2 * Copyright (C) 1994-1998, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.15 1998-03-05 08:45:14 adam
8 * New result set model and modular ranking system. Moved towards
9 * descent server API. System information stored as "SGML" records.
11 * Revision 1.14 1998/02/10 16:39:15 adam
14 * Revision 1.13 1998/02/10 12:03:06 adam
17 * Revision 1.12 1997/09/25 14:57:36 adam
20 * Revision 1.11 1996/12/23 15:30:46 adam
22 * Bug fix: result sets weren't deleted after server shut down.
24 * Revision 1.10 1995/10/30 15:08:08 adam
27 * Revision 1.9 1995/10/17 18:02:14 adam
28 * New feature: databases. Implemented as prefix to words in dictionary.
30 * Revision 1.8 1995/10/10 13:59:25 adam
31 * Function rset_open changed its wflag parameter to general flags.
33 * Revision 1.7 1995/10/06 14:38:01 adam
34 * New result set method: r_score.
35 * Local no (sysno) and score is transferred to retrieveCtrl.
37 * Revision 1.6 1995/09/28 09:19:49 adam
38 * xfree/xmalloc used everywhere.
39 * Extract/retrieve method seems to work for text records.
41 * Revision 1.5 1995/09/27 16:17:32 adam
42 * More work on retrieve.
44 * Revision 1.4 1995/09/07 13:58:36 adam
45 * New parameter: result-set file descriptor (RSFD) to support multiple
46 * positions within the same result-set.
47 * Boolean operators: and, or, not implemented.
48 * Result-set references.
50 * Revision 1.3 1995/09/06 16:11:19 adam
51 * Option: only one word key per file.
53 * Revision 1.2 1995/09/06 10:33:04 adam
54 * More work on present. Some log messages removed.
56 * Revision 1.1 1995/09/05 15:28:40 adam
57 * More work on search engine.
71 #define SORT_IDX_ENTRYSIZE 64
72 #define ZSET_SORT_MAX_LEVEL 3
78 struct zset_sort_info *sort_info;
79 struct zebra_set *next;
82 struct zset_sort_entry {
85 char buf[ZSET_SORT_MAX_LEVEL][SORT_IDX_ENTRYSIZE];
88 struct zset_sort_info {
91 struct zset_sort_entry **entries;
94 void resultSetRank (ZebraHandle zh, struct zset_sort_info *si,
95 RSET rset, int *hits);
97 ZebraSet resultSetAdd (ZebraHandle zh, const char *name, int ov,
103 for (s = zh->sets; s; s = s->next)
104 if (!strcmp (s->name, name))
106 logf (LOG_DEBUG, "updating result set %s", name);
109 rset_delete (s->rset);
111 resultSetRank (zh, s->sort_info, rset, hits);
114 logf (LOG_DEBUG, "adding result set %s", name);
115 s = xmalloc (sizeof(*s));
118 s->name = xmalloc (strlen(name)+1);
119 strcpy (s->name, name);
122 s->sort_info = xmalloc (sizeof(*s->sort_info));
123 s->sort_info->max_entries = 1000;
124 s->sort_info->entries = xmalloc (sizeof(*s->sort_info->entries) *
125 s->sort_info->max_entries);
126 for (i = 0; i < s->sort_info->max_entries; i++)
127 s->sort_info->entries[i] = xmalloc (sizeof(**s->sort_info->entries));
128 resultSetRank (zh, s->sort_info, rset, hits);
132 ZebraSet resultSetGet (ZebraHandle zh, const char *name)
136 for (s = zh->sets; s; s = s->next)
137 if (!strcmp (s->name, name))
143 void resultSetDestroy (ZebraHandle zh)
147 for (s = zh->sets; s; s = s1)
152 for (i = 0; i < s->sort_info->max_entries; i++)
153 xfree (s->sort_info->entries[i]);
154 xfree (s->sort_info);
156 rset_delete (s->rset);
163 ZebraPosSet zebraPosSetCreate (ZebraHandle zh, const char *name,
164 int num, int *positions)
170 struct zset_sort_info *sort_info;
172 if (!(sset = resultSetGet (zh, name)))
174 if (!(rset = sset->rset))
176 sr = xmalloc (sizeof(*sr) * num);
177 for (i = 0; i<num; i++)
