2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.39 1996-01-03 16:22:13 quinn
10 * Revision 1.38 1995/12/11 09:12:55 adam
11 * The rec_get function returns NULL if record doesn't exist - will
12 * happen in the server if the result set records have been deleted since
13 * the creation of the set (i.e. the search).
14 * The server saves a result temporarily if it is 'volatile', i.e. the
15 * set is register dependent.
17 * Revision 1.37 1995/12/06 15:05:28 adam
18 * More verbose in count_set.
20 * Revision 1.36 1995/12/06 12:41:27 adam
21 * New command 'stat' for the index program.
22 * Filenames can be read from stdin by specifying '-'.
23 * Bug fix/enhancement of the transformation from terms to regular
24 * expressons in the search engine.
26 * Revision 1.35 1995/11/27 09:29:00 adam
27 * Bug fixes regarding conversion to regular expressions.
29 * Revision 1.34 1995/11/16 17:00:56 adam
30 * Better logging of rpn query.
32 * Revision 1.33 1995/11/01 13:58:28 quinn
33 * Moving data1 to yaz/retrieval
35 * Revision 1.32 1995/10/27 14:00:11 adam
36 * Implemented detection of database availability.
38 * Revision 1.31 1995/10/17 18:02:10 adam
39 * New feature: databases. Implemented as prefix to words in dictionary.
41 * Revision 1.30 1995/10/16 09:32:38 adam
42 * More work on relational op.
44 * Revision 1.29 1995/10/13 16:01:49 adam
47 * Revision 1.28 1995/10/13 12:26:43 adam
48 * Optimization of truncation.
50 * Revision 1.27 1995/10/12 17:07:22 adam
53 * Revision 1.26 1995/10/12 12:40:54 adam
54 * Bug fixes in rpn_prox.
56 * Revision 1.25 1995/10/10 13:59:24 adam
57 * Function rset_open changed its wflag parameter to general flags.
59 * Revision 1.24 1995/10/09 16:18:37 adam
60 * Function dict_lookup_grep got extra client data parameter.
62 * Revision 1.23 1995/10/06 16:33:37 adam
63 * Use attribute mappings.
65 * Revision 1.22 1995/10/06 15:07:39 adam
66 * Structure 'local-number' handled.
68 * Revision 1.21 1995/10/06 13:52:06 adam
69 * Bug fixes. Handler may abort further scanning.
71 * Revision 1.20 1995/10/06 11:06:33 adam
72 * Scan entries include 'occurrences' now.
74 * Revision 1.19 1995/10/06 10:43:56 adam
75 * Scan added. 'occurrences' in scan entries not set yet.
77 * Revision 1.18 1995/10/04 16:57:20 adam
78 * Key input and merge sort in one pass.
80 * Revision 1.17 1995/10/04 12:55:17 adam
81 * Bug fix in ranked search. Use=Any keys inserted.
83 * Revision 1.16 1995/10/02 16:24:40 adam
84 * Use attribute actually used in search requests.
86 * Revision 1.15 1995/10/02 15:18:52 adam
87 * New member in recRetrieveCtrl: diagnostic.
89 * Revision 1.14 1995/09/28 12:10:32 adam
90 * Bug fixes. Field prefix used in queries.
92 * Revision 1.13 1995/09/18 14:17:50 adam
95 * Revision 1.12 1995/09/15 14:45:21 adam
99 * Revision 1.11 1995/09/14 11:53:27 adam
100 * First work on regular expressions/truncations.
102 * Revision 1.10 1995/09/11 15:23:26 adam
103 * More work on relevance search.
105 * Revision 1.9 1995/09/11 13:09:35 adam
106 * More work on relevance feedback.
108 * Revision 1.8 1995/09/08 14:52:27 adam
109 * Minor changes. Dictionary is lower case now.
111 * Revision 1.7 1995/09/07 13:58:36 adam
112 * New parameter: result-set file descriptor (RSFD) to support multiple
113 * positions within the same result-set.
