X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=rset%2Frsrel.c;h=a02fb232226c7fe3af03dcef3a7cfce007a91087;hb=d79f0901b184c1bcf383be8500aee3cfd197ca1d;hp=0d522500539901ee08e15a923b840670f1687e30;hpb=40ca1d08c1d83b92e0b90951d918b252c61c63b2;p=idzebra-moved-to-github.git

diff --git a/rset/rsrel.c b/rset/rsrel.c
index 0d52250..a02fb23 100644
--- a/rset/rsrel.c
+++ b/rset/rsrel.c
@@ -1,10 +1,71 @@
 /*
- * Copyright (C) 1994-1995, Index Data I/S 
+ * Copyright (C) 1994-1998, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: rsrel.c,v $
- * Revision 1.5  1995-10-06 14:38:06  adam
+ * Revision 1.23  1998-01-07 13:53:41  adam
+ * Queries using simple ranked operands returns right number of hits.
+ *
+ * Revision 1.22  1997/12/18 10:54:25  adam
+ * New method result set method rs_hits that returns the number of
+ * hits in result-set (if known). The ranked result set returns real
+ * number of hits but only when not combined with other operands.
+ *
+ * Revision 1.21  1997/11/18 10:05:08  adam
+ * Changed character map facility so that admin can specify character
+ * mapping files for each register type, w, p, etc.
+ *
+ * Revision 1.20  1997/10/31 12:37:55  adam
+ * Code calls xfree() instead of free().
+ *
+ * Revision 1.19  1997/10/01 11:44:06  adam
+ * Small improvement of new ranking.
+ *
+ * Revision 1.18  1997/09/24 13:36:41  adam
+ * More work on new ranking algorithm.
+ *
+ * Revision 1.17  1997/09/22 12:39:07  adam
+ * Added get_pos method for the ranked result sets.
+ *
+ * Revision 1.16  1997/09/17 12:19:23  adam
+ * Zebra version corresponds to YAZ version 1.4.
+ * Changed Zebra server so that it doesn't depend on global common_resource.
+ *
+ * Revision 1.15  1997/09/09 13:38:16  adam
+ * Partial port to WIN95/NT.
+ *
+ * Revision 1.14  1996/11/08 11:15:58  adam
+ * Compressed isam fully supported.
+ *
+ * Revision 1.13  1996/10/29 13:55:26  adam
+ * Include of zebrautl.h instead of alexutil.h.
+ *
+ * Revision 1.12  1996/10/08 13:00:40  adam
+ * Bug fix: result sets with ranked operands in boolean operations weren't
+ * sorted.
+ *
+ * Revision 1.11  1996/10/07  16:05:29  quinn
+ * Work.
+ *
+ * Revision 1.9  1995/12/11  09:15:26  adam
+ * New set types: sand/sor/snot - ranked versions of and/or/not in
+ * ranked/semi-ranked result sets.
+ * Note: the snot not finished yet.
+ * New rset member: flag.
+ * Bug fix: r_delete in rsrel.c did free bad memory block.
+ *
+ * Revision 1.8  1995/12/05  11:25:45  adam
+ * Doesn't include math.h.
+ *
+ * Revision 1.7  1995/10/12  12:41:57  adam
+ * Private info (buf) moved from struct rset_control to struct rset.
+ * Bug fixes in relevance.
+ *
+ * Revision 1.6  1995/10/10  14:00:04  adam
+ * Function rset_open changed its wflag parameter to general flags.
+ *
+ * Revision 1.5  1995/10/06  14:38:06  adam
  * New result set method: r_score.
  * Local no (sysno) and score is transferred to retrieveCtrl.
  *
@@ -24,33 +85,36 @@
 
 #include <stdio.h>
 #include <stdlib.h>
-#include <math.h>
+#include <string.h>
 #include <assert.h>
 
 #include <isam.h>
+#include <isamc.h>
 #include <rsrel.h>
-#include <alexutil.h>
+#include <zebrautl.h>
 
