2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.43 1996-05-09 09:54:43 adam
8 * Server supports maps from one logical attributes to a list of physical
10 * The extraction process doesn't make space consuming 'any' keys.
12 * Revision 1.42 1996/05/09 07:28:56 quinn
13 * Work towards phrases and multiple registers
15 * Revision 1.41 1996/03/20 09:36:43 adam
16 * Function dict_lookup_grep got extra parameter, init_pos, which marks
17 * from which position in pattern approximate pattern matching should occur.
18 * Approximate pattern matching is used in relevance=re-2.
20 * Revision 1.40 1996/02/02 13:44:44 adam
21 * The public dictionary functions simply use char instead of Dict_char
22 * to represent search strings. Dict_char is used internally only.
24 * Revision 1.39 1996/01/03 16:22:13 quinn
27 * Revision 1.38 1995/12/11 09:12:55 adam
28 * The rec_get function returns NULL if record doesn't exist - will
29 * happen in the server if the result set records have been deleted since
30 * the creation of the set (i.e. the search).
31 * The server saves a result temporarily if it is 'volatile', i.e. the
32 * set is register dependent.
34 * Revision 1.37 1995/12/06 15:05:28 adam
35 * More verbose in count_set.
37 * Revision 1.36 1995/12/06 12:41:27 adam
38 * New command 'stat' for the index program.
39 * Filenames can be read from stdin by specifying '-'.
40 * Bug fix/enhancement of the transformation from terms to regular
41 * expressons in the search engine.
43 * Revision 1.35 1995/11/27 09:29:00 adam
44 * Bug fixes regarding conversion to regular expressions.
46 * Revision 1.34 1995/11/16 17:00:56 adam
47 * Better logging of rpn query.
49 * Revision 1.33 1995/11/01 13:58:28 quinn
50 * Moving data1 to yaz/retrieval
52 * Revision 1.32 1995/10/27 14:00:11 adam
53 * Implemented detection of database availability.
55 * Revision 1.31 1995/10/17 18:02:10 adam
56 * New feature: databases. Implemented as prefix to words in dictionary.
58 * Revision 1.30 1995/10/16 09:32:38 adam
59 * More work on relational op.
61 * Revision 1.29 1995/10/13 16:01:49 adam
64 * Revision 1.28 1995/10/13 12:26:43 adam
65 * Optimization of truncation.
67 * Revision 1.27 1995/10/12 17:07:22 adam
70 * Revision 1.26 1995/10/12 12:40:54 adam
71 * Bug fixes in rpn_prox.
73 * Revision 1.25 1995/10/10 13:59:24 adam
74 * Function rset_open changed its wflag parameter to general flags.
76 * Revision 1.24 1995/10/09 16:18:37 adam
77 * Function dict_lookup_grep got extra client data parameter.
79 * Revision 1.23 1995/10/06 16:33:37 adam
80 * Use attribute mappings.
82 * Revision 1.22 1995/10/06 15:07:39 adam
83 * Structure 'local-number' handled.
85 * Revision 1.21 1995/10/06 13:52:06 adam
86 * Bug fixes. Handler may abort further scanning.
88 * Revision 1.20 1995/10/06 11:06:33 adam
89 * Scan entries include 'occurrences' now.
91 * Revision 1.19 1995/10/06 10:43:56 adam
92 * Scan added. 'occurrences' in scan entries not set yet.
94 * Revision 1.18 1995/10/04 16:57:20 adam
95 * Key input and merge sort in one pass.
97 * Revision 1.17 1995/10/04 12:55:17 adam
98 * Bug fix in ranked search. Use=Any keys inserted.
100 * Revision 1.16 1995/10/02 16:24:40 adam
101 * Use attribute actually used in search requests.
103 * Revision 1.15 1995/10/02 15:18:52 adam
104 * New member in recRetrieveCtrl: diagnostic.
106 * Revision 1.14 1995/09/28 12:10:32 adam
107 * Bug fixes. Field prefix used in queries.
109 * Revision 1.13 1995/09/18 14:17:50 adam
112 * Revision 1.12 1995/09/15 14:45:21 adam
114 * Work on truncation.
116 * Revision 1.11 1995/09/14 11:53:27 adam
117 * First work on regular expressions/truncations.
119 * Revision 1.10 1995/09/11 15:23:26 adam
120 * More work on relevance search.
122 * Revision 1.9 1995/09/11 13:09:35 adam
123 * More work on relevance feedback.
