2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.35 1995-11-27 09:29:00 adam
8 * Bug fixes regarding conversion to regular expressions.
10 * Revision 1.34 1995/11/16 17:00:56 adam
11 * Better logging of rpn query.
13 * Revision 1.33 1995/11/01 13:58:28 quinn
14 * Moving data1 to yaz/retrieval
16 * Revision 1.32 1995/10/27 14:00:11 adam
17 * Implemented detection of database availability.
19 * Revision 1.31 1995/10/17 18:02:10 adam
20 * New feature: databases. Implemented as prefix to words in dictionary.
22 * Revision 1.30 1995/10/16 09:32:38 adam
23 * More work on relational op.
25 * Revision 1.29 1995/10/13 16:01:49 adam
28 * Revision 1.28 1995/10/13 12:26:43 adam
29 * Optimization of truncation.
31 * Revision 1.27 1995/10/12 17:07:22 adam
34 * Revision 1.26 1995/10/12 12:40:54 adam
35 * Bug fixes in rpn_prox.
37 * Revision 1.25 1995/10/10 13:59:24 adam
38 * Function rset_open changed its wflag parameter to general flags.
40 * Revision 1.24 1995/10/09 16:18:37 adam
41 * Function dict_lookup_grep got extra client data parameter.
43 * Revision 1.23 1995/10/06 16:33:37 adam
44 * Use attribute mappings.
46 * Revision 1.22 1995/10/06 15:07:39 adam
47 * Structure 'local-number' handled.
49 * Revision 1.21 1995/10/06 13:52:06 adam
50 * Bug fixes. Handler may abort further scanning.
52 * Revision 1.20 1995/10/06 11:06:33 adam
53 * Scan entries include 'occurrences' now.
55 * Revision 1.19 1995/10/06 10:43:56 adam
56 * Scan added. 'occurrences' in scan entries not set yet.
58 * Revision 1.18 1995/10/04 16:57:20 adam
59 * Key input and merge sort in one pass.
61 * Revision 1.17 1995/10/04 12:55:17 adam
62 * Bug fix in ranked search. Use=Any keys inserted.
64 * Revision 1.16 1995/10/02 16:24:40 adam
65 * Use attribute actually used in search requests.
67 * Revision 1.15 1995/10/02 15:18:52 adam
68 * New member in recRetrieveCtrl: diagnostic.
70 * Revision 1.14 1995/09/28 12:10:32 adam
71 * Bug fixes. Field prefix used in queries.
73 * Revision 1.13 1995/09/18 14:17:50 adam
76 * Revision 1.12 1995/09/15 14:45:21 adam
80 * Revision 1.11 1995/09/14 11:53:27 adam
81 * First work on regular expressions/truncations.
83 * Revision 1.10 1995/09/11 15:23:26 adam
84 * More work on relevance search.
86 * Revision 1.9 1995/09/11 13:09:35 adam
87 * More work on relevance feedback.
89 * Revision 1.8 1995/09/08 14:52:27 adam
90 * Minor changes. Dictionary is lower case now.
92 * Revision 1.7 1995/09/07 13:58:36 adam
93 * New parameter: result-set file descriptor (RSFD) to support multiple
94 * positions within the same result-set.
95 * Boolean operators: and, or, not implemented.
96 * Result-set references.
98 * Revision 1.6 1995/09/06 16:11:18 adam
99 * Option: only one word key per file.
101 * Revision 1.5 1995/09/06 10:33:04 adam
102 * More work on present. Some log messages removed.
104 * Revision 1.4 1995/09/05 15:28:40 adam
105 * More work on search engine.
107 * Revision 1.3 1995/09/04 15:20:22 adam
110 * Revision 1.2 1995/09/04 12:33:43 adam
111 * Various cleanup. YAZ util used instead.
113 * Revision 1.1 1995/09/04 09:10:40 adam
114 * More work on index add/del/update.
115 * Merge sort implemented.
116 * Initial work on z39 server.
