2 * Copyright (C) 1994-2000, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.29 2001-01-22 10:42:56 adam
8 * Added numerical sort.
10 * Revision 1.28 2000/07/07 12:49:20 adam
11 * Optimized resultSetInsert{Rank,Sort}.
13 * Revision 1.27 2000/04/05 09:49:36 adam
14 * On Unix, zebra/z'mbol uses automake.
16 * Revision 1.26 2000/03/20 19:08:36 adam
17 * Added remote record import using Z39.50 extended services and Segment
20 * Revision 1.25 2000/03/15 15:00:31 adam
21 * First work on threaded version.
23 * Revision 1.24 1999/11/04 15:00:45 adam
24 * Implemented delete result set(s).
26 * Revision 1.23 1999/05/26 07:49:13 adam
29 * Revision 1.22 1999/02/02 14:51:15 adam
30 * Updated WIN32 code specific sections. Changed header.
32 * Revision 1.21 1998/11/16 16:03:46 adam
33 * Moved loggin utilities to Yaz. Was implemented in file zlogs.c.
35 * Revision 1.20 1998/11/16 10:10:53 adam
36 * Fixed problem with zebraPosSetCreate that occurred when positions were
39 * Revision 1.19 1998/09/22 10:48:22 adam
40 * Minor changes in search API.
42 * Revision 1.18 1998/09/22 10:03:45 adam
43 * Changed result sets to be persistent in the sense that they can
44 * be re-searched if needed.
45 * Fixed memory leak in rsm_or.
47 * Revision 1.17 1998/06/23 15:33:36 adam
48 * Added feature to specify sort criteria in query (type 7 specifies
51 * Revision 1.16 1998/05/20 10:12:24 adam
52 * Implemented automatic EXPLAIN database maintenance.
53 * Modified Zebra to work with ASN.1 compiled version of YAZ.
55 * Revision 1.15 1998/03/05 08:45:14 adam
56 * New result set model and modular ranking system. Moved towards
57 * descent server API. System information stored as "SGML" records.
59 * Revision 1.14 1998/02/10 16:39:15 adam
62 * Revision 1.13 1998/02/10 12:03:06 adam
65 * Revision 1.12 1997/09/25 14:57:36 adam
68 * Revision 1.11 1996/12/23 15:30:46 adam
70 * Bug fix: result sets weren't deleted after server shut down.
72 * Revision 1.10 1995/10/30 15:08:08 adam
75 * Revision 1.9 1995/10/17 18:02:14 adam
76 * New feature: databases. Implemented as prefix to words in dictionary.
78 * Revision 1.8 1995/10/10 13:59:25 adam
79 * Function rset_open changed its wflag parameter to general flags.
81 * Revision 1.7 1995/10/06 14:38:01 adam
82 * New result set method: r_score.
83 * Local no (sysno) and score is transferred to retrieveCtrl.
85 * Revision 1.6 1995/09/28 09:19:49 adam
86 * xfree/xmalloc used everywhere.
87 * Extract/retrieve method seems to work for text records.
89 * Revision 1.5 1995/09/27 16:17:32 adam
90 * More work on retrieve.
92 * Revision 1.4 1995/09/07 13:58:36 adam
93 * New parameter: result-set file descriptor (RSFD) to support multiple
94 * positions within the same result-set.
95 * Boolean operators: and, or, not implemented.
96 * Result-set references.
98 * Revision 1.3 1995/09/06 16:11:19 adam
99 * Option: only one word key per file.
101 * Revision 1.2 1995/09/06 10:33:04 adam
102 * More work on present. Some log messages removed.
104 * Revision 1.1 1995/09/05 15:28:40 adam
105 * More work on search engine.
