Round error converting from double to string without using %f - c

Following this question:
If you dont have %f in C, how to write a C program to print decimal number upto 2 decimal places without %f?
This is my attempt:
#include <stdio.h>
#include <math.h>
void dbl2str(char *s, double number, int decimals)
{
double integral, fractional;
int n, i;
fractional = modf(number, &integral);
n = sprintf(s, "%d%c", (int)integral, decimals ? '.' : 0);
for (i = 0; i < decimals; i++) {
fractional *= 10;
s[n + i] = '0' + (int)fractional;
fractional = modf(fractional, &integral);
}
s[n + i] = '\0';
}
int main(void)
{
char s[32];
dbl2str(s, 3.1416, 4);
printf("%s\n", s);
dbl2str(s, 3.159, 4);
printf("%s\n", s);
dbl2str(s, 3.04, 2);
printf("%s\n", s);
return 0;
}
Output:
3.1415
3.1589
3.04
As you can see there are round errors, is there a way to get the correct output?

The problem is that 0.0006 isn't exactly representable in binary instead the representation will exactly match something like 0.0005999... This means that when you do the multiplication by 10 four times you get 5 as your last digit instead of 10. You need to look at the next digit in your sequence and round appropriately (the next digit will be greater than 5 in this case).
Be careful with the rounding because rounding up the last digit may cause the one before to round up as well (if the last digit was also a 9).

The line
s[n + i] = '0' + (int)fractional;
truncates the fractional part to an integer. You want the last digit to be correctly rounded, so you have to treat it different than the others:
void dbl2str(char *s, double number, int decimals)
{
double integral, fractional;
int n, i;
fractional = modf(number, &integral);
if (fractional < 0)
fractional = -fractional;
n = sprintf(s, "%d%c", (int)integral, decimals ? '.' : 0);
for (i = 0; i < decimals-1; i++) {
fractional *= 10;
s[n++] = '0' + (int)fractional;
fractional = modf(fractional, &integral);
}
fractional *= 10;
s[n++] = '0' + (int)(fractional+0.5f);
fractional = modf(fractional, &integral);
s[n] = '\0';
}
I added a test for negative numbers as well. Without, the fractional part goes wild.

I would do something like
void dbl2str(char *s, double number, int decimals)
{
double integral, fractional;
int n, i;
/* use an epsilon and add the correct rounding value */
double eps= 1e-9;
double round = 0.5*pow(10,-decimals);
fractional = modf(number+round+eps, &integral);
n = sprintf(s, "%d%c", (int)integral, decimals ? '.' : 0);
for (i = 0; i < decimals; i++) {
fractional *= 10;
s[n + i] = '0' + (int)fractional;
fractional = modf(fractional, &integral);
}
s[n + i] = '\0';
}
This way, I add 0.5*10^-decimals to get the (int) fractional to round correctly up to an epsilon.
It works for your input:
3.1416
3.1590
3.04
Of course, you may need to adjust the epsilon, depending on the digits, and then think about the case where you have exhausted your 16 digit precision. Overall, this is very hard to do right.

Add this:
if(i == decimals - 1)
fractional = round(fractional);
after fractional *= 10;

