2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.30 1995-10-16 09:32:38 adam
8 * More work on relational op.
10 * Revision 1.29 1995/10/13 16:01:49 adam
13 * Revision 1.28 1995/10/13 12:26:43 adam
14 * Optimization of truncation.
16 * Revision 1.27 1995/10/12 17:07:22 adam
19 * Revision 1.26 1995/10/12 12:40:54 adam
20 * Bug fixes in rpn_prox.
22 * Revision 1.25 1995/10/10 13:59:24 adam
23 * Function rset_open changed its wflag parameter to general flags.
25 * Revision 1.24 1995/10/09 16:18:37 adam
26 * Function dict_lookup_grep got extra client data parameter.
28 * Revision 1.23 1995/10/06 16:33:37 adam
29 * Use attribute mappings.
31 * Revision 1.22 1995/10/06 15:07:39 adam
32 * Structure 'local-number' handled.
34 * Revision 1.21 1995/10/06 13:52:06 adam
35 * Bug fixes. Handler may abort further scanning.
37 * Revision 1.20 1995/10/06 11:06:33 adam
38 * Scan entries include 'occurrences' now.
40 * Revision 1.19 1995/10/06 10:43:56 adam
41 * Scan added. 'occurrences' in scan entries not set yet.
43 * Revision 1.18 1995/10/04 16:57:20 adam
44 * Key input and merge sort in one pass.
46 * Revision 1.17 1995/10/04 12:55:17 adam
47 * Bug fix in ranked search. Use=Any keys inserted.
49 * Revision 1.16 1995/10/02 16:24:40 adam
50 * Use attribute actually used in search requests.
52 * Revision 1.15 1995/10/02 15:18:52 adam
53 * New member in recRetrieveCtrl: diagnostic.
55 * Revision 1.14 1995/09/28 12:10:32 adam
56 * Bug fixes. Field prefix used in queries.
58 * Revision 1.13 1995/09/18 14:17:50 adam
61 * Revision 1.12 1995/09/15 14:45:21 adam
65 * Revision 1.11 1995/09/14 11:53:27 adam
66 * First work on regular expressions/truncations.
68 * Revision 1.10 1995/09/11 15:23:26 adam
69 * More work on relevance search.
71 * Revision 1.9 1995/09/11 13:09:35 adam
72 * More work on relevance feedback.
74 * Revision 1.8 1995/09/08 14:52:27 adam
75 * Minor changes. Dictionary is lower case now.
77 * Revision 1.7 1995/09/07 13:58:36 adam
78 * New parameter: result-set file descriptor (RSFD) to support multiple
79 * positions within the same result-set.
80 * Boolean operators: and, or, not implemented.
81 * Result-set references.
83 * Revision 1.6 1995/09/06 16:11:18 adam
84 * Option: only one word key per file.
86 * Revision 1.5 1995/09/06 10:33:04 adam
87 * More work on present. Some log messages removed.
89 * Revision 1.4 1995/09/05 15:28:40 adam
90 * More work on search engine.
92 * Revision 1.3 1995/09/04 15:20:22 adam
95 * Revision 1.2 1995/09/04 12:33:43 adam
96 * Various cleanup. YAZ util used instead.
98 * Revision 1.1 1995/09/04 09:10:40 adam
99 * More work on index add/del/update.
100 * Merge sort implemented.
101 * Initial work on z39 server.
