#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>
#include <arpa/inet.h>
#include <string.h>

const unsigned int symbolGreekTranslation[256] = {
	/* 000 (00) */	0xFFFD /*' '*/, 0xFFFD /*'
'*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 
	/* 004 (04) */	0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 
	/* 008 (08) */	0xFFFD /*'backspace? '*/, 0xFFFD /*'\t'*/, 0x000a /*\n*/, 0xFFFD /*''*/, 
	/* 012 (0c) */	0xFFFD /*''*/, 0x000d /*'\r'*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 
	/* 016 (10) */	0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 
	/* 020 (14) */	0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 
	/* 024 (18) */	0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 
	/* 028 (1c) */	0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 0xFFFD /*''*/, 
	/* 032 (20) */	0x0020 /*' '*/, 0x0021 /*'!'*/, 0x03C2 /*'"'*/, 0x0023 /*'#'*/, 
	/* 036 (24) */	0x0024 /*'$'*/, 0x0025 /*'%'*/, 0x0026 /*'&'*/, 0x0342 /* circumflex perispomeni'''*/, 
	/* 040 (28) */	0x0028 /*'('*/, 0x0029 /*')'*/, 0x002a /*'*'*/, 0x002b /*'+'*/, 
	/* 044 (2c) */	0x002c /*','*/, 0x002d /*'-'*/, 0x002e /*'.'*/, 0x002f /*'/'*/, 
	/* 048 (30) */	0x0030 /*'0'*/, 0x0031 /*'1'*/, 0x0032 /*'2'*/, 0x0033 /*'3'*/, 
	/* 052 (34) */	0x0034 /*'4'*/, 0x0035 /*'5'*/, 0x0036 /*'6'*/, 0x0037 /*'7'*/, 
	/* 056 (38) */	0x0038 /*'8'*/, 0x0039 /*'9'*/, 0x00b7 /* middle dot ':'*/, ';' /* possibly a grave varia (see code below) */, 
	/* 060 (3c) */	0x003c /*'<'*/, 0x003d /*'='*/, 0x0308 /*'>'*/, 0x003f /*'?'*/, 
	/* 064 (40) */	0x0040 /*'@'*/, 0x0391 /*'A'*/, 0x0392 /*'B'*/, 0x03A7 /*'C'*/, 
	/* 068 (44) */	0x0394 /*'D'*/, 0x0395 /*'E'*/, 0x03A6 /*'F'*/, 0x0393 /*'G'*/, 
	/* 072 (48) */	0x0397 /*'H'*/, 0x0399 /*'I'*/, 0x0314 /*'J'*/, 0x039a /*'K' bug in .py */, 
	/* 076 (4c) */	0x039b /*'L'*/, 0x039c /*'M'*/, 0x039d /*'N'*/, 0x039f /*'O'*/, 
	/* 080 (50) */	0x03a0 /*'P'*/, 0x0398 /*'Q'*/, 0x03a1 /*'R'*/, 0x03a3 /*'S'*/, 
	/* 084 (54) */	0x03a4 /*'T'*/, 0x03a5 /*'U'*/, 0x0384 /*'V'*/, 0x03a9 /*'W'*/, 
	/* 088 (58) */	0x039e /*'X'*/, 0x03a8 /*'Y'*/, 0x005a /*'Z'*/, 0x005b /*'['*/, 
	/* 092 (5c) */	0x005c /*'\'*/, 0x005d /*']'*/, 0x005e /*'^'*/, 0x005f /*'_'*/, 
	/* 096 (60) */	0x0060 /*'`'*/, 0x03b1 /*'a'*/, 0x03b2 /*'b'*/, 0x03c7 /*'c'*/, 
	/* 100 (64) */	0x03b4 /*'d'*/, 0x03b5 /*'e'*/, 0x03c6 /*'f'*/, 0x03b3 /*'g'*/, 
	/* 104 (68) */	0x03b7 /*'h'*/, 0x03b9 /*'i'*/, 0x0313 /* smooth breathing 'j'*/, 0x03ba /*'k'*/, 
	/* 108 (6c) */	0x03bb /*'l'*/, 0x03bc /*'m'*/, 0x03bd /*'n'*/, 0x03bf /*'o'*/,            
	/* 112 (70) */	0x03c0 /*'p'*/, 0x03b8 /*'q'*/, 0x03c1 /*'r'*/, 0x03c3 /*'s'*/, 
	/* 116 (74) */	0x03c4 /*'t'*/, 0x03c5 /*'u'*/, 0x0301 /* acute tonos 'v'*/, 0x03c9 /*'w'*/, 
	/* 120 (78) */	0x03be /*'x'*/, 0x03c8 /*'y'*/, 0x03b6 /*'z'*/, 0x007b /*'{'*/, 
