Re-estabslished rank-1. Gets same order of results, but slightly different

scores, due to using a better estimate for term occurrences.

diff --git a/index/index.h b/index/index.h
index 0e873de..39ad880 100644 (file)
--- a/index/index.h
+++ b/index/index.h
@@ -1,4 +1,4 @@
-/* $Id: index.h,v 1.123 2004-09-28 12:39:55 adam Exp $
+/* $Id: index.h,v 1.124 2004-10-26 15:32:11 heikki Exp $

    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 

@@ -318,7 +318,9 @@ struct rank_control {
     char *name;
     void *(*create)(ZebraHandle zh);
     void (*destroy)(struct zebra_register *reg, void *class_handle);
     char *name;
     void *(*create)(ZebraHandle zh);
     void (*destroy)(struct zebra_register *reg, void *class_handle);

-    void *(*begin)(struct zebra_register *reg, void *class_handle, RSET rset);
+    void *(*begin)(struct zebra_register *reg, 
+                   void *class_handle, RSET rset, NMEM nmem,
+                   TERMID *terms, int numterms);

     /* ### Could add parameters to begin:
      * char *index;    // author, title, etc.
      * int dbsize;     // number of records in database
     /* ### Could add parameters to begin:
      * char *index;    // author, title, etc.
      * int dbsize;     // number of records in database

@@ -326,7 +328,7 @@ struct rank_control {
      */
     void (*end)(struct zebra_register *reg, void *set_handle);
     int (*calc)(void *set_handle, zint sysno);
      */
     void (*end)(struct zebra_register *reg, void *set_handle);
     int (*calc)(void *set_handle, zint sysno);

-    void (*add)(void *set_handle, int seqno, int term_index);
+    void (*add)(void *set_handle, int seqno, TERMID term);

 };
 
 struct term_set_entry {
 };
 
 struct term_set_entry {

@@ -379,7 +381,7 @@ void resultSetSort (ZebraHandle zh, NMEM nmem,
 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
                          ZebraSet sset, RSET rset,
                          Z_SortKeySpecList *sort_sequence, int *sort_status);
 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
                          ZebraSet sset, RSET rset,
                          Z_SortKeySpecList *sort_sequence, int *sort_status);

-void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset);
+void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset, NMEM nmem);

 void resultSetInvalidate (ZebraHandle zh);
 
 int zebra_server_lock_init (ZebraService zh);
 void resultSetInvalidate (ZebraHandle zh);
 
 int zebra_server_lock_init (ZebraService zh);

diff --git a/index/livcode.c b/index/livcode.c
index cd85e34..ba86b2e 100644 (file)
--- a/index/livcode.c
+++ b/index/livcode.c
@@ -11,10 +11,12 @@ rights reserved.
 Licensed under the Academic Free License version 1.1.
 http://opensource.org/licenses/academic.php
 
 Licensed under the Academic Free License version 1.1.
 http://opensource.org/licenses/academic.php
 

-$Id: livcode.c,v 1.3 2004-08-20 14:44:46 heikki Exp $
+$Id: livcode.c,v 1.4 2004-10-26 15:32:11 heikki Exp $

 
 */
 
 
 */
 

+#ifdef SKIPTHIS  /* Need to fix the interface - FIXME */
+

 #include <stdlib.h>
 #include <stdio.h>
 #ifdef WIN32
 #include <stdlib.h>
 #include <stdio.h>
 #ifdef WIN32

@@ -538,6 +540,7 @@ struct rank_set_info {
     int last_pos;
     int no_entries;
     int no_rank_entries;
     int last_pos;
     int no_entries;
     int no_rank_entries;

+    NMEM nmem;

     struct rank_term_info *entries;
 };
 
     struct rank_term_info *entries;
 };
 

@@ -584,7 +587,8 @@ static void destroy (struct zebra_register *reg, void *class_handle)
  *  each result set. The returned handle is a "set handle" and
  *  will be used in each of the handlers below.
  */
  *  each result set. The returned handle is a "set handle" and
  *  will be used in each of the handlers below.
  */

-static void *begin (struct zebra_register *reg, void *class_handle, RSET rset)
+static void *begin (struct zebra_register *reg, void *class_handle, 
+                    RSET rset, NMEM nmem)

