Chris@0: <?php
Chris@0: 
Chris@0: /*
Chris@0:  * This file is part of the Symfony package.
Chris@0:  *
Chris@0:  * (c) Fabien Potencier <fabien@symfony.com>
Chris@0:  *
Chris@0:  * For the full copyright and license information, please view the LICENSE
Chris@0:  * file that was distributed with this source code.
Chris@0:  */
Chris@0: 
Chris@0: namespace Symfony\Component\DomCrawler;
Chris@0: 
Chris@0: use Symfony\Component\CssSelector\CssSelectorConverter;
Chris@0: 
Chris@0: /**
Chris@0:  * Crawler eases navigation of a list of \DOMNode objects.
Chris@0:  *
Chris@0:  * @author Fabien Potencier <fabien@symfony.com>
Chris@0:  */
Chris@0: class Crawler implements \Countable, \IteratorAggregate
Chris@0: {
Chris@0:     protected $uri;
Chris@0: 
Chris@0:     /**
Chris@0:      * @var string The default namespace prefix to be used with XPath and CSS expressions
Chris@0:      */
Chris@0:     private $defaultNamespacePrefix = 'default';
Chris@0: 
Chris@0:     /**
Chris@0:      * @var array A map of manually registered namespaces
Chris@0:      */
Chris@17:     private $namespaces = [];
Chris@0: 
Chris@0:     /**
Chris@0:      * @var string The base href value
Chris@0:      */
Chris@0:     private $baseHref;
Chris@0: 
Chris@0:     /**
Chris@0:      * @var \DOMDocument|null
Chris@0:      */
Chris@0:     private $document;
Chris@0: 
Chris@0:     /**
Chris@0:      * @var \DOMElement[]
Chris@0:      */
Chris@17:     private $nodes = [];
Chris@0: 
Chris@0:     /**
Chris@0:      * Whether the Crawler contains HTML or XML content (used when converting CSS to XPath).
Chris@0:      *
Chris@0:      * @var bool
Chris@0:      */
Chris@0:     private $isHtml = true;
Chris@0: 
Chris@0:     /**
Chris@12:      * @param mixed  $node     A Node to use as the base for the crawling
Chris@12:      * @param string $uri      The current URI
Chris@12:      * @param string $baseHref The base href value
Chris@0:      */
Chris@12:     public function __construct($node = null, $uri = null, $baseHref = null)
Chris@0:     {
Chris@12:         $this->uri = $uri;
Chris@12:         $this->baseHref = $baseHref ?: $uri;
Chris@0: 
Chris@0:         $this->add($node);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the current URI.
Chris@0:      *
Chris@0:      * @return string
Chris@0:      */
Chris@0:     public function getUri()
Chris@0:     {
Chris@0:         return $this->uri;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns base href.
Chris@0:      *
Chris@0:      * @return string
Chris@0:      */
Chris@0:     public function getBaseHref()
Chris@0:     {
Chris@0:         return $this->baseHref;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Removes all the nodes.
Chris@0:      */
Chris@0:     public function clear()
Chris@0:     {
Chris@17:         $this->nodes = [];
Chris@0:         $this->document = null;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Adds a node to the current list of nodes.
Chris@0:      *
Chris@0:      * This method uses the appropriate specialized add*() method based
Chris@0:      * on the type of the argument.
Chris@0:      *
Chris@0:      * @param \DOMNodeList|\DOMNode|array|string|null $node A node
Chris@0:      *
Chris@12:      * @throws \InvalidArgumentException when node is not the expected type
Chris@0:      */
Chris@0:     public function add($node)
Chris@0:     {
Chris@0:         if ($node instanceof \DOMNodeList) {
Chris@0:             $this->addNodeList($node);
Chris@0:         } elseif ($node instanceof \DOMNode) {
Chris@0:             $this->addNode($node);
Chris@17:         } elseif (\is_array($node)) {
Chris@0:             $this->addNodes($node);
Chris@17:         } elseif (\is_string($node)) {
Chris@0:             $this->addContent($node);
Chris@0:         } elseif (null !== $node) {
Chris@17:             throw new \InvalidArgumentException(sprintf('Expecting a DOMNodeList or DOMNode instance, an array, a string, or null, but got "%s".', \is_object($node) ? \get_class($node) : \gettype($node)));
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Adds HTML/XML content.
Chris@0:      *
Chris@12:      * If the charset is not set via the content type, it is assumed to be UTF-8,
Chris@12:      * or ISO-8859-1 as a fallback, which is the default charset defined by the
Chris@0:      * HTTP 1.1 specification.
