2 * Copyright (C) 1995-1998, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.78 1998-06-08 14:43:17 adam
8 * Added suport for EXPLAIN Proxy servers - added settings databasePath
9 * and explainDatabase to facilitate this. Increased maximum number
10 * of databases and attributes in one register.
12 * Revision 1.77 1998/05/20 10:12:22 adam
13 * Implemented automatic EXPLAIN database maintenance.
14 * Modified Zebra to work with ASN.1 compiled version of YAZ.
16 * Revision 1.76 1998/04/02 14:35:29 adam
17 * First version of Zebra that works with compiled ASN.1.
19 * Revision 1.75 1998/03/05 08:45:13 adam
20 * New result set model and modular ranking system. Moved towards
21 * descent server API. System information stored as "SGML" records.
23 * Revision 1.74 1998/02/10 12:03:06 adam
26 * Revision 1.73 1998/01/29 13:40:11 adam
27 * Better logging for scan service.
29 * Revision 1.72 1998/01/07 13:53:41 adam
30 * Queries using simple ranked operands returns right number of hits.
32 * Revision 1.71 1997/12/18 10:54:24 adam
33 * New method result set method rs_hits that returns the number of
34 * hits in result-set (if known). The ranked result set returns real
35 * number of hits but only when not combined with other operands.
37 * Revision 1.70 1997/10/31 12:34:43 adam
38 * Changed a few log statements.
40 * Revision 1.69 1997/10/29 12:05:02 adam
41 * Server produces diagnostic "Unsupported Attribute Set" when appropriate.
43 * Revision 1.68 1997/10/27 14:33:06 adam
44 * Moved towards generic character mapping depending on "structure"
45 * field in abstract syntax file. Fixed a few memory leaks. Fixed
46 * bug with negative integers when doing searches with relational
49 * Revision 1.67 1997/09/29 09:06:10 adam
50 * Removed one static var in order to make this module thread safe.
52 * Revision 1.66 1997/09/25 14:58:03 adam
55 * Revision 1.65 1997/09/22 12:39:06 adam
56 * Added get_pos method for the ranked result sets.
58 * Revision 1.64 1997/09/18 08:59:20 adam
59 * Extra generic handle for the character mapping routines.
61 * Revision 1.63 1997/09/17 12:19:18 adam
62 * Zebra version corresponds to YAZ version 1.4.
63 * Changed Zebra server so that it doesn't depend on global common_resource.
65 * Revision 1.62 1997/09/05 15:30:09 adam
66 * Changed prototype for chr_map_input - added const.
67 * Added support for C++, headers uses extern "C" for public definitions.
69 * Revision 1.61 1997/02/10 10:21:14 adam
70 * Bug fix: in search terms character (^) wasn't observed.
72 * Revision 1.60 1997/01/31 11:10:34 adam
73 * Bug fix: Leading and trailing white space weren't removed in scan tokens.
75 * Revision 1.59 1997/01/17 11:31:46 adam
76 * Bug fix: complete phrase search didn't work.
78 * Revision 1.58 1996/12/23 15:30:45 adam
80 * Bug fix: result sets weren't deleted after server shut down.
82 * Revision 1.57 1996/11/11 13:38:02 adam
83 * Added proximity support in search.
85 * Revision 1.56 1996/11/08 11:10:32 adam
86 * Buffers used during file match got bigger.
87 * Compressed ISAM support everywhere.
88 * Bug fixes regarding masking characters in queries.
89 * Redesigned Regexp-2 queries.
91 * Revision 1.55 1996/11/04 14:07:44 adam
92 * Moved truncation code to trunc.c.
94 * Revision 1.54 1996/10/29 14:09:52 adam
95 * Use of cisam system - enabled if setting isamc is 1.
97 * Revision 1.53 1996/06/26 09:21:43 adam
98 * Bug fix: local attribute set wasn't obeyed in scan.
100 * Revision 1.52 1996/06/17 14:26:20 adam
101 * Function gen_regular_rel changed to handle negative numbers.
103 * Revision 1.51 1996/06/11 10:54:15 quinn
106 * Revision 1.50 1996/06/07 08:51:53 adam
107 * Bug fix: Character mapping was broken (introducued by last revision).
109 * Revision 1.49 1996/06/04 10:18:11 adam
110 * Search/scan uses character mapping module.
112 * Revision 1.48 1996/05/28 15:15:01 adam
113 * Bug fix: Didn't handle unknown database correctly.
115 * Revision 1.47 1996/05/15 18:36:28 adam
116 * Function trans_term transforms unsearchable characters to blanks.
118 * Revision 1.46 1996/05/15 11:57:56 adam
119 * Fixed bug introduced by set/field mapping in search operations.
121 * Revision 1.45 1996/05/14 11:34:00 adam
122 * Scan support in multiple registers/databases.
124 * Revision 1.44 1996/05/14 06:16:44 adam
125 * Compact use/set bytes used in search service.
127 * Revision 1.43 1996/05/09 09:54:43 adam
128 * Server supports maps from one logical attributes to a list of physical
130 * The extraction process doesn't make space consuming 'any' keys.
132 * Revision 1.42 1996/05/09 07:28:56 quinn
133 * Work towards phrases and multiple registers
135 * Revision 1.41 1996/03/20 09:36:43 adam
136 * Function dict_lookup_grep got extra parameter, init_pos, which marks
137 * from which position in pattern approximate pattern matching should occur.
138 * Approximate pattern matching is used in relevance=re-2.
140 * Revision 1.40 1996/02/02 13:44:44 adam
141 * The public dictionary functions simply use char instead of Dict_char
142 * to represent search strings. Dict_char is used internally only.
144 * Revision 1.39 1996/01/03 16:22:13 quinn
145 * operator->roperator
147 * Revision 1.38 1995/12/11 09:12:55 adam
148 * The rec_get function returns NULL if record doesn't exist - will
149 * happen in the server if the result set records have been deleted since
150 * the creation of the set (i.e. the search).
