source: trunk/gsdl/src/recpt/receptionist.cpp@ 10636

Last change on this file since 10636 was 10636, checked in by kjdon, 18 years ago

don't crap out if a macro file is missing

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 54.2 KB
Line 
1/**********************************************************************
2 *
3 * receptionist.cpp -- a web interface for the gsdl
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// following line required to get fstream.filedesc() on darwin (Mac OS X)
27// gcc 2.91 automatically defines this in stream.h
28#define _STREAM_COMPAT 1
29
30#include "receptionist.h"
31#include "fileutil.h"
32#include "cgiutils.h"
33#include "htmlutils.h"
34#include "gsdltools.h"
35#include "gsdltimes.h"
36#include "OIDtools.h"
37#include <assert.h>
38#include <time.h>
39#include <stdio.h> // for open()
40#include <fcntl.h> // for open() flags
41// following 2 are for printing Last-Modified http header.
42#include <sys/stat.h>
43#include <time.h>
44
45#if defined (GSDL_USE_IOS_H)
46#include <fstream.h>
47#else
48#include <fstream>
49#endif
50
51void recptconf::clear () {
52 gsdlhome.clear();
53 gdbmhome.clear();
54 collectinfo.erase(collectinfo.begin(), collectinfo.end());
55 collection.clear();
56 collectdir.clear();
57 httpprefix.clear();
58 httpimg = "/images";
59 gwcgi.clear();
60 macrofiles.erase(macrofiles.begin(), macrofiles.end());
61 saveconf.clear();
62 usecookies = false;
63 logcgiargs = false;
64 LogDateFormat = LocalTime;
65
66 maintainer.clear();
67 MailServer.clear();
68 LogEvents = Disabled;
69 EmailEvents = Disabled;
70 EmailUserEvents = false;
71
72 languages.erase(languages.begin(), languages.end());
73 encodings.erase(encodings.begin(), encodings.end());
74
75 site_auth = false;
76 HomePageType = "images";
77 HomePageCols = 3;
78
79 // these default page parameters can always be overriden
80 // in the configuration file
81 pageparams.erase(pageparams.begin(), pageparams.end());
82 pageparams["c"] = "";
83 pageparams["l"] = "en";
84
85#ifdef MACROPRECEDENCE
86 macroprecedence = MACROPRECEDENCE;
87#else
88 macroprecedence.clear();
89#endif
90}
91
92void collectioninfo_t::clear () {
93 gsdl_gsdlhome.clear();
94 gsdl_gdbmhome.clear();
95
96 info_loaded = false;
97 info.clear();
98}
99
100void languageinfo_t::clear () {
101 longname.clear();
102 defaultencoding.clear();
103}
104
105receptionist::receptionist () {
106 // create a list of cgi arguments
107 // this must be done before the configuration
108
109 cgiarginfo ainfo;
110
111 ainfo.shortname = "e";
112 ainfo.longname = "compressed arguments";
113 ainfo.multiplechar = true;
114 ainfo.defaultstatus = cgiarginfo::good;
115 ainfo.argdefault = g_EmptyText;
116 ainfo.savedarginfo = cgiarginfo::mustnot;
117 argsinfo.addarginfo (NULL, ainfo);
118
119 ainfo.shortname = "a";
120 ainfo.longname = "action";
121 ainfo.multiplechar = true;
122 ainfo.defaultstatus = cgiarginfo::none;
123 ainfo.argdefault = g_EmptyText;
124 ainfo.savedarginfo = cgiarginfo::must;
125 argsinfo.addarginfo (NULL, ainfo);
126
127 // w=western
128 ainfo.shortname = "w";
129 ainfo.longname = "encoding";
130 ainfo.multiplechar = true;
131 ainfo.defaultstatus = cgiarginfo::none;
132 ainfo.argdefault = g_EmptyText;
133 ainfo.savedarginfo = cgiarginfo::must;
134 argsinfo.addarginfo (NULL, ainfo);
135
136 ainfo.shortname = "nw";
137 ainfo.longname = "new encoding";
138 ainfo.multiplechar = true;
139 ainfo.defaultstatus = cgiarginfo::none;
140 ainfo.argdefault = g_EmptyText;
141 ainfo.savedarginfo = cgiarginfo::mustnot;
142 argsinfo.addarginfo (NULL, ainfo);
143
144 ainfo.shortname = "c";
145 ainfo.longname = "collection";
146 ainfo.multiplechar = true;
147 ainfo.defaultstatus = cgiarginfo::none;
148 ainfo.argdefault = g_EmptyText;
149 ainfo.savedarginfo = cgiarginfo::must;
150 argsinfo.addarginfo (NULL, ainfo);
151
152 // the interface language name should use the ISO 639
153 // standard
154 ainfo.shortname = "l";
155 ainfo.longname = "interface language";
156 ainfo.multiplechar = true;
157 ainfo.defaultstatus = cgiarginfo::weak;
158 ainfo.argdefault = "en";
159 ainfo.savedarginfo = cgiarginfo::must;
160 argsinfo.addarginfo (NULL, ainfo);
161
162 ainfo.shortname = "nl";
163 ainfo.longname = "new language";
164 ainfo.multiplechar = false;
165 ainfo.defaultstatus = cgiarginfo::none;
166 ainfo.argdefault = "0";
167 ainfo.savedarginfo = cgiarginfo::mustnot;
168 argsinfo.addarginfo (NULL, ainfo);
169
170 // the GSDL_UID (cookie)
171 ainfo.shortname = "z";
172 ainfo.longname = "gsdl uid";
173 ainfo.multiplechar = true;
174 ainfo.defaultstatus = cgiarginfo::none;
175 ainfo.argdefault = g_EmptyText;
176 ainfo.savedarginfo = cgiarginfo::mustnot;
177 argsinfo.addarginfo (NULL, ainfo);
178}
179
180
181void receptionist::add_action (action *theaction) {
182 // make sure we have an action to add
183 if (theaction == NULL) return;
184
185 // add this action to the list of actions
186 actions.addaction(theaction);
187
188 // add the cgi arguments from this action
189 argsinfo.addarginfo (NULL, theaction->getargsinfo());
190}
191
192
193void receptionist::add_browser (browserclass *thebrowser) {
194 // make sure we have a browser to add
195 if (thebrowser == NULL) return;
196
197 // add this browser to the list of browsers
198 browsers.addbrowser(thebrowser);
199}
200
201
202void receptionist::setdefaultbrowser (const text_t &browsername) {
203 browsers.setdefaultbrowser (browsername);
204}
205
206
207// configure should be called for each line in the
208// configuration files to configure the receptionist and everything
209// it contains. The configuration should take place after everything
210// has been added but before the initialisation.
211
212void receptionist::configure (const text_t &key, const text_tarray &cfgline) {
213 // configure the receptionist
214
215
216
217 if (cfgline.size() >= 1) {
218 cgiarginfo *info = NULL;
219 if (key == "gsdlhome") {
220 configinfo.gsdlhome = cfgline[0];
221 if (configinfo.gdbmhome.empty()) configinfo.gdbmhome = cfgline[0];
222 }
223 else if (key == "gdbmhome") configinfo.gdbmhome = cfgline[0];
224 else if (key == "collection") {
225 configinfo.collection = cfgline[0];
226 // also need to set the default arg to this collection
227 if ((info = argsinfo.getarginfo("c")) != NULL) {
228 info->defaultstatus = cgiarginfo::good;
229 info->argdefault = cfgline[0];
230 }
231
232 } else if (key == "collectdir") configinfo.collectdir = cfgline[0];
233 else if (key == "httpprefix") configinfo.httpprefix = cfgline[0];
234 else if (key == "httpimg") configinfo.httpimg = cfgline[0];
235 else if (key == "gwcgi") configinfo.gwcgi = cfgline[0];
236 else if (key == "macrofiles") {
237 // want to append to macrofiles (i.e. may be several config files
238 // contributing, maybe from several collections).
