[release/6.0] Detect the default locale name during startup on Apple platforms (#68934)

* Detect the default locale name during startup on Apple platforms

This change adds a function to lookup the current NSLocale and extract the language + country code to load into ICU by default.  Previously, we would defer to uloc_getDefault in ICU, which would return a value we would ignore (en_US_POSIX) and result in falling back to invariant mode.

Fixes #68321

* Link in foundation and address feedback

* Link in foundation in coreclr

* Check LANG env variable first

* Defer to icu first, since it's looking for LANG and certain values.

* #ifdef not #if

* PR feedback

* Remove redundant string replace

* pal_locale.m needs to be in a different location than upstream

* cmake tweak to make sure foundation gets linked in

Co-authored-by: Steve Pfister <[email protected]>

Author: github-actions[bot]

src/coreclr/dlls/mscoree/coreclr/CMakeLists.txt

@@ -167,10 +167,15 @@ if(FEATURE_MERGE_JIT_AND_ENGINE)
     set(CLRJIT_STATIC clrjit_static)
 endif(FEATURE_MERGE_JIT_AND_ENGINE)
 
+if (CLR_CMAKE_TARGET_OSX OR CLR_CMAKE_TARGET_MACCATALYST)
+    include(CMakeFindFrameworks)
+    find_library(FOUNDATION Foundation REQUIRED)
+endif()
+
 target_sources(coreclr PUBLIC $<TARGET_OBJECTS:cee_wks_core>)
-target_link_libraries(coreclr PUBLIC ${CORECLR_LIBRARIES} ${CLRJIT_STATIC} cee_wks)
+target_link_libraries(coreclr PUBLIC ${CORECLR_LIBRARIES} ${CLRJIT_STATIC} cee_wks ${FOUNDATION})
 target_sources(coreclr_static PUBLIC $<TARGET_OBJECTS:cee_wks_core>)
-target_link_libraries(coreclr_static PUBLIC ${CORECLR_LIBRARIES} clrjit_static cee_wks_mergeable)
+target_link_libraries(coreclr_static PUBLIC ${CORECLR_LIBRARIES} clrjit_static cee_wks_mergeable ${FOUNDATION})
 target_compile_definitions(coreclr_static PUBLIC CORECLR_EMBEDDED)
 
 if(CLR_CMAKE_TARGET_WIN32)

src/libraries/Native/Unix/System.Globalization.Native/CMakeLists.txt

@@ -64,6 +64,10 @@ set(NATIVEGLOBALIZATION_SOURCES
     pal_icushim.c
 )
 
+if (CLR_CMAKE_TARGET_OSX OR CLR_CMAKE_TARGET_MACCATALYST OR CLR_CMAKE_TARGET_IOS OR CLR_CMAKE_TARGET_TVOS)
+    set(NATIVEGLOBALIZATION_SOURCES ${NATIVEGLOBALIZATION_SOURCES} pal_locale.m)
+endif()
+
 # time zone names are filtered out of icu data for the browser and associated functionality is disabled
 if (NOT CLR_CMAKE_TARGET_BROWSER)
     set(NATIVEGLOBALIZATION_SOURCES ${NATIVEGLOBALIZATION_SOURCES} pal_timeZoneInfo.c)
@@ -80,6 +84,11 @@ endif()
 include_directories("../Common")
 
 if (GEN_SHARED_LIB)
+    if (CLR_CMAKE_TARGET_OSX OR CLR_CMAKE_TARGET_MACCATALYST OR  CLR_CMAKE_TARGET_IOS OR CLR_CMAKE_TARGET_TVOS)
+        include(CMakeFindFrameworks)
+        find_library(FOUNDATION Foundation REQUIRED)
+    endif()
+
     add_library(System.Globalization.Native
         SHARED
         ${NATIVEGLOBALIZATION_SOURCES}
@@ -88,12 +97,12 @@ if (GEN_SHARED_LIB)
 
     target_link_libraries(System.Globalization.Native
         dl
+        ${FOUNDATION}
     )
 
     install_with_stripped_symbols (System.Globalization.Native PROGRAMS .)
 endif()
 
