tweaks regex

onthegomap · Nov 26, 2024 · c43db46 · c43db46
1 parent e049876
commit c43db46
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 1 deletion.
diff --git a/planetiler-core/src/main/java/com/onthegomap/planetiler/util/LanguageUtils.java b/planetiler-core/src/main/java/com/onthegomap/planetiler/util/LanguageUtils.java
@@ -10,7 +10,7 @@ public class LanguageUtils {
   // See https://wiki.openstreetmap.org/wiki/Multilingual_names
   public static final Predicate<String> VALID_NAME_TAGS =
     Pattern
-      .compile("^name:[a-z]{2,3}(-[a-z]{4})?([-_](x-)?[a-z]{2,})?(-([a-z]{2}|\\d{3}))?$", Pattern.CASE_INSENSITIVE)
+      .compile("^name:[a-z]{2,3}(-[A-Z][a-z]{3})?([-_](x-)?[a-z]{2,})?(-([A-Z]{2}|\\d{3}))?$")
       .asMatchPredicate();
   // See https://github.com/onthegomap/planetiler/issues/86
   // Match strings that only contain latin characters.

diff --git a/planetiler-core/src/test/java/com/onthegomap/planetiler/util/LanguageUtilsTest.java b/planetiler-core/src/test/java/com/onthegomap/planetiler/util/LanguageUtilsTest.java
@@ -29,6 +29,7 @@ void testRemoveNonLatin(String in, String out) {
   @CsvSource(value = {
     "name:es, true",
     "name:en-US, true",
+    "name:en-001, true",
     "name:fr-x-gallo, true",
     "name:ko-Latn, true",
     "name:be-tarask, true",
@@ -45,6 +46,11 @@ void testRemoveNonLatin(String in, String out) {
     "nombre, false",
     "name:, false",
     "name:xxxxx, false",
+    "name:tec, true",
+    "name:TEC, false",
+    "name:be-tarask, true",
+    "name:nan-Latn-pehoeji, true",
+    "name:zh-Latn-pinyin, true",
   }, nullValues = "null")
   void testIsValidOsmNameTag(String in, boolean out) {
     assertEquals(out, LanguageUtils.isValidOsmNameTag(in));