root/main/trunk/greenstone2/runtime-src/src/recpt/cgiutils.cpp @ 22796

Revision 22796, 26.4 KB (checked in by mdewsnip, 9 years ago)

Need to add <time.h> so my last change compiles on Windows.

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
Line 
1/**********************************************************************
2 *
3 * cgiutils.cpp -- general cgi utilities
4 * Copyright (C) 1999  The New Zealand Digital Library Project
5 *
6 * A component of the Greenstone digital library software
7 * from the New Zealand Digital Library Project at the
8 * University of Waikato, New Zealand.
9 *
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
14 *
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18 * GNU General Public License for more details.
19 *
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
23 *
24 *********************************************************************/
25
26#include "cgiutils.h"
27#include "fileutil.h"
28#include "gsdlunicode.h"
29#include "fileutil.h"
30#include "unitool.h" // in mg, for output_utf8_char
31#include <cstdlib>
32#include <time.h>
33
34#if defined(GSDL_USE_OBJECTSPACE)
35#  include <ospace\std\iostream>
36#  include <ospace\std\fstream>
37#elif defined(GSDL_USE_IOS_H)
38#  include <iostream.h>
39#  include <fstream.h>
40#else
41#  include <iostream>
42#  include <fstream>
43#endif
44
45
46static unsigned short hexdigit (unsigned short c) {
47  if (c >= '0' && c <= '9') return (c-'0');
48  if (c >= 'a' && c <= 'f') return (c-'a'+10);
49  if (c >= 'A' && c <= 'F') return (c-'A'+10);
50  return c;
51}
52
53
54static void c2hex (unsigned short c, text_t &t) {
55  t.clear();
56 
57  if (c >= 256) {
58    t = "20"; // ' '
59    return;
60  }
61 
62  unsigned short o1, o2;
63 
64  o1 = (c/16) % 16;
65  o2 = c % 16;
66  if (o1 >= 10) o1 += 'a' - 10;
67  else o1 += '0';
68  if (o2 >= 10) o2 += 'a' - 10;
69  else o2 += '0';
70 
71  t.push_back(o1);
72  t.push_back(o2);
73}
74
75static text_t::iterator getline (text_t::iterator first,
76                 text_t::iterator last,
77                 bool include_crlf) {
78  while (first != last) {
79    if (((first+1) != last) && (*first == 13) && (*(first+1) == 10)) {
80      // found <CRLF>
81      if (include_crlf) first += 2;
82      break;
83    }
84
85    first++;
86  }
87
88  return first;
89}
90
91static void process_post_section (text_t &argname, text_t &argdata, text_t &filename, text_t &filedata,
92                  text_t &filetype, bool &isfile, text_t &argstr,
93                  fileupload_tmap &fileuploads, const text_t &gsdlhome) {
94
95  if (!argname.empty()) {
96
97    if (!isfile) {
98      // argdata includes a trailing <CRLF> that we must remove
99      if ((argdata.size() > 1) && (*(argdata.end()-2) == 13) && (*(argdata.end()-1) == 10)) {
100    argdata.erase(argdata.end()-2, argdata.end());
101      }
102      if (!argstr.empty()) argstr += "&";
103      argstr += argname + "=" + argdata;
104
105    } else if (!filename.empty()) {
106      // filedata includes a trailing <CRLF> that we must remove
107      if ((filedata.size() > 1) && (*(filedata.end()-2) == 13) && (*(filedata.end()-1) == 10)) {
108    filedata.erase(filedata.end()-2, filedata.end());
109      }
110
111      // create tmp_name for storing the file on disk, using the current timestamp
112      text_t tmp_name(time(NULL));
113      tmp_name = filename_cat(gsdlhome, "tmp", tmp_name);
114
115      char *tmp_name_c = tmp_name.getcstr();
116
117      // write the file data to disk
118      outconvertclass out;
119      ofstream filestream(tmp_name_c, ios::out | ios::binary);
120      filestream << out << filedata;
121      filestream.close();
122      delete tmp_name_c;
123
124      // populate the fields of a fileupload_t and put it in the
125      // fileuploads map
126      fileupload_t fu;
127      // note that filename currently may or may not include the path since
128      // some browsers (e.g. IE) include the path while others
129      // (e.g. mozilla) do not. we should probably remove the path from
130      // this field here to get a consistent value across all browsers.
