source: trunk/gsdl/src/recpt/receptionist.cpp@ 9674

Last change on this file since 9674 was 9674, checked in by jrm21, 17 years ago

rename utf-16 to utf-16be, otherwise browsers might get confused

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 53.9 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()) return "";
534
535 return default_encoding;
536}
537
538// parse_cgi_args parses cgi arguments into an argument class.
539// This function should be called for each page request. It returns false
540// if there was a major problem with the cgi arguments.
541bool receptionist::parse_cgi_args (const text_t &argstr, cgiargsclass &args,
542 ostream &logout, text_tmap &fcgienv) {
543
544 // get an initial list of cgi arguments
545 args.clear();
546 split_cgi_args (argsinfo, argstr, args);
547
548 // expand the compressed argument (if there was one)
549 if (!expand_save_args (argsinfo, configinfo.saveconf, args, logout)) return false;
550
551 // add the defaults
552 add_default_args (argsinfo, args, logout);
553
554 // get the cookie
555 if (configinfo.usecookies) get_cookie(args["z"], fcgienv);
556
557 // if we're changing languages, set the encoding to the default for the new language
558 if (args["nl"] == "1") {
559 args["nw"] = get_default_encoding(args["l"]);
560 }
561
562 // get the input encoding
563 // if encoding isn't set, set it to the default for the current language
564 if ((args.getarg("w") == NULL) || args["w"].empty()) {
565 args["w"] = get_default_encoding(args["l"]);
566 }
567
568 text_t &arg_w = args["w"];
569
570 inconvertclass defaultinconvert;
571 inconvertclass *inconvert = converters.get_inconverter (arg_w);
572 if (inconvert == NULL) inconvert = &defaultinconvert;
573
574 // see if the next page will have a different encoding
575 if (args.getarg("nw") != NULL) arg_w = args["nw"];
576
577 // convert arguments which aren't in unicode to unicode
578 args_tounicode (args, *inconvert);
579
580
581 // decide on the output conversion class (needed for checking the external
582 // cgi arguments)
583 rzwsoutconvertclass defaultoutconverter;
584 rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
585 if (outconverter == NULL) outconverter = &defaultoutconverter;
586 outconverter->reset();
587
588 // check the main cgi arguments
589 if (!check_mainargs (args, logout)) return false;
590
591 // check the arguments for the action
592 action *a = actions.getaction (args["a"]);
593 if (a != NULL) {
594 if (!a->check_cgiargs (argsinfo, args, &protocols, logout)) return false;
595 } else {
596 // the action was not found!!
597 outconvertclass text_t2ascii;
598 logout << text_t2ascii << "Error: the action \"" << args["a"]
599 << "\" could not be found.\n";
600 return false;
601 }
602
603 // check external cgi arguments for each action
604 actionptrmap::iterator actionhere = actions.begin ();
605 actionptrmap::iterator actionend = actions.end ();
606 while (actionhere != actionend) {
607 assert ((*actionhere).second.a != NULL);
608 if ((*actionhere).second.a != NULL) {
609 if (!(*actionhere).second.a->check_external_cgiargs (argsinfo, args, *outconverter,
610 configinfo.saveconf, logout))
611 return false;
612 }
613 ++actionhere;
614 }
615
616 // the action might have changed but we will assume that
617 // the cgiargs were checked properly when the change was made
618
619 return true;
620}
621
622// returns true if cookie already existed, false
623// if it was generated
624bool receptionist::get_cookie (text_t &cookie, text_tmap &fcgienv) {
625
626 text_t cookiestring = gsdl_getenv ("HTTP_COOKIE", fcgienv);
627
628 text_t::const_iterator end = cookiestring.end();
629 text_t::const_iterator here = findchar ((text_t::const_iterator)cookiestring.begin(), end, 'G');
630
631 while (here+9 < end) {
632
633 if (substr(here, here+8) == "GSDL_UID") {
634 cookie = substr (here+9, findchar (here+9, end, ';'));
635 return true;
636 }
637 ++here;
638 here = findchar (here, end, 'G');
639 }
640
641 cookie.clear();
642 text_t host = gsdl_getenv("REMOTE_ADDR", fcgienv);
643 time_t ttime = time(NULL);
644 if (!host.empty()) {
645 cookie += host;
646 cookie.push_back ('-');
647 }
648 cookie += text_t(ttime);
649
650 return false;
651}
652
653// as above but just tests if cookie exists
654bool receptionist::get_cookie (text_tmap &fcgienv) {
655
656 text_t c = gsdl_getenv("HTTP_COOKIE", fcgienv);
657 if (!c.empty()) {
658 text_t cookiestring = c;
659
660 text_t::const_iterator end = cookiestring.end();
661 text_t::const_iterator here = findchar ((text_t::const_iterator)cookiestring.begin(), end, 'G');
662
663 while (here+9 < end) {
664 if (substr(here, here+8) == "GSDL_UID") return true;
665 ++here;
666 here = findchar (here, end, 'G');
667 }
668 }
669 return false;
670}
671
672bool receptionist::log_cgi_args (cgiargsclass &args, ostream &logout, text_tmap &fcgienv) {
673
674 // see if we want to log the cgi arguments
675 if (!configinfo.logcgiargs) return true;
676
677 text_t host = gsdl_getenv ("REMOTE_HOST", fcgienv);
678 text_t script_name = gsdl_getenv ("SCRIPT_NAME", fcgienv);
679 if (host.empty()) host = gsdl_getenv ("REMOTE_ADDR", fcgienv);