109 #include <attribute.h>
117 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse)
121 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
122 attp = att_getentbyatt (attrSet, attrUse);
125 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
126 return index_word_prefix (string, attp->attset_ordinal,
127 attp->local_attribute);
131 * attr_print: log attributes
133 static void attr_print (Z_AttributesPlusTerm *t)
136 for (of = 0; of < t->num_attributes; of++)
138 Z_AttributeElement *element;
139 element = t->attributeList[of];
141 switch (element->which)
143 case Z_AttributeValue_numeric:
144 logf (LOG_DEBUG, "attributeType=%d value=%d",
145 *element->attributeType,
146 *element->value.numeric);
148 case Z_AttributeValue_complex:
149 logf (LOG_DEBUG, "attributeType=%d complex",
150 *element->attributeType);
151 for (i = 0; i<element->value.complex->num_list; i++)
153 if (element->value.complex->list[i]->which ==
154 Z_StringOrNumeric_string)
155 logf (LOG_DEBUG, " string: '%s'",
156 element->value.complex->list[i]->u.string);
157 else if (element->value.complex->list[i]->which ==
158 Z_StringOrNumeric_numeric)
159 logf (LOG_DEBUG, " numeric: '%d'",
160 *element->value.complex->list[i]->u.numeric);
173 Z_AttributesPlusTerm *zapt;
176 static int attr_find (AttrType *src, oid_value *attributeSetP)
178 while (src->major < src->zapt->num_attributes)
180 Z_AttributeElement *element;
182 element = src->zapt->attributeList[src->major];
183 if (src->type == *element->attributeType)
185 switch (element->which)
187 case Z_AttributeValue_numeric:
189 if (element->attributeSet && attributeSetP)
193 attrset = oid_getentbyoid (element->attributeSet);
194 *attributeSetP = attrset->value;
196 return *element->value.numeric;
198 case Z_AttributeValue_complex:
199 if (src->minor >= element->value.complex->num_list ||
200 element->value.complex->list[src->minor]->which !=
201 Z_StringOrNumeric_numeric)
204 if (element->attributeSet && attributeSetP)
208 attrset = oid_getentbyoid (element->attributeSet);
209 *attributeSetP = attrset->value;
211 return *element->value.complex->list[src->minor-1]->u.numeric;
221 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
235 int (*cmp)(const void *p1, const void *p2);
242 static void heap_swap (struct trunc_info *ti, int i1, int i2)
247 ti->ptr[i1] = ti->ptr[i2];
251 static void heap_delete (struct trunc_info *ti)
253 int cur = 1, child = 2;
255 heap_swap (ti, 1, ti->heapnum--);
256 while (child <= ti->heapnum) {
257 if (child < ti->heapnum &&
258 (*ti->cmp)(ti->heap[ti->ptr[child]],
259 ti->heap[ti->ptr[1+child]]) > 0)
261 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
262 ti->heap[ti->ptr[child]]) > 0)
264 heap_swap (ti, cur, child);
273 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
277 cur = ++(ti->heapnum);
278 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
279 ti->indx[ti->ptr[cur]] = indx;
281 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
282 ti->heap[ti->ptr[cur]]) > 0)
284 heap_swap (ti, cur, parent);
291 struct trunc_info *heap_init (int size, int key_size,
292 int (*cmp)(const void *p1, const void *p2))
294 struct trunc_info *ti = xmalloc (sizeof(*ti));
299 ti->keysize = key_size;
301 ti->indx = xmalloc (size * sizeof(*ti->indx));
302 ti->heap = xmalloc (size * sizeof(*ti->heap));
303 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
304 ti->swapbuf = xmalloc (ti->keysize);
305 ti->tmpbuf = xmalloc (ti->keysize);
306 ti->buf = xmalloc (size * ti->keysize);
307 for (i = size; --i >= 0; )
310 ti->heap[i] = ti->buf + ti->keysize * i;
315 static void heap_close (struct trunc_info *ti)
325 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
330 rset_temp_parms parms;
332 parms.key_size = sizeof(struct it_key);
333 result = rset_create (rset_kind_temp, &parms);
334 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
336 if (to - from > merge_chunk)
340 int i, i_add = (to-from)/merge_chunk + 1;
341 struct trunc_info *ti;
343 int rsmax = (to-from)/i_add + 1;
345 rset = xmalloc (sizeof(*rset) * rsmax);
346 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
348 for (i = from; i < to; i += i_add)
