-/*
- * Copyright (C) 1994-2000, Index Data
- * All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Log: zsets.c,v $
- * Revision 1.31 2001-11-19 23:05:22 adam
- * Added a few prototypes.
- *
- * Revision 1.30 2001/10/15 19:53:43 adam
- * POSIX thread updates. First work on term sets.
- *
- * Revision 1.29 2001/01/22 10:42:56 adam
- * Added numerical sort.
- *
- * Revision 1.28 2000/07/07 12:49:20 adam
- * Optimized resultSetInsert{Rank,Sort}.
- *
- * Revision 1.27 2000/04/05 09:49:36 adam
- * On Unix, zebra/z'mbol uses automake.
- *
- * Revision 1.26 2000/03/20 19:08:36 adam
- * Added remote record import using Z39.50 extended services and Segment
- * Requests.
- *
- * 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 1.108 2006-06-22 15:07:20 adam Exp $
+ Copyright (C) 1995-2006
+ 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 <rstemp.h>
+#include "index.h"
+#include "rank.h"
+#include <yaz/diagbib1.h>
+#include <rset.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 *index_name;
char *term;
};
+
struct zebra_set {
char *name;
RSET rset;
NMEM nmem;
- int hits;
+ NMEM rset_nmem; /* for creating the rsets in */
+ zint hits;
int num_bases;
char **basenames;
Z_RPNQuery *rpn;
+ Z_SortKeySpecList *sortSpec;
struct zset_sort_info *sort_info;
struct zebra_set_term_entry *term_entries;
int term_entries_max;
struct zebra_set *next;
int locked;
+
+ zint cache_position; /* last position */
+ RSFD cache_rfd; /* rfd (NULL if not existing) */
+ zint cache_psysno; /* sysno for last position */
+ zint approx_limit; /* limit before we do approx */
};
struct zset_sort_entry {
- int sysno;
+ zint sysno;
int score;
char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
};
struct zset_sort_entry **entries;
};
-ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
- Z_RPNQuery *rpn, int num_bases, char **basenames,
- const char *setname)
+static int log_level_set=0;
+static int log_level_sort=0;
+static int log_level_searchhits=0;
+static int log_level_searchterms=0;
+static int log_level_resultsets=0;
+
+static void loglevels()
+{
+ if (log_level_set)
+ return;
+ log_level_sort = yaz_log_module_level("sorting");
+ log_level_searchhits = yaz_log_module_level("searchhits");
+ log_level_searchterms = yaz_log_module_level("searchterms");
+ log_level_resultsets = yaz_log_module_level("resultsets");
+ log_level_set = 1;
+}
+
+ZEBRA_RES resultSetSearch(ZebraHandle zh, NMEM nmem, NMEM rset_nmem,
+ Z_RPNQuery *rpn, ZebraSet sset)
+{
+ RSET rset = 0;
+ oident *attrset;
+ Z_SortKeySpecList *sort_sequence;
+ int sort_status, i;
+ ZEBRA_RES res = ZEBRA_OK;
+
+ zh->hits = 0;
+
+ sort_sequence = (Z_SortKeySpecList *)
+ nmem_malloc(nmem, sizeof(*sort_sequence));
+ sort_sequence->num_specs = 10; /* FIXME - Hard-coded number */
+ sort_sequence->specs = (Z_SortKeySpec **)
+ nmem_malloc(nmem, sort_sequence->num_specs *
+ sizeof(*sort_sequence->specs));
+ for (i = 0; i<sort_sequence->num_specs; i++)
+ sort_sequence->specs[i] = 0;
+
+ attrset = oid_getentbyoid (rpn->attributeSetId);
+ res = rpn_search_top(zh, rpn->RPNStructure, attrset->value,
+ nmem, rset_nmem,
+ sort_sequence,
+ sset->num_bases, sset->basenames,
+ &rset);
+ if (res != ZEBRA_OK)
+ {
+ sset->rset = 0;
+ return res;
+ }
+ for (i = 0; sort_sequence->specs[i]; i++)
+ ;
+ sort_sequence->num_specs = i;
+ rset->hits_limit = sset->approx_limit;
+ if (!i)
+ {
+ res = resultSetRank (zh, sset, rset, rset_nmem);
+ }
+ else
+ {
+ res = resultSetSortSingle (zh, nmem, sset, rset,
+ sort_sequence, &sort_status);
+ }
+ sset->rset = rset;
+ return res;
+}
+
+
+ZEBRA_RES resultSetAddRPN (ZebraHandle zh, NMEM m, Z_RPNQuery *rpn,
+ int num_bases, char **basenames,
+ const char *setname)
{
ZebraSet zebraSet;
+ int i;
+ ZEBRA_RES res;
- zh->errCode = 0;
- zh->errString = NULL;
zh->hits = 0;
- zebraSet = resultSetAdd (zh, setname, 1);
+ zebraSet = resultSetAdd(zh, setname, 1);
if (!zebraSet)
- return 0;
+ return ZEBRA_FAIL;
zebraSet->locked = 1;
zebraSet->rpn = 0;
+ zebraSet->nmem = m;
+ zebraSet->rset_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,
- zebraSet->basenames, zebraSet->name,
- zebraSet);
