2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.41 1996-03-20 09:36:43 adam
8 * Function dict_lookup_grep got extra parameter, init_pos, which marks
9 * from which position in pattern approximate pattern matching should occur.
10 * Approximate pattern matching is used in relevance=re-2.
12 * Revision 1.40 1996/02/02 13:44:44 adam
13 * The public dictionary functions simply use char instead of Dict_char
14 * to represent search strings. Dict_char is used internally only.
16 * Revision 1.39 1996/01/03 16:22:13 quinn
19 * Revision 1.38 1995/12/11 09:12:55 adam
20 * The rec_get function returns NULL if record doesn't exist - will
21 * happen in the server if the result set records have been deleted since
22 * the creation of the set (i.e. the search).
23 * The server saves a result temporarily if it is 'volatile', i.e. the
24 * set is register dependent.
26 * Revision 1.37 1995/12/06 15:05:28 adam
27 * More verbose in count_set.
29 * Revision 1.36 1995/12/06 12:41:27 adam
30 * New command 'stat' for the index program.
31 * Filenames can be read from stdin by specifying '-'.
32 * Bug fix/enhancement of the transformation from terms to regular
33 * expressons in the search engine.
35 * Revision 1.35 1995/11/27 09:29:00 adam
36 * Bug fixes regarding conversion to regular expressions.
38 * Revision 1.34 1995/11/16 17:00:56 adam
39 * Better logging of rpn query.
41 * Revision 1.33 1995/11/01 13:58:28 quinn
42 * Moving data1 to yaz/retrieval
44 * Revision 1.32 1995/10/27 14:00:11 adam
45 * Implemented detection of database availability.
47 * Revision 1.31 1995/10/17 18:02:10 adam
48 * New feature: databases. Implemented as prefix to words in dictionary.
50 * Revision 1.30 1995/10/16 09:32:38 adam
51 * More work on relational op.
53 * Revision 1.29 1995/10/13 16:01:49 adam
56 * Revision 1.28 1995/10/13 12:26:43 adam
57 * Optimization of truncation.
59 * Revision 1.27 1995/10/12 17:07:22 adam
62 * Revision 1.26 1995/10/12 12:40:54 adam
63 * Bug fixes in rpn_prox.
65 * Revision 1.25 1995/10/10 13:59:24 adam
66 * Function rset_open changed its wflag parameter to general flags.
68 * Revision 1.24 1995/10/09 16:18:37 adam
69 * Function dict_lookup_grep got extra client data parameter.
71 * Revision 1.23 1995/10/06 16:33:37 adam
72 * Use attribute mappings.
74 * Revision 1.22 1995/10/06 15:07:39 adam
75 * Structure 'local-number' handled.
77 * Revision 1.21 1995/10/06 13:52:06 adam
78 * Bug fixes. Handler may abort further scanning.
80 * Revision 1.20 1995/10/06 11:06:33 adam
81 * Scan entries include 'occurrences' now.
83 * Revision 1.19 1995/10/06 10:43:56 adam
84 * Scan added. 'occurrences' in scan entries not set yet.
86 * Revision 1.18 1995/10/04 16:57:20 adam
87 * Key input and merge sort in one pass.
89 * Revision 1.17 1995/10/04 12:55:17 adam
90 * Bug fix in ranked search. Use=Any keys inserted.
92 * Revision 1.16 1995/10/02 16:24:40 adam
93 * Use attribute actually used in search requests.
95 * Revision 1.15 1995/10/02 15:18:52 adam
96 * New member in recRetrieveCtrl: diagnostic.
98 * Revision 1.14 1995/09/28 12:10:32 adam
99 * Bug fixes. Field prefix used in queries.
101 * Revision 1.13 1995/09/18 14:17:50 adam
104 * Revision 1.12 1995/09/15 14:45:21 adam
106 * Work on truncation.
108 * Revision 1.11 1995/09/14 11:53:27 adam
109 * First work on regular expressions/truncations.
111 * Revision 1.10 1995/09/11 15:23:26 adam
112 * More work on relevance search.
