root/main/trunk/greenstone2/runtime-src/src/recpt/queryaction.cpp @ 28956

Revision 28956, 49.9 KB (checked in by kjdon, 6 years ago)

if no current value, use the default from the option - for sortfield.

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
Line 
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
41queryaction::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
374queryaction::~queryaction ()
375{
376}
377
378void queryaction::configure (const text_t &key, const text_tarray &cfgline) {
379  basequeryaction::configure (key, cfgline);
380}
381
382bool queryaction::init (ostream &logout) {
383  return basequeryaction::init (logout);
384}
385
386bool 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
458void 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
467void 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
501void 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
602void queryaction::set_sfselection_macro(text_t current_value,
603                       const FilterOption_t &option,
604                       displayclass &disp) {
605
606  // we need at least two options here to continue
607  if (option.validValues.size() < 2) {
608    return;
609  }
610
611  text_t macrovalue = "<select name=\"sf\">\n";
612 
613  if (current_value.empty()) {
614    current_value = option.defaultValue;
615  }
616  text_tarray::const_iterator thisvalue = option.validValues.begin();
617  text_tarray::const_iterator endvalue = option.validValues.end();
618  //  int valid_count = 0;
619  while (thisvalue != endvalue) {
620    if (*thisvalue == "rank") {
621       macrovalue += "<option value=\"rank\"";
622       if (current_value == "rank") {
623     macrovalue += " selected";
624       }
625       macrovalue += ">_query:textsortbyrank_\n";
626    } else {
627      macrovalue += "<option value=\"" + *thisvalue + "\"";
628      if (current_value == *thisvalue)
629    macrovalue += " selected";
630      macrovalue += ">_" + *thisvalue + "_\n";
631    }
632    ++thisvalue;
633  }
634  macrovalue += "</select>"; 
635  disp.setmacro ("sfselection", displayclass::defaultpackage, macrovalue);
636 
637}
638
639// sets the selection box macro _gformselection_.
640// the default for _gformselection_ is _gselection_
641void queryaction::set_gformselection_macro (text_t current_value,
642                    const FilterOption_t &option,
643                    displayclass &disp) {
644 
645  if (option.validValues.size() <= 1) {
646    return;
647  }
648  // we need to check to see if there is paragraph present
649  text_tarray::const_iterator thisvalue = option.validValues.begin();
650  text_tarray::const_iterator endvalue = option.validValues.end();
651
652  bool has_paras = false;
653  while (thisvalue != endvalue) {
654    if (*thisvalue == "Para") {
655      has_paras = true;
656      break;
657    }
658    ++thisvalue;
659  }
660  if (!has_paras) return; // there is no difference between the form selection and the normal one
661
662  if (option.validValues.size() == 2) {
663    // we will only have one value, but we will still put it in as a text string
664    int opt = 0;
665    if (option.validValues[0] == "Para") {
666      opt = 1;
667    }
668    disp.setmacro ("gformselection", displayclass::defaultpackage, "_"+option.validValues[opt]+"_");
669    return;
670  }
671
672  // there will be a select box
673  text_t macrovalue = "<select name=\"g\">\n";
674
675  if (current_value.empty()) current_value = option.defaultValue;
676 
677  thisvalue = option.validValues.begin();
678 
679  while (thisvalue != endvalue) {
680    if (*thisvalue != "Para") {
681      macrovalue += "<option value=\"" + *thisvalue + "\"";
682      if (*thisvalue == current_value)
683    macrovalue += " selected";
684      macrovalue += ">_" + *thisvalue + "_\n";
685    }
686    ++thisvalue;
687  }
688  macrovalue += "</select>\n";
689  disp.setmacro ("gformselection", displayclass::defaultpackage, macrovalue);
690}
691
692void queryaction::define_form_macros (displayclass &disp, cgiargsclass &args,
693                      recptprotolistclass *protos,
