2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.42 1996-05-09 07:28:56 quinn
8 * Work towards phrases and multiple registers
10 * Revision 1.41 1996/03/20 09:36:43 adam
11 * Function dict_lookup_grep got extra parameter, init_pos, which marks
12 * from which position in pattern approximate pattern matching should occur.
13 * Approximate pattern matching is used in relevance=re-2.
15 * Revision 1.40 1996/02/02 13:44:44 adam
16 * The public dictionary functions simply use char instead of Dict_char
17 * to represent search strings. Dict_char is used internally only.
19 * Revision 1.39 1996/01/03 16:22:13 quinn
22 * Revision 1.38 1995/12/11 09:12:55 adam
23 * The rec_get function returns NULL if record doesn't exist - will
24 * happen in the server if the result set records have been deleted since
25 * the creation of the set (i.e. the search).
26 * The server saves a result temporarily if it is 'volatile', i.e. the
27 * set is register dependent.
29 * Revision 1.37 1995/12/06 15:05:28 adam
30 * More verbose in count_set.
32 * Revision 1.36 1995/12/06 12:41:27 adam
33 * New command 'stat' for the index program.
34 * Filenames can be read from stdin by specifying '-'.
35 * Bug fix/enhancement of the transformation from terms to regular
36 * expressons in the search engine.
38 * Revision 1.35 1995/11/27 09:29:00 adam
39 * Bug fixes regarding conversion to regular expressions.
41 * Revision 1.34 1995/11/16 17:00:56 adam
42 * Better logging of rpn query.
44 * Revision 1.33 1995/11/01 13:58:28 quinn
45 * Moving data1 to yaz/retrieval
47 * Revision 1.32 1995/10/27 14:00:11 adam
48 * Implemented detection of database availability.
50 * Revision 1.31 1995/10/17 18:02:10 adam
51 * New feature: databases. Implemented as prefix to words in dictionary.
53 * Revision 1.30 1995/10/16 09:32:38 adam
54 * More work on relational op.
56 * Revision 1.29 1995/10/13 16:01:49 adam
59 * Revision 1.28 1995/10/13 12:26:43 adam
60 * Optimization of truncation.
62 * Revision 1.27 1995/10/12 17:07:22 adam
65 * Revision 1.26 1995/10/12 12:40:54 adam
66 * Bug fixes in rpn_prox.
68 * Revision 1.25 1995/10/10 13:59:24 adam
69 * Function rset_open changed its wflag parameter to general flags.
71 * Revision 1.24 1995/10/09 16:18:37 adam
72 * Function dict_lookup_grep got extra client data parameter.
74 * Revision 1.23 1995/10/06 16:33:37 adam
75 * Use attribute mappings.
77 * Revision 1.22 1995/10/06 15:07:39 adam
78 * Structure 'local-number' handled.
80 * Revision 1.21 1995/10/06 13:52:06 adam
81 * Bug fixes. Handler may abort further scanning.
83 * Revision 1.20 1995/10/06 11:06:33 adam
84 * Scan entries include 'occurrences' now.
86 * Revision 1.19 1995/10/06 10:43:56 adam
87 * Scan added. 'occurrences' in scan entries not set yet.
89 * Revision 1.18 1995/10/04 16:57:20 adam
90 * Key input and merge sort in one pass.
92 * Revision 1.17 1995/10/04 12:55:17 adam
93 * Bug fix in ranked search. Use=Any keys inserted.
95 * Revision 1.16 1995/10/02 16:24:40 adam
96 * Use attribute actually used in search requests.
98 * Revision 1.15 1995/10/02 15:18:52 adam
99 * New member in recRetrieveCtrl: diagnostic.
101 * Revision 1.14 1995/09/28 12:10:32 adam
102 * Bug fixes. Field prefix used in queries.
104 * Revision 1.13 1995/09/18 14:17:50 adam
107 * Revision 1.12 1995/09/15 14:45:21 adam
109 * Work on truncation.
111 * Revision 1.11 1995/09/14 11:53:27 adam
112 * First work on regular expressions/truncations.