Related

My decimal to hex conversion function only woks with positive nums

I'm having problems converting negative numbers, from decimal base to hexadecimal base, with the following function:
#include <stdio.h>
int main()
{
int quotient, remainder;
int i, j = 0;
char hexadecimalnum[100];
quotient = -50;
while (quotient != 0)
{
remainder = quotient % 16;
if (remainder < 10)
hexadecimalnum[j++] = 48 + remainder;
else
hexadecimalnum[j++] = 55 + remainder;
quotient = quotient / 16;
}
strrev(hexadecimalnum);
printf("%s", hexadecimalnum);
return 0;
}
For quotient = -50; the correct output should be:
ffffffce
But this function's output is:
.
With positive numbers the output is always correct but with negative numbers not.
I'm having a hard time understanding to why it doesn't work with negative numbers.
Some fixes:
unsigned int quotient - you need to convert -50 to a large hex number in two's complement or you'll get the wrong number of iterations (2) in the loop, instead of 8 as required.
Removal of "magic numbers": '0' + remainder and 'A' + remainder - 10.
Zero initialize hexadecimalnum becaues it needs to be null terminated before printing a string from there. Better yet, add the null termination explicitly.
Use for loops when possible.
Might as well store the characters from the back to front and save the extra call of reversing the string.
Result:
#include <stdio.h>
// 4 bytes*2 = 8 nibbles
#define HEX_STRLEN (sizeof(int)*2)
int main()
{
unsigned int remainder;
int i = 0;
char hex[100];
for(unsigned int q = -50; q!=0; q/=16)
{
remainder = q % 16;
if (remainder < 10)
hex[HEX_STRLEN-i-1] = '0' + remainder;
else
hex[HEX_STRLEN-i-1] = 'A' + remainder - 10;
i++;
}
hex[HEX_STRLEN] = '\0'; // explict null termination
printf("%s\n", hex);
}
(There's lots of improvements than can be made still, this is just to be considered as the first draft.)
You can use printf's format specifier "%08x", then you can print any number in their respective hexadecimal representation.
#include <stdio.h>
void num_to_hex(int a, char *ptr) { snprintf(ptr, 9, "%08x", a); }
int main() {
char hex[10] = {};
num_to_hex(-50, hex);
printf("%s\n", hex);
return 0;
}
Output:
ffffffce

convert a decimal number to string in c

I need to redo printf for a projet, so I actually have a problem with the conversion of float.
I managed to convert almost everything but for the number 1254451555.6
I got an issue: I got 1254451555.59999.
I think it's the calculation to keep the part after the . that doesnt work.
nbr = ((n - nbr) * 100000000);
I tried different things but I haven't managed to fix it yet.
Do you have any idea?
int getlenghtitoa(long long n, int nbase)
{
int i;
i = 0;
while (n >= 0)
{
n /= nbase;
i++;
if (n == 0)
break ;
}
return (i);
}
float ft_nbconv(float n, int i)
{
while (i-- > 0)
n = n *10;
return (n);
}
int ft_power(long long nbr)
{
int i;
i = 1;
while(nbr > 10)
{
i *= 10;
nbr = nbr / 10;
}
return (i);
}
char *ft_conver_f(long double n)
{
char *dest;
int i;
int a;
long long int nbr;
int power;
nbr = (long long) n;
i = getlenghtitoa((long long )n, 10);
if (!(dest = malloc(sizeof(char) * (i + 8))))
return (0);
a = i;
i = 0;
power = ft_power(nbr);
while (a--)
{
dest[i++] = ((nbr / power) % 10) + '0';
if (power != 1)
power /= 10;
}
dest[i++] = '.';
nbr = ((n - nbr) * 100000000);
power = 10000000;
while (a++ < 5)
{
if (a == 5)
if ((((nbr / power)) % 10) >= 5)
{
dest[i++] = ((nbr / power) % 10 + 1) + '0';
break;
}
dest[i++] = ((nbr / power) % 10) + '0';
power /= 10;
}
dest[i] = '\0';
return (dest);
}
Most decimal fractions cannot be represented exactly as binary fractions. A consequence is that, in general, the decimal floating-point numbers you enter are only approximated by the binary floating-point numbers actually stored in the machine.
That's why when implementing a printf, the only way to really be able to convert a floating number to a 2-seperated-by-point integers, is by using the precision factor and rounding manually.
If you are not required to implement the precision, the default is 6.
(Precision is the number of places to print after the dot (and it's rounded)).
And that's what's missing in your implementation.
Let's call the digits before the dot the ipart and the digits after the fpart .
nbr = ((n - nbr) * 100000000);
This should be
nbr = ((n - nbr) * 10000000); // 7 zeros
// nbr is now equal to 5999999
if (nbr % 10 >= 5)
{
nbr = nbr / 10 + 1;
}
else
nbr = nbr / 10;
This way, you get 7 digits after the dot, see if the last one is higher than 5, if it is, you add +1 to nbr (after dividing by 10 to make sure nbr has 6 digits), if it's not, you just divide by 10.
One more note about this rounding method, It will not be able to carry the rounding from the fpart to the ipart .
what if you want to print 3.9999999 ? It should print 4.000000. That means that can't just convert the ipart to a string from the beginning, because sometimes rounding the fpart will add +1 to your ipart
So think about creating a function ltoa for example that takes a long long int and converts it to a string, complete the piece of code about rounding i just gave you to make sure rounding can be carried to the ipart , then convert the whole thing to string using something like
dest = join(ltoa(ipart), ".", ltoa(fpart)).
A couple more notes, your function does not handle negative numbers.
And your int ft_pow can be easily flooded, so consider changing to long long ft_pow

