← All changes | vendor/symfony/dom-crawler/Crawler.php +28 -11 4.0.5 → 7.2.0 View file →

		@@ -80,9 +80,9 @@
80	80
81	81	/**
82	82	* @param \DOMNodeList|\DOMNode|\DOMNode[]|string|null $node A Node to use as the base for the crawling
83	83	*/
84		- public function __construct($node = null, string $uri = null, string $baseHref = null)
	84	+ public function __construct($node = null, ?string $uri = null, ?string $baseHref = null)
85	85	{
86	86	$this->uri = $uri;
87	87	$this->baseHref = $baseHref ?: $uri;
88	88	$this->html5Parser = class_exists(HTML5::class) ? new HTML5(['disable_html_ns' => true]) : null;
		@@ -152,9 +152,9 @@
152	152	* If the charset is not set via the content type, it is assumed to be UTF-8,
153	153	* or ISO-8859-1 as a fallback, which is the default charset defined by the
154	154	* HTTP 1.1 specification.
155	155	*/
156		- public function addContent(string $content, string $type = null)
	156	+ public function addContent(string $content, ?string $type = null)
157	157	{
158	158	if (empty($type)) {
159	159	$type = str_starts_with($content, '<?xml') ? 'application/xml' : 'text/html';
160	160	}
		@@ -239,9 +239,8 @@
239	239	$disableEntities = libxml_disable_entity_loader(true);
240	240	}
241	241
242	242	$dom = new \DOMDocument('1.0', $charset);
243		- $dom->validateOnParse = true;
244	243
245	244	if ('' !== trim($content)) {
246	245	@$dom->loadXML($content, $options);
247	246	}
		@@ -365,9 +364,9 @@
365	364	* Slices the list of nodes by $offset and $length.
366	365	*
367	366	* @return static
368	367	*/
369		- public function slice(int $offset = 0, int $length = null)
	368	+ public function slice(int $offset = 0, ?int $length = null)
370	369	{
371	370	return $this->createSubCrawler(\array_slice($this->nodes, $offset, $length));
372	371	}
373	372
		@@ -545,9 +544,9 @@
545	544	*
546	545	* @throws \InvalidArgumentException When current node is empty
547	546	* @throws \RuntimeException If the CssSelector Component is not available and $selector is provided
548	547	*/
549		- public function children(string $selector = null)
	548	+ public function children(?string $selector = null)
550	549	{
551	550	if (!$this->nodes) {
552	551	throw new \InvalidArgumentException('The current node list is empty.');
553	552	}
		@@ -609,9 +608,9 @@
609	608	* @return string
610	609	*
611	610	* @throws \InvalidArgumentException When current node is empty
612	611	*/
613		- public function text(string $default = null, bool $normalizeWhitespace = true)
	612	+ public function text(?string $default = null, bool $normalizeWhitespace = true)
614	613	{
615	614	if (!$this->nodes) {
616	615	if (null !== $default) {
617	616	return $default;
		@@ -645,9 +644,9 @@
645	644	* @return string
646	645	*
647	646	* @throws \InvalidArgumentException When current node is empty
648	647	*/
649		- public function html(string $default = null)
	648	+ public function html(?string $default = null)
650	649	{
651	650	if (!$this->nodes) {
652	651	if (null !== $default) {
653	652	return $default;
		@@ -777,9 +776,9 @@
777	776	* This method only works if you have installed the CssSelector Symfony Component.
778	777	*
779	778	* @return static
780	779	*
781		- * @throws \RuntimeException if the CssSelector Component is not available
	780	+ * @throws \LogicException if the CssSelector Component is not available
782	781	*/
783	782	public function filter(string $selector)
784	783	{
785	784	$converter = $this->createCssSelectorConverter();
		@@ -914,9 +913,9 @@
914	913	* @return Form
915	914	*
916	915	* @throws \InvalidArgumentException If the current node list is empty or the selected node is not instance of DOMElement
917	916	*/
918		- public function form(array $values = null, string $method = null)
	917	+ public function form(?array $values = null, ?string $method = null)
919	918	{
920	919	if (!$this->nodes) {
921	920	throw new \InvalidArgumentException('The current node list is empty.');
922	921	}
		@@ -1150,14 +1149,32 @@
1150	1149	}
1151	1150
1152	1151	private function parseHtml5(string $htmlContent, string $charset = 'UTF-8'): \DOMDocument
1153	1152	{
1154		- return $this->html5Parser->parse($this->convertToHtmlEntities($htmlContent, $charset));
	1153	+ if (!$this->supportsEncoding($charset)) {
	1154	+ $htmlContent = $this->convertToHtmlEntities($htmlContent, $charset);
	1155	+ $charset = 'UTF-8';
	1156	+ }
	1157	+
	1158	+ return $this->html5Parser->parse($htmlContent, ['encoding' => $charset]);
1155	1159	}
1156	1160
	1161	+ private function supportsEncoding(string $encoding): bool
	1162	+ {
	1163	+ try {
	1164	+ return '' === @mb_convert_encoding('', $encoding, 'UTF-8');
	1165	+ } catch (\Throwable $e) {
	1166	+ return false;
	1167	+ }
	1168	+ }
	1169	+
1157	1170	private function parseXhtml(string $htmlContent, string $charset = 'UTF-8'): \DOMDocument
1158	1171	{
1159		- $htmlContent = $this->convertToHtmlEntities($htmlContent, $charset);
	1172	+ if ('UTF-8' === $charset && preg_match('//u', $htmlContent)) {
	1173	+ $htmlContent = '<?xml encoding="UTF-8">'.$htmlContent;
	1174	+ } else {
	1175	+ $htmlContent = $this->convertToHtmlEntities($htmlContent, $charset);
	1176	+ }
1160	1177
1161	1178	$internalErrors = libxml_use_internal_errors(true);
1162	1179	if (\LIBXML_VERSION < 20900) {
1163	1180	$disableEntities = libxml_disable_entity_loader(true);