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 0a12e628298b9398924d56f37347237958a62a08..4611cf8efe5f25378caa699044543af6269bab4c 100644 (file)
--- 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);
 
@@ -425,8 +429,6 @@ class ParseUrl
                                case 'news_keywords':
                                        $keywords = explode(',', $meta_tag['content']);
                                        break;
-                               default:
-                                       Logger::debug('Unsupported meta-tag found', ['meta-tag' => $meta_tag]);
                        }
                }
 
@@ -643,6 +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
@@ -664,6 +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)
@@ -727,6 +731,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseParts(array $siteinfo, array $jsonld): array
@@ -772,6 +777,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseJsonLd(array $siteinfo, array $jsonld): array
@@ -865,6 +871,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseJsonLdAuthor(array $siteinfo, array $jsonld): array
@@ -949,6 +956,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseJsonLdArticle(array $siteinfo, array $jsonld): array
@@ -1019,6 +1027,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseJsonLdWebPage(array $siteinfo, array $jsonld): array
@@ -1031,17 +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);
                }
 
@@ -1058,6 +1067,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseJsonLdWebSite(array $siteinfo, array $jsonld): array
@@ -1065,22 +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);
                }
 
@@ -1096,6 +1106,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseJsonLdWebOrganization(array $siteinfo, array $jsonld): array
@@ -1103,32 +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);
                }
 
@@ -1142,6 +1153,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseJsonLdWebPerson(array $siteinfo, array $jsonld): array
@@ -1149,12 +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);
                }
 
@@ -1164,7 +1176,7 @@ class ParseUrl
                }
 
                $content = JsonLD::fetchElement($jsonld, 'url');
-               if (!empty($content)) {
+               if (!empty($content) && is_string($content)) {
                        $jsonldinfo['author_url'] = trim($content);
                }
 
@@ -1187,6 +1199,7 @@ class ParseUrl
         *
         * @param array $siteinfo
         * @param array $jsonld
+        *
         * @return array siteinfo
         */
        private static function parseJsonLdMediaObject(array $siteinfo, array $jsonld, string $name): array
@@ -1194,22 +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);
                }
 
@@ -1219,27 +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);
                }