124 #include "attribute.h"
132 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
137 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
138 attp = att_getentbyatt (attrSet, attrUse);
141 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
142 return index_word_prefix (string, attp->attset_ordinal,
143 attp->local_attribute, basename);
150 Z_AttributesPlusTerm *zapt;
153 static int attr_find (AttrType *src, oid_value *attributeSetP)
155 while (src->major < src->zapt->num_attributes)
157 Z_AttributeElement *element;
159 element = src->zapt->attributeList[src->major];
160 if (src->type == *element->attributeType)
162 switch (element->which)
164 case Z_AttributeValue_numeric:
166 if (element->attributeSet && attributeSetP)
170 attrset = oid_getentbyoid (element->attributeSet);
171 *attributeSetP = attrset->value;
173 return *element->value.numeric;
175 case Z_AttributeValue_complex:
176 if (src->minor >= element->value.complex->num_list ||
177 element->value.complex->list[src->minor]->which !=
178 Z_StringOrNumeric_numeric)
181 if (element->attributeSet && attributeSetP)
185 attrset = oid_getentbyoid (element->attributeSet);
186 *attributeSetP = attrset->value;
188 return *element->value.complex->list[src->minor-1]->u.numeric;
198 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
212 int (*cmp)(const void *p1, const void *p2);
219 static void heap_swap (struct trunc_info *ti, int i1, int i2)
224 ti->ptr[i1] = ti->ptr[i2];
228 static void heap_delete (struct trunc_info *ti)
230 int cur = 1, child = 2;
232 heap_swap (ti, 1, ti->heapnum--);
233 while (child <= ti->heapnum) {
234 if (child < ti->heapnum &&
235 (*ti->cmp)(ti->heap[ti->ptr[child]],
236 ti->heap[ti->ptr[1+child]]) > 0)
238 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
239 ti->heap[ti->ptr[child]]) > 0)
241 heap_swap (ti, cur, child);
250 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
254 cur = ++(ti->heapnum);
255 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
256 ti->indx[ti->ptr[cur]] = indx;
258 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
259 ti->heap[ti->ptr[cur]]) > 0)
261 heap_swap (ti, cur, parent);
268 struct trunc_info *heap_init (int size, int key_size,
269 int (*cmp)(const void *p1, const void *p2))
271 struct trunc_info *ti = xmalloc (sizeof(*ti));
276 ti->keysize = key_size;
278 ti->indx = xmalloc (size * sizeof(*ti->indx));
279 ti->heap = xmalloc (size * sizeof(*ti->heap));
280 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
281 ti->swapbuf = xmalloc (ti->keysize);
282 ti->tmpbuf = xmalloc (ti->keysize);
283 ti->buf = xmalloc (size * ti->keysize);
284 for (i = size; --i >= 0; )
287 ti->heap[i] = ti->buf + ti->keysize * i;
292 static void heap_close (struct trunc_info *ti)
302 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
307 rset_temp_parms parms;
309 parms.key_size = sizeof(struct it_key);
310 result = rset_create (rset_kind_temp, &parms);
311 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
313 if (to - from > merge_chunk)
317 int i, i_add = (to-from)/merge_chunk + 1;
318 struct trunc_info *ti;
320 int rsmax = (to-from)/i_add + 1;
322 rset = xmalloc (sizeof(*rset) * rsmax);
323 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
325 for (i = from; i < to; i += i_add)
328 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
331 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
335 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
336 for (i = rscur; --i >= 0; )
338 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
339 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
340 heap_insert (ti, ti->tmpbuf, i);
343 rset_close (rset[i], rsfd[i]);
344 rset_delete (rset[i]);
349 int n = ti->indx[ti->ptr[1]];
351 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
355 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
358 rset_close (rset[n], rsfd[n]);
359 rset_delete (rset[n]);
362 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
365 heap_insert (ti, ti->tmpbuf, n);
378 struct trunc_info *ti;
380 ispt = xmalloc (sizeof(*ispt) * (to-from));
382 ti = heap_init (to-from, sizeof(struct it_key),
384 for (i = to-from; --i >= 0; )
386 ispt[i] = is_position (isam, isam_p[from+i]);
387 if (is_readkey (ispt[i], ti->tmpbuf))
388 heap_insert (ti, ti->tmpbuf, i);
390 is_pt_free (ispt[i]);
394 int n = ti->indx[ti->ptr[1]];
