2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.36 1995-12-06 12:41:27 adam
8 * New command 'stat' for the index program.
9 * Filenames can be read from stdin by specifying '-'.
10 * Bug fix/enhancement of the transformation from terms to regular
11 * expressons in the search engine.
13 * Revision 1.35 1995/11/27 09:29:00 adam
14 * Bug fixes regarding conversion to regular expressions.
16 * Revision 1.34 1995/11/16 17:00:56 adam
17 * Better logging of rpn query.
19 * Revision 1.33 1995/11/01 13:58:28 quinn
20 * Moving data1 to yaz/retrieval
22 * Revision 1.32 1995/10/27 14:00:11 adam
23 * Implemented detection of database availability.
25 * Revision 1.31 1995/10/17 18:02:10 adam
26 * New feature: databases. Implemented as prefix to words in dictionary.
28 * Revision 1.30 1995/10/16 09:32:38 adam
29 * More work on relational op.
31 * Revision 1.29 1995/10/13 16:01:49 adam
34 * Revision 1.28 1995/10/13 12:26:43 adam
35 * Optimization of truncation.
37 * Revision 1.27 1995/10/12 17:07:22 adam
40 * Revision 1.26 1995/10/12 12:40:54 adam
41 * Bug fixes in rpn_prox.
43 * Revision 1.25 1995/10/10 13:59:24 adam
44 * Function rset_open changed its wflag parameter to general flags.
46 * Revision 1.24 1995/10/09 16:18:37 adam
47 * Function dict_lookup_grep got extra client data parameter.
49 * Revision 1.23 1995/10/06 16:33:37 adam
50 * Use attribute mappings.
52 * Revision 1.22 1995/10/06 15:07:39 adam
53 * Structure 'local-number' handled.
55 * Revision 1.21 1995/10/06 13:52:06 adam
56 * Bug fixes. Handler may abort further scanning.
58 * Revision 1.20 1995/10/06 11:06:33 adam
59 * Scan entries include 'occurrences' now.
61 * Revision 1.19 1995/10/06 10:43:56 adam
62 * Scan added. 'occurrences' in scan entries not set yet.
64 * Revision 1.18 1995/10/04 16:57:20 adam
65 * Key input and merge sort in one pass.
67 * Revision 1.17 1995/10/04 12:55:17 adam
68 * Bug fix in ranked search. Use=Any keys inserted.
70 * Revision 1.16 1995/10/02 16:24:40 adam
71 * Use attribute actually used in search requests.
73 * Revision 1.15 1995/10/02 15:18:52 adam
74 * New member in recRetrieveCtrl: diagnostic.
76 * Revision 1.14 1995/09/28 12:10:32 adam
77 * Bug fixes. Field prefix used in queries.
79 * Revision 1.13 1995/09/18 14:17:50 adam
82 * Revision 1.12 1995/09/15 14:45:21 adam
86 * Revision 1.11 1995/09/14 11:53:27 adam
87 * First work on regular expressions/truncations.
89 * Revision 1.10 1995/09/11 15:23:26 adam
90 * More work on relevance search.
92 * Revision 1.9 1995/09/11 13:09:35 adam
93 * More work on relevance feedback.
95 * Revision 1.8 1995/09/08 14:52:27 adam
96 * Minor changes. Dictionary is lower case now.
98 * Revision 1.7 1995/09/07 13:58:36 adam
99 * New parameter: result-set file descriptor (RSFD) to support multiple
100 * positions within the same result-set.
101 * Boolean operators: and, or, not implemented.
102 * Result-set references.
104 * Revision 1.6 1995/09/06 16:11:18 adam
105 * Option: only one word key per file.
107 * Revision 1.5 1995/09/06 10:33:04 adam
108 * More work on present. Some log messages removed.
110 * Revision 1.4 1995/09/05 15:28:40 adam
111 * More work on search engine.
113 * Revision 1.3 1995/09/04 15:20:22 adam
116 * Revision 1.2 1995/09/04 12:33:43 adam
117 * Various cleanup. YAZ util used instead.
119 * Revision 1.1 1995/09/04 09:10:40 adam
120 * More work on index add/del/update.
121 * Merge sort implemented.
122 * Initial work on z39 server.