114 * Revision 1.9 1995/09/11 13:09:35 adam
115 * More work on relevance feedback.
117 * Revision 1.8 1995/09/08 14:52:27 adam
118 * Minor changes. Dictionary is lower case now.
120 * Revision 1.7 1995/09/07 13:58:36 adam
121 * New parameter: result-set file descriptor (RSFD) to support multiple
122 * positions within the same result-set.
123 * Boolean operators: and, or, not implemented.
124 * Result-set references.
126 * Revision 1.6 1995/09/06 16:11:18 adam
127 * Option: only one word key per file.
129 * Revision 1.5 1995/09/06 10:33:04 adam
130 * More work on present. Some log messages removed.
132 * Revision 1.4 1995/09/05 15:28:40 adam
133 * More work on search engine.
135 * Revision 1.3 1995/09/04 15:20:22 adam
138 * Revision 1.2 1995/09/04 12:33:43 adam
139 * Various cleanup. YAZ util used instead.
141 * Revision 1.1 1995/09/04 09:10:40 adam
142 * More work on index add/del/update.
143 * Merge sort implemented.
144 * Initial work on z39 server.
153 #include "attribute.h"
161 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
166 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
167 attp = att_getentbyatt (attrSet, attrUse);
170 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
171 return index_word_prefix (string, attp->attset_ordinal,
172 attp->local_attribute, basename);
179 Z_AttributesPlusTerm *zapt;
182 static int attr_find (AttrType *src, oid_value *attributeSetP)
184 while (src->major < src->zapt->num_attributes)
186 Z_AttributeElement *element;
188 element = src->zapt->attributeList[src->major];
189 if (src->type == *element->attributeType)
191 switch (element->which)
193 case Z_AttributeValue_numeric:
195 if (element->attributeSet && attributeSetP)
199 attrset = oid_getentbyoid (element->attributeSet);
200 *attributeSetP = attrset->value;
202 return *element->value.numeric;
204 case Z_AttributeValue_complex:
205 if (src->minor >= element->value.complex->num_list ||
206 element->value.complex->list[src->minor]->which !=
207 Z_StringOrNumeric_numeric)
210 if (element->attributeSet && attributeSetP)
214 attrset = oid_getentbyoid (element->attributeSet);
215 *attributeSetP = attrset->value;
217 return *element->value.complex->list[src->minor-1]->u.numeric;
227 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
241 int (*cmp)(const void *p1, const void *p2);
248 static void heap_swap (struct trunc_info *ti, int i1, int i2)
253 ti->ptr[i1] = ti->ptr[i2];
257 static void heap_delete (struct trunc_info *ti)
259 int cur = 1, child = 2;
261 heap_swap (ti, 1, ti->heapnum--);
262 while (child <= ti->heapnum) {
263 if (child < ti->heapnum &&
264 (*ti->cmp)(ti->heap[ti->ptr[child]],
265 ti->heap[ti->ptr[1+child]]) > 0)
267 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
268 ti->heap[ti->ptr[child]]) > 0)
270 heap_swap (ti, cur, child);
279 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
283 cur = ++(ti->heapnum);
284 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
285 ti->indx[ti->ptr[cur]] = indx;
287 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
288 ti->heap[ti->ptr[cur]]) > 0)
290 heap_swap (ti, cur, parent);
297 struct trunc_info *heap_init (int size, int key_size,
298 int (*cmp)(const void *p1, const void *p2))
300 struct trunc_info *ti = xmalloc (sizeof(*ti));
305 ti->keysize = key_size;
307 ti->indx = xmalloc (size * sizeof(*ti->indx));
308 ti->heap = xmalloc (size * sizeof(*ti->heap));
309 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
310 ti->swapbuf = xmalloc (ti->keysize);
311 ti->tmpbuf = xmalloc (ti->keysize);
312 ti->buf = xmalloc (size * ti->keysize);
313 for (i = size; --i >= 0; )
316 ti->heap[i] = ti->buf + ti->keysize * i;
321 static void heap_close (struct trunc_info *ti)
331 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
336 rset_temp_parms parms;
338 parms.key_size = sizeof(struct it_key);
339 result = rset_create (rset_kind_temp, &parms);
340 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
