2 * Copyright (C) 1994-1998, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.16 1998-05-20 10:12:24 adam
8 * Implemented automatic EXPLAIN database maintenance.
9 * Modified Zebra to work with ASN.1 compiled version of YAZ.
11 * Revision 1.15 1998/03/05 08:45:14 adam
12 * New result set model and modular ranking system. Moved towards
13 * descent server API. System information stored as "SGML" records.
15 * Revision 1.14 1998/02/10 16:39:15 adam
18 * Revision 1.13 1998/02/10 12:03:06 adam
21 * Revision 1.12 1997/09/25 14:57:36 adam
24 * Revision 1.11 1996/12/23 15:30:46 adam
26 * Bug fix: result sets weren't deleted after server shut down.
28 * Revision 1.10 1995/10/30 15:08:08 adam
31 * Revision 1.9 1995/10/17 18:02:14 adam
32 * New feature: databases. Implemented as prefix to words in dictionary.
34 * Revision 1.8 1995/10/10 13:59:25 adam
35 * Function rset_open changed its wflag parameter to general flags.
37 * Revision 1.7 1995/10/06 14:38:01 adam
38 * New result set method: r_score.
39 * Local no (sysno) and score is transferred to retrieveCtrl.
41 * Revision 1.6 1995/09/28 09:19:49 adam
42 * xfree/xmalloc used everywhere.
43 * Extract/retrieve method seems to work for text records.
45 * Revision 1.5 1995/09/27 16:17:32 adam
46 * More work on retrieve.
48 * Revision 1.4 1995/09/07 13:58:36 adam
49 * New parameter: result-set file descriptor (RSFD) to support multiple
50 * positions within the same result-set.
51 * Boolean operators: and, or, not implemented.
52 * Result-set references.
54 * Revision 1.3 1995/09/06 16:11:19 adam
55 * Option: only one word key per file.
57 * Revision 1.2 1995/09/06 10:33:04 adam
58 * More work on present. Some log messages removed.
60 * Revision 1.1 1995/09/05 15:28:40 adam
61 * More work on search engine.
75 #define SORT_IDX_ENTRYSIZE 64
76 #define ZSET_SORT_MAX_LEVEL 3
82 struct zset_sort_info *sort_info;
83 struct zebra_set *next;
86 struct zset_sort_entry {
89 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
92 struct zset_sort_info {
95 struct zset_sort_entry *all_entries;
96 struct zset_sort_entry **entries;
99 void resultSetRank (ZebraHandle zh, struct zset_sort_info *si,
100 RSET rset, int *hits);
102 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov,
103 RSET rset, int *hits)
108 for (s = zh->sets; s; s = s->next)
109 if (!strcmp (s->name, name))
111 logf (LOG_DEBUG, "updating result set %s", name);
114 rset_delete (s->rset);
116 resultSetRank (zh, s->sort_info, rset, hits);
119 logf (LOG_DEBUG, "adding result set %s", name);
120 s = xmalloc (sizeof(*s));
123 s->name = xmalloc (strlen(name)+1);
124 strcpy (s->name, name);
127 s->sort_info = xmalloc (sizeof(*s->sort_info));
128 s->sort_info->max_entries = 1000;
129 s->sort_info->entries = xmalloc (sizeof(*s->sort_info->entries) *
130 s->sort_info->max_entries);
131 s->sort_info->all_entries = xmalloc (sizeof(*s->sort_info->all_entries) *
132 s->sort_info->max_entries);
133 for (i = 0; i < s->sort_info->max_entries; i++)
134 s->sort_info->entries[i] = s->sort_info->all_entries + i;
135 resultSetRank (zh, s->sort_info, rset, hits);
139 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
143 for (s = zh->sets; s; s = s->next)
144 if (!strcmp (s->name, name))
150 void resultSetDestroy (ZebraHandle zh)
154 for (s = zh->sets; s; s = s1)
158 xfree (s->sort_info->all_entries);
159 xfree (s->sort_info->entries);
160 xfree (s->sort_info);
162 rset_delete (s->rset);
169 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
170 int num, int *positions)
176 struct zset_sort_info *sort_info;
178 if (!(sset = resultSetGet (zh, name)))
180 if (!(rset = sset->rset))
182 sr = xmalloc (sizeof(*sr) * num);
183 for (i = 0; i<num; i++)
188 sort_info = sset->sort_info;
193 for (i = 0; i<num; i++)
195 position = positions[i];
196 if (position <= sort_info->num_entries)
198 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
199 sr[i].sysno = sort_info->entries[position-1]->sysno;
200 sr[i].score = sort_info->entries[position-1]->score;
204 /* did we really get all entries using sort ? */
205 for (i = 0; i<num; i++)
210 if (i < num) /* nope, get the rest, unsorted - sorry */
220 position = sort_info->num_entries;
221 while (num_i < num && positions[num_i] < position)
223 rfd = rset_open (rset, RSETF_READ);
224 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
226 if (key.sysno != psysno)
231 /* determine we alreay have this in our set */
232 for (i = sort_info->num_entries; --i >= 0; )
233 if (psysno == sort_info->entries[i]->sysno)
239 assert (num_i < num);
240 if (position == positions[num_i])
242 sr[num_i].sysno = psysno;
243 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
244 sr[num_i].score = -1;
249 rset_close (rset, rfd);
254 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
264 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
265 struct sortKey *criteria, int num_criteria,
268 struct zset_sort_entry this_entry;
269 struct zset_sort_entry *new_entry = NULL;
270 struct zset_sort_info *sort_info = sset->sort_info;
273 sortIdx_sysno (zh->sortIdx, sysno);
274 for (i = 0; i<num_criteria; i++)
276 sortIdx_type (zh->sortIdx, criteria[i].attrUse);
277 sortIdx_read (zh->sortIdx, this_entry.buf[i]);
