/*
Program: MERGEALL.C
Author : Kim Moser
Date   : 19 January, 1991
System : IBM PC / Borland Turbo C 2.0
Descrip: Acts on files of words, "a.txt", "b.txt", "c.txt" ... "z.txt".
	 First searches for words of length 1, then 2, then 3, etc.
	 (Stops when no words found in pass.)
	 For each pass, writes (sorted) words found from each file
	 to file "n.out" where 'n' is the length of the words in
	 that file.

	 E.g. given files:

	 A.TXT:		B.TXT		C.TXT

	    A              BE           C
	    AARON          BORE         CAT
	    APPLE          BARN         CONE
	    ARGH                        CY

	 Produces files:

	 1.OUT		2.OUT		3.OUT		4.OUT		5.OUT

	    A              BE           CAT                ARGH            AARON
	    C              CY                              BORE            APPLE
							   BARN
							   CONE

Usage  : MERGEALL
*/

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

#define MAX_LEN		28
/* Max word len */

static int LENGTH;
/* Current length being worked on */

#define BUF_SIZE	100

static char words[26][BUF_SIZE];
static long int fps[26];  /* Last file positions */

static FILE *OUTFILE;
static char OUTFILENAME[40];

static void initvars(void);
static void initvars(void)
/* Initialize variables */
{
	int i;

	for (i=0; i<26; i++) {
		words[i][0] = '\0';
		fps[i] = 0;
	}
}

static void readnext(int i);
static void readnext(int i)
/* Reads next word of length LENGTH from file "<i+'A'>.txt". */
{
	FILE *fp;
	char *fname = "?.txt";

	if (fps[i] == -1L) return;

	fname[0] = i + 'A';
	if ((fp = fopen(fname, "r")) == NULL) {
		fprintf(stderr, "fopen() failed for file '%s'.\n", fname);
		exit(-1);
	}

	if (fseek(fp, fps[i], SEEK_SET)) {
		fprintf(stderr, "fseek() failed for file '%s' (pos=%ld).\n", fname, fps[i]);
		exit(-1);
	}

	do {
		if (fgets(words[i], BUF_SIZE, fp) == NULL) {
			words[i][0] = '\0';
			break;
		}
		/* Remove trailing newline: */
		words[i][strlen(words[i])-1] = '\0';
/*		fprintf(stderr, "%s\n", words[i]); */
	} while (strlen(words[i]) != LENGTH);

	if (words[i][0]) {
		if ((fps[i] = ftell(fp)) == -1L) {
			fprintf(stderr, "ftell() failed for '%s'.\n", fname);
			exit(-1);
		}
	} else {
		fps[i] = -1L;
	}

	if (fclose(fp)) {
		fprintf(stderr, "fclose() failed for '%s'.\n", fname);
		exit(-1);
	}
}

static int out(void);
static int out(void)
/* Find "smallest" (alphabetically) word, output it, and read next
   word (of length LENGTH) from file that it came from.

   Returns whether there are any words. */
{
	int i;

	/* Start 'smallest' at first non-empty word: */
	for (i=0; i<26 && !words[i][0]; i++);
	if (i >= 26) return (0);

	fprintf(OUTFILE, "%s\n", words[i]);
	readnext(i);

#if 0
	/* Now find smallest: */
	for (i=smallest+1; i<26; i++) {
		if (words[i][0] && (strcmp(words[i], words[smallest]) < 0)) smallest = i;
	}

	fprintf(OUTFILE, "%s\n", words[smallest]);

	readnext(words[smallest][0]-'A');
#endif

	return (1);
}

void main(void)
{
	int i;
	unsigned long count;

	for (LENGTH=1; LENGTH<MAX_LEN; LENGTH++) {
		fprintf(stderr, "Working on length %d.\n", LENGTH);

		initvars();
		count = 0;
		sprintf(OUTFILENAME, "%d.out", LENGTH);
		if ((OUTFILE = fopen(OUTFILENAME, "w")) == NULL) {
			fprintf(stderr, "fopen failed for OUTFILE (%s).\n", OUTFILENAME);
			exit(-1);
		}

		/* Find first words of length 'LENGTH' in all 26 files: */
		for (i=0; i<26; i++) {
			fprintf(stderr, "%c", i+'A');
			readnext(i);
		}
		fprintf(stderr, "\n");

		while (out()) {
			if (!(++count % 100)) {
				fprintf(stderr, "%lu\r", count);
			}
		}

		if (fclose(OUTFILE)) {
			fprintf(stderr, "fclose() failed for OUTFILE (%s).\n", OUTFILENAME);
			exit(-1);
		}
	}
}