2 * Copyright (C) 1994-1998, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.21 1998-11-16 16:03:46 adam
8 * Moved loggin utilities to Yaz. Was implemented in file zlogs.c.
10 * Revision 1.20 1998/11/16 10:10:53 adam
11 * Fixed problem with zebraPosSetCreate that occurred when positions were
14 * Revision 1.19 1998/09/22 10:48:22 adam
15 * Minor changes in search API.
17 * Revision 1.18 1998/09/22 10:03:45 adam
18 * Changed result sets to be persistent in the sense that they can
19 * be re-searched if needed.
20 * Fixed memory leak in rsm_or.
22 * Revision 1.17 1998/06/23 15:33:36 adam
23 * Added feature to specify sort criteria in query (type 7 specifies
26 * Revision 1.16 1998/05/20 10:12:24 adam
27 * Implemented automatic EXPLAIN database maintenance.
28 * Modified Zebra to work with ASN.1 compiled version of YAZ.
30 * Revision 1.15 1998/03/05 08:45:14 adam
31 * New result set model and modular ranking system. Moved towards
32 * descent server API. System information stored as "SGML" records.
34 * Revision 1.14 1998/02/10 16:39:15 adam
37 * Revision 1.13 1998/02/10 12:03:06 adam
40 * Revision 1.12 1997/09/25 14:57:36 adam
43 * Revision 1.11 1996/12/23 15:30:46 adam
45 * Bug fix: result sets weren't deleted after server shut down.
47 * Revision 1.10 1995/10/30 15:08:08 adam
50 * Revision 1.9 1995/10/17 18:02:14 adam
51 * New feature: databases. Implemented as prefix to words in dictionary.
53 * Revision 1.8 1995/10/10 13:59:25 adam
54 * Function rset_open changed its wflag parameter to general flags.
56 * Revision 1.7 1995/10/06 14:38:01 adam
57 * New result set method: r_score.
58 * Local no (sysno) and score is transferred to retrieveCtrl.
60 * Revision 1.6 1995/09/28 09:19:49 adam
61 * xfree/xmalloc used everywhere.
62 * Extract/retrieve method seems to work for text records.
64 * Revision 1.5 1995/09/27 16:17:32 adam
65 * More work on retrieve.
67 * Revision 1.4 1995/09/07 13:58:36 adam
68 * New parameter: result-set file descriptor (RSFD) to support multiple
69 * positions within the same result-set.
70 * Boolean operators: and, or, not implemented.
71 * Result-set references.
73 * Revision 1.3 1995/09/06 16:11:19 adam
74 * Option: only one word key per file.
76 * Revision 1.2 1995/09/06 10:33:04 adam
77 * More work on present. Some log messages removed.
79 * Revision 1.1 1995/09/05 15:28:40 adam
80 * More work on search engine.
94 #define SORT_IDX_ENTRYSIZE 64
95 #define ZSET_SORT_MAX_LEVEL 3
105 struct zset_sort_info *sort_info;
106 struct zebra_set *next;
109 struct zset_sort_entry {
112 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
115 struct zset_sort_info {
118 struct zset_sort_entry *all_entries;
119 struct zset_sort_entry **entries;
122 ZebraSet resultSetAddRPN (ZebraHandle zh, ODR input, ODR output,
123 Z_RPNQuery *rpn, int num_bases, char **basenames,
129 zh->errString = NULL;
132 zebraSet = resultSetAdd (zh, setname, 1);
136 zebraSet->num_bases = num_bases;
137 zebraSet->basenames = basenames;
138 zebraSet->nmem = odr_extract_mem (input);
140 zebraSet->rset = rpn_search (zh, output->mem, rpn,
142 zebraSet->basenames, zebraSet->name,
144 zh->hits = zebraSet->hits;
150 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov)
155 for (s = zh->sets; s; s = s->next)
156 if (!strcmp (s->name, name))
160 logf (LOG_DEBUG, "updating result set %s", name);
164 rset_delete (s->rset);
166 nmem_destroy (s->nmem);
170 logf (LOG_DEBUG, "adding result set %s", name);
171 s = xmalloc (sizeof(*s));
174 s->name = xmalloc (strlen(name)+1);
175 strcpy (s->name, name);
177 s->sort_info = xmalloc (sizeof(*s->sort_info));
178 s->sort_info->max_entries = 1000;
179 s->sort_info->entries =
180 xmalloc (sizeof(*s->sort_info->entries) *
181 s->sort_info->max_entries);
182 s->sort_info->all_entries =
183 xmalloc (sizeof(*s->sort_info->all_entries) *
184 s->sort_info->max_entries);
185 for (i = 0; i < s->sort_info->max_entries; i++)
186 s->sort_info->entries[i] = s->sort_info->all_entries + i;