342 if (to - from > merge_chunk)
346 int i, i_add = (to-from)/merge_chunk + 1;
347 struct trunc_info *ti;
349 int rsmax = (to-from)/i_add + 1;
351 rset = xmalloc (sizeof(*rset) * rsmax);
352 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
354 for (i = from; i < to; i += i_add)
357 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
360 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
364 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
365 for (i = rscur; --i >= 0; )
367 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
368 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
369 heap_insert (ti, ti->tmpbuf, i);
372 rset_close (rset[i], rsfd[i]);
373 rset_delete (rset[i]);
378 int n = ti->indx[ti->ptr[1]];
380 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
384 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
387 rset_close (rset[n], rsfd[n]);
388 rset_delete (rset[n]);
391 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
394 heap_insert (ti, ti->tmpbuf, n);
407 struct trunc_info *ti;
409 ispt = xmalloc (sizeof(*ispt) * (to-from));
411 ti = heap_init (to-from, sizeof(struct it_key),
413 for (i = to-from; --i >= 0; )
415 ispt[i] = is_position (isam, isam_p[from+i]);
416 if (is_readkey (ispt[i], ti->tmpbuf))
417 heap_insert (ti, ti->tmpbuf, i);
419 is_pt_free (ispt[i]);
423 int n = ti->indx[ti->ptr[1]];
425 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
427 /* section that preserve all keys */
429 if (is_readkey (ispt[n], ti->tmpbuf))
430 heap_insert (ti, ti->tmpbuf, n);
432 is_pt_free (ispt[n]);
434 /* section that preserve all keys with unique sysnos */
437 if (!is_readkey (ispt[n], ti->tmpbuf))
440 is_pt_free (ispt[n]);
443 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
446 heap_insert (ti, ti->tmpbuf, n);
455 rset_close (result, result_rsfd);
459 static int isam_trunc_cmp (const void *p1, const void *p2)
461 ISAM_P i1 = *(ISAM_P*) p1;
462 ISAM_P i2 = *(ISAM_P*) p2;
465 d = is_type (i1) - is_type (i2);
468 return is_block (i1) - is_block (i2);
471 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
474 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
475 return rset_trunc_r (isam, isam_p, 0, no, 100);
484 static void add_isam_p (const char *info, struct grep_info *p)
486 if (p->isam_p_indx == p->isam_p_size)
488 ISAM_P *new_isam_p_buf;
490 p->isam_p_size = 2*p->isam_p_size + 100;
491 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
495 memcpy (new_isam_p_buf, p->isam_p_buf,
496 p->isam_p_indx * sizeof(*p->isam_p_buf));
497 xfree (p->isam_p_buf);
499 p->isam_p_buf = new_isam_p_buf;
501 assert (*info == sizeof(*p->isam_p_buf));
502 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
506 static int grep_handle (char *name, const char *info, void *p)
508 logf (LOG_DEBUG, "dict name: %s", name);
509 add_isam_p (info, p);
513 static void gen_regular_rel (char *dst, int val, int islt)
521 sprintf (numstr, "%d", val);
522 for (w = strlen(numstr); --w >= 0; pos++)
541 strcpy (dst + dst_p, numstr);
542 dst_p = strlen(dst) - pos - 1;
570 for (i = 0; i<pos; i++)
583 for (i=1; i<pos; i++)
584 strcat (dst, "[0-9]?");
588 for (i = 0; i <= pos; i++)
589 strcat (dst, "[0-9]");
590 strcat (dst, "[0-9]*");
595 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
596 const char *term_sub,
598 oid_value attributeSet,
599 struct grep_info *grep_info,
607 attr_init (&relation, zapt, 2);
608 relation_value = attr_find (&relation, NULL);
609 term_value = atoi (term_sub);
611 switch (relation_value)
616 logf (LOG_DEBUG, "Relation <");
617 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
622 logf (LOG_DEBUG, "Relation <=");
623 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
628 logf (LOG_DEBUG, "Relation >=");
629 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
634 logf (LOG_DEBUG, "Relation >");
635 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
640 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
