From b31934c7c5788cb3df9e9f4878bcacca67a1a67c Mon Sep 17 00:00:00 2001
From: Tarek Mahmoud Sayed <tarekms@microsoft.com>
Date: Tue, 29 Apr 2025 14:33:52 -0700
Subject: [PATCH] Fix culture creation with undetermined lang tag

---
 .../src/System/Globalization/CultureData.Icu.cs   | 15 +++++++++++++--
 .../CultureInfo/CultureInfoCtor.cs                | 12 ++++++++++++
 2 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/src/libraries/System.Private.CoreLib/src/System/Globalization/CultureData.Icu.cs b/src/libraries/System.Private.CoreLib/src/System/Globalization/CultureData.Icu.cs
index a95226e089686e..4b976bd92456a1 100644
--- a/src/libraries/System.Private.CoreLib/src/System/Globalization/CultureData.Icu.cs
+++ b/src/libraries/System.Private.CoreLib/src/System/Globalization/CultureData.Icu.cs
@@ -19,6 +19,7 @@ internal sealed partial class CultureData
         /// </summary>
         /// <param name="name">The locale name that ICU returns.</param>
         /// <param name="extension">The extension part in the original culture name.</param>
+        /// <param name="originalName">The original input culture name.</param>
         /// <param name="collationStart">The index of the collation in the name.</param>
         /// <remarks>
         /// BCP 47 specifications allow for extensions in the locale name, following the format language-script-region-extensions-collation. However,
@@ -29,7 +30,7 @@ internal sealed partial class CultureData
         /// between names with extensions and those without. For example, we may have a name like en-US and en-US-u-xx. Although .NET doesn't support the extension xx,
         /// we still include it in the name to distinguish it from the name without the extension.
         /// </remarks>
-        private static string NormalizeCultureName(string name, ReadOnlySpan<char> extension, out int collationStart)
+        private static string NormalizeCultureName(string name, ReadOnlySpan<char> extension, string originalName, out int collationStart)
         {
             Debug.Assert(name is not null);
             Debug.Assert(name.Length <= ICU_ULOC_FULLNAME_CAPACITY);
@@ -39,6 +40,16 @@ private static string NormalizeCultureName(string name, ReadOnlySpan<char> exten
             Span<char> buffer = stackalloc char[ICU_ULOC_FULLNAME_CAPACITY];
             int bufferIndex = 0;
 
+            // Using `Undetermined` language (e.g. `und-US`), the returned name from ICU getName is `_US`. We return back the original language `und`
+            if (name.Length > 1 && (name[0] == '-' || name[0] == '_') && originalName.StartsWith("und", StringComparison.OrdinalIgnoreCase))
+            {
+                buffer[bufferIndex++] = 'u';
+                buffer[bufferIndex++] = 'n';
+                buffer[bufferIndex++] = 'd';
+
+                changed = true;
+            }
+
             for (int i = 0; i < name.Length && bufferIndex < ICU_ULOC_FULLNAME_CAPACITY; i++)
             {
                 char c = name[i];
@@ -141,7 +152,7 @@ private bool InitIcuCultureDataCore()
 
             Debug.Assert(_sWindowsName != null);
 
-            _sRealName = NormalizeCultureName(_sWindowsName, indexOfExtensions > 0 ? _sRealName.AsSpan(indexOfExtensions) : ReadOnlySpan<char>.Empty, out int collationStart);
+            _sRealName = NormalizeCultureName(_sWindowsName, indexOfExtensions > 0 ? _sRealName.AsSpan(indexOfExtensions) : ReadOnlySpan<char>.Empty, _sRealName, out int collationStart);
 
             _iLanguage = LCID;
             if (_iLanguage == 0)
diff --git a/src/libraries/System.Runtime/tests/System.Globalization.Tests/CultureInfo/CultureInfoCtor.cs b/src/libraries/System.Runtime/tests/System.Globalization.Tests/CultureInfo/CultureInfoCtor.cs
index e5fd8c7fd0a123..8fa5ba6f53a153 100644
--- a/src/libraries/System.Runtime/tests/System.Globalization.Tests/CultureInfo/CultureInfoCtor.cs
+++ b/src/libraries/System.Runtime/tests/System.Globalization.Tests/CultureInfo/CultureInfoCtor.cs
@@ -388,6 +388,18 @@ public static IEnumerable<object[]> Ctor_String_TestData()
             }
         }
 
+        [ConditionalTheory(typeof(PlatformDetection), nameof(PlatformDetection.IsIcuGlobalization))]
+        [InlineData("und-us", "und-US", "und-US")]
+        [InlineData("und-us_tradnl", "und-US", "und-US_tradnl")]
+        [InlineData("und-es-u-co-phoneb", "und-ES", "und-ES_phoneb")]
+        [InlineData("und-es-t-something", "und-ES", "und-ES")]
+        public void CtorUndeterminedLanguageTag(string cultureName, string expectedCultureName, string expectedSortName)
+        {
+            CultureInfo culture = new CultureInfo(cultureName);
+            Assert.Equal(expectedCultureName, culture.Name);
+            Assert.Equal(expectedSortName, culture.CompareInfo.Name);
+        }
+
         [Theory]
         [MemberData(nameof(Ctor_String_TestData))]
         public void Ctor_String(string name, string[] expectedNames)