239 text_tarray::const_iterator here = cfgline.begin();
240 text_tarray::const_iterator end = cfgline.end();
241 while (here != end) {
242 configinfo.macrofiles.insert (*here);
243 ++here;
244 }
245 }
246 else if (key == "saveconf") configinfo.saveconf = cfgline[0];
247 else if (key == "usecookies") configinfo.usecookies = (cfgline[0] == "true");
248 else if (key == "logcgiargs") configinfo.logcgiargs = (cfgline[0] == "true");
249 else if (key == "maintainer") configinfo.maintainer = cfgline[0];
250 else if (key == "MailServer") configinfo.MailServer = cfgline[0];
251 else if (key == "LogDateFormat") {
252 if (cfgline[0] == "UTCTime") configinfo.LogDateFormat = UTCTime;
253 else if (cfgline[0] == "Absolute") configinfo.LogDateFormat = Absolute;
254 }
255 else if (key == "LogEvents") {
256 if (cfgline[0] == "CollectorEvents") configinfo.LogEvents = CollectorEvents;
257 else if (cfgline[0] == "AllEvents") configinfo.LogEvents = AllEvents;
258 }
259 else if (key == "EmailEvents") {
260 if (cfgline[0] == "CollectorEvents") configinfo.EmailEvents = CollectorEvents;
261 else if (cfgline[0] == "AllEvents") configinfo.EmailEvents = AllEvents;
262 }
263 else if (key == "EmailUserEvents") configinfo.EmailUserEvents = (cfgline[0] == "true");
264 else if (key == "pageparam") {
265 if (cfgline.size() >= 2) configinfo.pageparams[cfgline[0]] = cfgline[1];
266 else configinfo.pageparams[cfgline[0]] = "";
267 }
268 else if (key == "macroprecedence") configinfo.macroprecedence = cfgline[0];
269 else if (key == "collectinfo") {
270 if (cfgline.size() >= 3) {
271 collectioninfo_t cinfo;
272 cinfo.gsdl_gsdlhome = cfgline[1];
273 cinfo.gsdl_gdbmhome = cfgline[2];
274 configinfo.collectinfo[cfgline[0]] = cinfo;
275 }
276 }
277
278 // Read in the value for the site_auth directive either true or false
279 else if (key == "site_auth") configinfo.site_auth = (cfgline[0] == "true");
280
281 else if (key == "site_group")
282 joinchar(cfgline,',',configinfo.site_group);
283
284 else if (key == "SiteFormat") {
285 if (cfgline[0] == "HomePageType") {
286 configinfo.HomePageType = cfgline[1];
287 } else if (cfgline[0] == "HomePageCols") {
288 configinfo.HomePageCols = cfgline[1].getint();
289 }
290 }
291
292 else if (key == "cgiarg") {
293 // get shortname
294 bool seen_defaultstatus = false;
295 text_t subkey, subvalue;
296 text_t shortname;
297 text_t::const_iterator cfglinesub_here;
298 text_tarray::const_iterator cfgline_here = cfgline.begin();
299 text_tarray::const_iterator cfgline_end = cfgline.end();
300 while (cfgline_here != cfgline_end) {
301 cfglinesub_here = getdelimitstr((*cfgline_here).begin(),
302 (*cfgline_here).end(), '=', subkey);
303 if (subkey == "shortname") {
304 shortname = substr (cfglinesub_here, (*cfgline_here).end());
305 }
306 ++cfgline_here;
307 }
308
309 // if we found the shortname process the line again filling in values
310 if (!shortname.empty()) {
311 cgiarginfo &chinfo = argsinfo[shortname];
312 chinfo.shortname = shortname; // in case this is a new argument
313
314 cfgline_here = cfgline.begin();
315 while (cfgline_here != cfgline_end) {
316 cfglinesub_here = getdelimitstr((*cfgline_here).begin(),
317 (*cfgline_here).end(), '=', subkey);
318 subvalue = substr (cfglinesub_here, (*cfgline_here).end());
319
320 if (subkey == "longname") chinfo.longname = subvalue;
321 else if (subkey == "multiplechar") chinfo.multiplechar = (subvalue == "true");
322 else if (subkey == "defaultstatus") {
323 seen_defaultstatus = true;
324 if (subvalue == "none") chinfo.defaultstatus = cgiarginfo::none;
325 else if (subvalue == "weak") chinfo.defaultstatus = cgiarginfo::weak;
326 else if (subvalue == "good") chinfo.defaultstatus = cgiarginfo::good;
327 else if (subvalue == "config") chinfo.defaultstatus = cgiarginfo::config;
328 else if (subvalue == "imperative") chinfo.defaultstatus = cgiarginfo::imperative;
329 }
330 else if (subkey == "argdefault") {
331 chinfo.argdefault = subvalue;
332 if (!seen_defaultstatus) chinfo.defaultstatus = cgiarginfo::config;
333 }
334 else if (subkey == "savedarginfo") {
335 if (subvalue == "mustnot") chinfo.savedarginfo = cgiarginfo::mustnot;
336 else if (subvalue == "can") chinfo.savedarginfo = cgiarginfo::can;
337 else if (subvalue == "must") chinfo.savedarginfo = cgiarginfo::must;
338 }
339
340 ++cfgline_here;
341 }
342 }
343
344 } else if (key == "Encoding") {
345
346 configure_encoding (cfgline);
347
348 } else if (key == "Language") {
349 text_t subkey, subvalue, shortname;
350 languageinfo_t lang;
351 text_t::const_iterator cfglinesub_here;
352 text_tarray::const_iterator cfgline_here = cfgline.begin();
353 text_tarray::const_iterator cfgline_end = cfgline.end();
354 while (cfgline_here != cfgline_end) {
355 cfglinesub_here = getdelimitstr((*cfgline_here).begin(),
356 (*cfgline_here).end(), '=', subkey);
357 if (subkey == "shortname") {
358 shortname = substr (cfglinesub_here, (*cfgline_here).end());
359 } else if (subkey == "longname") {
360 lang.longname = substr (cfglinesub_here, (*cfgline_here).end());
361 } else if (subkey == "default_encoding") {
362 lang.defaultencoding = substr (cfglinesub_here, (*cfgline_here).end());
363 }
364 ++cfgline_here;
365 }
366 if (!shortname.empty()) {
367 if (lang.longname.empty()) lang.longname = shortname;
368 configinfo.languages[shortname] = lang;
369 }
370 }
371 }
372
373 // configure the actions
374 actionptrmap::iterator actionhere = actions.begin ();
375 actionptrmap::iterator actionend = actions.end ();
376
377 while (actionhere != actionend) {
378 assert ((*actionhere).second.a != NULL);
379 if ((*actionhere).second.a != NULL)
380 (*actionhere).second.a->configure(key, cfgline);
381
382 ++actionhere;
383 }
384
385 // configure the protocols
386 recptprotolistclass::iterator protohere = protocols.begin ();
387 recptprotolistclass::iterator protoend = protocols.end ();
388
389 while (protohere != protoend) {
390 assert ((*protohere).p != NULL);
391 comerror_t err;
392 if ((*protohere).p != NULL)
393 (*protohere).p->configure(key, cfgline, err);
394
395 ++protohere;
396 }
397
398 // configure the browsers
399 browserptrmap::iterator browserhere = browsers.begin ();
400 browserptrmap::iterator browserend = browsers.end ();
401
402 while (browserhere != browserend) {
403 assert ((*browserhere).second.b != NULL);
404 if ((*browserhere).second.b != NULL)
405 (*browserhere).second.b->configure(key, cfgline);
406
407 ++browserhere;
408 }
409}
410
411
412void receptionist::configure (const text_t &key, const text_t &value) {
413 text_tarray cfgline;
414 cfgline.push_back (value);
415 configure(key, cfgline);
416}
417
418
419// init should be called after all the actions and protocols have been
420// added to the receptionist and after everything has been configured but
421// before any pages are created. It returns true on success and false on
422// failure. If false is returned getpage should not be called (without
423// producing meaningless output), instead an error page should be produced
424// by the calling code.