Chris@0:      *
Chris@0:      * @param string      $content A string to parse as HTML/XML
Chris@17:      * @param string|null $type    The content type of the string
Chris@0:      */
Chris@0:     public function addContent($content, $type = null)
Chris@0:     {
Chris@0:         if (empty($type)) {
Chris@0:             $type = 0 === strpos($content, '<?xml') ? 'application/xml' : 'text/html';
Chris@0:         }
Chris@0: 
Chris@0:         // DOM only for HTML/XML content
Chris@0:         if (!preg_match('/(x|ht)ml/i', $type, $xmlMatches)) {
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         $charset = null;
Chris@0:         if (false !== $pos = stripos($type, 'charset=')) {
Chris@0:             $charset = substr($type, $pos + 8);
Chris@0:             if (false !== $pos = strpos($charset, ';')) {
Chris@0:                 $charset = substr($charset, 0, $pos);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         // http://www.w3.org/TR/encoding/#encodings
Chris@0:         // http://www.w3.org/TR/REC-xml/#NT-EncName
Chris@0:         if (null === $charset &&
Chris@0:             preg_match('/\<meta[^\>]+charset *= *["\']?([a-zA-Z\-0-9_:.]+)/i', $content, $matches)) {
Chris@0:             $charset = $matches[1];
Chris@0:         }
Chris@0: 
Chris@0:         if (null === $charset) {
Chris@12:             $charset = preg_match('//u', $content) ? 'UTF-8' : 'ISO-8859-1';
Chris@0:         }
Chris@0: 
Chris@0:         if ('x' === $xmlMatches[1]) {
Chris@0:             $this->addXmlContent($content, $charset);
Chris@0:         } else {
Chris@0:             $this->addHtmlContent($content, $charset);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Adds an HTML content to the list of nodes.
Chris@0:      *
Chris@0:      * The libxml errors are disabled when the content is parsed.
Chris@0:      *
Chris@0:      * If you want to get parsing errors, be sure to enable
Chris@0:      * internal errors via libxml_use_internal_errors(true)
Chris@0:      * and then, get the errors via libxml_get_errors(). Be
Chris@0:      * sure to clear errors with libxml_clear_errors() afterward.
Chris@0:      *
Chris@0:      * @param string $content The HTML content
Chris@0:      * @param string $charset The charset
Chris@0:      */
Chris@0:     public function addHtmlContent($content, $charset = 'UTF-8')
Chris@0:     {
Chris@0:         $internalErrors = libxml_use_internal_errors(true);
Chris@0:         $disableEntities = libxml_disable_entity_loader(true);
Chris@0: 
Chris@0:         $dom = new \DOMDocument('1.0', $charset);
Chris@0:         $dom->validateOnParse = true;
Chris@0: 
Chris@0:         set_error_handler(function () { throw new \Exception(); });
Chris@0: 
Chris@0:         try {
Chris@0:             // Convert charset to HTML-entities to work around bugs in DOMDocument::loadHTML()
Chris@0:             $content = mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
Chris@0:         } catch (\Exception $e) {
Chris@0:         }
Chris@0: 
Chris@0:         restore_error_handler();
Chris@0: 
Chris@0:         if ('' !== trim($content)) {
Chris@0:             @$dom->loadHTML($content);
Chris@0:         }
Chris@0: 
Chris@0:         libxml_use_internal_errors($internalErrors);
Chris@0:         libxml_disable_entity_loader($disableEntities);
Chris@0: 
Chris@0:         $this->addDocument($dom);
Chris@0: 
Chris@17:         $base = $this->filterRelativeXPath('descendant-or-self::base')->extract(['href']);
Chris@0: 
Chris@0:         $baseHref = current($base);
Chris@17:         if (\count($base) && !empty($baseHref)) {
Chris@0:             if ($this->baseHref) {
Chris@0:                 $linkNode = $dom->createElement('a');
Chris@0:                 $linkNode->setAttribute('href', $baseHref);
Chris@0:                 $link = new Link($linkNode, $this->baseHref);
Chris@0:                 $this->baseHref = $link->getUri();
Chris@0:             } else {
Chris@0:                 $this->baseHref = $baseHref;
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Adds an XML content to the list of nodes.
Chris@0:      *
Chris@0:      * The libxml errors are disabled when the content is parsed.
Chris@0:      *
Chris@0:      * If you want to get parsing errors, be sure to enable
Chris@0:      * internal errors via libxml_use_internal_errors(true)
Chris@0:      * and then, get the errors via libxml_get_errors(). Be
Chris@0:      * sure to clear errors with libxml_clear_errors() afterward.
Chris@0:      *
Chris@0:      * @param string $content The XML content
Chris@0:      * @param string $charset The charset
Chris@0:      * @param int    $options Bitwise OR of the libxml option constants
Chris@0:      *                        LIBXML_PARSEHUGE is dangerous, see
Chris@0:      *                        http://symfony.com/blog/security-release-symfony-2-0-17-released
Chris@0:      */
Chris@0:     public function addXmlContent($content, $charset = 'UTF-8', $options = LIBXML_NONET)
Chris@0:     {
Chris@0:         // remove the default namespace if it's the only namespace to make XPath expressions simpler
Chris@0:         if (!preg_match('/xmlns:/', $content)) {
Chris@0:             $content = str_replace('xmlns', 'ns', $content);
Chris@0:         }
Chris@0: 
Chris@0:         $internalErrors = libxml_use_internal_errors(true);
Chris@0:         $disableEntities = libxml_disable_entity_loader(true);
Chris@0: 
Chris@0:         $dom = new \DOMDocument('1.0', $charset);
Chris@0:         $dom->validateOnParse = true;
Chris@0: 
Chris@0:         if ('' !== trim($content)) {
Chris@0:             @$dom->loadXML($content, $options);
Chris@0:         }
Chris@0: 
Chris@0:         libxml_use_internal_errors($internalErrors);
Chris@0:         libxml_disable_entity_loader($disableEntities);
Chris@0: 
Chris@0:         $this->addDocument($dom);
Chris@0: 
Chris@0:         $this->isHtml = false;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Adds a \DOMDocument to the list of nodes.
