1 /* $Id: zsets.c,v 1.74 2004-12-10 11:56:22 heikki Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
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 #define SORT_IDX_ENTRYSIZE 64
36 #define ZSET_SORT_MAX_LEVEL 3
38 struct zebra_set_term_entry {
49 NMEM rset_nmem; /* for creating the rsets in */
54 struct zset_sort_info *sort_info;
55 struct zebra_set_term_entry *term_entries;
57 struct zebra_set *next;
61 struct zset_sort_entry {
64 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
67 struct zset_sort_info {
70 struct zset_sort_entry *all_entries;
71 struct zset_sort_entry **entries;
74 static int log_level_set=0;
75 static int log_level_sorting=0;
76 static int log_level_searchhits=0;
77 static int log_level_searchterms=0;
78 static int log_level_resultsets=0;
80 static void loglevels()
84 log_level_sorting = yaz_log_module_level("sorting");
85 log_level_searchhits = yaz_log_module_level("searchhits");
86 log_level_searchterms = yaz_log_module_level("searchterms");
87 log_level_resultsets = yaz_log_module_level("resultsets");
90 ZebraSet resultSetAddRPN (ZebraHandle zh, NMEM m,
91 Z_RPNQuery *rpn, int num_bases,
102 zebraSet = resultSetAdd (zh, setname, 1);
105 zebraSet->locked = 1;
108 zebraSet->rset_nmem=nmem_create();
110 zebraSet->num_bases = num_bases;
111 zebraSet->basenames =
112 nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames));
113 for (i = 0; i<num_bases; i++)
114 zebraSet->basenames[i] = nmem_strdup (zebraSet->nmem, basenames[i]);
117 zebraSet->rset = rpn_search (zh, zebraSet->nmem, zebraSet->rset_nmem,
118 rpn, zebraSet->num_bases,
119 zebraSet->basenames, zebraSet->name,
121 zh->hits = zebraSet->hits;
124 zebraSet->locked = 0;
128 void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type,
129 const char *db, int set,
130 int use, const char *term)
132 assert(zh); /* compiler shut up */
134 s->nmem = nmem_create ();
135 if (!s->term_entries)
138 s->term_entries_max = 1000;
140 nmem_malloc (s->nmem, s->term_entries_max *
141 sizeof(*s->term_entries));
142 for (i = 0; i < s->term_entries_max; i++)
143 s->term_entries[i].term = 0;
145 if (s->hits < s->term_entries_max)
147 s->term_entries[s->hits].reg_type = reg_type;
148 s->term_entries[s->hits].db = nmem_strdup (s->nmem, db);
149 s->term_entries[s->hits].set = set;
150 s->term_entries[s->hits].use = use;
151 s->term_entries[s->hits].term = nmem_strdup (s->nmem, term);
156 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
161 for (s = zh->sets; s; s = s->next)
162 if (!strcmp (s->name, name))
169 yaz_log(log_level_resultsets, "updating result set %s", name);
170 if (!ov || s->locked)
173 rset_delete (s->rset);
175 nmem_destroy (s->rset_nmem);
177 nmem_destroy (s->nmem);
181 const char *sort_max_str = zebra_get_resource(zh, "sortmax", "1000");
183 yaz_log(log_level_resultsets, "adding result set %s", name);
184 s = (ZebraSet) xmalloc (sizeof(*s));
187 s->name = (char *) xmalloc (strlen(name)+1);
188 strcpy (s->name, name);
190 s->sort_info = (struct zset_sort_info *)
191 xmalloc (sizeof(*s->sort_info));
192 s->sort_info->max_entries = atoi(sort_max_str);
193 if (s->sort_info->max_entries < 2)
194 s->sort_info->max_entries = 2;
196 s->sort_info->entries = (struct zset_sort_entry **)
197 xmalloc (sizeof(*s->sort_info->entries) *
198 s->sort_info->max_entries);
199 s->sort_info->all_entries = (struct zset_sort_entry *)
200 xmalloc (sizeof(*s->sort_info->all_entries) *
201 s->sort_info->max_entries);
202 for (i = 0; i < s->sort_info->max_entries; i++)
203 s->sort_info->entries[i] = s->sort_info->all_entries + i;
215 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
