2 * Copyright (C) 1995-1998, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.79 1998-06-22 11:35:09 adam
10 * Revision 1.78 1998/06/08 14:43:17 adam
11 * Added suport for EXPLAIN Proxy servers - added settings databasePath
12 * and explainDatabase to facilitate this. Increased maximum number
13 * of databases and attributes in one register.
15 * Revision 1.77 1998/05/20 10:12:22 adam
16 * Implemented automatic EXPLAIN database maintenance.
17 * Modified Zebra to work with ASN.1 compiled version of YAZ.
19 * Revision 1.76 1998/04/02 14:35:29 adam
20 * First version of Zebra that works with compiled ASN.1.
22 * Revision 1.75 1998/03/05 08:45:13 adam
23 * New result set model and modular ranking system. Moved towards
24 * descent server API. System information stored as "SGML" records.
26 * Revision 1.74 1998/02/10 12:03:06 adam
29 * Revision 1.73 1998/01/29 13:40:11 adam
30 * Better logging for scan service.
32 * Revision 1.72 1998/01/07 13:53:41 adam
33 * Queries using simple ranked operands returns right number of hits.
35 * Revision 1.71 1997/12/18 10:54:24 adam
36 * New method result set method rs_hits that returns the number of
37 * hits in result-set (if known). The ranked result set returns real
38 * number of hits but only when not combined with other operands.
40 * Revision 1.70 1997/10/31 12:34:43 adam
41 * Changed a few log statements.
43 * Revision 1.69 1997/10/29 12:05:02 adam
44 * Server produces diagnostic "Unsupported Attribute Set" when appropriate.
46 * Revision 1.68 1997/10/27 14:33:06 adam
47 * Moved towards generic character mapping depending on "structure"
48 * field in abstract syntax file. Fixed a few memory leaks. Fixed
49 * bug with negative integers when doing searches with relational
52 * Revision 1.67 1997/09/29 09:06:10 adam
53 * Removed one static var in order to make this module thread safe.
55 * Revision 1.66 1997/09/25 14:58:03 adam
58 * Revision 1.65 1997/09/22 12:39:06 adam
59 * Added get_pos method for the ranked result sets.
61 * Revision 1.64 1997/09/18 08:59:20 adam
62 * Extra generic handle for the character mapping routines.
64 * Revision 1.63 1997/09/17 12:19:18 adam
65 * Zebra version corresponds to YAZ version 1.4.
66 * Changed Zebra server so that it doesn't depend on global common_resource.
68 * Revision 1.62 1997/09/05 15:30:09 adam
69 * Changed prototype for chr_map_input - added const.
70 * Added support for C++, headers uses extern "C" for public definitions.
72 * Revision 1.61 1997/02/10 10:21:14 adam
73 * Bug fix: in search terms character (^) wasn't observed.
75 * Revision 1.60 1997/01/31 11:10:34 adam
76 * Bug fix: Leading and trailing white space weren't removed in scan tokens.
78 * Revision 1.59 1997/01/17 11:31:46 adam
79 * Bug fix: complete phrase search didn't work.
81 * Revision 1.58 1996/12/23 15:30:45 adam
83 * Bug fix: result sets weren't deleted after server shut down.
85 * Revision 1.57 1996/11/11 13:38:02 adam
86 * Added proximity support in search.
88 * Revision 1.56 1996/11/08 11:10:32 adam
89 * Buffers used during file match got bigger.
90 * Compressed ISAM support everywhere.
91 * Bug fixes regarding masking characters in queries.
92 * Redesigned Regexp-2 queries.
94 * Revision 1.55 1996/11/04 14:07:44 adam
95 * Moved truncation code to trunc.c.
97 * Revision 1.54 1996/10/29 14:09:52 adam
98 * Use of cisam system - enabled if setting isamc is 1.
100 * Revision 1.53 1996/06/26 09:21:43 adam
101 * Bug fix: local attribute set wasn't obeyed in scan.
103 * Revision 1.52 1996/06/17 14:26:20 adam
104 * Function gen_regular_rel changed to handle negative numbers.
106 * Revision 1.51 1996/06/11 10:54:15 quinn
109 * Revision 1.50 1996/06/07 08:51:53 adam
110 * Bug fix: Character mapping was broken (introducued by last revision).
112 * Revision 1.49 1996/06/04 10:18:11 adam
113 * Search/scan uses character mapping module.
115 * Revision 1.48 1996/05/28 15:15:01 adam
116 * Bug fix: Didn't handle unknown database correctly.
118 * Revision 1.47 1996/05/15 18:36:28 adam
119 * Function trans_term transforms unsearchable characters to blanks.
121 * Revision 1.46 1996/05/15 11:57:56 adam
122 * Fixed bug introduced by set/field mapping in search operations.
124 * Revision 1.45 1996/05/14 11:34:00 adam
125 * Scan support in multiple registers/databases.
127 * Revision 1.44 1996/05/14 06:16:44 adam
128 * Compact use/set bytes used in search service.
130 * Revision 1.43 1996/05/09 09:54:43 adam
131 * Server supports maps from one logical attributes to a list of physical
133 * The extraction process doesn't make space consuming 'any' keys.
135 * Revision 1.42 1996/05/09 07:28:56 quinn
136 * Work towards phrases and multiple registers
138 * Revision 1.41 1996/03/20 09:36:43 adam
139 * Function dict_lookup_grep got extra parameter, init_pos, which marks
140 * from which position in pattern approximate pattern matching should occur.
141 * Approximate pattern matching is used in relevance=re-2.
