2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.23 1999-05-26 07:49:13 adam
10 * Revision 1.22 1999/02/02 14:51:15 adam
11 * Updated WIN32 code specific sections. Changed header.
13 * Revision 1.21 1998/11/16 16:03:46 adam
14 * Moved loggin utilities to Yaz. Was implemented in file zlogs.c.
16 * Revision 1.20 1998/11/16 10:10:53 adam
17 * Fixed problem with zebraPosSetCreate that occurred when positions were
20 * Revision 1.19 1998/09/22 10:48:22 adam
21 * Minor changes in search API.
23 * Revision 1.18 1998/09/22 10:03:45 adam
24 * Changed result sets to be persistent in the sense that they can
25 * be re-searched if needed.
26 * Fixed memory leak in rsm_or.
28 * Revision 1.17 1998/06/23 15:33:36 adam
29 * Added feature to specify sort criteria in query (type 7 specifies
32 * Revision 1.16 1998/05/20 10:12:24 adam
33 * Implemented automatic EXPLAIN database maintenance.
34 * Modified Zebra to work with ASN.1 compiled version of YAZ.
36 * Revision 1.15 1998/03/05 08:45:14 adam
37 * New result set model and modular ranking system. Moved towards
38 * descent server API. System information stored as "SGML" records.
40 * Revision 1.14 1998/02/10 16:39:15 adam
43 * Revision 1.13 1998/02/10 12:03:06 adam
46 * Revision 1.12 1997/09/25 14:57:36 adam
49 * Revision 1.11 1996/12/23 15:30:46 adam
51 * Bug fix: result sets weren't deleted after server shut down.
53 * Revision 1.10 1995/10/30 15:08:08 adam
56 * Revision 1.9 1995/10/17 18:02:14 adam
57 * New feature: databases. Implemented as prefix to words in dictionary.
59 * Revision 1.8 1995/10/10 13:59:25 adam
60 * Function rset_open changed its wflag parameter to general flags.
62 * Revision 1.7 1995/10/06 14:38:01 adam
63 * New result set method: r_score.
64 * Local no (sysno) and score is transferred to retrieveCtrl.
66 * Revision 1.6 1995/09/28 09:19:49 adam
67 * xfree/xmalloc used everywhere.
68 * Extract/retrieve method seems to work for text records.
70 * Revision 1.5 1995/09/27 16:17:32 adam
71 * More work on retrieve.
73 * Revision 1.4 1995/09/07 13:58:36 adam
74 * New parameter: result-set file descriptor (RSFD) to support multiple
75 * positions within the same result-set.
76 * Boolean operators: and, or, not implemented.
77 * Result-set references.
79 * Revision 1.3 1995/09/06 16:11:19 adam
80 * Option: only one word key per file.
82 * Revision 1.2 1995/09/06 10:33:04 adam
83 * More work on present. Some log messages removed.
85 * Revision 1.1 1995/09/05 15:28:40 adam
86 * More work on search engine.
100 #define SORT_IDX_ENTRYSIZE 64
101 #define ZSET_SORT_MAX_LEVEL 3
111 struct zset_sort_info *sort_info;
112 struct zebra_set *next;
115 struct zset_sort_entry {
118 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
121 struct zset_sort_info {
124 struct zset_sort_entry *all_entries;
125 struct zset_sort_entry **entries;
128 ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
129 Z_RPNQuery *rpn, int num_bases, char **basenames,
135 zh->errString = NULL;
138 zebraSet = resultSetAdd (zh, setname, 1);
142 zebraSet->num_bases = num_bases;
143 zebraSet->basenames = basenames;
144 zebraSet->nmem = odr_extract_mem (input);
146 zebraSet->rset = rpn_search (zh, output->mem, rpn,
148 zebraSet->basenames, zebraSet->name,
150 zh->hits = zebraSet->hits;
156 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
161 for (s = zh->sets; s; s = s->next)
162 if (!strcmp (s->name, name))
166 logf (LOG_DEBUG, "updating result set %s", name);
170 rset_delete (s->rset);
172 nmem_destroy (s->nmem);
176 logf (LOG_DEBUG, "adding result set %s", name);
177 s = (ZebraSet) xmalloc (sizeof(*s));
180 s->name = (char *) xmalloc (strlen(name)+1);
181 strcpy (s->name, name);
183 s->sort_info = (struct zset_sort_info *)
184 xmalloc (sizeof(*s->sort_info));
185 s->sort_info->max_entries = 1000;
186 s->sort_info->entries = (struct zset_sort_entry **)
187 xmalloc (sizeof(*s->sort_info->entries) *
188 s->sort_info->max_entries);
189 s->sort_info->all_entries = (struct zset_sort_entry *)
190 xmalloc (sizeof(*s->sort_info->all_entries) *
191 s->sort_info->max_entries);