396 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
398 /* section that preserve all keys */
400 if (is_readkey (ispt[n], ti->tmpbuf))
401 heap_insert (ti, ti->tmpbuf, n);
403 is_pt_free (ispt[n]);
405 /* section that preserve all keys with unique sysnos */
408 if (!is_readkey (ispt[n], ti->tmpbuf))
411 is_pt_free (ispt[n]);
414 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
417 heap_insert (ti, ti->tmpbuf, n);
426 rset_close (result, result_rsfd);
430 static int isam_trunc_cmp (const void *p1, const void *p2)
432 ISAM_P i1 = *(ISAM_P*) p1;
433 ISAM_P i2 = *(ISAM_P*) p2;
436 d = is_type (i1) - is_type (i2);
439 return is_block (i1) - is_block (i2);
442 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
445 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
446 return rset_trunc_r (isam, isam_p, 0, no, 100);
455 static void add_isam_p (const char *info, struct grep_info *p)
457 if (p->isam_p_indx == p->isam_p_size)
459 ISAM_P *new_isam_p_buf;
461 p->isam_p_size = 2*p->isam_p_size + 100;
462 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
466 memcpy (new_isam_p_buf, p->isam_p_buf,
467 p->isam_p_indx * sizeof(*p->isam_p_buf));
468 xfree (p->isam_p_buf);
470 p->isam_p_buf = new_isam_p_buf;
472 assert (*info == sizeof(*p->isam_p_buf));
473 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
477 static int grep_handle (Dict_char *name, const char *info, void *p)
479 logf (LOG_DEBUG, "dict name: %s", name);
480 add_isam_p (info, p);
484 static void gen_regular_rel (char *dst, int val, int islt)
492 sprintf (numstr, "%d", val);
493 for (w = strlen(numstr); --w >= 0; pos++)
512 strcpy (dst + dst_p, numstr);
513 dst_p = strlen(dst) - pos - 1;
541 for (i = 0; i<pos; i++)
554 for (i=1; i<pos; i++)
555 strcat (dst, "[0-9]?");
559 for (i = 0; i <= pos; i++)
560 strcat (dst, "[0-9]");
561 strcat (dst, "[0-9]*");
566 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
567 const char *term_sub,
569 oid_value attributeSet,
570 struct grep_info *grep_info,
578 attr_init (&relation, zapt, 2);
579 relation_value = attr_find (&relation, NULL);
580 term_value = atoi (term_sub);
582 switch (relation_value)
587 logf (LOG_DEBUG, "Relation <");
588 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
593 logf (LOG_DEBUG, "Relation <=");
594 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
599 logf (LOG_DEBUG, "Relation >=");
600 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
605 logf (LOG_DEBUG, "Relation >");
606 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
611 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
612 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
615 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
616 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
620 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
621 const char *term_sub,
622 oid_value attributeSet, struct grep_info *grep_info,
623 int num_bases, char **basenames)
625 char term_dict[2*IT_MAX_WORD+2];
626 int i, j, r, base_no;
628 int truncation_value;
631 oid_value curAttributeSet = attributeSet;
633 attr_init (&use, zapt, 1);
634 use_value = attr_find (&use, &curAttributeSet);
635 logf (LOG_DEBUG, "use value %d", use_value);
636 attr_init (&truncation, zapt, 5);
637 truncation_value = attr_find (&truncation, NULL);
638 logf (LOG_DEBUG, "truncation value %d", truncation_value);
643 for (base_no = 0; base_no < num_bases; base_no++)
646 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
654 if (!relational_term (zi, zapt, term_sub, term_dict,
655 attributeSet, grep_info, &max_pos))
657 switch (truncation_value)
659 case -1: /* not specified */
660 case 100: /* do not truncate */
661 sprintf (term_dict + strlen(term_dict),
662 "([]%d %s)", strlen(term_sub), term_sub);
663 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
664 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
665 &max_pos, grep_handle);
667 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
669 case 1: /* right truncation */
670 sprintf (term_dict + strlen(term_dict),
671 "([]%d %s.*)", strlen(term_sub), term_sub);
672 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
673 &max_pos, grep_handle);
675 case 2: /* left truncation */
676 case 3: /* left&right truncation */
679 case 101: /* process # in term */
680 strcat (term_dict, "(");
681 j = strlen(term_dict);
