1 | /**********************************************************************
|
---|
2 | *
|
---|
3 | * queryaction.cpp --
|
---|
4 | * Copyright (C) 1999 The New Zealand Digital Library Project
|
---|
5 | *
|
---|
6 | * A component of the Greenstone digital library software
|
---|
7 | * from the New Zealand Digital Library Project at the
|
---|
8 | * University of Waikato, New Zealand.
|
---|
9 | *
|
---|
10 | * This program is free software; you can redistribute it and/or modify
|
---|
11 | * it under the terms of the GNU General Public License as published by
|
---|
12 | * the Free Software Foundation; either version 2 of the License, or
|
---|
13 | * (at your option) any later version.
|
---|
14 | *
|
---|
15 | * This program is distributed in the hope that it will be useful,
|
---|
16 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
17 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
18 | * GNU General Public License for more details.
|
---|
19 | *
|
---|
20 | * You should have received a copy of the GNU General Public License
|
---|
21 | * along with this program; if not, write to the Free Software
|
---|
22 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
23 | *
|
---|
24 | *********************************************************************/
|
---|
25 |
|
---|
26 | #include "queryaction.h"
|
---|
27 | #include "querytools.h"
|
---|
28 | #include "formattools.h"
|
---|
29 | #include "cgiutils.h"
|
---|
30 | #include "OIDtools.h"
|
---|
31 | #include "fileutil.h"
|
---|
32 | #include "text_t.h"
|
---|
33 | #include "historydb.h"
|
---|
34 | #include "htmlutils.h" // for html_safe in do_action
|
---|
35 | #include "gsdltools.h"
|
---|
36 | #include "phrases.h" // for get_phrases
|
---|
37 | #include <stdlib.h> // for strtol
|
---|
38 | #include <assert.h>
|
---|
39 |
|
---|
40 |
|
---|
41 | queryaction::queryaction ()
|
---|
42 | : basequeryaction()
|
---|
43 | {
|
---|
44 | num_phrases = 0;
|
---|
45 |
|
---|
46 | cgiarginfo arg_ainfo;
|
---|
47 |
|
---|
48 | // this action uses cgi variable "a"
|
---|
49 | arg_ainfo.shortname = "a";
|
---|
50 | arg_ainfo.longname = "action";
|
---|
51 | arg_ainfo.multiplechar = true;
|
---|
52 | arg_ainfo.multiplevalue = false;
|
---|
53 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
54 | arg_ainfo.argdefault = "q";
|
---|
55 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
56 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
57 |
|
---|
58 | // "ct" - 0 = mg, 1 = mgpp, 2=lucene
|
---|
59 | arg_ainfo.shortname = "ct";
|
---|
60 | arg_ainfo.longname = "collection type";
|
---|
61 | arg_ainfo.multiplechar = true; // can be empty or single char
|
---|
62 | arg_ainfo.multiplevalue = false;
|
---|
63 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
64 | arg_ainfo.argdefault = g_EmptyText;
|
---|
65 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
66 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
67 |
|
---|
68 | // "b" - 0 = simple, 1 = advanced
|
---|
69 | arg_ainfo.shortname = "b";
|
---|
70 | arg_ainfo.longname = "query mode";
|
---|
71 | arg_ainfo.multiplechar = false;
|
---|
72 | arg_ainfo.multiplevalue = false;
|
---|
73 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
74 | arg_ainfo.argdefault = "0";
|
---|
75 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
76 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
77 |
|
---|
78 | // "h"
|
---|
79 | arg_ainfo.shortname = "h";
|
---|
80 | arg_ainfo.longname = "main index";
|
---|
81 | arg_ainfo.multiplechar = true;
|
---|
82 | arg_ainfo.multiplevalue = false;
|
---|
83 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
84 | arg_ainfo.argdefault = g_EmptyText;
|
---|
85 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
86 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
87 |
|
---|
88 | // "h2"
|
---|
89 | arg_ainfo.shortname = "h2";
|
---|
90 | arg_ainfo.longname = "main index for second query";
|
---|
91 | arg_ainfo.multiplechar = true;
|
---|
92 | arg_ainfo.multiplevalue = false;
|
---|
93 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
94 | arg_ainfo.argdefault = g_EmptyText;
|
---|
95 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
96 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
97 |
|
---|
98 | // "j"
|
---|
99 | arg_ainfo.shortname = "j";
|
---|
100 | arg_ainfo.longname = "sub collection index";
|
---|
101 | arg_ainfo.multiplechar = true;
|
---|
102 | arg_ainfo.multiplevalue = false;
|
---|
103 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
104 | arg_ainfo.argdefault = g_EmptyText;
|
---|
105 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
106 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
107 |
|
---|
108 | // "j2"
|
---|
109 | arg_ainfo.shortname = "j2";
|
---|
110 | arg_ainfo.longname = "sub collection index for second query";
|
---|
111 | arg_ainfo.multiplechar = true;
|
---|
112 | arg_ainfo.multiplevalue = false;
|
---|
113 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
114 | arg_ainfo.argdefault = g_EmptyText;
|
---|
115 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
116 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
117 |
|
---|
118 | // "n"
|
---|
119 | arg_ainfo.shortname = "n";
|
---|
120 | arg_ainfo.longname = "language index";
|
---|
121 | arg_ainfo.multiplechar = true;
|
---|
122 | arg_ainfo.multiplevalue = false;
|
---|
123 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
124 | arg_ainfo.argdefault = g_EmptyText;
|
---|
125 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
126 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
127 |
|
---|
128 | // "n2"
|
---|
129 | arg_ainfo.shortname = "n2";
|
---|
130 | arg_ainfo.longname = "language index for second query";
|
---|
131 | arg_ainfo.multiplechar = true;
|
---|
132 | arg_ainfo.multiplevalue = false;
|
---|
133 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
134 | arg_ainfo.argdefault = g_EmptyText;
|
---|
135 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
136 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
137 |
|
---|
138 |
|
---|
139 | // "t" - 1 = ranked 0 = boolean
|
---|
140 | arg_ainfo.shortname = "t";
|
---|
141 | arg_ainfo.longname = "search type";
|
---|
142 | arg_ainfo.multiplechar = false;
|
---|
143 | arg_ainfo.multiplevalue = false;
|
---|
144 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
145 | arg_ainfo.argdefault = "1";
|
---|
146 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
147 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
148 |
|
---|
149 | // "k"
|
---|
150 | arg_ainfo.shortname = "k";
|
---|
151 | arg_ainfo.longname = "casefolding";
|
---|
152 | arg_ainfo.multiplechar = false;
|
---|
153 | arg_ainfo.multiplevalue = false;
|
---|
154 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
155 | arg_ainfo.argdefault = "1";
|
---|
156 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
157 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
158 |
|
---|
159 | // "ks"
|
---|
160 | arg_ainfo.shortname = "ks";
|
---|
161 | arg_ainfo.longname = "casefolding support";
|
---|
162 | arg_ainfo.multiplechar = false;
|
---|
163 | arg_ainfo.multiplevalue = false;
|
---|
164 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
165 | arg_ainfo.argdefault = "0";
|
---|
166 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
167 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
168 |
|
---|
169 | // "s"
|
---|
170 | arg_ainfo.shortname = "s";
|
---|
171 | arg_ainfo.longname = "stemming";
|
---|
172 | arg_ainfo.multiplechar = false;
|
---|
173 | arg_ainfo.multiplevalue = false;
|
---|
174 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
175 | arg_ainfo.argdefault = "0";
|
---|
176 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
177 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
178 |
|
---|
179 | // "ss"
|
---|
180 | arg_ainfo.shortname = "ss";
|
---|
181 | arg_ainfo.longname = "stemming support";
|
---|
182 | arg_ainfo.multiplechar = false;
|
---|
183 | arg_ainfo.multiplevalue = false;
|
---|
184 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
185 | arg_ainfo.argdefault = "0";
|
---|
186 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
187 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
188 |
|
---|
189 | // "af"
|
---|
190 | arg_ainfo.shortname = "af";
|
---|
191 | arg_ainfo.longname = "accentfolding";
|
---|
192 | arg_ainfo.multiplechar = false;
|
---|
193 | arg_ainfo.multiplevalue = false;
|
---|
194 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
195 | arg_ainfo.argdefault = "0";
|
---|
196 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
197 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
198 |
|
---|
199 | // "afs"
|
---|
200 | arg_ainfo.shortname = "afs";
|
---|
201 | arg_ainfo.longname = "accentfolding support";
|
---|
202 | arg_ainfo.multiplechar = false;
|
---|
203 | arg_ainfo.multiplevalue = false;
|
---|
204 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
205 | arg_ainfo.argdefault = "0";
|
---|
206 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
207 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
208 |
|
---|
209 |
|
---|
210 | // "ccs"
|
---|
211 | arg_ainfo.shortname = "ccs";
|
---|
212 | arg_ainfo.longname = "cross collection searching";
|
---|
213 | arg_ainfo.multiplechar = false;
|
---|
214 | arg_ainfo.multiplevalue = false;
|
---|
215 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
216 | arg_ainfo.argdefault = "0";
|
---|
217 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
218 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
219 |
|
---|
220 | // "ccp"
|
---|
221 | arg_ainfo.shortname = "ccp";
|
---|
222 | arg_ainfo.longname = "cross collection page";
|
---|