219 for (s = zh->sets; s; s = s->next)
220 if (!strcmp (s->name, name))
222 if (!s->term_entries && !s->rset && s->rpn)
224 NMEM nmem = nmem_create ();
225 yaz_log(log_level_resultsets, "research %s", name);
227 s->rset_nmem=nmem_create();
229 rpn_search (zh, nmem, s->rset_nmem, s->rpn, s->num_bases,
230 s->basenames, s->name, s);
238 void resultSetInvalidate (ZebraHandle zh)
240 ZebraSet s = zh->sets;
242 yaz_log(log_level_resultsets, "invalidating result sets");
243 for (; s; s = s->next)
246 rset_delete (s->rset);
249 nmem_destroy(s->rset_nmem);
254 void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
256 ZebraSet * ss = &zh->sets;
260 for (i = 0; i<num; i++)
261 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
263 zh->errString = NULL;
270 for (i = 0; i<num; i++)
271 if (!strcmp (s->name, names[i]))
274 statuses[i] = Z_DeleteStatus_success;
283 xfree (s->sort_info->all_entries);
284 xfree (s->sort_info->entries);
285 xfree (s->sort_info);
288 nmem_destroy (s->nmem);
290 rset_delete (s->rset);
292 nmem_destroy(s->rset_nmem);
301 ZebraMetaRecord *zebra_meta_records_create_range (ZebraHandle zh,
306 zint *pos = pos_small;
310 if (num > 10000 || num <= 0)
314 pos = xmalloc(sizeof(*pos) * num);
316 for (i = 0; i<num; i++)
319 mr = zebra_meta_records_create(zh, name, num, pos);
326 ZebraMetaRecord *zebra_meta_records_create (ZebraHandle zh, const char *name,
327 int num, zint *positions)
330 ZebraMetaRecord *sr = 0;
333 struct zset_sort_info *sort_info;
337 if (!(sset = resultSetGet (zh, name)))
339 if (!(rset = sset->rset))
341 if (!sset->term_entries)
343 sr = (ZebraMetaRecord *) xmalloc (sizeof(*sr) * num);
344 for (i = 0; i<num; i++)
351 if (positions[i] <= sset->term_entries_max)
353 sr[i].term = sset->term_entries[positions[i]-1].term;
354 sr[i].db = sset->term_entries[positions[i]-1].db;
360 sr = (ZebraMetaRecord *) xmalloc (sizeof(*sr) * num);
361 for (i = 0; i<num; i++)
368 sort_info = sset->sort_info;
373 for (i = 0; i<num; i++)
375 position = positions[i];
376 if (position > 0 && position <= sort_info->num_entries)
378 yaz_log(log_level_sorting, "got pos=%d (sorted)", position);
379 sr[i].sysno = sort_info->entries[position-1]->sysno;
380 sr[i].score = sort_info->entries[position-1]->score;
384 /* did we really get all entries using sort ? */
385 for (i = 0; i<num; i++)
390 if (i < num) /* nope, get the rest, unsorted - sorry */
399 position = sort_info->num_entries;
400 while (num_i < num && positions[num_i] < position)
402 rfd = rset_open (rset, RSETF_READ);
403 while (num_i < num && rset_read (rfd, &key, 0))
405 zint this_sys = key.mem[0];
406 if (this_sys != psysno)
411 /* determine we alreay have this in our set */
412 for (i = sort_info->num_entries; --i >= 0; )
413 if (psysno == sort_info->entries[i]->sysno)
419 assert (num_i < num);
420 if (position == positions[num_i])
422 sr[num_i].sysno = psysno;
423 yaz_log(log_level_sorting, "got pos=%d (unsorted)", position);
424 sr[num_i].score = -1;
435 void zebra_meta_records_destroy (ZebraHandle zh, ZebraMetaRecord *records,
438 assert(zh); /* compiler shut up about unused arg */
448 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
449 struct sortKeyInfo *criteria, int num_criteria,
452 struct zset_sort_entry this_entry;
453 struct zset_sort_entry *new_entry = NULL;
454 struct zset_sort_info *sort_info = sset->sort_info;
457 sortIdx_sysno (zh->reg->sortIdx, sysno);
458 for (i = 0; i<num_criteria; i++)
460 sortIdx_type (zh->reg->sortIdx, criteria[i].attrUse);
461 sortIdx_read (zh->reg->sortIdx, this_entry.buf[i]);
463 i = sort_info->num_entries;
467 for (j = 0; j<num_criteria; j++)
469 if (criteria[j].numerical)