680 text_t browser = gsdl_getenv ("HTTP_USER_AGENT", fcgienv);
681
682 cgiargsclass::const_iterator args_here = args.begin();
683 cgiargsclass::const_iterator args_end = args.end();
684
685 text_t argstr;
686 bool first = true;
687 while (args_here != args_end) {
688 if (!first) argstr += ", ";
689 argstr += (*args_here).first + "=" + (*args_here).second.value;
690 first = false;
691 ++args_here;
692 }
693
694 text_t logfile = filename_cat (configinfo.gdbmhome, "etc", "usage.txt");
695
696 text_t logstr = script_name;
697 logstr += " " + host;
698 logstr += " [";
699 if (configinfo.LogDateFormat == UTCTime) {
700 logstr += get_date (false);
701 } else if (configinfo.LogDateFormat == Absolute) {
702 time_t ttime = time(NULL);
703 logstr += ttime;
704 } else {
705 // LocalTime
706 logstr += get_date (true);
707 }
708 logstr += "] (" + argstr + ") \"";
709 logstr += browser;
710 logstr += "\"\n";
711
712 return append_logstr (logfile, logstr, logout);
713}
714
715bool receptionist::append_logstr (const text_t &filename, const text_t &logstr,
716 ostream &logout) {
717
718 utf8outconvertclass text_t2utf8;
719 char *lfile = filename.getcstr();
720
721 int fd = open(lfile, O_WRONLY | O_APPEND);
722
723 if (fd == -1) {
724 logout << "Error: Couldn't open file " << lfile << "\n";
725 delete []lfile;
726 return false;
727 }
728
729 // lock_val is set to 0 if file is locked successfully
730 int lock_val = 1;
731 GSDL_LOCK_FILE (fd);
732 if (lock_val == 0) {
733 text_t tmp_log_str(logstr); // so we don't pass a const to setinput...
734 text_t2utf8.setinput(&tmp_log_str);
735 char *buffer=new char[logstr.size()];
736 size_t num_chars;
737 convertclass::status_t status;
738 text_t2utf8.convert(buffer, logstr.size(), num_chars, status);
739 // ignore status - assume it is "finished" as buffer is big enough
740 write(fd, buffer, num_chars);
741 GSDL_UNLOCK_FILE (fd);
742 delete []buffer;
743 } else {
744 logout << "Error: Couldn't lock file " << lfile << "\n";
745 close(fd);
746 delete []lfile;
747 return false;
748 }
749
750 close(fd);
751
752 delete []lfile;
753 return true;
754}
755
756text_t receptionist::expandmacros (const text_t &astring, cgiargsclass &args,
757 ostream &logout) {
758 text_t outstring;
759 outconvertclass text_t2ascii;
760
761 action *a = actions.getaction (args["a"]);
762 prepare_page (a, args, text_t2ascii, logout);
763 disp.expandstring (displayclass::defaultpackage, astring, outstring);
764 return outstring;
765}
766
767// produce_cgi_page will call get_cgihead_info and
768// produce_content in the appropriate way to output a cgi header and
769// the page content (if needed). If a page could not be created it
770// will return false
771bool receptionist::produce_cgi_page (cgiargsclass &args, ostream &contentout,
772 ostream &logout, text_tmap &fcgienv) {
773 outconvertclass text_t2ascii;
774
775 response_t response;
776 text_t response_data;
777
778 // produce cgi header
779 get_cgihead_info (args, response, response_data, logout, fcgienv);
780 if (response == location) {
781 // location response (url may contain macros!!)
782 response_data = expandmacros (response_data, args, logout);
783 contentout << text_t2ascii << "Location: " << response_data << "\n\n";
784 contentout << flush;
785 return true;
786 } else if (response == content) {
787 // content response
788
789#ifdef GSDL_NOCACHE
790 contentout << "Expires: Mon, 26 Jul 1997 05:00:00 GMT\n"; // date in the past
791 tm *tm_ptr = NULL;
792 time_t t = time(NULL);
793 tm_ptr = gmtime (&t);
794 if (tm_ptr != NULL) {
795 char *timestr = new char[128];
796 strftime (timestr, 128, "%a, %d %b %Y %H:%M:%S", tm_ptr);
797 contentout << "Last-Modified: " << timestr << " GMT\n"; // always modified
798 delete []timestr;
799 }
800 contentout << "Cache-Control: no-cache, must-revalidate\n"; // HTTP/1.1
801 contentout << "Pragma: no-cache\n"; // HTTP/1.0
802
803#else
804
805 // use the later of build.cfg and collect.cfg modification times
806 // as the Last-Modified: header, for caching values
807 struct stat file_info;
808 time_t latest=0;
809
810 text_t collectname="";
811 collectname=args["c"];
812 if (collectname != "") {
813 text_t collectdir=filename_cat(configinfo.gsdlhome,"collect");
814 collectdir=filename_cat(collectdir,collectname);
815 text_t buildcfg=filename_cat(collectdir,"index");
816 buildcfg=filename_cat(buildcfg,"build.cfg");
817 char *buildcfg_ptr=buildcfg.getcstr();
818 text_t collectcfg=filename_cat(collectdir,"etc");
819 collectcfg=filename_cat(collectcfg,"collect.cfg");
820 char *collectcfg_ptr=collectcfg.getcstr();
821
822 if (stat(buildcfg_ptr, &file_info)) {
823 // we got an error. Currently don't handle error :(
824 // logout <<
825 } else {
826 latest=file_info.st_mtime;
827 }
828
829 if (stat(collectcfg_ptr, &file_info)) {
830 // error - unhandled for now
831 } else {
832 if (latest<file_info.st_mtime) latest=file_info.st_mtime;
833 }
834 delete []buildcfg_ptr;
835 delete []collectcfg_ptr;
836
837 if (latest>0) {
838 // print out modified time, "DDD, dd MMM YYYY hh:mm:ss" format
839 // c library takes care of mem for this string... (has \n at end!!!!)
