Use regex to strip similar chars

NDLANO · Nov 25, 2024 · 5ad5062 · 5ad5062
1 parent 8cc7ee7
commit 5ad5062
Showing 1 changed file with 6 additions and 10 deletions.
diff --git a/src/main/java/no/ndla/taxonomy/util/PrettyUrlUtil.java b/src/main/java/no/ndla/taxonomy/util/PrettyUrlUtil.java
@@ -65,17 +65,13 @@ private static String cleanString(String name) {
         return Jsoup.parse(name)
                 .text()
                 .toLowerCase()
-                .replaceAll("[.,!?()/«»'¡:`’#]", "")
-                .replaceAll("æ", "a")
-                .replaceAll("ø", "o")
-                .replaceAll("ö", "o")
-                .replaceAll("å", "a")
-                .replaceAll("ä", "a")
-                .replaceAll("á", "a")
-                .replaceAll("ï", "i")
-                .replaceAll("í", "i")
-                .replaceAll("é", "e")
+                .replaceAll("[.,!?()/«»'¡:`’#°π]", "")
+                .replaceAll("[æåäáǎ]", "a")
+                .replaceAll("[øöô]", "o")
+                .replaceAll("[ïíǐ]", "i")
+                .replaceAll("[éè]", "e")
                 .replaceAll("š", "s")
+                .replaceAll("č", "c")
                 .replaceAll("ü", "u")
                 .replaceAll("1D45-B7C5", "d")
                 .replaceAll("–", "-")