/* letters.c */

/* This program performs a case-insensitive analysis of frequency of occurrence in a
 * text file.  The analysis ignores all non-alphabetic characters.
 *
 * Input is from stdin; Output is to stdout; Error is to stderr.
 *
 * One optional command line parameter allows specification of the floor for
 * printing out double and triple letter occurrences.  Occurrence statistics less
 * than the floor will not be displayed except for single letters of which all 26
 * are printed.  The default floor (used when none is specified) is given by DEFFL.
*/

#include <stdio.h>

#define LET 26
#define DEFFL 0.005  /* Default floor of 0.5% */

/* Letter occurrence counters */
static int singles[LET] = {0};
static int doubles[LET][LET] = {0};
static int triples[LET][LET][LET] = {0};

int main(int argc, char **argv)
{
	double lowperc = DEFFL;	/* floor for printing doubles and triples */
	int prev, last, next;	/* characters read */
	int i, j, k;		/* loop counters */
	int letcount;		/* number of letters */
	double perc;		/* intermediate temporary storage of percentage */

	if( argc == 2 )
		sscanf(argv[1], "%lf", &lowperc);  
	else if( argc > 2 )
	{
		fprintf(stderr, "\nUsage: %s [floor]\n", argv[0]);
		return 1;
	}  /* end if */

	/* Process first letter */
firstchar:
	if( (prev = tolower(getchar())) == EOF )
	{
		fprintf(stderr, "\nThe file contains no letters.");
		return 1;
	}
	else if( (prev >= (int)'a') && (prev <= (int)'z') )
		++singles[prev -= (int)'a'];
	else
		goto firstchar;

	/* Process second letter */
secondchar:
	if( (last = tolower(getchar())) == EOF )
	{
		fprintf(stderr, "\nThe file contains only one letter: %c",
			(char)(prev + (int)'a'));
		return 1;
	}
	else if( (last >= (int)'a') && (last <= (int)'z') )
	{
		++doubles[prev][last -= (int)'a'];
		++singles[last];
	}
	else
		goto secondchar;

	/* Process rest of input */
	letcount = 2;
	while( (next = tolower(getchar())) != EOF )
	{
		if( (next < (int)'a') || (next > (int)'z') )
			continue;
		/* process this character */
		++singles[next -= (int)'a'];
		++doubles[last][next];
		++triples[prev][last][next];
		prev = last;
		last = next;
		++letcount;
	}  /* end while */

	/* Print stats */
	printf("\nThe file contains %d letters.", letcount);
	printf("\n\nSINGLE LETTERS:\n");
	printf("----------------------------------------------------------------------------");
	for( i = 0; i < LET; i++ )
		printf("\n%c:  %lf", (char)(i + (int)'a'), 
			(double)singles[i]/letcount);
	printf("\n\nDOUBLE LETTERS:\n");
	printf("----------------------------------------------------------------------------");
	for( i = 0; i < LET; i++ )
		for( j = 0; j < LET; j++ )
			if( (perc = (double)doubles[i][j]/(letcount-1)) >= lowperc )
				printf("\n%c%c:  %lf", (char)(i + (int)'a'), 
					(char)(j + (int)'a'), perc); 
	printf("\n\nTRIPLE LETTERS:\n");
	printf("----------------------------------------------------------------------------");
	for( i = 0; i < LET; i++ )
		for( j = 0; j < LET; j++ )
			for( k = 0; k < LET; k++ )
				if( (perc = (double)triples[i][j][k]/(letcount-2)) >= lowperc )
					printf("\n%c%c%c:  %lf", (char)(i + (int)'a'), 
						(char)(j + (int)'a'), (char)(k + (int)'a'),
						 perc); 
	return 0;
}