143 * Revision 1.40 1996/02/02 13:44:44 adam
144 * The public dictionary functions simply use char instead of Dict_char
145 * to represent search strings. Dict_char is used internally only.
147 * Revision 1.39 1996/01/03 16:22:13 quinn
148 * operator->roperator
150 * Revision 1.38 1995/12/11 09:12:55 adam
151 * The rec_get function returns NULL if record doesn't exist - will
152 * happen in the server if the result set records have been deleted since
153 * the creation of the set (i.e. the search).
154 * The server saves a result temporarily if it is 'volatile', i.e. the
155 * set is register dependent.
157 * Revision 1.37 1995/12/06 15:05:28 adam
158 * More verbose in count_set.
160 * Revision 1.36 1995/12/06 12:41:27 adam
161 * New command 'stat' for the index program.
162 * Filenames can be read from stdin by specifying '-'.
163 * Bug fix/enhancement of the transformation from terms to regular
164 * expressons in the search engine.
166 * Revision 1.35 1995/11/27 09:29:00 adam
167 * Bug fixes regarding conversion to regular expressions.
169 * Revision 1.34 1995/11/16 17:00:56 adam
170 * Better logging of rpn query.
172 * Revision 1.33 1995/11/01 13:58:28 quinn
173 * Moving data1 to yaz/retrieval
175 * Revision 1.32 1995/10/27 14:00:11 adam
176 * Implemented detection of database availability.
178 * Revision 1.31 1995/10/17 18:02:10 adam
179 * New feature: databases. Implemented as prefix to words in dictionary.
181 * Revision 1.30 1995/10/16 09:32:38 adam
182 * More work on relational op.
184 * Revision 1.29 1995/10/13 16:01:49 adam
187 * Revision 1.28 1995/10/13 12:26:43 adam
188 * Optimization of truncation.
190 * Revision 1.27 1995/10/12 17:07:22 adam
193 * Revision 1.26 1995/10/12 12:40:54 adam
194 * Bug fixes in rpn_prox.
196 * Revision 1.25 1995/10/10 13:59:24 adam
197 * Function rset_open changed its wflag parameter to general flags.
199 * Revision 1.24 1995/10/09 16:18:37 adam
200 * Function dict_lookup_grep got extra client data parameter.
202 * Revision 1.23 1995/10/06 16:33:37 adam
203 * Use attribute mappings.
205 * Revision 1.22 1995/10/06 15:07:39 adam
206 * Structure 'local-number' handled.
208 * Revision 1.21 1995/10/06 13:52:06 adam
209 * Bug fixes. Handler may abort further scanning.
211 * Revision 1.20 1995/10/06 11:06:33 adam
212 * Scan entries include 'occurrences' now.
214 * Revision 1.19 1995/10/06 10:43:56 adam
215 * Scan added. 'occurrences' in scan entries not set yet.
217 * Revision 1.18 1995/10/04 16:57:20 adam
218 * Key input and merge sort in one pass.
220 * Revision 1.17 1995/10/04 12:55:17 adam
221 * Bug fix in ranked search. Use=Any keys inserted.
223 * Revision 1.16 1995/10/02 16:24:40 adam
224 * Use attribute actually used in search requests.
226 * Revision 1.15 1995/10/02 15:18:52 adam
227 * New member in recRetrieveCtrl: diagnostic.
229 * Revision 1.14 1995/09/28 12:10:32 adam
230 * Bug fixes. Field prefix used in queries.
232 * Revision 1.13 1995/09/18 14:17:50 adam
235 * Revision 1.12 1995/09/15 14:45:21 adam
237 * Work on truncation.
239 * Revision 1.11 1995/09/14 11:53:27 adam
240 * First work on regular expressions/truncations.
242 * Revision 1.10 1995/09/11 15:23:26 adam
243 * More work on relevance search.
245 * Revision 1.9 1995/09/11 13:09:35 adam
246 * More work on relevance feedback.
248 * Revision 1.8 1995/09/08 14:52:27 adam
249 * Minor changes. Dictionary is lower case now.
251 * Revision 1.7 1995/09/07 13:58:36 adam
252 * New parameter: result-set file descriptor (RSFD) to support multiple
253 * positions within the same result-set.
254 * Boolean operators: and, or, not implemented.
255 * Result-set references.
257 * Revision 1.6 1995/09/06 16:11:18 adam
258 * Option: only one word key per file.
260 * Revision 1.5 1995/09/06 10:33:04 adam
261 * More work on present. Some log messages removed.
263 * Revision 1.4 1995/09/05 15:28:40 adam
264 * More work on search engine.
266 * Revision 1.3 1995/09/04 15:20:22 adam
269 * Revision 1.2 1995/09/04 12:33:43 adam
270 * Various cleanup. YAZ util used instead.
272 * Revision 1.1 1995/09/04 09:10:40 adam
273 * More work on index add/del/update.
274 * Merge sort implemented.
275 * Initial work on z39 server.