114 * Boolean operators: and, or, not implemented.
115 * Result-set references.
117 * Revision 1.6 1995/09/06 16:11:18 adam
118 * Option: only one word key per file.
120 * Revision 1.5 1995/09/06 10:33:04 adam
121 * More work on present. Some log messages removed.
123 * Revision 1.4 1995/09/05 15:28:40 adam
124 * More work on search engine.
126 * Revision 1.3 1995/09/04 15:20:22 adam
129 * Revision 1.2 1995/09/04 12:33:43 adam
130 * Various cleanup. YAZ util used instead.
132 * Revision 1.1 1995/09/04 09:10:40 adam
133 * More work on index add/del/update.
134 * Merge sort implemented.
135 * Initial work on z39 server.
144 #include "attribute.h"
152 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
157 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
158 attp = att_getentbyatt (attrSet, attrUse);
161 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
162 return index_word_prefix (string, attp->attset_ordinal,
163 attp->local_attribute, basename);
170 Z_AttributesPlusTerm *zapt;
173 static int attr_find (AttrType *src, oid_value *attributeSetP)
175 while (src->major < src->zapt->num_attributes)
177 Z_AttributeElement *element;
179 element = src->zapt->attributeList[src->major];
180 if (src->type == *element->attributeType)
182 switch (element->which)
184 case Z_AttributeValue_numeric:
186 if (element->attributeSet && attributeSetP)
190 attrset = oid_getentbyoid (element->attributeSet);
191 *attributeSetP = attrset->value;
193 return *element->value.numeric;
195 case Z_AttributeValue_complex:
196 if (src->minor >= element->value.complex->num_list ||
197 element->value.complex->list[src->minor]->which !=
198 Z_StringOrNumeric_numeric)
201 if (element->attributeSet && attributeSetP)
205 attrset = oid_getentbyoid (element->attributeSet);
206 *attributeSetP = attrset->value;
208 return *element->value.complex->list[src->minor-1]->u.numeric;
218 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
232 int (*cmp)(const void *p1, const void *p2);
239 static void heap_swap (struct trunc_info *ti, int i1, int i2)
244 ti->ptr[i1] = ti->ptr[i2];
248 static void heap_delete (struct trunc_info *ti)
250 int cur = 1, child = 2;
252 heap_swap (ti, 1, ti->heapnum--);
253 while (child <= ti->heapnum) {
254 if (child < ti->heapnum &&
255 (*ti->cmp)(ti->heap[ti->ptr[child]],
256 ti->heap[ti->ptr[1+child]]) > 0)
258 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
259 ti->heap[ti->ptr[child]]) > 0)
261 heap_swap (ti, cur, child);
270 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
274 cur = ++(ti->heapnum);
275 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
276 ti->indx[ti->ptr[cur]] = indx;
278 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
279 ti->heap[ti->ptr[cur]]) > 0)
281 heap_swap (ti, cur, parent);
288 struct trunc_info *heap_init (int size, int key_size,
289 int (*cmp)(const void *p1, const void *p2))
291 struct trunc_info *ti = xmalloc (sizeof(*ti));
296 ti->keysize = key_size;
298 ti->indx = xmalloc (size * sizeof(*ti->indx));
299 ti->heap = xmalloc (size * sizeof(*ti->heap));
300 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
301 ti->swapbuf = xmalloc (ti->keysize);
302 ti->tmpbuf = xmalloc (ti->keysize);
303 ti->buf = xmalloc (size * ti->keysize);
304 for (i = size; --i >= 0; )
307 ti->heap[i] = ti->buf + ti->keysize * i;
312 static void heap_close (struct trunc_info *ti)
322 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
327 rset_temp_parms parms;
329 parms.key_size = sizeof(struct it_key);
330 result = rset_create (rset_kind_temp, &parms);
331 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
333 if (to - from > merge_chunk)
337 int i, i_add = (to-from)/merge_chunk + 1;
338 struct trunc_info *ti;
340 int rsmax = (to-from)/i_add + 1;
342 rset = xmalloc (sizeof(*rset) * rsmax);
343 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