125 * Revision 1.8 1995/09/08 14:52:27 adam
126 * Minor changes. Dictionary is lower case now.
128 * Revision 1.7 1995/09/07 13:58:36 adam
129 * New parameter: result-set file descriptor (RSFD) to support multiple
130 * positions within the same result-set.
131 * Boolean operators: and, or, not implemented.
132 * Result-set references.
134 * Revision 1.6 1995/09/06 16:11:18 adam
135 * Option: only one word key per file.
137 * Revision 1.5 1995/09/06 10:33:04 adam
138 * More work on present. Some log messages removed.
140 * Revision 1.4 1995/09/05 15:28:40 adam
141 * More work on search engine.
143 * Revision 1.3 1995/09/04 15:20:22 adam
146 * Revision 1.2 1995/09/04 12:33:43 adam
147 * Various cleanup. YAZ util used instead.
149 * Revision 1.1 1995/09/04 09:10:40 adam
150 * More work on index add/del/update.
151 * Merge sort implemented.
152 * Initial work on z39 server.
161 #include "attribute.h"
169 int index_word_prefix_map (char *string, oid_value attrSet, int attrUse,
174 logf (LOG_DEBUG, "oid_value attrSet = %d, attrUse = %d", attrSet, attrUse);
175 attp = att_getentbyatt (attrSet, attrUse);
178 logf (LOG_DEBUG, "ord=%d", attp->attset_ordinal);
179 return index_word_prefix (string, attp->attset_ordinal,
180 attp->local_attributes->local, basename);
187 Z_AttributesPlusTerm *zapt;
190 static int attr_find (AttrType *src, oid_value *attributeSetP)
192 while (src->major < src->zapt->num_attributes)
194 Z_AttributeElement *element;
196 element = src->zapt->attributeList[src->major];
197 if (src->type == *element->attributeType)
199 switch (element->which)
201 case Z_AttributeValue_numeric:
203 if (element->attributeSet && attributeSetP)
207 attrset = oid_getentbyoid (element->attributeSet);
208 *attributeSetP = attrset->value;
210 return *element->value.numeric;
212 case Z_AttributeValue_complex:
213 if (src->minor >= element->value.complex->num_list ||
214 element->value.complex->list[src->minor]->which !=
215 Z_StringOrNumeric_numeric)
218 if (element->attributeSet && attributeSetP)
222 attrset = oid_getentbyoid (element->attributeSet);
223 *attributeSetP = attrset->value;
225 return *element->value.complex->list[src->minor-1]->u.numeric;
235 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
249 int (*cmp)(const void *p1, const void *p2);
256 static void heap_swap (struct trunc_info *ti, int i1, int i2)
261 ti->ptr[i1] = ti->ptr[i2];
265 static void heap_delete (struct trunc_info *ti)
267 int cur = 1, child = 2;
269 heap_swap (ti, 1, ti->heapnum--);
270 while (child <= ti->heapnum) {
271 if (child < ti->heapnum &&
272 (*ti->cmp)(ti->heap[ti->ptr[child]],
273 ti->heap[ti->ptr[1+child]]) > 0)
275 if ((*ti->cmp)(ti->heap[ti->ptr[cur]],
276 ti->heap[ti->ptr[child]]) > 0)
278 heap_swap (ti, cur, child);
287 static void heap_insert (struct trunc_info *ti, const char *buf, int indx)
291 cur = ++(ti->heapnum);
292 memcpy (ti->heap[ti->ptr[cur]], buf, ti->keysize);
293 ti->indx[ti->ptr[cur]] = indx;
295 while (parent && (*ti->cmp)(ti->heap[ti->ptr[parent]],
296 ti->heap[ti->ptr[cur]]) > 0)
298 heap_swap (ti, cur, parent);
305 struct trunc_info *heap_init (int size, int key_size,
306 int (*cmp)(const void *p1, const void *p2))
308 struct trunc_info *ti = xmalloc (sizeof(*ti));
313 ti->keysize = key_size;
315 ti->indx = xmalloc (size * sizeof(*ti->indx));
316 ti->heap = xmalloc (size * sizeof(*ti->heap));
317 ti->ptr = xmalloc (size * sizeof(*ti->ptr));
318 ti->swapbuf = xmalloc (ti->keysize);
319 ti->tmpbuf = xmalloc (ti->keysize);
320 ti->buf = xmalloc (size * ti->keysize);
321 for (i = size; --i >= 0; )
324 ti->heap[i] = ti->buf + ti->keysize * i;
329 static void heap_close (struct trunc_info *ti)
339 static RSET rset_trunc_r (ISAM isam, ISAM_P *isam_p, int from, int to,
344 rset_temp_parms parms;
