2 * Copyright (C) 1994-1996, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.51 1996-06-11 10:54:15 quinn
10 * Revision 1.50 1996/06/07 08:51:53 adam
11 * Bug fix: Character mapping was broken (introducued by last revision).
13 * Revision 1.49 1996/06/04 10:18:11 adam
14 * Search/scan uses character mapping module.
16 * Revision 1.48 1996/05/28 15:15:01 adam
17 * Bug fix: Didn't handle unknown database correctly.
19 * Revision 1.47 1996/05/15 18:36:28 adam
20 * Function trans_term transforms unsearchable characters to blanks.
22 * Revision 1.46 1996/05/15 11:57:56 adam
23 * Fixed bug introduced by set/field mapping in search operations.
25 * Revision 1.45 1996/05/14 11:34:00 adam
26 * Scan support in multiple registers/databases.
28 * Revision 1.44 1996/05/14 06:16:44 adam
29 * Compact use/set bytes used in search service.
31 * Revision 1.43 1996/05/09 09:54:43 adam
32 * Server supports maps from one logical attributes to a list of physical
34 * The extraction process doesn't make space consuming 'any' keys.
36 * Revision 1.42 1996/05/09 07:28:56 quinn
37 * Work towards phrases and multiple registers
39 * Revision 1.41 1996/03/20 09:36:43 adam
40 * Function dict_lookup_grep got extra parameter, init_pos, which marks
41 * from which position in pattern approximate pattern matching should occur.
42 * Approximate pattern matching is used in relevance=re-2.
44 * Revision 1.40 1996/02/02 13:44:44 adam
45 * The public dictionary functions simply use char instead of Dict_char
46 * to represent search strings. Dict_char is used internally only.
48 * Revision 1.39 1996/01/03 16:22:13 quinn
51 * Revision 1.38 1995/12/11 09:12:55 adam
52 * The rec_get function returns NULL if record doesn't exist - will
53 * happen in the server if the result set records have been deleted since
54 * the creation of the set (i.e. the search).
55 * The server saves a result temporarily if it is 'volatile', i.e. the
56 * set is register dependent.
58 * Revision 1.37 1995/12/06 15:05:28 adam
59 * More verbose in count_set.
61 * Revision 1.36 1995/12/06 12:41:27 adam
62 * New command 'stat' for the index program.
63 * Filenames can be read from stdin by specifying '-'.
64 * Bug fix/enhancement of the transformation from terms to regular
65 * expressons in the search engine.
67 * Revision 1.35 1995/11/27 09:29:00 adam
68 * Bug fixes regarding conversion to regular expressions.
70 * Revision 1.34 1995/11/16 17:00:56 adam
71 * Better logging of rpn query.
73 * Revision 1.33 1995/11/01 13:58:28 quinn
74 * Moving data1 to yaz/retrieval
76 * Revision 1.32 1995/10/27 14:00:11 adam
77 * Implemented detection of database availability.
79 * Revision 1.31 1995/10/17 18:02:10 adam
80 * New feature: databases. Implemented as prefix to words in dictionary.
82 * Revision 1.30 1995/10/16 09:32:38 adam
83 * More work on relational op.
85 * Revision 1.29 1995/10/13 16:01:49 adam
88 * Revision 1.28 1995/10/13 12:26:43 adam
89 * Optimization of truncation.
91 * Revision 1.27 1995/10/12 17:07:22 adam
94 * Revision 1.26 1995/10/12 12:40:54 adam
95 * Bug fixes in rpn_prox.
97 * Revision 1.25 1995/10/10 13:59:24 adam
98 * Function rset_open changed its wflag parameter to general flags.
100 * Revision 1.24 1995/10/09 16:18:37 adam
101 * Function dict_lookup_grep got extra client data parameter.
103 * Revision 1.23 1995/10/06 16:33:37 adam
104 * Use attribute mappings.
106 * Revision 1.22 1995/10/06 15:07:39 adam
107 * Structure 'local-number' handled.
109 * Revision 1.21 1995/10/06 13:52:06 adam
110 * Bug fixes. Handler may abort further scanning.
112 * Revision 1.20 1995/10/06 11:06:33 adam
113 * Scan entries include 'occurrences' now.
115 * Revision 1.19 1995/10/06 10:43:56 adam
116 * Scan added. 'occurrences' in scan entries not set yet.
118 * Revision 1.18 1995/10/04 16:57:20 adam
119 * Key input and merge sort in one pass.
121 * Revision 1.17 1995/10/04 12:55:17 adam
122 * Bug fix in ranked search. Use=Any keys inserted.