131      text_t::iterator slash = findlastchar(filename.begin(), filename.end(), '\\');
132      if (slash != filename.end()) {
133    filename = substr(slash+1, filename.end());
134      }
135      fu.name = filename;
136      fu.type = filetype;
137      // size has yet to be implemented
138      fu.size = filedata.size();
139
140      fu.tmp_name = tmp_name;
141      fileuploads[argname] = fu;
142    }
143  }
144  isfile = false;
145  argname.clear();
146  argdata.clear();
147  filename.clear();
148  filedata.clear();
149  filetype.clear();
150}
151
152// parse data obtained through a CGI POST request
153text_t parse_post_data (text_t &content_type, text_t &raw_post_data,
154            fileupload_tmap &fileuploads, const text_t &gsdlhome) {
155
156  text_t argstr;
157
158  text_t::iterator content_type_begin = content_type.begin();
159  text_t::iterator content_type_end = content_type.end();
160  if (findword(content_type_begin, content_type_end, "multipart/form-data") == content_type_end) {
161    // a simple post request
162
163    return raw_post_data;
164
165  } else {
166    // multipart/form data - may contain one or more uploaded files
167
168    /*
169      content_type should look something like the following
170        multipart/form-data; boundary=---------------------------7d411e1a50330
171   
172      while raw_post_data will be as follows
173        -----------------------------7d43e73450330CRLF
174    Content-Disposition: form-data; name="e"<CRLF>
175    <CRLF>
176    d-0testss--1-0-00---4----0--0-110--1en-Zz-1---10-about-0--00031-001utfZz-8-0<CRLF>
177    -----------------------------7d43e73450330<CRLF>
178    Content-Disposition: form-data; name="afile"; filename="C:\somedoc.doc"<CRLF>
179    Content-Type: application/msword<CRLF>
180    <CRLF>
181    <Content of file><CRLF>
182   
183    */
184
185    // first get the boundary from content-type
186    text_t::iterator boundary_begin = findword(content_type_begin, content_type_end, "boundary=");
187    if (boundary_begin+9 < content_type_end)
188      {
189    // skip over "boundary=" part of string
190    boundary_begin += 9;
191      }
192    else {
193      // error
194      cerr << "Error: malformed boundary? '" <<  content_type << "'" << endl;
195      return "";
196    }
197    text_t boundary = substr(boundary_begin, getline(boundary_begin, content_type_end, false));
198    int boundary_len = boundary.size();
199
200
201    text_t argname, argdata, filename, filedata, filetype;
202    bool isfile = false;
203    text_t::iterator data_here = raw_post_data.begin();
204    text_t::iterator data_end = raw_post_data.end();
205    while (data_here != data_end) {
206     
207      // get the next available line (including the trailing <CRLF>
208      text_t line = substr(data_here, getline(data_here, data_end, true));
209
210      data_here += line.size();
211      text_t::iterator line_begin = line.begin();
212      text_t::iterator line_end = line.end();
213      if (findword(line_begin, line_end, boundary) != line_end) {
214    // we've found a boundary
215    process_post_section(argname, argdata, filename, filedata, filetype,
216                 isfile, argstr, fileuploads, gsdlhome);
217
218      } else if (findword(line_begin, line_end, "Content-Disposition: form-data") != line_end) {
219    // we've found the the beginning of a new section
220    argname.clear();
221    argdata.clear();
222
223    // get the name of this piece of form data
224    text_t::iterator it = findword(line_begin, line_end, "name=\"");
225    if (it == line_end) break; // error - this shouldn't happen
226    it = findchar(it, line_end, '"');
227    if ((it != line_end) && (it+1 != line_end)) {
228      argname = substr(it+1, findchar(it+1, line_end, '"'));
229    }
230
231    // if this piece of form data contains filename="" it's a file
232    // upload and needs to be treated special
233    it = (findword(line_begin, line_end, "filename=\""));
234    if (it != line_end) {
235      // we've found a file upload
236      isfile = true;
237      it = findchar(it, line_end, '"');
238      if ((it != line_end) && (it+1 != line_end)) {
239        filename = substr(it+1, findchar(it+1, line_end, '"'));
240      }
241
242      // the next line is the content-type of this section
243      line = substr(data_here, getline(data_here, data_end, true));
244      data_here += line.size();
245      line_begin = line.begin();
246      line_end = line.end();
247      it = (findword(line_begin, line_end, "Content-Type: "));
248      if (it != line_end) {
249        filetype = substr(it+14, getline(it, line_end, false));
250      }
251    }
252
253    // eat up the next line as it's just a <CRLF> on it's own
254    data_here += 2;
255
256      } else {
257    if (isfile) filedata += line;
258    else argdata += line;
259      }
260
261    }
262
263    // process last section
264    process_post_section(argname, argdata, filename, filedata, filetype,
265             isfile, argstr, fileuploads, gsdlhome);
266
267    return argstr;
268  }
269}
270
271// convert %xx and + to their appropriate equivalents
272// IE 6.0 and later use "%u" followed by 4 hex digits... MS IIS extension!