	/* 124 (7c) */	0x007c /*'|'*/, 0x007d /*'}'*/, 0x03c2 /*'~'*/, 0x007f /*''*/, 
	/* 128 (80) */	' '    /*'�'*/, 0x0081 /*'�'*/, 'C',            0x0083 /*'�'*/, 
	/* 132 (84) */	0x0084 /*'�'*/, 0x003e /* blockend '�'*/, 0x1fbd /*elusionszeichen'�'*/, 0x0087 /*'�'*/, 
	/* 136 (88) */	0x0088 /*'�'*/, 0x003b /*; '�'*/, ' '  /*'�'*/, 0x008b /*'�'*/, 
	/* 140 (8c) */	'a',            'c',            0x008e /*'�'*/, 0x008f /*'�'*/, 
	/* 144 (90) */	0x0090 /*'�'*/, 0x0091 /*'�'*/, 0x0092 /*'�'*/, 0x0093 /*'�'*/, 
	/* 148 (94) */	0x0094 /*'�'*/, 0x0095 /*'�'*/, 0x0096 /*'�'*/, 0x0056 /*'�'*/, 
	/* 152 (98) */	0x0098 /*'�'*/, 0x0099 /*'�'*/, 0x002f /* / '�'*/, 0x009b /*'�'*/, 
	/* 156 (9c) */	0x009c /*'�'*/, 0x009d /*'�'*/, 0x009e /*'�'*/, 0x009f /*'�'*/, 
	/* 160 (a0) */	't',            0x00a1 /*'�'*/, 0x00a2 /*'�'*/, 0x00a3 /*'�'*/, 
	/* 164 (a4) */	0x00a4 /*'�'*/, 0x00a5 /*'�'*/, 0x00a6 /*'�'*/, 's',            
	/* 168 (a8) */	'r',            'g',            0x00aa /*'�'*/, 'e',             
	/* 172 (ac) */	'u',            0x0047 /* G '�'*/, 0x00ae /*'�'*/, '}' /*'�'*/, 
	/* 176 (b0) */	0x00b0 /*'�'*/, 0x0306 /*'�'*/, 'j',            0x0323 /*unterpunkt'�'*/, 
	/* 180 (b4) */	'y',            'm',            'd',            'w',            
	/* 184 (b8) */	0x00b8 /*'�'*/, 'p',            'b' /*'�'*/, 0x005b /* [ '�'*/, 
	/* 188 (bc) */	0x005d /*]'�'*/, 0x00bd /*'�'*/, 0x00be /*'�'*/, '{' /*'�'*/, 
	/* 192 (c0) */	0x00c0 /*'�'*/, 0x00BA /* degree '�'*/, 'l'/* 0x006c */, 'v' /*'�'*/, 
	/* 196 (c4) */	'f',            'x',            0x02bc/* apostrophe */, 0x00c7 /*'�'*/, 
	/* 200 (c8) */	0x00c8 /*'�'*/, ';' /*'�'*/, ' ' /*'�'*/, 0x00cb /*'�'*/, 
	/* 204 (cc) */	0x00cc /*'�'*/, 0x00cd /*'�'*/, 'Q',            'q',           
	/* 208 (d0) */	0x0304,         0x0305 /* intf uses: 0x00af oberstrich'�'*/, 0x00d2 /*'�'*/, 0x00d3 /*'�'*/, 
	/* 212 (d4) */	0x00d4 /*'�'*/, 0x00d5 /*'�'*/, 0x00d6 /*'�'*/, 'V',             
	/* 216 (d8) */	0x00d8 /*'�'*/, 0x00d9 /*'�'*/, 0x007c /*|'�'*/, 0x005c /*`'�'*/, 
	/* 220 (dc) */	0x00dc /*'�'*/, 0x00dd /*'�'*/, 0x00de /*'�'*/, 0x0302 /*'�'*/, 
	/* 224 (e0) */	0x00e0 /*'�'*/, 0x00e1 /*'�'*/, ' '    /*'�'*/, 'W',            
	/* 228 (e4) */	0x00e4 /*'�'*/, 'R',            0x00e6 /*'�'*/, 'Y',            
	/* 232 (e8) */	'U',            'i',            'S',            'D',         
	/* 236 (ec) */	'F',            'G',            0x00ee /*'�'*/, ' '    /*'�'*/, 
	/* 240 (f0) */	0x00f0 /*'�'*/, 'L',            ':' /*'�'*/, 0x00f3 /*'�'*/, 
	/* 244 (f4) */	0x00f4 /*'�'*/, 0x03DB /*stigma*/, 'n',            0x00f7 /*'�'*/, 
	/* 248 (f8) */	'J',            0x00f9 /*'�'*/, 'h',            'k',            
	/* 252 (fc) */	0x00fc /*'�'*/, 0x00fd /*'�'*/, 0x00fe /*'�'*/, 0x00ff /*'?'*/
};