124 * Revision 1.16 1995/10/02 16:24:40 adam
125 * Use attribute actually used in search requests.
127 * Revision 1.15 1995/10/02 15:18:52 adam
128 * New member in recRetrieveCtrl: diagnostic.
130 * Revision 1.14 1995/09/28 12:10:32 adam
131 * Bug fixes. Field prefix used in queries.
133 * Revision 1.13 1995/09/18 14:17:50 adam
136 * Revision 1.12 1995/09/15 14:45:21 adam
138 * Work on truncation.
140 * Revision 1.11 1995/09/14 11:53:27 adam
141 * First work on regular expressions/truncations.
143 * Revision 1.10 1995/09/11 15:23:26 adam
144 * More work on relevance search.
146 * Revision 1.9 1995/09/11 13:09:35 adam
147 * More work on relevance feedback.
149 * Revision 1.8 1995/09/08 14:52:27 adam
150 * Minor changes. Dictionary is lower case now.
152 * Revision 1.7 1995/09/07 13:58:36 adam
153 * New parameter: result-set file descriptor (RSFD) to support multiple
154 * positions within the same result-set.
155 * Boolean operators: and, or, not implemented.
156 * Result-set references.
158 * Revision 1.6 1995/09/06 16:11:18 adam
159 * Option: only one word key per file.
161 * Revision 1.5 1995/09/06 10:33:04 adam
162 * More work on present. Some log messages removed.
164 * Revision 1.4 1995/09/05 15:28:40 adam
165 * More work on search engine.
167 * Revision 1.3 1995/09/04 15:20:22 adam
170 * Revision 1.2 1995/09/04 12:33:43 adam
171 * Various cleanup. YAZ util used instead.
173 * Revision 1.1 1995/09/04 09:10:40 adam
174 * More work on index add/del/update.
175 * Merge sort implemented.
176 * Initial work on z39 server.
185 #include "attribute.h"
198 Z_AttributesPlusTerm *zapt;
201 static int attr_find (AttrType *src, oid_value *attributeSetP)
203 while (src->major < src->zapt->num_attributes)
205 Z_AttributeElement *element;
207 element = src->zapt->attributeList[src->major];
208 if (src->type == *element->attributeType)
210 switch (element->which)
212 case Z_AttributeValue_numeric:
214 if (element->attributeSet && attributeSetP)
218 attrset = oid_getentbyoid (element->attributeSet);
219 *attributeSetP = attrset->value;
221 return *element->value.numeric;
223 case Z_AttributeValue_complex:
224 if (src->minor >= element->value.complex->num_list ||
225 element->value.complex->list[src->minor]->which !=
226 Z_StringOrNumeric_numeric)
229 if (element->attributeSet && attributeSetP)
233 attrset = oid_getentbyoid (element->attributeSet);
234 *attributeSetP = attrset->value;
236 return *element->value.complex->list[src->minor-1]->u.numeric;
246 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
260 int (*cmp)(const void *p1, const void *p2);
267 static void heap_swap (struct trunc_info *ti, int i1, int i2)
272 ti->ptr[i1] = ti->ptr[i2];
276 static void heap_delete (struct trunc_info *ti)
278 int cur = 1, child = 2;
280 heap_swap (ti, 1, ti->heapnum--);
281 while (child <= ti->heapnum) {
282 if (child < ti->heapnum &&
283 (*ti->cmp)(ti->heap[ti->ptr[child]],
284 ti->heap[ti->ptr[1+child]]) > 0)
286 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
287 ti->heap[ti->ptr[child]]) > 0)
289 heap_swap (ti, cur, child);
298 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
302 cur = ++(ti->heapnum);
303 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
304 ti->indx[ti->ptr[cur]] = indx;
306 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
307 ti->heap[ti->ptr[cur]]) > 0)
309 heap_swap (ti, cur, parent);
316 struct trunc_info *heap_init (int size, int key_size,
317 int (*cmp)(const void *p1, const void *p2))
319 struct trunc_info *ti = xmalloc (sizeof(*ti));
324 ti->keysize = key_size;
326 ti->indx = xmalloc (size * sizeof(*ti->indx));
327 ti->heap = xmalloc (size * sizeof(*ti->heap));
328 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
329 ti->swapbuf = xmalloc (ti->keysize);
330 ti->tmpbuf = xmalloc (ti->keysize);
331 ti->buf = xmalloc (size * ti->keysize);
332 for (i = size; --i >= 0; )
335 ti->heap[i] = ti->buf + ti->keysize * i;
340 static void heap_close (struct trunc_info *ti)
350 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
355 rset_temp_parms parms;
357 parms.key_size = sizeof(struct it_key);
358 result = rset_create (rset_kind_temp, &parms);