694                      ostream &logout)
695{
696
697  // defines the following macros
698  // _regformlist_
699  // _advformlist_
700
701  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
702    return; // dont need these macros
703
704  text_t form = "";
705  int argfqn = args.getintarg("fqn");
706 
707  if (args["b"] == "1") { // advanced form
708    form += "_firstadvformelement_\n";
709    for (int i=1; i<argfqn; ++i) {     
710    form += "_advformelement_\n";
711    }
712    disp.setmacro("advformlist", "query", form);
713  }
714  else { // simple form
715    for (int i=0; i<argfqn; ++i) {
716      form += "_regformelement_\n";
717    }
718    disp.setmacro("regformlist", "query", form);
719  }
720 
721}
722
723void queryaction::output_ccp (cgiargsclass &args, recptprotolistclass *protos,
724                  displayclass &disp, outconvertclass &outconvert,
725                  ostream &textout, ostream &logout) {
726
727  ColInfoResponse_t *cinfo = NULL;
728  comerror_t err;
729  InfoFilterOptionsResponse_t fresponse;
730  InfoFilterOptionsRequest_t frequest;
731  frequest.filterName = "QueryFilter";
732
733  text_t &index = args["h"];
734  text_t &subcollection = args["j"];
735  text_t &language = args["n"];
736
737  text_tset collections;
738  text_t arg_cc = args["cc"];
739  decode_cgi_arg (arg_cc);
740  splitchar (arg_cc.begin(), arg_cc.end(), ',', collections);
741
742  textout << outconvert << disp << "_query:header_\n"
743      << "<center>_navigationbar_</center><br>\n"
744      << "<form name=\"QueryForm\" method=\"get\" action=\"_gwcgi_\">\n"
745      << "<input type=\"hidden\" name=\"a\" value=\"q\">\n"
746          << "<input type=\"hidden\" name=\"site\" value=\"_cgiargsiteAttrsafe_\"\n"
747      << "<input type=\"hidden\" name=\"e\" value=\"_compressedoptions_\">\n"
748      << "<input type=\"hidden\" name=\"ccp\" value=\"1\">\n"
749      << "<center><table width=\"_pagewidth_\"><tr valign=\"top\">\n"
750      << "<td>Select collections to search for \"" << encodeForHTML(args["q"])
751      << "\" <i>(index=" << encodeForHTML(index) << " subcollection=" << encodeForHTML(subcollection)
752      << " language=" << encodeForHTML(language) << ")</i></td>\n"
753      << "<td><input type=\"submit\" value=\"_query:textbeginsearch_\"></td>\n"
754      << "</tr></table></center>\n"
755      << "<center><table width=\"_pagewidth_\">\n"
756      << "<tr><td>\n";
757 
758  recptprotolistclass::iterator rprotolist_here = protos->begin();
759  recptprotolistclass::iterator rprotolist_end = protos->end();
760  while (rprotolist_here != rprotolist_end) {
761    if ((*rprotolist_here).p != NULL) {
762     
763      text_tarray collist;
764      (*rprotolist_here).p->get_collection_list (collist, err, logout);
765      if (err == noError) {
766    text_tarray::iterator collist_here = collist.begin();
767    text_tarray::iterator collist_end = collist.end();
768    while (collist_here != collist_end) {
769     
770      cinfo = recpt->get_collectinfo_ptr ((*rprotolist_here).p, *collist_here, logout);
771      //      if (err == noError && cinfo.isPublic && (cinfo.buildDate > 0)) {
772      if (cinfo != NULL && (cinfo->buildDate > 0)) {
773       
774        (*rprotolist_here).p->get_filteroptions (*collist_here, frequest, fresponse, err, logout);
775        if (err == noError) {
776         
777          FilterOption_tmap::const_iterator it;
778          FilterOption_tmap::const_iterator end = fresponse.filterOptions.end();
779          if (!index.empty()) {
780        it = fresponse.filterOptions.find ("Index");
781        if (it == end) {++collist_here; continue;}
782        text_tarray::const_iterator there = (*it).second.validValues.begin();
783        text_tarray::const_iterator tend = (*it).second.validValues.end();
784        while (there != tend) {
785          if (*there == index) break;
786          ++there;
787        }
788        if (there == tend) {++collist_here; continue;}
789          }
790          if (!subcollection.empty()) {
791        it = fresponse.filterOptions.find ("Subcollection");
792        if (it == end) {++collist_here; continue;}