682 for (i=0; term_sub[i]; i++)
683 if (i > 2 && term_sub[i] == '#')
685 term_dict[j++] = '.';
686 term_dict[j++] = '*';
690 term_dict[j++] = '\\';
691 term_dict[j++] = term_sub[i];
693 strcpy (term_dict+j, ")");
694 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
695 &max_pos, grep_handle);
697 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
700 case 102: /* regular expression */
701 sprintf (term_dict + strlen(term_dict), "(%s)", term_sub);
702 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
703 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
704 &max_pos, grep_handle);
706 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
711 if (max_pos <= strlen(basenames[base_no]))
713 zi->errCode = 109; /* Database unavailable */
714 zi->errString = basenames[base_no];
718 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
722 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
726 Z_Term *term = zapt->term;
728 sizez = term->u.general->len;
729 if (sizez > IT_MAX_WORD)
731 for (i = 0; i < sizez; i++)
732 termz[i] = index_char_cvt (term->u.general->buf[i]);
736 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
737 Z_AttributesPlusTerm *zapt,
738 oid_value attributeSet,
739 int num_bases, char **basenames)
741 rset_relevance_parms parms;
742 char termz[IT_MAX_WORD+1];
743 char term_sub[IT_MAX_WORD+1];
744 struct grep_info grep_info;
745 char *p0 = termz, *p1 = NULL;
748 parms.key_size = sizeof(struct it_key);
750 parms.cmp = key_compare;
751 parms.is = zi->wordIsam;
753 if (zapt->term->which != Z_Term_general)
758 trans_term (zi, zapt, termz);
759 grep_info.isam_p_indx = 0;
760 grep_info.isam_p_size = 0;
761 grep_info.isam_p_buf = NULL;
764 if ((p1 = strchr (p0, ' ')))
766 memcpy (term_sub, p0, p1-p0);
767 term_sub[p1-p0] = '\0';
770 strcpy (term_sub, p0);
771 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
772 num_bases, basenames))
780 parms.isam_positions = grep_info.isam_p_buf;
781 parms.no_isam_positions = grep_info.isam_p_indx;
782 if (grep_info.isam_p_indx > 0)
783 result = rset_create (rset_kind_relevance, &parms);
785 result = rset_create (rset_kind_null, NULL);
786 xfree (grep_info.isam_p_buf);
790 static RSET rpn_search_APT_word (ZServerInfo *zi,
791 Z_AttributesPlusTerm *zapt,
792 oid_value attributeSet,
793 int num_bases, char **basenames)
795 rset_isam_parms parms;
796 char termz[IT_MAX_WORD+1];
797 struct grep_info grep_info;
800 if (zapt->term->which != Z_Term_general)
805 trans_term (zi, zapt, termz);
807 grep_info.isam_p_indx = 0;
808 grep_info.isam_p_size = 0;
809 grep_info.isam_p_buf = NULL;
811 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
812 num_bases, basenames))
814 if (grep_info.isam_p_indx < 1)
815 result = rset_create (rset_kind_null, NULL);
816 else if (grep_info.isam_p_indx == 1)
818 parms.is = zi->wordIsam;
819 parms.pos = *grep_info.isam_p_buf;
820 result = rset_create (rset_kind_isam, &parms);
823 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
824 grep_info.isam_p_indx);
825 xfree (grep_info.isam_p_buf);
829 static RSET rpn_prox (RSET *rset, int rset_no)
837 rset_temp_parms parms;
839 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
840 more = xmalloc (sizeof(*more)*rset_no);
841 buf = xmalloc (sizeof(*buf)*rset_no);
843 for (i = 0; i<rset_no; i++)
845 buf[i] = xmalloc (sizeof(**buf));
846 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
847 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
851 rset_close (rset[i], rsfd[i]);
858 return rset_create (rset_kind_null, NULL);
861 parms.key_size = sizeof (struct it_key);
862 result = rset_create (rset_kind_temp, &parms);
863 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
867 for (i = 1; i<rset_no; i++)
876 cmp = key_compare (buf[i], buf[i-1]);
879 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
884 if (buf[i-1]->seqno+1 != buf[i]->seqno)
886 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
892 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
898 rset_write (result, rsfd_result, buf[0]);
899 more[0] = rset_read (*rset, *rsfd, *buf);
903 for (i = 0; i<rset_no; i++)
905 rset_close (rset[i], rsfd[i]);
908 rset_close (result, rsfd_result);
915 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
916 Z_AttributesPlusTerm *zapt,
917 oid_value attributeSet,
918 int num_bases, char **basenames)