359 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
361 if (to - from > merge_chunk)
365 int i, i_add = (to-from)/merge_chunk + 1;
366 struct trunc_info *ti;
368 int rsmax = (to-from)/i_add + 1;
370 rset = xmalloc (sizeof(*rset) * rsmax);
371 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
373 for (i = from; i < to; i += i_add)
376 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
379 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
383 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
384 for (i = rscur; --i >= 0; )
386 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
387 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
388 heap_insert (ti, ti->tmpbuf, i);
391 rset_close (rset[i], rsfd[i]);
392 rset_delete (rset[i]);
397 int n = ti->indx[ti->ptr[1]];
399 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
403 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
406 rset_close (rset[n], rsfd[n]);
407 rset_delete (rset[n]);
410 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
413 heap_insert (ti, ti->tmpbuf, n);
426 struct trunc_info *ti;
428 ispt = xmalloc (sizeof(*ispt) * (to-from));
430 ti = heap_init (to-from, sizeof(struct it_key),
432 for (i = to-from; --i >= 0; )
434 ispt[i] = is_position (isam, isam_p[from+i]);
435 if (is_readkey (ispt[i], ti->tmpbuf))
436 heap_insert (ti, ti->tmpbuf, i);
438 is_pt_free (ispt[i]);
442 int n = ti->indx[ti->ptr[1]];
444 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
446 /* section that preserve all keys */
448 if (is_readkey (ispt[n], ti->tmpbuf))
449 heap_insert (ti, ti->tmpbuf, n);
451 is_pt_free (ispt[n]);
453 /* section that preserve all keys with unique sysnos */
456 if (!is_readkey (ispt[n], ti->tmpbuf))
459 is_pt_free (ispt[n]);
462 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
465 heap_insert (ti, ti->tmpbuf, n);
474 rset_close (result, result_rsfd);
478 static int isam_trunc_cmp (const void *p1, const void *p2)
480 ISAM_P i1 = *(ISAM_P*) p1;
481 ISAM_P i2 = *(ISAM_P*) p2;
484 d = is_type (i1) - is_type (i2);
487 return is_block (i1) - is_block (i2);
490 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
493 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
494 return rset_trunc_r (isam, isam_p, 0, no, 100);
508 static void add_isam_p (const char *info, struct grep_info *p)
510 if (p->isam_p_indx == p->isam_p_size)
512 ISAM_P *new_isam_p_buf;
517 p->isam_p_size = 2*p->isam_p_size + 100;
518 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
522 memcpy (new_isam_p_buf, p->isam_p_buf,
523 p->isam_p_indx * sizeof(*p->isam_p_buf));
524 xfree (p->isam_p_buf);
526 p->isam_p_buf = new_isam_p_buf;
529 new_term_no = xmalloc (sizeof(*new_term_no) *
533 memcpy (new_term_no, p->isam_p_buf,
534 p->isam_p_indx * sizeof(*p->term_no));
537 p->term_no = new_term_no;
540 assert (*info == sizeof(*p->isam_p_buf));
541 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
545 static int grep_handle (char *name, const char *info, void *p)
547 logf (LOG_DEBUG, "dict name: %s", name);
548 add_isam_p (info, p);
552 static void gen_regular_rel (char *dst, int val, int islt)
560 sprintf (numstr, "%d", val);
561 for (w = strlen(numstr); --w >= 0; pos++)
580 strcpy (dst + dst_p, numstr);
581 dst_p = strlen(dst) - pos - 1;
609 for (i = 0; i<pos; i++)
622 for (i=1; i<pos; i++)
623 strcat (dst, "[0-9]?");
627 for (i = 0; i <= pos; i++)
628 strcat (dst, "[0-9]");
629 strcat (dst, "[0-9]*");
634 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
635 const char *term_sub,
637 oid_value attributeSet,
638 struct grep_info *grep_info,
646 attr_init (&relation, zapt, 2);
647 relation_value = attr_find (&relation, NULL);
648 term_value = atoi (term_sub);
650 switch (relation_value)
655 logf (LOG_DEBUG, "Relation <");
656 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
661 logf (LOG_DEBUG, "Relation <=");
662 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
667 logf (LOG_DEBUG, "Relation >=");
668 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
673 logf (LOG_DEBUG, "Relation >");
674 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
679 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
680 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
