modal/src/modal.c

#include <stdio.h>

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

static int flip, quiet, rmin = 0xff, rmax = 0x00, cycles = 0x10000;
static Rule rules[0x1000], *rules_ = rules + 1;
static char dict[0x8000], *dict_ = dict;
static char bank_a[0x4000], *src_ = bank_a;
static char bank_b[0x4000], *dst_ = 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 void
write_reg(char r, char *reg)
{
	char c, *cap = walk(reg);
	if(r == '*') {
		int i, depth = 0;
		if(*reg == '(') { /* special explode tuple */
			reg++;
			while(reg < cap) {
				while((c = *reg) && !spacer(c))
					*dst_++ = c, reg++;
				*dst_++ = ' ';
				*dst_++ = '(', reg++, depth++;
			}
		} else { /* special explode token */
			while((c = *reg++) && !spacer(c))
				*dst_++ = c, *dst_++ = ' ', *dst_++ = '(', depth++;
		}
		for(i = 0; i < depth; i++)
			*dst_++ = ')';
	} else if(r == '.') { /* special unpack */
		if(*reg == '(') reg++, --cap;
		while(reg < cap) *dst_++ = *reg++;
	} else if(r == '^') { /* special join */
		if(*reg == '(') reg++, --cap;
		while(reg < cap && (c = *reg++))
			if(!spacer(c)) *dst_++ = c;
	} else if(r == '~') { /* special stdin */
		while(fread(&c, 1, 1, stdin) && c >= ' ')
			*dst_++ = c;
	} else if(r == ':') { /* special stdout */
		if(*reg == '(') reg++, --cap;
		while(reg < cap) {
			c = *reg++;
			if(c == '\\') {
				switch(*reg++) {
				case 't': putc(0x09, stdout); break;
				case 'n': putc(0x0a, stdout); break;
				case 's': putc(0x20, stdout); break;
				}
			} else
				putc(c, stdout);
		}
	} else
		while(reg < cap) *dst_++ = *reg++;
}

static int
write_rule(Rule *r, char *s, int create)
{
	while((*dst_++ = *s++))
		;
	*dst_++ = 0;
	if((flip = !flip))
		src_ = bank_b, dst_ = bank_a;
	else
		src_ = bank_a, dst_ = bank_b;
	if(!quiet) {
		if(create)
			fprintf(stderr, "<> (%s) (%s)\n", r->a, r->b);
		else
			fprintf(stderr, "%02d %s\n", r->id, src_), ++r->refs;
	}
	return 1;
}

static int
run_rule(Rule *r, char *s)
{
	int i;
	char c, *a = r->a, *p = s;
	if(rmax) {
		for(i = rmin; i <= rmax; i++)
			regs[i] = 0;
		rmin = 0xff, rmax = 0x00;
	}
	while((c = *a)) {
		if(c == '?') {
			int regid = (int)*(++a);
			char *pcap = walk(p), *reg = regs[regid];
			if(reg) { /* reg cmp */
				char *rcap = walk(reg), *pp = p;
				while(reg < rcap || pp < pcap)
					if(*reg++ != *pp++) return 0;
			} else { /* reg set */
				regs[regid] = p;
				if(regid < rmin) rmin = regid;
				if(regid > rmax) rmax = regid;
			}
			a++, p = pcap;
			if(!spacer(*a))
				while((c = *a) && !spacer(c)) a++;
			continue;
		}
		if(c != *p) return 0;
		a++, p++;
	}
	c = *p;
	if(spacer(c)) {
		char *b = r->b, *reg, *origin = dst_;
		while((c = *b++))
			if(c == '?' && (reg = regs[(int)*b]))
				write_reg(*b++, reg);
			else
				*dst_++ = c;
		if(dst_ == origin) {
			while(*p == ' ') p++;
			if(*p == ')' && *(dst_ - 1) == ' ') dst_--;
		}
		return write_rule(r, p, 0);
	}
	return 0;
}

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

static int
rewrite(void)
{
	char c, last = 0, *cap, *s = src_;
	while(*s == ' ') s++;
	while((c = *s)) {
		if(spacer(last)) {
			Rule *r;
			if(c == '<' && s[1] == '>') { /* rule */
				r = rules_++, r->id = rules_ - rules - 1;
				r->a = dict_, s = parse_frag(s + 2);
				r->b = dict_, s = parse_frag(s);
				while(*s == ' ') s++;
				return write_rule(r, s, 1);
			}
			if(c == '?' && s[1] == '(') { /* lambda */
				cap = walk(s + 1);
				r = &rules[0], r->id = -1;
				r->a = dict_, s = parse_frag(s + 2);
				r->b = dict_, parse_frag(s), s = cap;
				while(*s == ' ') s++;
				return run_rule(r, s);
			}

			for(r = rules + 1; r < rules_; r++)
				if(run_rule(r, s)) return 1;
		}
		*dst_++ = last = c;
		s++;
	}
	*dst_++ = 0;
	return 0;
}

int
main(int argc, char **argv)
{
	FILE *f;
	int i, pl = 0, pr = 0;
	char c, *w = bank_a;
	if(argc < 2)
		return !printf("usage: modal [-vqn] source.modal\n");
	for(i = 1; i < argc && *argv[i] == '-'; i++) {
		switch(argv[i][1]) {
		case 'v': /* version */ return !printf("Modal Interpreter, 22 Apr 2024.\n");
		case 'q': /* quiet */ quiet = 1; break;
		case 'n': /* infinite */ cycles = 0xffffffff; break;
		}
	}
	if(!(f = fopen(argv[i], "r")))
		return !fprintf(stdout, "Modal file invalid: %s.\n", argv[i]);
	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--;
			if(c == '(') pl++;
			if(c == ')') pr++;
		}
		*w++ = c;
	}
	while(*(--w) <= ' ') *w = 0;
	fclose(f);
	if(pr != pl)
		return !fprintf(stdout, "Modal program imbalanced.\n");
	while(rewrite())
		if(!cycles--) return !fprintf(stdout, "Modal rewrites exceeded.\n");
	while(rules_-- > rules && !quiet)
		if(!rules_->refs) printf("-- Unused rule: %d <> (%s) (%s)\n", rules_->refs, rules_->a, rules_->b);
	return 0;
}