840 // latest is currently local time, convert to UTC.
841 struct tm* utc_latest;
842 utc_latest=gmtime(&latest);
843 contentout << "Last-Modified: " << asctime(utc_latest);
844 }
845 } // end of collection != ""
846
847#endif
848
849 contentout << text_t2ascii << "Content-type: " << response_data << "\n\n";
850 }
851 else if (response == undecided_location) {
852 // Wait until later to output the target location
853 // Used for the "I'm feeling lucky" functionality
854 }
855 else {
856 // unknown response
857 logout << "Error: get_cgihead_info returned an unknown response type.\n";
858 return false;
859 }
860
861 // produce cgi page
862 if (!produce_content (args, contentout, logout)) return false;
863
864 // flush contentout
865 contentout << flush;
866 return true;
867}
868
869
870// get_cgihead_info determines the cgi header information for
871// a set of cgi arguments. If response contains location then
872// response_data contains the redirect address. If reponse
873// contains content then reponse_data contains the content-type.
874// Note that images can now be produced by the receptionist.
875// Note also, alternative for get_cgihead_info below which
876// stores the information in a text_tmap so it is more easily digested
877
878void receptionist::get_cgihead_info (cgiargsclass &args, response_t &response,
879 text_t &response_data, ostream &logout,
880 text_tmap &fcgienv) {
881 outconvertclass text_t2ascii;
882
883 // get the action
884 action *a = actions.getaction (args["a"]);
885 if (a != NULL) {
886 a->get_cgihead_info (args, &protocols, response, response_data, logout);
887
888 } else {
889 // the action was not found!!
890 logout << text_t2ascii << "Error receptionist::get_cgihead_info: the action \""
891 << args["a"] << "\" could not be found.\n";
892 response = content;
893 response_data = "text/html";
894 }
895
896 // add the encoding information
897 if (response == content) {
898 if (converters.find(args["w"]) != converters.end()) {
899 response_data += "; charset=" + args["w"];
900 } else {
901 // default to latin 1
902 response_data += "; charset=ISO-8859-1";
903 }
904
905 // add cookie if required
906 if (configinfo.usecookies && !get_cookie(fcgienv))
907 response_data += "\nSet-Cookie: GSDL_UID=" + args["z"]
908 + "; expires=25-Dec-37 00:00:00 GMT";
909 }
910}
911
912
913// Alternative version of get_cgihead_info, stores fielded infomation
914// in text_tmap rather than concatenated string
915void receptionist::get_cgihead_info (cgiargsclass &args, text_tmap &headers,
916 ostream &logout, text_tmap &fcgienv) {
917
918 response_t response;
919 text_t response_data;
920
921 // get the action
922 action *a = actions.getaction (args["a"]);
923 if (a != NULL) {
924 a->get_cgihead_info (args, &protocols, response, response_data, logout);
925
926 } else {
927 // the action was not found!!
928 outconvertclass text_t2ascii;
929 logout << text_t2ascii << "Error receptionist::get_cgihead_info: the action \""
930 << args["a"] << "\" could not be found.\n";
931 response = content;
932 response_data = "text/html";
933 }
934
935 if (response == location) {
936 response_data = expandmacros(response_data, args, logout);
937 headers["Location"] = response_data;
938 return;
939 }
940
941 // add the encoding information
942 if (response == content) {
943
944 if (converters.find(args["w"]) != converters.end()) {
945 headers["content-encoding"] = args["w"];
946 response_data += "; charset=" + args["w"];
947 } else {
948 // default to utf-8
949 headers["content-encoding"] = "utf-8";
950 response_data += "; charset=utf-8";
951 }
952
953 headers["content-type"] = response_data;
954
955 }
956
957}
958
959
960
961// produce the page content
962bool receptionist::produce_content (cgiargsclass &args, ostream &contentout,
963 ostream &logout) {
964
965 // decide on the output conversion class
966 text_t &arg_w = args["w"];
967 rzwsoutconvertclass defaultoutconverter;
968 rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
969 if (outconverter == NULL) outconverter = &defaultoutconverter;
970 outconverter->reset();
971
972 // needed for 16-bit unicode only - big endian marker 0xfeff (RFC 2781)
973 if (arg_w=="utf-16be") {
974 contentout << '\xfe' << '\xff' ;
975 }
976
977 recptproto *collectproto = protocols.getrecptproto (args["c"], logout);
978 if (collectproto != NULL) {
979 // get browsers to process OID
980 text_t OID = args["d"];
981 if (OID.empty()) OID = args["cl"];
982 if (!OID.empty()) {
983 text_tset metadata;
984 text_tarray OIDs;
985 OIDs.push_back (OID);
986 if (!is_top(OID)) OIDs.push_back (OID + ".pr");
987 FilterResponse_t response;
988 metadata.insert ("childtype");
989 if (get_info (OIDs, args["c"], args["l"], metadata, false, collectproto, response, logout)) {
990 text_t classifytype;
991 if (!response.docInfo[0].metadata["childtype"].values[0].empty())
992 classifytype = response.docInfo[0].metadata["childtype"].values[0];
993 else if (!is_top (OID)) {
994 // not sure why this is occasionally not set, but it will
995 // cause a segfault... possibly if built with no_text? jrm21
996 if (response.docInfo[1].metadata.find("childtype")
997 == response.docInfo[1].metadata.end()) {
998 cerr << "receptionist: no childtype element in metadata map!"
