-/*
- * Copyright (C) 1994-1999, Index Data
- * All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Log: zsets.c,v $
- * Revision 1.25 2000-03-15 15:00:31 adam
- * First work on threaded version.
- *
- * Revision 1.24 1999/11/04 15:00:45 adam
- * Implemented delete result set(s).
- *
- * Revision 1.23 1999/05/26 07:49:13 adam
- * C++ compilation.
- *
- * Revision 1.22 1999/02/02 14:51:15 adam
- * Updated WIN32 code specific sections. Changed header.
- *
- * Revision 1.21 1998/11/16 16:03:46 adam
- * Moved loggin utilities to Yaz. Was implemented in file zlogs.c.
- *
- * Revision 1.20 1998/11/16 10:10:53 adam
- * Fixed problem with zebraPosSetCreate that occurred when positions were
- * less than 1.
- *
- * Revision 1.19 1998/09/22 10:48:22 adam
- * Minor changes in search API.
- *
- * Revision 1.18 1998/09/22 10:03:45 adam
- * Changed result sets to be persistent in the sense that they can
- * be re-searched if needed.
- * Fixed memory leak in rsm_or.
- *
- * Revision 1.17 1998/06/23 15:33:36 adam
- * Added feature to specify sort criteria in query (type 7 specifies
- * sort flags).
- *
- * Revision 1.16 1998/05/20 10:12:24 adam
- * Implemented automatic EXPLAIN database maintenance.
- * Modified Zebra to work with ASN.1 compiled version of YAZ.
- *
- * Revision 1.15 1998/03/05 08:45:14 adam
- * New result set model and modular ranking system. Moved towards
- * descent server API. System information stored as "SGML" records.
- *
- * Revision 1.14 1998/02/10 16:39:15 adam
- * Minor change.
- *
- * Revision 1.13 1998/02/10 12:03:06 adam
- * Implemented Sort.
- *
- * Revision 1.12 1997/09/25 14:57:36 adam
- * Windows NT port.
- *
- * Revision 1.11 1996/12/23 15:30:46 adam
- * Work on truncation.
- * Bug fix: result sets weren't deleted after server shut down.
- *
- * Revision 1.10 1995/10/30 15:08:08 adam
- * Bug fixes.
- *
- * Revision 1.9 1995/10/17 18:02:14 adam
- * New feature: databases. Implemented as prefix to words in dictionary.
- *
- * Revision 1.8 1995/10/10 13:59:25 adam
- * Function rset_open changed its wflag parameter to general flags.
- *
- * Revision 1.7 1995/10/06 14:38:01 adam
- * New result set method: r_score.
- * Local no (sysno) and score is transferred to retrieveCtrl.
- *
- * Revision 1.6 1995/09/28 09:19:49 adam
- * xfree/xmalloc used everywhere.
- * Extract/retrieve method seems to work for text records.
- *
- * Revision 1.5 1995/09/27 16:17:32 adam
- * More work on retrieve.
- *
- * Revision 1.4 1995/09/07 13:58:36 adam
- * New parameter: result-set file descriptor (RSFD) to support multiple
- * positions within the same result-set.
- * Boolean operators: and, or, not implemented.
- * Result-set references.
- *
- * Revision 1.3 1995/09/06 16:11:19 adam
- * Option: only one word key per file.
- *
- * Revision 1.2 1995/09/06 10:33:04 adam
- * More work on present. Some log messages removed.
- *
- * Revision 1.1 1995/09/05 15:28:40 adam
- * More work on search engine.
