1 /* $Id: zsets.c,v 1.53 2004-08-06 12:55:01 adam 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 {
53 struct zset_sort_info *sort_info;
54 struct zebra_set_term_entry *term_entries;
56 struct zebra_set *next;
60 struct zset_sort_entry {
63 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
66 struct zset_sort_info {
69 struct zset_sort_entry *all_entries;
70 struct zset_sort_entry **entries;
73 ZebraSet resultSetAddRPN (ZebraHandle zh, NMEM m,
74 Z_RPNQuery *rpn, int num_bases,
85 zebraSet = resultSetAdd (zh, setname, 1);
92 zebraSet->num_bases = num_bases;
94 nmem_malloc (zebraSet->nmem, num_bases * sizeof(*zebraSet->basenames));
95 for (i = 0; i<num_bases; i++)
96 zebraSet->basenames[i] = nmem_strdup (zebraSet->nmem, basenames[i]);
99 zebraSet->rset = rpn_search (zh, zebraSet->nmem, rpn,
101 zebraSet->basenames, zebraSet->name,
103 zh->hits = zebraSet->hits;
106 zebraSet->locked = 0;
110 void resultSetAddTerm (ZebraHandle zh, ZebraSet s, int reg_type,
111 const char *db, int set,
112 int use, const char *term)
115 s->nmem = nmem_create ();
116 if (!s->term_entries)
119 s->term_entries_max = 1000;
121 nmem_malloc (s->nmem, s->term_entries_max *
122 sizeof(*s->term_entries));
123 for (i = 0; i < s->term_entries_max; i++)
124 s->term_entries[i].term = 0;
126 if (s->hits < s->term_entries_max)
128 s->term_entries[s->hits].reg_type = reg_type;
129 s->term_entries[s->hits].db = nmem_strdup (s->nmem, db);
130 s->term_entries[s->hits].set = set;
131 s->term_entries[s->hits].use = use;
132 s->term_entries[s->hits].term = nmem_strdup (s->nmem, term);
138 int zebra_resultSetTerms (ZebraHandle zh, const char *setname,
140 int *type, char *out, size_t *len)
142 ZebraSet s = resultSetGet (zh, setname);
149 no_max = s->rset->no_rset_terms;
150 if (no < 0 || no >= no_max)
153 *count = s->rset->rset_terms[no]->count;
155 *type = s->rset->rset_terms[no]->type;
159 char *inbuf = s->rset->rset_terms[no]->name;
160 size_t inleft = strlen(inbuf);
161 size_t outleft = *len - 1;
164 if (zh->iconv_from_utf8 != 0)
169 ret = yaz_iconv(zh->iconv_from_utf8, &inbuf, &inleft,
171 if (ret == (size_t)(-1))
179 if (inleft > outleft)
182 memcpy (out, inbuf, *len);
190 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
195 for (s = zh->sets; s; s = s->next)
196 if (!strcmp (s->name, name))
200 yaz_log (LOG_DEBUG, "updating result set %s", name);
201 if (!ov || s->locked)
204 rset_delete (s->rset);
206 nmem_destroy (s->nmem);
210 const char *sort_max_str = zebra_get_resource(zh, "sortmax", "1000");
212 yaz_log (LOG_DEBUG, "adding result set %s", name);
213 s = (ZebraSet) xmalloc (sizeof(*s));
216 s->name = (char *) xmalloc (strlen(name)+1);
217 strcpy (s->name, name);
219 s->sort_info = (struct zset_sort_info *)
220 xmalloc (sizeof(*s->sort_info));
221 s->sort_info->max_entries = atoi(sort_max_str);
222 if (s->sort_info->max_entries < 2)
223 s->sort_info->max_entries = 2;
225 s->sort_info->entries = (struct zset_sort_entry **)
226 xmalloc (sizeof(*s->sort_info->entries) *
227 s->sort_info->max_entries);
228 s->sort_info->all_entries = (struct zset_sort_entry *)
229 xmalloc (sizeof(*s->sort_info->all_entries) *
230 s->sort_info->max_entries);
231 for (i = 0; i < s->sort_info->max_entries; i++)
232 s->sort_info->entries[i] = s->sort_info->all_entries + i;
243 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