 {
     struct rank_set_info *si = (struct rank_set_info *) xmalloc (sizeof(*si));
     int i;
 {
     struct rank_set_info *si = (struct rank_set_info *) xmalloc (sizeof(*si));
     int i;

@@ -594,6 +598,7 @@ static void *begin (struct zebra_register *reg, void *class_handle, RSET rset)
     /* do about this ??? */
     si->no_entries = 0; /* rset->no_rset_terms; */ /* FIXME ??? */
     si->no_rank_entries = 0;
     /* do about this ??? */
     si->no_entries = 0; /* rset->no_rset_terms; */ /* FIXME ??? */
     si->no_rank_entries = 0;

+    si->nmem=nmem;

     si->entries = (struct rank_term_info *)
        xmalloc (sizeof(*si->entries)*si->no_entries);
     for (i = 0; i < si->no_entries; i++)
     si->entries = (struct rank_term_info *)
        xmalloc (sizeof(*si->entries)*si->no_entries);
     for (i = 0; i < si->no_entries; i++)

@@ -706,3 +711,5 @@ static struct rank_control rank_control = {
 };
  
 struct rank_control *rankliv_class = &rank_control;
 };
  
 struct rank_control *rankliv_class = &rank_control;

+#endif
+

diff --git a/index/rank1.c b/index/rank1.c
index 05f0761..581e78a 100644 (file)
--- a/index/rank1.c
+++ b/index/rank1.c
@@ -1,4 +1,4 @@
-/* $Id: rank1.c,v 1.17 2004-08-20 14:44:46 heikki Exp $
+/* $Id: rank1.c,v 1.18 2004-10-26 15:32:11 heikki Exp $

    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
    Index Data Aps
 
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
    Index Data Aps
 

@@ -30,7 +30,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 #include <unistd.h>
 #endif
 
 #include <unistd.h>
 #endif
 

-#define DEBUG_RANK 0
+#define DEBUG_RANK 1

 
 #include "index.h"
 
 
 #include "index.h"
 

@@ -44,6 +44,8 @@ struct rank_term_info {
     int global_inv;
     int rank_flag;
     int rank_weight;
     int global_inv;
     int rank_flag;
     int rank_weight;

+    TERMID term;
+    int term_index;

 };
 
 struct rank_set_info {
 };
 
 struct rank_set_info {

@@ -51,6 +53,7 @@ struct rank_set_info {
     int no_entries;
     int no_rank_entries;
     struct rank_term_info *entries;
     int no_entries;
     int no_rank_entries;
     struct rank_term_info *entries;

+    NMEM nmem;

 };
 
 static int log2_int (unsigned g)
 };
 
 static int log2_int (unsigned g)

@@ -67,8 +70,8 @@ static int log2_int (unsigned g)
  */
 static void *create (ZebraHandle zh)
 {
  */
 static void *create (ZebraHandle zh)
 {

-    struct rank_class_info *ci = (struct rank_class_info *)
-       xmalloc (sizeof(*ci));
+    struct rank_class_info *ci = 
+        (struct rank_class_info *) xmalloc (sizeof(*ci));

 
     yaz_log (LOG_DEBUG, "rank-1 create");
     return ci;
 
     yaz_log (LOG_DEBUG, "rank-1 create");
     return ci;

@@ -88,54 +91,58 @@ static void destroy (struct zebra_register *reg, void *class_handle)
 }
 
 
 }
 
 

-/*
+/**

  * begin: Prepares beginning of "real" ranking. Called once for
  *  each result set. The returned handle is a "set handle" and
  *  will be used in each of the handlers below.
  */
  * begin: Prepares beginning of "real" ranking. Called once for
  *  each result set. The returned handle is a "set handle" and
  *  will be used in each of the handlers below.
  */

-static void *begin (struct zebra_register *reg, void *class_handle, RSET rset)
+static void *begin (struct zebra_register *reg, 
+                    void *class_handle, RSET rset, NMEM nmem,
+                    TERMID *terms, int numterms)

 {
 {

-    struct rank_set_info *si = (struct rank_set_info *) xmalloc (sizeof(*si));
+    struct rank_set_info *si = 
+        (struct rank_set_info *) nmem_malloc (nmem,sizeof(*si));

     int i;
 