- *
- */
+/* $Id: zsets.c,v 1.39 2002-08-28 19:52:29 adam Exp $
+ Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+ 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 <stdio.h>
#include <assert.h>
#ifdef WIN32
#include <unistd.h>
#endif
-#include "zserver.h"
+#include "index.h"
#include <rstemp.h>
#define SORT_IDX_ENTRYSIZE 64
#define ZSET_SORT_MAX_LEVEL 3
+struct zebra_set_term_entry {
+ int reg_type;
+ char *db;
+ int set;
+ int use;
+ char *term;
+};
struct zebra_set {
char *name;
RSET rset;
char **basenames;
Z_RPNQuery *rpn;
struct zset_sort_info *sort_info;
+ struct zebra_set_term_entry *term_entries;
+ int term_entries_max;
struct zebra_set *next;
+ int locked;
};
struct zset_sort_entry {
};
ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
- Z_RPNQuery *rpn, int num_bases, char **basenames,
+ Z_RPNQuery *rpn, int num_bases,
+ char **basenames,
const char *setname)
{
ZebraSet zebraSet;
+ int i;
zh->errCode = 0;
zh->errString = NULL;
zebraSet = resultSetAdd (zh, setname, 1);
if (!zebraSet)
return 0;
+ zebraSet->locked = 1;
zebraSet->rpn = 0;
+ zebraSet->nmem = nmem_create ();
+
zebraSet->num_bases = num_bases;
- zebraSet->basenames = basenames;
- zebraSet->nmem = odr_extract_mem (input);
+ zebraSet->basenames =
+ nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames));
+ for (i = 0; i<num_bases; i++)
+ zebraSet->basenames[i] = nmem_strdup (zebraSet->nmem, basenames[i]);
+
zebraSet->rset = rpn_search (zh, output->mem, rpn,
zebraSet->num_bases,
zh->hits = zebraSet->hits;
if (zebraSet->rset)
zebraSet->rpn = rpn;
+ zebraSet->locked = 0;
return zebraSet;
}
+void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type,
+ const char *db, int set,
+ int use, const char *term)
+{
+ if (!s->nmem)
+ s->nmem = nmem_create ();
+ if (!s->term_entries)
+ {
+ int i;
+ s->term_entries_max = 1000;
+ s->term_entries =
+ nmem_malloc (s->nmem, s->term_entries_max *
+ sizeof(*s->term_entries));
+ for (i = 0; i < s->term_entries_max; i++)
+ s->term_entries[i].term = 0;
+ }
+ if (s->hits < s->term_entries_max)
+ {
+ s->term_entries[s->hits].reg_type = reg_type;
+ s->term_entries[s->hits].db = nmem_strdup (s->nmem, db);
+ s->term_entries[s->hits].set = set;
+ s->term_entries[s->hits].use = use;
+ s->term_entries[s->hits].term = nmem_strdup (s->nmem, term);
+ }
+ (s->hits)++;
+}
+
+
+int zebra_resultSetTerms (ZebraHandle zh, const char *setname,
+ int no, int *count,
+ int *type, char *out, size_t *len)
+{
+ ZebraSet s = resultSetGet (zh, setname);
+ int no_max = 0;
+
+ if (count)
+ *count = 0;
+ if (!s || !s->rset)
+ return 0;
+ no_max = s->rset->no_rset_terms;
+ if (no < 0 || no >= no_max)
+ return 0;
+ if (count)
+ *count = s->rset->rset_terms[no]->count;
+ if (type)
+ *type = s->rset->rset_terms[no]->type;
+
+ if (out)
+ {
+ char *inbuf = s->rset->rset_terms[no]->name;
+ size_t inleft = strlen(inbuf);
+ size_t outleft = *len - 1;
+ int converted = 0;
+
+ if (zh->iconv_from_utf8 != 0)
+ {
+ char *outbuf = out;
+ size_t ret;
+
+ ret = yaz_iconv(zh->iconv_from_utf8, &inbuf, &inleft,
+ &outbuf, &outleft);
+ if (ret == (size_t)(-1))
+ *len = 0;
+ else
+ *len = outbuf - out;
+ converted = 1;
+ }
+ if (!converted)
+ {
+ if (inleft > outleft)
+ inleft = outleft;
+ *len = inleft;
+ memcpy (out, inbuf, *len);
+ }
+ out[*len] = 0;
+ }
+ return no_max;
+}
+
+
ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
{
ZebraSet s;
if (s)
{
logf (LOG_DEBUG, "updating result set %s", name);
- if (!ov)
+ if (!ov || s->locked)
return NULL;
if (s->rset)
rset_delete (s->rset);
