/* Frequency count program for CS 638.  Reads standard input and
   reports:

        Number of times each character occurs
        Number of letters (A-Z) and distinct letters
        Entropy of A-Z distribution
        Index of coincidence for A-Z distribution
        Estimate for number of alphabets, based on index of coincidence 

   Eric Bach 10/19/98

*/

#include <stdio.h>

extern double log();
#define LOG2    0.69314718055994530941

#define NUMCHARS 256

#define SPACE   ' '
#define TAB     '\t'
#define NEWLINE '\n'

main()
{
char c;
int F[NUMCHARS];
int i;
int lettercount, distinctlett;
double p, entropy, chi, IC;
double n, alphabets;

/* Read and echo text, and count characters */
for (i=0;i<NUMCHARS;i++) F[i] = 0;
while ( (c = getchar()) != EOF) {
        printf("%c",c);
        F[ (int) c ] ++;
        }
printf("\n");

/* Print out frequency counts */
for (i=0;i<NUMCHARS;i++) {
        if (F[i]) { /* Ad hack method of handling unprintable characters
... */
                if (i == NEWLINE) printf("%d\tnewline\n",F[i]);
                else if (i == SPACE) printf("%d\tspace\n",F[i]);
                else if (i == TAB) printf("%d\ttab\n",F[i]);
                else printf("%d\t%c\n",F[i],(char)i);
                }
        }

/* Count total letters and number of distinct letters */
distinctlett  = 0;
lettercount = 0;
for (i='A';i<='Z';i++) {
        if (F[i]) distinctlett++;
        lettercount += F[i];
        }

/* Evaluate frequency statistics */
entropy = 0;
IC = 0;
for (i='A';i<='Z';i++) {
        if (F[i]) {
                p = (double) F[i]/lettercount;
                entropy += p*log(1/p)/LOG2;
                IC += (double) F[i]*(F[i]-1);
                }
        }
IC = IC/(lettercount*(lettercount-1));

printf("Text contains %d A-Z, %d distinct.\n",
         lettercount, distinctlett);

printf("Entropy for A-Z = %lf bits/letter.\n",entropy);

printf("Index of coincidence for A-Z: %lf\n",IC);

n = lettercount;
alphabets = 0.027*n / ( IC*(n-1) - 0.038*n + 0.065 ); /* from Beker/Piper
*/
printf("Estimated number of alphabets: %lf\n",alphabets);

}