+ res = resultSetSearch(zh, zebraSet->nmem, zebraSet->rset_nmem,
+ rpn, zebraSet);
zh->hits = zebraSet->hits;
if (zebraSet->rset)
zebraSet->rpn = rpn;
zebraSet->locked = 0;
- return zebraSet;
+ if (!zebraSet->rset)
+ return ZEBRA_FAIL;
+ return res;
}
-void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type,
- const char *db, int set,
- int use, const char *term)
+void resultSetAddTerm(ZebraHandle zh, ZebraSet s, int reg_type,
+ const char *db, const char *index_name,
+ const char *term)
{
+ assert(zh); /* compiler shut up */
if (!s->nmem)
- s->nmem = nmem_create ();
+ 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;
+ 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->term_entries[s->hits].reg_type = reg_type;
+ s->term_entries[s->hits].db = nmem_strdup (s->nmem, db);
+ s->term_entries[s->hits].index_name = nmem_strdup(s->nmem, index_name);
+ s->term_entries[s->hits].term = nmem_strdup(s->nmem, term);
}
(s->hits)++;
}
-ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
+ZebraSet resultSetAdd(ZebraHandle zh, const char *name, int ov)
{
ZebraSet s;
int i;
for (s = zh->sets; s; s = s->next)
if (!strcmp (s->name, name))
- break;
+ break;
+
+ if (!log_level_set)
+ loglevels();
if (s)
{
- logf (LOG_DEBUG, "updating result set %s", name);
- if (!ov || s->locked)
- return NULL;
- if (s->rset)
- rset_delete (s->rset);
- if (s->nmem)
- nmem_destroy (s->nmem);
+ yaz_log(log_level_resultsets, "updating result set %s", name);
+ if (!ov || s->locked)
+ return NULL;
+ if (s->rset)
+ {
+ if (s->cache_rfd)
+ rset_close(s->cache_rfd);
+ rset_delete (s->rset);
+ }
+ if (s->rset_nmem)
+ nmem_destroy (s->rset_nmem);
+ if (s->nmem)
+ nmem_destroy (s->nmem);
}
else
{
- logf (LOG_DEBUG, "adding result set %s", name);
- s = (ZebraSet) xmalloc (sizeof(*s));
- s->next = zh->sets;
- zh->sets = s;
- s->name = (char *) xmalloc (strlen(name)+1);
- strcpy (s->name, name);
-
- s->sort_info = (struct zset_sort_info *)
- xmalloc (sizeof(*s->sort_info));
- s->sort_info->max_entries = 1000;
- s->sort_info->entries = (struct zset_sort_entry **)
- xmalloc (sizeof(*s->sort_info->entries) *
- s->sort_info->max_entries);
- s->sort_info->all_entries = (struct zset_sort_entry *)
- xmalloc (sizeof(*s->sort_info->all_entries) *
- s->sort_info->max_entries);
- for (i = 0; i < s->sort_info->max_entries; i++)
- s->sort_info->entries[i] = s->sort_info->all_entries + i;
+ const char *sort_max_str = zebra_get_resource(zh, "sortmax", "1000");
+
+ yaz_log(log_level_resultsets, "adding result set %s", name);
+ s = (ZebraSet) xmalloc (sizeof(*s));
+ s->next = zh->sets;
+ zh->sets = s;
+ s->name = (char *) xmalloc (strlen(name)+1);
+ strcpy (s->name, name);
+
+ s->sort_info = (struct zset_sort_info *)
+ xmalloc (sizeof(*s->sort_info));
+ s->sort_info->max_entries = atoi(sort_max_str);
+ if (s->sort_info->max_entries < 2)
+ s->sort_info->max_entries = 2;
+
+ s->sort_info->entries = (struct zset_sort_entry **)
+ xmalloc (sizeof(*s->sort_info->entries) *
+ s->sort_info->max_entries);
+ s->sort_info->all_entries = (struct zset_sort_entry *)
+ xmalloc (sizeof(*s->sort_info->all_entries) *
+ s->sort_info->max_entries);
+ 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->rset_nmem = 0;
s->nmem = 0;
s->rpn = 0;
+ s->sortSpec = 0;
+ s->cache_position = 0;
+ s->cache_rfd = 0;
+ s->approx_limit = zh->approx_limit;
return s;
}
-ZebraSet resultSetGet (ZebraHandle zh, const char *name)
+ZebraSet resultSetGet(ZebraHandle zh, const char *name)
{
ZebraSet s;
if (!s->term_entries && !s->rset && s->rpn)
{
NMEM nmem = nmem_create ();
- s->rset =
- rpn_search (zh, nmem, s->rpn, s->num_bases,
- s->basenames, s->name, s);
+ yaz_log(log_level_resultsets, "research %s", name);
+ if (!s->rset_nmem)
+ s->rset_nmem=nmem_create();
+ resultSetSearch(zh, nmem, s->rset_nmem, s->rpn, s);
+ if (s->rset && s->sortSpec)
+ {
+ int sort_status;
+ yaz_log(log_level_resultsets, "resort %s", name);
+ resultSetSortSingle (zh, nmem, s, s->rset, s->sortSpec,
+ &sort_status);
+ }
nmem_destroy (nmem);
}
return s;
return NULL;
}
-void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
+void resultSetInvalidate (ZebraHandle zh)
+{