182 sort_info = sset->sort_info;
187 for (i = 0; i<num; i++)
189 position = positions[i];
190 if (position <= sort_info->num_entries)
192 logf (LOG_DEBUG, "got pos=%d (sorted)", position);
193 sr[i].sysno = sort_info->entries[position-1]->sysno;
194 sr[i].score = sort_info->entries[position-1]->score;
198 /* did we really get all entries using sort ? */
199 for (i = 0; i<num; i++)
204 if (i < num) /* nope, get the rest, unsorted - sorry */
214 position = sort_info->num_entries;
215 while (num_i < num && positions[num_i] < position)
217 rfd = rset_open (rset, RSETF_READ);
218 while (num_i < num && rset_read (rset, rfd, &key, &term_index))
220 if (key.sysno != psysno)
225 /* determine we alreay have this in our set */
226 for (i = sort_info->num_entries; --i >= 0; )
227 if (psysno == sort_info->entries[i]->sysno)
233 assert (num_i < num);
234 if (position == positions[num_i])
236 sr[num_i].sysno = psysno;
237 logf (LOG_DEBUG, "got pos=%d (unsorted)", position);
238 sr[num_i].score = -1;
243 rset_close (rset, rfd);
248 void zebraPosSetDestroy (ZebraHandle zh, ZebraPosSet records, int num)
258 void resultSetInsertSort (ZebraHandle zh, ZebraSet sset,
259 struct sortKey *criteria, int num_criteria,
262 struct zset_sort_entry this_entry;
263 struct zset_sort_entry *new_entry = NULL;
264 struct zset_sort_info *sort_info = sset->sort_info;
267 sortIdx_sysno (zh->sortIdx, sysno);
268 for (i = 0; i<num_criteria; i++)
270 sortIdx_type (zh->sortIdx, criteria[i].attrUse);
271 sortIdx_read (zh->sortIdx, this_entry.buf[i]);
273 i = sort_info->num_entries;
277 for (j = 0; j<num_criteria; j++)
279 rel = memcmp (this_entry.buf[j], sort_info->entries[i]->buf[j],
286 if (criteria[j].relation == 'D')
291 else if (criteria[j].relation == 'A')
297 j = sort_info->max_entries-1;
301 new_entry = sort_info->entries[j];
304 sort_info->entries[j] = sort_info->entries[j-1];
307 sort_info->entries[j] = new_entry;
309 if (sort_info->num_entries != sort_info->max_entries)
310 (sort_info->num_entries)++;
311 for (i = 0; i<num_criteria; i++)
312 memcpy (new_entry->buf[i], this_entry.buf[i], SORT_IDX_ENTRYSIZE);
313 new_entry->sysno = sysno;
314 new_entry->score = -1;
317 void resultSetInsertRank (ZebraHandle zh, struct zset_sort_info *sort_info,
318 int sysno, int score, int relation)
320 struct zset_sort_entry *new_entry = NULL;
323 i = sort_info->num_entries;
328 rel = score - sort_info->entries[i]->score;
335 else if (relation == 'A')
341 j = sort_info->max_entries-1;
345 new_entry = sort_info->entries[j];
348 sort_info->entries[j] = sort_info->entries[j-1];
351 sort_info->entries[j] = new_entry;
353 if (sort_info->num_entries != sort_info->max_entries)
354 (sort_info->num_entries)++;
355 new_entry->sysno = sysno;
356 new_entry->score = score;
359 void resultSetSort (ZebraHandle zh, ODR stream,
360 int num_input_setnames, char **input_setnames,
361 char *output_setname, Z_SortKeySpecList *sort_sequence,
368 struct sortKey sort_criteria[3];
373 if (num_input_setnames == 0)
378 if (num_input_setnames > 1)
383 sset = resultSetGet (zh, input_setnames[0]);
387 zh->errString = input_setnames[0];
390 if (!(rset = sset->rset))
393 zh->errString = input_setnames[0];
396 num_criteria = sort_sequence->num_specs;
397 if (num_criteria > 3)
399 for (i = 0; i < num_criteria; i++)
401 Z_SortKeySpec *sks = sort_sequence->specs[i];
404 if (*sks->sortRelation == Z_SortRelation_ascending)