683 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
684 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
688 static void verbatim_char (int ch, int *indx, char *dst)
691 dst[(*indx)++] = '\\';
695 static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
696 const char *term_sub, int regType,
697 oid_value attributeSet, struct grep_info *grep_info,
698 int num_bases, char **basenames)
700 char term_dict[2*IT_MAX_WORD+2];
701 int i, j, r, base_no;
703 int truncation_value;
706 oid_value curAttributeSet = attributeSet;
708 attr_init (&use, zapt, 1);
709 use_value = attr_find (&use, &curAttributeSet);
710 logf (LOG_DEBUG, "use value %d", use_value);
711 attr_init (&truncation, zapt, 5);
712 truncation_value = attr_find (&truncation, NULL);
713 logf (LOG_DEBUG, "truncation value %d", truncation_value);
718 for (base_no = 0; base_no < num_bases; base_no++)
721 data1_local_attribute *local_attr;
722 int max_pos, prefix_len = 0;
724 attp = att_getentbyatt (curAttributeSet, use_value);
727 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
728 curAttributeSet, use_value);
732 if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no]))
734 zi->errCode = 109; /* Database unavailable */
735 zi->errString = basenames[base_no];
738 for (local_attr = attp->local_attributes; local_attr;
739 local_attr = local_attr->next)
743 ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal,
748 term_dict[prefix_len++] = '|';
750 term_dict[prefix_len++] = '(';
751 term_dict[prefix_len++] = 1;
752 term_dict[prefix_len++] = ord;
759 term_dict[prefix_len++] = ')';
760 term_dict[prefix_len++] = 1;
761 term_dict[prefix_len++] = regType;
762 term_dict[prefix_len] = '\0';
763 if (!relational_term (zi, zapt, term_sub, term_dict,
764 attributeSet, grep_info, &max_pos))
769 switch (truncation_value)
771 case -1: /* not specified */
772 case 100: /* do not truncate */
773 term_dict[j++] = '(';
774 for (i = 0; term_sub[i]; i++)
775 verbatim_char (term_sub[i], &j, term_dict);
776 strcpy (term_dict+j, ")");
777 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
778 &max_pos, 0, grep_handle);
780 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
782 case 1: /* right truncation */
783 term_dict[j++] = '(';
784 for (i = 0; term_sub[i]; i++)
785 verbatim_char (term_sub[i], &j, term_dict);
786 strcpy (term_dict+j, ".*)");
787 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
788 &max_pos, 0, grep_handle);
790 case 2: /* left truncation */
791 case 3: /* left&right truncation */
794 case 101: /* process # in term */
795 term_dict[j++] = '(';
796 for (i=0; term_sub[i]; i++)
797 if (term_sub[i] == '#' && i > 2)
799 term_dict[j++] = '.';
800 term_dict[j++] = '*';
803 verbatim_char (term_sub[i], &j, term_dict);
804 strcpy (term_dict+j, ")");
805 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
806 &max_pos, 0, grep_handle);
808 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
811 case 102: /* regular expression */
812 sprintf (term_dict + j, "(%s)", term_sub);
813 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
814 &max_pos, 0, grep_handle);
816 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
819 case 103: /* regular expression with error correction */
822 if (*cp == '*' && cp[1] && cp[2])
827 sprintf (term_dict + j, "(%s)", cp);
828 r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info,
829 &max_pos, j, grep_handle);
831 logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
837 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
841 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
845 Z_Term *term = zapt->term;
847 sizez = term->u.general->len;
848 if (sizez > IT_MAX_WORD-1)
849 sizez = IT_MAX_WORD-1;
850 memcpy (termz, term->u.general->buf, sizez);
854 static void trans_scan_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
857 Z_Term *term = zapt->term;
859 const char *cp = (const char *) term->u.general->buf;
860 const char *cp_end = cp + term->u.general->len;
866 while ((len = (cp_end - cp)) > 0)
868 map = map_chrs_input (&cp, len);
869 if (**map == *CHR_SPACE)
877 for (src = *map; *src; src++)
883 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
