#!/bin/bash
# -*- sh -*-

# Takes on stdin a file formated like the official UnicodeData list;
# Writes on stdout a fallback table suitable for `consolechars -k'

# Note: we assume all diacrs match a non-diacr !

if [ $# != 3 ]
then
    echo >&2 "Usage: $(basename $0) FULL_PATTERN REDUCTED_PATTERN TRANSLATION"
    echo >&2 "          < UnicodeData-2.txt >out.fallback"
    exit 1
fi

FULL_GREP_PATTERN=$1
REDUCTED_PATTERN=$2
TRANSLATION=$3

FULL_SED_PATTERN=$(echo ${FULL_GREP_PATTERN} | sed -e 's/?/\\{0,1\\}/g' -e 's/\([()|]\)/\\\1/g')

FILE=/tmp/file.$$
cut -d\; -f1,2 >$FILE

cat << EOF
#
# Fallback file from chars matching:
# '${FULL_GREP_PATTERN}'
# to chars matching:
# '${REDUCTED_PATTERN}'
# using as a translation rule:
# '${TRANSLATION}'
#
# Auto-generated by $(basename $0) from UnicodeData list 
#
# ** DO NOT EDIT MANUALLY **
#

EOF

cat < $FILE |
    # select only targeted lines
    egrep -f <(echo ";${FULL_GREP_PATTERN}$") |
    # filter out those we want to reduct to
    egrep -v -f <(echo ";${REDUCTED_PATTERN}$") |
    # translate into wished reduction, and keep orig name as 3rd field
    sed -f <(echo "s/;${FULL_SED_PATTERN}$/;${TRANSLATION}\0/") |
    # translate into wished unicode
    sed -f <(
	# construct a sed filter that adds their unicodes to reduction names
	egrep < $FILE ";${REDUCTED_PATTERN}$" |
	    awk -F\; '{print "s/;" $2 ";/;U+" $1 "\\0/"}'
    ) |
    # filter out lines that did not give a reducted pattern
    grep ';U+' |
    # format lines in fallback-file syntax
    awk -F\; '{print "# " $4 " : " $3 "\nU+" $1 " " $2}'

rm $FILE