2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.11 1995-09-14 11:53:27 adam
8 * First work on regular expressions/truncations.
10 * Revision 1.10 1995/09/11 15:23:26 adam
11 * More work on relevance search.
13 * Revision 1.9 1995/09/11 13:09:35 adam
14 * More work on relevance feedback.
16 * Revision 1.8 1995/09/08 14:52:27 adam
17 * Minor changes. Dictionary is lower case now.
19 * Revision 1.7 1995/09/07 13:58:36 adam
20 * New parameter: result-set file descriptor (RSFD) to support multiple
21 * positions within the same result-set.
22 * Boolean operators: and, or, not implemented.
23 * Result-set references.
25 * Revision 1.6 1995/09/06 16:11:18 adam
26 * Option: only one word key per file.
28 * Revision 1.5 1995/09/06 10:33:04 adam
29 * More work on present. Some log messages removed.
31 * Revision 1.4 1995/09/05 15:28:40 adam
32 * More work on search engine.
34 * Revision 1.3 1995/09/04 15:20:22 adam
37 * Revision 1.2 1995/09/04 12:33:43 adam
38 * Various cleanup. YAZ util used instead.
40 * Revision 1.1 1995/09/04 09:10:40 adam
41 * More work on index add/del/update.
42 * Merge sort implemented.
43 * Initial work on z39 server.
59 * attr_print: log attributes
61 static void attr_print (Z_AttributesPlusTerm *t)
64 for (of = 0; of < t->num_attributes; of++)
66 Z_AttributeElement *element;
67 element = t->attributeList[of];
69 switch (element->which)
71 case Z_AttributeValue_numeric:
72 logf (LOG_DEBUG, "attributeType=%d value=%d",
73 *element->attributeType,
74 *element->value.numeric);
76 case Z_AttributeValue_complex:
77 logf (LOG_DEBUG, "attributeType=%d complex",
78 *element->attributeType);
79 for (i = 0; i<element->value.complex->num_list; i++)
81 if (element->value.complex->list[i]->which ==
82 Z_StringOrNumeric_string)
83 logf (LOG_DEBUG, " string: '%s'",
84 element->value.complex->list[i]->u.string);
85 else if (element->value.complex->list[i]->which ==
86 Z_StringOrNumeric_numeric)
87 logf (LOG_DEBUG, " numeric: '%d'",
88 *element->value.complex->list[i]->u.numeric);
101 Z_AttributesPlusTerm *zapt;
104 static int attr_find (AttrType *src)
106 while (src->major < src->zapt->num_attributes)
108 Z_AttributeElement *element;
109 element = src->zapt->attributeList[src->major];
111 if (src->type == *element->attributeType)
113 switch (element->which)
115 case Z_AttributeValue_numeric:
117 return *element->value.numeric;
119 case Z_AttributeValue_complex:
120 if (src->minor >= element->value.complex->num_list ||
121 element->value.complex->list[src->minor]->which !=
122 Z_StringOrNumeric_numeric)
125 return *element->value.complex->list[src->minor-1]->u.numeric;
135 static void attr_init (AttrType *src, Z_AttributesPlusTerm *zapt,
144 static ISAM_P *isam_p_buf = NULL;
145 static int isam_p_size = 0;
146 static int isam_p_indx;
148 static void add_isam_p (const char *info)
150 if (isam_p_indx == isam_p_size)
152 ISAM_P *new_isam_p_buf;
154 isam_p_size = 2*isam_p_size + 100;
155 new_isam_p_buf = xmalloc (sizeof(*new_isam_p_buf) *
159 memcpy (new_isam_p_buf, isam_p_buf,
160 isam_p_indx * sizeof(*isam_p_buf));
163 isam_p_buf = new_isam_p_buf;
165 assert (*info == sizeof(*isam_p_buf));