Concatenating binary numbers

I am trying to code a program that will take a floating point number in base 10 and convert its fractional part in base 2. In the following code, I am intending to call my converting function into a printf, and format the output; the issue I have lies in my fra_binary() where I can't figure out the best way to return an integer made of the result of the conversion at each turn respectively (concatenation). Here is what I have done now (the code is not optimized because I am still working on it) :
#include <stdio.h>
#include <math.h>
int fra_binary(double fract) ;
int main()
{
long double n ;
double fract, deci ;
printf("base 10 :\n") ;
scanf("%Lf", &n) ;
fract = modf(n, &deci) ;
int d = deci ;
printf("base 2: %d.%d\n", d, fra_binary(fract)) ;
return(0) ;
}
int fra_binary(double F)
{
double fl ;
double decimal ;
int array[30] ;
for (int i = 0 ; i < 30 ; i++) {
fl = F * 2 ;
F = modf(fl, &decimal) ;
array[i] = decimal ;
if (F == 0) break ;
}
return array[0] ;
}
Obviously this returns partly the desired output, because I would need the whole array concatenated as one int or char to display the series of 1 and 0s I need. So at each turn, I want to use the decimal part of the number I work on as the binary number to concatenate (1 + 0 = 10 and not 1). How would I go about it?
Hope this makes sense!
return array[0] ; is only the first value of int array[30] set in fra_binary(). Code discards all but the first calculation of the loop for (int i = 0 ; i < 30 ; i++).
convert its fractional part in base 2
OP's loop idea is a good starting point. Yet int array[30] is insufficient to encode the fractional portion of all double into a "binary".
can't figure out the best way to return an integer
Returning an int will be insufficient. Instead consider using a string - or manage an integer array in a likewise fashion.
Use defines from <float.h> to drive the buffer requirements.
#include <stdio.h>
#include <math.h>
#include <float.h>
char *fra_binary(char *dest, double x) {
_Static_assert(FLT_RADIX == 2, "Unexpected FP base");
double deci;
double fract = modf(x, &deci);
fract = fabs(fract);
char *s = dest;
do {
double d;
fract = modf(fract * 2.0, &d);
*s++ = "01"[(int) d];
} while (fract);
*s = '\0';
// For debug
printf("%*.*g --> %.0f and .", DBL_DECIMAL_DIG + 8, DBL_DECIMAL_DIG, x,
deci);
return dest;
}
int main(void) {
// Perhaps 53 - -1021 + 1
char fraction_string[DBL_MANT_DIG - DBL_MIN_EXP + 1];
puts(fra_binary(fraction_string, -0.0));
puts(fra_binary(fraction_string, 1.0));
puts(fra_binary(fraction_string, asin(-1))); // machine pi
puts(fra_binary(fraction_string, -0.1));
puts(fra_binary(fraction_string, DBL_MAX));
puts(fra_binary(fraction_string, DBL_MIN));
puts(fra_binary(fraction_string, DBL_TRUE_MIN));
}
Output
-0 --> -0 and .0
1 --> 1 and .0
3.1415926535897931 --> 3 and .001001000011111101101010100010001000010110100011
-0.10000000000000001 --> -0 and .0001100110011001100110011001100110011001100110011001101
1.7976931348623157e+308 --> 179769313486231570814527423731704356798070600000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000 and .0