223 | arg_ainfo.multiplechar = false;
|
---|
224 | arg_ainfo.multiplevalue = false;
|
---|
225 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
226 | arg_ainfo.argdefault = "0";
|
---|
227 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
228 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
229 |
|
---|
230 | // "g" - new arg for granularity, for mgpp collections
|
---|
231 | arg_ainfo.shortname = "g";
|
---|
232 | arg_ainfo.longname = "granularity";
|
---|
233 | arg_ainfo.multiplechar = true;
|
---|
234 | arg_ainfo.multiplevalue = false;
|
---|
235 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
236 | arg_ainfo.argdefault = g_EmptyText;
|
---|
237 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
238 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
239 |
|
---|
240 | // "ds" - start date
|
---|
241 | arg_ainfo.shortname = "ds";
|
---|
242 | arg_ainfo.longname = "start date";
|
---|
243 | arg_ainfo.multiplechar = true;
|
---|
244 | arg_ainfo.multiplevalue = false;
|
---|
245 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
246 | arg_ainfo.argdefault = g_EmptyText;
|
---|
247 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
248 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
249 |
|
---|
250 | // "de" - end date
|
---|
251 | arg_ainfo.shortname = "de";
|
---|
252 | arg_ainfo.longname = "end date";
|
---|
253 | arg_ainfo.multiplechar = true;
|
---|
254 | arg_ainfo.multiplevalue = false;
|
---|
255 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
256 | arg_ainfo.argdefault = g_EmptyText;
|
---|
257 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
258 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
259 |
|
---|
260 | // "dsbc" - whether or not start date is prechristian
|
---|
261 | arg_ainfo.shortname = "dsbc";
|
---|
262 | arg_ainfo.longname = "start date bc";
|
---|
263 | arg_ainfo.multiplechar = false;
|
---|
264 | arg_ainfo.multiplevalue = false;
|
---|
265 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
266 | arg_ainfo.argdefault = "0";
|
---|
267 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
268 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
269 |
|
---|
270 | // "debc" - whether or not end date is prechristian
|
---|
271 | arg_ainfo.shortname = "debc";
|
---|
272 | arg_ainfo.longname = "end date bc";
|
---|
273 | arg_ainfo.multiplechar = false;
|
---|
274 | arg_ainfo.multiplevalue = false;
|
---|
275 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
276 | arg_ainfo.argdefault = "0";
|
---|
277 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
278 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
279 |
|
---|
280 | // "qt" - 0 = text, 1 = form
|
---|
281 | arg_ainfo.shortname = "qt";
|
---|
282 | arg_ainfo.longname = "query type";
|
---|
283 | arg_ainfo.multiplechar = true; // can be empty or single char
|
---|
284 | arg_ainfo.multiplevalue = false;
|
---|
285 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
286 | arg_ainfo.argdefault = g_EmptyText;
|
---|
287 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
288 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
289 |
|
---|
290 | // "qto" - 1 = text only, 2 = form only, 3 = text and form
|
---|
291 | arg_ainfo.shortname = "qto";
|
---|
292 | arg_ainfo.longname = "query type options";
|
---|
293 | arg_ainfo.multiplechar = true; // can be empty or single char
|
---|
294 | arg_ainfo.multiplevalue = false;
|
---|
295 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
296 | arg_ainfo.argdefault = g_EmptyText;
|
---|
297 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
298 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
299 |
|
---|
300 | // "qb" - 0 = regular, 1 = large
|
---|
301 | arg_ainfo.shortname = "qb";
|
---|
302 | arg_ainfo.longname = "query box type";
|
---|
303 | arg_ainfo.multiplechar = false;
|
---|
304 | arg_ainfo.multiplevalue = false;
|
---|
305 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
306 | arg_ainfo.argdefault = "0";
|
---|
307 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
308 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
309 |
|
---|
310 |
|
---|
311 | // "fqs" - the list of stemming options in the form query
|
---|
312 | // - a comma separated list
|
---|
313 | arg_ainfo.shortname = "fqs";
|
---|
314 | arg_ainfo.longname = "form query stems";
|
---|
315 | arg_ainfo.multiplechar = true;
|
---|
316 | arg_ainfo.multiplevalue = false;
|
---|
317 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
318 | arg_ainfo.argdefault = g_EmptyText;
|
---|
319 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
320 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
321 |
|
---|
322 |
|
---|
323 | // "fqk" - the list of casefolding options in the form query
|
---|
324 | // - a comma separated list
|
---|
325 | arg_ainfo.shortname = "fqk";
|
---|
326 | arg_ainfo.longname = "form query casefolds";
|
---|
327 | arg_ainfo.multiplechar = true;
|
---|
328 | arg_ainfo.multiplevalue = false;
|
---|
329 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
330 | arg_ainfo.argdefault = g_EmptyText;
|
---|
331 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
332 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
333 |
|
---|
334 | // "fqaf" - the list of accentfolding options in the form query
|
---|
335 | // - a comma separated list
|
---|
336 | arg_ainfo.shortname = "fqaf";
|
---|
337 | arg_ainfo.longname = "form query accentfold";
|
---|
338 | arg_ainfo.multiplechar = true;
|
---|
339 | arg_ainfo.multiplevalue = false;
|
---|
340 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
341 | arg_ainfo.argdefault = g_EmptyText;
|
---|
342 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
343 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
344 |
|
---|
345 | // "cc"
|
---|
346 | arg_ainfo.shortname = "cc";
|
---|
347 | arg_ainfo.longname = "collections to search";
|
---|
348 | arg_ainfo.multiplechar = true;
|
---|
349 | arg_ainfo.multiplevalue = true;
|
---|
350 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
351 | arg_ainfo.argdefault = g_EmptyText;
|
---|
352 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
353 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
354 |
|
---|
355 |
|
---|
356 | // ****
|
---|
357 | // should this even be here???
|
---|
358 | // seems to be mixed up between "sf" and "sqlsf"
|
---|
359 |
|
---|
360 | // "sf" - Sort field. Set to field to be used for sorting search reult
|
---|
361 | // set (only implemented for lucene collections at present).
|
---|
362 | arg_ainfo.shortname = "sqlsf";
|
---|
363 | arg_ainfo.longname = "sql sort field";
|
---|
364 | arg_ainfo.multiplechar = true;
|
---|
365 | arg_ainfo.multiplevalue = false;
|
---|
366 | arg_ainfo.defaultstatus = cgiarginfo::weak;
|
---|
367 | arg_ainfo.argdefault = g_EmptyText;
|
---|
368 | arg_ainfo.savedarginfo = cgiarginfo::must;
|
---|
369 | argsinfo.addarginfo (NULL, arg_ainfo);
|
---|
370 |
|
---|
371 |
|
---|
372 | }
|
---|
373 |
|
---|
374 | queryaction::~queryaction ()
|
---|
375 | {
|
---|
376 | }
|
---|
377 |
|
---|
378 | void queryaction::configure (const text_t &key, const text_tarray &cfgline) {
|
---|
379 | basequeryaction::configure (key, cfgline);
|
---|
380 | }
|
---|
381 |
|
---|
382 | bool queryaction::init (ostream &logout) {
|
---|
383 | return basequeryaction::init (logout);
|
---|
384 | }
|
---|
385 |
|
---|
386 | bool queryaction::check_cgiargs (cgiargsinfoclass &argsinfo, cgiargsclass &args,
|
---|
387 | recptprotolistclass* protos, ostream &logout) {
|
---|
388 |
|
---|
389 | // check t argument
|
---|
390 | int arg_t = args.getintarg("t");
|
---|
391 | if (arg_t != 0 && arg_t != 1) {
|
---|
392 | logout << "Warning: \"t\" argument out of range (" << arg_t << ")\n";
|
---|
393 | cgiarginfo *tinfo = argsinfo.getarginfo ("t");
|
---|
394 | if (tinfo != NULL) args["t"] = tinfo->argdefault;
|
---|
395 | }
|
---|
396 |
|
---|
397 | // check k argument
|
---|
398 | int arg_k = args.getintarg("k");
|
---|
399 | if (arg_k != 0 && arg_k != 1) {
|
---|
400 | logout << "Warning: \"k\" argument out of range (" << arg_k << ")\n";
|
---|
401 | cgiarginfo *kinfo = argsinfo.getarginfo ("k");
|
---|
402 | if (kinfo != NULL) args["k"] = kinfo->argdefault;
|
---|
403 | }
|
---|
404 |
|
---|
405 | // check s argument
|
---|
406 | int arg_s = args.getintarg("s");
|
---|
407 | if (arg_s != 0 && arg_s != 1) {
|
---|
408 | logout << "Warning: \"s\" argument out of range (" << arg_s << ")\n";
|
---|
409 | cgiarginfo *sinfo = argsinfo.getarginfo ("s");
|
---|
410 | if (sinfo != NULL) args["s"] = sinfo->argdefault;
|
---|
411 | }
|
---|
412 |
|
---|
413 |
|
---|
414 | // check ct argument
|
---|
415 | int arg_ct = args.getintarg("ct");
|
---|
416 | if (arg_ct < 0 || arg_ct > 2) {
|
---|
417 | logout << "Warning: \"ct\" argument out of range (" << arg_ct << ")\n";
|
---|
418 | cgiarginfo *ctinfo = argsinfo.getarginfo ("ct");
|
---|
419 | if (ctinfo != NULL) args["ct"] = ctinfo->argdefault;
|
---|
420 | }
|
---|
421 |
|
---|
422 | // check qt argument
|
---|
423 | int arg_qt = args.getintarg("qt");
|
---|
424 | if (arg_qt<0 || arg_qt>2) {
|
---|
425 | logout << "Warning: \"qt\" argument out of range (" << arg_qt << ")\n";
|
---|
426 | cgiarginfo *qtinfo = argsinfo.getarginfo ("qt");
|
---|
427 | if (qtinfo != NULL) args["qt"] = qtinfo->argdefault;
|
---|
428 | }
|
---|
429 |
|
---|
430 | // check qb argument
|
---|
431 | int arg_qb = args.getintarg("qb");
|
---|
432 | if (arg_qb !=0 && arg_qb !=1) {
|
---|
433 | logout << "Warning: \"qb\" argument out of range (" << arg_qb << ")\n";
|
---|
434 | cgiarginfo *qbinfo = argsinfo.getarginfo ("qb");
|
---|
435 | if (qbinfo != NULL) args["qb"] = qbinfo->argdefault;
|
---|
436 | }
|
---|
