1 /* $Id: rank1.c,v 1.10 2002-08-02 19:26:55 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
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
35 struct rank_class_info {
39 struct rank_term_info {
47 struct rank_set_info {
51 struct rank_term_info *entries;
54 static int log2_int (unsigned g)
63 * create: Creates/Initialises this rank handler. This routine is
64 * called exactly once. The routine returns the class_handle.
66 static void *create (struct zebra_register *reg)
68 struct rank_class_info *ci = (struct rank_class_info *)
69 xmalloc (sizeof(*ci));
71 logf (LOG_DEBUG, "rank-1 create");
76 * destroy: Destroys this rank handler. This routine is called
77 * when the handler is no longer needed - i.e. when the server
78 * dies. The class_handle was previously returned by create.
80 static void destroy (struct zebra_register *reg, void *class_handle)
82 struct rank_class_info *ci = (struct rank_class_info *) class_handle;
84 logf (LOG_DEBUG, "rank-1 destroy");
90 * begin: Prepares beginning of "real" ranking. Called once for
91 * each result set. The returned handle is a "set handle" and
92 * will be used in each of the handlers below.
94 static void *begin (struct zebra_register *reg, void *class_handle, RSET rset)
96 struct rank_set_info *si = (struct rank_set_info *) xmalloc (sizeof(*si));
99 logf (LOG_DEBUG, "rank-1 begin");
100 si->no_entries = rset->no_rset_terms;
101 si->no_rank_entries = 0;
102 si->entries = (struct rank_term_info *)
103 xmalloc (sizeof(*si->entries)*si->no_entries);
104 for (i = 0; i < si->no_entries; i++)
106 int g = rset->rset_terms[i]->nn;
107 if (!strncmp (rset->rset_terms[i]->flags, "rank,", 5))
109 yaz_log (LOG_LOG, "%s", rset->rset_terms[i]->flags);
110 si->entries[i].rank_flag = 1;
111 si->entries[i].rank_weight = atoi (rset->rset_terms[i]->flags+5);
112 yaz_log (LOG_LOG, "i=%d weight=%d", i, si->entries[i].rank_weight);
113 (si->no_rank_entries)++;
116 si->entries[i].rank_flag = 0;
117 si->entries[i].local_occur = 0;
118 si->entries[i].global_occur = g;
119 si->entries[i].global_inv = 32 - log2_int (g);
120 logf (LOG_DEBUG, "-------- %d ------", 32 - log2_int (g));
126 * end: Terminates ranking process. Called after a result set
129 static void end (struct zebra_register *reg, void *set_handle)
131 struct rank_set_info *si = (struct rank_set_info *) set_handle;
132 logf (LOG_DEBUG, "rank-1 end");
138 * add: Called for each word occurence in a result set. This routine
139 * should be as fast as possible. This routine should "incrementally"
142 static void add (void *set_handle, int seqno, int term_index)
144 struct rank_set_info *si = (struct rank_set_info *) set_handle;
145 /*logf (LOG_DEBUG, "rank-1 add seqno=%d term_index=%d", seqno, term_index);*/
146 si->last_pos = seqno;
147 si->entries[term_index].local_occur++;
151 * calc: Called for each document in a result. This handler should
152 * produce a score based on previous call(s) to the add handler. The
153 * score should be between 0 and 1000. If score cannot be obtained
154 * -1 should be returned.
156 static int calc (void *set_handle, int sysno)
158 int i, lo, divisor, score = 0;
159 struct rank_set_info *si = (struct rank_set_info *) set_handle;
161 if (!si->no_rank_entries)
164 for (i = 0; i < si->no_entries; i++)
165 if (si->entries[i].rank_flag && (lo = si->entries[i].local_occur))
166 score += (8+log2_int (lo)) * si->entries[i].global_inv *
167 si->entries[i].rank_weight;
168 divisor = si->no_rank_entries * (8+log2_int (si->last_pos/si->no_entries));
169 score = score / divisor;
170 yaz_log (LOG_LOG, "sysno=%d score=%d", sysno, score);
173 for (i = 0; i < si->no_entries; i++)
174 si->entries[i].local_occur = 0;
179 * Pseudo-meta code with sequence of calls as they occur in a
180 * server. Handlers are prefixed by --:
196 static struct rank_control rank_control = {
206 struct rank_control *rank1_class = &rank_control;