resolved conflicts

This commit is contained in:
2025-03-31 03:46:05 +03:00
6454 changed files with 1520539 additions and 9 deletions
@@ -0,0 +1,29 @@
CHANGELOG
=========
7.1
---
* Add support for `:is()`
* Add support for `:where()`
6.3
---
* Add support for `:scope`
4.4.0
-----
* Added support for `*:only-of-type`
2.8.0
-----
* Added the `CssSelectorConverter` class as a non-static API for the component.
* Deprecated the `CssSelector` static API of the component.
2.1.0
-----
* none
@@ -0,0 +1,55 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Exception;
use Symfony\Component\CssSelector\Parser\Token;
/**
* ParseException is thrown when a CSS selector syntax is not valid.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*/
class SyntaxErrorException extends ParseException
{
public static function unexpectedToken(string $expectedValue, Token $foundToken): self
{
return new self(\sprintf('Expected %s, but %s found.', $expectedValue, $foundToken));
}
public static function pseudoElementFound(string $pseudoElement, string $unexpectedLocation): self
{
return new self(\sprintf('Unexpected pseudo-element "::%s" found %s.', $pseudoElement, $unexpectedLocation));
}
public static function unclosedString(int $position): self
{
return new self(\sprintf('Unclosed/invalid string at %s.', $position));
}
public static function nestedNot(): self
{
return new self('Got nested ::not().');
}
public static function notAtTheStartOfASelector(string $pseudoElement): self
{
return new self(\sprintf('Got immediate child pseudo-element ":%s" not at the start of a selector', $pseudoElement));
}
public static function stringAsFunctionArgument(): self
{
return new self('String not allowed as function argument.');
}
}
@@ -0,0 +1,73 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a "<selector>[<namespace>|<attribute> <operator> <value>]" node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class AttributeNode extends AbstractNode
{
public function __construct(
private NodeInterface $selector,
private ?string $namespace,
private string $attribute,
private string $operator,
private ?string $value,
) {
}
public function getSelector(): NodeInterface
{
return $this->selector;
}
public function getNamespace(): ?string
{
return $this->namespace;
}
public function getAttribute(): string
{
return $this->attribute;
}
public function getOperator(): string
{
return $this->operator;
}
public function getValue(): ?string
{
return $this->value;
}
public function getSpecificity(): Specificity
{
return $this->selector->getSpecificity()->plus(new Specificity(0, 1, 0));
}
public function __toString(): string
{
$attribute = $this->namespace ? $this->namespace.'|'.$this->attribute : $this->attribute;
return 'exists' === $this->operator
? \sprintf('%s[%s[%s]]', $this->getNodeName(), $this->selector, $attribute)
: \sprintf("%s[%s[%s %s '%s']]", $this->getNodeName(), $this->selector, $attribute, $this->operator, $this->value);
}
}
@@ -0,0 +1,51 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a "<selector>.<name>" node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class ClassNode extends AbstractNode
{
public function __construct(
private NodeInterface $selector,
private string $name,
) {
}
public function getSelector(): NodeInterface
{
return $this->selector;
}
public function getName(): string
{
return $this->name;
}
public function getSpecificity(): Specificity
{
return $this->selector->getSpecificity()->plus(new Specificity(0, 1, 0));
}
public function __toString(): string
{
return \sprintf('%s[%s.%s]', $this->getNodeName(), $this->selector, $this->name);
}
}
@@ -0,0 +1,59 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a combined node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class CombinedSelectorNode extends AbstractNode
{
public function __construct(
private NodeInterface $selector,
private string $combinator,
private NodeInterface $subSelector,
) {
}
public function getSelector(): NodeInterface
{
return $this->selector;
}
public function getCombinator(): string
{
return $this->combinator;
}
public function getSubSelector(): NodeInterface
{
return $this->subSelector;
}
public function getSpecificity(): Specificity
{
return $this->selector->getSpecificity()->plus($this->subSelector->getSpecificity());
}
public function __toString(): string
{
$combinator = ' ' === $this->combinator ? '<followed>' : $this->combinator;
return \sprintf('%s[%s %s %s]', $this->getNodeName(), $this->selector, $combinator, $this->subSelector);
}
}
@@ -0,0 +1,53 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a "<namespace>|<element>" node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class ElementNode extends AbstractNode
{
public function __construct(
private ?string $namespace = null,
private ?string $element = null,
) {
}
public function getNamespace(): ?string
{
return $this->namespace;
}
public function getElement(): ?string
{
return $this->element;
}
public function getSpecificity(): Specificity
{
return new Specificity(0, 0, $this->element ? 1 : 0);
}
public function __toString(): string
{
$element = $this->element ?: '*';
return \sprintf('%s[%s]', $this->getNodeName(), $this->namespace ? $this->namespace.'|'.$element : $element);
}
}
@@ -0,0 +1,70 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
use Symfony\Component\CssSelector\Parser\Token;
/**
* Represents a "<selector>:<name>(<arguments>)" node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class FunctionNode extends AbstractNode
{
private string $name;
/**
* @param Token[] $arguments
*/
public function __construct(
private NodeInterface $selector,
string $name,
private array $arguments = [],
) {
$this->name = strtolower($name);
}
public function getSelector(): NodeInterface
{
return $this->selector;
}
public function getName(): string
{
return $this->name;
}
/**
* @return Token[]
*/
public function getArguments(): array
{
return $this->arguments;
}
public function getSpecificity(): Specificity
{
return $this->selector->getSpecificity()->plus(new Specificity(0, 1, 0));
}
public function __toString(): string
{
$arguments = implode(', ', array_map(fn (Token $token) => "'".$token->getValue()."'", $this->arguments));
return \sprintf('%s[%s:%s(%s)]', $this->getNodeName(), $this->selector, $this->name, $arguments ? '['.$arguments.']' : '');
}
}
@@ -0,0 +1,51 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a "<selector>#<id>" node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class HashNode extends AbstractNode
{
public function __construct(
private NodeInterface $selector,
private string $id,
) {
}
public function getSelector(): NodeInterface
{
return $this->selector;
}
public function getId(): string
{
return $this->id;
}
public function getSpecificity(): Specificity
{
return $this->selector->getSpecificity()->plus(new Specificity(1, 0, 0));
}
public function __toString(): string
{
return \sprintf('%s[%s#%s]', $this->getNodeName(), $this->selector, $this->id);
}
}
@@ -0,0 +1,51 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a "<selector>:not(<identifier>)" node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class NegationNode extends AbstractNode
{
public function __construct(
private NodeInterface $selector,
private NodeInterface $subSelector,
) {
}
public function getSelector(): NodeInterface
{
return $this->selector;
}
public function getSubSelector(): NodeInterface
{
return $this->subSelector;
}
public function getSpecificity(): Specificity
{
return $this->selector->getSpecificity()->plus($this->subSelector->getSpecificity());
}
public function __toString(): string
{
return \sprintf('%s[%s:not(%s)]', $this->getNodeName(), $this->selector, $this->subSelector);
}
}
@@ -0,0 +1,54 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a "<selector>:<identifier>" node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class PseudoNode extends AbstractNode
{
private string $identifier;
public function __construct(
private NodeInterface $selector,
string $identifier,
) {
$this->identifier = strtolower($identifier);
}
public function getSelector(): NodeInterface
{
return $this->selector;
}
public function getIdentifier(): string
{
return $this->identifier;
}
public function getSpecificity(): Specificity
{
return $this->selector->getSpecificity()->plus(new Specificity(0, 1, 0));
}
public function __toString(): string
{
return \sprintf('%s[%s:%s]', $this->getNodeName(), $this->selector, $this->identifier);
}
}
@@ -0,0 +1,54 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a "<selector>(::|:)<pseudoElement>" node.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class SelectorNode extends AbstractNode
{
private ?string $pseudoElement;
public function __construct(
private NodeInterface $tree,
?string $pseudoElement = null,
) {
$this->pseudoElement = $pseudoElement ? strtolower($pseudoElement) : null;
}
public function getTree(): NodeInterface
{
return $this->tree;
}
public function getPseudoElement(): ?string
{
return $this->pseudoElement;
}
public function getSpecificity(): Specificity
{
return $this->tree->getSpecificity()->plus(new Specificity(0, 0, $this->pseudoElement ? 1 : 0));
}
public function __toString(): string
{
return \sprintf('%s[%s%s]', $this->getNodeName(), $this->tree, $this->pseudoElement ? '::'.$this->pseudoElement : '');
}
}
@@ -0,0 +1,69 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Node;
/**
* Represents a node specificity.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @see http://www.w3.org/TR/selectors/#specificity
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class Specificity
{
public const A_FACTOR = 100;
public const B_FACTOR = 10;
public const C_FACTOR = 1;
public function __construct(
private int $a,
private int $b,
private int $c,
) {
}
public function plus(self $specificity): self
{
return new self($this->a + $specificity->a, $this->b + $specificity->b, $this->c + $specificity->c);
}
public function getValue(): int
{
return $this->a * self::A_FACTOR + $this->b * self::B_FACTOR + $this->c * self::C_FACTOR;
}
/**
* Returns -1 if the object specificity is lower than the argument,
* 0 if they are equal, and 1 if the argument is lower.