114 * Revision 1.10 1995/09/11 15:23:26 adam
115 * More work on relevance search.
117 * Revision 1.9 1995/09/11 13:09:35 adam
118 * More work on relevance feedback.
120 * Revision 1.8 1995/09/08 14:52:27 adam
121 * Minor changes. Dictionary is lower case now.
123 * Revision 1.7 1995/09/07 13:58:36 adam
124 * New parameter: result-set file descriptor (RSFD) to support multiple
125 * positions within the same result-set.
126 * Boolean operators: and, or, not implemented.
127 * Result-set references.
129 * Revision 1.6 1995/09/06 16:11:18 adam
130 * Option: only one word key per file.
132 * Revision 1.5 1995/09/06 10:33:04 adam
133 * More work on present. Some log messages removed.
135 * Revision 1.4 1995/09/05 15:28:40 adam
136 * More work on search engine.
138 * Revision 1.3 1995/09/04 15:20:22 adam
141 * Revision 1.2 1995/09/04 12:33:43 adam
142 * Various cleanup. YAZ util used instead.
144 * Revision 1.1 1995/09/04 09:10:40 adam
145 * More work on index add/del/update.
146 * Merge sort implemented.
147 * Initial work on z39 server.
156 #include "attribute.h"
164 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
169 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
170 attp = att_getentbyatt (attrSet, attrUse);
173 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
174 return index_word_prefix (string, attp->attset_ordinal,
175 attp->local_attributes->local, basename);
182 Z_AttributesPlusTerm *zapt;
185 static int attr_find (AttrType *src, oid_value *attributeSetP)
187 while (src->major < src->zapt->num_attributes)
189 Z_AttributeElement *element;
191 element = src->zapt->attributeList[src->major];
192 if (src->type == *element->attributeType)
194 switch (element->which)
196 case Z_AttributeValue_numeric:
198 if (element->attributeSet && attributeSetP)
202 attrset = oid_getentbyoid (element->attributeSet);
203 *attributeSetP = attrset->value;
205 return *element->value.numeric;
207 case Z_AttributeValue_complex:
208 if (src->minor >= element->value.complex->num_list ||
209 element->value.complex->list[src->minor]->which !=
210 Z_StringOrNumeric_numeric)
213 if (element->attributeSet && attributeSetP)
217 attrset = oid_getentbyoid (element->attributeSet);
218 *attributeSetP = attrset->value;
220 return *element->value.complex->list[src->minor-1]->u.numeric;
230 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
244 int (*cmp)(const void *p1, const void *p2);
251 static void heap_swap (struct trunc_info *ti, int i1, int i2)
256 ti->ptr[i1] = ti->ptr[i2];
260 static void heap_delete (struct trunc_info *ti)
262 int cur = 1, child = 2;
264 heap_swap (ti, 1, ti->heapnum--);
265 while (child <= ti->heapnum) {
266 if (child < ti->heapnum &&
267 (*ti->cmp)(ti->heap[ti->ptr[child]],
268 ti->heap[ti->ptr[1+child]]) > 0)
270 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
271 ti->heap[ti->ptr[child]]) > 0)
273 heap_swap (ti, cur, child);
282 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
286 cur = ++(ti->heapnum);
287 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
288 ti->indx[ti->ptr[cur]] = indx;
290 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
291 ti->heap[ti->ptr[cur]]) > 0)
293 heap_swap (ti, cur, parent);
300 struct trunc_info *heap_init (int size, int key_size,
301 int (*cmp)(const void *p1, const void *p2))
303 struct trunc_info *ti = xmalloc (sizeof(*ti));
308 ti->keysize = key_size;
310 ti->indx = xmalloc (size * sizeof(*ti->indx));
311 ti->heap = xmalloc (size * sizeof(*ti->heap));
312 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
313 ti->swapbuf = xmalloc (ti->keysize);
314 ti->tmpbuf = xmalloc (ti->keysize);
315 ti->buf = xmalloc (size * ti->keysize);
316 for (i = size; --i >= 0; )
319 ti->heap[i] = ti->buf + ti->keysize * i;
324 static void heap_close (struct trunc_info *ti)
334 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
339 rset_temp_parms parms;
341 parms.key_size = sizeof(struct it_key);
342 result = rset_create (rset_kind_temp, &parms);