345 for (i = from; i < to; i += i_add)
348 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
351 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
355 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
356 for (i = rscur; --i >= 0; )
358 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
359 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
360 heap_insert (ti, ti->tmpbuf, i);
363 rset_close (rset[i], rsfd[i]);
364 rset_delete (rset[i]);
369 int n = ti->indx[ti->ptr[1]];
371 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
375 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
378 rset_close (rset[n], rsfd[n]);
379 rset_delete (rset[n]);
382 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
385 heap_insert (ti, ti->tmpbuf, n);
398 struct trunc_info *ti;
400 ispt = xmalloc (sizeof(*ispt) * (to-from));
402 ti = heap_init (to-from, sizeof(struct it_key),
404 for (i = to-from; --i >= 0; )
406 ispt[i] = is_position (isam, isam_p[from+i]);
407 if (is_readkey (ispt[i], ti->tmpbuf))
408 heap_insert (ti, ti->tmpbuf, i);
410 is_pt_free (ispt[i]);
414 int n = ti->indx[ti->ptr[1]];
416 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
418 /* section that preserve all keys */
420 if (is_readkey (ispt[n], ti->tmpbuf))
421 heap_insert (ti, ti->tmpbuf, n);
423 is_pt_free (ispt[n]);
425 /* section that preserve all keys with unique sysnos */
428 if (!is_readkey (ispt[n], ti->tmpbuf))
431 is_pt_free (ispt[n]);
434 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
437 heap_insert (ti, ti->tmpbuf, n);
446 rset_close (result, result_rsfd);
450 static int isam_trunc_cmp (const void *p1, const void *p2)
452 ISAM_P i1 = *(ISAM_P*) p1;
453 ISAM_P i2 = *(ISAM_P*) p2;
456 d = is_type (i1) - is_type (i2);
459 return is_block (i1) - is_block (i2);
462 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
465 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
466 return rset_trunc_r (isam, isam_p, 0, no, 100);
475 static void add_isam_p (const char *info, struct grep_info *p)
477 if (p->isam_p_indx == p->isam_p_size)
479 ISAM_P *new_isam_p_buf;
481 p->isam_p_size = 2*p->isam_p_size + 100;
482 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
486 memcpy (new_isam_p_buf, p->isam_p_buf,
487 p->isam_p_indx * sizeof(*p->isam_p_buf));
488 xfree (p->isam_p_buf);
490 p->isam_p_buf = new_isam_p_buf;
492 assert (*info == sizeof(*p->isam_p_buf));
493 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
497 static int grep_handle (Dict_char *name, const char *info, void *p)
499 logf (LOG_DEBUG, "dict name: %s", name);
500 add_isam_p (info, p);
504 static void gen_regular_rel (char *dst, int val, int islt)
512 sprintf (numstr, "%d", val);
513 for (w = strlen(numstr); --w >= 0; pos++)
532 strcpy (dst + dst_p, numstr);
533 dst_p = strlen(dst) - pos - 1;
561 for (i = 0; i<pos; i++)
574 for (i=1; i<pos; i++)
575 strcat (dst, "[0-9]?");
579 for (i = 0; i <= pos; i++)
580 strcat (dst, "[0-9]");
581 strcat (dst, "[0-9]*");
586 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
587 const char *term_sub,
589 oid_value attributeSet,
590 struct grep_info *grep_info,
598 attr_init (&relation, zapt, 2);
599 relation_value = attr_find (&relation, NULL);
600 term_value = atoi (term_sub);
602 switch (relation_value)
607 logf (LOG_DEBUG, "Relation <");
608 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
613 logf (LOG_DEBUG, "Relation <=");
614 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
619 logf (LOG_DEBUG, "Relation >=");
620 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
625 logf (LOG_DEBUG, "Relation >");
626 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
631 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
632 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
635 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
636 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