920 char termz[IT_MAX_WORD+1];
921 char term_sub[IT_MAX_WORD+1];
922 char *p0 = termz, *p1 = NULL;
923 RSET rset[60], result;
925 struct grep_info grep_info;
927 if (zapt->term->which != Z_Term_general)
932 trans_term (zi, zapt, termz);
934 grep_info.isam_p_size = 0;
935 grep_info.isam_p_buf = NULL;
939 if ((p1 = strchr (p0, ' ')))
941 memcpy (term_sub, p0, p1-p0);
942 term_sub[p1-p0] = '\0';
945 strcpy (term_sub, p0);
947 grep_info.isam_p_indx = 0;
948 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
949 num_bases, basenames))
951 if (grep_info.isam_p_indx == 0)
952 rset[rset_no] = rset_create (rset_kind_null, NULL);
953 else if (grep_info.isam_p_indx > 1)
954 rset[rset_no] = rset_trunc (zi->wordIsam,
955 grep_info.isam_p_buf,
956 grep_info.isam_p_indx);
959 rset_isam_parms parms;
961 parms.is = zi->wordIsam;
962 parms.pos = *grep_info.isam_p_buf;
963 rset[rset_no] = rset_create (rset_kind_isam, &parms);
965 assert (rset[rset_no]);
966 if (++rset_no >= sizeof(rset)/sizeof(*rset))
974 xfree (grep_info.isam_p_buf);
976 return rset_create (rset_kind_null, NULL);
977 else if (rset_no == 1)
980 result = rpn_prox (rset, rset_no);
981 for (i = 0; i<rset_no; i++)
982 rset_delete (rset[i]);
986 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
987 oid_value attributeSet)
992 rset_temp_parms parms;
993 char termz[IT_MAX_WORD+1];
995 if (zapt->term->which != Z_Term_general)
1000 parms.key_size = sizeof (struct it_key);
1001 result = rset_create (rset_kind_temp, &parms);
1002 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1004 trans_term (zi, zapt, termz);
1005 key.sysno = atoi (termz);
1008 rset_write (result, rsfd, &key);
1009 rset_close (result, rsfd);
1013 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1014 oid_value attributeSet,
1015 int num_bases, char **basenames)
1019 int relation_value, structure_value;
1021 attr_init (&relation, zapt, 2);
1022 attr_init (&structure, zapt, 4);
1024 relation_value = attr_find (&relation, NULL);
1025 structure_value = attr_find (&structure, NULL);
1026 switch (structure_value)
1029 if (relation_value == 102) /* relevance relation */
1030 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1031 num_bases, basenames);
1032 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1033 num_bases, basenames);
1034 case 1: /* phrase */
1035 if (relation_value == 102) /* relevance relation */
1036 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1037 num_bases, basenames);
1038 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1039 num_bases, basenames);
1042 if (relation_value == 102) /* relevance relation */
1043 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1044 num_bases, basenames);
1045 return rpn_search_APT_word (zi, zapt, attributeSet,
1046 num_bases, basenames);
1051 case 5: /* date - normalized */
1053 case 6: /* word list */
1054 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1055 num_bases, basenames);
1056 case 100: /* date - un-normalized */
1058 case 101: /* name - normalized */
1060 case 102: /* date - un-normalized */
1062 case 103: /* structure */
1066 case 105: /* free-form-text */
1067 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1068 num_bases, basenames);
1069 case 106: /* document-text */
1070 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1071 num_bases, basenames);
1072 case 107: /* local-number */
1073 return rpn_search_APT_local (zi, zapt, attributeSet);
1074 case 108: /* string */
1075 return rpn_search_APT_word (zi, zapt, attributeSet,
1076 num_bases, basenames);
1077 case 109: /* numeric string */
1084 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1088 if (!(s = resultSetGet (zi, resultSetId)))
1089 return rset_create (rset_kind_null, NULL);
1093 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1094 oid_value attributeSet,
1095 int num_bases, char **basenames)
1098 if (zs->which == Z_RPNStructure_complex)
1100 rset_bool_parms bool_parms;
1102 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1104 num_bases, basenames);
1105 if (bool_parms.rset_l == NULL)
1107 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1109 num_bases, basenames);
1110 if (bool_parms.rset_r == NULL)
1112 rset_delete (bool_parms.rset_l);
1115 bool_parms.key_size = sizeof(struct it_key);
1116 bool_parms.cmp = key_compare;
1118 switch (zs->u.complex->operator->which)
