@0: <?php
Chris@0: 
Chris@0: /*
Chris@0:  * This file is part of the Symfony package.
Chris@0:  *
Chris@0:  * (c) Fabien Potencier <fabien@symfony.com>
Chris@0:  *
Chris@0:  * For the full copyright and license information, please view the LICENSE
Chris@0:  * file that was distributed with this source code.
Chris@0:  */
Chris@0: 
Chris@0: namespace Symfony\Component\CssSelector\Parser;
Chris@0: 
Chris@0: use Symfony\Component\CssSelector\Exception\SyntaxErrorException;
Chris@0: use Symfony\Component\CssSelector\Node;
Chris@0: use Symfony\Component\CssSelector\Parser\Tokenizer\Tokenizer;
Chris@0: 
Chris@0: /**
Chris@0:  * CSS selector parser.
Chris@0:  *
Chris@0:  * This component is a port of the Python cssselect library,
Chris@0:  * which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
Chris@0:  *
Chris@0:  * @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
Chris@0:  *
Chris@0:  * @internal
Chris@0:  */
Chris@0: class Parser implements ParserInterface
Chris@0: {
Chris@0:     private $tokenizer;
Chris@0: 
Chris@0:     public function __construct(Tokenizer $tokenizer = null)
Chris@0:     {
Chris@0:         $this->tokenizer = $tokenizer ?: new Tokenizer();
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * {@inheritdoc}
Chris@0:      */
Chris@0:     public function parse($source)
Chris@0:     {
Chris@0:         $reader = new Reader($source);
Chris@0:         $stream = $this->tokenizer->tokenize($reader);
Chris@0: 
Chris@0:         return $this->parseSelectorList($stream);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses the arguments for ":nth-child()" and friends.
Chris@0:      *
Chris@0:      * @param Token[] $tokens
Chris@0:      *
Chris@0:      * @return array
Chris@0:      *
Chris@0:      * @throws SyntaxErrorException
Chris@0:      */
Chris@0:     public static function parseSeries(array $tokens)
Chris@0:     {
Chris@0:         foreach ($tokens as $token) {
Chris@0:             if ($token->isString()) {
Chris@0:                 throw SyntaxErrorException::stringAsFunctionArgument();
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         $joined = trim(implode('', array_map(function (Token $token) {
Chris@0:             return $token->getValue();
Chris@0:         }, $tokens)));
Chris@0: 
Chris@0:         $int = function ($string) {
Chris@0:             if (!is_numeric($string)) {
Chris@0:                 throw SyntaxErrorException::stringAsFunctionArgument();
Chris@0:             }
Chris@0: 
Chris@0:             return (int) $string;
Chris@0:         };
Chris@0: 
Chris@0:         switch (true) {
Chris@0:             case 'odd' === $joined:
Chris@17:                 return [2, 1];
Chris@0:             case 'even' === $joined:
Chris@17:                 return [2, 0];
Chris@0:             case 'n' === $joined:
Chris@17:                 return [1, 0];
Chris@0:             case false === strpos($joined, 'n'):
Chris@17:                 return [0, $int($joined)];
Chris@0:         }
Chris@0: 
Chris@0:         $split = explode('n', $joined);
Chris@0:         $first = isset($split[0]) ? $split[0] : null;
Chris@0: 
Chris@17:         return [
Chris@0:             $first ? ('-' === $first || '+' === $first ? $int($first.'1') : $int($first)) : 1,
Chris@0:             isset($split[1]) && $split[1] ? $int($split[1]) : 0,
Chris@17:         ];
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses selector nodes.
Chris@0:      *
Chris@0:      * @return array
Chris@0:      */
Chris@0:     private function parseSelectorList(TokenStream $stream)
Chris@0:     {
Chris@0:         $stream->skipWhitespace();
Chris@17:         $selectors = [];
Chris@0: 
Chris@0:         while (true) {
Chris@0:             $selectors[] = $this->parserSelectorNode($stream);
Chris@0: 
Chris@17:             if ($stream->getPeek()->isDelimiter([','])) {
Chris@0:                 $stream->getNext();
Chris@0:                 $stream->skipWhitespace();
Chris@0:             } else {
Chris@0:                 break;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return $selectors;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses next selector or combined node.
