#include <stdio.h>
#include <strings.h>
#include <string.h>
#include "efc_structs.h"

#define The_Stars_Twinkle 1

int	cur_len,	/* length of current word in input stream */
	wd_len,		/* length of current target word */
	tok_high = -1,	/* most tokens used to date */
	new_wd;		/* flag used so that statistics are gathered only
			   on first token follower to match */

struct stats wd_info[20];	/* structure to store EFC statistics */

float WDS = 0;	/* total words sampled */

float wd_occ,			/* relative occurrence of target word */
	wd_cur_match = 0.0,	/* number of matches against current target */
	matching_tok[5*TOK_NO+1],	/* stats on tok fol activation */
	real_tok[TOK_NO+1];	/* statistics on physical token follower use */

/**************************************************************************

Routine: main

Purpose: Check for proper parameter passing to program and then
	call top-level EFC simulator routine.

Inputs: data base containing words and their relative frequencies of
	occurrence.

Outputs: none.

Other Side Effects: none.

**************************************************************************/

main(argc, argv)
int argc;
char *argv[];
{
	if (argc != 2)
		error("Usage: efc wordfile", NULL);
	efc(argv[1]);
}

/**************************************************************************

Subroutine: efc

Purpose: Simulate one string matcher of an Electronic Filing Cabinet (EFC)
	system using a data base of words. The data base is formatted in two
	columns where the first is made up of words and the second is made up
	their relative frequencies of occurrence (integers). The sum of these
	frequencies represents the total number of words sampled. For the
	simulation, every word is compared to every other word in the data base.

Inputs: pointer to data base of words.

Outputs: Statistics of the simulation.

**************************************************************************/

efc(fname)
char *fname;
{
	struct str_matcher ss[SS_NO];
	struct state *st_tbl, *make_state_table(), *get_next_state();
	char wd[50];
	int ss_result, i, j, ta, match, newest_tok;
	float occ;
	FILE *fptr, *fopen();
	void rewind(), deactivate_toks();

	for (i = 0; i < 20 ;i++)
		init_stats(&wd_info[i]);
	st_tbl = make_state_table();
	if ((fptr= fopen(fname, "r")) == NULL)
		error("efc: can't open %s", fname);
	while (The_Stars_Twinkle) {
	init_ss_matcher(fname, &ss[0]);
	while (getwd(fptr, wd, &occ, 1) != EOF ) {
		newest_tok = match = 0;
		new_wd = 1;
		for (i = 0; !match && (i < cur_len); i++) {
			activate_tok(st_tbl, &(ss[0]), &newest_tok);
			for (j = 0; !match && (j < TOK_NO); j++)
				if (ss[0].tok[j].active) {
					ss_result =
						ss_match(&ss[0], wd[i],
						&(ss[0].tok[j]));
					ss[0].tok[j].current_st =
						get_next_state(ss[0].tok[j].current_st,
						ss_result);
					do_state_actions(&(ss[0].tok[j]));
					match = check_tok(&(ss[0].tok[j]),
						ss[0].str_len, occ, wd, j);
				}
		}
		if (match)
			wd_info[wd_len].mismatches += wd_occ * wd_cur_match;
		if (tok_high >= 0) {
			wd_info[wd_len].toks_used[tok_high] += occ * wd_occ;
		}
		deactivate_toks(&ss[0]);
	}
	rewind(fptr);
	}
}

/**************************************************************************

Subroutine: getwd

Purpose: Retrieve a word and its relative frequency of occurrence
	from the word data base used in the simulation.

Inputs: fptr	- pointer to word data base.
	buf 	- where to store newly read word.
	occ	- frequency of occurrence of new word.
	chg_stats - update frequency of occurrence of either
			target word or input word.

Outputs: none.

Returns: EOF upon end of data base file.

Other Side Effects: input words are truncated to 16 characters.

**************************************************************************/

getwd(fptr, buf, occ, chg_stats)
FILE *fptr;
char *buf;
int chg_stats;
float *occ;
{
	int res, len, i;
	res = fscanf(fptr, "%s %f\n", buf, occ);
	if (res != EOF) {
		len = strlen(buf);
		len = (len > 15) ? 15: len;
		buf[len] = '\0';
		if (chg_stats)
			cur_len = (float)len;
		else
			wd_len = (float)len;
	}
	return(res);
}