path: root/locale-gen

#!/bin/bash

#
# Based upon Debian's locale-gen, fetched from glibc_2.3.6-7.diff.gz, but completely rewritten.
#

# NB: Bash-4.0+ required.  We use += and ${var,,} features.

unset POSIXLY_CORRECT IFS
umask 0022

argv0=${0##*/}

EPREFIX="@GENTOO_PORTAGE_EPREFIX@"
if [[ ${EPREFIX} == "@"GENTOO_PORTAGE_EPREFIX"@" ]] ; then
	EPREFIX=""
fi

FUNCTIONS_SH="/lib/gentoo/functions.sh"
source "${EPREFIX}"${FUNCTIONS_SH} || {
	echo "${argv0}: Could not source ${FUNCTIONS_SH}!" 1>&2
	exit 1
}

COMPILED_LOCALES=""

show_usage() {
	cat <<-EOF
	Usage: ${HILITE}${argv0}${NORMAL} ${GOOD}[options]${NORMAL} -- ${GOOD}[localedef options]${NORMAL}

	Generate locales based upon the config file /etc/locale.gen.

	${HILITE}Options:${NORMAL}
	    ${GOOD}-k, --keep${NORMAL}               Don't nuke existing locales
	    ${GOOD}-d, --destdir <dir>${NORMAL}      Use locale data in specified DESTDIR tree
	    ${GOOD}-c, --config <config>${NORMAL}    Use specified config instead of default locale.gen
	    ${GOOD}-l, --list${NORMAL}               List all the locales to be generated
	    ${GOOD}-a, --ask${NORMAL}                Ask before generating each locale
	    ${GOOD}-A, --all${NORMAL}                Pretend the locale list contains all locales
	    ${GOOD}-u, --update${NORMAL}             Only generate locales that are missing
	    ${GOOD}-G, --generate <locale>${NORMAL}  Generate specified locale (one shot; implies -k -u)
	    ${GOOD}-j, --jobs <num>${NORMAL}         Number of locales to generate at a time (parallel)
	    ${GOOD}-q, --quiet${NORMAL}              Only show errors
	    ${GOOD}-V, --version${NORMAL}            Meaningless version information
	    ${GOOD}-h, --help${NORMAL}               Show this help cruft

	${HILITE}Localedef Options:${NORMAL}
	    By default, ${GOOD}${LOCALEDEF_OPTS}${NORMAL} is passed to localedef.

	For more info, see the ${HILITE}locale-gen${NORMAL}(1) and ${HILITE}locale.gen${NORMAL}(8) manpages.
	EOF
	[[ $# -eq 0 ]] && exit 0
	echo ""
	eerror "Unknown option '$1'"
	exit 1
}

show_version() {
	echo "locale-gen-2.xxx"
	exit 0
}

LOCALEDEF_OPTS=""
KEEP=""
DESTDIR=""
CONFIG=""
JUST_LIST=""
ASK=""
ALL=""
UPDATE=""
GENERATE=""
JOBS_MAX=""
QUIET=0
SET_X=""
LOCALE_ARCHIVE=true
INPLACE_GLIBC=""
while [[ $# -gt 0 ]] ; do
	case $1 in
		--inplace-glibc)           INPLACE_GLIBC=$1;;
		-k|--keep|--keep-existing) KEEP=$1;;
		-d|--destdir)              shift; DESTDIR=$1; unset ROOT;;
		-c|--config)               shift; CONFIG=$1;;
		-l|--list)                 JUST_LIST=$1;;
		-a|--ask)                  ASK=$1;;
		-A|--all)                  ALL=$1;;
		-u|--update)               UPDATE=$1;;
		-G|--generate)             shift; GENERATE=$1;;
		-j|--jobs)                 shift; JOBS_MAX=$(( $1 ));;
		-j*)                       : $(( JOBS_MAX = ${1#-j} ));;
		-q|--quiet)                : $(( ++QUIET ));;
		-x|--debug)                SET_X="true";;
		-V|--version)              show_version;;
		-h|--help)                 show_usage;;
		--)                        shift; LOCALEDEF_OPTS=$*; break;;
		*)                         show_usage $1;;
	esac
	shift
done