425bool receptionist::init (ostream &logout) {
426
427 // first configure collectdir
428 text_t thecollectdir = configinfo.gsdlhome;
429 if (!configinfo.collection.empty()) {
430 // collection specific mode
431 if (!configinfo.collectdir.empty()) {
432 // has already been configured
433 thecollectdir = configinfo.collectdir;
434 } else {
435 // decide where collectdir is by searching for collect.cfg
436 // look in $GSDLHOME/collect/collection-name/etc/collect.cfg and
437 // then $GSDLHOME/etc/collect.cfg
438 thecollectdir = filename_cat (configinfo.gsdlhome, "collect");
439 thecollectdir = filename_cat (thecollectdir, configinfo.collection);
440 text_t filename = filename_cat (thecollectdir, "etc");
441 filename = filename_cat (filename, "collect.cfg");
442
443 if (!file_exists(filename)) thecollectdir = configinfo.gsdlhome;
444 }
445 }
446 configure("collectdir", thecollectdir);
447
448 // read in the macro files
449 if (!read_macrofiles (logout)) return false;
450
451 // there must be at least one action defined
452 if (actions.empty()) {
453 logout << "Error: no actions have been added to the receptionist\n";
454 return false;
455 }
456
457 // there must be at least one browser defined
458 if (browsers.empty()) {
459 logout << "Error: no browsers have been added to the receptionist\n";
460 return false;
461 }
462
463 // create a saveconf string if there isn't one already
464 if (configinfo.saveconf.empty())
465 configinfo.saveconf = create_save_conf_str (argsinfo, logout);
466
467 // check the saveconf string
468 if (!check_save_conf_str (configinfo.saveconf, argsinfo, logout))
469 return false;
470
471 // set a random seed
472 srand (time(NULL));
473
474 // if maintainer email address is something dodgy (for now I'll define
475 // dodgy as being anything that doesn't contain '@') disable EmailEvents
476 // and EmailUserEvents (we don't strictly need to disable EmailUserEvents
477 // in this case but we will as it seems likely that MailServer will also
478 // be screwed up if maintainer is).
479 text_t::const_iterator maintainer_end = configinfo.maintainer.end ();
480 text_t::const_iterator maintainer_here = findchar ((text_t::const_iterator)configinfo.maintainer.begin(),
481 maintainer_end, '@');
482 if (maintainer_here == maintainer_end) {
483 configinfo.EmailEvents = Disabled;
484 configinfo.EmailUserEvents = Disabled;
485 } else {
486 // if MailServer isn't set it should default to mail.maintainer-domain
487 if (configinfo.MailServer.empty()) {
488 configinfo.MailServer = "mail." + substr (maintainer_here+1, maintainer_end);
489 }
490 }
491
492 // init the actions
493 actionptrmap::iterator actionhere = actions.begin ();
494 actionptrmap::iterator actionend = actions.end ();
495 while (actionhere != actionend) {
496 if (((*actionhere).second.a == NULL) ||
497 !(*actionhere).second.a->init(logout)) return false;
498 ++actionhere;
499 }
500
501 // init the protocols
502 recptprotolistclass::iterator protohere = protocols.begin ();
503 recptprotolistclass::iterator protoend = protocols.end ();
504 while (protohere != protoend) {
505 comerror_t err;
506 if (((*protohere).p == NULL) ||
507 !(*protohere).p->init(err, logout)) return false;
508 ++protohere;
509 }
510
511 // init the browsers
512 browserptrmap::iterator browserhere = browsers.begin ();
513 browserptrmap::iterator browserend = browsers.end ();
514 while (browserhere != browserend) {
515 if (((*browserhere).second.b == NULL) ||
516 !(*browserhere).second.b->init(logout)) return false;
517 ++browserhere;
518 }
519
520 return true;
521}
522
523// get the default encoding for the given language - if it fails for any
524// reason return ""
525text_t receptionist::get_default_encoding (const text_t &language) {
526
527 // make sure language is valid
528 if (configinfo.languages.find(language) == configinfo.languages.end()) return "";
529
530 text_t default_encoding = configinfo.languages[language].defaultencoding;
531
532 // make sure the encoding is valid
533 if (converters.find(default_encoding) == converters.end()) {
534 // we don't support the encoding specified as default for this language
535 if (configinfo.encodings.size()==1) {
536 // only 1 encoding specified in main.cfg, so use it
537 return configinfo.encodings.begin()->second;
538 }
539 return "";
540 }
541
542 return default_encoding;
543}
544
545// parse_cgi_args parses cgi arguments into an argument class.
546// This function should be called for each page request. It returns false
547// if there was a major problem with the cgi arguments.
548bool receptionist::parse_cgi_args (const text_t &argstr, cgiargsclass &args,
549 ostream &logout, text_tmap &fcgienv) {
550
551 // get an initial list of cgi arguments
552 args.clear();
553 split_cgi_args (argsinfo, argstr, args);
554
555 // expand the compressed argument (if there was one)
556 if (!expand_save_args (argsinfo, configinfo.saveconf, args, logout)) return false;
557
558 // add the defaults
559 add_default_args (argsinfo, args, logout);
560
561 // get the cookie
562 if (configinfo.usecookies) get_cookie(args["z"], fcgienv);
563
564 // if we're changing languages, set the encoding to the default for the new language
565 if (args["nl"] == "1") {
566 args["nw"] = get_default_encoding(args["l"]);
567 }
568
569 // get the input encoding
570 // if encoding isn't set, set it to the default for the current language
571 if ((args.getarg("w") == NULL) || args["w"].empty()) {
572 args["w"] = get_default_encoding(args["l"]);
573 }
574
575 text_t &arg_w = args["w"];
576
577 inconvertclass defaultinconvert;
578 inconvertclass *inconvert = converters.get_inconverter (arg_w);
579 if (inconvert == NULL) inconvert = &defaultinconvert;
580
581 // see if the next page will have a different encoding
582 if (args.getarg("nw") != NULL) arg_w = args["nw"];
583
584 // convert arguments which aren't in unicode to unicode
585 args_tounicode (args, *inconvert);
586
587
588 // decide on the output conversion class (needed for checking the external
589 // cgi arguments)
590 rzwsoutconvertclass defaultoutconverter;
591 rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
592 if (outconverter == NULL) outconverter = &defaultoutconverter;
593 outconverter->reset();
594
595 // check the main cgi arguments
596 if (!check_mainargs (args, logout)) return false;
597
598 // check the arguments for the action
599 action *a = actions.getaction (args["a"]);
600 if (a != NULL) {
601 if (!a->check_cgiargs (argsinfo, args, &protocols, logout)) return false;
602 } else {
603 // the action was not found!!