343 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
345 if (to - from > merge_chunk)
349 int i, i_add = (to-from)/merge_chunk + 1;
350 struct trunc_info *ti;
352 int rsmax = (to-from)/i_add + 1;
354 rset = xmalloc (sizeof(*rset) * rsmax);
355 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
357 for (i = from; i < to; i += i_add)
360 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
363 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
367 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
368 for (i = rscur; --i >= 0; )
370 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
371 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
372 heap_insert (ti, ti->tmpbuf, i);
375 rset_close (rset[i], rsfd[i]);
376 rset_delete (rset[i]);
381 int n = ti->indx[ti->ptr[1]];
383 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
387 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
390 rset_close (rset[n], rsfd[n]);
391 rset_delete (rset[n]);
394 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
397 heap_insert (ti, ti->tmpbuf, n);
410 struct trunc_info *ti;
412 ispt = xmalloc (sizeof(*ispt) * (to-from));
414 ti = heap_init (to-from, sizeof(struct it_key),
416 for (i = to-from; --i >= 0; )
418 ispt[i] = is_position (isam, isam_p[from+i]);
419 if (is_readkey (ispt[i], ti->tmpbuf))
420 heap_insert (ti, ti->tmpbuf, i);
422 is_pt_free (ispt[i]);
426 int n = ti->indx[ti->ptr[1]];
428 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
430 /* section that preserve all keys */
432 if (is_readkey (ispt[n], ti->tmpbuf))
433 heap_insert (ti, ti->tmpbuf, n);
435 is_pt_free (ispt[n]);
437 /* section that preserve all keys with unique sysnos */
440 if (!is_readkey (ispt[n], ti->tmpbuf))
443 is_pt_free (ispt[n]);
446 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
449 heap_insert (ti, ti->tmpbuf, n);
458 rset_close (result, result_rsfd);
462 static int isam_trunc_cmp (const void *p1, const void *p2)
464 ISAM_P i1 = *(ISAM_P*) p1;
465 ISAM_P i2 = *(ISAM_P*) p2;
468 d = is_type (i1) - is_type (i2);
471 return is_block (i1) - is_block (i2);
474 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
477 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
478 return rset_trunc_r (isam, isam_p, 0, no, 100);
487 static void add_isam_p (const char *info, struct grep_info *p)
489 if (p->isam_p_indx == p->isam_p_size)
491 ISAM_P *new_isam_p_buf;
493 p->isam_p_size = 2*p->isam_p_size + 100;
494 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
498 memcpy (new_isam_p_buf, p->isam_p_buf,
499 p->isam_p_indx * sizeof(*p->isam_p_buf));
500 xfree (p->isam_p_buf);
502 p->isam_p_buf = new_isam_p_buf;
504 assert (*info == sizeof(*p->isam_p_buf));
505 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
509 static int grep_handle (char *name, const char *info, void *p)
511 logf (LOG_DEBUG, "dict name: %s", name);
512 add_isam_p (info, p);
516 static void gen_regular_rel (char *dst, int val, int islt)
524 sprintf (numstr, "%d", val);
525 for (w = strlen(numstr); --w >= 0; pos++)
544 strcpy (dst + dst_p, numstr);
545 dst_p = strlen(dst) - pos - 1;
573 for (i = 0; i<pos; i++)
586 for (i=1; i<pos; i++)
587 strcat (dst, "[0-9]?");
591 for (i = 0; i <= pos; i++)
592 strcat (dst, "[0-9]");
593 strcat (dst, "[0-9]*");
598 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
599 const char *term_sub,
601 oid_value attributeSet,
602 struct grep_info *grep_info,
610 attr_init (&relation, zapt, 2);
611 relation_value = attr_find (&relation, NULL);
612 term_value = atoi (term_sub);
614 switch (relation_value)
619 logf (LOG_DEBUG, "Relation <");
620 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
625 logf (LOG_DEBUG, "Relation <=");
626 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
631 logf (LOG_DEBUG, "Relation >=");
632 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
637 logf (LOG_DEBUG, "Relation >");
638 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
643 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
