2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.34 1995-11-16 17:00:56 adam
8 * Better logging of rpn query.
10 * Revision 1.33 1995/11/01 13:58:28 quinn
11 * Moving data1 to yaz/retrieval
13 * Revision 1.32 1995/10/27 14:00:11 adam
14 * Implemented detection of database availability.
16 * Revision 1.31 1995/10/17 18:02:10 adam
17 * New feature: databases. Implemented as prefix to words in dictionary.
19 * Revision 1.30 1995/10/16 09:32:38 adam
20 * More work on relational op.
22 * Revision 1.29 1995/10/13 16:01:49 adam
25 * Revision 1.28 1995/10/13 12:26:43 adam
26 * Optimization of truncation.
28 * Revision 1.27 1995/10/12 17:07:22 adam
31 * Revision 1.26 1995/10/12 12:40:54 adam
32 * Bug fixes in rpn_prox.
34 * Revision 1.25 1995/10/10 13:59:24 adam
35 * Function rset_open changed its wflag parameter to general flags.
37 * Revision 1.24 1995/10/09 16:18:37 adam
38 * Function dict_lookup_grep got extra client data parameter.
40 * Revision 1.23 1995/10/06 16:33:37 adam
41 * Use attribute mappings.
43 * Revision 1.22 1995/10/06 15:07:39 adam
44 * Structure 'local-number' handled.
46 * Revision 1.21 1995/10/06 13:52:06 adam
47 * Bug fixes. Handler may abort further scanning.
49 * Revision 1.20 1995/10/06 11:06:33 adam
50 * Scan entries include 'occurrences' now.
52 * Revision 1.19 1995/10/06 10:43:56 adam
53 * Scan added. 'occurrences' in scan entries not set yet.
55 * Revision 1.18 1995/10/04 16:57:20 adam
56 * Key input and merge sort in one pass.
58 * Revision 1.17 1995/10/04 12:55:17 adam
59 * Bug fix in ranked search. Use=Any keys inserted.
61 * Revision 1.16 1995/10/02 16:24:40 adam
62 * Use attribute actually used in search requests.
64 * Revision 1.15 1995/10/02 15:18:52 adam
65 * New member in recRetrieveCtrl: diagnostic.
67 * Revision 1.14 1995/09/28 12:10:32 adam
68 * Bug fixes. Field prefix used in queries.
70 * Revision 1.13 1995/09/18 14:17:50 adam
73 * Revision 1.12 1995/09/15 14:45:21 adam
77 * Revision 1.11 1995/09/14 11:53:27 adam
78 * First work on regular expressions/truncations.
80 * Revision 1.10 1995/09/11 15:23:26 adam
81 * More work on relevance search.
83 * Revision 1.9 1995/09/11 13:09:35 adam
84 * More work on relevance feedback.
86 * Revision 1.8 1995/09/08 14:52:27 adam
87 * Minor changes. Dictionary is lower case now.
89 * Revision 1.7 1995/09/07 13:58:36 adam
90 * New parameter: result-set file descriptor (RSFD) to support multiple
91 * positions within the same result-set.
92 * Boolean operators: and, or, not implemented.
93 * Result-set references.
95 * Revision 1.6 1995/09/06 16:11:18 adam
96 * Option: only one word key per file.
98 * Revision 1.5 1995/09/06 10:33:04 adam
99 * More work on present. Some log messages removed.
101 * Revision 1.4 1995/09/05 15:28:40 adam
102 * More work on search engine.
104 * Revision 1.3 1995/09/04 15:20:22 adam
107 * Revision 1.2 1995/09/04 12:33:43 adam
108 * Various cleanup. YAZ util used instead.
110 * Revision 1.1 1995/09/04 09:10:40 adam
111 * More work on index add/del/update.
112 * Merge sort implemented.
113 * Initial work on z39 server.