999 << endl;
1000 } else {
1001 if (!response.docInfo[1].metadata["childtype"].values[0].empty())
1002 classifytype = response.docInfo[1].metadata["childtype"].values[0];
1003 }
1004 }
1005 browserclass *b = browsers.getbrowser (classifytype);
1006 b->processOID (args, collectproto, logout);
1007 }
1008 }
1009
1010 // translate "d" and "cl" arguments if required
1011 translate_OIDs (args, collectproto, logout);
1012 }
1013
1014 // produce the page using the desired action
1015 action *a = actions.getaction (args["a"]);
1016 if (a != NULL) {
1017 if (a->uses_display(args)) prepare_page (a, args, (*outconverter), logout);
1018 if (!a->do_action (args, &protocols, &browsers, disp, (*outconverter), contentout, logout))
1019 return false;
1020 } else {
1021 // the action was not found!!
1022 outconvertclass text_t2ascii;
1023
1024 logout << text_t2ascii << "Error receptionist::produce_content: the action \""
1025 << args["a"] << "\" could not be found.\n";
1026
1027 contentout << (*outconverter)
1028 << "<html>\n"
1029 << "<head>\n"
1030 << "<title>Error</title>\n"
1031 << "</head>\n"
1032 << "<body>\n"
1033 << "<h2>Oops!</h2>\n"
1034 << "Undefined Page. The action \""
1035 << args["a"] << "\" could not be found.\n"
1036 << "</body>\n"
1037 << "</html>\n";
1038 }
1039 return true;
1040}
1041
1042
1043// returns the compressed argument ("e") corresponding to the argument
1044// list. This can be used to save preferences between sessions.
1045text_t receptionist::get_compressed_arg (cgiargsclass &args, ostream &logout) {
1046 // decide on the output conversion class
1047 text_t &arg_w = args["w"];
1048 rzwsoutconvertclass defaultoutconverter;
1049 rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
1050 if (outconverter == NULL) outconverter = &defaultoutconverter;
1051 outconverter->reset();
1052
1053 text_t compressed_args;
1054 if (compress_save_args (argsinfo, configinfo.saveconf, args,
1055 compressed_args, *outconverter, logout))
1056 return compressed_args;
1057
1058 return g_EmptyText;
1059}
1060
1061
1062// will read in all the macro files. If one is not found an
1063// error message will be written to logout and the method will
1064// return false.
1065bool receptionist::read_macrofiles (ostream &logout) {
1066 outconvertclass text_t2ascii;
1067
1068 // redirect the error output to logout
1069 ostream *savedlogout = disp.setlogout (&logout);
1070
1071 // unload any macros that were previously loaded - this allows us to call
1072 // this function a second time to reload all the macro files (useful for
1073 // reading in changed macro files in server versions of greenstone)
1074 disp.unloaddefaultmacros();
1075
1076 // load up the default macro files, the collection directory
1077 // is searched first for the file (if this is being used in
1078 // collection specific mode) and then the main directory(s)
1079 text_t colmacrodir = filename_cat (configinfo.collectdir, "macros");
1080
1081 text_tset maindirs;
1082 text_t gsdlmacrodir = filename_cat (configinfo.gsdlhome, "macros");
1083 maindirs.insert (gsdlmacrodir);
1084 colinfo_tmap::iterator colhere = configinfo.collectinfo.begin();
1085 colinfo_tmap::iterator colend = configinfo.collectinfo.end();
1086 while (colhere != colend) {
1087 if (!((*colhere).second.gsdl_gsdlhome).empty()) {
1088 gsdlmacrodir = filename_cat ((*colhere).second.gsdl_gsdlhome, "macros");
1089 maindirs.insert (gsdlmacrodir);
1090 }
1091 ++colhere;
1092 }
1093
1094 text_tset::iterator arrhere = configinfo.macrofiles.begin();
1095 text_tset::iterator arrend = configinfo.macrofiles.end();
1096 text_t filename;
1097 while (arrhere != arrend) {
1098 bool foundfile = false;
1099
1100 // try in the collection directory if this is being
1101 // run in collection specific mode
1102 if (!configinfo.collection.empty()) {
1103 filename = filename_cat (colmacrodir, *arrhere);
1104 if (file_exists (filename)) {
1105 disp.loaddefaultmacros(filename);
1106 foundfile = true;
1107 }
1108 }
1109
1110 // if we haven't found the macro file yet try in
1111 // the main macro directory(s)
1112 // if file is found in more than one main directory
1113 // we'll load all copies
1114 if (!foundfile) {
1115 text_tset::const_iterator dirhere = maindirs.begin();
1116 text_tset::const_iterator dirend = maindirs.end();
1117 while (dirhere != dirend) {
1118 filename = filename_cat (*dirhere, *arrhere);
1119 if (file_exists (filename)) {
1120 disp.loaddefaultmacros(filename);
1121 foundfile = true;
1122 }
1123 ++dirhere;
1124 }
1125 }
1126
1127 // see if we found the file or not
1128 if (!foundfile) {
1129 logout << text_t2ascii
