2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.33 1995-11-01 13:58:28 quinn
8 * Moving data1 to yaz/retrieval
10 * Revision 1.32 1995/10/27 14:00:11 adam
11 * Implemented detection of database availability.
13 * Revision 1.31 1995/10/17 18:02:10 adam
14 * New feature: databases. Implemented as prefix to words in dictionary.
16 * Revision 1.30 1995/10/16 09:32:38 adam
17 * More work on relational op.
19 * Revision 1.29 1995/10/13 16:01:49 adam
22 * Revision 1.28 1995/10/13 12:26:43 adam
23 * Optimization of truncation.
25 * Revision 1.27 1995/10/12 17:07:22 adam
28 * Revision 1.26 1995/10/12 12:40:54 adam
29 * Bug fixes in rpn_prox.
31 * Revision 1.25 1995/10/10 13:59:24 adam
32 * Function rset_open changed its wflag parameter to general flags.
34 * Revision 1.24 1995/10/09 16:18:37 adam
35 * Function dict_lookup_grep got extra client data parameter.
37 * Revision 1.23 1995/10/06 16:33:37 adam
38 * Use attribute mappings.
40 * Revision 1.22 1995/10/06 15:07:39 adam
41 * Structure 'local-number' handled.
43 * Revision 1.21 1995/10/06 13:52:06 adam
44 * Bug fixes. Handler may abort further scanning.
46 * Revision 1.20 1995/10/06 11:06:33 adam
47 * Scan entries include 'occurrences' now.
49 * Revision 1.19 1995/10/06 10:43:56 adam
50 * Scan added. 'occurrences' in scan entries not set yet.
52 * Revision 1.18 1995/10/04 16:57:20 adam
53 * Key input and merge sort in one pass.
55 * Revision 1.17 1995/10/04 12:55:17 adam
56 * Bug fix in ranked search. Use=Any keys inserted.
58 * Revision 1.16 1995/10/02 16:24:40 adam
59 * Use attribute actually used in search requests.
61 * Revision 1.15 1995/10/02 15:18:52 adam
62 * New member in recRetrieveCtrl: diagnostic.
64 * Revision 1.14 1995/09/28 12:10:32 adam
65 * Bug fixes. Field prefix used in queries.
67 * Revision 1.13 1995/09/18 14:17:50 adam
70 * Revision 1.12 1995/09/15 14:45:21 adam
74 * Revision 1.11 1995/09/14 11:53:27 adam
75 * First work on regular expressions/truncations.
77 * Revision 1.10 1995/09/11 15:23:26 adam
78 * More work on relevance search.
80 * Revision 1.9 1995/09/11 13:09:35 adam
81 * More work on relevance feedback.
83 * Revision 1.8 1995/09/08 14:52:27 adam
84 * Minor changes. Dictionary is lower case now.
86 * Revision 1.7 1995/09/07 13:58:36 adam
87 * New parameter: result-set file descriptor (RSFD) to support multiple
88 * positions within the same result-set.
89 * Boolean operators: and, or, not implemented.
90 * Result-set references.
92 * Revision 1.6 1995/09/06 16:11:18 adam
93 * Option: only one word key per file.
95 * Revision 1.5 1995/09/06 10:33:04 adam
96 * More work on present. Some log messages removed.
98 * Revision 1.4 1995/09/05 15:28:40 adam
99 * More work on search engine.
101 * Revision 1.3 1995/09/04 15:20:22 adam
104 * Revision 1.2 1995/09/04 12:33:43 adam
105 * Various cleanup. YAZ util used instead.
107 * Revision 1.1 1995/09/04 09:10:40 adam
108 * More work on index add/del/update.
109 * Merge sort implemented.
110 * Initial work on z39 server.