Chris@0:      *
Chris@0:      * @param \DOMDocument $dom A \DOMDocument instance
Chris@0:      */
Chris@0:     public function addDocument(\DOMDocument $dom)
Chris@0:     {
Chris@0:         if ($dom->documentElement) {
Chris@0:             $this->addNode($dom->documentElement);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Adds a \DOMNodeList to the list of nodes.
Chris@0:      *
Chris@0:      * @param \DOMNodeList $nodes A \DOMNodeList instance
Chris@0:      */
Chris@0:     public function addNodeList(\DOMNodeList $nodes)
Chris@0:     {
Chris@0:         foreach ($nodes as $node) {
Chris@0:             if ($node instanceof \DOMNode) {
Chris@0:                 $this->addNode($node);
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Adds an array of \DOMNode instances to the list of nodes.
Chris@0:      *
Chris@0:      * @param \DOMNode[] $nodes An array of \DOMNode instances
Chris@0:      */
Chris@0:     public function addNodes(array $nodes)
Chris@0:     {
Chris@0:         foreach ($nodes as $node) {
Chris@0:             $this->add($node);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Adds a \DOMNode instance to the list of nodes.
Chris@0:      *
Chris@0:      * @param \DOMNode $node A \DOMNode instance
Chris@0:      */
Chris@0:     public function addNode(\DOMNode $node)
Chris@0:     {
Chris@0:         if ($node instanceof \DOMDocument) {
Chris@0:             $node = $node->documentElement;
Chris@0:         }
Chris@0: 
Chris@0:         if (null !== $this->document && $this->document !== $node->ownerDocument) {
Chris@0:             throw new \InvalidArgumentException('Attaching DOM nodes from multiple documents in the same crawler is forbidden.');
Chris@0:         }
Chris@0: 
Chris@0:         if (null === $this->document) {
Chris@0:             $this->document = $node->ownerDocument;
Chris@0:         }
Chris@0: 
Chris@0:         // Don't add duplicate nodes in the Crawler
Chris@17:         if (\in_array($node, $this->nodes, true)) {
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         $this->nodes[] = $node;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns a node given its position in the node list.
Chris@0:      *
Chris@0:      * @param int $position The position
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     public function eq($position)
Chris@0:     {
Chris@0:         if (isset($this->nodes[$position])) {
Chris@0:             return $this->createSubCrawler($this->nodes[$position]);
Chris@0:         }
Chris@0: 
Chris@0:         return $this->createSubCrawler(null);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Calls an anonymous function on each node of the list.
Chris@0:      *
Chris@0:      * The anonymous function receives the position and the node wrapped
Chris@0:      * in a Crawler instance as arguments.
Chris@0:      *
Chris@0:      * Example:
Chris@0:      *
Chris@0:      *     $crawler->filter('h1')->each(function ($node, $i) {
Chris@0:      *         return $node->text();
Chris@0:      *     });
Chris@0:      *
Chris@0:      * @param \Closure $closure An anonymous function
Chris@0:      *
Chris@0:      * @return array An array of values returned by the anonymous function
Chris@0:      */
Chris@0:     public function each(\Closure $closure)
Chris@0:     {
Chris@17:         $data = [];
Chris@0:         foreach ($this->nodes as $i => $node) {
Chris@0:             $data[] = $closure($this->createSubCrawler($node), $i);
Chris@0:         }
Chris@0: 
Chris@0:         return $data;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Slices the list of nodes by $offset and $length.
Chris@0:      *
Chris@0:      * @param int $offset
Chris@0:      * @param int $length
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     public function slice($offset = 0, $length = null)
Chris@0:     {
Chris@17:         return $this->createSubCrawler(\array_slice($this->nodes, $offset, $length));
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Reduces the list of nodes by calling an anonymous function.
Chris@0:      *
Chris@0:      * To remove a node from the list, the anonymous function must return false.
Chris@0:      *
Chris@0:      * @param \Closure $closure An anonymous function
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     public function reduce(\Closure $closure)
Chris@0:     {
Chris@17:         $nodes = [];
Chris@0:         foreach ($this->nodes as $i => $node) {
Chris@0:             if (false !== $closure($this->createSubCrawler($node), $i)) {
Chris@0:                 $nodes[] = $node;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return $this->createSubCrawler($nodes);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the first node of the current selection.
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     public function first()
Chris@0:     {
Chris@0:         return $this->eq(0);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the last node of the current selection.
