2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.19 1995-10-06 10:43:56 adam
8 * Scan added. 'occurrences' in scan entries not set yet.
10 * Revision 1.18 1995/10/04 16:57:20 adam
11 * Key input and merge sort in one pass.
13 * Revision 1.17 1995/10/04 12:55:17 adam
14 * Bug fix in ranked search. Use=Any keys inserted.
16 * Revision 1.16 1995/10/02 16:24:40 adam
17 * Use attribute actually used in search requests.
19 * Revision 1.15 1995/10/02 15:18:52 adam
20 * New member in recRetrieveCtrl: diagnostic.
22 * Revision 1.14 1995/09/28 12:10:32 adam
23 * Bug fixes. Field prefix used in queries.
25 * Revision 1.13 1995/09/18 14:17:50 adam
28 * Revision 1.12 1995/09/15 14:45:21 adam
32 * Revision 1.11 1995/09/14 11:53:27 adam
33 * First work on regular expressions/truncations.
35 * Revision 1.10 1995/09/11 15:23:26 adam
36 * More work on relevance search.
38 * Revision 1.9 1995/09/11 13:09:35 adam
39 * More work on relevance feedback.
41 * Revision 1.8 1995/09/08 14:52:27 adam
42 * Minor changes. Dictionary is lower case now.
44 * Revision 1.7 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 1.6 1995/09/06 16:11:18 adam
51 * Option: only one word key per file.
53 * Revision 1.5 1995/09/06 10:33:04 adam
54 * More work on present. Some log messages removed.
56 * Revision 1.4 1995/09/05 15:28:40 adam
57 * More work on search engine.
59 * Revision 1.3 1995/09/04 15:20:22 adam
62 * Revision 1.2 1995/09/04 12:33:43 adam
63 * Various cleanup. YAZ util used instead.
65 * Revision 1.1 1995/09/04 09:10:40 adam
66 * More work on index add/del/update.
67 * Merge sort implemented.
68 * Initial work on z39 server.
84 * attr_print: log attributes
86 static void attr_print (Z_AttributesPlusTerm *t)
89 for (of = 0; of < t->num_attributes; of++)
91 Z_AttributeElement *element;
92 element = t->attributeList[of];
94 switch (element->which)
96 case Z_AttributeValue_numeric:
97 logf (LOG_DEBUG, "attributeType=%d value=%d",
98 *element->attributeType,
99 *element->value.numeric);
101 case Z_AttributeValue_complex:
102 logf (LOG_DEBUG, "attributeType=%d complex",
103 *element->attributeType);
104 for (i = 0; i<element->value.complex->num_list; i++)
106 if (element->value.complex->list[i]->which ==
107 Z_StringOrNumeric_string)
108 logf (LOG_DEBUG, " string: '%s'",
109 element->value.complex->list[i]->u.string);
110 else if (element->value.complex->list[i]->which ==
111 Z_StringOrNumeric_numeric)
112 logf (LOG_DEBUG, " numeric: '%d'",
113 *element->value.complex->list[i]->u.numeric);
126 Z_AttributesPlusTerm *zapt;
129 static int attr_find (AttrType *src)
131 while (src->major < src->zapt->num_attributes)
133 Z_AttributeElement *element;
134 element = src->zapt->attributeList[src->major];
136 if (src->type == *element->attributeType)
138 switch (element->which)
140 case Z_AttributeValue_numeric:
142 return *element->value.numeric;
144 case Z_AttributeValue_complex:
145 if (src->minor >= element->value.complex->num_list ||
146 element->value.complex->list[src->minor]->which !=
147 Z_StringOrNumeric_numeric)
150 return *element->value.complex->list[src->minor-1]->u.numeric;
160 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
173 int (*cmp)(const void *p1, const void *p2);
180 static void heap_swap (struct trunc_info *ti, int i1, int i2)
184 memcpy (ti->swapbuf, ti->heap[i1], ti->keysize);
185 memcpy (ti->heap[i1], ti->heap[i2], ti->keysize);
186 memcpy (ti->heap[i2], ti->swapbuf, ti->keysize);
189 ti->indx[i1] = ti->indx[i2];
193 static void heap_delete (struct trunc_info *ti)
195 int cur = 1, child = 2;
197 assert (ti->heapnum > 0);
