str_conv

Integers and 32-bit floating-point numbers conversion functions to and from strings of binary, decimal and hexadecimal characters.

#include "to_dec_str.h"
#include <Windows.h>

void print_numbers(HANDLE* output, schar8* output_buffer, f32* numbers, u32 numbers_count)
{
  schar8* begin = output_buffer;
  schar8* end   = output_buffer;
  
  for (u32 i = 0; i < numbers_count; i++)
  {
    f32 num          = numbers[i];
    u32 num_str_size = f32_to_dec_str(end, num);
    
    end[num_str_size] = '\n';
    end += num_str_size + 1u;
  }

  u32 char_written_count = (u32)(end - begin);
  WriteFile(output, output_buffer, char_written_count, 0, 0);
}

You can compile this example yourself by running build.bat (requires Visual Studio Build Tools or the Native Desktop workload):

build && build run

Features

• Signed and unsigned integers up to 64-bit to binary, decimal and hexadecimal ASCII strings, including from a pre-determined count of numerals

  • Capital letters are used when writing hexadecimal strings

• 32-bit floating point numbers to binary, decimal and hexadecimal ASCII strings, including from a pre-determined count of numerals or a pre-determined minimum integer part size and fixed fractional part size, with a few particularities:

  • Values outside of [-2^32 + 1, 2^32 - 1] are converted to string -big or big, depending on their sign (see rational in f32_number_to_dec_str comments, in to_dec_str.c)
  • "-qnan", "qnan", "-snan", "snan", "-inf" or "inf" may be output for matching non-number values
  • -0 will be converted to "0"
  • By default, 6 digits past the period are written to the output string. This can be increased up to 9 (see F32_DEC_FRAC_DEFAULT_STR_SIZE and F32_DEC_FRAC_MULT in types_max_str_size.h)
  • When passing sizes for the integer and fractional parts, the integer part is padded withs 0 to the left if necessary, and the fractional part's will be truncated to the maximum allowed size of 9 if it exceeds this value (see F32_DEC_FRAC_MAX_STR_SIZE in types_max_str_size.h)
  • Values are written in full. Scientific notation and other notations are not supported

• Binary, decimal and hexadecimal ASCII strings to signed and unsigned integers up to 64-bit, including from a pre-determined string size

• Binary, decimal and hexadecimal ASCII strings to 32-bit floating point numbers, including from a pre-determined string size, with the following particularities:

  • Values exceeding the bounds of the 32-bit floating point type are not handled
  • Strings -big and big are converted to -2^32 and 2^32, respectively
  • Strings -qnan, qnan, -snan, snan, -inf and inf are converted to their matching bits representation. For +/- qnan, the leftmost, highest mantissa bit is set, while for -/+ snan and -/+ inf, the mantissa is all zeros
  • "-0" will be converted to 0
  • Numbers must start with either - or a numeral, but not with .
  • Scientific notation and other notations are not supported

• A few utilities for counting numerals in a number and qualify a single character

• No allocation, only user-provided memory buffers

• No C runtime, no C standard library

Supported targets

• Windows XP and above
• 64-bit and 32-bit x86 CPU architectures
• MSVC compiler

Repository files

• [from|to]_<base_abbreviation>_str.[h|c]: conversion functions
• utilities.h / utilities.c: numerals counting and identification functions
• types.h: custom typedefs, for personal convenience
• types_max_str_size.h: macros defining the maximum amount of characters written when converting to strings, to make it easier to initialise a destination buffer of sufficient size on the stack for instance
• isa.h: macros to identify which compiler is being used and for which CPU instruction set code is being compiled for

License

The code in this repository is released in the public domain. You can use it with no constraint.