2 * Copyright (C) 1994-2000, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.27 2000-04-05 09:49:36 adam
8 * On Unix, zebra/z'mbol uses automake.
10 * Revision 1.26 2000/03/20 19:08:36 adam
11 * Added remote record import using Z39.50 extended services and Segment
14 * Revision 1.25 2000/03/15 15:00:31 adam
15 * First work on threaded version.
17 * Revision 1.24 1999/11/04 15:00:45 adam
18 * Implemented delete result set(s).
20 * Revision 1.23 1999/05/26 07:49:13 adam
23 * Revision 1.22 1999/02/02 14:51:15 adam
24 * Updated WIN32 code specific sections. Changed header.
26 * Revision 1.21 1998/11/16 16:03:46 adam
27 * Moved loggin utilities to Yaz. Was implemented in file zlogs.c.
29 * Revision 1.20 1998/11/16 10:10:53 adam
30 * Fixed problem with zebraPosSetCreate that occurred when positions were
33 * Revision 1.19 1998/09/22 10:48:22 adam
34 * Minor changes in search API.
36 * Revision 1.18 1998/09/22 10:03:45 adam
37 * Changed result sets to be persistent in the sense that they can
38 * be re-searched if needed.
39 * Fixed memory leak in rsm_or.
41 * Revision 1.17 1998/06/23 15:33:36 adam
42 * Added feature to specify sort criteria in query (type 7 specifies
45 * Revision 1.16 1998/05/20 10:12:24 adam
46 * Implemented automatic EXPLAIN database maintenance.
47 * Modified Zebra to work with ASN.1 compiled version of YAZ.
49 * Revision 1.15 1998/03/05 08:45:14 adam
50 * New result set model and modular ranking system. Moved towards
51 * descent server API. System information stored as "SGML" records.
53 * Revision 1.14 1998/02/10 16:39:15 adam
56 * Revision 1.13 1998/02/10 12:03:06 adam
59 * Revision 1.12 1997/09/25 14:57:36 adam
62 * Revision 1.11 1996/12/23 15:30:46 adam
64 * Bug fix: result sets weren't deleted after server shut down.
66 * Revision 1.10 1995/10/30 15:08:08 adam
69 * Revision 1.9 1995/10/17 18:02:14 adam
70 * New feature: databases. Implemented as prefix to words in dictionary.
72 * Revision 1.8 1995/10/10 13:59:25 adam
73 * Function rset_open changed its wflag parameter to general flags.
75 * Revision 1.7 1995/10/06 14:38:01 adam
76 * New result set method: r_score.
77 * Local no (sysno) and score is transferred to retrieveCtrl.
79 * Revision 1.6 1995/09/28 09:19:49 adam
80 * xfree/xmalloc used everywhere.
81 * Extract/retrieve method seems to work for text records.
83 * Revision 1.5 1995/09/27 16:17:32 adam
84 * More work on retrieve.
86 * Revision 1.4 1995/09/07 13:58:36 adam
87 * New parameter: result-set file descriptor (RSFD) to support multiple
88 * positions within the same result-set.
89 * Boolean operators: and, or, not implemented.
90 * Result-set references.
92 * Revision 1.3 1995/09/06 16:11:19 adam
93 * Option: only one word key per file.
95 * Revision 1.2 1995/09/06 10:33:04 adam
96 * More work on present. Some log messages removed.
98 * Revision 1.1 1995/09/05 15:28:40 adam
99 * More work on search engine.
