Crawler.php

<?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\DomCrawler;

use Symfony\Component\CssSelector\CssSelectorConverter;

/**
 * Crawler eases navigation of a list of \DOMNode objects.
 *
 * @author Fabien Potencier <fabien@symfony.com>
 */
class Crawler implements \Countable, \IteratorAggregate
{
    /**
     * @var string The current URI
     */
    protected $uri;

    /**
     * @var string The default namespace prefix to be used with XPath and CSS expressions
     */
    private $defaultNamespacePrefix = 'default';

    /**
     * @var array A map of manually registered namespaces
     */
    private $namespaces = array();

    /**
     * @var string The base href value
     */
    private $baseHref;

    /**
     * @var \DOMDocument|null
     */
    private $document;

    /**
     * @var \DOMElement[]
     */
    private $nodes = array();

    /**
     * Whether the Crawler contains HTML or XML content (used when converting CSS to XPath).
     *
     * @var bool
     */
    private $isHtml = true;

    /**
     * @param mixed  $node       A Node to use as the base for the crawling
     * @param string $currentUri The current URI
     * @param string $baseHref   The base href value
     */
    public function __construct($node = null, $currentUri = null, $baseHref = null)
    {
        $this->uri = $currentUri;
        $this->baseHref = $baseHref ?: $currentUri;

        $this->add($node);
    }

    /**
     * Returns the current URI.
     *
     * @return string
     */
    public function getUri()
    {
        return $this->uri;
    }

    /**
     * Returns base href.
     *
     * @return string
     */
    public function getBaseHref()
    {
        return $this->baseHref;
    }

    /**
     * Removes all the nodes.
     */
    public function clear()
    {
        $this->nodes = array();
        $this->document = null;
    }

    /**
     * Adds a node to the current list of nodes.
     *
     * This method uses the appropriate specialized add*() method based
     * on the type of the argument.
     *
     * @param \DOMNodeList|\DOMNode|array|string|null $node A node
     *
     * @throws \InvalidArgumentException When node is not the expected type.
     */
    public function add($node)
    {
        if ($node instanceof \DOMNodeList) {
            $this->addNodeList($node);
        } elseif ($node instanceof \DOMNode) {
            $this->addNode($node);
        } elseif (is_array($node)) {
            $this->addNodes($node);
        } elseif (is_string($node)) {
            $this->addContent($node);
        } elseif (null !== $node) {
            throw new \InvalidArgumentException(sprintf('Expecting a DOMNodeList or DOMNode instance, an array, a string, or null, but got "%s".', is_object($node) ? get_class($node) : gettype($node)));
        }
    }

    /**
     * Adds HTML/XML content.
     *
     * If the charset is not set via the content type, it is assumed
     * to be ISO-8859-1, which is the default charset defined by the
     * HTTP 1.1 specification.
     *
     * @param string      $content A string to parse as HTML/XML
     * @param null|string $type    The content type of the string
     */
    public function addContent($content, $type = null)
    {
        if (empty($type)) {
            $type = 0 === strpos($content, '<?xml') ? 'application/xml' : 'text/html';
        }

        // DOM only for HTML/XML content
        if (!preg_match('/(x|ht)ml/i', $type, $xmlMatches)) {
            return;
        }

        $charset = null;
        if (false !== $pos = stripos($type, 'charset=')) {
            $charset = substr($type, $pos + 8);
            if (false !== $pos = strpos($charset, ';')) {
                $charset = substr($charset, 0, $pos);
            }
        }

        // http://www.w3.org/TR/encoding/#encodings
        // http://www.w3.org/TR/REC-xml/#NT-EncName
        if (null === $charset &&
            preg_match('/\<meta[^\>]+charset *= *["\']?([a-zA-Z\-0-9_:.]+)/i', $content, $matches)) {
            $charset = $matches[1];
        }

        if (null === $charset) {
            $charset = 'ISO-8859-1';
        }

        if ('x' === $xmlMatches[1]) {
            $this->addXmlContent($content, $charset);
        } else {
            $this->addHtmlContent($content, $charset);
        }
    }