118 #include "attribute.h"
126 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
131 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
132 attp = att_getentbyatt (attrSet, attrUse);
135 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
136 return index_word_prefix (string, attp->attset_ordinal,
137 attp->local_attribute, basename);
141 * attr_print: log attributes
143 static void attr_print (Z_AttributesPlusTerm *t)
146 for (of = 0; of < t->num_attributes; of++)
148 Z_AttributeElement *element;
149 element = t->attributeList[of];
151 switch (element->which)
153 case Z_AttributeValue_numeric:
154 logf (LOG_DEBUG, "attributeType=%d value=%d",
155 *element->attributeType,
156 *element->value.numeric);
158 case Z_AttributeValue_complex:
159 logf (LOG_DEBUG, "attributeType=%d complex",
160 *element->attributeType);
161 for (i = 0; i<element->value.complex->num_list; i++)
163 if (element->value.complex->list[i]->which ==
164 Z_StringOrNumeric_string)
165 logf (LOG_DEBUG, " string: '%s'",
166 element->value.complex->list[i]->u.string);
167 else if (element->value.complex->list[i]->which ==
168 Z_StringOrNumeric_numeric)
169 logf (LOG_DEBUG, " numeric: '%d'",
170 *element->value.complex->list[i]->u.numeric);
183 Z_AttributesPlusTerm *zapt;
186 static int attr_find (AttrType *src, oid_value *attributeSetP)
188 while (src->major < src->zapt->num_attributes)
190 Z_AttributeElement *element;
192 element = src->zapt->attributeList[src->major];
193 if (src->type == *element->attributeType)
195 switch (element->which)
197 case Z_AttributeValue_numeric:
199 if (element->attributeSet && attributeSetP)
203 attrset = oid_getentbyoid (element->attributeSet);
204 *attributeSetP = attrset->value;
206 return *element->value.numeric;
208 case Z_AttributeValue_complex:
209 if (src->minor >= element->value.complex->num_list ||
210 element->value.complex->list[src->minor]->which !=
211 Z_StringOrNumeric_numeric)
214 if (element->attributeSet && attributeSetP)
218 attrset = oid_getentbyoid (element->attributeSet);
219 *attributeSetP = attrset->value;
221 return *element->value.complex->list[src->minor-1]->u.numeric;
231 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
245 int (*cmp)(const void *p1, const void *p2);
252 static void heap_swap (struct trunc_info *ti, int i1, int i2)
257 ti->ptr[i1] = ti->ptr[i2];
261 static void heap_delete (struct trunc_info *ti)
263 int cur = 1, child = 2;
265 heap_swap (ti, 1, ti->heapnum--);
266 while (child <= ti->heapnum) {
267 if (child < ti->heapnum &&
268 (*ti->cmp)(ti->heap[ti->ptr[child]],
269 ti->heap[ti->ptr[1+child]]) > 0)
271 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
272 ti->heap[ti->ptr[child]]) > 0)
274 heap_swap (ti, cur, child);
283 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
287 cur = ++(ti->heapnum);
288 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
289 ti->indx[ti->ptr[cur]] = indx;
291 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
292 ti->heap[ti->ptr[cur]]) > 0)
294 heap_swap (ti, cur, parent);
301 struct trunc_info *heap_init (int size, int key_size,
302 int (*cmp)(const void *p1, const void *p2))
304 struct trunc_info *ti = xmalloc (sizeof(*ti));
309 ti->keysize = key_size;
311 ti->indx = xmalloc (size * sizeof(*ti->indx));
312 ti->heap = xmalloc (size * sizeof(*ti->heap));
313 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
314 ti->swapbuf = xmalloc (ti->keysize);
315 ti->tmpbuf = xmalloc (ti->keysize);
316 ti->buf = xmalloc (size * ti->keysize);
317 for (i = size; --i >= 0; )
320 ti->heap[i] = ti->buf + ti->keysize * i;
325 static void heap_close (struct trunc_info *ti)
335 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
340 rset_temp_parms parms;
342 parms.key_size = sizeof(struct it_key);
343 result = rset_create (rset_kind_temp, &parms);
344 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
346 if (to - from > merge_chunk)
350 int i, i_add = (to-from)/merge_chunk + 1;
351 struct trunc_info *ti;
353 int rsmax = (to-from)/i_add + 1;
355 rset = xmalloc (sizeof(*rset) * rsmax);
356 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
358 for (i = from; i < to; i += i_add)
361 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
364 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
368 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
369 for (i = rscur; --i >= 0; )
371 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
372 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
373 heap_insert (ti, ti->tmpbuf, i);