113 #define SORT_IDX_ENTRYSIZE 64
114 #define ZSET_SORT_MAX_LEVEL 3
124 struct zset_sort_info *sort_info;
125 struct zebra_set *next;
128 struct zset_sort_entry {
131 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
134 struct zset_sort_info {
137 struct zset_sort_entry *all_entries;
138 struct zset_sort_entry **entries;
141 ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
142 Z_RPNQuery *rpn, int num_bases, char **basenames,
148 zh->errString = NULL;
151 zebraSet = resultSetAdd (zh, setname, 1);
155 zebraSet->num_bases = num_bases;
156 zebraSet->basenames = basenames;
157 zebraSet->nmem = odr_extract_mem (input);
159 zebraSet->rset = rpn_search (zh, output->mem, rpn,
161 zebraSet->basenames, zebraSet->name,
163 zh->hits = zebraSet->hits;
169 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
174 for (s = zh->sets; s; s = s->next)
175 if (!strcmp (s->name, name))
179 logf (LOG_DEBUG, "updating result set %s", name);
183 rset_delete (s->rset);
185 nmem_destroy (s->nmem);
189 logf (LOG_DEBUG, "adding result set %s", name);
190 s = (ZebraSet) xmalloc (sizeof(*s));
193 s->name = (char *) xmalloc (strlen(name)+1);
194 strcpy (s->name, name);
196 s->sort_info = (struct zset_sort_info *)
197 xmalloc (sizeof(*s->sort_info));
198 s->sort_info->max_entries = 1000;
199 s->sort_info->entries = (struct zset_sort_entry **)
200 xmalloc (sizeof(*s->sort_info->entries) *
201 s->sort_info->max_entries);
202 s->sort_info->all_entries = (struct zset_sort_entry *)
203 xmalloc (sizeof(*s->sort_info->all_entries) *
204 s->sort_info->max_entries);
205 for (i = 0; i < s->sort_info->max_entries; i++)
206 s->sort_info->entries[i] = s->sort_info->all_entries + i;
213 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
217 for (s = zh->sets; s; s = s->next)
218 if (!strcmp (s->name, name))
220 if (!s->rset && s->rpn)
222 NMEM nmem = nmem_create ();
224 rpn_search (zh, nmem, s->rpn, s->num_bases,
225 s->basenames, s->name, s);
234 void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
236 ZebraSet *ss = &zh->sets;
240 for (i = 0; i<num; i++)
241 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
243 zh->errString = NULL;
250 for (i = 0; i<num; i++)
251 if (!strcmp (s->name, names[i]))
254 statuses[i] = Z_DeleteStatus_success;
263 xfree (s->sort_info->all_entries);
264 xfree (s->sort_info->entries);
265 xfree (s->sort_info);
268 nmem_destroy (s->nmem);
269 rset_delete (s->rset);
278 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
279 int num, int *positions)
285 struct zset_sort_info *sort_info;
287 if (!(sset = resultSetGet (zh, name)))
289 if (!(rset = sset->rset))
291 sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
292 for (i = 0; i<num; i++)
297 sort_info = sset->sort_info;
302 for (i = 0; i<num; i++)
304 position = positions[i];
305 if (position > 0 && position <= sort_info->num_entries)
307 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
308 sr[i].sysno = sort_info->entries[position-1]->sysno;
309 sr[i].score = sort_info->entries[position-1]->score;
313 /* did we really get all entries using sort ? */
314 for (i = 0; i<num; i++)
319 if (i < num) /* nope, get the rest, unsorted - sorry */
329 position = sort_info->num_entries;
330 while (num_i < num && positions[num_i] < position)
332 rfd = rset_open (rset, RSETF_READ);
333 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
335 if (key.sysno != psysno)
340 /* determine we alreay have this in our set */
341 for (i = sort_info->num_entries; --i >= 0; )
342 if (psysno == sort_info->entries[i]->sysno)
348 assert (num_i < num);
349 if (position == positions[num_i])
351 sr[num_i].sysno = psysno;
352 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
353 sr[num_i].score = -1;
358 rset_close (rset, rfd);
363 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
373 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
374 struct sortKeyInfo *criteria, int num_criteria,
377 struct zset_sort_entry this_entry;
378 struct zset_sort_entry *new_entry = NULL;
379 struct zset_sort_info *sort_info = sset->sort_info;
382 sortIdx_sysno (zh->service->sortIdx, sysno);
383 for (i = 0; i<num_criteria; i++)
385 sortIdx_type (zh->service->sortIdx, criteria[i].attrUse);
386 sortIdx_read (zh->service->sortIdx, this_entry.buf[i]);
388 i = sort_info->num_entries;
392 for (j = 0; j<num_criteria; j++)
394 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
401 if (criteria[j].relation == 'A')
406 else if (criteria[j].relation == 'D')
412 j = sort_info->max_entries-1;
416 new_entry = sort_info->entries[j];
419 sort_info->entries[j] = sort_info->entries[j-1];
422 sort_info->entries[j] = new_entry;
424 if (sort_info->num_entries != sort_info->max_entries)
425 (sort_info->num_entries)++;
426 for (i = 0; i<num_criteria; i++)
427 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
428 new_entry->sysno = sysno;
429 new_entry->score = -1;
432 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
433 int sysno, int score, int relation)