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     public function last()
Chris@0:     {
Chris@17:         return $this->eq(\count($this->nodes) - 1);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the siblings nodes of the current selection.
Chris@0:      *
Chris@0:      * @return self
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException When current node is empty
Chris@0:      */
Chris@0:     public function siblings()
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         return $this->createSubCrawler($this->sibling($this->getNode(0)->parentNode->firstChild));
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the next siblings nodes of the current selection.
Chris@0:      *
Chris@0:      * @return self
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException When current node is empty
Chris@0:      */
Chris@0:     public function nextAll()
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         return $this->createSubCrawler($this->sibling($this->getNode(0)));
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the previous sibling nodes of the current selection.
Chris@0:      *
Chris@0:      * @return self
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException
Chris@0:      */
Chris@0:     public function previousAll()
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         return $this->createSubCrawler($this->sibling($this->getNode(0), 'previousSibling'));
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the parents nodes of the current selection.
Chris@0:      *
Chris@0:      * @return self
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException When current node is empty
Chris@0:      */
Chris@0:     public function parents()
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         $node = $this->getNode(0);
Chris@17:         $nodes = [];
Chris@0: 
Chris@0:         while ($node = $node->parentNode) {
Chris@0:             if (XML_ELEMENT_NODE === $node->nodeType) {
Chris@0:                 $nodes[] = $node;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return $this->createSubCrawler($nodes);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the children nodes of the current selection.
Chris@0:      *
Chris@0:      * @return self
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException When current node is empty
Chris@0:      */
Chris@0:     public function children()
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         $node = $this->getNode(0)->firstChild;
Chris@0: 
Chris@17:         return $this->createSubCrawler($node ? $this->sibling($node) : []);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the attribute value of the first node of the list.
Chris@0:      *
Chris@0:      * @param string $attribute The attribute name
Chris@0:      *
Chris@0:      * @return string|null The attribute value or null if the attribute does not exist
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException When current node is empty
Chris@0:      */
Chris@0:     public function attr($attribute)
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         $node = $this->getNode(0);
Chris@0: 
Chris@0:         return $node->hasAttribute($attribute) ? $node->getAttribute($attribute) : null;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the node name of the first node of the list.
Chris@0:      *
Chris@0:      * @return string The node name
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException When current node is empty
Chris@0:      */
Chris@0:     public function nodeName()
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         return $this->getNode(0)->nodeName;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the node value of the first node of the list.
Chris@0:      *
Chris@0:      * @return string The node value
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException When current node is empty
Chris@0:      */
Chris@0:     public function text()
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         return $this->getNode(0)->nodeValue;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the first node of the list as HTML.
Chris@0:      *
Chris@0:      * @return string The node html
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException When current node is empty
Chris@0:      */
Chris@0:     public function html()
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         $html = '';
Chris@0:         foreach ($this->getNode(0)->childNodes as $child) {
Chris@0:             $html .= $child->ownerDocument->saveHTML($child);
Chris@0:         }
Chris@0: 
Chris@0:         return $html;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Evaluates an XPath expression.
Chris@0:      *
Chris@0:      * Since an XPath expression might evaluate to either a simple type or a \DOMNodeList,
Chris@0:      * this method will return either an array of simple types or a new Crawler instance.
Chris@0:      *
Chris@0:      * @param string $xpath An XPath expression
Chris@0:      *
Chris@0:      * @return array|Crawler An array of evaluation results or a new Crawler instance
Chris@0:      */
Chris@0:     public function evaluate($xpath)
Chris@0:     {
Chris@0:         if (null === $this->document) {
Chris@0:             throw new \LogicException('Cannot evaluate the expression on an uninitialized crawler.');
Chris@0:         }
Chris@0: 
Chris@17:         $data = [];
Chris@0:         $domxpath = $this->createDOMXPath($this->document, $this->findNamespacePrefixes($xpath));
Chris@0: 
Chris@0:         foreach ($this->nodes as $node) {
Chris@0:             $data[] = $domxpath->evaluate($xpath, $node);
Chris@0:         }
Chris@0: 
Chris@0:         if (isset($data[0]) && $data[0] instanceof \DOMNodeList) {
Chris@0:             return $this->createSubCrawler($data);
Chris@0:         }
Chris@0: 
Chris@0:         return $data;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Extracts information from the list of nodes.
Chris@0:      *
Chris@0:      * You can extract attributes or/and the node value (_text).