437 |
|
---|
438 | // check fqa argument
|
---|
439 | int arg_fqa = args.getintarg("fqa");
|
---|
440 | if (arg_fqa !=0 && arg_fqa !=1) {
|
---|
441 | logout << "Warning: \"fqa\" argument out of range (" << arg_fqa << ")\n";
|
---|
442 | cgiarginfo *fqainfo = argsinfo.getarginfo ("fqa");
|
---|
443 | if (fqainfo != NULL) args["fqa"] = fqainfo->argdefault;
|
---|
444 | }
|
---|
445 |
|
---|
446 | // check fqn argument
|
---|
447 | int arg_fqn = args.getintarg("fqn");
|
---|
448 | if (arg_fqn < -1) {
|
---|
449 | logout << "Warning: \"fqn\" argument less than -1 (" << arg_fqn << ")\n";
|
---|
450 | cgiarginfo *fqninfo = argsinfo.getarginfo ("fqn");
|
---|
451 | if (fqninfo != NULL) args["fqn"] = fqninfo->argdefault;
|
---|
452 | }
|
---|
453 |
|
---|
454 | return basequeryaction::check_cgiargs(argsinfo,args,protos,logout);
|
---|
455 |
|
---|
456 | }
|
---|
457 |
|
---|
458 | void queryaction::define_internal_macros (displayclass &disp, cgiargsclass &args,
|
---|
459 | recptprotolistclass * protos,
|
---|
460 | ostream &logout)
|
---|
461 | {
|
---|
462 | basequeryaction::define_internal_macros(disp,args,protos,logout);
|
---|
463 |
|
---|
464 | define_query_interface(disp, args, protos, logout);
|
---|
465 | }
|
---|
466 |
|
---|
467 | void queryaction::define_query_interface(displayclass &disp,
|
---|
468 | cgiargsclass &args,
|
---|
469 | recptprotolistclass * protos,
|
---|
470 | ostream &logout){
|
---|
471 | text_t collection = args["c"];
|
---|
472 |
|
---|
473 | //check that the protocol is alive
|
---|
474 | recptproto* colproto = protos->getrecptproto (collection, logout);
|
---|
475 | if(colproto == NULL) {
|
---|
476 | logout << "ERROR: Null collection protocol trying to query"
|
---|
477 | << collection.getcstr() << "\n";
|
---|
478 | return;
|
---|
479 | }
|
---|
480 |
|
---|
481 | //check the collection is responding/in place
|
---|
482 | ColInfoResponse_t *colinfo = recpt->get_collectinfo_ptr(colproto, collection,
|
---|
483 | logout);
|
---|
484 | if(colinfo == NULL){
|
---|
485 | logout << "ERROR: Null returned for get_collectinfo_ptr on "
|
---|
486 | << collection.getcstr() << "in queryaction::define_query_interface\n";
|
---|
487 | return;
|
---|
488 | }
|
---|
489 |
|
---|
490 | text_tmap::iterator check = colinfo->format.find("QueryInterface");
|
---|
491 | if(check != colinfo->format.end()){
|
---|
492 | if((*check).second=="DateSearch"){
|
---|
493 | text_t current = "_datesearch_";
|
---|
494 | disp.setmacro("optdatesearch","query",current);
|
---|
495 | }
|
---|
496 | }
|
---|
497 | }
|
---|
498 |
|
---|
499 |
|
---|
500 |
|
---|
501 | void queryaction::define_external_macros (displayclass &disp, cgiargsclass &args,
|
---|
502 | recptprotolistclass *protos, ostream &logout) {
|
---|
503 |
|
---|
504 | // define_external_macros sets the following macros:
|
---|
505 |
|
---|
506 | // some or all of these may not be required to be set
|
---|
507 | // _hselection_, _h2selection_ the selection box for the main part of the index
|
---|
508 | // _jselection_, _j2selection_ the selection box for the subcollection part of the index
|
---|
509 | // _nselection_, _n2selection_ the selection box for the language part of the index
|
---|
510 | // _cq2selection the selection box for combining two queries
|
---|
511 |
|
---|
512 | // _gselection_, the selection box forlevels (mgpp)
|
---|
513 | // _fqfselection_, the selection box for index/fields (mgpp)
|
---|
514 | // can't do anything if collectproto is null (i.e. no collection was specified)
|
---|
515 | recptproto *collectproto = protos->getrecptproto (args["c"], logout);
|
---|
516 | if (collectproto == NULL) return;
|
---|
517 |
|
---|
518 | ColInfoResponse_t *colinfo = recpt->get_collectinfo_ptr(collectproto,
|
---|
519 | args["c"],
|
---|
520 | logout);
|
---|
521 | set_query_type_args(colinfo, args);
|
---|
522 | set_stem_index_args(colinfo, args);
|
---|
523 |
|
---|
524 | comerror_t err;
|
---|
525 | InfoFilterOptionsResponse_t response;
|
---|
526 | InfoFilterOptionsRequest_t request;
|
---|
527 | request.filterName = "QueryFilter";
|
---|
528 |
|
---|
529 | collectproto->get_filteroptions (args["c"], request, response, err, logout);
|
---|
530 | if (err == noError) {
|
---|
531 |
|
---|
532 | FilterOption_tmap::const_iterator it;
|
---|
533 | FilterOption_tmap::const_iterator end = response.filterOptions.end();
|
---|
534 |
|
---|
535 | // _hselection_ and _h2selection_ (Index)
|
---|
536 | it = response.filterOptions.find ("Index");
|
---|
537 | if (it != end) set_option_macro ("h", args["h"], true, false, (*it).second, disp);
|
---|
538 | if (it != end) set_option_macro ("h2", args["h2"], true,false, (*it).second, disp);
|
---|
539 |
|
---|
540 | // _jselection_ and _j2selection_ (Subcollection)
|
---|
541 | it = response.filterOptions.find ("Subcollection");
|
---|
542 | if (it != end) set_option_macro ("j", args["j"], true,false, (*it).second, disp);
|
---|
543 | if (it != end) set_option_macro ("j2", args["j2"], true,false, (*it).second, disp);
|
---|
544 |
|
---|
545 | // _nselection_ and _n2selection_ (Language)
|
---|
546 | it = response.filterOptions.find ("Language");
|
---|
547 | if (it != end) set_option_macro ("n", args["n"], true,false, (*it).second, disp);
|
---|
548 | if (it != end) set_option_macro ("n2", args["n2"], true,false, (*it).second, disp);
|
---|
549 |
|
---|
550 | // _cq2selection_ (CombineQuery)
|
---|
551 | it = response.filterOptions.find ("CombineQuery");
|
---|
552 | if (it != end) set_option_macro ("cq2", args["cq2"], true,false, (*it).second, disp);
|
---|
553 |
|
---|
554 | if ((args["ct"] == "1") || (args["ct"] == "2")) { // mgpp/lucene collections
|
---|
555 | // _gselection_ (Level)
|
---|
556 | it = response.filterOptions.find("Level");
|
---|
557 | if (it!=end) {
|
---|
558 | set_option_macro("g", args["g"], false, false, (*it).second, disp);
|
---|
559 | if (args["qt"]=="1") { // form search
|
---|
560 | set_gformselection_macro(args["g"], (*it).second, disp);
|
---|
561 | }
|
---|
562 | }
|
---|
563 | // _fqfselection_ field list
|
---|
564 | it = response.filterOptions.find("IndexField");
|
---|
565 | if (it!=end) {
|
---|
566 | bool form_search = false;
|
---|
567 | if (args["qto"]=="2" || args["qt"]=="1") {
|
---|
568 | form_search = true;
|
---|
569 | }
|
---|
570 | set_option_macro ("fqf", args["fqf"], true, form_search, (*it).second, disp);
|
---|
571 | }
|
---|
572 | if (args["ct"] == "2") {// lucene
|
---|
573 | it = response.filterOptions.find("SortField");
|
---|
574 | // set the sort field macro
|
---|
575 | set_sfselection_macro(args["sf"], (*it).second, disp);
|
---|
576 | }
|
---|
577 | }
|
---|
578 |
|
---|
579 |
|
---|
580 | // add a queryterms macro for plain version of search terms
|
---|
581 | if (!args["q"].empty()|| !args["fqv"].empty()) {
|
---|
582 | text_t query_arg = "";
|
---|
583 | if (args["qt"]=="0" && args["qto"] != "2") { // normal text search
|
---|
584 | query_arg = args["q"];
|
---|
585 | }
|
---|
586 | else if (args["qt"]=="1" || args["qto"]=="2"){ // form search
|
---|
587 |
|
---|
588 | if (args["b"]=="1" && args["fqa"]=="1") { // explicit query
|
---|
589 | query_arg = args["q"];
|
---|
590 | }
|
---|
591 | else { // form search
|
---|
592 | query_arg = args["fqv"];
|
---|
593 | }
|
---|
594 | }
|
---|
595 | disp.setmacro ("queryterms", displayclass::defaultpackage, get_plain_query_terms(query_arg, args["ct"]));
|
---|
596 | }
|
---|
597 |
|
---|
598 | }
|
---|
599 | } // define external macros
|
---|
600 |
|
---|
601 |
|
---|
602 | void queryaction::set_sfselection_macro(text_t current_value,
|
---|
603 | const FilterOption_t &option,
|
---|
604 | displayclass &disp) {
|
---|
605 |
|
---|
606 | // we need at one sort option here to continue
|
---|
607 | if (option.validValues.size() < 1) {
|
---|
608 | return;
|
---|
609 | }
|
---|
610 | if (option.validValues.size() == 1) {
|
---|
611 | // we don't need a drop down list, just the value
|
---|
612 | text_t value = option.defaultValue;
|
---|
613 | text_t macrovalue = "";
|
---|
614 | if (value == "rank") {
|
---|
615 | macrovalue = "_query:textsortbyrank_";
|
---|
616 | } else if (value == "none") {
|
---|
617 | return; // no sorting is the only option, so don't display anything
|
---|
618 | } else {
|
---|
619 | macrovalue = "_"+value+"_";
|
---|
620 | }
|
---|
621 | disp.setmacro ("sfselection", displayclass::defaultpackage, macrovalue);
|
---|
622 | return;
|
---|
623 | }
|
---|
624 | // if we have more than two options, make a drop down list
|
---|
625 | text_t macrovalue = "<select name=\"sf\">\n";
|
---|
626 |
|
---|
627 | if (current_value.empty()) {
|
---|
628 | current_value = option.defaultValue;
|
---|
629 | }
|
---|
630 | text_tarray::const_iterator thisvalue = option.validValues.begin();
|
---|
631 | text_tarray::const_iterator endvalue = option.validValues.end();
|
---|
632 | // int valid_count = 0;
|
---|
633 | while (thisvalue != endvalue) {
|
---|
634 | macrovalue += "<option value=\"" + *thisvalue + "\"";
|
---|
635 | if (current_value == *thisvalue)
|
---|
636 | macrovalue += " selected";
|
---|
637 | macrovalue += ">";
|
---|
638 | if (*thisvalue == "rank") {
|
---|
639 | macrovalue += "_query:textsortbyrank_";
|
---|
640 | } else if (*thisvalue == "none") {
|
---|
641 | macrovalue += "_query:textsortbynone_";
|
---|
642 | } else {
|
---|
643 | macrovalue += "_" + *thisvalue + "_";
|
---|
644 | }
|
---|
645 | macrovalue += "</option>/n";
|
---|
646 | ++thisvalue;
|
---|
647 | }
|
---|
648 | macrovalue += "</select>";
|
---|
649 | disp.setmacro ("sfselection", displayclass::defaultpackage, macrovalue);
|
---|
650 |
|
---|
651 | }
|
---|
652 |
|
---|
653 | // sets the selection box macro _gformselection_.
