2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.37 1995-12-06 15:05:28 adam
8 * More verbose in count_set.
10 * Revision 1.36 1995/12/06 12:41:27 adam
11 * New command 'stat' for the index program.
12 * Filenames can be read from stdin by specifying '-'.
13 * Bug fix/enhancement of the transformation from terms to regular
14 * expressons in the search engine.
16 * Revision 1.35 1995/11/27 09:29:00 adam
17 * Bug fixes regarding conversion to regular expressions.
19 * Revision 1.34 1995/11/16 17:00:56 adam
20 * Better logging of rpn query.
22 * Revision 1.33 1995/11/01 13:58:28 quinn
23 * Moving data1 to yaz/retrieval
25 * Revision 1.32 1995/10/27 14:00:11 adam
26 * Implemented detection of database availability.
28 * Revision 1.31 1995/10/17 18:02:10 adam
29 * New feature: databases. Implemented as prefix to words in dictionary.
31 * Revision 1.30 1995/10/16 09:32:38 adam
32 * More work on relational op.
34 * Revision 1.29 1995/10/13 16:01:49 adam
37 * Revision 1.28 1995/10/13 12:26:43 adam
38 * Optimization of truncation.
40 * Revision 1.27 1995/10/12 17:07:22 adam
43 * Revision 1.26 1995/10/12 12:40:54 adam
44 * Bug fixes in rpn_prox.
46 * Revision 1.25 1995/10/10 13:59:24 adam
47 * Function rset_open changed its wflag parameter to general flags.
49 * Revision 1.24 1995/10/09 16:18:37 adam
50 * Function dict_lookup_grep got extra client data parameter.
52 * Revision 1.23 1995/10/06 16:33:37 adam
53 * Use attribute mappings.
55 * Revision 1.22 1995/10/06 15:07:39 adam
56 * Structure 'local-number' handled.
58 * Revision 1.21 1995/10/06 13:52:06 adam
59 * Bug fixes. Handler may abort further scanning.
61 * Revision 1.20 1995/10/06 11:06:33 adam
62 * Scan entries include 'occurrences' now.
64 * Revision 1.19 1995/10/06 10:43:56 adam
65 * Scan added. 'occurrences' in scan entries not set yet.
67 * Revision 1.18 1995/10/04 16:57:20 adam
68 * Key input and merge sort in one pass.
70 * Revision 1.17 1995/10/04 12:55:17 adam
71 * Bug fix in ranked search. Use=Any keys inserted.
73 * Revision 1.16 1995/10/02 16:24:40 adam
74 * Use attribute actually used in search requests.
76 * Revision 1.15 1995/10/02 15:18:52 adam
77 * New member in recRetrieveCtrl: diagnostic.
79 * Revision 1.14 1995/09/28 12:10:32 adam
80 * Bug fixes. Field prefix used in queries.
82 * Revision 1.13 1995/09/18 14:17:50 adam
85 * Revision 1.12 1995/09/15 14:45:21 adam
89 * Revision 1.11 1995/09/14 11:53:27 adam
90 * First work on regular expressions/truncations.
92 * Revision 1.10 1995/09/11 15:23:26 adam
93 * More work on relevance search.
95 * Revision 1.9 1995/09/11 13:09:35 adam
96 * More work on relevance feedback.
98 * Revision 1.8 1995/09/08 14:52:27 adam
99 * Minor changes. Dictionary is lower case now.
101 * Revision 1.7 1995/09/07 13:58:36 adam
102 * New parameter: result-set file descriptor (RSFD) to support multiple
103 * positions within the same result-set.
104 * Boolean operators: and, or, not implemented.
105 * Result-set references.
107 * Revision 1.6 1995/09/06 16:11:18 adam
108 * Option: only one word key per file.
110 * Revision 1.5 1995/09/06 10:33:04 adam
111 * More work on present. Some log messages removed.
113 * Revision 1.4 1995/09/05 15:28:40 adam
114 * More work on search engine.
116 * Revision 1.3 1995/09/04 15:20:22 adam
119 * Revision 1.2 1995/09/04 12:33:43 adam
120 * Various cleanup. YAZ util used instead.
122 * Revision 1.1 1995/09/04 09:10:40 adam
123 * More work on index add/del/update.
124 * Merge sort implemented.
125 * Initial work on z39 server.