351 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
354 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
358 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
359 for (i = rscur; --i >= 0; )
361 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
362 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
363 heap_insert (ti, ti->tmpbuf, i);
366 rset_close (rset[i], rsfd[i]);
367 rset_delete (rset[i]);
372 int n = ti->indx[ti->ptr[1]];
374 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
378 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
381 rset_close (rset[n], rsfd[n]);
382 rset_delete (rset[n]);
385 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
388 heap_insert (ti, ti->tmpbuf, n);
401 struct trunc_info *ti;
403 ispt = xmalloc (sizeof(*ispt) * (to-from));
405 ti = heap_init (to-from, sizeof(struct it_key),
407 for (i = to-from; --i >= 0; )
409 ispt[i] = is_position (isam, isam_p[from+i]);
410 if (is_readkey (ispt[i], ti->tmpbuf))
411 heap_insert (ti, ti->tmpbuf, i);
413 is_pt_free (ispt[i]);
417 int n = ti->indx[ti->ptr[1]];
419 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
421 /* section that preserve all keys */
423 if (is_readkey (ispt[n], ti->tmpbuf))
424 heap_insert (ti, ti->tmpbuf, n);
426 is_pt_free (ispt[n]);
428 /* section that preserve all keys with unique sysnos */
431 if (!is_readkey (ispt[n], ti->tmpbuf))
434 is_pt_free (ispt[n]);
437 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
440 heap_insert (ti, ti->tmpbuf, n);
449 rset_close (result, result_rsfd);
453 static int isam_trunc_cmp (const void *p1, const void *p2)
455 ISAM_P i1 = *(ISAM_P*) p1;
456 ISAM_P i2 = *(ISAM_P*) p2;
459 d = is_type (i1) - is_type (i2);
462 return is_block (i1) - is_block (i2);
465 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
468 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
469 return rset_trunc_r (isam, isam_p, 0, no, 100);
478 static void add_isam_p (const char *info, struct grep_info *p)
480 if (p->isam_p_indx == p->isam_p_size)
482 ISAM_P *new_isam_p_buf;
484 p->isam_p_size = 2*p->isam_p_size + 100;
485 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
489 memcpy (new_isam_p_buf, p->isam_p_buf,
490 p->isam_p_indx * sizeof(*p->isam_p_buf));
491 xfree (p->isam_p_buf);
493 p->isam_p_buf = new_isam_p_buf;
495 assert (*info == sizeof(*p->isam_p_buf));
496 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
500 static int grep_handle (Dict_char *name, const char *info, void *p)
502 logf (LOG_DEBUG, "dict name: %s", name);
503 add_isam_p (info, p);
507 static void gen_regular_ge (char *dst, int val)
517 sprintf (numstr, "%d", val);
518 for (w = strlen(numstr); --w >= 0; pos++)
528 strcpy (dst + dst_p, numstr);
529 dst_p = strlen(dst) - pos - 1;
542 for (i = 0; i<pos; i++)
553 for (i = 0; i <= pos; i++)
554 strcat (dst, "[0-9]");
555 strcat (dst, "[0-9]*)");
558 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
559 const char *term_sub,
560 oid_value attributeSet, struct grep_info *grep_info)
562 char term_dict[10*IT_MAX_WORD+2];
566 int truncation_value;
571 oid_value curAttributeSet = attributeSet;
573 attr_init (&use, zapt, 1);
574 use_value = attr_find (&use, &curAttributeSet);
575 logf (LOG_DEBUG, "use value %d", use_value);
576 attr_init (&truncation, zapt, 5);
577 truncation_value = attr_find (&truncation, NULL);
578 logf (LOG_DEBUG, "truncation value %d", truncation_value);
580 attr_init (&relation, zapt, 2);
581 relation_value = attr_find (&relation, NULL);
585 i = index_word_prefix_map (term_dict, curAttributeSet, use_value);
591 switch (relation_value)
597 logf (LOG_LOG, "Relation ge");
598 gen_regular_ge (term_dict + strlen(term_dict), atoi(term_sub));
599 logf (LOG_LOG, "dict_lookup_grep: %s", term_dict);
600 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
603 logf (LOG_WARN, "dict_lookup_grep fail, rel=ge: %d", r);
604 logf (LOG_LOG, "%d positions", grep_info->isam_p_indx);
607 logf (LOG_LOG, "Relation gt");
608 gen_regular_ge (term_dict + strlen(term_dict), atoi(term_sub)+1);
609 logf (LOG_LOG, "dict_lookup_grep: %s", term_dict);
610 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
