ICU: support @attr 5=2, @attr 5=3 in searches

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

diff --git a/index/rpnsearch.c b/index/rpnsearch.c
index b0c0ce3..f119eb5 100644 (file)
--- a/index/rpnsearch.c
+++ b/index/rpnsearch.c
@@ -1,5 +1,5 @@
 /* This file is part of the Zebra server.
-   Copyright (C) 1994-2009 Index Data
+   Copyright (C) 1994-2011 Index Data
 
 Zebra is free software; you can redistribute it and/or modify it under
 the terms of the GNU General Public License as published by the Free
@@ -17,6 +17,9 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 
 */
 
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
 #include <stdio.h>
 #include <assert.h>
 #ifdef WIN32
@@ -154,7 +157,7 @@ static int grep_handle(char *name, const char *info, void *p)
 }
 
 static int term_pre(zebra_map_t zm, const char **src,
-                   const char *ct1, const char *ct2, int first)
+                   const char *ct1, int first)
 {
     const char *s1, *s0 = *src;
     const char **map;
@@ -164,8 +167,6 @@ static int term_pre(zebra_map_t zm, const char **src,
     {
         if (ct1 && strchr(ct1, *s0))
             break;
-        if (ct2 && strchr(ct2, *s0))
-            break;
         s1 = s0;
         map = zebra_maps_input(zm, &s1, strlen(s1), first);
         if (**map != *CHR_SPACE)
@@ -236,7 +237,7 @@ static void add_non_space(const char *start, const char *end,
 static int term_100_icu(zebra_map_t zm,
                         const char **src, WRBUF term_dict, int space_split,
                         WRBUF display_term,
-                        int right_trunc)
+                        int mode)
 {
     int i;
     const char *res_buf = 0;
@@ -250,7 +251,7 @@ static int term_100_icu(zebra_map_t zm,
         return 0;
     }
     wrbuf_write(display_term, display_buf, display_len);
-    if (right_trunc)
+    if (mode)
     {
         /* ICU sort keys seem to be of the form
            basechars \x01 accents \x01 length
@@ -271,6 +272,8 @@ static int term_100_icu(zebra_map_t zm,
         }
         res_len = i; /* reduce res_len */
     }
+    if (mode & 2)
+        wrbuf_puts(term_dict, ".*");
     for (i = 0; i < res_len; i++)
     {
         if (strchr(REGEX_CHARS "\\", res_buf[i]))
@@ -280,8 +283,11 @@ static int term_100_icu(zebra_map_t zm,
             
         wrbuf_putc(term_dict, res_buf[i]);
     }
-    if (right_trunc)
+    if (mode & 1)
         wrbuf_puts(term_dict, ".*");
+    else if (mode)
+        wrbuf_puts(term_dict, "\x01\x01.*");
+        
     return 1;
 }
 
@@ -297,7 +303,7 @@ static int term_100(zebra_map_t zm,
     const char *space_start = 0;
     const char *space_end = 0;
 
-    if (!term_pre(zm, src, NULL, NULL, !space_split))
+    if (!term_pre(zm, src, 0, !space_split))
         return 0;
     s0 = *src;
     while (*s0)
@@ -350,7 +356,7 @@ static int term_101(zebra_map_t zm,
     const char **map;
     int i = 0;
 
-    if (!term_pre(zm, src, "#", "#", !space_split))
+    if (!term_pre(zm, src, "#", !space_split))
         return 0;
     s0 = *src;
     while (*s0)
@@ -387,7 +393,7 @@ static int term_103(zebra_map_t zm, const char **src,
     const char *s0;
     const char **map;
 
-    if (!term_pre(zm, src, "^\\()[].*+?|", "(", !space_split))
+    if (!term_pre(zm, src, "^\\()[].*+?|", !space_split))
         return 0;
     s0 = *src;
     if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] &&
@@ -432,7 +438,7 @@ static int term_102(zebra_map_t zm, const char **src,
 }
 
 
-/* term_104: handle term, process # and ! */
+/* term_104: handle term, process ?n * # */
 static int term_104(zebra_map_t zm, const char **src, 
                     WRBUF term_dict, int space_split, WRBUF display_term)
 {
@@ -440,7 +446,7 @@ static int term_104(zebra_map_t zm, const char **src,
     const char **map;
     int i = 0;
 
-    if (!term_pre(zm, src, "?*#", "?*#", !space_split))
+    if (!term_pre(zm, src, "?*#", !space_split))
         return 0;
     s0 = *src;
     while (*s0)
@@ -501,7 +507,7 @@ static int term_104(zebra_map_t zm, const char **src,
     return i;
 }
 