151 * The server saves a result temporarily if it is 'volatile', i.e. the
152 * set is register dependent.
154 * Revision 1.37 1995/12/06 15:05:28 adam
155 * More verbose in count_set.
157 * Revision 1.36 1995/12/06 12:41:27 adam
158 * New command 'stat' for the index program.
159 * Filenames can be read from stdin by specifying '-'.
160 * Bug fix/enhancement of the transformation from terms to regular
161 * expressons in the search engine.
163 * Revision 1.35 1995/11/27 09:29:00 adam
164 * Bug fixes regarding conversion to regular expressions.
166 * Revision 1.34 1995/11/16 17:00:56 adam
167 * Better logging of rpn query.
169 * Revision 1.33 1995/11/01 13:58:28 quinn
170 * Moving data1 to yaz/retrieval
172 * Revision 1.32 1995/10/27 14:00:11 adam
173 * Implemented detection of database availability.
175 * Revision 1.31 1995/10/17 18:02:10 adam
176 * New feature: databases. Implemented as prefix to words in dictionary.
178 * Revision 1.30 1995/10/16 09:32:38 adam
179 * More work on relational op.
181 * Revision 1.29 1995/10/13 16:01:49 adam
184 * Revision 1.28 1995/10/13 12:26:43 adam
185 * Optimization of truncation.
187 * Revision 1.27 1995/10/12 17:07:22 adam
190 * Revision 1.26 1995/10/12 12:40:54 adam
191 * Bug fixes in rpn_prox.
193 * Revision 1.25 1995/10/10 13:59:24 adam
194 * Function rset_open changed its wflag parameter to general flags.
196 * Revision 1.24 1995/10/09 16:18:37 adam
197 * Function dict_lookup_grep got extra client data parameter.
199 * Revision 1.23 1995/10/06 16:33:37 adam
200 * Use attribute mappings.
202 * Revision 1.22 1995/10/06 15:07:39 adam
203 * Structure 'local-number' handled.
205 * Revision 1.21 1995/10/06 13:52:06 adam
206 * Bug fixes. Handler may abort further scanning.
208 * Revision 1.20 1995/10/06 11:06:33 adam
209 * Scan entries include 'occurrences' now.
211 * Revision 1.19 1995/10/06 10:43:56 adam
212 * Scan added. 'occurrences' in scan entries not set yet.
214 * Revision 1.18 1995/10/04 16:57:20 adam
215 * Key input and merge sort in one pass.
217 * Revision 1.17 1995/10/04 12:55:17 adam
218 * Bug fix in ranked search. Use=Any keys inserted.
220 * Revision 1.16 1995/10/02 16:24:40 adam
221 * Use attribute actually used in search requests.
223 * Revision 1.15 1995/10/02 15:18:52 adam
224 * New member in recRetrieveCtrl: diagnostic.
226 * Revision 1.14 1995/09/28 12:10:32 adam
227 * Bug fixes. Field prefix used in queries.
229 * Revision 1.13 1995/09/18 14:17:50 adam
232 * Revision 1.12 1995/09/15 14:45:21 adam
234 * Work on truncation.
236 * Revision 1.11 1995/09/14 11:53:27 adam
237 * First work on regular expressions/truncations.
239 * Revision 1.10 1995/09/11 15:23:26 adam
240 * More work on relevance search.
242 * Revision 1.9 1995/09/11 13:09:35 adam
243 * More work on relevance feedback.
245 * Revision 1.8 1995/09/08 14:52:27 adam
246 * Minor changes. Dictionary is lower case now.
248 * Revision 1.7 1995/09/07 13:58:36 adam
249 * New parameter: result-set file descriptor (RSFD) to support multiple
250 * positions within the same result-set.
251 * Boolean operators: and, or, not implemented.
252 * Result-set references.
254 * Revision 1.6 1995/09/06 16:11:18 adam
255 * Option: only one word key per file.
257 * Revision 1.5 1995/09/06 10:33:04 adam
258 * More work on present. Some log messages removed.
260 * Revision 1.4 1995/09/05 15:28:40 adam
261 * More work on search engine.
263 * Revision 1.3 1995/09/04 15:20:22 adam
266 * Revision 1.2 1995/09/04 12:33:43 adam
267 * Various cleanup. YAZ util used instead.
269 * Revision 1.1 1995/09/04 09:10:40 adam
270 * More work on index add/del/update.
271 * Merge sort implemented.
272 * Initial work on z39 server.