1130 << "Error: the macro file \"" << *arrhere << "\" could not be found.\n";
1131 if (configinfo.collection.empty()) {
1132 text_t dirs;
1133 joinchar (maindirs, ", ", dirs);
1134 logout << text_t2ascii
1135 << "It should be in either of the following directories ("
1136 << dirs << ").\n\n";
1137
1138 } else {
1139 logout << text_t2ascii
1140 << "It should be in either " << colmacrodir << " or in "
1141 << gsdlmacrodir << ".\n\n";
1142 }
1143 // reset logout to what it was
1144 disp.setlogout (savedlogout);
1145 return false;
1146 }
1147 ++arrhere;
1148 }
1149
1150 // success
1151
1152 // reset logout to what it was
1153 disp.setlogout (savedlogout);
1154 return true;
1155}
1156
1157
1158
1159
1160// Go through the list of macro files looking to see
1161// if any exist in the collectoin specific area. If they
1162// do then read them in and add them to the set of existing
1163// current macros
1164
1165void receptionist::read_collection_macrofiles (const text_t& collection, ostream &logout)
1166{
1167 outconvertclass text_t2ascii;
1168
1169 // disp.unloadcollectionmacros();
1170
1171 // redirect the error output to logout
1172 ostream *savedlogout = disp.setlogout (&logout);
1173
1174 text_t colmacrodir
1175 = filename_cat (configinfo.gsdlhome,"collect",collection, "macros");
1176
1177 if (directory_exists (colmacrodir)) {
1178
1179 text_tset::iterator arrhere = configinfo.macrofiles.begin();
1180 text_tset::iterator arrend = configinfo.macrofiles.end();
1181 text_t filename;
1182 while (arrhere != arrend) {
1183
1184 filename = filename_cat (colmacrodir, *arrhere);
1185 if (file_exists (filename)) {
1186 disp.loadcollectionmacros(filename);
1187 }
1188
1189 ++arrhere;
1190 }
1191 }
1192
1193 // reset logout to what it was
1194 disp.setlogout (savedlogout);
1195}
1196
1197
1198
1199
1200// check_mainargs will check all the main arguments. If a major
1201// error is found it will return false and no cgi page should
1202// be created using the arguments.
1203
1204bool receptionist::check_mainargs (cgiargsclass &args, ostream &logout) {
1205
1206 if(configinfo.site_auth)
1207 {
1208 args["uan"] = "1";
1209 args["ug"] = configinfo.site_group;
1210 }
1211
1212
1213 // if this receptionist is running in collection dependant mode
1214 // then it should always set the collection argument to the
1215 // collection
1216 if (!configinfo.collection.empty()) args["c"] = configinfo.collection;
1217
1218 // if current collection uses ccscols make sure
1219 // "ccs" argument is set and make "cc" default to
1220 // all collections in "ccs"
1221 if (args["a"] != "config" && !args["c"].empty()) {
1222
1223 text_t &arg_c = args["c"];
1224 recptproto *collectproto = protocols.getrecptproto (arg_c, logout);
1225 if (collectproto == NULL) {
1226 // oops, this collection isn't valid
1227 outconvertclass text_t2ascii;
1228 logout << text_t2ascii << "ERROR: Invalid collection: " << arg_c << "\n";
1229 // args["c"].clear();
1230
1231 } else {
1232
1233 ColInfoResponse_t *cinfo = get_collectinfo_ptr (collectproto, arg_c, logout);
1234
1235 if(cinfo->authenticate == "collection")
1236 {
1237 args["uan"] = "1";
1238 args["ug"] = cinfo->auth_group;
1239 }
1240
1241
1242 if (cinfo != NULL) {
1243 if (!cinfo->ccsCols.empty()) {
1244 args["ccs"] = 1;
1245 if (args["cc"].empty()) {
1246 text_tarray::const_iterator col_here = cinfo->ccsCols.begin();
1247 text_tarray::const_iterator col_end = cinfo->ccsCols.end();
1248 bool first = true;
1249 while (col_here != col_end) {
1250 // make sure it's a valid collection
1251 if (protocols.getrecptproto (*col_here, logout) != NULL) {
1252 if (!first) args["cc"].push_back (',');
1253 args["cc"] += *col_here;
1254 first = false;
1255 }
1256 ++col_here;
1257 }
1258 }
1259 }
1260 } else {
1261 logout << "ERROR (receptionist::check_mainargs): get_collectinfo_ptr returned NULL\n";
1262 }
1263 }
1264 }
1265
1266 // argument "v" can only be 0 or 1. Use the default value
1267 // if it is out of range
1268 int arg_v = args.getintarg ("v");
1269 if (arg_v != 0 && arg_v != 1) {
1270 cgiarginfo *vinfo = argsinfo.getarginfo ("v");
1271 if (vinfo != NULL) args["v"] = vinfo->argdefault;
1272 }
1273
1274 // argument "f" can only be 0 or 1. Use the default value
1275 // if it is out of range
1276 int arg_f = args.getintarg ("f");
1277 if (arg_f != 0 && arg_f != 1) {
1278 cgiarginfo *finfo = argsinfo.getarginfo ("f");
1279 if (finfo != NULL) args["f"] = finfo->argdefault;
1280 }
1281
1282 return true;
1283}
1284
1285// translate_OIDs translates the "d" and "cl" arguments to their correct values
1286// if they use the tricky ".fc", ".lc" type syntax.