|
---|
654 | // the default for _gformselection_ is _gselection_
|
---|
655 | void queryaction::set_gformselection_macro (text_t current_value,
|
---|
656 | const FilterOption_t &option,
|
---|
657 | displayclass &disp) {
|
---|
658 |
|
---|
659 | if (option.validValues.size() <= 1) {
|
---|
660 | return;
|
---|
661 | }
|
---|
662 | // we need to check to see if there is paragraph present
|
---|
663 | text_tarray::const_iterator thisvalue = option.validValues.begin();
|
---|
664 | text_tarray::const_iterator endvalue = option.validValues.end();
|
---|
665 |
|
---|
666 | bool has_paras = false;
|
---|
667 | while (thisvalue != endvalue) {
|
---|
668 | if (*thisvalue == "Para") {
|
---|
669 | has_paras = true;
|
---|
670 | break;
|
---|
671 | }
|
---|
672 | ++thisvalue;
|
---|
673 | }
|
---|
674 | if (!has_paras) return; // there is no difference between the form selection and the normal one
|
---|
675 |
|
---|
676 | if (option.validValues.size() == 2) {
|
---|
677 | // we will only have one value, but we will still put it in as a text string
|
---|
678 | int opt = 0;
|
---|
679 | if (option.validValues[0] == "Para") {
|
---|
680 | opt = 1;
|
---|
681 | }
|
---|
682 | disp.setmacro ("gformselection", displayclass::defaultpackage, "_"+option.validValues[opt]+"_");
|
---|
683 | return;
|
---|
684 | }
|
---|
685 |
|
---|
686 | // there will be a select box
|
---|
687 | text_t macrovalue = "<select name=\"g\">\n";
|
---|
688 |
|
---|
689 | if (current_value.empty()) current_value = option.defaultValue;
|
---|
690 |
|
---|
691 | thisvalue = option.validValues.begin();
|
---|
692 |
|
---|
693 | while (thisvalue != endvalue) {
|
---|
694 | if (*thisvalue != "Para") {
|
---|
695 | macrovalue += "<option value=\"" + *thisvalue + "\"";
|
---|
696 | if (*thisvalue == current_value)
|
---|
697 | macrovalue += " selected";
|
---|
698 | macrovalue += ">_" + *thisvalue + "_\n";
|
---|
699 | }
|
---|
700 | ++thisvalue;
|
---|
701 | }
|
---|
702 | macrovalue += "</select>\n";
|
---|
703 | disp.setmacro ("gformselection", displayclass::defaultpackage, macrovalue);
|
---|
704 | }
|
---|
705 |
|
---|
706 | void queryaction::define_form_macros (displayclass &disp, cgiargsclass &args,
|
---|
707 | recptprotolistclass *protos,
|
---|
708 | ostream &logout)
|
---|
709 | {
|
---|
710 |
|
---|
711 | // defines the following macros
|
---|
712 | // _regformlist_
|
---|
713 | // _advformlist_
|
---|
714 |
|
---|
715 | if (args["ct"]=="0" || args["qto"]=="1" || (args["qto"]=="3" && args["qt"] == "0") ) // mg, or mgpp/lucene with plain only, or mgpp with both, but set to plain
|
---|
716 | return; // dont need these macros
|
---|
717 |
|
---|
718 | text_t form = "";
|
---|
719 | int argfqn = args.getintarg("fqn");
|
---|
720 |
|
---|
721 | if (args["b"] == "1") { // advanced form
|
---|
722 | form += "_firstadvformelement_\n";
|
---|
723 | for (int i=1; i<argfqn; ++i) {
|
---|
724 | form += "_advformelement_\n";
|
---|
725 | }
|
---|
726 | disp.setmacro("advformlist", "query", form);
|
---|
727 | }
|
---|
728 | else { // simple form
|
---|
729 | for (int i=0; i<argfqn; ++i) {
|
---|
730 | form += "_regformelement_\n";
|
---|
731 | }
|
---|
732 | disp.setmacro("regformlist", "query", form);
|
---|
733 | }
|
---|
734 |
|
---|
735 | }
|
---|
736 |
|
---|
737 | void queryaction::output_ccp (cgiargsclass &args, recptprotolistclass *protos,
|
---|
738 | displayclass &disp, outconvertclass &outconvert,
|
---|
739 | ostream &textout, ostream &logout) {
|
---|
740 |
|
---|
741 | ColInfoResponse_t *cinfo = NULL;
|
---|
742 | comerror_t err;
|
---|
743 | InfoFilterOptionsResponse_t fresponse;
|
---|
744 | InfoFilterOptionsRequest_t frequest;
|
---|
745 | frequest.filterName = "QueryFilter";
|
---|
746 |
|
---|
747 | text_t &index = args["h"];
|
---|
748 | text_t &subcollection = args["j"];
|
---|
749 | text_t &language = args["n"];
|
---|
750 |
|
---|
751 | text_tset collections;
|
---|
752 | text_t arg_cc = args["cc"];
|
---|
753 | decode_cgi_arg (arg_cc);
|
---|
754 | splitchar (arg_cc.begin(), arg_cc.end(), ',', collections);
|
---|
755 |
|
---|
756 | textout << outconvert << disp << "_query:header_\n"
|
---|
757 | << "<center>_navigationbar_</center><br>\n"
|
---|
758 | << "<form name=\"QueryForm\" method=\"get\" action=\"_gwcgi_\">\n"
|
---|
759 | << "<input type=\"hidden\" name=\"a\" value=\"q\">\n"
|
---|
760 | << "<input type=\"hidden\" name=\"site\" value=\"_cgiargsiteAttrsafe_\"\n"
|
---|
761 | << "<input type=\"hidden\" name=\"e\" value=\"_compressedoptions_\">\n"
|
---|
762 | << "<input type=\"hidden\" name=\"ccp\" value=\"1\">\n"
|
---|
763 | << "<center><table width=\"_pagewidth_\"><tr valign=\"top\">\n"
|
---|
764 | << "<td>Select collections to search for \"" << encodeForHTML(args["q"])
|
---|
765 | << "\" <i>(index=" << encodeForHTML(index) << " subcollection=" << encodeForHTML(subcollection)
|
---|
766 | << " language=" << encodeForHTML(language) << ")</i></td>\n"
|
---|
767 | << "<td><input type=\"submit\" value=\"_query:textbeginsearch_\"></td>\n"
|
---|
768 | << "</tr></table></center>\n"
|
---|
769 | << "<center><table width=\"_pagewidth_\">\n"
|
---|
770 | << "<tr><td>\n";
|
---|
771 |
|
---|
772 | recptprotolistclass::iterator rprotolist_here = protos->begin();
|
---|
773 | recptprotolistclass::iterator rprotolist_end = protos->end();
|
---|
774 | while (rprotolist_here != rprotolist_end) {
|
---|
775 | if ((*rprotolist_here).p != NULL) {
|
---|
776 |
|
---|
777 | text_tarray collist;
|
---|
778 | (*rprotolist_here).p->get_collection_list (collist, err, logout);
|
---|
779 | if (err == noError) {
|
---|
780 | text_tarray::iterator collist_here = collist.begin();
|
---|
781 | text_tarray::iterator collist_end = collist.end();
|
---|
782 | while (collist_here != collist_end) {
|
---|
783 |
|
---|
784 | cinfo = recpt->get_collectinfo_ptr ((*rprotolist_here).p, *collist_here, logout);
|
---|