376 rset_close (rset[i], rsfd[i]);
377 rset_delete (rset[i]);
382 int n = ti->indx[ti->ptr[1]];
384 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
388 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
391 rset_close (rset[n], rsfd[n]);
392 rset_delete (rset[n]);
395 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
398 heap_insert (ti, ti->tmpbuf, n);
411 struct trunc_info *ti;
413 ispt = xmalloc (sizeof(*ispt) * (to-from));
415 ti = heap_init (to-from, sizeof(struct it_key),
417 for (i = to-from; --i >= 0; )
419 ispt[i] = is_position (isam, isam_p[from+i]);
420 if (is_readkey (ispt[i], ti->tmpbuf))
421 heap_insert (ti, ti->tmpbuf, i);
423 is_pt_free (ispt[i]);
427 int n = ti->indx[ti->ptr[1]];
429 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
431 /* section that preserve all keys */
433 if (is_readkey (ispt[n], ti->tmpbuf))
434 heap_insert (ti, ti->tmpbuf, n);
436 is_pt_free (ispt[n]);
438 /* section that preserve all keys with unique sysnos */
441 if (!is_readkey (ispt[n], ti->tmpbuf))
444 is_pt_free (ispt[n]);
447 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
450 heap_insert (ti, ti->tmpbuf, n);
459 rset_close (result, result_rsfd);
463 static int isam_trunc_cmp (const void *p1, const void *p2)
465 ISAM_P i1 = *(ISAM_P*) p1;
466 ISAM_P i2 = *(ISAM_P*) p2;
469 d = is_type (i1) - is_type (i2);
472 return is_block (i1) - is_block (i2);
475 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
478 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
479 return rset_trunc_r (isam, isam_p, 0, no, 100);
488 static void add_isam_p (const char *info, struct grep_info *p)
490 if (p->isam_p_indx == p->isam_p_size)
492 ISAM_P *new_isam_p_buf;
494 p->isam_p_size = 2*p->isam_p_size + 100;
495 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
499 memcpy (new_isam_p_buf, p->isam_p_buf,
500 p->isam_p_indx * sizeof(*p->isam_p_buf));
501 xfree (p->isam_p_buf);
503 p->isam_p_buf = new_isam_p_buf;
505 assert (*info == sizeof(*p->isam_p_buf));
506 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
510 static int grep_handle (Dict_char *name, const char *info, void *p)
512 logf (LOG_DEBUG, "dict name: %s", name);
513 add_isam_p (info, p);
517 static void gen_regular_rel (char *dst, int val, int islt)
525 sprintf (numstr, "%d", val);
526 for (w = strlen(numstr); --w >= 0; pos++)
545 strcpy (dst + dst_p, numstr);
546 dst_p = strlen(dst) - pos - 1;
574 for (i = 0; i<pos; i++)
587 for (i=1; i<pos; i++)
588 strcat (dst, "[0-9]?");
592 for (i = 0; i <= pos; i++)
593 strcat (dst, "[0-9]");
594 strcat (dst, "[0-9]*");
599 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
600 const char *term_sub,
602 oid_value attributeSet,
603 struct grep_info *grep_info,
611 attr_init (&relation, zapt, 2);
612 relation_value = attr_find (&relation, NULL);
613 term_value = atoi (term_sub);
615 switch (relation_value)
620 logf (LOG_DEBUG, "Relation <");
621 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
626 logf (LOG_DEBUG, "Relation <=");
627 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
632 logf (LOG_DEBUG, "Relation >=");
633 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
638 logf (LOG_DEBUG, "Relation >");
639 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
644 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
645 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
648 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
649 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
653 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
654 const char *term_sub,
655 oid_value attributeSet, struct grep_info *grep_info,
656 int num_bases, char **basenames)
658 char term_dict[2*IT_MAX_WORD+2];
659 int i, j, r, base_no;
661 int truncation_value;
664 oid_value curAttributeSet = attributeSet;
666 attr_init (&use, zapt, 1);
667 use_value = attr_find (&use, &curAttributeSet);
668 logf (LOG_DEBUG, "use value %d", use_value);
669 attr_init (&truncation, zapt, 5);
670 truncation_value = attr_find (&truncation, NULL);
671 logf (LOG_DEBUG, "truncation value %d", truncation_value);
676 for (base_no = 0; base_no < num_bases; base_no++)
679 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
687 if (!relational_term (zi, zapt, term_sub, term_dict,
688 attributeSet, grep_info, &max_pos))
690 switch (truncation_value)
692 case -1: /* not specified */
693 case 100: /* do not truncate */
694 strcat (term_dict, "(");
695 strcat (term_dict, term_sub);
696 strcat (term_dict, ")");