613 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
614 logf (LOG_LOG, "%d positions", grep_info->isam_p_indx);
617 switch (truncation_value)
619 case -1: /* not specified */
620 case 100: /* do not truncate */
621 strcat (term_dict, term_sub);
622 logf (LOG_DEBUG, "dict_lookup: %s", term_dict);
623 if ((info = dict_lookup (zi->wordDict, term_dict)))
624 add_isam_p (info, grep_info);
626 case 1: /* right truncation */
627 strcat (term_dict, term_sub);
628 strcat (term_dict, ".*");
629 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, grep_handle);
631 case 2: /* left truncation */
632 case 3: /* left&right truncation */
635 case 101: /* process # in term */
636 for (j = strlen(term_dict), i = 0; term_sub[i] && i < 2; i++)
637 term_dict[j++] = term_sub[i];
638 for (; term_sub[i]; i++)
639 if (term_sub[i] == '#')
641 term_dict[j++] = '.';
642 term_dict[j++] = '*';
645 term_dict[j++] = term_sub[i];
647 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, grep_handle);
649 case 102: /* regular expression */
650 strcat (term_dict, "(");
651 strcat (term_dict, term_sub);
652 strcat (term_dict, ")");
653 logf (LOG_LOG, "dict_lookup_grep: %s", term_dict);
654 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
657 logf (LOG_WARN, "dict_lookup_grep fail, truncation=regular: %d", r);
660 logf (LOG_LOG, "%d positions", grep_info->isam_p_indx);
664 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
668 Z_Term *term = zapt->term;
670 sizez = term->u.general->len;
671 if (sizez > IT_MAX_WORD)
673 for (i = 0; i < sizez; i++)
674 termz[i] = index_char_cvt (term->u.general->buf[i]);
678 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
679 Z_AttributesPlusTerm *zapt,
680 oid_value attributeSet)
682 rset_relevance_parms parms;
683 char termz[IT_MAX_WORD+1];
684 char term_sub[IT_MAX_WORD+1];
685 struct grep_info grep_info;
686 char *p0 = termz, *p1 = NULL;
689 parms.key_size = sizeof(struct it_key);
691 parms.cmp = key_compare;
692 parms.is = zi->wordIsam;
694 if (zapt->term->which != Z_Term_general)
699 trans_term (zi, zapt, termz);
700 grep_info.isam_p_indx = 0;
701 grep_info.isam_p_size = 0;
702 grep_info.isam_p_buf = NULL;
705 if ((p1 = strchr (p0, ' ')))
707 memcpy (term_sub, p0, p1-p0);
708 term_sub[p1-p0] = '\0';
711 strcpy (term_sub, p0);
712 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info))
720 parms.isam_positions = grep_info.isam_p_buf;
721 parms.no_isam_positions = grep_info.isam_p_indx;
722 if (grep_info.isam_p_indx > 0)
723 result = rset_create (rset_kind_relevance, &parms);
725 result = rset_create (rset_kind_null, NULL);
726 xfree (grep_info.isam_p_buf);
730 static RSET rpn_search_APT_word (ZServerInfo *zi,
731 Z_AttributesPlusTerm *zapt,
732 oid_value attributeSet)
734 rset_isam_parms parms;
735 char termz[IT_MAX_WORD+1];
736 struct grep_info grep_info;
739 if (zapt->term->which != Z_Term_general)
744 trans_term (zi, zapt, termz);
746 grep_info.isam_p_indx = 0;
747 grep_info.isam_p_size = 0;
748 grep_info.isam_p_buf = NULL;
750 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info))
752 if (grep_info.isam_p_indx < 1)
753 result = rset_create (rset_kind_null, NULL);
754 else if (grep_info.isam_p_indx == 1)
756 parms.is = zi->wordIsam;
757 parms.pos = *grep_info.isam_p_buf;
758 result = rset_create (rset_kind_isam, &parms);
761 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
762 grep_info.isam_p_indx);
763 xfree (grep_info.isam_p_buf);
767 static RSET rpn_prox (RSET *rset, int rset_no)
775 rset_temp_parms parms;
777 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
778 more = xmalloc (sizeof(*more)*rset_no);
779 buf = xmalloc (sizeof(*buf)*rset_no);
781 for (i = 0; i<rset_no; i++)
783 buf[i] = xmalloc (sizeof(**buf));
784 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
785 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
789 rset_close (rset[i], rsfd[i]);
796 return rset_create (rset_kind_null, NULL);
799 parms.key_size = sizeof (struct it_key);
800 result = rset_create (rset_kind_temp, &parms);
801 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
805 for (i = 1; i<rset_no; i++)
814 cmp = key_compare (buf[i], buf[i-1]);
817 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