131 #include "attribute.h"
139 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
144 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
145 attp = att_getentbyatt (attrSet, attrUse);
148 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
149 return index_word_prefix (string, attp->attset_ordinal,
150 attp->local_attribute, basename);
157 Z_AttributesPlusTerm *zapt;
160 static int attr_find (AttrType *src, oid_value *attributeSetP)
162 while (src->major < src->zapt->num_attributes)
164 Z_AttributeElement *element;
166 element = src->zapt->attributeList[src->major];
167 if (src->type == *element->attributeType)
169 switch (element->which)
171 case Z_AttributeValue_numeric:
173 if (element->attributeSet && attributeSetP)
177 attrset = oid_getentbyoid (element->attributeSet);
178 *attributeSetP = attrset->value;
180 return *element->value.numeric;
182 case Z_AttributeValue_complex:
183 if (src->minor >= element->value.complex->num_list ||
184 element->value.complex->list[src->minor]->which !=
185 Z_StringOrNumeric_numeric)
188 if (element->attributeSet && attributeSetP)
192 attrset = oid_getentbyoid (element->attributeSet);
193 *attributeSetP = attrset->value;
195 return *element->value.complex->list[src->minor-1]->u.numeric;
205 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
219 int (*cmp)(const void *p1, const void *p2);
226 static void heap_swap (struct trunc_info *ti, int i1, int i2)
231 ti->ptr[i1] = ti->ptr[i2];
235 static void heap_delete (struct trunc_info *ti)
237 int cur = 1, child = 2;
239 heap_swap (ti, 1, ti->heapnum--);
240 while (child <= ti->heapnum) {
241 if (child < ti->heapnum &&
242 (*ti->cmp)(ti->heap[ti->ptr[child]],
243 ti->heap[ti->ptr[1+child]]) > 0)
245 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
246 ti->heap[ti->ptr[child]]) > 0)
248 heap_swap (ti, cur, child);
257 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
261 cur = ++(ti->heapnum);
262 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
263 ti->indx[ti->ptr[cur]] = indx;
265 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
266 ti->heap[ti->ptr[cur]]) > 0)
268 heap_swap (ti, cur, parent);
275 struct trunc_info *heap_init (int size, int key_size,
276 int (*cmp)(const void *p1, const void *p2))
278 struct trunc_info *ti = xmalloc (sizeof(*ti));
283 ti->keysize = key_size;
285 ti->indx = xmalloc (size * sizeof(*ti->indx));
286 ti->heap = xmalloc (size * sizeof(*ti->heap));
287 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
288 ti->swapbuf = xmalloc (ti->keysize);
289 ti->tmpbuf = xmalloc (ti->keysize);
290 ti->buf = xmalloc (size * ti->keysize);
291 for (i = size; --i >= 0; )
294 ti->heap[i] = ti->buf + ti->keysize * i;
299 static void heap_close (struct trunc_info *ti)
309 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
314 rset_temp_parms parms;
316 parms.key_size = sizeof(struct it_key);
317 result = rset_create (rset_kind_temp, &parms);
318 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
320 if (to - from > merge_chunk)
324 int i, i_add = (to-from)/merge_chunk + 1;
325 struct trunc_info *ti;
327 int rsmax = (to-from)/i_add + 1;
329 rset = xmalloc (sizeof(*rset) * rsmax);
330 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
332 for (i = from; i < to; i += i_add)
335 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
338 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
342 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
343 for (i = rscur; --i >= 0; )
345 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
346 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
347 heap_insert (ti, ti->tmpbuf, i);
350 rset_close (rset[i], rsfd[i]);
351 rset_delete (rset[i]);
356 int n = ti->indx[ti->ptr[1]];
358 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
362 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
365 rset_close (rset[n], rsfd[n]);
366 rset_delete (rset[n]);
369 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
372 heap_insert (ti, ti->tmpbuf, n);
385 struct trunc_info *ti;
387 ispt = xmalloc (sizeof(*ispt) * (to-from));
389 ti = heap_init (to-from, sizeof(struct it_key),
391 for (i = to-from; --i >= 0; )
393 ispt[i] = is_position (isam, isam_p[from+i]);
394 if (is_readkey (ispt[i], ti->tmpbuf))
395 heap_insert (ti, ti->tmpbuf, i);
397 is_pt_free (ispt[i]);
401 int n = ti->indx[ti->ptr[1]];
