From 413227a5c16a2c16e50442f228365ea3125c9a0b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Robert=20M=C3=BCller?= <robytemueller@gmail.com>
Date: Sun, 19 Mar 2023 16:04:42 +0100
Subject: [PATCH] Add `os_locale_str` to get user locale

This function determines the preferred user locale setting.
---
 src/base/system.cpp | 68 +++++++++++++++++++++++++++++++++++++++++++++
 src/base/system.h   | 22 +++++++++++++++
 src/test/os.cpp     |  7 +++++
 3 files changed, 97 insertions(+)

diff --git a/src/base/system.cpp b/src/base/system.cpp
index 057176f8a..4f6142289 100644
--- a/src/base/system.cpp
+++ b/src/base/system.cpp
@@ -26,6 +26,7 @@
 
 #if defined(CONF_FAMILY_UNIX)
 #include <csignal>
+#include <locale>
 #include <sys/stat.h>
 #include <sys/time.h>
 #include <sys/utsname.h>
@@ -51,6 +52,7 @@
 #define _task_user_
 
 #include <Carbon/Carbon.h>
+#include <CoreFoundation/CoreFoundation.h>
 #include <mach-o/dyld.h>
 #include <mach/mach_time.h>
 
@@ -4266,6 +4268,72 @@ int os_version_str(char *version, int length)
 #endif
 }
 
+void os_locale_str(char *locale, size_t length)
+{
+#if defined(CONF_FAMILY_WINDOWS)
+	wchar_t wide_buffer[LOCALE_NAME_MAX_LENGTH];
+	dbg_assert(GetUserDefaultLocaleName(wide_buffer, std::size(wide_buffer)) > 0, "GetUserDefaultLocaleName failure");
+
+	// Assume maximum possible length for encoding as UTF-8.
+	char buffer[UTF8_BYTE_LENGTH * LOCALE_NAME_MAX_LENGTH + 1];
+	dbg_assert(WideCharToMultiByte(CP_UTF8, 0, wide_buffer, -1, buffer, sizeof(buffer), NULL, NULL) > 0, "WideCharToMultiByte failure");
+
+	str_copy(locale, buffer, length);
+#elif defined(CONF_PLATFORM_MACOS)
+	CFLocaleRef locale_ref = CFLocaleCopyCurrent();
+	CFStringRef locale_identifier_ref = static_cast<CFStringRef>(CFLocaleGetValue(locale_ref, kCFLocaleIdentifier));
+
+	// Count number of UTF16 codepoints, +1 for zero-termination.
+	// Assume maximum possible length for encoding as UTF-8.
+	CFIndex locale_identifier_size = (UTF8_BYTE_LENGTH * CFStringGetLength(locale_identifier_ref) + 1) * sizeof(char);
+	char *locale_identifier = (char *)malloc(locale_identifier_size);
+	dbg_assert(CFStringGetCString(locale_identifier_ref, locale_identifier, locale_identifier_size, kCFStringEncodingUTF8), "CFStringGetCString failure");
+
+	str_copy(locale, locale_identifier, length);
+
+	free(locale_identifier);
+	CFRelease(locale_ref);
+#else
+	static const char *ENV_VARIABLES[] = {
+		"LC_ALL",
+		"LC_MESSAGES",
+		"LANG",
+	};
+
+	locale[0] = '\0';
+	for(const char *env_variable : ENV_VARIABLES)
+	{
+		const char *env_value = getenv(env_variable);
+		if(env_value)
+		{
+			str_copy(locale, env_value, length);
+			break;
+		}
+	}
+#endif
+
+	// Ensure RFC 3066 format:
+	// - use hyphens instead of underscores
+	// - truncate locale string after first non-standard letter
+	for(int i = 0; i < str_length(locale); ++i)
+	{
+		if(locale[i] == '_')
+		{
+			locale[i] = '-';
+		}
+		else if(locale[i] != '-' && !(locale[i] >= 'a' && locale[i] <= 'z') && !(locale[i] >= 'A' && locale[i] <= 'Z') && !(locale[i] >= '0' && locale[i] <= '9'))
+		{
+			locale[i] = '\0';
+			break;
+		}
+	}
+
+	// Use default if we could not determine the locale,
+	// i.e. if only the C or POSIX locale is available.
+	if(locale[0] == '\0' || str_comp(locale, "C") == 0 || str_comp(locale, "POSIX") == 0)
+		str_copy(locale, "en-US", length);
+}
+
 #if defined(CONF_EXCEPTION_HANDLING)
 #if defined(CONF_FAMILY_WINDOWS)
 static HMODULE exception_handling_module = nullptr;
diff --git a/src/base/system.h b/src/base/system.h
index b6599f448..ee6331f99 100644
--- a/src/base/system.h
+++ b/src/base/system.h
@@ -2143,6 +2143,14 @@ char str_uppercase(char c);
 int str_isallnum(const char *str);
 unsigned str_quickhash(const char *str);
 
+enum
+{
+	/**
+	 * The maximum bytes necessary to encode one Unicode codepoint with UTF-8.
+	 */
+	UTF8_BYTE_LENGTH = 4,
+};
+
 int str_utf8_to_skeleton(const char *str, int *buf, int buf_len);
 
 /*
@@ -2586,6 +2594,20 @@ int secure_rand_below(int below);
 */
 int os_version_str(char *version, int length);
 
+/**
+ * Returns a string of the preferred locale of the user / operating system.
+ * The string conforms to [RFC 3066](https://www.ietf.org/rfc/rfc3066.txt)
+ * and only contains the characters `a`-`z`, `A`-`Z`, `0`-`9` and `-`.
+ * If the preferred locale could not be determined this function
+ * falls back to the locale `"en-US"`.
+ *
+ * @param locale Buffer to use for the output.
+ * @param length Length of the output buffer.
+ *
+ * @remark The destination buffer will be zero-terminated.
+ */
+void os_locale_str(char *locale, size_t length);
+
 #if defined(CONF_EXCEPTION_HANDLING)
 void init_exception_handler();
 void set_exception_handler_log_file(const char *log_file_path);
diff --git a/src/test/os.cpp b/src/test/os.cpp
index 4b83a9acc..21ee5177a 100644
--- a/src/test/os.cpp
+++ b/src/test/os.cpp
@@ -9,3 +9,10 @@ TEST(Os, VersionStr)
 	EXPECT_FALSE(os_version_str(aVersion, sizeof(aVersion)));
 	EXPECT_STRNE(aVersion, "");
 }
+
+TEST(Os, LocaleStr)
+{
+	char aLocale[128];
+	os_locale_str(aLocale, sizeof(aLocale));
+	EXPECT_STRNE(aLocale, "");
+}