#! /bin/sh

# generate mnemonics for accented letters

if make UnicodeData.txt >&2
then	true
else	echo Could not acquire Unicode data file UnicodeData.txt
	exit 1
fi


script=`echo ${1-GREEK} | tr '[a-z]' '[A-Z]'`

file=mkaccents.$script
export file
case "$script" in
ARABIC) mnemofile=mnemos.ara;;
CYRILLIC) mnemofile=mnemos.cyr;;
GREEK) mnemofile=mnemos.grk;;
HEBREW) mnemofile=mnemos.heb;;
LATIN) mnemofile=mnemos.lat;;
-) mnemofile=mnemos.com;;
esac


#############################################################################
# generate C program header
(
cat <<\/eoc
#include <stdio.h>

int b;

void
printutf8 (unichar)
	unsigned long unichar;
{
	if (unichar == '\\' || unichar == '"') {
		printf ("\\");
	}

	if (unichar < 0x80) {
		printf ("%c", unichar);
	} else if (unichar < 0x800) {
		printf ("%c", 0xC0 | (unichar >> 6));
		printf ("%c", 0x80 | (unichar & 0x3F));
	} else if (unichar < 0x10000) {
		printf ("%c", 0xE0 | (unichar >> 12));
		printf ("%c", 0x80 | ((unichar >> 6) & 0x3F));
		printf ("%c", 0x80 | (unichar & 0x3F));
	} else if (unichar < 0x200000) {
		printf ("%c", 0xF0 | (unichar >> 18));
		printf ("%c", 0x80 | ((unichar >> 12) & 0x3F));
		printf ("%c", 0x80 | ((unichar >> 6) & 0x3F));
		printf ("%c", 0x80 | (unichar & 0x3F));
	} else if (unichar < 0x4000000) {
		printf ("%c", 0xF8 | (unichar >> 24));
		printf ("%c", 0x80 | ((unichar >> 18) & 0x3F));
		printf ("%c", 0x80 | ((unichar >> 12) & 0x3F));
		printf ("%c", 0x80 | ((unichar >> 6) & 0x3F));
		printf ("%c", 0x80 | (unichar & 0x3F));
	} else if (unichar < 0x80000000) {
		printf ("%c", 0xFC | (unichar >> 30));
		printf ("%c", 0x80 | ((unichar >> 24) & 0x3F));
		printf ("%c", 0x80 | ((unichar >> 18) & 0x3F));
		printf ("%c", 0x80 | ((unichar >> 12) & 0x3F));
		printf ("%c", 0x80 | ((unichar >> 6) & 0x3F));
		printf ("%c", 0x80 | (unichar & 0x3F));
	}
}

void
compose (base, postfix, code, name, combining)
	unsigned long base;
	char * postfix;
	unsigned long code;
	char * name;
	int combining;
{
/*	{"α'", 0x03AC / * ά GREEK SMALL LETTER ALPHA WITH TONOS * /},	*/
	printf ("	{\"");
	printutf8 (base);
	printf ("%s\", 0x%04X /* %s", postfix, code, combining ? "◌" : "");
	printutf8 (code);
	printf (" %s */},\n", name);
}

int
main () {
/eoc
) > $file.c


#############################################################################
# generate data into C program

case "$script" in
-) sed -e "s,^\([^;]*\);\(COMBINING [^;]*\);[^;]*;[^;]*;[^;]*;\([^;]*\);.*,\1;\2;\3," \
	-e t -e d UnicodeData.txt
	;;
*) sed -e "s,^\([^;]*\);\($script[^;]* LETTER [^;]*\);[^;]*;[^;]*;[^;]*;\([^;]*\);.*,\1;\2;\3," \
	-e t -e d UnicodeData.txt > $file
	egrep -v " WITH " $file > $file.base
	egrep " WITH " $file
	;;
esac |
(cat <<\/EOS
#char "03AC" "GREEK SMALL LETTER ALPHA WITH TONOS"
#compose "GREEK SMALL LETTER ALPHA" "'" "03B1 0301"

char () {
	code=$1
	name="$2"
}

compose () {
	basename="$1"
	postfix="$2"
#	0391;GREEK CAPITAL LETTER ALPHA;
	base=`sed -e "s,^\([^;]*\);$basename;.*,\1," -e t -e d $file.base`
	echo "	compose (0x$base, \"$postfix\", 0x$code, \"$name\", 0);"
}

combining () {
	basename="$1"
	postfix="$2"
	echo "	compose ('	', \"$postfix\", 0x$code, \"$name\", 1);"
}

/EOS

sed -e "s,\(..*\)	\(..*\),s= WITH \2:=\1=g," \
	-e 's,[&],\\&,g' accents.cfg > accents.sed

case "$script" in
-) sed -e h -e 's,^\([^;]*\);\([^;]*\);.*,char "\1" "\2",' -e p -e g \
	-e 's,;COMBINING ,;SPACE WITH ,' \
	-e 's,^\([^;]*\);\(.*\)\( WITH [^;]*\);\([^;]*\),combining "\2" "\3:" "\4",' \
	-e 's, AND ,: WITH ,g' > $file.with
	sed -e "/^char/ b" -f accents.sed -e t -e d $file.with
	;;
*) sed -e h -e 's,^\([^;]*\);\([^;]*\);.*,char "\1" "\2",' -e p \
	-e g \
	-e 's,^\([^;]*\);\(.*\)\( WITH [^;]*\);\([^;]*\),compose "\2" "\3:" "\4",' \
	-e 's, AND ,: WITH ,g' > $file.with
	sed -f accents.sed $file.with
	;;
esac

) | sh >> $file.c


rm -f $file $file.base $file.with


#############################################################################
# finish C program
(
cat <</eoc
	return 0;
}
/eoc
) >> $file.c


if cc -o $file.exe $file.c
then	echo "	/* $script accented mnemos - generated automatically */" > $file.mnemo
	./$file.exe >> $file.mnemo
	rm -f $file.exe $file.c
fi

yellow=
green=
red=
normal=

case "$mnemofile" in
mnemos.*)	if cmp $file.mnemo $mnemofile
		then	echo "$yellow $mnemofile unchanged $normal"
			#rm -f $file.mnemo
		else	echo "$red $mnemofile changed $normal"
		fi
		mv $file.mnemo $mnemofile
		;;
esac