119 #define SORT_IDX_ENTRYSIZE 64
120 #define ZSET_SORT_MAX_LEVEL 3
130 struct zset_sort_info *sort_info;
131 struct zebra_set *next;
134 struct zset_sort_entry {
137 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
140 struct zset_sort_info {
143 struct zset_sort_entry *all_entries;
144 struct zset_sort_entry **entries;
147 ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
148 Z_RPNQuery *rpn, int num_bases, char **basenames,
154 zh->errString = NULL;
157 zebraSet = resultSetAdd (zh, setname, 1);
161 zebraSet->num_bases = num_bases;
162 zebraSet->basenames = basenames;
163 zebraSet->nmem = odr_extract_mem (input);
165 zebraSet->rset = rpn_search (zh, output->mem, rpn,
167 zebraSet->basenames, zebraSet->name,
169 zh->hits = zebraSet->hits;
175 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
180 for (s = zh->sets; s; s = s->next)
181 if (!strcmp (s->name, name))
185 logf (LOG_DEBUG, "updating result set %s", name);
189 rset_delete (s->rset);
191 nmem_destroy (s->nmem);
195 logf (LOG_DEBUG, "adding result set %s", name);
196 s = (ZebraSet) xmalloc (sizeof(*s));
199 s->name = (char *) xmalloc (strlen(name)+1);
200 strcpy (s->name, name);
202 s->sort_info = (struct zset_sort_info *)
203 xmalloc (sizeof(*s->sort_info));
204 s->sort_info->max_entries = 1000;
205 s->sort_info->entries = (struct zset_sort_entry **)
206 xmalloc (sizeof(*s->sort_info->entries) *
207 s->sort_info->max_entries);
208 s->sort_info->all_entries = (struct zset_sort_entry *)
209 xmalloc (sizeof(*s->sort_info->all_entries) *
210 s->sort_info->max_entries);
211 for (i = 0; i < s->sort_info->max_entries; i++)
212 s->sort_info->entries[i] = s->sort_info->all_entries + i;
219 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
223 for (s = zh->sets; s; s = s->next)
224 if (!strcmp (s->name, name))
226 if (!s->rset && s->rpn)
228 NMEM nmem = nmem_create ();
230 rpn_search (zh, nmem, s->rpn, s->num_bases,
231 s->basenames, s->name, s);
240 void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
242 ZebraSet *ss = &zh->sets;
246 for (i = 0; i<num; i++)
247 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
249 zh->errString = NULL;
256 for (i = 0; i<num; i++)
257 if (!strcmp (s->name, names[i]))
260 statuses[i] = Z_DeleteStatus_success;
269 xfree (s->sort_info->all_entries);
270 xfree (s->sort_info->entries);
271 xfree (s->sort_info);
274 nmem_destroy (s->nmem);
275 rset_delete (s->rset);
284 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
285 int num, int *positions)
291 struct zset_sort_info *sort_info;
293 if (!(sset = resultSetGet (zh, name)))
295 if (!(rset = sset->rset))
297 sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
298 for (i = 0; i<num; i++)
303 sort_info = sset->sort_info;
308 for (i = 0; i<num; i++)
310 position = positions[i];
311 if (position > 0 && position <= sort_info->num_entries)
313 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
314 sr[i].sysno = sort_info->entries[position-1]->sysno;
315 sr[i].score = sort_info->entries[position-1]->score;
319 /* did we really get all entries using sort ? */
320 for (i = 0; i<num; i++)
325 if (i < num) /* nope, get the rest, unsorted - sorry */
335 position = sort_info->num_entries;
336 while (num_i < num && positions[num_i] < position)
338 rfd = rset_open (rset, RSETF_READ);
339 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
341 if (key.sysno != psysno)
346 /* determine we alreay have this in our set */
347 for (i = sort_info->num_entries; --i >= 0; )
348 if (psysno == sort_info->entries[i]->sysno)
354 assert (num_i < num);
355 if (position == positions[num_i])
357 sr[num_i].sysno = psysno;
358 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
359 sr[num_i].score = -1;
364 rset_close (rset, rfd);
369 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
380 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
381 struct sortKeyInfo *criteria, int num_criteria,
384 struct zset_sort_entry this_entry;
385 struct zset_sort_entry *new_entry = NULL;
386 struct zset_sort_info *sort_info = sset->sort_info;
389 sortIdx_sysno (zh->service->sortIdx, sysno);
390 for (i = 0; i<num_criteria; i++)
392 sortIdx_type (zh->service->sortIdx, criteria[i].attrUse);
393 sortIdx_read (zh->service->sortIdx, this_entry.buf[i]);
395 i = sort_info->num_entries;
399 for (j = 0; j<num_criteria; j++)
401 if (criteria[j].numerical)
403 double diff = atof(this_entry.buf[j]) -
404 atof(sort_info->entries[i]->buf[j]);
413 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
421 if (criteria[j].relation == 'A')
426 else if (criteria[j].relation == 'D')
433 j = sort_info->max_entries;
437 if (sort_info->num_entries == j)
440 j = (sort_info->num_entries)++;
441 new_entry = sort_info->entries[j];
444 sort_info->entries[j] = sort_info->entries[j-1];
447 sort_info->entries[i] = new_entry;
449 for (i = 0; i<num_criteria; i++)
450 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
451 new_entry->sysno = sysno;
452 new_entry->score = -1;
455 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
456 int sysno, int score, int relation)