134 #include "attribute.h"
142 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
147 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
148 attp = att_getentbyatt (attrSet, attrUse);
151 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
152 return index_word_prefix (string, attp->attset_ordinal,
153 attp->local_attribute, basename);
160 Z_AttributesPlusTerm *zapt;
163 static int attr_find (AttrType *src, oid_value *attributeSetP)
165 while (src->major < src->zapt->num_attributes)
167 Z_AttributeElement *element;
169 element = src->zapt->attributeList[src->major];
170 if (src->type == *element->attributeType)
172 switch (element->which)
174 case Z_AttributeValue_numeric:
176 if (element->attributeSet && attributeSetP)
180 attrset = oid_getentbyoid (element->attributeSet);
181 *attributeSetP = attrset->value;
183 return *element->value.numeric;
185 case Z_AttributeValue_complex:
186 if (src->minor >= element->value.complex->num_list ||
187 element->value.complex->list[src->minor]->which !=
188 Z_StringOrNumeric_numeric)
191 if (element->attributeSet && attributeSetP)
195 attrset = oid_getentbyoid (element->attributeSet);
196 *attributeSetP = attrset->value;
198 return *element->value.complex->list[src->minor-1]->u.numeric;
208 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
222 int (*cmp)(const void *p1, const void *p2);
229 static void heap_swap (struct trunc_info *ti, int i1, int i2)
234 ti->ptr[i1] = ti->ptr[i2];
238 static void heap_delete (struct trunc_info *ti)
240 int cur = 1, child = 2;
242 heap_swap (ti, 1, ti->heapnum--);
243 while (child <= ti->heapnum) {
244 if (child < ti->heapnum &&
245 (*ti->cmp)(ti->heap[ti->ptr[child]],
246 ti->heap[ti->ptr[1+child]]) > 0)
248 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
249 ti->heap[ti->ptr[child]]) > 0)
251 heap_swap (ti, cur, child);
260 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
264 cur = ++(ti->heapnum);
265 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
266 ti->indx[ti->ptr[cur]] = indx;
268 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
269 ti->heap[ti->ptr[cur]]) > 0)
271 heap_swap (ti, cur, parent);
278 struct trunc_info *heap_init (int size, int key_size,
279 int (*cmp)(const void *p1, const void *p2))
281 struct trunc_info *ti = xmalloc (sizeof(*ti));
286 ti->keysize = key_size;
288 ti->indx = xmalloc (size * sizeof(*ti->indx));
289 ti->heap = xmalloc (size * sizeof(*ti->heap));
290 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
291 ti->swapbuf = xmalloc (ti->keysize);
292 ti->tmpbuf = xmalloc (ti->keysize);
293 ti->buf = xmalloc (size * ti->keysize);
294 for (i = size; --i >= 0; )
297 ti->heap[i] = ti->buf + ti->keysize * i;
302 static void heap_close (struct trunc_info *ti)
312 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
317 rset_temp_parms parms;
319 parms.key_size = sizeof(struct it_key);
320 result = rset_create (rset_kind_temp, &parms);
321 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
323 if (to - from > merge_chunk)
327 int i, i_add = (to-from)/merge_chunk + 1;
328 struct trunc_info *ti;
330 int rsmax = (to-from)/i_add + 1;
332 rset = xmalloc (sizeof(*rset) * rsmax);
333 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
335 for (i = from; i < to; i += i_add)
338 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
341 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
345 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
346 for (i = rscur; --i >= 0; )
348 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
349 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
350 heap_insert (ti, ti->tmpbuf, i);
353 rset_close (rset[i], rsfd[i]);
354 rset_delete (rset[i]);
359 int n = ti->indx[ti->ptr[1]];
361 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
365 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
368 rset_close (rset[n], rsfd[n]);
369 rset_delete (rset[n]);
372 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
375 heap_insert (ti, ti->tmpbuf, n);
388 struct trunc_info *ti;
390 ispt = xmalloc (sizeof(*ispt) * (to-from));
392 ti = heap_init (to-from, sizeof(struct it_key),
394 for (i = to-from; --i >= 0; )
396 ispt[i] = is_position (isam, isam_p[from+i]);
397 if (is_readkey (ispt[i], ti->tmpbuf))
398 heap_insert (ti, ti->tmpbuf, i);
400 is_pt_free (ispt[i]);
404 int n = ti->indx[ti->ptr[1]];
