xref: /freebsd-11-stable/tools/tools/locale/tools/finalize

#!/bin/sh
#
# $FreeBSD: stable/11/tools/tools/locale/tools/finalize 367262 2020-11-02 01:05:41Z lwhsu $
#
# This is a helper script for the Makefile in the parent directory.
# When the localization definitions are generated in the draft area,
# this script will copy base ones that others symlink to, and rearrange
# the generate makefile to pull the LOCALES first.
#

set -e

usage ()
{
	echo "finalize <type>' to package standard localization"
	echo "type must be one of { monetdef, msgdef, numericdef, timedef, colldef, ctypedef }"
	exit 1
}

[ $# -ne 1 ] && usage
[ $1 = "monetdef" -o $1 = "msgdef" -o $1 = "colldef" -o \
  $1 = "numericdef" -o $1 = "timedef" -o $1 = "ctypedef" ] || usage

self=$(realpath $0)
base=$(dirname ${self})
old=${base}/../${1}.draft
new=${base}/../${1}
TEMP=/tmp/${1}.locales
TEMP2=/tmp/${1}.hashes
TEMP3=/tmp/${1}.symlinks
TEMP4=/tmp/${1}.mapped
FULLMAP=/tmp/utf8-map
FULLEXTRACT=/tmp/extracted-names
AWKCMD="/## PLACEHOLDER/ { \
	  while ( getline line < \"${TEMP}\" ) {print line} } \
	/## SYMPAIRS/ { \
	  while ( getline line < \"${TEMP3}\" ) {print line} } \
	/## LOCALES_MAPPED/ { \
	  while ( getline line < \"${TEMP4}\" ) {print line} } \
	!/## / { print \$0 }"

# Rename the sources with 3 components name into the POSIX version of the name using @modifier
cd $old
for i in *_*_*.*.src; do
	oldname=${i%.*}
	nname=`echo $oldname | awk '{ split($0, a, "_"); print a[1]"_"a[3]"@"a[2];} '`
	mv -i ${oldname}.src ${nname}.src
done
	sed -i '' -Ee "s/([a-zA-Z]{2})_([a-zA-Z]+)_([a-zA-Z]{2}).([a-zA-Z0-9-]+)/\1_\3.\4@\2/g" ${old}/Makefile

# For variable without @modifier ambiguity do not keep the @modifier
for i in *@*.src; do
	oldname=${i%.*}
	shortname=${oldname%@*}
	if [ $(ls ${shortname}@* | wc -l) -eq 1 -a ! -f ${shortname}.src ] ; then
		mv -i $i ${shortname}.src
		sed -i '' -e "s/${oldname}/${shortname}/g" ${old}/Makefile
	fi
done

# Rename the modifiers into non abbreviated version
for i in *@Latn.src; do
	if [ "$i" = "*@Latn.src" ]; then
		break
	fi
	mv ${i} ${i%@*}@latin.src
done
	sed -i '' -e "s/@Latn/@latin/g" ${old}/Makefile
for i in *@Cyrl.src; do
	if [ "$i" = "*@Cyrl.src" ]; then
		break
	fi
	mv ${i} ${i%@*}@cyrillic.src
done
	sed -i '' -e "s/@Cyrl/@cyrillic/g" ${old}/Makefile

# On locales with multiple modifiers rename the "default" version without the @modifier
default_locales="sr_RS@cyrillic"
for i in ${default_locales}; do
	localename=${i%@*}
	mod=${i#*@}
	for l in ${localename}.*@${mod}.src; do
		if [ "$l" = "${localename}.*@${mod}.src" ]; then
			break
		fi
		mv ${l} ${l%@*}.src
		sed -i '' -e "s/${l%.*}/${l%@*}/g" ${old}/Makefile
	done
done
cd -

grep '^LOCALES+' ${old}/Makefile > ${TEMP}

if [ $1 = "ctypedef" ]
then
	keep=$(cat ${TEMP} | awk '{ print $2 ".src" }')
	(cd ${old} && md5 -r ${keep} | sort) > ${TEMP2}
	keep=$(awk '{ if ($1 != last1) print $2; last1 = $1; }' ${TEMP2})
	for original in ${keep}
	do
		cp ${old}/${original} ${new}/
	done
	awk '{ if ($1 == last1) { print "SYMPAIRS+=\t" last2 " " $2 } \
	else {last1 = $1; last2 = $2}}' ${TEMP2} > ${TEMP3}
	rm -f ${TEMP2}
	/usr/bin/sed -E -e 's/[ ]+/ /g' \
		${CLDRDIR}/posix/UTF-8.cm \
		> ${base}/../etc/final-maps/map.UTF-8
	/usr/bin/sed -E -e 's/[ ]+/ /g' \
		${CLDRDIR}/posix/eucCN.cm \
		> ${base}/../etc/final-maps/map.eucCN
	/usr/bin/sed -E -e 's/[ ]+/ /g' \
		${CLDRDIR}/posix/eucCN.cm \
		> ${base}/../etc/final-maps/map.GB2312

	# GB18030 and Big5 are pre-generated from CLDR data
	CHARMAPS="ARMSCII-8 CP1131 CP1251 \
		CP866 GBK ISCII-DEV ISO8859-1 \
		ISO8859-13 ISO8859-15 ISO8859-2 ISO8859-4 \
		ISO8859-5 ISO8859-7 ISO8859-9 KOI8-R KOI8-U \
		PT154 SJIS US-ASCII eucJP eucKR"

	for map in ${CHARMAPS}
	do
		encoding=${map}
		/usr/local/bin/perl ${base}/convert_map.pl \
			${base}/../etc/charmaps/${map}.TXT ${encoding} \
			| /usr/bin/sed -E -e 's/	+/ /g' \
		> ${base}/../etc/final-maps/map.${map}
		echo map ${map} converted.
	done

elif [ $1 = "colldef" ]
then
	awk -v tmp4=${TEMP4} '$1 == "SAME+=" && $0 !~ /legacy/ {
		orig=$2
		dest=$3
		gsub(/.*\./, "", orig)
		gsub(/.*\./, "", dest)
		if (orig != dest )
			print "LOCALES_MAPPED+=\t"$2 " "$3 > tmp4
		}' ${old}/Makefile

	for line in $(awk '{ print $3 }' ${TEMP4}); do
		sed -i '' "/^SAME.*$line$/d" ${old}/Makefile
	done
	echo "" >> ${TEMP4}
	for enc in ${COLLATIONS_SPECIAL}; do
		nname=`echo $enc | sed -e 's/_Hans//g'`
		sed -i '' "/^.*${nname}$/d" ${TEMP4}
		echo "LOCALES+=	${nname}" >> ${TEMP4}
	done

	keep=$(cat ${TEMP} | awk '{ print $2 }')
	for original in ${keep}
	do
		cp ${old}/${original}.src ${new}/
	done
else  # below is everything but ctypedef

	keep=$(cat ${TEMP} | awk '{ print $2 }')
	for original in ${keep}
	do
		cp ${old}/${original}.src ${new}/
	done

fi

grep -v '^LOCALES+' ${old}/Makefile | awk "${AWKCMD}" > ${new}/Makefile

rm -f ${TEMP} ${TEMP3} ${TEMP4}