198 memcpy (ti->heap[1], ti->heap[ti->heapnum], ti->keysize);
199 ti->indx[1] = ti->indx[ti->heapnum--];
200 while (child <= ti->heapnum) {
201 if (child < ti->heapnum &&
202 (*ti->cmp)(ti->heap[child], ti->heap[1+child]) > 0)
204 if ((*ti->cmp)(ti->heap[cur], ti->heap[child]) > 0)
206 heap_swap (ti, cur, child);
215 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
219 cur = ++(ti->heapnum);
220 memcpy (ti->heap[cur], buf, ti->keysize);
221 ti->indx[cur] = indx;
223 while (parent && (*ti->cmp)(ti->heap[parent], ti->heap[cur]) > 0)
225 heap_swap (ti, cur, parent);
232 struct trunc_info *heap_init (int size, int key_size,
233 int (*cmp)(const void *p1, const void *p2))
235 struct trunc_info *ti = xmalloc (sizeof(*ti));
240 ti->keysize = key_size;
242 ti->indx = xmalloc (size * sizeof(*ti->indx));
243 ti->heap = xmalloc (size * sizeof(*ti->heap));
244 ti->swapbuf = xmalloc (ti->keysize);
245 ti->tmpbuf = xmalloc (ti->keysize);
246 ti->buf = xmalloc (size * ti->keysize);
247 for (i = size; --i >= 0; )
248 ti->heap[i] = ti->buf + ti->keysize * i;
252 static void heap_close (struct trunc_info *ti)
261 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int from, int to,
264 logf (LOG_DEBUG, "rset_trunc, range=%d-%d", from, to-1);
265 if (to - from > merge_chunk)
273 struct trunc_info *ti;
276 rset_temp_parms parms;
278 ispt = xmalloc (sizeof(*ispt) * (to-from));
279 parms.key_size = sizeof (struct it_key);
280 result = rset_create (rset_kind_temp, &parms);
281 rsfd = rset_open (result, 1);
283 ti = heap_init (to-from, sizeof(struct it_key),
285 for (i = to-from; --i >= 0; )
287 ispt[i] = is_position (isam, isam_p[from+i]);
288 if (is_readkey (ispt[i], ti->tmpbuf))
289 heap_insert (ti, ti->tmpbuf, i);
295 rset_write (result, rsfd, ti->heap[1]);
297 if (is_readkey (ispt[n], ti->tmpbuf))
298 heap_insert (ti, ti->tmpbuf, n);
300 for (i = to-from; --i >= 0; )
301 is_pt_free (ispt[i]);
302 rset_close (result, rsfd);
309 static ISAM_P *isam_p_buf = NULL;
310 static int isam_p_size = 0;
311 static int isam_p_indx;
313 static void add_isam_p (const char *info)
315 if (isam_p_indx == isam_p_size)
317 ISAM_P *new_isam_p_buf;
319 isam_p_size = 2*isam_p_size + 100;
320 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
324 memcpy (new_isam_p_buf, isam_p_buf,
325 isam_p_indx * sizeof(*isam_p_buf));
328 isam_p_buf = new_isam_p_buf;
330 assert (*info == sizeof(*isam_p_buf));
331 memcpy (isam_p_buf + isam_p_indx, info+1, sizeof(*isam_p_buf));
335 static int grep_handle (Dict_char *name, const char *info)
337 logf (LOG_DEBUG, "dict name: %s", name);
342 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
343 const char *term_sub, ISAM_P **isam_ps)
345 char term_dict[2*IT_MAX_WORD+2];
349 int truncation_value;
353 attr_init (&use, zapt, 1);
354 use_value = attr_find (&use);
355 logf (LOG_DEBUG, "use value %d", use_value);
356 attr_init (&truncation, zapt, 5);
357 truncation_value = attr_find (&truncation);
358 logf (LOG_DEBUG, "truncation value %d", truncation_value);
362 i = index_word_prefix (term_dict, 1, use_value);
364 switch (truncation_value)
366 case -1: /* not specified */
367 case 100: /* do not truncate */
368 strcat (term_dict, term_sub);
369 logf (LOG_DEBUG, "dict_lookup: %s", term_dict);
370 if ((info = dict_lookup (zi->wordDict, term_dict)))
373 case 1: /* right truncation */
374 strcat (term_dict, term_sub);
375 strcat (term_dict, ".*");
376 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
378 case 2: /* left truncation */
379 case 3: /* left&right truncation */
382 case 101: /* process # in term */