for (i = 0; i < s->sort_info->max_entries; i++)
s->sort_info->entries[i] = s->sort_info->all_entries + i;
}
+ s->locked = 0;
+ s->term_entries = 0;
+ s->hits = 0;
s->rset = 0;
- s->nmem = 0;
+ s->nmem = 0;
+ s->rpn = 0;
return s;
}
for (s = zh->sets; s; s = s->next)
if (!strcmp (s->name, name))
{
- if (!s->rset && s->rpn)
+ if (!s->term_entries && !s->rset && s->rpn)
{
NMEM nmem = nmem_create ();
+ yaz_log (LOG_LOG, "research %s", name);
s->rset =
rpn_search (zh, nmem, s->rpn, s->num_bases,
s->basenames, s->name, s);
nmem_destroy (nmem);
-
}
return s;
}
return NULL;
}
+void resultSetInvalidate (ZebraHandle zh)
+{
+ ZebraSet s = zh->sets;
+
+ for (; s; s = s->next)
+ {
+ if (s->rset)
+ rset_delete (s->rset);
+ s->rset = 0;
+ }
+}
+
void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
{
- ZebraSet *ss = &zh->sets;
+ ZebraSet * ss = &zh->sets;
int i;
if (statuses)
if (s->nmem)
nmem_destroy (s->nmem);
- rset_delete (s->rset);
+ if (s->rset)
+ rset_delete (s->rset);
xfree (s->name);
xfree (s);
}
int num, int *positions)
{
ZebraSet sset;
- ZebraPosSet sr;
+ ZebraPosSet sr = 0;
RSET rset;
int i;
struct zset_sort_info *sort_info;
if (!(sset = resultSetGet (zh, name)))
return NULL;
if (!(rset = sset->rset))
- return NULL;
- sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
- for (i = 0; i<num; i++)
- {
- sr[i].sysno = 0;
- sr[i].score = -1;
- }
- sort_info = sset->sort_info;
- if (sort_info)
{
- int position;
-
+ if (!sset->term_entries)
+ return 0;
+ sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
for (i = 0; i<num; i++)
{
- position = positions[i];
- if (position > 0 && position <= sort_info->num_entries)
+ sr[i].sysno = 0;
+ sr[i].score = -1;
+ sr[i].term = 0;
+ sr[i].db = 0;
+
+ if (positions[i] <= sset->term_entries_max)
{
- logf (LOG_DEBUG, "got pos=%d (sorted)", position);
- sr[i].sysno = sort_info->entries[position-1]->sysno;
- sr[i].score = sort_info->entries[position-1]->score;
+ sr[i].term = sset->term_entries[positions[i]-1].term;
+ sr[i].db = sset->term_entries[positions[i]-1].db;
}
}
}
- /* did we really get all entries using sort ? */
- for (i = 0; i<num; i++)
- {
- if (!sr[i].sysno)
- break;
- }
- if (i < num) /* nope, get the rest, unsorted - sorry */
+ else
{
- int position = 0;
- int num_i = 0;
- int psysno = 0;
- int term_index;
- RSFD rfd;
- struct it_key key;
-
+ sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
+ for (i = 0; i<num; i++)
+ {
+ sr[i].sysno = 0;
+ sr[i].score = -1;
+ sr[i].term = 0;
+ sr[i].db = 0;
+ }
+ sort_info = sset->sort_info;
if (sort_info)
- position = sort_info->num_entries;
- while (num_i < num && positions[num_i] < position)
- num_i++;
- rfd = rset_open (rset, RSETF_READ);
- while (num_i < num && rset_read (rset, rfd, &key, &term_index))
{
- if (key.sysno != psysno)
+ int position;
+
+ for (i = 0; i<num; i++)
{
- psysno = key.sysno;
- if (sort_info)
+ position = positions[i];
+ if (position > 0 && position <= sort_info->num_entries)
{
- /* determine we alreay have this in our set */
- for (i = sort_info->num_entries; --i >= 0; )
- if (psysno == sort_info->entries[i]->sysno)
- break;
- if (i >= 0)
- continue;
+ logf (LOG_DEBUG, "got pos=%d (sorted)", position);
+ sr[i].sysno = sort_info->entries[position-1]->sysno;
+ sr[i].score = sort_info->entries[position-1]->score;
}
- position++;
- assert (num_i < num);
- if (position == positions[num_i])
+ }
+ }
+ /* did we really get all entries using sort ? */
+ for (i = 0; i<num; i++)
+ {
+ if (!sr[i].sysno)
+ break;
+ }
+ if (i < num) /* nope, get the rest, unsorted - sorry */
+ {