884 Z_AttributesPlusTerm *zapt,
885 oid_value attributeSet,
886 int num_bases, char **basenames)
888 rset_relevance_parms parms;
889 char termz[IT_MAX_WORD+1];
890 char term_sub[IT_MAX_WORD+1];
891 struct grep_info grep_info;
896 parms.key_size = sizeof(struct it_key);
898 parms.cmp = key_compare;
899 parms.is = zi->wordIsam;
902 if (zapt->term->which != Z_Term_general)
907 trans_term (zi, zapt, termz);
910 grep_info.term_no = 0;
912 grep_info.isam_p_indx = 0;
913 grep_info.isam_p_size = 0;
914 grep_info.isam_p_buf = NULL;
923 map = map_chrs_input (&p1, strlen(p1));
924 if (**map != *CHR_SPACE)
933 map = map_chrs_input (&p1, strlen(p1));
934 if (**map == *CHR_SPACE)
939 memcpy (term_sub, p0, p2-p0);
940 term_sub[p2-p0] = '\0';
942 if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info,
943 num_bases, basenames))
946 for (; term_index < grep_info.isam_p_indx; term_index++)
947 grep_info.term_no[term_index] = parms.no_terms;
951 parms.term_no = grep_info.term_no;
952 parms.isam_positions = grep_info.isam_p_buf;
953 parms.no_isam_positions = grep_info.isam_p_indx;
954 if (grep_info.isam_p_indx > 0)
955 result = rset_create (rset_kind_relevance, &parms);
957 result = rset_create (rset_kind_null, NULL);
959 xfree(grep_info.term_no);
961 xfree (grep_info.isam_p_buf);
965 static RSET rpn_search_APT_cphrase (ZServerInfo *zi,
966 Z_AttributesPlusTerm *zapt,
967 oid_value attributeSet,
968 int num_bases, char **basenames)
970 rset_isam_parms parms;
971 char termz[IT_MAX_WORD+1];
972 struct grep_info grep_info;
975 if (zapt->term->which != Z_Term_general)
980 trans_term (zi, zapt, termz);
983 grep_info.term_no = 0;
985 grep_info.isam_p_indx = 0;
986 grep_info.isam_p_size = 0;
987 grep_info.isam_p_buf = NULL;
989 if (field_term (zi, zapt, termz, 'p', attributeSet, &grep_info,
990 num_bases, basenames))
992 if (grep_info.isam_p_indx < 1)
993 result = rset_create (rset_kind_null, NULL);
994 else if (grep_info.isam_p_indx == 1)
996 parms.is = zi->wordIsam;
997 parms.pos = *grep_info.isam_p_buf;
998 result = rset_create (rset_kind_isam, &parms);
1001 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
1002 grep_info.isam_p_indx);
1004 xfree(grep_info.term_no);
1006 xfree (grep_info.isam_p_buf);
1010 static RSET rpn_prox (RSET *rset, int rset_no)
1015 struct it_key **buf;
1018 rset_temp_parms parms;
1020 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
1021 more = xmalloc (sizeof(*more)*rset_no);
1022 buf = xmalloc (sizeof(*buf)*rset_no);
1024 for (i = 0; i<rset_no; i++)
1026 buf[i] = xmalloc (sizeof(**buf));
1027 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
1028 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
1032 rset_close (rset[i], rsfd[i]);
1039 return rset_create (rset_kind_null, NULL);
1042 parms.key_size = sizeof (struct it_key);
1043 result = rset_create (rset_kind_temp, &parms);
1044 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1048 for (i = 1; i<rset_no; i++)
1057 cmp = key_compare (buf[i], buf[i-1]);
1060 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
1065 if (buf[i-1]->seqno+1 != buf[i]->seqno)
1067 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
1073 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
1079 rset_write (result, rsfd_result, buf[0]);
1080 more[0] = rset_read (*rset, *rsfd, *buf);
1084 for (i = 0; i<rset_no; i++)
1086 rset_close (rset[i], rsfd[i]);
1089 rset_close (result, rsfd_result);
1096 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
1097 Z_AttributesPlusTerm *zapt,
1098 oid_value attributeSet,
1099 int num_bases, char **basenames)
1101 char termz[IT_MAX_WORD+1];
1102 char term_sub[IT_MAX_WORD+1];
1104 RSET rset[60], result;
1106 struct grep_info grep_info;
1108 if (zapt->term->which != Z_Term_general)
1113 trans_term (zi, zapt, termz);
1116 grep_info.term_no = 0;
1118 grep_info.isam_p_size = 0;
1119 grep_info.isam_p_buf = NULL;
1129 map = map_chrs_input (&p1, strlen(p1));
1130 if (**map != *CHR_SPACE)
1139 map = map_chrs_input (&p1, strlen(p1));
1140 if (**map == *CHR_SPACE)
1146 memcpy (term_sub, p0, p2-p0);
1147 term_sub[p2-p0] = '\0';
1150 grep_info.isam_p_indx = 0;
1151 if (field_term (zi, zapt, term_sub, 'w', attributeSet, &grep_info,