641 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
644 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
645 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
649 static void verbatim_char (int ch, int *indx, char *dst)
652 dst[(*indx)++] = '\\';
656 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
657 const char *term_sub,
658 oid_value attributeSet, struct grep_info *grep_info,
659 int num_bases, char **basenames)
661 char term_dict[2*IT_MAX_WORD+2];
662 int i, j, r, base_no;
664 int truncation_value;
667 oid_value curAttributeSet = attributeSet;
669 attr_init (&use, zapt, 1);
670 use_value = attr_find (&use, &curAttributeSet);
671 logf (LOG_DEBUG, "use value %d", use_value);
672 attr_init (&truncation, zapt, 5);
673 truncation_value = attr_find (&truncation, NULL);
674 logf (LOG_DEBUG, "truncation value %d", truncation_value);
679 for (base_no = 0; base_no < num_bases; base_no++)
682 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
690 if (!relational_term (zi, zapt, term_sub, term_dict,
691 attributeSet, grep_info, &max_pos))
696 switch (truncation_value)
698 case -1: /* not specified */
699 case 100: /* do not truncate */
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 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
705 &max_pos, 0, grep_handle);
707 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
709 case 1: /* right truncation */
710 term_dict[j++] = '(';
711 for (i = 0; term_sub[i]; i++)
712 verbatim_char (term_sub[i], &j, term_dict);
713 strcpy (term_dict+j, ".*)");
714 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
715 &max_pos, 0, grep_handle);
717 case 2: /* left truncation */
718 case 3: /* left&right truncation */
721 case 101: /* process # in term */
722 term_dict[j++] = '(';
723 for (i=0; term_sub[i]; i++)
724 if (term_sub[i] == '#' && i > 2)
726 term_dict[j++] = '.';
727 term_dict[j++] = '*';
730 verbatim_char (term_sub[i], &j, term_dict);
731 strcpy (term_dict+j, ")");
732 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
733 &max_pos, 0, grep_handle);
735 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
738 case 102: /* regular expression */
739 sprintf (term_dict + j, "(%s)", term_sub);
740 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
741 &max_pos, 0, grep_handle);
743 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
746 case 103: /* regular expression with error correction */
749 if (*cp == '*' && cp[1] && cp[2])
754 sprintf (term_dict + j, "(%s)", cp);
755 r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info,
756 &max_pos, j, grep_handle);
758 logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
763 if (max_pos <= strlen(basenames[base_no]))
765 zi->errCode = 109; /* Database unavailable */
766 zi->errString = basenames[base_no];
770 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
774 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
778 Z_Term *term = zapt->term;
780 sizez = term->u.general->len;
781 if (sizez > IT_MAX_WORD)
783 for (i = 0; i < sizez; i++)
784 termz[i] = index_char_cvt (term->u.general->buf[i]);
788 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
789 Z_AttributesPlusTerm *zapt,
790 oid_value attributeSet,
791 int num_bases, char **basenames)
793 rset_relevance_parms parms;
794 char termz[IT_MAX_WORD+1];
795 char term_sub[IT_MAX_WORD+1];
796 struct grep_info grep_info;
797 char *p0 = termz, *p1 = NULL;
800 parms.key_size = sizeof(struct it_key);
802 parms.cmp = key_compare;
803 parms.is = zi->wordIsam;
805 if (zapt->term->which != Z_Term_general)
810 trans_term (zi, zapt, termz);
811 grep_info.isam_p_indx = 0;
812 grep_info.isam_p_size = 0;
813 grep_info.isam_p_buf = NULL;
816 if ((p1 = strchr (p0, ' ')))
818 memcpy (term_sub, p0, p1-p0);
819 term_sub[p1-p0] = '\0';
822 strcpy (term_sub, p0);
823 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
824 num_bases, basenames))
832 parms.isam_positions = grep_info.isam_p_buf;
833 parms.no_isam_positions = grep_info.isam_p_indx;