    /**
     * Adds an HTML content to the list of nodes.
     *
     * The libxml errors are disabled when the content is parsed.
     *
     * If you want to get parsing errors, be sure to enable
     * internal errors via libxml_use_internal_errors(true)
     * and then, get the errors via libxml_get_errors(). Be
     * sure to clear errors with libxml_clear_errors() afterward.
     *
     * @param string $content The HTML content
     * @param string $charset The charset
     */
    public function addHtmlContent($content, $charset = 'UTF-8')
    {
        $internalErrors = libxml_use_internal_errors(true);
        $disableEntities = libxml_disable_entity_loader(true);

        $dom = new \DOMDocument('1.0', $charset);
        $dom->validateOnParse = true;

        set_error_handler(function () {throw new \Exception();});

        try {
            // Convert charset to HTML-entities to work around bugs in DOMDocument::loadHTML()
            $content = mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
        } catch (\Exception $e) {
        }

        restore_error_handler();

        if ('' !== trim($content)) {
            @$dom->loadHTML($content);
        }

        libxml_use_internal_errors($internalErrors);
        libxml_disable_entity_loader($disableEntities);

        $this->addDocument($dom);

        $base = $this->filterRelativeXPath('descendant-or-self::base')->extract(array('href'));

        $baseHref = current($base);
        if (count($base) && !empty($baseHref)) {
            if ($this->baseHref) {
                $linkNode = $dom->createElement('a');
                $linkNode->setAttribute('href', $baseHref);
                $link = new Link($linkNode, $this->baseHref);
                $this->baseHref = $link->getUri();
            } else {
                $this->baseHref = $baseHref;
            }
        }
    }

    /**
     * Adds an XML content to the list of nodes.
     *
     * The libxml errors are disabled when the content is parsed.
     *
     * If you want to get parsing errors, be sure to enable
     * internal errors via libxml_use_internal_errors(true)
     * and then, get the errors via libxml_get_errors(). Be
     * sure to clear errors with libxml_clear_errors() afterward.
     *
     * @param string $content The XML content
     * @param string $charset The charset
     * @param int    $options Bitwise OR of the libxml option constants
     *                        LIBXML_PARSEHUGE is dangerous, see
     *                        http://symfony.com/blog/security-release-symfony-2-0-17-released
     */
    public function addXmlContent($content, $charset = 'UTF-8', $options = LIBXML_NONET)
    {
        // remove the default namespace if it's the only namespace to make XPath expressions simpler
        if (!preg_match('/xmlns:/', $content)) {
            $content = str_replace('xmlns', 'ns', $content);
        }

        $internalErrors = libxml_use_internal_errors(true);
        $disableEntities = libxml_disable_entity_loader(true);

        $dom = new \DOMDocument('1.0', $charset);
        $dom->validateOnParse = true;

        if ('' !== trim($content)) {
            @$dom->loadXML($content, $options);
        }

        libxml_use_internal_errors($internalErrors);
        libxml_disable_entity_loader($disableEntities);

        $this->addDocument($dom);

        $this->isHtml = false;
    }

    /**
     * Adds a \DOMDocument to the list of nodes.
     *
     * @param \DOMDocument $dom A \DOMDocument instance
     */
    public function addDocument(\DOMDocument $dom)
    {
        if ($dom->documentElement) {
            $this->addNode($dom->documentElement);
        }
    }

    /**
     * Adds a \DOMNodeList to the list of nodes.
     *
     * @param \DOMNodeList $nodes A \DOMNodeList instance
     */
    public function addNodeList(\DOMNodeList $nodes)
    {
        foreach ($nodes as $node) {
            if ($node instanceof \DOMNode) {
                $this->addNode($node);
            }
        }
    }

    /**
     * Adds an array of \DOMNode instances to the list of nodes.
     *
     * @param \DOMNode[] $nodes An array of \DOMNode instances
     */
    public function addNodes(array $nodes)
    {
        foreach ($nodes as $node) {
            $this->add($node);
        }
    }

    /**
     * Adds a \DOMNode instance to the list of nodes.
     *
     * @param \DOMNode $node A \DOMNode instance
     */
    public function addNode(\DOMNode $node)
    {
        if ($node instanceof \DOMDocument) {
            $node = $node->documentElement;
        }

        if (null !== $this->document && $this->document !== $node->ownerDocument) {
            throw new \InvalidArgumentException('Attaching DOM nodes from multiple documents in the same crawler is forbidden.');
        }

        if (null === $this->document) {
            $this->document = $node->ownerDocument;
        }