383 for (j = strlen(term_dict), i = 0; term_sub[i] && i < 2; i++)
384 term_dict[j++] = term_sub[i];
385 for (; term_sub[i]; i++)
386 if (term_sub[i] == '#')
388 term_dict[j++] = '.';
389 term_dict[j++] = '*';
392 term_dict[j++] = term_sub[i];
394 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
396 case 102: /* regular expression */
397 strcat (term_dict, term_sub);
398 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
401 *isam_ps = isam_p_buf;
402 logf (LOG_DEBUG, "%d positions", isam_p_indx);
406 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
410 Z_Term *term = zapt->term;
412 sizez = term->u.general->len;
413 if (sizez > IT_MAX_WORD)
415 for (i = 0; i < sizez; i++)
416 termz[i] = index_char_cvt (term->u.general->buf[i]);
420 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
421 Z_AttributesPlusTerm *zapt)
423 rset_relevance_parms parms;
424 char termz[IT_MAX_WORD+1];
425 char term_sub[IT_MAX_WORD+1];
426 char *p0 = termz, *p1 = NULL;
427 Z_Term *term = zapt->term;
429 parms.key_size = sizeof(struct it_key);
431 parms.cmp = key_compare;
432 parms.is = zi->wordIsam;
434 if (term->which != Z_Term_general)
439 trans_term (zi, zapt, termz);
440 isam_p_indx = 0; /* global, set by trunc_term - see below */
443 if ((p1 = strchr (p0, ' ')))
445 memcpy (term_sub, p0, p1-p0);
446 term_sub[p1-p0] = '\0';
449 strcpy (term_sub, p0);
450 if (trunc_term (zi, zapt, term_sub, &parms.isam_positions))
456 parms.no_isam_positions = isam_p_indx;
458 return rset_create (rset_kind_relevance, &parms);
460 return rset_create (rset_kind_null, NULL);
463 static RSET rpn_search_APT_word (ZServerInfo *zi,
464 Z_AttributesPlusTerm *zapt)
466 ISAM_P *isam_positions;
467 rset_isam_parms parms;
469 char termz[IT_MAX_WORD+1];
470 Z_Term *term = zapt->term;
472 if (term->which != Z_Term_general)
477 trans_term (zi, zapt, termz);
478 isam_p_indx = 0; /* global, set by trunc_term - see below */
479 if (trunc_term (zi, zapt, termz, &isam_positions))
482 return rset_create (rset_kind_null, NULL);
483 else if (isam_p_indx == 1)
485 parms.is = zi->wordIsam;
486 parms.pos = *isam_positions;
487 return rset_create (rset_kind_isam, &parms);
490 return rset_trunc (zi->wordIsam, isam_positions, 0, isam_p_indx, 400);
493 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
494 Z_AttributesPlusTerm *zapt)
496 ISAM_P *isam_positions;
497 rset_isam_parms parms;
499 char termz[IT_MAX_WORD+1];
500 Z_Term *term = zapt->term;
502 if (term->which != Z_Term_general)
507 trans_term (zi, zapt, termz);
508 isam_p_indx = 0; /* global, set by trunc_term - see below */
509 if (trunc_term (zi, zapt, termz, &isam_positions))
511 if (isam_p_indx != 1)
512 return rset_create (rset_kind_null, NULL);
513 parms.is = zi->wordIsam;
514 parms.pos = *isam_positions;
515 return rset_create (rset_kind_isam, &parms);
518 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt)
522 int relation_value, structure_value;
524 attr_init (&relation, zapt, 2);
525 attr_init (&structure, zapt, 4);
527 relation_value = attr_find (&relation);
528 structure_value = attr_find (&structure);
529 switch (structure_value)
532 if (relation_value == 102) /* relevance relation */
533 return rpn_search_APT_relevance (zi, zapt);
534 return rpn_search_APT_word (zi, zapt);
536 if (relation_value == 102) /* relevance relation */
537 return rpn_search_APT_relevance (zi, zapt);
538 return rpn_search_APT_phrase (zi, zapt);
541 if (relation_value == 102) /* relevance relation */
542 return rpn_search_APT_relevance (zi, zapt);
543 return rpn_search_APT_word (zi, zapt);