294 struct rpn_char_map_info {
299 static const char **rpn_char_map_handler (void *vp, const char **from, int len)
301 struct rpn_char_map_info *p = vp;
302 return zebra_maps_input (p->zm, p->reg_type, from, len);
305 static void rpn_char_map_prepare (ZebraHandle zh, int reg_type,
306 struct rpn_char_map_info *map_info)
308 map_info->zm = zh->zebra_maps;
309 map_info->reg_type = reg_type;
310 dict_grep_cmap (zh->dict, map_info, rpn_char_map_handler);
317 Z_AttributesPlusTerm *zapt;
320 static int attr_find (AttrType *src, oid_value *attributeSetP)
325 num_attributes = src->zapt->attributes->num_attributes;
327 num_attributes = src->zapt->num_attributes;
329 while (src->major < num_attributes)
331 Z_AttributeElement *element;
334 element = src->zapt->attributes->attributes[src->major];
336 element = src->zapt->attributeList[src->major];
338 if (src->type == *element->attributeType)
340 switch (element->which)
342 case Z_AttributeValue_numeric:
344 if (element->attributeSet && attributeSetP)
348 attrset = oid_getentbyoid (element->attributeSet);
349 *attributeSetP = attrset->value;
351 return *element->value.numeric;
353 case Z_AttributeValue_complex:
354 if (src->minor >= element->value.complex->num_list ||
355 element->value.complex->list[src->minor]->which !=
356 Z_StringOrNumeric_numeric)
359 if (element->attributeSet && attributeSetP)
363 attrset = oid_getentbyoid (element->attributeSet);
364 *attributeSetP = attrset->value;
366 return *element->value.complex->list[src->minor-1]->u.numeric;
376 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
398 static void term_untrans (ZebraHandle zh, int reg_type,
399 char *dst, const char *src)
403 const char *cp = zebra_maps_output (zh->zebra_maps, reg_type, &src);
410 static void add_isam_p (const char *name, const char *info,
414 if (p->isam_p_indx == p->isam_p_size)
416 ISAM_P *new_isam_p_buf;
420 p->isam_p_size = 2*p->isam_p_size + 100;
421 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
425 memcpy (new_isam_p_buf, p->isam_p_buf,
426 p->isam_p_indx * sizeof(*p->isam_p_buf));
427 xfree (p->isam_p_buf);
429 p->isam_p_buf = new_isam_p_buf;
432 new_term_no = xmalloc (sizeof(*new_term_no) *
436 memcpy (new_term_no, p->isam_p_buf,
437 p->isam_p_indx * sizeof(*p->term_no));
440 p->term_no = new_term_no;
443 assert (*info == sizeof(*p->isam_p_buf));
444 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
447 term_untrans (p->zh, p->reg_type, term_tmp, name+2);
448 logf (LOG_DEBUG, "grep: %s", term_tmp);
453 static int grep_handle (char *name, const char *info, void *p)
455 add_isam_p (name, info, p);
459 static int term_pre (ZebraMaps zebra_maps, int reg_type, const char **src,
460 const char *ct1, const char *ct2)
462 const char *s1, *s0 = *src;
465 /* skip white space */
468 if (ct1 && strchr (ct1, *s0))
470 if (ct2 && strchr (ct2, *s0))
473 map = zebra_maps_input (zebra_maps, reg_type, &s1, strlen(s1));
474 if (**map != *CHR_SPACE)
482 static int term_100 (ZebraMaps zebra_maps, int reg_type,
483 const char **src, char *dst, int space_split,
491 if (!term_pre (zebra_maps, reg_type, src, NULL, NULL))
497 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
498 if (space_split && **map == *CHR_SPACE)
502 if (!isalnum (*s1) && *s1 != '-')
514 static int term_101 (ZebraMaps zebra_maps, int reg_type,
515 const char **src, char *dst, int space_split,
523 if (!term_pre (zebra_maps, reg_type, src, "#", "#"))
532 dst_term[j++] = *s0++;
537 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
538 if (space_split && **map == *CHR_SPACE)
550 dst_term[j++] = '\0';
556 static int term_103 (ZebraMaps zebra_maps, int reg_type, const char **src,
557 char *dst, int *errors, int space_split,
565 if (!term_pre (zebra_maps, reg_type, src, "^\\()[].*+?|", "("))
568 if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] &&
571 *errors = s0[1] - '0';
578 if (strchr ("^\\()[].*+?|-", *s0))
586 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
587 if (**map == *CHR_SPACE)
604 static int term_102 (ZebraMaps zebra_maps, int reg_type, const char **src,
605 char *dst, int space_split, char *dst_term)
607 return term_103 (zebra_maps, reg_type, src, dst, NULL, space_split,
611 /* gen_regular_rel - generate regular expression from relation
612 * val: border value (inclusive)
613 * islt: 1 if <=; 0 if >=.