291 struct rpn_char_map_info {
296 static const char **rpn_char_map_handler (void *vp, const char **from, int len)
298 struct rpn_char_map_info *p = vp;
299 return zebra_maps_input (p->zm, p->reg_type, from, len);
302 static void rpn_char_map_prepare (ZebraHandle zh, int reg_type,
303 struct rpn_char_map_info *map_info)
305 map_info->zm = zh->zebra_maps;
306 map_info->reg_type = reg_type;
307 dict_grep_cmap (zh->dict, map_info, rpn_char_map_handler);
314 Z_AttributesPlusTerm *zapt;
317 static int attr_find (AttrType *src, oid_value *attributeSetP)
322 num_attributes = src->zapt->attributes->num_attributes;
324 num_attributes = src->zapt->num_attributes;
326 while (src->major < num_attributes)
328 Z_AttributeElement *element;
331 element = src->zapt->attributes->attributes[src->major];
333 element = src->zapt->attributeList[src->major];
335 if (src->type == *element->attributeType)
337 switch (element->which)
339 case Z_AttributeValue_numeric:
341 if (element->attributeSet && attributeSetP)
345 attrset = oid_getentbyoid (element->attributeSet);
346 *attributeSetP = attrset->value;
348 return *element->value.numeric;
350 case Z_AttributeValue_complex:
351 if (src->minor >= element->value.complex->num_list ||
352 element->value.complex->list[src->minor]->which !=
353 Z_StringOrNumeric_numeric)
356 if (element->attributeSet && attributeSetP)
360 attrset = oid_getentbyoid (element->attributeSet);
361 *attributeSetP = attrset->value;
363 return *element->value.complex->list[src->minor-1]->u.numeric;
373 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
395 static void term_untrans (ZebraHandle zh, int reg_type,
396 char *dst, const char *src)
400 const char *cp = zebra_maps_output (zh->zebra_maps, reg_type, &src);
407 static void add_isam_p (const char *name, const char *info,
411 if (p->isam_p_indx == p->isam_p_size)
413 ISAM_P *new_isam_p_buf;
417 p->isam_p_size = 2*p->isam_p_size + 100;
418 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
422 memcpy (new_isam_p_buf, p->isam_p_buf,
423 p->isam_p_indx * sizeof(*p->isam_p_buf));
424 xfree (p->isam_p_buf);
426 p->isam_p_buf = new_isam_p_buf;
429 new_term_no = xmalloc (sizeof(*new_term_no) *
433 memcpy (new_term_no, p->isam_p_buf,
434 p->isam_p_indx * sizeof(*p->term_no));
437 p->term_no = new_term_no;
440 assert (*info == sizeof(*p->isam_p_buf));
441 memcpy (p->isam_p_buf + p->isam_p_indx, info+1, sizeof(*p->isam_p_buf));
444 term_untrans (p->zh, p->reg_type, term_tmp, name+2);
445 logf (LOG_DEBUG, "grep: %s", term_tmp);
450 static int grep_handle (char *name, const char *info, void *p)
452 add_isam_p (name, info, p);
456 static int term_pre (ZebraMaps zebra_maps, int reg_type, const char **src,
457 const char *ct1, const char *ct2)
459 const char *s1, *s0 = *src;
462 /* skip white space */
465 if (ct1 && strchr (ct1, *s0))
467 if (ct2 && strchr (ct2, *s0))
470 map = zebra_maps_input (zebra_maps, reg_type, &s1, strlen(s1));
471 if (**map != *CHR_SPACE)
479 static int term_100 (ZebraMaps zebra_maps, int reg_type,
480 const char **src, char *dst, int space_split,
488 if (!term_pre (zebra_maps, reg_type, src, NULL, NULL))
494 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
495 if (space_split && **map == *CHR_SPACE)
499 if (!isalnum (*s1) && *s1 != '-')
511 static int term_101 (ZebraMaps zebra_maps, int reg_type,
512 const char **src, char *dst, int space_split,
520 if (!term_pre (zebra_maps, reg_type, src, "#", "#"))
529 dst_term[j++] = *s0++;
534 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
535 if (space_split && **map == *CHR_SPACE)
547 dst_term[j++] = '\0';
553 static int term_103 (ZebraMaps zebra_maps, int reg_type, const char **src,
554 char *dst, int *errors, int space_split,
562 if (!term_pre (zebra_maps, reg_type, src, "^\\()[].*+?|", "("))
565 if (errors && *s0 == '+' && s0[1] && s0[2] == '+' && s0[3] &&
568 *errors = s0[1] - '0';
575 if (strchr ("^\\()[].*+?|-", *s0))
583 map = zebra_maps_input (zebra_maps, reg_type, &s0, strlen(s0));
584 if (**map == *CHR_SPACE)
601 static int term_102 (ZebraMaps zebra_maps, int reg_type, const char **src,
602 char *dst, int space_split, char *dst_term)
604 return term_103 (zebra_maps, reg_type, src, dst, NULL, space_split,
608 /* gen_regular_rel - generate regular expression from relation
609 * val: border value (inclusive)
610 * islt: 1 if <=; 0 if >=.
612 static void gen_regular_rel (char *dst, int val, int islt)
619 logf (LOG_DEBUG, "gen_regular_rel. val=%d, islt=%d", val, islt);
623 strcpy (dst, "(-[0-9]+|(");
631 strcpy (dst, "([0-9]+|-(");
632 dst_p = strlen (dst);
642 dst_p = strlen (dst);
643 sprintf (numstr, "%d", val);
644 for (w = strlen(numstr); --w >= 0; pos++)
663 strcpy (dst + dst_p, numstr);
664 dst_p = strlen(dst) - pos - 1;
692 for (i = 0; i<pos; i++)
705 for (i=1; i<pos; i++)
706 strcat (dst, "[0-9]?");
710 for (i = 0; i <= pos; i++)