192 for (i = 0; i < s->sort_info->max_entries; i++)
193 s->sort_info->entries[i] = s->sort_info->all_entries + i;
200 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
204 for (s = zh->sets; s; s = s->next)
205 if (!strcmp (s->name, name))
207 if (!s->rset && s->rpn)
209 NMEM nmem = nmem_create ();
211 rpn_search (zh, nmem, s->rpn, s->num_bases,
212 s->basenames, s->name, s);
222 void resultSetDestroy (ZebraHandle zh)
226 for (s = zh->sets; s; s = s1)
230 xfree (s->sort_info->all_entries);
231 xfree (s->sort_info->entries);
232 xfree (s->sort_info);
235 nmem_destroy (s->nmem);
236 rset_delete (s->rset);
243 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
244 int num, int *positions)
250 struct zset_sort_info *sort_info;
252 if (!(sset = resultSetGet (zh, name)))
254 if (!(rset = sset->rset))
256 sr = (ZebraPosSet) xmalloc (sizeof(*sr) * num);
257 for (i = 0; i<num; i++)
262 sort_info = sset->sort_info;
267 for (i = 0; i<num; i++)
269 position = positions[i];
270 if (position > 0 && position <= sort_info->num_entries)
272 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
273 sr[i].sysno = sort_info->entries[position-1]->sysno;
274 sr[i].score = sort_info->entries[position-1]->score;
278 /* did we really get all entries using sort ? */
279 for (i = 0; i<num; i++)
284 if (i < num) /* nope, get the rest, unsorted - sorry */
294 position = sort_info->num_entries;
295 while (num_i < num && positions[num_i] < position)
297 rfd = rset_open (rset, RSETF_READ);
298 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
300 if (key.sysno != psysno)
305 /* determine we alreay have this in our set */
306 for (i = sort_info->num_entries; --i >= 0; )
307 if (psysno == sort_info->entries[i]->sysno)
313 assert (num_i < num);
314 if (position == positions[num_i])
316 sr[num_i].sysno = psysno;
317 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
318 sr[num_i].score = -1;
323 rset_close (rset, rfd);
328 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
338 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
339 struct sortKey *criteria, int num_criteria,
342 struct zset_sort_entry this_entry;
343 struct zset_sort_entry *new_entry = NULL;
344 struct zset_sort_info *sort_info = sset->sort_info;
347 sortIdx_sysno (zh->sortIdx, sysno);
348 for (i = 0; i<num_criteria; i++)
350 sortIdx_type (zh->sortIdx, criteria[i].attrUse);
351 sortIdx_read (zh->sortIdx, this_entry.buf[i]);
353 i = sort_info->num_entries;
357 for (j = 0; j<num_criteria; j++)
359 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
366 if (criteria[j].relation == 'A')
371 else if (criteria[j].relation == 'D')
377 j = sort_info->max_entries-1;
381 new_entry = sort_info->entries[j];
384 sort_info->entries[j] = sort_info->entries[j-1];
387 sort_info->entries[j] = new_entry;
389 if (sort_info->num_entries != sort_info->max_entries)
390 (sort_info->num_entries)++;
391 for (i = 0; i<num_criteria; i++)
392 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
393 new_entry->sysno = sysno;
394 new_entry->score = -1;
397 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
398 int sysno, int score, int relation)
400 struct zset_sort_entry *new_entry = NULL;
403 i = sort_info->num_entries;
408 rel = score - sort_info->entries[i]->score;
415 else if (relation == 'A')
421 j = sort_info->max_entries-1;
425 new_entry = sort_info->entries[j];
428 sort_info->entries[j] = sort_info->entries[j-1];
431 sort_info->entries[j] = new_entry;
433 if (sort_info->num_entries != sort_info->max_entries)
434 (sort_info->num_entries)++;
435 new_entry->sysno = sysno;
436 new_entry->score = score;
439 void resultSetSort (ZebraHandle zh, NMEM nmem,
440 int num_input_setnames, const char **input_setnames,
441 const char *output_setname,
442 Z_SortKeySpecList *sort_sequence, int *sort_status)
447 if (num_input_setnames == 0)
452 if (num_input_setnames > 1)
457 logf (LOG_DEBUG, "result set sort input=%s output=%s",
458 *input_setnames, output_setname);
459 sset = resultSetGet (zh, input_setnames[0]);
463 zh->errString = nmem_strdup (nmem, input_setnames[0]);