Chris@0:      *
Chris@0:      * Example:
Chris@0:      *
Chris@17:      *     $crawler->filter('h1 a')->extract(['_text', 'href']);
Chris@0:      *
Chris@0:      * @param array $attributes An array of attributes
Chris@0:      *
Chris@0:      * @return array An array of extracted values
Chris@0:      */
Chris@0:     public function extract($attributes)
Chris@0:     {
Chris@0:         $attributes = (array) $attributes;
Chris@17:         $count = \count($attributes);
Chris@0: 
Chris@17:         $data = [];
Chris@0:         foreach ($this->nodes as $node) {
Chris@17:             $elements = [];
Chris@0:             foreach ($attributes as $attribute) {
Chris@0:                 if ('_text' === $attribute) {
Chris@0:                     $elements[] = $node->nodeValue;
Chris@0:                 } else {
Chris@0:                     $elements[] = $node->getAttribute($attribute);
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@13:             $data[] = 1 === $count ? $elements[0] : $elements;
Chris@0:         }
Chris@0: 
Chris@0:         return $data;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Filters the list of nodes with an XPath expression.
Chris@0:      *
Chris@0:      * The XPath expression is evaluated in the context of the crawler, which
Chris@0:      * is considered as a fake parent of the elements inside it.
Chris@0:      * This means that a child selector "div" or "./div" will match only
Chris@0:      * the div elements of the current crawler, not their children.
Chris@0:      *
Chris@0:      * @param string $xpath An XPath expression
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     public function filterXPath($xpath)
Chris@0:     {
Chris@0:         $xpath = $this->relativize($xpath);
Chris@0: 
Chris@0:         // If we dropped all expressions in the XPath while preparing it, there would be no match
Chris@0:         if ('' === $xpath) {
Chris@0:             return $this->createSubCrawler(null);
Chris@0:         }
Chris@0: 
Chris@0:         return $this->filterRelativeXPath($xpath);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Filters the list of nodes with a CSS selector.
Chris@0:      *
Chris@0:      * This method only works if you have installed the CssSelector Symfony Component.
Chris@0:      *
Chris@0:      * @param string $selector A CSS selector
Chris@0:      *
Chris@0:      * @return self
Chris@0:      *
Chris@0:      * @throws \RuntimeException if the CssSelector Component is not available
Chris@0:      */
Chris@0:     public function filter($selector)
Chris@0:     {
Chris@12:         if (!class_exists(CssSelectorConverter::class)) {
Chris@12:             throw new \RuntimeException('To filter with a CSS selector, install the CssSelector component ("composer require symfony/css-selector"). Or use filterXpath instead.');
Chris@0:         }
Chris@0: 
Chris@0:         $converter = new CssSelectorConverter($this->isHtml);
Chris@0: 
Chris@0:         // The CssSelector already prefixes the selector with descendant-or-self::
Chris@0:         return $this->filterRelativeXPath($converter->toXPath($selector));
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Selects links by name or alt value for clickable images.
Chris@0:      *
Chris@0:      * @param string $value The link text
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     public function selectLink($value)
Chris@0:     {
Chris@0:         $xpath = sprintf('descendant-or-self::a[contains(concat(\' \', normalize-space(string(.)), \' \'), %s) ', static::xpathLiteral(' '.$value.' ')).
Chris@0:                             sprintf('or ./img[contains(concat(\' \', normalize-space(string(@alt)), \' \'), %s)]]', static::xpathLiteral(' '.$value.' '));
Chris@0: 
Chris@0:         return $this->filterRelativeXPath($xpath);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Selects images by alt value.
Chris@0:      *
Chris@0:      * @param string $value The image alt
Chris@0:      *
Chris@0:      * @return self A new instance of Crawler with the filtered list of nodes
Chris@0:      */
Chris@0:     public function selectImage($value)
Chris@0:     {
Chris@0:         $xpath = sprintf('descendant-or-self::img[contains(normalize-space(string(@alt)), %s)]', static::xpathLiteral($value));
Chris@0: 
Chris@0:         return $this->filterRelativeXPath($xpath);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Selects a button by name or alt value for images.
Chris@0:      *
Chris@0:      * @param string $value The button text
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     public function selectButton($value)
Chris@0:     {
Chris@0:         $translate = 'translate(@type, "ABCDEFGHIJKLMNOPQRSTUVWXYZ", "abcdefghijklmnopqrstuvwxyz")';
Chris@12:         $xpath = sprintf('descendant-or-self::input[((contains(%s, "submit") or contains(%1$s, "button")) and contains(concat(\' \', normalize-space(string(@value)), \' \'), %s)) ', $translate, static::xpathLiteral(' '.$value.' ')).
Chris@0:                          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)).
Chris@0:                          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));
Chris@0: 
Chris@0:         return $this->filterRelativeXPath($xpath);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns a Link object for the first node in the list.
Chris@0:      *
Chris@0:      * @param string $method The method for the link (get by default)
Chris@0:      *
Chris@0:      * @return Link A Link instance
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException If the current node list is empty or the selected node is not instance of DOMElement
Chris@0:      */
Chris@0:     public function link($method = 'get')
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         $node = $this->getNode(0);
Chris@0: 
Chris@0:         if (!$node instanceof \DOMElement) {
Chris@17:             throw new \InvalidArgumentException(sprintf('The selected node should be instance of DOMElement, got "%s".', \get_class($node)));
Chris@0:         }
Chris@0: 
Chris@0:         return new Link($node, $this->baseHref, $method);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns an array of Link objects for the nodes in the list.