406 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
408 /* section that preserve all keys */
410 if (is_readkey (ispt[n], ti->tmpbuf))
411 heap_insert (ti, ti->tmpbuf, n);
413 is_pt_free (ispt[n]);
415 /* section that preserve all keys with unique sysnos */
418 if (!is_readkey (ispt[n], ti->tmpbuf))
421 is_pt_free (ispt[n]);
424 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
427 heap_insert (ti, ti->tmpbuf, n);
436 rset_close (result, result_rsfd);
440 static int isam_trunc_cmp (const void *p1, const void *p2)
442 ISAM_P i1 = *(ISAM_P*) p1;
443 ISAM_P i2 = *(ISAM_P*) p2;
446 d = is_type (i1) - is_type (i2);
449 return is_block (i1) - is_block (i2);
452 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
455 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
456 return rset_trunc_r (isam, isam_p, 0, no, 100);
465 static void add_isam_p (const char *info, struct grep_info *p)
467 if (p->isam_p_indx == p->isam_p_size)
469 ISAM_P *new_isam_p_buf;
471 p->isam_p_size = 2*p->isam_p_size + 100;
472 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
476 memcpy (new_isam_p_buf, p->isam_p_buf,
477 p->isam_p_indx * sizeof(*p->isam_p_buf));
478 xfree (p->isam_p_buf);
480 p->isam_p_buf = new_isam_p_buf;
482 assert (*info == sizeof(*p->isam_p_buf));
483 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
487 static int grep_handle (Dict_char *name, const char *info, void *p)
489 logf (LOG_DEBUG, "dict name: %s", name);
490 add_isam_p (info, p);
494 static void gen_regular_rel (char *dst, int val, int islt)
502 sprintf (numstr, "%d", val);
503 for (w = strlen(numstr); --w >= 0; pos++)
522 strcpy (dst + dst_p, numstr);
523 dst_p = strlen(dst) - pos - 1;
551 for (i = 0; i<pos; i++)
564 for (i=1; i<pos; i++)
565 strcat (dst, "[0-9]?");
569 for (i = 0; i <= pos; i++)
570 strcat (dst, "[0-9]");
571 strcat (dst, "[0-9]*");
576 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
577 const char *term_sub,
579 oid_value attributeSet,
580 struct grep_info *grep_info,
588 attr_init (&relation, zapt, 2);
589 relation_value = attr_find (&relation, NULL);
590 term_value = atoi (term_sub);
592 switch (relation_value)
597 logf (LOG_DEBUG, "Relation <");
598 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
603 logf (LOG_DEBUG, "Relation <=");
604 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
609 logf (LOG_DEBUG, "Relation >=");
610 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
615 logf (LOG_DEBUG, "Relation >");
616 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
621 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
622 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
625 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
626 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
630 static void verbatim_char (int ch, int *indx, char *dst)
633 dst[(*indx)++] = '\\';
637 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
638 const char *term_sub,
639 oid_value attributeSet, struct grep_info *grep_info,
640 int num_bases, char **basenames)
642 char term_dict[2*IT_MAX_WORD+2];
643 int i, j, r, base_no;
645 int truncation_value;
648 oid_value curAttributeSet = attributeSet;
650 attr_init (&use, zapt, 1);
651 use_value = attr_find (&use, &curAttributeSet);
652 logf (LOG_DEBUG, "use value %d", use_value);
653 attr_init (&truncation, zapt, 5);
654 truncation_value = attr_find (&truncation, NULL);
655 logf (LOG_DEBUG, "truncation value %d", truncation_value);
660 for (base_no = 0; base_no < num_bases; base_no++)
663 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
671 if (!relational_term (zi, zapt, term_sub, term_dict,
672 attributeSet, grep_info, &max_pos))
674 switch (truncation_value)
676 case -1: /* not specified */
677 case 100: /* do not truncate */
678 j = strlen(term_dict);
679 term_dict[j++] = '(';
680 for (i = 0; term_sub[i]; i++)
681 verbatim_char (term_sub[i], &j, term_dict);
682 strcpy (term_dict+j, ")");
683 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
684 &max_pos, grep_handle);
686 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
688 case 1: /* right truncation */
689 j = strlen(term_dict);
690 term_dict[j++] = '(';
691 for (i = 0; term_sub[i]; i++)
692 verbatim_char (term_sub[i], &j, term_dict);
693 strcpy (term_dict+j, ".*)");
694 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
695 &max_pos, grep_handle);
697 case 2: /* left truncation */