644 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
647 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
648 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
652 static void verbatim_char (int ch, int *indx, char *dst)
655 dst[(*indx)++] = '\\';
659 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
660 const char *term_sub,
661 oid_value attributeSet, struct grep_info *grep_info,
662 int num_bases, char **basenames)
664 char term_dict[2*IT_MAX_WORD+2];
665 int i, j, r, base_no;
667 int truncation_value;
670 oid_value curAttributeSet = attributeSet;
672 attr_init (&use, zapt, 1);
673 use_value = attr_find (&use, &curAttributeSet);
674 logf (LOG_DEBUG, "use value %d", use_value);
675 attr_init (&truncation, zapt, 5);
676 truncation_value = attr_find (&truncation, NULL);
677 logf (LOG_DEBUG, "truncation value %d", truncation_value);
682 for (base_no = 0; base_no < num_bases; base_no++)
685 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
693 if (!relational_term (zi, zapt, term_sub, term_dict,
694 attributeSet, grep_info, &max_pos))
699 switch (truncation_value)
701 case -1: /* not specified */
702 case 100: /* do not truncate */
703 term_dict[j++] = '(';
704 for (i = 0; term_sub[i]; i++)
705 verbatim_char (term_sub[i], &j, term_dict);
706 strcpy (term_dict+j, ")");
707 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
708 &max_pos, 0, grep_handle);
710 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
712 case 1: /* right truncation */
713 term_dict[j++] = '(';
714 for (i = 0; term_sub[i]; i++)
715 verbatim_char (term_sub[i], &j, term_dict);
716 strcpy (term_dict+j, ".*)");
717 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
718 &max_pos, 0, grep_handle);
720 case 2: /* left truncation */
721 case 3: /* left&right truncation */
724 case 101: /* process # in term */
725 term_dict[j++] = '(';
726 for (i=0; term_sub[i]; i++)
727 if (term_sub[i] == '#' && i > 2)
729 term_dict[j++] = '.';
730 term_dict[j++] = '*';
733 verbatim_char (term_sub[i], &j, term_dict);
734 strcpy (term_dict+j, ")");
735 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
736 &max_pos, 0, grep_handle);
738 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
741 case 102: /* regular expression */
742 sprintf (term_dict + j, "(%s)", term_sub);
743 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
744 &max_pos, 0, grep_handle);
746 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
749 case 103: /* regular expression with error correction */
752 if (*cp == '*' && cp[1] && cp[2])
757 sprintf (term_dict + j, "(%s)", cp);
758 r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info,
759 &max_pos, j, grep_handle);
761 logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
766 if (max_pos <= strlen(basenames[base_no]))
768 zi->errCode = 109; /* Database unavailable */
769 zi->errString = basenames[base_no];
773 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
777 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
781 Z_Term *term = zapt->term;
783 sizez = term->u.general->len;
784 if (sizez > IT_MAX_WORD)
786 for (i = 0; i < sizez; i++)
787 termz[i] = index_char_cvt (term->u.general->buf[i]);
791 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
792 Z_AttributesPlusTerm *zapt,
793 oid_value attributeSet,
794 int num_bases, char **basenames)
796 rset_relevance_parms parms;
797 char termz[IT_MAX_WORD+1];
798 char term_sub[IT_MAX_WORD+1];
799 struct grep_info grep_info;
800 char *p0 = termz, *p1 = NULL;
803 parms.key_size = sizeof(struct it_key);
805 parms.cmp = key_compare;
806 parms.is = zi->wordIsam;
808 if (zapt->term->which != Z_Term_general)
813 trans_term (zi, zapt, termz);
814 grep_info.isam_p_indx = 0;
815 grep_info.isam_p_size = 0;
816 grep_info.isam_p_buf = NULL;
819 if ((p1 = strchr (p0, ' ')))
821 memcpy (term_sub, p0, p1-p0);
822 term_sub[p1-p0] = '\0';
825 strcpy (term_sub, p0);
826 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
827 num_bases, basenames))