822 if (buf[i-1]->seqno+1 != buf[i]->seqno)
824 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
830 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
836 rset_write (result, rsfd_result, buf[0]);
837 more[0] = rset_read (*rset, *rsfd, *buf);
841 for (i = 0; i<rset_no; i++)
843 rset_close (rset[i], rsfd[i]);
846 rset_close (result, rsfd_result);
853 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
854 Z_AttributesPlusTerm *zapt,
855 oid_value attributeSet)
857 char termz[IT_MAX_WORD+1];
858 char term_sub[IT_MAX_WORD+1];
859 char *p0 = termz, *p1 = NULL;
860 RSET rset[60], result;
862 struct grep_info grep_info;
864 if (zapt->term->which != Z_Term_general)
869 trans_term (zi, zapt, termz);
871 grep_info.isam_p_size = 0;
872 grep_info.isam_p_buf = NULL;
876 if ((p1 = strchr (p0, ' ')))
878 memcpy (term_sub, p0, p1-p0);
879 term_sub[p1-p0] = '\0';
882 strcpy (term_sub, p0);
884 grep_info.isam_p_indx = 0;
885 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info))
887 if (grep_info.isam_p_indx == 0)
888 rset[rset_no] = rset_create (rset_kind_null, NULL);
889 else if (grep_info.isam_p_indx > 1)
890 rset[rset_no] = rset_trunc (zi->wordIsam,
891 grep_info.isam_p_buf,
892 grep_info.isam_p_indx);
895 rset_isam_parms parms;
897 parms.is = zi->wordIsam;
898 parms.pos = *grep_info.isam_p_buf;
899 rset[rset_no] = rset_create (rset_kind_isam, &parms);
901 assert (rset[rset_no]);
902 if (++rset_no >= sizeof(rset)/sizeof(*rset))
910 xfree (grep_info.isam_p_buf);
912 return rset_create (rset_kind_null, NULL);
913 else if (rset_no == 1)
916 result = rpn_prox (rset, rset_no);
917 for (i = 0; i<rset_no; i++)
918 rset_delete (rset[i]);
922 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
923 oid_value attributeSet)
928 rset_temp_parms parms;
929 char termz[IT_MAX_WORD+1];
931 if (zapt->term->which != Z_Term_general)
936 parms.key_size = sizeof (struct it_key);
937 result = rset_create (rset_kind_temp, &parms);
938 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
940 trans_term (zi, zapt, termz);
941 key.sysno = atoi (termz);
944 rset_write (result, rsfd, &key);
945 rset_close (result, rsfd);
949 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
950 oid_value attributeSet)
954 int relation_value, structure_value;
956 attr_init (&relation, zapt, 2);
957 attr_init (&structure, zapt, 4);
959 relation_value = attr_find (&relation, NULL);
960 structure_value = attr_find (&structure, NULL);
961 switch (structure_value)
964 if (relation_value == 102) /* relevance relation */
965 return rpn_search_APT_relevance (zi, zapt, attributeSet);
966 return rpn_search_APT_phrase (zi, zapt, attributeSet);
968 if (relation_value == 102) /* relevance relation */
969 return rpn_search_APT_relevance (zi, zapt, attributeSet);
970 return rpn_search_APT_phrase (zi, zapt, attributeSet);
973 if (relation_value == 102) /* relevance relation */
974 return rpn_search_APT_relevance (zi, zapt, attributeSet);
975 return rpn_search_APT_word (zi, zapt, attributeSet);
980 case 5: /* date - normalized */
982 case 6: /* word list */
983 return rpn_search_APT_relevance (zi, zapt, attributeSet);
984 case 100: /* date - un-normalized */
986 case 101: /* name - normalized */
988 case 102: /* date - un-normalized */
990 case 103: /* structure */
994 case 105: /* free-form-text */
995 return rpn_search_APT_relevance (zi, zapt, attributeSet);
996 case 106: /* document-text */
997 return rpn_search_APT_relevance (zi, zapt, attributeSet);
998 case 107: /* local-number */
999 return rpn_search_APT_local (zi, zapt, attributeSet);
1000 case 108: /* string */
1001 return rpn_search_APT_word (zi, zapt, attributeSet);
1002 case 109: /* numeric string */
1009 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1013 if (!(s = resultSetGet (zi, resultSetId)))
1014 return rset_create (rset_kind_null, NULL);
1018 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1019 oid_value attributeSet)
1022 if (zs->which == Z_RPNStructure_complex)
1024 rset_bool_parms bool_parms;
1026 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1028 if (bool_parms.rset_l == NULL)