273// NOTE: this method is crap. It assumes the input encoding is utf-8. If it
274// actually was, then this returns utf-8, and needs to_uni on the
275// result to get it back to unicode. If the encoding wasn't utf-8, then the
276// output may be crap. Seems to work for 8 bit encodings.
277// Really, this should be given the encoding, and should always return unicode.
278void decode_cgi_arg (text_t &argstr) {
279  text_t::iterator in = argstr.begin();
280  text_t::iterator out = in;
281  text_t::iterator end = argstr.end();
282 
283  while (in != end) {
284    if (*in == '+') *out = ' ';
285   
286    else if (*in == '%') {
287      unsigned short c = '%';
288      ++in;
289      if (in != end) { // this is an encoding...
290    if (*in == 'u') { // convert %uHHHH to unicode then current encoding
291      // this assumes a short int is at least 16 bits...
292      ++in; 
293      if (in != end)
294        c=hexdigit(*in++) << 12;
295      if (in != end)
296        c+=hexdigit(*in++) << 8;
297      if (in != end)
298        c+=hexdigit(*in++) << 4;
299      if (in != end)
300        c+=hexdigit(*in);
301      /* BAD!! The following assumes the interface is using utf-8. But
302         at this point we don't know what encoding we are using, unless
303         we can parse it out of the string we are currently decoding... */
304      text_t uni=" ";
305      uni[0]=c;
306      text_t utf8=to_utf8(uni);
307      int last_byte=utf8.size()-1;
308      for (int i=0;i<last_byte;++i)
309        *out++ = utf8[i];
310      c=utf8[last_byte];
311    } else {  // convert %HH to hex value
312      c = hexdigit (*in);
313      ++in;
314      if (in != end && c < 16) { // sanity check on the previous character
315        c = c*16 + hexdigit (*in);
316      }
317    }
318      }
319      *out = c;
320    } else *out = *in;
321   
322    if (in != end) ++in;
323    ++out;
324  }
325 
326  // remove the excess characters
327  argstr.erase (out, end);
328 
329}
330
331
332// split up the cgi arguments
333void split_cgi_args (const cgiargsinfoclass &argsinfo, text_t argstr,
334             cgiargsclass &args) {
335  args.clear();
336 
337  text_t::const_iterator here = argstr.begin();
338  text_t::const_iterator end = argstr.end();
339  text_t key, value;
340 
341  // extract out the key=value pairs
342  while (here != end) {
343    // get the next key and value pair
344    here = getdelimitstr (here, end, '=', key);
345    here = getdelimitstr (here, end, '&', value);
346   
347    // convert %xx and + to their appropriate equivalents
348    decode_cgi_arg (value);
349    value.setencoding(1); // other encoding
350    // store this key=value pair
351    if (!key.empty()) {
352
353      // if arg occurs multiple times (as is the case with multiple
354      // checkboxes using the same name) we'll create a comma separated
355      // list of all the values (this uses a hack that encodes naturally
356      // occurring commas as %2C - values will therefore need to be decoded
357      // again before use) - it should use an array instead
358      const cgiarginfo *info = argsinfo.getarginfo (key);
359      if (info==NULL) {
360    // If info is NULL, we can't tell if the arg is multiple value or not
361    // Because we need to have dynamically named arguments multivalued, we
362    // will always assume multiplevalue = true
363    // If the arg is not multi valued, then you need to decode the commas.
