WRBUF updates.
[idzebra-moved-to-github.git] / index / rpnsearch.c
index 3057580..aaae58a 100644 (file)
@@ -1,5 +1,5 @@
-/* $Id: rpnsearch.c,v 1.2 2006-10-12 12:28:42 adam Exp $
-   Copyright (C) 1995-2006
+/* $Id: rpnsearch.c,v 1.10 2007-03-19 21:50:39 adam Exp $
+   Copyright (C) 1995-2007
    Index Data ApS
 
 This file is part of the Zebra server.
@@ -78,19 +78,25 @@ struct grep_info {
     ISAM_P *isam_p_buf;
     int isam_p_size;        
     int isam_p_indx;
+    int trunc_max;
     ZebraHandle zh;
     int reg_type;
     ZebraSet termset;
 };        
 
-static void add_isam_p(const char *name, const char *info,
-                      struct grep_info *p)
+static int add_isam_p(const char *name, const char *info,
+                      struct grep_info *p)
 {
     if (!log_level_set)
     {
         log_level_rpn = yaz_log_module_level("rpn");
         log_level_set = 1;
     }
+    /* we may have to stop this madness.. NOTE: -1 so that if
+       truncmax == trunxlimit we do *not* generate result sets */
+    if (p->isam_p_indx >= p->trunc_max - 1)
+        return 1;
+
     if (p->isam_p_indx == p->isam_p_size)
     {
         ISAM_P *new_isam_p_buf;
@@ -140,12 +146,12 @@ static void add_isam_p(const char *name, const char *info,
                         index_name, term_tmp);
     }
     (p->isam_p_indx)++;
+    return 0;
 }
 
 static int grep_handle(char *name, const char *info, void *p)
 {
-    add_isam_p(name, info, (struct grep_info *) p);
-    return 0;
+    return add_isam_p(name, info, (struct grep_info *) p);
 }
 
 static int term_pre(ZebraMaps zebra_maps, int reg_type, const char **src,
@@ -1177,7 +1183,9 @@ static ZEBRA_RES string_term(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
                              grep_info, &max_pos, 
                              ord_len /* number of "exact" chars */,
                              grep_handle);
-        if (r)
+        if (r == 1)
+            zebra_set_partial_result(zh);
+        else if (r)
             yaz_log(YLOG_WARN, "dict_lookup_grep fail %d", r);
     }
     if (!bases_ok)
@@ -1202,47 +1210,59 @@ static ZEBRA_RES grep_info_prepare(ZebraHandle zh,
                                   struct grep_info *grep_info,
                                   int reg_type)
 {
-    AttrType termset;
-    int termset_value_numeric;
-    const char *termset_value_string;
-
 #ifdef TERM_COUNT
     grep_info->term_no = 0;
 #endif
+    grep_info->trunc_max = atoi(res_get_def(zh->res, "truncmax", "10000"));
     grep_info->isam_p_size = 0;
     grep_info->isam_p_buf = NULL;
     grep_info->zh = zh;
     grep_info->reg_type = reg_type;
     grep_info->termset = 0;
-    if (!zapt)
-        return ZEBRA_OK;
-    attr_init_APT(&termset, zapt, 8);
-    termset_value_numeric =
-        attr_find_ex(&termset, NULL, &termset_value_string);
-    if (termset_value_numeric != -1)
+    if (zapt)
     {
+        AttrType truncmax;
+        int truncmax_value;
+
+        attr_init_APT(&truncmax, zapt, 13);
+        truncmax_value = attr_find(&truncmax, NULL);
+        if (truncmax_value != -1)
+            grep_info->trunc_max = truncmax_value;
+    }
+    if (zapt)
+    {
+        AttrType termset;
+        int termset_value_numeric;
+        const char *termset_value_string;
+
+        attr_init_APT(&termset, zapt, 8);
+        termset_value_numeric =
+            attr_find_ex(&termset, NULL, &termset_value_string);
+        if (termset_value_numeric != -1)
+        {
 #if TERMSET_DISABLE
-        zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, "termset");
-        return ZEBRA_FAIL;
+            zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, "termset");
+            return ZEBRA_FAIL;
 #else
-        char resname[32];
-        const char *termset_name = 0;
-        if (termset_value_numeric != -2)
-        {
-    
-            sprintf(resname, "%d", termset_value_numeric);
-            termset_name = resname;
-        }
-        else
+            char resname[32];
+            const char *termset_name = 0;
+            if (termset_value_numeric != -2)
+            {
+                
+                sprintf(resname, "%d", termset_value_numeric);
+                termset_name = resname;
+            }
+            else
             termset_name = termset_value_string;
-        yaz_log(log_level_rpn, "creating termset set %s", termset_name);
-        grep_info->termset = resultSetAdd(zh, termset_name, 1);
-        if (!grep_info->termset)
-        {
-           zebra_setError(zh, YAZ_BIB1_ILLEGAL_RESULT_SET_NAME, termset_name);
-            return ZEBRA_FAIL;
-        }
+            yaz_log(log_level_rpn, "creating termset set %s", termset_name);
+            grep_info->termset = resultSetAdd(zh, termset_name, 1);
+            if (!grep_info->termset)
+            {
+                zebra_setError(zh, YAZ_BIB1_ILLEGAL_RESULT_SET_NAME, termset_name);
+                return ZEBRA_FAIL;
+            }
 #endif
+        }
     }
     return ZEBRA_OK;
 }
@@ -1692,7 +1712,10 @@ static int numeric_relation(ZebraHandle zh, Z_AttributesPlusTerm *zapt,
     yaz_log(log_level_rpn, "dict_lookup_grep: %s", term_tmp);
     r = dict_lookup_grep(zh->reg->dict, term_dict, 0, grep_info, max_pos,
                           0, grep_handle);
-    if (r)
+
+    if (r == 1)
+        zebra_set_partial_result(zh);
+    else if (r)
         yaz_log(YLOG_WARN, "dict_lookup_grep fail, rel = gt: %d", r);
     yaz_log(log_level_rpn, "%d positions", grep_info->isam_p_indx);
     return 1;
@@ -1863,21 +1886,35 @@ static ZEBRA_RES rpn_search_APT_local(ZebraHandle zh,
                                      RSET *rset,
                                      struct rset_key_control *kc)
 {
-    RSFD rsfd;
-    struct it_key key;
-    int sys;
-    *rset = rset_create_temp(rset_nmem, kc, kc->scope,
-                             res_get (zh->res, "setTmpDir"),0 );
-    rsfd = rset_open(*rset, RSETF_WRITE);
+    Record rec;
+    zint sysno = atozint(termz);
     