1030 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1032 if (bool_parms.rset_r == NULL)
1034 rset_delete (bool_parms.rset_l);
1037 bool_parms.key_size = sizeof(struct it_key);
1038 bool_parms.cmp = key_compare;
1040 switch (zs->u.complex->operator->which)
1042 case Z_Operator_and:
1043 r = rset_create (rset_kind_and, &bool_parms);
1046 r = rset_create (rset_kind_or, &bool_parms);
1048 case Z_Operator_and_not:
1049 r = rset_create (rset_kind_not, &bool_parms);
1055 else if (zs->which == Z_RPNStructure_simple)
1057 if (zs->u.simple->which == Z_Operand_APT)
1059 logf (LOG_DEBUG, "rpn_search_APT");
1060 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1063 else if (zs->u.simple->which == Z_Operand_resultSetId)
1065 logf (LOG_DEBUG, "rpn_search_ref");
1066 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1080 static void count_set (RSET r, int *count)
1086 logf (LOG_DEBUG, "rpn_save_set");
1088 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1089 while (rset_read (r, rfd, &key))
1091 if (key.sysno != psysno)
1097 rset_close (r, rfd);
1098 logf (LOG_DEBUG, "%d distinct sysnos", *count);
1101 int rpn_search (ZServerInfo *zi,
1102 Z_RPNQuery *rpn, int num_bases, char **basenames,
1103 const char *setname, int *hits)
1107 oid_value attributeSet;
1110 zi->errString = NULL;
1112 attrset = oid_getentbyoid (rpn->attributeSetId);
1113 attributeSet = attrset->value;
1115 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet);
1118 count_set (rset, hits);
1119 resultSetAdd (zi, setname, 1, rset);
1121 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1126 struct scan_entry *list;
1133 static int scan_handle (Dict_char *name, const char *info, int pos,
1136 int len_prefix, idx;
1139 struct scan_info *scan_info = client;
1141 rset_isam_parms parms;
1143 len_prefix = strlen(scan_info->prefix);
1144 if (memcmp (name, scan_info->prefix, len_prefix))
1147 idx = scan_info->after - pos + scan_info->before;
1150 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1151 strlen(name + len_prefix)+1);
1152 strcpy (scan_info->list[idx].term, name + len_prefix);
1153 assert (*info == sizeof(isam_p));
1154 memcpy (&isam_p, info+1, sizeof(isam_p));
1155 parms.is = scan_info->isam;
1158 rset = rset_create (rset_kind_isam, &parms);
1159 count_set (rset, &scan_info->list[idx].occurrences);
1162 scan_info->list[idx].occurrences = 1;
1164 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1168 int rpn_scan (ZServerInfo *zi, ODR odr, Z_AttributesPlusTerm *zapt,
1169 int *position, int *num_entries, struct scan_entry **list,
1173 int pos = *position;
1174 int num = *num_entries;
1177 char termz[IT_MAX_WORD+20];
1180 Z_Term *term = zapt->term;
1181 struct scan_info scan_info;
1183 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1184 scan_info.before = before = pos-1;
1185 scan_info.after = after = 1+num-pos;
1186 scan_info.odr = odr;
1188 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1190 scan_info.isam = zi->wordIsam;
1191 scan_info.list = odr_malloc (odr, (before+after)*sizeof(*scan_info.list));
1192 for (j = 0; j<before+after; j++)
1193 scan_info.list[j].term = NULL;
1194 attr_init (&use, zapt, 1);
1195 use_value = attr_find (&use, NULL);
1196 logf (LOG_DEBUG, "use value %d", use_value);
1198 if (use_value == -1)
1200 i = index_word_prefix (termz, 1, use_value);
1201 strcpy (scan_info.prefix, termz);
1202 sizez = term->u.general->len;
1203 if (sizez > IT_MAX_WORD)
1204 sizez = IT_MAX_WORD;
1205 for (j = 0; j<sizez; j++)
1206 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1209 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1211 *status = BEND_SCAN_SUCCESS;
1213 for (i = 0; i<scan_info.after; i++)
1214 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1218 *status = BEND_SCAN_PARTIAL;
1220 for (i = 0; i<scan_info.before; i++)
1221 if (scan_info.list[i].term)
1224 *status = BEND_SCAN_PARTIAL;
1228 *list = scan_info.list+i; /* list is set to first 'real' entry */
1230 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1231 zi->errCode = 114; /* if no entries was found */
1232 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1233 *position, *num_entries);
1235 logf (LOG_DEBUG, "scan error: %d", zi->errCode);