711 strcat (dst, "[0-9]");
712 strcat (dst, "[0-9]*");
717 static int relational_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
718 const char **term_sub,
720 oid_value attributeSet,
721 struct grep_info *grep_info,
730 char *term_tmp = term_dict + strlen(term_dict);
732 attr_init (&relation, zapt, 2);
733 relation_value = attr_find (&relation, NULL);
735 logf (LOG_DEBUG, "relation value=%d", relation_value);
736 switch (relation_value)
739 if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
742 term_value = atoi (term_tmp);
743 logf (LOG_DEBUG, "Relation <");
744 gen_regular_rel (term_tmp, term_value-1, 1);
747 if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
750 term_value = atoi (term_tmp);
751 logf (LOG_DEBUG, "Relation <=");
752 gen_regular_rel (term_tmp, term_value, 1);
755 if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
758 term_value = atoi (term_tmp);
759 logf (LOG_DEBUG, "Relation >=");
760 gen_regular_rel (term_tmp, term_value, 0);
763 if (!term_100 (zh->zebra_maps, reg_type, term_sub, term_tmp, 1,
766 term_value = atoi (term_tmp);
767 logf (LOG_DEBUG, "Relation >");
768 gen_regular_rel (term_tmp, term_value+1, 0);
773 logf (LOG_DEBUG, "dict_lookup_grep: %s", term_tmp);
774 r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info, max_pos,
777 logf (LOG_WARN, "dict_lookup_grep fail, rel=gt: %d", r);
778 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
782 static int field_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
783 const char **term_sub,
784 oid_value attributeSet, struct grep_info *grep_info,
785 int reg_type, int complete_flag,
786 int num_bases, char **basenames,
789 char term_dict[2*IT_MAX_WORD+2];
792 int truncation_value;
795 oid_value curAttributeSet = attributeSet;
797 struct rpn_char_map_info rcmi;
798 int space_split = complete_flag ? 0 : 1;
800 rpn_char_map_prepare (zh, reg_type, &rcmi);
801 attr_init (&use, zapt, 1);
802 use_value = attr_find (&use, &curAttributeSet);
803 logf (LOG_DEBUG, "field_term, use value %d", use_value);
804 attr_init (&truncation, zapt, 5);
805 truncation_value = attr_find (&truncation, NULL);
806 logf (LOG_DEBUG, "truncation value %d", truncation_value);
811 for (base_no = 0; base_no < num_bases; base_no++)
814 data1_local_attribute *local_attr;
815 int max_pos, prefix_len = 0;
818 if ((r=att_getentbyatt (zh, &attp, curAttributeSet, use_value)))
820 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d r=%d",
821 curAttributeSet, use_value, r);
828 if (zebraExplain_curDatabase (zh->zei, basenames[base_no]))
830 zh->errCode = 109; /* Database unavailable */
831 zh->errString = basenames[base_no];
834 for (local_attr = attp.local_attributes; local_attr;
835 local_attr = local_attr->next)
841 ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal,
846 term_dict[prefix_len++] = '|';
848 term_dict[prefix_len++] = '(';
850 ord_len = key_SU_code (ord, ord_buf);
851 for (i = 0; i<ord_len; i++)
853 term_dict[prefix_len++] = 1;
854 term_dict[prefix_len++] = ord_buf[i];
862 term_dict[prefix_len++] = ')';
863 term_dict[prefix_len++] = 1;
864 term_dict[prefix_len++] = reg_type;
865 logf (LOG_DEBUG, "reg_type = %d", term_dict[prefix_len-1]);
866 term_dict[prefix_len] = '\0';
867 if (!relational_term (zh, zapt, &termp, term_dict,
868 attributeSet, grep_info, &max_pos, reg_type,
872 switch (truncation_value)
874 case -1: /* not specified */
875 case 100: /* do not truncate */
876 term_dict[j++] = '(';
877 if (!term_100 (zh->zebra_maps, reg_type,
878 &termp, term_dict + j, space_split, term_dst))
880 strcat (term_dict, ")");
881 r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
882 &max_pos, 0, grep_handle);
884 logf (LOG_WARN, "dict_lookup_grep err, trunc=none:%d", r);
886 case 1: /* right truncation */
887 term_dict[j++] = '(';
888 if (!term_100 (zh->zebra_maps, reg_type,
889 &termp, term_dict + j, space_split, term_dst))
891 strcat (term_dict, ".*)");
892 dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
893 &max_pos, 0, grep_handle);
895 case 2: /* left truncation */
896 case 3: /* left&right truncation */
899 case 101: /* process # in term */
900 term_dict[j++] = '(';
901 if (!term_101 (zh->zebra_maps, reg_type,
902 &termp, term_dict + j, space_split, term_dst))
904 strcat (term_dict, ")");
905 r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
906 &max_pos, 0, grep_handle);
908 logf (LOG_WARN, "dict_lookup_grep err, trunc=#: %d", r);
910 case 102: /* Regexp-1 */
911 term_dict[j++] = '(';
912 if (!term_102 (zh->zebra_maps, reg_type,
913 &termp, term_dict + j, space_split, term_dst))
915 strcat (term_dict, ")");
916 logf (LOG_DEBUG, "Regexp-1 tolerance=%d", r);
917 r = dict_lookup_grep (zh->dict, term_dict, 0, grep_info,
918 &max_pos, 0, grep_handle);
920 logf (LOG_WARN, "dict_lookup_grep err, trunc=regular: %d",
923 case 103: /* Regexp-2 */
925 term_dict[j++] = '(';