403 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
405 /* section that preserve all keys */
407 if (is_readkey (ispt[n], ti->tmpbuf))
408 heap_insert (ti, ti->tmpbuf, n);
410 is_pt_free (ispt[n]);
412 /* section that preserve all keys with unique sysnos */
415 if (!is_readkey (ispt[n], ti->tmpbuf))
418 is_pt_free (ispt[n]);
421 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
424 heap_insert (ti, ti->tmpbuf, n);
433 rset_close (result, result_rsfd);
437 static int isam_trunc_cmp (const void *p1, const void *p2)
439 ISAM_P i1 = *(ISAM_P*) p1;
440 ISAM_P i2 = *(ISAM_P*) p2;
443 d = is_type (i1) - is_type (i2);
446 return is_block (i1) - is_block (i2);
449 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
452 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
453 return rset_trunc_r (isam, isam_p, 0, no, 100);
462 static void add_isam_p (const char *info, struct grep_info *p)
464 if (p->isam_p_indx == p->isam_p_size)
466 ISAM_P *new_isam_p_buf;
468 p->isam_p_size = 2*p->isam_p_size + 100;
469 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
473 memcpy (new_isam_p_buf, p->isam_p_buf,
474 p->isam_p_indx * sizeof(*p->isam_p_buf));
475 xfree (p->isam_p_buf);
477 p->isam_p_buf = new_isam_p_buf;
479 assert (*info == sizeof(*p->isam_p_buf));
480 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
484 static int grep_handle (Dict_char *name, const char *info, void *p)
486 logf (LOG_DEBUG, "dict name: %s", name);
487 add_isam_p (info, p);
491 static void gen_regular_rel (char *dst, int val, int islt)
499 sprintf (numstr, "%d", val);
500 for (w = strlen(numstr); --w >= 0; pos++)
519 strcpy (dst + dst_p, numstr);
520 dst_p = strlen(dst) - pos - 1;
548 for (i = 0; i<pos; i++)
561 for (i=1; i<pos; i++)
562 strcat (dst, "[0-9]?");
566 for (i = 0; i <= pos; i++)
567 strcat (dst, "[0-9]");
568 strcat (dst, "[0-9]*");
573 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
574 const char *term_sub,
576 oid_value attributeSet,
577 struct grep_info *grep_info,
585 attr_init (&relation, zapt, 2);
586 relation_value = attr_find (&relation, NULL);
587 term_value = atoi (term_sub);
589 switch (relation_value)
594 logf (LOG_DEBUG, "Relation <");
595 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
600 logf (LOG_DEBUG, "Relation <=");
601 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
606 logf (LOG_DEBUG, "Relation >=");
607 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
612 logf (LOG_DEBUG, "Relation >");
613 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
618 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
619 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
622 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
623 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
627 static void verbatim_char (int ch, int *indx, char *dst)
630 dst[(*indx)++] = '\\';
634 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
635 const char *term_sub,
636 oid_value attributeSet, struct grep_info *grep_info,
637 int num_bases, char **basenames)
639 char term_dict[2*IT_MAX_WORD+2];
640 int i, j, r, base_no;
642 int truncation_value;
645 oid_value curAttributeSet = attributeSet;
647 attr_init (&use, zapt, 1);
648 use_value = attr_find (&use, &curAttributeSet);
649 logf (LOG_DEBUG, "use value %d", use_value);
650 attr_init (&truncation, zapt, 5);
651 truncation_value = attr_find (&truncation, NULL);
652 logf (LOG_DEBUG, "truncation value %d", truncation_value);
657 for (base_no = 0; base_no < num_bases; base_no++)
660 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
668 if (!relational_term (zi, zapt, term_sub, term_dict,
669 attributeSet, grep_info, &max_pos))
671 switch (truncation_value)
673 case -1: /* not specified */
674 case 100: /* do not truncate */
675 j = strlen(term_dict);
676 term_dict[j++] = '(';
677 for (i = 0; term_sub[i]; i++)
678 verbatim_char (term_sub[i], &j, term_dict);
679 strcpy (term_dict+j, ")");
680 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
681 &max_pos, grep_handle);
683 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
685 case 1: /* right truncation */
686 j = strlen(term_dict);
687 term_dict[j++] = '(';
688 for (i = 0; term_sub[i]; i++)
689 verbatim_char (term_sub[i], &j, term_dict);
690 strcpy (term_dict+j, ".*)");
691 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
692 &max_pos, grep_handle);
694 case 2: /* left truncation */