 #if DEBUG_RANK
     yaz_log (LOG_LOG, "rank-1 begin");
 #endif
     int i;
 
 #if DEBUG_RANK
     yaz_log (LOG_LOG, "rank-1 begin");
 #endif

-    si->no_entries = 0; /* rset->no_rset_terms; */ /* FIXME - what to do here*/
-       /* Now that we don't count term occurrences, ranking will have to */
-       /* different! */
+    si->no_entries = numterms;

     si->no_rank_entries = 0;
     si->no_rank_entries = 0;

+    si->nmem=nmem;

     si->entries = (struct rank_term_info *)
     si->entries = (struct rank_term_info *)

-       xmalloc (sizeof(*si->entries)*si->no_entries);
-    for (i = 0; i < si->no_entries; i++)
+       nmem_malloc (si->nmem, sizeof(*si->entries)*numterms); 
+    for (i = 0; i < numterms; i++)

     {
     {

-       zint g = 0; /* rset->rset_terms[i]->nn; */ /* FIXME ??? */
+       zint g = rset_count(terms[i]->rset);

 #if DEBUG_RANK
 #if DEBUG_RANK

-        yaz_log(LOG_LOG, "i=%d flags=%s", i, rset->rset_terms[i]->flags);
+        yaz_log(LOG_LOG, "i=%d flags=%s '%s'", i, 
+                terms[i]->flags, terms[i]->name );

 #endif
 #endif

-       if (0) /* (!strncmp (rset->rset_terms[i]->flags, "rank,", 5)) */
-               /* FIXME */ /* ??? */
+       if  (!strncmp (terms[i]->flags, "rank,", 5)) 

        {
        {

-            const char *cp = "w"; 
-                         /*= strstr(rset->rset_terms[i]->flags+4, ",w=");*/
-                         /* FIXME ??? */
+            const char *cp = strstr(terms[i]->flags+4, ",w=");

            si->entries[i].rank_flag = 1;
             if (cp)
                 si->entries[i].rank_weight = atoi (cp+3);
             else
                 si->entries[i].rank_weight = 34;
 #if DEBUG_RANK
            si->entries[i].rank_flag = 1;
             if (cp)
                 si->entries[i].rank_weight = atoi (cp+3);
             else
                 si->entries[i].rank_weight = 34;
 #if DEBUG_RANK

-            yaz_log (LOG_LOG, " i=%d weight=%d", i,
-                     si->entries[i].rank_weight);
+            yaz_log (LOG_LOG, " i=%d weight=%d g="ZINT_FORMAT, i,
+                     si->entries[i].rank_weight, g);

 #endif
            (si->no_rank_entries)++;
        }
        else
            si->entries[i].rank_flag = 0;
 #endif
            (si->no_rank_entries)++;
        }
        else
            si->entries[i].rank_flag = 0;

-       si->entries[i].local_occur = 0;
+       si->entries[i].local_occur = 0;  /* FIXME */

        si->entries[i].global_occur = g;
        si->entries[i].global_inv = 32 - log2_int (g);
        si->entries[i].global_occur = g;
        si->entries[i].global_inv = 32 - log2_int (g);

-       yaz_log (LOG_DEBUG, " global_inv = %d g = " ZINT_FORMAT, (int) (32-log2_int (g)), g);
+       yaz_log (LOG_DEBUG, " global_inv = %d g = " ZINT_FORMAT, 
+                (int) (32-log2_int (g)), g);
+        si->entries[i].term=terms[i];
+        si->entries[i].term_index=i;
+        terms[i]->rankpriv=&(si->entries[i]);

     }
     return si;
 }
     }
     return si;
 }

@@ -146,25 +153,28 @@ static void *begin (struct zebra_register *reg, void *class_handle, RSET rset)
  */
 static void end (struct zebra_register *reg, void *set_handle)
 {
  */
 static void end (struct zebra_register *reg, void *set_handle)
 {

-    struct rank_set_info *si = (struct rank_set_info *) set_handle;

     yaz_log (LOG_DEBUG, "rank-1 end");
     yaz_log (LOG_DEBUG, "rank-1 end");

-    xfree (si->entries);
-    xfree (si);
+    /* no need to free anything, they are in nmems */

