2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.25 2000-03-15 15:00:31 adam
8 * First work on threaded version.
10 * Revision 1.24 1999/11/04 15:00:45 adam
11 * Implemented delete result set(s).
13 * Revision 1.23 1999/05/26 07:49:13 adam
16 * Revision 1.22 1999/02/02 14:51:15 adam
17 * Updated WIN32 code specific sections. Changed header.
19 * Revision 1.21 1998/11/16 16:03:46 adam
20 * Moved loggin utilities to Yaz. Was implemented in file zlogs.c.
22 * Revision 1.20 1998/11/16 10:10:53 adam
23 * Fixed problem with zebraPosSetCreate that occurred when positions were
26 * Revision 1.19 1998/09/22 10:48:22 adam
27 * Minor changes in search API.
29 * Revision 1.18 1998/09/22 10:03:45 adam
30 * Changed result sets to be persistent in the sense that they can
31 * be re-searched if needed.
32 * Fixed memory leak in rsm_or.
34 * Revision 1.17 1998/06/23 15:33:36 adam
35 * Added feature to specify sort criteria in query (type 7 specifies
38 * Revision 1.16 1998/05/20 10:12:24 adam
39 * Implemented automatic EXPLAIN database maintenance.
40 * Modified Zebra to work with ASN.1 compiled version of YAZ.
42 * Revision 1.15 1998/03/05 08:45:14 adam
43 * New result set model and modular ranking system. Moved towards
44 * descent server API. System information stored as "SGML" records.
46 * Revision 1.14 1998/02/10 16:39:15 adam
49 * Revision 1.13 1998/02/10 12:03:06 adam
52 * Revision 1.12 1997/09/25 14:57:36 adam
55 * Revision 1.11 1996/12/23 15:30:46 adam
57 * Bug fix: result sets weren't deleted after server shut down.
59 * Revision 1.10 1995/10/30 15:08:08 adam
62 * Revision 1.9 1995/10/17 18:02:14 adam
63 * New feature: databases. Implemented as prefix to words in dictionary.
65 * Revision 1.8 1995/10/10 13:59:25 adam
66 * Function rset_open changed its wflag parameter to general flags.
68 * Revision 1.7 1995/10/06 14:38:01 adam
69 * New result set method: r_score.
70 * Local no (sysno) and score is transferred to retrieveCtrl.
72 * Revision 1.6 1995/09/28 09:19:49 adam
73 * xfree/xmalloc used everywhere.
74 * Extract/retrieve method seems to work for text records.
76 * Revision 1.5 1995/09/27 16:17:32 adam
77 * More work on retrieve.
79 * Revision 1.4 1995/09/07 13:58:36 adam
80 * New parameter: result-set file descriptor (RSFD) to support multiple
81 * positions within the same result-set.
82 * Boolean operators: and, or, not implemented.
83 * Result-set references.
85 * Revision 1.3 1995/09/06 16:11:19 adam
86 * Option: only one word key per file.
88 * Revision 1.2 1995/09/06 10:33:04 adam
89 * More work on present. Some log messages removed.
91 * Revision 1.1 1995/09/05 15:28:40 adam
92 * More work on search engine.
106 #define SORT_IDX_ENTRYSIZE 64
107 #define ZSET_SORT_MAX_LEVEL 3
117 struct zset_sort_info *sort_info;
118 struct zebra_set *next;
121 struct zset_sort_entry {
124 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
127 struct zset_sort_info {
130 struct zset_sort_entry *all_entries;
131 struct zset_sort_entry **entries;
134 ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
135 Z_RPNQuery *rpn, int num_bases, char **basenames,
141 zh->errString = NULL;
144 zebraSet = resultSetAdd (zh, setname, 1);
148 zebraSet->num_bases = num_bases;
149 zebraSet->basenames = basenames;
150 zebraSet->nmem = odr_extract_mem (input);
152 zebraSet->rset = rpn_search (zh, output->mem, rpn,
154 zebraSet->basenames, zebraSet->name,
156 zh->hits = zebraSet->hits;
162 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
167 for (s = zh->sets; s; s = s->next)
168 if (!strcmp (s->name, name))
172 logf (LOG_DEBUG, "updating result set %s", name);
176 rset_delete (s->rset);
178 nmem_destroy (s->nmem);
182 logf (LOG_DEBUG, "adding result set %s", name);
183 s = (ZebraSet) xmalloc (sizeof(*s));
186 s->name = (char *) xmalloc (strlen(name)+1);
187 strcpy (s->name, name);
189 s->sort_info = (struct zset_sort_info *)
190 xmalloc (sizeof(*s->sort_info));
191 s->sort_info->max_entries = 1000;
192 s->sort_info->entries = (struct zset_sort_entry **)
193 xmalloc (sizeof(*s->sort_info->entries) *
194 s->sort_info->max_entries);
195 s->sort_info->all_entries = (struct zset_sort_entry *)
196 xmalloc (sizeof(*s->sort_info->all_entries) *
197 s->sort_info->max_entries);