364    if (args.getarg(key)==NULL) {
365      args.setarg (key, encode_commas(value), cgiarg_t::cgi_arg);
366    }
367    else {
368      text_t newvalue = args[key];
369
370      newvalue += "," + encode_commas(value);
371      newvalue.setencoding(1); // other encoding
372      args.setarg (key, newvalue, cgiarg_t::cgi_arg);
373    }
374      }
375      else {
376    if (info->multiplevalue) {
377
378      text_t newvalue = args[key];
379      if (args.lookupcgiarg(key).source == cgiarg_t::cgi_arg) newvalue += ",";
380      newvalue += encode_commas(value);
381      newvalue.setencoding(1); // other encoding
382      args.setarg (key, newvalue, cgiarg_t::cgi_arg);
383
384    } else {
385      args.setarg (key, value, cgiarg_t::cgi_arg);
386    }
387      }
388    }
389  }
390}
391
392text_t encode_commas (const text_t &intext) {
393
394  text_t outtext;
395
396  text_t::const_iterator here = intext.begin ();
397  text_t::const_iterator end = intext.end ();
398
399  while (here != end) {
400    if (*here == ',') outtext += "%2C";
401    else outtext.push_back (*here);
402    ++here;
403  }
404  return outtext;
405}
406
407text_t decode_commas (const text_t &intext) {
408
409  text_t outtext;
410 
411  text_t::const_iterator here = intext.begin ();
412  text_t::const_iterator end = intext.end ();
413 
414  while (here != end) {
415    if ((here+2<end) && *here == '%' && *(here+1) == '2' &&
416    (*(here+2) == 'C' || *(here+2) == 'c')) {
417      here += 2;
418      outtext.push_back(',');
419     
420    }else outtext.push_back (*here);
421    ++here;
422  }
423  return outtext;
424}
425
426// set utf8 to true if input is in utf-8, otherwise expects input in unicode
427text_t minus_safe (const text_t &intext, bool utf8) {
428
429  text_t outtext;
430
431  text_t::const_iterator here = intext.begin ();
432  text_t::const_iterator end = intext.end ();
433
434  while (here != end) {
435    if (*here == '-') outtext += "Zz-";
436    else outtext.push_back (*here);
437    ++here;
438  }
439  if (utf8) {
440    outtext = cgi_safe_utf8 (outtext);
441  } else {
442    outtext = cgi_safe_unicode (outtext);
443  }
444  return outtext;
445}
446
447// takes utf-8 input
448text_t cgi_safe_utf8 (const text_t &intext) {
449  text_t outtext;
450 
451  text_t::const_iterator here = intext.begin ();
452  text_t::const_iterator end = intext.end ();
453  unsigned short c;
454  text_t ttmp;
455 
456  while (here != end) {
457    c = *here;
458    if (((c >= 'a') && (c <= 'z')) ||
459    ((c >= 'A') && (c <= 'Z')) ||
460    ((c >= '0') && (c <= '9')) ||
461    (c == '%') || (c == '-')) {
462      // alphanumeric character
463      outtext.push_back(c);
464    } else if (c == ' ') {
465      // space
466      outtext.push_back('+');
467    } else if (c > 255) { // not utf-8 character
468      cerr << "WARNING: expected utf-8 char, but got unicode!!\n";
469    } else {
470      // everything else
471      outtext.push_back('%');
472      c2hex(c, ttmp);
473      outtext += ttmp;
474    }
475   
476    ++here;
477  }
478 
479  return outtext;
480}
481// takes unicode input
482text_t cgi_safe_unicode (const text_t &intext) {
483  text_t outtext;
484 
485  text_t::const_iterator here = intext.begin ();
486  text_t::const_iterator end = intext.end ();
487  unsigned short c;
488  text_t ttmp;
489 
490  while (here != end) {
491    c = *here;
492    if (((c >= 'a') && (c <= 'z')) ||
493    ((c >= 'A') && (c <= 'Z')) ||
494    ((c >= '0') && (c <= '9')) ||
495    (c == '%') || (c == '-')) {
496      // alphanumeric character
497      outtext.push_back(c);
498    } else if (c == ' ') {
499      // space
500      outtext.push_back('+');
501    } else if (c > 127) { // unicode character
502      unsigned char buf[3]; // up to 3 bytes