+ ZebraSet s = zh->sets;
+
+ yaz_log(log_level_resultsets, "invalidating result sets");
+ for (; s; s = s->next)
+ {
+ if (s->rset)
+ {
+ if (s->cache_rfd)
+ rset_close(s->cache_rfd);
+ rset_delete (s->rset);
+ }
+ s->rset = 0;
+ s->cache_rfd = 0;
+ s->cache_position = 0;
+ if (s->rset_nmem)
+ nmem_destroy(s->rset_nmem);
+ s->rset_nmem=0;
+ }
+}
+
+void resultSetDestroy(ZebraHandle zh, int num, char **names,int *statuses)
{
- ZebraSet *ss = &zh->sets;
+ ZebraSet * ss = &zh->sets;
int i;
if (statuses)
- for (i = 0; i<num; i++)
- statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
- zh->errCode = 0;
- zh->errString = NULL;
+ for (i = 0; i<num; i++)
+ statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
while (*ss)
{
- int i = -1;
- ZebraSet s = *ss;
- if (num >= 0)
- {
- for (i = 0; i<num; i++)
- if (!strcmp (s->name, names[i]))
- {
- if (statuses)
- statuses[i] = Z_DeleteStatus_success;
- i = -1;
- break;
- }
- }
- if (i < 0)
- {
- *ss = s->next;
-
- xfree (s->sort_info->all_entries);
- xfree (s->sort_info->entries);
- xfree (s->sort_info);
-
- if (s->nmem)
- nmem_destroy (s->nmem);
- if (s->rset)
- rset_delete (s->rset);
- xfree (s->name);
- xfree (s);
- }
- else
- ss = &s->next;
+ int i = -1;
+ ZebraSet s = *ss;
+ if (num >= 0)
+ {
+ for (i = 0; i<num; i++)
+ if (!strcmp (s->name, names[i]))
+ {
+ if (statuses)
+ statuses[i] = Z_DeleteStatus_success;
+ i = -1;
+ break;
+ }
+ }
+ if (i < 0)
+ {
+ *ss = s->next;
+
+ xfree (s->sort_info->all_entries);
+ xfree (s->sort_info->entries);
+ xfree (s->sort_info);
+
+ if (s->nmem)
+ nmem_destroy (s->nmem);
+ if (s->rset)
+ {
+ if (s->cache_rfd)
+ rset_close(s->cache_rfd);
+ rset_delete (s->rset);
+ }
+ if (s->rset_nmem)
+ nmem_destroy(s->rset_nmem);
+ xfree (s->name);
+ xfree (s);
+ }
+ else
+ ss = &s->next;
}
}
-ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
- int num, int *positions)
+ZebraMetaRecord *zebra_meta_records_create_range(ZebraHandle zh,
+ const char *name,
+ zint start, int num)
+{
+ zint pos_small[10];
+ zint *pos = pos_small;
+ ZebraMetaRecord *mr;
+ int i;
+
+ if (num > 10000 || num <= 0)
+ return 0;
+
+ if (num > 10)
+ pos = xmalloc(sizeof(*pos) * num);
+
+ for (i = 0; i<num; i++)
+ pos[i] = start+i;
+
+ mr = zebra_meta_records_create(zh, name, num, pos);
+
+ if (num > 10)
+ xfree(pos);
+ return mr;
+}
+
+ZebraMetaRecord *zebra_meta_records_create(ZebraHandle zh, const char *name,
+ int num, zint *positions)
{
ZebraSet sset;
- ZebraPosSet sr = 0;
+ ZebraMetaRecord *sr = 0;
RSET rset;
int i;
struct zset_sort_info *sort_info;
+ size_t sysno_mem_index = 0;
+ if (zh->m_staticrank)
+ sysno_mem_index = 1;
+
+ if (!log_level_set)
+ loglevels();
if (!(sset = resultSetGet (zh, name)))
return NULL;
if (!(rset = sset->rset))
{
- if (!sset->term_entries)
- return 0;
- sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
- for (i = 0; i<num; i++)
- {
- struct zebra_set_term_entry *entry = sset->term_entries;
-
- sr[i].sysno = 0;
- sr[i].score = -1;
- sr[i].term = 0;
- sr[i].db = 0;
+ if (!sset->term_entries)
+ return 0;
+ sr = (ZebraMetaRecord *) 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;
- if (positions[i] <= sset->term_entries_max)
- {
- sr[i].term = sset->term_entries[positions[i]-1].term;
- sr[i].db = sset->term_entries[positions[i]-1].db;
- }
- }
+ if (positions[i] <= sset->term_entries_max)
+ {
+ sr[i].term = sset->term_entries[positions[i]-1].term;
+ sr[i].db = sset->term_entries[positions[i]-1].db;
+ }
+ }
}
else
{
- 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)
- {
- int position;
+ sr = (ZebraMetaRecord *) 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)
+ {
+ zint position;
+
+ for (i = 0; i<num; i++)
+ {
+ position = positions[i];
+ if (position > 0 && position <= sort_info->num_entries)
+ {
+ yaz_log(log_level_sort, "got pos=" ZINT_FORMAT
+ " (sorted)", position);
+ sr[i].sysno = sort_info->entries[position-1]->sysno;
+ sr[i].score = sort_info->entries[position-1]->score;
+ }
+ }
+ }
+ /* 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 */
+ {
+ zint position = 0;
+ int num_i = 0;
+ zint psysno = 0;
+ RSFD rfd;
+ struct it_key key;
+
+ if (sort_info)