*/
public function compareTo(self $specificity): int
{
if ($this->a !== $specificity->a) {
return $this->a > $specificity->a ? 1 : -1;
}
if ($this->b !== $specificity->b) {
return $this->b > $specificity->b ? 1 : -1;
}
if ($this->c !== $specificity->c) {
return $this->c > $specificity->c ? 1 : -1;
}
return 0;
}
}
@@ -0,0 +1,52 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser\Handler;
use Symfony\Component\CssSelector\Parser\Reader;
use Symfony\Component\CssSelector\Parser\Token;
use Symfony\Component\CssSelector\Parser\Tokenizer\TokenizerEscaping;
use Symfony\Component\CssSelector\Parser\Tokenizer\TokenizerPatterns;
use Symfony\Component\CssSelector\Parser\TokenStream;
/**
* CSS selector comment handler.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class HashHandler implements HandlerInterface
{
public function __construct(
private TokenizerPatterns $patterns,
private TokenizerEscaping $escaping,
) {
}
public function handle(Reader $reader, TokenStream $stream): bool
{
$match = $reader->findPattern($this->patterns->getHashPattern());
if (!$match) {
return false;
}
$value = $this->escaping->escapeUnicode($match[1]);
$stream->push(new Token(Token::TYPE_HASH, $value, $reader->getPosition()));
$reader->moveForward(\strlen($match[0]));
return true;
}
}
@@ -0,0 +1,52 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser\Handler;
use Symfony\Component\CssSelector\Parser\Reader;
use Symfony\Component\CssSelector\Parser\Token;
use Symfony\Component\CssSelector\Parser\Tokenizer\TokenizerEscaping;
use Symfony\Component\CssSelector\Parser\Tokenizer\TokenizerPatterns;
use Symfony\Component\CssSelector\Parser\TokenStream;
/**
* CSS selector comment handler.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class IdentifierHandler implements HandlerInterface
{
public function __construct(
private TokenizerPatterns $patterns,
private TokenizerEscaping $escaping,
) {
}
public function handle(Reader $reader, TokenStream $stream): bool
{
$match = $reader->findPattern($this->patterns->getIdentifierPattern());
if (!$match) {
return false;
}
$value = $this->escaping->escapeUnicode($match[0]);
$stream->push(new Token(Token::TYPE_IDENTIFIER, $value, $reader->getPosition()));
$reader->moveForward(\strlen($match[0]));
return true;
}
}
@@ -0,0 +1,49 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser\Handler;
use Symfony\Component\CssSelector\Parser\Reader;
use Symfony\Component\CssSelector\Parser\Token;
use Symfony\Component\CssSelector\Parser\Tokenizer\TokenizerPatterns;
use Symfony\Component\CssSelector\Parser\TokenStream;
/**
* CSS selector comment handler.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class NumberHandler implements HandlerInterface
{
public function __construct(
private TokenizerPatterns $patterns,
) {
}
public function handle(Reader $reader, TokenStream $stream): bool
{
$match = $reader->findPattern($this->patterns->getNumberPattern());
if (!$match) {
return false;
}
$stream->push(new Token(Token::TYPE_NUMBER, $match[0], $reader->getPosition()));
$reader->moveForward(\strlen($match[0]));
return true;
}
}
@@ -0,0 +1,71 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser\Handler;
use Symfony\Component\CssSelector\Exception\InternalErrorException;
use Symfony\Component\CssSelector\Exception\SyntaxErrorException;
use Symfony\Component\CssSelector\Parser\Reader;
use Symfony\Component\CssSelector\Parser\Token;
use Symfony\Component\CssSelector\Parser\Tokenizer\TokenizerEscaping;
use Symfony\Component\CssSelector\Parser\Tokenizer\TokenizerPatterns;
use Symfony\Component\CssSelector\Parser\TokenStream;
/**
* CSS selector comment handler.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class StringHandler implements HandlerInterface
{
public function __construct(
private TokenizerPatterns $patterns,
private TokenizerEscaping $escaping,
) {
}
public function handle(Reader $reader, TokenStream $stream): bool
{
$quote = $reader->getSubstring(1);
if (!\in_array($quote, ["'", '"'])) {
return false;
}
$reader->moveForward(1);
$match = $reader->findPattern($this->patterns->getQuotedStringPattern($quote));
if (!$match) {
throw new InternalErrorException(\sprintf('Should have found at least an empty match at %d.', $reader->getPosition()));
}
// check unclosed strings
if (\strlen($match[0]) === $reader->getRemainingLength()) {
throw SyntaxErrorException::unclosedString($reader->getPosition() - 1);
}
// check quotes pairs validity
if ($quote !== $reader->getSubstring(1, \strlen($match[0]))) {
throw SyntaxErrorException::unclosedString($reader->getPosition() - 1);
}
$string = $this->escaping->escapeUnicodeAndNewLine($match[0]);
$stream->push(new Token(Token::TYPE_STRING, $string, $reader->getPosition()));
$reader->moveForward(\strlen($match[0]) + 1);
return true;
}
}
@@ -0,0 +1,385 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser;
use Symfony\Component\CssSelector\Exception\SyntaxErrorException;
use Symfony\Component\CssSelector\Node;
use Symfony\Component\CssSelector\Parser\Tokenizer\Tokenizer;
/**
* CSS selector parser.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/scrapy/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class Parser implements ParserInterface
{
private Tokenizer $tokenizer;
public function __construct(?Tokenizer $tokenizer = null)
{
$this->tokenizer = $tokenizer ?? new Tokenizer();
}
public function parse(string $source): array
{
$reader = new Reader($source);
$stream = $this->tokenizer->tokenize($reader);
return $this->parseSelectorList($stream);
}
/**
* Parses the arguments for ":nth-child()" and friends.