471 double diff = atof(this_entry.buf[j]) -
472 atof(sort_info->entries[i]->buf[j]);
481 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
489 if (criteria[j].relation == 'A')
494 else if (criteria[j].relation == 'D')
501 j = sort_info->max_entries;
505 if (sort_info->num_entries == j)
508 j = (sort_info->num_entries)++;
509 new_entry = sort_info->entries[j];
512 sort_info->entries[j] = sort_info->entries[j-1];
515 sort_info->entries[i] = new_entry;
517 for (i = 0; i<num_criteria; i++)
518 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
519 new_entry->sysno = sysno;
520 new_entry->score = -1;
523 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
524 zint sysno, int score, int relation)
526 struct zset_sort_entry *new_entry = NULL;
528 assert(zh); /* compiler shut up about unused arg */
530 i = sort_info->num_entries;
535 rel = score - sort_info->entries[i]->score;
542 else if (relation == 'A')
549 j = sort_info->max_entries;
553 if (sort_info->num_entries == j)
556 j = (sort_info->num_entries)++;
558 new_entry = sort_info->entries[j];
561 sort_info->entries[j] = sort_info->entries[j-1];
564 sort_info->entries[i] = new_entry;
566 new_entry->sysno = sysno;
567 new_entry->score = score;
570 void resultSetSort (ZebraHandle zh, NMEM nmem,
571 int num_input_setnames, const char **input_setnames,
572 const char *output_setname,
573 Z_SortKeySpecList *sort_sequence, int *sort_status)
578 if (num_input_setnames == 0)
583 if (num_input_setnames > 1)
590 yaz_log(log_level_sorting, "result set sort input=%s output=%s",
591 *input_setnames, output_setname);
592 sset = resultSetGet (zh, input_setnames[0]);
596 zh->errString = nmem_strdup (nmem, input_setnames[0]);
599 if (!(rset = sset->rset))
602 zh->errString = nmem_strdup (nmem, input_setnames[0]);
605 if (strcmp (output_setname, input_setnames[0]))
607 rset = rset_dup (rset);
608 sset = resultSetAdd (zh, output_setname, 1);
611 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
614 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
615 ZebraSet sset, RSET rset,
616 Z_SortKeySpecList *sort_sequence, int *sort_status)
623 struct sortKeyInfo sort_criteria[3];
630 assert(nmem); /* compiler shut up about unused param */
631 sset->sort_info->num_entries = 0;
633 rset_getterms(rset, 0, 0, &n);
634 terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
635 rset_getterms(rset, terms, n, &numTerms);
638 num_criteria = sort_sequence->num_specs;
639 if (num_criteria > 3)
641 for (i = 0; i < num_criteria; i++)
643 Z_SortKeySpec *sks = sort_sequence->specs[i];
646 if (*sks->sortRelation == Z_SortKeySpec_ascending)
647 sort_criteria[i].relation = 'A';
648 else if (*sks->sortRelation == Z_SortKeySpec_descending)
649 sort_criteria[i].relation = 'D';
655 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
660 else if (sks->sortElement->which != Z_SortElement_generic)
665 sk = sks->sortElement->u.generic;
668 case Z_SortKey_sortField:
669 yaz_log(log_level_sorting, "Sort: key %d is of type sortField", i+1);
672 case Z_SortKey_elementSpec:
673 yaz_log(log_level_sorting, "Sort: key %d is of type elementSpec", i+1);
676 case Z_SortKey_sortAttributes:
677 yaz_log(log_level_sorting, "Sort: key %d is of type sortAttributes", i+1);
678 sort_criteria[i].attrUse =
679 zebra_maps_sort (zh->reg->zebra_maps,
680 sk->u.sortAttributes,
681 &sort_criteria[i].numerical);
682 yaz_log(log_level_sorting, "use value = %d", sort_criteria[i].attrUse);
683 if (sort_criteria[i].attrUse == -1)
688 if (sortIdx_type (zh->reg->sortIdx, sort_criteria[i].attrUse))
696 rfd = rset_open (rset, RSETF_READ);
697 while (rset_read (rfd, &key, &termid))