 }
 
 }
 

-/*
+
+/**

  * add: Called for each word occurence in a result set. This routine
  *  should be as fast as possible. This routine should "incrementally"
  *  update the score.
  */
  * add: Called for each word occurence in a result set. This routine
  *  should be as fast as possible. This routine should "incrementally"
  *  update the score.
  */

-static void add (void *set_handle, int seqno, int term_index)
+static void add (void *set_handle, int seqno, TERMID term)

 {
     struct rank_set_info *si = (struct rank_set_info *) set_handle;
 {
     struct rank_set_info *si = (struct rank_set_info *) set_handle;

+    struct rank_term_info *ti= (struct rank_term_info *) term->rankpriv;
+    assert(si);
+    assert(term);
+    assert(ti);

 #if DEBUG_RANK
 #if DEBUG_RANK

-    yaz_log (LOG_LOG, "rank-1 add seqno=%d term_index=%d", seqno, term_index);
+    yaz_log (LOG_LOG, "rank-1 add seqno=%d term=%s", seqno, term->name);

 #endif
     si->last_pos = seqno;
 #endif
     si->last_pos = seqno;

-    si->entries[term_index].local_occur++;
+    ti->local_occur++;

 }
 
 /*
 }
 
 /*

@@ -187,7 +197,7 @@ static int calc (void *set_handle, zint sysno)
     for (i = 0; i < si->no_entries; i++)
     {
 #if DEBUG_RANK
     for (i = 0; i < si->no_entries; i++)
     {
 #if DEBUG_RANK

-        yaz_log(LOG_LOG, "i=%d rank_flag=%d lo=%d",
+        yaz_log(LOG_LOG, "calc: i=%d rank_flag=%d lo=%d",

                 i, si->entries[i].rank_flag, si->entries[i].local_occur);
 #endif
        if (si->entries[i].rank_flag && (lo = si->entries[i].local_occur))
                 i, si->entries[i].rank_flag, si->entries[i].local_occur);
 #endif
        if (si->entries[i].rank_flag && (lo = si->entries[i].local_occur))

@@ -197,7 +207,7 @@ static int calc (void *set_handle, zint sysno)
     divisor = si->no_rank_entries * (8+log2_int (si->last_pos/si->no_entries));
     score = score / divisor;
 #if DEBUG_RANK
     divisor = si->no_rank_entries * (8+log2_int (si->last_pos/si->no_entries));
     score = score / divisor;
 #if DEBUG_RANK

-    yaz_log (LOG_LOG, "sysno=" ZINT_FORMAT " score=%d", sysno, score);
+    yaz_log (LOG_LOG, "calc sysno=" ZINT_FORMAT " score=%d", sysno, score);

 #endif
     if (score > 1000)
        score = 1000;
 #endif
     if (score > 1000)
        score = 1000;

diff --git a/index/zebraapi.c b/index/zebraapi.c
index 9896bf6..e97e27c 100644 (file)
--- a/index/zebraapi.c
+++ b/index/zebraapi.c
@@ -1,4 +1,4 @@
-/* $Id: zebraapi.c,v 1.136 2004-10-15 10:07:32 heikki Exp $
+/* $Id: zebraapi.c,v 1.137 2004-10-26 15:32:11 heikki Exp $

    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 

@@ -292,8 +292,10 @@ struct zebra_register *zebra_register_open (ZebraService zs, const char *name,
     reg->ptr_i=0;
     
     zebraRankInstall (reg, rank1_class);
     reg->ptr_i=0;
     
     zebraRankInstall (reg, rank1_class);

+#ifdef SKIPTHIS /* FIXME - those ranks not yet converted to new interface */

     zebraRankInstall (reg, rankzv_class);
     zebraRankInstall (reg, rankliv_class);
     zebraRankInstall (reg, rankzv_class);
     zebraRankInstall (reg, rankliv_class);

+#endif

 
     recordCompression = res_get_def (res, "recordCompression", "none");
     if (!strcmp (recordCompression, "none"))
 
     recordCompression = res_get_def (res, "recordCompression", "none");
     if (!strcmp (recordCompression, "none"))

@@ -812,7 +814,9 @@ int zebra_search_RPN (ZebraHandle zh, ODR o,
     if (zebra_begin_read (zh))
        return 1;
 
     if (zebra_begin_read (zh))
        return 1;
 