604 outconvertclass text_t2ascii;
605 logout << text_t2ascii << "Error: the action \"" << args["a"]
606 << "\" could not be found.\n";
607 return false;
608 }
609
610 // check external cgi arguments for each action
611 actionptrmap::iterator actionhere = actions.begin ();
612 actionptrmap::iterator actionend = actions.end ();
613 while (actionhere != actionend) {
614 assert ((*actionhere).second.a != NULL);
615 if ((*actionhere).second.a != NULL) {
616 if (!(*actionhere).second.a->check_external_cgiargs (argsinfo, args, *outconverter,
617 configinfo.saveconf, logout))
618 return false;
619 }
620 ++actionhere;
621 }
622
623 // the action might have changed but we will assume that
624 // the cgiargs were checked properly when the change was made
625
626 return true;
627}
628
629// returns true if cookie already existed, false
630// if it was generated
631bool receptionist::get_cookie (text_t &cookie, text_tmap &fcgienv) {
632
633 text_t cookiestring = gsdl_getenv ("HTTP_COOKIE", fcgienv);
634
635 text_t::const_iterator end = cookiestring.end();
636 text_t::const_iterator here = findchar ((text_t::const_iterator)cookiestring.begin(), end, 'G');
637
638 while (here+9 < end) {
639
640 if (substr(here, here+8) == "GSDL_UID") {
641 cookie = substr (here+9, findchar (here+9, end, ';'));
642 return true;
643 }
644 ++here;
645 here = findchar (here, end, 'G');
646 }
647
648 cookie.clear();
649 text_t host = gsdl_getenv("REMOTE_ADDR", fcgienv);
650 time_t ttime = time(NULL);
651 if (!host.empty()) {
652 cookie += host;
653 cookie.push_back ('-');
654 }
655 cookie += text_t(ttime);
656
657 return false;
658}
659
660// as above but just tests if cookie exists
661bool receptionist::get_cookie (text_tmap &fcgienv) {
662
663 text_t c = gsdl_getenv("HTTP_COOKIE", fcgienv);
664 if (!c.empty()) {
665 text_t cookiestring = c;
666
667 text_t::const_iterator end = cookiestring.end();
668 text_t::const_iterator here = findchar ((text_t::const_iterator)cookiestring.begin(), end, 'G');
669
670 while (here+9 < end) {
671 if (substr(here, here+8) == "GSDL_UID") return true;
672 ++here;
673 here = findchar (here, end, 'G');
674 }
675 }
676 return false;
677}
678
679bool receptionist::log_cgi_args (cgiargsclass &args, ostream &logout, text_tmap &fcgienv) {
680
681 // see if we want to log the cgi arguments
682 if (!configinfo.logcgiargs) return true;
683
684 text_t host = gsdl_getenv ("REMOTE_HOST", fcgienv);
685 text_t script_name = gsdl_getenv ("SCRIPT_NAME", fcgienv);
686 if (host.empty()) host = gsdl_getenv ("REMOTE_ADDR", fcgienv);
687 text_t browser = gsdl_getenv ("HTTP_USER_AGENT", fcgienv);
688
689 cgiargsclass::const_iterator args_here = args.begin();
690 cgiargsclass::const_iterator args_end = args.end();
691
692 text_t argstr;
693 bool first = true;
694 while (args_here != args_end) {
695 if (!first) argstr += ", ";
696 argstr += (*args_here).first + "=" + (*args_here).second.value;
697 first = false;
698 ++args_here;
699 }
700
701 text_t logfile = filename_cat (configinfo.gdbmhome, "etc", "usage.txt");
702
703 text_t logstr = script_name;
704 logstr += " " + host;
705 logstr += " [";
706 if (configinfo.LogDateFormat == UTCTime) {
707 logstr += get_date (false);
708 } else if (configinfo.LogDateFormat == Absolute) {
709 time_t ttime = time(NULL);
710 logstr += ttime;
711 } else {
712 // LocalTime
713 logstr += get_date (true);
714 }
715 logstr += "] (" + argstr + ") \"";
716 logstr += browser;
717 logstr += "\"\n";
718
719 return append_logstr (logfile, logstr, logout);
720}
721
722bool receptionist::append_logstr (const text_t &filename, const text_t &logstr,
723 ostream &logout) {
724
725 utf8outconvertclass text_t2utf8;
726 char *lfile = filename.getcstr();
727
728 int fd = open(lfile, O_WRONLY | O_APPEND);
729
730 if (fd == -1) {
731 logout << "Error: Couldn't open file " << lfile << "\n";
732 delete []lfile;
733 return false;
734 }
735
736 // lock_val is set to 0 if file is locked successfully
737 int lock_val = 1;
738 GSDL_LOCK_FILE (fd);
739 if (lock_val == 0) {
740 text_t tmp_log_str(logstr); // so we don't pass a const to setinput...
741 text_t2utf8.setinput(&tmp_log_str);
742 char *buffer=new char[logstr.size()];
743 size_t num_chars;
744 convertclass::status_t status;
745 text_t2utf8.convert(buffer, logstr.size(), num_chars, status);
746 // ignore status - assume it is "finished" as buffer is big enough
747 write(fd, buffer, num_chars);
748 GSDL_UNLOCK_FILE (fd);
749 delete []buffer;
750 } else {
751 logout << "Error: Couldn't lock file " << lfile << "\n";
752 close(fd);
753 delete []lfile;
754 return false;
755 }
756
757 close(fd);
758
759 delete []lfile;
760 return true;
761}
762
763text_t receptionist::expandmacros (const text_t &astring, cgiargsclass &args,
764 ostream &logout) {
765 text_t outstring;
766 outconvertclass text_t2ascii;
767
768 action *a = actions.getaction (args["a"]);
769 prepare_page (a, args, text_t2ascii, logout);
770 disp.expandstring (displayclass::defaultpackage, astring, outstring);
771 return outstring;
772}
773
774// produce_cgi_page will call get_cgihead_info and
775// produce_content in the appropriate way to output a cgi header and
776// the page content (if needed). If a page could not be created it
777// will return false
778bool receptionist::produce_cgi_page (cgiargsclass &args, ostream &contentout,
779 ostream &logout, text_tmap &fcgienv) {
780 outconvertclass text_t2ascii;
781
782 response_t response;
783 text_t response_data;
784
785 // produce cgi header
786 get_cgihead_info (args, response, response_data, logout, fcgienv);
787 if (response == location) {
788 // location response (url may contain macros!!)
789 response_data = expandmacros (response_data, args, logout);
790 contentout << text_t2ascii << "Location: " << response_data << "\n\n";
791 contentout << flush;
792 return true;
793 } else if (response == content) {
794 // content response
795
796#ifdef GSDL_NOCACHE
797 contentout << "Expires: Mon, 26 Jul 1997 05:00:00 GMT\n"; // date in the past
798 tm *tm_ptr = NULL;
799 time_t t = time(NULL);
800 tm_ptr = gmtime (&t);
801 if (tm_ptr != NULL) {
802 char *timestr = new char[128];
803 strftime (timestr, 128, "%a, %d %b %Y %H:%M:%S", tm_ptr);
804 contentout << "Last-Modified: " << timestr << " GMT\n"; // always modified
805 delete []timestr;
806 }
807 contentout << "Cache-Control: no-cache, must-revalidate\n"; // HTTP/1.1
808 contentout << "Pragma: no-cache\n"; // HTTP/1.0
809
810#else
811
812 // use the later of build.cfg and collect.cfg modification times
813 // as the Last-Modified: header, for caching values
814 struct stat file_info;
815 time_t latest=0;
816
817 text_t collectname="";
818 collectname=args["c"];
819 if (collectname != "") {
820 text_t collectdir=filename_cat(configinfo.gsdlhome,"collect");
821 collectdir=filename_cat(collectdir,collectname);
822 text_t buildcfg=filename_cat(collectdir,"index");
823 buildcfg=filename_cat(buildcfg,"build.cfg");
824 char *buildcfg_ptr=buildcfg.getcstr();
825 text_t collectcfg=filename_cat(collectdir,"etc");
826 collectcfg=filename_cat(collectcfg,"collect.cfg");
827 char *collectcfg_ptr=collectcfg.getcstr();
828
829 if (stat(buildcfg_ptr, &file_info)) {
830 // we got an error. Currently don't handle error :(
831 // logout <<
832 } else {
833 latest=file_info.st_mtime;
834 }
835
836 if (stat(collectcfg_ptr, &file_info)) {
837 // error - unhandled for now
838 } else {
839 if (latest<file_info.st_mtime) latest=file_info.st_mtime;
840 }
841 delete []buildcfg_ptr;
842 delete []collectcfg_ptr;
843
844 if (latest>0) {
845 // print out modified time, "DDD, dd MMM YYYY hh:mm:ss" format
846 // c library takes care of mem for this string... (has \n at end!!!!)
847 // latest is currently local time, convert to UTC.