198 for (i = 0; i < s->sort_info->max_entries; i++)
199 s->sort_info->entries[i] = s->sort_info->all_entries + i;
206 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
210 for (s = zh->sets; s; s = s->next)
211 if (!strcmp (s->name, name))
213 if (!s->rset && s->rpn)
215 NMEM nmem = nmem_create ();
217 rpn_search (zh, nmem, s->rpn, s->num_bases,
218 s->basenames, s->name, s);
227 void resultSetDestroy (ZebraHandle zh, int num, char **names,int *statuses)
229 ZebraSet *ss = &zh->sets;
233 for (i = 0; i<num; i++)
234 statuses[i] = Z_DeleteStatus_resultSetDidNotExist;
236 zh->errString = NULL;
243 for (i = 0; i<num; i++)
244 if (!strcmp (s->name, names[i]))
247 statuses[i] = Z_DeleteStatus_success;
256 xfree (s->sort_info->all_entries);
257 xfree (s->sort_info->entries);
258 xfree (s->sort_info);
261 nmem_destroy (s->nmem);
262 rset_delete (s->rset);
271 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
272 int num, int *positions)
278 struct zset_sort_info *sort_info;
280 if (!(sset = resultSetGet (zh, name)))
282 if (!(rset = sset->rset))
284 sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
285 for (i = 0; i<num; i++)
290 sort_info = sset->sort_info;
295 for (i = 0; i<num; i++)
297 position = positions[i];
298 if (position > 0 && position <= sort_info->num_entries)
300 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
301 sr[i].sysno = sort_info->entries[position-1]->sysno;
302 sr[i].score = sort_info->entries[position-1]->score;
306 /* did we really get all entries using sort ? */
307 for (i = 0; i<num; i++)
312 if (i < num) /* nope, get the rest, unsorted - sorry */
322 position = sort_info->num_entries;
323 while (num_i < num && positions[num_i] < position)
325 rfd = rset_open (rset, RSETF_READ);
326 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
328 if (key.sysno != psysno)
333 /* determine we alreay have this in our set */
334 for (i = sort_info->num_entries; --i >= 0; )
335 if (psysno == sort_info->entries[i]->sysno)
341 assert (num_i < num);
342 if (position == positions[num_i])
344 sr[num_i].sysno = psysno;
345 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
346 sr[num_i].score = -1;
351 rset_close (rset, rfd);
356 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
366 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
367 struct sortKey *criteria, int num_criteria,
370 struct zset_sort_entry this_entry;
371 struct zset_sort_entry *new_entry = NULL;
372 struct zset_sort_info *sort_info = sset->sort_info;
375 sortIdx_sysno (zh->service->sortIdx, sysno);
376 for (i = 0; i<num_criteria; i++)
378 sortIdx_type (zh->service->sortIdx, criteria[i].attrUse);
379 sortIdx_read (zh->service->sortIdx, this_entry.buf[i]);
381 i = sort_info->num_entries;
385 for (j = 0; j<num_criteria; j++)
387 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
394 if (criteria[j].relation == 'A')
399 else if (criteria[j].relation == 'D')
405 j = sort_info->max_entries-1;
409 new_entry = sort_info->entries[j];
412 sort_info->entries[j] = sort_info->entries[j-1];
415 sort_info->entries[j] = new_entry;
417 if (sort_info->num_entries != sort_info->max_entries)
418 (sort_info->num_entries)++;
419 for (i = 0; i<num_criteria; i++)
420 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
421 new_entry->sysno = sysno;
422 new_entry->score = -1;
425 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
426 int sysno, int score, int relation)
428 struct zset_sort_entry *new_entry = NULL;
431 i = sort_info->num_entries;
436 rel = score - sort_info->entries[i]->score;
443 else if (relation == 'A')
449 j = sort_info->max_entries-1;
453 new_entry = sort_info->entries[j];
456 sort_info->entries[j] = sort_info->entries[j-1];
459 sort_info->entries[j] = new_entry;
461 if (sort_info->num_entries != sort_info->max_entries)
462 (sort_info->num_entries)++;
463 new_entry->sysno = sysno;
464 new_entry->score = score;
467 void resultSetSort (ZebraHandle zh, NMEM nmem,
468 int num_input_setnames, const char **input_setnames,
469 const char *output_setname,
470 Z_SortKeySpecList *sort_sequence, int *sort_status)
475 if (num_input_setnames == 0)
480 if (num_input_setnames > 1)
485 logf (LOG_DEBUG, "result set sort input=%s output=%s",