Chris@0:      *
Chris@0:      * @return Link[] An array of Link instances
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException If the current node list contains non-DOMElement instances
Chris@0:      */
Chris@0:     public function links()
Chris@0:     {
Chris@17:         $links = [];
Chris@0:         foreach ($this->nodes as $node) {
Chris@0:             if (!$node instanceof \DOMElement) {
Chris@17:                 throw new \InvalidArgumentException(sprintf('The current node list should contain only DOMElement instances, "%s" found.', \get_class($node)));
Chris@0:             }
Chris@0: 
Chris@0:             $links[] = new Link($node, $this->baseHref, 'get');
Chris@0:         }
Chris@0: 
Chris@0:         return $links;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns an Image object for the first node in the list.
Chris@0:      *
Chris@0:      * @return Image An Image instance
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException If the current node list is empty
Chris@0:      */
Chris@0:     public function image()
Chris@0:     {
Chris@17:         if (!\count($this)) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         $node = $this->getNode(0);
Chris@0: 
Chris@0:         if (!$node instanceof \DOMElement) {
Chris@17:             throw new \InvalidArgumentException(sprintf('The selected node should be instance of DOMElement, got "%s".', \get_class($node)));
Chris@0:         }
Chris@0: 
Chris@0:         return new Image($node, $this->baseHref);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns an array of Image objects for the nodes in the list.
Chris@0:      *
Chris@0:      * @return Image[] An array of Image instances
Chris@0:      */
Chris@0:     public function images()
Chris@0:     {
Chris@17:         $images = [];
Chris@0:         foreach ($this as $node) {
Chris@0:             if (!$node instanceof \DOMElement) {
Chris@17:                 throw new \InvalidArgumentException(sprintf('The current node list should contain only DOMElement instances, "%s" found.', \get_class($node)));
Chris@0:             }
Chris@0: 
Chris@0:             $images[] = new Image($node, $this->baseHref);
Chris@0:         }
Chris@0: 
Chris@0:         return $images;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns a Form object for the first node in the list.
Chris@0:      *
Chris@0:      * @param array  $values An array of values for the form fields
Chris@0:      * @param string $method The method for the form
Chris@0:      *
Chris@0:      * @return Form A Form instance
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException If the current node list is empty or the selected node is not instance of DOMElement
Chris@0:      */
Chris@0:     public function form(array $values = null, $method = null)
Chris@0:     {
Chris@0:         if (!$this->nodes) {
Chris@0:             throw new \InvalidArgumentException('The current node list is empty.');
Chris@0:         }
Chris@0: 
Chris@0:         $node = $this->getNode(0);
Chris@0: 
Chris@0:         if (!$node instanceof \DOMElement) {
Chris@17:             throw new \InvalidArgumentException(sprintf('The selected node should be instance of DOMElement, got "%s".', \get_class($node)));
Chris@0:         }
Chris@0: 
Chris@0:         $form = new Form($node, $this->uri, $method, $this->baseHref);
Chris@0: 
Chris@0:         if (null !== $values) {
Chris@0:             $form->setValues($values);
Chris@0:         }
Chris@0: 
Chris@0:         return $form;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Overloads a default namespace prefix to be used with XPath and CSS expressions.
Chris@0:      *
Chris@0:      * @param string $prefix
Chris@0:      */
Chris@0:     public function setDefaultNamespacePrefix($prefix)
Chris@0:     {
Chris@0:         $this->defaultNamespacePrefix = $prefix;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param string $prefix
Chris@0:      * @param string $namespace
Chris@0:      */
Chris@0:     public function registerNamespace($prefix, $namespace)
Chris@0:     {
Chris@0:         $this->namespaces[$prefix] = $namespace;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Converts string for XPath expressions.
Chris@0:      *
Chris@0:      * Escaped characters are: quotes (") and apostrophe (').
Chris@0:      *
Chris@0:      *  Examples:
Chris@17:      *
Chris@0:      *     echo Crawler::xpathLiteral('foo " bar');
Chris@0:      *     //prints 'foo " bar'
Chris@0:      *
Chris@0:      *     echo Crawler::xpathLiteral("foo ' bar");
Chris@0:      *     //prints "foo ' bar"
Chris@0:      *
Chris@0:      *     echo Crawler::xpathLiteral('a\'b"c');
Chris@0:      *     //prints concat('a', "'", 'b"c')
Chris@17:      *
Chris@0:      *
Chris@0:      * @param string $s String to be escaped
Chris@0:      *
Chris@0:      * @return string Converted string
Chris@0:      */
Chris@0:     public static function xpathLiteral($s)
Chris@0:     {
Chris@0:         if (false === strpos($s, "'")) {
Chris@0:             return sprintf("'%s'", $s);
Chris@0:         }
Chris@0: 
Chris@0:         if (false === strpos($s, '"')) {
Chris@0:             return sprintf('"%s"', $s);
Chris@0:         }
Chris@0: 
Chris@0:         $string = $s;
Chris@17:         $parts = [];
Chris@0:         while (true) {
Chris@0:             if (false !== $pos = strpos($string, "'")) {
Chris@0:                 $parts[] = sprintf("'%s'", substr($string, 0, $pos));
Chris@0:                 $parts[] = "\"'\"";
Chris@0:                 $string = substr($string, $pos + 1);
Chris@0:             } else {
Chris@0:                 $parts[] = "'$string'";
Chris@0:                 break;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return sprintf('concat(%s)', implode(', ', $parts));
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Filters the list of nodes with an XPath expression.
