2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.24 1999-11-04 15:00:45 adam
8 * Implemented delete result set(s).
10 * Revision 1.23 1999/05/26 07:49:13 adam
13 * Revision 1.22 1999/02/02 14:51:15 adam
14 * Updated WIN32 code specific sections. Changed header.
16 * Revision 1.21 1998/11/16 16:03:46 adam
17 * Moved loggin utilities to Yaz. Was implemented in file zlogs.c.
19 * Revision 1.20 1998/11/16 10:10:53 adam
20 * Fixed problem with zebraPosSetCreate that occurred when positions were
23 * Revision 1.19 1998/09/22 10:48:22 adam
24 * Minor changes in search API.
26 * Revision 1.18 1998/09/22 10:03:45 adam
27 * Changed result sets to be persistent in the sense that they can
28 * be re-searched if needed.
29 * Fixed memory leak in rsm_or.
31 * Revision 1.17 1998/06/23 15:33:36 adam
32 * Added feature to specify sort criteria in query (type 7 specifies
35 * Revision 1.16 1998/05/20 10:12:24 adam
36 * Implemented automatic EXPLAIN database maintenance.
37 * Modified Zebra to work with ASN.1 compiled version of YAZ.
39 * Revision 1.15 1998/03/05 08:45:14 adam
40 * New result set model and modular ranking system. Moved towards
41 * descent server API. System information stored as "SGML" records.
43 * Revision 1.14 1998/02/10 16:39:15 adam
46 * Revision 1.13 1998/02/10 12:03:06 adam
49 * Revision 1.12 1997/09/25 14:57:36 adam
52 * Revision 1.11 1996/12/23 15:30:46 adam
54 * Bug fix: result sets weren't deleted after server shut down.
56 * Revision 1.10 1995/10/30 15:08:08 adam
59 * Revision 1.9 1995/10/17 18:02:14 adam
60 * New feature: databases. Implemented as prefix to words in dictionary.
62 * Revision 1.8 1995/10/10 13:59:25 adam
63 * Function rset_open changed its wflag parameter to general flags.
65 * Revision 1.7 1995/10/06 14:38:01 adam
66 * New result set method: r_score.
67 * Local no (sysno) and score is transferred to retrieveCtrl.
69 * Revision 1.6 1995/09/28 09:19:49 adam
70 * xfree/xmalloc used everywhere.
71 * Extract/retrieve method seems to work for text records.
73 * Revision 1.5 1995/09/27 16:17:32 adam
74 * More work on retrieve.
76 * Revision 1.4 1995/09/07 13:58:36 adam
77 * New parameter: result-set file descriptor (RSFD) to support multiple
78 * positions within the same result-set.
79 * Boolean operators: and, or, not implemented.
80 * Result-set references.
82 * Revision 1.3 1995/09/06 16:11:19 adam
83 * Option: only one word key per file.
85 * Revision 1.2 1995/09/06 10:33:04 adam
86 * More work on present. Some log messages removed.
88 * Revision 1.1 1995/09/05 15:28:40 adam
89 * More work on search engine.
103 #define SORT_IDX_ENTRYSIZE 64
104 #define ZSET_SORT_MAX_LEVEL 3
114 struct zset_sort_info *sort_info;
115 struct zebra_set *next;
118 struct zset_sort_entry {
121 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
124 struct zset_sort_info {
127 struct zset_sort_entry *all_entries;
128 struct zset_sort_entry **entries;
131 ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
132 Z_RPNQuery *rpn, int num_bases, char **basenames,
138 zh->errString = NULL;
141 zebraSet = resultSetAdd (zh, setname, 1);
145 zebraSet->num_bases = num_bases;
146 zebraSet->basenames = basenames;
147 zebraSet->nmem = odr_extract_mem (input);
149 zebraSet->rset = rpn_search (zh, output->mem, rpn,
151 zebraSet->basenames, zebraSet->name,
153 zh->hits = zebraSet->hits;
159 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
164 for (s = zh->sets; s; s = s->next)
165 if (!strcmp (s->name, name))
169 logf (LOG_DEBUG, "updating result set %s", name);
173 rset_delete (s->rset);
175 nmem_destroy (s->nmem);
179 logf (LOG_DEBUG, "adding result set %s", name);
180 s = (ZebraSet) xmalloc (sizeof(*s));
183 s->name = (char *) xmalloc (strlen(name)+1);
184 strcpy (s->name, name);
186 s->sort_info = (struct zset_sort_info *)
187 xmalloc (sizeof(*s->sort_info));
188 s->sort_info->max_entries = 1000;
189 s->sort_info->entries = (struct zset_sort_entry **)
190 xmalloc (sizeof(*s->sort_info->entries) *
191 s->sort_info->max_entries);
192 s->sort_info->all_entries = (struct zset_sort_entry *)
193 xmalloc (sizeof(*s->sort_info->all_entries) *
194 s->sort_info->max_entries);