166 memcpy (isam_p_buf + isam_p_indx, info+1, sizeof(*isam_p_buf));
170 static int grep_handle (Dict_char *name, const char *info)
172 logf (LOG_DEBUG, "dict name: %s", name);
177 static int trunc_term (ZServerInfo *zi, Z_AttributesPlusTerm *zapt,
179 int *no, int split_flag)
181 char termz[IT_MAX_WORD+1];
182 char term_sub[IT_MAX_WORD+1];
183 char term_dict[2*IT_MAX_WORD+2];
185 char *p0 = termz, *p1 = NULL;
188 int truncation_value;
189 Z_Term *term = zapt->term;
192 attr_init (&truncation, zapt, 5);
193 truncation_value = attr_find (&truncation);
194 logf (LOG_DEBUG, "truncation value %d", truncation_value);
196 if (term->which != Z_Term_general)
201 sizez = term->u.general->len;
202 if (sizez > IT_MAX_WORD)
204 for (i = 0; i<sizez; i++)
205 termz[i] = index_char_cvt (term->u.general->buf[i]);
210 if (split_flag && (p1 = strchr (p0, ' ')))
212 memcpy (term_sub, p0, p1-p0);
213 term_sub[p1-p0] = '\0';
216 strcpy (term_sub, p0);
217 switch (truncation_value)
219 case -1: /* not specified */
220 case 100: /* do not truncate */
221 strcpy (term_dict, term_sub);
222 logf (LOG_DEBUG, "dict_lookup: %s", term_dict);
223 if ((info = dict_lookup (zi->wordDict, term_dict)))
226 case 1: /* right truncation */
227 strcpy (term_dict, term_sub);
228 strcat (term_dict, ".*");
229 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
231 case 2: /* left truncation */
232 case 3: /* left&right truncation */
235 case 101: /* process # in term */
236 for (j = 0, i = 0; term_sub[i] && i < 3; i++)
237 term_dict[j++] = term_sub[i];
238 for (; term_sub[i]; i++)
239 if (term_sub[i] == '#')
241 term_dict[j++] = '.';
242 term_dict[j++] = '*';
245 term_dict[j++] = term_sub[i];
247 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
249 case 102: /* regular expression */
250 strcpy (term_dict, term_sub);
251 dict_lookup_grep (zi->wordDict, term_dict, 0, grep_handle);
258 *isam_ps = isam_p_buf;
260 logf (LOG_DEBUG, "%d positions", *no);
264 static RSET rpn_search_APT_relevance (ZServerInfo *zi,
265 Z_AttributesPlusTerm *zapt)
267 rset_relevance_parms parms;
269 parms.key_size = sizeof(struct it_key);
271 parms.cmp = key_compare;
272 parms.is = zi->wordIsam;
273 if (trunc_term (zi, zapt, &parms.isam_positions,
274 &parms.no_isam_positions, 1))
276 if (parms.no_isam_positions > 0)
277 return rset_create (rset_kind_relevance, &parms);
279 return rset_create (rset_kind_null, NULL);
282 static RSET rpn_search_APT_word (ZServerInfo *zi,
283 Z_AttributesPlusTerm *zapt)
285 ISAM_P *isam_positions;
286 int no_isam_positions;
287 rset_isam_parms parms;
289 if (trunc_term (zi, zapt, &isam_positions,
290 &no_isam_positions, 0))
292 if (no_isam_positions != 1)
293 return rset_create (rset_kind_null, NULL);
294 parms.is = zi->wordIsam;
295 parms.pos = *isam_positions;
296 return rset_create (rset_kind_isam, &parms);
299 static RSET rpn_search_APT_phrase (ZServerInfo *zi,
300 Z_AttributesPlusTerm *zapt)
302 ISAM_P *isam_positions;
303 int no_isam_positions;
304 rset_isam_parms parms;
306 if (trunc_term (zi, zapt, &isam_positions,
307 &no_isam_positions, 1))
309 if (no_isam_positions != 1)
310 return rset_create (rset_kind_null, NULL);