466 if (!(rset = sset->rset))
469 zh->errString = nmem_strdup (nmem, input_setnames[0]);
472 if (strcmp (output_setname, input_setnames[0]))
474 rset = rset_dup (rset);
475 sset = resultSetAdd (zh, output_setname, 1);
478 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
481 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
482 ZebraSet sset, RSET rset,
483 Z_SortKeySpecList *sort_sequence, int *sort_status)
487 struct sortKey sort_criteria[3];
492 logf (LOG_DEBUG, "resultSetSortSingle start");
493 sset->sort_info->num_entries = 0;
496 num_criteria = sort_sequence->num_specs;
497 if (num_criteria > 3)
499 for (i = 0; i < num_criteria; i++)
501 Z_SortKeySpec *sks = sort_sequence->specs[i];
504 if (*sks->sortRelation == Z_SortRelation_ascending)
505 sort_criteria[i].relation = 'A';
506 else if (*sks->sortRelation == Z_SortRelation_descending)
507 sort_criteria[i].relation = 'D';
513 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
518 else if (sks->sortElement->which != Z_SortElement_generic)
523 sk = sks->sortElement->u.generic;
526 case Z_SortKey_sortField:
527 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
530 case Z_SortKey_elementSpec:
531 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
534 case Z_SortKey_sortAttributes:
535 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
536 sort_criteria[i].attrUse =
537 zebra_maps_sort (zh->zebra_maps, sk->u.sortAttributes);
538 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
539 if (sort_criteria[i].attrUse == -1)
544 if (sortIdx_type (zh->sortIdx, sort_criteria[i].attrUse))
552 rfd = rset_open (rset, RSETF_READ);
553 while (rset_read (rset, rfd, &key, &term_index))
555 if (key.sysno != psysno)
559 resultSetInsertSort (zh, sset,
560 sort_criteria, num_criteria, psysno);
563 rset_close (rset, rfd);
566 *sort_status = Z_SortStatus_success;
567 logf (LOG_DEBUG, "resultSetSortSingle end");
570 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
574 if ((s = resultSetGet (zh, resultSetId)))
579 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
585 ZebraRankClass rank_class;
586 struct rank_control *rc;
587 struct zset_sort_info *sort_info;
589 sort_info = zebraSet->sort_info;
590 sort_info->num_entries = 0;
592 rfd = rset_open (rset, RSETF_READ);
594 logf (LOG_DEBUG, "resultSetRank");
595 for (i = 0; i < rset->no_rset_terms; i++)
596 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
597 rset->rset_terms[i]->name,
598 rset->rset_terms[i]->nn,
599 rset->rset_terms[i]->flags);
601 rank_class = zebraRankLookup (zh, "rank-1");
602 rc = rank_class->control;
604 if (rset_read (rset, rfd, &key, &term_index))
606 int psysno = key.sysno;
609 (*rc->begin) (zh, rank_class->class_handle, rset);
614 if (key.sysno != psysno)
616 score = (*rc->calc) (handle, psysno);
618 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
622 (*rc->add) (handle, key.seqno, term_index);
624 while (rset_read (rset, rfd, &key, &term_index));
625 score = (*rc->calc) (handle, psysno);
626 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
627 (*rc->end) (zh, handle);
629 rset_close (rset, rfd);
630 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
633 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
635 ZebraRankClass p = zh->rank_classes;
636 while (p && strcmp (p->control->name, name))
638 if (p && !p->init_flag)
640 if (p->control->create)
641 p->class_handle = (*p->control->create)(zh);
647 void zebraRankInstall (ZebraHandle zh, struct rank_control *ctrl)
649 ZebraRankClass p = (ZebraRankClass) xmalloc (sizeof(*p));
650 p->control = (struct rank_control *) xmalloc (sizeof(*p->control));
651 memcpy (p->control, ctrl, sizeof(*p->control));
652 p->control->name = xstrdup (ctrl->name);
654 p->next = zh->rank_classes;
655 zh->rank_classes = p;
658 void zebraRankDestroy (ZebraHandle zh)
660 ZebraRankClass p = zh->rank_classes;
663 ZebraRankClass p_next = p->next;
664 if (p->init_flag && p->control->destroy)
665 (*p->control->destroy)(zh, p->class_handle);
666 xfree (p->control->name);
671 zh->rank_classes = NULL;