-static rset_control *r_create(const struct rset_control *sel, void *parms);
-static RSFD r_open (rset_control *ct, int wflag);
+static void *r_create(const struct rset_control *sel, void *parms,
+                      int *flags);
+static RSFD r_open (RSET ct, int flag);
 static void r_close (RSFD rfd);
-static void r_delete (rset_control *ct);
+static void r_delete (RSET ct);
 static void r_rewind (RSFD rfd);
-static int r_count (rset_control *ct);
+static int r_count (RSET ct);
+static int r_hits (RSET ct, void *oi);
 static int r_read (RSFD rfd, void *buf);
 static int r_write (RSFD rfd, const void *buf);
 static int r_score (RSFD rfd, int *score);
 
 static const rset_control control = 
 {
-    "relevance set type",
-    0,
+    "relevance",
     r_create,
     r_open,
     r_close,
     r_delete,
     r_rewind,
     r_count,
+    r_hits,
     r_read,
     r_write,
     r_score
@@ -62,16 +126,28 @@ struct rset_rel_info {
     int     key_size;
     int     max_rec;
     int     no_rec;
+    int     hits;                       /* hits count */
     int     (*cmp)(const void *p1, const void *p2);
-    char    *key_buf;
-    float   *score_buf;
-    int     *sort_idx;
-
+    int     (*get_pos)(const void *p);
+    char    *key_buf;                   /* key buffer */
+    float   *score_buf;                 /* score buffer */
+    int     *sort_idx;                  /* score sorted index */
+    int     *sysno_idx;                 /* sysno sorted index (ring buffer) */
+    int     no_isam_positions;
+    ISAM    is;
+    ISAMC   isc;
+    ISAM_P  *isam_positions;
+    int     no_terms;
+    int     *term_no;
+
+    int     method;
     struct rset_rel_rfd *rfd_list;
 };
 
 struct rset_rel_rfd {
+    int     last_read_pos;
     int     position;
+    int     flag;
     struct rset_rel_rfd *next;
     struct rset_rel_info *info;
 };
@@ -80,7 +156,7 @@ static void add_rec (struct rset_rel_info *info, double score, void *key)
 {
     int idx, i, j;
 
-    logf (LOG_DEBUG, "add %f", score);
+    (info->hits)++;
     for (i = 0; i<info->no_rec; i++)
     {
         idx = info->sort_idx[i];
@@ -88,99 +164,233 @@ static void add_rec (struct rset_rel_info *info, double score, void *key)
             break;
     }
     if (info->no_rec < info->max_rec)
-    {
+    {                                        /* there is room for this entry */
         for (j = info->no_rec; j > i; --j)
             info->sort_idx[j] = info->sort_idx[j-1];
         idx = info->sort_idx[j] = info->no_rec;
         ++(info->no_rec);
     }
     else if (i == 0)
-        return;
+        return;                              /* score too low */
     else
     {
-        idx = info->sort_idx[0];
+        idx = info->sort_idx[0];             /* remove this entry */
 
         --i;
-        for (j = 0; j < i; ++j)
+        for (j = 0; j < i; ++j)              /* make room */
             info->sort_idx[j] = info->sort_idx[j+1];
-        info->sort_idx[j] = idx;
+        info->sort_idx[j] = idx;             /* allocate sort entry */
     }
     memcpy (info->key_buf + idx*info->key_size, key, info->key_size);
     info->score_buf[idx] = score;
 }
 