-
 add_library(System.Globalization.Native-Static
     STATIC
     ${NATIVEGLOBALIZATION_SOURCES}

src/libraries/Native/Unix/System.Globalization.Native/pal_locale.c

@@ -129,12 +129,15 @@ int32_t FixupLocaleName(UChar* value, int32_t valueLength)
     return i;
 }
 
-// We use whatever ICU give us as the default locale except if it is en_US_POSIX. We'll map
-// this POSIX locale to Invariant instead. The reason is POSIX locale collation behavior
-// is not desirable at all because it doesn't support case insensitive string comparisons.
+// We use whatever ICU give us as the default locale except if it is en_US_POSIX. 
+//
+// On Apple related platforms (OSX, iOS, tvOS, MacCatalyst), we'll take what the system locale is.  
+// On all other platforms we'll map this POSIX locale to Invariant instead. 
+// The reason is POSIX locale collation behavior is not desirable at all because it doesn't support case insensitive string comparisons.
 const char* DetectDefaultLocaleName()
 {
     const char* icuLocale = uloc_getDefault();
+
     if (strcmp(icuLocale, "en_US_POSIX") == 0)
     {
         return "";
@@ -216,6 +219,16 @@ int32_t GlobalizationNative_GetDefaultLocaleName(UChar* value, int32_t valueLeng
 
     const char* defaultLocale = DetectDefaultLocaleName();
 
+#ifdef __APPLE__
+    char* appleLocale = NULL;
+    
+    if (strcmp(defaultLocale, "") == 0)
+    {
+        appleLocale = DetectDefaultAppleLocaleName();
+        defaultLocale = appleLocale;
+    }
+#endif
+
     uloc_getBaseName(defaultLocale, localeNameBuffer, ULOC_FULLNAME_CAPACITY, &status);
     u_charsToUChars_safe(localeNameBuffer, value, valueLength, &status);
 
@@ -236,6 +249,11 @@ int32_t GlobalizationNative_GetDefaultLocaleName(UChar* value, int32_t valueLeng
         }
     }
 
+#ifdef __APPLE__
+    if (appleLocale)
+        free(appleLocale);
+#endif
+
     return UErrorCodeToBool(status);
 }
 

src/libraries/Native/Unix/System.Globalization.Native/pal_locale.m

@@ -0,0 +1,30 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+
+#include <stdlib.h>
+#include "pal_locale_internal.h"
+
+#import <Foundation/Foundation.h>
+
+char* DetectDefaultAppleLocaleName()
+{
+    NSLocale *currentLocale = [NSLocale currentLocale];
+    NSString *localeName = @"";
+    
+    if (!currentLocale)
+    {
+        return strdup([localeName UTF8String]);
+    }
+
+    if ([currentLocale.languageCode length] > 0 && [currentLocale.countryCode length] > 0)
+    {
+        localeName = [NSString stringWithFormat:@"%@-%@", currentLocale.languageCode, currentLocale.countryCode];
+    }
+    else
+    {
+        localeName = currentLocale.localeIdentifier;
+    }
+    
+    return strdup([localeName UTF8String]);
+}
+

src/libraries/Native/Unix/System.Globalization.Native/pal_locale_internal.h

@@ -51,3 +51,13 @@ Detect the default locale for the machine, defaulting to Invaraint if
 we can't compute one (different from uloc_getDefault()) would do.
 */
 const char* DetectDefaultLocaleName(void);
+
+#ifdef __APPLE__
+/*
+Function:
+DetectDefaultSystemLocaleName
+
+Detects the default locale string for Apple platforms
+*/
+char* DetectDefaultAppleLocaleName(void);
+#endif

src/libraries/System.Globalization/tests/CultureInfo/CultureInfoCurrentCulture.cs