848 struct tm* utc_latest;
849 utc_latest=gmtime(&latest);
850 contentout << "Last-Modified: " << asctime(utc_latest);
851 }
852 } // end of collection != ""
853
854#endif
855
856 contentout << text_t2ascii << "Content-type: " << response_data << "\n\n";
857 }
858 else if (response == undecided_location) {
859 // Wait until later to output the target location
860 // Used for the "I'm feeling lucky" functionality
861 }
862 else {
863 // unknown response
864 logout << "Error: get_cgihead_info returned an unknown response type.\n";
865 return false;
866 }
867
868 // produce cgi page
869 if (!produce_content (args, contentout, logout)) return false;
870
871 // flush contentout
872 contentout << flush;
873 return true;
874}
875
876
877// get_cgihead_info determines the cgi header information for
878// a set of cgi arguments. If response contains location then
879// response_data contains the redirect address. If reponse
880// contains content then reponse_data contains the content-type.
881// Note that images can now be produced by the receptionist.
882// Note also, alternative for get_cgihead_info below which
883// stores the information in a text_tmap so it is more easily digested
884
885void receptionist::get_cgihead_info (cgiargsclass &args, response_t &response,
886 text_t &response_data, ostream &logout,
887 text_tmap &fcgienv) {
888 outconvertclass text_t2ascii;
889
890 // get the action
891 action *a = actions.getaction (args["a"]);
892 if (a != NULL) {
893 a->get_cgihead_info (args, &protocols, response, response_data, logout);
894
895 } else {
896 // the action was not found!!
897 logout << text_t2ascii << "Error receptionist::get_cgihead_info: the action \""
898 << args["a"] << "\" could not be found.\n";
899 response = content;
900 response_data = "text/html";
901 }
902
903 // add the encoding information
904 if (response == content) {
905 if (converters.find(args["w"]) != converters.end()) {
906 response_data += "; charset=" + args["w"];
907 } else {
908 // default to latin 1
909 response_data += "; charset=ISO-8859-1";
910 }
911
912 // add cookie if required
913 if (configinfo.usecookies && !get_cookie(fcgienv))
914 response_data += "\nSet-Cookie: GSDL_UID=" + args["z"]
915 + "; expires=25-Dec-37 00:00:00 GMT";
916 }
917}
918
919
920// Alternative version of get_cgihead_info, stores fielded infomation
921// in text_tmap rather than concatenated string
922void receptionist::get_cgihead_info (cgiargsclass &args, text_tmap &headers,
923 ostream &logout, text_tmap &fcgienv) {
924
925 response_t response;
926 text_t response_data;
927
928 // get the action
929 action *a = actions.getaction (args["a"]);
930 if (a != NULL) {
931 a->get_cgihead_info (args, &protocols, response, response_data, logout);
932
933 } else {
934 // the action was not found!!
935 outconvertclass text_t2ascii;
936 logout << text_t2ascii << "Error receptionist::get_cgihead_info: the action \""
937 << args["a"] << "\" could not be found.\n";
938 response = content;
939 response_data = "text/html";
940 }
941
942 if (response == location) {
943 response_data = expandmacros(response_data, args, logout);
944 headers["Location"] = response_data;
945 return;
946 }
947
948 // add the encoding information
949 if (response == content) {
950
951 if (converters.find(args["w"]) != converters.end()) {
952 headers["content-encoding"] = args["w"];
953 response_data += "; charset=" + args["w"];
954 } else {
955 // default to utf-8
956 headers["content-encoding"] = "utf-8";
957 response_data += "; charset=utf-8";
958 }
959
960 headers["content-type"] = response_data;
961
962 }
963
964}
965
966
967
968// produce the page content
969bool receptionist::produce_content (cgiargsclass &args, ostream &contentout,
970 ostream &logout) {
971
972 // decide on the output conversion class
973 text_t &arg_w = args["w"];
974 rzwsoutconvertclass defaultoutconverter;
975 rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
976 if (outconverter == NULL) outconverter = &defaultoutconverter;
977 outconverter->reset();
978
979 // needed for 16-bit unicode only - big endian marker 0xfeff (RFC 2781)
980 if (arg_w=="utf-16be") {
981 contentout << '\xfe' << '\xff' ;
982 }
983
984 recptproto *collectproto = protocols.getrecptproto (args["c"], logout);
985 if (collectproto != NULL) {
986 // get browsers to process OID
987 text_t OID = args["d"];
988 if (OID.empty()) OID = args["cl"];
989 if (!OID.empty()) {
990 text_tset metadata;
991 text_tarray OIDs;
992 OIDs.push_back (OID);
993 if (!is_top(OID)) OIDs.push_back (OID + ".pr");
994 FilterResponse_t response;
995 metadata.insert ("childtype");
996 if (get_info (OIDs, args["c"], args["l"], metadata, false, collectproto, response, logout)) {
997 text_t classifytype;
998 if (!response.docInfo[0].metadata["childtype"].values[0].empty())
999 classifytype = response.docInfo[0].metadata["childtype"].values[0];
1000 else if (!is_top (OID)) {
1001 // not sure why this is occasionally not set, but it will
1002 // cause a segfault... possibly if built with no_text? jrm21
1003 if (response.docInfo[1].metadata.find("childtype")
1004 == response.docInfo[1].metadata.end()) {
1005 cerr << "receptionist: no childtype element in metadata map!"
1006 << endl;
1007 } else {
1008 if (!response.docInfo[1].metadata["childtype"].values[0].empty())
1009 classifytype = response.docInfo[1].metadata["childtype"].values[0];
1010 }
1011 }
1012 browserclass *b = browsers.getbrowser (classifytype);
1013 b->processOID (args, collectproto, logout);
1014 }
1015 }
1016
1017 // translate "d" and "cl" arguments if required
1018 translate_OIDs (args, collectproto, logout);
1019 }
1020
1021 // produce the page using the desired action
1022 action *a = actions.getaction (args["a"]);
1023 if (a != NULL) {
1024 if (a->uses_display(args)) prepare_page (a, args, (*outconverter), logout);
1025 if (!a->do_action (args, &protocols, &browsers, disp, (*outconverter), contentout, logout))
1026 return false;
1027 } else {
1028 // the action was not found!!
1029 outconvertclass text_t2ascii;
1030
1031 logout << text_t2ascii << "Error receptionist::produce_content: the action \""
1032 << args["a"] << "\" could not be found.\n";
1033
1034 contentout << (*outconverter)
1035 << "<html>\n"
1036 << "<head>\n"
1037 << "<title>Error</title>\n"
1038 << "</head>\n"
1039 << "<body>\n"
1040 << "<h2>Oops!</h2>\n"
1041 << "Undefined Page. The action \""
1042 << args["a"] << "\" could not be found.\n"
1043 << "</body>\n"
1044 << "</html>\n";
1045 }
1046 return true;
1047}
1048
1049
1050// returns the compressed argument ("e") corresponding to the argument
1051// list. This can be used to save preferences between sessions.
1052text_t receptionist::get_compressed_arg (cgiargsclass &args, ostream &logout) {
1053 // decide on the output conversion class
1054 text_t &arg_w = args["w"];
1055 rzwsoutconvertclass defaultoutconverter;
1056 rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
1057 if (outconverter == NULL) outconverter = &defaultoutconverter;
1058 outconverter->reset();
1059
1060 text_t compressed_args;
1061 if (compress_save_args (argsinfo, configinfo.saveconf, args,
1062 compressed_args, *outconverter, logout))
1063 return compressed_args;
1064
1065 return g_EmptyText;
1066}
1067
1068
1069// will read in all the macro files. If one is not found an
1070// error message will be written to logout and the method will
1071// return false.