if [[ -n ${COMPILED_LOCALES} ]] ; then
	ewarn "All locales have been installed and registered by the package manager. If you"
	ewarn "rebuild the locale archive now, file integrity tools may show it as corrupted."
	ewarn "This is not really a big problem, but a better solution is to disable"
	ewarn "USE=compile-locales and re-install glibc if you dont need all locales."
	echo
fi

if [[ -z ${JOBS_MAX} ]] ; then
	JOBS_MAX=$(getconf _NPROCESSORS_ONLN 2>/dev/null)
	: "${JOBS_MAX:=1}"
fi
[[ ${JOBS_MAX} -lt 1 ]] && JOBS_MAX=1
[[ -n ${SET_X} ]] && set -x
: "${KEEP:=${JUST_LIST}}"
[[ -n ${GENERATE} ]] && UPDATE="true" && KEEP="true"

: "${ROOT:=/}"
ROOT="${ROOT%/}/"

if [[ ${ROOT} != "/" ]] ; then
	eerror "Sorry, but ROOT is not supported."
	exit 0
fi

: "${EROOT:=${EPREFIX}/}"
if [[ ${EROOT} != "/" ]] ; then
	einfo "Using locale.gen from ${EROOT%/}/etc/"
fi

if [[ -n ${DESTDIR} ]] ; then
	DESTDIR="${DESTDIR%/}/"
	einfo "Building locales in DESTDIR '${DESTDIR}'"
else
	DESTDIR="${EROOT%/}/"
fi

: "${CONFIG:=${EROOT%/}/etc/locale.gen}"
LOCALES=${DESTDIR}usr/share/i18n/locales
CHARMAPS=${DESTDIR}usr/share/i18n/charmaps
SUPPORTED=${DESTDIR}usr/share/i18n/SUPPORTED
ALIAS=${DESTDIR}usr/share/locale/locale.alias

#
# Grab any user options in their config file
options=$(sed -n \
	-e '/^[[:space:]]*#%/s:^[[:space:]]*#%[[:space:]]*::p'\
	"${CONFIG}" 2>/dev/null
)
IFS=$'\n'
for option in ${options} ; do
	case ${option} in
		no-locale-archive)
			LOCALE_ARCHIVE=false
			;;
		*)
			ewarn "Unrecognized option '${option}'"
			;;
	esac
done
unset IFS

[[ -n ${ALL} ]] && CONFIG=${SUPPORTED}

# Extract the location of the locale dir on the fly as `localedef --help` has:
#                        locale path    : /usr/lib64/locale:/usr/share/i18n
# For long paths, the line may get wrapped into two, in which case space (' ') is replaced
# by newline (\n).
LOCALEDIR=$(LC_ALL="C" "${DESTDIR}"usr/bin/localedef --help | sed -n -r '/locale path/{N;s|.*:[ \n](.*):/.*|\1|;p}')
LOCALEDIR="${DESTDIR}${LOCALEDIR#${EPREFIX}}"
if [[ $? -ne 0 ]] || [[ -z ${LOCALEDIR} ]] || [[ ${LOCALEDIR} != ${DESTDIR}/usr/lib*/locale ]] ; then
	eerror "Unable to parse the output of your localedef utility." 1>&2
	eerror "File a bug about this issue and include the output of 'localedef --help'." 1>&2
	exit 1
fi