640 static void verbatim_char (int ch, int *indx, char *dst)
643 dst[(*indx)++] = '\\';
647 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
648 const char *term_sub,
649 oid_value attributeSet, struct grep_info *grep_info,
650 int num_bases, char **basenames)
652 char term_dict[2*IT_MAX_WORD+2];
653 int i, j, r, base_no;
655 int truncation_value;
658 oid_value curAttributeSet = attributeSet;
660 attr_init (&use, zapt, 1);
661 use_value = attr_find (&use, &curAttributeSet);
662 logf (LOG_DEBUG, "use value %d", use_value);
663 attr_init (&truncation, zapt, 5);
664 truncation_value = attr_find (&truncation, NULL);
665 logf (LOG_DEBUG, "truncation value %d", truncation_value);
670 for (base_no = 0; base_no < num_bases; base_no++)
673 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
681 if (!relational_term (zi, zapt, term_sub, term_dict,
682 attributeSet, grep_info, &max_pos))
684 switch (truncation_value)
686 case -1: /* not specified */
687 case 100: /* do not truncate */
688 j = strlen(term_dict);
689 term_dict[j++] = '(';
690 for (i = 0; term_sub[i]; i++)
691 verbatim_char (term_sub[i], &j, term_dict);
692 strcpy (term_dict+j, ")");
693 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
694 &max_pos, grep_handle);
696 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
698 case 1: /* right truncation */
699 j = strlen(term_dict);
700 term_dict[j++] = '(';
701 for (i = 0; term_sub[i]; i++)
702 verbatim_char (term_sub[i], &j, term_dict);
703 strcpy (term_dict+j, ".*)");
704 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
705 &max_pos, grep_handle);
707 case 2: /* left truncation */
708 case 3: /* left&right truncation */
711 case 101: /* process # in term */
712 j = strlen(term_dict);
713 term_dict[j++] = '(';
714 for (i=0; term_sub[i]; i++)
715 if (term_sub[i] == '#' && i > 2)
717 term_dict[j++] = '.';
718 term_dict[j++] = '*';
721 verbatim_char (term_sub[i], &j, term_dict);
722 strcpy (term_dict+j, ")");
723 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
724 &max_pos, grep_handle);
726 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
729 case 102: /* regular expression */
730 sprintf (term_dict + strlen(term_dict), "(%s)", term_sub);
731 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
732 &max_pos, grep_handle);
734 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
739 if (max_pos <= strlen(basenames[base_no]))
741 zi->errCode = 109; /* Database unavailable */
742 zi->errString = basenames[base_no];
746 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
750 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
754 Z_Term *term = zapt->term;
756 sizez = term->u.general->len;
757 if (sizez > IT_MAX_WORD)
759 for (i = 0; i < sizez; i++)
760 termz[i] = index_char_cvt (term->u.general->buf[i]);
764 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
765 Z_AttributesPlusTerm *zapt,
766 oid_value attributeSet,
767 int num_bases, char **basenames)
769 rset_relevance_parms parms;
770 char termz[IT_MAX_WORD+1];
771 char term_sub[IT_MAX_WORD+1];
772 struct grep_info grep_info;
773 char *p0 = termz, *p1 = NULL;
776 parms.key_size = sizeof(struct it_key);
778 parms.cmp = key_compare;
779 parms.is = zi->wordIsam;
781 if (zapt->term->which != Z_Term_general)
786 trans_term (zi, zapt, termz);
787 grep_info.isam_p_indx = 0;
788 grep_info.isam_p_size = 0;
789 grep_info.isam_p_buf = NULL;
792 if ((p1 = strchr (p0, ' ')))
794 memcpy (term_sub, p0, p1-p0);
795 term_sub[p1-p0] = '\0';
798 strcpy (term_sub, p0);
799 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
800 num_bases, basenames))
808 parms.isam_positions = grep_info.isam_p_buf;
809 parms.no_isam_positions = grep_info.isam_p_indx;
810 if (grep_info.isam_p_indx > 0)
811 result = rset_create (rset_kind_relevance, &parms);
813 result = rset_create (rset_kind_null, NULL);
814 xfree (grep_info.isam_p_buf);