793        text_tarray::const_iterator there = (*it).second.validValues.begin();
794        text_tarray::const_iterator tend = (*it).second.validValues.end();
795        while (there != tend) {
796          if (*there == subcollection) break;
797          ++there;
798        }
799        if (there == tend) {++collist_here; continue;}
800          }
801          if (!language.empty()) {
802        it = fresponse.filterOptions.find ("Language");
803        if (it == end) {++collist_here; continue;}
804        text_tarray::const_iterator there = (*it).second.validValues.begin();
805        text_tarray::const_iterator tend = (*it).second.validValues.end();
806        while (there != tend) {
807          if (*there == language) break;
808          ++there;
809        }
810        if (there == tend) {++collist_here; continue;}
811          }
812       
813          // we've got a matching collection
814          textout << outconvert << "<input type=\"checkbox\"";
815
816          text_tset::const_iterator t = collections.find (*collist_here);
817          if (t != collections.end()) textout << outconvert << " checked";
818         
819          text_t collectionname = cinfo->get_collectionmeta("collectionname", args["l"]);
820          if (collectionname.empty()) {
821        collectionname = *collist_here;
822          }
823          textout << outconvert << disp
824              << " name=\"cc\" value=\"" << *collist_here << "\">"
825              << collectionname << "<br>\n";
826
827         
828        }
829      }
830      ++collist_here;
831    }
832      }
833    }
834    ++rprotolist_here;
835  }
836  textout << outconvert << disp
837      << "</td></tr></table></center>\n"
838      << "</form>\n"
839      << "_query:footer_\n";
840 
841}
842
843bool queryaction::user_groups_match(const text_t &collection_groups, const text_t &user_groups) {
844
845      text_tset splitgrps;
846      text_t::const_iterator split_here = collection_groups.begin();
847      text_t::const_iterator split_end = collection_groups.end();
848
849      splitchar(split_here,split_end,',',splitgrps);
850
851     text_t::const_iterator ugroup_here = user_groups.begin();
852     text_t::const_iterator ugroup_end = user_groups.end();
853     text_t thisugroup;
854    while (ugroup_here != ugroup_end) {
855       ugroup_here = getdelimitstr (ugroup_here, ugroup_end, ',', thisugroup);
856       if (splitgrps.find(thisugroup) != splitgrps.end() )
857         { // we have permission!
858           return true;
859         }
860    }
861    return false;
862}
863
864// If we are currently authenticated to be in this collection, then check all
865// 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
866// If there had been no authentication needed to get to this collection, then
867// we'll ignore any collections that have collection level authentication
868void queryaction::validate_ccs_collection_list(cgiargsclass &args, recptprotolistclass *protos, ostream &logout) {
869
870  text_tarray collections;
871  text_t arg_cc = args["cc"];
872  text_t arg_c = args["c"];
873  decode_cgi_arg (arg_cc);
874  splitchar (arg_cc.begin(), arg_cc.end(), ',', collections);
875  bool currently_authenticated = false;
876  if (!args["uan"].empty()) {
877    // 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
878    currently_authenticated = true;
879  }
880  args["cc"] = ""; // we will add colls in one by one if they are valid
881  text_tarray::iterator col_here = collections.begin();
882  text_tarray::iterator col_end = collections.end();
883  bool first = true;
884  text_t current_user_name = args["un"];
885  userinfo_t thisuser;
886  if (currently_authenticated) {
887    int status = user_database->get_user_info (current_user_name, thisuser);
888    if (status != ERRNO_SUCCEED) { // something has gone wrong, so assume not
889      // authenticated
890      currently_authenticated = false;
891    }
892  }
893 
894  while (col_here != col_end) {
895    bool include_coll = false;
896    if (*col_here == arg_c) {
897      // current collection must be accessible otherwise we wouldn't be here.
