X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fzrpn.c;h=9927949690ce70432bb8a6d878f98d3c848fab0d;hb=47eeb5384a8fae8bcac8afb8a84782ca094683f9;hp=1b74b9a1be64fdd8462a0fceda83029982bfc149;hpb=7a2d0f25682890bde5d8f2883d6020df2ed0b365;p=idzebra-moved-to-github.git

diff --git a/index/zrpn.c b/index/zrpn.c
index 1b74b9a..9927949 100644
--- a/index/zrpn.c
+++ b/index/zrpn.c
@@ -1,4 +1,4 @@
-/* $Id: zrpn.c,v 1.197 2005-06-07 14:53:39 adam Exp $
+/* $Id: zrpn.c,v 1.200 2005-06-14 20:28:54 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -24,7 +24,8 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 #include <assert.h>
 #ifdef WIN32
 #include <io.h>
-#else
+#endif
+#if HAVE_UNISTD_H
 #include <unistd.h>
 #endif
 #include <ctype.h>
@@ -1432,50 +1433,6 @@ static ZEBRA_RES trans_scan_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
     return ZEBRA_OK;
 }
 
-char *normalize_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
-                     const char *termz, NMEM stream, unsigned reg_id)
-{
-    WRBUF wrbuf = 0;
-    AttrType truncation;
-    int truncation_value;
-    char *ex_list = 0;
-
-    attr_init(&truncation, zapt, 5);
-    truncation_value = attr_find(&truncation, NULL);
-
-    switch (truncation_value)
-    {
-    default:
-        ex_list = "";
-        break;
-    case 101:
-        ex_list = "#";
-        break;
-    case 102:
-    case 103:
-        ex_list = 0;
-        break;
-    case 104:
-        ex_list = "!#";
-        break;
-    case 105:
-        ex_list = "!*";
-        break;
-    }
-    if (ex_list)
-        wrbuf = zebra_replace(zh->reg->zebra_maps, reg_id, ex_list,
-                              termz, strlen(termz));
-    if (!wrbuf)
-        return nmem_strdup(stream, termz);
-    else
-    {
-        char *buf = (char*) nmem_malloc(stream, wrbuf_len(wrbuf)+1);
-        memcpy (buf, wrbuf_buf(wrbuf), wrbuf_len(wrbuf));
-        buf[wrbuf_len(wrbuf)] = '\0';
-        return buf;
-    }
-}
-
 static void grep_info_delete(struct grep_info *grep_info)
 {
 #ifdef TERM_COUNT
@@ -1533,7 +1490,7 @@ static ZEBRA_RES grep_info_prepare(ZebraHandle zh,
 /**
   \brief Create result set(s) for list of terms
   \param zh Zebra Handle
-  \param termz_org term as used in query but converted to UTF-8
+  \param termz term as used in query but converted to UTF-8
   \param attributeSet default attribute set
   \param stream memory for result
   \param reg_type register type ('w', 'p',..)
@@ -1549,7 +1506,7 @@ static ZEBRA_RES grep_info_prepare(ZebraHandle zh,
 */
 static ZEBRA_RES term_list_trunc(ZebraHandle zh,
 				 Z_AttributesPlusTerm *zapt,
-				 const char *termz_org,
+				 const char *termz,
 				 oid_value attributeSet,
 				 NMEM stream,
 				 int reg_type, int complete_flag,
@@ -1561,7 +1518,6 @@ static ZEBRA_RES term_list_trunc(ZebraHandle zh,
 {
     char term_dst[IT_MAX_WORD+1];
     struct grep_info grep_info;
-    char *termz = normalize_term(zh, zapt, termz_org, stream, reg_type);
     const char *termp = termz;
     int alloc_sets = 0;
 
@@ -2698,28 +2654,29 @@ void zebra_term_untrans_iconv(ZebraHandle zh, NMEM stream, int reg_type,
         *dst = nmem_strdup(stream, term_src);
 }
 
-static void count_set (RSET r, int *count)
+static void count_set(ZebraHandle zh, RSET rset, zint *count)
 {
     zint psysno = 0;
-    int kno = 0;
     struct it_key key;
     RSFD rfd;
 
     yaz_log(YLOG_DEBUG, "count_set");
 
+    rset->hits_limit = zh->approx_limit;
+
     *count = 0;
-    rfd = rset_open (r, RSETF_READ);
-    while (rset_read (rfd, &key,0 /* never mind terms */))
+    rfd = rset_open(rset, RSETF_READ);
+    while (rset_read(rfd, &key,0 /* never mind terms */))
     {
         if (key.mem[0] != psysno)
         {
             psysno = key.mem[0];
-            (*count)++;
+	    if (rfd->counted_items >= rset->hits_limit)
+		break;
         }
-        kno++;
     }
     rset_close (rfd);
-    yaz_log(YLOG_DEBUG, "%d keys, %d records", kno, *count);
+    *count = rset->hits_count;
 }
 
 ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
@@ -2994,6 +2951,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
         }
 	if (lo >= 0)
 	{
+	    zint count;
 	    /* merge with limit_set if given */
 	    if (limit_set)
 	    {
@@ -3006,7 +2964,8 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
 					  2, rsets);
 	    }
 	    /* count it */
-	    count_set(rset, &glist[lo].occurrences);
+	    count_set(zh, rset, &count);
+	    glist[lo].occurrences = count;
 	    rset_delete(rset);
 	}
     }
@@ -3033,6 +2992,7 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
 	const char *tst;
 	RSET rset;
 	int lo = before-1-i; /* offset in result list */
+	zint count;
 	
 	for (j = 0; j <ord_no; j++)
 	{
@@ -3089,7 +3049,8 @@ ZEBRA_RES rpn_scan(ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
 	    rset = rsmulti_and_create(rset_nmem, kc,
 				      kc->scope, 2, rsets);
 	}
-	count_set (rset, &glist[lo].occurrences);
+	count_set(zh, rset, &count);
+	glist[lo].occurrences = count;
 	rset_delete (rset);
     }
     (*kc->dec)(kc);