ascii-histogram.c

#include <stdio.h>
#include <stdlib.h>
#include <strings.h>

#define BUFSIZE (1023 * 1024 * 4)
#define HISTSIZE (0x7f - ' ')

typedef struct hist_entry_s {
  uint64_t count;
  char codepoint;
} hist_entry;

int32_t fold_case = 0;
int32_t sort_direction = 1;
int32_t do_sort = 1;
int32_t print_empty_sets = 0;

int32_t compare_hist_entry(hist_entry * a, hist_entry * b) {
  uint64_t ac = a->count;
  uint64_t bc = b->count;

  return (ac == bc ? 0 : ac < bc ? -1 : 1) * sort_direction * do_sort;
}

int32_t main(int argc, char * argv[]) {
  size_t r;
  char buf[BUFSIZE];
  hist_entry h[HISTSIZE];
  int32_t i;

  for(i = 0; i < argc; i++) {
    size_t l = strlen(argv[i]);

    if(l > 1 && argv[i][0] == '-')
      while(--l)
        switch (argv[i][l]) {
        case 'c':
          fold_case = 1;
          break;
        case 'r':
          sort_direction = -1;
          break;
        case 'n':
          do_sort = 0;
          break;
        case 'e':
          print_empty_sets = 1;
          break;
        }
  }

  for(i = ' '; i < 0x7f; i++)
    h[i - ' '].codepoint = i;

  while((r = fread((void *) buf, 1, BUFSIZE, stdin)))
    while (r--) {
      char v = buf[r];
      if(v >= ' ' && v < 0x7f) {
        if(fold_case && (v >= 'A' && v <= 'Z')) v += 32;
        h[v - ' '].count++;
      }
    }

  if (ferror(stdin)) {
    fprintf(stderr, "Something went horribly wrong with reading stdin.");
    exit(1);
  }

  qsort(h, HISTSIZE, sizeof(hist_entry),
        (int (*)(const void *, const void *))compare_hist_entry);

  for (i = 0; i < HISTSIZE; i++)
    if (print_empty_sets || h[i].count != 0)
      printf(" %c : %llu\n", h[i].codepoint, h[i].count);

  exit(0);
}