926 if (!term_103 (zh->zebra_maps, reg_type,
927 &termp, term_dict + j, &r, space_split, term_dst))
929 strcat (term_dict, ")");
930 logf (LOG_DEBUG, "Regexp-2 tolerance=%d", r);
931 r = dict_lookup_grep (zh->dict, term_dict, r, grep_info,
932 &max_pos, 2, grep_handle);
934 logf (LOG_WARN, "dict_lookup_grep err, trunc=eregular: %d",
941 logf (LOG_DEBUG, "%d positions", grep_info->isam_p_indx);
945 static void trans_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
949 Z_Term *term = zapt->term;
951 sizez = term->u.general->len;
952 if (sizez > IT_MAX_WORD-1)
953 sizez = IT_MAX_WORD-1;
954 memcpy (termz, term->u.general->buf, sizez);
958 static void trans_scan_term (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
959 char *termz, int reg_type)
961 Z_Term *term = zapt->term;
963 const char *cp = (const char *) term->u.general->buf;
964 const char *cp_end = cp + term->u.general->len;
967 const char *space_map = NULL;
970 while ((len = (cp_end - cp)) > 0)
972 map = zebra_maps_input (zh->zebra_maps, reg_type, &cp, len);
973 if (**map == *CHR_SPACE)
978 for (src = space_map; *src; src++)
981 for (src = *map; *src; src++)
988 static RSET rpn_proximity (ZebraHandle zh, RSET rset1, RSET rset2,
990 int exclusion, int relation, int distance)
995 struct it_key buf1, buf2;
998 rset_temp_parms parms;
1001 rsfd1 = rset_open (rset1, RSETF_READ);
1002 more1 = rset_read (rset1, rsfd1, &buf1, &term_index);
1004 rsfd2 = rset_open (rset2, RSETF_READ);
1005 more2 = rset_read (rset2, rsfd2, &buf2, &term_index);
1007 parms.key_size = sizeof (struct it_key);
1008 parms.temp_path = res_get (zh->res, "setTmpDir");
1009 result = rset_create (rset_kind_temp, &parms);
1010 rsfd_result = rset_open (result, RSETF_WRITE);
1012 logf (LOG_DEBUG, "rpn_proximity excl=%d ord=%d rel=%d dis=%d",
1013 exclusion, ordered, relation, distance);
1014 while (more1 && more2)
1016 int cmp = key_compare_it (&buf1, &buf2);
1018 more1 = rset_read (rset1, rsfd1, &buf1, &term_index);
1020 more2 = rset_read (rset2, rsfd2, &buf2, &term_index);
1023 int sysno = buf1.sysno;
1027 seqno[n++] = buf1.seqno;
1028 while ((more1 = rset_read (rset1, rsfd1, &buf1, &term_index)) &&
1029 sysno == buf1.sysno)
1031 seqno[n++] = buf1.seqno;
1034 for (i = 0; i<n; i++)
1036 int diff = buf2.seqno - seqno[i];
1037 int excl = exclusion;
1038 if (!ordered && diff < 0)
1043 if (diff < distance)
1047 if (diff <= distance)
1051 if (diff == distance)
1055 if (diff >= distance)
1059 if (diff > distance)
1063 if (diff != distance)
1068 rset_write (result, rsfd_result, &buf2);
1070 } while ((more2 = rset_read (rset2, rsfd2, &buf2, &term_index)) &&
1071 sysno == buf2.sysno);
1074 rset_close (result, rsfd_result);
1075 rset_close (rset1, rsfd1);
1076 rset_close (rset2, rsfd2);
1080 static RSET rpn_prox (ZebraHandle zh, RSET *rset, int rset_no)
1085 struct it_key **buf;
1087 char prox_term[1024];
1088 int length_prox_term = 0;
1089 int min_nn = 10000000;
1091 const char *flags = NULL;
1093 rsfd = xmalloc (sizeof(*rsfd)*rset_no);
1094 more = xmalloc (sizeof(*more)*rset_no);
1095 buf = xmalloc (sizeof(*buf)*rset_no);
1097 for (i = 0; i<rset_no; i++)
1100 buf[i] = xmalloc (sizeof(**buf));
1101 rsfd[i] = rset_open (rset[i], RSETF_READ);
1102 if (!(more[i] = rset_read (rset[i], rsfd[i], buf[i], &term_index)))
1104 for (j = 0; j<rset[i]->no_rset_terms; j++)
1106 const char *nflags = rset[i]->rset_terms[j]->flags;
1107 char *term = rset[i]->rset_terms[j]->name;
1108 int lterm = strlen(term);
1109 if (length_prox_term)
1110 prox_term[length_prox_term++] = ' ';
1111 strcpy (prox_term + length_prox_term, term);
1112 length_prox_term += lterm;
1113 if (min_nn > rset[i]->rset_terms[j]->nn)
1114 min_nn = rset[i]->rset_terms[j]->nn;
1120 rset_null_parms parms;
1124 rset_close (rset[i], rsfd[i]);
1128 parms.rset_term = rset_term_create (prox_term, -1, flags);
1129 parms.rset_term->nn = 0;
1130 result = rset_create (rset_kind_null, &parms);
1134 rset_temp_parms parms;
1137 parms.rset_term = rset_term_create (prox_term, -1, flags);
1138 parms.rset_term->nn = min_nn;
1139 parms.key_size = sizeof (struct it_key);
1140 parms.temp_path = res_get (zh->res, "setTmpDir");
1141 result = rset_create (rset_kind_temp, &parms);
1142 rsfd_result = rset_open (result, RSETF_WRITE);
1146 for (i = 1; i<rset_no; i++)
1155 cmp = key_compare_it (buf[i], buf[i-1]);
1158 more[i-1] = rset_read (rset[i-1], rsfd[i-1],
1159 buf[i-1], &term_index);
1164 if (buf[i-1]->seqno+1 != buf[i]->seqno)
1166 more[i-1] = rset_read (rset[i-1], rsfd[i-1],
1167 buf[i-1], &term_index);
1173 more[i] = rset_read (rset[i], rsfd[i], buf[i],
1180 rset_write (result, rsfd_result, buf[0]);
1181 more[0] = rset_read (*rset, *rsfd, *buf, &term_index);
1185 for (i = 0; i<rset_no; i++)
1187 rset_close (rset[i], rsfd[i]);
1190 rset_close (result, rsfd_result);
1198 static RSET rpn_search_APT_phrase (ZebraHandle zh,
