int to char

firstly, I have no idea now that I look at ti again, how I managed to get away with that.
I think I'll post a couple notes on what I noticed I did wrong, and then post a better version, with comments included (if you also want it in an email just ask, it might be a bit harder to read on here)
ok here goes for the errors I made yesterday(at least the ones I noticed):
first of all, adding the char at the beggining of buf was pointess since I only added a '-' to the beggining of b, never buf.
secondly, I was using j to keep track of which place I was in.. but how did that work?? j is only an int, so it's surprising i didn't get errors - i suppose I should have made sure the function could handle even large ints properly.
thirdly, I was putting a null terminator on the char *after* the end of b
and another thing - after using the first char of buf for something else, i never made buf long enough, so it was actually only using sizeint-1 chars, which means it couldn't handle a large int anyway.
Well, I don't know how that slipped past me, but here's a hopefully better solution:

Code:

/i is the int we want to convert, b is the string we'll put it into,
//and sizeb is the length of string b
//(actually since the string starts at 0, sizeb should be
//the last element of b
void myitoa(int i, char* b, int sizeb)
{
  //if the string doesn't point to anything,
  //or if it's 0 length, return
	if(!b || !sizeb) return;
	//j is for the loops, r is the temp variable for getting the value out of i
	int j=0, r=i;
	//this tells us how many bits an int has (assuming a byte has 8 bits.. I *think* this is safe? I may be wrong tho)
	int szint = sizeof(int)*8;
	//buf is the string we'll use to work on, s will point to the current spot we're working on
	char* buf=0;
	char* s=0;
	//allocate enough memory to handle the size of an int.. + 1 for the /0 at the end
	try {
		buf = new char[szint+1];
	} catch(...) {
		cout << "error allocating memory\n";
		buf=0;
	}
	//this way if there was an error we won't go trying to write an invalid pointer
	if(buf)
	{
		int c;
		//initialize the string to be all '0's and slap a null terminator on it
		for(j=0;j<szint;j++) *(buf+j) = '0';
		*(buf+szint) = '\0';
		//now if i is a negative number, reverse the sign on r
		//and then just treat it as a normal positive number
		//this was a hack cause I totally forgot to include support for
		//negative numbers untill like 2 minutes before I was going to post the code ;)
		if(i < 0)
		{
			r = (-(i));
		}
		//have no idea how the old way wokred at all..
		//s starts at the end of buf and works backwards
		s=buf+szint;
		//until it hits the beggining
		while(s >= buf)
		{
		  //make the char at s's current address = '0' + whatever is in the current place;
		  //if r == 756 for instance, r%756 == 6; and '0' + 6 = '6'
		  *s = (char)('0' + r%10);
		  //then switch to the next place by dividing by ten
		  r /= 10;
		  //and switch to the next(or rather previous ^.^) char by decreasing s
		  s--;
		}
		s=buf; c=0;
		//remove the leading 0s since they look stupid -
		//what I'm actually doing is just having s point to the first character that is not 0 -
		//unless it's the only one in the string, in which case it will just point to the last one
		while(*s == '0' && c++ < szint) s++;
		//now copy it into the string we were passed, making sure not to write too far
		strncpy(b, s, sizeb);
		//and, if i was a negative number, shift the characters in b over to the right and make the first one a '-'
		//(might be a more efficient way of doing this but like I said, I just hacked it in two minutes before posting :P
		//and it actually works too hehe)
		if(i<0)
		{
			for(j=sizeb;j>0;j--) *(b+j) = *(b+j-1);
			*b = '-';
		}
		//now if the b string is long enough to hold our whole string, cap it at the end of our string,
		//so we won't have gibberish after the number
		if(c < sizeb) *(b+c) = '\0';
	}
	//and also be sure to cap the end of b once we're done with it, even if it doesn't hold our whole string.
	*(b+sizeb) = '\0';
}

edit {
re:Hammer
quite correct, it was, which is part of the reason why I'm not sure how I got away with it ^.^
}

edit {
re:Hammer
quite correct, it was, which is part of the reason why I'm not sure how I got away with it ^.^
}

... you didn't, in a way: it was spotted by another programmer (ie me). The computer running the program will show this type of problem in undefined ways, maybe not even at all. That's the problem with overflows, you never know when they're going to crash your program.

quite true
in fact if I had tested the code a little better I probably would have seen it sooner - which just goes to show that even when coding just a quick little bit of code you should make sure to check it out properly

finally made one myself:

very similar to yours cobra..
(I know.. didnt make comments either..)

Code:

char* itoa(int integer, char* abc)
{
      int t_int = integer;
      char* buffer = abc;
      while(t_int)	  
      {
	  *buffer++;
	  t_int/= 10;
      }
      t_int = integer;
      if(integer<0)
      {
	  *buffer++;
	   t_int = (-(integer));
      }
	  *buffer = '\0';	  
	  do {
	        *--buffer = ((char)('0'+t_int%10));
	        t_int/= 10;
	  } while(t_int);
	   if(integer<0) *--buffer = '-';
      return buffer;
}

Luigi