898      include_coll = true;
899    } else {
900      recptproto *collectproto = protos->getrecptproto (*col_here, logout);
901      if (collectproto != NULL) {
902    ColInfoResponse_t *cinfo = recpt->get_collectinfo_ptr (collectproto, *col_here, logout);
903    text_t authenticate = cinfo->authenticate;
904    if (authenticate == "collection") {
905      if (currently_authenticated) {
906        text_t collection_groups = cinfo->auth_group;
907        if (user_groups_match(collection_groups, thisuser.groups)) {
908          include_coll = true;
909        }
910      } // else we'll not include it
911    } else { // not authenticated, or document level authentication - can include in the list
912      include_coll = true;
913    }
914      }
915    }
916    if (include_coll) {
917      if (!first) args["cc"].push_back (',');
918      args["cc"] += *col_here;
919      first = false;
920    }
921       
922    ++col_here;
923  }
924
925}
926
927bool queryaction::do_action (cgiargsclass &args, recptprotolistclass *protos,
928                 browsermapclass *browsers, displayclass &disp,
929                 outconvertclass &outconvert, ostream &textout,
930                 ostream &logout) {
931 
932  if (recpt == NULL) {
933    logout << "ERROR (queryaction::do_action): This action does not contain information\n"
934       << "      about any receptionists. The method set_receptionist was probably\n"
935       << "      not called from the module which instantiated this action.\n";
936    return true;
937  }
938
939  if (args["ccs"] == "1") {
940    if (!args["cc"].empty()) {
941      validate_ccs_collection_list(args, protos, logout); // include only those which current user has access to
942      // query the selected collections
943      text_t::const_iterator b = args["cc"].begin();
944      text_t::const_iterator e = args["cc"].end();
945      if (findchar (b, e, ',') != e) {
946    if (!search_multiple_collections (args, protos, browsers, disp, outconvert,
947                      textout, logout)) return false;
948    return true;
949      } else {
950    if (!search_single_collection (args, args["cc"], protos, browsers, disp,
951                       outconvert, textout, logout)) return false;
952    return true;
953      }
954    }
955  }
956
957  // simply query the current collection
958  if (!search_single_collection (args, args["c"], protos, browsers, disp,
959                 outconvert, textout, logout)) return false;
960  return true;
961}
962
963
964
965// request.filterResultOptions and request.fields (if required) should
966// be set from the calling code
967void queryaction::set_queryfilter_options (FilterRequest_t &request,
968                       const text_t &querystring,
969                       cgiargsclass &args)
970{
971  set_fulltext_queryfilter_options(request,querystring,args);
972}
973
974
975
976void queryaction::set_queryfilter_options (FilterRequest_t &request,
977                       const text_t &querystring1,
978                       const text_t &querystring2,
979                       cgiargsclass &args)
980{
981  set_fulltext_queryfilter_options(request,querystring1,querystring2,args);
982}
983
984
985
986bool queryaction::search_multiple_collections (cgiargsclass &args, recptprotolistclass *protos,
987                           browsermapclass *browsers, displayclass &disp,
988                           outconvertclass &outconvert, ostream &textout,
989                           ostream &logout) {
990
991  text_tarray collections;
992 
993  text_t arg_cc = args["cc"];
994  decode_cgi_arg (arg_cc);
995  splitchar (arg_cc.begin(), arg_cc.end(), ',', collections);
996
997  if (collections.empty()) {
998    logout << "queryaction::search_multiple_collections: No collections "
999       << "set for doing multiple query - will search current collection\n";
1000    textout << outconvert << disp << "_query:textwarningnocollections_\n";
1001    return search_single_collection (args, args["c"], protos, browsers, disp,
1002                     outconvert, textout, logout);
1003  }
1004
1005  // check the main coll
1006  text_t main_collection = args["c"];
1007  recptproto *collectproto = protos->getrecptproto (main_collection, logout);
1008  if (collectproto == NULL) {
1009    logout << outconvert << "queryaction::search_multiple_collection: " << main_collection
1010       << " collection has a NULL collectproto\n";
1011
1012    // Display the "this collection is not installed on this system" page
1013    disp.setmacro("cvariable", displayclass::defaultpackage, encodeForHTML(main_collection));
1014    disp.setmacro("content", "query", "<p>_textbadcollection_<p>");
1015
1016    textout << outconvert << disp << "_query:header_\n"
1017        << "_query:content_\n" << "_query:footer_\n";
1018    return true;
1019  }
1020
1021  ColInfoResponse_t *cinfo = recpt->get_collectinfo_ptr (collectproto, main_collection, logout);
1022  if (cinfo == NULL) {
1023    logout << "ERROR (query_action::search_multiple_collections): get_collectinfo_ptr returned NULL for '"<<main_collection<<"'\n";
1024    return false;
1025  }
1026
1027  // get the formatted query string
1028  // what to do about segmentation for multiple colls??
1029  // TODO
1030  bool segment = false;
1031  text_t formattedstring = "";
1032  get_formatted_query_string(formattedstring, segment, args, disp, logout);
1033
1034  if (formattedstring.empty()) {
1035    // dont bother doing a query if no query string
1036    define_history_macros (disp, args, protos, logout);
1037    textout << outconvert << disp << "_query:header_\n"
1038      << "_query:content_";
1039    textout << outconvert << disp << "_query:footer_";
1040
1041    return true;
1042  }
1043
1044
1045  // queryaction uses "VList" browser to display results,
1046  // a queries clasification is "Search"
1047  text_t browsertype = "VList";
1048  text_t classification = "Search";
1049
1050  QueryResult_tset results;
1051  map<text_t, colinfo_t, lttext_t> colinfomap;
1052
1053  comerror_t err;
1054  FilterRequest_t request;
1055  FilterResponse_t response;
1056  request.filterResultOptions = FROID | FRmetadata | FRtermFreq | FRranking;
1057  text_t freqmsg = "_textfreqmsg1_";
1058  int numdocs = 0;
1059  isapprox isApprox = Exact;
1060
1061
1062  bool syntax_error = false;
1063
1064  set_queryfilter_options (request, formattedstring, args);
1065
1066  // need to retrieve maxdocs matches for each collection
1067  // (will eventually want to tidy this up, do so caching etc.)