615 static void gen_regular_rel (char *dst, int val, int islt)
622 logf (LOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt);
626 strcpy (dst, "(-[0-9]+|(");
634 strcpy (dst, "([0-9]+|-(");
635 dst_p = strlen (dst);
645 dst_p = strlen (dst);
646 sprintf (numstr, "%d", val);
647 for (w = strlen(numstr); --w >= 0; pos++)
666 strcpy (dst + dst_p, numstr);
667 dst_p = strlen(dst) - pos - 1;
695 for (i = 0; i<pos; i++)
708 for (i=1; i<pos; i++)
709 strcat (dst, "[0-9]?");
713 for (i = 0; i <= pos; i++)
714 strcat (dst, "[0-9]");
715 strcat (dst, "[0-9]*");
720 static int relational_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
721 const char **term_sub,
723 oid_value attributeSet,
724 struct grep_info *grep_info,
733 char *term_tmp = term_dict + strlen(term_dict);
735 attr_init (&relation, zapt, 2);
736 relation_value = attr_find (&relation, NULL);
738 logf (LOG_DEBUG, "relation value=%d", relation_value);
739 switch (relation_value)
742 if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
745 term_value = atoi (term_tmp);
746 logf (LOG_DEBUG, "Relation <");
747 gen_regular_rel (term_tmp, term_value-1, 1);
750 if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
753 term_value = atoi (term_tmp);
754 logf (LOG_DEBUG, "Relation <=");
755 gen_regular_rel (term_tmp, term_value, 1);
758 if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
761 term_value = atoi (term_tmp);
762 logf (LOG_DEBUG, "Relation >=");
763 gen_regular_rel (term_tmp, term_value, 0);
766 if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
769 term_value = atoi (term_tmp);
770 logf (LOG_DEBUG, "Relation >");
771 gen_regular_rel (term_tmp, term_value+1, 0);
776 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_tmp);
777 r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info, max_pos,
780 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
781 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
785 static int field_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
786 const char **term_sub,
787 oid_value attributeSet, struct grep_info *grep_info,
788 int reg_type, int complete_flag,
789 int num_bases, char **basenames,
792 char term_dict[2*IT_MAX_WORD+2];
795 int truncation_value;
798 oid_value curAttributeSet = attributeSet;
800 struct rpn_char_map_info rcmi;
801 int space_split = complete_flag ? 0 : 1;
803 rpn_char_map_prepare (zh, reg_type, &rcmi);
804 attr_init (&use, zapt, 1);
805 use_value = attr_find (&use, &curAttributeSet);
806 logf (LOG_DEBUG, "field_term, use value %d", use_value);
807 attr_init (&truncation, zapt, 5);
808 truncation_value = attr_find (&truncation, NULL);
809 logf (LOG_DEBUG, "truncation value %d", truncation_value);
814 for (base_no = 0; base_no < num_bases; base_no++)
817 data1_local_attribute *local_attr;
818 int max_pos, prefix_len = 0;
821 if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value)))
823 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
824 curAttributeSet, use_value, r);
831 if (zebraExplain_curDatabase (zh->zei, basenames[base_no]))
833 zh->errCode = 109; /* Database unavailable */
834 zh->errString = basenames[base_no];
837 for (local_attr = attp.local_attributes; local_attr;
838 local_attr = local_attr->next)
844 ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal,
849 term_dict[prefix_len++] = '|';
851 term_dict[prefix_len++] = '(';
853 ord_len = key_SU_code (ord, ord_buf);
854 for (i = 0; i<ord_len; i++)
856 term_dict[prefix_len++] = 1;
857 term_dict[prefix_len++] = ord_buf[i];
865 term_dict[prefix_len++] = ')';
866 term_dict[prefix_len++] = 1;
867 term_dict[prefix_len++] = reg_type;
868 logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
869 term_dict[prefix_len] = '\0';
870 if (!relational_term (zh, zapt, &termp, term_dict,
871 attributeSet, grep_info, &max_pos, reg_type,
875 switch (truncation_value)
877 case -1: /* not specified */
878 case 100: /* do not truncate */
879 term_dict[j++] = '(';
880 if (!term_100 (zh->zebra_maps, reg_type,
881 &termp, term_dict + j, space_split, term_dst))
883 strcat (term_dict, ")");
884 r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
885 &max_pos, 0, grep_handle);
887 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
889 case 1: /* right truncation */
890 term_dict[j++] = '(';
891 if (!term_100 (zh->zebra_maps, reg_type,
892 &termp, term_dict + j, space_split, term_dst))
894 strcat (term_dict, ".*)");
895 dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
896 &max_pos, 0, grep_handle);
898 case 2: /* left truncation */
899 case 3: /* left&right truncation */
902 case 101: /* process # in term */
903 term_dict[j++] = '(';
904 if (!term_101 (zh->zebra_maps, reg_type,
905 &termp, term_dict + j, space_split, term_dst))
907 strcat (term_dict, ")");
908 r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
909 &max_pos, 0, grep_handle);
911 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", r);
913 case 102: /* Regexp-1 */
914 term_dict[j++] = '(';