-static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
+
+static struct rset_rel_info *qsort_info;
+
+static int qcomp (const void *p1, const void *p2)
+{
+    int i1 = *(int*) p1;
+    int i2 = *(int*) p2;
+
+    return qsort_info->cmp (qsort_info->key_buf + i1*qsort_info->key_size,
+                            qsort_info->key_buf + i2*qsort_info->key_size);
+}
+
+#define NEW_RANKING 0
+
+#define SCORE_SHOW 0.0                       /* base score for showing up */
+#define SCORE_COOC 0.3                       /* component dependent on co-oc */
+#define SCORE_DYN  (1-(SCORE_SHOW+SCORE_COOC)) /* dynamic component of score */
+
+static void relevance (struct rset_rel_info *info)
 {
     char **isam_buf;
     char *isam_tmp_buf;
     int  *isam_r;
-    int  *max_tf;
-    ISPT *isam_pt;
-    double *wgt;
+    int  *max_tf, *tf;
+
+    int  *pos_tf = NULL;
+    int score_sum = 0;
+    int no_occur = 0;
+    char *isam_prev_buf = NULL;
+    int fact1, fact2;
+
+    ISPT *isam_pt = NULL;
+    ISAMC_PP *isamc_pp = NULL;
     int i;
 
     logf (LOG_DEBUG, "relevance");
-    isam_buf = xmalloc (parms->no_isam_positions * sizeof(*isam_buf));
-    isam_r = xmalloc (sizeof (*isam_r) * parms->no_isam_positions);
-    isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
+    isam_buf = xmalloc (info->no_isam_positions * sizeof(*isam_buf));
+    isam_r = xmalloc (sizeof (*isam_r) * info->no_isam_positions);
+    if (info->is)
+        isam_pt = xmalloc (sizeof (*isam_pt) * info->no_isam_positions);
+    else if (info->isc)
+        isamc_pp = xmalloc (sizeof (*isamc_pp) * info->no_isam_positions);
+    else
+    {
+        logf (LOG_FATAL, "No isamc or isam in rs_rel");
+        abort ();
+    }
     isam_tmp_buf = xmalloc (info->key_size);
-    max_tf = xmalloc (sizeof (*max_tf) * parms->no_isam_positions);
-    wgt = xmalloc (sizeof (*wgt) * parms->no_isam_positions);
+    max_tf = xmalloc (sizeof (*max_tf) * info->no_terms);
+    tf = xmalloc (sizeof (*tf) * info->no_terms);
 