Chris@0:      *
Chris@0:      * @return Node\SelectorNode
Chris@0:      *
Chris@0:      * @throws SyntaxErrorException
Chris@0:      */
Chris@0:     private function parserSelectorNode(TokenStream $stream)
Chris@0:     {
Chris@0:         list($result, $pseudoElement) = $this->parseSimpleSelector($stream);
Chris@0: 
Chris@0:         while (true) {
Chris@0:             $stream->skipWhitespace();
Chris@0:             $peek = $stream->getPeek();
Chris@0: 
Chris@17:             if ($peek->isFileEnd() || $peek->isDelimiter([','])) {
Chris@0:                 break;
Chris@0:             }
Chris@0: 
Chris@0:             if (null !== $pseudoElement) {
Chris@0:                 throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector');
Chris@0:             }
Chris@0: 
Chris@17:             if ($peek->isDelimiter(['+', '>', '~'])) {
Chris@0:                 $combinator = $stream->getNext()->getValue();
Chris@0:                 $stream->skipWhitespace();
Chris@0:             } else {
Chris@0:                 $combinator = ' ';
Chris@0:             }
Chris@0: 
Chris@0:             list($nextSelector, $pseudoElement) = $this->parseSimpleSelector($stream);
Chris@0:             $result = new Node\CombinedSelectorNode($result, $combinator, $nextSelector);
Chris@0:         }
Chris@0: 
Chris@0:         return new Node\SelectorNode($result, $pseudoElement);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses next simple node (hash, class, pseudo, negation).
Chris@0:      *
Chris@0:      * @param TokenStream $stream
Chris@0:      * @param bool        $insideNegation
Chris@0:      *
Chris@0:      * @return array
Chris@0:      *
Chris@0:      * @throws SyntaxErrorException
Chris@0:      */
Chris@0:     private function parseSimpleSelector(TokenStream $stream, $insideNegation = false)
Chris@0:     {
Chris@0:         $stream->skipWhitespace();
Chris@0: 
Chris@17:         $selectorStart = \count($stream->getUsed());
Chris@0:         $result = $this->parseElementNode($stream);
Chris@0:         $pseudoElement = null;
Chris@0: 
Chris@0:         while (true) {
Chris@0:             $peek = $stream->getPeek();
Chris@0:             if ($peek->isWhitespace()
Chris@0:                 || $peek->isFileEnd()
Chris@17:                 || $peek->isDelimiter([',', '+', '>', '~'])
Chris@17:                 || ($insideNegation && $peek->isDelimiter([')']))
Chris@0:             ) {
Chris@0:                 break;
Chris@0:             }
Chris@0: 
Chris@0:             if (null !== $pseudoElement) {
Chris@0:                 throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector');
Chris@0:             }
Chris@0: 
Chris@0:             if ($peek->isHash()) {
Chris@0:                 $result = new Node\HashNode($result, $stream->getNext()->getValue());
Chris@17:             } elseif ($peek->isDelimiter(['.'])) {
Chris@0:                 $stream->getNext();
Chris@0:                 $result = new Node\ClassNode($result, $stream->getNextIdentifier());
Chris@17:             } elseif ($peek->isDelimiter(['['])) {
Chris@0:                 $stream->getNext();
Chris@0:                 $result = $this->parseAttributeNode($result, $stream);
Chris@17:             } elseif ($peek->isDelimiter([':'])) {
Chris@0:                 $stream->getNext();
Chris@0: 
Chris@17:                 if ($stream->getPeek()->isDelimiter([':'])) {
Chris@0:                     $stream->getNext();
Chris@0:                     $pseudoElement = $stream->getNextIdentifier();
Chris@0: 
Chris@0:                     continue;
Chris@0:                 }
Chris@0: 
Chris@0:                 $identifier = $stream->getNextIdentifier();
Chris@17:                 if (\in_array(strtolower($identifier), ['first-line', 'first-letter', 'before', 'after'])) {
Chris@0:                     // Special case: CSS 2.1 pseudo-elements can have a single ':'.