# Only generate locales the user specified before falling back to the config.
locales_to_generate=${GENERATE}

if [[ -z ${locales_to_generate} ]] && [[ -e ${CONFIG} ]] ; then
	locales_to_generate=$(sed \
		-e 's:#.*::' \
		-e '/^[[:space:]]*$/d' \
		"${CONFIG}" | sort)
	# Sanity check to make sure people did not duplicate entries. #550884
	# The first column must be unique specifically. #235555
	dup_locales_to_generate=$(
		echo "${locales_to_generate}" | \
			awk '{ if ($1 == last) { print lastline; print; } else { lastline = $0; last = $1; } }')
	if [[ -n ${dup_locales_to_generate} ]] ; then
		ewarn "These locales have been duplicated in your config:\n${dup_locales_to_generate}"
		ewarn "Some might be filtered, but you must fix it."
		locales_to_generate=$(echo "${locales_to_generate}" | uniq)
	fi
fi

# Transform the name in locales.gen to the name used when storing the locale data in
# /usr/lib/locale/.  This normalize algo is taken out of the glibc localedef source:
# https://sourceware.org/git/?p=glibc.git;a=blob;f=locale/programs/localedef.c;hb=glibc-2.34#l562
normalize() {
	if [[ $1 == *.* ]] ; then
		local ret=${1##*.}
		ret=${ret,,}
		echo "${1%%.*}.${ret//-}"
	else
		echo "$1"
	fi
}

# These funky sed's are based on the stuff in glibc's localedata/Makefile
# Basically we want to rewrite the display like so:
# <locale without a . or @>.<charmap>[@extra stuff after the @ in the locale]
# en_US       ISO-8859-1  -> en_US.ISO-8859-1
# en_US.UTF-8 UTF-8       -> en_US.UTF-8
# de_DE@euro  ISO-8859-15 -> de_DE.ISO-8859-15@euro
locales_disp=$(echo "${locales_to_generate}" | sed \
	-e '   /@/   s:[[:space:]]*\([^@[:space:]]*\)\([^[:space:]]*\)[[:space:]]\+\([^[:space:]]*\):\1.\3\2:' \
	-e '/^[^@]*$/s:[[:space:]]*\([^.[:space:]]*\)\([^[:space:]]*\)[[:space:]]\+\([^[:space:]]*\):\1.\3:')

# Now check the normalized version for C.UTF-8, and add it if not present
if [[ -z ${locales_to_generate} ]] ; then
	if [[ -z ${JUST_LIST} ]] ; then
		[[ ${QUIET} -eq 0 ]] && \
			ewarn "No locales to generate found, keeping archive but ensuring C.UTF-8 is present"
		KEEP=1
		UPDATE=1
		locales_disp='C.UTF-8'
		locales_to_generate='C.UTF-8 UTF-8'
	fi
else
	if echo ${locales_disp} | grep -vqi 'C.UTF-8' ; then
		locales_to_generate=$(echo "${locales_to_generate}" ; echo -n 'C.UTF-8 UTF-8')
		locales_disp=$(echo "${locales_disp}" ; echo -n 'C.UTF-8')
	fi
fi

mkdir -p "${LOCALEDIR}"
if [[ -z ${KEEP} && -z ${UPDATE} ]] ; then
	# Remove all old locale dir and locale-archive before generating new
	# locale data.  Scrubbing via update is done elsewhere.
	rm -rf "${LOCALEDIR}"/* &> /dev/null || true
fi

eval declare -a locales_disp=(${locales_disp})
eval declare -a locales_to_generate=(${locales_to_generate})
total=$(( ${#locales_to_generate[*]} / 2 ))