695 case 3: /* left&right truncation */
698 case 101: /* process # in term */
699 j = strlen(term_dict);
700 term_dict[j++] = '(';
701 for (i=0; term_sub[i]; i++)
702 if (term_sub[i] == '#' && i > 2)
704 term_dict[j++] = '.';
705 term_dict[j++] = '*';
708 verbatim_char (term_sub[i], &j, term_dict);
709 strcpy (term_dict+j, ")");
710 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
711 &max_pos, grep_handle);
713 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
716 case 102: /* regular expression */
717 sprintf (term_dict + strlen(term_dict), "(%s)", term_sub);
718 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
719 &max_pos, grep_handle);
721 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
726 if (max_pos <= strlen(basenames[base_no]))
728 zi->errCode = 109; /* Database unavailable */
729 zi->errString = basenames[base_no];
733 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
737 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
741 Z_Term *term = zapt->term;
743 sizez = term->u.general->len;
744 if (sizez > IT_MAX_WORD)
746 for (i = 0; i < sizez; i++)
747 termz[i] = index_char_cvt (term->u.general->buf[i]);
751 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
752 Z_AttributesPlusTerm *zapt,
753 oid_value attributeSet,
754 int num_bases, char **basenames)
756 rset_relevance_parms parms;
757 char termz[IT_MAX_WORD+1];
758 char term_sub[IT_MAX_WORD+1];
759 struct grep_info grep_info;
760 char *p0 = termz, *p1 = NULL;
763 parms.key_size = sizeof(struct it_key);
765 parms.cmp = key_compare;
766 parms.is = zi->wordIsam;
768 if (zapt->term->which != Z_Term_general)
773 trans_term (zi, zapt, termz);
774 grep_info.isam_p_indx = 0;
775 grep_info.isam_p_size = 0;
776 grep_info.isam_p_buf = NULL;
779 if ((p1 = strchr (p0, ' ')))
781 memcpy (term_sub, p0, p1-p0);
782 term_sub[p1-p0] = '\0';
785 strcpy (term_sub, p0);
786 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
787 num_bases, basenames))
795 parms.isam_positions = grep_info.isam_p_buf;
796 parms.no_isam_positions = grep_info.isam_p_indx;
797 if (grep_info.isam_p_indx > 0)
798 result = rset_create (rset_kind_relevance, &parms);
800 result = rset_create (rset_kind_null, NULL);
801 xfree (grep_info.isam_p_buf);
805 static RSET rpn_search_APT_word (ZServerInfo *zi,
806 Z_AttributesPlusTerm *zapt,
807 oid_value attributeSet,
808 int num_bases, char **basenames)
810 rset_isam_parms parms;
811 char termz[IT_MAX_WORD+1];
812 struct grep_info grep_info;
815 if (zapt->term->which != Z_Term_general)
820 trans_term (zi, zapt, termz);
822 grep_info.isam_p_indx = 0;
823 grep_info.isam_p_size = 0;
824 grep_info.isam_p_buf = NULL;
826 if (trunc_term (zi, zapt, termz, attributeSet, &grep_info,
827 num_bases, basenames))
829 if (grep_info.isam_p_indx < 1)
830 result = rset_create (rset_kind_null, NULL);
831 else if (grep_info.isam_p_indx == 1)
833 parms.is = zi->wordIsam;
834 parms.pos = *grep_info.isam_p_buf;
835 result = rset_create (rset_kind_isam, &parms);
838 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
839 grep_info.isam_p_indx);
840 xfree (grep_info.isam_p_buf);
844 static RSET rpn_prox (RSET *rset, int rset_no)
852 rset_temp_parms parms;
854 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
855 more = xmalloc (sizeof(*more)*rset_no);
856 buf = xmalloc (sizeof(*buf)*rset_no);
858 for (i = 0; i<rset_no; i++)
860 buf[i] = xmalloc (sizeof(**buf));
861 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
862 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
866 rset_close (rset[i], rsfd[i]);
873 return rset_create (rset_kind_null, NULL);
876 parms.key_size = sizeof (struct it_key);
877 result = rset_create (rset_kind_temp, &parms);
878 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
882 for (i = 1; i<rset_no; i++)
891 cmp = key_compare (buf[i], buf[i-1]);
894 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
899 if (buf[i-1]->seqno+1 != buf[i]->seqno)
901 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
907 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
913 rset_write (result, rsfd_result, buf[0]);
914 more[0] = rset_read (*rset, *rsfd, *buf);
918 for (i = 0; i<rset_no; i++)
920 rset_close (rset[i], rsfd[i]);
923 rset_close (result, rsfd_result);
930 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
931 Z_AttributesPlusTerm *zapt,
932 oid_value attributeSet,