834 if (grep_info.isam_p_indx > 0)
835 result = rset_create (rset_kind_relevance, &parms);
837 result = rset_create (rset_kind_null, NULL);
838 xfree (grep_info.isam_p_buf);
842 static RSET rpn_search_APT_word (ZServerInfo *zi,
843 Z_AttributesPlusTerm *zapt,
844 oid_value attributeSet,
845 int num_bases, char **basenames)
847 rset_isam_parms parms;
848 char termz[IT_MAX_WORD+1];
849 struct grep_info grep_info;
852 if (zapt->term->which != Z_Term_general)
857 trans_term (zi, zapt, termz);
859 grep_info.isam_p_indx = 0;
860 grep_info.isam_p_size = 0;
861 grep_info.isam_p_buf = NULL;
863 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
864 num_bases, basenames))
866 if (grep_info.isam_p_indx < 1)
867 result = rset_create (rset_kind_null, NULL);
868 else if (grep_info.isam_p_indx == 1)
870 parms.is = zi->wordIsam;
871 parms.pos = *grep_info.isam_p_buf;
872 result = rset_create (rset_kind_isam, &parms);
875 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
876 grep_info.isam_p_indx);
877 xfree (grep_info.isam_p_buf);
881 static RSET rpn_prox (RSET *rset, int rset_no)
889 rset_temp_parms parms;
891 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
892 more = xmalloc (sizeof(*more)*rset_no);
893 buf = xmalloc (sizeof(*buf)*rset_no);
895 for (i = 0; i<rset_no; i++)
897 buf[i] = xmalloc (sizeof(**buf));
898 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
899 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
903 rset_close (rset[i], rsfd[i]);
910 return rset_create (rset_kind_null, NULL);
913 parms.key_size = sizeof (struct it_key);
914 result = rset_create (rset_kind_temp, &parms);
915 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
919 for (i = 1; i<rset_no; i++)
928 cmp = key_compare (buf[i], buf[i-1]);
931 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
936 if (buf[i-1]->seqno+1 != buf[i]->seqno)
938 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
944 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
950 rset_write (result, rsfd_result, buf[0]);
951 more[0] = rset_read (*rset, *rsfd, *buf);
955 for (i = 0; i<rset_no; i++)
957 rset_close (rset[i], rsfd[i]);
960 rset_close (result, rsfd_result);
967 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
968 Z_AttributesPlusTerm *zapt,
969 oid_value attributeSet,
970 int num_bases, char **basenames)
972 char termz[IT_MAX_WORD+1];
973 char term_sub[IT_MAX_WORD+1];
974 char *p0 = termz, *p1 = NULL;
975 RSET rset[60], result;
977 struct grep_info grep_info;
979 if (zapt->term->which != Z_Term_general)
984 trans_term (zi, zapt, termz);
986 grep_info.isam_p_size = 0;
987 grep_info.isam_p_buf = NULL;
991 if ((p1 = strchr (p0, ' ')))
993 memcpy (term_sub, p0, p1-p0);
994 term_sub[p1-p0] = '\0';
997 strcpy (term_sub, p0);
999 grep_info.isam_p_indx = 0;
1000 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
1001 num_bases, basenames))
1003 if (grep_info.isam_p_indx == 0)
1004 rset[rset_no] = rset_create (rset_kind_null, NULL);
1005 else if (grep_info.isam_p_indx > 1)
1006 rset[rset_no] = rset_trunc (zi->wordIsam,
1007 grep_info.isam_p_buf,
1008 grep_info.isam_p_indx);
1011 rset_isam_parms parms;
1013 parms.is = zi->wordIsam;
1014 parms.pos = *grep_info.isam_p_buf;
1015 rset[rset_no] = rset_create (rset_kind_isam, &parms);
1017 assert (rset[rset_no]);
1018 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1023 while (*++p0 == ' ')
1026 xfree (grep_info.isam_p_buf);
1028 return rset_create (rset_kind_null, NULL);
1029 else if (rset_no == 1)
1032 result = rpn_prox (rset, rset_no);
1033 for (i = 0; i<rset_no; i++)
1034 rset_delete (rset[i]);
1038 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1039 oid_value attributeSet)
1044 rset_temp_parms parms;
1045 char termz[IT_MAX_WORD+1];
1047 if (zapt->term->which != Z_Term_general)
1052 parms.key_size = sizeof (struct it_key);
1053 result = rset_create (rset_kind_temp, &parms);