835 parms.isam_positions = grep_info.isam_p_buf;
836 parms.no_isam_positions = grep_info.isam_p_indx;
837 if (grep_info.isam_p_indx > 0)
838 result = rset_create (rset_kind_relevance, &parms);
840 result = rset_create (rset_kind_null, NULL);
841 xfree (grep_info.isam_p_buf);
845 static RSET rpn_search_APT_word (ZServerInfo *zi,
846 Z_AttributesPlusTerm *zapt,
847 oid_value attributeSet,
848 int num_bases, char **basenames)
850 rset_isam_parms parms;
851 char termz[IT_MAX_WORD+1];
852 struct grep_info grep_info;
855 if (zapt->term->which != Z_Term_general)
860 trans_term (zi, zapt, termz);
862 grep_info.isam_p_indx = 0;
863 grep_info.isam_p_size = 0;
864 grep_info.isam_p_buf = NULL;
866 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
867 num_bases, basenames))
869 if (grep_info.isam_p_indx < 1)
870 result = rset_create (rset_kind_null, NULL);
871 else if (grep_info.isam_p_indx == 1)
873 parms.is = zi->wordIsam;
874 parms.pos = *grep_info.isam_p_buf;
875 result = rset_create (rset_kind_isam, &parms);
878 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
879 grep_info.isam_p_indx);
880 xfree (grep_info.isam_p_buf);
884 static RSET rpn_prox (RSET *rset, int rset_no)
892 rset_temp_parms parms;
894 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
895 more = xmalloc (sizeof(*more)*rset_no);
896 buf = xmalloc (sizeof(*buf)*rset_no);
898 for (i = 0; i<rset_no; i++)
900 buf[i] = xmalloc (sizeof(**buf));
901 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
902 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
906 rset_close (rset[i], rsfd[i]);
913 return rset_create (rset_kind_null, NULL);
916 parms.key_size = sizeof (struct it_key);
917 result = rset_create (rset_kind_temp, &parms);
918 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
922 for (i = 1; i<rset_no; i++)
931 cmp = key_compare (buf[i], buf[i-1]);
934 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
939 if (buf[i-1]->seqno+1 != buf[i]->seqno)
941 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
947 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
953 rset_write (result, rsfd_result, buf[0]);
954 more[0] = rset_read (*rset, *rsfd, *buf);
958 for (i = 0; i<rset_no; i++)
960 rset_close (rset[i], rsfd[i]);
963 rset_close (result, rsfd_result);
970 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
971 Z_AttributesPlusTerm *zapt,
972 oid_value attributeSet,
973 int num_bases, char **basenames)
975 char termz[IT_MAX_WORD+1];
976 char term_sub[IT_MAX_WORD+1];
977 char *p0 = termz, *p1 = NULL;
978 RSET rset[60], result;
980 struct grep_info grep_info;
982 if (zapt->term->which != Z_Term_general)
987 trans_term (zi, zapt, termz);
989 grep_info.isam_p_size = 0;
990 grep_info.isam_p_buf = NULL;
994 if ((p1 = strchr (p0, ' ')))
996 memcpy (term_sub, p0, p1-p0);
997 term_sub[p1-p0] = '\0';
1000 strcpy (term_sub, p0);
1002 grep_info.isam_p_indx = 0;
1003 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
1004 num_bases, basenames))
1006 if (grep_info.isam_p_indx == 0)
1007 rset[rset_no] = rset_create (rset_kind_null, NULL);
1008 else if (grep_info.isam_p_indx > 1)
1009 rset[rset_no] = rset_trunc (zi->wordIsam,
1010 grep_info.isam_p_buf,
1011 grep_info.isam_p_indx);
1014 rset_isam_parms parms;
1016 parms.is = zi->wordIsam;
1017 parms.pos = *grep_info.isam_p_buf;
1018 rset[rset_no] = rset_create (rset_kind_isam, &parms);
1020 assert (rset[rset_no]);
1021 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1026 while (*++p0 == ' ')
1029 xfree (grep_info.isam_p_buf);
1031 return rset_create (rset_kind_null, NULL);
1032 else if (rset_no == 1)
1035 result = rpn_prox (rset, rset_no);
1036 for (i = 0; i<rset_no; i++)
1037 rset_delete (rset[i]);
1041 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1042 oid_value attributeSet)
1047 rset_temp_parms parms;
1048 char termz[IT_MAX_WORD+1];
1050 if (zapt->term->which != Z_Term_general)
1055 parms.key_size = sizeof (struct it_key);
1056 result = rset_create (rset_kind_temp, &parms);
