For estimated hit counts, Zebra returns resultSetStatus=estimtate as

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

diff --git a/index/zsets.c b/index/zsets.c
index c4ee2ff..9b045d4 100644 (file)
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,5 +1,5 @@
-/* $Id: zsets.c,v 1.111 2006-09-20 10:51:25 adam Exp $
-   Copyright (C) 1995-2006
+/* $Id: zsets.c,v 1.116 2007-01-16 15:01:15 adam Exp $
+   Copyright (C) 1995-2007
    Index Data ApS
 
 This file is part of the Zebra server.
@@ -58,6 +58,7 @@ struct zebra_set {
     int term_entries_max;
     struct zebra_set *next;
     int locked;
+    int estimated_hit_count;
 
     zint cache_position;  /* last position */
     RSFD cache_rfd;       /* rfd (NULL if not existing) */
@@ -83,7 +84,7 @@ static int log_level_searchhits=0;
 static int log_level_searchterms=0;
 static int log_level_resultsets=0;
 
-static void loglevels()
+static void loglevels(void)
 {
     if (log_level_set)
         return;
@@ -94,8 +95,9 @@ static void loglevels()
     log_level_set = 1;
 }
 
-ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
-                         Z_RPNQuery *rpn, ZebraSet sset)
+
+static ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
+                                 Z_RPNQuery *rpn, ZebraSet sset)
 {
     RSET rset = 0;
     oident *attrset;
@@ -103,8 +105,6 @@ ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
     int sort_status, i;
     ZEBRA_RES res = ZEBRA_OK;
 
-    zh->hits = 0;
-
     sort_sequence = (Z_SortKeySpecList *)
         nmem_malloc(nmem, sizeof(*sort_sequence));
     sort_sequence->num_specs = 10; /* FIXME - Hard-coded number */
@@ -115,6 +115,9 @@ ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
         sort_sequence->specs[i] = 0;
     
     attrset = oid_getentbyoid (rpn->attributeSetId);
+
+    rpn_get_top_approx_limit(zh, rpn->RPNStructure, &sset->approx_limit);
+
     res = rpn_search_top(zh, rpn->RPNStructure, attrset->value,
                         nmem, rset_nmem,
                         sort_sequence,
@@ -145,13 +148,17 @@ ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
 
 ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
                           int num_bases, char **basenames,
-                          const char *setname)
+                          const char *setname,
+                           zint *hits, int *estimated_hit_count,
+                           int *partial_resultset)
 {
     ZebraSet zebraSet;
     int i;
     ZEBRA_RES res;
 
-    zh->hits = 0;
+    *hits = 0;
+    *estimated_hit_count = 0;
+    *partial_resultset = 0;
 
     zebraSet = resultSetAdd(zh, setname, 1);
     if (!zebraSet)
@@ -169,7 +176,10 @@ ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
 
     res = resultSetSearch(zh, zebraSet->nmem, zebraSet->rset_nmem,
                          rpn, zebraSet);
-    zh->hits = zebraSet->hits;
+    *hits = zebraSet->hits;
+    if (zebraSet->estimated_hit_count)
+        *estimated_hit_count = 1;
+
     if (zebraSet->rset)
         zebraSet->rpn = rpn;
     zebraSet->locked = 0;
@@ -269,6 +279,7 @@ ZebraSet resultSetAdd(ZebraHandle zh, const char *name, int ov)
     s->cache_position = 0;
     s->cache_rfd = 0;
     s->approx_limit = zh->approx_limit;
+    s->estimated_hit_count = 0;
     return s;
 }
 
@@ -548,15 +559,15 @@ void resultSetInsertSort(ZebraHandle zh, ZebraSet sset,
     struct zset_sort_info *sort_info = sset->sort_info;
     int i, j;
 
-    sortIdx_sysno (zh->reg->sortIdx, sysno);
+    zebra_sort_sysno(zh->reg->sort_index, sysno);
     for (i = 0; i<num_criteria; i++)
     {
         char *this_entry_buf = tmp_cmp_buf[i];
         memset(this_entry_buf, '\0', SORT_IDX_ENTRYSIZE);
         if (criteria[i].ord != -1)
         {
-            sortIdx_type(zh->reg->sortIdx, criteria[i].ord);
-            sortIdx_read(zh->reg->sortIdx, this_entry_buf);
+            zebra_sort_type(zh->reg->sort_index, criteria[i].ord);
+            zebra_sort_read(zh->reg->sort_index, this_entry_buf);
         }
     }
     i = sort_info->num_entries;
@@ -983,6 +994,7 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
     sort_info = zebraSet->sort_info;
     sort_info->num_entries = 0;
     zebraSet->hits = 0;
+    zebraSet->estimated_hit_count = 0;
     rset_getterms(rset, 0, 0, &n);
     terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
     rset_getterms(rset, terms, n, &numTerms);
@@ -1019,7 +1031,10 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
            if (this_sys != psysno) 
            {   /* new record .. */
                if (rfd->counted_items > rset->hits_limit)
+                {
+                    zebraSet->estimated_hit_count = 1;
                    break;
+                }
                if (psysno)
                {   /* only if we did have a previous record */
                    score = (*rc->calc) (handle, psysno, pstaticrank,