698 case 3: /* left&right truncation */
701 case 101: /* process # in term */
702 j = strlen(term_dict);
703 term_dict[j++] = '(';
704 for (i=0; term_sub[i]; i++)
705 if (term_sub[i] == '#' && i > 2)
707 term_dict[j++] = '.';
708 term_dict[j++] = '*';
711 verbatim_char (term_sub[i], &j, term_dict);
712 strcpy (term_dict+j, ")");
713 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
714 &max_pos, grep_handle);
716 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
719 case 102: /* regular expression */
720 sprintf (term_dict + strlen(term_dict), "(%s)", term_sub);
721 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
722 &max_pos, grep_handle);
724 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
729 if (max_pos <= strlen(basenames[base_no]))
731 zi->errCode = 109; /* Database unavailable */
732 zi->errString = basenames[base_no];
736 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
740 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
744 Z_Term *term = zapt->term;
746 sizez = term->u.general->len;
747 if (sizez > IT_MAX_WORD)
749 for (i = 0; i < sizez; i++)
750 termz[i] = index_char_cvt (term->u.general->buf[i]);
754 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
755 Z_AttributesPlusTerm *zapt,
756 oid_value attributeSet,
757 int num_bases, char **basenames)
759 rset_relevance_parms parms;
760 char termz[IT_MAX_WORD+1];
761 char term_sub[IT_MAX_WORD+1];
762 struct grep_info grep_info;
763 char *p0 = termz, *p1 = NULL;
766 parms.key_size = sizeof(struct it_key);
768 parms.cmp = key_compare;
769 parms.is = zi->wordIsam;
771 if (zapt->term->which != Z_Term_general)
776 trans_term (zi, zapt, termz);
777 grep_info.isam_p_indx = 0;
778 grep_info.isam_p_size = 0;
779 grep_info.isam_p_buf = NULL;
782 if ((p1 = strchr (p0, ' ')))
784 memcpy (term_sub, p0, p1-p0);
785 term_sub[p1-p0] = '\0';
788 strcpy (term_sub, p0);
789 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
790 num_bases, basenames))
798 parms.isam_positions = grep_info.isam_p_buf;
799 parms.no_isam_positions = grep_info.isam_p_indx;
800 if (grep_info.isam_p_indx > 0)
801 result = rset_create (rset_kind_relevance, &parms);
803 result = rset_create (rset_kind_null, NULL);
804 xfree (grep_info.isam_p_buf);
808 static RSET rpn_search_APT_word (ZServerInfo *zi,
809 Z_AttributesPlusTerm *zapt,
810 oid_value attributeSet,
811 int num_bases, char **basenames)
813 rset_isam_parms parms;
814 char termz[IT_MAX_WORD+1];
815 struct grep_info grep_info;
818 if (zapt->term->which != Z_Term_general)
823 trans_term (zi, zapt, termz);
825 grep_info.isam_p_indx = 0;
826 grep_info.isam_p_size = 0;
827 grep_info.isam_p_buf = NULL;
829 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
830 num_bases, basenames))
832 if (grep_info.isam_p_indx < 1)
833 result = rset_create (rset_kind_null, NULL);
834 else if (grep_info.isam_p_indx == 1)
836 parms.is = zi->wordIsam;
837 parms.pos = *grep_info.isam_p_buf;
838 result = rset_create (rset_kind_isam, &parms);
841 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
842 grep_info.isam_p_indx);
843 xfree (grep_info.isam_p_buf);
847 static RSET rpn_prox (RSET *rset, int rset_no)
855 rset_temp_parms parms;
857 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
858 more = xmalloc (sizeof(*more)*rset_no);
859 buf = xmalloc (sizeof(*buf)*rset_no);
861 for (i = 0; i<rset_no; i++)
863 buf[i] = xmalloc (sizeof(**buf));
864 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
865 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
869 rset_close (rset[i], rsfd[i]);
876 return rset_create (rset_kind_null, NULL);
879 parms.key_size = sizeof (struct it_key);
880 result = rset_create (rset_kind_temp, &parms);
881 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
885 for (i = 1; i<rset_no; i++)
894 cmp = key_compare (buf[i], buf[i-1]);
897 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
902 if (buf[i-1]->seqno+1 != buf[i]->seqno)
904 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
910 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
916 rset_write (result, rsfd_result, buf[0]);
917 more[0] = rset_read (*rset, *rsfd, *buf);
921 for (i = 0; i<rset_no; i++)
923 rset_close (rset[i], rsfd[i]);
926 rset_close (result, rsfd_result);
933 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
934 Z_AttributesPlusTerm *zapt,
935 oid_value attributeSet,