1152 num_bases, basenames))
1154 if (grep_info.isam_p_indx == 0)
1155 rset[rset_no] = rset_create (rset_kind_null, NULL);
1156 else if (grep_info.isam_p_indx > 1)
1157 rset[rset_no] = rset_trunc (zi->wordIsam,
1158 grep_info.isam_p_buf,
1159 grep_info.isam_p_indx);
1162 rset_isam_parms parms;
1164 parms.is = zi->wordIsam;
1165 parms.pos = *grep_info.isam_p_buf;
1166 rset[rset_no] = rset_create (rset_kind_isam, &parms);
1168 assert (rset[rset_no]);
1169 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1173 xfree(grep_info.term_no);
1175 xfree (grep_info.isam_p_buf);
1177 return rset_create (rset_kind_null, NULL);
1178 else if (rset_no == 1)
1181 result = rpn_prox (rset, rset_no);
1182 for (i = 0; i<rset_no; i++)
1183 rset_delete (rset[i]);
1187 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1188 oid_value attributeSet)
1193 rset_temp_parms parms;
1194 char termz[IT_MAX_WORD+1];
1196 if (zapt->term->which != Z_Term_general)
1201 parms.key_size = sizeof (struct it_key);
1202 result = rset_create (rset_kind_temp, &parms);
1203 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1205 trans_term (zi, zapt, termz);
1207 key.sysno = atoi (termz);
1210 rset_write (result, rsfd, &key);
1211 rset_close (result, rsfd);
1215 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1216 oid_value attributeSet,
1217 int num_bases, char **basenames)
1221 AttrType completeness;
1222 int relation_value, structure_value, completeness_value;
1224 attr_init (&relation, zapt, 2);
1225 attr_init (&structure, zapt, 4);
1226 attr_init (&completeness, zapt, 6);
1228 relation_value = attr_find (&relation, NULL);
1229 structure_value = attr_find (&structure, NULL);
1230 completeness_value = attr_find (&completeness, NULL);
1231 switch (structure_value)
1234 if (relation_value == 102) /* relevance relation */
1235 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1236 num_bases, basenames);
1237 if (completeness_value == 2 || completeness_value == 3)
1238 return rpn_search_APT_cphrase (zi, zapt, attributeSet,
1239 num_bases, basenames);
1240 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1241 num_bases, basenames);
1242 case 1: /* phrase */
1243 if (relation_value == 102) /* relevance relation */
1244 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1245 num_bases, basenames);
1246 if (completeness_value == 2 || completeness_value == 3)
1247 return rpn_search_APT_cphrase (zi, zapt, attributeSet,
1248 num_bases, basenames);
1249 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1250 num_bases, basenames);
1253 if (relation_value == 102) /* relevance relation */
1254 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1255 num_bases, basenames);
1256 if (completeness_value == 2 || completeness_value == 3)
1257 return rpn_search_APT_cphrase (zi, zapt, attributeSet,
1258 num_bases, basenames);
1259 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1260 num_bases, basenames);
1265 case 5: /* date - normalized */
1267 case 6: /* word list */
1268 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1269 num_bases, basenames);
1270 case 100: /* date - un-normalized */
1272 case 101: /* name - normalized */
1274 case 102: /* date - un-normalized */
1276 case 103: /* structure */
1280 case 105: /* free-form-text */
1281 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1282 num_bases, basenames);
1283 case 106: /* document-text */
1284 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1285 num_bases, basenames);
1286 case 107: /* local-number */
1287 return rpn_search_APT_local (zi, zapt, attributeSet);
1288 case 108: /* string */
1289 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1290 num_bases, basenames);
1291 case 109: /* numeric string */
1298 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1302 if (!(s = resultSetGet (zi, resultSetId)))
1303 return rset_create (rset_kind_null, NULL);
1307 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1308 oid_value attributeSet,
1309 int num_bases, char **basenames)
1312 if (zs->which == Z_RPNStructure_complex)
1314 rset_bool_parms bool_parms;
1317 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1319 num_bases, basenames);