[[ ${QUIET} -eq 0 ]] && [[ -z ${JUST_LIST} ]] && \
einfo "Generating ${total} locales (this might take a while) with ${JOBS_MAX} jobs"

if [[ -n ${UPDATE} ]] ; then
	# normalize newlines into spaces
	existing_locales=" $(echo $(locale -a 2>/dev/null)) "
fi

generate_locale() {
	local output=""

	if [[ -z ${ASK} ]] && [[ ${QUIET} -eq 0 ]] ; then
		output=" (${cnt_fmt}/${total}) Generating ${disp}"
	fi

	if [[ $(( JOB_IDX_E - JOB_IDX_S )) == ${JOBS_MAX} ]] ; then
		wait ${JOB_PIDS[$(( JOB_IDX_S++ ))]}
		JOB_RETS+=( $? )
	fi
	(
		# Accumulate all the output in one go so the parallel
		# jobs don't tromp on each other
		x=$(
			[[ -n ${output} ]] && ebegin "${output}"
			# In most cases, localedef can just use the system glibc.
			# However, if we are within a major glibc upgrade, this may fail
			# in src_* phases since the new localedef links against the new
			# glibc, but the new glibc is not installed yet...
			if [[ -z ${INPLACE_GLIBC} ]] ; then
				"${DESTDIR}"usr/bin/localedef ${LOCALEDEF_OPTS} \
					--no-archive \
					-i "${input}" \
					-f "${charmap}" \
					-A "${ALIAS}" \
					--prefix "${DESTDIR%${EPREFIX}/}/" \
					"${locale}" 2>&1
			else
				# We assume that the current directory is "${ED}"/$(get_libdir),
				# see the glibc ebuild, function glibc_sanity_check(), for why.
				LC_ALL=C ./ld-*.so --library-path . \
					"${DESTDIR}"usr/bin/localedef ${LOCALEDEF_OPTS} \
					--no-archive \
					-i "${input}" \
					-f "${charmap}" \
					-A "${ALIAS}" \
					--prefix "${DESTDIR%${EPREFIX}/}/" \
					"${locale}" 2>&1
			fi
			ret=$?
			[[ -n ${output} ]] && eend ${ret}
			exit ${ret}
		)
		ret=$?
		if [[ -n ${output} ]] ; then
			echo "${x}"
		elif [[ ${ret} -ne 0 ]] ; then
			eerror "${disp}: ${x}"
		fi

		if [[ ${ret} -ne 0 && ${locale} == */* ]] ; then
			ewarn "Perhaps you meant to use a space instead of a / in your config file ?"
		fi
		exit ${ret}
	) &
	JOB_PIDS+=( $! )
	: $(( ++JOB_IDX_E ))
}

JOB_PIDS=()
JOB_RETS=()
JOB_IDX_S=0
JOB_IDX_E=0
cnt=0
lidx=0
# Keep track of (normalized) locales generated in case the request has different inputs that
# normalize down to the same value.  We trim $existing_locales as we go for later use which
# prevents its direct use.
generated_locales=" "
while [[ -n ${locales_to_generate[${lidx}]} ]] ; do
	: $(( ++cnt ))
	locale=${locales_to_generate[$((lidx++))]}
	charmap=${locales_to_generate[$((lidx++))]}

	# XXX: if we wanted to, we could check existence of
	#      ${LOCALES}/${locale} and ${CHARMAPS}/${charmap}
	#      this would fail for things like "en_US.UTF8", but
	#      in that case we could fall back to checking the
	#      SUPPORTED file ... then again, the localedef
	#      below will abort nicely for us ...
	if [[ -z ${locale} || -z ${charmap} ]] ; then
		eerror "Bad entry in locale.gen: '${locale} ${charmap}'; skipping"
		continue
	fi

	disp=${locales_disp[$(( cnt - 1 ))]}

	normalized_locale=$(normalize ${locale})
	if [[ ${generated_locales} == *" ${normalized_locale} "* ]] ; then
		already_generated="true"
	else
		already_generated="false"
	fi
	generated_locales+="${normalized_locale} "
	if ${already_generated} || \
	   [[ -n ${UPDATE} && ${existing_locales} == *" ${normalized_locale} "* ]] ; then
		existing_locales=${existing_locales/ ${normalized_locale} / }
		if [[ ${QUIET} -eq 0 ]] ; then
			cnt_fmt=$(printf "%${#total}i" ${cnt})
			einfo " (${cnt_fmt}/${total}) Skipping ${disp}"
		fi
		continue
	fi