1057 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1059 trans_term (zi, zapt, termz);
1060 key.sysno = atoi (termz);
1063 rset_write (result, rsfd, &key);
1064 rset_close (result, rsfd);
1068 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1069 oid_value attributeSet,
1070 int num_bases, char **basenames)
1074 int relation_value, structure_value;
1076 attr_init (&relation, zapt, 2);
1077 attr_init (&structure, zapt, 4);
1079 relation_value = attr_find (&relation, NULL);
1080 structure_value = attr_find (&structure, NULL);
1081 switch (structure_value)
1084 if (relation_value == 102) /* relevance relation */
1085 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1086 num_bases, basenames);
1087 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1088 num_bases, basenames);
1089 case 1: /* phrase */
1090 if (relation_value == 102) /* relevance relation */
1091 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1092 num_bases, basenames);
1093 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1094 num_bases, basenames);
1097 if (relation_value == 102) /* relevance relation */
1098 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1099 num_bases, basenames);
1100 return rpn_search_APT_word (zi, zapt, attributeSet,
1101 num_bases, basenames);
1106 case 5: /* date - normalized */
1108 case 6: /* word list */
1109 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1110 num_bases, basenames);
1111 case 100: /* date - un-normalized */
1113 case 101: /* name - normalized */
1115 case 102: /* date - un-normalized */
1117 case 103: /* structure */
1121 case 105: /* free-form-text */
1122 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1123 num_bases, basenames);
1124 case 106: /* document-text */
1125 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1126 num_bases, basenames);
1127 case 107: /* local-number */
1128 return rpn_search_APT_local (zi, zapt, attributeSet);
1129 case 108: /* string */
1130 return rpn_search_APT_word (zi, zapt, attributeSet,
1131 num_bases, basenames);
1132 case 109: /* numeric string */
1139 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1143 if (!(s = resultSetGet (zi, resultSetId)))
1144 return rset_create (rset_kind_null, NULL);
1148 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1149 oid_value attributeSet,
1150 int num_bases, char **basenames)
1153 if (zs->which == Z_RPNStructure_complex)
1155 rset_bool_parms bool_parms;
1158 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1160 num_bases, basenames);
1161 if (bool_parms.rset_l == NULL)
1163 if (rset_is_ranked(bool_parms.rset_l))
1165 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1167 num_bases, basenames);
1168 if (bool_parms.rset_r == NULL)
1170 rset_delete (bool_parms.rset_l);
1173 if (rset_is_ranked(bool_parms.rset_r))
1175 bool_parms.key_size = sizeof(struct it_key);
1176 bool_parms.cmp = key_compare;
1178 switch (zs->u.complex->roperator->which)
1180 case Z_Operator_and:
1181 r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms);
1184 r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms);
1186 case Z_Operator_and_not:
1187 r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms);
1193 else if (zs->which == Z_RPNStructure_simple)
1195 if (zs->u.simple->which == Z_Operand_APT)
1197 logf (LOG_DEBUG, "rpn_search_APT");
1198 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1199 attributeSet, num_bases, basenames);
1201 else if (zs->u.simple->which == Z_Operand_resultSetId)
1203 logf (LOG_DEBUG, "rpn_search_ref");
1204 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1218 void count_set_save (RSET *r, int *count)
1225 rset_temp_parms parms;
1227 logf (LOG_DEBUG, "count_set_save");
1229 parms.key_size = sizeof(struct it_key);
1230 w = rset_create (rset_kind_temp, &parms);
1231 wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO);
1232 rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO);
1233 while (rset_read (*r, rfd, &key))
1235 if (key.sysno != psysno)