-    for (i = 0; i<parms->no_isam_positions; i++)
+    for (i = 0; i<info->no_terms; i++)
+	max_tf[i] = 0;
+    for (i = 0; i < info->no_isam_positions; i++)
     {
         isam_buf[i] = xmalloc (info->key_size);
-        isam_pt[i] = is_position (parms->is, parms->isam_positions[i]);
-        max_tf [i] = is_numkeys (isam_pt[i]);
-        isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
-        logf (LOG_DEBUG, "max tf %d = %d", i, max_tf[i]);
-    }
-    while (1)
-    {
-        int min = -1, i;
-        double score;
-
-        /* find min with lowest sysno */
-        for (i = 0; i<parms->no_isam_positions; i++)
-            if (isam_r[i] && 
-               (min < 0 || (*parms->cmp)(isam_buf[i], isam_buf[min]) < 1))
-                min = i;
-        if (min < 0)
-            break;
-        memcpy (isam_tmp_buf, isam_buf[min], info->key_size);
-        /* calculate for all with those sysno */
-        for (i = 0; i<parms->no_isam_positions; i++)
+        if (isam_pt)
+        {
+            isam_pt[i] = is_position (info->is, info->isam_positions[i]);
+            max_tf [info->term_no[i]] = is_numkeys (isam_pt[i]);
+            isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
+        } 
+        else if (isamc_pp)
         {
-            int r;
-            
-            if (isam_r[i])
-                r = (*parms->cmp)(isam_buf[i], isam_tmp_buf);
-            else 
-                r = 2;
-            if (r > 1 || r < -1)
-                wgt[i] = 0.0;
-            else
-            {
-                int tf = 0;
-                do
-                {
-                    tf++;
-                    isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
-                } while (isam_r[i] && 
-                         (*parms->cmp)(isam_buf[i], isam_tmp_buf) <= 1);
-                wgt[i] = 0.1+tf*0.9/max_tf[i];
-            }
+            isamc_pp[i] = isc_pp_open (info->isc, info->isam_positions[i]);
+            max_tf [info->term_no[i]] = isc_pp_num (isamc_pp[i]);
+            isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
         }
-        /* calculate relevance value */
-        score = 0.0;
-        for (i = 0; i<parms->no_isam_positions; i++)
-            score += wgt[i];
-        /* if value is in the top score, then save it - don't emit yet */
-        add_rec (info, score, isam_tmp_buf);
+        logf (LOG_DEBUG, "max tf %d = %d", i, max_tf[i]);
     }
-    for (i = 0; i<parms->no_isam_positions; i++)
+    switch (info->method)
+    {
+    case RSREL_METHOD_B:
+	while (1)
+	{
+	    int r, min = -1;
+	    int pos = 0;
+	    for (i = 0; i<info->no_isam_positions; i++)
+		if (isam_r[i] && 
+		    (min < 0 ||
+		     (r = (*info->cmp)(isam_buf[i], isam_buf[min])) < 1))
+		    min = i;
+	    if (!isam_prev_buf)
+	    {
+		pos_tf = xmalloc (sizeof(*pos_tf) * info->no_isam_positions);
+		isam_prev_buf = xmalloc (info->key_size);
+		fact1 = 100000/info->no_isam_positions;
+		fact2 = 100000/
+		    (info->no_isam_positions*info->no_isam_positions);
+		
+		no_occur = score_sum = 0;
+		memcpy (isam_prev_buf, isam_buf[min], info->key_size);
+		for (i = 0; i<info->no_isam_positions; i++)
+		    pos_tf[i] = -10;
+	    }
+	    else if (min < 0 ||
+		     (*info->cmp)(isam_buf[min], isam_prev_buf) > 1)
+	    {
+		logf (LOG_LOG, "final occur = %d ratio=%d",
+		      no_occur, score_sum / no_occur);
+		add_rec (info, score_sum / (10000.0*no_occur), isam_prev_buf);
+		if (min < 0)
+		    break;
+		no_occur = score_sum = 0;
+		memcpy (isam_prev_buf, isam_buf[min], info->key_size);
+		for (i = 0; i<info->no_isam_positions; i++)
+		    pos_tf[i] = -10;
+	    }
+	    pos = (*info->get_pos)(isam_buf[min]);
+	    logf (LOG_LOG, "pos=%d", pos);
+	    for (i = 0; i<info->no_isam_positions; i++)
+	    {
+		int d = pos - pos_tf[i];
+		
+		no_occur++;
+		if (pos_tf[i] < 0 && i != min)
+		    continue;
+		if (d < 10)
+		    d = 10;
+		if (i == min)
+		    score_sum += fact2 / d;
+		else
+		    score_sum += fact1 / d;
+	    }
+	    pos_tf[min] = pos;
+	    logf (LOG_LOG, "score_sum = %d", score_sum);
+	    i = min;
+	    if (isam_pt)
+		isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
+	    else if (isamc_pp)
+		isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
+	} /* while */
+	xfree (isam_prev_buf);
+	xfree (pos_tf);
+	break;
+    case RSREL_METHOD_A:
+	while (1)
+	{
+	    int min = -1, i, r;
+	    double score;
+	    int co_oc, last_term;    /* Number of co-occurrences */
+	    
+	    last_term = -1;
+	    /* find min with lowest sysno */
+	    for (i = 0; i<info->no_isam_positions; i++)
+	    {
+		if (isam_r[i] && 
+		    (min < 0
+		     || (r = (*info->cmp)(isam_buf[i], isam_buf[min])) < 2))
+		{
+		    min = i;
+		    co_oc = 1;
+		}
+		else if (!r && last_term != info->term_no[i]) 
+		    co_oc++;  /* new occurrence */
+		last_term = info->term_no[i];
+	    }
+	    
+	    if (min < 0)
+		break;
+	    memcpy (isam_tmp_buf, isam_buf[min], info->key_size);
+	    /* calculate for all with those sysno */
+	    for (i = 0; i < info->no_terms; i++)
+		tf[i] = 0;
+	    for (i = 0; i<info->no_isam_positions; i++)
+	    {
+		int r;
+		
+		if (isam_r[i])
+		    r = (*info->cmp)(isam_buf[i], isam_tmp_buf);
+		else 
+		    r = 2;
+		if (r <= 1 && r >= -1)
+		{
+		    do
+		    {
+			tf[info->term_no[i]]++;
+			if (isam_pt)
+			    isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
+			else if (isamc_pp)
+			    isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
+		    } while (isam_r[i] && 
+			     (*info->cmp)(isam_buf[i], isam_tmp_buf) <= 1);
+		}
+	    }
+	    /* calculate relevance value */
+	    score = 0.0;
+	    for (i = 0; i<info->no_terms; i++)
+		if (tf[i])
+		    score += SCORE_SHOW + SCORE_COOC*co_oc/info->no_terms +
+			SCORE_DYN*tf[i]/max_tf[i];
+	    /* if value is in the top score, then save it - don't emit yet */
+	    add_rec (info, score/info->no_terms, isam_tmp_buf);
+	} /* while */
+	break;
+    } /* switch */
+    for (i = 0; i<info->no_rec; i++)
+        info->sysno_idx[i] = i;
+    qsort_info = info;
+    qsort (info->sysno_idx, info->no_rec, sizeof(*info->sysno_idx), qcomp);
+    for (i = 0; i<info->no_isam_positions; i++)
     {
-        is_pt_free (isam_pt[i]);
+        if (isam_pt)
+            is_pt_free (isam_pt[i]);
+        if (isamc_pp)
+            isc_pp_close (isamc_pp[i]);
         xfree (isam_buf[i]);
     }
     xfree (max_tf);
@@ -188,51 +398,66 @@ static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
     xfree (isam_buf);
     xfree (isam_r);
     xfree (isam_pt);
-    xfree (wgt);
+    xfree (isamc_pp);
+    xfree(tf);
 }
 