1068  OptionValue_t option;
1069  option.name = "StartResults";
1070  option.value = "1";
1071  request.filterOptions.push_back (option);
1072 
1073  option.name = "EndResults";
1074  option.value = args["m"];
1075  request.filterOptions.push_back (option);
1076 
1077  // check the main collection for uniform formatting info - do we use
1078  // individual format statements, or just the main one?
1079  bool use_main_col_format = false;
1080  if (cinfo->ccsOptions & CCSUniformSearchResultsFormatting) {
1081    use_main_col_format = true;
1082  }
1083
1084  browserclass *bptr = browsers->getbrowser (browsertype);
1085 
1086  request.fields.erase (request.fields.begin(), request.fields.end());
1087  request.getParents = false;
1088  bptr->load_metadata_defaults (request.fields);
1089 
1090  text_t formatstring;
1091  format_t *formatlistptr = new format_t();
1092  if (use_main_col_format) {
1093    // just get one format for main coll and use it for each subcol
1094    if (!get_formatstring (classification, browsertype,
1095               cinfo->format, formatstring)) {
1096      formatstring = bptr->get_default_formatstring();
1097    }
1098   
1099    parse_formatstring (formatstring, formatlistptr, request.fields, request.getParents);
1100  }
1101 
1102  text_tarray::iterator col_here = collections.begin();
1103  text_tarray::iterator col_end = collections.end();
1104
1105  map<text_t, int, lttext_t> termfreqs;
1106
1107  ColInfoResponse_t *tmp_cinfo;
1108  while (col_here != col_end) {
1109
1110    collectproto = protos->getrecptproto (*col_here, logout);
1111    if (collectproto == NULL) {
1112      // skip this collection
1113      logout << outconvert << "queryaction::search_multiple_collections: " << *col_here
1114         << " collection has a NULL collectproto, ignoring\n";
1115      ++col_here;
1116      continue;
1117    }
1118    tmp_cinfo = recpt->get_collectinfo_ptr (collectproto, *col_here, logout);
1119    if (tmp_cinfo == NULL) {
1120      // skip this collection
1121      logout << "ERROR (query_action::search_multiple_collections): get_collectinfo_ptr returned NULL\n";
1122      ++col_here;
1123      continue;
1124    }
1125   
1126    if (!use_main_col_format) {
1127      request.fields.erase (request.fields.begin(), request.fields.end());
1128      request.getParents = false;
1129      bptr->load_metadata_defaults (request.fields);
1130     
1131      // get the formatstring if there is one
1132      if (!get_formatstring (classification, browsertype,
1133                 tmp_cinfo->format, formatstring)) {
1134    formatstring = bptr->get_default_formatstring();
1135      }
1136
1137      formatlistptr = new format_t();
1138      parse_formatstring (formatstring, formatlistptr, request.fields, request.getParents);
1139    }
1140   
1141    colinfo_t thiscolinfo;
1142    thiscolinfo.formatlistptr = formatlistptr;
1143    thiscolinfo.browserptr = bptr;
1144    colinfomap[*col_here] = thiscolinfo;
1145   
1146    // do the query
1147    collectproto->filter (*col_here, request, response, err, logout);
1148    if (err != noError && err != syntaxError) {
1149      outconvertclass text_t2ascii;
1150      logout << text_t2ascii
1151         << "queryaction::search_multiple_collections: call to QueryFilter failed "
1152         << "for " << *col_here << " collection (" << get_comerror_string (err) << ")\n";
1153      return false;
1154    }
1155
1156    if (err == syntaxError) {
1157      syntax_error = true;
1158      freqmsg = "_textinvalidquery_";
1159      // assume the syntax will be invalid for all colls
1160      break;
1161    }
1162    if (response.error_message == "TOO_MANY_CLAUSES") {
1163      freqmsg = "_textlucenetoomanyclauses_";
1164      break;
1165    }
1166    if (isApprox == Exact)
1167      isApprox = response.isApprox;
1168    else if (isApprox == MoreThan)
1169      if (response.isApprox == Approximate)
