X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=src%2FUtil%2FParseUrl.php;h=4611cf8efe5f25378caa699044543af6269bab4c;hb=fe442683efb8b75394d7df041716c47c81b05fbc;hp=9f0fd8a9bca82be9681597314bdd52324a341e68;hpb=78bf7f187c9fd0effdcfb6bbc2390b8ff74dcb76;p=friendica.git

diff --git a/src/Util/ParseUrl.php b/src/Util/ParseUrl.php
index 9f0fd8a9bc..4611cf8efe 100644
--- a/src/Util/ParseUrl.php
+++ b/src/Util/ParseUrl.php
@@ -57,15 +57,22 @@ class ParseUrl
 	 * Fetch the content type of the given url
 	 * @param string $url    URL of the page
 	 * @param string $accept content-type to accept
+	 * @param int    $timeout
 	 * @return array content type
 	 */
-	public static function getContentType(string $url, string $accept = HttpClientAccept::DEFAULT): array
+	public static function getContentType(string $url, string $accept = HttpClientAccept::DEFAULT, int $timeout = 0): array
 	{
-		$curlResult = DI::httpClient()->head($url, [HttpClientOptions::ACCEPT_CONTENT => $accept]);
+		if (!empty($timeout)) {
+			$options = [HttpClientOptions::TIMEOUT => $timeout];
+		} else {
+			$options = [];
+		}
+
+		$curlResult = DI::httpClient()->head($url, array_merge([HttpClientOptions::ACCEPT_CONTENT => $accept], $options));
 
-		// Workaround for systems that can't handle a HEAD request
-		if (!$curlResult->isSuccess() && ($curlResult->getReturnCode() == 405)) {
-			$curlResult = DI::httpClient()->get($url, $accept, [HttpClientOptions::CONTENT_LENGTH => 1000000]);
+		// Workaround for systems that can't handle a HEAD request. Don't retry on timeouts.
+		if (!$curlResult->isSuccess() && ($curlResult->getReturnCode() >= 400) && !in_array($curlResult->getReturnCode(), [408, 504])) {
+			$curlResult = DI::httpClient()->get($url, $accept, array_merge([HttpClientOptions::CONTENT_LENGTH => 1000000], $options));
 		}
 
 		if (!$curlResult->isSuccess()) {
@@ -180,7 +187,7 @@ class ParseUrl
 	 * </body>
 	 * @endverbatim
 	 */
-	public static function getSiteinfo(string $url, bool $do_oembed = true, int $count = 1)
+	public static function getSiteinfo(string $url, bool $do_oembed = true, int $count = 1): array
 	{
 		if (empty($url)) {
 			return [
@@ -207,7 +214,7 @@ class ParseUrl
 		];
 
 		if ($count > 10) {
-			Logger::notice('Endless loop detected', ['url' => $url]);
+			Logger::warning('Endless loop detected', ['url' => $url]);
 			return $siteinfo;
 		}
 
@@ -280,22 +287,22 @@ class ParseUrl
 		// Expected form: Content-Type: text/html; charset=ISO-8859-4
 		if (preg_match('/charset=([a-z0-9-_.\/]+)/i', $curlResult->getContentType(), $matches)) {
 			$charset = trim(trim(trim(array_pop($matches)), ';,'));
+		} else {
+			// Then in body that gets precedence
+			// Expected forms:
+			// - <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
+			// - <meta charset="utf-8">
+			// - <meta charset=utf-8>
+			// - <meta charSet="utf-8">
+			// We escape <style> and <script> tags since they can contain irrelevant charset information
+			// (see https://github.com/friendica/friendica/issues/9251#issuecomment-698636806)
+			Strings::performWithEscapedBlocks($body, '#<(?:style|script).*?</(?:style|script)>#ism', function ($body) use (&$charset) {
+				if (preg_match('/charset=["\']?([a-z0-9-_.\/]+)/i', $body, $matches)) {
+					$charset = trim(trim(trim(array_pop($matches)), ';,'));
+				}
+			});
 		}
 
-		// Then in body that gets precedence
-		// Expected forms:
-		// - <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
-		// - <meta charset="utf-8">
-		// - <meta charset=utf-8>
-		// - <meta charSet="utf-8">
-		// We escape <style> and <script> tags since they can contain irrelevant charset information
-		// (see https://github.com/friendica/friendica/issues/9251#issuecomment-698636806)
-		Strings::performWithEscapedBlocks($body, '#<(?:style|script).*?</(?:style|script)>#ism', function ($body) use (&$charset) {
-			if (preg_match('/charset=["\']?([a-z0-9-_.\/]+)/i', $body, $matches)) {
-				$charset = trim(trim(trim(array_pop($matches)), ';,'));
-			}
-		});
-
 		$siteinfo['charset'] = $charset;
 
 		if ($charset && strtoupper($charset) != 'UTF-8') {
@@ -308,6 +315,10 @@ class ParseUrl
 