697 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
698 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
699 &max_pos, grep_handle);
701 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
703 case 1: /* right truncation */
704 strcat (term_dict, term_sub);
705 strcat (term_dict, ".*");
706 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
707 &max_pos, grep_handle);
709 case 2: /* left truncation */
710 case 3: /* left&right truncation */
713 case 101: /* process # in term */
714 for (j = strlen(term_dict), i = 0; term_sub[i] && i < 2; i++)
715 term_dict[j++] = term_sub[i];
716 for (; term_sub[i]; i++)
717 if (term_sub[i] == '#')
719 term_dict[j++] = '.';
720 term_dict[j++] = '*';
723 term_dict[j++] = term_sub[i];
725 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
726 &max_pos, grep_handle);
728 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
731 case 102: /* regular expression */
732 strcat (term_dict, "(");
733 strcat (term_dict, term_sub);
734 strcat (term_dict, ")");
735 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
736 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
737 &max_pos, grep_handle);
739 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
744 logf (LOG_DEBUG, "max_pos = %d", max_pos);
745 if (max_pos <= strlen(basenames[base_no]))
747 zi->errCode = 109; /* Database unavailable */
748 zi->errString = basenames[base_no];
752 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
756 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
760 Z_Term *term = zapt->term;
762 sizez = term->u.general->len;
763 if (sizez > IT_MAX_WORD)
765 for (i = 0; i < sizez; i++)
766 termz[i] = index_char_cvt (term->u.general->buf[i]);
770 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
771 Z_AttributesPlusTerm *zapt,
772 oid_value attributeSet,
773 int num_bases, char **basenames)
775 rset_relevance_parms parms;
776 char termz[IT_MAX_WORD+1];
777 char term_sub[IT_MAX_WORD+1];
778 struct grep_info grep_info;
779 char *p0 = termz, *p1 = NULL;
782 parms.key_size = sizeof(struct it_key);
784 parms.cmp = key_compare;
785 parms.is = zi->wordIsam;
787 if (zapt->term->which != Z_Term_general)
792 trans_term (zi, zapt, termz);
793 grep_info.isam_p_indx = 0;
794 grep_info.isam_p_size = 0;
795 grep_info.isam_p_buf = NULL;
798 if ((p1 = strchr (p0, ' ')))
800 memcpy (term_sub, p0, p1-p0);
801 term_sub[p1-p0] = '\0';
804 strcpy (term_sub, p0);
805 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
806 num_bases, basenames))
814 parms.isam_positions = grep_info.isam_p_buf;
815 parms.no_isam_positions = grep_info.isam_p_indx;
816 if (grep_info.isam_p_indx > 0)
817 result = rset_create (rset_kind_relevance, &parms);
819 result = rset_create (rset_kind_null, NULL);
820 xfree (grep_info.isam_p_buf);
824 static RSET rpn_search_APT_word (ZServerInfo *zi,
825 Z_AttributesPlusTerm *zapt,
826 oid_value attributeSet,
827 int num_bases, char **basenames)
829 rset_isam_parms parms;
830 char termz[IT_MAX_WORD+1];
831 struct grep_info grep_info;
834 if (zapt->term->which != Z_Term_general)
839 trans_term (zi, zapt, termz);
841 grep_info.isam_p_indx = 0;
842 grep_info.isam_p_size = 0;
843 grep_info.isam_p_buf = NULL;
845 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
846 num_bases, basenames))
848 if (grep_info.isam_p_indx < 1)
849 result = rset_create (rset_kind_null, NULL);
850 else if (grep_info.isam_p_indx == 1)
852 parms.is = zi->wordIsam;
853 parms.pos = *grep_info.isam_p_buf;
854 result = rset_create (rset_kind_isam, &parms);
857 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
858 grep_info.isam_p_indx);
859 xfree (grep_info.isam_p_buf);
863 static RSET rpn_prox (RSET *rset, int rset_no)
871 rset_temp_parms parms;
873 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
874 more = xmalloc (sizeof(*more)*rset_no);
875 buf = xmalloc (sizeof(*buf)*rset_no);
877 for (i = 0; i<rset_no; i++)
879 buf[i] = xmalloc (sizeof(**buf));
880 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
881 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
885 rset_close (rset[i], rsfd[i]);
892 return rset_create (rset_kind_null, NULL);
895 parms.key_size = sizeof (struct it_key);
896 result = rset_create (rset_kind_temp, &parms);
897 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
901 for (i = 1; i<rset_no; i++)
910 cmp = key_compare (buf[i], buf[i-1]);
913 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
918 if (buf[i-1]->seqno+1 != buf[i]->seqno)
920 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