195 for (i = 0; i < s->sort_info->max_entries; i++)
196 s->sort_info->entries[i] = s->sort_info->all_entries + i;
203 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
207 for (s = zh->sets; s; s = s->next)
208 if (!strcmp (s->name, name))
210 if (!s->rset && s->rpn)
212 NMEM nmem = nmem_create ();
214 rpn_search (zh, nmem, s->rpn, s->num_bases,
215 s->basenames, s->name, s);
224 void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
226 ZebraSet *ss = &zh->sets;
230 for (i = 0; i<num; i++)
231 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
233 zh->errString = NULL;
240 for (i = 0; i<num; i++)
241 if (!strcmp (s->name, names[i]))
244 statuses[i] = Z_DeleteStatus_success;
253 xfree (s->sort_info->all_entries);
254 xfree (s->sort_info->entries);
255 xfree (s->sort_info);
258 nmem_destroy (s->nmem);
259 rset_delete (s->rset);
268 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
269 int num, int *positions)
275 struct zset_sort_info *sort_info;
277 if (!(sset = resultSetGet (zh, name)))
279 if (!(rset = sset->rset))
281 sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
282 for (i = 0; i<num; i++)
287 sort_info = sset->sort_info;
292 for (i = 0; i<num; i++)
294 position = positions[i];
295 if (position > 0 && position <= sort_info->num_entries)
297 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
298 sr[i].sysno = sort_info->entries[position-1]->sysno;
299 sr[i].score = sort_info->entries[position-1]->score;
303 /* did we really get all entries using sort ? */
304 for (i = 0; i<num; i++)
309 if (i < num) /* nope, get the rest, unsorted - sorry */
319 position = sort_info->num_entries;
320 while (num_i < num && positions[num_i] < position)
322 rfd = rset_open (rset, RSETF_READ);
323 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
325 if (key.sysno != psysno)
330 /* determine we alreay have this in our set */
331 for (i = sort_info->num_entries; --i >= 0; )
332 if (psysno == sort_info->entries[i]->sysno)
338 assert (num_i < num);
339 if (position == positions[num_i])
341 sr[num_i].sysno = psysno;
342 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
343 sr[num_i].score = -1;
348 rset_close (rset, rfd);
353 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
363 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
364 struct sortKey *criteria, int num_criteria,
367 struct zset_sort_entry this_entry;
368 struct zset_sort_entry *new_entry = NULL;
369 struct zset_sort_info *sort_info = sset->sort_info;
372 sortIdx_sysno (zh->sortIdx, sysno);
373 for (i = 0; i<num_criteria; i++)
375 sortIdx_type (zh->sortIdx, criteria[i].attrUse);
376 sortIdx_read (zh->sortIdx, this_entry.buf[i]);
378 i = sort_info->num_entries;
382 for (j = 0; j<num_criteria; j++)
384 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
391 if (criteria[j].relation == 'A')
396 else if (criteria[j].relation == 'D')
402 j = sort_info->max_entries-1;
406 new_entry = sort_info->entries[j];
409 sort_info->entries[j] = sort_info->entries[j-1];
412 sort_info->entries[j] = new_entry;
414 if (sort_info->num_entries != sort_info->max_entries)
415 (sort_info->num_entries)++;
416 for (i = 0; i<num_criteria; i++)
417 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
418 new_entry->sysno = sysno;
419 new_entry->score = -1;
422 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
423 int sysno, int score, int relation)
425 struct zset_sort_entry *new_entry = NULL;
428 i = sort_info->num_entries;
433 rel = score - sort_info->entries[i]->score;
440 else if (relation == 'A')
446 j = sort_info->max_entries-1;
450 new_entry = sort_info->entries[j];
453 sort_info->entries[j] = sort_info->entries[j-1];
456 sort_info->entries[j] = new_entry;
458 if (sort_info->num_entries != sort_info->max_entries)
459 (sort_info->num_entries)++;
460 new_entry->sysno = sysno;
461 new_entry->score = score;
464 void resultSetSort (ZebraHandle zh, NMEM nmem,
465 int num_input_setnames, const char **input_setnames,
466 const char *output_setname,
467 Z_SortKeySpecList *sort_sequence, int *sort_status)
472 if (num_input_setnames == 0)
477 if (num_input_setnames > 1)
482 logf (LOG_DEBUG, "result set sort input=%s output=%s",