1287void receptionist::translate_OIDs (cgiargsclass &args, recptproto *collectproto,
1288 ostream &logout) {
1289
1290 FilterResponse_t response;
1291 FilterRequest_t request;
1292 comerror_t err;
1293 text_t &arg_d = args["d"];
1294 text_t &arg_cl = args["cl"];
1295 text_t &collection = args["c"];
1296
1297 // do a call to translate OIDs if required
1298 request.filterName = "NullFilter";
1299 request.filterResultOptions = FROID;
1300 if (!arg_d.empty() && needs_translating (arg_d)) {
1301 request.docSet.push_back (arg_d);
1302 collectproto->filter (collection, request, response, err, logout);
1303 arg_d = response.docInfo[0].OID;
1304 request.clear();
1305 }
1306 // we'll also check here that the "cl" argument has a "classify" doctype
1307 // (in case ".fc" or ".lc" have screwed up)
1308 if (needs_translating (arg_cl)) {
1309 request.fields.insert ("doctype");
1310 request.docSet.push_back (arg_cl);
1311 request.filterResultOptions = FRmetadata;
1312 collectproto->filter (collection, request, response, err, logout);
1313 // set to original value (without .xx stuff) if doctype isn't "classify"
1314 if (response.docInfo[0].metadata["doctype"].values[0] != "classify")
1315 strip_suffix (arg_cl);
1316 else
1317 arg_cl = response.docInfo[0].OID;
1318 }
1319}
1320
1321// prepare_page sets up page parameters, sets display macros
1322// and opens the page ready for output
1323void receptionist::prepare_page (action *a, cgiargsclass &args,
1324 outconvertclass &outconvert,
1325 ostream &logout) {
1326 // set up page parameters
1327 text_t pageparams;
1328 bool first = true;
1329
1330 text_tmap::iterator params_here = configinfo.pageparams.begin();
1331 text_tmap::iterator params_end = configinfo.pageparams.end();
1332 while (params_here != params_end) {
1333 // page params are those from main.cfg (eg pageparam v 0) plus
1334 // two defaults set in recptconf.clear() (c="" and l=en)
1335 // This used to check if the current value of the page param
1336 // == the default value, then don't add in it the list
1337 // but if l=en, and there is a macro with [l=en], then it doesn't
1338 // find it.
1339 // so now all page params will go into the list. I assume this will
1340 // mean more attempts to find each macro, but nothing worsee than
1341 // that. --kjdon
1342 //if (args[(*params_here).first] != (*params_here).second) {
1343 if (first)
1344 first = false;
1345 else
1346 pageparams += ",";
1347
1348 pageparams += (*params_here).first;
1349 pageparams += "=";
1350 pageparams += args[(*params_here).first];
1351 // }
1352
1353 ++params_here;
1354 }
1355
1356
1357 // open the page
1358 disp.openpage(pageparams, configinfo.macroprecedence);
1359
1360 disp.unloadcollectionmacros();
1361
1362 text_t collection = args["c"];
1363 if (!collection.empty()) {
1364 read_collection_macrofiles(collection,logout);
1365 }
1366
1367 // define external macros for each action
1368 actionptrmap::iterator actionhere = actions.begin ();
1369 actionptrmap::iterator actionend = actions.end ();
1370
1371 while (actionhere != actionend) {
1372 assert ((*actionhere).second.a != NULL);
1373 if ((*actionhere).second.a != NULL) {
1374 (*actionhere).second.a->define_external_macros (disp, args, &protocols, logout);
1375 }
1376 ++actionhere;
1377 }
1378
1379
1380 // define internal macros for the current action
1381 a->define_internal_macros (disp, args, &protocols, logout);
1382
1383 // define general macros. the defining of general macros is done here so that
1384 // the last possible version of the cgi arguments are used
1385 define_general_macros (args, outconvert, logout);
1386}
1387
1388void receptionist::define_general_macros (cgiargsclass &args, outconvertclass &/*outconvert*/,
1389 ostream &logout) {
1390
1391 text_t &collection = args["c"];
1392
1393 disp.setmacro ("gsdlhome", displayclass::defaultpackage, dm_safe(configinfo.gsdlhome));
1394 disp.setmacro ("gwcgi", displayclass::defaultpackage, configinfo.gwcgi);
1395 disp.setmacro ("httpimg", displayclass::defaultpackage, configinfo.httpimg);
1396 disp.setmacro ("httpprefix", displayclass::defaultpackage, configinfo.httpprefix);
1397
1398 // get relevant info from the collection
1399 if (!collection.empty()) {
1400 ColInfoResponse_t cinfo;
1401 comerror_t err;
1402 recptproto *collectproto = protocols.getrecptproto (collection, logout);
1403 if (collectproto != NULL) {
1404 collectproto->get_collectinfo (collection, cinfo, err, logout);
1405 text_t httpcollection;
1406 if (!cinfo.httpdomain.empty()) httpcollection = "http://";
1407 httpcollection += cinfo.httpdomain + cinfo.httpprefix + "/collect/"
1408 + collection;
1409 disp.setmacro ("httpcollection", displayclass::defaultpackage,
1410 httpcollection);
1411 // as of gsdl 2.53, collect.cfg can specify macros
1412 if (cinfo.collection_macros.size() > 0) {
