Thread: Manually reading a character literal

So I've started with the easiest to deal with annoyance, escape characters, this is the function I got so far:

Code:

int rdEscChr( int_c_voidp_t getchr, void *source, int c, char32_t *c32 ) {
	uint_least64_t num = 0;
	char text[5] = {0};
	str_t str = {0};
	if ( c !== '\\' ) {
		c = -1;
		goto rdEscChr_done;
	}
	str.txt = text;
	str.cap = 5;
	switch ( (c = getchr(source)) ) {
	case '0':
		c = getchr(source);
		if ( !isdigit(c) )
			break;
		c = rdU64_base62( getchr, source, c, 8, 0, &num );
		goto rdEscChr_done;
	case 'b': num = '\b'; break;
	case 'f': num = '\f': break;
	case 'n': num = '\n'; break;
	case 'r': num = '\r'; break;
	case 't': num = '\t'; break;
	case 'u':
		text[0] = getchr(source);
		text[1] = getchr(source);
		text[2] = getchr(source);
		text[3] = getchr(source);
		str.len = 4;
		(void)rdU64_base62( (int_c_voidp_t)sgetc, &str, sgetc(&str), 16, 0, &num );
		break;
	case 'v': num = '\v'; break;
	case 'x':
		text[0] = getchr(source);
		text[1] = getchr(source);
		str.len = 3;
		(void)rdU64_base62( (int_c_voidp_t)sgetc, &str, sgetc(&str), 16, 0, &num );
		break;
	// Escape characters \ ' " and ^ do not need to be explicitly defined here
	default: num = c;
	}
	c = getchr(source);
	rdEscChr_done:
	if ( c32 ) *c32 = (char32_t)num;
	return c;
}

If you want the details of rdU64_base62, str_t and sgetc then refer back to a previous thread of mine: Manually reading an integer literal
I'll worry about making those C89 compatible later if I decide to put them in a library (which I am considering atm).

Anyways does anyone see any potential problems or missing stuff?

Edit: Within a minute or 2 of posting this I noticed a bunch of problems myself and fixed them, I've now replaced the above code with that.

You may want to add '\e', it's not exactly standard but it's widely used for ANSI escape sequences.

Originally Posted by awsdert

Within a minute or 2 of posting this I noticed a bunch of problems myself and fixed them, I've now replaced the above code with that.

Write unit tests?

Originally Posted by awsdert

I'll be doing that after I made sure nothing is missing like the below

You should write the tests before making further changes, to help you think through what might be missing, and detect any regressions due to your changes.

Originally Posted by awsdert

Well I don't expect there to be regressions on a 1st test nor further on given that this is a self contained function that should never need to be modified in the C environment

That implies that you're saying your code is perfect so you're not going to change it any more... but then you're asking for people to suggest improvements. I mean, whatever. If you don't want to do something that is beneficial to you, that's your own problem, and so you will continue to find yourself in situations like this: "Anyways does anyone see any potential problems or missing stuff?

Edit: Within a minute or 2 of posting this I noticed a bunch of problems myself"

Originally Posted by Hodor

Is it safe assigning -1 to 'num' (or c32 for that matter) considering num is unsigned? Other than that I guess the function is perfect. Except, why is 'source' a void*?

again Hodor the getchr() and source variable are abstracted to indicate the function doesn't care how the characters are provided only that they are provided, as for that -1 I'll have a look what for what you mean 1st, in mean time I'll post my updated function:

Code:

int ishex( int c ) {
	return ((c >= '0' && c <='9')
				|| (c >= 'A' && c <= 'F')
				|| (c >= 'a' && c <= 'f'));
}

int rdEscChr( int_c_voidp_t getchr, void *source, int c, char32_t *c32 ) {
	uint_least64_t num = 0;
	char *mbc = (char*)c32, *var;
	int i, m;
	if ( c != '\\' ) {
		c = -1;
		goto rdEscChr_done;
	}
	switch ( (m = getchr(source)) ) {
	case 'a': num = '\a'; break;
	case 'b': num = '\b'; break;
	case 'e': num = '\e'; break;
	case 'f': num = '\f'; break;
	case 'n': num = '\n'; break;
	case 'r': num = '\r'; break;
	case 't': num = '\t'; break;
	case 'u': c = rdU64_base62( getchr, source, getchr(source), 4, 16, 0, &num ); goto rdEscChr_done;
	case 'U': c = rdU64_base62( getchr, source, getchr(source), 8, 16, 0, &num ); goto rdEscChr_done;
	case 'v': num = '\v'; break;
	case 'x': c = rdU64_base62( getchr, source, getchr(source), 4, 16, 0, &num ); goto rdEscChr_done;
	// Escape characters \ ' " and ^ do not need to be explicitly defined here
	default:
		if ( isdigit(m) ) {
			c = rdU64_base62( getchr, source, m, 3, 8, 0, &num );
			goto rdEscChr_done;
		}
		else num = m;
	}
	c = getchr(source);
	rdEscChr_done:
	if ( c32 ) {
		var = getenv("MBCHAR_DIR");
		if ( var && var[0] == '>') {
			for ( i = 0; i < 4; ++i ) {
				if ( mbc[i] == 0 ) {
					break;
				}
			}
			if ( i == 4 ) {
				*c32 = num;
				return -1;
			}
			*c32 |= (num << (i * 8));
		}
		else *c32 = (((*c32) << 8) | (char32_t)num);
	}
	return c;
}

In the middle of manually testing stuff, most of escaped characters work fine, I think I will be adding a new parameter to rdU64_base62 though to limit the number of characters read (or set infinite if the parameter is 0)

Edit: made some small changes, updated the above to reflect