785 | // if (err == noError && cinfo.isPublic && (cinfo.buildDate > 0)) {
|
---|
786 | if (cinfo != NULL && (cinfo->buildDate > 0)) {
|
---|
787 |
|
---|
788 | (*rprotolist_here).p->get_filteroptions (*collist_here, frequest, fresponse, err, logout);
|
---|
789 | if (err == noError) {
|
---|
790 |
|
---|
791 | FilterOption_tmap::const_iterator it;
|
---|
792 | FilterOption_tmap::const_iterator end = fresponse.filterOptions.end();
|
---|
793 | if (!index.empty()) {
|
---|
794 | it = fresponse.filterOptions.find ("Index");
|
---|
795 | if (it == end) {++collist_here; continue;}
|
---|
796 | text_tarray::const_iterator there = (*it).second.validValues.begin();
|
---|
797 | text_tarray::const_iterator tend = (*it).second.validValues.end();
|
---|
798 | while (there != tend) {
|
---|
799 | if (*there == index) break;
|
---|
800 | ++there;
|
---|
801 | }
|
---|
802 | if (there == tend) {++collist_here; continue;}
|
---|
803 | }
|
---|
804 | if (!subcollection.empty()) {
|
---|
805 | it = fresponse.filterOptions.find ("Subcollection");
|
---|
806 | if (it == end) {++collist_here; continue;}
|
---|
807 | text_tarray::const_iterator there = (*it).second.validValues.begin();
|
---|
808 | text_tarray::const_iterator tend = (*it).second.validValues.end();
|
---|
809 | while (there != tend) {
|
---|
810 | if (*there == subcollection) break;
|
---|
811 | ++there;
|
---|
812 | }
|
---|
813 | if (there == tend) {++collist_here; continue;}
|
---|
814 | }
|
---|
815 | if (!language.empty()) {
|
---|
816 | it = fresponse.filterOptions.find ("Language");
|
---|
817 | if (it == end) {++collist_here; continue;}
|
---|
818 | text_tarray::const_iterator there = (*it).second.validValues.begin();
|
---|
819 | text_tarray::const_iterator tend = (*it).second.validValues.end();
|
---|
820 | while (there != tend) {
|
---|
821 | if (*there == language) break;
|
---|
822 | ++there;
|
---|
823 | }
|
---|
824 | if (there == tend) {++collist_here; continue;}
|
---|
825 | }
|
---|
826 |
|
---|
827 | // we've got a matching collection
|
---|
828 | textout << outconvert << "<input type=\"checkbox\"";
|
---|
829 |
|
---|
830 | text_tset::const_iterator t = collections.find (*collist_here);
|
---|
831 | if (t != collections.end()) textout << outconvert << " checked";
|
---|
832 |
|
---|
833 | text_t collectionname = cinfo->get_collectionmeta("collectionname", args["l"]);
|
---|
834 | if (collectionname.empty()) {
|
---|
835 | collectionname = *collist_here;
|
---|
836 | }
|
---|
837 | textout << outconvert << disp
|
---|
838 | << " name=\"cc\" value=\"" << *collist_here << "\">"
|
---|
839 | << collectionname << "<br>\n";
|
---|
840 |
|
---|
841 |
|
---|
842 | }
|
---|
843 | }
|
---|
844 | ++collist_here;
|
---|
845 | }
|
---|
846 | }
|
---|
847 | }
|
---|
848 | ++rprotolist_here;
|
---|
849 | }
|
---|
850 | textout << outconvert << disp
|
---|
851 | << "</td></tr></table></center>\n"
|
---|
852 | << "</form>\n"
|
---|
853 | << "_query:footer_\n";
|
---|
854 |
|
---|
855 | }
|
---|
856 |
|
---|
857 | bool queryaction::user_groups_match(const text_t &collection_groups, const text_t &user_groups) {
|
---|
858 |
|
---|
859 | text_tset splitgrps;
|
---|
860 | text_t::const_iterator split_here = collection_groups.begin();
|
---|
861 | text_t::const_iterator split_end = collection_groups.end();
|
---|
862 |
|
---|
863 | splitchar(split_here,split_end,',',splitgrps);
|
---|
864 |
|
---|
865 | text_t::const_iterator ugroup_here = user_groups.begin();
|
---|
866 | text_t::const_iterator ugroup_end = user_groups.end();
|
---|
867 | text_t thisugroup;
|
---|
868 | while (ugroup_here != ugroup_end) {
|
---|
869 | ugroup_here = getdelimitstr (ugroup_here, ugroup_end, ',', thisugroup);
|
---|
870 | if (splitgrps.find(thisugroup) != splitgrps.end() )
|
---|
871 | { // we have permission!
|
---|
872 | return true;
|
---|
873 | }
|
---|
874 | }
|
---|
875 | return false;
|
---|
876 | }
|
---|
877 |
|
---|
878 | // If we are currently authenticated to be in this collection, then check all
|
---|
879 | // collections in the list against the groups of the current user - if there is an overlap of groups, then add the collection into ccs list
|
---|
880 | // If there had been no authentication needed to get to this collection, then
|
---|
881 | // we'll ignore any collections that have collection level authentication
|
---|
882 | void queryaction::validate_ccs_collection_list(cgiargsclass &args, recptprotolistclass *protos, ostream &logout) {
|
---|
883 |
|
---|
884 | text_tarray collections;
|
---|
885 | text_t arg_cc = args["cc"];
|
---|
886 | text_t arg_c = args["c"];
|
---|
887 | decode_cgi_arg (arg_cc);
|
---|
888 | splitchar (arg_cc.begin(), arg_cc.end(), ',', collections);
|
---|
889 | bool currently_authenticated = false;
|
---|
890 | if (!args["uan"].empty()) {
|
---|
891 | // uan=1 means needs authentication. We'll only get here if we have passed authentication, otherwise the page would have been redirected to login page
|
---|
892 | currently_authenticated = true;
|
---|
893 | }
|
---|
894 | args["cc"] = ""; // we will add colls in one by one if they are valid
|
---|
895 | text_tarray::iterator col_here = collections.begin();
|
---|
896 | text_tarray::iterator col_end = collections.end();
|
---|
897 | bool first = true;
|
---|
898 | text_t current_user_name = args["un"];
|
---|
899 | userinfo_t thisuser;
|
---|
900 | if (currently_authenticated) {
|
---|
901 | int status = user_database->get_user_info (current_user_name, thisuser);
|
---|
902 | if (status != ERRNO_SUCCEED) { // something has gone wrong, so assume not
|
---|
903 | // authenticated
|
---|
904 | currently_authenticated = false;
|
---|
905 | }
|
---|
906 | }
|
---|
907 |
|
---|
908 | while (col_here != col_end) {
|
---|
909 | bool include_coll = false;
|
---|
910 | if (*col_here == arg_c) {
|
---|
911 | // current collection must be accessible otherwise we wouldn't be here.