2.2250738585072014e-308 --> 0 and .00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001
4.9406564584124654e-324 --> 0 and .000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001
Also unclear why input is long double, yet processing is with double. Recommend using just one FP type.
Note that your algorithm finds out the binary representation of the fraction most significant bit first.
One way to convert the fractional part to a binary string, would be to supply the function with a string and a string length, and have the function fill it with up to that many binary digits:
/* This function returns the number of chars needed in dst
to describe the fractional part of value in binary,
not including the trailing NUL ('\0').
Returns zero in case of an error (non-finite value).
*/
size_t fractional_bits(char *dst, size_t len, double value)
{
double fraction, integral;
size_t i = 0;
if (!isfinite(value))
return 0;
if (value > 0.0)
fraction = modf(value, &integral);
else
if (value < 0.0)
fraction = modf(-value, &integral);
else {
/* Zero fraction. */
if (len > 1) {
dst[0] = '0';
dst[1] = '\0';
} else
if (len > 0)
dst[0] = '\0';
/* One binary digit was needed for exact representation. */
return 1;
}
while (fraction > 0.0) {
fraction = fraction * 2.0;
if (fraction >= 1.0) {
fraction = fraction - 1.0;
if (i < len)
dst[i] = '1';
} else
if (i < len)
dst[i] = '0';
i++;
}
if (i < len)
dst[i] = '\0';
else
if (len > 0)
dst[len - 1] = '\0';
return i;
}
The above function works very much like snprintf(), except it takes only the double whose fractional bits are to be stored as a string of binary digits (0 or 1). and returns 0 in case of an error (non-finite double value).
Another option is to use an unsigned integer type to hold the bits. For example, if your code is intended to work on architectures where double is an IEEE-754 Binary64 type or similar, the mantissa has up to 53 bits of precision, and an uint64_t would suffice.
Here is an example of that:
uint64_t fractional_bits(const double val, size_t bits)
{
double fraction, integral;
uint64_t result = 0;
if (bits < 1 || bits > 64) {
errno = EINVAL;
return 0;
}
if (!isfinite(val)) {
errno = EDOM;
return 0;
}
if (val > 0.0)
fraction = modf(val, &integral);
else
if (val < 0.0)
fraction = modf(-val, &integral);
else {
errno = 0;
return 0;
}
while (bits-->0) {
result = result << 1;
fraction = fraction * 2.0;
if (fraction >= 1.0) {
fraction = fraction - 1.0;
result = result + 1;
}
}
errno = 0;
return result;
}
The return value is the binary representation of the fractional part: [i]fractional_part[/i] ≈ [i]result[/i] / 2[sup][i]bits[/i][/sup], where [i]bits[/i] is between 1 and 64, inclusive.
In order for the caller to detect an error, the function clears errno to zero if no error occurred. If an error does occur, the function returns zero with errno set to EDOM if the value is not finite, or to EINVAL if bits is less than 1 or greater than 64.
You can combine the two approaches, if you implement an arbitrary-size unsigned integer type, or a bitmap type.