915 if (!term_102 (zh->zebra_maps, reg_type,
916 &termp, term_dict + j, space_split, term_dst))
918 strcat (term_dict, ")");
919 logf (LOG_DEBUG, "Regexp-1 tolerance=%d", r);
920 r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
921 &max_pos, 0, grep_handle);
923 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
926 case 103: /* Regexp-2 */
928 term_dict[j++] = '(';
929 if (!term_103 (zh->zebra_maps, reg_type,
930 &termp, term_dict + j, &r, space_split, term_dst))
932 strcat (term_dict, ")");
933 logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r);
934 r = dict_lookup_grep (zh->dict, term_dict, r, grep_info,
935 &max_pos, 2, grep_handle);
937 logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
944 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
948 static void trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
952 Z_Term *term = zapt->term;
954 sizez = term->u.general->len;
955 if (sizez > IT_MAX_WORD-1)
956 sizez = IT_MAX_WORD-1;
957 memcpy (termz, term->u.general->buf, sizez);
961 static void trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
962 char *termz, int reg_type)
964 Z_Term *term = zapt->term;
966 const char *cp = (const char *) term->u.general->buf;
967 const char *cp_end = cp + term->u.general->len;
970 const char *space_map = NULL;
973 while ((len = (cp_end - cp)) > 0)
975 map = zebra_maps_input (zh->zebra_maps, reg_type, &cp, len);
976 if (**map == *CHR_SPACE)
981 for (src = space_map; *src; src++)
984 for (src = *map; *src; src++)
991 static RSET rpn_proximity (ZebraHandle zh, RSET rset1, RSET rset2,
993 int exclusion, int relation, int distance)
998 struct it_key buf1, buf2;
1001 rset_temp_parms parms;
1004 rsfd1 = rset_open (rset1, RSETF_READ);
1005 more1 = rset_read (rset1, rsfd1, &buf1, &term_index);
1007 rsfd2 = rset_open (rset2, RSETF_READ);
1008 more2 = rset_read (rset2, rsfd2, &buf2, &term_index);
1010 parms.key_size = sizeof (struct it_key);
1011 parms.temp_path = res_get (zh->res, "setTmpDir");
1012 result = rset_create (rset_kind_temp, &parms);
1013 rsfd_result = rset_open (result, RSETF_WRITE);
1015 logf (LOG_DEBUG, "rpn_proximity excl=%d ord=%d rel=%d dis=%d",
1016 exclusion, ordered, relation, distance);
1017 while (more1 && more2)
1019 int cmp = key_compare_it (&buf1, &buf2);
1021 more1 = rset_read (rset1, rsfd1, &buf1, &term_index);
1023 more2 = rset_read (rset2, rsfd2, &buf2, &term_index);
1026 int sysno = buf1.sysno;
1030 seqno[n++] = buf1.seqno;
1031 while ((more1 = rset_read (rset1, rsfd1, &buf1, &term_index)) &&
1032 sysno == buf1.sysno)
1034 seqno[n++] = buf1.seqno;
1037 for (i = 0; i<n; i++)
1039 int diff = buf2.seqno - seqno[i];
1040 int excl = exclusion;
1041 if (!ordered && diff < 0)
1046 if (diff < distance)
1050 if (diff <= distance)
1054 if (diff == distance)
1058 if (diff >= distance)
1062 if (diff > distance)
1066 if (diff != distance)
1071 rset_write (result, rsfd_result, &buf2);
1073 } while ((more2 = rset_read (rset2, rsfd2, &buf2, &term_index)) &&
1074 sysno == buf2.sysno);
1077 rset_close (result, rsfd_result);
1078 rset_close (rset1, rsfd1);
1079 rset_close (rset2, rsfd2);
1083 static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no)
1088 struct it_key **buf;
1090 char prox_term[1024];
1091 int length_prox_term = 0;
1092 int min_nn = 10000000;
1094 const char *flags = NULL;
1096 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
1097 more = xmalloc (sizeof(*more)*rset_no);
1098 buf = xmalloc (sizeof(*buf)*rset_no);
1100 for (i = 0; i<rset_no; i++)
1103 buf[i] = xmalloc (sizeof(**buf));
1104 rsfd[i] = rset_open (rset[i], RSETF_READ);
1105 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i], &term_index)))
1107 for (j = 0; j<rset[i]->no_rset_terms; j++)
1109 const char *nflags = rset[i]->rset_terms[j]->flags;
1110 char *term = rset[i]->rset_terms[j]->name;
1111 int lterm = strlen(term);
1112 if (length_prox_term)
1113 prox_term[length_prox_term++] = ' ';
1114 strcpy (prox_term + length_prox_term, term);
1115 length_prox_term += lterm;
1116 if (min_nn > rset[i]->rset_terms[j]->nn)
1117 min_nn = rset[i]->rset_terms[j]->nn;
1123 rset_null_parms parms;
1127 rset_close (rset[i], rsfd[i]);
1131 parms.rset_term = rset_term_create (prox_term, -1, flags);
1132 parms.rset_term->nn = 0;
1133 result = rset_create (rset_kind_null, &parms);
1137 rset_temp_parms parms;
1140 parms.rset_term = rset_term_create (prox_term, -1, flags);
1141 parms.rset_term->nn = min_nn;
1142 parms.key_size = sizeof (struct it_key);
1143 parms.temp_path = res_get (zh->res, "setTmpDir");
1144 result = rset_create (rset_kind_temp, &parms);
1145 rsfd_result = rset_open (result, RSETF_WRITE);
1149 for (i = 1; i<rset_no; i++)
1158 cmp = key_compare_it (buf[i], buf[i-1]);
1161 more[i-1] = rset_read (rset[i-1], rsfd[i-1],
1162 buf[i-1], &term_index);
1167 if (buf[i-1]->seqno+1 != buf[i]->seqno)
1169 more[i-1] = rset_read (rset[i-1], rsfd[i-1],
1170 buf[i-1], &term_index);
1176 more[i] = rset_read (rset[i], rsfd[i], buf[i],