        // Don't add duplicate nodes in the Crawler
        if (in_array($node, $this->nodes, true)) {
            return;
        }

        $this->nodes[] = $node;
    }

    /**
     * Returns a node given its position in the node list.
     *
     * @param int $position The position
     *
     * @return Crawler A new instance of the Crawler with the selected node, or an empty Crawler if it does not exist
     */
    public function eq($position)
    {
        if (isset($this->nodes[$position])) {
            return $this->createSubCrawler($this->nodes[$position]);
        }

        return $this->createSubCrawler(null);
    }

    /**
     * Calls an anonymous function on each node of the list.
     *
     * The anonymous function receives the position and the node wrapped
     * in a Crawler instance as arguments.
     *
     * Example:
     *
     *     $crawler->filter('h1')->each(function ($node, $i) {
     *         return $node->text();
     *     });
     *
     * @param \Closure $closure An anonymous function
     *
     * @return array An array of values returned by the anonymous function
     */
    public function each(\Closure $closure)
    {
        $data = array();
        foreach ($this->nodes as $i => $node) {
            $data[] = $closure($this->createSubCrawler($node), $i);
        }

        return $data;
    }

    /**
     * Slices the list of nodes by $offset and $length.
     *
     * @param int $offset
     * @param int $length
     *
     * @return Crawler A Crawler instance with the sliced nodes
     */
    public function slice($offset = 0, $length = null)
    {
        return $this->createSubCrawler(array_slice($this->nodes, $offset, $length));
    }

    /**
     * Reduces the list of nodes by calling an anonymous function.
     *
     * To remove a node from the list, the anonymous function must return false.
     *
     * @param \Closure $closure An anonymous function
     *
     * @return Crawler A Crawler instance with the selected nodes
     */
    public function reduce(\Closure $closure)
    {
        $nodes = array();
        foreach ($this->nodes as $i => $node) {
            if (false !== $closure($this->createSubCrawler($node), $i)) {
                $nodes[] = $node;
            }
        }

        return $this->createSubCrawler($nodes);
    }

    /**
     * Returns the first node of the current selection.
     *
     * @return Crawler A Crawler instance with the first selected node
     */
    public function first()
    {
        return $this->eq(0);
    }

    /**
     * Returns the last node of the current selection.
     *
     * @return Crawler A Crawler instance with the last selected node
     */
    public function last()
    {
        return $this->eq(count($this->nodes) - 1);
    }

    /**
     * Returns the siblings nodes of the current selection.
     *
     * @return Crawler A Crawler instance with the sibling nodes
     *
     * @throws \InvalidArgumentException When current node is empty
     */
    public function siblings()
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        return $this->createSubCrawler($this->sibling($this->getNode(0)->parentNode->firstChild));
    }

    /**
     * Returns the next siblings nodes of the current selection.
     *
     * @return Crawler A Crawler instance with the next sibling nodes
     *
     * @throws \InvalidArgumentException When current node is empty
     */
    public function nextAll()
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        return $this->createSubCrawler($this->sibling($this->getNode(0)));
    }

    /**
     * Returns the previous sibling nodes of the current selection.
     *
     * @return Crawler A Crawler instance with the previous sibling nodes
     *
     * @throws \InvalidArgumentException
     */
    public function previousAll()
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        return $this->createSubCrawler($this->sibling($this->getNode(0), 'previousSibling'));
    }

    /**
     * Returns the parents nodes of the current selection.
     *
     * @return Crawler A Crawler instance with the parents nodes of the current selection
     *
     * @throws \InvalidArgumentException When current node is empty
     */
    public function parents()
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        $node = $this->getNode(0);
        $nodes = array();

        while ($node = $node->parentNode) {
            if (XML_ELEMENT_NODE === $node->nodeType) {
                $nodes[] = $node;
            }
        }

        return $this->createSubCrawler($nodes);
    }

    /**
     * Returns the children nodes of the current selection.
     *
     * @return Crawler A Crawler instance with the children nodes
     *
     * @throws \InvalidArgumentException When current node is empty
     */
    public function children()
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        $node = $this->getNode(0)->firstChild;

        return $this->createSubCrawler($node ? $this->sibling($node) : array());
    }

    /**
     * Returns the attribute value of the first node of the list.
     *
     * @param string $attribute The attribute name
     *
     * @return string|null The attribute value or null if the attribute does not exist
     *
     * @throws \InvalidArgumentException When current node is empty
     */
    public function attr($attribute)
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        $node = $this->getNode(0);

        return $node->hasAttribute($attribute) ? $node->getAttribute($attribute) : null;
    }