*
* @param Token[] $tokens
*
* @throws SyntaxErrorException
*/
public static function parseSeries(array $tokens): array
{
foreach ($tokens as $token) {
if ($token->isString()) {
throw SyntaxErrorException::stringAsFunctionArgument();
}
}
$joined = trim(implode('', array_map(fn (Token $token) => $token->getValue(), $tokens)));
$int = function ($string) {
if (!is_numeric($string)) {
throw SyntaxErrorException::stringAsFunctionArgument();
}
return (int) $string;
};
switch (true) {
case 'odd' === $joined:
return [2, 1];
case 'even' === $joined:
return [2, 0];
case 'n' === $joined:
return [1, 0];
case !str_contains($joined, 'n'):
return [0, $int($joined)];
}
$split = explode('n', $joined);
$first = $split[0] ?? null;
return [
$first ? ('-' === $first || '+' === $first ? $int($first.'1') : $int($first)) : 1,
isset($split[1]) && $split[1] ? $int($split[1]) : 0,
];
}
private function parseSelectorList(TokenStream $stream, bool $isArgument = false): array
{
$stream->skipWhitespace();
$selectors = [];
while (true) {
if ($isArgument && $stream->getPeek()->isDelimiter([')'])) {
break;
}
$selectors[] = $this->parserSelectorNode($stream, $isArgument);
if ($stream->getPeek()->isDelimiter([','])) {
$stream->getNext();
$stream->skipWhitespace();
} else {
break;
}
}
return $selectors;
}
private function parserSelectorNode(TokenStream $stream, bool $isArgument = false): Node\SelectorNode
{
[$result, $pseudoElement] = $this->parseSimpleSelector($stream, false, $isArgument);
while (true) {
$stream->skipWhitespace();
$peek = $stream->getPeek();
if (
$peek->isFileEnd()
|| $peek->isDelimiter([','])
|| ($isArgument && $peek->isDelimiter([')']))
) {
break;
}
if (null !== $pseudoElement) {
throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector');
}
if ($peek->isDelimiter(['+', '>', '~'])) {
$combinator = $stream->getNext()->getValue();
$stream->skipWhitespace();
} else {
$combinator = ' ';
}
[$nextSelector, $pseudoElement] = $this->parseSimpleSelector($stream, false, $isArgument);
$result = new Node\CombinedSelectorNode($result, $combinator, $nextSelector);
}
return new Node\SelectorNode($result, $pseudoElement);
}
/**
* Parses next simple node (hash, class, pseudo, negation).
*
* @throws SyntaxErrorException
*/
private function parseSimpleSelector(TokenStream $stream, bool $insideNegation = false, bool $isArgument = false): array
{
$stream->skipWhitespace();
$selectorStart = \count($stream->getUsed());
$result = $this->parseElementNode($stream);
$pseudoElement = null;
while (true) {
$peek = $stream->getPeek();
if ($peek->isWhitespace()
|| $peek->isFileEnd()
|| $peek->isDelimiter([',', '+', '>', '~'])
|| ($isArgument && $peek->isDelimiter([')']))
) {
break;
}
if (null !== $pseudoElement) {
throw SyntaxErrorException::pseudoElementFound($pseudoElement, 'not at the end of a selector');
}
if ($peek->isHash()) {
$result = new Node\HashNode($result, $stream->getNext()->getValue());
} elseif ($peek->isDelimiter(['.'])) {
$stream->getNext();
$result = new Node\ClassNode($result, $stream->getNextIdentifier());
} elseif ($peek->isDelimiter(['['])) {
$stream->getNext();
$result = $this->parseAttributeNode($result, $stream);
} elseif ($peek->isDelimiter([':'])) {
$stream->getNext();
if ($stream->getPeek()->isDelimiter([':'])) {
$stream->getNext();
$pseudoElement = $stream->getNextIdentifier();
continue;
}
$identifier = $stream->getNextIdentifier();
if (\in_array(strtolower($identifier), ['first-line', 'first-letter', 'before', 'after'])) {
// Special case: CSS 2.1 pseudo-elements can have a single ':'.
// Any new pseudo-element must have two.
$pseudoElement = $identifier;
continue;
}
if (!$stream->getPeek()->isDelimiter(['('])) {
$result = new Node\PseudoNode($result, $identifier);
if ('Pseudo[Element[*]:scope]' === $result->__toString()) {
$used = \count($stream->getUsed());
if (!(2 === $used
|| 3 === $used && $stream->getUsed()[0]->isWhiteSpace()
|| $used >= 3 && $stream->getUsed()[$used - 3]->isDelimiter([','])
|| $used >= 4
&& $stream->getUsed()[$used - 3]->isWhiteSpace()
&& $stream->getUsed()[$used - 4]->isDelimiter([','])
)) {
throw SyntaxErrorException::notAtTheStartOfASelector('scope');
}
}
continue;
}
$stream->getNext();
$stream->skipWhitespace();
if ('not' === strtolower($identifier)) {
if ($insideNegation) {
throw SyntaxErrorException::nestedNot();
}
[$argument, $argumentPseudoElement] = $this->parseSimpleSelector($stream, true, true);
$next = $stream->getNext();
if (null !== $argumentPseudoElement) {
throw SyntaxErrorException::pseudoElementFound($argumentPseudoElement, 'inside ::not()');
}
if (!$next->isDelimiter([')'])) {
throw SyntaxErrorException::unexpectedToken('")"', $next);
}
$result = new Node\NegationNode($result, $argument);
} elseif ('is' === strtolower($identifier)) {
$selectors = $this->parseSelectorList($stream, true);
$next = $stream->getNext();
if (!$next->isDelimiter([')'])) {
throw SyntaxErrorException::unexpectedToken('")"', $next);
}
$result = new Node\MatchingNode($result, $selectors);
} elseif ('where' === strtolower($identifier)) {
$selectors = $this->parseSelectorList($stream, true);
$next = $stream->getNext();
if (!$next->isDelimiter([')'])) {
throw SyntaxErrorException::unexpectedToken('")"', $next);
}
$result = new Node\SpecificityAdjustmentNode($result, $selectors);
} else {
$arguments = [];
$next = null;
while (true) {
$stream->skipWhitespace();
$next = $stream->getNext();
if ($next->isIdentifier()
|| $next->isString()
|| $next->isNumber()
|| $next->isDelimiter(['+', '-'])
) {
$arguments[] = $next;
} elseif ($next->isDelimiter([')'])) {
break;
} else {
throw SyntaxErrorException::unexpectedToken('an argument', $next);
}
}
if (!$arguments) {
throw SyntaxErrorException::unexpectedToken('at least one argument', $next);
}
$result = new Node\FunctionNode($result, $identifier, $arguments);
}
} else {
throw SyntaxErrorException::unexpectedToken('selector', $peek);
}
}
if (\count($stream->getUsed()) === $selectorStart) {
throw SyntaxErrorException::unexpectedToken('selector', $stream->getPeek());
}
return [$result, $pseudoElement];
}
private function parseElementNode(TokenStream $stream): Node\ElementNode
{
$peek = $stream->getPeek();
if ($peek->isIdentifier() || $peek->isDelimiter(['*'])) {
if ($peek->isIdentifier()) {
$namespace = $stream->getNext()->getValue();
} else {
$stream->getNext();
$namespace = null;
}
if ($stream->getPeek()->isDelimiter(['|'])) {
$stream->getNext();
$element = $stream->getNextIdentifierOrStar();
} else {
$element = $namespace;
$namespace = null;
}
} else {
$element = $namespace = null;
}
return new Node\ElementNode($namespace, $element);
}
private function parseAttributeNode(Node\NodeInterface $selector, TokenStream $stream): Node\AttributeNode
{
$stream->skipWhitespace();
$attribute = $stream->getNextIdentifierOrStar();
if (null === $attribute && !$stream->getPeek()->isDelimiter(['|'])) {
throw SyntaxErrorException::unexpectedToken('"|"', $stream->getPeek());
}
if ($stream->getPeek()->isDelimiter(['|'])) {
$stream->getNext();
if ($stream->getPeek()->isDelimiter(['='])) {
$namespace = null;
$stream->getNext();
$operator = '|=';
} else {
$namespace = $attribute;
$attribute = $stream->getNextIdentifier();
$operator = null;
}
} else {
$namespace = $operator = null;
}
if (null === $operator) {
$stream->skipWhitespace();
$next = $stream->getNext();
if ($next->isDelimiter([']'])) {
return new Node\AttributeNode($selector, $namespace, $attribute, 'exists', null);
} elseif ($next->isDelimiter(['='])) {
$operator = '=';
} elseif ($next->isDelimiter(['^', '$', '*', '~', '|', '!'])