503      buf[0]='\0';buf[1]='\0';buf[2]='\0';
504      output_utf8_char(c,buf, buf+2);
505      outtext.push_back('%');
506      c2hex(buf[0], ttmp);
507      outtext += ttmp;
508      outtext.push_back('%');
509      c2hex(buf[1], ttmp);
510      outtext += ttmp;
511      if (buf[2]) {
512      outtext.push_back('%');
513      c2hex(buf[2], ttmp);
514      outtext += ttmp;
515      }
516    } else {
517      // everything else
518      outtext.push_back('%');
519      c2hex(c, ttmp);
520      outtext += ttmp;
521    }
522   
523    ++here;
524  }
525 
526  return outtext;
527}
528
529
530
531
532static text_t::const_iterator get_next_save_arg (text_t::const_iterator first,
533                       text_t::const_iterator last,
534                       text_t &argname) {
535  first = getdelimitstr (first, last, '-', argname);
536  return first;
537}
538
539
540// check_save_conf_str checks the configuration string for
541// the saved args and makes sure it does not conflict with
542// the information about the arguments. If an error is encountered
543// it will return false and the program should not produce any
544// output.
545bool check_save_conf_str (const text_t &saveconf,
546              const cgiargsinfoclass &argsinfo,
547              ostream &logout) {
548  outconvertclass text_t2ascii;
549
550  text_tset argsset;
551  text_t::const_iterator saveconfhere = saveconf.begin ();
552  text_t::const_iterator saveconfend = saveconf.end ();
553  text_t argname;
554  const cgiarginfo *info;
555
556  // first check to make sure all saved arguments can be saved
557 
558  while (saveconfhere != saveconfend) {
559    saveconfhere = get_next_save_arg (saveconfhere, saveconfend, argname);
560
561    if (!argname.empty()) {
562      // save the argument name for later
563      argsset.insert (argname);
564
565      // check the argument
566      info =  argsinfo.getarginfo (argname);
567      if (info == NULL) {
568    logout << text_t2ascii << "Error: the cgi argument \"" << argname
569           << "\" is used in the configuration string for the\n"
570           << "saved arguments but does not exist as a valid argument.\n\n";
571    return false;
572      }
573      if (info->savedarginfo == cgiarginfo::mustnot) {
574    logout << text_t2ascii << "Error: the cgi argument \"" << argname
575           << "\" is used in the configuration string for the\n"
576           << "saved arguments but has been specified as an argument whose\n"
577           << "state must not be saved.\n\n";
578    return false;
579      }
580    }
581  }
582
583
584  // next check that all saved arguments that should be saved
585  // are saved
586  cgiargsinfoclass::const_iterator argsinfohere = argsinfo.begin ();
587  cgiargsinfoclass::const_iterator argsinfoend = argsinfo.end ();
588
589  while (argsinfohere != argsinfoend) {
590    if (((*argsinfohere).second.savedarginfo == cgiarginfo::must) &&
591    (argsset.find((*argsinfohere).second.shortname) == argsset.end())) {
592      logout << text_t2ascii << "Error: the cgi argument \""
593         << (*argsinfohere).second.shortname << "\" was specified as needing to\n"
594         << "be save but was not listed in the saved arguments.\n\n";
595      return false;
596    }
597
598    ++argsinfohere;
599  }
600 
601  return true; // made it, no clashes
602}
603
604
605// create_save_conf_str will create a configuration string
606// based on the information in argsinfo. This method of configuration
607// is not recomended as small changes can produce large changes in
608// the resulting configuration string (for instance a totally different
609// ordering). Only arguments which "must" be saved are included in
610// the resulting string.
