1 /* $Id: zsets.c,v 1.75 2004-12-10 12:37:07 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");
91 ZebraSet resultSetAddRPN (ZebraHandle zh, NMEM m,
92 Z_RPNQuery *rpn, int num_bases,
100 zh->errString = NULL;
103 zebraSet = resultSetAdd (zh, setname, 1);
106 zebraSet->locked = 1;
109 zebraSet->rset_nmem=nmem_create();
111 zebraSet->num_bases = num_bases;
112 zebraSet->basenames =
113 nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames));
114 for (i = 0; i<num_bases; i++)
115 zebraSet->basenames[i] = nmem_strdup (zebraSet->nmem, basenames[i]);
118 zebraSet->rset = rpn_search (zh, zebraSet->nmem, zebraSet->rset_nmem,
119 rpn, zebraSet->num_bases,
120 zebraSet->basenames, zebraSet->name,
122 zh->hits = zebraSet->hits;
125 zebraSet->locked = 0;
129 void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type,
130 const char *db, int set,
131 int use, const char *term)
133 assert(zh); /* compiler shut up */
135 s->nmem = nmem_create ();
136 if (!s->term_entries)
139 s->term_entries_max = 1000;
141 nmem_malloc (s->nmem, s->term_entries_max *
142 sizeof(*s->term_entries));
143 for (i = 0; i < s->term_entries_max; i++)
144 s->term_entries[i].term = 0;
146 if (s->hits < s->term_entries_max)
148 s->term_entries[s->hits].reg_type = reg_type;
149 s->term_entries[s->hits].db = nmem_strdup (s->nmem, db);
150 s->term_entries[s->hits].set = set;
151 s->term_entries[s->hits].use = use;
152 s->term_entries[s->hits].term = nmem_strdup (s->nmem, term);
157 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
162 for (s = zh->sets; s; s = s->next)
163 if (!strcmp (s->name, name))
170 yaz_log(log_level_resultsets, "updating result set %s", name);
171 if (!ov || s->locked)
174 rset_delete (s->rset);
176 nmem_destroy (s->rset_nmem);
178 nmem_destroy (s->nmem);
182 const char *sort_max_str = zebra_get_resource(zh, "sortmax", "1000");
184 yaz_log(log_level_resultsets, "adding result set %s", name);
185 s = (ZebraSet) xmalloc (sizeof(*s));
188 s->name = (char *) xmalloc (strlen(name)+1);
189 strcpy (s->name, name);
191 s->sort_info = (struct zset_sort_info *)
192 xmalloc (sizeof(*s->sort_info));
193 s->sort_info->max_entries = atoi(sort_max_str);
194 if (s->sort_info->max_entries < 2)
195 s->sort_info->max_entries = 2;
197 s->sort_info->entries = (struct zset_sort_entry **)
198 xmalloc (sizeof(*s->sort_info->entries) *
199 s->sort_info->max_entries);
200 s->sort_info->all_entries = (struct zset_sort_entry *)
201 xmalloc (sizeof(*s->sort_info->all_entries) *
202 s->sort_info->max_entries);
203 for (i = 0; i < s->sort_info->max_entries; i++)
204 s->sort_info->entries[i] = s->sort_info->all_entries + i;
216 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
220 for (s = zh->sets; s; s = s->next)
221 if (!strcmp (s->name, name))
223 if (!s->term_entries && !s->rset && s->rpn)
225 NMEM nmem = nmem_create ();
226 yaz_log(log_level_resultsets, "research %s", name);
228 s->rset_nmem=nmem_create();
230 rpn_search (zh, nmem, s->rset_nmem, s->rpn, s->num_bases,
231 s->basenames, s->name, s);
239 void resultSetInvalidate (ZebraHandle zh)
241 ZebraSet s = zh->sets;
243 yaz_log(log_level_resultsets, "invalidating result sets");
244 for (; s; s = s->next)
247 rset_delete (s->rset);
250 nmem_destroy(s->rset_nmem);
255 void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
257 ZebraSet * ss = &zh->sets;
261 for (i = 0; i<num; i++)
262 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