1054 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1056 trans_term (zi, zapt, termz);
1057 key.sysno = atoi (termz);
1060 rset_write (result, rsfd, &key);
1061 rset_close (result, rsfd);
1065 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1066 oid_value attributeSet,
1067 int num_bases, char **basenames)
1071 int relation_value, structure_value;
1073 attr_init (&relation, zapt, 2);
1074 attr_init (&structure, zapt, 4);
1076 relation_value = attr_find (&relation, NULL);
1077 structure_value = attr_find (&structure, NULL);
1078 switch (structure_value)
1081 if (relation_value == 102) /* relevance relation */
1082 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1083 num_bases, basenames);
1084 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1085 num_bases, basenames);
1086 case 1: /* phrase */
1087 if (relation_value == 102) /* relevance relation */
1088 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1089 num_bases, basenames);
1090 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1091 num_bases, basenames);
1094 if (relation_value == 102) /* relevance relation */
1095 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1096 num_bases, basenames);
1097 return rpn_search_APT_word (zi, zapt, attributeSet,
1098 num_bases, basenames);
1103 case 5: /* date - normalized */
1105 case 6: /* word list */
1106 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1107 num_bases, basenames);
1108 case 100: /* date - un-normalized */
1110 case 101: /* name - normalized */
1112 case 102: /* date - un-normalized */
1114 case 103: /* structure */
1118 case 105: /* free-form-text */
1119 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1120 num_bases, basenames);
1121 case 106: /* document-text */
1122 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1123 num_bases, basenames);
1124 case 107: /* local-number */
1125 return rpn_search_APT_local (zi, zapt, attributeSet);
1126 case 108: /* string */
1127 return rpn_search_APT_word (zi, zapt, attributeSet,
1128 num_bases, basenames);
1129 case 109: /* numeric string */
1136 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1140 if (!(s = resultSetGet (zi, resultSetId)))
1141 return rset_create (rset_kind_null, NULL);
1145 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1146 oid_value attributeSet,
1147 int num_bases, char **basenames)
1150 if (zs->which == Z_RPNStructure_complex)
1152 rset_bool_parms bool_parms;
1155 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1157 num_bases, basenames);
1158 if (bool_parms.rset_l == NULL)
1160 if (rset_is_ranked(bool_parms.rset_l))
1162 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1164 num_bases, basenames);
1165 if (bool_parms.rset_r == NULL)
1167 rset_delete (bool_parms.rset_l);
1170 if (rset_is_ranked(bool_parms.rset_r))
1172 bool_parms.key_size = sizeof(struct it_key);
1173 bool_parms.cmp = key_compare;
1175 switch (zs->u.complex->roperator->which)
1177 case Z_Operator_and:
1178 r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms);
1181 r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms);
1183 case Z_Operator_and_not:
1184 r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms);
1190 else if (zs->which == Z_RPNStructure_simple)
1192 if (zs->u.simple->which == Z_Operand_APT)
1194 logf (LOG_DEBUG, "rpn_search_APT");
1195 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1196 attributeSet, num_bases, basenames);
1198 else if (zs->u.simple->which == Z_Operand_resultSetId)
1200 logf (LOG_DEBUG, "rpn_search_ref");
1201 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1215 void count_set_save (RSET *r, int *count)
1222 rset_temp_parms parms;
1224 logf (LOG_DEBUG, "count_set_save");
1226 parms.key_size = sizeof(struct it_key);
1227 w = rset_create (rset_kind_temp, &parms);
1228 wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO);
1229 rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO);
1230 while (rset_read (*r, rfd, &key))
1232 if (key.sysno != psysno)