346 parms.key_size = sizeof(struct it_key);
347 result = rset_create (rset_kind_temp, &parms);
348 result_rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
350 if (to - from > merge_chunk)
354 int i, i_add = (to-from)/merge_chunk + 1;
355 struct trunc_info *ti;
357 int rsmax = (to-from)/i_add + 1;
359 rset = xmalloc (sizeof(*rset) * rsmax);
360 rsfd = xmalloc (sizeof(*rsfd) * rsmax);
362 for (i = from; i < to; i += i_add)
365 rset[rscur] = rset_trunc_r (isam, isam_p, i, i+i_add,
368 rset[rscur] = rset_trunc_r (isam, isam_p, i, to,
372 ti = heap_init (rscur, sizeof(struct it_key), key_compare);
373 for (i = rscur; --i >= 0; )
375 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
376 if (rset_read (rset[i], rsfd[i], ti->tmpbuf))
377 heap_insert (ti, ti->tmpbuf, i);
380 rset_close (rset[i], rsfd[i]);
381 rset_delete (rset[i]);
386 int n = ti->indx[ti->ptr[1]];
388 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
392 if (!rset_read (rset[n], rsfd[n], ti->tmpbuf))
395 rset_close (rset[n], rsfd[n]);
396 rset_delete (rset[n]);
399 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
402 heap_insert (ti, ti->tmpbuf, n);
415 struct trunc_info *ti;
417 ispt = xmalloc (sizeof(*ispt) * (to-from));
419 ti = heap_init (to-from, sizeof(struct it_key),
421 for (i = to-from; --i >= 0; )
423 ispt[i] = is_position (isam, isam_p[from+i]);
424 if (is_readkey (ispt[i], ti->tmpbuf))
425 heap_insert (ti, ti->tmpbuf, i);
427 is_pt_free (ispt[i]);
431 int n = ti->indx[ti->ptr[1]];
433 rset_write (result, result_rsfd, ti->heap[ti->ptr[1]]);
435 /* section that preserve all keys */
437 if (is_readkey (ispt[n], ti->tmpbuf))
438 heap_insert (ti, ti->tmpbuf, n);
440 is_pt_free (ispt[n]);
442 /* section that preserve all keys with unique sysnos */
445 if (!is_readkey (ispt[n], ti->tmpbuf))
448 is_pt_free (ispt[n]);
451 if ((*ti->cmp)(ti->tmpbuf, ti->heap[ti->ptr[1]]) > 1)
454 heap_insert (ti, ti->tmpbuf, n);
463 rset_close (result, result_rsfd);
467 static int isam_trunc_cmp (const void *p1, const void *p2)
469 ISAM_P i1 = *(ISAM_P*) p1;
470 ISAM_P i2 = *(ISAM_P*) p2;
473 d = is_type (i1) - is_type (i2);
476 return is_block (i1) - is_block (i2);
479 static RSET rset_trunc (ISAM isam, ISAM_P *isam_p, int no)
482 qsort (isam_p, no, sizeof(*isam_p), isam_trunc_cmp);
483 return rset_trunc_r (isam, isam_p, 0, no, 100);
492 static void add_isam_p (const char *info, struct grep_info *p)
494 if (p->isam_p_indx == p->isam_p_size)
496 ISAM_P *new_isam_p_buf;
498 p->isam_p_size = 2*p->isam_p_size + 100;
499 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
503 memcpy (new_isam_p_buf, p->isam_p_buf,
504 p->isam_p_indx * sizeof(*p->isam_p_buf));
505 xfree (p->isam_p_buf);
507 p->isam_p_buf = new_isam_p_buf;
509 assert (*info == sizeof(*p->isam_p_buf));
510 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
514 static int grep_handle (char *name, const char *info, void *p)
516 logf (LOG_DEBUG, "dict name: %s", name);
517 add_isam_p (info, p);
521 static void gen_regular_rel (char *dst, int val, int islt)
529 sprintf (numstr, "%d", val);
530 for (w = strlen(numstr); --w >= 0; pos++)
549 strcpy (dst + dst_p, numstr);
550 dst_p = strlen(dst) - pos - 1;
578 for (i = 0; i<pos; i++)
591 for (i=1; i<pos; i++)
592 strcat (dst, "[0-9]?");
596 for (i = 0; i <= pos; i++)
597 strcat (dst, "[0-9]");
598 strcat (dst, "[0-9]*");
603 static int relational_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
604 const char *term_sub,
606 oid_value attributeSet,
607 struct grep_info *grep_info,
615 attr_init (&relation, zapt, 2);
616 relation_value = attr_find (&relation, NULL);
617 term_value = atoi (term_sub);
619 switch (relation_value)
624 logf (LOG_DEBUG, "Relation <");
625 gen_regular_rel (term_dict + strlen(term_dict), term_value-1, 1);
