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((c = *s) && !spacer(c) && *s++)
		;
	return s;
}

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++, c = *s;
			if(!spacer(c))
				*outp_++ = ' ';
		}
		for(i = 0; i < depth; i++)
			*outp_++ = ')';
	}
	return s;
}

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

static int
write_reg(char r)
{
	char *s = regs[(int)r];
	if(r == '*')
		s = plode(s);
	else if(s) {
		char *ss = walk(s);
		while((s < ss) && (*outp_++ = *s++))
			;
	} else
		*outp_++ = r;
	return 1;
}

static char *
match_rule(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_reg(*(++a), b)) return NULL;
			a++, b = walk(b);
			continue;
		}
		if(*a != *b) return NULL;
		a++, b++;
	}
	c = *b;
	return spacer(c) ? b : NULL;
}

static int
commit_rule(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_rule(char *s)
{
	char *ss = walk(s), *d = dict_;
	if(*s == '(') s++, ss--;
	while((s < ss) && (*dict_++ = *s++))
		;
	*dict_++ = 0;
	return d;
}

static int
add_rule(char *p)
{
	char c;
	Rule *r = &rules[rules_len++];
	p += 2;
	while((c = *p) && c <= ' ') p++;
	r->a = parse_rule(p), p = walk(p);
	while((c = *p) && c <= ' ') p++;
	r->b = parse_rule(p), p = walk(p);
	return commit_rule(-1, p);
}

static int
rewrite(void)
{
	char c, c_ = 0, last = 0, *p = direction ? bank_b : bank_a, *o = p;
	while((c = *p) && c <= ' ') p++;
	while((c = *p)) {
		int i;
		if(p[0] == '<' && p[1] == '>') return add_rule(p);
		if(p != o) c_ = *(p - 1);
		if(p == o || spacer(c_)) {
			for(i = 0; i < rules_len; i++) {
				Rule *r = &rules[i];
				char *res = match_rule(p, r);
				if(res != NULL) {
					char cc, *b = r->b;
					if(!*b && last == ' ') outp_--;
					while((cc = *b++)) {
						if(cc == '?')
							write_reg(*b++);
						else
							*outp_++ = cc;
					}
					return commit_rule(i, res);
				}
			}
		}
		*outp_++ = last = 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, 8 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;
}