    /**
     * Returns the node name of the first node of the list.
     *
     * @return string The node name
     *
     * @throws \InvalidArgumentException When current node is empty
     */
    public function nodeName()
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        return $this->getNode(0)->nodeName;
    }

    /**
     * Returns the node value of the first node of the list.
     *
     * @return string The node value
     *
     * @throws \InvalidArgumentException When current node is empty
     */
    public function text()
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        return $this->getNode(0)->nodeValue;
    }

    /**
     * Returns the first node of the list as HTML.
     *
     * @return string The node html
     *
     * @throws \InvalidArgumentException When current node is empty
     */
    public function html()
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        $html = '';
        foreach ($this->getNode(0)->childNodes as $child) {
            $html .= $child->ownerDocument->saveHTML($child);
        }

        return $html;
    }

    /**
     * Evaluates an XPath expression.
     *
     * Since an XPath expression might evaluate to either a simple type or a \DOMDoneList,
     * this method will return either an array of simple types or a new Crawler instance.
     *
     * @param string $xpath An XPath expression
     *
     * @return array|Crawler An array of evaluation results or a new Crawler instance
     */
    public function evaluate($xpath)
    {
        if (null === $this->document) {
            throw new \LogicException('Cannot evaluate the expression on an uninitialized crawler.');
        }

        $data = array();
        $domxpath = $this->createDOMXPath($this->document, $this->findNamespacePrefixes($xpath));

        foreach ($this->nodes as $node) {
            $data[] = $domxpath->evaluate($xpath, $node);
        }

        if (isset($data[0]) && $data[0] instanceof \DOMNodeList) {
            return $this->createSubCrawler($data);
        }

        return $data;
    }

    /**
     * Extracts information from the list of nodes.
     *
     * You can extract attributes or/and the node value (_text).
     *
     * Example:
     *
     * $crawler->filter('h1 a')->extract(array('_text', 'href'));
     *
     * @param array $attributes An array of attributes
     *
     * @return array An array of extracted values
     */
    public function extract($attributes)
    {
        $attributes = (array) $attributes;
        $count = count($attributes);

        $data = array();
        foreach ($this->nodes as $node) {
            $elements = array();
            foreach ($attributes as $attribute) {
                if ('_text' === $attribute) {
                    $elements[] = $node->nodeValue;
                } else {
                    $elements[] = $node->getAttribute($attribute);
                }
            }

            $data[] = $count > 1 ? $elements : $elements[0];
        }

        return $data;
    }

    /**
     * Filters the list of nodes with an XPath expression.
     *
     * The XPath expression is evaluated in the context of the crawler, which
     * is considered as a fake parent of the elements inside it.
     * This means that a child selector "div" or "./div" will match only
     * the div elements of the current crawler, not their children.
     *
     * @param string $xpath An XPath expression
     *
     * @return Crawler A new instance of Crawler with the filtered list of nodes
     */
    public function filterXPath($xpath)
    {
        $xpath = $this->relativize($xpath);

        // If we dropped all expressions in the XPath while preparing it, there would be no match
        if ('' === $xpath) {
            return $this->createSubCrawler(null);
        }

        return $this->filterRelativeXPath($xpath);
    }

    /**
     * Filters the list of nodes with a CSS selector.
     *
     * This method only works if you have installed the CssSelector Symfony Component.
     *
     * @param string $selector A CSS selector
     *
     * @return Crawler A new instance of Crawler with the filtered list of nodes
     *
     * @throws \RuntimeException if the CssSelector Component is not available
     */
    public function filter($selector)
    {
        if (!class_exists('Symfony\\Component\\CssSelector\\CssSelectorConverter')) {
            throw new \RuntimeException('Unable to filter with a CSS selector as the Symfony CssSelector 2.8+ is not installed (you can use filterXPath instead).');
        }