818 static RSET rpn_search_APT_word (ZServerInfo *zi,
819 Z_AttributesPlusTerm *zapt,
820 oid_value attributeSet,
821 int num_bases, char **basenames)
823 rset_isam_parms parms;
824 char termz[IT_MAX_WORD+1];
825 struct grep_info grep_info;
828 if (zapt->term->which != Z_Term_general)
833 trans_term (zi, zapt, termz);
835 grep_info.isam_p_indx = 0;
836 grep_info.isam_p_size = 0;
837 grep_info.isam_p_buf = NULL;
839 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
840 num_bases, basenames))
842 if (grep_info.isam_p_indx < 1)
843 result = rset_create (rset_kind_null, NULL);
844 else if (grep_info.isam_p_indx == 1)
846 parms.is = zi->wordIsam;
847 parms.pos = *grep_info.isam_p_buf;
848 result = rset_create (rset_kind_isam, &parms);
851 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
852 grep_info.isam_p_indx);
853 xfree (grep_info.isam_p_buf);
857 static RSET rpn_prox (RSET *rset, int rset_no)
865 rset_temp_parms parms;
867 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
868 more = xmalloc (sizeof(*more)*rset_no);
869 buf = xmalloc (sizeof(*buf)*rset_no);
871 for (i = 0; i<rset_no; i++)
873 buf[i] = xmalloc (sizeof(**buf));
874 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
875 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
879 rset_close (rset[i], rsfd[i]);
886 return rset_create (rset_kind_null, NULL);
889 parms.key_size = sizeof (struct it_key);
890 result = rset_create (rset_kind_temp, &parms);
891 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
895 for (i = 1; i<rset_no; i++)
904 cmp = key_compare (buf[i], buf[i-1]);
907 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
912 if (buf[i-1]->seqno+1 != buf[i]->seqno)
914 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
920 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
926 rset_write (result, rsfd_result, buf[0]);
927 more[0] = rset_read (*rset, *rsfd, *buf);
931 for (i = 0; i<rset_no; i++)
933 rset_close (rset[i], rsfd[i]);
936 rset_close (result, rsfd_result);
943 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
944 Z_AttributesPlusTerm *zapt,
945 oid_value attributeSet,
946 int num_bases, char **basenames)
948 char termz[IT_MAX_WORD+1];
949 char term_sub[IT_MAX_WORD+1];
950 char *p0 = termz, *p1 = NULL;
951 RSET rset[60], result;
953 struct grep_info grep_info;
955 if (zapt->term->which != Z_Term_general)
960 trans_term (zi, zapt, termz);
962 grep_info.isam_p_size = 0;
963 grep_info.isam_p_buf = NULL;
967 if ((p1 = strchr (p0, ' ')))
969 memcpy (term_sub, p0, p1-p0);
970 term_sub[p1-p0] = '\0';
973 strcpy (term_sub, p0);
975 grep_info.isam_p_indx = 0;
976 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
977 num_bases, basenames))
979 if (grep_info.isam_p_indx == 0)
980 rset[rset_no] = rset_create (rset_kind_null, NULL);
981 else if (grep_info.isam_p_indx > 1)
982 rset[rset_no] = rset_trunc (zi->wordIsam,
983 grep_info.isam_p_buf,
984 grep_info.isam_p_indx);
987 rset_isam_parms parms;
989 parms.is = zi->wordIsam;
990 parms.pos = *grep_info.isam_p_buf;
991 rset[rset_no] = rset_create (rset_kind_isam, &parms);
993 assert (rset[rset_no]);
994 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1002 xfree (grep_info.isam_p_buf);
1004 return rset_create (rset_kind_null, NULL);
1005 else if (rset_no == 1)
1008 result = rpn_prox (rset, rset_no);
1009 for (i = 0; i<rset_no; i++)
1010 rset_delete (rset[i]);
1014 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1015 oid_value attributeSet)
1020 rset_temp_parms parms;
1021 char termz[IT_MAX_WORD+1];
1023 if (zapt->term->which != Z_Term_general)
1028 parms.key_size = sizeof (struct it_key);
1029 result = rset_create (rset_kind_temp, &parms);
1030 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1032 trans_term (zi, zapt, termz);
1033 key.sysno = atoi (termz);
1036 rset_write (result, rsfd, &key);
1037 rset_close (result, rsfd);