+#ifdef SKIPTHIS /* FIXME - livcode rank not yet available */

     zebra_livcode_transform(zh, query);
     zebra_livcode_transform(zh, query);

+#endif

 
     resultSetAddRPN (zh, odr_extract_mem(o), query, 
                      zh->num_basenames, zh->basenames, setname);
 
     resultSetAddRPN (zh, odr_extract_mem(o), query, 
                      zh->num_basenames, zh->basenames, setname);

diff --git a/index/zrpn.c b/index/zrpn.c
index 18e6dcc..c185b95 100644 (file)
--- a/index/zrpn.c
+++ b/index/zrpn.c
@@ -1,4 +1,4 @@
-/* $Id: zrpn.c,v 1.157 2004-10-15 10:07:32 heikki Exp $
+/* $Id: zrpn.c,v 1.158 2004-10-26 15:32:11 heikki Exp $

    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 

@@ -2304,7 +2304,7 @@ RSET rpn_search (ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
         ;
     sort_sequence->num_specs = i;
     if (!i)
         ;
     sort_sequence->num_specs = i;
     if (!i)

-        resultSetRank (zh, sset, rset);
+        resultSetRank (zh, sset, rset, rset_nmem);

     else
     {
         logf (LOG_DEBUG, "resultSetSortSingle in rpn_search");
     else
     {
         logf (LOG_DEBUG, "resultSetSortSingle in rpn_search");

diff --git a/index/zsets.c b/index/zsets.c
index c284b96..efe14c7 100644 (file)
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,4 +1,4 @@
-/* $Id: zsets.c,v 1.68 2004-10-22 11:33:28 heikki Exp $
+/* $Id: zsets.c,v 1.69 2004-10-26 15:32:11 heikki Exp $

    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 

@@ -727,13 +727,14 @@ RSET resultSetRef (ZebraHandle zh, const char *resultSetId)
     return NULL;
 }
 
     return NULL;
 }
 

-void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
+void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset, NMEM nmem)

 {
     zint kno = 0;
     struct it_key key;
     RSFD rfd;
     TERMID termid;
     TERMID *terms;
 {
     zint kno = 0;
     struct it_key key;
     RSFD rfd;
     TERMID termid;
     TERMID *terms;

+    int numterms;

     int i,n;
     ZebraRankClass rank_class;
     struct rank_control *rc;
     int i,n;
     ZebraRankClass rank_class;
     struct rank_control *rc;

@@ -747,6 +748,12 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
     sort_info = zebraSet->sort_info;
     sort_info->num_entries = 0;
     zebraSet->hits = 0;
     sort_info = zebraSet->sort_info;
     sort_info->num_entries = 0;
     zebraSet->hits = 0;

+    n=0;
+    rset_getterms(rset,0,0,&n);
+    terms=malloc( sizeof(*terms)*n);
+    numterms=0;
+    rset_getterms(rset,terms,n,&numterms);
+

     rfd = rset_open (rset, RSETF_READ);
 
     rank_class = zebraRankLookup (zh, rank_handler_name);
     rfd = rset_open (rset, RSETF_READ);
 
     rank_class = zebraRankLookup (zh, rank_handler_name);

@@ -762,14 +769,16 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
         zint psysno = key.mem[0];
         int score;
         void *handle =
         zint psysno = key.mem[0];
         int score;
         void *handle =

-            (*rc->begin) (zh->reg, rank_class->class_handle, rset);
+            (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem,
+                          terms, numterms);

         (zebraSet->hits)++;
         esthits=atoi(res_get_def(zh->res,"estimatehits","0"));
         if (!esthits) 
             est=-1; /* can not do */
         do
         {
         (zebraSet->hits)++;
         esthits=atoi(res_get_def(zh->res,"estimatehits","0"));
         if (!esthits) 
             est=-1; /* can not do */
         do
         {

-            zint this_sys = key.mem[0];
+            zint this_sys = key.mem[0]; /* FIXME - assumes scope==2 */
+            zint seqno = key.mem[1]; /* FIXME - assumes scope==2 */

             kno++;
             if (this_sys != psysno)
             {
             kno++;
             if (this_sys != psysno)
             {