        $converter = new CssSelectorConverter($this->isHtml);

        // The CssSelector already prefixes the selector with descendant-or-self::
        return $this->filterRelativeXPath($converter->toXPath($selector));
    }

    /**
     * Selects links by name or alt value for clickable images.
     *
     * @param string $value The link text
     *
     * @return Crawler A new instance of Crawler with the filtered list of nodes
     */
    public function selectLink($value)
    {
        $xpath = sprintf('descendant-or-self::a[contains(concat(\' \', normalize-space(string(.)), \' \'), %s) ', static::xpathLiteral(' '.$value.' ')).
                            sprintf('or ./img[contains(concat(\' \', normalize-space(string(@alt)), \' \'), %s)]]', static::xpathLiteral(' '.$value.' '));

        return $this->filterRelativeXPath($xpath);
    }

    /**
     * Selects images by alt value.
     *
     * @param string $value The image alt
     *
     * @return Crawler A new instance of Crawler with the filtered list of nodes
     */
    public function selectImage($value)
    {
        $xpath = sprintf('descendant-or-self::img[contains(normalize-space(string(@alt)), %s)]', static::xpathLiteral($value));

        return $this->filterRelativeXPath($xpath);
    }

    /**
     * Selects a button by name or alt value for images.
     *
     * @param string $value The button text
     *
     * @return Crawler A new instance of Crawler with the filtered list of nodes
     */
    public function selectButton($value)
    {
        $translate = 'translate(@type, "ABCDEFGHIJKLMNOPQRSTUVWXYZ", "abcdefghijklmnopqrstuvwxyz")';
        $xpath = sprintf('descendant-or-self::input[((contains(%s, "submit") or contains(%s, "button")) and contains(concat(\' \', normalize-space(string(@value)), \' \'), %s)) ', $translate, $translate, static::xpathLiteral(' '.$value.' ')).
                         sprintf('or (contains(%s, "image") and contains(concat(\' \', normalize-space(string(@alt)), \' \'), %s)) or @id=%s or @name=%s] ', $translate, static::xpathLiteral(' '.$value.' '), static::xpathLiteral($value), static::xpathLiteral($value)).
                         sprintf('| descendant-or-self::button[contains(concat(\' \', normalize-space(string(.)), \' \'), %s) or @id=%s or @name=%s]', static::xpathLiteral(' '.$value.' '), static::xpathLiteral($value), static::xpathLiteral($value));

        return $this->filterRelativeXPath($xpath);
    }

    /**
     * Returns a Link object for the first node in the list.
     *
     * @param string $method The method for the link (get by default)
     *
     * @return Link A Link instance
     *
     * @throws \InvalidArgumentException If the current node list is empty or the selected node is not instance of DOMElement
     */
    public function link($method = 'get')
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        $node = $this->getNode(0);

        if (!$node instanceof \DOMElement) {
            throw new \InvalidArgumentException(sprintf('The selected node should be instance of DOMElement, got "%s".', get_class($node)));
        }

        return new Link($node, $this->baseHref, $method);
    }

    /**
     * Returns an array of Link objects for the nodes in the list.
     *
     * @return Link[] An array of Link instances
     *
     * @throws \InvalidArgumentException If the current node list contains non-DOMElement instances
     */
    public function links()
    {
        $links = array();
        foreach ($this->nodes as $node) {
            if (!$node instanceof \DOMElement) {
                throw new \InvalidArgumentException(sprintf('The current node list should contain only DOMElement instances, "%s" found.', get_class($node)));
            }

            $links[] = new Link($node, $this->baseHref, 'get');
        }

        return $links;
    }

    /**
     * Returns an Image object for the first node in the list.
     *
     * @return Image An Image instance
     *
     * @throws \InvalidArgumentException If the current node list is empty
     */
    public function image()
    {
        if (!count($this)) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        $node = $this->getNode(0);

        if (!$node instanceof \DOMElement) {
            throw new \InvalidArgumentException(sprintf('The selected node should be instance of DOMElement, got "%s".', get_class($node)));
        }

        return new Image($node, $this->baseHref);
    }

    /**
     * Returns an array of Image objects for the nodes in the list.
     *
     * @return Image[] An array of Image instances
     */
    public function images()
    {
        $images = array();
        foreach ($this as $node) {
            if (!$node instanceof \DOMElement) {
                throw new \InvalidArgumentException(sprintf('The current node list should contain only DOMElement instances, "%s" found.', get_class($node)));
            }