311 parms.is = zi->wordIsam;
312 parms.pos = *isam_positions;
313 return rset_create (rset_kind_isam, &parms);
316 static RSET rpn_search_APT (ZServerInfo *zi, Z_AttributesPlusTerm *zapt)
320 int relation_value, structure_value;
322 attr_init (&relation, zapt, 2);
323 attr_init (&structure, zapt, 4);
325 relation_value = attr_find (&relation);
326 structure_value = attr_find (&structure);
327 switch (structure_value)
330 if (relation_value == 102) /* relevance relation */
331 return rpn_search_APT_relevance (zi, zapt);
332 return rpn_search_APT_word (zi, zapt);
334 if (relation_value == 102) /* relevance relation */
335 return rpn_search_APT_relevance (zi, zapt);
336 return rpn_search_APT_phrase (zi, zapt);
339 if (relation_value == 102) /* relevance relation */
340 return rpn_search_APT_relevance (zi, zapt);
341 return rpn_search_APT_word (zi, zapt);
346 case 5: /* date - normalized */
348 case 6: /* word list */
349 return rpn_search_APT_relevance (zi, zapt);
350 case 100: /* date - un-normalized */
352 case 101: /* name - normalized */
354 case 102: /* date - un-normalized */
356 case 103: /* structure */
360 case 105: /* free-form-text */
361 return rpn_search_APT_relevance (zi, zapt);
362 case 106: /* document-text */
363 return rpn_search_APT_relevance (zi, zapt);
364 case 107: /* local-number */
366 case 108: /* string */
367 return rpn_search_APT_word (zi, zapt);
368 case 109: /* numeric string */
375 static RSET rpn_search_ref (ZServerInfo *zi, Z_ResultSetId *resultSetId)
379 if (!(s = resultSetGet (zi, resultSetId)))
380 return rset_create (rset_kind_null, NULL);
384 static RSET rpn_search_structure (ZServerInfo *zi, Z_RPNStructure *zs)
387 if (zs->which == Z_RPNStructure_complex)
389 rset_bool_parms bool_parms;
391 bool_parms.rset_l = rpn_search_structure (zi, zs->u.complex->s1);
392 if (bool_parms.rset_l == NULL)
394 bool_parms.rset_r = rpn_search_structure (zi, zs->u.complex->s2);
395 if (bool_parms.rset_r == NULL)
397 rset_delete (bool_parms.rset_l);
400 bool_parms.key_size = sizeof(struct it_key);
401 bool_parms.cmp = key_compare;
403 switch (zs->u.complex->operator->which)
406 r = rset_create (rset_kind_and, &bool_parms);
409 r = rset_create (rset_kind_or, &bool_parms);
411 case Z_Operator_and_not:
412 r = rset_create (rset_kind_not, &bool_parms);
418 else if (zs->which == Z_RPNStructure_simple)
420 if (zs->u.simple->which == Z_Operand_APT)
422 logf (LOG_DEBUG, "rpn_search_APT");
423 r = rpn_search_APT (zi, zs->u.simple->u.attributesPlusTerm);
425 else if (zs->u.simple->which == Z_Operand_resultSetId)
427 logf (LOG_DEBUG, "rpn_search_ref");
428 r = rpn_search_ref (zi, zs->u.simple->u.resultSetId);
442 static void count_set (RSET r, int *count)
448 logf (LOG_DEBUG, "rpn_save_set");
450 rfd = rset_open (r, 0);
451 while (rset_read (r, rfd, &key))
453 if (key.sysno != psysno)
460 logf (LOG_DEBUG, "%d distinct sysnos", *count);
463 int rpn_search (ZServerInfo *zi,
464 Z_RPNQuery *rpn, int num_bases, char **basenames,
465 const char *setname, int *hits)
470 zi->errString = NULL;
471 rset = rpn_search_structure (zi, rpn->RPNStructure);
474 count_set (rset, hits);
475 resultSetAdd (zi, setname, 1, rset);