Writing IEEE 754-1985 double as ASCII on a limited 16 bytes string

This is a follow-up to my original post. But I'll repeat it for clarity:
As per DICOM standard, a type of floating point can be stored using a Value Representation of Decimal String. See Table 6.2-1. DICOM Value Representations:
Decimal String: A string of characters representing either a fixed
point number or a floating point number. A fixed point number shall
contain only the characters 0-9 with an optional leading "+" or "-"
and an optional "." to mark the decimal point. A floating point number
shall be conveyed as defined in ANSI X3.9, with an "E" or "e" to
indicate the start of the exponent. Decimal Strings may be padded with
leading or trailing spaces. Embedded spaces are not allowed.
"0"-"9", "+", "-", "E", "e", "." and the SPACE character of Default
Character Repertoire. 16 bytes maximum
The standard is saying that the textual representation is fixed point vs. floating point. The standard only refers to how the values are represented within in the DICOM data set itself. As such there is not requirement to load a fixed point textual representation into a fixed-point variable.
So now that this is clear that DICOM standard implicitely recommend double (IEEE 754-1985) for representing a Value Representation of type Decimal String (maximum of 16 significant digits). My question is how do I use the standard C I/O library to convert back this binary representation from memory into ASCII onto this limited sized string ?
From random source on internet, this is non-trivial, but a generally accepted solution is either:
printf("%1.16e\n", d); // Round-trippable double, always with an exponent
or
printf("%.17g\n", d); // Round-trippable double, shortest possible
Of course both expression are invalid in my case since they can produce output much longer than my limited maximum of 16 bytes. So what is the solution to minimizing the loss in precision when writing out an arbitrary double value to a limited 16 bytes string ?
Edit: if this is not clear, I am required to follow the standard. I cannot use hex/uuencode encoding.
Edit 2: I am running the comparison using travis-ci see: here
So far the suggested codes are:
Serge Ballesta
chux
Mark Dickinson
chux
Results I see over here are:
compute1.c leads to a total sum error of: 0.0095729050923877828
compute2.c leads to a total sum error of: 0.21764383725715469
compute3.c leads to a total sum error of: 4.050031792674619
compute4.c leads to a total sum error of: 0.001287056579548422
So compute4.c leads to the best possible precision (0.001287056579548422 < 4.050031792674619), but triple (x3) the overall execution time (only tested in debug mode using time command).
It is trickier than first thought.
Given the various corner cases, it seems best to try at a high precision and then work down as needed.
Any negative number prints the same as a positive number with 1 less precision due to the '-'.
'+' sign not needed at the beginning of the string nor after the 'e'.
'.' not needed.
Dangerous to use anything other than sprintf() to do the mathematical part given so many corner cases. Given various rounding modes, FLT_EVAL_METHOD, etc., leave the heavy coding to well established functions.
When an attempt is too long by more than 1 character, iterations can be saved. E.g. If an attempt, with precision 14, resulted with a width of 20, no need to try precision 13 and 12, just go to 11.
Scaling of the exponent due to the removal of the '.', must be done after sprintf() to 1) avoid injecting computational error 2) decrementing a double to below its minimum exponent.
Maximum relative error is less than 1 part in 2,000,000,000 as with -1.00000000049999e-200. Average relative error about 1 part in 50,000,000,000.
14 digit precision, the highest, occurs with numbers like 12345678901234e1 so start with 16-2 digits.
static size_t shrink(char *fp_buffer) {
int lead, expo;
long long mant;
int n0, n1;
int n = sscanf(fp_buffer, "%d.%n%lld%ne%d", &lead, &n0, &mant, &n1, &expo);
assert(n == 3);
return sprintf(fp_buffer, "%d%0*llde%d", lead, n1 - n0, mant,
expo - (n1 - n0));
}
int x16printf(char *dest, size_t width, double value) {
if (!isfinite(value)) return 1;
if (width < 5) return 2;
if (signbit(value)) {
value = -value;
strcpy(dest++, "-");
width--;
}
int precision = width - 2;
while (precision > 0) {
char buffer[width + 10];
// %.*e prints 1 digit, '.' and then `precision - 1` digits
snprintf(buffer, sizeof buffer, "%.*e", precision - 1, value);
