Fixed bug #252: Sort does not work.

[idzebra-moved-to-github.git] / index / zsets.c

diff --git a/index/zsets.c b/index/zsets.c
index a26ac40..313a868 100644 (file)
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,5 +1,5 @@
-/* $Id: zsets.c,v 1.50 2004-08-04 08:35:24 adam Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
+/* $Id: zsets.c,v 1.49.2.3 2005-01-21 11:35:49 adam Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004,2005
    Index Data Aps
 
 This file is part of the Zebra server.
@@ -55,10 +55,14 @@ struct zebra_set {
     int term_entries_max;
     struct zebra_set *next;
     int locked;
+
+    int cache_position;  /* last position */
+    RSFD cache_rfd;      /* rfd (NULL if not existing) */
+    int cache_psysno;
 };
 
 struct zset_sort_entry {
-    zint sysno;
+    int sysno;
     int score;
     char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
 };
@@ -88,7 +92,7 @@ ZebraSet resultSetAddRPN (ZebraHandle zh, NMEM m,
     zebraSet->locked = 1;
     zebraSet->rpn = 0;
     zebraSet->nmem = m;
-
+    
     zebraSet->num_bases = num_bases;
     zebraSet->basenames = 
         nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames));
@@ -201,7 +205,11 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
        if (!ov || s->locked)
            return NULL;
        if (s->rset)
+       {
+           if (s->cache_rfd)
+               rset_close(s->rset, s->cache_rfd);
            rset_delete (s->rset);
+       }
        if (s->nmem)
            nmem_destroy (s->nmem);
     }
@@ -237,6 +245,8 @@ ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
     s->rset = 0;
     s->nmem = 0;
     s->rpn = 0;
+    s->cache_position = 0;
+    s->cache_rfd = 0;
     return s;
 }
 
@@ -268,8 +278,14 @@ void resultSetInvalidate (ZebraHandle zh)
     for (; s; s = s->next)
     {
         if (s->rset)
+       {
+           if (s->cache_rfd)
+               rset_close(s->rset, s->cache_rfd);
             rset_delete (s->rset);
+       }
         s->rset = 0;
+       s->cache_rfd = 0;
+       s->cache_position = 0;
     }
 }
 
@@ -309,7 +325,11 @@ void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
            if (s->nmem)
                nmem_destroy (s->nmem);
            if (s->rset)
+           {
+               if (s->cache_rfd)
+                   rset_close (s->rset, s->cache_rfd);
                rset_delete (s->rset);
+           }
            xfree (s->name);
            xfree (s);
        }
@@ -384,7 +404,7 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
        {
            int position = 0;
            int num_i = 0;
-           zint psysno = 0;
+           int psysno = 0;
            int term_index;
            RSFD rfd;
            struct it_key key;
@@ -393,17 +413,25 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
                position = sort_info->num_entries;
            while (num_i < num && positions[num_i] < position)
                num_i++;
-           rfd = rset_open (rset, RSETF_READ);
+           
+           if (sset->cache_rfd && 
+               num_i < num && positions[num_i] > sset->cache_position)
+           {
+               position = sset->cache_position;
+               rfd = sset->cache_rfd;
+               psysno = sset->cache_psysno;
+           } 
+           else
+           {
+               if (sset->cache_rfd)
+                   rset_close(rset, sset->cache_rfd);
+               rfd = rset_open (rset, RSETF_READ);
+           }
            while (num_i < num && rset_read (rset, rfd, &key, &term_index))
            {
-#if IT_KEY_NEW
-               zint this_sys = key.mem[0];
-#else
-               zint this_sys = key.sysno;
-#endif
-               if (this_sys != psysno)
+               if (key.sysno != psysno)
                {
-                   psysno = this_sys;
+                   psysno = key.sysno;
                    if (sort_info)
                    {
                        /* determine we alreay have this in our set */
@@ -424,7 +452,9 @@ ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
                    }
                }
            }
-           rset_close (rset, rfd);
+           sset->cache_position = position;
+           sset->cache_psysno = psysno;
+           sset->cache_rfd = rfd;
        }
     }
     return sr;
@@ -443,7 +473,7 @@ struct sortKeyInfo {
 