1120 case Z_Operator_and:
1121 r = rset_create (rset_kind_and, &bool_parms);
1124 r = rset_create (rset_kind_or, &bool_parms);
1126 case Z_Operator_and_not:
1127 r = rset_create (rset_kind_not, &bool_parms);
1133 else if (zs->which == Z_RPNStructure_simple)
1135 if (zs->u.simple->which == Z_Operand_APT)
1137 logf (LOG_DEBUG, "rpn_search_APT");
1138 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1139 attributeSet, num_bases, basenames);
1141 else if (zs->u.simple->which == Z_Operand_resultSetId)
1143 logf (LOG_DEBUG, "rpn_search_ref");
1144 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1158 static void count_set (RSET r, int *count)
1164 logf (LOG_DEBUG, "rpn_save_set");
1166 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1167 while (rset_read (r, rfd, &key))
1169 if (key.sysno != psysno)
1175 rset_close (r, rfd);
1176 logf (LOG_DEBUG, "%d distinct sysnos", *count);
1179 int rpn_search (ZServerInfo *zi,
1180 Z_RPNQuery *rpn, int num_bases, char **basenames,
1181 const char *setname, int *hits)
1185 oid_value attributeSet;
1190 zi->errString = NULL;
1192 attrset = oid_getentbyoid (rpn->attributeSetId);
1193 attributeSet = attrset->value;
1194 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1195 num_bases, basenames);
1198 count_set (rset, hits);
1199 resultSetAdd (zi, setname, 1, rset);
1201 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1206 struct scan_entry *list;
1213 static int scan_handle (Dict_char *name, const char *info, int pos,
1216 int len_prefix, idx;
1219 struct scan_info *scan_info = client;
1221 rset_isam_parms parms;
1223 len_prefix = strlen(scan_info->prefix);
1224 if (memcmp (name, scan_info->prefix, len_prefix))
1227 idx = scan_info->after - pos + scan_info->before;
1230 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1231 strlen(name + len_prefix)+1);
1232 strcpy (scan_info->list[idx].term, name + len_prefix);
1233 assert (*info == sizeof(isam_p));
1234 memcpy (&isam_p, info+1, sizeof(isam_p));
1235 parms.is = scan_info->isam;
1238 rset = rset_create (rset_kind_isam, &parms);
1239 count_set (rset, &scan_info->list[idx].occurrences);
1242 scan_info->list[idx].occurrences = 1;
1244 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1249 static int dummy_handle (Dict_char *name, const char *info, void *p)
1254 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1255 int num_bases, char **basenames,
1256 int *position, int *num_entries, struct scan_entry **list,
1259 int i, j, sizez, max_pos;
1260 int pos = *position;
1261 int num = *num_entries;
1264 char termz[IT_MAX_WORD+20];
1267 Z_Term *term = zapt->term;
1268 struct scan_info scan_info;
1270 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1274 scan_info.before = before = pos-1;
1275 scan_info.after = after = 1+num-pos;
1276 scan_info.odr = zi->odr;
1278 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1280 scan_info.isam = zi->wordIsam;
1281 scan_info.list = odr_malloc (zi->odr, (before+after)*
1282 sizeof(*scan_info.list));
1283 for (j = 0; j<before+after; j++)
1284 scan_info.list[j].term = NULL;
1285 attr_init (&use, zapt, 1);
1286 use_value = attr_find (&use, NULL);
1287 logf (LOG_DEBUG, "use value %d", use_value);
1289 if (use_value == -1)
1291 i = index_word_prefix (termz, 1, use_value, *basenames);
1293 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
1295 if (max_pos <= strlen(*basenames))
1297 zi->errString = *basenames;
1298 return zi->errCode = 109; /* Database unavailable */
1300 strcpy (scan_info.prefix, termz);
1301 sizez = term->u.general->len;
1302 if (sizez > IT_MAX_WORD)
1303 sizez = IT_MAX_WORD;
1304 for (j = 0; j<sizez; j++)
1305 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1308 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1310 *status = BEND_SCAN_SUCCESS;
1312 for (i = 0; i<scan_info.after; i++)
1313 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1317 *status = BEND_SCAN_PARTIAL;
1319 for (i = 0; i<scan_info.before; i++)
1320 if (scan_info.list[i].term)
1323 *status = BEND_SCAN_PARTIAL;
1327 *list = scan_info.list+i; /* list is set to first 'real' entry */
1329 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1330 zi->errCode = 114; /* if no entries was found */
1331 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1332 *position, *num_entries);
1334 logf (LOG_DEBUG, "scan error: %d", zi->errCode);