193 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
197 for (s = zh->sets; s; s = s->next)
198 if (!strcmp (s->name, name))
200 if (!s->rset && s->rpn)
202 NMEM nmem = nmem_create ();
204 rpn_search (zh, nmem, s->rpn, s->num_bases,
205 s->basenames, s->name, s);
215 void resultSetDestroy (ZebraHandle zh)
219 for (s = zh->sets; s; s = s1)
223 xfree (s->sort_info->all_entries);
224 xfree (s->sort_info->entries);
225 xfree (s->sort_info);
228 nmem_destroy (s->nmem);
229 rset_delete (s->rset);
236 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
237 int num, int *positions)
243 struct zset_sort_info *sort_info;
245 if (!(sset = resultSetGet (zh, name)))
247 if (!(rset = sset->rset))
249 sr = xmalloc (sizeof(*sr) * num);
250 for (i = 0; i<num; i++)
255 sort_info = sset->sort_info;
260 for (i = 0; i<num; i++)
262 position = positions[i];
263 if (position > 0 && position <= sort_info->num_entries)
265 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
266 sr[i].sysno = sort_info->entries[position-1]->sysno;
267 sr[i].score = sort_info->entries[position-1]->score;
271 /* did we really get all entries using sort ? */
272 for (i = 0; i<num; i++)
277 if (i < num) /* nope, get the rest, unsorted - sorry */
287 position = sort_info->num_entries;
288 while (num_i < num && positions[num_i] < position)
290 rfd = rset_open (rset, RSETF_READ);
291 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
293 if (key.sysno != psysno)
298 /* determine we alreay have this in our set */
299 for (i = sort_info->num_entries; --i >= 0; )
300 if (psysno == sort_info->entries[i]->sysno)
306 assert (num_i < num);
307 if (position == positions[num_i])
309 sr[num_i].sysno = psysno;
310 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
311 sr[num_i].score = -1;
316 rset_close (rset, rfd);
321 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
331 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
332 struct sortKey *criteria, int num_criteria,
335 struct zset_sort_entry this_entry;
336 struct zset_sort_entry *new_entry = NULL;
337 struct zset_sort_info *sort_info = sset->sort_info;
340 sortIdx_sysno (zh->sortIdx, sysno);
341 for (i = 0; i<num_criteria; i++)
343 sortIdx_type (zh->sortIdx, criteria[i].attrUse);
344 sortIdx_read (zh->sortIdx, this_entry.buf[i]);
346 i = sort_info->num_entries;
350 for (j = 0; j<num_criteria; j++)
352 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
359 if (criteria[j].relation == 'A')
364 else if (criteria[j].relation == 'D')
370 j = sort_info->max_entries-1;
374 new_entry = sort_info->entries[j];
377 sort_info->entries[j] = sort_info->entries[j-1];
380 sort_info->entries[j] = new_entry;
382 if (sort_info->num_entries != sort_info->max_entries)
383 (sort_info->num_entries)++;
384 for (i = 0; i<num_criteria; i++)
385 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
386 new_entry->sysno = sysno;
387 new_entry->score = -1;
390 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
391 int sysno, int score, int relation)
393 struct zset_sort_entry *new_entry = NULL;
396 i = sort_info->num_entries;
401 rel = score - sort_info->entries[i]->score;
408 else if (relation == 'A')
414 j = sort_info->max_entries-1;
418 new_entry = sort_info->entries[j];
421 sort_info->entries[j] = sort_info->entries[j-1];
424 sort_info->entries[j] = new_entry;
426 if (sort_info->num_entries != sort_info->max_entries)
427 (sort_info->num_entries)++;
428 new_entry->sysno = sysno;
429 new_entry->score = score;
432 void resultSetSort (ZebraHandle zh, NMEM nmem,
433 int num_input_setnames, const char **input_setnames,
434 const char *output_setname,
435 Z_SortKeySpecList *sort_sequence, int *sort_status)
440 if (num_input_setnames == 0)
445 if (num_input_setnames > 1)
450 logf (LOG_DEBUG, "result set sort input=%s output=%s",
451 *input_setnames, output_setname);
452 sset = resultSetGet (zh, input_setnames[0]);
456 zh->errString = nmem_strdup (nmem, input_setnames[0]);
459 if (!(rset = sset->rset))
462 zh->errString = nmem_strdup (nmem, input_setnames[0]);
