modal/src/modal.c

#include <stdio.h>

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

static int dst;
static Rule rules[0x1000], lambda, *rules_ = rules;
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 <= ' ' || c == '(' || c == ')')

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

static int
set_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
		regs[r] = b;
	return 1;
}

static void
put_reg(char r)
{
	char c, *s = regs[(int)r];
	if(r == '~') {
		/* special stdin */
		while(fread(&c, 1, 1, stdin) && c >= ' ')
			*outp_++ = c;
	} else if(s) {
		char *ss = walk(s);
		if(r == '*') {
			/* special implode */
			if(*s == '(') {
				while(s < ss && (c = *s++))
					if(!spacer(c)) *outp_++ = c;
			}
			/* special explode */
			else {
				int i, depth = 0;
				while((c = *s++) && !spacer(c))
					*outp_++ = c, *outp_++ = ' ', *outp_++ = '(', depth++;
				for(i = 0; i < depth; i++)
					*outp_++ = ')';
			}
		} else if(r == ':') {
			/* special stdout */
			if(*s == '(') s++, --ss;
			while(s < ss) {
				c = *s++;
				if(c == '\\') {
					switch(*s++) {
					case 't': putc(0x09, stdout); break;
					case 'n': putc(0x0a, stdout); break;
					case 's': putc(0x20, stdout); break;
					}
				} else
					putc(c, stdout);
			}
		} else
			while(s < ss) *outp_++ = *s++;
	} else
		*outp_++ = '?', *outp_++ = r;
}

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

static int
commit_rule(Rule *r, char *s, int create)
{
	while((*outp_++ = *s++))
		;
	*outp_++ = 0;
	if((dst = !dst))
		prog_ = bank_b, outp_ = bank_a;
	else
		prog_ = bank_a, outp_ = bank_b;
	if(create)
		fprintf(stderr, "<> (%s) (%s)\n", r->a, r->b);
	else
		fprintf(stderr, "%02d %s\n", r->id, prog_);
	return 1;
}

static int
write_rule(Rule *r, char last, char *res)
{
	char c, *b = r->b, *origin = outp_;
	while((c = *b++))
		if(spacer(last) && c == '?')
			put_reg(*b++);
		else
			*outp_++ = c, last = c;
	if(last == ' ' && outp_ - origin == 0) outp_--;
	return commit_rule(r, res, 0);
}

static char *
parse_frag(char *s)
{
	char c, *ss;
	while((c = *s) && c <= ' ') s++;
	if(*s != ')' && *s != '<' && s[1] != '>') {
		ss = walk(s);
		if(*s == '(') s++, ss--;
		while(s < ss) *dict_++ = *s++;
		s++;
	}
	*dict_++ = 0;
	return s;
}

static char *
create_rule(Rule *r, int id, char *s)
{
	r->id = id, s += 2;
	r->a = dict_, s = parse_frag(s);
	r->b = dict_, s = parse_frag(s);
	return s;
}

static int
rewrite(void)
{
	char c, last = 0, *cap, *s = dst ? bank_b : bank_a, *res;
	while((c = *s) && c <= ' ') s++;
	while((c = *s)) {
		if(spacer(last)) {
			Rule *r;
			if(*s == '<' && s[1] == '>') {
				r = rules_++;
				s = create_rule(r, rules_ - rules - 1, s);
				while((c = *s) && c <= ' ') s++;
				return commit_rule(r, s, 1);
			}
			if(*s == '?' && s[1] == '(') {
				r = &lambda, cap = walk(s + 1);
				create_rule(&lambda, -1, s), s = cap;
				while((c = *s) && c <= ' ') s++;
				if((res = match_rule(&lambda, s)) != NULL)
					return write_rule(&lambda, last, res);
			}
			for(r = rules; r < rules_; r++)
				if((res = match_rule(r, s)) != NULL)
					return write_rule(r, last, res);
		}
		*outp_++ = last = c;
		s++;
	}
	*outp_++ = 0;
	return 0;
}

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 Interpreter, 12 Apr 2024.\n");
	if(!(f = fopen(argv[1], "r")))
		return !printf("Invalid Modal file: %s.\n", argv[1]);
	while(fread(&c, 1, 1, f)) {
		c = c <= 0x20 ? 0x20 : c;
		if(w > bank_a) {
			if(c == ' ' && *(w - 1) == '(') continue;
			if(c == ')' && *(w - 1) == ' ') w--;
			if(c == ' ' && *(w - 1) == ' ') w--;
		}
		*w++ = c;
	}
	while(*(--w) <= ' ') *w = 0;
	fclose(f);
	while(rewrite())
		;
	return 0;
}