1183 rset_write (result, rsfd_result, buf[0]);
1184 more[0] = rset_read (*rset, *rsfd, *buf, &term_index);
1188 for (i = 0; i<rset_no; i++)
1190 rset_close (rset[i], rsfd[i]);
1193 rset_close (result, rsfd_result);
1201 static RSET rpn_search_APT_phrase (ZebraHandle zh,
1202 Z_AttributesPlusTerm *zapt,
1204 oid_value attributeSet,
1205 int reg_type, int complete_flag,
1206 const char *rank_type,
1207 int num_bases, char **basenames)
1209 char term_dst[IT_MAX_WORD+1];
1210 const char *termp = termz;
1211 RSET rset[60], result;
1212 int i, r, rset_no = 0;
1213 struct grep_info grep_info;
1216 grep_info.term_no = 0;
1218 grep_info.isam_p_size = 0;
1219 grep_info.isam_p_buf = NULL;
1221 grep_info.reg_type = reg_type;
1225 logf (LOG_DEBUG, "APT_phrase termp=%s", termp);
1226 grep_info.isam_p_indx = 0;
1227 r = field_term (zh, zapt, &termp, attributeSet, &grep_info,
1228 reg_type, complete_flag, num_bases, basenames,
1232 logf (LOG_DEBUG, "term: %s", term_dst);
1233 rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
1234 grep_info.isam_p_indx, term_dst,
1235 strlen(term_dst), rank_type);
1236 assert (rset[rset_no]);
1237 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1241 xfree(grep_info.term_no);
1243 xfree (grep_info.isam_p_buf);
1246 rset_null_parms parms;
1248 parms.rset_term = rset_term_create (term_dst, -1, rank_type);
1249 return rset_create (rset_kind_null, &parms);
1251 else if (rset_no == 1)
1253 result = rpn_prox (zh, rset, rset_no);
1254 for (i = 0; i<rset_no; i++)
1255 rset_delete (rset[i]);
1259 static RSET rpn_search_APT_or_list (ZebraHandle zh,
1260 Z_AttributesPlusTerm *zapt,
1262 oid_value attributeSet,
1263 int reg_type, int complete_flag,
1264 const char *rank_type,
1265 int num_bases, char **basenames)
1267 char term_dst[IT_MAX_WORD+1];
1268 const char *termp = termz;
1269 RSET rset[60], result;
1270 int i, r, rset_no = 0;
1271 struct grep_info grep_info;
1274 grep_info.term_no = 0;
1276 grep_info.isam_p_size = 0;
1277 grep_info.isam_p_buf = NULL;
1279 grep_info.reg_type = reg_type;
1283 logf (LOG_DEBUG, "APT_or_list termp=%s", termp);
1284 grep_info.isam_p_indx = 0;
1285 r = field_term (zh, zapt, &termp, attributeSet, &grep_info,
1286 reg_type, complete_flag, num_bases, basenames,
1290 logf (LOG_DEBUG, "term: %s", term_dst);
1291 rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
1292 grep_info.isam_p_indx, term_dst,
1293 strlen(term_dst), rank_type);
1294 assert (rset[rset_no]);
1295 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1299 xfree(grep_info.term_no);
1301 xfree (grep_info.isam_p_buf);
1304 rset_null_parms parms;
1306 parms.rset_term = rset_term_create (term_dst, -1, rank_type);
1307 return rset_create (rset_kind_null, &parms);
1310 for (i = 1; i<rset_no; i++)
1312 rset_bool_parms bool_parms;
1314 bool_parms.rset_l = result;
1315 bool_parms.rset_r = rset[i];
1316 bool_parms.key_size = sizeof(struct it_key);
1317 bool_parms.cmp = key_compare_it;
1318 result = rset_create (rset_kind_or, &bool_parms);
1323 static RSET rpn_search_APT_and_list (ZebraHandle zh,
1324 Z_AttributesPlusTerm *zapt,
1326 oid_value attributeSet,
1327 int reg_type, int complete_flag,
1328 const char *rank_type,
1329 int num_bases, char **basenames)
1331 char term_dst[IT_MAX_WORD+1];
1332 const char *termp = termz;
1333 RSET rset[60], result;
1334 int i, r, rset_no = 0;
1335 struct grep_info grep_info;
1338 grep_info.term_no = 0;
1340 grep_info.isam_p_size = 0;
1341 grep_info.isam_p_buf = NULL;
1343 grep_info.reg_type = reg_type;
1347 logf (LOG_DEBUG, "APT_and_list termp=%s", termp);
1348 grep_info.isam_p_indx = 0;
1349 r = field_term (zh, zapt, &termp, attributeSet, &grep_info,
1350 reg_type, complete_flag, num_bases, basenames,
1354 logf (LOG_DEBUG, "term: %s", term_dst);
1355 rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
1356 grep_info.isam_p_indx, term_dst,
1357 strlen(term_dst), rank_type);
1358 assert (rset[rset_no]);
1359 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1363 xfree(grep_info.term_no);
1365 xfree (grep_info.isam_p_buf);
1368 rset_null_parms parms;
1370 parms.rset_term = rset_term_create (term_dst, -1, rank_type);
1371 return rset_create (rset_kind_null, &parms);
1374 for (i = 1; i<rset_no; i++)
1376 rset_bool_parms bool_parms;
1378 bool_parms.rset_l = result;
1379 bool_parms.rset_r = rset[i];
1380 bool_parms.key_size = sizeof(struct it_key);
1381 bool_parms.cmp = key_compare_it;
1382 result = rset_create (rset_kind_and, &bool_parms);
1388 static RSET rpn_search_APT_local (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1390 oid_value attributeSet,
1391 const char *rank_type)
1396 rset_temp_parms parms;
1398 parms.rset_term = rset_term_create (termz, -1, rank_type);
1399 parms.key_size = sizeof (struct it_key);
1400 parms.temp_path = res_get (zh->res, "setTmpDir");
1401 result = rset_create (rset_kind_temp, &parms);
1402 rsfd = rset_open (result, RSETF_WRITE);
1404 key.sysno = atoi (termz);