Chris@0:                     // Any new pseudo-element must have two.
Chris@0:                     $pseudoElement = $identifier;
Chris@0: 
Chris@0:                     continue;
Chris@0:                 }
Chris@0: 
Chris@17:                 if (!$stream->getPeek()->isDelimiter(['('])) {
Chris@0:                     $result = new Node\PseudoNode($result, $identifier);
Chris@0: 
Chris@0:                     continue;
Chris@0:                 }
Chris@0: 
Chris@0:                 $stream->getNext();
Chris@0:                 $stream->skipWhitespace();
Chris@0: 
Chris@0:                 if ('not' === strtolower($identifier)) {
Chris@0:                     if ($insideNegation) {
Chris@0:                         throw SyntaxErrorException::nestedNot();
Chris@0:                     }
Chris@0: 
Chris@0:                     list($argument, $argumentPseudoElement) = $this->parseSimpleSelector($stream, true);
Chris@0:                     $next = $stream->getNext();
Chris@0: 
Chris@0:                     if (null !== $argumentPseudoElement) {
Chris@0:                         throw SyntaxErrorException::pseudoElementFound($argumentPseudoElement, 'inside ::not()');
Chris@0:                     }
Chris@0: 
Chris@17:                     if (!$next->isDelimiter([')'])) {
Chris@0:                         throw SyntaxErrorException::unexpectedToken('")"', $next);
Chris@0:                     }
Chris@0: 
Chris@0:                     $result = new Node\NegationNode($result, $argument);
Chris@0:                 } else {
Chris@17:                     $arguments = [];
Chris@0:                     $next = null;
Chris@0: 
Chris@0:                     while (true) {
Chris@0:                         $stream->skipWhitespace();
Chris@0:                         $next = $stream->getNext();
Chris@0: 
Chris@0:                         if ($next->isIdentifier()
Chris@0:                             || $next->isString()
Chris@0:                             || $next->isNumber()
Chris@17:                             || $next->isDelimiter(['+', '-'])
Chris@0:                         ) {
Chris@0:                             $arguments[] = $next;
Chris@17:                         } elseif ($next->isDelimiter([')'])) {
Chris@0:                             break;
Chris@0:                         } else {
Chris@0:                             throw SyntaxErrorException::unexpectedToken('an argument', $next);
Chris@0:                         }
Chris@0:                     }
Chris@0: 
Chris@0:                     if (empty($arguments)) {
Chris@0:                         throw SyntaxErrorException::unexpectedToken('at least one argument', $next);
Chris@0:                     }
Chris@0: 
Chris@0:                     $result = new Node\FunctionNode($result, $identifier, $arguments);
Chris@0:                 }
Chris@0:             } else {
Chris@0:                 throw SyntaxErrorException::unexpectedToken('selector', $peek);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@17:         if (\count($stream->getUsed()) === $selectorStart) {
Chris@0:             throw SyntaxErrorException::unexpectedToken('selector', $stream->getPeek());
Chris@0:         }
Chris@0: 
Chris@17:         return [$result, $pseudoElement];
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses next element node.