Chris@0:      *
Chris@0:      * The XPath expression should already be processed to apply it in the context of each node.
Chris@0:      *
Chris@0:      * @param string $xpath
Chris@0:      *
Chris@0:      * @return self
Chris@0:      */
Chris@0:     private function filterRelativeXPath($xpath)
Chris@0:     {
Chris@0:         $prefixes = $this->findNamespacePrefixes($xpath);
Chris@0: 
Chris@0:         $crawler = $this->createSubCrawler(null);
Chris@0: 
Chris@0:         foreach ($this->nodes as $node) {
Chris@0:             $domxpath = $this->createDOMXPath($node->ownerDocument, $prefixes);
Chris@0:             $crawler->add($domxpath->query($xpath, $node));
Chris@0:         }
Chris@0: 
Chris@0:         return $crawler;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Make the XPath relative to the current context.
Chris@0:      *
Chris@0:      * The returned XPath will match elements matching the XPath inside the current crawler
Chris@0:      * when running in the context of a node of the crawler.
Chris@0:      *
Chris@0:      * @param string $xpath
Chris@0:      *
Chris@0:      * @return string
Chris@0:      */
Chris@0:     private function relativize($xpath)
Chris@0:     {
Chris@17:         $expressions = [];
Chris@0: 
Chris@0:         // An expression which will never match to replace expressions which cannot match in the crawler
Chris@18:         // We cannot drop
Chris@0:         $nonMatchingExpression = 'a[name() = "b"]';
Chris@0: 
Chris@17:         $xpathLen = \strlen($xpath);
Chris@0:         $openedBrackets = 0;
Chris@0:         $startPosition = strspn($xpath, " \t\n\r\0\x0B");
Chris@0: 
Chris@0:         for ($i = $startPosition; $i <= $xpathLen; ++$i) {
Chris@0:             $i += strcspn($xpath, '"\'[]|', $i);
Chris@0: 
Chris@0:             if ($i < $xpathLen) {
Chris@0:                 switch ($xpath[$i]) {
Chris@0:                     case '"':
Chris@0:                     case "'":
Chris@0:                         if (false === $i = strpos($xpath, $xpath[$i], $i + 1)) {
Chris@0:                             return $xpath; // The XPath expression is invalid
Chris@0:                         }
Chris@0:                         continue 2;
Chris@0:                     case '[':
Chris@0:                         ++$openedBrackets;
Chris@0:                         continue 2;
Chris@0:                     case ']':
Chris@0:                         --$openedBrackets;
Chris@0:                         continue 2;
Chris@0:                 }
Chris@0:             }
Chris@0:             if ($openedBrackets) {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             if ($startPosition < $xpathLen && '(' === $xpath[$startPosition]) {
Chris@0:                 // If the union is inside some braces, we need to preserve the opening braces and apply
Chris@0:                 // the change only inside it.
Chris@0:                 $j = 1 + strspn($xpath, "( \t\n\r\0\x0B", $startPosition + 1);
Chris@0:                 $parenthesis = substr($xpath, $startPosition, $j);
Chris@0:                 $startPosition += $j;
Chris@0:             } else {
Chris@0:                 $parenthesis = '';
Chris@0:             }
Chris@0:             $expression = rtrim(substr($xpath, $startPosition, $i - $startPosition));
Chris@0: 
Chris@0:             if (0 === strpos($expression, 'self::*/')) {
Chris@0:                 $expression = './'.substr($expression, 8);
Chris@0:             }
Chris@0: 
Chris@0:             // add prefix before absolute element selector
Chris@0:             if ('' === $expression) {
Chris@0:                 $expression = $nonMatchingExpression;
Chris@0:             } elseif (0 === strpos($expression, '//')) {
Chris@0:                 $expression = 'descendant-or-self::'.substr($expression, 2);
Chris@0:             } elseif (0 === strpos($expression, './/')) {
Chris@0:                 $expression = 'descendant-or-self::'.substr($expression, 3);
Chris@0:             } elseif (0 === strpos($expression, './')) {
Chris@0:                 $expression = 'self::'.substr($expression, 2);
Chris@0:             } elseif (0 === strpos($expression, 'child::')) {
Chris@0:                 $expression = 'self::'.substr($expression, 7);
Chris@0:             } elseif ('/' === $expression[0] || '.' === $expression[0] || 0 === strpos($expression, 'self::')) {
Chris@0:                 $expression = $nonMatchingExpression;
Chris@0:             } elseif (0 === strpos($expression, 'descendant::')) {
Chris@0:                 $expression = 'descendant-or-self::'.substr($expression, 12);
Chris@0:             } elseif (preg_match('/^(ancestor|ancestor-or-self|attribute|following|following-sibling|namespace|parent|preceding|preceding-sibling)::/', $expression)) {
