UKHAS Wiki

UK High Altitude Society

User Tools

Site Tools


guides:fixedpoint

Fixed-point arithmetic

Fixed-point arithmetic provides a way to handle floating point numbers using only simple integer data types. As an example, the value 3.141 could be represented as an integer with a value of 3141 - in this case the value has a fixed-precision of 3 decimal places.

String to a fixed-point integer

#include <stdint.h>
 
int32_t strntofp(const char *s, char **endptr, size_t n, char dp)
{
        int32_t i = 0;
        char neg = 0, fp = 0;
 
        if(n <= 0) goto out;
 
        /* Test for a + or - sign */
        switch(*s)
        {
        case '-': neg = !neg;
        case '+': s++; n--;
        }
 
        /* Read in the number */
        while(*s && n && (!fp || dp))
        {
                char d = *s;
 
                if(d >= '0' && d <= '9')
                {
                        /* Add the digit */
                        i *= 10;
                        i += d - '0';
 
                        if(fp) dp--;
                }
                else if(dp > 0 && d == '.') fp = 1;
                else break;
 
                /* Next... */
                s++;
                n--;
        }
 
        while(dp > 0)
        {
                i *= 10;
                dp--;
        }
 
        /* Fix result if it's negative */
        if(neg) i = -i;
 
out:
        /* Set the end pointer if needed */
        if(endptr) *endptr = (char *) s;
 
        return(i);
}

Fixed-point integer to string

#include <stdio.h>
#include <stdint.h>
 
const char *fptostr(char *s, size_t n, int32_t value, int8_t dp)
{
        int32_t sh;
        char i, neg;
 
        neg = value < 0;
        if(neg) value = -value;
 
        for(sh = 1, i = 0; i < dp; i++) sh *= 10;
 
        snprintf(s, n, "%s%u.%0*u",
                (neg ? "-" : ""),
                value / sh, dp,
                value % sh);
 
        return(s);
}
guides/fixedpoint.txt ยท Last modified: 2011/07/17 19:28 by fsphil