264 zh->errString = NULL;
271 for (i = 0; i<num; i++)
272 if (!strcmp (s->name, names[i]))
275 statuses[i] = Z_DeleteStatus_success;
284 xfree (s->sort_info->all_entries);
285 xfree (s->sort_info->entries);
286 xfree (s->sort_info);
289 nmem_destroy (s->nmem);
291 rset_delete (s->rset);
293 nmem_destroy(s->rset_nmem);
302 ZebraMetaRecord *zebra_meta_records_create_range (ZebraHandle zh,
307 zint *pos = pos_small;
311 if (num > 10000 || num <= 0)
315 pos = xmalloc(sizeof(*pos) * num);
317 for (i = 0; i<num; i++)
320 mr = zebra_meta_records_create(zh, name, num, pos);
327 ZebraMetaRecord *zebra_meta_records_create (ZebraHandle zh, const char *name,
328 int num, zint *positions)
331 ZebraMetaRecord *sr = 0;
334 struct zset_sort_info *sort_info;
338 if (!(sset = resultSetGet (zh, name)))
340 if (!(rset = sset->rset))
342 if (!sset->term_entries)
344 sr = (ZebraMetaRecord *) xmalloc (sizeof(*sr) * num);
345 for (i = 0; i<num; i++)
352 if (positions[i] <= sset->term_entries_max)
354 sr[i].term = sset->term_entries[positions[i]-1].term;
355 sr[i].db = sset->term_entries[positions[i]-1].db;
361 sr = (ZebraMetaRecord *) xmalloc (sizeof(*sr) * num);
362 for (i = 0; i<num; i++)
369 sort_info = sset->sort_info;
374 for (i = 0; i<num; i++)
376 position = positions[i];
377 if (position > 0 && position <= sort_info->num_entries)
379 yaz_log(log_level_sorting, "got pos=%d (sorted)", position);
380 sr[i].sysno = sort_info->entries[position-1]->sysno;
381 sr[i].score = sort_info->entries[position-1]->score;
385 /* did we really get all entries using sort ? */
386 for (i = 0; i<num; i++)
391 if (i < num) /* nope, get the rest, unsorted - sorry */
400 position = sort_info->num_entries;
401 while (num_i < num && positions[num_i] < position)
403 rfd = rset_open (rset, RSETF_READ);
404 while (num_i < num && rset_read (rfd, &key, 0))
406 zint this_sys = key.mem[0];
407 if (this_sys != psysno)
412 /* determine we alreay have this in our set */
413 for (i = sort_info->num_entries; --i >= 0; )
414 if (psysno == sort_info->entries[i]->sysno)
420 assert (num_i < num);
421 if (position == positions[num_i])
423 sr[num_i].sysno = psysno;
424 yaz_log(log_level_sorting, "got pos=%d (unsorted)", position);
425 sr[num_i].score = -1;
436 void zebra_meta_records_destroy (ZebraHandle zh, ZebraMetaRecord *records,
439 assert(zh); /* compiler shut up about unused arg */
449 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
450 struct sortKeyInfo *criteria, int num_criteria,
453 struct zset_sort_entry this_entry;
454 struct zset_sort_entry *new_entry = NULL;
455 struct zset_sort_info *sort_info = sset->sort_info;
458 sortIdx_sysno (zh->reg->sortIdx, sysno);
459 for (i = 0; i<num_criteria; i++)
461 sortIdx_type (zh->reg->sortIdx, criteria[i].attrUse);
462 sortIdx_read (zh->reg->sortIdx, this_entry.buf[i]);
464 i = sort_info->num_entries;
468 for (j = 0; j<num_criteria; j++)
470 if (criteria[j].numerical)
472 double diff = atof(this_entry.buf[j]) -
473 atof(sort_info->entries[i]->buf[j]);
482 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
490 if (criteria[j].relation == 'A')
495 else if (criteria[j].relation == 'D')
502 j = sort_info->max_entries;
506 if (sort_info->num_entries == j)
509 j = (sort_info->num_entries)++;
510 new_entry = sort_info->entries[j];
513 sort_info->entries[j] = sort_info->entries[j-1];
516 sort_info->entries[i] = new_entry;