-static rset_control *r_create (const struct rset_control *sel, void *parms)
+static void *r_create (const struct rset_control *sel, void *v_parms,
+                       int *flags)
 {
-    rset_control *newct;
-    rset_relevance_parms *r_parms = parms;
+    rset_relevance_parms *parms = v_parms;
     struct rset_rel_info *info;
 
-    newct = xmalloc(sizeof(*newct));
-    memcpy(newct, sel, sizeof(*sel));
-    newct->buf = xmalloc (sizeof(struct rset_rel_info));
-
-    info = newct->buf;
-    info->key_size = r_parms->key_size;
+    *flags |= RSET_FLAG_RANKED;
+    info = xmalloc (sizeof(struct rset_rel_info));
+    info->key_size = parms->key_size;
     assert (info->key_size > 1);
-    info->max_rec = r_parms->max_rec;
+    info->max_rec = parms->max_rec;
     assert (info->max_rec > 1);
-    info->cmp = r_parms->cmp;
-
+    info->cmp = parms->cmp;
+    info->get_pos = parms->get_pos;
+
+    info->method = parms->method;
+    info->no_isam_positions = parms->no_isam_positions;
+    info->isam_positions =
+	xmalloc (sizeof(*info->isam_positions)*parms->no_isam_positions);
+    memcpy (info->isam_positions, parms->isam_positions,
+	    sizeof(*info->isam_positions) * parms->no_isam_positions);
+    info->is = parms->is;
+    info->isc = parms->isc;
+    info->no_terms = parms->no_terms;
+    info->term_no = xmalloc (sizeof(*info->term_no)*parms->no_isam_positions);
+    memcpy (info->term_no, parms->term_no,
+	    sizeof(*info->term_no)*parms->no_isam_positions);
+    
     info->key_buf = xmalloc (info->key_size * info->max_rec);
     info->score_buf = xmalloc (sizeof(*info->score_buf) * info->max_rec);
     info->sort_idx = xmalloc (sizeof(*info->sort_idx) * info->max_rec);
+    info->sysno_idx = xmalloc (sizeof(*info->sysno_idx) * info->max_rec);
     info->no_rec = 0;
+    info->hits = 0;
     info->rfd_list = NULL;
 