279 i = sort_info->num_entries;
283 for (j = 0; j<num_criteria; j++)
285 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
292 if (criteria[j].relation == 'D')
297 else if (criteria[j].relation == 'A')
303 j = sort_info->max_entries-1;
307 new_entry = sort_info->entries[j];
310 sort_info->entries[j] = sort_info->entries[j-1];
313 sort_info->entries[j] = new_entry;
315 if (sort_info->num_entries != sort_info->max_entries)
316 (sort_info->num_entries)++;
317 for (i = 0; i<num_criteria; i++)
318 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
319 new_entry->sysno = sysno;
320 new_entry->score = -1;
323 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
324 int sysno, int score, int relation)
326 struct zset_sort_entry *new_entry = NULL;
329 i = sort_info->num_entries;
334 rel = score - sort_info->entries[i]->score;
341 else if (relation == 'A')
347 j = sort_info->max_entries-1;
351 new_entry = sort_info->entries[j];
354 sort_info->entries[j] = sort_info->entries[j-1];
357 sort_info->entries[j] = new_entry;
359 if (sort_info->num_entries != sort_info->max_entries)
360 (sort_info->num_entries)++;
361 new_entry->sysno = sysno;
362 new_entry->score = score;
365 void resultSetSort (ZebraHandle zh, ODR stream,
366 int num_input_setnames, char **input_setnames,
367 char *output_setname, Z_SortKeySpecList *sort_sequence,
374 struct sortKey sort_criteria[3];
379 if (num_input_setnames == 0)
384 if (num_input_setnames > 1)
389 sset = resultSetGet (zh, input_setnames[0]);
393 zh->errString = input_setnames[0];
396 if (!(rset = sset->rset))
399 zh->errString = input_setnames[0];
402 num_criteria = sort_sequence->num_specs;
403 if (num_criteria > 3)
405 for (i = 0; i < num_criteria; i++)
407 Z_SortKeySpec *sks = sort_sequence->specs[i];
410 if (*sks->sortRelation == Z_SortRelation_ascending)
411 sort_criteria[i].relation = 'A';
412 else if (*sks->sortRelation == Z_SortRelation_descending)
413 sort_criteria[i].relation = 'D';
419 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
424 else if (sks->sortElement->which != Z_SortElement_generic)
429 sk = sks->sortElement->u.generic;
432 case Z_SortKey_sortField:
433 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
436 case Z_SortKey_elementSpec:
437 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
440 case Z_SortKey_sortAttributes:
441 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
442 sort_criteria[i].attrUse =
443 zebra_maps_sort (zh->zebra_maps, sk->u.sortAttributes);
444 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
445 if (sort_criteria[i].attrUse == -1)
450 if (sortIdx_type (zh->sortIdx, sort_criteria[i].attrUse))
458 if (strcmp (output_setname, input_setnames[0]))
461 rset = rset_dup (rset);
462 sset = resultSetAdd (zh, output_setname, 1, rset, &hits);
464 sset->sort_info->num_entries = 0;
466 rfd = rset_open (rset, RSETF_READ);
467 while (rset_read (rset, rfd, &key, &term_index))
469 if (key.sysno != psysno)
472 resultSetInsertSort (zh, sset,
473 sort_criteria, num_criteria, psysno);
476 rset_close (rset, rfd);
479 *sort_status = Z_SortStatus_success;
484 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
488 if ((s = resultSetGet (zh, resultSetId)))
493 void resultSetRank (ZebraHandle zh, struct zset_sort_info *sort_info,
494 RSET rset, int *hits)
500 ZebraRankClass rank_class;
501 struct rank_control *rc;
503 sort_info->num_entries = 0;
505 rfd = rset_open (rset, RSETF_READ);
507 logf (LOG_DEBUG, "resultSetRank");
508 for (i = 0; i < rset->no_rset_terms; i++)
509 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
510 rset->rset_terms[i]->name,
511 rset->rset_terms[i]->nn,
512 rset->rset_terms[i]->flags);
514 rank_class = zebraRankLookup (zh, "rank-1");
515 rc = rank_class->control;
517 if (rset_read (rset, rfd, &key, &term_index))
519 int psysno = key.sysno;
522 (*rc->begin) (zh, rank_class->class_handle, rset);
527 if (key.sysno != psysno)
529 score = (*rc->calc) (handle, psysno);
531 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
535 (*rc->add) (handle, key.seqno, term_index);
537 while (rset_read (rset, rfd, &key, &term_index));
538 score = (*rc->calc) (handle, psysno);
539 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
540 (*rc->end) (zh, handle);
542 rset_close (rset, rfd);
543 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *hits);
546 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
548 ZebraRankClass p = zh->rank_classes;
549 while (p && strcmp (p->control->name, name))
551 if (p && !p->init_flag)
553 if (p->control->create)
554 p->class_handle = (*p->control->create)(zh);
560 void zebraRankInstall (ZebraHandle zh, struct rank_control *ctrl)
562 ZebraRankClass p = xmalloc (sizeof(*p));
563 p->control = xmalloc (sizeof(*p->control));
564 memcpy (p->control, ctrl, sizeof(*p->control));
565 p->control->name = xstrdup (ctrl->name);
567 p->next = zh->rank_classes;
568 zh->rank_classes = p;
571 void zebraRankDestroy (ZebraHandle zh)
573 ZebraRankClass p = zh->rank_classes;
576 ZebraRankClass p_next = p->next;
577 if (p->init_flag && p->control->destroy)
578 (*p->control->destroy)(zh, p->class_handle);
579 xfree (p->control->name);
584 zh->rank_classes = NULL;