@@ -33,6 +33,17 @@ public void CurrentCulture()
             }
         }
 
+        [Fact]
+        [PlatformSpecific(TestPlatforms.OSX | TestPlatforms.iOS | TestPlatforms.MacCatalyst | TestPlatforms.tvOS)]
+        public void CurrentCulture_Default_Not_Invariant()
+        {
+            // On OSX-like platforms, it should default to what the default system culture is 
+            // set to.  Since we shouldn't assume en-US, we just test if it's not the invariant
+            // culture.
+            Assert.NotEqual(CultureInfo.CurrentCulture, CultureInfo.InvariantCulture);
+            Assert.NotEqual(CultureInfo.CurrentUICulture, CultureInfo.InvariantCulture);
+        }
+
         [Fact]
         public void CurrentCulture_Set_Null_ThrowsArgumentNullException()
         {
@@ -125,7 +136,7 @@ public void CurrentCulture_BasedOnLangEnvVar(string langEnvVar, string expectedC
             }, expectedCultureName, new RemoteInvokeOptions { StartInfo = psi }).Dispose();
         }
 
-        [PlatformSpecific(TestPlatforms.AnyUnix)]  // When LANG is empty or unset, should default to the invariant culture on Unix.
+        [PlatformSpecific(TestPlatforms.AnyUnix)]
         [ConditionalTheory(typeof(RemoteExecutor), nameof(RemoteExecutor.IsSupported))]
         [InlineData("")]
         [InlineData(null)]
@@ -141,13 +152,28 @@ public void CurrentCulture_DefaultWithNoLang(string langEnvVar)
                psi.Environment["LANG"] = langEnvVar;
             }
 
+            // When LANG is empty or unset, on Unix it should default to the invariant culture.
+            // On OSX-like platforms, it should default to what the default system culture is 
+            // set to.  Since we shouldn't assume en-US, we just test if it's not the invariant
+            // culture.
             RemoteExecutor.Invoke(() =>
             {
                 Assert.NotNull(CultureInfo.CurrentCulture);
                 Assert.NotNull(CultureInfo.CurrentUICulture);
 
-                Assert.Equal("", CultureInfo.CurrentCulture.Name);
-                Assert.Equal("", CultureInfo.CurrentUICulture.Name);
+                if (PlatformDetection.IsOSXLike)
+                {
+                    Assert.NotEqual("", CultureInfo.CurrentCulture.Name);
+                    Assert.NotEqual("", CultureInfo.CurrentUICulture.Name);
+
+                    Assert.NotEqual(CultureInfo.CurrentCulture, CultureInfo.InvariantCulture);
+                    Assert.NotEqual(CultureInfo.CurrentUICulture, CultureInfo.InvariantCulture);
+                }
+                else
+                {
+                    Assert.Equal("", CultureInfo.CurrentCulture.Name);
+                    Assert.Equal("", CultureInfo.CurrentUICulture.Name);
+                }
             }, new RemoteInvokeOptions { StartInfo = psi }).Dispose();
         }
 

src/mono/mono/mini/CMakeLists.txt

@@ -23,7 +23,7 @@ if(HOST_DARWIN)
   set(OS_LIBS "-framework CoreFoundation" "-framework Foundation")
 
   if(CMAKE_SYSTEM_VARIANT STREQUAL "MacCatalyst")
-    set(OS_LIBS "-lobjc" "-lc++")
+    set(OS_LIBS ${OS_LIBS} "-lobjc" "-lc++")
   endif()
 elseif(HOST_IOS)
 set(OS_LIBS "-framework CoreFoundation" "-lobjc" "-lc++")
@@ -79,6 +79,13 @@ if(HAVE_SYS_ICU)
       pal_timeZoneInfo.c
       entrypoints.c
       ${pal_icushim_sources_base})
+
+  if (TARGET_DARWIN)
+    set(icu_shim_sources_base
+        ${icu_shim_sources_base}
+        pal_locale.m)
+  endif()
+
   addprefix(icu_shim_sources "${ICU_SHIM_PATH}" "${icu_shim_sources_base}")
   set_source_files_properties(${icu_shim_sources} PROPERTIES COMPILE_DEFINITIONS OSX_ICU_LIBRARY_PATH="${OSX_ICU_LIBRARY_PATH}")
   set_source_files_properties(${icu_shim_sources} PROPERTIES COMPILE_FLAGS "-I\"${ICU_INCLUDEDIR}\" -I\"${CMAKE_CURRENT_SOURCE_DIR}/../../../libraries/Native/Unix/System.Globalization.Native/\" -I\"${CMAKE_CURRENT_SOURCE_DIR}/../../../libraries/Native/Unix/Common/\" ${ICU_FLAGS}")