435 struct zset_sort_entry *new_entry = NULL;
438 i = sort_info->num_entries;
443 rel = score - sort_info->entries[i]->score;
450 else if (relation == 'A')
456 j = sort_info->max_entries-1;
460 new_entry = sort_info->entries[j];
463 sort_info->entries[j] = sort_info->entries[j-1];
466 sort_info->entries[j] = new_entry;
468 if (sort_info->num_entries != sort_info->max_entries)
469 (sort_info->num_entries)++;
470 new_entry->sysno = sysno;
471 new_entry->score = score;
474 void resultSetSort (ZebraHandle zh, NMEM nmem,
475 int num_input_setnames, const char **input_setnames,
476 const char *output_setname,
477 Z_SortKeySpecList *sort_sequence, int *sort_status)
482 if (num_input_setnames == 0)
487 if (num_input_setnames > 1)
492 logf (LOG_DEBUG, "result set sort input=%s output=%s",
493 *input_setnames, output_setname);
494 sset = resultSetGet (zh, input_setnames[0]);
498 zh->errString = nmem_strdup (nmem, input_setnames[0]);
501 if (!(rset = sset->rset))
504 zh->errString = nmem_strdup (nmem, input_setnames[0]);
507 if (strcmp (output_setname, input_setnames[0]))
509 rset = rset_dup (rset);
510 sset = resultSetAdd (zh, output_setname, 1);
513 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
516 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
517 ZebraSet sset, RSET rset,
518 Z_SortKeySpecList *sort_sequence, int *sort_status)
522 struct sortKeyInfo sort_criteria[3];
527 logf (LOG_DEBUG, "resultSetSortSingle start");
528 sset->sort_info->num_entries = 0;
531 num_criteria = sort_sequence->num_specs;
532 if (num_criteria > 3)
534 for (i = 0; i < num_criteria; i++)
536 Z_SortKeySpec *sks = sort_sequence->specs[i];
539 if (*sks->sortRelation == Z_SortRelation_ascending)
540 sort_criteria[i].relation = 'A';
541 else if (*sks->sortRelation == Z_SortRelation_descending)
542 sort_criteria[i].relation = 'D';
548 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
553 else if (sks->sortElement->which != Z_SortElement_generic)
558 sk = sks->sortElement->u.generic;
561 case Z_SortKey_sortField:
562 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
565 case Z_SortKey_elementSpec:
566 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
569 case Z_SortKey_sortAttributes:
570 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
571 sort_criteria[i].attrUse =
572 zebra_maps_sort (zh->service->zebra_maps,
573 sk->u.sortAttributes);
574 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
575 if (sort_criteria[i].attrUse == -1)
580 if (sortIdx_type (zh->service->sortIdx, sort_criteria[i].attrUse))
588 rfd = rset_open (rset, RSETF_READ);
589 while (rset_read (rset, rfd, &key, &term_index))
591 if (key.sysno != psysno)
595 resultSetInsertSort (zh, sset,
596 sort_criteria, num_criteria, psysno);
599 rset_close (rset, rfd);
601 *sort_status = Z_SortStatus_success;
602 logf (LOG_DEBUG, "resultSetSortSingle end");
605 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
609 if ((s = resultSetGet (zh, resultSetId)))
614 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
620 ZebraRankClass rank_class;
621 struct rank_control *rc;
622 struct zset_sort_info *sort_info;
624 sort_info = zebraSet->sort_info;
625 sort_info->num_entries = 0;
627 rfd = rset_open (rset, RSETF_READ);
629 logf (LOG_DEBUG, "resultSetRank");
630 for (i = 0; i < rset->no_rset_terms; i++)
631 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
632 rset->rset_terms[i]->name,
633 rset->rset_terms[i]->nn,
634 rset->rset_terms[i]->flags);
636 rank_class = zebraRankLookup (zh, "rank-1");
637 rc = rank_class->control;
639 if (rset_read (rset, rfd, &key, &term_index))
641 int psysno = key.sysno;
644 (*rc->begin) (zh, rank_class->class_handle, rset);
649 if (key.sysno != psysno)
651 score = (*rc->calc) (handle, psysno);
653 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
657 (*rc->add) (handle, key.seqno, term_index);
659 while (rset_read (rset, rfd, &key, &term_index));
660 score = (*rc->calc) (handle, psysno);
661 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
662 (*rc->end) (zh, handle);
664 rset_close (rset, rfd);
665 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
668 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
670 ZebraRankClass p = zh->service->rank_classes;
671 while (p && strcmp (p->control->name, name))
673 if (p && !p->init_flag)
675 if (p->control->create)
676 p->class_handle = (*p->control->create)(zh->service);
682 void zebraRankInstall (ZebraService zh, struct rank_control *ctrl)
684 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
685 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
686 memcpy (p->control, ctrl, sizeof(*p->control));
687 p->control->name = xstrdup (ctrl->name);
689 p->next = zh->rank_classes;
690 zh->rank_classes = p;
693 void zebraRankDestroy (ZebraService zh)
695 ZebraRankClass p = zh->rank_classes;
698 ZebraRankClass p_next = p->next;
699 if (p->init_flag && p->control->destroy)
700 (*p->control->destroy)(zh, p->class_handle);
701 xfree (p->control->name);
706 zh->rank_classes = NULL;