&& $stream->getPeek()->isDelimiter(['='])
) {
$operator = $next->getValue().'=';
$stream->getNext();
} else {
throw SyntaxErrorException::unexpectedToken('operator', $next);
}
}
$stream->skipWhitespace();
$value = $stream->getNext();
if ($value->isNumber()) {
// if the value is a number, it's casted into a string
$value = new Token(Token::TYPE_STRING, (string) $value->getValue(), $value->getPosition());
}
if (!($value->isIdentifier() || $value->isString())) {
throw SyntaxErrorException::unexpectedToken('string or identifier', $value);
}
$stream->skipWhitespace();
$next = $stream->getNext();
if (!$next->isDelimiter([']'])) {
throw SyntaxErrorException::unexpectedToken('"]"', $next);
}
return new Node\AttributeNode($selector, $namespace, $attribute, $operator, $value->getValue());
}
}
@@ -0,0 +1,82 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser;
/**
* CSS selector reader.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class Reader
{
private int $length;
private int $position = 0;
public function __construct(
private string $source,
) {
$this->length = \strlen($source);
}
public function isEOF(): bool
{
return $this->position >= $this->length;
}
public function getPosition(): int
{
return $this->position;
}
public function getRemainingLength(): int
{
return $this->length - $this->position;
}
public function getSubstring(int $length, int $offset = 0): string
{
return substr($this->source, $this->position + $offset, $length);
}
public function getOffset(string $string): int|false
{
$position = strpos($this->source, $string, $this->position);
return false === $position ? false : $position - $this->position;
}
public function findPattern(string $pattern): array|false
{
$source = substr($this->source, $this->position);
if (preg_match($pattern, $source, $matches)) {
return $matches;
}
return false;
}
public function moveForward(int $length): void
{
$this->position += $length;
}
public function moveToEnd(): void
{
$this->position = $this->length;
}
}
@@ -0,0 +1,107 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser;
/**
* CSS selector token.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class Token
{
public const TYPE_FILE_END = 'eof';
public const TYPE_DELIMITER = 'delimiter';
public const TYPE_WHITESPACE = 'whitespace';
public const TYPE_IDENTIFIER = 'identifier';
public const TYPE_HASH = 'hash';
public const TYPE_NUMBER = 'number';
public const TYPE_STRING = 'string';
public function __construct(
private ?string $type,
private ?string $value,
private ?int $position,
) {
}
public function getType(): ?int
{
return $this->type;
}
public function getValue(): ?string
{
return $this->value;
}
public function getPosition(): ?int
{
return $this->position;
}
public function isFileEnd(): bool
{
return self::TYPE_FILE_END === $this->type;
}
public function isDelimiter(array $values = []): bool
{
if (self::TYPE_DELIMITER !== $this->type) {
return false;
}
if (!$values) {
return true;
}
return \in_array($this->value, $values, true);
}
public function isWhitespace(): bool
{
return self::TYPE_WHITESPACE === $this->type;
}
public function isIdentifier(): bool
{
return self::TYPE_IDENTIFIER === $this->type;
}
public function isHash(): bool
{
return self::TYPE_HASH === $this->type;
}
public function isNumber(): bool
{
return self::TYPE_NUMBER === $this->type;
}
public function isString(): bool
{
return self::TYPE_STRING === $this->type;
}
public function __toString(): string
{
if ($this->value) {
return \sprintf('<%s "%s" at %s>', $this->type, $this->value, $this->position);
}
return \sprintf('<%s at %s>', $this->type, $this->position);
}
}
@@ -0,0 +1,63 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser\Tokenizer;
/**
* CSS selector tokenizer escaping applier.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class TokenizerEscaping
{
public function __construct(
private TokenizerPatterns $patterns,
) {
}
public function escapeUnicode(string $value): string
{
$value = $this->replaceUnicodeSequences($value);
return preg_replace($this->patterns->getSimpleEscapePattern(), '$1', $value);
}
public function escapeUnicodeAndNewLine(string $value): string
{
$value = preg_replace($this->patterns->getNewLineEscapePattern(), '', $value);
return $this->escapeUnicode($value);
}
private function replaceUnicodeSequences(string $value): string
{
return preg_replace_callback($this->patterns->getUnicodeEscapePattern(), function ($match) {
$c = hexdec($match[1]);
if (0x80 > $c %= 0x200000) {
return \chr($c);
}
if (0x800 > $c) {
return \chr(0xC0 | $c >> 6).\chr(0x80 | $c & 0x3F);
}
if (0x10000 > $c) {
return \chr(0xE0 | $c >> 12).\chr(0x80 | $c >> 6 & 0x3F).\chr(0x80 | $c & 0x3F);
}
return '';
}, $value);
}
}
@@ -0,0 +1,89 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\Parser\Tokenizer;
/**
* CSS selector tokenizer patterns builder.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class TokenizerPatterns
{
private string $unicodeEscapePattern;
private string $simpleEscapePattern;
private string $newLineEscapePattern;
private string $escapePattern;
private string $stringEscapePattern;
private string $nonAsciiPattern;
private string $nmCharPattern;
private string $nmStartPattern;
private string $identifierPattern;
private string $hashPattern;
private string $numberPattern;
private string $quotedStringPattern;
public function __construct()
{
$this->unicodeEscapePattern = '\\\\([0-9a-f]{1,6})(?:\r\n|[ \n\r\t\f])?';
$this->simpleEscapePattern = '\\\\(.)';
$this->newLineEscapePattern = '\\\\(?:\n|\r\n|\r|\f)';
$this->escapePattern = $this->unicodeEscapePattern.'|\\\\[^\n\r\f0-9a-f]';
$this->stringEscapePattern = $this->newLineEscapePattern.'|'.$this->escapePattern;
$this->nonAsciiPattern = '[^\x00-\x7F]';
$this->nmCharPattern = '[_a-z0-9-]|'.$this->escapePattern.'|'.$this->nonAsciiPattern;
$this->nmStartPattern = '[_a-z]|'.$this->escapePattern.'|'.$this->nonAsciiPattern;
$this->identifierPattern = '-?(?:'.$this->nmStartPattern.')(?:'.$this->nmCharPattern.')*';
$this->hashPattern = '#((?:'.$this->nmCharPattern.')+)';
$this->numberPattern = '[+-]?(?:[0-9]*\.[0-9]+|[0-9]+)';
$this->quotedStringPattern = '([^\n\r\f\\\\%s]|'.$this->stringEscapePattern.')*';
}
public function getNewLineEscapePattern(): string
{
return '~'.$this->newLineEscapePattern.'~';
}
public function getSimpleEscapePattern(): string
{
return '~'.$this->simpleEscapePattern.'~';
}
public function getUnicodeEscapePattern(): string
{
return '~'.$this->unicodeEscapePattern.'~i';
}
public function getIdentifierPattern(): string
{
return '~^'.$this->identifierPattern.'~i';
}
public function getHashPattern(): string
{