630 logf (LOG_DEBUG, "Relation <=");
631 gen_regular_rel (term_dict + strlen(term_dict), term_value, 1);
636 logf (LOG_DEBUG, "Relation >=");
637 gen_regular_rel (term_dict + strlen(term_dict), term_value, 0);
642 logf (LOG_DEBUG, "Relation >");
643 gen_regular_rel (term_dict + strlen(term_dict), term_value+1, 0);
648 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_dict);
649 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info, max_pos,
652 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
653 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
657 static void verbatim_char (int ch, int *indx, char *dst)
660 dst[(*indx)++] = '\\';
664 static int field_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
665 const char *term_sub,
666 oid_value attributeSet, struct grep_info *grep_info,
667 int num_bases, char **basenames)
669 char term_dict[2*IT_MAX_WORD+2];
670 int i, j, r, base_no;
672 int truncation_value;
675 oid_value curAttributeSet = attributeSet;
677 attr_init (&use, zapt, 1);
678 use_value = attr_find (&use, &curAttributeSet);
679 logf (LOG_DEBUG, "use value %d", use_value);
680 attr_init (&truncation, zapt, 5);
681 truncation_value = attr_find (&truncation, NULL);
682 logf (LOG_DEBUG, "truncation value %d", truncation_value);
687 for (base_no = 0; base_no < num_bases; base_no++)
692 data1_local_attribute *local_attr;
695 attp = att_getentbyatt (curAttributeSet, use_value);
701 for (local_attr = attp->local_attributes; local_attr;
702 local_attr = local_attr->next)
704 prefix_len = index_word_prefix (term_dict, attp->attset_ordinal,
708 if (!relational_term (zi, zapt, term_sub, term_dict,
709 attributeSet, grep_info, &max_pos))
714 switch (truncation_value)
716 case -1: /* not specified */
717 case 100: /* do not truncate */
718 term_dict[j++] = '(';
719 for (i = 0; term_sub[i]; i++)
720 verbatim_char (term_sub[i], &j, term_dict);
721 strcpy (term_dict+j, ")");
722 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
723 &max_pos, 0, grep_handle);
725 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
727 case 1: /* right truncation */
728 term_dict[j++] = '(';
729 for (i = 0; term_sub[i]; i++)
730 verbatim_char (term_sub[i], &j, term_dict);
731 strcpy (term_dict+j, ".*)");
732 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
733 &max_pos, 0, grep_handle);
735 case 2: /* left truncation */
736 case 3: /* left&right truncation */
739 case 101: /* process # in term */
740 term_dict[j++] = '(';
741 for (i=0; term_sub[i]; i++)
742 if (term_sub[i] == '#' && i > 2)
744 term_dict[j++] = '.';
745 term_dict[j++] = '*';
748 verbatim_char (term_sub[i], &j, term_dict);
749 strcpy (term_dict+j, ")");
750 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
751 &max_pos, 0, grep_handle);
753 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
756 case 102: /* regular expression */
757 sprintf (term_dict + j, "(%s)", term_sub);
758 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
759 &max_pos, 0, grep_handle);
761 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
764 case 103: /* regular expression with error correction */
767 if (*cp == '*' && cp[1] && cp[2])
772 sprintf (term_dict + j, "(%s)", cp);
773 r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info,
774 &max_pos, j, grep_handle);
776 logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
781 if (max_pos <= strlen(basenames[base_no]))
783 zi->errCode = 109; /* Database unavailable */
784 zi->errString = basenames[base_no];
789 int prefix_len = index_word_prefix_map (term_dict, curAttributeSet,
797 if (!relational_term (zi, zapt, term_sub, term_dict,
798 attributeSet, grep_info, &max_pos))
803 switch (truncation_value)
805 case -1: /* not specified */
