2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.10 1995-09-11 15:23:26 adam
8 * More work on relevance search.
10 * Revision 1.9 1995/09/11 13:09:35 adam
11 * More work on relevance feedback.
13 * Revision 1.8 1995/09/08 14:52:27 adam
14 * Minor changes. Dictionary is lower case now.
16 * Revision 1.7 1995/09/07 13:58:36 adam
17 * New parameter: result-set file descriptor (RSFD) to support multiple
18 * positions within the same result-set.
19 * Boolean operators: and, or, not implemented.
20 * Result-set references.
22 * Revision 1.6 1995/09/06 16:11:18 adam
23 * Option: only one word key per file.
25 * Revision 1.5 1995/09/06 10:33:04 adam
26 * More work on present. Some log messages removed.
28 * Revision 1.4 1995/09/05 15:28:40 adam
29 * More work on search engine.
31 * Revision 1.3 1995/09/04 15:20:22 adam
34 * Revision 1.2 1995/09/04 12:33:43 adam
35 * Various cleanup. YAZ util used instead.
37 * Revision 1.1 1995/09/04 09:10:40 adam
38 * More work on index add/del/update.
39 * Merge sort implemented.
40 * Initial work on z39 server.
55 int split_term (ZServerInfo *zi, Z_Term *term, ISAM_P **isam_ps, int *no)
57 static ISAM_P isam_p[16];
59 char termz[IT_MAX_WORD+1];
60 char term_sub[IT_MAX_WORD+1];
65 if (term->which != Z_Term_general)
67 sizez = term->u.general->len;
68 if (sizez > IT_MAX_WORD)
70 for (i = 0; i<sizez; i++)
71 termz[i] = index_char_cvt (term->u.general->buf[i]);
77 if ((p1 = strchr (p0, ' ')))
79 memcpy (term_sub, p0, p1-p0);
80 term_sub[p1-p0] = '\0';
83 strcpy (term_sub, p0);
84 logf (LOG_DEBUG, "dict_lookup: %s", term_sub);
85 if ((info = dict_lookup (zi->wordDict, term_sub)))
87 logf (LOG_DEBUG, " found");
88 assert (*info == sizeof(*isam_p));
89 memcpy (isam_p + isam_p_indx, info+1, sizeof(*isam_p));
98 logf (LOG_DEBUG, "%d positions", *no);
102 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
103 Z_AttributesPlusTerm *zapt)
105 rset_relevance_parms parms;
107 parms.key_size = sizeof(struct it_key);
109 parms.cmp = key_compare;
110 parms.is = zi->wordIsam;
111 split_term (zi, zapt->term, &parms.isam_positions,
112 &parms.no_isam_positions);
113 if (parms.no_isam_positions > 0)
114 return rset_create (rset_kind_relevance, &parms);
116 return rset_create (rset_kind_null, NULL);
119 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt)
122 Z_Term *term = zapt->term;
123 char termz[IT_MAX_WORD+1];
125 struct rset_isam_parms parms;
129 if (term->which != Z_Term_general)
131 sizez = term->u.general->len;
132 if (sizez > IT_MAX_WORD)
134 for (i = 0; i<sizez; i++)
135 termz[i] = index_char_cvt (term->u.general->buf[i]);
137 logf (LOG_DEBUG, "dict_lookup: %s", termz);
138 if (!(info = dict_lookup (zi->wordDict, termz)))
139 return rset_create (rset_kind_null, NULL);
140 assert (*info == sizeof(parms.pos));
141 memcpy (&parms.pos, info+1, sizeof(parms.pos));
142 parms.is = zi->wordIsam;
143 logf (LOG_DEBUG, "rset_create isam");
144 return rset_create (rset_kind_isam, &parms);
146 return rpn_search_APT_relevance (zi, zapt);
150 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
154 if (!(s = resultSetGet (zi, resultSetId)))
155 return rset_create (rset_kind_null, NULL);
159 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs)
162 if (zs->which == Z_RPNStructure_complex)
164 rset_bool_parms bool_parms;
166 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1);
167 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2);
168 bool_parms.key_size = sizeof(struct it_key);
169 bool_parms.cmp = key_compare;
171 switch (zs->u.complex->operator->which)
174 r = rset_create (rset_kind_and, &bool_parms);
177 r = rset_create (rset_kind_or, &bool_parms);
179 case Z_Operator_and_not:
180 r = rset_create (rset_kind_not, &bool_parms);
186 else if (zs->which == Z_RPNStructure_simple)
188 if (zs->u.simple->which == Z_Operand_APT)
190 logf (LOG_DEBUG, "rpn_search_APT");
191 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm);
193 else if (zs->u.simple->which == Z_Operand_resultSetId)
195 logf (LOG_DEBUG, "rpn_search_ref");
196 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
210 static RSET rpn_save_set (RSET r, int *count)
214 rset_temp_parms parms;
220 logf (LOG_DEBUG, "rpn_save_set");
223 parms.key_size = sizeof(struct it_key);
224 d = rset_create (rset_kind_temp, &parms);
228 rfd = rset_open (r, 0);
229 while (rset_read (r, rfd, &key))
231 if (key.sysno != psysno)
237 rset_write (d, &key);
244 logf (LOG_DEBUG, "%d distinct sysnos", *count);
250 int rpn_search (ZServerInfo *zi,
251 Z_RPNQuery *rpn, int num_bases, char **basenames,
252 const char *setname, int *hits)
254 RSET rset, result_rset;
256 rset = rpn_search_structure (zi, rpn->RPNStructure);
259 result_rset = rpn_save_set (rset, hits);
261 rset_delete (result_rset);
264 resultSetAdd (zi, setname, 1, rset);