926 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
932 rset_write (result, rsfd_result, buf[0]);
933 more[0] = rset_read (*rset, *rsfd, *buf);
937 for (i = 0; i<rset_no; i++)
939 rset_close (rset[i], rsfd[i]);
942 rset_close (result, rsfd_result);
949 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
950 Z_AttributesPlusTerm *zapt,
951 oid_value attributeSet,
952 int num_bases, char **basenames)
954 char termz[IT_MAX_WORD+1];
955 char term_sub[IT_MAX_WORD+1];
956 char *p0 = termz, *p1 = NULL;
957 RSET rset[60], result;
959 struct grep_info grep_info;
961 if (zapt->term->which != Z_Term_general)
966 trans_term (zi, zapt, termz);
968 grep_info.isam_p_size = 0;
969 grep_info.isam_p_buf = NULL;
973 if ((p1 = strchr (p0, ' ')))
975 memcpy (term_sub, p0, p1-p0);
976 term_sub[p1-p0] = '\0';
979 strcpy (term_sub, p0);
981 grep_info.isam_p_indx = 0;
982 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
983 num_bases, basenames))
985 if (grep_info.isam_p_indx == 0)
986 rset[rset_no] = rset_create (rset_kind_null, NULL);
987 else if (grep_info.isam_p_indx > 1)
988 rset[rset_no] = rset_trunc (zi->wordIsam,
989 grep_info.isam_p_buf,
990 grep_info.isam_p_indx);
993 rset_isam_parms parms;
995 parms.is = zi->wordIsam;
996 parms.pos = *grep_info.isam_p_buf;
997 rset[rset_no] = rset_create (rset_kind_isam, &parms);
999 assert (rset[rset_no]);
1000 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1005 while (*++p0 == ' ')
1008 xfree (grep_info.isam_p_buf);
1010 return rset_create (rset_kind_null, NULL);
1011 else if (rset_no == 1)
1014 result = rpn_prox (rset, rset_no);
1015 for (i = 0; i<rset_no; i++)
1016 rset_delete (rset[i]);
1020 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1021 oid_value attributeSet)
1026 rset_temp_parms parms;
1027 char termz[IT_MAX_WORD+1];
1029 if (zapt->term->which != Z_Term_general)
1034 parms.key_size = sizeof (struct it_key);
1035 result = rset_create (rset_kind_temp, &parms);
1036 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1038 trans_term (zi, zapt, termz);
1039 key.sysno = atoi (termz);
1042 rset_write (result, rsfd, &key);
1043 rset_close (result, rsfd);
1047 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1048 oid_value attributeSet,
1049 int num_bases, char **basenames)
1053 int relation_value, structure_value;
1055 attr_init (&relation, zapt, 2);
1056 attr_init (&structure, zapt, 4);
1058 relation_value = attr_find (&relation, NULL);
1059 structure_value = attr_find (&structure, NULL);
1060 switch (structure_value)
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);
1068 case 1: /* phrase */
1069 if (relation_value == 102) /* relevance relation */
1070 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1071 num_bases, basenames);
1072 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1073 num_bases, basenames);
1076 if (relation_value == 102) /* relevance relation */
1077 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1078 num_bases, basenames);
1079 return rpn_search_APT_word (zi, zapt, attributeSet,
1080 num_bases, basenames);
1085 case 5: /* date - normalized */
1087 case 6: /* word list */
1088 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1089 num_bases, basenames);
1090 case 100: /* date - un-normalized */
1092 case 101: /* name - normalized */
1094 case 102: /* date - un-normalized */
1096 case 103: /* structure */
1100 case 105: /* free-form-text */
1101 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1102 num_bases, basenames);
1103 case 106: /* document-text */
1104 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1105 num_bases, basenames);
1106 case 107: /* local-number */
1107 return rpn_search_APT_local (zi, zapt, attributeSet);
1108 case 108: /* string */
1109 return rpn_search_APT_word (zi, zapt, attributeSet,
1110 num_bases, basenames);
1111 case 109: /* numeric string */
1118 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1122 if (!(s = resultSetGet (zi, resultSetId)))
1123 return rset_create (rset_kind_null, NULL);
1127 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1128 oid_value attributeSet,
1129 int num_bases, char **basenames)
1132 if (zs->which == Z_RPNStructure_complex)
1134 rset_bool_parms bool_parms;
1136 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1138 num_bases, basenames);
1139 if (bool_parms.rset_l == NULL)
1141 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1143 num_bases, basenames);
1144 if (bool_parms.rset_r == NULL)
1146 rset_delete (bool_parms.rset_l);