1199 Z_AttributesPlusTerm *zapt,
1201 oid_value attributeSet,
1202 int reg_type, int complete_flag,
1203 const char *rank_type,
1204 int num_bases, char **basenames)
1206 char term_dst[IT_MAX_WORD+1];
1207 const char *termp = termz;
1208 RSET rset[60], result;
1209 int i, r, rset_no = 0;
1210 struct grep_info grep_info;
1213 grep_info.term_no = 0;
1215 grep_info.isam_p_size = 0;
1216 grep_info.isam_p_buf = NULL;
1218 grep_info.reg_type = reg_type;
1222 logf (LOG_DEBUG, "APT_phrase termp=%s", termp);
1223 grep_info.isam_p_indx = 0;
1224 r = field_term (zh, zapt, &termp, attributeSet, &grep_info,
1225 reg_type, complete_flag, num_bases, basenames,
1229 logf (LOG_DEBUG, "term: %s", term_dst);
1230 rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
1231 grep_info.isam_p_indx, term_dst,
1232 strlen(term_dst), rank_type);
1233 assert (rset[rset_no]);
1234 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1238 xfree(grep_info.term_no);
1240 xfree (grep_info.isam_p_buf);
1243 rset_null_parms parms;
1245 parms.rset_term = rset_term_create (term_dst, -1, rank_type);
1246 return rset_create (rset_kind_null, &parms);
1248 else if (rset_no == 1)
1250 result = rpn_prox (zh, rset, rset_no);
1251 for (i = 0; i<rset_no; i++)
1252 rset_delete (rset[i]);
1256 static RSET rpn_search_APT_or_list (ZebraHandle zh,
1257 Z_AttributesPlusTerm *zapt,
1259 oid_value attributeSet,
1260 int reg_type, int complete_flag,
1261 const char *rank_type,
1262 int num_bases, char **basenames)
1264 char term_dst[IT_MAX_WORD+1];
1265 const char *termp = termz;
1266 RSET rset[60], result;
1267 int i, r, rset_no = 0;
1268 struct grep_info grep_info;
1271 grep_info.term_no = 0;
1273 grep_info.isam_p_size = 0;
1274 grep_info.isam_p_buf = NULL;
1276 grep_info.reg_type = reg_type;
1280 logf (LOG_DEBUG, "APT_or_list termp=%s", termp);
1281 grep_info.isam_p_indx = 0;
1282 r = field_term (zh, zapt, &termp, attributeSet, &grep_info,
1283 reg_type, complete_flag, num_bases, basenames,
1287 logf (LOG_DEBUG, "term: %s", term_dst);
1288 rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
1289 grep_info.isam_p_indx, term_dst,
1290 strlen(term_dst), rank_type);
1291 assert (rset[rset_no]);
1292 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1296 xfree(grep_info.term_no);
1298 xfree (grep_info.isam_p_buf);
1301 rset_null_parms parms;
1303 parms.rset_term = rset_term_create (term_dst, -1, rank_type);
1304 return rset_create (rset_kind_null, &parms);
1307 for (i = 1; i<rset_no; i++)
1309 rset_bool_parms bool_parms;
1311 bool_parms.rset_l = result;
1312 bool_parms.rset_r = rset[i];
1313 bool_parms.key_size = sizeof(struct it_key);
1314 bool_parms.cmp = key_compare_it;
1315 result = rset_create (rset_kind_or, &bool_parms);
1320 static RSET rpn_search_APT_and_list (ZebraHandle zh,
1321 Z_AttributesPlusTerm *zapt,
1323 oid_value attributeSet,
1324 int reg_type, int complete_flag,
1325 const char *rank_type,
1326 int num_bases, char **basenames)
1328 char term_dst[IT_MAX_WORD+1];
1329 const char *termp = termz;
1330 RSET rset[60], result;
1331 int i, r, rset_no = 0;
1332 struct grep_info grep_info;
1335 grep_info.term_no = 0;
1337 grep_info.isam_p_size = 0;
1338 grep_info.isam_p_buf = NULL;
1340 grep_info.reg_type = reg_type;
1344 logf (LOG_DEBUG, "APT_and_list termp=%s", termp);
1345 grep_info.isam_p_indx = 0;
1346 r = field_term (zh, zapt, &termp, attributeSet, &grep_info,
1347 reg_type, complete_flag, num_bases, basenames,
1351 logf (LOG_DEBUG, "term: %s", term_dst);
1352 rset[rset_no] = rset_trunc (zh, grep_info.isam_p_buf,
1353 grep_info.isam_p_indx, term_dst,
1354 strlen(term_dst), rank_type);
1355 assert (rset[rset_no]);
1356 if (++rset_no >= sizeof(rset)/sizeof(*rset))
1360 xfree(grep_info.term_no);
1362 xfree (grep_info.isam_p_buf);
1365 rset_null_parms parms;
1367 parms.rset_term = rset_term_create (term_dst, -1, rank_type);
1368 return rset_create (rset_kind_null, &parms);
1371 for (i = 1; i<rset_no; i++)
1373 rset_bool_parms bool_parms;
1375 bool_parms.rset_l = result;
1376 bool_parms.rset_r = rset[i];
1377 bool_parms.key_size = sizeof(struct it_key);
1378 bool_parms.cmp = key_compare_it;
1379 result = rset_create (rset_kind_and, &bool_parms);
1385 static RSET rpn_search_APT_local (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1387 oid_value attributeSet,
1388 const char *rank_type)
1393 rset_temp_parms parms;
1395 parms.rset_term = rset_term_create (termz, -1, rank_type);
1396 parms.key_size = sizeof (struct it_key);
1397 parms.temp_path = res_get (zh->res, "setTmpDir");
1398 result = rset_create (rset_kind_temp, &parms);
1399 rsfd = rset_open (result, RSETF_WRITE);
1401 key.sysno = atoi (termz);
1405 rset_write (result, rsfd, &key);
1406 rset_close (result, rsfd);
1410 static RSET rpn_search_APT (ZebraHandle zh, Z_AttributesPlusTerm *zapt,
1411 oid_value attributeSet,
