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)
{
	int depth = 0;
	char c;
	while((c = *s++)) {
		if(spacer(c) && !depth) break;
		if(c == '(') depth++;
		if(c == ')') --depth;
	}
	return s - 1;
}

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(c != *b) return NULL;
		a++, b++;
	}
	return b;
}

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

static void
save(void)
{
	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(".. %s\n", prog);
}

static int
rewrite(void)
{
	int success = 0;
	char c, *p = prog;
	while((c = *p)) {
		int i, found = 0;
		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;
				found = success = 1;
				p = res;
			}
		}
		if(!found) {
			*outp_++ = c;
			p++;
		}
	}
	*outp_++ = 0;
	if(success)
		save();
	return success;
}

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

static char *
parse_rulefrag(FILE *f)
{
	int depth = 0;
	char c, *origin = dict_;
	while(f && fread(&c, 1, 1, f) && c && c != 0xa) {
		if(c == '(') depth++;
		if(c == ')') --depth;
		if(c == ' ' && !depth) break;
		*dict_++ = c;
	}
	*dict_++ = 0;
	return origin;
}

static void
tokenize(char *t, FILE *f)
{
	char c;
	if(!t[0]) return;
	if(t[0] == '<' && t[1] == '>') {
		Rule *r = &rules[rules_len++];
		r->a = parse_rulefrag(f), r->b = parse_rulefrag(f);
		return;
	}
	while((c = *t++))
		*prog_++ = c;
	*prog_++ = ' ';
	while(f && fread(&c, 1, 1, f) && c)
		*prog_++ = c == 0xa ? ' ' : c;
}

static int
parse(char *path)
{
	FILE *f;
	char c, token[0x40], *tokptr;
	if(!(f = fopen(path, "r")))
		return !printf("Invalid file: %s\n", path);
	tokptr = token;
	while(f && fread(&c, 1, 1, f)) {
		if(c < 0x21)
			*tokptr++ = 0x00, tokenize(token, f), tokptr = token;
		else if(tokptr - token < 0x3f)
			*tokptr++ = c;
		else
			return printf("Token too long: %s\n", token);
	}
	*tokptr++ = 0x00, tokenize(token, f), tokptr = token;
	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;
}