cmake_host_system_information: Add query for the locale's charset

Issue: #27562

.gitlab/ci/configure_fedora43_makefiles.cmake

@@ -108,6 +108,7 @@ set(CMake_TEST_IPO_WORKS_Fortran "ON" CACHE BOOL "")
 if (NOT "$ENV{CMAKE_CI_NIGHTLY}" STREQUAL "")
   set(CMake_TEST_ISPC "ON" CACHE STRING "")
 endif()
+set(CMake_TEST_LOCALE_CHARSET "ISO-8859-1" CACHE STRING "")
 set(CMake_TEST_Qt5 "ON" CACHE BOOL "")
 set(CMake_TEST_TLS_VERIFY_URL "https://gitlab.kitware.com" CACHE STRING "")
 set(CMake_TEST_TLS_VERIFY_URL_BAD "https://badtls-expired.kitware.com" CACHE STRING "")

.gitlab/ci/configure_fedora43_ninja.cmake

@@ -4,6 +4,7 @@ if (NOT "$ENV{CMAKE_CI_NIGHTLY}" STREQUAL "")
   set(CMake_TEST_CPACK_APPIMAGE_RUNTIME_FILE "$ENV{CI_PROJECT_DIR}/.gitlab/appimagetool/lib/appimagetool/runtime" CACHE FILEPATH "")
   set(CMake_TEST_ISPC "ON" CACHE STRING "")
 endif()
+set(CMake_TEST_LOCALE_CHARSET "UTF-8" CACHE STRING "")
 set(CMake_TEST_MODULE_COMPILATION "named,compile_commands,collation,partitions,internal_partitions,export_bmi,install_bmi,shared,bmionly,build_database" CACHE STRING "")
 set(CMake_TEST_Rust "ON" CACHE STRING "")
 set(CMake_TEST_TLS_VERIFY_URL "https://gitlab.kitware.com" CACHE STRING "")

.gitlab/ci/configure_macos_arm64_ninja.cmake

@@ -6,6 +6,7 @@ set(CMake_TEST_FindOpenMP "ON" CACHE BOOL "")
 set(CMake_TEST_FindOpenMP_C "ON" CACHE BOOL "")
 set(CMake_TEST_FindOpenMP_CXX "ON" CACHE BOOL "")
 set(CMake_TEST_GUI "ON" CACHE BOOL "")
+set(CMake_TEST_LOCALE_CHARSET "UTF-8" CACHE STRING "")
 set(CMake_TEST_TLS_VERIFY_URL "https://gitlab.kitware.com" CACHE STRING "")
 set(CMake_TEST_TLS_VERIFY_URL_BAD "https://badtls-expired.kitware.com" CACHE STRING "")
 set(CMake_TEST_TLS_VERSION "1.2" CACHE STRING "")

.gitlab/ci/configure_macos_x86_64_ninja.cmake

@@ -9,6 +9,7 @@ set(CMake_TEST_GUI "ON" CACHE BOOL "")
 if (NOT "$ENV{CMAKE_CI_NIGHTLY}" STREQUAL "")
   set(CMake_TEST_ISPC "ON" CACHE STRING "")
 endif()
+set(CMake_TEST_LOCALE_CHARSET "UTF-8" CACHE STRING "")
 set(CMake_TEST_TLS_VERIFY_URL "https://gitlab.kitware.com" CACHE STRING "")
 set(CMake_TEST_TLS_VERIFY_URL_BAD "https://badtls-expired.kitware.com" CACHE STRING "")
 set(CMake_TEST_TLS_VERSION "1.2" CACHE STRING "")

.gitlab/ci/configure_windows_common.cmake

@@ -2,6 +2,7 @@ set(BUILD_QtDialog ON CACHE BOOL "")
 set(BUILD_CursesDialog ON CACHE BOOL "")
 set(CMAKE_PREFIX_PATH "$ENV{CI_PROJECT_DIR}/.gitlab/qt" CACHE STRING "")
 set(CMake_TEST_Java OFF CACHE BOOL "")
+set(CMake_TEST_LOCALE_CHARSET "UTF-8" CACHE STRING "")
 set(Python_FIND_REGISTRY NEVER CACHE STRING "")
 
 include("${CMAKE_CURRENT_LIST_DIR}/configure_common.cmake")

.gitlab/ci/env_fedora43_makefiles.sh

@@ -1,3 +1,6 @@
 if test "$CMAKE_CI_NIGHTLY" = "true"; then
   source .gitlab/ci/ispc-env.sh
 fi
+
+# Test in a Latin-1 locale.
+export LANG=en_US.ISO-8859-1

.gitlab/ci/env_fedora43_ninja.sh

@@ -2,3 +2,6 @@ if test "$CMAKE_CI_NIGHTLY" = "true"; then
   source .gitlab/ci/appimagetool-env.sh
   source .gitlab/ci/ispc-env.sh
 fi
+
+# Test in a UTF-8 locale.
+export LANG=en_US.UTF-8

Help/command/cmake_host_system_information.rst

@@ -137,6 +137,17 @@ queried.  The list of queried values is stored in ``<variable>``.
 