121 #include "attribute.h"
129 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
134 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
135 attp = att_getentbyatt (attrSet, attrUse);
138 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
139 return index_word_prefix (string, attp->attset_ordinal,
140 attp->local_attribute, basename);
147 Z_AttributesPlusTerm *zapt;
150 static int attr_find (AttrType *src, oid_value *attributeSetP)
152 while (src->major < src->zapt->num_attributes)
154 Z_AttributeElement *element;
156 element = src->zapt->attributeList[src->major];
157 if (src->type == *element->attributeType)
159 switch (element->which)
161 case Z_AttributeValue_numeric:
163 if (element->attributeSet && attributeSetP)
167 attrset = oid_getentbyoid (element->attributeSet);
168 *attributeSetP = attrset->value;
170 return *element->value.numeric;
172 case Z_AttributeValue_complex:
173 if (src->minor >= element->value.complex->num_list ||
174 element->value.complex->list[src->minor]->which !=
175 Z_StringOrNumeric_numeric)
178 if (element->attributeSet && attributeSetP)
182 attrset = oid_getentbyoid (element->attributeSet);
183 *attributeSetP = attrset->value;
185 return *element->value.complex->list[src->minor-1]->u.numeric;
195 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
209 int (*cmp)(const void *p1, const void *p2);
216 static void heap_swap (struct trunc_info *ti, int i1, int i2)
221 ti->ptr[i1] = ti->ptr[i2];
225 static void heap_delete (struct trunc_info *ti)
227 int cur = 1, child = 2;
229 heap_swap (ti, 1, ti->heapnum--);
230 while (child <= ti->heapnum) {
231 if (child < ti->heapnum &&
232 (*ti->cmp)(ti->heap[ti->ptr[child]],
233 ti->heap[ti->ptr[1+child]]) > 0)
235 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
236 ti->heap[ti->ptr[child]]) > 0)
238 heap_swap (ti, cur, child);
247 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
251 cur = ++(ti->heapnum);
252 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
253 ti->indx[ti->ptr[cur]] = indx;
255 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
256 ti->heap[ti->ptr[cur]]) > 0)
258 heap_swap (ti, cur, parent);
265 struct trunc_info *heap_init (int size, int key_size,
266 int (*cmp)(const void *p1, const void *p2))
268 struct trunc_info *ti = xmalloc (sizeof(*ti));
273 ti->keysize = key_size;
275 ti->indx = xmalloc (size * sizeof(*ti->indx));
276 ti->heap = xmalloc (size * sizeof(*ti->heap));
277 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
278 ti->swapbuf = xmalloc (ti->keysize);
279 ti->tmpbuf = xmalloc (ti->keysize);
280 ti->buf = xmalloc (size * ti->keysize);
281 for (i = size; --i >= 0; )
284 ti->heap[i] = ti->buf + ti->keysize * i;
289 static void heap_close (struct trunc_info *ti)
299 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
304 rset_temp_parms parms;
306 parms.key_size = sizeof(struct it_key);
307 result = rset_create (rset_kind_temp, &parms);
308 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
310 if (to - from > merge_chunk)
314 int i, i_add = (to-from)/merge_chunk + 1;
315 struct trunc_info *ti;
317 int rsmax = (to-from)/i_add + 1;
319 rset = xmalloc (sizeof(*rset) * rsmax);
320 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
322 for (i = from; i < to; i += i_add)
325 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
328 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
332 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
333 for (i = rscur; --i >= 0; )
335 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
336 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
337 heap_insert (ti, ti->tmpbuf, i);
340 rset_close (rset[i], rsfd[i]);
341 rset_delete (rset[i]);
346 int n = ti->indx[ti->ptr[1]];
348 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
352 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
355 rset_close (rset[n], rsfd[n]);
356 rset_delete (rset[n]);
359 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
362 heap_insert (ti, ti->tmpbuf, n);
375 struct trunc_info *ti;
377 ispt = xmalloc (sizeof(*ispt) * (to-from));
379 ti = heap_init (to-from, sizeof(struct it_key),
381 for (i = to-from; --i >= 0; )
383 ispt[i] = is_position (isam, isam_p[from+i]);
384 if (is_readkey (ispt[i], ti->tmpbuf))
385 heap_insert (ti, ti->tmpbuf, i);
387 is_pt_free (ispt[i]);
391 int n = ti->indx[ti->ptr[1]];