-    relevance (info, r_parms);
-    return newct;
+    relevance (info);
+    return info;
 }
 
-static RSFD r_open (rset_control *ct, int wflag)
+static RSFD r_open (RSET ct, int flag)
 {
     struct rset_rel_rfd *rfd;
     struct rset_rel_info *info = ct->buf;
 
-    if (wflag)
+    if (flag & RSETF_WRITE)
     {
 	logf (LOG_FATAL, "relevance set type is read-only");
 	return NULL;
     }
     rfd = xmalloc (sizeof(*rfd));
+    rfd->flag = flag;
     rfd->next = info->rfd_list;
-    info->rfd_list = rfd;
-    rfd->position = info->no_rec;
     rfd->info = info;
+    info->rfd_list = rfd;
+    r_rewind (rfd);
     return rfd;
 }
 
@@ -245,14 +470,14 @@ static void r_close (RSFD rfd)
         if (*rfdp == rfd)
         {
             *rfdp = (*rfdp)->next;
-            free (rfd);
+            xfree (rfd);
             return;
         }
     logf (LOG_FATAL, "r_close but no rfd match!");
     assert (0);
 }
 
-static void r_delete (rset_control *ct)
+static void r_delete (RSET ct)
 {
     struct rset_rel_info *info = ct->buf;
 
@@ -260,35 +485,58 @@ static void r_delete (rset_control *ct)
     xfree (info->key_buf);
     xfree (info->score_buf);
     xfree (info->sort_idx);
+    xfree (info->sysno_idx);
+    xfree (info->isam_positions);
+    xfree (info->term_no);
     xfree (info);
-    xfree (ct);
 }
 
 static void r_rewind (RSFD rfd)
 {
     struct rset_rel_rfd *p = rfd;
     struct rset_rel_info *info = p->info;
-    
-    p->position = info->no_rec;
+
+    if (p->flag & RSETF_SORT_RANK)
+        p->position = info->no_rec;
+    else
+        p->position = 0;
 }
 
-static int r_count (rset_control *ct)
+static int r_count (RSET ct)
 {
     struct rset_rel_info *info = ct->buf;
 
     return info->no_rec;
 }
 
+static int r_hits (RSET ct, void *oi)
+{
+    struct rset_rel_info *info = ct->buf;
+
+    return info->hits;
+}
+
 static int r_read (RSFD rfd, void *buf)
 {
     struct rset_rel_rfd *p = rfd;
     struct rset_rel_info *info = p->info;
 
-    if (p->position <= 0)
-        return 0;
-    --(p->position);
+    if (p->flag & RSETF_SORT_RANK)
+    {
+        if (p->position <= 0)
+            return 0;
+        --(p->position);
+        p->last_read_pos = info->sort_idx[p->position];
+    }
+    else
+    {
+        if (p->position == info->no_rec)
+            return 0;
+        p->last_read_pos = info->sysno_idx[p->position];
+        ++(p->position);
+    }
     memcpy ((char*) buf,
-            info->key_buf + info->key_size * info->sort_idx[p->position],
+            info->key_buf + info->key_size * p->last_read_pos,
             info->key_size);
     return 1;
 }
@@ -298,9 +546,7 @@ static int r_score (RSFD rfd, int *score)
     struct rset_rel_rfd *p = rfd;
     struct rset_rel_info *info = p->info;
 
-    if (p->position < 0)
-        return 0;
-    *score = (int) (1000*info->score_buf[info->sort_idx[p->position]]);
+    *score = (int) (1000*info->score_buf[p->last_read_pos]);
     return 1;
 }