From b169e50ef204402fe272c2ba4df0597e279fe4d7 Mon Sep 17 00:00:00 2001
From: Heikki Levanto <heikki@indexdata.dk>
Date: Fri, 22 Oct 2004 11:33:28 +0000
Subject: [PATCH] Term counts right in the log (if not in searchresponse)

---
 include/rset.h |   12 +++++++++---
 index/zsets.c  |   10 +++-------
 rset/rset.c    |   26 ++++++++++++++++++++++----
 3 files changed, 34 insertions(+), 14 deletions(-)

diff --git a/include/rset.h b/include/rset.h
index 4f28919..0fc191f 100644
--- a/include/rset.h
+++ b/include/rset.h
@@ -1,4 +1,4 @@
-/* $Id: rset.h,v 1.40 2004-10-22 10:58:28 heikki Exp $
+/* $Id: rset.h,v 1.41 2004-10-22 11:33:28 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 
@@ -48,11 +48,14 @@ typedef struct rset *RSET; /* Result set */
  */
 
 struct rset_term {
+    /** the term itself */
     char *name;
-    zint  nn; /* FIXME - what is this used for */
     char *flags;
-    zint  count;
     int  type;
+    /** the rset corresponding to this term */
+    RSET rset;
+    /** private stuff for the ranking algorithm */
+    void *rankpriv;
 };
 
 typedef struct rset_term *TERMID; 
@@ -208,6 +211,9 @@ RSET rset_dup (RSET rs);
 /* int rset_type (RSET) */
 #define rset_type(rs) ((rs)->control->desc)
 
+/** rset_count counts or estimates the keys in it*/
+zint rset_count(RSET rs);
+
 RSET rstemp_create( NMEM nmem, const struct key_control *kcontrol,
                     int scope, 
                     const char *temp_path);
diff --git a/index/zsets.c b/index/zsets.c
index 3fd85a0..c284b96 100644
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,4 +1,4 @@
-/* $Id: zsets.c,v 1.67 2004-10-22 10:58:28 heikki Exp $
+/* $Id: zsets.c,v 1.68 2004-10-22 11:33:28 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -821,15 +821,11 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
 
     for (i = 0; i < n; i++)
     {
-        if (est>0)
-            terms[i]->count = 
-                est=(zint)(terms[i]->count/ratio);
-        yaz_log (LOG_LOG, "term=\"%s\" nn=" ZINT_FORMAT 
+        yaz_log (LOG_LOG, "term=\"%s\" "
                     " type=%s count=" ZINT_FORMAT,
                  terms[i]->name,
-                 terms[i]->nn,
                  terms[i]->flags,
-                 terms[i]->count);
+                 rset_count(terms[i]->rset));
     }
     xfree(terms);
     yaz_log (LOG_DEBUG, ZINT_FORMAT " keys, "ZINT_FORMAT" distinct sysnos", 
diff --git a/rset/rset.c b/rset/rset.c
index bbc3fab..0691af1 100644
--- a/rset/rset.c
+++ b/rset/rset.c
@@ -1,4 +1,4 @@
-/* $Id: rset.c,v 1.37 2004-10-22 10:58:29 heikki Exp $
+/* $Id: rset.c,v 1.38 2004-10-22 11:33:29 heikki Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -86,12 +86,14 @@ RSET rset_create_base(const struct rset_control *sel,
     else 
         rnew->my_nmem=1;
     rnew->control = sel;
-    rnew->count = 1;
+    rnew->count = 1; /* refcount! */
     rnew->priv = 0;
     rnew->free_list=NULL;
     rnew->keycontrol=kcontrol;
     rnew->scope=scope;
     rnew->term=term;
+    if (term)
+        term->rset=rnew;
     return rnew;
 }
 
@@ -148,6 +150,22 @@ int rset_default_forward(RSFD rfd, void *buf, TERMID *term,
     return more;
 }
 
+/** 
+ * rset_count uses rset_pos to get the total and returns that.
+ * This is ok for rsisamb, and for some other rsets, but in case of
+ * booleans etc it will give bad estimate, as nothing has been read
+ * from that rset
+ */
+zint rset_count(RSET rs)
+{
+    double cur,tot;
+    RSFD rfd=rset_open(rs,0);
+    rset_pos(rfd,&cur,&tot);
+    rset_close(rfd);
+    return (zint)(tot);
+}
+
+
 /** rset_get_no_terms is a getterms function for those that don't have any */
 void rset_get_no_terms(RSET ct, TERMID *terms, int maxterms, int *curterm)
 {
@@ -188,9 +206,9 @@ TERMID rset_term_create (const char *name, int length, const char *flags,
         t->flags = NULL;
     else
         t->flags = nmem_strdup(nmem,flags);
-    t->nn = -1;
-    t->count = 0;
     t->type = type;
+    t->rankpriv=0;
+    t->rset=0;
     return t;
 }
 
-- 
1.7.10.4