Chris@0:                 // the fake root has no parent, preceding or following nodes and also no attributes (even no namespace attributes)
Chris@0:                 $expression = $nonMatchingExpression;
Chris@0:             } elseif (0 !== strpos($expression, 'descendant-or-self::')) {
Chris@0:                 $expression = 'self::'.$expression;
Chris@0:             }
Chris@0:             $expressions[] = $parenthesis.$expression;
Chris@0: 
Chris@0:             if ($i === $xpathLen) {
Chris@0:                 return implode(' | ', $expressions);
Chris@0:             }
Chris@0: 
Chris@0:             $i += strspn($xpath, " \t\n\r\0\x0B", $i + 1);
Chris@0:             $startPosition = $i + 1;
Chris@0:         }
Chris@0: 
Chris@0:         return $xpath; // The XPath expression is invalid
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param int $position
Chris@0:      *
Chris@0:      * @return \DOMElement|null
Chris@0:      */
Chris@0:     public function getNode($position)
Chris@0:     {
Chris@0:         if (isset($this->nodes[$position])) {
Chris@0:             return $this->nodes[$position];
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @return int
Chris@0:      */
Chris@0:     public function count()
Chris@0:     {
Chris@17:         return \count($this->nodes);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@12:      * @return \ArrayIterator|\DOMElement[]
Chris@0:      */
Chris@0:     public function getIterator()
Chris@0:     {
Chris@0:         return new \ArrayIterator($this->nodes);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param \DOMElement $node
Chris@0:      * @param string      $siblingDir
Chris@0:      *
Chris@0:      * @return array
Chris@0:      */
Chris@0:     protected function sibling($node, $siblingDir = 'nextSibling')
Chris@0:     {
Chris@17:         $nodes = [];
Chris@0: 
Chris@0:         do {
Chris@12:             if ($node !== $this->getNode(0) && 1 === $node->nodeType) {
Chris@0:                 $nodes[] = $node;
Chris@0:             }
Chris@0:         } while ($node = $node->$siblingDir);
Chris@0: 
Chris@0:         return $nodes;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param \DOMDocument $document
Chris@0:      * @param array        $prefixes
Chris@0:      *
Chris@0:      * @return \DOMXPath
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException
Chris@0:      */
Chris@17:     private function createDOMXPath(\DOMDocument $document, array $prefixes = [])
Chris@0:     {
Chris@0:         $domxpath = new \DOMXPath($document);
Chris@0: 
Chris@0:         foreach ($prefixes as $prefix) {
Chris@0:             $namespace = $this->discoverNamespace($domxpath, $prefix);
Chris@0:             if (null !== $namespace) {
Chris@0:                 $domxpath->registerNamespace($prefix, $namespace);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         return $domxpath;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param \DOMXPath $domxpath
Chris@0:      * @param string    $prefix
Chris@0:      *
Chris@0:      * @return string
Chris@0:      *
Chris@0:      * @throws \InvalidArgumentException
Chris@0:      */
Chris@0:     private function discoverNamespace(\DOMXPath $domxpath, $prefix)
Chris@0:     {
Chris@0:         if (isset($this->namespaces[$prefix])) {
Chris@0:             return $this->namespaces[$prefix];
Chris@0:         }
Chris@0: 
Chris@0:         // ask for one namespace, otherwise we'd get a collection with an item for each node
Chris@0:         $namespaces = $domxpath->query(sprintf('(//namespace::*[name()="%s"])[last()]', $this->defaultNamespacePrefix === $prefix ? '' : $prefix));
Chris@0: 
Chris@0:         if ($node = $namespaces->item(0)) {
Chris@0:             return $node->nodeValue;
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param string $xpath
Chris@0:      *
Chris@0:      * @return array
Chris@0:      */
Chris@0:     private function findNamespacePrefixes($xpath)
Chris@0:     {
Chris@0:         if (preg_match_all('/(?P<prefix>[a-z_][a-z_0-9\-\.]*+):[^"\/:]/i', $xpath, $matches)) {
Chris@0:             return array_unique($matches['prefix']);
Chris@0:         }
Chris@0: 
Chris@17:         return [];
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Creates a crawler for some subnodes.
Chris@0:      *
Chris@0:      * @param \DOMElement|\DOMElement[]|\DOMNodeList|null $nodes
Chris@0:      *
Chris@0:      * @return static
Chris@0:      */
Chris@0:     private function createSubCrawler($nodes)
Chris@0:     {
Chris@0:         $crawler = new static($nodes, $this->uri, $this->baseHref);
Chris@0:         $crawler->isHtml = $this->isHtml;
Chris@0:         $crawler->document = $this->document;
Chris@0:         $crawler->namespaces = $this->namespaces;
Chris@0: 
Chris@0:         return $crawler;
Chris@0:     }
Chris@0: }