Merge pull request #14708 from annando/exception

Fix exception "Argument #2 ($jsonld) must be of type array, string given"

src/Util/ParseUrl.php

@@ -116,9 +116,7 @@ class ParseUrl
 
 		$urlHash = hash('sha256', $url);
 
-		$parsed_url = DBA::selectFirst('parsed_url', ['content'],
+		$parsed_url = DBA::selectFirst('parsed_url', ['content'], ['url_hash' => $urlHash, 'oembed' => false]);
-			['url_hash' => $urlHash, 'oembed' => false]
-		);
 		if (!empty($parsed_url['content'])) {
 			$data = unserialize($parsed_url['content']);
 			return $data;
@@ -246,11 +244,13 @@ class ParseUrl
 		if ($cacheControlHeader = $curlResult->getHeader('Cache-Control')[0] ?? '') {
 			if (preg_match('/max-age=([0-9]+)/i', $cacheControlHeader, $matches)) {
 				$maxAge = max(86400, (int)array_pop($matches));
+
 				$siteinfo['expires'] = DateTimeFormat::utc("now + $maxAge seconds");
 			}
 		}
 
 		$body = $curlResult->getBodyString();
+
 		$siteinfo['size'] = mb_strlen($body);
 
 		$charset = '';
@@ -260,7 +260,8 @@ class ParseUrl
 			if (isset($mediaType->parameters['charset'])) {
 				$charset = $mediaType->parameters['charset'];
 			}
-		} catch(\InvalidArgumentException $e) {}
+		} catch(\InvalidArgumentException $e) {
+		}
 
 		$siteinfo['charset'] = $charset;
 
@@ -307,9 +308,8 @@ class ParseUrl
 
 			if (@$meta_tag['http-equiv'] == 'refresh') {
 				$path    = $meta_tag['content'];
-				$pathinfo = explode(';', $path);
 				$content = '';
-				foreach ($pathinfo as $value) {
+				foreach (explode(';', $path) as $value) {
 					if (substr(strtolower($value), 0, 4) == 'url=') {
 						$content = substr($value, 4);
 					}
@@ -455,7 +455,8 @@ class ParseUrl
 		$list = $xpath->query("//script[@type='application/ld+json']");
 		foreach ($list as $node) {
 			if (!empty($node->nodeValue)) {
-				if ($jsonld = json_decode($node->nodeValue, true)) {
+				$jsonld = json_decode($node->nodeValue, true);
+				if (is_array($jsonld)) {
 					$siteinfo = self::parseParts($siteinfo, $jsonld);
 				}
 			}
@@ -488,6 +489,7 @@ class ParseUrl
 
 		if (!empty($siteinfo['text']) && mb_strlen($siteinfo['text']) > self::MAX_DESC_COUNT) {
 			$siteinfo['text'] = mb_substr($siteinfo['text'], 0, self::MAX_DESC_COUNT) . '…';
+
 			$pos = mb_strrpos($siteinfo['text'], '.');
 			if ($pos > self::MIN_DESC_COUNT) {
 				$siteinfo['text'] = mb_substr($siteinfo['text'], 0, $pos + 1);
@@ -522,6 +524,7 @@ class ParseUrl
 				 */
 				if (!empty($image['url'])) {
 					$image['url'] = self::completeUrl($image['url'], $page_url);
+
 					$photodata = Images::getInfoFromURLCached($image['url']);
 					if (($photodata) && ($photodata[0] > 50) && ($photodata[1] > 50)) {
 						$image['src']         = $image['url'];
@@ -552,6 +555,7 @@ class ParseUrl
 					foreach (['embed', 'content', 'url'] as $field) {
 						if (!empty($media[$field])) {
 							$media[$field] = self::completeUrl($media[$field], $page_url);
+
 							$type = self::getContentType($media[$field]);
 							if (($type[0] ?? '') == 'text') {
 								if ($field == 'embed') {
@@ -811,7 +815,7 @@ class ParseUrl
 			case 'Person':
 			case 'Patient':
 			case 'PerformingGroup':
-			case 'DanceGroup';
+			case 'DanceGroup':
 			case 'MusicGroup':
 			case 'TheaterGroup':
 				return self::parseJsonLdWebPerson($siteinfo, $jsonld);
@@ -954,8 +958,7 @@ class ParseUrl
 			$content = JsonLD::fetchElement($jsonld, 'keywords');
 			if (!empty($content)) {
 				$siteinfo['keywords'] = [];
-				$keywords = explode(',', $content);
+				foreach (explode(',', $content) as $keyword) {
-				foreach ($keywords as $keyword) {
 					$siteinfo['keywords'][] = trim($keyword);
 				}
 			}