+ position = sort_info->num_entries;
+ while (num_i < num && positions[num_i] <= position)
+ num_i++;
- for (i = 0; i<num; i++)
+ if (sset->cache_rfd &&
+ num_i < num && positions[num_i] > sset->cache_position)
{
- position = positions[i];
- if (position > 0 && position <= sort_info->num_entries)
- {
- 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 = sset->cache_position;
+ rfd = sset->cache_rfd;
+ psysno = sset->cache_psysno;
}
- }
- /* 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))
+ else
{
- if (key.sysno != psysno)
- {
- 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++;
- }
- }
+ if (sset->cache_rfd)
+ rset_close(sset->cache_rfd);
+ rfd = rset_open (rset, RSETF_READ);
}
- rset_close (rset, rfd);
- }
+ while (num_i < num && rset_read (rfd, &key, 0))
+ {
+ zint this_sys = key.mem[sysno_mem_index];
+ if (this_sys != psysno)
+ {
+ psysno = this_sys;
+ 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;
+ yaz_log(log_level_sort, "got pos=" ZINT_FORMAT " (unsorted)", position);
+ sr[num_i].score = -1;
+ num_i++;
+ }
+ }
+ }
+ sset->cache_position = position;
+ sset->cache_psysno = psysno;
+ sset->cache_rfd = rfd;
+ }
}
return sr;
}
-void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
+void zebra_meta_records_destroy (ZebraHandle zh, ZebraMetaRecord *records,
+ int num)
{
+ assert(zh); /* compiler shut up about unused arg */
xfree (records);
}
struct sortKeyInfo {
int relation;
+#if 0
int attrUse;
+#else
+ int ord;
+#endif
int numerical;
};
void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
- struct sortKeyInfo *criteria, int num_criteria,
- int sysno)
+ struct sortKeyInfo *criteria, int num_criteria,
+ zint sysno)
{
struct zset_sort_entry this_entry;
struct zset_sort_entry *new_entry = NULL;
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]);
+ memset(this_entry.buf[i], '\0', SORT_IDX_ENTRYSIZE);
+ if (criteria[i].ord != -1)
+ {
+ sortIdx_type (zh->reg->sortIdx, criteria[i].ord);
+ 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++)
- {
+ int rel = 0;
+ for (j = 0; j<num_criteria; j++)
+ {
if (criteria[j].numerical)
{
double diff = atof(this_entry.buf[j]) -
}
else
{
- rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
- SORT_IDX_ENTRYSIZE);
+ rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
+ SORT_IDX_ENTRYSIZE);
}
- if (rel)
- break;
- }
- if (!rel)
- break;
- if (criteria[j].relation == 'A')
+ if (rel)
+ break;
+ }
+ if (!rel)
+ break;
+ if (criteria[j].relation == 'A')
{
- if (rel > 0)
- break;
- }
+ if (rel > 0)
+ break;
+ }
else if (criteria[j].relation == 'D')
- {
- if (rel < 0)
- break;
- }
+ {
+ if (rel < 0)
+ break;
+ }
}
++i;
j = sort_info->max_entries;
if (i == j)
- return;
+ return;
if (sort_info->num_entries == j)
- --j;
+ --j;
else
- j = (sort_info->num_entries)++;
+ 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] = sort_info->entries[j-1];
+ --j;
}
sort_info->entries[i] = new_entry;
assert (new_entry);
for (i = 0; i<num_criteria; i++)
- memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
+ memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
new_entry->sysno = sysno;
new_entry->score = -1;
}
void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
- int sysno, int score, int relation)
+ zint sysno, int score, int relation)
{
struct zset_sort_entry *new_entry = NULL;
int i, j;
+ assert(zh); /* compiler shut up about unused arg */
i = sort_info->num_entries;
while (--i >= 0)
{
- int rel = 0;
+ int rel = 0;
- rel = score - sort_info->entries[i]->score;
+ rel = score - sort_info->entries[i]->score;
- if (relation == 'D')
- {
- if (rel >= 0)
- break;
- }
- else if (relation == 'A')
- {
- if (rel <= 0)
- break;
- }
+ if (relation == 'D')
+ {
+ if (rel >= 0)
+ break;
+ }
+ else if (relation == 'A')
+ {
+ if (rel <= 0)
+ break;
+ }
}
++i;
j = sort_info->max_entries;
if (i == j)
- return;
+ return;
if (sort_info->num_entries == j)
- --j;
+ --j;
else
- j = (sort_info->num_entries)++;
+ 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] = sort_info->entries[j-1];
+ --j;
}
sort_info->entries[i] = new_entry;
assert (new_entry);
new_entry->score = score;
}