247 for (s = zh->sets; s; s = s->next)
248 if (!strcmp (s->name, name))
250 if (!s->term_entries && !s->rset && s->rpn)
252 NMEM nmem = nmem_create ();
253 yaz_log (LOG_LOG, "research %s", name);
255 rpn_search (zh, nmem, s->rpn, s->num_bases,
256 s->basenames, s->name, s);
264 void resultSetInvalidate (ZebraHandle zh)
266 ZebraSet s = zh->sets;
268 for (; s; s = s->next)
271 rset_delete (s->rset);
276 void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
278 ZebraSet * ss = &zh->sets;
282 for (i = 0; i<num; i++)
283 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
285 zh->errString = NULL;
292 for (i = 0; i<num; i++)
293 if (!strcmp (s->name, names[i]))
296 statuses[i] = Z_DeleteStatus_success;
305 xfree (s->sort_info->all_entries);
306 xfree (s->sort_info->entries);
307 xfree (s->sort_info);
310 nmem_destroy (s->nmem);
312 rset_delete (s->rset);
321 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
322 int num, int *positions)
328 struct zset_sort_info *sort_info;
330 if (!(sset = resultSetGet (zh, name)))
332 if (!(rset = sset->rset))
334 if (!sset->term_entries)
336 sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
337 for (i = 0; i<num; i++)
344 if (positions[i] <= sset->term_entries_max)
346 sr[i].term = sset->term_entries[positions[i]-1].term;
347 sr[i].db = sset->term_entries[positions[i]-1].db;
353 sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
354 for (i = 0; i<num; i++)
361 sort_info = sset->sort_info;
366 for (i = 0; i<num; i++)
368 position = positions[i];
369 if (position > 0 && position <= sort_info->num_entries)
371 yaz_log (LOG_DEBUG, "got pos=%d (sorted)", position);
372 sr[i].sysno = sort_info->entries[position-1]->sysno;
373 sr[i].score = sort_info->entries[position-1]->score;
377 /* did we really get all entries using sort ? */
378 for (i = 0; i<num; i++)
383 if (i < num) /* nope, get the rest, unsorted - sorry */
393 position = sort_info->num_entries;
394 while (num_i < num && positions[num_i] < position)
396 rfd = rset_open (rset, RSETF_READ);
397 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
400 zint this_sys = key.mem[0];
402 zint this_sys = key.sysno;
404 if (this_sys != psysno)
409 /* determine we alreay have this in our set */
410 for (i = sort_info->num_entries; --i >= 0; )
411 if (psysno == sort_info->entries[i]->sysno)
417 assert (num_i < num);
418 if (position == positions[num_i])
420 sr[num_i].sysno = psysno;
421 yaz_log (LOG_DEBUG, "got pos=%d (unsorted)", position);
422 sr[num_i].score = -1;
427 rset_close (rset, rfd);
433 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
444 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
445 struct sortKeyInfo *criteria, int num_criteria,
448 struct zset_sort_entry this_entry;
449 struct zset_sort_entry *new_entry = NULL;
450 struct zset_sort_info *sort_info = sset->sort_info;
453 sortIdx_sysno (zh->reg->sortIdx, sysno);
454 for (i = 0; i<num_criteria; i++)
456 sortIdx_type (zh->reg->sortIdx, criteria[i].attrUse);
457 sortIdx_read (zh->reg->sortIdx, this_entry.buf[i]);
459 i = sort_info->num_entries;
463 for (j = 0; j<num_criteria; j++)
465 if (criteria[j].numerical)
467 double diff = atof(this_entry.buf[j]) -
468 atof(sort_info->entries[i]->buf[j]);
477 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
485 if (criteria[j].relation == 'A')
490 else if (criteria[j].relation == 'D')
