Merge pull request #11973 from MrPetovan/task/test-fixDateFormat

[friendica.git] / src / Util / ParseUrl.php

diff --git a/src/Util/ParseUrl.php b/src/Util/ParseUrl.php
index d0b168278563e24be39d6a79c01939846e7e1d98..4611cf8efe5f25378caa699044543af6269bab4c 100644 (file)
--- a/src/Util/ParseUrl.php
+++ b/src/Util/ParseUrl.php
@@ -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') {