Question

I’d like to get the first three digits from large floats or integers and on some insert a decimal. For example:

KB
----------
32589 >> 325
43266 >> 432

MB
----------
1234599 >> 1.23
3422847 >> 3.42

For the particular number, I will have the “KB” and “MB” strings. This will let me know if the decimal is required, as in “MB” examples. I looked at NSNumberFormatter but wasn’t sure what on there would help. Any suggestions?

Answer 1

Actually, decided to chime in with my own answer. This is my routine for generating strings for outputting human sizes from byte counts:

#include <math.h>    // for isgreater()
static NSString * MemorySizeString( mach_vm_size_t size )
{
    enum
    {
        kSizeIsBytes        = 0,
        kSizeIsKilobytes,
        kSizeIsMegabytes,
        kSizeIsGigabytes,
        kSizeIsTerabytes,
        kSizeIsPetabytes,
        kSizeIsExabytes
    };

    int sizeType = kSizeIsBytes;
    double dSize = (double) size;

    while ( isgreater(dSize, 1024.0) )
    {
        dSize = dSize / 1024.0;
        sizeType++;
    }

    NSMutableString * str = [[NSMutableString alloc] initWithFormat: (sizeType == kSizeIsBytes ? @"%.00f" : @"%.02f"), dSize];
    switch ( sizeType )
    {
        default:
        case kSizeIsBytes:
            [str appendString: @" bytes"];
            break;

        case kSizeIsKilobytes:
            [str appendString: @"KB"];
            break;

        case kSizeIsMegabytes:
            [str appendString: @"MB"];
            break;

        case kSizeIsGigabytes:
            [str appendString: @"GB"];
            break;

        case kSizeIsTerabytes:
            [str appendString: @"TB"];
            break;

        case kSizeIsPetabytes:
            [str appendString: @"PB"];
            break;

        case kSizeIsExabytes:
            [str appendString: @"EB"];
            break;
    }

    NSString * result = [str copy];
    [str release];

    return ( [result autorelease] );
}

It works out the correct size by seeing what order of binary magnitude it is, using 1024 as the base (1024 bytes = 1KB, 1024KB = 1MB, etc.). While doing this, it shrinks the input value (using floating-point arithmetic) such that once it’s below 1024, it has both a human-readable value and a magnitude specifier. It then generates a string containing the formatted value (no decimal places for bytes, 2 decimal places for any larger magnitude), and inspects the magnitude constant to determine what suffix to attach.