1234 rset_write (w, wfd, &key);
1240 rset_close (*r, rfd);
1242 rset_close (w, wfd);
1244 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1247 static void count_set (RSET r, int *count)
1254 logf (LOG_DEBUG, "count_set");
1256 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1257 while (rset_read (r, rfd, &key))
1259 if (key.sysno != psysno)
1266 rset_close (r, rfd);
1267 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1270 int rpn_search (ZServerInfo *zi,
1271 Z_RPNQuery *rpn, int num_bases, char **basenames,
1272 const char *setname, int *hits)
1276 oid_value attributeSet;
1281 zi->errString = NULL;
1283 attrset = oid_getentbyoid (rpn->attributeSetId);
1284 attributeSet = attrset->value;
1285 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1286 num_bases, basenames);
1289 if (rset_is_volatile(rset))
1290 count_set_save(&rset,hits);
1292 count_set (rset, hits);
1293 resultSetAdd (zi, setname, 1, rset);
1295 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1300 struct scan_entry *list;
1307 static int scan_handle (char *name, const char *info, int pos, void *client)
1309 int len_prefix, idx;
1312 struct scan_info *scan_info = client;
1314 rset_isam_parms parms;
1316 len_prefix = strlen(scan_info->prefix);
1317 if (memcmp (name, scan_info->prefix, len_prefix))
1320 idx = scan_info->after - pos + scan_info->before;
1323 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1324 strlen(name + len_prefix)+1);
1325 strcpy (scan_info->list[idx].term, name + len_prefix);
1326 assert (*info == sizeof(isam_p));
1327 memcpy (&isam_p, info+1, sizeof(isam_p));
1328 parms.is = scan_info->isam;
1331 rset = rset_create (rset_kind_isam, &parms);
1332 count_set (rset, &scan_info->list[idx].occurrences);
1335 scan_info->list[idx].occurrences = 1;
1337 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1342 static int dummy_handle (char *name, const char *info, void *p)
1347 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1348 int num_bases, char **basenames,
1349 int *position, int *num_entries, struct scan_entry **list,
1352 int i, j, sizez, max_pos;
1353 int pos = *position;
1354 int num = *num_entries;
1357 char termz[IT_MAX_WORD+20];
1360 Z_Term *term = zapt->term;
1361 struct scan_info scan_info;
1363 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1367 scan_info.before = before = pos-1;
1368 scan_info.after = after = 1+num-pos;
1369 scan_info.odr = zi->odr;
1371 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1373 scan_info.isam = zi->wordIsam;
1374 scan_info.list = odr_malloc (zi->odr, (before+after)*
1375 sizeof(*scan_info.list));
1376 for (j = 0; j<before+after; j++)
1377 scan_info.list[j].term = NULL;
1378 attr_init (&use, zapt, 1);
1379 use_value = attr_find (&use, NULL);
1380 logf (LOG_DEBUG, "use value %d", use_value);
1382 if (use_value == -1)
1384 i = index_word_prefix (termz, 1, use_value, *basenames);
1386 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos, 0,
1388 if (max_pos <= strlen(*basenames))
1390 zi->errString = *basenames;
1391 return zi->errCode = 109; /* Database unavailable */
1393 strcpy (scan_info.prefix, termz);
1394 sizez = term->u.general->len;
1395 if (sizez > IT_MAX_WORD)
1396 sizez = IT_MAX_WORD;
1397 for (j = 0; j<sizez; j++)
1398 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1401 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1403 *status = BEND_SCAN_SUCCESS;
1405 for (i = 0; i<scan_info.after; i++)
1406 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1410 *status = BEND_SCAN_PARTIAL;
1412 for (i = 0; i<scan_info.before; i++)
1413 if (scan_info.list[i].term)
1416 *status = BEND_SCAN_PARTIAL;
1420 *list = scan_info.list+i; /* list is set to first 'real' entry */
1422 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1423 zi->errCode = 114; /* if no entries was found */
1424 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1425 *position, *num_entries);
1427 logf (LOG_DEBUG, "scan error: %d", zi->errCode);