-void resultSetSort (ZebraHandle zh, NMEM nmem,
- int num_input_setnames, const char **input_setnames,
- const char *output_setname,
- Z_SortKeySpecList *sort_sequence, int *sort_status)
+static Z_RPNQuery *copy_RPNQuery(Z_RPNQuery *src, NMEM nmem)
+{
+ Z_RPNQuery *dst = 0;
+ ODR encode = odr_createmem(ODR_ENCODE);
+ ODR decode = odr_createmem(ODR_DECODE);
+
+ if (z_RPNQuery(encode, &src, 0, 0))
+ {
+ int len;
+ char *buf = odr_getbuf(encode, &len, 0);
+
+ if (buf)
+ {
+ odr_setbuf(decode, buf, len, 0);
+ z_RPNQuery(decode, &dst, 0, 0);
+ }
+ }
+ nmem_transfer(nmem, decode->mem);
+ odr_destroy(encode);
+ odr_destroy(decode);
+ return dst;
+}
+
+static Z_SortKeySpecList *copy_SortKeySpecList(Z_SortKeySpecList *src, NMEM nmem)
+{
+ Z_SortKeySpecList *dst = 0;
+ ODR encode = odr_createmem(ODR_ENCODE);
+ ODR decode = odr_createmem(ODR_DECODE);
+
+ if (z_SortKeySpecList(encode, &src, 0, 0))
+ {
+ int len;
+ char *buf = odr_getbuf(encode, &len, 0);
+
+ if (buf)
+ {
+ odr_setbuf(decode, buf, len, 0);
+ z_SortKeySpecList(decode, &dst, 0, 0);
+ }
+ }
+ nmem_transfer(nmem, decode->mem);
+ odr_destroy(encode);
+ odr_destroy(decode);
+ return dst;
+}
+
+ZebraSet resultSetClone(ZebraHandle zh, const char *setname,
+ ZebraSet rset)
+{
+ ZebraSet nset;
+ int i;
+
+ nset = resultSetAdd(zh, setname, 1);
+ if (!nset)
+ return 0;
+
+ nset->nmem = nmem_create();
+
+ nset->num_bases = rset->num_bases;
+ nset->basenames =
+ nmem_malloc (nset->nmem, nset->num_bases * sizeof(*rset->basenames));
+ for (i = 0; i<rset->num_bases; i++)
+ nset->basenames[i] = nmem_strdup(nset->nmem, rset->basenames[i]);
+
+ if (rset->rset)
+ nset->rset = rset_dup(rset->rset);
+ if (rset->rpn)
+ nset->rpn = copy_RPNQuery(rset->rpn, nset->nmem);
+ return nset;
+}
+
+ZEBRA_RES resultSetSort(ZebraHandle zh, NMEM nmem,
+ int num_input_setnames, const char **input_setnames,
+ const char *output_setname,
+ Z_SortKeySpecList *sort_sequence, int *sort_status)
{
ZebraSet sset;
RSET rset;
if (num_input_setnames == 0)
{
- zh->errCode = 208;
- return ;
+ zebra_setError(zh, YAZ_BIB1_NO_RESULT_SET_NAME_SUPPLIED_ON_SORT, 0);
+ return ZEBRA_FAIL;
}
if (num_input_setnames > 1)
{
- zh->errCode = 230;
- return;
+ zebra_setError(zh, YAZ_BIB1_SORT_TOO_MANY_INPUT_RESULTS, 0);
+ return ZEBRA_FAIL;
}
- logf (LOG_DEBUG, "result set sort input=%s output=%s",
- *input_setnames, output_setname);
+ if (!log_level_set)
+ loglevels();
+ yaz_log(log_level_sort, "result set sort input=%s output=%s",
+ *input_setnames, output_setname);
sset = resultSetGet (zh, input_setnames[0]);
if (!sset)
{
- zh->errCode = 30;
- zh->errString = nmem_strdup (nmem, input_setnames[0]);
- return;
+ zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
+ input_setnames[0]);
+ return ZEBRA_FAIL;
}
if (!(rset = sset->rset))
{
- zh->errCode = 30;
- zh->errString = nmem_strdup (nmem, input_setnames[0]);
- return;
+ zebra_setError(zh, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST,
+ input_setnames[0]);
+ return ZEBRA_FAIL;
}
if (strcmp (output_setname, input_setnames[0]))
- {
- rset = rset_dup (rset);
- sset = resultSetAdd (zh, output_setname, 1);
- sset->rset = rset;
- }
- resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
+ sset = resultSetClone(zh, output_setname, sset);
+ sset->sortSpec = copy_SortKeySpecList(sort_sequence, sset->nmem);
+ return resultSetSortSingle (zh, nmem, sset, rset, sort_sequence,
+ sort_status);
}
-void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
- ZebraSet sset, RSET rset,
- Z_SortKeySpecList *sort_sequence, int *sort_status)
+ZEBRA_RES resultSetSortSingle(ZebraHandle zh, NMEM nmem,
+ ZebraSet sset, RSET rset,
+ Z_SortKeySpecList *sort_sequence,
+ int *sort_status)
{
- int i, psysno = 0;
+ int i;
+ int n = 0;
+ zint kno = 0;
+ zint psysno = 0;
struct it_key key;
struct sortKeyInfo sort_criteria[3];
int num_criteria;
- int term_index;
RSFD rfd;
+ TERMID termid;
+ TERMID *terms;
+ int numTerms = 0;
+ size_t sysno_mem_index = 0;
+
+ if (zh->m_staticrank)
+ sysno_mem_index = 1;
- logf (LOG_DEBUG, "resultSetSortSingle start");
+ assert(nmem); /* compiler shut up about unused param */
sset->sort_info->num_entries = 0;