393 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
395 /* section that preserve all keys */
397 if (is_readkey (ispt[n], ti->tmpbuf))
398 heap_insert (ti, ti->tmpbuf, n);
400 is_pt_free (ispt[n]);
402 /* section that preserve all keys with unique sysnos */
405 if (!is_readkey (ispt[n], ti->tmpbuf))
408 is_pt_free (ispt[n]);
411 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
414 heap_insert (ti, ti->tmpbuf, n);
423 rset_close (result, result_rsfd);
427 static int isam_trunc_cmp (const void *p1, const void *p2)
429 ISAM_P i1 = *(ISAM_P*) p1;
430 ISAM_P i2 = *(ISAM_P*) p2;
433 d = is_type (i1) - is_type (i2);
436 return is_block (i1) - is_block (i2);
439 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
442 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
443 return rset_trunc_r (isam, isam_p, 0, no, 100);
452 static void add_isam_p (const char *info, struct grep_info *p)
454 if (p->isam_p_indx == p->isam_p_size)
456 ISAM_P *new_isam_p_buf;
458 p->isam_p_size = 2*p->isam_p_size + 100;
459 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
463 memcpy (new_isam_p_buf, p->isam_p_buf,
464 p->isam_p_indx * sizeof(*p->isam_p_buf));
465 xfree (p->isam_p_buf);
467 p->isam_p_buf = new_isam_p_buf;
469 assert (*info == sizeof(*p->isam_p_buf));
470 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
474 static int grep_handle (Dict_char *name, const char *info, void *p)
476 logf (LOG_DEBUG, "dict name: %s", name);
477 add_isam_p (info, p);
481 static void gen_regular_rel (char *dst, int val, int islt)
489 sprintf (numstr, "%d", val);
490 for (w = strlen(numstr); --w >= 0; pos++)
509 strcpy (dst + dst_p, numstr);
510 dst_p = strlen(dst) - pos - 1;
538 for (i = 0; i<pos; i++)
551 for (i=1; i<pos; i++)
552 strcat (dst, "[0-9]?");
556 for (i = 0; i <= pos; i++)
557 strcat (dst, "[0-9]");
558 strcat (dst, "[0-9]*");
563 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
564 const char *term_sub,
566 oid_value attributeSet,
567 struct grep_info *grep_info,
575 attr_init (&relation, zapt, 2);
576 relation_value = attr_find (&relation, NULL);
577 term_value = atoi (term_sub);
579 switch (relation_value)
584 logf (LOG_DEBUG, "Relation <");
585 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
590 logf (LOG_DEBUG, "Relation <=");
591 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
596 logf (LOG_DEBUG, "Relation >=");
597 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
602 logf (LOG_DEBUG, "Relation >");
603 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
608 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
609 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
612 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
613 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
617 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
618 const char *term_sub,
619 oid_value attributeSet, struct grep_info *grep_info,
620 int num_bases, char **basenames)
622 char term_dict[2*IT_MAX_WORD+2];
623 int i, j, r, base_no;
625 int truncation_value;
628 oid_value curAttributeSet = attributeSet;
630 attr_init (&use, zapt, 1);
631 use_value = attr_find (&use, &curAttributeSet);
632 logf (LOG_DEBUG, "use value %d", use_value);
633 attr_init (&truncation, zapt, 5);
634 truncation_value = attr_find (&truncation, NULL);
635 logf (LOG_DEBUG, "truncation value %d", truncation_value);
640 for (base_no = 0; base_no < num_bases; base_no++)
643 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
651 if (!relational_term (zi, zapt, term_sub, term_dict,
652 attributeSet, grep_info, &max_pos))
654 switch (truncation_value)
656 case -1: /* not specified */
657 case 100: /* do not truncate */
658 strcat (term_dict, "(");
659 strcat (term_dict, term_sub);
660 strcat (term_dict, ")");
661 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
662 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
663 &max_pos, grep_handle);
665 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
667 case 1: /* right truncation */
668 strcat (term_dict, term_sub);
669 strcat (term_dict, ".*");
670 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
671 &max_pos, grep_handle);
673 case 2: /* left truncation */
674 case 3: /* left&right truncation */
677 case 101: /* process # in term */
678 for (j = strlen(term_dict), i = 0; term_sub[i] && i < 2; i++)
