LibWeb: Add more HTML entities

®, ß and all the lowercase and uppercase umlaut characters.
Author: https://github.com/linusg Commit: https://github.com/SerenityOS/serenity/commit/72c52466e0c Pull-request: https://github.com/SerenityOS/serenity/pull/2407
2025-01-24 02:03:06 -05:00 · 2020-05-26 21:20:28 +01:00 · 2020-05-26 21:20:28 +01:00 · 72c52466e0 · 2024-07-19 06:06:05 +09:00
commit 72c52466e0
parent f01af62313
1 changed files with 15 additions and 1 deletions
--- a/Libraries/LibWeb/Parser/HTMLParser.cpp
+++ b/Libraries/LibWeb/Parser/HTMLParser.cpp
@ -229,7 +229,21 @@ static bool parse_html_document(const StringView& html, Document& document, Pare
                    { "&quot;", "\"" },
                    { "&raquo;", ">>" },
                    { "&laquo;", "<<" },
-                    { "&copy;", "\xC2\xA9" },
+                    { "&copy;", "©" },
+                    { "&reg;", "®" },
+                    { "&szlig;", "ß" },
+                    { "&auml;", "ä" },
+                    { "&euml;", "ë" },
+                    { "&iuml;", "ï" },
+                    { "&ouml;", "ö" },
+                    { "&uuml;", "ü" },
+                    { "&yuml;", "ÿ" },
+                    { "&Auml;", "Ä" },
+                    { "&Euml;", "Ë" },
+                    { "&Iuml;", "Ï" },
+                    { "&Ouml;", "Ö" },
+                    { "&Uuml;", "Ü" },
+                    { "&Yuml;", "Ÿ" },
                };
                auto rest_of_html = html.substring_view(i, html.length() - i);
                bool found = false;