465 if (strcmp (output_setname, input_setnames[0]))
467 rset = rset_dup (rset);
468 sset = resultSetAdd (zh, output_setname, 1);
471 resultSetSortSingle (zh, nmem, sset, rset, sort_sequence, sort_status);
474 void resultSetSortSingle (ZebraHandle zh, NMEM nmem,
475 ZebraSet sset, RSET rset,
476 Z_SortKeySpecList *sort_sequence, int *sort_status)
480 struct sortKey sort_criteria[3];
485 logf (LOG_DEBUG, "resultSetSortSingle start");
486 sset->sort_info->num_entries = 0;
489 num_criteria = sort_sequence->num_specs;
490 if (num_criteria > 3)
492 for (i = 0; i < num_criteria; i++)
494 Z_SortKeySpec *sks = sort_sequence->specs[i];
497 if (*sks->sortRelation == Z_SortRelation_ascending)
498 sort_criteria[i].relation = 'A';
499 else if (*sks->sortRelation == Z_SortRelation_descending)
500 sort_criteria[i].relation = 'D';
506 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
511 else if (sks->sortElement->which != Z_SortElement_generic)
516 sk = sks->sortElement->u.generic;
519 case Z_SortKey_sortField:
520 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
523 case Z_SortKey_elementSpec:
524 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
527 case Z_SortKey_sortAttributes:
528 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
529 sort_criteria[i].attrUse =
530 zebra_maps_sort (zh->zebra_maps, sk->u.sortAttributes);
531 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
532 if (sort_criteria[i].attrUse == -1)
537 if (sortIdx_type (zh->sortIdx, sort_criteria[i].attrUse))
545 rfd = rset_open (rset, RSETF_READ);
546 while (rset_read (rset, rfd, &key, &term_index))
548 if (key.sysno != psysno)
552 resultSetInsertSort (zh, sset,
553 sort_criteria, num_criteria, psysno);
556 rset_close (rset, rfd);
559 *sort_status = Z_SortStatus_success;
560 logf (LOG_DEBUG, "resultSetSortSingle end");
563 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
567 if ((s = resultSetGet (zh, resultSetId)))
572 void resultSetRank (ZebraHandle zh, ZebraSet zebraSet, RSET rset)
578 ZebraRankClass rank_class;
579 struct rank_control *rc;
580 struct zset_sort_info *sort_info;
582 sort_info = zebraSet->sort_info;
583 sort_info->num_entries = 0;
585 rfd = rset_open (rset, RSETF_READ);
587 logf (LOG_DEBUG, "resultSetRank");
588 for (i = 0; i < rset->no_rset_terms; i++)
589 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
590 rset->rset_terms[i]->name,
591 rset->rset_terms[i]->nn,
592 rset->rset_terms[i]->flags);
594 rank_class = zebraRankLookup (zh, "rank-1");
595 rc = rank_class->control;
597 if (rset_read (rset, rfd, &key, &term_index))
599 int psysno = key.sysno;
602 (*rc->begin) (zh, rank_class->class_handle, rset);
607 if (key.sysno != psysno)
609 score = (*rc->calc) (handle, psysno);
611 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
615 (*rc->add) (handle, key.seqno, term_index);
617 while (rset_read (rset, rfd, &key, &term_index));
618 score = (*rc->calc) (handle, psysno);
619 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
620 (*rc->end) (zh, handle);
622 rset_close (rset, rfd);
623 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, zebraSet->hits);
626 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
628 ZebraRankClass p = zh->rank_classes;
629 while (p && strcmp (p->control->name, name))
631 if (p && !p->init_flag)
633 if (p->control->create)
634 p->class_handle = (*p->control->create)(zh);
640 void zebraRankInstall (ZebraHandle zh, struct rank_control *ctrl)
642 ZebraRankClass p = xmalloc (sizeof(*p));
643 p->control = xmalloc (sizeof(*p->control));
644 memcpy (p->control, ctrl, sizeof(*p->control));
645 p->control->name = xstrdup (ctrl->name);
647 p->next = zh->rank_classes;
648 zh->rank_classes = p;
651 void zebraRankDestroy (ZebraHandle zh)
653 ZebraRankClass p = zh->rank_classes;
656 ZebraRankClass p_next = p->next;
657 if (p->init_flag && p->control->destroy)
658 (*p->control->destroy)(zh, p->class_handle);
659 xfree (p->control->name);
664 zh->rank_classes = NULL;