size_t n = shrink(buffer);
if (n <= width) {
strcpy(dest, buffer);
return 0;
}
if (n > width + 1) precision -= n - width - 1;
else precision--;
}
return 3;
}
Test code
double rand_double(void) {
union {
double d;
unsigned char uc[sizeof(double)];
} u;
do {
for (size_t i = 0; i < sizeof(double); i++) {
u.uc[i] = rand();
}
} while (!isfinite(u.d));
return u.d;
}
void x16printf_test(double value) {
printf("%-27.*e", 17, value);
char buf[16+1];
buf[0] = 0;
int y = x16printf(buf, sizeof buf - 1, value);
printf(" %d\n", y);
printf("'%s'\n", buf);
}
int main(void) {
for (int i = 0; i < 10; i++)
x16printf_test(rand_double());
}
Output
-1.55736829786841915e+118 0
'-15573682979e108'
-3.06117209691283956e+125 0
'-30611720969e115'
8.05005611774356367e+175 0
'805005611774e164'
-1.06083057094522472e+132 0
'-10608305709e122'
3.39265065244054607e-209 0
'33926506524e-219'
-2.36818580315246204e-244 0
'-2368185803e-253'
7.91188576978592497e+301 0
'791188576979e290'
-1.40513111051994779e-53 0
'-14051311105e-63'
-1.37897140950449389e-14 0
'-13789714095e-24'
-2.15869805640288206e+125 0
'-21586980564e115'
For finite floating point values the printf() format specifier "%e" well matches
"A floating point number shall be ... with an "E" or "e" to indicate the start of the exponent"
[−]d.ddd...ddde±dd
The sign is present with negative numbers and likely -0.0. The exponent is at least 2 digits.
If we assume DBL_MAX < 1e1000, (safe for IEEE 754-1985 double), then the below works in all cases: 1 optional sign, 1 lead digit, '.', 8 digits, 'e', sign, up to 3 digits.
(Note: the "16 bytes maximum" does not seem to refer to C string null character termination. Adjust by 1 if needed.)
// Room for 16 printable characters.
char buf[16+1];
int n = snprintf(buf, sizeof buf, "%.*e", 8, x);
assert(n >= 0 && n < sizeof buf);
puts(buf);
But this reserves room for the optional sign and 2 to 3 exponent digits.
The trick is the boundary, due to rounding, of when a number uses 2 or uses 3 exponent digits is fuzzy. Even testing for negative numbers, the -0.0 is an issue.
[Edit] Also needed test for very small numbers.
Candidate:
// Room for 16 printable characters.
char buf[16+1];
assert(isfinite(x)); // for now, only address finite numbers
int precision = 8+1+1;
if (signbit(x)) precision--; // Or simply `if (x <= 0.0) precision--;`
if (fabs(x) >= 9.99999999e99) precision--; // some refinement possible here.
else if (fabs(x) <= 1.0e-99) precision--;
int n = snprintf(buf, sizeof buf, "%.*e", precision, x);
assert(n >= 0 && n < sizeof buf);
puts(buf);
Additional concerns:
Some compilers print at least 3 exponent digits.
The maximum number of decimal significant digits for IEEE 754-1985 double needed varies on definition of need, but likely about 15-17. Printf width specifier to maintain precision of floating-point value
Candidate 2: One time test for too long an output
// Room for N printable characters.
#define N 16
char buf[N+1];
assert(isfinite(x)); // for now, only address finite numbers
int precision = N - 2 - 4; // 1.xxxxxxxxxxe-dd
if (signbit(x)) precision--;
int n = snprintf(buf, sizeof buf, "%.*e", precision, x);
if (n >= sizeof buf) {
n = snprintf(buf, sizeof buf, "%.*e", precision - (n - sizeof buf) - 1, x);
}
assert(n >= 0 && n < sizeof buf);
puts(buf);
C library formatter has no direct format for your requirement. At a simple level, if you can accept the waste of characters of the standard %g format (e20 is written e+020: 2 chars wasted), you can:
generate the output for the %.17g format
if it is greater the 16 characters, compute the precision that would lead to 16
generate the output for that format.
Code could look like:
void encode(double f, char *buf) {
char line[40];
char format[8];
int prec;
int l;
l = sprintf(line, "%.17g", f);
if (l > 16) {
prec = 33 - strlen(line);
l = sprintf(line, "%.*g", prec, f);
while(l > 16) {
/* putc('.', stdout);*/
prec -=1;
l = sprintf(line, "%.*g", prec, f);
}
}
strcpy(buf, line);
}
If you really try to be optimal (meaning write e30 instead of e+030), you could try to use %1.16e format and post-process the output. Rationale (for positive numbers):
the %1.16e format allows you to separate the mantissa and the exponent (base 10)
if the exponenent is between size-2 (included) and size (excluded): just correctly round the mantissa to the int part and display it
if the exponent is between 0 and size-2 (both included): display the rounded mantissa with the dot correctly placed