518 for (i = 0; i<num_criteria; i++)
519 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
520 new_entry->sysno = sysno;
521 new_entry->score = -1;
524 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
525 zint sysno, int score, int relation)
527 struct zset_sort_entry *new_entry = NULL;
529 assert(zh); /* compiler shut up about unused arg */
531 i = sort_info->num_entries;
536 rel = score - sort_info->entries[i]->score;
543 else if (relation == 'A')
550 j = sort_info->max_entries;
554 if (sort_info->num_entries == j)
557 j = (sort_info->num_entries)++;
559 new_entry = sort_info->entries[j];
562 sort_info->entries[j] = sort_info->entries[j-1];
565 sort_info->entries[i] = new_entry;
567 new_entry->sysno = sysno;
568 new_entry->score = score;
571 void resultSetSort (ZebraHandle zh, NMEM nmem,
572 int num_input_setnames, const char **input_setnames,
573 const char *output_setname,
574 Z_SortKeySpecList *sort_sequence, int *sort_status)
579 if (num_input_setnames == 0)
584 if (num_input_setnames > 1)
591 yaz_log(log_level_sorting, "result set sort input=%s output=%s",
592 *input_setnames, output_setname);
593 sset = resultSetGet (zh, input_setnames[0]);
597 zh->errString = nmem_strdup (nmem, input_setnames[0]);
600 if (!(rset = sset->rset))
603 zh->errString = nmem_strdup (nmem, input_setnames[0]);
606 if (strcmp (output_setname, input_setnames[0]))
608 rset = rset_dup (rset);
609 sset = resultSetAdd (zh, output_setname, 1);
612 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
615 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
616 ZebraSet sset, RSET rset,
617 Z_SortKeySpecList *sort_sequence, int *sort_status)
624 struct sortKeyInfo sort_criteria[3];
631 assert(nmem); /* compiler shut up about unused param */
632 sset->sort_info->num_entries = 0;
634 rset_getterms(rset, 0, 0, &n);
635 terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
636 rset_getterms(rset, terms, n, &numTerms);
639 num_criteria = sort_sequence->num_specs;
640 if (num_criteria > 3)
642 for (i = 0; i < num_criteria; i++)
644 Z_SortKeySpec *sks = sort_sequence->specs[i];
647 if (*sks->sortRelation == Z_SortKeySpec_ascending)
648 sort_criteria[i].relation = 'A';
649 else if (*sks->sortRelation == Z_SortKeySpec_descending)
650 sort_criteria[i].relation = 'D';
656 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
661 else if (sks->sortElement->which != Z_SortElement_generic)
666 sk = sks->sortElement->u.generic;
669 case Z_SortKey_sortField:
670 yaz_log(log_level_sorting, "Sort: key %d is of type sortField", i+1);
673 case Z_SortKey_elementSpec:
674 yaz_log(log_level_sorting, "Sort: key %d is of type elementSpec", i+1);
677 case Z_SortKey_sortAttributes:
678 yaz_log(log_level_sorting, "Sort: key %d is of type sortAttributes", i+1);
679 sort_criteria[i].attrUse =
680 zebra_maps_sort (zh->reg->zebra_maps,
681 sk->u.sortAttributes,
682 &sort_criteria[i].numerical);
683 yaz_log(log_level_sorting, "use value = %d", sort_criteria[i].attrUse);
684 if (sort_criteria[i].attrUse == -1)
689 if (sortIdx_type (zh->reg->sortIdx, sort_criteria[i].attrUse))
697 rfd = rset_open (rset, RSETF_READ);
698 while (rset_read (rfd, &key, &termid))
699 /* FIXME - pass a TERMID *, and use it for something below !! */
701 zint this_sys = key.mem[0];
703 if (this_sys != psysno)
707 resultSetInsertSort (zh, sset,
708 sort_criteria, num_criteria, psysno);