1412 int num_bases, char **basenames)
1415 char *search_type = NULL;
1416 char *rank_type = NULL;
1418 char termz[IT_MAX_WORD+1];
1420 zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type,
1421 &rank_type, &complete_flag);
1423 logf (LOG_DEBUG, "reg_id=%c", reg_id);
1424 logf (LOG_DEBUG, "complete_flag=%d", complete_flag);
1425 logf (LOG_DEBUG, "search_type=%s", search_type);
1426 logf (LOG_DEBUG, "rank_type=%s", rank_type);
1428 if (zapt->term->which != Z_Term_general)
1433 trans_term (zh, zapt, termz);
1435 if (!strcmp (search_type, "phrase"))
1437 return rpn_search_APT_phrase (zh, zapt, termz, attributeSet,
1438 reg_id, complete_flag, rank_type,
1439 num_bases, basenames);
1441 else if (!strcmp (search_type, "and-list"))
1443 return rpn_search_APT_and_list (zh, zapt, termz, attributeSet,
1444 reg_id, complete_flag, rank_type,
1445 num_bases, basenames);
1447 else if (!strcmp (search_type, "or-list"))
1449 return rpn_search_APT_or_list (zh, zapt, termz, attributeSet,
1450 reg_id, complete_flag, rank_type,
1451 num_bases, basenames);
1453 else if (!strcmp (search_type, "local"))
1455 return rpn_search_APT_local (zh, zapt, termz, attributeSet,
1462 static RSET rpn_search_structure (ZebraHandle zh, Z_RPNStructure *zs,
1463 oid_value attributeSet, ODR stream,
1464 int num_bases, char **basenames)
1467 if (zs->which == Z_RPNStructure_complex)
1469 Z_Operator *zop = zs->u.complex->roperator;
1470 rset_bool_parms bool_parms;
1472 bool_parms.rset_l = rpn_search_structure (zh, zs->u.complex->s1,
1473 attributeSet, stream,
1474 num_bases, basenames);
1475 if (bool_parms.rset_l == NULL)
1477 bool_parms.rset_r = rpn_search_structure (zh, zs->u.complex->s2,
1478 attributeSet, stream,
1479 num_bases, basenames);
1480 if (bool_parms.rset_r == NULL)
1482 rset_delete (bool_parms.rset_l);
1485 bool_parms.key_size = sizeof(struct it_key);
1486 bool_parms.cmp = key_compare_it;
1490 case Z_Operator_and:
1491 r = rset_create (rset_kind_and, &bool_parms);
1494 r = rset_create (rset_kind_or, &bool_parms);
1496 case Z_Operator_and_not:
1497 r = rset_create (rset_kind_not, &bool_parms);
1499 case Z_Operator_prox:
1501 if (zop->u.prox->which != Z_ProximityOperator_known)
1507 if (zop->u.prox->which != Z_ProxCode_known)
1515 if (*zop->u.prox->u.known != Z_ProxUnit_word)
1517 char *val = odr_malloc (stream, 16);
1519 zh->errString = val;
1520 sprintf (val, "%d", *zop->u.prox->u.known);
1524 if (*zop->u.prox->proximityUnitCode != Z_ProxUnit_word)
1526 char *val = odr_malloc (stream, 16);
1528 zh->errString = val;
1529 sprintf (val, "%d", *zop->u.prox->proximityUnitCode);
1533 r = rpn_proximity (zh, bool_parms.rset_l, bool_parms.rset_r,
1534 *zop->u.prox->ordered,
1535 (!zop->u.prox->exclusion ? 0 :
1536 *zop->u.prox->exclusion),
1537 *zop->u.prox->relationType,
1538 *zop->u.prox->distance);
1545 else if (zs->which == Z_RPNStructure_simple)
1547 if (zs->u.simple->which == Z_Operand_APT)
1549 logf (LOG_DEBUG, "rpn_search_APT");
1550 r = rpn_search_APT (zh, zs->u.simple->u.attributesPlusTerm,
1551 attributeSet, num_bases, basenames);
1553 else if (zs->u.simple->which == Z_Operand_resultSetId)
1555 logf (LOG_DEBUG, "rpn_search_ref");
1556 r = resultSetRef (zh, zs->u.simple->u.resultSetId);
1558 r = rset_create (rset_kind_null, NULL);
1574 void rpn_search (ZebraHandle zh, ODR stream,
1575 Z_RPNQuery *rpn, int num_bases, char **basenames,
1576 const char *setname)
1580 oid_value attributeSet;
1585 zh->errString = NULL;
1588 attrset = oid_getentbyoid (rpn->attributeSetId);
1589 attributeSet = attrset->value;
1590 rset = rpn_search_structure (zh, rpn->RPNStructure, attributeSet, stream,
1591 num_bases, basenames);
1595 resultSetAdd (zh, setname, 1, rset, &zh->hits);
1597 logf (LOG_DEBUG, "search error: %d", zh->errCode);
1600 struct scan_info_entry {
1606 struct scan_info_entry *list;
1612 static int scan_handle (char *name, const char *info, int pos, void *client)
1614 int len_prefix, idx;
1615 struct scan_info *scan_info = client;
1617 len_prefix = strlen(scan_info->prefix);
1618 if (memcmp (name, scan_info->prefix, len_prefix))
1621 idx = scan_info->after - pos + scan_info->before;
1624 scan_info->list[idx].term = odr_malloc (scan_info->odr,
1625 strlen(name + len_prefix)+1);
1626 strcpy (scan_info->list[idx].term, name + len_prefix);
1627 assert (*info == sizeof(ISAM_P));
1628 memcpy (&scan_info->list[idx].isam_p, info+1, sizeof(ISAM_P));
1632 static void scan_term_untrans (ZebraHandle zh, ODR stream, int reg_type,
1633 char **dst, const char *src)
1635 char term_dst[1024];
1637 term_untrans (zh, reg_type, term_dst, src);
1639 *dst = odr_malloc (stream, strlen(term_dst)+1);
1640 strcpy (*dst, term_dst);
1643 static void count_set (RSET r, int *count)
1651 logf (LOG_DEBUG, "count_set");
1654 rfd = rset_open (r, RSETF_READ);
1655 while (rset_read (r, rfd, &key, &term_index))