933 int num_bases, char **basenames)
935 char termz[IT_MAX_WORD+1];
936 char term_sub[IT_MAX_WORD+1];
937 char *p0 = termz, *p1 = NULL;
938 RSET rset[60], result;
940 struct grep_info grep_info;
942 if (zapt->term->which != Z_Term_general)
947 trans_term (zi, zapt, termz);
949 grep_info.isam_p_size = 0;
950 grep_info.isam_p_buf = NULL;
954 if ((p1 = strchr (p0, ' ')))
956 memcpy (term_sub, p0, p1-p0);
957 term_sub[p1-p0] = '\0';
960 strcpy (term_sub, p0);
962 grep_info.isam_p_indx = 0;
963 if (trunc_term (zi, zapt, term_sub, attributeSet, &grep_info,
964 num_bases, basenames))
966 if (grep_info.isam_p_indx == 0)
967 rset[rset_no] = rset_create (rset_kind_null, NULL);
968 else if (grep_info.isam_p_indx > 1)
969 rset[rset_no] = rset_trunc (zi->wordIsam,
970 grep_info.isam_p_buf,
971 grep_info.isam_p_indx);
974 rset_isam_parms parms;
976 parms.is = zi->wordIsam;
977 parms.pos = *grep_info.isam_p_buf;
978 rset[rset_no] = rset_create (rset_kind_isam, &parms);
980 assert (rset[rset_no]);
981 if (++rset_no >= sizeof(rset)/sizeof(*rset))
989 xfree (grep_info.isam_p_buf);
991 return rset_create (rset_kind_null, NULL);
992 else if (rset_no == 1)
995 result = rpn_prox (rset, rset_no);
996 for (i = 0; i<rset_no; i++)
997 rset_delete (rset[i]);
1001 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1002 oid_value attributeSet)
1007 rset_temp_parms parms;
1008 char termz[IT_MAX_WORD+1];
1010 if (zapt->term->which != Z_Term_general)
1015 parms.key_size = sizeof (struct it_key);
1016 result = rset_create (rset_kind_temp, &parms);
1017 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1019 trans_term (zi, zapt, termz);
1020 key.sysno = atoi (termz);
1023 rset_write (result, rsfd, &key);
1024 rset_close (result, rsfd);
1028 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1029 oid_value attributeSet,
1030 int num_bases, char **basenames)
1034 int relation_value, structure_value;
1036 attr_init (&relation, zapt, 2);
1037 attr_init (&structure, zapt, 4);
1039 relation_value = attr_find (&relation, NULL);
1040 structure_value = attr_find (&structure, NULL);
1041 switch (structure_value)
1044 if (relation_value == 102) /* relevance relation */
1045 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1046 num_bases, basenames);
1047 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1048 num_bases, basenames);
1049 case 1: /* phrase */
1050 if (relation_value == 102) /* relevance relation */
1051 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1052 num_bases, basenames);
1053 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1054 num_bases, basenames);
1057 if (relation_value == 102) /* relevance relation */
1058 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1059 num_bases, basenames);
1060 return rpn_search_APT_word (zi, zapt, attributeSet,
1061 num_bases, basenames);
1066 case 5: /* date - normalized */
1068 case 6: /* word list */
1069 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1070 num_bases, basenames);
1071 case 100: /* date - un-normalized */
1073 case 101: /* name - normalized */
1075 case 102: /* date - un-normalized */
1077 case 103: /* structure */
1081 case 105: /* free-form-text */
1082 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1083 num_bases, basenames);
1084 case 106: /* document-text */
1085 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1086 num_bases, basenames);
1087 case 107: /* local-number */
1088 return rpn_search_APT_local (zi, zapt, attributeSet);
1089 case 108: /* string */
1090 return rpn_search_APT_word (zi, zapt, attributeSet,
1091 num_bases, basenames);
1092 case 109: /* numeric string */
1099 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1103 if (!(s = resultSetGet (zi, resultSetId)))
1104 return rset_create (rset_kind_null, NULL);
1108 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1109 oid_value attributeSet,
1110 int num_bases, char **basenames)
1113 if (zs->which == Z_RPNStructure_complex)
1115 rset_bool_parms bool_parms;
1117 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1119 num_bases, basenames);
1120 if (bool_parms.rset_l == NULL)
1122 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1124 num_bases, basenames);
1125 if (bool_parms.rset_r == NULL)
1127 rset_delete (bool_parms.rset_l);
1130 bool_parms.key_size = sizeof(struct it_key);
1131 bool_parms.cmp = key_compare;
1133 switch (zs->u.complex->operator->which)