1408 rset_write (result, rsfd, &key);
1409 rset_close (result, rsfd);
1413 static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1414 oid_value attributeSet,
1415 int num_bases, char **basenames)
1418 char *search_type = NULL;
1419 char *rank_type = NULL;
1421 char termz[IT_MAX_WORD+1];
1423 zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type,
1424 &rank_type, &complete_flag);
1426 logf (LOG_DEBUG, "reg_id=%c", reg_id);
1427 logf (LOG_DEBUG, "complete_flag=%d", complete_flag);
1428 logf (LOG_DEBUG, "search_type=%s", search_type);
1429 logf (LOG_DEBUG, "rank_type=%s", rank_type);
1431 if (zapt->term->which != Z_Term_general)
1436 trans_term (zh, zapt, termz);
1438 if (!strcmp (search_type, "phrase"))
1440 return rpn_search_APT_phrase (zh, zapt, termz, attributeSet,
1441 reg_id, complete_flag, rank_type,
1442 num_bases, basenames);
1444 else if (!strcmp (search_type, "and-list"))
1446 return rpn_search_APT_and_list (zh, zapt, termz, attributeSet,
1447 reg_id, complete_flag, rank_type,
1448 num_bases, basenames);
1450 else if (!strcmp (search_type, "or-list"))
1452 return rpn_search_APT_or_list (zh, zapt, termz, attributeSet,
1453 reg_id, complete_flag, rank_type,
1454 num_bases, basenames);
1456 else if (!strcmp (search_type, "local"))
1458 return rpn_search_APT_local (zh, zapt, termz, attributeSet,
1465 static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs,
1466 oid_value attributeSet, ODR stream,
1467 int num_bases, char **basenames)
1470 if (zs->which == Z_RPNStructure_complex)
1472 Z_Operator *zop = zs->u.complex->roperator;
1473 rset_bool_parms bool_parms;
1475 bool_parms.rset_l = rpn_search_structure (zh, zs->u.complex->s1,
1476 attributeSet, stream,
1477 num_bases, basenames);
1478 if (bool_parms.rset_l == NULL)
1480 bool_parms.rset_r = rpn_search_structure (zh, zs->u.complex->s2,
1481 attributeSet, stream,
1482 num_bases, basenames);
1483 if (bool_parms.rset_r == NULL)
1485 rset_delete (bool_parms.rset_l);
1488 bool_parms.key_size = sizeof(struct it_key);
1489 bool_parms.cmp = key_compare_it;
1493 case Z_Operator_and:
1494 r = rset_create (rset_kind_and, &bool_parms);
1497 r = rset_create (rset_kind_or, &bool_parms);
1499 case Z_Operator_and_not:
1500 r = rset_create (rset_kind_not, &bool_parms);
1502 case Z_Operator_prox:
1504 if (zop->u.prox->which != Z_ProximityOperator_known)
1510 if (zop->u.prox->which != Z_ProxCode_known)
1518 if (*zop->u.prox->u.known != Z_ProxUnit_word)
1520 char *val = odr_malloc (stream, 16);
1522 zh->errString = val;
1523 sprintf (val, "%d", *zop->u.prox->u.known);
1527 if (*zop->u.prox->proximityUnitCode != Z_ProxUnit_word)
1529 char *val = odr_malloc (stream, 16);
1531 zh->errString = val;
1532 sprintf (val, "%d", *zop->u.prox->proximityUnitCode);
1536 r = rpn_proximity (zh, bool_parms.rset_l, bool_parms.rset_r,
1537 *zop->u.prox->ordered,
1538 (!zop->u.prox->exclusion ? 0 :
1539 *zop->u.prox->exclusion),
1540 *zop->u.prox->relationType,
1541 *zop->u.prox->distance);
1548 else if (zs->which == Z_RPNStructure_simple)
1550 if (zs->u.simple->which == Z_Operand_APT)
1552 logf (LOG_DEBUG, "rpn_search_APT");
1553 r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm,
1554 attributeSet, num_bases, basenames);
1556 else if (zs->u.simple->which == Z_Operand_resultSetId)
1558 logf (LOG_DEBUG, "rpn_search_ref");
1559 r = resultSetRef (zh, zs->u.simple->u.resultSetId);
1561 r = rset_create (rset_kind_null, NULL);
1577 void rpn_search (ZebraHandle zh, ODR stream,
1578 Z_RPNQuery *rpn, int num_bases, char **basenames,
1579 const char *setname)
1583 oid_value attributeSet;
1588 zh->errString = NULL;
1591 attrset = oid_getentbyoid (rpn->attributeSetId);
1592 attributeSet = attrset->value;
1593 rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet, stream,
1594 num_bases, basenames);
1598 resultSetAdd (zh, setname, 1, rset, &zh->hits);
1600 logf (LOG_DEBUG, "search error: %d", zh->errCode);
1603 struct scan_info_entry {
1609 struct scan_info_entry *list;
1615 static int scan_handle (char *name, const char *info, int pos, void *client)
1617 int len_prefix, idx;
1618 struct scan_info *scan_info = client;
1620 len_prefix = strlen(scan_info->prefix);
1621 if (memcmp (name, scan_info->prefix, len_prefix))
1624 idx = scan_info->after - pos + scan_info->before;
1627 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1628 strlen(name + len_prefix)+1);
1629 strcpy (scan_info->list[idx].term, name + len_prefix);
1630 assert (*info == sizeof(ISAM_P));
1631 memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P));
1635 static void scan_term_untrans (ZebraHandle zh, ODR stream, int reg_type,
1636 char **dst, const char *src)
1638 char term_dst[1024];
1640 term_untrans (zh, reg_type, term_dst, src);
1642 *dst = odr_malloc (stream, strlen(term_dst)+1);
1643 strcpy (*dst, term_dst);
1646 static void count_set (RSET r, int *count)
1654 logf (LOG_DEBUG, "count_set");
1657 rfd = rset_open (r, RSETF_READ);
1658 while (rset_read (r, rfd, &key, &term_index))