+ int position = 0;
+ int num_i = 0;
+ int psysno = 0;
+ int term_index;
+ RSFD rfd;
+ struct it_key key;
+
+ if (sort_info)
+ position = sort_info->num_entries;
+ while (num_i < num && positions[num_i] < position)
+ num_i++;
+ rfd = rset_open (rset, RSETF_READ);
+ while (num_i < num && rset_read (rset, rfd, &key, &term_index))
+ {
+ if (key.sysno != psysno)
{
- sr[num_i].sysno = psysno;
- logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
- sr[num_i].score = -1;
- num_i++;
+ psysno = key.sysno;
+ if (sort_info)
+ {
+ /* determine we alreay have this in our set */
+ for (i = sort_info->num_entries; --i >= 0; )
+ if (psysno == sort_info->entries[i]->sysno)
+ break;
+ if (i >= 0)
+ continue;
+ }
+ position++;
+ assert (num_i < num);
+ if (position == positions[num_i])
+ {
+ sr[num_i].sysno = psysno;
+ logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
+ sr[num_i].score = -1;
+ num_i++;
+ }
}
}
+ rset_close (rset, rfd);
}
- rset_close (rset, rfd);
}
return sr;
}
xfree (records);
}
-struct sortKey {
+struct sortKeyInfo {
int relation;
int attrUse;
+ int numerical;
};
void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
- struct sortKey *criteria, int num_criteria,
+ struct sortKeyInfo *criteria, int num_criteria,
int sysno)
{
struct zset_sort_entry this_entry;
struct zset_sort_info *sort_info = sset->sort_info;
int i, j;
- sortIdx_sysno (zh->service->sortIdx, sysno);
+ sortIdx_sysno (zh->reg->sortIdx, sysno);
for (i = 0; i<num_criteria; i++)
{
- sortIdx_type (zh->service->sortIdx, criteria[i].attrUse);
- sortIdx_read (zh->service->sortIdx, this_entry.buf[i]);
+ sortIdx_type (zh->reg->sortIdx, criteria[i].attrUse);
+ sortIdx_read (zh->reg->sortIdx, this_entry.buf[i]);
}
i = sort_info->num_entries;
while (--i >= 0)
int rel = 0;
for (j = 0; j<num_criteria; j++)
{
- rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
+ if (criteria[j].numerical)
+ {
+ double diff = atof(this_entry.buf[j]) -
+ atof(sort_info->entries[i]->buf[j]);
+ rel = 0;
+ if (diff > 0.0)
+ rel = 1;
+ else if (diff < 0.0)
+ rel = -1;
+ }
+ else
+ {
+ rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
SORT_IDX_ENTRYSIZE);
+ }
if (rel)
break;
}
break;
}
}
- j = sort_info->max_entries-1;
+ ++i;
+ j = sort_info->max_entries;
if (i == j)
return;
- ++i;
+
+ if (sort_info->num_entries == j)
+ --j;
+ else
+ j = (sort_info->num_entries)++;
new_entry = sort_info->entries[j];
while (j != i)
{
sort_info->entries[j] = sort_info->entries[j-1];
--j;
}
- sort_info->entries[j] = new_entry;
+ sort_info->entries[i] = new_entry;
assert (new_entry);
- if (sort_info->num_entries != sort_info->max_entries)
- (sort_info->num_entries)++;
for (i = 0; i<num_criteria; i++)
memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
new_entry->sysno = sysno;
break;
}
}
- j = sort_info->max_entries-1;
+ ++i;
+ j = sort_info->max_entries;
if (i == j)
return;
- ++i;
+
+ if (sort_info->num_entries == j)
+ --j;
+ else
+ j = (sort_info->num_entries)++;
+
new_entry = sort_info->entries[j];
while (j != i)
{
sort_info->entries[j] = sort_info->entries[j-1];
--j;
}
- sort_info->entries[j] = new_entry;
+ sort_info->entries[i] = new_entry;
assert (new_entry);
- if (sort_info->num_entries != sort_info->max_entries)
- (sort_info->num_entries)++;
new_entry->sysno = sysno;
new_entry->score = score;
}
{
int i, psysno = 0;
struct it_key key;
- struct sortKey sort_criteria[3];
+ struct sortKeyInfo sort_criteria[3];
int num_criteria;
int term_index;
RSFD rfd;
- logf (LOG_DEBUG, "resultSetSortSingle start");
+ logf (LOG_LOG, "resultSetSortSingle start");
