Thread: how to read a digit of a floating point number?????

Few ways of doing it:

1. 1. I think you can call floor() on the float, and then cast it to an int. This should take something like 34.54 and convert it to 34.
   2. Get the remainder of the int divided by 10 (Hint: % operator) 34 / 10 = 3, R = 4
   3. Print.
      
   
   
2. 1. Read in float as a string with fgets().
   2. Find the '.' char.
   3. If it exists, print the char before it.
   4. If it doesn't exist, print the last char of the string.
      

I think you're right, but for whatever reason I was under the impression the floor() was needed. I don't remember what prompted me to think that.

You could make your job easier with %.0f:

Code:

double get_first_digit(double number) {
    char buffer[BUFSIZ];

    sprintf(buffer, "%.0f", number);

    if(*buffer) {
        return buffer[strlen(buffer) - 1] - '0';
    }

    return 0;
}

But anyway, using a string is a bad idea in my opinion. You use lots of extra memory, and sprintf()'s probably pretty slow too. The mathematical way isn't much harder, if any.

I think you're right, but for whatever reason I was under the impression the floor() was needed. I don't remember what prompted me to think that.

Casting a floating point number to an int would indeed have the same effect as flooring it. But using floor() and modf() instead of a cast to an integral type is a much better idea, because a floating point number can easily store values that cannot be represented in any integral type. If you casted one of those values to (int), you'd get an integer overflow.

So floor() exists so that you can round a number without risk of overflow.

Anyway, you could also use something like this:

Code:

double get_first_digit(double number) {
    return floor(floor(number/10)*10 - number);
}

That's probably a bad way to do it . . . perhaps something like this, then.

Code:

double get_first_digit(double number) {
    return floor(fmod(number, 10));
}

To get the first decimal digit, of course.

Originally Posted by brewbuck

Again though, not always. Consider a negative number. On my system, (int)(-3.5) == -3, not -4. A floor() operation would have taken that value to -4.0.

Floor()ing the float under that condition is not what the OP wants I take it. Straight casting might be better if he chooses to do conversions to an int.

The string method still sounds good. lol...

I like

Code:

double get_first_digit(double number) {
    if(number < 0) {
        return ceil(fmod(number, 10));
    }
    else {
        return floor(fmod(number, 10));
    }
}

Or if the digit alone was required, without a sign:

Code:

floor(fmod(fabs(number), 10));

[edit] The C99 round() would work too, replacing floor() and ceil().

Code:

round(fmod(number, 10));

[/edit]

Using floor explicitly also ensures correct operation if the /QIfist compiler option is turned on. (Which I currently do for my software 3D engine)

Code:

#include <stdio.h>

int main(void)
{
	float fNum = 123.456;
	
	printf("fNum = &#37;1.0f\n",fNum);
	
	return 0;
}

Output:

Code:

fNum = 123

In other words, no. That's wrong.