if the exponent is between -1 and -3 (both included): start with a dot, add eventual 0 and fill with rounded mantissa
else use a e format with minimal size for the exponent part and fill with the rounded mantissa
Corner cases:
for negative numbers, put a starting - and add the display for the opposite number and size-1
rounding : if first rejected digit is >=5, increase preceding number and iterate if it was a 9. Process 9.9999999999... as a special case rounded to 10
Possible code:
void clean(char *mant) {
char *ix = mant + strlen(mant) - 1;
while(('0' == *ix) && (ix > mant)) {
*ix-- = '\0';
}
if ('.' == *ix) {
*ix = '\0';
}
}
int add1(char *buf, int n) {
if (n < 0) return 1;
if (buf[n] == '9') {
buf[n] = '0';
return add1(buf, n-1);
}
else {
buf[n] += 1;
}
return 0;
}
int doround(char *buf, unsigned int n) {
char c;
if (n >= strlen(buf)) return 0;
c = buf[n];
buf[n] = 0;
if ((c >= '5') && (c <= '9')) return add1(buf, n-1);
return 0;
}
int roundat(char *buf, unsigned int i, int iexp) {
if (doround(buf, i) != 0) {
iexp += 1;
switch(iexp) {
case -2:
strcpy(buf, ".01");
break;
case -1:
strcpy(buf, ".1");
break;
case 0:
strcpy(buf, "1.");
break;
case 1:
strcpy(buf, "10");
break;
case 2:
strcpy(buf, "100");
break;
default:
sprintf(buf, "1e%d", iexp);
}
return 1;
}
return 0;
}
void encode(double f, char *buf, int size) {
char line[40];
char *mant = line + 1;
int iexp, lexp, i;
char exp[6];
if (f < 0) {
f = -f;
size -= 1;
*buf++ = '-';
}
sprintf(line, "%1.16e", f);
if (line[0] == '-') {
f = -f;
size -= 1;
*buf++ = '-';
sprintf(line, "%1.16e", f);
}
*mant = line[0];
i = strcspn(mant, "eE");
mant[i] = '\0';
iexp = strtol(mant + i + 1, NULL, 10);
lexp = sprintf(exp, "e%d", iexp);
if ((iexp >= size) || (iexp < -3)) {
i = roundat(mant, size - 1 -lexp, iexp);
if(i == 1) {
strcpy(buf, mant);
return;
}
buf[0] = mant[0];
buf[1] = '.';
strncpy(buf + i + 2, mant + 1, size - 2 - lexp);
buf[size-lexp] = 0;
clean(buf);
strcat(buf, exp);
}
else if (iexp >= size - 2) {
roundat(mant, iexp + 1, iexp);
strcpy(buf, mant);
}
else if (iexp >= 0) {
i = roundat(mant, size - 1, iexp);
if (i == 1) {
strcpy(buf, mant);
return;
}
strncpy(buf, mant, iexp + 1);
buf[iexp + 1] = '.';
strncpy(buf + iexp + 2, mant + iexp + 1, size - iexp - 1);
buf[size] = 0;
clean(buf);
}
else {
int j;
i = roundat(mant, size + 1 + iexp, iexp);
if (i == 1) {
strcpy(buf, mant);
return;
}
buf[0] = '.';
for(j=0; j< -1 - iexp; j++) {
buf[j+1] = '0';
}
if ((i == 1) && (iexp != -1)) {
buf[-iexp] = '1';
buf++;
}
strncpy(buf - iexp, mant, size + 1 + iexp);
buf[size] = 0;
clean(buf);
}
}
I think your best option is to use printf("%.17g\n", d); to generate an initial answer and then trim it. The simplest way to trim it is to drop digits from the end of the mantissa until it fits. This actually works very well but will not minimize the error because you are truncating instead of rounding to nearest.
A better solution would be to examine the digits to be removed, treating them as an n-digit number between 0.0 and 1.0, so '49' would be 0.49. If their value is less than 0.5 then just remove them. If their value is greater than 0.50 then increment the printed value in its decimal form. That is, add one to the last digit, with wrap-around and carry as needed. Any trailing zeroes that are created should be trimmed.
The only time this becomes a problem is if the carry propagates all the way to the first digit and overflows it from 9 to zero. This might be impossible, but I don't know for sure. In this case (+9.99999e17) the answer would be +1e18, so as long as you have tests for that case you should be fine.
So, print the number, split it into sign/mantissa strings and an exponent integer, and string manipulate them to get your result.
Printing in decimal cannot work because for some numbers a 17 digit mantissa is needed which uses up all of your space without printing the exponent. To be more precise, printing a double in decimal sometimes requires more than 16 characters to guarantee accurate round-tripping.
Instead you should print the underlying binary representation using hexadecimal. This will use exactly 16 bytes, assuming that a null-terminator isn't needed.
If you want to print the results using fewer than 16 bytes then you can basically uuencode it. That is, use more than 16 digits so that you can squeeze more bits into each digit. If you use 64 different characters (six bits) then a 64-bit double can be printed in eleven characters. Not very readable, but tradeoffs must be made.