806 case 100: /* do not truncate */
807 term_dict[j++] = '(';
808 for (i = 0; term_sub[i]; i++)
809 verbatim_char (term_sub[i], &j, term_dict);
810 strcpy (term_dict+j, ")");
811 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
812 &max_pos, 0, grep_handle);
814 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
816 case 1: /* right truncation */
817 term_dict[j++] = '(';
818 for (i = 0; term_sub[i]; i++)
819 verbatim_char (term_sub[i], &j, term_dict);
820 strcpy (term_dict+j, ".*)");
821 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
822 &max_pos, 0, grep_handle);
824 case 2: /* left truncation */
825 case 3: /* left&right truncation */
828 case 101: /* process # in term */
829 term_dict[j++] = '(';
830 for (i=0; term_sub[i]; i++)
831 if (term_sub[i] == '#' && i > 2)
833 term_dict[j++] = '.';
834 term_dict[j++] = '*';
837 verbatim_char (term_sub[i], &j, term_dict);
838 strcpy (term_dict+j, ")");
839 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
840 &max_pos, 0, grep_handle);
842 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d",
845 case 102: /* regular expression */
846 sprintf (term_dict + j, "(%s)", term_sub);
847 r = dict_lookup_grep (zi->wordDict, term_dict, 0, grep_info,
848 &max_pos, 0, grep_handle);
850 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
853 case 103: /* regular expression with error correction */
856 if (*cp == '*' && cp[1] && cp[2])
861 sprintf (term_dict + j, "(%s)", cp);
862 r = dict_lookup_grep (zi->wordDict, term_dict, r, grep_info,
863 &max_pos, j, grep_handle);
865 logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
870 if (max_pos <= strlen(basenames[base_no]))
872 zi->errCode = 109; /* Database unavailable */
873 zi->errString = basenames[base_no];
878 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
882 static void trans_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
886 Z_Term *term = zapt->term;
888 sizez = term->u.general->len;
889 if (sizez > IT_MAX_WORD)
891 for (i = 0; i < sizez; i++)
892 termz[i] = index_char_cvt (term->u.general->buf[i]);
896 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
897 Z_AttributesPlusTerm *zapt,
898 oid_value attributeSet,
899 int num_bases, char **basenames)
901 rset_relevance_parms parms;
902 char termz[IT_MAX_WORD+1];
903 char term_sub[IT_MAX_WORD+1];
904 struct grep_info grep_info;
905 char *p0 = termz, *p1 = NULL;
908 parms.key_size = sizeof(struct it_key);
910 parms.cmp = key_compare;
911 parms.is = zi->wordIsam;
913 if (zapt->term->which != Z_Term_general)
918 trans_term (zi, zapt, termz);
919 grep_info.isam_p_indx = 0;
920 grep_info.isam_p_size = 0;
921 grep_info.isam_p_buf = NULL;
924 if ((p1 = strchr (p0, ' ')))
926 memcpy (term_sub, p0, p1-p0);
927 term_sub[p1-p0] = '\0';
930 strcpy (term_sub, p0);
931 if (field_term (zi, zapt, term_sub, attributeSet, &grep_info,
932 num_bases, basenames))
940 parms.isam_positions = grep_info.isam_p_buf;
941 parms.no_isam_positions = grep_info.isam_p_indx;
942 if (grep_info.isam_p_indx > 0)
943 result = rset_create (rset_kind_relevance, &parms);
945 result = rset_create (rset_kind_null, NULL);
946 xfree (grep_info.isam_p_buf);
950 static RSET rpn_search_APT_word (ZServerInfo *zi,
951 Z_AttributesPlusTerm *zapt,
952 oid_value attributeSet,
953 int num_bases, char **basenames)
955 rset_isam_parms parms;
956 char termz[IT_MAX_WORD+1];
957 struct grep_info grep_info;
960 if (zapt->term->which != Z_Term_general)
965 trans_term (zi, zapt, termz);
967 grep_info.isam_p_indx = 0;
968 grep_info.isam_p_size = 0;
969 grep_info.isam_p_buf = NULL;
971 if (field_term (zi, zapt, termz, attributeSet, &grep_info,
972 num_bases, basenames))
974 if (grep_info.isam_p_indx < 1)
975 result = rset_create (rset_kind_null, NULL);
976 else if (grep_info.isam_p_indx == 1)
