Perform a simple statistical analysis of arbitrary byte streams

character-distribution.c

#include <stdlib.h>
#include <stdint.h>
#include <stdio.h>
#include <ctype.h>
#include <limits.h>
#include <math.h>
#include <sysexits.h>


inline static double square(double x)
{
	return x * x;
}


int main()
{
	static const unsigned distribution_size = 1 << CHAR_BIT;

	int rv = EX_OK;
	uintmax_t *distribution = calloc(distribution_size, sizeof(*distribution));

	{
		int c;
		while ((c = getchar()) != EOF)
			distribution[c]++;

		if (ferror(stdin)) {
			perror("I/O error on standard input");
			rv = EX_IOERR;
		}
	}

	uintmax_t sum = 0;
	for (unsigned i = 0; i != distribution_size; i++)
		sum += distribution[i];
	double avg = (double) sum / distribution_size;

	double var_accum = 0.0;
	for (unsigned i = 0; i != distribution_size; i++)
	{
		const uintmax_t x = distribution[i];

		printf("'%c' (%02X): %20ju", isprint((int) i) ? i : ' ', i, x);
		if (x != 0) {
			var_accum += square((double) x - avg);
			printf(" (%+.2e %%)\n", ((double) x / avg - 1.0) * 100.0);
		} else {
			var_accum += square(avg);
			putchar('\n');
		}
	}

	double stdev = sqrt(var_accum / distribution_size);
	double varcoeff = stdev / avg;
	printf(
		"total: %ju\n"
		"average: %e\n"
		"standard deviation: %e\n"
		"variation coefficient: %e\n",
		sum, avg, stdev, varcoeff);

	free(distribution);
	return rv;
}