1320 if (bool_parms.rset_l == NULL)
1322 if (rset_is_ranked(bool_parms.rset_l))
1324 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1326 num_bases, basenames);
1327 if (bool_parms.rset_r == NULL)
1329 rset_delete (bool_parms.rset_l);
1332 if (rset_is_ranked(bool_parms.rset_r))
1334 bool_parms.key_size = sizeof(struct it_key);
1335 bool_parms.cmp = key_compare;
1337 switch (zs->u.complex->roperator->which)
1339 case Z_Operator_and:
1340 r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms);
1343 r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms);
1345 case Z_Operator_and_not:
1346 r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms);
1352 else if (zs->which == Z_RPNStructure_simple)
1354 if (zs->u.simple->which == Z_Operand_APT)
1356 logf (LOG_DEBUG, "rpn_search_APT");
1357 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1358 attributeSet, num_bases, basenames);
1360 else if (zs->u.simple->which == Z_Operand_resultSetId)
1362 logf (LOG_DEBUG, "rpn_search_ref");
1363 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1377 void count_set_save (RSET *r, int *count)
1384 rset_temp_parms parms;
1386 logf (LOG_DEBUG, "count_set_save");
1388 parms.key_size = sizeof(struct it_key);
1389 w = rset_create (rset_kind_temp, &parms);
1390 wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO);
1391 rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO);
1392 while (rset_read (*r, rfd, &key))
1394 if (key.sysno != psysno)
1396 rset_write (w, wfd, &key);
1402 rset_close (*r, rfd);
1404 rset_close (w, wfd);
1406 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1409 static void count_set (RSET r, int *count)
1416 logf (LOG_DEBUG, "count_set");
1418 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1419 while (rset_read (r, rfd, &key))
1421 if (key.sysno != psysno)
1428 rset_close (r, rfd);
1429 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1432 int rpn_search (ZServerInfo *zi,
1433 Z_RPNQuery *rpn, int num_bases, char **basenames,
1434 const char *setname, int *hits)
1438 oid_value attributeSet;
1440 dict_grep_cmap (zi->wordDict, map_chrs_input);
1444 zi->errString = NULL;
1446 attrset = oid_getentbyoid (rpn->attributeSetId);
1447 attributeSet = attrset->value;
1448 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1449 num_bases, basenames);
1452 if (rset_is_volatile(rset))
1453 count_set_save(&rset,hits);
1455 count_set (rset, hits);
1456 resultSetAdd (zi, setname, 1, rset);
1458 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1462 struct scan_info_entry {
1468 struct scan_info_entry *list;
1474 static int scan_handle (char *name, const char *info, int pos, void *client)
1476 int len_prefix, idx;
1477 struct scan_info *scan_info = client;
1479 len_prefix = strlen(scan_info->prefix);
1480 if (memcmp (name, scan_info->prefix, len_prefix))
1483 idx = scan_info->after - pos + scan_info->before;
1486 logf (LOG_DEBUG, "%-3d %s", idx, name+len_prefix);
1487 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1488 strlen(name + len_prefix)+1);
1489 strcpy (scan_info->list[idx].term, name + len_prefix);
1490 assert (*info == sizeof(ISAM_P));
1491 memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P));
1496 static void scan_term_untrans (ODR odr, char **dstp, const char *src)
1498 char *dst = odr_malloc (odr, strlen(src)*2+1);
1503 const char *cp = map_chrs_output (&src);
1510 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1511 oid_value attributeset,
1512 int num_bases, char **basenames,
1513 int *position, int *num_entries, struct scan_entry **list,
1517 int pos = *position;
1518 int num = *num_entries;
1522 char termz[IT_MAX_WORD+20];
1525 AttrType completeness;
1526 int completeness_value;
1527 struct scan_info *scan_info_array;
1528 struct scan_entry *glist;
1529 int ords[32], ord_no = 0;
1532 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1534 attr_init (&use, zapt, 1);
1535 use_value = attr_find (&use, NULL);
1536 logf (LOG_DEBUG, "use value %d", use_value);
1538 attr_init (&completeness, zapt, 6);
1539 completeness_value = attr_find (&completeness, NULL);
1540 logf (LOG_DEBUG, "completeness value %d", completeness_value);
1542 if (attributeset == VAL_NONE)