978 parms.is = zi->wordIsam;
979 parms.pos = *grep_info.isam_p_buf;
980 result = rset_create (rset_kind_isam, &parms);
983 result = rset_trunc (zi->wordIsam, grep_info.isam_p_buf,
984 grep_info.isam_p_indx);
985 xfree (grep_info.isam_p_buf);
989 static RSET rpn_prox (RSET *rset, int rset_no)
997 rset_temp_parms parms;
999 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
1000 more = xmalloc (sizeof(*more)*rset_no);
1001 buf = xmalloc (sizeof(*buf)*rset_no);
1003 for (i = 0; i<rset_no; i++)
1005 buf[i] = xmalloc (sizeof(**buf));
1006 rsfd[i] = rset_open (rset[i], RSETF_READ|RSETF_SORT_SYSNO);
1007 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i])))
1011 rset_close (rset[i], rsfd[i]);
1018 return rset_create (rset_kind_null, NULL);
1021 parms.key_size = sizeof (struct it_key);
1022 result = rset_create (rset_kind_temp, &parms);
1023 rsfd_result = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1027 for (i = 1; i<rset_no; i++)
1036 cmp = key_compare (buf[i], buf[i-1]);
1039 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
1044 if (buf[i-1]->seqno+1 != buf[i]->seqno)
1046 more[i-1] = rset_read (rset[i-1], rsfd[i-1], buf[i-1]);
1052 more[i] = rset_read (rset[i], rsfd[i], buf[i]);
1058 rset_write (result, rsfd_result, buf[0]);
1059 more[0] = rset_read (*rset, *rsfd, *buf);
1063 for (i = 0; i<rset_no; i++)
1065 rset_close (rset[i], rsfd[i]);
1068 rset_close (result, rsfd_result);
1075 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
1076 Z_AttributesPlusTerm *zapt,
1077 oid_value attributeSet,
1078 int num_bases, char **basenames)
1080 char termz[IT_MAX_WORD+1];
1081 char term_sub[IT_MAX_WORD+1];
1082 char *p0 = termz, *p1 = NULL;
1083 RSET rset[60], result;
1085 struct grep_info grep_info;
1087 if (zapt->term->which != Z_Term_general)
1092 trans_term (zi, zapt, termz);
1094 grep_info.isam_p_size = 0;
1095 grep_info.isam_p_buf = NULL;
1099 if ((p1 = strchr (p0, ' ')))
1101 memcpy (term_sub, p0, p1-p0);
1102 term_sub[p1-p0] = '\0';
1105 strcpy (term_sub, p0);
1107 grep_info.isam_p_indx = 0;
1108 if (field_term (zi, zapt, term_sub, attributeSet, &grep_info,
1109 num_bases, basenames))
1111 if (grep_info.isam_p_indx == 0)
1112 rset[rset_no] = rset_create (rset_kind_null, NULL);
1113 else if (grep_info.isam_p_indx > 1)
1114 rset[rset_no] = rset_trunc (zi->wordIsam,
1115 grep_info.isam_p_buf,
1116 grep_info.isam_p_indx);
1119 rset_isam_parms parms;
1121 parms.is = zi->wordIsam;
1122 parms.pos = *grep_info.isam_p_buf;
1123 rset[rset_no] = rset_create (rset_kind_isam, &parms);
1125 assert (rset[rset_no]);
1126 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1131 while (*++p0 == ' ')
1134 xfree (grep_info.isam_p_buf);
1136 return rset_create (rset_kind_null, NULL);
1137 else if (rset_no == 1)
1140 result = rpn_prox (rset, rset_no);
1141 for (i = 0; i<rset_no; i++)
1142 rset_delete (rset[i]);
1146 static RSET rpn_search_APT_local (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1147 oid_value attributeSet)
1152 rset_temp_parms parms;
1153 char termz[IT_MAX_WORD+1];
1155 if (zapt->term->which != Z_Term_general)
1160 parms.key_size = sizeof (struct it_key);
1161 result = rset_create (rset_kind_temp, &parms);
1162 rsfd = rset_open (result, RSETF_WRITE|RSETF_SORT_SYSNO);
1164 trans_term (zi, zapt, termz);
1165 key.sysno = atoi (termz);
1168 rset_write (result, rsfd, &key);
1169 rset_close (result, rsfd);
1173 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1174 oid_value attributeSet,
1175 int num_bases, char **basenames)
1179 int relation_value, structure_value;
1181 attr_init (&relation, zapt, 2);
1182 attr_init (&structure, zapt, 4);
1184 relation_value = attr_find (&relation, NULL);
1185 structure_value = attr_find (&structure, NULL);
1186 switch (structure_value)