+ rset_getterms(rset, 0, 0, &n);
+ terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
+ rset_getterms(rset, terms, n, &numTerms);
+
sset->hits = 0;
num_criteria = sort_sequence->num_specs;
if (num_criteria > 3)
- num_criteria = 3;
+ num_criteria = 3;
for (i = 0; i < num_criteria; i++)
{
- Z_SortKeySpec *sks = sort_sequence->specs[i];
- Z_SortKey *sk;
+ Z_SortKeySpec *sks = sort_sequence->specs[i];
+ Z_SortKey *sk;
+ ZEBRA_RES res;
- if (*sks->sortRelation == Z_SortRelation_ascending)
- sort_criteria[i].relation = 'A';
- else if (*sks->sortRelation == Z_SortRelation_descending)
- sort_criteria[i].relation = 'D';
- else
- {
- zh->errCode = 214;
- return;
- }
- if (sks->sortElement->which == Z_SortElement_databaseSpecific)
- {
- zh->errCode = 210;
- return;
- }
- else if (sks->sortElement->which != Z_SortElement_generic)
- {
- zh->errCode = 237;
- return;
- }
- sk = sks->sortElement->u.generic;
- switch (sk->which)
- {
- case Z_SortKey_sortField:
- logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
- zh->errCode = 207;
- return;
- case Z_SortKey_elementSpec:
- logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
- zh->errCode = 207;
- return;
- 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,
- &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))
- {
- zh->errCode = 207;
- return;
- }
- break;
- }
+ sort_criteria[i].ord = -1;
+ sort_criteria[i].numerical = 0;
+
+ if (sks->which == Z_SortKeySpec_missingValueData)
+ {
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_MISSING_DATA_ACTION, 0);
+ return ZEBRA_FAIL;
+ }
+ if (*sks->sortRelation == Z_SortKeySpec_ascending)
+ sort_criteria[i].relation = 'A';
+ else if (*sks->sortRelation == Z_SortKeySpec_descending)
+ sort_criteria[i].relation = 'D';
+ else
+ {
+ zebra_setError(zh, YAZ_BIB1_ILLEGAL_SORT_RELATION, 0);
+ return ZEBRA_FAIL;
+ }
+ if (sks->sortElement->which == Z_SortElement_databaseSpecific)
+ {
+ zebra_setError(zh, YAZ_BIB1_DATABASE_SPECIFIC_SORT_UNSUPP, 0);
+ return ZEBRA_FAIL;
+ }
+ else if (sks->sortElement->which != Z_SortElement_generic)
+ {
+ zebra_setError(zh, YAZ_BIB1_SORT_ILLEGAL_SORT, 0);
+ return ZEBRA_FAIL;
+ }
+ sk = sks->sortElement->u.generic;
+ switch (sk->which)
+ {
+ case Z_SortKey_sortField:
+ yaz_log(log_level_sort, "key %d is of type sortField",
+ i+1);
+ sort_criteria[i].numerical = 0;
+ sort_criteria[i].ord =
+ zebraExplain_lookup_attr_str(zh->reg->zei,
+ zinfo_index_category_sort,
+ 's',
+ sk->u.sortField);
+ if (sks->which != Z_SortKeySpec_null
+ && sort_criteria[i].ord == -1)
+ {
+ zebra_setError(zh,
+ YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
+ return ZEBRA_FAIL;
+ }
+ break;
+ case Z_SortKey_elementSpec:
+ yaz_log(log_level_sort, "key %d is of type elementSpec",
+ i+1);
+ zebra_setError(zh, YAZ_BIB1_CANNOT_SORT_ACCORDING_TO_SEQUENCE, 0);
+ return ZEBRA_FAIL;
+ case Z_SortKey_sortAttributes:
+ yaz_log(log_level_sort, "key %d is of type sortAttributes", i+1);
+ res = zebra_sort_get_ord(zh, sk->u.sortAttributes,
+ &sort_criteria[i].ord,
+ &sort_criteria[i].numerical);
+ if (sks->which != Z_SortKeySpec_null && res != ZEBRA_OK)
+ return ZEBRA_FAIL;
+ break;
+ }
}
rfd = rset_open (rset, RSETF_READ);
- while (rset_read (rset, rfd, &key, &term_index))
+ while (rset_read (rfd, &key, &termid))
{
- if (key.sysno != psysno)
+ zint this_sys = key.mem[sysno_mem_index];
+ if (log_level_searchhits)
+ key_logdump_txt(log_level_searchhits, &key, termid->name);
+ kno++;
+ if (this_sys != psysno)
{
- (sset->hits)++;
- psysno = key.sysno;
- resultSetInsertSort (zh, sset,
- sort_criteria, num_criteria, psysno);
+ (sset->hits)++;
+ psysno = this_sys;
+ resultSetInsertSort (zh, sset,
+ sort_criteria, num_criteria, psysno);
}
}
- rset_close (rset, rfd);
-
- *sort_status = Z_SortStatus_success;
- logf (LOG_DEBUG, "resultSetSortSingle end");
+ rset_close (rfd);
+ yaz_log(log_level_sort, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, sort",
+ kno, sset->hits);
+ for (i = 0; i < numTerms; i++)
+ yaz_log(log_level_sort, "term=\"%s\" type=%s count=" ZINT_FORMAT,
+ terms[i]->name, terms[i]->flags, terms[i]->rset->hits_count);
