update-tables

#! /bin/bash
#
# update-tables
# Copyright (C) 2016 Adrian Perez <aperez@igalia.com>
#
# Distributed under terms of the MIT license.
#
set -e

WIDE_URL='http://www.unicode.org/Public/UNIDATA/EastAsianWidth.txt'
ZERO_URL='http://www.unicode.org/Public/UNIDATA/extracted/DerivedGeneralCategory.txt'

# The version numbers of the files above are parsed from the files themselves.
WIDE_VER='(unknown)'
ZERO_VER='(unknown)'

srcdir=$(dirname "$0")

WIDE_FILE="${srcdir}/${WIDE_URL##*/}"
ZERO_FILE="${srcdir}/${ZERO_URL##*/}"

[[ -r ${WIDE_FILE} ]] || wget -c -O "${WIDE_FILE}" "${WIDE_URL}"
[[ -r ${ZERO_FILE} ]] || wget -c -O "${ZERO_FILE}" "${ZERO_URL}"

# Pad hex values to be 8 characters wide, and prepend "0x" to them.
u32hex () {
	local -i len=${#1}
	local -i nzeros=$(( 8 - len ))
	echo -n '0x'
	for (( ; nzeros > 0 ; nzeros-- )) ; do
		echo -n '0'
	done
	echo -n "$1"
}

format_range () {
	if [[ $1 = *..* ]] ; then
		u32hex "${1%..*}"
		echo -n ','
		u32hex "${1#*..}"
		echo ','
	else
		u32hex "${1}"
		echo -n ','
		u32hex "${1}"
		echo ','
	fi
}

zero_table () {
	local tmpfile="${TMPDIR:-/tmp}/update-tables-$$-${RANDOM}"
	echo "-- Autogenerated from ${ZERO_FILE##*/}"
	echo 'return {'
	while read -r line ; do
		if [[ -z ${line} || ${line} = \#* ]] ; then
			read line rest <<< "${line}"
			if [[ ${rest} = *.txt ]] ; then
				ZERO_VER=${rest}
			elif [[ ${rest} = Date:\ 20* || ${rest} = ©\ 20* ]] ; then
				ZERO_VER="${ZERO_VER}, ${rest}"
			fi
			continue
		fi
		read -a items <<< "${line}"
		if [[ ${items[2]} == Mn || ${items[2]} == Me ]] ; then
			format_range "${items[0]}"
		fi
	done > "${tmpfile}"
	sort "${tmpfile}"
	rm "${tmpfile}"
	echo '}'
}

wide_table () {
	local tmpfile="${TMPDIR:-/tmp}/update-tables-$$-${RANDOM}"
	echo "-- Autogenerated from ${WIDE_FILE##*/}"
	echo 'return {'
	while read -r item rest ; do
		if [[ -z ${item} || ${item} = \#* ]] ; then
			if [[ ${rest} = *.txt ]] ; then
				WIDE_VER=${rest}
			elif [[ ${rest} = Date:\ 20* || ${rest} = ©\ 20* ]] ; then
				WIDE_VER="${WIDE_VER}, ${rest}"
			fi
			continue
		fi
		range=${item%;*}
		flags=${item#*;}
		if [[ ${flags} = W* || ${flags} = F* ]] ; then
			format_range "${range}"
		fi
	done > "${tmpfile}"
	sort "${tmpfile}"
	rm "${tmpfile}"
	echo '}'
}

make_readme () {
	# Pick newest LuaRocks version
	local V=$(find "${srcdir}/luarocks" -name 'wcwidth-[0-9]*.rockspec' \
		| sed 's,^.*/wcwidth-\([0-9\.]\+\).*$,\1,' | sort -g | tail -1)
	sed -e "s+@@WIDE_FILE@@+${WIDE_FILE##*/}+g" \
		-e "s+@@ZERO_FILE@@+${ZERO_FILE##*/}+g" \
		-e "s+@@WIDE_URL@@+${WIDE_URL}+g" \
		-e "s+@@ZERO_URL@@+${ZERO_URL}+g" \
		-e "s+@@WIDE_VER@@+${WIDE_VER}+g" \
		-e "s+@@ZERO_VER@@+${ZERO_VER}+g" \
		-e "s+@@LUAROCKS_VER@@+${V}+g"
}

wide_table  < "${WIDE_FILE}" > "${srcdir}/wcwidth/widetab.lua"
zero_table  < "${ZERO_FILE}" > "${srcdir}/wcwidth/zerotab.lua"
make_readme < "${srcdir}/README.md.in" > "${srcdir}/README.md"