936 int num_bases, char **basenames)
938 char termz[IT_MAX_WORD+1];
939 char term_sub[IT_MAX_WORD+1];
940 char *p0 = termz, *p1 = NULL;
941 RSET rset[60], result;
943 struct grep_info grep_info;
945 if (zapt->term->which != Z_Term_general)
950 trans_term (zi, zapt, termz);
952 grep_info.isam_p_size = 0;
953 grep_info.isam_p_buf = NULL;
957 if ((p1 = strchr (p0, ' ')))
959 memcpy (term_sub, p0, p1-p0);
960 term_sub[p1-p0] = '\0';
963 strcpy (term_sub, p0);
965 grep_info.isam_p_indx = 0;
966 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
967 num_bases, basenames))
969 if (grep_info.isam_p_indx == 0)
970 rset[rset_no] = rset_create (rset_kind_null, NULL);
971 else if (grep_info.isam_p_indx > 1)
972 rset[rset_no] = rset_trunc (zi->wordIsam,
973 grep_info.isam_p_buf,
974 grep_info.isam_p_indx);
977 rset_isam_parms parms;
979 parms.is = zi->wordIsam;
980 parms.pos = *grep_info.isam_p_buf;
981 rset[rset_no] = rset_create (rset_kind_isam, &parms);
983 assert (rset[rset_no]);
984 if (++rset_no >= sizeof(rset)/sizeof(*rset))
992 xfree (grep_info.isam_p_buf);
994 return rset_create (rset_kind_null, NULL);
995 else if (rset_no == 1)
998 result = rpn_prox (rset, rset_no);
999 for (i = 0; i<rset_no; i++)
1000 rset_delete (rset[i]);
1004 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1005 oid_value attributeSet)
1010 rset_temp_parms parms;
1011 char termz[IT_MAX_WORD+1];
1013 if (zapt->term->which != Z_Term_general)
1018 parms.key_size = sizeof (struct it_key);
1019 result = rset_create (rset_kind_temp, &parms);
1020 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1022 trans_term (zi, zapt, termz);
1023 key.sysno = atoi (termz);
1026 rset_write (result, rsfd, &key);
1027 rset_close (result, rsfd);
1031 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1032 oid_value attributeSet,
1033 int num_bases, char **basenames)
1037 int relation_value, structure_value;
1039 attr_init (&relation, zapt, 2);
1040 attr_init (&structure, zapt, 4);
1042 relation_value = attr_find (&relation, NULL);
1043 structure_value = attr_find (&structure, NULL);
1044 switch (structure_value)
1047 if (relation_value == 102) /* relevance relation */
1048 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1049 num_bases, basenames);
1050 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1051 num_bases, basenames);
1052 case 1: /* phrase */
1053 if (relation_value == 102) /* relevance relation */
1054 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1055 num_bases, basenames);
1056 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1057 num_bases, basenames);
1060 if (relation_value == 102) /* relevance relation */
1061 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1062 num_bases, basenames);
1063 return rpn_search_APT_word (zi, zapt, attributeSet,
1064 num_bases, basenames);
1069 case 5: /* date - normalized */
1071 case 6: /* word list */
1072 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1073 num_bases, basenames);
1074 case 100: /* date - un-normalized */
1076 case 101: /* name - normalized */
1078 case 102: /* date - un-normalized */
1080 case 103: /* structure */
1084 case 105: /* free-form-text */
1085 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1086 num_bases, basenames);
1087 case 106: /* document-text */
1088 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1089 num_bases, basenames);
1090 case 107: /* local-number */
1091 return rpn_search_APT_local (zi, zapt, attributeSet);
1092 case 108: /* string */
1093 return rpn_search_APT_word (zi, zapt, attributeSet,
1094 num_bases, basenames);
1095 case 109: /* numeric string */
1102 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1106 if (!(s = resultSetGet (zi, resultSetId)))
1107 return rset_create (rset_kind_null, NULL);
1111 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1112 oid_value attributeSet,
1113 int num_bases, char **basenames)
1116 if (zs->which == Z_RPNStructure_complex)
1118 rset_bool_parms bool_parms;
1120 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1122 num_bases, basenames);
1123 if (bool_parms.rset_l == NULL)
1125 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1127 num_bases, basenames);
1128 if (bool_parms.rset_r == NULL)
1130 rset_delete (bool_parms.rset_l);
1133 bool_parms.key_size = sizeof(struct it_key);
1134 bool_parms.cmp = key_compare;
1136 switch (zs->u.complex->operator->which)