1189 if (relation_value == 102) /* relevance relation */
1190 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1191 num_bases, basenames);
1192 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1193 num_bases, basenames);
1194 case 1: /* phrase */
1195 if (relation_value == 102) /* relevance relation */
1196 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1197 num_bases, basenames);
1198 return rpn_search_APT_phrase (zi, zapt, attributeSet,
1199 num_bases, basenames);
1202 if (relation_value == 102) /* relevance relation */
1203 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1204 num_bases, basenames);
1205 return rpn_search_APT_word (zi, zapt, attributeSet,
1206 num_bases, basenames);
1211 case 5: /* date - normalized */
1213 case 6: /* word list */
1214 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1215 num_bases, basenames);
1216 case 100: /* date - un-normalized */
1218 case 101: /* name - normalized */
1220 case 102: /* date - un-normalized */
1222 case 103: /* structure */
1226 case 105: /* free-form-text */
1227 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1228 num_bases, basenames);
1229 case 106: /* document-text */
1230 return rpn_search_APT_relevance (zi, zapt, attributeSet,
1231 num_bases, basenames);
1232 case 107: /* local-number */
1233 return rpn_search_APT_local (zi, zapt, attributeSet);
1234 case 108: /* string */
1235 return rpn_search_APT_word (zi, zapt, attributeSet,
1236 num_bases, basenames);
1237 case 109: /* numeric string */
1244 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
1248 if (!(s = resultSetGet (zi, resultSetId)))
1249 return rset_create (rset_kind_null, NULL);
1253 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs,
1254 oid_value attributeSet,
1255 int num_bases, char **basenames)
1258 if (zs->which == Z_RPNStructure_complex)
1260 rset_bool_parms bool_parms;
1263 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1,
1265 num_bases, basenames);
1266 if (bool_parms.rset_l == NULL)
1268 if (rset_is_ranked(bool_parms.rset_l))
1270 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2,
1272 num_bases, basenames);
1273 if (bool_parms.rset_r == NULL)
1275 rset_delete (bool_parms.rset_l);
1278 if (rset_is_ranked(bool_parms.rset_r))
1280 bool_parms.key_size = sizeof(struct it_key);
1281 bool_parms.cmp = key_compare;
1283 switch (zs->u.complex->roperator->which)
1285 case Z_Operator_and:
1286 r = rset_create (soft ? rset_kind_sand:rset_kind_and, &bool_parms);
1289 r = rset_create (soft ? rset_kind_sor:rset_kind_or, &bool_parms);
1291 case Z_Operator_and_not:
1292 r = rset_create (soft ? rset_kind_snot:rset_kind_not, &bool_parms);
1298 else if (zs->which == Z_RPNStructure_simple)
1300 if (zs->u.simple->which == Z_Operand_APT)
1302 logf (LOG_DEBUG, "rpn_search_APT");
1303 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm,
1304 attributeSet, num_bases, basenames);
1306 else if (zs->u.simple->which == Z_Operand_resultSetId)
1308 logf (LOG_DEBUG, "rpn_search_ref");
1309 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
1323 void count_set_save (RSET *r, int *count)
1330 rset_temp_parms parms;
1332 logf (LOG_DEBUG, "count_set_save");
1334 parms.key_size = sizeof(struct it_key);
1335 w = rset_create (rset_kind_temp, &parms);
1336 wfd = rset_open (w, RSETF_WRITE|RSETF_SORT_SYSNO);
1337 rfd = rset_open (*r, RSETF_READ|RSETF_SORT_SYSNO);
1338 while (rset_read (*r, rfd, &key))
1340 if (key.sysno != psysno)
1342 rset_write (w, wfd, &key);
1348 rset_close (*r, rfd);
1350 rset_close (w, wfd);
1352 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1355 static void count_set (RSET r, int *count)
1362 logf (LOG_DEBUG, "count_set");
1364 rfd = rset_open (r, RSETF_READ|RSETF_SORT_SYSNO);
1365 while (rset_read (r, rfd, &key))