1072bool receptionist::read_macrofiles (ostream &logout) {
1073 outconvertclass text_t2ascii;
1074
1075 // redirect the error output to logout
1076 ostream *savedlogout = disp.setlogout (&logout);
1077
1078 // unload any macros that were previously loaded - this allows us to call
1079 // this function a second time to reload all the macro files (useful for
1080 // reading in changed macro files in server versions of greenstone)
1081 disp.unloaddefaultmacros();
1082
1083 // load up the default macro files, the collection directory
1084 // is searched first for the file (if this is being used in
1085 // collection specific mode) and then the main directory(s)
1086 text_t colmacrodir = filename_cat (configinfo.collectdir, "macros");
1087
1088 text_tset maindirs;
1089 text_t gsdlmacrodir = filename_cat (configinfo.gsdlhome, "macros");
1090 maindirs.insert (gsdlmacrodir);
1091 colinfo_tmap::iterator colhere = configinfo.collectinfo.begin();
1092 colinfo_tmap::iterator colend = configinfo.collectinfo.end();
1093 while (colhere != colend) {
1094 if (!((*colhere).second.gsdl_gsdlhome).empty()) {
1095 gsdlmacrodir = filename_cat ((*colhere).second.gsdl_gsdlhome, "macros");
1096 maindirs.insert (gsdlmacrodir);
1097 }
1098 ++colhere;
1099 }
1100
1101 text_tset::iterator arrhere = configinfo.macrofiles.begin();
1102 text_tset::iterator arrend = configinfo.macrofiles.end();
1103 text_t filename;
1104 while (arrhere != arrend) {
1105 bool foundfile = false;
1106
1107 // try in the collection directory if this is being
1108 // run in collection specific mode
1109 if (!configinfo.collection.empty()) {
1110 filename = filename_cat (colmacrodir, *arrhere);
1111 if (file_exists (filename)) {
1112 disp.loaddefaultmacros(filename);
1113 foundfile = true;
1114 }
1115 }
1116
1117 // if we haven't found the macro file yet try in
1118 // the main macro directory(s)
1119 // if file is found in more than one main directory
1120 // we'll load all copies
1121 if (!foundfile) {
1122 text_tset::const_iterator dirhere = maindirs.begin();
1123 text_tset::const_iterator dirend = maindirs.end();
1124 while (dirhere != dirend) {
1125 filename = filename_cat (*dirhere, *arrhere);
1126 if (file_exists (filename)) {
1127 disp.loaddefaultmacros(filename);
1128 foundfile = true;
1129 }
1130 ++dirhere;
1131 }
1132 }
1133
1134 // see if we found the file or not
1135 if (!foundfile) {
1136 logout << text_t2ascii
1137 << "Error: the macro file \"" << *arrhere << "\" could not be found.\n";
1138 if (configinfo.collection.empty()) {
1139 text_t dirs;
1140 joinchar (maindirs, ", ", dirs);
1141 logout << text_t2ascii
1142 << "It should be in either of the following directories ("
1143 << dirs << ").\n\n";
1144
1145 } else {
1146 logout << text_t2ascii
1147 << "It should be in either " << colmacrodir << " or in "
1148 << gsdlmacrodir << ".\n\n";
1149 }
1150 // don't crap out if a macro file is missing
1151 //disp.setlogout (savedlogout);
1152 //return false;
1153 }
1154 ++arrhere;
1155 }
1156
1157 // success
1158
1159 // reset logout to what it was
1160 disp.setlogout (savedlogout);
1161 return true;
1162}
1163
1164
1165
1166
1167// Go through the list of macro files looking to see
1168// if any exist in the collectoin specific area. If they
1169// do then read them in and add them to the set of existing
1170// current macros
1171
1172void receptionist::read_collection_macrofiles (const text_t& collection, ostream &logout)
1173{
1174 outconvertclass text_t2ascii;
1175
1176 // disp.unloadcollectionmacros();
1177
1178 // redirect the error output to logout
1179 ostream *savedlogout = disp.setlogout (&logout);
1180
1181 text_t colmacrodir
1182 = filename_cat (configinfo.gsdlhome,"collect",collection, "macros");
1183
1184 if (directory_exists (colmacrodir)) {
1185
1186 text_tset::iterator arrhere = configinfo.macrofiles.begin();
1187 text_tset::iterator arrend = configinfo.macrofiles.end();
1188 text_t filename;
1189 while (arrhere != arrend) {
1190
1191 filename = filename_cat (colmacrodir, *arrhere);
1192 if (file_exists (filename)) {
1193 disp.loadcollectionmacros(filename);
1194 }
1195
1196 ++arrhere;
1197 }
1198 }
1199
1200 // reset logout to what it was
1201 disp.setlogout (savedlogout);
1202}
1203
1204
1205
1206
1207// check_mainargs will check all the main arguments. If a major
1208// error is found it will return false and no cgi page should
1209// be created using the arguments.
1210
1211bool receptionist::check_mainargs (cgiargsclass &args, ostream &logout) {
1212
1213 if(configinfo.site_auth)
1214 {
1215 args["uan"] = "1";
1216 args["ug"] = configinfo.site_group;
1217 }
1218
1219
1220 // if this receptionist is running in collection dependant mode
1221 // then it should always set the collection argument to the
1222 // collection
1223 if (!configinfo.collection.empty()) args["c"] = configinfo.collection;
1224
1225 // if current collection uses ccscols make sure
1226 // "ccs" argument is set and make "cc" default to
1227 // all collections in "ccs"
1228 if (args["a"] != "config" && !args["c"].empty()) {
1229
1230 text_t &arg_c = args["c"];
1231 recptproto *collectproto = protocols.getrecptproto (arg_c, logout);
1232 if (collectproto == NULL) {
1233 // oops, this collection isn't valid
1234 outconvertclass text_t2ascii;
1235 logout << text_t2ascii << "ERROR: Invalid collection: " << arg_c << "\n";
1236 // args["c"].clear();
1237
1238 } else {
1239
1240 ColInfoResponse_t *cinfo = get_collectinfo_ptr (collectproto, arg_c, logout);
1241
1242 if(cinfo->authenticate == "collection")
1243 {
1244 args["uan"] = "1";
1245 args["ug"] = cinfo->auth_group;
1246 }
1247
1248
1249 if (cinfo != NULL) {
1250 if (!cinfo->ccsCols.empty()) {
1251 args["ccs"] = 1;
1252 if (args["cc"].empty()) {
1253 text_tarray::const_iterator col_here = cinfo->ccsCols.begin();
1254 text_tarray::const_iterator col_end = cinfo->ccsCols.end();
1255 bool first = true;
1256 while (col_here != col_end) {
1257 // make sure it's a valid collection
1258 if (protocols.getrecptproto (*col_here, logout) != NULL) {
1259 if (!first) args["cc"].push_back (',');
1260 args["cc"] += *col_here;
1261 first = false;
1262 }
1263 ++col_here;
1264 }
1265 }
1266 }
1267 } else {
1268 logout << "ERROR (receptionist::check_mainargs): get_collectinfo_ptr returned NULL\n";
1269 }
1270 }
1271 }
1272
1273 // argument "v" can only be 0 or 1. Use the default value
1274 // if it is out of range
1275 int arg_v = args.getintarg ("v");
1276 if (arg_v != 0 && arg_v != 1) {
1277 cgiarginfo *vinfo = argsinfo.getarginfo ("v");
1278 if (vinfo != NULL) args["v"] = vinfo->argdefault;
1279 }
1280
1281 // argument "f" can only be 0 or 1. Use the default value
1282 // if it is out of range
1283 int arg_f = args.getintarg ("f");
1284 if (arg_f != 0 && arg_f != 1) {
1285 cgiarginfo *finfo = argsinfo.getarginfo ("f");
1286 if (finfo != NULL) args["f"] = finfo->argdefault;
1287 }
1288
1289 return true;
1290}
1291
1292// translate_OIDs translates the "d" and "cl" arguments to their correct values
1293// if they use the tricky ".fc", ".lc" type syntax.