483 *input_setnames, output_setname);
484 sset = resultSetGet (zh, input_setnames[0]);
488 zh->errString = nmem_strdup (nmem, input_setnames[0]);
491 if (!(rset = sset->rset))
494 zh->errString = nmem_strdup (nmem, input_setnames[0]);
497 if (strcmp (output_setname, input_setnames[0]))
499 rset = rset_dup (rset);
500 sset = resultSetAdd (zh, output_setname, 1);
503 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
506 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
507 ZebraSet sset, RSET rset,
508 Z_SortKeySpecList *sort_sequence, int *sort_status)
512 struct sortKey sort_criteria[3];
517 logf (LOG_DEBUG, "resultSetSortSingle start");
518 sset->sort_info->num_entries = 0;
521 num_criteria = sort_sequence->num_specs;
522 if (num_criteria > 3)
524 for (i = 0; i < num_criteria; i++)
526 Z_SortKeySpec *sks = sort_sequence->specs[i];
529 if (*sks->sortRelation == Z_SortRelation_ascending)
530 sort_criteria[i].relation = 'A';
531 else if (*sks->sortRelation == Z_SortRelation_descending)
532 sort_criteria[i].relation = 'D';
538 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
543 else if (sks->sortElement->which != Z_SortElement_generic)
548 sk = sks->sortElement->u.generic;
551 case Z_SortKey_sortField:
552 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
555 case Z_SortKey_elementSpec:
556 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
559 case Z_SortKey_sortAttributes:
560 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
561 sort_criteria[i].attrUse =
562 zebra_maps_sort (zh->zebra_maps, sk->u.sortAttributes);
563 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
564 if (sort_criteria[i].attrUse == -1)
569 if (sortIdx_type (zh->sortIdx, sort_criteria[i].attrUse))
577 rfd = rset_open (rset, RSETF_READ);
578 while (rset_read (rset, rfd, &key, &term_index))
580 if (key.sysno != psysno)
584 resultSetInsertSort (zh, sset,
585 sort_criteria, num_criteria, psysno);
588 rset_close (rset, rfd);
591 *sort_status = Z_SortStatus_success;
592 logf (LOG_DEBUG, "resultSetSortSingle end");
595 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
599 if ((s = resultSetGet (zh, resultSetId)))
604 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
610 ZebraRankClass rank_class;
611 struct rank_control *rc;
612 struct zset_sort_info *sort_info;
614 sort_info = zebraSet->sort_info;
615 sort_info->num_entries = 0;
617 rfd = rset_open (rset, RSETF_READ);
619 logf (LOG_DEBUG, "resultSetRank");
620 for (i = 0; i < rset->no_rset_terms; i++)
621 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
622 rset->rset_terms[i]->name,
623 rset->rset_terms[i]->nn,
624 rset->rset_terms[i]->flags);
626 rank_class = zebraRankLookup (zh, "rank-1");
627 rc = rank_class->control;
629 if (rset_read (rset, rfd, &key, &term_index))
631 int psysno = key.sysno;
634 (*rc->begin) (zh, rank_class->class_handle, rset);
639 if (key.sysno != psysno)
641 score = (*rc->calc) (handle, psysno);
643 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
647 (*rc->add) (handle, key.seqno, term_index);
649 while (rset_read (rset, rfd, &key, &term_index));
650 score = (*rc->calc) (handle, psysno);
651 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
652 (*rc->end) (zh, handle);
654 rset_close (rset, rfd);
655 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
658 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
660 ZebraRankClass p = zh->rank_classes;
661 while (p && strcmp (p->control->name, name))
663 if (p && !p->init_flag)
665 if (p->control->create)
666 p->class_handle = (*p->control->create)(zh);
672 void zebraRankInstall (ZebraHandle zh, struct rank_control *ctrl)
674 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
675 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
676 memcpy (p->control, ctrl, sizeof(*p->control));
677 p->control->name = xstrdup (ctrl->name);
679 p->next = zh->rank_classes;
680 zh->rank_classes = p;
683 void zebraRankDestroy (ZebraHandle zh)
685 ZebraRankClass p = zh->rank_classes;
688 ZebraRankClass p_next = p->next;
689 if (p->init_flag && p->control->destroy)
690 (*p->control->destroy)(zh, p->class_handle);
691 xfree (p->control->name);
696 zh->rank_classes = NULL;