1 /* $Id: rank1.c,v 1.15 2004-08-04 08:35:23 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
5 This file is part of the Zebra server.
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
37 struct rank_class_info {
41 struct rank_term_info {
49 struct rank_set_info {
53 struct rank_term_info *entries;
56 static int log2_int (unsigned g)
65 * create: Creates/Initialises this rank handler. This routine is
66 * called exactly once. The routine returns the class_handle.
68 static void *create (ZebraHandle zh)
70 struct rank_class_info *ci = (struct rank_class_info *)
71 xmalloc (sizeof(*ci));
73 yaz_log (LOG_DEBUG, "rank-1 create");
78 * destroy: Destroys this rank handler. This routine is called
79 * when the handler is no longer needed - i.e. when the server
80 * dies. The class_handle was previously returned by create.
82 static void destroy (struct zebra_register *reg, void *class_handle)
84 struct rank_class_info *ci = (struct rank_class_info *) class_handle;
86 yaz_log (LOG_DEBUG, "rank-1 destroy");
92 * begin: Prepares beginning of "real" ranking. Called once for
93 * each result set. The returned handle is a "set handle" and
94 * will be used in each of the handlers below.
96 static void *begin (struct zebra_register *reg, void *class_handle, RSET rset)
98 struct rank_set_info *si = (struct rank_set_info *) xmalloc (sizeof(*si));
102 yaz_log (LOG_LOG, "rank-1 begin");
104 si->no_entries = rset->no_rset_terms;
105 si->no_rank_entries = 0;
106 si->entries = (struct rank_term_info *)
107 xmalloc (sizeof(*si->entries)*si->no_entries);
108 for (i = 0; i < si->no_entries; i++)
110 int g = rset->rset_terms[i]->nn;
112 yaz_log(LOG_LOG, "i=%d flags=%s", i, rset->rset_terms[i]->flags);
114 if (!strncmp (rset->rset_terms[i]->flags, "rank,", 5))
116 const char *cp = strstr(rset->rset_terms[i]->flags+4, ",w=");
117 si->entries[i].rank_flag = 1;
119 si->entries[i].rank_weight = atoi (cp+3);
121 si->entries[i].rank_weight = 34;
123 yaz_log (LOG_LOG, " i=%d weight=%d", i,
124 si->entries[i].rank_weight);
126 (si->no_rank_entries)++;
129 si->entries[i].rank_flag = 0;
130 si->entries[i].local_occur = 0;
131 si->entries[i].global_occur = g;
132 si->entries[i].global_inv = 32 - log2_int (g);
133 yaz_log (LOG_DEBUG, " global_inv = %d g = %d", 32 - log2_int (g), g);
139 * end: Terminates ranking process. Called after a result set
142 static void end (struct zebra_register *reg, void *set_handle)
144 struct rank_set_info *si = (struct rank_set_info *) set_handle;
145 yaz_log (LOG_DEBUG, "rank-1 end");
151 * add: Called for each word occurence in a result set. This routine
152 * should be as fast as possible. This routine should "incrementally"
155 static void add (void *set_handle, int seqno, int term_index)
157 struct rank_set_info *si = (struct rank_set_info *) set_handle;
159 yaz_log (LOG_LOG, "rank-1 add seqno=%d term_index=%d", seqno, term_index);
161 si->last_pos = seqno;
162 si->entries[term_index].local_occur++;
166 * calc: Called for each document in a result. This handler should
167 * produce a score based on previous call(s) to the add handler. The
168 * score should be between 0 and 1000. If score cannot be obtained
169 * -1 should be returned.
171 static int calc (void *set_handle, zint sysno)
173 int i, lo, divisor, score = 0;
174 struct rank_set_info *si = (struct rank_set_info *) set_handle;
176 if (!si->no_rank_entries)
180 yaz_log(LOG_LOG, "calc");
182 for (i = 0; i < si->no_entries; i++)
185 yaz_log(LOG_LOG, "i=%d rank_flag=%d lo=%d",
186 i, si->entries[i].rank_flag, si->entries[i].local_occur);
188 if (si->entries[i].rank_flag && (lo = si->entries[i].local_occur))
189 score += (8+log2_int (lo)) * si->entries[i].global_inv *
190 si->entries[i].rank_weight;
192 divisor = si->no_rank_entries * (8+log2_int (si->last_pos/si->no_entries));
193 score = score / divisor;
195 yaz_log (LOG_LOG, "sysno=" ZINT_FORMAT " score=%d", sysno, score);
199 for (i = 0; i < si->no_entries; i++)
200 si->entries[i].local_occur = 0;
205 * Pseudo-meta code with sequence of calls as they occur in a
206 * server. Handlers are prefixed by --:
222 static struct rank_control rank_control = {
232 struct rank_control *rank1_class = &rank_control;