+ *sort_status = Z_SortResponse_success;
+ return ZEBRA_OK;
}
-RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
+RSET resultSetRef(ZebraHandle zh, const char *resultSetId)
{
ZebraSet s;
if ((s = resultSetGet (zh, resultSetId)))
- return s->rset;
+ return s->rset;
return NULL;
}
-void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
+ZEBRA_RES resultSetRank(ZebraHandle zh, ZebraSet zebraSet,
+ RSET rset, NMEM nmem)
{
- int kno = 0;
struct it_key key;
- RSFD rfd;
- int term_index, i;
+ TERMID termid;
+ TERMID *terms;
+ zint kno = 0;
+ int numTerms = 0;
+ int n = 0;
+ int i;
ZebraRankClass rank_class;
- struct rank_control *rc;
struct zset_sort_info *sort_info;
+ const char *rank_handler_name = res_get_def(zh->res, "rank", "rank-1");
+ size_t sysno_mem_index = 0;
+
+ if (zh->m_staticrank)
+ sysno_mem_index = 1;
+ if (!log_level_set)
+ loglevels();
sort_info = zebraSet->sort_info;
sort_info->num_entries = 0;
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);
+ rset_getterms(rset, 0, 0, &n);
+ terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
+ rset_getterms(rset, terms, n, &numTerms);
- rank_class = zebraRankLookup (zh, "rank-1");
- rc = rank_class->control;
- if (rset_read (rset, rfd, &key, &term_index))
+ rank_class = zebraRankLookup(zh, rank_handler_name);
+ if (!rank_class)
+ {
+ yaz_log(YLOG_WARN, "No such rank handler: %s", rank_handler_name);
+ zebra_setError(zh, YAZ_BIB1_UNSUPP_SEARCH, "Cannot find rank handler");
+ return ZEBRA_FAIL;
+ }
+ else
{
- int psysno = key.sysno;
+ RSFD rfd = rset_open(rset, RSETF_READ);
+ struct rank_control *rc = rank_class->control;
int score;
+ zint count = 0;
+
void *handle =
- (*rc->begin) (zh, rank_class->class_handle, rset);
- (zebraSet->hits)++;
- do
+ (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem,
+ terms, numTerms);
+ zint psysno = 0; /* previous doc id / sys no */
+ zint pstaticrank = 0; /* previous static rank */
+ int stop_flag = 0;
+ while (rset_read(rfd, &key, &termid))
{
- kno++;
- if (key.sysno != psysno)
- {
- score = (*rc->calc) (handle, psysno);
+ zint this_sys = key.mem[sysno_mem_index];
- resultSetInsertRank (zh, sort_info, psysno, score, 'A');
- (zebraSet->hits)++;
- psysno = key.sysno;
+ zint seqno = key.mem[key.len-1];
+ kno++;
+ if (log_level_searchhits)
+ key_logdump_txt(log_level_searchhits, &key, termid->name);
+ if (this_sys != psysno)
+ { /* new record .. */
+ if (rfd->counted_items > rset->hits_limit)
+ break;
+ if (psysno)
+ { /* only if we did have a previous record */
+ score = (*rc->calc) (handle, psysno, pstaticrank,
+ &stop_flag);
+ /* insert the hit. A=Ascending */
+ resultSetInsertRank (zh, sort_info, psysno, score, 'A');
+ count++;
+ if (stop_flag)
+ break;
+ }
+ psysno = this_sys;
+ if (zh->m_staticrank)
+ pstaticrank = key.mem[0];
}
- (*rc->add) (handle, key.seqno, term_index);
+ (*rc->add) (handle, CAST_ZINT_TO_INT(seqno), termid);
}
- while (rset_read (rset, rfd, &key, &term_index));
- score = (*rc->calc) (handle, psysno);
- resultSetInsertRank (zh, sort_info, psysno, score, 'A');
- (*rc->end) (zh, handle);
+ /* no more items */
+ if (psysno)
+ { /* we had - at least - one record */
+ score = (*rc->calc)(handle, psysno, pstaticrank, &stop_flag);
+ /* insert the hit. A=Ascending */
+ resultSetInsertRank(zh, sort_info, psysno, score, 'A');
+ count++;
+ }
+ (*rc->end) (zh->reg, handle);
+ rset_close (rfd);
+ }
+ zebraSet->hits = rset->hits_count;
+
+ yaz_log(log_level_searchterms, ZINT_FORMAT " keys, "
+ ZINT_FORMAT " sysnos, rank", kno, zebraSet->hits);
+ for (i = 0; i < numTerms; i++)
+ {
+ yaz_log(log_level_searchterms, "term=\"%s\" type=%s count="
+ ZINT_FORMAT,
+ terms[i]->name, terms[i]->flags, terms[i]->rset->hits_count);
}
- rset_close (rset, rfd);
- logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
+ return ZEBRA_OK;
}
-ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
+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;
+ p = p->next;
if (p && !p->init_flag)
{
- if (p->control->create)
- p->class_handle = (*p->control->create)(zh->service);
- p->init_flag = 1;
+ if (p->control->create)
+ p->class_handle = (*p->control->create)(zh);