548 case 5: /* date - normalized */
550 case 6: /* word list */
551 return rpn_search_APT_relevance (zi, zapt);
552 case 100: /* date - un-normalized */
554 case 101: /* name - normalized */
556 case 102: /* date - un-normalized */
558 case 103: /* structure */
562 case 105: /* free-form-text */
563 return rpn_search_APT_relevance (zi, zapt);
564 case 106: /* document-text */
565 return rpn_search_APT_relevance (zi, zapt);
566 case 107: /* local-number */
568 case 108: /* string */
569 return rpn_search_APT_word (zi, zapt);
570 case 109: /* numeric string */
577 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
581 if (!(s = resultSetGet (zi, resultSetId)))
582 return rset_create (rset_kind_null, NULL);
586 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs)
589 if (zs->which == Z_RPNStructure_complex)
591 rset_bool_parms bool_parms;
593 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1);
594 if (bool_parms.rset_l == NULL)
596 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2);
597 if (bool_parms.rset_r == NULL)
599 rset_delete (bool_parms.rset_l);
602 bool_parms.key_size = sizeof(struct it_key);
603 bool_parms.cmp = key_compare;
605 switch (zs->u.complex->operator->which)
608 r = rset_create (rset_kind_and, &bool_parms);
611 r = rset_create (rset_kind_or, &bool_parms);
613 case Z_Operator_and_not:
614 r = rset_create (rset_kind_not, &bool_parms);
620 else if (zs->which == Z_RPNStructure_simple)
622 if (zs->u.simple->which == Z_Operand_APT)
624 logf (LOG_DEBUG, "rpn_search_APT");
625 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm);
627 else if (zs->u.simple->which == Z_Operand_resultSetId)
629 logf (LOG_DEBUG, "rpn_search_ref");
630 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
644 static void count_set (RSET r, int *count)
650 logf (LOG_DEBUG, "rpn_save_set");
652 rfd = rset_open (r, 0);
653 while (rset_read (r, rfd, &key))
655 if (key.sysno != psysno)
662 logf (LOG_DEBUG, "%d distinct sysnos", *count);
665 int rpn_search (ZServerInfo *zi,
666 Z_RPNQuery *rpn, int num_bases, char **basenames,
667 const char *setname, int *hits)
672 zi->errString = NULL;
673 rset = rpn_search_structure (zi, rpn->RPNStructure);
676 count_set (rset, hits);
677 resultSetAdd (zi, setname, 1, rset);
681 static struct scan_entry *scan_list;
683 static int scan_before, scan_after;
684 static int scan_prefix;
686 static int scan_handle (Dict_char *name, const char *info, int pos)
691 idx = scan_after - pos + scan_before;
694 scan_list[idx].term = odr_malloc (scan_odr, strlen(name + scan_prefix)+1);
695 strcpy (scan_list[idx].term, name + scan_prefix);
696 scan_list[idx].occurrences = 1;
697 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
701 int rpn_scan (ZServerInfo *zi, ODR odr, Z_AttributesPlusTerm *zapt,
702 int *position, int *num_entries, struct scan_entry **list)
706 int num = *num_entries;
709 char termz[IT_MAX_WORD+20];
712 Z_Term *term = zapt->term;
714 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
715 scan_before = before = pos-1;
716 scan_after = after = 1+num-pos;
719 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
721 scan_list = *list = odr_malloc (odr, (before+after)*sizeof(**list));
722 attr_init (&use, zapt, 1);
723 use_value = attr_find (&use);
724 logf (LOG_DEBUG, "use value %d", use_value);
728 scan_prefix = i = index_word_prefix (termz, 1, use_value);
729 sizez = term->u.general->len;
730 if (sizez > IT_MAX_WORD)
732 for (j = 0; j<sizez; j++)
733 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
736 dict_scan (zi->wordDict, termz, &before, &after, scan_handle);