1543 attributeset = VAL_BIB1;
1545 if (use_value == -1)
1547 for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
1550 data1_local_attribute *local_attr;
1552 attp = att_getentbyatt (attributeset, use_value);
1555 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
1556 attributeset, use_value);
1557 return zi->errCode = 114;
1559 if (zebTargetInfo_curDatabase (zi->zti, basenames[base_no]))
1561 zi->errString = basenames[base_no];
1562 return zi->errCode = 109; /* Database unavailable */
1564 for (local_attr = attp->local_attributes; local_attr && ord_no < 32;
1565 local_attr = local_attr->next)
1569 ord = zebTargetInfo_lookupSU (zi->zti, attp->attset_ordinal,
1572 ords[ord_no++] = ord;
1576 return zi->errCode = 113;
1579 scan_info_array = odr_malloc (zi->odr, ord_no * sizeof(*scan_info_array));
1580 for (i = 0; i < ord_no; i++)
1582 int j, prefix_len = 0;
1583 int before_tmp = before, after_tmp = after;
1584 struct scan_info *scan_info = scan_info_array + i;
1586 scan_info->before = before;
1587 scan_info->after = after;
1588 scan_info->odr = zi->odr;
1590 scan_info->list = odr_malloc (zi->odr, (before+after)*
1591 sizeof(*scan_info->list));
1592 for (j = 0; j<before+after; j++)
1593 scan_info->list[j].term = NULL;
1594 termz[prefix_len++] = ords[i];
1595 termz[prefix_len++] =
1596 (completeness_value==2 || completeness_value==3) ? 'p': 'w';
1597 termz[prefix_len] = 0;
1598 strcpy (scan_info->prefix, termz);
1600 trans_scan_term (zi, zapt, termz+prefix_len);
1602 dict_scan (zi->wordDict, termz, &before_tmp, &after_tmp, scan_info,
1605 glist = odr_malloc (zi->odr, (before+after)*sizeof(*glist));
1606 for (i = 0; i < ord_no; i++)
1609 *status = BEND_SCAN_SUCCESS;
1610 for (i = 0; i<after; i++)
1613 const char *mterm = NULL;
1616 rset_isam_parms parms;
1618 for (j = 0; j < ord_no; j++)
1620 if (ptr[j] < before+after &&
1621 (tst=scan_info_array[j].list[ptr[j]].term) &&
1622 (!mterm || strcmp (tst, mterm) < 0))
1630 scan_term_untrans (zi->odr, &glist[i+before].term, mterm);
1631 parms.is = zi->wordIsam;
1632 parms.pos = scan_info_array[j0].list[ptr[j0]].isam_p;
1633 rset = rset_create (rset_kind_isam, &parms);
1636 for (j = j0+1; j<ord_no; j++)
1638 if (ptr[j] < before+after &&
1639 (tst=scan_info_array[j].list[ptr[j]].term) &&
1640 !strcmp (tst, mterm))
1642 rset_isam_parms parms;
1643 rset_bool_parms bool_parms;
1646 parms.is = zi->wordIsam;
1647 parms.pos = scan_info_array[j].list[ptr[j]].isam_p;
1648 rset2 = rset_create (rset_kind_isam, &parms);
1650 bool_parms.key_size = sizeof(struct it_key);
1651 bool_parms.cmp = key_compare;
1652 bool_parms.rset_l = rset;
1653 bool_parms.rset_r = rset2;
1655 rset = rset_create (rset_kind_or, &bool_parms);
1660 count_set (rset, &glist[i+before].occurrences);
1665 *num_entries -= (after-i);
1666 *status = BEND_SCAN_PARTIAL;
1669 for (i = 0; i<ord_no; i++)
1672 for (i = 0; i<before; i++)
1675 const char *mterm = NULL;
1678 rset_isam_parms parms;
1680 for (j = 0; j <ord_no; j++)
1682 if (ptr[j] < before &&
1683 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
1684 (!mterm || strcmp (tst, mterm) > 0))
1693 scan_term_untrans (zi->odr, &glist[before-1-i].term, mterm);
1695 parms.is = zi->wordIsam;
1696 parms.pos = scan_info_array[j0].list[before-1-ptr[j0]].isam_p;
1697 rset = rset_create (rset_kind_isam, &parms);
1701 for (j = j0+1; j<ord_no; j++)
1703 if (ptr[j] < before &&
1704 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
1705 !strcmp (tst, mterm))
1707 rset_isam_parms parms;
1708 rset_bool_parms bool_parms;
1711 parms.is = zi->wordIsam;
1712 parms.pos = scan_info_array[j].list[before-1-ptr[j]].isam_p;
1713 rset2 = rset_create (rset_kind_isam, &parms);
1715 bool_parms.key_size = sizeof(struct it_key);
1716 bool_parms.cmp = key_compare;
1717 bool_parms.rset_l = rset;
1718 bool_parms.rset_r = rset2;
1720 rset = rset_create (rset_kind_or, &bool_parms);
1725 count_set (rset, &glist[before-1-i].occurrences);
1731 *status = BEND_SCAN_PARTIAL;
1735 *list = glist + i; /* list is set to first 'real' entry */
1737 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1738 *position, *num_entries);
1740 logf (LOG_DEBUG, "scan error: %d", zi->errCode);