405 sort_criteria[i].relation = 'A';
406 else if (*sks->sortRelation == Z_SortRelation_descending)
407 sort_criteria[i].relation = 'D';
413 if (sks->sortElement->which == Z_SortElement_databaseSpecific)
418 else if (sks->sortElement->which != Z_SortElement_generic)
423 sk = sks->sortElement->u.generic;
426 case Z_SortKey_sortField:
427 logf (LOG_DEBUG, "Sort: key %d is of type sortField", i+1);
430 case Z_SortKey_elementSpec:
431 logf (LOG_DEBUG, "Sort: key %d is of type elementSpec", i+1);
434 case Z_SortKey_sortAttributes:
435 logf (LOG_DEBUG, "Sort: key %d is of type sortAttributes", i+1);
436 sort_criteria[i].attrUse =
437 zebra_maps_sort (zh->zebra_maps, sk->u.sortAttributes);
438 logf (LOG_DEBUG, "use value = %d", sort_criteria[i].attrUse);
439 if (sort_criteria[i].attrUse == -1)
444 if (sortIdx_type (zh->sortIdx, sort_criteria[i].attrUse))
452 if (strcmp (output_setname, input_setnames[0]))
455 rset = rset_dup (rset);
456 sset = resultSetAdd (zh, output_setname, 1, rset, &hits);
458 sset->sort_info->num_entries = 0;
460 rfd = rset_open (rset, RSETF_READ);
461 while (rset_read (rset, rfd, &key, &term_index))
463 if (key.sysno != psysno)
466 resultSetInsertSort (zh, sset,
467 sort_criteria, num_criteria, psysno);
470 rset_close (rset, rfd);
473 *sort_status = Z_SortStatus_success;
478 RSET resultSetRef (ZebraHandle zh, Z_ResultSetId *resultSetId)
482 if ((s = resultSetGet (zh, resultSetId)))
487 void resultSetRank (ZebraHandle zh, struct zset_sort_info *sort_info,
488 RSET rset, int *hits)
494 ZebraRankClass rank_class;
495 struct rank_control *rc;
497 sort_info->num_entries = 0;
499 rfd = rset_open (rset, RSETF_READ);
501 logf (LOG_DEBUG, "resultSetRank");
502 for (i = 0; i < rset->no_rset_terms; i++)
503 logf (LOG_DEBUG, "term=\"%s\" cnt=%d type=%s",
504 rset->rset_terms[i]->name,
505 rset->rset_terms[i]->nn,
506 rset->rset_terms[i]->flags);
508 rank_class = zebraRankLookup (zh, "rank-1");
509 rc = rank_class->control;
511 if (rset_read (rset, rfd, &key, &term_index))
513 int psysno = key.sysno;
516 (*rc->begin) (zh, rank_class->class_handle, rset);
521 if (key.sysno != psysno)
523 score = (*rc->calc) (handle, psysno);
525 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
529 (*rc->add) (handle, key.seqno, term_index);
531 while (rset_read (rset, rfd, &key, &term_index));
532 score = (*rc->calc) (handle, psysno);
533 resultSetInsertRank (zh, sort_info, psysno, score, 'A');
534 (*rc->end) (zh, handle);
536 rset_close (rset, rfd);
537 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *hits);
540 ZebraRankClass zebraRankLookup (ZebraHandle zh, const char *name)
542 ZebraRankClass p = zh->rank_classes;
543 while (p && strcmp (p->control->name, name))
545 if (p && !p->init_flag)
547 if (p->control->create)
548 p->class_handle = (*p->control->create)(zh);
554 void zebraRankInstall (ZebraHandle zh, struct rank_control *ctrl)
556 ZebraRankClass p = xmalloc (sizeof(*p));
557 p->control = xmalloc (sizeof(*p->control));
558 memcpy (p->control, ctrl, sizeof(*p->control));
559 p->control->name = xstrdup (ctrl->name);
561 p->next = zh->rank_classes;
562 zh->rank_classes = p;
565 void zebraRankDestroy (ZebraHandle zh)
567 ZebraRankClass p = zh->rank_classes;
570 ZebraRankClass p_next = p->next;
571 if (p->init_flag && p->control->destroy)
572 (*p->control->destroy)(zh, p->class_handle);
573 xfree (p->control->name);
578 zh->rank_classes = NULL;