/**
* @dataProvider getGetUriTests
*/
public function testGetUriOnArea($url, $currentUri, $expected)
{
$dom = new \DOMDocument();
$dom->loadHTML(sprintf('<html><map><area href="%s" /></map></html>', $url));
$link = new Link($dom->getElementsByTagName('area')->item(0), $currentUri);
$this->assertEquals($expected, $link->getUri());
}
/**
* @return PagesCollection New Collection with all found links
*/
public function findLinkedPages()
{
$pages = new PagesCollection();
$crawler = $this->getCrawler();
foreach ($crawler->filter('a') as $node) {
$link = new Link($node, $this->uri);
$page = new Page($link->getUri());
$page->setClient($this->client);
$pages->add($page);
}
return $pages;
}
public function testGetters()
{
$dom = new \DOMDocument();
$dom->loadHTML('<html><a href="/foo">foo</a></html>');
$node = $dom->getElementsByTagName('a')->item(0);
$link = new Link($node);
$this->assertEquals('/foo', $link->getUri(), '->getUri() returns the URI of the link');
$this->assertEquals($node, $link->getNode(), '->getNode() returns the node associated with the link');
$this->assertEquals('get', $link->getMethod(), '->getMethod() returns the method of the link');
$link = new Link($node, 'post');
$this->assertEquals('post', $link->getMethod(), '->getMethod() returns the method of the link');
$link = new Link($node, 'get', 'http://localhost', '/bar/');
$this->assertEquals('http://localhost/foo', $link->getUri(), '->getUri() returns the absolute URI of the link');
$this->assertEquals('/foo', $link->getUri(false), '->getUri() returns the relative URI of the link if false is the first argument');
$dom = new \DOMDocument();
$dom->loadHTML('<html><a href="foo">foo</a></html>');
$node = $dom->getElementsByTagName('a')->item(0);
$link = new Link($node, 'get', 'http://localhost', '/bar/');
$this->assertEquals('http://localhost/bar/foo', $link->getUri(), '->getUri() returns the absolute URI of the link for relative hrefs');
$this->assertEquals('/bar/foo', $link->getUri(false), '->getUri() returns the relative URI of the link if false is the first argument');
$dom = new \DOMDocument();
$dom->loadHTML('<html><a href="http://login.foo.com/foo">foo</a></html>');
$node = $dom->getElementsByTagName('a')->item(0);
$link = new Link($node, 'get', 'http://www.foo.com');
$this->assertEquals('http://login.foo.com/foo', $link->getUri(), '->getUri() returns the absolute URI of the link, regardless of the context of the object');
$link = new Link($node, 'get');
$this->assertEquals('http://login.foo.com/foo', $link->getUri(), '->getUri() returns the absolute URI of the link, regardless of the context of the object');
$link = new Link($node, 'get', null, '/bar/');
$this->assertEquals('http://login.foo.com/foo', $link->getUri(), '->getUri() returns the absolute URI of the link, regardless of the context of the object');
$link = new Link($node, 'get', 'http://www.foo.com', '/bar/');
$this->assertEquals('http://login.foo.com/foo', $link->getUri(), '->getUri() returns the absolute URI of the link, regardless of the context of the object');
$dom = new \DOMDocument();
$dom->loadHTML('<html><a href="?get=param">foo</a></html>');
$node = $dom->getElementsByTagName('a')->item(0);
$link = new Link($node, 'get', 'http://www.foo.com', '/foo/bar');
$this->assertEquals('http://www.foo.com/foo/bar?get=param', $link->getUri(), '->getUri() returns the absolute URI of the link, regardless of the context of the object');
$link = new Link($node, 'get', 'http://www.foo.com', '/foo/bar');
$this->assertEquals('/foo/bar?get=param', $link->getUri(false), '->getUri() returns the relative URI of the link if false is the first argument');
$dom = new \DOMDocument();
$dom->loadHTML('<html><a href="test.html">foo</a></html>');
$node = $dom->getElementsByTagName('a')->item(0);
$link = new Link($node, 'get', null, '/foo/bar', 'http://www.foo.com/');
$this->assertEquals('http://www.foo.com/test.html', $link->getUri());
}
/**
* @return null|string
*/
public function getUri()
{
switch ($this->type) {
case 'application/rss+xml':
case 'application/x.atom+xml':
case 'application/atom+xml':
case 'application/rdf+xml':
return parent::getUri();
default:
break;
}
return null;
}
/**
* Gets the URI of the form.
*
* The returned URI is not the same as the form "action" attribute.
* This method merges the value if the method is GET to mimics
* browser behavior.