Chris@0:      *
Chris@0:      * @return Node\ElementNode
Chris@0:      */
Chris@0:     private function parseElementNode(TokenStream $stream)
Chris@0:     {
Chris@0:         $peek = $stream->getPeek();
Chris@0: 
Chris@17:         if ($peek->isIdentifier() || $peek->isDelimiter(['*'])) {
Chris@0:             if ($peek->isIdentifier()) {
Chris@0:                 $namespace = $stream->getNext()->getValue();
Chris@0:             } else {
Chris@0:                 $stream->getNext();
Chris@0:                 $namespace = null;
Chris@0:             }
Chris@0: 
Chris@17:             if ($stream->getPeek()->isDelimiter(['|'])) {
Chris@0:                 $stream->getNext();
Chris@0:                 $element = $stream->getNextIdentifierOrStar();
Chris@0:             } else {
Chris@0:                 $element = $namespace;
Chris@0:                 $namespace = null;
Chris@0:             }
Chris@0:         } else {
Chris@0:             $element = $namespace = null;
Chris@0:         }
Chris@0: 
Chris@0:         return new Node\ElementNode($namespace, $element);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Parses next attribute node.
Chris@0:      *
Chris@0:      * @return Node\AttributeNode
Chris@0:      *
Chris@0:      * @throws SyntaxErrorException
Chris@0:      */
Chris@0:     private function parseAttributeNode(Node\NodeInterface $selector, TokenStream $stream)
Chris@0:     {
Chris@0:         $stream->skipWhitespace();
Chris@0:         $attribute = $stream->getNextIdentifierOrStar();
Chris@0: 
Chris@17:         if (null === $attribute && !$stream->getPeek()->isDelimiter(['|'])) {
Chris@0:             throw SyntaxErrorException::unexpectedToken('"|"', $stream->getPeek());
Chris@0:         }
Chris@0: 
Chris@17:         if ($stream->getPeek()->isDelimiter(['|'])) {
Chris@0:             $stream->getNext();
Chris@0: 
Chris@17:             if ($stream->getPeek()->isDelimiter(['='])) {
Chris@0:                 $namespace = null;
Chris@0:                 $stream->getNext();
Chris@0:                 $operator = '|=';
Chris@0:             } else {
Chris@0:                 $namespace = $attribute;
Chris@0:                 $attribute = $stream->getNextIdentifier();
Chris@0:                 $operator = null;
Chris@0:             }
Chris@0:         } else {
Chris@0:             $namespace = $operator = null;
Chris@0:         }
Chris@0: 
Chris@0:         if (null === $operator) {
Chris@0:             $stream->skipWhitespace();
Chris@0:             $next = $stream->getNext();
Chris@0: 
Chris@17:             if ($next->isDelimiter([']'])) {
Chris@0:                 return new Node\AttributeNode($selector, $namespace, $attribute, 'exists', null);
Chris@17:             } elseif ($next->isDelimiter(['='])) {
Chris@0:                 $operator = '=';
Chris@17:             } elseif ($next->isDelimiter(['^', '$', '*', '~', '|', '!'])
Chris@17:                 && $stream->getPeek()->isDelimiter(['='])
Chris@0:             ) {
Chris@0:                 $operator = $next->getValue().'=';
Chris@0:                 $stream->getNext();
Chris@0:             } else {
Chris@0:                 throw SyntaxErrorException::unexpectedToken('operator', $next);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         $stream->skipWhitespace();
Chris@0:         $value = $stream->getNext();
Chris@0: 
Chris@0:         if ($value->isNumber()) {
Chris@0:             // if the value is a number, it's casted into a string
Chris@0:             $value = new Token(Token::TYPE_STRING, (string) $value->getValue(), $value->getPosition());
Chris@0:         }
Chris@0: 
Chris@0:         if (!($value->isIdentifier() || $value->isString())) {
Chris@0:             throw SyntaxErrorException::unexpectedToken('string or identifier', $value);
Chris@0:         }
Chris@0: 
Chris@0:         $stream->skipWhitespace();
Chris@0:         $next = $stream->getNext();
Chris@0: 
Chris@17:         if (!$next->isDelimiter([']'])) {
Chris@0:             throw SyntaxErrorException::unexpectedToken('"]"', $next);
Chris@0:         }
Chris@0: 
Chris@0:         return new Node\AttributeNode($selector, $namespace, $attribute, $operator, $value->getValue());
Chris@0:     }
Chris@0: }