# include "stdio.h"
# include "error.h"
#include "inv.h"

void
main(int argc, char **argv)
{
/* make inverted file indexes.  Reads a stream from mkey which
   gives record pointer items and keys.  Generates a set of files
      a. NHASH pointers to file b.
      b. lists of record numbers.
      c. record pointer items.

   these files are named xxx.ia, xxx.ib, xxx.ic;
   where xxx is taken from arg1.
   If the files exist they are updated.
*/

	FILE * fa, *fb, *fc, *fta, *ftb, *fd;
	int	nhash = 256;
	int	appflg = 1;
	int	keepkey = 0, pipein = 0;
	char	nma[100], nmb[100], nmc[100], com[100], nmd[100];
	char	tmpa[20], tmpb[20], tmpc[20];
	char	*remf = NULL;
	int	chatty = 0, docs, hashes, status;
	long	keys;
	char	*sortdir;

	progname = mkprogname(argv[0]);
	sortdir = "/tmp";
	for (; argc > 1 && argv[1][0] == '-'; argc--, argv++) {
		switch (argv[1][1]) {
		case 'h': /* size of hash table */
			nhash = atoi (argv[1] + 2); 
			break;
		case 'n': /* new, don't append */
			appflg = 0; 
			break;
		case 'a': /* append to old file */
			appflg = 1; 
			break;
		case 'v': /* verbose output */
			chatty = 1; 
			break;
		case 'd': /* keep keys on file .id for check on searching */
			keepkey = 1; 
			break;
		case 'p': /* pipe into sort (saves space, costs time)*/
			pipein = 1; 
			break;
		case 'i': /* input is on file, not stdin */
			close(0);
			if (open(argv[2], 0) != 0)
				err("can't read input %s", argv[2]);
			if (argv[1][2] == 'u') /* remove */
				remf = argv[2];
			argc--; 
			argv++;
			break;
		case 'T':
			sortdir = &argv[1][2];
			break;
		}
	}

	strcpy(nma, argc >= 2 ? argv[1] : "Index");
	strcpy(nmb, nma);
	strcpy(nmc, nma);
	strcpy(nmd, nma);
	strcat(nma, ".ia");
	strcat(nmb, ".ib");
	strcat(nmc, ".ic");
	strcat(nmd, ".id");

	sprintf(tmpa, "junk%di", getpid());
	if (pipein) {
		sprintf(com, "/bin/sort -T %s -o %s", sortdir, tmpa);
		if ((fta = popen(com, "w")) == NULL)
			err("can't pipe into %s", com);
	} else
		fta = efopen(tmpa, "w");	/* use tmp file */
	if (appflg) {
		if ((fb = fopen(nmb, "r")) != NULL) {
			sprintf(tmpb, "junk%dj", getpid());
			ftb = efopen(tmpb, "w");
			nhash = recopy(ftb, fb, fopen(nma, "r"), nhash);
			fclose(ftb);
			fclose(fb);
		} else
			appflg = 0;
	}
	fc = efopen(nmc,  appflg ? "a" : "w");
	fd = keepkey? efopen(nmd, "w") : 0;
	docs = newkeys(fta, stdin, fc, nhash, fd);
	fclose(stdin);
	if (remf != NULL)
		remove(remf);
	if (pipein) {
		status = pclose(fta);
		if (status != 0)
			err("pipe into sort failed");
	} else {
		fclose(fta);
		sprintf(com, "/bin/sort -T %s %s -o %s", sortdir, tmpa, tmpa);
		system(com);
	}

	if (appflg) {
		sprintf(tmpc, "junk%dk", getpid());
		rename(tmpa, tmpc);
		sprintf(com, "/bin/sort -T %s  -m %s %s -o %s", sortdir, tmpb, tmpc, tmpa);
		system(com);
	}
	fta = efopen(tmpa, "r");
	fa = efopen(nma, "w");
	fb = efopen(nmb, "w");
	whash(fta, fa, fb, nhash, &keys, &hashes);
	fclose(fta);
# ifndef D1
	remove(tmpa);
# endif
	if (appflg) {
		remove(tmpb);
		remove(tmpc);
	}
	if (chatty)
		printf ("%ld key occurrences,  %d hashes, %d docs\n",
		    keys, hashes, docs);
	exit(0);
}