*
* @return string The URI
*
* @api
*/
public function getUri()
{
$uri = parent::getUri();
if (!in_array($this->getMethod(), array('POST', 'PUT', 'DELETE', 'PATCH'))) {
$query = parse_url($uri, PHP_URL_QUERY);
$currentParameters = array();
if ($query) {
parse_str($query, $currentParameters);
}
$queryString = http_build_query(array_merge($currentParameters, $this->getValues()), null, '&');
$pos = strpos($uri, '?');
$base = false === $pos ? $uri : substr($uri, 0, $pos);
$uri = rtrim($base . '?' . $queryString, '?');
}
return $uri;
}
/**
* Gets the URI of the form.
*
* The returned URI is not the same as the form "action" attribute.
* This method merges the value if the method is GET to mimics
* browser behavior.
*
* @return string The URI
*
* @api
*/
public function getUri()
{
$uri = parent::getUri();
if (!in_array($this->getMethod(), array('POST', 'PUT', 'DELETE', 'PATCH')) && ($queryString = http_build_query($this->getValues(), null, '&'))) {
$sep = false === strpos($uri, '?') ? '?' : '&';
$uri .= $sep . $queryString;
}
return $uri;
}
/**
* 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 $charset The charset
*
* @api
*/
public function addHtmlContent($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()
if (function_exists('mb_convert_encoding')) {
$content = mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
} elseif (function_exists('iconv')) {
$content = preg_replace_callback('/[\\x80-\\xFF]+/', function ($m) {
$m = unpack('C*', $m[0]);
$i = 1;
$entities = '';
while (isset($m[$i])) {
if (0xf0 <= $m[$i]) {
$c = ($m[$i++] - 0xf0 << 18) + ($m[$i++] - 0x80 << 12) + ($m[$i++] - 0x80 << 6) + $m[$i++] - 0x80;
} elseif (0xe0 <= $m[$i]) {
$c = ($m[$i++] - 0xe0 << 12) + ($m[$i++] - 0x80 << 6) + $m[$i++] - 0x80;
} else {
$c = ($m[$i++] - 0xc0 << 6) + $m[$i++] - 0x80;
}
$entities .= '&#' . $c . ';';
}
return $entities;
}, iconv($charset, 'UTF-8', $content));
}
} 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 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 $charset The charset
*
* @api
*/
public function addHtmlContent($content, $charset = 'UTF-8')
{
$current = libxml_use_internal_errors(true);
$disableEntities = libxml_disable_entity_loader(true);
$dom = new \DOMDocument('1.0', $charset);
$dom->validateOnParse = true;
if (function_exists('mb_convert_encoding') && in_array(strtolower($charset), array_map('strtolower', mb_list_encodings()))) {
$content = mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
}
@$dom->loadHTML($content);
libxml_use_internal_errors($current);
libxml_disable_entity_loader($disableEntities);
$this->addDocument($dom);
$base = $this->filterXPath('descendant-or-self::base')->extract(array('href'));
$baseHref = current($base);
if (count($base) && !empty($baseHref)) {
if ($this->uri) {
$linkNode = $dom->createElement('a');
$linkNode->setAttribute('href', $baseHref);
$link = new Link($linkNode, $this->uri);
$this->uri = $link->getUri();
} else {
$this->uri = $baseHref;
}
}
}
/**
* 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 $charset The charset
*
* @api
*/
public function addHtmlContent($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;
if (function_exists('mb_convert_encoding')) {
$hasError = false;
set_error_handler(function () use(&$hasError) {
$hasError = true;
});
$tmpContent = @mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
restore_error_handler();
if (!$hasError) {
$content = $tmpContent;
}
}
if ('' !== trim($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->uri) {
$linkNode = $dom->createElement('a');
$linkNode->setAttribute('href', $baseHref);
$link = new Link($linkNode, $this->uri);
$this->uri = $link->getUri();
} else {
$this->uri = $baseHref;
}
}
}
/**
* Clicks on a given link.
*
* @param Link $link A Link instance
*/
public function click(Link $link)
{
return $this->request($link->getMethod(), $link->getUri());
}
/**
* Gets the URI of the form.
*
* The returned URI is not the same as the form "action" attribute.
* This method merges the value if the method is GET to mimics
* browser behavior.
*
* @return string The URI
*
* @api
*/
public function getUri()
{
$uri = parent::getUri();
if (!in_array($this->getMethod(), array('post', 'put', 'delete')) && $queryString = http_build_query($this->getValues(), null, '&')) {
$sep = false === strpos($uri, '?') ? '?' : '&';
$uri .= $sep.$queryString;
}
return $uri;
}
/**
* Link constructor.
*
* @param DOMElement $node
* @param string $currentUri
*/
public function __construct(DOMElement $node, $currentUri)
{
parent::__construct($node, $currentUri);
}
/**
* 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;
}
}
}