Any decent way of printing out floats and doubles with commas?

I'm working on a program that regards with currency. Ive been finding a solution to display money values decently like this:
9,999.99 USD
Remember when assigning a certain variable with a value (money), you musn't insert commas.
I.e.:
double money=9999.99;
And when accessing it;
printf("%.2l USD",money);
Which will output:
9999.99 USD
This is not what I want, especially on bigger amounts exceeding the hundredth, thousandth, millionth, or even billionth place value.
Now I can't find any solution than printing out the desired output directly on the printf.
printf("9,999.99");
Which is undesirable with many variables.
Can anyone help me out?
Please take a look and printf manual page taking note of the following bit:
*"For some numeric conversions a radix character ("decimal point") or thousands' grouping character is used. The actual character used depends on the LC_NUMERIC part of the locale. The POSIX locale uses '.' as radix character, and does not have a grouping character. Thus,
printf("%'.2f", 1234567.89);
results in "1234567.89" in the POSIX locale, in "1234567,89" in the nl_NL locale, and in "1.234.567,89" in the da_DK locale."*
This can be changed by the function setlocale
There is a function, strfmon which might be able to help you
First, don't use floating-point types to represent money because normally floating-point types are binary and as such cannot represent all decimal fractions (cents) exactly, further these types are prone to rounding errors. Use integers instead and count cents instead of dollars.
#include <stdio.h>
#include <limits.h>
unsigned long long ConstructMoney(unsigned long long dollars, unsigned cents)
{
return dollars * 100 + cents;
}
void PrintWithCommas(unsigned long long n)
{
char s[sizeof n * CHAR_BIT + 1];
char *p = s + sizeof s;
unsigned count = 0;
*--p = '\0';
do
{
*--p = '0' + n % 10;
n /= 10;
if (++count == 3 && n)
{
*--p = ',';
count = 0;
}
} while (n);
printf("%s", p);
}
void PrintMoney(unsigned long long n)
{
PrintWithCommas(n / 100);
putchar('.');
n %= 100;
putchar('0' + n / 10);
putchar('0' + n % 10);
}
int main(void)
{
PrintMoney(ConstructMoney(0, 0)); puts("");
PrintMoney(ConstructMoney(0, 1)); puts("");
PrintMoney(ConstructMoney(1, 0)); puts("");
PrintMoney(ConstructMoney(1, 23)); puts("");
PrintMoney(ConstructMoney(12, 34)); puts("");
PrintMoney(ConstructMoney(123, 45)); puts("");
PrintMoney(ConstructMoney(1234, 56)); puts("");
PrintMoney(ConstructMoney(12345, 67)); puts("");
PrintMoney(ConstructMoney(123456, 78)); puts("");
PrintMoney(ConstructMoney(1234567, 89)); puts("");
return 0;
}
Output (ideone):
0.00
0.01
1.00
1.23
12.34
123.45
1,234.56
12,345.67
123,456.78
1,234,567.89
If you're using the standard library, there's no way to do this -- you have to write some code that does it by hand.
I would recommend multiplying the value by 100, casting to integer, and printing the digits with separators as needed -- it's much easier to handle individual digits on an integer.
The following code, for instance, will fill a char * buffer with the string representation of the value you have:
void formatString (double number, char * buffer) {
if (number < 0) {
*buffer = '-';
formatString(number, buffer + 1);
return;
}
unsigned long long num = (unsigned long long) (number * 100);
unsigned long long x; // temporary storage for counting the digits
unsigned char digits;
for (x = num / 1000, digits = 1; x; digits ++, x /= 10);
// counts the digits, also ensures that there's at least one digit
unsigned char pos; // digit position
for (pos = 1, x = 100; pos < digits; pos ++, x *= 10);
// reuses x as a value for extracting the digit in the needed position;
char * current = buffer;
for (pos = digits; pos; pos --) {
*(current ++) = 48 + (num / x);
// remember 48 + digit gives the ASCII for the digit
if (((pos % 3) == 1) && (pos > 1)) *(current ++) = ',';
num %= x;
x /= 10;
}
*(current ++) = '.';
*(current ++) = 48 + num / 10;
*(current ++) = 48 + num % 10;
*current = 0;
}

Resources