modal/src/modal.c

#include <stdio.h>

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

static int rules_len, direction;
static Rule rules[0x100];
static char dict[0x8000], *dict_ = dict;
static char bank_a[0x4000], *prog_ = bank_a;
static char bank_b[0x4000], *outp_ = bank_b;
static char *regs[0x100];

#define spacer(c) (c < 0x21 || 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]) && *s++)
		;
	return s;
}

static int
bind(int id, char *b)
{
	if(regs[id]) {
		char *a = regs[id], *aa = walk(a), *bb = walk(b);
		while(a < aa && b < bb)
			if(*a++ != *b++) return 0;
	} else if(id == ':') {
		char *bb = walk(b);
		if(*b == '(') b++, --bb;
		while(b < bb) putc(*(b++), stdout);
	} else
		regs[id] = b;
	return 1;
}

static char *
match(char *p, Rule *r)
{
	int i;
	char c, *a = r->a, *b = p;
	for(i = 0x21; i < 0x7f; i++)
		if((char)i != ':')
			regs[i] = 0;
	while((c = *a)) {
		if(c == '?') {
			if(!bind(*(++a), b)) return NULL;
			a++, b = walk(b);
		}
		if(!*a && spacer(*b)) return b;
		if(*a != *b) return NULL;
		a++, b++;
	}
	return spacer(*b) ? b : NULL;
}

static char *
plode(char *s)
{
	int i, depth = 0;
	char c;
	if(s[0] == '(') { /* implode */
		while((c = *s++)) {
			if(c == '(') depth++;
			if(!spacer(c)) *outp_++ = c;
			if(c == ')') --depth;
			if(!depth) return s;
		}
	} else { /* explode */
		*outp_++ = *s++, *outp_++ = ' ';
		while((c = *s++) && !spacer(c)) {
			*outp_++ = '(', *outp_++ = c, depth++;
			if(!spacer(*s))
				*outp_++ = ' ';
		}
		for(i = 0; i < depth; i++)
			*outp_++ = ')';
	}
	return s;
}

static int
commit(char r)
{
	int depth = 0;
	char c, *s = regs[(int)r];
	if(r == '*')
		s = plode(s);
	else if(s[0] == '(') {
		while((c = *s++)) {
			if(c == '(') depth++;
			*outp_++ = c;
			if(c == ')') --depth;
			if(!depth) return 1;
		}
	}
	while(!spacer(s[0]) && (*outp_++ = *s++))
		;
	return 1;
}

static int
save(int rule, char *s)
{
	while((*outp_++ = *s++))
		;
	*outp_++ = 0;
	if((direction = !direction))
		prog_ = bank_b, outp_ = bank_a;
	else
		prog_ = bank_a, outp_ = bank_b;
	if(rule >= 0)
		fprintf(stderr, "%02d %s\n", rule, prog_);
	return 1;
}

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
rewrite(void)
{
	char c, *p = direction ? bank_b : bank_a, *o = p;
	while((c = *p) && c <= ' ') p++;
	while((c = *p)) {
		int i;
		if(p[0] == '<' && p[1] == '>') {
			Rule *r = &rules[rules_len++];
			p += 3;
			r->a = parse_rulefrag(p), p = walk(p) + 1;
			r->b = parse_rulefrag(p), p = walk(p);
			return save(-1, p);
		}
		if(p == o || spacer(*(p - 1))) {
			for(i = 0; i < rules_len; i++) {
				Rule *r = &rules[i];
				char *res = match(p, r);
				if(res != NULL) {
					char cc, *b = r->b;
					if(!*b && outp_ != bank_a && outp_ != bank_b) outp_--;
					while((cc = *b++)) {
						if(cc == '?' && regs[(int)b[0]])
							commit(*b++);
						else
							*outp_++ = cc;
					}
					return save(i, res);
				}
			}
		}
		*outp_++ = c;
		p++;
	}
	*outp_++ = 0;
	return 0;
}

static void
print_rules(void)
{
	int i;
	fprintf(stderr, "\n");
	for(i = 0; i < rules_len; i++)
		fprintf(stderr, "<> (%s) (%s)\n", rules[i].a, rules[i].b);
	fprintf(stderr, "\n");
}

int
main(int argc, char **argv)
{
	FILE *f;
	char c, *w = bank_a;
	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, 7 Apr 2024.\n");
	if(!(f = fopen(argv[1], "r")))
		return !printf("Invalid Modal file: %s.\n", argv[1]);
	while(fread(&c, 1, 1, f)) {
		if(w > bank_a) {
			if(c == ' ' && *(w - 1) == '(') continue;
			if(c == ')' && *(w - 1) == ' ') w--;
			if(c == ' ' && *(w - 1) == ' ') w--;
		}
		*w++ = c;
	}
	*w++ = 0;
	fclose(f);
	regs[':'] = argv[2];
	while(rewrite())
		;
	print_rules();
	return 0;
}