1660 if (key.sysno != psysno)
1667 rset_close (r, rfd);
1668 logf (LOG_DEBUG, "%d keys, %d records", kno, *count);
1671 void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
1672 oid_value attributeset,
1673 int num_bases, char **basenames,
1674 int *position, int *num_entries, ZebraScanEntry **list,
1678 int pos = *position;
1679 int num = *num_entries;
1683 char termz[IT_MAX_WORD+20];
1686 struct scan_info *scan_info_array;
1687 ZebraScanEntry *glist;
1688 int ords[32], ord_no = 0;
1692 char *search_type = NULL;
1693 char *rank_type = NULL;
1696 if (attributeset == VAL_NONE)
1697 attributeset = VAL_BIB1;
1699 zlog_scan (zapt, attributeset);
1700 logf (LOG_DEBUG, "position = %d, num = %d", pos, num);
1702 attr_init (&use, zapt, 1);
1703 use_value = attr_find (&use, &attributeset);
1705 if (zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type,
1706 &rank_type, &complete_flag))
1712 if (use_value == -1)
1714 for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
1718 data1_local_attribute *local_attr;
1720 if ((r=att_getentbyatt (zh, &attp, attributeset, use_value)))
1722 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
1723 attributeset, use_value);
1729 if (zebraExplain_curDatabase (zh->zei, basenames[base_no]))
1731 zh->errString = basenames[base_no];
1732 zh->errCode = 109; /* Database unavailable */
1735 for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
1736 local_attr = local_attr->next)
1740 ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal,
1743 ords[ord_no++] = ord;
1751 /* prepare dictionary scanning */
1754 scan_info_array = odr_malloc (stream, ord_no * sizeof(*scan_info_array));
1755 for (i = 0; i < ord_no; i++)
1757 int j, prefix_len = 0;
1758 int before_tmp = before, after_tmp = after;
1759 struct scan_info *scan_info = scan_info_array + i;
1760 struct rpn_char_map_info rcmi;
1762 rpn_char_map_prepare (zh, reg_id, &rcmi);
1764 scan_info->before = before;
1765 scan_info->after = after;
1766 scan_info->odr = stream;
1768 scan_info->list = odr_malloc (stream, (before+after)*
1769 sizeof(*scan_info->list));
1770 for (j = 0; j<before+after; j++)
1771 scan_info->list[j].term = NULL;
1773 prefix_len += key_SU_code (ords[i], termz + prefix_len);
1774 termz[prefix_len++] = reg_id;
1775 termz[prefix_len] = 0;
1776 strcpy (scan_info->prefix, termz);
1778 trans_scan_term (zh, zapt, termz+prefix_len, reg_id);
1780 dict_scan (zh->dict, termz, &before_tmp, &after_tmp, scan_info,
1783 glist = odr_malloc (stream, (before+after)*sizeof(*glist));
1785 /* consider terms after main term */
1786 for (i = 0; i < ord_no; i++)
1790 for (i = 0; i<after; i++)
1793 const char *mterm = NULL;
1797 for (j = 0; j < ord_no; j++)
1799 if (ptr[j] < before+after &&
1800 (tst=scan_info_array[j].list[ptr[j]].term) &&
1801 (!mterm || strcmp (tst, mterm) < 0))
1809 scan_term_untrans (zh, stream, reg_id,
1810 &glist[i+before].term, mterm);
1811 rset = rset_trunc (zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
1812 glist[i+before].term, strlen(glist[i+before].term),
1816 for (j = j0+1; j<ord_no; j++)
1818 if (ptr[j] < before+after &&
1819 (tst=scan_info_array[j].list[ptr[j]].term) &&
1820 !strcmp (tst, mterm))
1822 rset_bool_parms bool_parms;
1826 rset_trunc (zh, &scan_info_array[j].list[ptr[j]].isam_p, 1,
1827 glist[i+before].term,
1828 strlen(glist[i+before].term), NULL);
1830 bool_parms.key_size = sizeof(struct it_key);
1831 bool_parms.cmp = key_compare_it;
1832 bool_parms.rset_l = rset;
1833 bool_parms.rset_r = rset2;
1835 rset = rset_create (rset_kind_or, &bool_parms);
1840 count_set (rset, &glist[i+before].occurrences);
1845 *num_entries -= (after-i);
1849 /* consider terms before main term */
1850 for (i = 0; i<ord_no; i++)
1853 for (i = 0; i<before; i++)
1856 const char *mterm = NULL;
1860 for (j = 0; j <ord_no; j++)
1862 if (ptr[j] < before &&
1863 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
1864 (!mterm || strcmp (tst, mterm) > 0))
1873 scan_term_untrans (zh, stream, reg_id,
1874 &glist[before-1-i].term, mterm);
1877 (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
1878 glist[before-1-i].term, strlen(glist[before-1-i].term),
1883 for (j = j0+1; j<ord_no; j++)
1885 if (ptr[j] < before &&
1886 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
1887 !strcmp (tst, mterm))
1889 rset_bool_parms bool_parms;
1892 rset2 = rset_trunc (zh,
1893 &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1,
1894 glist[before-1-i].term,
1895 strlen(glist[before-1-i].term), NULL);
1897 bool_parms.key_size = sizeof(struct it_key);
1898 bool_parms.cmp = key_compare_it;
1899 bool_parms.rset_l = rset;
1900 bool_parms.rset_r = rset2;
1902 rset = rset_create (rset_kind_or, &bool_parms);
1907 count_set (rset, &glist[before-1-i].occurrences);
1917 *list = glist + i; /* list is set to first 'real' entry */
1919 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1920 *position, *num_entries);
1922 logf (LOG_DEBUG, "scan error: %d", zh->errCode);