1170    isApprox = response.isApprox;
1171
1172    TermInfo_tarray::const_iterator this_term = response.termInfo.begin();
1173    TermInfo_tarray::const_iterator end_term = response.termInfo.end();
1174    while (this_term != end_term) {
1175      termfreqs[(*this_term).term] += (*this_term).freq;
1176      if ((col_here+1) == col_end) {
1177    freqmsg += (*this_term).term + ": " + termfreqs[(*this_term).term];
1178    if ((this_term+1) != end_term) freqmsg += ", ";
1179      }
1180      ++this_term;
1181    }
1182
1183    if (response.numDocs > 0) {
1184      numdocs += response.numDocs;
1185   
1186      QueryResult_t thisresult;
1187      thisresult.collection = *col_here;
1188      ResultDocInfo_tarray::iterator doc_here = response.docInfo.begin();
1189      ResultDocInfo_tarray::iterator doc_end = response.docInfo.end();
1190      while (doc_here != doc_end) {
1191    thisresult.doc = *doc_here;
1192    results.insert (thisresult); // this is ordering based on doc rank
1193    ++doc_here;
1194      }
1195    }
1196    ++col_here;
1197  } // for each coll
1198 
1199  // now we have an ordered list of results. If ifl (I feel lucky) is set, then pick out the one we want
1200  if (args["ifl"] == 1 || (args["ifl"] == 2 && numdocs == 1)) {
1201
1202    //Find whether DocumentSearchResultLinks is enabled
1203    bool show_links = false;
1204    text_tmap::const_iterator format_here = cinfo->format.begin();
1205    text_tmap::const_iterator format_end = cinfo->format.end();
1206   
1207    while (format_here != format_end) {
1208      if (((*format_here).first == "DocumentSearchResultLinks") &&
1209      ((*format_here).second == "true")){
1210    show_links = true;
1211    break;
1212      }
1213      ++format_here;
1214    }
1215
1216    // which doc do we want?
1217    int docnum;
1218    int ifl;
1219    int srn = 0;
1220    int srp = 0;
1221    if (args["ifl"] == 1) {
1222      ifl = args["ifln"].getint();
1223      docnum = ifl - 1;
1224      if (show_links) {
1225    // set the values for next and prev search result number
1226    srn = ifl + 1;
1227    if (srn > numdocs) {
1228      srn = 0;
1229    }
1230    srp = ifl - 1;
1231    if (srp < 0) {
1232      srp = 0;
1233    }
1234      }
1235    } else {
1236      // we just want the first (and only) result
1237      docnum = 0;
1238    }
1239   
1240    if (docnum >= 0 && docnum < numdocs) {
1241    // get the docnum'th item from the results
1242      QueryResult_tset::iterator res_here = results.begin();
1243      for (int i=0; i< docnum; i++) {
1244    ++res_here;
1245      }
1246     
1247      textout << outconvert << disp
1248          << "Location: _gwcgi_?e=_compressedoptions_&a=d&c="
1249          << (*res_here).collection << "&cl=search&d=" << (*res_here).doc.OID
1250          << "&srn=" << srn << "&srp=" << srp << "\n\n";
1251      textout << flush;
1252     
1253      return true;
1254     
1255    }
1256  }
1257 
1258  if (!args["ifl"].empty()) {
1259    // 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
1260    // done the right thing (because ifl was set), so we need to make sure the output is html
1261    textout << "Content-type: text/html\n\n";
1262  }
1263
1264
1265  text_t numdocs_t = numdocs; 
1266  args["nmd"] = numdocs_t;
1267
1268  disp.setmacro ("freqmsg", "query", freqmsg);
1269
1270  define_query_macros( args, disp, numdocs, isApprox);
1271  // save the query if appropriate
1272  save_search_history(args, numdocs, isApprox);
1273  define_history_macros (disp, args, protos, logout);
1274
1275  textout << outconvert << disp << "_query:header_\n"
1276      << "_query:content_";
1277
1278  if (!syntax_error) {
1279   
1280    // now go through each result and output it
1281    QueryResult_tset::iterator res_here = results.begin();
1282    QueryResult_tset::iterator res_end = results.end();
1283    text_tset metadata; // empty !!
1284    bool getParents = false; // don't care !!