458 struct zset_sort_entry *new_entry = NULL;
461 i = sort_info->num_entries;
466 rel = score - sort_info->entries[i]->score;
473 else if (relation == 'A')
480 j = sort_info->max_entries;
484 if (sort_info->num_entries == j)
487 j = (sort_info->num_entries)++;
489 new_entry = sort_info->entries[j];
492 sort_info->entries[j] = sort_info->entries[j-1];
495 sort_info->entries[i] = new_entry;
497 new_entry->sysno = sysno;
498 new_entry->score = score;
501 void resultSetSort (ZebraHandle zh, NMEM nmem,
502 int num_input_setnames, const char **input_setnames,
503 const char *output_setname,
504 Z_SortKeySpecList *sort_sequence, int *sort_status)
509 if (num_input_setnames == 0)
514 if (num_input_setnames > 1)
519 logf (LOG_DEBUG, "result set sort input=%s output=%s",
520 *input_setnames, output_setname);
521 sset = resultSetGet (zh, input_setnames[0]);
525 zh->errString = nmem_strdup (nmem, input_setnames[0]);
528 if (!(rset = sset->rset))
531 zh->errString = nmem_strdup (nmem, input_setnames[0]);
534 if (strcmp (output_setname, input_setnames[0]))
536 rset = rset_dup (rset);
537 sset = resultSetAdd (zh, output_setname, 1);
540 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
543 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
544 ZebraSet sset, RSET rset,
545 Z_SortKeySpecList *sort_sequence, int *sort_status)
549 struct sortKeyInfo sort_criteria[3];
554 logf (LOG_DEBUG, "resultSetSortSingle start");
555 sset->sort_info->num_entries = 0;
558 num_criteria = sort_sequence->num_specs;
559 if (num_criteria > 3)
561 for (i = 0; i < num_criteria; i++)
563 Z_SortKeySpec *sks = sort_sequence->specs[i];
566 if (*sks->sortRelation == Z_SortRelation_ascending)
567 sort_criteria[i].relation = 'A';
568 else if (*sks->sortRelation == Z_SortRelation_descending)
569 sort_criteria[i].relation = 'D';
575 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
580 else if (sks->sortElement->which != Z_SortElement_generic)
585 sk = sks->sortElement->u.generic;
588 case Z_SortKey_sortField:
589 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
592 case Z_SortKey_elementSpec:
593 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
596 case Z_SortKey_sortAttributes:
597 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
598 sort_criteria[i].attrUse =
599 zebra_maps_sort (zh->service->zebra_maps,
600 sk->u.sortAttributes,
601 &sort_criteria[i].numerical);
602 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
603 if (sort_criteria[i].attrUse == -1)
608 if (sortIdx_type (zh->service->sortIdx, sort_criteria[i].attrUse))
616 rfd = rset_open (rset, RSETF_READ);
617 while (rset_read (rset, rfd, &key, &term_index))
619 if (key.sysno != psysno)
623 resultSetInsertSort (zh, sset,
624 sort_criteria, num_criteria, psysno);
627 rset_close (rset, rfd);
629 *sort_status = Z_SortStatus_success;
630 logf (LOG_DEBUG, "resultSetSortSingle end");
633 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
637 if ((s = resultSetGet (zh, resultSetId)))
642 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
648 ZebraRankClass rank_class;
649 struct rank_control *rc;
650 struct zset_sort_info *sort_info;
652 sort_info = zebraSet->sort_info;
653 sort_info->num_entries = 0;
655 rfd = rset_open (rset, RSETF_READ);
657 logf (LOG_DEBUG, "resultSetRank");
658 for (i = 0; i < rset->no_rset_terms; i++)
659 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
660 rset->rset_terms[i]->name,
661 rset->rset_terms[i]->nn,
662 rset->rset_terms[i]->flags);
664 rank_class = zebraRankLookup (zh, "rank-1");
665 rc = rank_class->control;
667 if (rset_read (rset, rfd, &key, &term_index))
669 int psysno = key.sysno;
672 (*rc->begin) (zh, rank_class->class_handle, rset);
677 if (key.sysno != psysno)
679 score = (*rc->calc) (handle, psysno);
681 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
685 (*rc->add) (handle, key.seqno, term_index);
687 while (rset_read (rset, rfd, &key, &term_index));
688 score = (*rc->calc) (handle, psysno);
689 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
690 (*rc->end) (zh, handle);
692 rset_close (rset, rfd);
693 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
696 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
698 ZebraRankClass p = zh->service->rank_classes;
699 while (p && strcmp (p->control->name, name))
701 if (p && !p->init_flag)
703 if (p->control->create)
704 p->class_handle = (*p->control->create)(zh->service);
710 void zebraRankInstall (ZebraService zh, struct rank_control *ctrl)
712 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
713 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
714 memcpy (p->control, ctrl, sizeof(*p->control));
715 p->control->name = xstrdup (ctrl->name);
717 p->next = zh->rank_classes;
718 zh->rank_classes = p;
721 void zebraRankDestroy (ZebraService zh)
723 ZebraRankClass p = zh->rank_classes;
726 ZebraRankClass p_next = p->next;
727 if (p->init_flag && p->control->destroy)
728 (*p->control->destroy)(zh, p->class_handle);
729 xfree (p->control->name);
734 zh->rank_classes = NULL;