497 j = sort_info->max_entries;
501 if (sort_info->num_entries == j)
504 j = (sort_info->num_entries)++;
505 new_entry = sort_info->entries[j];
508 sort_info->entries[j] = sort_info->entries[j-1];
511 sort_info->entries[i] = new_entry;
513 for (i = 0; i<num_criteria; i++)
514 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
515 new_entry->sysno = sysno;
516 new_entry->score = -1;
519 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
520 zint sysno, int score, int relation)
522 struct zset_sort_entry *new_entry = NULL;
525 i = sort_info->num_entries;
530 rel = score - sort_info->entries[i]->score;
537 else if (relation == 'A')
544 j = sort_info->max_entries;
548 if (sort_info->num_entries == j)
551 j = (sort_info->num_entries)++;
553 new_entry = sort_info->entries[j];
556 sort_info->entries[j] = sort_info->entries[j-1];
559 sort_info->entries[i] = new_entry;
561 new_entry->sysno = sysno;
562 new_entry->score = score;
565 void resultSetSort (ZebraHandle zh, NMEM nmem,
566 int num_input_setnames, const char **input_setnames,
567 const char *output_setname,
568 Z_SortKeySpecList *sort_sequence, int *sort_status)
573 if (num_input_setnames == 0)
578 if (num_input_setnames > 1)
583 yaz_log (LOG_DEBUG, "result set sort input=%s output=%s",
584 *input_setnames, output_setname);
585 sset = resultSetGet (zh, input_setnames[0]);
589 zh->errString = nmem_strdup (nmem, input_setnames[0]);
592 if (!(rset = sset->rset))
595 zh->errString = nmem_strdup (nmem, input_setnames[0]);
598 if (strcmp (output_setname, input_setnames[0]))
600 rset = rset_dup (rset);
601 sset = resultSetAdd (zh, output_setname, 1);
604 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
607 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
608 ZebraSet sset, RSET rset,
609 Z_SortKeySpecList *sort_sequence, int *sort_status)
614 struct sortKeyInfo sort_criteria[3];
619 yaz_log (LOG_LOG, "resultSetSortSingle start");
620 sset->sort_info->num_entries = 0;
623 num_criteria = sort_sequence->num_specs;
624 if (num_criteria > 3)
626 for (i = 0; i < num_criteria; i++)
628 Z_SortKeySpec *sks = sort_sequence->specs[i];
631 if (*sks->sortRelation == Z_SortKeySpec_ascending)
632 sort_criteria[i].relation = 'A';
633 else if (*sks->sortRelation == Z_SortKeySpec_descending)
634 sort_criteria[i].relation = 'D';
640 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
645 else if (sks->sortElement->which != Z_SortElement_generic)
650 sk = sks->sortElement->u.generic;
653 case Z_SortKey_sortField:
654 yaz_log (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
657 case Z_SortKey_elementSpec:
658 yaz_log (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
661 case Z_SortKey_sortAttributes:
662 yaz_log (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
663 sort_criteria[i].attrUse =
664 zebra_maps_sort (zh->reg->zebra_maps,
665 sk->u.sortAttributes,
666 &sort_criteria[i].numerical);
667 yaz_log (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
668 if (sort_criteria[i].attrUse == -1)
673 if (sortIdx_type (zh->reg->sortIdx, sort_criteria[i].attrUse))
681 rfd = rset_open (rset, RSETF_READ);
682 while (rset_read (rset, rfd, &key, &term_index))
685 zint this_sys = key.mem[0];
687 zint this_sys = key.sysno;
689 if (this_sys != psysno)
693 resultSetInsertSort (zh, sset,
694 sort_criteria, num_criteria, psysno);
