/* lqunoverlap.c -- Copyright 1995, 1996 Liam R. E. Quin.
 * All Rights Reserved.
 * This code is NOT in the public domain.
 * See the file COPYRIGHT for full details.
 *
 * $Id: lqunoverlap.c,v 1.2 1996/07/01 21:37:26 lee Exp $
 *
 * lqunoverlap, part of Liam Quin's text retrieval package...
 *
 * coalesce overlapping matches in input.
 *
 */

#include "error.h"
#include "globals.h" /* defines and declarations for database filenames */

#include <stdio.h> /* stderr, also for fileinfo.h */
#include <sys/types.h>
#ifdef HAVE_SYSV_FCNTL_H
# include <sys/file.h>
#endif
#ifdef HAVE_FCNTL_H
#include <fcntl.h>
#endif
#include <malloc.h>
#ifdef HAVE_STRING_H
# include <string.h>
#else
# include <strings.h>
#endif

#include "emalloc.h" /* for efree() */
#include "fileinfo.h" /* for wordinfo.h */
#include "wordinfo.h"
#include "pblock.h"
#include "phrase.h"
#include "lqutil.h"
#include "lqtrace.h"
#include "liblqtext.h"


/** functions used before they're defined within this file: **/


/** **/

static char *Revision = "@(#) $Id: lqunoverlap.c,v 1.2 1996/07/01 21:37:26 lee Exp $";

char *progname = "$Revision: 1.2 $";

int SilentMode = 0; /* don't print matches if set to one */

static int MinimumNumberOfWordsInRankedPhrase = 3;

typedef struct s_OneFile {
    char *Name;
    struct s_OneFile *Next;
} t_OneFile;

static t_OneFile *FilesToProcess;
static t_OneFile **LastFile;

static void
AddOneFile(theName)
    char *theName;
{
    t_OneFile *F;

    F = (t_OneFile *) emalloc("remembering file name", sizeof(t_OneFile));
    F->Name = theName;
    F->Next = (t_OneFile *) 0;

    if (!LastFile) {
	LastFile = &FilesToProcess;
    }

    while (*LastFile) {
	LastFile = &(*LastFile)->Next;
    }

    *LastFile = F;
    LastFile = &F->Next;
}

static void ProcessFiles();

static void AddOneMatchString(
#ifdef HAVE_PROTO
    char *theString
#endif
);

int
main(argc, argv)
    int argc;
    char *argv[];
{
    extern int optind, getopt();
    extern char *optarg;
    int ch;
    int ErrorFlag = 0;
    char *InputFile = 0;

    progname = argv[0];

    LQT_InitFromArgv(argc, argv);

    while ((ch = getopt(argc, argv, "Zz:f:lNr:xVv")) != EOF) {
	switch (ch) {
	case 'z':
	case 'Z':
	    break; /* done by LQT_InitFromArgv(); */
	case 'V':
	    fprintf(stderr, "%s version %s\n", progname, Revision);
	    break;
	case 'f':
	    AddOneFile(optarg);
	    break;
	case 'l':
	    break; /* list mode is the default */
	case 'r':
	    /* ignored for compat. with lqrank*/
	    break;
	case 'x':
	    ErrorFlag = (-1);
	    break;
	case '?':
	    ErrorFlag = 1;
	}
    }

    if (ErrorFlag) {
	fprintf(stderr, "Usage: %s [options] [filename [...]]\n", progname);
	fprintf(stderr, "%s: options are:\n", progname);
	fputs("-f file	-- the named file contains matches to print\n\
\n", stderr);
	LQT_PrintDefaultUsage();
	exit( ErrorFlag > 0 ? 1 : 0); /* 0 means -x was used */
    }

    while (optind < argc) {
	AddOneFile(argv[optind]);
	++optind;
    }

    if (!FilesToProcess) {
	AddOneFile("-");
    }

    ProcessFiles();
}

#define STDIN_NAME "[standard input]"

static void
ProcessFiles()
{
    t_OneFile *thisFile;

    for (thisFile = FilesToProcess; thisFile; thisFile = thisFile->Next) {
	FILE *f;
	int NeedClose;

	if (STREQ(InputFile, "-")) {
	    f = stdin;
	    NeedClose = 0;
	    thisFile->Name = STDIN_NAME; /* for better error messages */
	} else {
	    f = LQU_fEopen(E_FATAL|E_SYS, InputFile, "List of phrases", "r");
	    NeedClose = 1;
	}

	while (LQU_fReadLine(f, &theLine, LQUF_NORMAL) >= 0) {
	    if (theLine && *theLine) {
		AddOneMatchString(theLine);
	    }
	}

	if (NeedClose) {
	    if (fclose(f) == EOF) {
		Error(E_WARN|E_SYSTEM,
		    "couldn't close file \"%s\"", thisFile->Name
		);
	    }
	}
    }
}

#define MATCHES_PER_LUMP 30

typedef struct s_MatchLump {
    t_MatchStart Matches[MATCHES_PER_LUMP];
    struct s_MatchLump *Next;
    short MatchesUsed;
} t_MatchLump;

typedef struct s_ResultsForOneFile {
    t_FID FID;
    t_MatchLump *Lumps;
    struct s_ResultsForOneFile *Next;
} t_ResultsForOneFile;

static void
AddOneMatchString(theString)
    char *theString;
{
    t_MatchStart *Start;

    /* convert the string to a match */

    /* find the right file */

    /* see if the match overlaps an existing one */
	/* yes -- combine them */
	/* no -- allocate a new match and save it */

}