+ 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);
- xfree (p->control->name);
- xfree (p->control);
- xfree (p);
- p = p_next;
+ ZebraRankClass p_next = p->next;
+ if (p->init_flag && p->control->destroy)
+ (*p->control->destroy)(reg, p->class_handle);
+ xfree(p->control->name);
+ xfree(p->control);
+ xfree(p);
+ p = p_next;
+ }
+ reg->rank_classes = NULL;
+}
+
+static int trav_rset_for_termids(RSET rset, TERMID *termid_array,
+ zint *hits_array, int *approx_array)
+{
+ int no = 0;
+ int i;
+ for (i = 0; i<rset->no_children; i++)
+ no += trav_rset_for_termids(rset->children[i],
+ (termid_array ? termid_array + no : 0),
+ (hits_array ? hits_array + no : 0),
+ (approx_array ? approx_array + no : 0));
+ if (rset->term)
+ {
+ if (termid_array)
+ termid_array[no] = rset->term;
+ if (hits_array)
+ hits_array[no] = rset->hits_count;
+ if (approx_array)
+ approx_array[no] = rset->hits_approx;
+#if 0
+ yaz_log(YLOG_LOG, "rset=%p term=%s limit=" ZINT_FORMAT
+ " count=" ZINT_FORMAT,
+ rset, rset->term->name, rset->hits_limit, rset->hits_count);
+#endif
+ no++;
+ }
+ return no;
+}
+
+ZEBRA_RES zebra_result_set_term_no(ZebraHandle zh, const char *setname,
+ int *num_terms)
+{
+ ZebraSet sset = resultSetGet(zh, setname);
+ *num_terms = 0;
+ if (sset)
+ {
+ *num_terms = trav_rset_for_termids(sset->rset, 0, 0, 0);
+ return ZEBRA_OK;
}
- zh->rank_classes = NULL;
+ return ZEBRA_FAIL;
}
+
+ZEBRA_RES zebra_result_set_term_info(ZebraHandle zh, const char *setname,
+ int no, zint *count, int *approx,
+ char *termbuf, size_t *termlen,
+ const char **term_ref_id)
+{
+ ZebraSet sset = resultSetGet(zh, setname);
+ if (sset)
+ {
+ int num_terms = trav_rset_for_termids(sset->rset, 0, 0, 0);
+ if (no >= 0 && no < num_terms)
+ {
+ TERMID *term_array = xmalloc(num_terms * sizeof(*term_array));
+ zint *hits_array = xmalloc(num_terms * sizeof(*hits_array));
+ int *approx_array = xmalloc(num_terms * sizeof(*approx_array));
+
+ trav_rset_for_termids(sset->rset, term_array,
+ hits_array, approx_array);
+
+ if (count)
+ *count = hits_array[no];
+ if (approx)
+ *approx = approx_array[no];
+ if (termbuf)
+ {
+ char *inbuf = term_array[no]->name;
+ size_t inleft = strlen(inbuf);
+ size_t outleft = *termlen - 1;
+
+ if (zh->iconv_from_utf8 != 0)
+ {
+ char *outbuf = termbuf;
+ size_t ret;
+
+ ret = yaz_iconv(zh->iconv_from_utf8, &inbuf, &inleft,
+ &outbuf, &outleft);
+ if (ret == (size_t)(-1))
+ *termlen = 0;
+ else
+ *termlen = outbuf - termbuf;
+ }
+ else
+ {
+ if (inleft > outleft)
+ inleft = outleft;
+ *termlen = inleft;
+ memcpy(termbuf, inbuf, *termlen);
+ }
+ termbuf[*termlen] = '\0';
+ }
+ if (term_ref_id)
+ *term_ref_id = term_array[no]->ref_id;
+
+ xfree(term_array);
+ xfree(hits_array);
+ xfree(approx_array);
+ return ZEBRA_OK;
+ }
+ }
+ return ZEBRA_FAIL;
+}
+
+ZEBRA_RES zebra_snippets_hit_vector(ZebraHandle zh, const char *setname,
+ zint sysno, zebra_snippets *snippets)
+{
+ ZebraSet sset = resultSetGet(zh, setname);
+ yaz_log(YLOG_DEBUG, "zebra_get_hit_vector setname=%s zysno=" ZINT_FORMAT,
+ setname, sysno);
+ if (!sset)
+ return ZEBRA_FAIL;
+ else
+ {
+ struct rset_key_control *kc = zebra_key_control_create(zh);
+ NMEM nmem = nmem_create();
+ struct it_key key;
+ RSET rsets[2], rset_comb;
+ RSET rset_temp = rset_create_temp(nmem, kc, kc->scope,
+ res_get (zh->res, "setTmpDir"),0 );
+
+ TERMID termid;
+ RSFD rsfd = rset_open(rset_temp, RSETF_WRITE);
+
+ key.mem[0] = sysno;
+ key.mem[1] = 0;
+ key.mem[2] = 0;
+ key.mem[3] = 0;
+ key.len = 2;
+ rset_write (rsfd, &key);
+ rset_close (rsfd);
+
+ rsets[0] = rset_temp;
+ rsets[1] = rset_dup(sset->rset);
+
+ rset_comb = rset_create_and(nmem, kc, kc->scope, 2, rsets);
+
+ rsfd = rset_open(rset_comb, RSETF_READ);
+
+ while (rset_read(rsfd, &key, &termid))
+ {
+ if (termid)
+ {
+ struct ord_list *ol;
+ for (ol = termid->ol; ol; ol = ol->next)
+ {
+ zebra_snippets_append(snippets, key.mem[key.len-1],
+ ol->ord, termid->name);
+ }
+ }
+ }
+ rset_close(rsfd);
+
+ rset_delete(rset_comb);
+ nmem_destroy(nmem);
+ kc->dec(kc);
+ }
+ return ZEBRA_OK;
+}
+
+/*
+ * Local variables:
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+