712 yaz_log(log_level_sorting, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, sort",
714 for (i = 0; i < numTerms; i++)
715 yaz_log(log_level_sorting, "term=\"%s\" type=%s count=" ZINT_FORMAT,
716 terms[i]->name, terms[i]->flags, rset_count(terms[i]->rset));
717 *sort_status = Z_SortResponse_success;
720 RSET resultSetRef (ZebraHandle zh, const char *resultSetId)
724 if ((s = resultSetGet (zh, resultSetId)))
729 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset, NMEM nmem)
739 ZebraRankClass rank_class;
740 struct rank_control *rc;
741 struct zset_sort_info *sort_info;
742 const char *rank_handler_name = res_get_def(zh->res, "rank", "rank-1");
744 zint est=-2; /* -2 not done, -1 can't do, >0 actual estimate*/
750 sort_info = zebraSet->sort_info;
751 sort_info->num_entries = 0;
753 rset_getterms(rset, 0, 0, &n);
754 terms = (TERMID *) nmem_malloc(nmem, sizeof(*terms)*n);
755 rset_getterms(rset, terms, n, &numTerms);
757 rfd = rset_open (rset, RSETF_READ);
759 rank_class = zebraRankLookup (zh, rank_handler_name);
762 yaz_log(YLOG_WARN, "No such rank handler: %s", rank_handler_name);
765 rc = rank_class->control;
767 if (rset_read (rfd, &key, &termid))
769 zint psysno = key.mem[0];
772 (*rc->begin) (zh->reg, rank_class->class_handle, rset, nmem,
775 esthits = atoi(res_get_def(zh->res, "estimatehits","0"));
777 est = -1; /* can not do */
780 zint this_sys = key.mem[0]; /* FIXME - assumes scope==2 */
781 zint seqno = key.mem[key.len-1]; /* FIXME - assumes scope==2 */
783 key_logdump_txt(log_level_searchhits,&key," Got hit");
784 if (this_sys != psysno)
786 score = (*rc->calc) (handle, psysno);
788 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
792 (*rc->add) (handle, seqno, termid);
794 if ((est==-2) && (zebraSet->hits==esthits))
795 { /* time to estimate the hits */
796 rset_pos(rfd,&cur,&tot);
799 est = (zint)(0.5+zebraSet->hits/ratio);
800 yaz_log(log_level_searchhits, "Estimating hits (%s) "
801 "%0.1f->" ZINT_FORMAT
802 "; %0.1f->" ZINT_FORMAT,
806 i = 0; /* round to 3 significant digits */
813 zebraSet->hits = est;
817 while (rset_read (rfd, &key,&termid) && (est<0) );
818 score = (*rc->calc)(handle, psysno);
819 resultSetInsertRank(zh, sort_info, psysno, score, 'A');
820 (*rc->end) (zh->reg, handle);
824 yaz_log(log_level_searchterms, ZINT_FORMAT " keys, " ZINT_FORMAT " sysnos, rank",
825 kno, zebraSet->hits);
826 for (i = 0; i < numTerms; i++)
828 yaz_log(log_level_searchterms, "term=\"%s\" type=%s count=" ZINT_FORMAT,
829 terms[i]->name, terms[i]->flags, rset_count(terms[i]->rset));
833 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
835 ZebraRankClass p = zh->reg->rank_classes;
836 while (p && strcmp (p->control->name, name))
838 if (p && !p->init_flag)
840 if (p->control->create)
841 p->class_handle = (*p->control->create)(zh);
847 void zebraRankInstall (struct zebra_register *reg, struct rank_control *ctrl)
849 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
850 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
851 memcpy (p->control, ctrl, sizeof(*p->control));
852 p->control->name = xstrdup (ctrl->name);
854 p->next = reg->rank_classes;
855 reg->rank_classes = p;
858 void zebraRankDestroy (struct zebra_register *reg)
860 ZebraRankClass p = reg->rank_classes;
863 ZebraRankClass p_next = p->next;
864 if (p->init_flag && p->control->destroy)
865 (*p->control->destroy)(reg, p->class_handle);
866 xfree(p->control->name);
871 reg->rank_classes = NULL;