Calculating number of digits?

[wordlife25]

Hello all, I've ran into a problem in my programming recently.

How do I calculate the number of digits in a double variable? For example, for a double with the value of 93938, I want it to return a value of 5, meaning 5 digits. Is there any inbuilt functions which returns the number of digits of a double variable or do I have to code it myself?

Thanks for the help!

[Skizmo]

You can use sprintf to write the number to a string and then get the length of the string

[indiocolifa]

For an integer :

int digits (T f)
{
   return (1+floor(log10(abs(x)))));
}

For floats, I will use some form of sprintf, excepting the point or comma.

[Ajay Vijay]

For a number (base 10) e.g:

int digits (T f)
{
   return (1+floor(log10(abs(x)))));
}

T can be double, float, int ,etc... just overload it as you want.

Wicked! You need to be mathematician, a computer scientist like me can't understand that code.

[ghovis]

or a 'brute force' method - use integer division, and divide by multiples of 10 until the result is zero - would be a simple loop to write:

(int)(93938/ (1*10)) = 9394
(int)(93938/ (2*10)) = 940
(int)(93938/ (3*10)) = 94
(int)(93938/ (4*10)) = 9
(int)(93938/ (5*10)) = 0 // answer is 5

[wildfrog]

indiocolifas solution will work, but it will only return the length of the numbers integral part.

- petter

[indiocolifa]

indiocolifas solution will work, but it will only return the length of the numbers integral part.

- petter

Exactly petter, I modified the post to point that. It's for integers.

For floats, may be a good math solution (not sprintf "hacks" ) is to iterate like ghovis proposed.

[krmed]

As noted, these solutions are good for integers, but all of them will have problems with a double (or float).

Printing to a string depends on the format used to print - any many doubles/floats cannot be represented exactly, so the number of digits would be infinite.

To make indiocolifa's solution a bit more generic, the number of digits required to represent a number is any number system (binary, octal, decimal, hex, etc), you can use

int digits (T f)
{
   return (1+(floor(log10(abs(x)) / log10(base))));
}

Hope that helps.