Fixed bug #108: File updates sometimes produce duplicates.

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

diff --git a/index/zsets.c b/index/zsets.c
index 57b7e4b..313a868 100644 (file)
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,5 +1,5 @@
-/* $Id: zsets.c,v 1.48 2004-06-07 22:09:32 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.
@@ -55,6 +55,10 @@ 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 {
@@ -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);
        }
@@ -393,7 +413,20 @@ 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 (key.sysno != psysno)
@@ -419,7 +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;
@@ -511,8 +546,8 @@ void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
     new_entry->score = -1;
 }
 
-void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
-                         int 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;
@@ -603,6 +638,7 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
                          ZebraSet sset, RSET rset,
                          Z_SortKeySpecList *sort_sequence, int *sort_status)
 {
+    int kno = 0;
     int i, psysno = 0;
     struct it_key key;
     struct sortKeyInfo sort_criteria[3];
@@ -610,7 +646,6 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
     int term_index;
     RSFD rfd;
 
-    yaz_log (LOG_LOG, "resultSetSortSingle start");
     sset->sort_info->num_entries = 0;
 
     sset->hits = 0;
@@ -675,6 +710,7 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
     rfd = rset_open (rset, RSETF_READ);
     while (rset_read (rset, rfd, &key, &term_index))
     {
+       kno++;
         if (key.sysno != psysno)
         {
            (sset->hits)++;
@@ -685,6 +721,7 @@ void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
     }
     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,
@@ -693,7 +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)
@@ -714,15 +750,19 @@ void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
     ZebraRankClass rank_class;
     struct rank_control *rc;
     struct zset_sort_info *sort_info;
+    const char *rank_handler_name = res_get_def(zh->res, "rank", "rank-1");
 
     sort_info = zebraSet->sort_info;
     sort_info->num_entries = 0;
     zebraSet->hits = 0;
     rfd = rset_open (rset, RSETF_READ);
 
-    yaz_log (LOG_LOG, "resultSetRank");
-
-    rank_class = zebraRankLookup (zh, res_get_def(zh->res, "rank", "rank-1"));
+    rank_class = zebraRankLookup (zh, rank_handler_name);
+    if (!rank_class)
+    {
+        yaz_log (LOG_WARN, "No such rank handler: %s", rank_handler_name);
+        return;
+    }
     rc = rank_class->control;
 
     if (rset_read (rset, rfd, &key, &term_index))
@@ -752,6 +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,
@@ -759,7 +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)