 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
                          struct sortKeyInfo *criteria, int num_criteria,
-                         zint sysno)
+                         int sysno)
 {
     struct zset_sort_entry this_entry;
     struct zset_sort_entry *new_entry = NULL;
@@ -516,8 +546,8 @@ void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
     new_entry->score = -1;
 }
 
-void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
-                         zint sysno, int score, int relation)
+void resultSetInsertRank(ZebraHandle zh, struct zset_sort_info *sort_info,
+                        int sysno, int score, int relation)
 {
     struct zset_sort_entry *new_entry = NULL;
     int i, j;
@@ -608,15 +638,14 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
                          ZebraSet sset, RSET rset,
                          Z_SortKeySpecList *sort_sequence, int *sort_status)
 {
-    int i;
-    zint psysno = 0;
+    int kno = 0;
+    int i, psysno = 0;
     struct it_key key;
     struct sortKeyInfo sort_criteria[3];
     int num_criteria;
     int term_index;
     RSFD rfd;
 
-    yaz_log (LOG_LOG, "resultSetSortSingle start");
     sset->sort_info->num_entries = 0;
 
     sset->hits = 0;
@@ -681,21 +710,18 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
     rfd = rset_open (rset, RSETF_READ);
     while (rset_read (rset, rfd, &key, &term_index))
     {
-#if IT_KEY_NEW
-       zint this_sys = key.mem[0];
-#else
-       zint this_sys = key.sysno;
-#endif
-        if (this_sys != psysno)
+       kno++;
+        if (key.sysno != psysno)
         {
            (sset->hits)++;
-            psysno = this_sys;
+            psysno = key.sysno;
            resultSetInsertSort (zh, sset,
                                 sort_criteria, num_criteria, psysno);
         }
     }
     rset_close (rset, rfd);
 
+    yaz_log (LOG_LOG, "%d keys, %d sysnos, sort", kno, sset->hits);
     for (i = 0; i < rset->no_rset_terms; i++)
        yaz_log (LOG_LOG, "term=\"%s\" nn=%d type=%s count=%d",
                  rset->rset_terms[i]->name,
@@ -704,7 +730,6 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
                  rset->rset_terms[i]->count);
 
     *sort_status = Z_SortResponse_success;
-    yaz_log (LOG_LOG, "resultSetSortSingle end");
 }
 
 RSET resultSetRef (ZebraHandle zh, const char *resultSetId)
@@ -732,8 +757,6 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
     zebraSet->hits = 0;
     rfd = rset_open (rset, RSETF_READ);
 
-    yaz_log (LOG_LOG, "resultSetRank");
-
     rank_class = zebraRankLookup (zh, rank_handler_name);
     if (!rank_class)
     {
@@ -744,32 +767,23 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
 
     if (rset_read (rset, rfd, &key, &term_index))
     {
-#if IT_KEY_NEW
-       zint psysno = key.mem[0];
-#else
-       zint psysno = key.sysno;
-#endif
+       int psysno = key.sysno;
        int score;
        void *handle =
            (*rc->begin) (zh->reg, rank_class->class_handle, rset);
        (zebraSet->hits)++;
        do
        {
-#if IT_KEY_NEW
-           zint this_sys = key.mem[0];
-#else
-           zint this_sys = key.sysno;
-#endif
            kno++;
-           if (this_sys != psysno)
+           if (key.sysno != psysno)
            {
                score = (*rc->calc) (handle, psysno);
 
                resultSetInsertRank (zh, sort_info, psysno, score, 'A');
                (zebraSet->hits)++;
-               psysno = this_sys;
+               psysno = key.sysno;
            }
-           (*rc->add) (handle, this_sys, term_index);
+           (*rc->add) (handle, key.seqno, term_index);
        }
        while (rset_read (rset, rfd, &key, &term_index));
        score = (*rc->calc) (handle, psysno);
@@ -778,6 +792,7 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
     }
     rset_close (rset, rfd);
 
+    yaz_log (LOG_LOG, "%d keys, %d sysnos, rank", kno, zebraSet->hits);
     for (i = 0; i < rset->no_rset_terms; i++)
        yaz_log (LOG_LOG, "term=\"%s\" nn=%d type=%s count=%d",
                  rset->rset_terms[i]->name,
@@ -785,7 +800,6 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
                  rset->rset_terms[i]->flags,
                  rset->rset_terms[i]->count);
     
-    yaz_log (LOG_LOG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
 }
 
 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)