Function to Split a String into Letters and Digits in C - arrays

Question

I'm pretty new to C, and I'm trying to write a function that takes a user input RAM size in B, kB, mB, or gB, and determines the address length. My test program is as follows:
int bitLength(char input[6]) {
char nums[4];
char letters[2];
for(int i = 0; i < (strlen(input)-1); i++){
if(isdigit(input[i])){
memmove(&nums[i], &input[i], 1);
} else {
//memmove(&letters[i], &input[i], 1);
}
}
int numsInt = atoi(nums);
int numExponent = log10(numsInt)/log10(2);
printf("%s\n", nums);
printf("%s\n", letters);
printf("%d", numExponent);
return numExponent;
}
This works correctly as it is, but only because I have that one line commented out. When I try to alter the 'letters' character array with that line, it changes the 'nums' character array to '5m2'
My string input is '512mB'
I need the letters to be able to tell if the user input is in B, kB, mB, or gB.
I am confused as to why the commented out line alters the 'nums' array.
Thank you.

Answer 1

In your input 512mB, "mB" is not digit and is supposed to handled in commented code. When handling those characters, i is 3 and 4. But because length of letters is only 2, when you execute memmove(&letters[i], &input[i], 1);, letters[i] access out of bounds of array so it does undefined behaviour - in this case, writing to memory of nums array.
To fix it, you have to keep unique index for letters. Or better, for both nums and letters since i is index of input.

Answer 2

There are several problems in your code. #MarkSolus have already pointed out that you access letters out-of-bounds because you are using i as index and i can be more than 1 when you do the memmove.
In this answer I'll address some of the other poroblems.
string size and termination
Strings in C needs a zero-termination. Therefore arrays must be 1 larger than the string you expect to store in the array. So
char nums[4]; // Can only hold a 3 char string
char letters[2]; // Can only hold a 1 char string
Most likely you want to increase both arrays by 1.
Further, your code never adds the zero-termination. So your strings are invalid.
You need code like:
nums[some_index] = '\0'; // Add zero-termination
Alternatively you can start by initializing the whole array to zero. Like:
char nums[5] = {0};
char letters[3] = {0};
Missing bounds checks
Your loop is a for-loop using strlen as stop-condition. Now what would happen if I gave the input "123456789BBBBBBBB" ? Well, the loop would go on and i would increment to values ..., 5, 6, 7, ... Then you would index the arrays with a value bigger than the array size, i.e. out-of-bounds access (which is real bad).
You need to make sure you never access the array out-of-bounds.
No format check
Now what if I gave an input without any digits, e.g. "HelloWorld" ? In this case nothin would be written to nums so it will be uninitialized when used in atoi(nums). Again - real bad.
Further, there should be a check to make sure that the non-digit input is one of B, kB, mB, or gB.
Performance
This is not that important but... using memmove for copy of a single character is slow. Just assign directly.
memmove(&nums[i], &input[i], 1); ---> nums[i] = input[i];
How to fix
There are many, many different ways to fix the code. Below is a simple solution. It's not the best way but it's done like this to keep the code simple:
#define DIGIT_LEN 4
#define FORMAT_LEN 2
int bitLength(char *input)
{
char nums[DIGIT_LEN + 1] = {0}; // Max allowed number is 9999
char letters[FORMAT_LEN + 1] = {0}; // Allow at max two non-digit chars
if (input == NULL) exit(1); // error - illegal input
if (!isdigit(input[0])) exit(1); // error - input must start with a digit
// parse digits (at max 4 digits)
int i = 0;
while(i < DIGITS && isdigit(input[i]))
{
nums[i] = input[i];
++i;
}
// parse memory format, i.e. rest of strin must be of of B, kB, mB, gB
if ((strcmp(&input[i], "B") != 0) &&
(strcmp(&input[i], "kB") != 0) &&
(strcmp(&input[i], "mB") != 0) &&
(strcmp(&input[i], "gB") != 0))
{
// error - illegal input
exit(1);
}
strcpy(letters, &input[i]);
// Now nums and letter are ready for further processing
...
...
}
}