1041 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1042 oid_value attributeSet,
1043 int num_bases, char **basenames)
1047 int relation_value, structure_value;
1049 attr_init (&relation, zapt, 2);
1050 attr_init (&structure, zapt, 4);
1052 relation_value = attr_find (&relation, NULL);
1053 structure_value = attr_find (&structure, NULL);
1054 switch (structure_value)
1057 if (relation_value == 102) /* relevance relation */
1058 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1059 num_bases, basenames);
1060 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1061 num_bases, basenames);
1062 case 1: /* phrase */
1063 if (relation_value == 102) /* relevance relation */
1064 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1065 num_bases, basenames);
1066 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1067 num_bases, basenames);
1070 if (relation_value == 102) /* relevance relation */
1071 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1072 num_bases, basenames);
1073 return rpn_search_APT_word (zi, zapt, attributeSet,
1074 num_bases, basenames);
1079 case 5: /* date - normalized */
1081 case 6: /* word list */
1082 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1083 num_bases, basenames);
1084 case 100: /* date - un-normalized */
1086 case 101: /* name - normalized */
1088 case 102: /* date - un-normalized */
1090 case 103: /* structure */
1094 case 105: /* free-form-text */
1095 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1096 num_bases, basenames);
1097 case 106: /* document-text */
1098 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1099 num_bases, basenames);
1100 case 107: /* local-number */
1101 return rpn_search_APT_local (zi, zapt, attributeSet);
1102 case 108: /* string */
1103 return rpn_search_APT_word (zi, zapt, attributeSet,
1104 num_bases, basenames);
1105 case 109: /* numeric string */
1112 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1116 if (!(s = resultSetGet (zi, resultSetId)))
1117 return rset_create (rset_kind_null, NULL);
1121 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1122 oid_value attributeSet,
1123 int num_bases, char **basenames)
1126 if (zs->which == Z_RPNStructure_complex)
1128 rset_bool_parms bool_parms;
1131 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1133 num_bases, basenames);
1134 if (bool_parms.rset_l == NULL)
1136 if (rset_is_ranked(bool_parms.rset_l))
1138 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1140 num_bases, basenames);
1141 if (bool_parms.rset_r == NULL)
1143 rset_delete (bool_parms.rset_l);
1146 if (rset_is_ranked(bool_parms.rset_r))
1148 bool_parms.key_size = sizeof(struct it_key);
1149 bool_parms.cmp = key_compare;
1151 switch (zs->u.complex->roperator->which)
1153 case Z_Operator_and:
1154 r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms);
1157 r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms);
1159 case Z_Operator_and_not:
1160 r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms);
1166 else if (zs->which == Z_RPNStructure_simple)
1168 if (zs->u.simple->which == Z_Operand_APT)
1170 logf (LOG_DEBUG, "rpn_search_APT");
1171 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1172 attributeSet, num_bases, basenames);
1174 else if (zs->u.simple->which == Z_Operand_resultSetId)
1176 logf (LOG_DEBUG, "rpn_search_ref");
1177 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1191 void count_set_save (RSET *r, int *count)
1198 rset_temp_parms parms;
1200 logf (LOG_DEBUG, "count_set_save");
1202 parms.key_size = sizeof(struct it_key);
1203 w = rset_create (rset_kind_temp, &parms);
1204 wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO);
1205 rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO);
1206 while (rset_read (*r, rfd, &key))
1208 if (key.sysno != psysno)
1210 rset_write (w, wfd, &key);
1216 rset_close (*r, rfd);
1218 rset_close (w, wfd);