|
---|
912 | include_coll = true;
|
---|
913 | } else {
|
---|
914 | recptproto *collectproto = protos->getrecptproto (*col_here, logout);
|
---|
915 | if (collectproto != NULL) {
|
---|
916 | ColInfoResponse_t *cinfo = recpt->get_collectinfo_ptr (collectproto, *col_here, logout);
|
---|
917 | text_t authenticate = cinfo->authenticate;
|
---|
918 | if (authenticate == "collection") {
|
---|
919 | if (currently_authenticated) {
|
---|
920 | text_t collection_groups = cinfo->auth_group;
|
---|
921 | if (user_groups_match(collection_groups, thisuser.groups)) {
|
---|
922 | include_coll = true;
|
---|
923 | }
|
---|
924 | } // else we'll not include it
|
---|
925 | } else { // not authenticated, or document level authentication - can include in the list
|
---|
926 | include_coll = true;
|
---|
927 | }
|
---|
928 | }
|
---|
929 | }
|
---|
930 | if (include_coll) {
|
---|
931 | if (!first) args["cc"].push_back (',');
|
---|
932 | args["cc"] += *col_here;
|
---|
933 | first = false;
|
---|
934 | }
|
---|
935 |
|
---|
936 | ++col_here;
|
---|
937 | }
|
---|
938 |
|
---|
939 | }
|
---|
940 |
|
---|
941 | bool queryaction::do_action (cgiargsclass &args, recptprotolistclass *protos,
|
---|
942 | browsermapclass *browsers, displayclass &disp,
|
---|
943 | outconvertclass &outconvert, ostream &textout,
|
---|
944 | ostream &logout) {
|
---|
945 |
|
---|
946 | if (recpt == NULL) {
|
---|
947 | logout << "ERROR (queryaction::do_action): This action does not contain information\n"
|
---|
948 | << " about any receptionists. The method set_receptionist was probably\n"
|
---|
949 | << " not called from the module which instantiated this action.\n";
|
---|
950 | return true;
|
---|
951 | }
|
---|
952 |
|
---|
953 | if (args["ccs"] == "1") {
|
---|
954 | if (!args["cc"].empty()) {
|
---|
955 | validate_ccs_collection_list(args, protos, logout); // include only those which current user has access to
|
---|
956 | // query the selected collections
|
---|
957 | text_t::const_iterator b = args["cc"].begin();
|
---|
958 | text_t::const_iterator e = args["cc"].end();
|
---|
959 | if (findchar (b, e, ',') != e) {
|
---|
960 | if (!search_multiple_collections (args, protos, browsers, disp, outconvert,
|
---|
961 | textout, logout)) return false;
|
---|
962 | return true;
|
---|
963 | } else {
|
---|
964 | if (!search_single_collection (args, args["cc"], protos, browsers, disp,
|
---|
965 | outconvert, textout, logout)) return false;
|
---|
966 | return true;
|
---|
967 | }
|
---|
968 | }
|
---|
969 | }
|
---|
970 |
|
---|
971 | // simply query the current collection
|
---|
972 | if (!search_single_collection (args, args["c"], protos, browsers, disp,
|
---|
973 | outconvert, textout, logout)) return false;
|
---|
974 | return true;
|
---|
975 | }
|
---|
976 |
|
---|
977 |
|
---|
978 |
|
---|
979 | // request.filterResultOptions and request.fields (if required) should
|
---|
980 | // be set from the calling code
|
---|
981 | void queryaction::set_queryfilter_options (FilterRequest_t &request,
|
---|
982 | const text_t &querystring,
|
---|
983 | cgiargsclass &args)
|
---|
984 | {
|
---|
985 | set_fulltext_queryfilter_options(request,querystring,args);
|
---|
986 | }
|
---|
987 |
|
---|
988 |
|
---|
989 |
|
---|
990 | void queryaction::set_queryfilter_options (FilterRequest_t &request,
|
---|
991 | const text_t &querystring1,
|
---|
992 | const text_t &querystring2,
|
---|
993 | cgiargsclass &args)
|
---|
994 | {
|
---|
995 | set_fulltext_queryfilter_options(request,querystring1,querystring2,args);
|
---|
996 | }
|
---|
997 |
|
---|
998 |
|
---|
999 |
|
---|
1000 | bool queryaction::search_multiple_collections (cgiargsclass &args, recptprotolistclass *protos,
|
---|
1001 | browsermapclass *browsers, displayclass &disp,
|
---|
1002 | outconvertclass &outconvert, ostream &textout,
|
---|
1003 | ostream &logout) {
|
---|
1004 |
|
---|
1005 | text_tarray collections;
|
---|
1006 |
|
---|
1007 | text_t arg_cc = args["cc"];
|
---|
1008 | decode_cgi_arg (arg_cc);
|
---|
1009 | splitchar (arg_cc.begin(), arg_cc.end(), ',', collections);
|
---|
1010 |
|
---|
1011 | if (collections.empty()) {
|
---|
1012 | logout << "queryaction::search_multiple_collections: No collections "
|
---|
1013 | << "set for doing multiple query - will search current collection\n";
|
---|
1014 | textout << outconvert << disp << "_query:textwarningnocollections_\n";
|
---|
1015 | return search_single_collection (args, args["c"], protos, browsers, disp,
|
---|
1016 | outconvert, textout, logout);
|
---|
1017 | }
|
---|
1018 |
|
---|
1019 | // check the main coll
|
---|
1020 | text_t main_collection = args["c"];
|
---|
1021 | recptproto *collectproto = protos->getrecptproto (main_collection, logout);
|
---|
1022 | if (collectproto == NULL) {
|
---|
1023 | logout << outconvert << "queryaction::search_multiple_collection: " << main_collection
|
---|
1024 | << " collection has a NULL collectproto\n";
|
---|
1025 |
|
---|
1026 | // Display the "this collection is not installed on this system" page
|
---|
1027 | disp.setmacro("cvariable", displayclass::defaultpackage, encodeForHTML(main_collection));
|
---|
1028 | disp.setmacro("content", "query", "<p>_textbadcollection_<p>");
|
---|
1029 |
|
---|
1030 | textout << outconvert << disp << "_query:header_\n"
|
---|
1031 | << "_query:content_\n" << "_query:footer_\n";
|
---|
1032 | return true;
|
---|
1033 | }
|
---|
1034 |
|
---|
1035 | ColInfoResponse_t *cinfo = recpt->get_collectinfo_ptr (collectproto, main_collection, logout);
|
---|
1036 | if (cinfo == NULL) {
|
---|
1037 | logout << "ERROR (query_action::search_multiple_collections): get_collectinfo_ptr returned NULL for '"<<main_collection<<"'\n";
|
---|
1038 | return false;
|
---|
1039 | }
|
---|
1040 |
|
---|
1041 | // get the formatted query string
|
---|
1042 | // what to do about segmentation for multiple colls??
|
---|
1043 | // TODO
|
---|
1044 | bool segment = false;
|
---|
1045 | text_t formattedstring = "";
|
---|
1046 | get_formatted_query_string(formattedstring, segment, args, disp, logout);
|
---|
1047 |
|
---|
1048 | if (formattedstring.empty()) {
|
---|
1049 | // dont bother doing a query if no query string
|
---|
1050 | define_history_macros (disp, args, protos, logout);
|
---|
1051 | textout << outconvert << disp << "_query:header_\n"
|
---|
1052 | << "_query:content_";
|
---|
1053 | textout << outconvert << disp << "_query:footer_";
|
---|
1054 |
|
---|
1055 | return true;
|
---|
1056 | }
|
---|
1057 |
|
---|
1058 |
|
---|
1059 | // queryaction uses "VList" browser to display results,
|
---|
1060 | // a queries clasification is "Search"
|
---|
1061 | text_t browsertype = "VList";
|
---|
1062 | text_t classification = "Search";
|
---|
1063 |
|
---|
1064 | QueryResult_tset results;
|
---|
1065 | map<text_t, colinfo_t, lttext_t> colinfomap;
|
---|
1066 |
|
---|
1067 | comerror_t err;
|
---|
1068 | FilterRequest_t request;
|
---|
1069 | FilterResponse_t response;
|
---|
1070 | request.filterResultOptions = FROID | FRmetadata | FRtermFreq | FRranking;
|
---|
1071 | text_t freqmsg = "_textfreqmsg1_";
|
---|
1072 | int numdocs = 0;
|
---|
1073 | isapprox isApprox = Exact;
|
---|
1074 |
|
---|
1075 |
|
---|
1076 | bool syntax_error = false;
|
---|
1077 |
|
---|
1078 | set_queryfilter_options (request, formattedstring, args);
|
---|
1079 |
|
---|
1080 | // need to retrieve maxdocs matches for each collection
|
---|
1081 | // (will eventually want to tidy this up, do so caching etc.)
|
---|
1082 | OptionValue_t option;
|
---|
1083 | option.name = "StartResults";
|
---|
1084 | option.value = "1";
|
---|
1085 | request.filterOptions.push_back (option);
|
---|
1086 |
|
---|
1087 | option.name = "EndResults";
|
---|
1088 | option.value = args["m"];
|
---|
1089 | request.filterOptions.push_back (option);
|
---|
1090 |
|
---|
1091 | // check the main collection for uniform formatting info - do we use
|
---|
1092 | // individual format statements, or just the main one?
|
---|
1093 | bool use_main_col_format = false;
|
---|
1094 | if (cinfo->ccsOptions & CCSUniformSearchResultsFormatting) {
|
---|
1095 | use_main_col_format = true;
|
---|
1096 | }
|
---|
1097 |
|
---|
1098 | browserclass *bptr = browsers->getbrowser (browsertype);
|
---|
1099 |
|
---|
1100 | request.fields.erase (request.fields.begin(), request.fields.end());
|
---|