697 rset_close (rset, rfd);
699 for (i = 0; i < rset->no_rset_terms; i++)
700 yaz_log (LOG_LOG, "term=\"%s\" nn=" ZINT_FORMAT " type=%s count=" ZINT_FORMAT,
701 rset->rset_terms[i]->name,
702 rset->rset_terms[i]->nn,
703 rset->rset_terms[i]->flags,
704 rset->rset_terms[i]->count);
706 *sort_status = Z_SortResponse_success;
707 yaz_log (LOG_LOG, "resultSetSortSingle end");
710 RSET resultSetRef (ZebraHandle zh, const char *resultSetId)
714 if ((s = resultSetGet (zh, resultSetId)))
719 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
725 ZebraRankClass rank_class;
726 struct rank_control *rc;
727 struct zset_sort_info *sort_info;
728 const char *rank_handler_name = res_get_def(zh->res, "rank", "rank-1");
730 zint est=-2; /* -2 not done, -1 can't do, >0 actual estimate*/
733 sort_info = zebraSet->sort_info;
734 sort_info->num_entries = 0;
736 rfd = rset_open (rset, RSETF_READ);
738 yaz_log (LOG_LOG, "resultSetRank");
740 rank_class = zebraRankLookup (zh, rank_handler_name);
743 yaz_log (LOG_WARN, "No such rank handler: %s", rank_handler_name);
746 rc = rank_class->control;
748 if (rset_read (rset, rfd, &key, &term_index))
751 zint psysno = key.mem[0];
753 zint psysno = key.sysno;
757 (*rc->begin) (zh->reg, rank_class->class_handle, rset);
759 esthits=atoi(res_get_def(zh->res,"estimatehits","0"));
760 if (!esthits) est=-1; /* can not do */
764 zint this_sys = key.mem[0];
766 zint this_sys = key.sysno;
769 if (this_sys != psysno)
771 score = (*rc->calc) (handle, psysno);
773 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
777 (*rc->add) (handle, this_sys, term_index);
778 if ( (est==-2) && (zebraSet->hits==esthits))
779 { /* time to estimate the hits */
781 rset_pos(rset,rfd,&cur,&tot);
784 est=(zint)(0.5+zebraSet->hits/f);
785 /* FIXME - round the guess to 3 digits */
786 logf(LOG_LOG, "Estimating hits (%s) "
788 "; %0.1f->"ZINT_FORMAT,
796 while (rset_read (rset, rfd, &key, &term_index) && (est<0) );
798 score = (*rc->calc) (handle, psysno);
799 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
800 (*rc->end) (zh->reg, handle);
802 rset_close (rset, rfd);
804 for (i = 0; i < rset->no_rset_terms; i++)
805 yaz_log (LOG_LOG, "term=\"%s\" nn=" ZINT_FORMAT " type=%s count=" ZINT_FORMAT,
806 rset->rset_terms[i]->name,
807 rset->rset_terms[i]->nn,
808 rset->rset_terms[i]->flags,
809 rset->rset_terms[i]->count);
811 yaz_log (LOG_LOG, ZINT_FORMAT " keys, %d distinct sysnos",
812 kno, zebraSet->hits);
815 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
817 ZebraRankClass p = zh->reg->rank_classes;
818 while (p && strcmp (p->control->name, name))
820 if (p && !p->init_flag)
822 if (p->control->create)
823 p->class_handle = (*p->control->create)(zh);
829 void zebraRankInstall (struct zebra_register *reg, struct rank_control *ctrl)
831 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
832 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
833 memcpy (p->control, ctrl, sizeof(*p->control));
834 p->control->name = xstrdup (ctrl->name);
836 p->next = reg->rank_classes;
837 reg->rank_classes = p;
840 void zebraRankDestroy (struct zebra_register *reg)
842 ZebraRankClass p = reg->rank_classes;
845 ZebraRankClass p_next = p->next;
846 if (p->init_flag && p->control->destroy)
847 (*p->control->destroy)(reg, p->class_handle);
848 xfree (p->control->name);
853 reg->rank_classes = NULL;