const unsigned int isGreekLetter[256] = {
	/* 000 (00) */	0 /*' '*/, 0 /*'
'*/, 0 /*''*/, 0 /*''*/, 
	/* 004 (04) */	0 /*''*/, 0 /*''*/, 0 /*''*/, 0 /*''*/, 
	/* 008 (08) */	0 /*'backspace? '*/, 0 /*'\t'*/, 0 /*\n*/, 0 /*''*/, 
	/* 012 (0c) */	0 /*''*/, 0 /*'\r'*/, 0 /*''*/, 0 /*''*/, 
	/* 016 (10) */	0 /*''*/, 0 /*''*/, 0 /*''*/, 0 /*''*/, 
	/* 020 (14) */	0 /*''*/, 0 /*''*/, 0 /*''*/, 0 /*''*/, 
	/* 024 (18) */	0 /*''*/, 0 /*''*/, 0 /*''*/, 0 /*''*/, 
	/* 028 (1c) */	0 /*''*/, 0 /*''*/, 0 /*''*/, 0 /*''*/, 
	/* 032 (20) */	0 /*' '*/, 0 /*'!'*/, 0x03C2 /*'"'*/, 0 /*'#'*/, 
	/* 036 (24) */	0 /*'$'*/, 0 /*'%'*/, 0 /*'&'*/, 0x0342 /* circumflex perispomeni'''*/, 
	/* 040 (28) */	0 /*'('*/, 0 /*')'*/, 0 /*'*'*/, 0 /*'+'*/, 
	/* 044 (2c) */	0 /*','*/, 0 /*'-'*/, 0 /*'.'*/, 0 /*'/'*/, 
	/* 048 (30) */	0 /*'0'*/, 0 /*'1'*/, 0 /*'2'*/, 0 /*'3'*/, 
	/* 052 (34) */	0 /*'4'*/, 0 /*'5'*/, 0 /*'6'*/, 0 /*'7'*/, 
	/* 056 (38) */	0 /*'8'*/, 0 /*'9'*/, 0 /* middle dot ':'*/, 0 /* possibly a grave varia (see code below) */, 
	/* 060 (3c) */	0 /*'<'*/, 0 /*'='*/, 0x0308 /*'>'*/, 0 /*'?'*/, 
	/* 064 (40) */	0 /*'@'*/, 0x0391 /*'A'*/, 0x0392 /*'B'*/, 0x03A7 /*'C'*/, 
	/* 068 (44) */	0x0394 /*'D'*/, 0x0395 /*'E'*/, 0x03A6 /*'F'*/, 0x0393 /*'G'*/, 
	/* 072 (48) */	0x0397 /*'H'*/, 0x0399 /*'I'*/, 0x0314 /*'J'*/, 0x039a /*'K' bug in .py */, 
	/* 076 (4c) */	0x039b /*'L'*/, 0x039c /*'M'*/, 0x039d /*'N'*/, 0x039f /*'O'*/, 
	/* 080 (50) */	0x03a0 /*'P'*/, 0x0398 /*'Q'*/, 0x03a1 /*'R'*/, 0x03a3 /*'S'*/, 
	/* 084 (54) */	0x03a4 /*'T'*/, 0x03a5 /*'U'*/, 0x0384 /*'V'*/, 0x03a9 /*'W'*/, 
	/* 088 (58) */	0x039e /*'X'*/, 0x03a8 /*'Y'*/, 0 /*'Z'*/, 0 /*'['*/, 
	/* 092 (5c) */	0 /*'\'*/, 0 /*']'*/, 0 /*'^'*/, 0 /*'_'*/, 
	/* 096 (60) */	0 /*'`'*/, 0x03b1 /*'a'*/, 0x03b2 /*'b'*/, 0x03c7 /*'c'*/, 
	/* 100 (64) */	0x03b4 /*'d'*/, 0x03b5 /*'e'*/, 0x03c6 /*'f'*/, 0x03b3 /*'g'*/, 
	/* 104 (68) */	0x03b7 /*'h'*/, 0x03b9 /*'i'*/, 0x0313 /* smooth breathing 'j'*/, 0x03ba /*'k'*/, 
	/* 108 (6c) */	0x03bb /*'l'*/, 0x03bc /*'m'*/, 0x03bd /*'n'*/, 0x03bf /*'o'*/,            
	/* 112 (70) */	0x03c0 /*'p'*/, 0x03b8 /*'q'*/, 0x03c1 /*'r'*/, 0x03c3 /*'s'*/, 
	/* 116 (74) */	0x03c4 /*'t'*/, 0x03c5 /*'u'*/, 0x0301 /* acute tonos 'v'*/, 0x03c9 /*'w'*/, 
	/* 120 (78) */	0x03be /*'x'*/, 0x03c8 /*'y'*/, 0x03b6 /*'z'*/, 0 /*'{'*/, 