return '~^'.$this->hashPattern.'~i';
}
public function getNumberPattern(): string
{
return '~^'.$this->numberPattern.'~';
}
public function getQuotedStringPattern(string $quote): string
{
return '~^'.\sprintf($this->quotedStringPattern, $quote).'~i';
}
}
@@ -0,0 +1,113 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\XPath\Extension;
use Symfony\Component\CssSelector\XPath\Translator;
use Symfony\Component\CssSelector\XPath\XPathExpr;
/**
* XPath expression translator attribute extension.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class AttributeMatchingExtension extends AbstractExtension
{
public function getAttributeMatchingTranslators(): array
{
return [
'exists' => $this->translateExists(...),
'=' => $this->translateEquals(...),
'~=' => $this->translateIncludes(...),
'|=' => $this->translateDashMatch(...),
'^=' => $this->translatePrefixMatch(...),
'$=' => $this->translateSuffixMatch(...),
'*=' => $this->translateSubstringMatch(...),
'!=' => $this->translateDifferent(...),
];
}
public function translateExists(XPathExpr $xpath, string $attribute, ?string $value): XPathExpr
{
return $xpath->addCondition($attribute);
}
public function translateEquals(XPathExpr $xpath, string $attribute, ?string $value): XPathExpr
{
return $xpath->addCondition(\sprintf('%s = %s', $attribute, Translator::getXpathLiteral($value)));
}
public function translateIncludes(XPathExpr $xpath, string $attribute, ?string $value): XPathExpr
{
return $xpath->addCondition($value ? \sprintf(
'%1$s and contains(concat(\' \', normalize-space(%1$s), \' \'), %2$s)',
$attribute,
Translator::getXpathLiteral(' '.$value.' ')
) : '0');
}
public function translateDashMatch(XPathExpr $xpath, string $attribute, ?string $value): XPathExpr
{
return $xpath->addCondition(\sprintf(
'%1$s and (%1$s = %2$s or starts-with(%1$s, %3$s))',
$attribute,
Translator::getXpathLiteral($value),
Translator::getXpathLiteral($value.'-')
));
}
public function translatePrefixMatch(XPathExpr $xpath, string $attribute, ?string $value): XPathExpr
{
return $xpath->addCondition($value ? \sprintf(
'%1$s and starts-with(%1$s, %2$s)',
$attribute,
Translator::getXpathLiteral($value)
) : '0');
}
public function translateSuffixMatch(XPathExpr $xpath, string $attribute, ?string $value): XPathExpr
{
return $xpath->addCondition($value ? \sprintf(
'%1$s and substring(%1$s, string-length(%1$s)-%2$s) = %3$s',
$attribute,
\strlen($value) - 1,
Translator::getXpathLiteral($value)
) : '0');
}
public function translateSubstringMatch(XPathExpr $xpath, string $attribute, ?string $value): XPathExpr
{
return $xpath->addCondition($value ? \sprintf(
'%1$s and contains(%1$s, %2$s)',
$attribute,
Translator::getXpathLiteral($value)
) : '0');
}
public function translateDifferent(XPathExpr $xpath, string $attribute, ?string $value): XPathExpr
{
return $xpath->addCondition(\sprintf(
$value ? 'not(%1$s) or %1$s != %2$s' : '%s != %s',
$attribute,
Translator::getXpathLiteral($value)
));
}
public function getName(): string
{
return 'attribute-matching';
}
}
@@ -0,0 +1,165 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\XPath\Extension;
use Symfony\Component\CssSelector\Exception\ExpressionErrorException;
use Symfony\Component\CssSelector\Exception\SyntaxErrorException;
use Symfony\Component\CssSelector\Node\FunctionNode;
use Symfony\Component\CssSelector\Parser\Parser;
use Symfony\Component\CssSelector\XPath\Translator;
use Symfony\Component\CssSelector\XPath\XPathExpr;
/**
* XPath expression translator function extension.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class FunctionExtension extends AbstractExtension
{
public function getFunctionTranslators(): array
{
return [
'nth-child' => $this->translateNthChild(...),
'nth-last-child' => $this->translateNthLastChild(...),
'nth-of-type' => $this->translateNthOfType(...),
'nth-last-of-type' => $this->translateNthLastOfType(...),
'contains' => $this->translateContains(...),
'lang' => $this->translateLang(...),
];
}
/**
* @throws ExpressionErrorException
*/
public function translateNthChild(XPathExpr $xpath, FunctionNode $function, bool $last = false, bool $addNameTest = true): XPathExpr
{
try {
[$a, $b] = Parser::parseSeries($function->getArguments());
} catch (SyntaxErrorException $e) {
throw new ExpressionErrorException(\sprintf('Invalid series: "%s".', implode('", "', $function->getArguments())), 0, $e);
}
$xpath->addStarPrefix();
if ($addNameTest) {
$xpath->addNameTest();
}
if (0 === $a) {
return $xpath->addCondition('position() = '.($last ? 'last() - '.($b - 1) : $b));
}
if ($a < 0) {
if ($b < 1) {
return $xpath->addCondition('false()');
}
$sign = '<=';
} else {
$sign = '>=';
}
$expr = 'position()';
if ($last) {
$expr = 'last() - '.$expr;
--$b;
}
if (0 !== $b) {
$expr .= ' - '.$b;
}
$conditions = [\sprintf('%s %s 0', $expr, $sign)];
if (1 !== $a && -1 !== $a) {
$conditions[] = \sprintf('(%s) mod %d = 0', $expr, $a);
}
return $xpath->addCondition(implode(' and ', $conditions));
// todo: handle an+b, odd, even
// an+b means every-a, plus b, e.g., 2n+1 means odd
// 0n+b means b
// n+0 means a=1, i.e., all elements
// an means every a elements, i.e., 2n means even
// -n means -1n
// -1n+6 means elements 6 and previous
}
public function translateNthLastChild(XPathExpr $xpath, FunctionNode $function): XPathExpr
{
return $this->translateNthChild($xpath, $function, true);
}
public function translateNthOfType(XPathExpr $xpath, FunctionNode $function): XPathExpr
{
return $this->translateNthChild($xpath, $function, false, false);
}
/**
* @throws ExpressionErrorException
*/
public function translateNthLastOfType(XPathExpr $xpath, FunctionNode $function): XPathExpr
{
if ('*' === $xpath->getElement()) {
throw new ExpressionErrorException('"*:nth-of-type()" is not implemented.');
}
return $this->translateNthChild($xpath, $function, true, false);
}
/**
* @throws ExpressionErrorException
*/
public function translateContains(XPathExpr $xpath, FunctionNode $function): XPathExpr
{
$arguments = $function->getArguments();
foreach ($arguments as $token) {
if (!($token->isString() || $token->isIdentifier())) {
throw new ExpressionErrorException('Expected a single string or identifier for :contains(), got '.implode(', ', $arguments));
}
}
return $xpath->addCondition(\sprintf(
'contains(string(.), %s)',
Translator::getXpathLiteral($arguments[0]->getValue())
));
}
/**
* @throws ExpressionErrorException
*/
public function translateLang(XPathExpr $xpath, FunctionNode $function): XPathExpr
{
$arguments = $function->getArguments();
foreach ($arguments as $token) {
if (!($token->isString() || $token->isIdentifier())) {