1413 macros_map::const_iterator this_macro=cinfo.collection_macros.begin();
1414 macros_map::const_iterator done_macro=cinfo.collection_macros.end();
1415 while (this_macro != done_macro) {
1416
1417 disp.setcollectionmacro("Global", // package
1418 this_macro->first, // macro name
1419 this_macro->second.first, // params
1420 this_macro->second.second); // value
1421 ++this_macro;
1422 }
1423 } // col macros
1424 } // collectproto != NULL
1425 }
1426
1427 text_t compressedoptions = get_compressed_arg(args, logout);
1428 disp.setmacro ("compressedoptions", displayclass::defaultpackage, dm_safe(compressedoptions));
1429 // need a decoded version of compressedoptions for use within forms
1430 // as browsers encode values from forms before sending to server
1431 // (e.g. %25 becomes %2525)
1432 decode_cgi_arg (compressedoptions);
1433 disp.setmacro ("decodedcompressedoptions", displayclass::defaultpackage, dm_safe(compressedoptions));
1434
1435#if defined (__WIN32__)
1436 disp.setmacro ("win32", displayclass::defaultpackage, "1");
1437#endif
1438
1439 // set _cgiargX_ macros for each cgi argument
1440 cgiargsclass::const_iterator argshere = args.begin();
1441 cgiargsclass::const_iterator argsend = args.end();
1442 while (argshere != argsend) {
1443 if (((*argshere).first == "q") ||
1444 ((*argshere).first == "qa") ||
1445 ((*argshere).first == "qtt") ||
1446 ((*argshere).first == "qty") ||
1447 ((*argshere).first == "qp") ||
1448 ((*argshere).first == "qpl") ||
1449 ((*argshere).first == "qr") ||
1450 ((*argshere).first == "q2"))
1451 // need to escape special characters from query string
1452 disp.setmacro ("cgiarg" + (*argshere).first,
1453 displayclass::defaultpackage, html_safe((*argshere).second.value));
1454 else if ((*argshere).first == "hp") {
1455 disp.setmacro ("cgiarg" + (*argshere).first, displayclass::defaultpackage, (*argshere).second.value);
1456 } else {
1457 disp.setmacro ("cgiarg" + (*argshere).first, displayclass::defaultpackage, dm_safe((*argshere).second.value));
1458 }
1459 ++argshere;
1460 }
1461
1462 // display text right to left if language is arabic/farsi (and if browser can support it)
1463 if (args["l"] == "ar" || args["l"] == "fa")
1464 disp.setmacro ("htmlextra", displayclass::defaultpackage, " dir=rtl");
1465
1466 // set collection specific macros
1467 if (!collection.empty()) {
1468 recptproto *collectproto = protocols.getrecptproto (collection, logout);
1469 if (collectproto != NULL) {
1470 FilterResponse_t response;
1471 text_tset metadata;
1472 get_info ("collection", collection, args["l"], metadata, false,
1473 collectproto, response, logout);
1474
1475 if (!response.docInfo[0].metadata.empty()) {
1476 MetadataInfo_tmap::const_iterator here = response.docInfo[0].metadata.begin();
1477 MetadataInfo_tmap::const_iterator end = response.docInfo[0].metadata.end();
1478 while (here != end) {
1479 if (((*here).first != "haschildren") && ((*here).first != "hasnext") &&
1480 ((*here).first != "hasprevious")) {
1481 // check for args in form name:lang
1482 text_t name = g_EmptyText;
1483 text_t lang = g_EmptyText;
1484 bool colonfound=false;
1485 text_t::const_iterator a = (*here).first.begin();
1486 text_t::const_iterator b = (*here).first.end();
1487 while (a !=b) {
1488 if (*a==':') {
1489 colonfound=true;
1490 }
1491 else {
1492 if (colonfound)
1493 lang.push_back(*a);
1494 else name.push_back(*a);
1495 }
1496 ++a;
1497 }
1498 if (!lang.empty()) {
1499 if (args["l"]==lang) {
1500 disp.setmacro (name, displayclass::defaultpackage, (*here).second.values[0]);
1501 }
1502 }
1503 else { // the default one
1504 disp.setmacro ((*here).first, displayclass::defaultpackage, (*here).second.values[0]);
1505 }
1506 }
1507 ++here;
1508 }
1509 }
1510
1511 text_t iconcollection;
1512 disp.expandstring (displayclass::defaultpackage, "_iconcollection_", iconcollection);
1513 if (!iconcollection.empty())
1514 {
1515 ColInfoResponse_t cinfo;
1516 comerror_t err;
1517 collectproto->get_collectinfo (collection, cinfo, err, logout);
1518 if (iconcollection[0]=='/' && !cinfo.httpdomain.empty())
1519 {
1520 // local but with full path
1521 iconcollection = "http://" + cinfo.httpdomain + iconcollection;
1522 disp.setmacro("iconcollection", displayclass::defaultpackage, iconcollection);
1523 }
1524 }
1525 }
1526 }
1527}
1528
1529// gets collection info from cache if found or
1530// calls collection server (and updates cache)
1531// returns NULL if there's an error
1532ColInfoResponse_t *receptionist::get_collectinfo_ptr (recptproto *collectproto,
1533 const text_t &collection,
1534 ostream &logout) {
1535
1536 // check the cache
1537 colinfo_tmap::iterator it = configinfo.collectinfo.find (collection);
1538 if ((it != configinfo.collectinfo.end()) && ((*it).second.info_loaded)) {
1539 // found it
1540 return &((*it).second.info);
1541 }
1542
1543 // not cached, get info from collection server
