X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=src%2FUtil%2FParseUrl.php;h=4611cf8efe5f25378caa699044543af6269bab4c;hb=b5ad8c3e153976cf3e63753597377f09852d98d7;hp=0183d6b14ae5be6785f90eaf5c16af01e6974de9;hpb=8b7cb5d9efeab580c2592e0fbe301f7142b73a3d;p=friendica.git

diff --git a/src/Util/ParseUrl.php b/src/Util/ParseUrl.php
index 0183d6b14a..4611cf8efe 100644
--- a/src/Util/ParseUrl.php
+++ b/src/Util/ParseUrl.php
@@ -187,7 +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 [
@@ -214,7 +214,7 @@ class ParseUrl
 		];
 
 		if ($count > 10) {
-			Logger::notice('Endless loop detected', ['url' => $url]);
+			Logger::warning('Endless loop detected', ['url' => $url]);
 			return $siteinfo;
 		}
 
@@ -287,22 +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') {
@@ -315,6 +315,10 @@ class ParseUrl
 
 		$body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
 
+		if (empty($body)) {
+			return $siteinfo;
+		}
+
 		$doc = new DOMDocument();
 		@$doc->loadHTML($body);
 
@@ -641,6 +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
@@ -662,6 +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)
@@ -725,6 +731,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseParts(array $siteinfo, array $jsonld): array
@@ -770,6 +777,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLd(array $siteinfo, array $jsonld): array
@@ -863,6 +871,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdAuthor(array $siteinfo, array $jsonld): array
@@ -947,6 +956,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdArticle(array $siteinfo, array $jsonld): array
@@ -1017,6 +1027,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdWebPage(array $siteinfo, array $jsonld): array
@@ -1029,17 +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);
 		}
 
@@ -1056,6 +1067,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdWebSite(array $siteinfo, array $jsonld): array
@@ -1063,22 +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);
 		}
 
@@ -1094,6 +1106,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdWebOrganization(array $siteinfo, array $jsonld): array
@@ -1101,32 +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);
 		}
 
@@ -1140,6 +1153,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdWebPerson(array $siteinfo, array $jsonld): array
@@ -1147,12 +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);
 		}
 
@@ -1162,7 +1176,7 @@ class ParseUrl
 		}
 
 		$content = JsonLD::fetchElement($jsonld, 'url');
-		if (!empty($content)) {
+		if (!empty($content) && is_string($content)) {
 			$jsonldinfo['author_url'] = trim($content);
 		}
 
@@ -1185,6 +1199,7 @@ class ParseUrl
 	 *
 	 * @param array $siteinfo
 	 * @param array $jsonld
+	 *
 	 * @return array siteinfo
 	 */
 	private static function parseJsonLdMediaObject(array $siteinfo, array $jsonld, string $name): array
@@ -1192,22 +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);
 		}
 
@@ -1217,27 +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);
 		}