2 * Copyright (C) 1994-2000, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.28 2000-07-07 12:49:20 adam
8 * Optimized resultSetInsert{Rank,Sort}.
10 * Revision 1.27 2000/04/05 09:49:36 adam
11 * On Unix, zebra/z'mbol uses automake.
13 * Revision 1.26 2000/03/20 19:08:36 adam
14 * Added remote record import using Z39.50 extended services and Segment
17 * Revision 1.25 2000/03/15 15:00:31 adam
18 * First work on threaded version.
20 * Revision 1.24 1999/11/04 15:00:45 adam
21 * Implemented delete result set(s).
23 * Revision 1.23 1999/05/26 07:49:13 adam
26 * Revision 1.22 1999/02/02 14:51:15 adam
27 * Updated WIN32 code specific sections. Changed header.
29 * Revision 1.21 1998/11/16 16:03:46 adam
30 * Moved loggin utilities to Yaz. Was implemented in file zlogs.c.
32 * Revision 1.20 1998/11/16 10:10:53 adam
33 * Fixed problem with zebraPosSetCreate that occurred when positions were
36 * Revision 1.19 1998/09/22 10:48:22 adam
37 * Minor changes in search API.
39 * Revision 1.18 1998/09/22 10:03:45 adam
40 * Changed result sets to be persistent in the sense that they can
41 * be re-searched if needed.
42 * Fixed memory leak in rsm_or.
44 * Revision 1.17 1998/06/23 15:33:36 adam
45 * Added feature to specify sort criteria in query (type 7 specifies
48 * Revision 1.16 1998/05/20 10:12:24 adam
49 * Implemented automatic EXPLAIN database maintenance.
50 * Modified Zebra to work with ASN.1 compiled version of YAZ.
52 * Revision 1.15 1998/03/05 08:45:14 adam
53 * New result set model and modular ranking system. Moved towards
54 * descent server API. System information stored as "SGML" records.
56 * Revision 1.14 1998/02/10 16:39:15 adam
59 * Revision 1.13 1998/02/10 12:03:06 adam
62 * Revision 1.12 1997/09/25 14:57:36 adam
65 * Revision 1.11 1996/12/23 15:30:46 adam
67 * Bug fix: result sets weren't deleted after server shut down.
69 * Revision 1.10 1995/10/30 15:08:08 adam
72 * Revision 1.9 1995/10/17 18:02:14 adam
73 * New feature: databases. Implemented as prefix to words in dictionary.
75 * Revision 1.8 1995/10/10 13:59:25 adam
76 * Function rset_open changed its wflag parameter to general flags.
78 * Revision 1.7 1995/10/06 14:38:01 adam
79 * New result set method: r_score.
80 * Local no (sysno) and score is transferred to retrieveCtrl.
82 * Revision 1.6 1995/09/28 09:19:49 adam
83 * xfree/xmalloc used everywhere.
84 * Extract/retrieve method seems to work for text records.
86 * Revision 1.5 1995/09/27 16:17:32 adam
87 * More work on retrieve.
89 * Revision 1.4 1995/09/07 13:58:36 adam
90 * New parameter: result-set file descriptor (RSFD) to support multiple
91 * positions within the same result-set.
92 * Boolean operators: and, or, not implemented.
93 * Result-set references.
95 * Revision 1.3 1995/09/06 16:11:19 adam
96 * Option: only one word key per file.
98 * Revision 1.2 1995/09/06 10:33:04 adam
99 * More work on present. Some log messages removed.
101 * Revision 1.1 1995/09/05 15:28:40 adam
102 * More work on search engine.