throw new ExpressionErrorException('Expected a single string or identifier for :lang(), got '.implode(', ', $arguments));
}
}
return $xpath->addCondition(\sprintf(
'lang(%s)',
Translator::getXpathLiteral($arguments[0]->getValue())
));
}
public function getName(): string
{
return 'function';
}
}
@@ -0,0 +1,178 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\XPath\Extension;
use Symfony\Component\CssSelector\Exception\ExpressionErrorException;
use Symfony\Component\CssSelector\Node\FunctionNode;
use Symfony\Component\CssSelector\XPath\Translator;
use Symfony\Component\CssSelector\XPath\XPathExpr;
/**
* XPath expression translator HTML extension.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class HtmlExtension extends AbstractExtension
{
public function __construct(Translator $translator)
{
$translator
->getExtension('node')
->setFlag(NodeExtension::ELEMENT_NAME_IN_LOWER_CASE, true)
->setFlag(NodeExtension::ATTRIBUTE_NAME_IN_LOWER_CASE, true);
}
public function getPseudoClassTranslators(): array
{
return [
'checked' => $this->translateChecked(...),
'link' => $this->translateLink(...),
'disabled' => $this->translateDisabled(...),
'enabled' => $this->translateEnabled(...),
'selected' => $this->translateSelected(...),
'invalid' => $this->translateInvalid(...),
'hover' => $this->translateHover(...),
'visited' => $this->translateVisited(...),
];
}
public function getFunctionTranslators(): array
{
return [
'lang' => $this->translateLang(...),
];
}
public function translateChecked(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition(
'(@checked '
."and (name(.) = 'input' or name(.) = 'command')"
."and (@type = 'checkbox' or @type = 'radio'))"
);
}
public function translateLink(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition("@href and (name(.) = 'a' or name(.) = 'link' or name(.) = 'area')");
}
public function translateDisabled(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition(
'('
.'@disabled and'
.'('
."(name(.) = 'input' and @type != 'hidden')"
." or name(.) = 'button'"
." or name(.) = 'select'"
." or name(.) = 'textarea'"
." or name(.) = 'command'"
." or name(.) = 'fieldset'"
." or name(.) = 'optgroup'"
." or name(.) = 'option'"
.')'
.') or ('
."(name(.) = 'input' and @type != 'hidden')"
." or name(.) = 'button'"
." or name(.) = 'select'"
." or name(.) = 'textarea'"
.')'
.' and ancestor::fieldset[@disabled]'
);
// todo: in the second half, add "and is not a descendant of that fieldset element's first legend element child, if any."
}
public function translateEnabled(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition(
'('
.'@href and ('
."name(.) = 'a'"
." or name(.) = 'link'"
." or name(.) = 'area'"
.')'
.') or ('
.'('
."name(.) = 'command'"
." or name(.) = 'fieldset'"
." or name(.) = 'optgroup'"
.')'
.' and not(@disabled)'
.') or ('
.'('
."(name(.) = 'input' and @type != 'hidden')"
." or name(.) = 'button'"
." or name(.) = 'select'"
." or name(.) = 'textarea'"
." or name(.) = 'keygen'"
.')'
.' and not (@disabled or ancestor::fieldset[@disabled])'
.') or ('
."name(.) = 'option' and not("
.'@disabled or ancestor::optgroup[@disabled]'
.')'
.')'
);
}
/**
* @throws ExpressionErrorException
*/
public function translateLang(XPathExpr $xpath, FunctionNode $function): XPathExpr
{
$arguments = $function->getArguments();
foreach ($arguments as $token) {
if (!($token->isString() || $token->isIdentifier())) {
throw new ExpressionErrorException('Expected a single string or identifier for :lang(), got '.implode(', ', $arguments));
}
}
return $xpath->addCondition(\sprintf(
'ancestor-or-self::*[@lang][1][starts-with(concat('
."translate(@%s, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz'), '-')"
.', %s)]',
'lang',
Translator::getXpathLiteral(strtolower($arguments[0]->getValue()).'-')
));
}
public function translateSelected(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition("(@selected and name(.) = 'option')");
}
public function translateInvalid(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition('0');
}
public function translateHover(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition('0');
}
public function translateVisited(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition('0');
}
public function getName(): string
{
return 'html';
}
}
@@ -0,0 +1,221 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\XPath\Extension;
use Symfony\Component\CssSelector\Node;
use Symfony\Component\CssSelector\XPath\Translator;
use Symfony\Component\CssSelector\XPath\XPathExpr;
/**
* XPath expression translator node extension.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class NodeExtension extends AbstractExtension
{
public const ELEMENT_NAME_IN_LOWER_CASE = 1;
public const ATTRIBUTE_NAME_IN_LOWER_CASE = 2;
public const ATTRIBUTE_VALUE_IN_LOWER_CASE = 4;
public function __construct(
private int $flags = 0,
) {
}
/**
* @return $this
*/
public function setFlag(int $flag, bool $on): static
{
if ($on && !$this->hasFlag($flag)) {
$this->flags += $flag;
}
if (!$on && $this->hasFlag($flag)) {
$this->flags -= $flag;
}
return $this;
}
public function hasFlag(int $flag): bool
{
return (bool) ($this->flags & $flag);
}
public function getNodeTranslators(): array
{
return [
'Selector' => $this->translateSelector(...),
'CombinedSelector' => $this->translateCombinedSelector(...),
'Negation' => $this->translateNegation(...),
'Matching' => $this->translateMatching(...),
'SpecificityAdjustment' => $this->translateSpecificityAdjustment(...),
'Function' => $this->translateFunction(...),
'Pseudo' => $this->translatePseudo(...),
'Attribute' => $this->translateAttribute(...),
'Class' => $this->translateClass(...),
'Hash' => $this->translateHash(...),
'Element' => $this->translateElement(...),
];
}
public function translateSelector(Node\SelectorNode $node, Translator $translator): XPathExpr
{
return $translator->nodeToXPath($node->getTree());
}
public function translateCombinedSelector(Node\CombinedSelectorNode $node, Translator $translator): XPathExpr
{
return $translator->addCombination($node->getCombinator(), $node->getSelector(), $node->getSubSelector());
}
public function translateNegation(Node\NegationNode $node, Translator $translator): XPathExpr
{
$xpath = $translator->nodeToXPath($node->getSelector());
$subXpath = $translator->nodeToXPath($node->getSubSelector());
$subXpath->addNameTest();
if ($subXpath->getCondition()) {
return $xpath->addCondition(\sprintf('not(%s)', $subXpath->getCondition()));
}
return $xpath->addCondition('0');
}
public function translateMatching(Node\MatchingNode $node, Translator $translator): XPathExpr
{