1657 if (key.sysno != psysno)
1664 rset_close (r, rfd);
1665 logf (LOG_DEBUG, "%d keys, %d records", kno, *count);
1668 void rpn_scan (ZebraHandle zh, ODR stream, Z_AttributesPlusTerm *zapt,
1669 oid_value attributeset,
1670 int num_bases, char **basenames,
1671 int *position, int *num_entries, ZebraScanEntry **list,
1675 int pos = *position;
1676 int num = *num_entries;
1680 char termz[IT_MAX_WORD+20];
1683 struct scan_info *scan_info_array;
1684 ZebraScanEntry *glist;
1685 int ords[32], ord_no = 0;
1689 char *search_type = NULL;
1690 char *rank_type = NULL;
1693 if (attributeset == VAL_NONE)
1694 attributeset = VAL_BIB1;
1696 zlog_scan (zapt, attributeset);
1697 logf (LOG_DEBUG, "position = %d, num = %d", pos, num);
1699 attr_init (&use, zapt, 1);
1700 use_value = attr_find (&use, &attributeset);
1702 if (zebra_maps_attr (zh->zebra_maps, zapt, ®_id, &search_type,
1703 &rank_type, &complete_flag))
1709 if (use_value == -1)
1711 for (base_no = 0; base_no < num_bases && ord_no < 32; base_no++)
1715 data1_local_attribute *local_attr;
1717 if ((r=att_getentbyatt (zh, &attp, attributeset, use_value)))
1719 logf (LOG_DEBUG, "att_getentbyatt fail. set=%d use=%d",
1720 attributeset, use_value);
1726 if (zebraExplain_curDatabase (zh->zei, basenames[base_no]))
1728 zh->errString = basenames[base_no];
1729 zh->errCode = 109; /* Database unavailable */
1732 for (local_attr = attp.local_attributes; local_attr && ord_no < 32;
1733 local_attr = local_attr->next)
1737 ord = zebraExplain_lookupSU (zh->zei, attp.attset_ordinal,
1740 ords[ord_no++] = ord;
1750 scan_info_array = odr_malloc (stream, ord_no * sizeof(*scan_info_array));
1751 for (i = 0; i < ord_no; i++)
1753 int j, prefix_len = 0;
1754 int before_tmp = before, after_tmp = after;
1755 struct scan_info *scan_info = scan_info_array + i;
1756 struct rpn_char_map_info rcmi;
1758 rpn_char_map_prepare (zh, reg_id, &rcmi);
1760 scan_info->before = before;
1761 scan_info->after = after;
1762 scan_info->odr = stream;
1764 scan_info->list = odr_malloc (stream, (before+after)*
1765 sizeof(*scan_info->list));
1766 for (j = 0; j<before+after; j++)
1767 scan_info->list[j].term = NULL;
1769 prefix_len += key_SU_code (ords[i], termz + prefix_len);
1770 termz[prefix_len++] = reg_id;
1771 termz[prefix_len] = 0;
1772 strcpy (scan_info->prefix, termz);
1774 trans_scan_term (zh, zapt, termz+prefix_len, reg_id);
1776 dict_scan (zh->dict, termz, &before_tmp, &after_tmp, scan_info,
1779 glist = odr_malloc (stream, (before+after)*sizeof(*glist));
1780 for (i = 0; i < ord_no; i++)
1784 for (i = 0; i<after; i++)
1787 const char *mterm = NULL;
1791 for (j = 0; j < ord_no; j++)
1793 if (ptr[j] < before+after &&
1794 (tst=scan_info_array[j].list[ptr[j]].term) &&
1795 (!mterm || strcmp (tst, mterm) < 0))
1803 scan_term_untrans (zh, stream, reg_id,
1804 &glist[i+before].term, mterm);
1805 rset = rset_trunc (zh, &scan_info_array[j0].list[ptr[j0]].isam_p, 1,
1806 glist[i+before].term, strlen(glist[i+before].term),
1810 for (j = j0+1; j<ord_no; j++)
1812 if (ptr[j] < before+after &&
1813 (tst=scan_info_array[j].list[ptr[j]].term) &&
1814 !strcmp (tst, mterm))
1816 rset_bool_parms bool_parms;
1820 rset_trunc (zh, &scan_info_array[j].list[ptr[j]].isam_p, 1,
1821 glist[i+before].term,
1822 strlen(glist[i+before].term), NULL);
1824 bool_parms.key_size = sizeof(struct it_key);
1825 bool_parms.cmp = key_compare_it;
1826 bool_parms.rset_l = rset;
1827 bool_parms.rset_r = rset2;
1829 rset = rset_create (rset_kind_or, &bool_parms);
1834 count_set (rset, &glist[i+before].occurrences);
1839 *num_entries -= (after-i);
1843 for (i = 0; i<ord_no; i++)
1846 for (i = 0; i<before; i++)
1849 const char *mterm = NULL;
1853 for (j = 0; j <ord_no; j++)
1855 if (ptr[j] < before &&
1856 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
1857 (!mterm || strcmp (tst, mterm) > 0))
1866 scan_term_untrans (zh, stream, reg_id,
1867 &glist[before-1-i].term, mterm);
1870 (zh, &scan_info_array[j0].list[before-1-ptr[j0]].isam_p, 1,
1871 glist[before-1-i].term, strlen(glist[before-1-i].term),
1876 for (j = j0+1; j<ord_no; j++)
1878 if (ptr[j] < before &&
1879 (tst=scan_info_array[j].list[before-1-ptr[j]].term) &&
1880 !strcmp (tst, mterm))
1882 rset_bool_parms bool_parms;
1885 rset2 = rset_trunc (zh,
1886 &scan_info_array[j].list[before-1-ptr[j]].isam_p, 1,
1887 glist[before-1-i].term,
1888 strlen(glist[before-1-i].term), NULL);
1890 bool_parms.key_size = sizeof(struct it_key);
1891 bool_parms.cmp = key_compare_it;
1892 bool_parms.rset_l = rset;
1893 bool_parms.rset_r = rset2;
1895 rset = rset_create (rset_kind_or, &bool_parms);
1900 count_set (rset, &glist[before-1-i].occurrences);
1910 *list = glist + i; /* list is set to first 'real' entry */
1912 logf (LOG_DEBUG, "position = %d, num_entries = %d",
1913 *position, *num_entries);
1915 logf (LOG_DEBUG, "scan error: %d", zh->errCode);