1237 rset_write (w, wfd, &key);
1243 rset_close (*r, rfd);
1245 rset_close (w, wfd);
1247 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1250 static void count_set (RSET r, int *count)
1257 logf (LOG_DEBUG, "count_set");
1259 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1260 while (rset_read (r, rfd, &key))
1262 if (key.sysno != psysno)
1269 rset_close (r, rfd);
1270 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1273 int rpn_search (ZServerInfo *zi,
1274 Z_RPNQuery *rpn, int num_bases, char **basenames,
1275 const char *setname, int *hits)
1279 oid_value attributeSet;
1284 zi->errString = NULL;
1286 attrset = oid_getentbyoid (rpn->attributeSetId);
1287 attributeSet = attrset->value;
1288 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1289 num_bases, basenames);
1292 if (rset_is_volatile(rset))
1293 count_set_save(&rset,hits);
1295 count_set (rset, hits);
1296 resultSetAdd (zi, setname, 1, rset);
1298 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1303 struct scan_entry *list;
1310 static int scan_handle (char *name, const char *info, int pos, void *client)
1312 int len_prefix, idx;
1315 struct scan_info *scan_info = client;
1317 rset_isam_parms parms;
1319 len_prefix = strlen(scan_info->prefix);
1320 if (memcmp (name, scan_info->prefix, len_prefix))
1323 idx = scan_info->after - pos + scan_info->before;
1326 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1327 strlen(name + len_prefix)+1);
1328 strcpy (scan_info->list[idx].term, name + len_prefix);
1329 assert (*info == sizeof(isam_p));
1330 memcpy (&isam_p, info+1, sizeof(isam_p));
1331 parms.is = scan_info->isam;
1334 rset = rset_create (rset_kind_isam, &parms);
1335 count_set (rset, &scan_info->list[idx].occurrences);
1338 scan_info->list[idx].occurrences = 1;
1340 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1345 static int dummy_handle (char *name, const char *info, void *p)
1350 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1351 int num_bases, char **basenames,
1352 int *position, int *num_entries, struct scan_entry **list,
1355 int i, j, sizez, max_pos;
1356 int pos = *position;
1357 int num = *num_entries;
1360 char termz[IT_MAX_WORD+20];
1363 Z_Term *term = zapt->term;
1364 struct scan_info scan_info;
1366 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1370 scan_info.before = before = pos-1;
1371 scan_info.after = after = 1+num-pos;
1372 scan_info.odr = zi->odr;
1374 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1376 scan_info.isam = zi->wordIsam;
1377 scan_info.list = odr_malloc (zi->odr, (before+after)*
1378 sizeof(*scan_info.list));
1379 for (j = 0; j<before+after; j++)
1380 scan_info.list[j].term = NULL;
1381 attr_init (&use, zapt, 1);
1382 use_value = attr_find (&use, NULL);
1383 logf (LOG_DEBUG, "use value %d", use_value);
1385 if (use_value == -1)
1387 i = index_word_prefix (termz, 1, use_value, *basenames);
1389 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos, 0,
1391 if (max_pos <= strlen(*basenames))
1393 zi->errString = *basenames;
1394 return zi->errCode = 109; /* Database unavailable */
1396 strcpy (scan_info.prefix, termz);
1397 sizez = term->u.general->len;
1398 if (sizez > IT_MAX_WORD)
1399 sizez = IT_MAX_WORD;
1400 for (j = 0; j<sizez; j++)
1401 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1404 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1406 *status = BEND_SCAN_SUCCESS;
1408 for (i = 0; i<scan_info.after; i++)
1409 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1413 *status = BEND_SCAN_PARTIAL;
1415 for (i = 0; i<scan_info.before; i++)
1416 if (scan_info.list[i].term)
1419 *status = BEND_SCAN_PARTIAL;
1423 *list = scan_info.list+i; /* list is set to first 'real' entry */
1425 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1426 zi->errCode = 114; /* if no entries was found */
1427 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1428 *position, *num_entries);
1430 logf (LOG_DEBUG, "scan error: %d", zi->errCode);