            $images[] = new Image($node, $this->baseHref);
        }

        return $images;
    }

    /**
     * Returns a Form object for the first node in the list.
     *
     * @param array  $values An array of values for the form fields
     * @param string $method The method for the form
     *
     * @return Form A Form instance
     *
     * @throws \InvalidArgumentException If the current node list is empty or the selected node is not instance of DOMElement
     */
    public function form(array $values = null, $method = null)
    {
        if (!$this->nodes) {
            throw new \InvalidArgumentException('The current node list is empty.');
        }

        $node = $this->getNode(0);

        if (!$node instanceof \DOMElement) {
            throw new \InvalidArgumentException(sprintf('The selected node should be instance of DOMElement, got "%s".', get_class($node)));
        }

        $form = new Form($node, $this->uri, $method, $this->baseHref);

        if (null !== $values) {
            $form->setValues($values);
        }

        return $form;
    }

    /**
     * Overloads a default namespace prefix to be used with XPath and CSS expressions.
     *
     * @param string $prefix
     */
    public function setDefaultNamespacePrefix($prefix)
    {
        $this->defaultNamespacePrefix = $prefix;
    }

    /**
     * @param string $prefix
     * @param string $namespace
     */
    public function registerNamespace($prefix, $namespace)
    {
        $this->namespaces[$prefix] = $namespace;
    }

    /**
     * Converts string for XPath expressions.
     *
     * Escaped characters are: quotes (") and apostrophe (').
     *
     *  Examples:
     *  <code>
     *     echo Crawler::xpathLiteral('foo " bar');
     *     //prints 'foo " bar'
     *
     *     echo Crawler::xpathLiteral("foo ' bar");
     *     //prints "foo ' bar"
     *
     *     echo Crawler::xpathLiteral('a\'b"c');
     *     //prints concat('a', "'", 'b"c')
     *  </code>
     *
     * @param string $s String to be escaped
     *
     * @return string Converted string
     */
    public static function xpathLiteral($s)
    {
        if (false === strpos($s, "'")) {
            return sprintf("'%s'", $s);
        }

        if (false === strpos($s, '"')) {
            return sprintf('"%s"', $s);
        }

        $string = $s;
        $parts = array();
        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, ', '));
    }

    /**
     * Filters the list of nodes with an XPath expression.
     *
     * The XPath expression should already be processed to apply it in the context of each node.
     *
     * @param string $xpath
     *
     * @return Crawler
     */
    private function filterRelativeXPath($xpath)
    {
        $prefixes = $this->findNamespacePrefixes($xpath);

        $crawler = $this->createSubCrawler(null);

        foreach ($this->nodes as $node) {
            $domxpath = $this->createDOMXPath($node->ownerDocument, $prefixes);
            $crawler->add($domxpath->query($xpath, $node));
        }

        return $crawler;
    }

    /**
     * Make the XPath relative to the current context.
     *
     * The returned XPath will match elements matching the XPath inside the current crawler
     * when running in the context of a node of the crawler.
     *
     * @param string $xpath
     *
     * @return string
     */
    private function relativize($xpath)
    {
        $expressions = array();

        $unionPattern = '/\|(?![^\[]*\])/';
        // An expression which will never match to replace expressions which cannot match in the crawler
        // We cannot simply drop
        $nonMatchingExpression = 'a[name() = "b"]';

        // Split any unions into individual expressions.
        foreach (preg_split($unionPattern, $xpath) as $expression) {
            $expression = trim($expression);
            $parenthesis = '';

            // If the union is inside some braces, we need to preserve the opening braces and apply
            // the change only inside it.
            if (preg_match('/^[\(\s*]+/', $expression, $matches)) {
                $parenthesis = $matches[0];
                $expression = substr($expression, strlen($parenthesis));
            }

            if (0 === strpos($expression, 'self::*/')) {
                $expression = './'.substr($expression, 8);
            }