1220 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1223 static void count_set (RSET r, int *count)
1230 logf (LOG_DEBUG, "count_set");
1232 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1233 while (rset_read (r, rfd, &key))
1235 if (key.sysno != psysno)
1242 rset_close (r, rfd);
1243 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1246 int rpn_search (ZServerInfo *zi,
1247 Z_RPNQuery *rpn, int num_bases, char **basenames,
1248 const char *setname, int *hits)
1252 oid_value attributeSet;
1257 zi->errString = NULL;
1259 attrset = oid_getentbyoid (rpn->attributeSetId);
1260 attributeSet = attrset->value;
1261 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1262 num_bases, basenames);
1265 if (rset_is_volatile(rset))
1266 count_set_save(&rset,hits);
1268 count_set (rset, hits);
1269 resultSetAdd (zi, setname, 1, rset);
1271 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1276 struct scan_entry *list;
1283 static int scan_handle (Dict_char *name, const char *info, int pos,
1286 int len_prefix, idx;
1289 struct scan_info *scan_info = client;
1291 rset_isam_parms parms;
1293 len_prefix = strlen(scan_info->prefix);
1294 if (memcmp (name, scan_info->prefix, len_prefix))
1297 idx = scan_info->after - pos + scan_info->before;
1300 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1301 strlen(name + len_prefix)+1);
1302 strcpy (scan_info->list[idx].term, name + len_prefix);
1303 assert (*info == sizeof(isam_p));
1304 memcpy (&isam_p, info+1, sizeof(isam_p));
1305 parms.is = scan_info->isam;
1308 rset = rset_create (rset_kind_isam, &parms);
1309 count_set (rset, &scan_info->list[idx].occurrences);
1312 scan_info->list[idx].occurrences = 1;
1314 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1319 static int dummy_handle (Dict_char *name, const char *info, void *p)
1324 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1325 int num_bases, char **basenames,
1326 int *position, int *num_entries, struct scan_entry **list,
1329 int i, j, sizez, max_pos;
1330 int pos = *position;
1331 int num = *num_entries;
1334 char termz[IT_MAX_WORD+20];
1337 Z_Term *term = zapt->term;
1338 struct scan_info scan_info;
1340 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1344 scan_info.before = before = pos-1;
1345 scan_info.after = after = 1+num-pos;
1346 scan_info.odr = zi->odr;
1348 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1350 scan_info.isam = zi->wordIsam;
1351 scan_info.list = odr_malloc (zi->odr, (before+after)*
1352 sizeof(*scan_info.list));
1353 for (j = 0; j<before+after; j++)
1354 scan_info.list[j].term = NULL;
1355 attr_init (&use, zapt, 1);
1356 use_value = attr_find (&use, NULL);
1357 logf (LOG_DEBUG, "use value %d", use_value);
1359 if (use_value == -1)
1361 i = index_word_prefix (termz, 1, use_value, *basenames);
1363 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
1365 if (max_pos <= strlen(*basenames))
1367 zi->errString = *basenames;
1368 return zi->errCode = 109; /* Database unavailable */
1370 strcpy (scan_info.prefix, termz);
1371 sizez = term->u.general->len;
1372 if (sizez > IT_MAX_WORD)
1373 sizez = IT_MAX_WORD;
1374 for (j = 0; j<sizez; j++)
1375 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1378 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1380 *status = BEND_SCAN_SUCCESS;
1382 for (i = 0; i<scan_info.after; i++)
1383 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1387 *status = BEND_SCAN_PARTIAL;
1389 for (i = 0; i<scan_info.before; i++)
1390 if (scan_info.list[i].term)
1393 *status = BEND_SCAN_PARTIAL;
1397 *list = scan_info.list+i; /* list is set to first 'real' entry */
1399 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1400 zi->errCode = 114; /* if no entries was found */
1401 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1402 *position, *num_entries);
1404 logf (LOG_DEBUG, "scan error: %d", zi->errCode);