486 *input_setnames, output_setname);
487 sset = resultSetGet (zh, input_setnames[0]);
491 zh->errString = nmem_strdup (nmem, input_setnames[0]);
494 if (!(rset = sset->rset))
497 zh->errString = nmem_strdup (nmem, input_setnames[0]);
500 if (strcmp (output_setname, input_setnames[0]))
502 rset = rset_dup (rset);
503 sset = resultSetAdd (zh, output_setname, 1);
506 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
509 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
510 ZebraSet sset, RSET rset,
511 Z_SortKeySpecList *sort_sequence, int *sort_status)
515 struct sortKey sort_criteria[3];
520 logf (LOG_DEBUG, "resultSetSortSingle start");
521 sset->sort_info->num_entries = 0;
524 num_criteria = sort_sequence->num_specs;
525 if (num_criteria > 3)
527 for (i = 0; i < num_criteria; i++)
529 Z_SortKeySpec *sks = sort_sequence->specs[i];
532 if (*sks->sortRelation == Z_SortRelation_ascending)
533 sort_criteria[i].relation = 'A';
534 else if (*sks->sortRelation == Z_SortRelation_descending)
535 sort_criteria[i].relation = 'D';
541 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
546 else if (sks->sortElement->which != Z_SortElement_generic)
551 sk = sks->sortElement->u.generic;
554 case Z_SortKey_sortField:
555 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
558 case Z_SortKey_elementSpec:
559 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
562 case Z_SortKey_sortAttributes:
563 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
564 sort_criteria[i].attrUse =
565 zebra_maps_sort (zh->service->zebra_maps,
566 sk->u.sortAttributes);
567 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
568 if (sort_criteria[i].attrUse == -1)
573 if (sortIdx_type (zh->service->sortIdx, sort_criteria[i].attrUse))
581 rfd = rset_open (rset, RSETF_READ);
582 while (rset_read (rset, rfd, &key, &term_index))
584 if (key.sysno != psysno)
588 resultSetInsertSort (zh, sset,
589 sort_criteria, num_criteria, psysno);
592 rset_close (rset, rfd);
595 *sort_status = Z_SortStatus_success;
596 logf (LOG_DEBUG, "resultSetSortSingle end");
599 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
603 if ((s = resultSetGet (zh, resultSetId)))
608 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
614 ZebraRankClass rank_class;
615 struct rank_control *rc;
616 struct zset_sort_info *sort_info;
618 sort_info = zebraSet->sort_info;
619 sort_info->num_entries = 0;
621 rfd = rset_open (rset, RSETF_READ);
623 logf (LOG_DEBUG, "resultSetRank");
624 for (i = 0; i < rset->no_rset_terms; i++)
625 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
626 rset->rset_terms[i]->name,
627 rset->rset_terms[i]->nn,
628 rset->rset_terms[i]->flags);
630 rank_class = zebraRankLookup (zh, "rank-1");
631 rc = rank_class->control;
633 if (rset_read (rset, rfd, &key, &term_index))
635 int psysno = key.sysno;
638 (*rc->begin) (zh, rank_class->class_handle, rset);
643 if (key.sysno != psysno)
645 score = (*rc->calc) (handle, psysno);
647 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
651 (*rc->add) (handle, key.seqno, term_index);
653 while (rset_read (rset, rfd, &key, &term_index));
654 score = (*rc->calc) (handle, psysno);
655 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
656 (*rc->end) (zh, handle);
658 rset_close (rset, rfd);
659 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
662 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
664 ZebraRankClass p = zh->service->rank_classes;
665 while (p && strcmp (p->control->name, name))
667 if (p && !p->init_flag)
669 if (p->control->create)
670 p->class_handle = (*p->control->create)(zh->service);
676 void zebraRankInstall (ZebraService zh, struct rank_control *ctrl)
678 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
679 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
680 memcpy (p->control, ctrl, sizeof(*p->control));
681 p->control->name = xstrdup (ctrl->name);
683 p->next = zh->rank_classes;
684 zh->rank_classes = p;
687 void zebraRankDestroy (ZebraService zh)
689 ZebraRankClass p = zh->rank_classes;
692 ZebraRankClass p_next = p->next;
693 if (p->init_flag && p->control->destroy)
694 (*p->control->destroy)(zh, p->class_handle);
695 xfree (p->control->name);
700 zh->rank_classes = NULL;