116 #define SORT_IDX_ENTRYSIZE 64
117 #define ZSET_SORT_MAX_LEVEL 3
127 struct zset_sort_info *sort_info;
128 struct zebra_set *next;
131 struct zset_sort_entry {
134 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
137 struct zset_sort_info {
140 struct zset_sort_entry *all_entries;
141 struct zset_sort_entry **entries;
144 ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
145 Z_RPNQuery *rpn, int num_bases, char **basenames,
151 zh->errString = NULL;
154 zebraSet = resultSetAdd (zh, setname, 1);
158 zebraSet->num_bases = num_bases;
159 zebraSet->basenames = basenames;
160 zebraSet->nmem = odr_extract_mem (input);
162 zebraSet->rset = rpn_search (zh, output->mem, rpn,
164 zebraSet->basenames, zebraSet->name,
166 zh->hits = zebraSet->hits;
172 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
177 for (s = zh->sets; s; s = s->next)
178 if (!strcmp (s->name, name))
182 logf (LOG_DEBUG, "updating result set %s", name);
186 rset_delete (s->rset);
188 nmem_destroy (s->nmem);
192 logf (LOG_DEBUG, "adding result set %s", name);
193 s = (ZebraSet) xmalloc (sizeof(*s));
196 s->name = (char *) xmalloc (strlen(name)+1);
197 strcpy (s->name, name);
199 s->sort_info = (struct zset_sort_info *)
200 xmalloc (sizeof(*s->sort_info));
201 s->sort_info->max_entries = 1000;
202 s->sort_info->entries = (struct zset_sort_entry **)
203 xmalloc (sizeof(*s->sort_info->entries) *
204 s->sort_info->max_entries);
205 s->sort_info->all_entries = (struct zset_sort_entry *)
206 xmalloc (sizeof(*s->sort_info->all_entries) *
207 s->sort_info->max_entries);
208 for (i = 0; i < s->sort_info->max_entries; i++)
209 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->rset && s->rpn)
225 NMEM nmem = nmem_create ();
227 rpn_search (zh, nmem, s->rpn, s->num_bases,
228 s->basenames, s->name, s);
237 void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
239 ZebraSet *ss = &zh->sets;
243 for (i = 0; i<num; i++)
244 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
246 zh->errString = NULL;
253 for (i = 0; i<num; i++)
254 if (!strcmp (s->name, names[i]))
257 statuses[i] = Z_DeleteStatus_success;
266 xfree (s->sort_info->all_entries);
267 xfree (s->sort_info->entries);
268 xfree (s->sort_info);
271 nmem_destroy (s->nmem);
272 rset_delete (s->rset);
281 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
282 int num, int *positions)
288 struct zset_sort_info *sort_info;
290 if (!(sset = resultSetGet (zh, name)))
292 if (!(rset = sset->rset))
294 sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
295 for (i = 0; i<num; i++)
300 sort_info = sset->sort_info;
305 for (i = 0; i<num; i++)
307 position = positions[i];
308 if (position > 0 && position <= sort_info->num_entries)
310 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
311 sr[i].sysno = sort_info->entries[position-1]->sysno;
312 sr[i].score = sort_info->entries[position-1]->score;
316 /* did we really get all entries using sort ? */
317 for (i = 0; i<num; i++)
322 if (i < num) /* nope, get the rest, unsorted - sorry */
332 position = sort_info->num_entries;
333 while (num_i < num && positions[num_i] < position)
335 rfd = rset_open (rset, RSETF_READ);
336 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
338 if (key.sysno != psysno)
343 /* determine we alreay have this in our set */
344 for (i = sort_info->num_entries; --i >= 0; )
345 if (psysno == sort_info->entries[i]->sysno)
351 assert (num_i < num);
352 if (position == positions[num_i])
354 sr[num_i].sysno = psysno;
355 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
356 sr[num_i].score = -1;
361 rset_close (rset, rfd);
366 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
376 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
377 struct sortKeyInfo *criteria, int num_criteria,
380 struct zset_sort_entry this_entry;
381 struct zset_sort_entry *new_entry = NULL;
382 struct zset_sort_info *sort_info = sset->sort_info;
385 sortIdx_sysno (zh->service->sortIdx, sysno);
386 for (i = 0; i<num_criteria; i++)
388 sortIdx_type (zh->service->sortIdx, criteria[i].attrUse);
389 sortIdx_read (zh->service->sortIdx, this_entry.buf[i]);
391 i = sort_info->num_entries;
395 for (j = 0; j<num_criteria; j++)
397 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
404 if (criteria[j].relation == 'A')
409 else if (criteria[j].relation == 'D')
416 j = sort_info->max_entries;
420 if (sort_info->num_entries == j)
423 j = (sort_info->num_entries)++;
424 new_entry = sort_info->entries[j];
427 sort_info->entries[j] = sort_info->entries[j-1];
430 sort_info->entries[i] = new_entry;
432 for (i = 0; i<num_criteria; i++)
433 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
434 new_entry->sysno = sysno;
435 new_entry->score = -1;
438 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
439 int sysno, int score, int relation)
441 struct zset_sort_entry *new_entry = NULL;
444 i = sort_info->num_entries;