1101 | request.getParents = false;
|
---|
1102 | bptr->load_metadata_defaults (request.fields);
|
---|
1103 |
|
---|
1104 | text_t formatstring;
|
---|
1105 | format_t *formatlistptr = new format_t();
|
---|
1106 | if (use_main_col_format) {
|
---|
1107 | // just get one format for main coll and use it for each subcol
|
---|
1108 | if (!get_formatstring (classification, browsertype,
|
---|
1109 | cinfo->format, formatstring)) {
|
---|
1110 | formatstring = bptr->get_default_formatstring();
|
---|
1111 | }
|
---|
1112 |
|
---|
1113 | parse_formatstring (formatstring, formatlistptr, request.fields, request.getParents);
|
---|
1114 | }
|
---|
1115 |
|
---|
1116 | text_tarray::iterator col_here = collections.begin();
|
---|
1117 | text_tarray::iterator col_end = collections.end();
|
---|
1118 |
|
---|
1119 | map<text_t, int, lttext_t> termfreqs;
|
---|
1120 |
|
---|
1121 | ColInfoResponse_t *tmp_cinfo;
|
---|
1122 | while (col_here != col_end) {
|
---|
1123 |
|
---|
1124 | collectproto = protos->getrecptproto (*col_here, logout);
|
---|
1125 | if (collectproto == NULL) {
|
---|
1126 | // skip this collection
|
---|
1127 | logout << outconvert << "queryaction::search_multiple_collections: " << *col_here
|
---|
1128 | << " collection has a NULL collectproto, ignoring\n";
|
---|
1129 | ++col_here;
|
---|
1130 | continue;
|
---|
1131 | }
|
---|
1132 | tmp_cinfo = recpt->get_collectinfo_ptr (collectproto, *col_here, logout);
|
---|
1133 | if (tmp_cinfo == NULL) {
|
---|
1134 | // skip this collection
|
---|
1135 | logout << "ERROR (query_action::search_multiple_collections): get_collectinfo_ptr returned NULL\n";
|
---|
1136 | ++col_here;
|
---|
1137 | continue;
|
---|
1138 | }
|
---|
1139 |
|
---|
1140 | if (!use_main_col_format) {
|
---|
1141 | request.fields.erase (request.fields.begin(), request.fields.end());
|
---|
1142 | request.getParents = false;
|
---|
1143 | bptr->load_metadata_defaults (request.fields);
|
---|
1144 |
|
---|
1145 | // get the formatstring if there is one
|
---|
1146 | if (!get_formatstring (classification, browsertype,
|
---|
1147 | tmp_cinfo->format, formatstring)) {
|
---|
1148 | formatstring = bptr->get_default_formatstring();
|
---|
1149 | }
|
---|
1150 |
|
---|
1151 | formatlistptr = new format_t();
|
---|
1152 | parse_formatstring (formatstring, formatlistptr, request.fields, request.getParents);
|
---|
1153 | }
|
---|
1154 |
|
---|
1155 | colinfo_t thiscolinfo;
|
---|
1156 | thiscolinfo.formatlistptr = formatlistptr;
|
---|
1157 | thiscolinfo.browserptr = bptr;
|
---|
1158 | colinfomap[*col_here] = thiscolinfo;
|
---|
1159 |
|
---|
1160 | // do the query
|
---|
1161 | collectproto->filter (*col_here, request, response, err, logout);
|
---|
1162 | if (err != noError && err != syntaxError) {
|
---|
1163 | outconvertclass text_t2ascii;
|
---|
1164 | logout << text_t2ascii
|
---|
1165 | << "queryaction::search_multiple_collections: call to QueryFilter failed "
|
---|
1166 | << "for " << *col_here << " collection (" << get_comerror_string (err) << ")\n";
|
---|
1167 | return false;
|
---|
1168 | }
|
---|
1169 |
|
---|
1170 | if (err == syntaxError) {
|
---|
1171 | syntax_error = true;
|
---|
1172 | freqmsg = "_textinvalidquery_";
|
---|
1173 | // assume the syntax will be invalid for all colls
|
---|
1174 | break;
|
---|
1175 | }
|
---|
1176 | if (response.error_message == "TOO_MANY_CLAUSES") {
|
---|
1177 | freqmsg = "_textlucenetoomanyclauses_";
|
---|
1178 | break;
|
---|
1179 | }
|
---|
1180 | if (isApprox == Exact)
|
---|
1181 | isApprox = response.isApprox;
|
---|
1182 | else if (isApprox == MoreThan)
|
---|
1183 | if (response.isApprox == Approximate)
|
---|
1184 | isApprox = response.isApprox;
|
---|
1185 |
|
---|
1186 | TermInfo_tarray::const_iterator this_term = response.termInfo.begin();
|
---|
1187 | TermInfo_tarray::const_iterator end_term = response.termInfo.end();
|
---|
1188 | while (this_term != end_term) {
|
---|
1189 | termfreqs[(*this_term).term] += (*this_term).freq;
|
---|
1190 | if ((col_here+1) == col_end) {
|
---|
1191 | freqmsg += (*this_term).term + ": " + termfreqs[(*this_term).term];
|
---|
1192 | if ((this_term+1) != end_term) freqmsg += ", ";
|
---|
1193 | }
|
---|
1194 | ++this_term;
|
---|
1195 | }
|
---|
1196 |
|
---|
1197 | if (response.numDocs > 0) {
|
---|
1198 | numdocs += response.numDocs;
|
---|
1199 |
|
---|
1200 | QueryResult_t thisresult;
|
---|
1201 | thisresult.collection = *col_here;
|
---|
1202 | ResultDocInfo_tarray::iterator doc_here = response.docInfo.begin();
|
---|
1203 | ResultDocInfo_tarray::iterator doc_end = response.docInfo.end();
|
---|
1204 | while (doc_here != doc_end) {
|
---|
1205 | thisresult.doc = *doc_here;
|
---|
1206 | results.insert (thisresult); // this is ordering based on doc rank
|
---|
1207 | ++doc_here;
|
---|
1208 | }
|
---|
1209 | }
|
---|
1210 | ++col_here;
|
---|
1211 | } // for each coll
|
---|
1212 |
|
---|
1213 | // now we have an ordered list of results. If ifl (I feel lucky) is set, then pick out the one we want
|
---|
1214 | if (args["ifl"] == 1 || (args["ifl"] == 2 && numdocs == 1)) {
|
---|
1215 |
|
---|
1216 | //Find whether DocumentSearchResultLinks is enabled
|
---|
1217 | bool show_links = false;
|
---|
1218 | text_tmap::const_iterator format_here = cinfo->format.begin();
|
---|
1219 | text_tmap::const_iterator format_end = cinfo->format.end();
|
---|
1220 |
|
---|
1221 | while (format_here != format_end) {
|
---|
1222 | if (((*format_here).first == "DocumentSearchResultLinks") &&
|
---|
1223 | ((*format_here).second == "true")){
|
---|
1224 | show_links = true;
|
---|
1225 | break;
|
---|
1226 | }
|
---|
1227 | ++format_here;
|
---|
1228 | }
|
---|
1229 |
|
---|
1230 | // which doc do we want?
|
---|
1231 | int docnum;
|
---|
1232 | int ifl;
|
---|
1233 | int srn = 0;
|
---|
1234 | int srp = 0;
|
---|
1235 | if (args["ifl"] == 1) {
|
---|
1236 | ifl = args["ifln"].getint();
|
---|
1237 | docnum = ifl - 1;
|
---|
1238 | if (show_links) {
|
---|
1239 | // set the values for next and prev search result number
|
---|
1240 | srn = ifl + 1;
|
---|
1241 | if (srn > numdocs) {
|
---|
1242 | srn = 0;
|
---|
1243 | }
|
---|
1244 | srp = ifl - 1;
|
---|
1245 | if (srp < 0) {
|
---|
1246 | srp = 0;
|
---|
1247 | }
|
---|
1248 | }
|
---|
1249 | } else {
|
---|
1250 | // we just want the first (and only) result
|
---|
1251 | docnum = 0;
|
---|
1252 | }
|
---|
1253 |
|
---|
1254 | if (docnum >= 0 && docnum < numdocs) {
|
---|
1255 | // get the docnum'th item from the results
|
---|
1256 | QueryResult_tset::iterator res_here = results.begin();
|
---|
1257 | for (int i=0; i< docnum; i++) {
|
---|
1258 | ++res_here;
|
---|
1259 | }
|
---|
1260 |
|
---|
1261 | textout << outconvert << disp
|
---|
1262 | << "Location: _gwcgi_?e=_compressedoptions_&a=d&c="
|
---|
1263 | << (*res_here).collection << "&cl=search&d=" << (*res_here).doc.OID
|
---|
1264 | << "&srn=" << srn << "&srp=" << srp << "\n\n";
|
---|
1265 | textout << flush;
|
---|
1266 |
|
---|
1267 | return true;
|
---|
1268 |
|
---|
1269 | }
|
---|
1270 | }
|
---|
1271 |
|
---|
1272 | if (!args["ifl"].empty()) {
|
---|
1273 | // if we get here, and ifl was set but we haven't output a document, then we'll carry on as if ifl wasn't set. The only catch is that get_cgihead_info won't have
|
---|
1274 | // done the right thing (because ifl was set), so we need to make sure the output is html
|
---|
1275 | textout << "Content-type: text/html\n\n";
|
---|
1276 | }
|
---|
1277 |
|
---|
1278 |
|
---|
1279 | text_t numdocs_t = numdocs;
|
---|
1280 | args["nmd"] = numdocs_t;
|
---|
1281 |
|
---|
1282 | disp.setmacro ("freqmsg", "query", freqmsg);
|
---|
1283 |
|
---|
1284 | define_query_macros( args, disp, numdocs, isApprox);
|
---|
1285 | // save the query if appropriate
|
---|
1286 | save_search_history(args, numdocs, isApprox);
|
---|
1287 | define_history_macros (disp, args, protos, logout);
|
---|
1288 |
|
---|
1289 | textout << outconvert << disp << "_query:header_\n"
|
---|
1290 | << "_query:content_";
|
---|
1291 |
|
---|
1292 | if (!syntax_error) {
|
---|
1293 |
|
---|
1294 | // now go through each result and output it
|
---|
1295 | QueryResult_tset::iterator res_here = results.begin();
|
---|
1296 | QueryResult_tset::iterator res_end = results.end();
|
---|
1297 | text_tset metadata; // empty !!
|
---|
1298 | bool getParents = false; // don't care !!