$xpath = $translator->nodeToXPath($node->selector);
foreach ($node->arguments as $argument) {
$expr = $translator->nodeToXPath($argument);
$expr->addNameTest();
if ($condition = $expr->getCondition()) {
$xpath->addCondition($condition, 'or');
}
}
return $xpath;
}
public function translateSpecificityAdjustment(Node\SpecificityAdjustmentNode $node, Translator $translator): XPathExpr
{
$xpath = $translator->nodeToXPath($node->selector);
foreach ($node->arguments as $argument) {
$expr = $translator->nodeToXPath($argument);
$expr->addNameTest();
if ($condition = $expr->getCondition()) {
$xpath->addCondition($condition, 'or');
}
}
return $xpath;
}
public function translateFunction(Node\FunctionNode $node, Translator $translator): XPathExpr
{
$xpath = $translator->nodeToXPath($node->getSelector());
return $translator->addFunction($xpath, $node);
}
public function translatePseudo(Node\PseudoNode $node, Translator $translator): XPathExpr
{
$xpath = $translator->nodeToXPath($node->getSelector());
return $translator->addPseudoClass($xpath, $node->getIdentifier());
}
public function translateAttribute(Node\AttributeNode $node, Translator $translator): XPathExpr
{
$name = $node->getAttribute();
$safe = $this->isSafeName($name);
if ($this->hasFlag(self::ATTRIBUTE_NAME_IN_LOWER_CASE)) {
$name = strtolower($name);
}
if ($node->getNamespace()) {
$name = \sprintf('%s:%s', $node->getNamespace(), $name);
$safe = $safe && $this->isSafeName($node->getNamespace());
}
$attribute = $safe ? '@'.$name : \sprintf('attribute::*[name() = %s]', Translator::getXpathLiteral($name));
$value = $node->getValue();
$xpath = $translator->nodeToXPath($node->getSelector());
if ($this->hasFlag(self::ATTRIBUTE_VALUE_IN_LOWER_CASE)) {
$value = strtolower($value);
}
return $translator->addAttributeMatching($xpath, $node->getOperator(), $attribute, $value);
}
public function translateClass(Node\ClassNode $node, Translator $translator): XPathExpr
{
$xpath = $translator->nodeToXPath($node->getSelector());
return $translator->addAttributeMatching($xpath, '~=', '@class', $node->getName());
}
public function translateHash(Node\HashNode $node, Translator $translator): XPathExpr
{
$xpath = $translator->nodeToXPath($node->getSelector());
return $translator->addAttributeMatching($xpath, '=', '@id', $node->getId());
}
public function translateElement(Node\ElementNode $node): XPathExpr
{
$element = $node->getElement();
if ($element && $this->hasFlag(self::ELEMENT_NAME_IN_LOWER_CASE)) {
$element = strtolower($element);
}
if ($element) {
$safe = $this->isSafeName($element);
} else {
$element = '*';
$safe = true;
}
if ($node->getNamespace()) {
$element = \sprintf('%s:%s', $node->getNamespace(), $element);
$safe = $safe && $this->isSafeName($node->getNamespace());
}
$xpath = new XPathExpr('', $element);
if (!$safe) {
$xpath->addNameTest();
}
return $xpath;
}
public function getName(): string
{
return 'node';
}
private function isSafeName(string $name): bool
{
return 0 < preg_match('~^[a-zA-Z_][a-zA-Z0-9_.-]*$~', $name);
}
}
@@ -0,0 +1,122 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\XPath\Extension;
use Symfony\Component\CssSelector\Exception\ExpressionErrorException;
use Symfony\Component\CssSelector\XPath\XPathExpr;
/**
* XPath expression translator pseudo-class extension.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class PseudoClassExtension extends AbstractExtension
{
public function getPseudoClassTranslators(): array
{
return [
'root' => $this->translateRoot(...),
'scope' => $this->translateScopePseudo(...),
'first-child' => $this->translateFirstChild(...),
'last-child' => $this->translateLastChild(...),
'first-of-type' => $this->translateFirstOfType(...),
'last-of-type' => $this->translateLastOfType(...),
'only-child' => $this->translateOnlyChild(...),
'only-of-type' => $this->translateOnlyOfType(...),
'empty' => $this->translateEmpty(...),
];
}
public function translateRoot(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition('not(parent::*)');
}
public function translateScopePseudo(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition('1');
}
public function translateFirstChild(XPathExpr $xpath): XPathExpr
{
return $xpath
->addStarPrefix()
->addNameTest()
->addCondition('position() = 1');
}
public function translateLastChild(XPathExpr $xpath): XPathExpr
{
return $xpath
->addStarPrefix()
->addNameTest()
->addCondition('position() = last()');
}
/**
* @throws ExpressionErrorException
*/
public function translateFirstOfType(XPathExpr $xpath): XPathExpr
{
if ('*' === $xpath->getElement()) {
throw new ExpressionErrorException('"*:first-of-type" is not implemented.');
}
return $xpath
->addStarPrefix()
->addCondition('position() = 1');
}
/**
* @throws ExpressionErrorException
*/
public function translateLastOfType(XPathExpr $xpath): XPathExpr
{
if ('*' === $xpath->getElement()) {
throw new ExpressionErrorException('"*:last-of-type" is not implemented.');
}
return $xpath
->addStarPrefix()
->addCondition('position() = last()');
}
public function translateOnlyChild(XPathExpr $xpath): XPathExpr
{
return $xpath
->addStarPrefix()
->addNameTest()
->addCondition('last() = 1');
}
public function translateOnlyOfType(XPathExpr $xpath): XPathExpr
{
$element = $xpath->getElement();
return $xpath->addCondition(\sprintf('count(preceding-sibling::%s)=0 and count(following-sibling::%s)=0', $element, $element));
}
public function translateEmpty(XPathExpr $xpath): XPathExpr
{
return $xpath->addCondition('not(*) and not(string-length())');
}
public function getName(): string
{
return 'pseudo-class';
}
}
@@ -0,0 +1,224 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\XPath;
use Symfony\Component\CssSelector\Exception\ExpressionErrorException;
use Symfony\Component\CssSelector\Node\FunctionNode;
use Symfony\Component\CssSelector\Node\NodeInterface;
use Symfony\Component\CssSelector\Node\SelectorNode;
use Symfony\Component\CssSelector\Parser\Parser;
use Symfony\Component\CssSelector\Parser\ParserInterface;
/**
* XPath expression translator interface.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class Translator implements TranslatorInterface
{
private ParserInterface $mainParser;
/**
* @var ParserInterface[]
*/
private array $shortcutParsers = [];
/**
* @var Extension\ExtensionInterface[]
*/
private array $extensions = [];
private array $nodeTranslators = [];
private array $combinationTranslators = [];
private array $functionTranslators = [];
private array $pseudoClassTranslators = [];
private array $attributeMatchingTranslators = [];
public function __construct(?ParserInterface $parser = null)
{
$this->mainParser = $parser ?? new Parser();