1544 if (collectproto == NULL) {
1545 logout << "ERROR: receptionist::get_collectinfo_ptr passed null collectproto\n";
1546 return NULL;
1547 }
1548
1549 comerror_t err;
1550 if (it == configinfo.collectinfo.end()) {
1551 collectioninfo_t cinfo;
1552 collectproto->get_collectinfo (collection, cinfo.info, err, logout);
1553 if (err != noError) {
1554 outconvertclass text_t2ascii;
1555 logout << text_t2ascii << "ERROR (receptionist::getcollectinfo_ptr): \""
1556 << get_comerror_string (err) << "\"while getting collectinfo\n";
1557 return NULL;
1558 }
1559 cinfo.info_loaded = true;
1560 configinfo.collectinfo[collection] = cinfo;
1561 return &(configinfo.collectinfo[collection].info);
1562 } else {
1563 collectproto->get_collectinfo (collection, (*it).second.info, err, logout);
1564 if (err != noError) {
1565 outconvertclass text_t2ascii;
1566 logout << text_t2ascii << "ERROR (receptionist::getcollectinfo_ptr): \""
1567 << get_comerror_string (err) << "\"while getting collectinfo\n";
1568 return NULL;
1569 }
1570 (*it).second.info_loaded = true;
1571 return &((*it).second.info);
1572 }
1573}
1574
1575// removes a collection from the cache so that the next
1576// call to get_collectinfo_ptr() for that collection will
1577// retrieve the collection info from the collection server
1578void receptionist::uncache_collection (const text_t &collection) {
1579
1580 colinfo_tmap::iterator it = configinfo.collectinfo.find (collection);
1581 if ((it != configinfo.collectinfo.end()) && ((*it).second.info_loaded)) {
1582
1583 (*it).second.info_loaded = false;
1584
1585 }
1586}
1587
1588// Handles an "Encoding" line from a configuration file - note that the
1589// configinfo.encodings map is a bit of a hack (to be fixed when the
1590// configuration files are tidied up).
1591void receptionist::configure_encoding (const text_tarray &cfgline) {
1592
1593 text_t subkey, subvalue, shortname, longname, mapfile;
1594 int multibyte = 0;
1595 text_t::const_iterator cfglinesub_here;
1596 text_tarray::const_iterator cfgline_here = cfgline.begin();
1597 text_tarray::const_iterator cfgline_end = cfgline.end();
1598 while (cfgline_here != cfgline_end) {
1599 if (*cfgline_here == "multibyte") {
1600 multibyte = 1;
1601 } else {
1602 cfglinesub_here = getdelimitstr((*cfgline_here).begin(),
1603 (*cfgline_here).end(), '=', subkey);
1604 if (subkey == "shortname") {
1605 shortname = substr (cfglinesub_here, (*cfgline_here).end());
1606 } else if (subkey == "longname") {
1607 longname = substr (cfglinesub_here, (*cfgline_here).end());
1608 } else if (subkey == "map") {
1609 mapfile = substr (cfglinesub_here, (*cfgline_here).end());
1610 }
1611 }
1612 ++cfgline_here;
1613 }
1614 if (!shortname.empty()) {
1615 if (longname.empty()) longname = shortname;
1616
1617 // add the converter
1618 if (shortname == "utf-8") {
1619 utf8inconvertclass *utf8inconvert = new utf8inconvertclass();
1620 utf8outconvertclass *utf8outconvert = new utf8outconvertclass();
1621 utf8outconvert->set_rzws(1);
1622 add_converter (shortname, utf8inconvert, utf8outconvert);
1623 configinfo.encodings[longname] = shortname;
1624
1625 } else if (shortname == "utf-16be") {
1626 // we use the default input converter as this shouldn't ever be used
1627 // for converting from unicode...
1628 inconvertclass *inconverter = new inconvertclass();
1629 utf16outconvertclass *outconverter = new utf16outconvertclass();
1630 add_converter (shortname, inconverter, outconverter);
1631 configinfo.encodings[longname] = shortname;
1632
1633 } else if (!mapfile.empty()) {
1634
1635 if (mapfile == "8859_1.ump") {
1636 // iso-8859-1 is a special case as it'll always be supported by the
1637 // standard converter class and therefore doesn't need to use its
1638 // mapping file
1639 inconvertclass *inconvert = new inconvertclass();
1640 rzwsoutconvertclass *outconvert = new rzwsoutconvertclass();
1641 outconvert->set_rzws(1);
1642 add_converter (shortname, inconvert, outconvert);
1643 configinfo.encodings[longname] = shortname;
1644
1645 } else {
1646 text_t to_uc_map = filename_cat(configinfo.gsdlhome, "mappings", "to_uc", mapfile);
1647 text_t from_uc_map = filename_cat(configinfo.gsdlhome, "mappings", "from_uc", mapfile);
1648 if (file_exists(to_uc_map) && file_exists(from_uc_map)) {
1649
1650 mapinconvertclass *mapinconvert = new mapinconvertclass();
1651 mapinconvert->setmapfile (to_uc_map, 0x003F);
1652 mapinconvert->set_multibyte (multibyte);
1653 mapoutconvertclass *mapoutconvert = new mapoutconvertclass();
1654 mapoutconvert->setmapfile (from_uc_map, 0x3F);
1655 mapoutconvert->set_multibyte (multibyte);
1656 mapoutconvert->set_rzws(1);
1657 add_converter (shortname, mapinconvert, mapoutconvert);
1658 configinfo.encodings[longname] = shortname;
1659 }
1660 }
1661 }
1662 }
1663}
Note: See TracBrowser for help on using the repository browser.