1135 case Z_Operator_and:
1136 r = rset_create (rset_kind_and, &bool_parms);
1139 r = rset_create (rset_kind_or, &bool_parms);
1141 case Z_Operator_and_not:
1142 r = rset_create (rset_kind_not, &bool_parms);
1148 else if (zs->which == Z_RPNStructure_simple)
1150 if (zs->u.simple->which == Z_Operand_APT)
1152 logf (LOG_DEBUG, "rpn_search_APT");
1153 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1154 attributeSet, num_bases, basenames);
1156 else if (zs->u.simple->which == Z_Operand_resultSetId)
1158 logf (LOG_DEBUG, "rpn_search_ref");
1159 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1173 static void count_set (RSET r, int *count)
1179 logf (LOG_DEBUG, "rpn_save_set");
1181 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1182 while (rset_read (r, rfd, &key))
1184 if (key.sysno != psysno)
1190 rset_close (r, rfd);
1191 logf (LOG_DEBUG, "%d distinct sysnos", *count);
1194 int rpn_search (ZServerInfo *zi,
1195 Z_RPNQuery *rpn, int num_bases, char **basenames,
1196 const char *setname, int *hits)
1200 oid_value attributeSet;
1205 zi->errString = NULL;
1207 attrset = oid_getentbyoid (rpn->attributeSetId);
1208 attributeSet = attrset->value;
1209 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1210 num_bases, basenames);
1213 count_set (rset, hits);
1214 resultSetAdd (zi, setname, 1, rset);
1216 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1221 struct scan_entry *list;
1228 static int scan_handle (Dict_char *name, const char *info, int pos,
1231 int len_prefix, idx;
1234 struct scan_info *scan_info = client;
1236 rset_isam_parms parms;
1238 len_prefix = strlen(scan_info->prefix);
1239 if (memcmp (name, scan_info->prefix, len_prefix))
1242 idx = scan_info->after - pos + scan_info->before;
1245 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1246 strlen(name + len_prefix)+1);
1247 strcpy (scan_info->list[idx].term, name + len_prefix);
1248 assert (*info == sizeof(isam_p));
1249 memcpy (&isam_p, info+1, sizeof(isam_p));
1250 parms.is = scan_info->isam;
1253 rset = rset_create (rset_kind_isam, &parms);
1254 count_set (rset, &scan_info->list[idx].occurrences);
1257 scan_info->list[idx].occurrences = 1;
1259 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1264 static int dummy_handle (Dict_char *name, const char *info, void *p)
1269 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1270 int num_bases, char **basenames,
1271 int *position, int *num_entries, struct scan_entry **list,
1274 int i, j, sizez, max_pos;
1275 int pos = *position;
1276 int num = *num_entries;
1279 char termz[IT_MAX_WORD+20];
1282 Z_Term *term = zapt->term;
1283 struct scan_info scan_info;
1285 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1289 scan_info.before = before = pos-1;
1290 scan_info.after = after = 1+num-pos;
1291 scan_info.odr = zi->odr;
1293 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1295 scan_info.isam = zi->wordIsam;
1296 scan_info.list = odr_malloc (zi->odr, (before+after)*
1297 sizeof(*scan_info.list));
1298 for (j = 0; j<before+after; j++)
1299 scan_info.list[j].term = NULL;
1300 attr_init (&use, zapt, 1);
1301 use_value = attr_find (&use, NULL);
1302 logf (LOG_DEBUG, "use value %d", use_value);
1304 if (use_value == -1)
1306 i = index_word_prefix (termz, 1, use_value, *basenames);
1308 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos,
1310 if (max_pos <= strlen(*basenames))
1312 zi->errString = *basenames;
1313 return zi->errCode = 109; /* Database unavailable */
1315 strcpy (scan_info.prefix, termz);
1316 sizez = term->u.general->len;
1317 if (sizez > IT_MAX_WORD)
1318 sizez = IT_MAX_WORD;
1319 for (j = 0; j<sizez; j++)
1320 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1323 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1325 *status = BEND_SCAN_SUCCESS;
1327 for (i = 0; i<scan_info.after; i++)
1328 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1332 *status = BEND_SCAN_PARTIAL;
1334 for (i = 0; i<scan_info.before; i++)
1335 if (scan_info.list[i].term)
1338 *status = BEND_SCAN_PARTIAL;
1342 *list = scan_info.list+i; /* list is set to first 'real' entry */
1344 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1345 zi->errCode = 114; /* if no entries was found */
1346 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1347 *position, *num_entries);
1349 logf (LOG_DEBUG, "scan error: %d", zi->errCode);