sset->sort_info->num_entries = 0;
sset->hits = 0;
case Z_SortKey_sortAttributes:
logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
sort_criteria[i].attrUse =
- zebra_maps_sort (zh->service->zebra_maps,
- sk->u.sortAttributes);
+ zebra_maps_sort (zh->reg->zebra_maps,
+ sk->u.sortAttributes,
+ &sort_criteria[i].numerical);
logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
if (sort_criteria[i].attrUse == -1)
{
zh->errCode = 116;
return;
}
- if (sortIdx_type (zh->service->sortIdx, sort_criteria[i].attrUse))
+ if (sortIdx_type (zh->reg->sortIdx, sort_criteria[i].attrUse))
{
zh->errCode = 207;
return;
}
rset_close (rset, rfd);
- zh->errCode = 0;
+ for (i = 0; i < rset->no_rset_terms; i++)
+ yaz_log (LOG_LOG, "term=\"%s\" nn=%d type=%s count=%d",
+ rset->rset_terms[i]->name,
+ rset->rset_terms[i]->nn,
+ rset->rset_terms[i]->flags,
+ rset->rset_terms[i]->count);
+
*sort_status = Z_SortStatus_success;
- logf (LOG_DEBUG, "resultSetSortSingle end");
+ logf (LOG_LOG, "resultSetSortSingle end");
}
RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
zebraSet->hits = 0;
rfd = rset_open (rset, RSETF_READ);
- logf (LOG_DEBUG, "resultSetRank");
- for (i = 0; i < rset->no_rset_terms; i++)
- logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
- rset->rset_terms[i]->name,
- rset->rset_terms[i]->nn,
- rset->rset_terms[i]->flags);
+ yaz_log (LOG_LOG, "resultSetRank");
rank_class = zebraRankLookup (zh, "rank-1");
rc = rank_class->control;
int psysno = key.sysno;
int score;
void *handle =
- (*rc->begin) (zh, rank_class->class_handle, rset);
+ (*rc->begin) (zh->reg, rank_class->class_handle, rset);
(zebraSet->hits)++;
do
{
while (rset_read (rset, rfd, &key, &term_index));
score = (*rc->calc) (handle, psysno);
resultSetInsertRank (zh, sort_info, psysno, score, 'A');
- (*rc->end) (zh, handle);
+ (*rc->end) (zh->reg, handle);
}
rset_close (rset, rfd);
- logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
+
+ for (i = 0; i < rset->no_rset_terms; i++)
+ yaz_log (LOG_LOG, "term=\"%s\" nn=%d type=%s count=%d",
+ rset->rset_terms[i]->name,
+ rset->rset_terms[i]->nn,
+ rset->rset_terms[i]->flags,
+ rset->rset_terms[i]->count);
+
+ yaz_log (LOG_LOG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
}
ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
{
- ZebraRankClass p = zh->service->rank_classes;
+ ZebraRankClass p = zh->reg->rank_classes;
while (p && strcmp (p->control->name, name))
p = p->next;
if (p && !p->init_flag)
{
if (p->control->create)
- p->class_handle = (*p->control->create)(zh->service);
+ p->class_handle = (*p->control->create)(zh->reg);
p->init_flag = 1;
}
return p;
}
-void zebraRankInstall (ZebraService zh, struct rank_control *ctrl)
+void zebraRankInstall (struct zebra_register *reg, struct rank_control *ctrl)
{
ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
memcpy (p->control, ctrl, sizeof(*p->control));
p->control->name = xstrdup (ctrl->name);
p->init_flag = 0;
- p->next = zh->rank_classes;
- zh->rank_classes = p;
+ p->next = reg->rank_classes;
+ reg->rank_classes = p;
}
-void zebraRankDestroy (ZebraService zh)
+void zebraRankDestroy (struct zebra_register *reg)
{
- ZebraRankClass p = zh->rank_classes;
+ ZebraRankClass p = reg->rank_classes;
while (p)
{
ZebraRankClass p_next = p->next;
if (p->init_flag && p->control->destroy)
- (*p->control->destroy)(zh, p->class_handle);
+ (*p->control->destroy)(reg, p->class_handle);
xfree (p->control->name);
xfree (p->control);
xfree (p);
p = p_next;
}
- zh->rank_classes = NULL;
+ reg->rank_classes = NULL;
}