Generic snippet support. Unlike previous versions of snippet
[idzebra-moved-to-github.git] / index / zsets.c
index 982e6d1..cdfc2d2 100644 (file)
@@ -1,4 +1,4 @@
-/* $Id: zsets.c,v 1.115 2007-01-15 15:10:17 adam Exp $
+/* $Id: zsets.c,v 1.122 2007-08-21 11:06:47 adam Exp $
    Copyright (C) 1995-2007
    Index Data ApS
 
@@ -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) */
@@ -95,17 +96,14 @@ static void loglevels(void)
 }
 
 
-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;
     Z_SortKeySpecList *sort_sequence;
     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 */
@@ -115,11 +113,9 @@ ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
     for (i = 0; i<sort_sequence->num_specs; i++)
         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,
+    res = rpn_search_top(zh, rpn->RPNStructure, rpn->attributeSetId,
                         nmem, rset_nmem,
                         sort_sequence,
                         sset->num_bases, sset->basenames,
@@ -147,15 +143,17 @@ 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)
+ZEBRA_RES resultSetAddRPN(ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
+                          int num_bases, char **basenames,
+                          const char *setname,
+                          zint *hits, int *estimated_hit_count)
 {
     ZebraSet zebraSet;
     int i;
     ZEBRA_RES res;
 
-    zh->hits = 0;
+    *hits = 0;
+    *estimated_hit_count = 0;
 
     zebraSet = resultSetAdd(zh, setname, 1);
     if (!zebraSet)
@@ -173,7 +171,10 @@ 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;
@@ -273,6 +274,7 @@ 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;
 }
 
@@ -931,6 +933,14 @@ ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem,
        kno++;
         if (this_sys != psysno)
         {
+            if ((sset->hits & 255) == 0 && zh->break_handler_func)
+            {
+                if (zh->break_handler_func(zh->break_handler_data))
+                {
+                    rset_set_hits_limit(rset, 0);
+                    break;
+                }
+            }
             (sset->hits)++;
             psysno = this_sys;
             resultSetInsertSort(zh, sset,
@@ -987,11 +997,11 @@ 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);
 
-
     rank_class = zebraRankLookup(zh, rank_handler_name);
     if (!rank_class)
     {
@@ -1005,10 +1015,8 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
        struct rank_control *rc = rank_class->control;
        int score;
        zint count = 0;
-       
-       void *handle =
-           (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem,
-                         terms, numTerms);
+        void *handle = (*rc->begin) (zh->reg, rank_class->class_handle, rset,
+                                     nmem, terms, numTerms);
        zint psysno = 0;  /* previous doc id / sys no */
        zint pstaticrank = 0; /* previous static rank */
        int stop_flag = 0;
@@ -1022,8 +1030,16 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
                key_logdump_txt(log_level_searchhits, &key, termid->name);
            if (this_sys != psysno) 
            {   /* new record .. */
+                if (!(rfd->counted_items & 255) && zh->break_handler_func)
+                {
+                    if (zh->break_handler_func(zh->break_handler_data))
+                    {
+                        yaz_log(YLOG_LOG, "Aborted search");
+                        stop_flag = 1;
+                    }
+                }
                if (rfd->counted_items > rset->hits_limit)
-                   break;
+                    stop_flag = 1;
                if (psysno)
                {   /* only if we did have a previous record */
                    score = (*rc->calc) (handle, psysno, pstaticrank,
@@ -1031,9 +1047,13 @@ ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
                    /* insert the hit. A=Ascending */
                    resultSetInsertRank (zh, sort_info, psysno, score, 'A');
                    count++;
-                   if (stop_flag)
-                       break;
                }
+                if (stop_flag)
+                {
+                    zebraSet->estimated_hit_count = 1;
+                    rset_set_hits_limit(rset, 0);
+                    break;
+                }
                psysno = this_sys;
                if (zh->m_staticrank)
                    pstaticrank = key.mem[0];
@@ -1184,7 +1204,11 @@ ZEBRA_RES zebra_result_set_term_info(ZebraHandle zh, const char *setname,
                    if (ret == (size_t)(-1))
                        *termlen = 0;
                    else
+                    {
+                       yaz_iconv(zh->iconv_from_utf8, 0, 0, 
+                                  &outbuf, &outleft);
                        *termlen = outbuf - termbuf;
+                    }
                }
                else
                {
@@ -1249,7 +1273,7 @@ ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname,
                struct ord_list *ol;
                for (ol = termid->ol; ol; ol = ol->next)
                {
-                   zebra_snippets_append(snippets, key.mem[key.len-1],
+                   zebra_snippets_append(snippets, key.mem[key.len-1], 0,
                                          ol->ord, termid->name);
                }
            }