1294void receptionist::translate_OIDs (cgiargsclass &args, recptproto *collectproto,
1295 ostream &logout) {
1296
1297 FilterResponse_t response;
1298 FilterRequest_t request;
1299 comerror_t err;
1300 text_t &arg_d = args["d"];
1301 text_t &arg_cl = args["cl"];
1302 text_t &collection = args["c"];
1303
1304 // do a call to translate OIDs if required
1305 request.filterName = "NullFilter";
1306 request.filterResultOptions = FROID;
1307 if (!arg_d.empty() && needs_translating (arg_d)) {
1308 request.docSet.push_back (arg_d);
1309 collectproto->filter (collection, request, response, err, logout);
1310 arg_d = response.docInfo[0].OID;
1311 request.clear();
1312 }
1313 // we'll also check here that the "cl" argument has a "classify" doctype
1314 // (in case ".fc" or ".lc" have screwed up)
1315 if (needs_translating (arg_cl)) {
1316 request.fields.insert ("doctype");
1317 request.docSet.push_back (arg_cl);
1318 request.filterResultOptions = FRmetadata;
1319 collectproto->filter (collection, request, response, err, logout);
1320 // set to original value (without .xx stuff) if doctype isn't "classify"
1321 if (response.docInfo[0].metadata["doctype"].values[0] != "classify")
1322 strip_suffix (arg_cl);
1323 else
1324 arg_cl = response.docInfo[0].OID;
1325 }
1326}
1327
1328// prepare_page sets up page parameters, sets display macros
1329// and opens the page ready for output
1330void receptionist::prepare_page (action *a, cgiargsclass &args,
1331 outconvertclass &outconvert,
1332 ostream &logout) {
1333 // set up page parameters
1334 text_t pageparams;
1335 bool first = true;
1336
1337 text_tmap::iterator params_here = configinfo.pageparams.begin();
1338 text_tmap::iterator params_end = configinfo.pageparams.end();
1339 while (params_here != params_end) {
1340 // page params are those from main.cfg (eg pageparam v 0) plus
1341 // two defaults set in recptconf.clear() (c="" and l=en)
1342 // This used to check if the current value of the page param
1343 // == the default value, then don't add in it the list
1344 // but if l=en, and there is a macro with [l=en], then it doesn't
1345 // find it.
1346 // so now all page params will go into the list. I assume this will
1347 // mean more attempts to find each macro, but nothing worsee than
1348 // that. --kjdon
1349 //if (args[(*params_here).first] != (*params_here).second) {
1350 if (first)
1351 first = false;
1352 else
1353 pageparams += ",";
1354
1355 pageparams += (*params_here).first;
1356 pageparams += "=";
1357 pageparams += args[(*params_here).first];
1358 // }
1359
1360 ++params_here;
1361 }
1362
1363
1364 // open the page
1365 disp.openpage(pageparams, configinfo.macroprecedence);
1366
1367 disp.unloadcollectionmacros();
1368
1369 text_t collection = args["c"];
1370 if (!collection.empty()) {
1371 read_collection_macrofiles(collection,logout);
1372 }
1373
1374 // define external macros for each action
1375 actionptrmap::iterator actionhere = actions.begin ();
1376 actionptrmap::iterator actionend = actions.end ();
1377
1378 while (actionhere != actionend) {
1379 assert ((*actionhere).second.a != NULL);
1380 if ((*actionhere).second.a != NULL) {
1381 (*actionhere).second.a->define_external_macros (disp, args, &protocols, logout);
1382 }
1383 ++actionhere;
1384 }
1385
1386
1387 // define internal macros for the current action
1388 a->define_internal_macros (disp, args, &protocols, logout);
1389
1390 // define general macros. the defining of general macros is done here so that
1391 // the last possible version of the cgi arguments are used
1392 define_general_macros (args, outconvert, logout);
1393}
1394
1395void receptionist::define_general_macros (cgiargsclass &args, outconvertclass &/*outconvert*/,
1396 ostream &logout) {
1397
1398 text_t &collection = args["c"];
1399
1400 disp.setmacro ("gsdlhome", displayclass::defaultpackage, dm_safe(configinfo.gsdlhome));
1401 disp.setmacro ("gwcgi", displayclass::defaultpackage, configinfo.gwcgi);
1402 disp.setmacro ("httpimg", displayclass::defaultpackage, configinfo.httpimg);
1403 disp.setmacro ("httpprefix", displayclass::defaultpackage, configinfo.httpprefix);
1404
1405
1406 text_t compressedoptions = get_compressed_arg(args, logout);
1407 disp.setmacro ("compressedoptions", displayclass::defaultpackage, dm_safe(compressedoptions));
1408 // need a decoded version of compressedoptions for use within forms
1409 // as browsers encode values from forms before sending to server
1410 // (e.g. %25 becomes %2525)
1411 decode_cgi_arg (compressedoptions);
1412 disp.setmacro ("decodedcompressedoptions", displayclass::defaultpackage, dm_safe(compressedoptions));
1413
1414#if defined (__WIN32__)
1415 disp.setmacro ("win32", displayclass::defaultpackage, "1");
1416#endif
1417
1418 // set _cgiargX_ macros for each cgi argument
1419 cgiargsclass::const_iterator argshere = args.begin();
1420 cgiargsclass::const_iterator argsend = args.end();
1421 while (argshere != argsend) {
1422 if (((*argshere).first == "q") ||
1423 ((*argshere).first == "qa") ||
1424 ((*argshere).first == "qtt") ||
1425 ((*argshere).first == "qty") ||
1426 ((*argshere).first == "qp") ||
1427 ((*argshere).first == "qpl") ||
1428 ((*argshere).first == "qr") ||
1429 ((*argshere).first == "q2"))
1430 // need to escape special characters from query string
1431 disp.setmacro ("cgiarg" + (*argshere).first,
1432 displayclass::defaultpackage, html_safe((*argshere).second.value));
1433 else if ((*argshere).first == "hp") {
1434 disp.setmacro ("cgiarg" + (*argshere).first, displayclass::defaultpackage, (*argshere).second.value);
1435 } else {
1436 disp.setmacro ("cgiarg" + (*argshere).first, displayclass::defaultpackage, dm_safe((*argshere).second.value));
1437 }
1438 ++argshere;
1439 }
1440
1441 // display text right to left if language is arabic/farsi (and if browser can support it)
1442 if (args["l"] == "ar" || args["l"] == "fa")
1443 disp.setmacro ("htmlextra", displayclass::defaultpackage, " dir=rtl");
1444
1445 // set collection specific macros
1446 if (!collection.empty()) {
1447 recptproto *collectproto = protocols.getrecptproto (collection, logout);
1448 if (collectproto != NULL) {
1449 FilterResponse_t response;
1450 text_tset metadata;
1451 get_info ("collection", collection, args["l"], metadata, false,
1452 collectproto, response, logout);
1453
1454 if (!response.docInfo[0].metadata.empty()) {
1455 MetadataInfo_tmap::const_iterator here = response.docInfo[0].metadata.begin();
1456 MetadataInfo_tmap::const_iterator end = response.docInfo[0].metadata.end();
1457 while (here != end) {
1458 if (((*here).first != "haschildren") && ((*here).first != "hasnext") &&
1459 ((*here).first != "hasprevious")) {
1460 // check for args in form name:lang
1461 text_t name = g_EmptyText;
1462 text_t lang = g_EmptyText;
1463 bool colonfound=false;
1464 text_t::const_iterator a = (*here).first.begin();
1465 text_t::const_iterator b = (*here).first.end();
1466 while (a !=b) {
1467 if (*a==':') {
1468 colonfound=true;
1469 }
1470 else {
1471 if (colonfound)
1472 lang.push_back(*a);
1473 else name.push_back(*a);
1474 }
1475 ++a;
1476 }
1477 if (!lang.empty()) {
1478 if (args["l"]==lang) {
1479 disp.setcollectionmacro(displayclass::defaultpackage, name, "", (*here).second.values[0]);
1480 }
1481 }