   Human readable full processor description
 
+``LOCALE_CHARSET``
+  .. versionadded:: 4.3
+
+  The locale's character set, if known, otherwise empty.
+
+  This is the :ref:`encoding <CMake Language Encoding>` expected
+  in :manual:`cmake-language(7)` files and scripts:
+
+  * On Windows, this is always ``UTF-8``.
+  * On UNIX and macOS, this is based on the current locale.
+
 ``OS_NAME``
   .. versionadded:: 3.10
 

Help/manual/cmake-language.7.rst

@@ -78,6 +78,11 @@ Furthermore, CMake 3.0 and above allow a leading UTF-8
 
 .. _Byte-Order Mark: https://en.wikipedia.org/wiki/Byte_order_mark
 
+.. versionadded:: 4.3
+
+  The :command:`cmake_host_system_information` command's ``LOCALE_CHARSET``
+  query returns the expected character set encoding.
+
 Source Files
 ------------
 

Help/release/dev/cmake-locale.rst

@@ -0,0 +1,6 @@
+cmake-locale
+------------
+
+* The :command:`cmake_host_system_information` command gained a
+  ``LOCALE_CHARSET`` query for the expected :manual:`cmake-language(7)`
+  script encoding.

Source/cmCMakeHostSystemInformationCommand.cxx

@@ -15,6 +15,10 @@
 #include <cm/type_traits>
 #include <cmext/string_view>
 
+#if !defined(_WIN32)
+#  include <langinfo.h>
+#endif
+
 #include "cmsys/FStream.hxx"
 #include "cmsys/Glob.hxx"
 #include "cmsys/String.h"
@@ -74,6 +78,22 @@ cm::optional<std::string> GetValue(cmsys::SystemInformation& info,
   if (key == "FQDN"_s) {
     return ValueToString(info.GetFullyQualifiedDomainName());
   }
+  if (key == "LOCALE_CHARSET"_s) {
+#if defined(_WIN32)
+    // On Windows we always use UTF-8.
+    return ValueToString("UTF-8");
+#else
+    // On non-Windows platforms we use the locale's encoding.
+    if (char const* charset = nl_langinfo(CODESET)) {
+      if (cmsysString_strcasecmp(charset, "utf-8") == 0 ||
+          cmsysString_strcasecmp(charset, "utf8") == 0) {
+        charset = "UTF-8";
+      }
+      return ValueToString(charset);
+    }
+    return ValueToString("");
+#endif
+  }
   if (key == "TOTAL_VIRTUAL_MEMORY"_s) {
     return ValueToString(info.GetTotalVirtualMemory());
   }

Tests/RunCMake/CMakeLists.txt

@@ -685,6 +685,9 @@ if(WIN32)
 endif()
 add_RunCMake_test(execute_process)
 add_RunCMake_test(export)
+if(DEFINED CMake_TEST_LOCALE_CHARSET)
+  list(APPEND cmake_host_system_information_ARGS -DCMake_TEST_LOCALE_CHARSET=${CMake_TEST_LOCALE_CHARSET})
+endif()
 if(CMake_TEST_MSYSTEM_PREFIX)
   list(APPEND cmake_host_system_information_ARGS -DCMake_TEST_MSYSTEM_PREFIX=${CMake_TEST_MSYSTEM_PREFIX})
 endif()

Tests/RunCMake/cmake_host_system_information/Locale-stdout-iso-8859-1.txt

@@ -0,0 +1 @@
+^-- LOCALE_CHARSET='ISO-8859-1'$

Tests/RunCMake/cmake_host_system_information/Locale-stdout-utf-8.txt

@@ -0,0 +1 @@
+^-- LOCALE_CHARSET='UTF-8'$

Tests/RunCMake/cmake_host_system_information/Locale-stdout.txt

@@ -0,0 +1 @@
+^-- LOCALE_CHARSET='[^']+'$

Tests/RunCMake/cmake_host_system_information/Locale.cmake

@@ -0,0 +1,2 @@
+cmake_host_system_information(RESULT charset QUERY LOCALE_CHARSET)
+message(STATUS "LOCALE_CHARSET='${charset}'")

Tests/RunCMake/cmake_host_system_information/QueryKeys-stdout.txt

@@ -23,6 +23,7 @@
 -- PROCESSOR_SERIAL_NUMBER=`.*`
 -- PROCESSOR_NAME=`.*`
 -- PROCESSOR_DESCRIPTION=`.*`
+-- LOCALE_CHARSET=`[^`]+`
 -- OS_NAME=`.*`
 -- OS_RELEASE=`.*`
 -- OS_VERSION=`.*`

Tests/RunCMake/cmake_host_system_information/QueryKeys.cmake

@@ -28,6 +28,7 @@ try_and_print(HAS_L1_CACHE)
 try_and_print(PROCESSOR_SERIAL_NUMBER)
 try_and_print(PROCESSOR_NAME)
 try_and_print(PROCESSOR_DESCRIPTION)
+try_and_print(LOCALE_CHARSET)
 try_and_print(OS_NAME)
 try_and_print(OS_RELEASE)
 try_and_print(OS_VERSION)

Tests/RunCMake/cmake_host_system_information/RunCMakeTest.cmake

@@ -7,6 +7,14 @@ run_cmake(BadArg3)
 run_cmake(QueryList)
 run_cmake(QueryKeys)
 
+if(CMake_TEST_LOCALE_CHARSET STREQUAL "UTF-8" OR CMAKE_HOST_WIN32)
+  set(RunCMake-stdout-file Locale-stdout-utf-8.txt)
+elseif(CMake_TEST_LOCALE_CHARSET STREQUAL "ISO-8859-1")
+  set(RunCMake-stdout-file Locale-stdout-iso-8859-1.txt)
+endif()
+run_cmake_script(Locale)
+unset(RunCMake-stdout-file)
+
 run_cmake(UnitTest)
 run_cmake(Exherbo)
 run_cmake(Ubuntu)