679 term_dict[j++] = term_sub[i];
680 for (; term_sub[i]; i++)
681 if (term_sub[i] == '#')
683 term_dict[j++] = '.';
684 term_dict[j++] = '*';
687 term_dict[j++] = term_sub[i];
689 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
690 &max_pos, grep_handle);
692 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
695 case 102: /* regular expression */
696 strcat (term_dict, "(");
697 strcat (term_dict, term_sub);
698 strcat (term_dict, ")");
699 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
700 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
701 &max_pos, grep_handle);
703 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
708 if (max_pos <= strlen(basenames[base_no]))
710 zi->errCode = 109; /* Database unavailable */
711 zi->errString = basenames[base_no];
715 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
719 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
723 Z_Term *term = zapt->term;
725 sizez = term->u.general->len;
726 if (sizez > IT_MAX_WORD)
728 for (i = 0; i < sizez; i++)
729 termz[i] = index_char_cvt (term->u.general->buf[i]);
733 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
734 Z_AttributesPlusTerm *zapt,
735 oid_value attributeSet,
736 int num_bases, char **basenames)
738 rset_relevance_parms parms;
739 char termz[IT_MAX_WORD+1];
740 char term_sub[IT_MAX_WORD+1];
741 struct grep_info grep_info;
742 char *p0 = termz, *p1 = NULL;
745 parms.key_size = sizeof(struct it_key);
747 parms.cmp = key_compare;
748 parms.is = zi->wordIsam;
750 if (zapt->term->which != Z_Term_general)
755 trans_term (zi, zapt, termz);
756 grep_info.isam_p_indx = 0;
757 grep_info.isam_p_size = 0;
758 grep_info.isam_p_buf = NULL;
761 if ((p1 = strchr (p0, ' ')))
763 memcpy (term_sub, p0, p1-p0);
764 term_sub[p1-p0] = '\0';
767 strcpy (term_sub, p0);
768 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
769 num_bases, basenames))
777 parms.isam_positions = grep_info.isam_p_buf;
778 parms.no_isam_positions = grep_info.isam_p_indx;
779 if (grep_info.isam_p_indx > 0)
780 result = rset_create (rset_kind_relevance, &parms);
782 result = rset_create (rset_kind_null, NULL);
783 xfree (grep_info.isam_p_buf);
787 static RSET rpn_search_APT_word (ZServerInfo *zi,
788 Z_AttributesPlusTerm *zapt,
789 oid_value attributeSet,
790 int num_bases, char **basenames)
792 rset_isam_parms parms;
793 char termz[IT_MAX_WORD+1];
794 struct grep_info grep_info;
797 if (zapt->term->which != Z_Term_general)
802 trans_term (zi, zapt, termz);
804 grep_info.isam_p_indx = 0;
805 grep_info.isam_p_size = 0;
806 grep_info.isam_p_buf = NULL;
808 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
809 num_bases, basenames))
811 if (grep_info.isam_p_indx < 1)
812 result = rset_create (rset_kind_null, NULL);
813 else if (grep_info.isam_p_indx == 1)
815 parms.is = zi->wordIsam;
816 parms.pos = *grep_info.isam_p_buf;
817 result = rset_create (rset_kind_isam, &parms);
820 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
821 grep_info.isam_p_indx);
822 xfree (grep_info.isam_p_buf);
826 static RSET rpn_prox (RSET *rset, int rset_no)
834 rset_temp_parms parms;
836 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
837 more = xmalloc (sizeof(*more)*rset_no);
838 buf = xmalloc (sizeof(*buf)*rset_no);
840 for (i = 0; i<rset_no; i++)
842 buf[i] = xmalloc (sizeof(**buf));
843 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
844 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
848 rset_close (rset[i], rsfd[i]);
855 return rset_create (rset_kind_null, NULL);
858 parms.key_size = sizeof (struct it_key);
859 result = rset_create (rset_kind_temp, &parms);
860 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
864 for (i = 1; i<rset_no; i++)
873 cmp = key_compare (buf[i], buf[i-1]);
876 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
881 if (buf[i-1]->seqno+1 != buf[i]->seqno)
883 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
889 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
895 rset_write (result, rsfd_result, buf[0]);
896 more[0] = rset_read (*rset, *rsfd, *buf);
900 for (i = 0; i<rset_no; i++)
902 rset_close (rset[i], rsfd[i]);
905 rset_close (result, rsfd_result);
912 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
913 Z_AttributesPlusTerm *zapt,
914 oid_value attributeSet,
915 int num_bases, char **basenames)
917 char termz[IT_MAX_WORD+1];