	/* 124 (7c) */	0 /*'|'*/, 0 /*'}'*/, 0x03c2 /*'~'*/, 0 /*''*/, 
	/* 128 (80) */	0 /*'�'*/, 0 /*'�'*/, 0, 0 /*'�'*/, 
	/* 132 (84) */	0 /*'�'*/, 0 /* blockend '�'*/, 0 /*elusionszeichen'�'*/, 0 /*'�'*/, 
	/* 136 (88) */	0 /*'�'*/, 0 /*; '�'*/, 0  /*'�'*/, 0 /*'�'*/, 
	/* 140 (8c) */	0,         0,           0 /*'�'*/, 0 /*'�'*/, 
	/* 144 (90) */	0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 
	/* 148 (94) */	0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 
	/* 152 (98) */	0 /*'�'*/, 0 /*'�'*/, 0 /* / '�'*/, 0 /*'�'*/, 
	/* 156 (9c) */	0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 
	/* 160 (a0) */	0,         0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 
	/* 164 (a4) */	0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0,            
	/* 168 (a8) */	0,         0,         0 /*'�'*/, 0,             
	/* 172 (ac) */	0,         0 /* G '�'*/, 0 /*'�'*/, 0 /*'�'*/, 
	/* 176 (b0) */	0 /*'�'*/, 0x0306 /*'�'*/, 'j',            0 /*unterpunkt'�'*/, 
	/* 180 (b4) */	0,            0,            0,            0,            
	/* 184 (b8) */	0 /*'�'*/, 0,            0 /*'�'*/, 0 /* [ '�'*/, 
	/* 188 (bc) */	0 /*]'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 
	/* 192 (c0) */	0 /*'�'*/, 0 /* degree '�'*/, 0 /* 0x006c */, 0 /*'�'*/, 
	/* 196 (c4) */	0,            0,            0/* apostrophe */, 0 /*'�'*/, 
	/* 200 (c8) */	0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 
	/* 204 (cc) */	0 /*'�'*/, 0 /*'�'*/, 0,            0,           
	/* 208 (d0) */	0x0304,    0 /* intf uses: 0x00af oberstrich'�'*/, 0 /*'�'*/, 0 /*'�'*/, 
	/* 212 (d4) */	0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0,             
	/* 216 (d8) */	0 /*'�'*/, 0 /*'�'*/, 0 /*|'�'*/, 0 /*`'�'*/, 
	/* 220 (dc) */	0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0x0302 /*'�'*/, 
	/* 224 (e0) */	0 /*'�'*/, 0 /*'�'*/, 0    /*'�'*/, 0,            
	/* 228 (e4) */	0 /*'�'*/, 0,         0 /*'�'*/, 0,            
	/* 232 (e8) */	0,         0,         0,         0,         
	/* 236 (ec) */	0,         0,         0 /*'�'*/, 0    /*'�'*/, 
	/* 240 (f0) */	0 /*'�'*/, 0,         0 /*'�'*/, 0 /*'�'*/, 
	/* 244 (f4) */	0 /*'�'*/, 0x03DB /*stigma*/, 0, 0 /*'�'*/, 
	/* 248 (f8) */	0,         0 /*'�'*/, 0,      0,            
	/* 252 (fc) */	0 /*'�'*/, 0 /*'�'*/, 0 /*'�'*/, 0 /*'?'*/
};