1285    bool use_table;
1286    ResultDocInfo_t thisdoc;
1287    format_t *formatlistptr = NULL;
1288    browserclass *browserptr = NULL;
1289   
1290    int count = 1;
1291    int firstdoc = args.getintarg("r");
1292    int hitsperpage = args.getintarg("o");
1293    int thislast = firstdoc + (hitsperpage - 1);
1294   
1295    // output results
1296    while (res_here != res_end) {
1297      if (count < firstdoc) {++count; ++res_here; continue;}
1298      if (count > thislast) break;
1299     
1300      formatlistptr = colinfomap[(*res_here).collection].formatlistptr;
1301      browserptr = colinfomap[(*res_here).collection].browserptr;
1302      thisdoc = (*res_here).doc;
1303      use_table = is_table_content (formatlistptr);
1304     
1305      collectproto = protos->getrecptproto ((*res_here).collection, logout);
1306      if (collectproto == NULL) {
1307    logout << outconvert << "queryaction::search_multiple_collections: " << (*res_here).collection
1308           << " collection has a NULL collectproto, ignoring results\n";
1309    ++res_here;
1310    continue;
1311      }
1312     
1313      browserptr->output_section_group (thisdoc, args, (*res_here).collection, 0,
1314                    formatlistptr, use_table, metadata, getParents,
1315                    collectproto, disp, outconvert, textout, logout);
1316      //    textout << outconvert << "(ranking: " << (*res_here).doc.ranking << ")\n";
1317      ++res_here;
1318      ++count;
1319    }
1320  }
1321  textout << outconvert << disp << "_query:footer_";
1322 
1323  // clean up the format_t pointers
1324  map<text_t, colinfo_t, lttext_t>::iterator here =  colinfomap.begin();
1325  map<text_t, colinfo_t, lttext_t>::iterator end =  colinfomap.end();
1326  while (here != end) {
1327    delete ((*here).second.formatlistptr);
1328    ++here;
1329  }
1330  return true;
1331}
1332
1333
1334// does the formatting of the query string - either uses q for a text search
1335// or the form values for an form search
1336// also adds dates if appropriate in text search
1337void queryaction::get_formatted_query_string (text_t &formattedstring,
1338                          bool segment,
1339                          cgiargsclass &args,
1340                          displayclass &disp,
1341                          ostream &logout) {
1342  if (args["qt"]=="0" && args["qto"] != "2") { // normal text search
1343    formattedstring = args["q"];
1344    // remove & | ! for simple search,do segmentation if necessary
1345    // To url-decode the '&', format_querystring() will call unsafe_cgi_arg() first
1346    format_querystring (formattedstring, args.getintarg("b"), segment);
1347    if (args["ct"]!=0) { // mgpp and lucene - need to add in tag info if appropriate
1348      format_field_info(formattedstring, args["fqf"], args.getintarg("ct"),
1349            args.getintarg("t"), args.getintarg("b"));
1350    }
1351
1352    add_dates(formattedstring, args.getintarg("ds"), args.getintarg("de"),
1353          args.getintarg("dsbc"), args.getintarg("debc"),
1354          args.getintarg("ct"));
1355    args["q"] = formattedstring;
1356   
1357  }
1358  else if (args["qt"]=="1" || args["qto"]=="2"){ // form search
1359
1360    if (args["b"]=="1" && args["fqa"]=="1") { // explicit query
1361      formattedstring = args["q"];
1362
1363      // Replace %22 and %26 with " and & respectively, since these characters have meaning
1364      // in queries: " are used in phrases and & is used in boolean advanced searches.
1365      // For form searches below, unsafe_cgi_arg is called in the parse_..._form() functions
1366
1367      unsafe_cgi_arg("ALL", formattedstring);
1368    }
1369    else { // form search
1370      if (args["b"]=="0") { // regular form
1371    parse_reg_query_form(formattedstring, args, segment); // will call unsafe_cgi_arg to decode url encoding
1372      }
1373      else  { // advanced form
1374    parse_adv_query_form(formattedstring, args, segment); // will call unsafe_cgi_arg to decode url encoding
1375      }
1376      args["q"] = formattedstring;
1377     
1378      // reset the cgiargfqv macro - need to escape any quotes in it
1379      disp.setmacro("cgiargfqv", "query", escape_quotes(args["fqv"]));
1380
1381      // also reset the _cgiargq_ macro as it has changed now
1382      disp.setmacro("cgiargq", displayclass::defaultpackage, html_safe(args["q"]));
1383
1384      // reset the compressed options to include the q arg
1385      text_t compressedoptions = recpt->get_compressed_arg(args, logout);
1386      if (!compressedoptions.empty()) {
1387    disp.setmacro ("compressedoptions", displayclass::defaultpackage, dm_safe(compressedoptions));
1388    // need a decoded version of compressedoptions for use within forms
1389    // as browsers encode values from forms before sending to server
1390    // (e.g. %25 becomes %2525)
1391    decode_cgi_arg (compressedoptions);
1392    if (args["w"] == "utf-8") { // if the encoding was utf-8, then compressed options was utf-8, and we need unicode.