1367 if (key.sysno != psysno)
1374 rset_close (r, rfd);
1375 logf (LOG_DEBUG, "%d keys, %d distinct sysnos", kno, *count);
1378 int rpn_search (ZServerInfo *zi,
1379 Z_RPNQuery *rpn, int num_bases, char **basenames,
1380 const char *setname, int *hits)
1384 oid_value attributeSet;
1389 zi->errString = NULL;
1391 attrset = oid_getentbyoid (rpn->attributeSetId);
1392 attributeSet = attrset->value;
1393 rset = rpn_search_structure (zi, rpn->RPNStructure, attributeSet,
1394 num_bases, basenames);
1397 if (rset_is_volatile(rset))
1398 count_set_save(&rset,hits);
1400 count_set (rset, hits);
1401 resultSetAdd (zi, setname, 1, rset);
1403 logf (LOG_DEBUG, "search error: %d", zi->errCode);
1408 struct scan_entry *list;
1415 static int scan_handle (char *name, const char *info, int pos, void *client)
1417 int len_prefix, idx;
1420 struct scan_info *scan_info = client;
1422 rset_isam_parms parms;
1424 len_prefix = strlen(scan_info->prefix);
1425 if (memcmp (name, scan_info->prefix, len_prefix))
1428 idx = scan_info->after - pos + scan_info->before;
1431 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1432 strlen(name + len_prefix)+1);
1433 strcpy (scan_info->list[idx].term, name + len_prefix);
1434 assert (*info == sizeof(isam_p));
1435 memcpy (&isam_p, info+1, sizeof(isam_p));
1436 parms.is = scan_info->isam;
1439 rset = rset_create (rset_kind_isam, &parms);
1440 count_set (rset, &scan_info->list[idx].occurrences);
1443 scan_info->list[idx].occurrences = 1;
1445 logf (LOG_DEBUG, "pos=%3d idx=%3d name=%s", pos, idx, name);
1450 static int dummy_handle (char *name, const char *info, void *p)
1455 int rpn_scan (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
1456 int num_bases, char **basenames,
1457 int *position, int *num_entries, struct scan_entry **list,
1460 int i, j, sizez, max_pos;
1461 int pos = *position;
1462 int num = *num_entries;
1465 char termz[IT_MAX_WORD+20];
1468 Z_Term *term = zapt->term;
1469 struct scan_info scan_info;
1471 logf (LOG_DEBUG, "scan, position = %d, num = %d", pos, num);
1475 scan_info.before = before = pos-1;
1476 scan_info.after = after = 1+num-pos;
1477 scan_info.odr = zi->odr;
1479 logf (LOG_DEBUG, "scan, before = %d, after = %d", before, after);
1481 scan_info.isam = zi->wordIsam;
1482 scan_info.list = odr_malloc (zi->odr, (before+after)*
1483 sizeof(*scan_info.list));
1484 for (j = 0; j<before+after; j++)
1485 scan_info.list[j].term = NULL;
1486 attr_init (&use, zapt, 1);
1487 use_value = attr_find (&use, NULL);
1488 logf (LOG_DEBUG, "use value %d", use_value);
1490 if (use_value == -1)
1492 i = index_word_prefix (termz, 1, use_value, *basenames);
1494 dict_lookup_grep (zi->wordDict, termz, 0, NULL, &max_pos, 0,
1496 if (max_pos <= strlen(*basenames))
1498 zi->errString = *basenames;
1499 return zi->errCode = 109; /* Database unavailable */
1501 strcpy (scan_info.prefix, termz);
1502 sizez = term->u.general->len;
1503 if (sizez > IT_MAX_WORD)
1504 sizez = IT_MAX_WORD;
1505 for (j = 0; j<sizez; j++)
1506 termz[j+i] = index_char_cvt (term->u.general->buf[j]);
1509 dict_scan (zi->wordDict, termz, &before, &after, &scan_info, scan_handle);
1511 *status = BEND_SCAN_SUCCESS;
1513 for (i = 0; i<scan_info.after; i++)
1514 if (scan_info.list[scan_info.before+scan_info.after-i-1].term)
1518 *status = BEND_SCAN_PARTIAL;
1520 for (i = 0; i<scan_info.before; i++)
1521 if (scan_info.list[i].term)
1524 *status = BEND_SCAN_PARTIAL;
1528 *list = scan_info.list+i; /* list is set to first 'real' entry */
1530 if (*num_entries == 0) /* signal 'unsupported use-attribute' */
1531 zi->errCode = 114; /* if no entries was found */
1532 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1533 *position, *num_entries);
1535 logf (LOG_DEBUG, "scan error: %d", zi->errCode);