611text_t create_save_conf_str (const cgiargsinfoclass &argsinfo,
612                 ostream &/*logout*/) {
613  cgiargsinfoclass::const_iterator argsinfohere = argsinfo.begin ();
614  cgiargsinfoclass::const_iterator argsinfoend = argsinfo.end ();
615  text_t saveconf;
616  bool first = true;
617
618  while (argsinfohere != argsinfoend) {
619    // save this argument if it must be saved
620    if ((*argsinfohere).second.savedarginfo == cgiarginfo::must) {
621      if (!first) saveconf.push_back ('-');
622      else first = false;
623      saveconf += (*argsinfohere).second.shortname;
624    }
625
626    ++argsinfohere;
627  }
628 
629  return saveconf;
630}
631
632
633// expand_save_args will expand the saved arguments based
634// on saveconf placing the results in args if they are not
635// already defined. If it encounters an error it will return false
636// and output more information to logout.
637bool expand_save_args (const cgiargsinfoclass &argsinfo,
638               const text_t &saveconf,
639               cgiargsclass &args,
640               ostream &logout) {
641  outconvertclass text_t2ascii;
642
643  text_t *arg_e = args.getarg("e");
644  if (arg_e == NULL) return true; // no compressed arguments
645  if (arg_e->empty()) return true; // no compressed arguments
646
647  text_t argname, argvalue;
648  const cgiarginfo *argnameinfo;
649 
650  text_t::const_iterator saveconfhere = saveconf.begin();
651  text_t::const_iterator saveconfend = saveconf.end();
652 
653  text_t::iterator arg_ebegin = arg_e->begin();
654  text_t::iterator arg_eend = arg_e->end();
655  text_t::iterator arg_ehere = arg_ebegin;
656  while (saveconfhere != saveconfend && arg_ehere != arg_eend) {
657    saveconfhere = get_next_save_arg (saveconfhere, saveconfend, argname);
658
659    if (!argname.empty()) {
660      // found another entry
661      argnameinfo = argsinfo.getarginfo (argname);
662
663      if (argnameinfo == NULL) {
664    // no information about the argument could be found
665    // we can't keep going because we don't know whether
666    // this argument is a single or multiple character value
667    logout << text_t2ascii << "Error: the cgi argument \"" << argname
668           << "\" was specified as being a compressed argument\n"
669           << "but no information about it could be found within the "
670           << "cgiargsinfoclass.\n";
671    return false;
672
673      } else {
674
675    // found the argument information
676    if (argnameinfo->multiplechar) {
677      text_t::const_iterator sav = arg_ehere;
678      arg_ehere = getdelimitstr (arg_ehere, arg_eend, '-', argvalue);
679      if (distance(arg_ebegin, arg_ehere) > 2) {
680        // replace any '-' chars escaped with 'Zz'
681        bool first = true;
682        while ((*(arg_ehere-3) == 'Z') && (*(arg_ehere-2) == 'z')) {
683          if (first) argvalue.clear();
684
685          // Hey, here's a wild idea. Why don't we check that there is
686          // another hyphen in the cgiarge before we get a pointer to it and
687          // add one. That way we are far less likely to wander off into
688          // random memory merrily parsing arguments that are then lovingly
689          // spewed all over the HTML page returned at the usage logs.
690          text_t::iterator minus_itr = findchar (arg_ehere, arg_eend, '-');
691          if (minus_itr == arg_eend)
692          {
693        logout << text_t2ascii << "Error: the cgi argument \"" << argname << "\" was specified as being a compressed argument but we have run out of cgiarge to decompress!\n";
694        return false;             
695          }
696          arg_ehere = minus_itr + 1;
697
698          while (sav != (arg_ehere-1)) {
699        if (!((*sav == 'Z') && (*(sav+1) == 'z') && (*(sav+2) == '-')) &&
700            !((*(sav-1) == 'Z') && (*sav == 'z') && (*(sav+1) == '-'))) argvalue.push_back (*sav);
701        ++sav;
702          }
703          first = false;
704        }
705      }
706      argvalue.setencoding(1); // other encoding
707      if (!argvalue.empty()) args.setdefaultarg (argname, argvalue, cgiarg_t::compressed_arg);
708    } else {
709      args.setdefaultcarg (argname,*arg_ehere, cgiarg_t::compressed_arg);
710      ++arg_ehere;
711    }
712      }
713    }
714  }
715
716  return true;
717}
718
719
720// adds the default values for those arguments which have not
721// been specified
722void add_default_args (const cgiargsinfoclass &argsinfo,