@@ -779,9 +788,7 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
                 (zebraSet->hits)++;
                 psysno = this_sys;
             }
                 (zebraSet->hits)++;
                 psysno = this_sys;
             }

-            /* FIXME - Ranking is broken, since rsets no longer have */
-            /* term lists! */
-            /* (*rc->add) (handle, this_sys, term_index); */
+            (*rc->add) (handle, seqno, termid);

             
             if ( (est==-2) && (zebraSet->hits==esthits))
             { /* time to estimate the hits */
             
             if ( (est==-2) && (zebraSet->hits==esthits))
             { /* time to estimate the hits */

@@ -811,15 +818,9 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
         (*rc->end) (zh->reg, handle);
     }
     rset_close (rfd);
         (*rc->end) (zh->reg, handle);
     }
     rset_close (rfd);

-    n=0;
-    rset_getterms(rset,0,0,&n);
-    terms=xmalloc( sizeof(*terms)*n);
-    i=n;
-    n=0;
-    rset_getterms(rset,terms,i,&n);

 
 
 
 

-    for (i = 0; i < n; i++)
+    for (i = 0; i < numterms; i++)

     {
         yaz_log (LOG_LOG, "term=\"%s\" "
                     " type=%s count=" ZINT_FORMAT,
     {
         yaz_log (LOG_LOG, "term=\"%s\" "
                     " type=%s count=" ZINT_FORMAT,

@@ -827,7 +828,6 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
                  terms[i]->flags,
                  rset_count(terms[i]->rset));
     }
                  terms[i]->flags,
                  rset_count(terms[i]->rset));
     }

-    xfree(terms);

     yaz_log (LOG_DEBUG, ZINT_FORMAT " keys, "ZINT_FORMAT" distinct sysnos", 
                     kno, zebraSet->hits);
 }
     yaz_log (LOG_DEBUG, ZINT_FORMAT " keys, "ZINT_FORMAT" distinct sysnos", 
                     kno, zebraSet->hits);
 }

diff --git a/index/zvrank.c b/index/zvrank.c
index e2c19c0..ecebce3 100644 (file)
--- a/index/zvrank.c
+++ b/index/zvrank.c
@@ -1,4 +1,4 @@
-/* $Id: zvrank.c,v 1.10 2004-08-20 14:44:46 heikki Exp $
+/* $Id: zvrank.c,v 1.11 2004-10-26 15:32:11 heikki Exp $

    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
    Index Data Aps
 
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
    Index Data Aps
 

@@ -40,6 +40,8 @@ fernuni-hagen.de>
 ** "ntc-atn", "atc-atn", etc.
 */
 
 ** "ntc-atn", "atc-atn", etc.
 */
 

+#if SKIPTHIS   /* FIXME - Disabled while changing the interface to ranking */
+

 #include <math.h>  /* for log */
 
 #include <stdio.h>
 #include <math.h>  /* for log */
 
 #include <stdio.h>

@@ -61,8 +63,7 @@ static double blog(double x) {
 
 /* structures */
 
 
 /* structures */
 

-struct rank_class_info { /* now we need this */
-    int dummy;
+struct rank_class_info {

     char rscheme[8];    /* name of weighting scheme */
 };
 
     char rscheme[8];    /* name of weighting scheme */
 };
 

@@ -76,6 +77,7 @@ struct rs_info {      /* for result set */
     char rscheme[8];    /* name of weighting scheme */
     /**/
     int veclen;
     char rscheme[8];    /* name of weighting scheme */
     /**/
     int veclen;

+    NMEM nmem;

     void (*d_tf_fct)(void *, void *);   /* doc term frequency function */
     void (*d_idf_fct)(void *, void *);  /* doc idf function */
     void (*d_norm_fct)(void *, void *); /* doc normalization function */
     void (*d_tf_fct)(void *, void *);   /* doc term frequency function */
     void (*d_idf_fct)(void *, void *);  /* doc idf function */
     void (*d_norm_fct)(void *, void *); /* doc normalization function */

@@ -684,7 +686,8 @@ static void zv_destroy (struct zebra_register *reg, void *class_handle) {
  *  each result set. The returned handle is a "set handle" and
  *  will be used in each of the handlers below.
  */
  *  each result set. The returned handle is a "set handle" and
  *  will be used in each of the handlers below.
  */

