Rework system locale

archinstall/lib/installer.py

@@ -12,7 +12,7 @@
 from .disk import get_partitions_in_use, Partition
 from .general import SysCommand, generate_password
 from .hardware import has_uefi, is_vm, cpu_vendor
-from .locale_helpers import verify_keyboard_layout, verify_x11_keyboard_layout
+from .locale_helpers import list_locales, verify_keyboard_layout, verify_x11_keyboard_layout
 from .disk.helpers import findmnt
 from .mirrors import use_mirrors
 from .plugins import plugins
@@ -440,10 +440,14 @@ def set_hostname(self, hostname: str, *args :str, **kwargs :str) -> None:
 			fh.write(hostname + '\n')
 
 	def set_locale(self, locale :str, encoding :str = 'UTF-8', *args :str, **kwargs :str) -> bool:
+		self.log(f"Setting locale '{locale}' with encoding '{encoding}'.", level=logging.INFO)
 		if not len(locale):
 			return True
 
-		modifier = ''
+		# Get locale entries from /etc/locale.gen
+		entries = list_locales()
+
+		entry = ''
 
 		# This is a temporary patch to fix #1200
 		if '.' in locale:
@@ -453,19 +457,119 @@ def set_locale(self, locale :str, encoding :str = 'UTF-8', *args :str, **kwargs
 			# and the "found" encoding differs.
 			if encoding == 'UTF-8' and encoding != potential_encoding:
 				encoding = potential_encoding
+		# - End patch
+
+		# Validate the locale with entries from /etc/locale.gen
+		for row in entries:
+			entry_locale_name, entry_encoding = row.split()
+			if '.' in entry_locale_name:
+				entry_locale = entry_locale_name.split('.', 1)[0]
+			else:
+				entry_locale = entry_locale_name
+			if locale == entry_locale and encoding == entry_encoding:
+				entry = row
+				locale_name = entry_locale_name
+				locale = entry_locale
+				break
+		else:
+			self.log(f"Locale '{locale}' with encoding '{encoding}' not found in /etc/locale.gen.", fg="red", level=logging.ERROR)
+			return False
+
+		# A user could modify /etc/locale.gen in the install environment and add an invalid entry before launching the installer.
+		# Verify the locale did not validate against an invalid entry due to a user modified /etc/locale.gen.
+		if not os.path.exists(f'{self.target}/usr/share/i18n/locales/{locale}'):
+			self.log(f'Invalid locale: {locale}.', fg="red", level=logging.ERROR)
+			return False
+
+		# Check if the locale is already installed using the output of localedef.
+		# Encodings are formatted differently in the output of localedef (no dashes and lowercase).
+		formatted_encoding = encoding.replace('-','').lower()
+		locale_formatted_encoding = f'{locale}.{formatted_encoding}'
+
+		installed = []
+
+		# Get installed locales from the output of localedef.
+		for line in SysCommand(f'/usr/bin/arch-chroot {self.target} localedef --list-archive').decode().split():
+			installed.append(line)
+
+		# Install the locale if it is not already installed.
+		if locale_formatted_encoding not in installed:
+			# Before installing the locale check if the locale archive already exists and remove it if it does.
+			try:
+				os.remove(f'{self.target}/usr/lib/locale/locale-archive')
+			except FileNotFoundError:
+				pass
+
+			# Use localdef rather than local-gen since local-gen is a wrapper of localdef for user convenience
+			# and all necessary parameters for localedef are avaliable.
+			command = f'/usr/bin/arch-chroot {self.target} localedef -i {locale} -c -f {encoding} -A /usr/share/locale/locale.alias {locale_name}'
+
+			if not SysCommand(command).exit_code == 0:
+				self.log(f"Failed to install locale '{locale}' with encoding '{encoding}'.", fg="red", level=logging.ERROR)
+				return False
+
+		modifier = ''
 
 		# Make sure we extract the modifier, that way we can put it in if needed.
 		if '@' in locale:
 			locale, modifier = locale.split('@', 1)
 			modifier = f"@{modifier}"
-		# - End patch
 
-		with open(f'{self.target}/etc/locale.gen', 'a') as fh:
-			fh.write(f'{locale}.{encoding}{modifier} {encoding}\n')
-		with open(f'{self.target}/etc/locale.conf', 'w') as fh:
-			fh.write(f'LANG={locale}.{encoding}{modifier}\n')
+		# A format of locale to check and set the system locale with.
+		formatted_locale = f'{locale}.{encoding}{modifier}'
+
+		locale_conf = f'{self.target}/etc/locale.conf'
+
+		found = ''
 
-		return True if SysCommand(f'/usr/bin/arch-chroot {self.target} locale-gen').exit_code == 0 else False
+		# Check if the system locale is already set correctly.
+		try:
+			with open(locale_conf, 'r') as fh:
+				# Set up a regular expression pattern of a line beginning with 'LANG=' followed by a locale
+				pattern = re.compile(rf'^LANG="?({locale_name}|{formatted_locale})"?$')
+
+				for line in fh:
+					found = pattern.search(line)
+					if found:
+						found = found.group(1)
+						break
+		except FileNotFoundError:
+			pass
+
+		# Change the system locale if it is not set or set incorrectly.
+		if not found:
+			with open(locale_conf, 'w') as fh:
+				fh.write(f'LANG={formatted_locale}\n')
+
+		locale_gen = f'{self.target}/etc/locale.gen'
+
+		# Update /etc/locale.gen if necessary so that locale-gen will function properly.
+		# Check if the locale is the only uncommented entry.
+		with open(locale_gen, 'r') as fh:
+			uncommented = []
+			for line in fh:
+				if line[0] != '#':
+					uncommented.append(line.strip())
+
+			if len(uncommented) == 1 and entry in uncommented:
+				return True
+
+		# Uncomment the entry for the locale and comment all other uncommented entries.
+		with open(locale_gen, 'r') as fh:
+			contents = fh.readlines()
+
+			index = 0
+			for index, line in enumerate(contents):
+				uncommented_line = line.replace('#', '')
+				if uncommented_line.rstrip() == entry:
+					contents[index] = uncommented_line
+					continue
+				if line[0] != '#':
+					contents[index] = f'#{contents[index]}'
+
+		# Open the file again in write mode, to replace the contents
+		with open(locale_gen, 'w') as fh:
+			return True if fh.writelines(contents) else False
 
 	def set_timezone(self, zone :str, *args :str, **kwargs :str) -> bool:
 		if not zone:

archinstall/lib/user_interaction/locale_conf.py

@@ -12,7 +12,7 @@
 
 def select_locale_lang(preset: str = None) -> str:
 	locales = list_locales()
-	locale_lang = set([locale.split()[0] for locale in locales])
+	locale_lang = set([locale.split('.', 1)[0] if '.' in locale else locale.split()[0] for locale in locales])
 
 	selected_locale = Menu(
 		_('Choose which locale language to use'),