modal/src/modal.c

#include <stdio.h>

typedef struct {
	char *a, *b;
} Rule;

static int rules_len;
static Rule rules[0x100];
static char dict[0x8000], *dict_ = dict;
static char prog[0x1000], *prog_ = prog;
static char outp[0x1000], *outp_ = outp;
static char *regs[0x100];

#define spacer(c) (c == ' ' || c == '(' || c == ')')

static char *
walk(char *s)
{
	char c;
	int depth = 0;
	if(s[0] == '(') {
		while((c = *s++)) {
			if(c == '(') depth++;
			if(c == ')') --depth;
			if(!depth) return s;
		}
	}
	while(!spacer(s[0]) && (c = *s++))
		;
	return s;
}

static char *
match(char *p, Rule *r)
{
	char c, *a = r->a, *b = p;
	while((c = *a)) {
		if(c == '?') regs[(int)*(++a)] = b, a++, b = walk(b), c = *b;
		if(!a[0]) return b;
		if(c != *b) return NULL;
		a++, b++;
	}
	return b;
}

static void
writereg(char r)
{
	int depth = 0;
	char c, *s = regs[(int)r];
	if(s[0] == '(') {
		while((c = *s++)) {
			if(c == '(') depth++;
			*outp_++ = c;
			if(c == ')') --depth;
			if(!depth) return;
		}
	}
	while(!spacer(s[0]) && (*outp_++ = *s++))
		;
	return;
}

static void
save(int rule)
{
	int i, end = outp_ - outp;
	/* todo: change pointer instead of copying memory */
	for(i = 0; i <= end; i++) prog[i] = outp[i];
	prog_ = prog, outp_ = outp;
	printf("..%02d %s\n", rule, prog);
}

static int
rewrite(void)
{
	char c, *p = prog;
	while((c = *p)) {
		int i;
		for(i = 0; i < rules_len; i++) {
			Rule *r = &rules[i];
			char *res = match(p, r);
			if(res != NULL) {
				char cc, *b = r->b;
				while((cc = *b++)) {
					if(cc == '?')
						writereg(*b++);
					else
						*outp_++ = cc;
				}
				while((*outp_++ = *res++))
					;
				*outp_++ = 0;
				save(i);
				return 1;
			}
		}
		*outp_++ = c;
		p++;
	}
	*outp_++ = 0;
	return 0;
}

static void
print_rules(void)
{
	int i;
	for(i = 0; i < rules_len; i++)
		printf("Rule #%d: <%s> -> <%s>\n", i, rules[i].a, rules[i].b);
	printf("\n");
}

static char *
parse_rulefrag(char *line)
{
	int depth = 0;
	char c, *s = line, *res = dict_;
	if(s[0] == '(') {
		while((c = *s++)) {
			if(c == '(') {
				depth++;
				if(depth == 1) continue;
			}
			if(c == ')') {
				--depth;
				if(!depth) {
					*dict_++ = 0;
					return res;
				}
			}
			*dict_++ = c;
		}
	}
	while(!spacer(s[0]) && (*dict_++ = *s++))
		;
	*dict_++ = 0;
	return res;
}

static int
parse_line(char *line)
{
	if(line[0] == 0) return 1;
	if(line[0] == '<' && line[1] == '>') {
		Rule *r = &rules[rules_len++];
		line += 3;
		r->a = parse_rulefrag(line);
		line = walk(line) + 1;
		r->b = parse_rulefrag(line);
		return 1;
	}
	while((*prog_++ = *line++))
		;
	*prog_++ = 0xa;
	return 1;
}

static int
parse(char *path)
{
	FILE *f;
	char c, line[0x400], *line_ = line;
	if(!(f = fopen(path, "r")))
		return !printf("Invalid file: %s\n", path);
	while(f && fread(&c, 1, 1, f)) {
		if(c == 0xa)
			*line_++ = 0x00, parse_line(line), line_ = line;
		else if(line_ - line < 0x400)
			*line_++ = c;
	}
	*line_++ = 0x00, parse_line(line), line_ = line;
	fclose(f);
	return 1;
}

int
main(int argc, char **argv)
{
	if(argc < 2)
		return !printf("usage: modal [-v] source.modal\n");
	if(argc < 3 && argv[1][0] == '-' && argv[1][1] == 'v')
		return !printf("Modal - Modal Interpreter, 3 Apr 2024.\n");
	parse(argv[1]);
	print_rules();
	printf(".... %s\n", prog);
	while(rewrite())
		;
	return 0;
}