-static void *zv_begin(struct zebra_register *reg, void *class_handle, RSET rset)
+static void *zv_begin(struct zebra_register *reg, void *class_handle, 
+                      RSET rset, NMEM nmem)

 {
     struct rs_info *rs=(struct rs_info *)xmalloc(sizeof(*rs));
     struct rank_class_info *ci=(struct rank_class_info *)class_handle;
 {
     struct rs_info *rs=(struct rs_info *)xmalloc(sizeof(*rs));
     struct rank_class_info *ci=(struct rank_class_info *)class_handle;

@@ -697,6 +700,7 @@ static void *zv_begin(struct zebra_register *reg, void *class_handle, RSET rset)
     /* FIXME - Now that we don't have term lists in rsets, what do */
     /* we do here ??? */
     zv_init(rs, ci->rscheme);
     /* FIXME - Now that we don't have term lists in rsets, what do */
     /* we do here ??? */
     zv_init(rs, ci->rscheme);

+    rs->nmem=nmem;

     rs->veclen=veclen;
     prn_rs(rs);
   
     rs->veclen=veclen;
     prn_rs(rs);
   

@@ -815,4 +819,5 @@ static struct rank_control rank_control_vsm = {
  
 struct rank_control *rankzv_class = &rank_control_vsm;
 
  
 struct rank_control *rankzv_class = &rank_control_vsm;
 

+#endif /* SKIPTHIS */

 /* EOF */
 /* EOF */

diff --git a/rset/rsmultiandor.c b/rset/rsmultiandor.c
index 4f4bdcd..9a88534 100644 (file)
--- a/rset/rsmultiandor.c
+++ b/rset/rsmultiandor.c
@@ -1,4 +1,4 @@
-/* $Id: rsmultiandor.c,v 1.7 2004-10-22 10:12:52 heikki Exp $
+/* $Id: rsmultiandor.c,v 1.8 2004-10-26 15:32:11 heikki Exp $

    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
    Index Data Aps
 

@@ -562,17 +562,19 @@ static void r_pos (RSFD rfd, double *current, double *total)
     int i;
     for (i=0; i<info->no_rsets; i++){
         rset_pos(mrfd->items[i].fd, &cur, &tot);
     int i;
     for (i=0; i<info->no_rsets; i++){
         rset_pos(mrfd->items[i].fd, &cur, &tot);

-        logf(LOG_DEBUG, "r_pos: %d %0.1f %0.1f", i, cur,tot);
+        /*logf(LOG_LOG, "r_pos: %d %0.1f %0.1f", i, cur,tot); */

         scur += cur;
         stot += tot;
     }
     if (stot <1.0) { /* nothing there */
         *current=0;
         *total=0;
         scur += cur;
         stot += tot;
     }
     if (stot <1.0) { /* nothing there */
         *current=0;
         *total=0;

+        /* logf(LOG_LOG, "r_pos: NULL  %0.1f %0.1f",  *current, *total);*/

         return;
     }
     *current=mrfd->hits;
     *total=*current*stot/scur;
         return;
     }
     *current=mrfd->hits;
     *total=*current*stot/scur;

+    /*logf(LOG_LOG, "r_pos: =  %0.1f %0.1f",  *current, *total);*/

 }
 
 
 }
 
 

@@ -595,12 +597,10 @@ static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm)
     for (i=0;i<info->no_rsets;i++)
     {
         rset_getterms(info->rsets[i], terms, maxterms, curterm);
     for (i=0;i<info->no_rsets;i++)
     {
         rset_getterms(info->rsets[i], terms, maxterms, curterm);

-        yaz_log(LOG_DEBUG,"rsmulti: getterms: i=%d *cur=%d",i,*curterm);
-        /* FIXME - remove this log once we know it works */

         if ( ( (*curterm) > firstterm+1 ) &&
              ( (*curterm) <= maxterms ) &&
              ( terms[(*curterm)-1] == terms[firstterm] ) )
         if ( ( (*curterm) > firstterm+1 ) &&
              ( (*curterm) <= maxterms ) &&
              ( terms[(*curterm)-1] == terms[firstterm] ) )

-            *curterm--; /* forget the term, seen that before */
+            (*curterm)--; /* forget the term, seen that before */

     }
 }
 
     }
 }