|
---|
1299 | bool use_table;
|
---|
1300 | ResultDocInfo_t thisdoc;
|
---|
1301 | format_t *formatlistptr = NULL;
|
---|
1302 | browserclass *browserptr = NULL;
|
---|
1303 |
|
---|
1304 | int count = 1;
|
---|
1305 | int firstdoc = args.getintarg("r");
|
---|
1306 | int hitsperpage = args.getintarg("o");
|
---|
1307 | int thislast = firstdoc + (hitsperpage - 1);
|
---|
1308 |
|
---|
1309 | // output results
|
---|
1310 | while (res_here != res_end) {
|
---|
1311 | if (count < firstdoc) {++count; ++res_here; continue;}
|
---|
1312 | if (count > thislast) break;
|
---|
1313 |
|
---|
1314 | formatlistptr = colinfomap[(*res_here).collection].formatlistptr;
|
---|
1315 | browserptr = colinfomap[(*res_here).collection].browserptr;
|
---|
1316 | thisdoc = (*res_here).doc;
|
---|
1317 | use_table = is_table_content (formatlistptr);
|
---|
1318 |
|
---|
1319 | collectproto = protos->getrecptproto ((*res_here).collection, logout);
|
---|
1320 | if (collectproto == NULL) {
|
---|
1321 | logout << outconvert << "queryaction::search_multiple_collections: " << (*res_here).collection
|
---|
1322 | << " collection has a NULL collectproto, ignoring results\n";
|
---|
1323 | ++res_here;
|
---|
1324 | continue;
|
---|
1325 | }
|
---|
1326 |
|
---|
1327 | browserptr->output_section_group (thisdoc, args, (*res_here).collection, 0,
|
---|
1328 | formatlistptr, use_table, metadata, getParents,
|
---|
1329 | collectproto, disp, outconvert, textout, logout);
|
---|
1330 | // textout << outconvert << "(ranking: " << (*res_here).doc.ranking << ")\n";
|
---|
1331 | ++res_here;
|
---|
1332 | ++count;
|
---|
1333 | }
|
---|
1334 | }
|
---|
1335 | textout << outconvert << disp << "_query:footer_";
|
---|
1336 |
|
---|
1337 | // clean up the format_t pointers
|
---|
1338 | map<text_t, colinfo_t, lttext_t>::iterator here = colinfomap.begin();
|
---|
1339 | map<text_t, colinfo_t, lttext_t>::iterator end = colinfomap.end();
|
---|
1340 | while (here != end) {
|
---|
1341 | delete ((*here).second.formatlistptr);
|
---|
1342 | ++here;
|
---|
1343 | }
|
---|
1344 | return true;
|
---|
1345 | }
|
---|
1346 |
|
---|
1347 |
|
---|
1348 | // does the formatting of the query string - either uses q for a text search
|
---|
1349 | // or the form values for an form search
|
---|
1350 | // also adds dates if appropriate in text search
|
---|
1351 | void queryaction::get_formatted_query_string (text_t &formattedstring,
|
---|
1352 | bool segment,
|
---|
1353 | cgiargsclass &args,
|
---|
1354 | displayclass &disp,
|
---|
1355 | ostream &logout) {
|
---|
1356 | if (args["qt"]=="0" && args["qto"] != "2") { // normal text search
|
---|
1357 | formattedstring = args["q"];
|
---|
1358 | // remove & | ! for simple search,do segmentation if necessary
|
---|
1359 | // To url-decode the '&', format_querystring() will call unsafe_cgi_arg() first
|
---|
1360 | format_querystring (formattedstring, args.getintarg("b"), segment);
|
---|
1361 | if (args["ct"]!=0) { // mgpp and lucene - need to add in tag info if appropriate
|
---|
1362 | format_field_info(formattedstring, args["fqf"], args.getintarg("ct"),
|
---|
1363 | args.getintarg("t"), args.getintarg("b"));
|
---|
1364 | }
|
---|
1365 |
|
---|
1366 | add_dates(formattedstring, args.getintarg("ds"), args.getintarg("de"),
|
---|
1367 | args.getintarg("dsbc"), args.getintarg("debc"),
|
---|
1368 | args.getintarg("ct"));
|
---|
1369 | args["q"] = formattedstring;
|
---|
1370 |
|
---|
1371 | }
|
---|
1372 | else if (args["qt"]=="1" || args["qto"]=="2"){ // form search
|
---|
1373 |
|
---|
1374 | if (args["b"]=="1" && args["fqa"]=="1") { // explicit query
|
---|
1375 | formattedstring = args["q"];
|
---|
1376 |
|
---|
1377 | // Replace %22 and %26 with " and & respectively, since these characters have meaning
|
---|
1378 | // in queries: " are used in phrases and & is used in boolean advanced searches.
|
---|
1379 | // For form searches below, unsafe_cgi_arg is called in the parse_..._form() functions
|
---|
1380 |
|
---|
1381 | unsafe_cgi_arg("ALL", formattedstring);
|
---|
1382 | }
|
---|
1383 | else { // form search
|
---|
1384 | if (args["b"]=="0") { // regular form
|
---|
1385 | parse_reg_query_form(formattedstring, args, segment); // will call unsafe_cgi_arg to decode url encoding
|
---|
1386 | }
|
---|
1387 | else { // advanced form
|
---|
1388 | parse_adv_query_form(formattedstring, args, segment); // will call unsafe_cgi_arg to decode url encoding
|
---|
1389 | }
|
---|
1390 | args["q"] = formattedstring;
|
---|
1391 |
|
---|
1392 | // reset the cgiargfqv macro - need to escape any quotes in it
|
---|
1393 | disp.setmacro("cgiargfqv", "query", escape_quotes(args["fqv"]));
|
---|
1394 |
|
---|
1395 | // also reset the _cgiargq_ macro as it has changed now
|
---|
1396 | disp.setmacro("cgiargq", displayclass::defaultpackage, html_safe(args["q"]));
|
---|
1397 |
|
---|
1398 | // reset the compressed options to include the q arg
|
---|
1399 | text_t compressedoptions = recpt->get_compressed_arg(args, logout);
|
---|
1400 | if (!compressedoptions.empty()) {
|
---|
1401 | disp.setmacro ("compressedoptions", displayclass::defaultpackage, dm_safe(compressedoptions));
|
---|
1402 | // need a decoded version of compressedoptions for use within forms
|
---|
1403 | // as browsers encode values from forms before sending to server
|
---|
1404 | // (e.g. %25 becomes %2525)
|
---|
1405 | decode_cgi_arg (compressedoptions);
|
---|
1406 | if (args["w"] == "utf-8") { // if the encoding was utf-8, then compressed options was utf-8, and we need unicode.
|
---|
1407 | // if encoding wasn't utf-8, then compressed options may be screwed up, but seems to work for 8 bit encodings?
|
---|
1408 | compressedoptions = to_uni(compressedoptions);
|
---|
1409 | }
|
---|
1410 |
|
---|
1411 | text_t macrovalue = dm_safe(compressedoptions);
|
---|
1412 | disp.setmacro ("decodedcompressedoptions", displayclass::defaultpackage, macrovalue);
|
---|
1413 | disp.setmacro ("decodedcompressedoptionsAttrsafe", displayclass::defaultpackage, encodeForHTMLAttr(macrovalue));
|
---|
1414 |
|
---|
1415 | }
|
---|
1416 | } // form search
|
---|
1417 | } // args["qt"]=1
|
---|
1418 | else {
|
---|
1419 | logout << "ERROR (queryaction::get_formatted_query_string): querytype not defined\n";
|
---|
1420 | }
|
---|
1421 | }
|
---|
1422 |
|
---|
1423 |
|
---|
1424 | // define_query_macros sets the macros that couldn't be set until the
|
---|
1425 | // query had been done. Those macros are
|
---|
1426 | // _resultline_, _nextfirst_, _nextlast_, _prevfirst_, _prevlast_,
|
---|
1427 | // _thisfirst_, and _thislast_ and _quotedquery_
|
---|
1428 | // this has been simplified so it can be used with both search_single_coll
|
---|
1429 | // and search_multiple_coll
|
---|
1430 | void queryaction::define_query_macros (cgiargsclass &args, displayclass &disp,
|
---|
1431 | int numdocs, isapprox isApprox)
|
---|
1432 | {
|
---|
1433 | // The following 'if' statatment is placed here to be keep the semantics
|
---|
1434 | // the same as the version before basequeryaction was introduced
|
---|
1435 |
|
---|
1436 | if (num_phrases > 0) isApprox = Exact;
|
---|
1437 |
|
---|
1438 | basequeryaction::define_query_macros(args,disp,numdocs,isApprox);
|
---|
1439 |
|
---|
1440 | if (args["ct"]==0) { // mg queries only, not mgpp
|
---|
1441 | // get the quoted bits of the query string and set _quotedquery_
|
---|
1442 | text_tarray phrases;
|
---|
1443 | get_phrases (args["q"], phrases);
|
---|
1444 | num_phrases = phrases.size();
|
---|
1445 | text_tarray::const_iterator phere = phrases.begin();
|
---|
1446 | text_tarray::const_iterator pend = phrases.end();
|
---|
1447 | bool first = true;
|
---|
1448 | text_t quotedquery;
|
---|
1449 | while (phere != pend) {
|
---|
1450 | if (!first)
|
---|
1451 | if ((phere +1) == pend) quotedquery += " and ";
|
---|
1452 | else quotedquery += ", ";
|
---|
1453 |
|
---|
1454 | quotedquery += "\"" + *phere + "\"";
|
---|
1455 | first = false;
|
---|
1456 | ++phere;
|
---|
1457 | }
|
---|
1458 | if (args.getintarg("s") && !quotedquery.empty()) quotedquery += "_textstemon_";
|
---|
1459 | disp.setmacro ("quotedquery", "query", quotedquery);
|
---|
1460 | }
|
---|
1461 |
|
---|
1462 | }
|
---|
1463 |
|
---|
1464 | // should this change for cross coll search??
|
---|
1465 | bool queryaction::save_search_history (cgiargsclass &args, int numdocs,
|
---|
1466 | isapprox isApprox) {
|
---|
1467 | if (args["q"]=="") return true; // null query, dont save
|
---|
1468 | if (args["hs"]=="0") return true; // only save when submit query pressed
|
---|
1469 |
|
---|
1470 | // get userid
|
---|
1471 | text_t userid = args["z"];
|
---|
1472 |
|
---|
1473 | // the number of docs goes on the front of the query string
|
---|
1474 | text_t query = text_t(numdocs);
|
---|
1475 | if (isApprox==MoreThan) { // there were more docs found
|
---|
1476 | query.push_back('+');
|
---|
1477 | }
|
---|
1478 | query += "c="+args["c"];
|
---|
1479 | query += ";h="+args["h"];
|
---|
1480 | query += ";t="+args["t"];
|
---|
1481 | query += ";b="+args["b"];
|
---|
1482 | query += ";j="+args["j"];
|
---|
1483 | query += ";n="+args["n"];
|
---|
1484 | query += ";s="+args["s"];
|
---|
1485 | query += ";k="+args["k"];
|
---|
1486 | query += ";g="+args["g"];
|
---|
1487 |
|
---|
1488 | text_t qstring = args["q"];
|
---|
1489 | //text_t formattedquery =cgi_safe(qstring);
|
---|
1490 | //query += "&q="+formattedquery;
|
---|
1491 | query += ";q="+qstring;
|
---|
1492 | bool display=false;
|
---|
1493 | int hd = args.getintarg("hd");
|
---|
1494 | if (hd > 0) display=true;
|
---|
1495 | if (set_history_info(userid, query, dbhome, display)) return true;
|
---|
1496 | else return false;
|
---|
1497 | }
|
---|
1498 |
|
---|