1482 else { // the default one
1483 disp.setcollectionmacro(displayclass::defaultpackage, (*here).first, "", (*here).second.values[0]);
1484 }
1485 }
1486 ++here;
1487 }
1488 }
1489
1490 text_t iconcollection;
1491 disp.expandstring (displayclass::defaultpackage, "_iconcollection_", iconcollection);
1492 if (!iconcollection.empty())
1493 {
1494 ColInfoResponse_t cinfo;
1495 comerror_t err;
1496 collectproto->get_collectinfo (collection, cinfo, err, logout);
1497 if (iconcollection[0]=='/' && !cinfo.httpdomain.empty())
1498 {
1499 // local but with full path
1500 iconcollection = "http://" + cinfo.httpdomain + iconcollection;
1501 disp.setmacro("iconcollection", displayclass::defaultpackage, iconcollection);
1502 }
1503 }
1504 }
1505 }
1506
1507 if (!collection.empty()) {
1508 ColInfoResponse_t cinfo;
1509 comerror_t err;
1510 recptproto *collectproto = protocols.getrecptproto (collection, logout);
1511 if (collectproto != NULL) {
1512 collectproto->get_collectinfo (collection, cinfo, err, logout);
1513 text_t httpcollection;
1514 if (!cinfo.httpdomain.empty()) httpcollection = "http://";
1515 httpcollection += cinfo.httpdomain + cinfo.httpprefix + "/collect/"
1516 + collection;
1517 disp.setmacro ("httpcollection", displayclass::defaultpackage,
1518 httpcollection);
1519 // as of gsdl 2.53, collect.cfg can specify macros
1520 if (cinfo.collection_macros.size() > 0) {
1521 macros_map::const_iterator this_macro=cinfo.collection_macros.begin();
1522 macros_map::const_iterator done_macro=cinfo.collection_macros.end();
1523 while (this_macro != done_macro) {
1524
1525 disp.setcollectionmacro("Global", // package
1526 this_macro->first, // macro name
1527 this_macro->second.first, // params
1528 this_macro->second.second); // value
1529 ++this_macro;
1530 }
1531 } // col macros
1532 } // collectproto != NULL
1533 }
1534
1535}
1536
1537// gets collection info from cache if found or
1538// calls collection server (and updates cache)
1539// returns NULL if there's an error
1540ColInfoResponse_t *receptionist::get_collectinfo_ptr (recptproto *collectproto,
1541 const text_t &collection,
1542 ostream &logout) {
1543
1544 // check the cache
1545 colinfo_tmap::iterator it = configinfo.collectinfo.find (collection);
1546 if ((it != configinfo.collectinfo.end()) && ((*it).second.info_loaded)) {
1547 // found it
1548 return &((*it).second.info);
1549 }
1550
1551 // not cached, get info from collection server
1552 if (collectproto == NULL) {
1553 logout << "ERROR: receptionist::get_collectinfo_ptr passed null collectproto\n";
1554 return NULL;
1555 }
1556
1557 comerror_t err;
1558 if (it == configinfo.collectinfo.end()) {
1559 collectioninfo_t cinfo;
1560 collectproto->get_collectinfo (collection, cinfo.info, err, logout);
1561 if (err != noError) {
1562 outconvertclass text_t2ascii;
1563 logout << text_t2ascii << "ERROR (receptionist::getcollectinfo_ptr): \""
1564 << get_comerror_string (err) << "\"while getting collectinfo\n";
1565 return NULL;
1566 }
1567 cinfo.info_loaded = true;
1568 configinfo.collectinfo[collection] = cinfo;
1569 return &(configinfo.collectinfo[collection].info);
1570 } else {
1571 collectproto->get_collectinfo (collection, (*it).second.info, err, logout);
1572 if (err != noError) {
1573 outconvertclass text_t2ascii;
1574 logout << text_t2ascii << "ERROR (receptionist::getcollectinfo_ptr): \""
1575 << get_comerror_string (err) << "\"while getting collectinfo\n";
1576 return NULL;
1577 }
1578 (*it).second.info_loaded = true;
1579 return &((*it).second.info);
1580 }
1581}
1582
1583// removes a collection from the cache so that the next
1584// call to get_collectinfo_ptr() for that collection will
1585// retrieve the collection info from the collection server
1586void receptionist::uncache_collection (const text_t &collection) {
1587
1588 colinfo_tmap::iterator it = configinfo.collectinfo.find (collection);
1589 if ((it != configinfo.collectinfo.end()) && ((*it).second.info_loaded)) {
1590
1591 (*it).second.info_loaded = false;
1592
1593 }
1594}
1595
1596// Handles an "Encoding" line from a configuration file - note that the
1597// configinfo.encodings map is a bit of a hack (to be fixed when the
1598// configuration files are tidied up).
1599void receptionist::configure_encoding (const text_tarray &cfgline) {
1600
1601 text_t subkey, subvalue, shortname, longname, mapfile;
1602 int multibyte = 0;
1603 text_t::const_iterator cfglinesub_here;
1604 text_tarray::const_iterator cfgline_here = cfgline.begin();
1605 text_tarray::const_iterator cfgline_end = cfgline.end();
1606 while (cfgline_here != cfgline_end) {
1607 if (*cfgline_here == "multibyte") {
1608 multibyte = 1;
1609 } else {
1610 cfglinesub_here = getdelimitstr((*cfgline_here).begin(),
1611 (*cfgline_here).end(), '=', subkey);
1612 if (subkey == "shortname") {
1613 shortname = substr (cfglinesub_here, (*cfgline_here).end());
1614 } else if (subkey == "longname") {
1615 longname = substr (cfglinesub_here, (*cfgline_here).end());
1616 } else if (subkey == "map") {
1617 mapfile = substr (cfglinesub_here, (*cfgline_here).end());
1618 }
1619 }
1620 ++cfgline_here;
1621 }
1622 if (!shortname.empty()) {
1623 if (longname.empty()) longname = shortname;
1624
1625 // add the converter
1626 if (shortname == "utf-8") {
1627 utf8inconvertclass *utf8inconvert = new utf8inconvertclass();
1628 utf8outconvertclass *utf8outconvert = new utf8outconvertclass();
1629 utf8outconvert->set_rzws(1);
1630 add_converter (shortname, utf8inconvert, utf8outconvert);
1631 configinfo.encodings[longname] = shortname;
1632
1633 } else if (shortname == "utf-16be") {
1634 // we use the default input converter as this shouldn't ever be used
1635 // for converting from unicode...
1636 inconvertclass *inconverter = new inconvertclass();
1637 utf16outconvertclass *outconverter = new utf16outconvertclass();
1638 add_converter (shortname, inconverter, outconverter);
1639 configinfo.encodings[longname] = shortname;
1640
1641 } else if (!mapfile.empty()) {
1642
1643 if (mapfile == "8859_1.ump") {
1644 // iso-8859-1 is a special case as it'll always be supported by the
1645 // standard converter class and therefore doesn't need to use its
1646 // mapping file
1647 inconvertclass *inconvert = new inconvertclass();
1648 rzwsoutconvertclass *outconvert = new rzwsoutconvertclass();
1649 outconvert->set_rzws(1);
1650 add_converter (shortname, inconvert, outconvert);
1651 configinfo.encodings[longname] = shortname;
1652
1653 } else {
1654 text_t to_uc_map = filename_cat(configinfo.gsdlhome, "mappings", "to_uc", mapfile);
1655 text_t from_uc_map = filename_cat(configinfo.gsdlhome, "mappings", "from_uc", mapfile);
1656 if (file_exists(to_uc_map) && file_exists(from_uc_map)) {
1657
1658 mapinconvertclass *mapinconvert = new mapinconvertclass();
1659 mapinconvert->setmapfile (to_uc_map, 0x003F);
1660 mapinconvert->set_multibyte (multibyte);
1661 mapoutconvertclass *mapoutconvert = new mapoutconvertclass();
1662 mapoutconvert->setmapfile (from_uc_map, 0x3F);
1663 mapoutconvert->set_multibyte (multibyte);
1664 mapoutconvert->set_rzws(1);
1665 add_converter (shortname, mapinconvert, mapoutconvert);
1666 configinfo.encodings[longname] = shortname;
1667 }
1668 }
1669 }
1670 }
1671}
Note: See TracBrowser for help on using the repository browser.