X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Frank1.c;h=581e78a4b33d1d5458ca31777c4174f47be8b484;hb=73bedd5ba9152c9c107b502fae65723b551aff09;hp=84d41592a1bc4671d8cae7f85eb22751f7d76663;hpb=ef696645cc3b7e0f4027008d1dc589c0f0f90c1f;p=idzebra-moved-to-github.git diff --git a/index/rank1.c b/index/rank1.c index 84d4159..581e78a 100644 --- a/index/rank1.c +++ b/index/rank1.c @@ -1,26 +1,26 @@ -/* - * Copyright (C) 1998-1999, Index Data - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: rank1.c,v $ - * Revision 1.5 1999-05-26 07:49:13 adam - * C++ compilation. - * - * Revision 1.4 1999/02/02 14:51:01 adam - * Updated WIN32 code specific sections. Changed header. - * - * Revision 1.3 1998/06/12 12:21:53 adam - * Fixed memory-leak. - * - * Revision 1.2 1998/03/05 13:03:29 adam - * Improved ranking. - * - * Revision 1.1 1998/03/05 08:45:12 adam - * New result set model and modular ranking system. Moved towards - * descent server API. System information stored as "SGML" records. - * - */ +/* $Id: rank1.c,v 1.18 2004-10-26 15:32:11 heikki Exp $ + Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003 + Index Data Aps + +This file is part of the Zebra server. + +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 +Software Foundation; either version 2, or (at your option) any later +version. + +Zebra is distributed in the hope that it will be useful, but WITHOUT ANY +WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +for more details. + +You should have received a copy of the GNU General Public License +along with Zebra; see the file LICENSE.zebra. If not, write to the +Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA +02111-1307, USA. +*/ + + #include #include @@ -30,7 +30,9 @@ #include #endif -#include "zserver.h" +#define DEBUG_RANK 1 + +#include "index.h" struct rank_class_info { int dummy; @@ -38,9 +40,12 @@ struct rank_class_info { struct rank_term_info { int local_occur; - int global_occur; + zint global_occur; int global_inv; int rank_flag; + int rank_weight; + TERMID term; + int term_index; }; struct rank_set_info { @@ -48,6 +53,7 @@ struct rank_set_info { int no_entries; int no_rank_entries; struct rank_term_info *entries; + NMEM nmem; }; static int log2_int (unsigned g) @@ -64,10 +70,10 @@ static int log2_int (unsigned g) */ static void *create (ZebraHandle zh) { - struct rank_class_info *ci = (struct rank_class_info *) - xmalloc (sizeof(*ci)); + struct rank_class_info *ci = + (struct rank_class_info *) xmalloc (sizeof(*ci)); - logf (LOG_DEBUG, "rank-1 create"); + yaz_log (LOG_DEBUG, "rank-1 create"); return ci; } @@ -76,44 +82,67 @@ static void *create (ZebraHandle zh) * when the handler is no longer needed - i.e. when the server * dies. The class_handle was previously returned by create. */ -static void destroy (ZebraHandle zh, void *class_handle) +static void destroy (struct zebra_register *reg, void *class_handle) { struct rank_class_info *ci = (struct rank_class_info *) class_handle; - logf (LOG_DEBUG, "rank-1 destroy"); + yaz_log (LOG_DEBUG, "rank-1 destroy"); xfree (ci); } -/* +/** * begin: Prepares beginning of "real" ranking. Called once for * each result set. The returned handle is a "set handle" and * will be used in each of the handlers below. */ -static void *begin (ZebraHandle zh, void *class_handle, RSET rset) +static void *begin (struct zebra_register *reg, + void *class_handle, RSET rset, NMEM nmem, + TERMID *terms, int numterms) { - struct rank_set_info *si = (struct rank_set_info *) xmalloc (sizeof(*si)); + struct rank_set_info *si = + (struct rank_set_info *) nmem_malloc (nmem,sizeof(*si)); int i; - logf (LOG_DEBUG, "rank-1 begin"); - si->no_entries = rset->no_rset_terms; +#if DEBUG_RANK + yaz_log (LOG_LOG, "rank-1 begin"); +#endif + si->no_entries = numterms; si->no_rank_entries = 0; + si->nmem=nmem; si->entries = (struct rank_term_info *) - xmalloc (sizeof(*si->entries)*si->no_entries); - for (i = 0; i < si->no_entries; i++) + nmem_malloc (si->nmem, sizeof(*si->entries)*numterms); + for (i = 0; i < numterms; i++) { - int g = rset->rset_terms[i]->nn; - if (!strcmp (rset->rset_terms[i]->flags, "rank")) + zint g = rset_count(terms[i]->rset); +#if DEBUG_RANK + yaz_log(LOG_LOG, "i=%d flags=%s '%s'", i, + terms[i]->flags, terms[i]->name ); +#endif + if (!strncmp (terms[i]->flags, "rank,", 5)) { + const char *cp = strstr(terms[i]->flags+4, ",w="); si->entries[i].rank_flag = 1; + if (cp) + si->entries[i].rank_weight = atoi (cp+3); + else + si->entries[i].rank_weight = 34; +#if DEBUG_RANK + yaz_log (LOG_LOG, " i=%d weight=%d g="ZINT_FORMAT, i, + si->entries[i].rank_weight, g); +#endif (si->no_rank_entries)++; } else si->entries[i].rank_flag = 0; - si->entries[i].local_occur = 0; + si->entries[i].local_occur = 0; /* FIXME */ si->entries[i].global_occur = g; si->entries[i].global_inv = 32 - log2_int (g); - logf (LOG_DEBUG, "-------- %d ------", 32 - log2_int (g)); + yaz_log (LOG_DEBUG, " global_inv = %d g = " ZINT_FORMAT, + (int) (32-log2_int (g)), g); + si->entries[i].term=terms[i]; + si->entries[i].term_index=i; + terms[i]->rankpriv=&(si->entries[i]); } return si; } @@ -122,25 +151,30 @@ static void *begin (ZebraHandle zh, void *class_handle, RSET rset) * end: Terminates ranking process. Called after a result set * has been ranked. */ -static void end (ZebraHandle zh, void *set_handle) +static void end (struct zebra_register *reg, void *set_handle) { - struct rank_set_info *si = (struct rank_set_info *) set_handle; - logf (LOG_DEBUG, "rank-1 end"); - xfree (si->entries); - xfree (si); + yaz_log (LOG_DEBUG, "rank-1 end"); + /* no need to free anything, they are in nmems */ } -/* + +/** * add: Called for each word occurence in a result set. This routine * should be as fast as possible. This routine should "incrementally" * update the score. */ -static void add (void *set_handle, int seqno, int term_index) +static void add (void *set_handle, int seqno, TERMID term) { struct rank_set_info *si = (struct rank_set_info *) set_handle; - logf (LOG_DEBUG, "rank-1 add seqno=%d term_index=%d", seqno, term_index); + struct rank_term_info *ti= (struct rank_term_info *) term->rankpriv; + assert(si); + assert(term); + assert(ti); +#if DEBUG_RANK + yaz_log (LOG_LOG, "rank-1 add seqno=%d term=%s", seqno, term->name); +#endif si->last_pos = seqno; - si->entries[term_index].local_occur++; + ti->local_occur++; } /* @@ -149,21 +183,32 @@ static void add (void *set_handle, int seqno, int term_index) * score should be between 0 and 1000. If score cannot be obtained * -1 should be returned. */ -static int calc (void *set_handle, int sysno) +static int calc (void *set_handle, zint sysno) { int i, lo, divisor, score = 0; struct rank_set_info *si = (struct rank_set_info *) set_handle; - logf (LOG_DEBUG, "rank-1 calc sysno=%d", sysno); - if (!si->no_rank_entries) return -1; + +#if DEBUG_RANK + yaz_log(LOG_LOG, "calc"); +#endif for (i = 0; i < si->no_entries; i++) + { +#if DEBUG_RANK + yaz_log(LOG_LOG, "calc: i=%d rank_flag=%d lo=%d", + i, si->entries[i].rank_flag, si->entries[i].local_occur); +#endif if (si->entries[i].rank_flag && (lo = si->entries[i].local_occur)) - score += (8+log2_int (lo)) * si->entries[i].global_inv; - score *= 34; + score += (8+log2_int (lo)) * si->entries[i].global_inv * + si->entries[i].rank_weight; + } divisor = si->no_rank_entries * (8+log2_int (si->last_pos/si->no_entries)); score = score / divisor; +#if DEBUG_RANK + yaz_log (LOG_LOG, "calc sysno=" ZINT_FORMAT " score=%d", sysno, score); +#endif if (score > 1000) score = 1000; for (i = 0; i < si->no_entries; i++)