1138 case Z_Operator_and:
1139 r = rset_create (rset_kind_and, &bool_parms);
1142 r = rset_create (rset_kind_or, &bool_parms);
1144 case Z_Operator_and_not:
1145 r = rset_create (rset_kind_not, &bool_parms);
1151 else if (zs->which == Z_RPNStructure_simple)
1153 if (zs->u.simple->which == Z_Operand_APT)
1155 logf (LOG_DEBUG, "rpn_search_APT");
1156 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1157 attributeSet, num_bases, basenames);
1159 else if (zs->u.simple->which == Z_Operand_resultSetId)
1161 logf (LOG_DEBUG, "rpn_search_ref");
1162 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1176 static void count_set (RSET r, int *count)
1183 logf (LOG_DEBUG, "rpn_save_set");
1185 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1186 while (rset_read (r, rfd, &key))
1188 if (key.sysno != psysno)
1195 rset_close (r, rfd);
1196 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1199 int rpn_search (ZServerInfo *zi,
1200 Z_RPNQuery *rpn, int num_bases, char **basenames,
1201 const char *setname, int *hits)
1205 oid_value attributeSet;
1210 zi->errString = NULL;
1212 attrset = oid_getentbyoid (rpn->attributeSetId);
1213 attributeSet = attrset->value;
1214 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1215 num_bases, basenames);
1218 count_set (rset, hits);
1219 resultSetAdd (zi, setname, 1, rset);
1221 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1226 struct scan_entry *list;
1233 static int scan_handle (Dict_char *name, const char *info, int pos,
1236 int len_prefix, idx;
1239 struct scan_info *scan_info = client;
1241 rset_isam_parms parms;
1243 len_prefix = strlen(scan_info->prefix);
1244 if (memcmp (name, scan_info->prefix, len_prefix))
1247 idx = scan_info->after - pos + scan_info->before;
1250 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1251 strlen(name + len_prefix)+1);
1252 strcpy (scan_info->list[idx].term, name + len_prefix);
1253 assert (*info == sizeof(isam_p));
1254 memcpy (&isam_p, info+1, sizeof(isam_p));
1255 parms.is = scan_info->isam;
1258 rset = rset_create (rset_kind_isam, &parms);
1259 count_set (rset, &scan_info->list[idx].occurrences);
1262 scan_info->list[idx].occurrences = 1;
1264 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1269 static int dummy_handle (Dict_char *name, const char *info, void *p)
1274 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1275 int num_bases, char **basenames,
1276 int *position, int *num_entries, struct scan_entry **list,
1279 int i, j, sizez, max_pos;
1280 int pos = *position;
1281 int num = *num_entries;
1284 char termz[IT_MAX_WORD+20];
1287 Z_Term *term = zapt->term;
1288 struct scan_info scan_info;
1290 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1294 scan_info.before = before = pos-1;
1295 scan_info.after = after = 1+num-pos;
1296 scan_info.odr = zi->odr;
1298 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1300 scan_info.isam = zi->wordIsam;
1301 scan_info.list = odr_malloc (zi->odr, (before+after)*
1302 sizeof(*scan_info.list));
1303 for (j = 0; j<before+after; j++)
1304 scan_info.list[j].term = NULL;
1305 attr_init (&use, zapt, 1);
1306 use_value = attr_find (&use, NULL);
1307 logf (LOG_DEBUG, "use value %d", use_value);
1309 if (use_value == -1)
1311 i = index_word_prefix (termz, 1, use_value, *basenames);
1313 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
1315 if (max_pos <= strlen(*basenames))
1317 zi->errString = *basenames;
1318 return zi->errCode = 109; /* Database unavailable */
1320 strcpy (scan_info.prefix, termz);
1321 sizez = term->u.general->len;
1322 if (sizez > IT_MAX_WORD)
1323 sizez = IT_MAX_WORD;
1324 for (j = 0; j<sizez; j++)
1325 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1328 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1330 *status = BEND_SCAN_SUCCESS;
1332 for (i = 0; i<scan_info.after; i++)
1333 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1337 *status = BEND_SCAN_PARTIAL;
1339 for (i = 0; i<scan_info.before; i++)
1340 if (scan_info.list[i].term)
1343 *status = BEND_SCAN_PARTIAL;
1347 *list = scan_info.list+i; /* list is set to first 'real' entry */
1349 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1350 zi->errCode = 114; /* if no entries was found */
1351 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1352 *position, *num_entries);
1354 logf (LOG_DEBUG, "scan error: %d", zi->errCode);