	# If the locale is like 'en_US.UTF8', then we really want 'en_US'
	if [[ -f ${LOCALES}/${locale} ]] ; then
		input=${locale}
	else
		input=${locale%%.*}
	fi

	if [[ -z ${JUST_LIST} ]] ; then
		# Format the output for the question/status
		cnt_fmt=$(printf "%${#total}i" ${cnt})
		if [[ -n ${ASK} ]] ; then
			einfon " (${cnt_fmt}/${total}) Generate ${disp} ? (Y/n) "
			read user_answer
			[[ ${user_answer} == [nN]* ]] && continue
		fi
		generate_locale
	else
		echo "${disp}"
	fi
done

for (( i = JOB_IDX_S; i < JOB_IDX_E; ++i )) ; do
	wait ${JOB_PIDS[i]}
	JOB_RETS+=( $? )
done
ret=$(( 0 ${JOB_RETS[@]/#/+} ))

[[ ${QUIET} -eq 0 ]] && [[ -z ${JUST_LIST} ]] && \
einfo "Generation complete"

if ${LOCALE_ARCHIVE} && [[ -z ${JUST_LIST} ]] ; then
	# need to check that at least one locale has to be added
	if [[ $(echo "${LOCALEDIR}"/*/) != "${LOCALEDIR}"'/*/' ]] ; then
		[[ ${QUIET} -eq 0 ]] && ebegin "Adding locales to archive"
		# The pattern ends with / on purpose: we don't care about files (like
		# locale-archive) in the locale subdir, and we definitely don't want to
		# delete them!
		for LOC in "${LOCALEDIR}"/*/; do
			LOC=${LOC%/} # Strip trailing /, since localedef doesn't like it
			x=$(
				# In most cases, localedef can just use the system glibc.
				# However, if we are within a major glibc upgrade, this may fail
				# in src_* phases since the new localedef links against the new
				# glibc, but the new glibc is not installed yet...
				if [[ -z ${INPLACE_GLIBC} ]] ; then
					"${DESTDIR}"usr/bin/localedef \
						--add-to-archive "${LOC}" \
						--replace \
						--prefix "${DESTDIR%${EPREFIX}/}/"
				else
					# We assume that the current directory is "${ED}"/$(get_libdir),
					# see the glibc ebuild, function glibc_sanity_check(), for why.
					LC_ALL=C ./ld-*.so --library-path . \
						"${DESTDIR}"usr/bin/localedef \
						--add-to-archive "${LOC}" \
						--replace \
						--prefix "${DESTDIR%${EPREFIX}/}/"
				fi
				ret=$?
				if [[ -n ${output} ]] ; then
					echo "${x}"
				elif [[ ${ret} -ne 0 ]] ; then
					eerror "${disp}: ${x}"
				fi
				if [[ $ret -eq 0 ]]; then
					rm -r "${LOC}"
				fi
				exit ${ret}
			)
		done
		[[ ${QUIET} -eq 0 ]] && eend ${ret}
	elif [[ ${QUIET} -eq 0 ]] ; then
		einfo "No locales are to be added to the archive."
	fi
fi

# Remove locales that existed but were not requested
if [[ -n ${UPDATE} ]] && [[ -z ${JUST_LIST} ]] ; then
	# Ignore these pseudo locales
	existing_locales=${existing_locales/ C / }
	existing_locales=${existing_locales/ POSIX / }
	if [[ -n ${existing_locales// } ]] ; then
		if [[ -z ${KEEP} ]] ; then
			[[ ${QUIET} -eq 0 ]] && einfo "Scrubbing old locales:"${existing_locales}
			cd "${LOCALEDIR}" && rm -rf ${existing_locales}
		else
			[[ ${QUIET} -eq 0 ]] && einfo "Keeping old locales:"${existing_locales}
		fi
	fi
fi

exit ${ret}