-/* term_105/106: handle term, where trunc = Process * and ! and right trunc */
+/* term_105/106: handle term, process * ! and possibly right_truncate */
 static int term_105(zebra_map_t zm, const char **src, 
                     WRBUF term_dict, int space_split,
                    WRBUF display_term, int right_truncate)
@@ -510,7 +516,7 @@ static int term_105(zebra_map_t zm, const char **src,
     const char **map;
     int i = 0;
 
-    if (!term_pre(zm, src, "*!", "*!", !space_split))
+    if (!term_pre(zm, src, "\\*!", !space_split))
         return 0;
     s0 = *src;
     while (*s0)
@@ -529,6 +535,13 @@ static int term_105(zebra_map_t zm, const char **src,
             wrbuf_putc(display_term, *s0);
             s0++;
         }
+        else if (*s0 == '\\')
+        {
+            i++;
+            wrbuf_puts(term_dict, "\\\\");
+            wrbuf_putc(display_term, *s0);
+            s0++;
+        }
        else
         {
            const char *s1 = s0;
@@ -1047,6 +1060,20 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
                     return ZEBRA_OK;
                 }
                 break;
+            case 2:
+                if (!term_100_icu(zm, &termp, term_dict, space_split, display_term, 2))
+                {
+                    *term_sub = 0;
+                    return ZEBRA_OK;
+                }
+                break;
+            case 3:
+                if (!term_100_icu(zm, &termp, term_dict, space_split, display_term, 3))
+                {
+                    *term_sub = 0;
+                    return ZEBRA_OK;
+                }
+                break;
             default:
                 zebra_setError_zint(zh,
                                     YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE,
@@ -1139,7 +1166,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
             }
             wrbuf_putc(term_dict, ')');
             break;
-        case 104:        /* process # and ! in term */
+        case 104:        /* process ?n * # term */
             wrbuf_putc(term_dict, '(');
             if (!term_104(zm, &termp, term_dict, space_split, display_term))
             {
@@ -1148,7 +1175,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
             }
             wrbuf_putc(term_dict, ')');
             break;
-        case 105:        /* process * and ! in term */
+        case 105:        /* process * ! in term and right truncate */
             wrbuf_putc(term_dict, '(');
             if (!term_105(zm, &termp, term_dict, space_split, display_term, 1))
             {
@@ -1157,7 +1184,7 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
             }
             wrbuf_putc(term_dict, ')');
             break;
-        case 106:        /* process * and ! in term */
+        case 106:        /* process * ! in term */
             wrbuf_putc(term_dict, '(');
             if (!term_105(zm, &termp, term_dict, space_split, display_term, 0))
             {
@@ -1803,6 +1830,7 @@ static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh,
                                        Z_AttributesPlusTerm *zapt,
                                        const char *termz,
                                        const Odr_oid *attributeSet,
+                                        zint hits_limit,
                                        NMEM stream,
                                        const char *index_type, 
                                         int complete_flag,
@@ -1818,7 +1846,7 @@ static ZEBRA_RES rpn_search_APT_numeric(ZebraHandle zh,
     ZEBRA_RES res;
     struct grep_info grep_info;
     int alloc_sets = 0;
-    zint hits_limit_value;
+    zint hits_limit_value = hits_limit;
     const char *term_ref_id_str = 0;
 
     zebra_term_limits_APT(zh, zapt, &hits_limit_value, &term_ref_id_str,
@@ -1977,7 +2005,7 @@ static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
     sk->u.sortAttributes->id = odr_oiddup_nmem(stream, attributeSet);
     sk->u.sortAttributes->list = zapt->attributes;
 
-    sks->sortRelation = (int *)
+    sks->sortRelation = (Odr_int *)
         nmem_malloc(stream, sizeof(*sks->sortRelation));
     if (sort_relation_value == 1)
         *sks->sortRelation = Z_SortKeySpec_ascending;
@@ -1986,7 +2014,7 @@ static ZEBRA_RES rpn_sort_spec(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
     else 
         *sks->sortRelation = Z_SortKeySpec_ascending;
 
-    sks->caseSensitivity = (int *)
+    sks->caseSensitivity = (Odr_int *)
         nmem_malloc(stream, sizeof(*sks->caseSensitivity));
     *sks->caseSensitivity = 0;
 
@@ -2375,7 +2403,8 @@ static ZEBRA_RES rpn_search_database(ZebraHandle zh,
     }
     else if (!strcmp(search_type, "numeric"))
     {
-        res = rpn_search_APT_numeric(zh, zapt, termz, attributeSet, stream,
+        res = rpn_search_APT_numeric(zh, zapt, termz, attributeSet, hits_limit,
+                                     stream,
                                     index_type, complete_flag, rank_type,
                                     xpath_use,
                                     rset_nmem,