-    sys = atoi(termz);
-    if (sys <= 0)
-        sys = 1;
-    key.mem[0] = sys;
-    key.mem[1] = 1;
-    key.len = 2;
-    rset_write (rsfd, &key);
-    rset_close (rsfd);
+    if (sysno <= 0)
+        sysno = 0;
+    rec = rec_get(zh->reg->records, sysno);
+    if (!rec)
+        sysno = 0;
+
+    rec_free(&rec);
+
+    if (sysno <= 0)
+    {
+        *rset = rset_create_null(rset_nmem, kc, 0);
+    }
+    else
+    {
+        RSFD rsfd;
+        struct it_key key;
+        *rset = rset_create_temp(rset_nmem, kc, kc->scope,
+                                 res_get(zh->res, "setTmpDir"), 0);
+        rsfd = rset_open(*rset, RSETF_WRITE);
+        
+        key.mem[0] = sysno;
+        key.mem[1] = 1;
+        key.len = 2;
+        rset_write(rsfd, &key);
+        rset_close(rsfd);
+    }
     return ZEBRA_OK;
 }
 
@@ -2138,26 +2175,25 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh,
                         cp++;
                     }
                 }
-                wrbuf_puts(wbuf, "");
                 rset_attr = xpath_trunc(
-                    zh, stream, '0', wrbuf_buf(wbuf), ZEBRA_XPATH_ATTR_NAME, 
+                    zh, stream, '0', wrbuf_cstr(wbuf), ZEBRA_XPATH_ATTR_NAME, 
                     rset_nmem, kc);
-                wrbuf_free(wbuf, 1);
+                wrbuf_destroy(wbuf);
             } 
             else 
             {
                 if (!first_path)
                 {
-                    wrbuf_free(xpath_rev, 1);
+                    wrbuf_destroy(xpath_rev);
                     continue;
                 }
             }
-            yaz_log(log_level_rpn, "xpath_rev (%d) = %.*s", level, 
-                    wrbuf_len(xpath_rev), wrbuf_buf(xpath_rev));
+            yaz_log(log_level_rpn, "xpath_rev (%d) = %s", level, 
+                    wrbuf_cstr(xpath_rev));
             if (wrbuf_len(xpath_rev))
             {
                 rset_start_tag = xpath_trunc(zh, stream, '0', 
-                                             wrbuf_buf(xpath_rev),
+                                             wrbuf_cstr(xpath_rev),
                                              ZEBRA_XPATH_ELM_BEGIN, 
                                              rset_nmem, kc);
                 if (always_matches)
@@ -2165,7 +2201,7 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh,
                 else
                 {
                     rset_end_tag = xpath_trunc(zh, stream, '0', 
-                                               wrbuf_buf(xpath_rev),
+                                               wrbuf_cstr(xpath_rev),
                                                ZEBRA_XPATH_ELM_END, 
                                                rset_nmem, kc);
                     
@@ -2174,7 +2210,7 @@ ZEBRA_RES rpn_search_xpath(ZebraHandle zh,
                                                rset_end_tag, rset_attr);
                 }
             }
-            wrbuf_free(xpath_rev, 1);
+            wrbuf_destroy(xpath_rev);
             first_path = 0;
         }
     }
@@ -2318,6 +2354,37 @@ static ZEBRA_RES rpn_search_structure(ZebraHandle zh, Z_RPNStructure *zs,
                                      Z_Operator *parent_op,
                                      struct rset_key_control *kc);
 
+ZEBRA_RES rpn_get_top_approx_limit(ZebraHandle zh, Z_RPNStructure *zs,
+                                   zint *approx_limit)
+{
+    ZEBRA_RES res = ZEBRA_OK;
+    if (zs->which == Z_RPNStructure_complex)
+    {
+        if (res == ZEBRA_OK)
+            res = rpn_get_top_approx_limit(zh, zs->u.complex->s1,
+                                           approx_limit);
+        if (res == ZEBRA_OK)
+            res = rpn_get_top_approx_limit(zh, zs->u.complex->s2,
+                                           approx_limit);
+    }
+    else if (zs->which == Z_RPNStructure_simple)
+    {
+        if (zs->u.simple->which == Z_Operand_APT)
+        {
+            Z_AttributesPlusTerm *zapt = zs->u.simple->u.attributesPlusTerm;
+            AttrType global_hits_limit_attr;
+            int l;
+            
+            attr_init_APT(&global_hits_limit_attr, zapt, 12);
+            
+            l = attr_find(&global_hits_limit_attr, NULL);
+            if (l != -1)
+                *approx_limit = l;
+        }
+    }
+    return res;
+}
+
 ZEBRA_RES rpn_search_top(ZebraHandle zh, Z_RPNStructure *zs,
                         oid_value attributeSet, 
                         NMEM stream, NMEM rset_nmem,