449 rel = score - sort_info->entries[i]->score;
456 else if (relation == 'A')
463 j = sort_info->max_entries;
467 if (sort_info->num_entries == j)
470 j = (sort_info->num_entries)++;
472 new_entry = sort_info->entries[j];
475 sort_info->entries[j] = sort_info->entries[j-1];
478 sort_info->entries[i] = new_entry;
480 new_entry->sysno = sysno;
481 new_entry->score = score;
484 void resultSetSort (ZebraHandle zh, NMEM nmem,
485 int num_input_setnames, const char **input_setnames,
486 const char *output_setname,
487 Z_SortKeySpecList *sort_sequence, int *sort_status)
492 if (num_input_setnames == 0)
497 if (num_input_setnames > 1)
502 logf (LOG_DEBUG, "result set sort input=%s output=%s",
503 *input_setnames, output_setname);
504 sset = resultSetGet (zh, input_setnames[0]);
508 zh->errString = nmem_strdup (nmem, input_setnames[0]);
511 if (!(rset = sset->rset))
514 zh->errString = nmem_strdup (nmem, input_setnames[0]);
517 if (strcmp (output_setname, input_setnames[0]))
519 rset = rset_dup (rset);
520 sset = resultSetAdd (zh, output_setname, 1);
523 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
526 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
527 ZebraSet sset, RSET rset,
528 Z_SortKeySpecList *sort_sequence, int *sort_status)
532 struct sortKeyInfo sort_criteria[3];
537 logf (LOG_DEBUG, "resultSetSortSingle start");
538 sset->sort_info->num_entries = 0;
541 num_criteria = sort_sequence->num_specs;
542 if (num_criteria > 3)
544 for (i = 0; i < num_criteria; i++)
546 Z_SortKeySpec *sks = sort_sequence->specs[i];
549 if (*sks->sortRelation == Z_SortRelation_ascending)
550 sort_criteria[i].relation = 'A';
551 else if (*sks->sortRelation == Z_SortRelation_descending)
552 sort_criteria[i].relation = 'D';
558 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
563 else if (sks->sortElement->which != Z_SortElement_generic)
568 sk = sks->sortElement->u.generic;
571 case Z_SortKey_sortField:
572 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
575 case Z_SortKey_elementSpec:
576 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
579 case Z_SortKey_sortAttributes:
580 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
581 sort_criteria[i].attrUse =
582 zebra_maps_sort (zh->service->zebra_maps,
583 sk->u.sortAttributes);
584 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
585 if (sort_criteria[i].attrUse == -1)
590 if (sortIdx_type (zh->service->sortIdx, sort_criteria[i].attrUse))
598 rfd = rset_open (rset, RSETF_READ);
599 while (rset_read (rset, rfd, &key, &term_index))
601 if (key.sysno != psysno)
605 resultSetInsertSort (zh, sset,
606 sort_criteria, num_criteria, psysno);
609 rset_close (rset, rfd);
611 *sort_status = Z_SortStatus_success;
612 logf (LOG_DEBUG, "resultSetSortSingle end");
615 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
619 if ((s = resultSetGet (zh, resultSetId)))
624 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
630 ZebraRankClass rank_class;
631 struct rank_control *rc;
632 struct zset_sort_info *sort_info;
634 sort_info = zebraSet->sort_info;
635 sort_info->num_entries = 0;
637 rfd = rset_open (rset, RSETF_READ);
639 logf (LOG_DEBUG, "resultSetRank");
640 for (i = 0; i < rset->no_rset_terms; i++)
641 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
642 rset->rset_terms[i]->name,
643 rset->rset_terms[i]->nn,
644 rset->rset_terms[i]->flags);
646 rank_class = zebraRankLookup (zh, "rank-1");
647 rc = rank_class->control;
649 if (rset_read (rset, rfd, &key, &term_index))
651 int psysno = key.sysno;
654 (*rc->begin) (zh, rank_class->class_handle, rset);
659 if (key.sysno != psysno)
661 score = (*rc->calc) (handle, psysno);
663 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
667 (*rc->add) (handle, key.seqno, term_index);
669 while (rset_read (rset, rfd, &key, &term_index));
670 score = (*rc->calc) (handle, psysno);
671 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
672 (*rc->end) (zh, handle);
674 rset_close (rset, rfd);
675 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
678 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
680 ZebraRankClass p = zh->service->rank_classes;
681 while (p && strcmp (p->control->name, name))
683 if (p && !p->init_flag)
685 if (p->control->create)
686 p->class_handle = (*p->control->create)(zh->service);
692 void zebraRankInstall (ZebraService zh, struct rank_control *ctrl)
694 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
695 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
696 memcpy (p->control, ctrl, sizeof(*p->control));
697 p->control->name = xstrdup (ctrl->name);
699 p->next = zh->rank_classes;
700 zh->rank_classes = p;
703 void zebraRankDestroy (ZebraService zh)
705 ZebraRankClass p = zh->rank_classes;
708 ZebraRankClass p_next = p->next;
709 if (p->init_flag && p->control->destroy)
710 (*p->control->destroy)(zh, p->class_handle);
711 xfree (p->control->name);
716 zh->rank_classes = NULL;