$this
->registerExtension(new Extension\NodeExtension())
->registerExtension(new Extension\CombinationExtension())
->registerExtension(new Extension\FunctionExtension())
->registerExtension(new Extension\PseudoClassExtension())
->registerExtension(new Extension\AttributeMatchingExtension())
;
}
public static function getXpathLiteral(string $element): string
{
if (!str_contains($element, "'")) {
return "'".$element."'";
}
if (!str_contains($element, '"')) {
return '"'.$element.'"';
}
$string = $element;
$parts = [];
while (true) {
if (false !== $pos = strpos($string, "'")) {
$parts[] = \sprintf("'%s'", substr($string, 0, $pos));
$parts[] = "\"'\"";
$string = substr($string, $pos + 1);
} else {
$parts[] = "'$string'";
break;
}
}
return \sprintf('concat(%s)', implode(', ', $parts));
}
public function cssToXPath(string $cssExpr, string $prefix = 'descendant-or-self::'): string
{
$selectors = $this->parseSelectors($cssExpr);
/** @var SelectorNode $selector */
foreach ($selectors as $index => $selector) {
if (null !== $selector->getPseudoElement()) {
throw new ExpressionErrorException('Pseudo-elements are not supported.');
}
$selectors[$index] = $this->selectorToXPath($selector, $prefix);
}
return implode(' | ', $selectors);
}
public function selectorToXPath(SelectorNode $selector, string $prefix = 'descendant-or-self::'): string
{
return ($prefix ?: '').$this->nodeToXPath($selector);
}
/**
* @return $this
*/
public function registerExtension(Extension\ExtensionInterface $extension): static
{
$this->extensions[$extension->getName()] = $extension;
$this->nodeTranslators = array_merge($this->nodeTranslators, $extension->getNodeTranslators());
$this->combinationTranslators = array_merge($this->combinationTranslators, $extension->getCombinationTranslators());
$this->functionTranslators = array_merge($this->functionTranslators, $extension->getFunctionTranslators());
$this->pseudoClassTranslators = array_merge($this->pseudoClassTranslators, $extension->getPseudoClassTranslators());
$this->attributeMatchingTranslators = array_merge($this->attributeMatchingTranslators, $extension->getAttributeMatchingTranslators());
return $this;
}
/**
* @throws ExpressionErrorException
*/
public function getExtension(string $name): Extension\ExtensionInterface
{
if (!isset($this->extensions[$name])) {
throw new ExpressionErrorException(\sprintf('Extension "%s" not registered.', $name));
}
return $this->extensions[$name];
}
/**
* @return $this
*/
public function registerParserShortcut(ParserInterface $shortcut): static
{
$this->shortcutParsers[] = $shortcut;
return $this;
}
/**
* @throws ExpressionErrorException
*/
public function nodeToXPath(NodeInterface $node): XPathExpr
{
if (!isset($this->nodeTranslators[$node->getNodeName()])) {
throw new ExpressionErrorException(\sprintf('Node "%s" not supported.', $node->getNodeName()));
}
return $this->nodeTranslators[$node->getNodeName()]($node, $this);
}
/**
* @throws ExpressionErrorException
*/
public function addCombination(string $combiner, NodeInterface $xpath, NodeInterface $combinedXpath): XPathExpr
{
if (!isset($this->combinationTranslators[$combiner])) {
throw new ExpressionErrorException(\sprintf('Combiner "%s" not supported.', $combiner));
}
return $this->combinationTranslators[$combiner]($this->nodeToXPath($xpath), $this->nodeToXPath($combinedXpath));
}
/**
* @throws ExpressionErrorException
*/
public function addFunction(XPathExpr $xpath, FunctionNode $function): XPathExpr
{
if (!isset($this->functionTranslators[$function->getName()])) {
throw new ExpressionErrorException(\sprintf('Function "%s" not supported.', $function->getName()));
}
return $this->functionTranslators[$function->getName()]($xpath, $function);
}
/**
* @throws ExpressionErrorException
*/
public function addPseudoClass(XPathExpr $xpath, string $pseudoClass): XPathExpr
{
if (!isset($this->pseudoClassTranslators[$pseudoClass])) {
throw new ExpressionErrorException(\sprintf('Pseudo-class "%s" not supported.', $pseudoClass));
}
return $this->pseudoClassTranslators[$pseudoClass]($xpath);
}
/**
* @throws ExpressionErrorException
*/
public function addAttributeMatching(XPathExpr $xpath, string $operator, string $attribute, ?string $value): XPathExpr
{
if (!isset($this->attributeMatchingTranslators[$operator])) {
throw new ExpressionErrorException(\sprintf('Attribute matcher operator "%s" not supported.', $operator));
}
return $this->attributeMatchingTranslators[$operator]($xpath, $attribute, $value);
}
/**
* @return SelectorNode[]
*/
private function parseSelectors(string $css): array
{
foreach ($this->shortcutParsers as $shortcut) {
$tokens = $shortcut->parse($css);
if ($tokens) {
return $tokens;
}
}
return $this->mainParser->parse($css);
}
}
@@ -0,0 +1,107 @@
<?php
/*
* This file is part of the Symfony package.
*
* (c) Fabien Potencier <fabien@symfony.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace Symfony\Component\CssSelector\XPath;
/**
* XPath expression translator interface.
*
* This component is a port of the Python cssselect library,
* which is copyright Ian Bicking, @see https://github.com/SimonSapin/cssselect.
*
* @author Jean-François Simon <jeanfrancois.simon@sensiolabs.com>
*
* @internal
*/
class XPathExpr
{
public function __construct(
private string $path = '',
private string $element = '*',
private string $condition = '',
bool $starPrefix = false,
) {
if ($starPrefix) {
$this->addStarPrefix();
}
}
public function getElement(): string
{
return $this->element;
}
/**
* @return $this
*/
public function addCondition(string $condition, string $operator = 'and'): static
{
$this->condition = $this->condition ? \sprintf('(%s) %s (%s)', $this->condition, $operator, $condition) : $condition;
return $this;
}
public function getCondition(): string
{
return $this->condition;
}
/**
* @return $this
*/
public function addNameTest(): static
{
if ('*' !== $this->element) {
$this->addCondition('name() = '.Translator::getXpathLiteral($this->element));
$this->element = '*';
}
return $this;
}
/**
* @return $this
*/
public function addStarPrefix(): static
{
$this->path .= '*/';
return $this;
}
/**
* Joins another XPathExpr with a combiner.
*
* @return $this
*/
public function join(string $combiner, self $expr): static
{
$path = $this->__toString().$combiner;
if ('*/' !== $expr->path) {
$path .= $expr->path;
}
$this->path = $path;
$this->element = $expr->element;
$this->condition = $expr->condition;
return $this;
}
public function __toString(): string
{
$path = $this->path.$this->element;
$condition = '' === $this->condition ? '' : '['.$this->condition.']';
return $path.$condition;
}
}