698 /* FIXME - pass a TERMID *, and use it for something below !! */
700 zint this_sys = key.mem[0];
702 if (this_sys != psysno)
706 resultSetInsertSort (zh, sset,
707 sort_criteria, num_criteria, psysno);
711 yaz_log(log_level_sorting, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, sort",
713 for (i = 0; i < numTerms; i++)
714 yaz_log(log_level_sorting, "term=\"%s\" type=%s count=" ZINT_FORMAT,
715 terms[i]->name, terms[i]->flags, rset_count(terms[i]->rset));
716 *sort_status = Z_SortResponse_success;
719 RSET resultSetRef (ZebraHandle zh, const char *resultSetId)
723 if ((s = resultSetGet (zh, resultSetId)))
728 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset, NMEM nmem)
738 ZebraRankClass rank_class;
739 struct rank_control *rc;
740 struct zset_sort_info *sort_info;
741 const char *rank_handler_name = res_get_def(zh->res, "rank", "rank-1");
743 zint est=-2; /* -2 not done, -1 can't do, >0 actual estimate*/
749 sort_info = zebraSet->sort_info;
750 sort_info->num_entries = 0;
752 rset_getterms(rset, 0, 0, &n);
753 terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
754 rset_getterms(rset, terms, n, &numTerms);
756 rfd = rset_open (rset, RSETF_READ);
758 rank_class = zebraRankLookup (zh, rank_handler_name);
761 yaz_log(YLOG_WARN, "No such rank handler: %s", rank_handler_name);
764 rc = rank_class->control;
766 if (rset_read (rfd, &key, &termid))
768 zint psysno = key.mem[0];
771 (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem,
774 esthits = atoi(res_get_def(zh->res, "estimatehits","0"));
776 est = -1; /* can not do */
779 zint this_sys = key.mem[0]; /* FIXME - assumes scope==2 */
780 zint seqno = key.mem[key.len-1]; /* FIXME - assumes scope==2 */
782 key_logdump_txt(log_level_searchhits,&key," Got hit");
783 if (this_sys != psysno)
785 score = (*rc->calc) (handle, psysno);
787 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
791 (*rc->add) (handle, seqno, termid);
793 if ((est==-2) && (zebraSet->hits==esthits))
794 { /* time to estimate the hits */
795 rset_pos(rfd,&cur,&tot);
798 est = (zint)(0.5+zebraSet->hits/ratio);
799 yaz_log(log_level_searchhits, "Estimating hits (%s) "
800 "%0.1f->" ZINT_FORMAT
801 "; %0.1f->" ZINT_FORMAT,
805 i = 0; /* round to 3 significant digits */
812 zebraSet->hits = est;
816 while (rset_read (rfd, &key,&termid) && (est<0) );
817 score = (*rc->calc)(handle, psysno);
818 resultSetInsertRank(zh, sort_info, psysno, score, 'A');
819 (*rc->end) (zh->reg, handle);
823 yaz_log(log_level_searchterms, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, rank",
824 kno, zebraSet->hits);
825 for (i = 0; i < numTerms; i++)
827 yaz_log(log_level_searchterms, "term=\"%s\" type=%s count=" ZINT_FORMAT,
828 terms[i]->name, terms[i]->flags, rset_count(terms[i]->rset));
832 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
834 ZebraRankClass p = zh->reg->rank_classes;
835 while (p && strcmp (p->control->name, name))
837 if (p && !p->init_flag)
839 if (p->control->create)
840 p->class_handle = (*p->control->create)(zh);
846 void zebraRankInstall (struct zebra_register *reg, struct rank_control *ctrl)
848 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
849 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
850 memcpy (p->control, ctrl, sizeof(*p->control));
851 p->control->name = xstrdup (ctrl->name);
853 p->next = reg->rank_classes;
854 reg->rank_classes = p;
857 void zebraRankDestroy (struct zebra_register *reg)
859 ZebraRankClass p = reg->rank_classes;
862 ZebraRankClass p_next = p->next;
863 if (p->init_flag && p->control->destroy)
864 (*p->control->destroy)(reg, p->class_handle);
865 xfree(p->control->name);
870 reg->rank_classes = NULL;