918 char term_sub[IT_MAX_WORD+1];
919 char *p0 = termz, *p1 = NULL;
920 RSET rset[60], result;
922 struct grep_info grep_info;
924 if (zapt->term->which != Z_Term_general)
929 trans_term (zi, zapt, termz);
931 grep_info.isam_p_size = 0;
932 grep_info.isam_p_buf = NULL;
936 if ((p1 = strchr (p0, ' ')))
938 memcpy (term_sub, p0, p1-p0);
939 term_sub[p1-p0] = '\0';
942 strcpy (term_sub, p0);
944 grep_info.isam_p_indx = 0;
945 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
946 num_bases, basenames))
948 if (grep_info.isam_p_indx == 0)
949 rset[rset_no] = rset_create (rset_kind_null, NULL);
950 else if (grep_info.isam_p_indx > 1)
951 rset[rset_no] = rset_trunc (zi->wordIsam,
952 grep_info.isam_p_buf,
953 grep_info.isam_p_indx);
956 rset_isam_parms parms;
958 parms.is = zi->wordIsam;
959 parms.pos = *grep_info.isam_p_buf;
960 rset[rset_no] = rset_create (rset_kind_isam, &parms);
962 assert (rset[rset_no]);
963 if (++rset_no >= sizeof(rset)/sizeof(*rset))
971 xfree (grep_info.isam_p_buf);
973 return rset_create (rset_kind_null, NULL);
974 else if (rset_no == 1)
977 result = rpn_prox (rset, rset_no);
978 for (i = 0; i<rset_no; i++)
979 rset_delete (rset[i]);
983 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
984 oid_value attributeSet)
989 rset_temp_parms parms;
990 char termz[IT_MAX_WORD+1];
992 if (zapt->term->which != Z_Term_general)
997 parms.key_size = sizeof (struct it_key);
998 result = rset_create (rset_kind_temp, &parms);
999 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1001 trans_term (zi, zapt, termz);
1002 key.sysno = atoi (termz);
1005 rset_write (result, rsfd, &key);
1006 rset_close (result, rsfd);
1010 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1011 oid_value attributeSet,
1012 int num_bases, char **basenames)
1016 int relation_value, structure_value;
1018 attr_init (&relation, zapt, 2);
1019 attr_init (&structure, zapt, 4);
1021 relation_value = attr_find (&relation, NULL);
1022 structure_value = attr_find (&structure, NULL);
1023 switch (structure_value)
1026 if (relation_value == 102) /* relevance relation */
1027 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1028 num_bases, basenames);
1029 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1030 num_bases, basenames);
1031 case 1: /* phrase */
1032 if (relation_value == 102) /* relevance relation */
1033 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1034 num_bases, basenames);
1035 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1036 num_bases, basenames);
1039 if (relation_value == 102) /* relevance relation */
1040 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1041 num_bases, basenames);
1042 return rpn_search_APT_word (zi, zapt, attributeSet,
1043 num_bases, basenames);
1048 case 5: /* date - normalized */
1050 case 6: /* word list */
1051 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1052 num_bases, basenames);
1053 case 100: /* date - un-normalized */
1055 case 101: /* name - normalized */
1057 case 102: /* date - un-normalized */
1059 case 103: /* structure */
1063 case 105: /* free-form-text */
1064 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1065 num_bases, basenames);
1066 case 106: /* document-text */
1067 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1068 num_bases, basenames);
1069 case 107: /* local-number */
1070 return rpn_search_APT_local (zi, zapt, attributeSet);
1071 case 108: /* string */
1072 return rpn_search_APT_word (zi, zapt, attributeSet,
1073 num_bases, basenames);
1074 case 109: /* numeric string */
1081 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1085 if (!(s = resultSetGet (zi, resultSetId)))
1086 return rset_create (rset_kind_null, NULL);
1090 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1091 oid_value attributeSet,
1092 int num_bases, char **basenames)
1095 if (zs->which == Z_RPNStructure_complex)
1097 rset_bool_parms bool_parms;
1099 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1101 num_bases, basenames);
1102 if (bool_parms.rset_l == NULL)
1104 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1106 num_bases, basenames);
1107 if (bool_parms.rset_r == NULL)
1109 rset_delete (bool_parms.rset_l);
1112 bool_parms.key_size = sizeof(struct it_key);
1113 bool_parms.cmp = key_compare;
1115 switch (zs->u.complex->operator->which)
1117 case Z_Operator_and:
1118 r = rset_create (rset_kind_and, &bool_parms);
1121 r = rset_create (rset_kind_or, &bool_parms);
1123 case Z_Operator_and_not:
1124 r = rset_create (rset_kind_not, &bool_parms);
1130 else if (zs->which == Z_RPNStructure_simple)
1132 if (zs->u.simple->which == Z_Operand_APT)
1134 logf (LOG_DEBUG, "rpn_search_APT");
1135 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1136 attributeSet, num_bases, basenames);
1138 else if (zs->u.simple->which == Z_Operand_resultSetId)
1140 logf (LOG_DEBUG, "rpn_search_ref");
1141 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1155 static void count_set (RSET r, int *count)
1161 logf (LOG_DEBUG, "rpn_save_set");
1163 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1164 while (rset_read (r, rfd, &key))
1166 if (key.sysno != psysno)
1172 rset_close (r, rfd);
1173 logf (LOG_DEBUG, "%d distinct sysnos", *count);
1176 int rpn_search (ZServerInfo *zi,
1177 Z_RPNQuery *rpn, int num_bases, char **basenames,
1178 const char *setname, int *hits)
1182 oid_value attributeSet;
1187 zi->errString = NULL;
1189 attrset = oid_getentbyoid (rpn->attributeSetId);
1190 attributeSet = attrset->value;
1191 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1192 num_bases, basenames);
1195 count_set (rset, hits);
1196 resultSetAdd (zi, setname, 1, rset);
1198 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1203 struct scan_entry *list;
1210 static int scan_handle (Dict_char *name, const char *info, int pos,
1213 int len_prefix, idx;
1216 struct scan_info *scan_info = client;
1218 rset_isam_parms parms;
1220 len_prefix = strlen(scan_info->prefix);
1221 if (memcmp (name, scan_info->prefix, len_prefix))
1224 idx = scan_info->after - pos + scan_info->before;
1227 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1228 strlen(name + len_prefix)+1);
1229 strcpy (scan_info->list[idx].term, name + len_prefix);
1230 assert (*info == sizeof(isam_p));
1231 memcpy (&isam_p, info+1, sizeof(isam_p));
1232 parms.is = scan_info->isam;
1235 rset = rset_create (rset_kind_isam, &parms);
1236 count_set (rset, &scan_info->list[idx].occurrences);
1239 scan_info->list[idx].occurrences = 1;
1241 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1246 static int dummy_handle (Dict_char *name, const char *info, void *p)
1251 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1252 int num_bases, char **basenames,
1253 int *position, int *num_entries, struct scan_entry **list,
1256 int i, j, sizez, max_pos;
1257 int pos = *position;
1258 int num = *num_entries;
1261 char termz[IT_MAX_WORD+20];
1264 Z_Term *term = zapt->term;
1265 struct scan_info scan_info;
1267 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1271 scan_info.before = before = pos-1;
1272 scan_info.after = after = 1+num-pos;
1273 scan_info.odr = zi->odr;
1275 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1277 scan_info.isam = zi->wordIsam;
1278 scan_info.list = odr_malloc (zi->odr, (before+after)*
1279 sizeof(*scan_info.list));
1280 for (j = 0; j<before+after; j++)
1281 scan_info.list[j].term = NULL;
1282 attr_init (&use, zapt, 1);
1283 use_value = attr_find (&use, NULL);
1284 logf (LOG_DEBUG, "use value %d", use_value);
1286 if (use_value == -1)
1288 i = index_word_prefix (termz, 1, use_value, *basenames);
1290 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
1292 if (max_pos <= strlen(*basenames))
1294 zi->errString = *basenames;
1295 return zi->errCode = 109; /* Database unavailable */
1297 strcpy (scan_info.prefix, termz);
1298 sizez = term->u.general->len;
1299 if (sizez > IT_MAX_WORD)
1300 sizez = IT_MAX_WORD;
1301 for (j = 0; j<sizez; j++)
1302 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1305 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1307 *status = BEND_SCAN_SUCCESS;
1309 for (i = 0; i<scan_info.after; i++)
1310 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1314 *status = BEND_SCAN_PARTIAL;
1316 for (i = 0; i<scan_info.before; i++)
1317 if (scan_info.list[i].term)
1320 *status = BEND_SCAN_PARTIAL;
1324 *list = scan_info.list+i; /* list is set to first 'real' entry */
1326 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1327 zi->errCode = 114; /* if no entries was found */
1328 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1329 *position, *num_entries);
1331 logf (LOG_DEBUG, "scan error: %d", zi->errCode);