/* c equiv to put_file_str_in_hndl */
int readLongString(int fd, char **txt) {
	if (*txt) free(*txt);
        size_t n_writ;
	unsigned int size;
        n_writ = read(fd, &size, 4);
	size = ntohl(size);
        if (n_writ != 4) return -1;
	*txt = (char *)calloc(1, size+1);
        n_writ = read(fd, *txt, size);
        if (n_writ != size) return -1;
        (*txt)[size] = 0;
        return size;
}
 

const char *toUTF8(unsigned long unicode) {
	static char result[8];
	int x = 0;
	if (unicode < 0x80) {
		x = 1;
		result[0] = unicode;
	}
	else if (unicode < 0x800) {
		x = 2;
		result[0] = 0xc0 | (unicode >> 6);
	}
	else if (unicode < 0x10000) {
		x = 3;
		result[0] = 0xe0 | (unicode >> 12);
	}
	else if (unicode < 0x200000) {
		x = 4;
		result[0] = 0xf0 | (unicode >> 18);
	}
	if (x > 1) {
		int y, i=0;
		for (y = x; y > 1; y--) {
			result[++i] = 0x80 | (unicode & (0x3f << ((y-2)*6)));
		}
	}
	result[x] = 0;
	return result;
}


void symbolGreekToUTF8(char **txt) {
	bool nomSac = false;
	for (int i = 0; (*txt)[i]; ++i) if ((unsigned char)(*txt)[i] == 209) { nomSac = true; break; }
	
	char *buf = (char *)calloc(6, strlen(*txt)+1);
	for (int i = 0; (*txt)[i]; ++i) {
		if ((unsigned char)(*txt)[i] != 209) {
			strcat(buf, toUTF8(symbolGreekTranslation[(unsigned char)(*txt)[i]]));
			if (nomSac && isGreekLetter[(unsigned char)(*txt)[i]]) strcat(buf, toUTF8(0x0305));
		}
	}
	free(*txt);
	*txt = buf;
}


int main(int argc, char **argv)
{
	long ioError = 0;
 	int fd;
 	char *p = 0, *q = 0;
	char another;
	int count;
	char *sel_mss = 0, *sel_plcs = 0;
	short mss, blocks, places;
	if (argc < 2) {
		fd = dup(0);
/*
		fprintf(stderr, "usage: %s <reg_file_name>\n", *argv);
		exit(-1);
*/
	}
	else fd = open(argv[1], O_RDONLY);
	if (fd < 0) {
		fprintf(stderr, "file not found: %s\n", argv[1]);
		exit(-1);
	}

	printf("<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n");
	printf("<rules>\n");

	count = sizeof(char);
	ioError = (read(fd, &another, count) != count);
	while ((!ioError) && (another)) {
			readLongString(fd, &p);
			symbolGreekToUTF8(&p);
			readLongString(fd, &q);
			symbolGreekToUTF8(&q);
			count = sizeof(short);
			read(fd, &mss, count);
			mss = ntohs(mss);
			read(fd, &blocks, count);
			blocks = ntohs(blocks);
			read(fd, &places, count);
			places = ntohs(places);
			readLongString(fd, &sel_mss);
			symbolGreekToUTF8(&sel_mss);
			readLongString(fd, &sel_plcs);
			symbolGreekToUTF8(&sel_plcs);

			printf("\t<rule");
			printf(" mss=\"%d\"", mss);
			printf(" blocks=\"%d\"", blocks);
			printf(" places=\"%d\">\n", places);
			printf("\t\t<p>%s</p>\n", p);
			printf("\t\t<q>%s</q>\n", q);
			printf("\t\t<sel_mss>%s</sel_mss>\n", sel_mss);
			printf("\t\t<sel_plcs>%s</sel_plcs>\n", sel_plcs);
			printf("\t</rule>\n");

			count = sizeof(char);
			ioError = (read(fd, &another, count) != count);
	}

	printf("</rules>\n");

	if (p) free(p);
	if (q) free(q);
	if (sel_mss) free(sel_mss);
	if (sel_plcs) free(sel_plcs);
	close(fd);

	return 0;
}