#!/bin/rc

. common.rc

Unicode | awk -F';' '
' ^ $hex ^ '

function init() {
	mark=0;		# decimal of 1st matching in range
	mdes="";	# description of mark.
	last=0;		# last in range
	ldes="";	# last desc.
}

BEGIN {
	print "static";
	print "Rune\t*__isrune[] =";
	print "{";

	init();
}

{
	codepoint = hex($1);
	d = tolower($2)

	if(d ~ /<(plane [0-9]+ )?private use/)
		next;
	key = ""
	if(d ~ /<[^,>]+, (first|last)>/){
		key = d
		sub(/,.*/, "", key)
	}
	if(last + 1 == codepoint || length(key)>0 && key == okey){
		if(key != okey)
			okey=""
		#if(length(key)>0)printf "KEY [%s] [%s]\n", key, okey
		last = codepoint;
		ldes = d;
		next;
	}

	if (last && last > mark){
		printf("\t0x%04x, 0x%04x,\t/* %s - %s */\n", mark, last, mdes, ldes);
	} else if (last) {
		printf("\t0x%04x, 0x%04x,\t/* %s */\n", mark, last, ldes);
	}

	init();
	okey = key
	mark=last=codepoint;
	ldes=mdes=d;
}

END {
	if (last && last > mark){
		printf("\t0x%04x, 0x%04x,\t/* %s - %s */\n", mark, last, mdes, ldes);
	} else if (last) {
		printf("\t0x%04x, 0x%04x,\t/* %s */\n", mark, last, ldes);
	}
	print "};"
	print ""
}
 ' | Sprint