histogram.c

#include <cstddef>
#include <cstdint>
#include <cstdio>
#include <cstdlib>
#include <cstring>

size_t
megabytes_size(uint32_t x)
{
  return x * 1024 * 1024;
}

struct occurence_histograms
{
  size_t bits[2];
  size_t nibbles[16];
  size_t bytes[256];
};

void
collect_occurences(uint8_t* data, size_t data_size, occurence_histograms* occurence_histograms)
{
  memset(occurence_histograms->bytes, 0, sizeof occurence_histograms->bytes);
  memset(occurence_histograms->nibbles, 0, sizeof occurence_histograms->nibbles);
  memset(occurence_histograms->bits, 0, sizeof occurence_histograms->bits);
  {
    uint8_t* f = data;
    uint8_t* l = data + data_size;
    while (f != l) {
      uint8_t byte = *f;
      ++occurence_histograms->bytes[byte];
      uint8_t nibble = byte;
      ++occurence_histograms->nibbles[nibble & 0xF];
      nibble >>= 4;
      ++occurence_histograms->nibbles[nibble & 0xF];
      uint8_t bit = byte;
      for (size_t bit_index = 0; bit_index < 8; ++bit_index) {
        ++occurence_histograms->bits[bit & 1];
        bit >>= 1;
      }
      ++f;
    }
  }
}

void
print_occurences(occurence_histograms const* occurence_histograms)
{
    printf("'byte';'occurrence'\n");
    {
      size_t byte = 0;
      while (byte < 256) {
        printf("'%lx';%lu\n", byte, occurence_histograms->bytes[byte]);
        ++byte;
      }
    }
    printf("'nibble';'occurrence'\n");
    {
      size_t nibble = 0;
      while (nibble < 16) {
        printf("'%lx';%lu\n", nibble, occurence_histograms->nibbles[nibble]);
        ++nibble;
      }
    }
    printf("'bits';'occurrence'\n");
    {
      size_t bit = 0;
      while (bit < 2) {
        printf("'%lx';%lu\n", bit, occurence_histograms->bits[bit]);
        ++bit;
      }
    }
}

int
main(int argc, char** argv)
{
  size_t words_size = megabytes_size(64);
  uint8_t *words = (uint8_t*)malloc(words_size);
  occurence_histograms words_histogram;
  collect_occurences(words, words_size, &words_histogram);
  puts("bytes returned by malloc");
  print_occurences(&words_histogram);
  puts("---");
  occurence_histograms histogram_histogram;
  collect_occurences((uint8_t*)&words_histogram, sizeof words_histogram, &histogram_histogram);
  puts("bytes inside the memory histogram");
  print_occurences(&histogram_histogram);
  puts("---");
  puts("bytes inside main function");
  occurence_histograms main_histogram;
  collect_occurences((uint8_t*)
                     main, ((uint8_t*)&&exit)-((uint8_t*)main), &main_histogram);
  print_occurences(&main_histogram);
 exit:
  return 0;
}