 		$body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
 
+		if (empty($body)) {
+			return $siteinfo;
+		}
+
 		$doc = new DOMDocument();
 		@$doc->loadHTML($body);
 
@@ -543,12 +554,15 @@ class ParseUrl
 	{
 		if (!empty($siteinfo['images'])) {
 			array_walk($siteinfo['images'], function (&$image) use ($page_url) {
-				// According to the specifications someone could place a picture url into the content field as well.
-				// But this doesn't seem to happen in the wild, so we don't cover it here.
+				/*
+				 * According to the specifications someone could place a picture
+				 * URL into the content field as well. But this doesn't seem to
+				 * happen in the wild, so we don't cover it here.
+				 */
 				if (!empty($image['url'])) {
 					$image['url'] = self::completeUrl($image['url'], $page_url);
 					$photodata = Images::getInfoFromURLCached($image['url']);
-					if (!empty($photodata) && ($photodata[0] > 50) && ($photodata[1] > 50)) {
+					if (($photodata) && ($photodata[0] > 50) && ($photodata[1] > 50)) {
 						$image['src'] = $image['url'];
 						$image['width'] = $photodata[0];
 						$image['height'] = $photodata[1];
@@ -631,6 +645,7 @@ class ParseUrl
 	 * Convert tags from CSV to an array
 	 *
 	 * @param string $string Tags
+	 *
 	 * @return array with formatted Hashtags
 	 */
 	public static function convertTagsToArray(string $string): array
@@ -652,6 +667,7 @@ class ParseUrl
 	 *
 	 * @param string $tag The pure tag name
 	 * @param int    $k   Counter for internal use
+	 *
 	 * @return void
 	 */
 	private static function arrAddHashes(string &$tag, int $k)
@@ -715,6 +731,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseParts(array $siteinfo, array $jsonld): array
@@ -760,6 +777,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLd(array $siteinfo, array $jsonld): array
@@ -853,6 +871,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdAuthor(array $siteinfo, array $jsonld): array
@@ -937,6 +956,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdArticle(array $siteinfo, array $jsonld): array
@@ -1007,6 +1027,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdWebPage(array $siteinfo, array $jsonld): array
@@ -1019,17 +1040,17 @@ class ParseUrl
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'description');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['text'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'image');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['image'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'thumbnailUrl');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['image'] = trim($content);
 		}
 
@@ -1046,6 +1067,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdWebSite(array $siteinfo, array $jsonld): array
@@ -1053,22 +1075,22 @@ class ParseUrl
 		$jsonldinfo = [];
 
 		$content = JsonLD::fetchElement($jsonld, 'name');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_name'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'description');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_description'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'url');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_url'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'thumbnailUrl');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['image'] = trim($content);
 		}
 
@@ -1084,6 +1106,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdWebOrganization(array $siteinfo, array $jsonld): array
@@ -1091,32 +1114,32 @@ class ParseUrl
 		$jsonldinfo = [];
 
 		$content = JsonLD::fetchElement($jsonld, 'name');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_name'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'description');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_description'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'url');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_url'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'logo', 'url', '@type', 'ImageObject');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_img'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'brand', 'name', '@type', 'Organization');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_name'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'brand', 'url', '@type', 'Organization');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['publisher_url'] = trim($content);
 		}
 
@@ -1130,6 +1153,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdWebPerson(array $siteinfo, array $jsonld): array
@@ -1137,12 +1161,12 @@ class ParseUrl
 		$jsonldinfo = [];
 
 		$content = JsonLD::fetchElement($jsonld, 'name');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['author_name'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'description');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['author_description'] = trim($content);
 		}
 
@@ -1152,7 +1176,7 @@ class ParseUrl
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'url');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['author_url'] = trim($content);
 		}
 
@@ -1175,6 +1199,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdMediaObject(array $siteinfo, array $jsonld, string $name): array
@@ -1182,22 +1207,22 @@ class ParseUrl
 		$media = [];
 
 		$content = JsonLD::fetchElement($jsonld, 'caption');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['caption'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'url');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['url'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'mainEntityOfPage');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['main'] = Strings::compareLink($content, $siteinfo['url']);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'description');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['description'] = trim($content);
 		}
 
@@ -1207,27 +1232,27 @@ class ParseUrl
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'contentUrl');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['content'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'embedUrl');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['embed'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'height');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['height'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'width');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['width'] = trim($content);
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'image');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$media['image'] = trim($content);
 		}