1393    // if encoding wasn't utf-8, then compressed options may be screwed up, but seems to work for 8 bit encodings?
1394      compressedoptions = to_uni(compressedoptions);
1395    }
1396
1397    text_t macrovalue = dm_safe(compressedoptions);
1398    disp.setmacro ("decodedcompressedoptions", displayclass::defaultpackage, macrovalue);
1399    disp.setmacro ("decodedcompressedoptionsAttrsafe", displayclass::defaultpackage, encodeForHTMLAttr(macrovalue));
1400   
1401      }
1402    } // form search
1403  } // args["qt"]=1
1404  else {
1405    logout << "ERROR (queryaction::get_formatted_query_string): querytype not defined\n";
1406  }
1407}
1408
1409
1410// define_query_macros sets the macros that couldn't be set until the
1411// query had been done. Those macros are
1412// _resultline_, _nextfirst_, _nextlast_, _prevfirst_, _prevlast_,
1413// _thisfirst_, and _thislast_ and _quotedquery_
1414// this has been simplified so it can be used with both search_single_coll
1415// and search_multiple_coll
1416void queryaction::define_query_macros (cgiargsclass &args, displayclass &disp,
1417                       int numdocs, isapprox isApprox)
1418{
1419  // The following 'if' statatment is placed here to be keep the semantics
1420  // the same as the version before basequeryaction was introduced
1421
1422  if (num_phrases > 0) isApprox = Exact;
1423
1424  basequeryaction::define_query_macros(args,disp,numdocs,isApprox);
1425
1426  if (args["ct"]==0) { // mg queries only, not mgpp
1427    // get the quoted bits of the query string and set _quotedquery_
1428    text_tarray phrases;
1429    get_phrases (args["q"], phrases);
1430    num_phrases = phrases.size();
1431    text_tarray::const_iterator phere = phrases.begin();
1432    text_tarray::const_iterator pend = phrases.end();
1433    bool first = true;
1434    text_t quotedquery;
1435    while (phere != pend) {
1436      if (!first)
1437    if ((phere +1) == pend) quotedquery += " and ";
1438    else quotedquery += ", ";
1439     
1440      quotedquery += "\"" + *phere + "\"";
1441      first = false;
1442      ++phere;
1443    }
1444    if (args.getintarg("s") && !quotedquery.empty()) quotedquery += "_textstemon_";
1445    disp.setmacro ("quotedquery", "query", quotedquery);
1446  }
1447
1448}
1449
1450// should this change for cross coll search??
1451bool queryaction::save_search_history (cgiargsclass &args, int numdocs,
1452                       isapprox isApprox) {
1453  if (args["q"]=="") return true; // null query, dont save
1454  if (args["hs"]=="0") return true; // only save when submit query pressed
1455 
1456  // get userid
1457  text_t userid = args["z"];
1458
1459  // the number of docs goes on the front of the query string
1460  text_t query = text_t(numdocs);
1461  if (isApprox==MoreThan) { // there were more docs found
1462    query.push_back('+');
1463  }
1464  query += "c="+args["c"];
1465  query += ";h="+args["h"];
1466  query += ";t="+args["t"];
1467  query += ";b="+args["b"];
1468  query += ";j="+args["j"];
1469  query += ";n="+args["n"];
1470  query += ";s="+args["s"];
1471  query += ";k="+args["k"];
1472  query += ";g="+args["g"];
1473 
1474  text_t qstring = args["q"];
1475  //text_t formattedquery =cgi_safe(qstring);
1476  //query += "&amp;q="+formattedquery;
1477  query += ";q="+qstring;
1478  bool display=false;
1479  int hd = args.getintarg("hd");
1480  if (hd > 0) display=true;
1481  if (set_history_info(userid, query, dbhome, display)) return true;
1482  else return false;
1483}
1484
Note: See TracBrowser for help on using the browser.