1149 bool_parms.key_size = sizeof(struct it_key);
1150 bool_parms.cmp = key_compare;
1152 switch (zs->u.complex->operator->which)
1154 case Z_Operator_and:
1155 r = rset_create (rset_kind_and, &bool_parms);
1158 r = rset_create (rset_kind_or, &bool_parms);
1160 case Z_Operator_and_not:
1161 r = rset_create (rset_kind_not, &bool_parms);
1167 else if (zs->which == Z_RPNStructure_simple)
1169 if (zs->u.simple->which == Z_Operand_APT)
1171 logf (LOG_DEBUG, "rpn_search_APT");
1172 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1173 attributeSet, num_bases, basenames);
1175 else if (zs->u.simple->which == Z_Operand_resultSetId)
1177 logf (LOG_DEBUG, "rpn_search_ref");
1178 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1192 static void count_set (RSET r, int *count)
1198 logf (LOG_DEBUG, "rpn_save_set");
1200 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1201 while (rset_read (r, rfd, &key))
1203 if (key.sysno != psysno)
1209 rset_close (r, rfd);
1210 logf (LOG_DEBUG, "%d distinct sysnos", *count);
1213 int rpn_search (ZServerInfo *zi,
1214 Z_RPNQuery *rpn, int num_bases, char **basenames,
1215 const char *setname, int *hits)
1219 oid_value attributeSet;
1222 zi->errString = NULL;
1224 attrset = oid_getentbyoid (rpn->attributeSetId);
1225 attributeSet = attrset->value;
1227 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1228 num_bases, basenames);
1231 count_set (rset, hits);
1232 resultSetAdd (zi, setname, 1, rset);
1234 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1239 struct scan_entry *list;
1246 static int scan_handle (Dict_char *name, const char *info, int pos,
1249 int len_prefix, idx;
1252 struct scan_info *scan_info = client;
1254 rset_isam_parms parms;
1256 len_prefix = strlen(scan_info->prefix);
1257 if (memcmp (name, scan_info->prefix, len_prefix))
1260 idx = scan_info->after - pos + scan_info->before;
1263 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1264 strlen(name + len_prefix)+1);
1265 strcpy (scan_info->list[idx].term, name + len_prefix);
1266 assert (*info == sizeof(isam_p));
1267 memcpy (&isam_p, info+1, sizeof(isam_p));
1268 parms.is = scan_info->isam;
1271 rset = rset_create (rset_kind_isam, &parms);
1272 count_set (rset, &scan_info->list[idx].occurrences);
1275 scan_info->list[idx].occurrences = 1;
1277 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1282 static int dummy_handle (Dict_char *name, const char *info, void *p)
1287 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1288 int num_bases, char **basenames,
1289 int *position, int *num_entries, struct scan_entry **list,
1292 int i, j, sizez, max_pos;
1293 int pos = *position;
1294 int num = *num_entries;
1297 char termz[IT_MAX_WORD+20];
1300 Z_Term *term = zapt->term;
1301 struct scan_info scan_info;
1303 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1307 scan_info.before = before = pos-1;
1308 scan_info.after = after = 1+num-pos;
1309 scan_info.odr = zi->odr;
1311 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1313 scan_info.isam = zi->wordIsam;
1314 scan_info.list = odr_malloc (zi->odr, (before+after)*
1315 sizeof(*scan_info.list));
1316 for (j = 0; j<before+after; j++)
1317 scan_info.list[j].term = NULL;
1318 attr_init (&use, zapt, 1);
1319 use_value = attr_find (&use, NULL);
1320 logf (LOG_DEBUG, "use value %d", use_value);
1322 if (use_value == -1)
1324 i = index_word_prefix (termz, 1, use_value, *basenames);
1326 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
1328 if (max_pos <= strlen(*basenames))
1330 zi->errString = *basenames;
1331 return zi->errCode = 109; /* Database unavailable */
1333 strcpy (scan_info.prefix, termz);
1334 sizez = term->u.general->len;
1335 if (sizez > IT_MAX_WORD)
1336 sizez = IT_MAX_WORD;
1337 for (j = 0; j<sizez; j++)
1338 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1341 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1343 *status = BEND_SCAN_SUCCESS;
1345 for (i = 0; i<scan_info.after; i++)
1346 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1350 *status = BEND_SCAN_PARTIAL;
1352 for (i = 0; i<scan_info.before; i++)
1353 if (scan_info.list[i].term)
1356 *status = BEND_SCAN_PARTIAL;
1360 *list = scan_info.list+i; /* list is set to first 'real' entry */
1362 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1363 zi->errCode = 114; /* if no entries was found */
1364 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1365 *position, *num_entries);
1367 logf (LOG_DEBUG, "scan error: %d", zi->errCode);