            // add prefix before absolute element selector
            if (empty($expression)) {
                $expression = $nonMatchingExpression;
            } elseif (0 === strpos($expression, '//')) {
                $expression = 'descendant-or-self::'.substr($expression, 2);
            } elseif (0 === strpos($expression, './/')) {
                $expression = 'descendant-or-self::'.substr($expression, 3);
            } elseif (0 === strpos($expression, './')) {
                $expression = 'self::'.substr($expression, 2);
            } elseif (0 === strpos($expression, 'child::')) {
                $expression = 'self::'.substr($expression, 7);
            } elseif ('/' === $expression[0] || '.' === $expression[0] || 0 === strpos($expression, 'self::')) {
                $expression = $nonMatchingExpression;
            } elseif (0 === strpos($expression, 'descendant::')) {
                $expression = 'descendant-or-self::'.substr($expression, strlen('descendant::'));
            } elseif (preg_match('/^(ancestor|ancestor-or-self|attribute|following|following-sibling|namespace|parent|preceding|preceding-sibling)::/', $expression)) {
                // the fake root has no parent, preceding or following nodes and also no attributes (even no namespace attributes)
                $expression = $nonMatchingExpression;
            } elseif (0 !== strpos($expression, 'descendant-or-self::')) {
                $expression = 'self::'.$expression;
            }
            $expressions[] = $parenthesis.$expression;
        }

        return implode(' | ', $expressions);
    }

    /**
     * @param int $position
     *
     * @return \DOMElement|null
     */
    public function getNode($position)
    {
        if (isset($this->nodes[$position])) {
            return $this->nodes[$position];
        }
    }

    /**
     * @return int
     */
    public function count()
    {
        return count($this->nodes);
    }

    /**
     * @return \ArrayIterator
     */
    public function getIterator()
    {
        return new \ArrayIterator($this->nodes);
    }

    /**
     * @param \DOMElement $node
     * @param string      $siblingDir
     *
     * @return array
     */
    protected function sibling($node, $siblingDir = 'nextSibling')
    {
        $nodes = array();

        do {
            if ($node !== $this->getNode(0) && $node->nodeType === 1) {
                $nodes[] = $node;
            }
        } while ($node = $node->$siblingDir);

        return $nodes;
    }

    /**
     * @param \DOMDocument $document
     * @param array        $prefixes
     *
     * @return \DOMXPath
     *
     * @throws \InvalidArgumentException
     */
    private function createDOMXPath(\DOMDocument $document, array $prefixes = array())
    {
        $domxpath = new \DOMXPath($document);

        foreach ($prefixes as $prefix) {
            $namespace = $this->discoverNamespace($domxpath, $prefix);
            if (null !== $namespace) {
                $domxpath->registerNamespace($prefix, $namespace);
            }
        }

        return $domxpath;
    }

    /**
     * @param \DOMXPath $domxpath
     * @param string    $prefix
     *
     * @return string
     *
     * @throws \InvalidArgumentException
     */
    private function discoverNamespace(\DOMXPath $domxpath, $prefix)
    {
        if (isset($this->namespaces[$prefix])) {
            return $this->namespaces[$prefix];
        }

        // ask for one namespace, otherwise we'd get a collection with an item for each node
        $namespaces = $domxpath->query(sprintf('(//namespace::*[name()="%s"])[last()]', $this->defaultNamespacePrefix === $prefix ? '' : $prefix));

        if ($node = $namespaces->item(0)) {
            return $node->nodeValue;
        }
    }

    /**
     * @param string $xpath
     *
     * @return array
     */
    private function findNamespacePrefixes($xpath)
    {
        if (preg_match_all('/(?P<prefix>[a-z_][a-z_0-9\-\.]*+):[^"\/:]/i', $xpath, $matches)) {
            return array_unique($matches['prefix']);
        }

        return array();
    }

    /**
     * Creates a crawler for some subnodes.
     *
     * @param \DOMElement|\DOMElement[]|\DOMNodeList|null $nodes
     *
     * @return static
     */
    private function createSubCrawler($nodes)
    {
        $crawler = new static($nodes, $this->uri, $this->baseHref);
        $crawler->isHtml = $this->isHtml;
        $crawler->document = $this->document;
        $crawler->namespaces = $this->namespaces;

        return $crawler;
    }
}