/* lqphrase.c -- Copyright 1989, 1990, 1994, 1995 Liam R. E. Quin.
 * All Rights Reserved.
 * This code is NOT in the public domain.
 * See the file COPYRIGHT for full details.
 *
 * $Id: lqphrase.c,v 1.12 2001/05/31 03:50:13 liam Exp $
 *
 * lqphrase, part of Liam Quin's text retrieval package...
 *
 * lqphrase is intended to be an example of one way to use the programming
 * interface to lq-text.
 *
 * The idea is quite simple:
 * Simply take a phrase p, held in a string (char *p), and call
 *	t_Phrase *Phrase = LQT_StringToPhrase(p);
 * The result, if not null, contains only one interesting thing at this
 * point:
 *	Phrase->ModifiedString
 * is the canonical version of p -- with common and short words removed.
 * for example,
 *	p = "The boy sat down in His Boat and playd with his toes.";
 * might result in Phrase->ModifiedString containing
 *	"[*the] boy sat down [in] [*his] boat [*and] [?playd] with [*his] toe"
 * Common words are marked with a *, and unknown words with ?.
 * [NOTE: modifiedPhrase is no longer be supported]
 *
 * An attempt may have been made to reduce plurals.
 *
 * Since this phrase contains a word not in the database (playd), it will
 * never match anything.  As a result, it is a good idea to print this string
 * (possibly massaging it first) so users can see what is going on.  If you
 * have it, the curses-based "lqtext" does this.
 *
 * If we change "playd" to "played", the above string is equivalent to
 *	"[*the] boy sat down [xx] [*the] boat [*the] played with [*the] toe"
 * In other words, all common words are equivalent.  The package remembers
 * that one or more common words were skipped, and also that one or more
 * lumps of letters too small to make up a word were skipped.
 * Compare also h/wordrules.h for more info, and liblqtext/Phrase.c for
 * the code that actually does the matching.
 *
 * Now, having sorted that out, we have our canonical string (and lots of
 * other things) in Phrase, so we can now call
 *	LQT_MakeMatches(Phrase);
 * This will return the number of matches (*NOT* the number of files) for
 * the given ModifiedPhrase in the database.
 * This can take several seconds, so again, it can be worth printing out
 * the modified string as soon as it is available, so the user is looking at
 * that whilst LQT_MakeMatches is working!
 *
 * Now we have done LQT_MakeMatches, we can march along the linked list of
 * pointers to linked lists of arrays of matches.  See MatchOnePhrase() below.
 *
 * As an optimisation, LQT_MakeMatchesWhere() lets us call a function to
 * print (in out casr) each match as it is found.  This is slower overall,
 * because of the extra function calls, but not much slower, and the results
 * start to appear much sooner, so it feels much faster.
 *
 * Now, each match currently gives us
 * t_FID FID; Files are numbered from 1 in the database
 * unsigned long BlockInFile; -- the block in the file
 * unsigned char WordInBlock; -- the word in the block
 * unsigned char StuffBefore; -- the amount of leading garbage
 * unsigned char Flags, including (see wordrules.h):
 *
 * WPF_WASPLURAL		The word...  ended in s
 * WPF_UPPERCASE		...Started with a capital letter
 * WPF_POSSESSIVE		...ended in 's
 * WPF_ENDEDINING		...ended in ing
 * WPF_LASTWASCOMMON	the previous word was common
 * WPF_LASTHADLETTERS	we skipped some letters to get here
 * WPF_LASTINBLOCK	I'm the last word in this block
 *
 */

#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
#ifdef HAVE_STRING_H
# include <string.h>
#else
# include <strings.h>
#endif

#ifdef HAVE_STDLIB_H
# include <stdlib.h>
#else
# include <malloc.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 "liblqtext.h"
#include "lqtrace.h"

/** System calls and functions... **/
/** Unix system calls used in this file: **/
extern void exit();

/** Unix Library Functions used: **/
/** lqtext library functions: **/

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

PRIVATE void MatchOnePhrase(
#ifdef HAVE_PROTO
    t_LQTEXT_Database *db,
    char *Phrase
#endif
);

/** **/

static char *Revision = "@(#) $Id: lqphrase.c,v 1.12 2001/05/31 03:50:13 liam Exp $";

char *progname = "lqphrase";

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

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


    progname = argv[0];

    Options = LQT_InitFromArgv(argc, argv);

    while ((ch = getopt(argc, argv, "Zz:af:hNpr:slxVv")) != 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':
	    InputFile = optarg;
	    break;
	case 'l':
	    break; /* list mode is the default */
	case 'r':
	    /* ignored for compat. with lqrank*/
	    break;
	case 's':
	    SilentMode = 1;
	    break;
	case 'x':
	    ErrorFlag = (-1);
	    break;
	case '?':
	    ErrorFlag = 1;
	}
    }

    if (ErrorFlag) {
	fprintf(stderr, "Usage: %s [options] \"phrase\" [...]\n", progname);
	fprintf(stderr, "%s: options are:\n", progname);
	fputs("\
	-l	-- list mode, suitable for lqshow (the default)\n\
	-s	-- silent mode; exit status indicates success of matching\n\
\n", stderr);

	LQT_PrintDefaultUsage(Options);
	exit( ErrorFlag > 0 ? 1 : 0); /* 0 means -x was used */
    }
    
    db = LQT_OpenDatabase(Options, O_RDONLY, 0);
    if (!db || LQT_ObtainReadOnlyAccess(db) < 0) {
	Error(E_FATAL, "couldn't open lq-text database");
    }

    if (InputFile) {
	FILE *f;
	char NeedClose = 1;
	char *theLine;

	if (STREQ(InputFile, "-")) {
	    f = stdin;
	    NeedClose = 0;
	} else {
	    f = LQU_fEopen(E_FATAL|E_SYS, InputFile, "List of phrases", "r");
	}

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

	if (NeedClose) {
	    (void) fclose(f);
	}
    }

    while (optind < argc) {
	MatchOnePhrase(db, argv[optind++]);
    }

    if (SilentMode) {
	/* if we got to here we didn't find anything */
	exit(1);
    }
    return 0;
}

static int gotOne = 0;

static int
ExitSilently(db, Phrase, Match)
    t_LQTEXT_Database *db;
    t_Phrase *Phrase;
    t_Match *Match;
{
    ++gotOne;
    return LQMATCH_QUIT;
}

typedef struct s_Rank {
    t_FID FID;
    unsigned long Sum;
} t_Rank;

PRIVATE void
MatchOnePhrase(db, Phrase)
    t_LQTEXT_Database *db;
    char *Phrase;
{
    t_Phrase *P;

    if (!Phrase || !*Phrase) {
	/* ignore an empty phrase */
	return;
    }

    if ((P = LQT_StringToPhrase(db, Phrase)) == (t_Phrase *) 0) return;

    if (SilentMode) {
	(void) LQT_MakeMatchesWhere(db, P, ExitSilently);
	if (gotOne) {
	    exit(0);
	}
    } else {
	(void) LQT_MakeMatchesWhere(db, P, LQT_PrintAndRejectOneMatch);
    }
    LQT_DestroyPhrase(db, P);
}