Fixes for hit estimates. Added zebra_set_approx_limit.

diff --git a/NEWS b/NEWS
index 1952f13..993570a 100644 (file)
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,6 @@
+Added zebra_set_approx_limit for a ZebraHandle (session). Results
+will be approximate if hit count is greater than the limit specified.
+
 Added support for term hit counts. This was not in place for earlier
 1.4 versions, but is present in the 1.3 series. Bug #124.
 

diff --git a/include/idzebra/api.h b/include/idzebra/api.h
index bd6516a..1b9dd22 100644 (file)
--- a/include/idzebra/api.h
+++ b/include/idzebra/api.h
@@ -1,4 +1,4 @@
-/* $Id: api.h,v 1.25 2005-06-02 11:59:53 adam Exp $
+/* $Id: api.h,v 1.26 2005-06-09 10:39:52 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -175,6 +175,17 @@ void zebra_result(ZebraHandle zh, int *code, char **addinfo);
 YAZ_EXPORT
 void zebra_clearError(ZebraHandle zh);
 
+
+/**
+   \brief Set limit before Zebra does approx hit count
+   \param zh session handle
+   \param approx_limit the limit
+   
+   Results will be approximiate if hit count is greater than the
+   limit specified. By default there is a high-limit (no limit).
+*/
+ZEBRA_RES zebra_set_approx_limit(ZebraHandle zh, zint approx_limit);
+
 /**
    \brief Search using PQF Query 
    \param zh session handle

diff --git a/index/index.h b/index/index.h
index 95371fa..131f12c 100644 (file)
--- a/index/index.h
+++ b/index/index.h
@@ -1,4 +1,4 @@
-/* $Id: index.h,v 1.142 2005-06-07 14:53:39 adam Exp $
+/* $Id: index.h,v 1.143 2005-06-09 10:39:53 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -266,6 +266,7 @@ struct zebra_session {
 
     char **basenames;
     int num_basenames;
+    zint approx_limit;
     char *reg_name;
     char *path_reg;
 

diff --git a/index/zebraapi.c b/index/zebraapi.c
index 5ee16a2..d536064 100644 (file)
--- a/index/zebraapi.c
+++ b/index/zebraapi.c
@@ -1,4 +1,4 @@
-/* $Id: zebraapi.c,v 1.174 2005-06-07 11:36:38 adam Exp $
+/* $Id: zebraapi.c,v 1.175 2005-06-09 10:39:53 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -115,6 +115,7 @@ ZebraHandle zebra_open (ZebraService zs)
     zh->num_basenames = 0;
     zh->basenames = 0;
 
+    zh->approx_limit = 1000000000;
     zh->trans_no = 0;
     zh->trans_w_no = 0;
 
@@ -864,6 +865,12 @@ ZEBRA_RES zebra_select_databases (ZebraHandle zh, int num_bases,
     return ZEBRA_OK;
 }
 
+ZEBRA_RES zebra_set_approx_limit(ZebraHandle zh, zint approx_limit)
+{
+    zh->approx_limit = approx_limit;
+    return ZEBRA_OK;
+}
+
 ZEBRA_RES zebra_search_RPN(ZebraHandle zh, ODR o, Z_RPNQuery *query,
                           const char *setname, zint *hits)
 {

diff --git a/index/zrpn.c b/index/zrpn.c
index 1b74b9a..179f210 100644 (file)
--- a/index/zrpn.c
+++ b/index/zrpn.c
@@ -1,4 +1,4 @@
-/* $Id: zrpn.c,v 1.197 2005-06-07 14:53:39 adam Exp $
+/* $Id: zrpn.c,v 1.198 2005-06-09 10:39:53 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -2698,28 +2698,29 @@ void zebra_term_untrans_iconv(ZebraHandle zh, NMEM stream, int reg_type,
         *dst = nmem_strdup(stream, term_src);
 }
 
-static void count_set (RSET r, int *count)
+static void count_set(ZebraHandle zh, RSET rset, zint *count)
 {
     zint psysno = 0;
-    int kno = 0;
     struct it_key key;
     RSFD rfd;
 
     yaz_log(YLOG_DEBUG, "count_set");
 
+    rset->hits_limit = zh->approx_limit;
+
     *count = 0;
-    rfd = rset_open (r, RSETF_READ);
-    while (rset_read (rfd, &key,0 /* never mind terms */))
+    rfd = rset_open(rset, RSETF_READ);
+    while (rset_read(rfd, &key,0 /* never mind terms */))
     {
         if (key.mem[0] != psysno)
         {
             psysno = key.mem[0];
-            (*count)++;
+           if (rfd->counted_items >= rset->hits_limit)
+               break;
         }
-        kno++;
     }
     rset_close (rfd);
-    yaz_log(YLOG_DEBUG, "%d keys, %d records", kno, *count);
+    *count = rset->hits_count;
 }
 
 ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
@@ -2994,6 +2995,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
         }
        if (lo >= 0)
        {
+           zint count;
            /* merge with limit_set if given */
            if (limit_set)
            {
@@ -3006,7 +3008,8 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
                                          2, rsets);
            }
            /* count it */
-           count_set(rset, &glist[lo].occurrences);
+           count_set(zh, rset, &count);
+           glist[lo].occurrences = count;
            rset_delete(rset);
        }
     }
@@ -3033,6 +3036,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
        const char *tst;
        RSET rset;
        int lo = before-1-i; /* offset in result list */
+       zint count;
        
        for (j = 0; j <ord_no; j++)
        {
@@ -3089,7 +3093,8 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
            rset = rsmulti_and_create(rset_nmem, kc,
                                      kc->scope, 2, rsets);
        }
-       count_set (rset, &glist[lo].occurrences);
+       count_set(zh, rset, &count);
+       glist[lo].occurrences = count;
        rset_delete (rset);
     }
     (*kc->dec)(kc);

diff --git a/index/zsets.c b/index/zsets.c
index b20041f..024113e 100644 (file)
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,4 +1,4 @@
-/* $Id: zsets.c,v 1.88 2005-06-07 14:53:39 adam Exp $
+/* $Id: zsets.c,v 1.89 2005-06-09 10:39:53 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -62,6 +62,7 @@ struct zebra_set {
     zint cache_position;  /* last position */
     RSFD cache_rfd;       /* rfd (NULL if not existing) */
     zint cache_psysno;    /* sysno for last position */
+    zint approx_limit;    /* limit before we do approx */
 };
 
 struct zset_sort_entry {
@@ -128,6 +129,7 @@ ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
     for (i = 0; sort_sequence->specs[i]; i++)
         ;
     sort_sequence->num_specs = i;
+    rset->hits_limit = sset->approx_limit;
     if (!i)
     {
         res = resultSetRank (zh, sset, rset, rset_nmem);
@@ -152,7 +154,7 @@ ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
 
     zh->hits = 0;
 
-    zebraSet = resultSetAdd (zh, setname, 1);
+    zebraSet = resultSetAdd(zh, setname, 1);
     if (!zebraSet)
         return ZEBRA_FAIL;
     zebraSet->locked = 1;
@@ -164,7 +166,7 @@ ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
     zebraSet->basenames = 
         nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames));
     for (i = 0; i<num_bases; i++)
-        zebraSet->basenames[i] = nmem_strdup (zebraSet->nmem, basenames[i]);
+        zebraSet->basenames[i] = nmem_strdup(zebraSet->nmem, basenames[i]);
 
     res = resultSetSearch(zh, zebraSet->nmem, zebraSet->rset_nmem,
                          rpn, zebraSet);
@@ -205,7 +207,7 @@ void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type,
     (s->hits)++;
 }
 
-ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
+ZebraSet resultSetAdd(ZebraHandle zh, const char *name, int ov)
 {
     ZebraSet s;
     int i;
@@ -267,10 +269,11 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
     s->rpn = 0;
     s->cache_position = 0;
     s->cache_rfd = 0;
+    s->approx_limit = zh->approx_limit;
     return s;
 }
 
-ZebraSet resultSetGet (ZebraHandle zh, const char *name)
+ZebraSet resultSetGet(ZebraHandle zh, const char *name)
 {
     ZebraSet s;
 
@@ -313,7 +316,7 @@ void resultSetInvalidate (ZebraHandle zh)
     }
 }
 
-void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
+void resultSetDestroy(ZebraHandle zh, int num, char **names,int *statuses)
 {
     ZebraSet * ss = &zh->sets;
     int i;
@@ -848,6 +851,7 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
        RSFD rfd = rset_open(rset, RSETF_READ);
        struct rank_control *rc = rank_class->control;
        double score;
+       zint count = 0;
        
        void *handle =
            (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem,
@@ -868,6 +872,7 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
                {
                    score = (*rc->calc) (handle, psysno);
                    resultSetInsertRank (zh, sort_info, psysno, score, 'A');
+                   count++;
                }
                psysno = this_sys;
            }
@@ -877,6 +882,7 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
        {
            score = (*rc->calc)(handle, psysno);
            resultSetInsertRank(zh, sort_info, psysno, score, 'A');
+           count++;
        }
        (*rc->end) (zh->reg, handle);
        rset_close (rfd);

diff --git a/rset/rset.c b/rset/rset.c
index 52f3033..648193a 100644 (file)
--- a/rset/rset.c
+++ b/rset/rset.c
@@ -1,4 +1,4 @@
-/* $Id: rset.c,v 1.50 2005-06-07 14:53:39 adam Exp $
+/* $Id: rset.c,v 1.51 2005-06-09 10:39:53 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -166,7 +166,7 @@ RSET rset_create_base(const struct rset_control *sel,
     rset->free_list = NULL;
     rset->use_list = NULL;
     rset->hits_count = 0;
-    rset->hits_limit = 1000;
+    rset->hits_limit = 0;
     rset->hits_round = 1000;
     rset->keycontrol = kcontrol;
     (*kcontrol->inc)(kcontrol);
@@ -347,7 +347,7 @@ int rset_default_read(RSFD rfd, void *buf, TERMID *term)
     if (rc > 0)
     {
        if (rfd->counted_items == 0 ||
-           (rset->keycontrol->cmp)(buf, rfd->counted_buf) >= rset->scope)
+           (rset->keycontrol->cmp)(buf, rfd->counted_buf) > rset->scope)
        {
            memcpy(rfd->counted_buf, buf, rset->keycontrol->key_size);
            rfd->counted_items++;