723               cgiargsclass &args,
724               ostream &/*logout*/) {
725  cgiargsinfoclass::const_iterator argsinfohere = argsinfo.begin ();
726  cgiargsinfoclass::const_iterator argsinfoend = argsinfo.end ();
727
728  while (argsinfohere != argsinfoend) {
729    if ((*argsinfohere).second.defaultstatus != cgiarginfo::none) {
730      args.setdefaultarg ((*argsinfohere).second.shortname,
731              (*argsinfohere).second.argdefault, cgiarg_t::default_arg);
732    }
733    ++argsinfohere;
734  }
735}
736
737void add_fileupload_args (const cgiargsinfoclass &argsinfo,
738              cgiargsclass &args,
739              fileupload_tmap &fileuploads,
740              ostream &logout) {
741 
742  const cgiarginfo *info = argsinfo.getarginfo("a");
743  fileupload_tmap::const_iterator this_file = fileuploads.begin();
744  fileupload_tmap::const_iterator end_file = fileuploads.end();
745  while (this_file != end_file) {
746    const cgiarginfo *info = argsinfo.getarginfo((*this_file).first);
747    if (info != NULL) {
748
749      if ((*info).fileupload && (file_exists((*this_file).second.tmp_name))) {
750
751    args.setargfile((*this_file).first, (*this_file).second);
752      }
753    }
754    this_file++;
755  }
756}
757
758// compress_save_args will compress the arguments and return
759// them in compressed_args. If an error was encountered
760// compressed_args will be set to to "", an error will be
761// written to logout, and the function will return false.
762bool compress_save_args (const cgiargsinfoclass &argsinfo,
763             const text_t &saveconf,
764             cgiargsclass &args,
765             text_t &compressed_args,
766             outconvertclass &outconvert,
767             ostream &logout) {
768  outconvertclass text_t2ascii;
769
770  compressed_args.clear();
771
772  text_t argname, argvalue;
773  const cgiarginfo *argnameinfo;
774 
775  text_t::const_iterator saveconfhere = saveconf.begin();
776  text_t::const_iterator saveconfend = saveconf.end();
777 
778  while (saveconfhere != saveconfend) {
779    saveconfhere = get_next_save_arg (saveconfhere, saveconfend, argname);
780
781    if (!argname.empty()) {
782      // found another entry
783      argnameinfo = argsinfo.getarginfo (argname);
784
785      if (argnameinfo == NULL) {
786    // no information about the argument could be found
787    // we can't keep going because we don't know whether
788    // this argument is a single or multiple character value
789    logout << text_t2ascii << "Error: the cgi argument \"" << argname
790           << "\" was specified as being a compressed argument\n"
791           << "but no information about it could be found within the "
792           << "cgiargsinfoclass.\n";
793    compressed_args.clear();
794    return false;
795
796      } else {
797    // found the argument information
798    if (argnameinfo->multiplechar) {
799      // multiple character argument -- sort out any '-' chars
800      if (args["w"]=="utf-16be") // browsers don't like \0 in urls...
801        compressed_args += minus_safe (args[argname], false);
802      else
803        compressed_args += minus_safe (outconvert.convert(args[argname]), true);
804     
805      if (saveconfhere != saveconfend) compressed_args.push_back ('-');
806
807    } else {
808      // single character argument
809      if (args[argname].size() == 0) {
810        logout << text_t2ascii << "Error: the cgi argument \"" << argname
811           << "\" was specified as being a compressed argument which\n"
812           << "should have a one character value but it was empty.\n\n";
813        compressed_args.clear ();
814        return false;
815
816      } else if (args[argname].size() > 1) {
817        logout << text_t2ascii << "Error: the cgi argument \"" << argname
818           << "\" was specified as being a compressed argument which\n"
819           << "should have a one character value but it had multiple characters.\n\n";
820        compressed_args.clear ();
821        return false;
822      }
823
824      // everything is ok
825      compressed_args += args[argname];
826    }
827      }
828    }
829  }
830
831  return true;
832}
833
834
835// args_tounicode converts any arguments which are not in unicode
836// to unicode using inconvert
837void args_tounicode (cgiargsclass &args, inconvertclass &inconvert) {
838  cgiargsclass::iterator here = args.begin();
839  cgiargsclass::iterator end = args.end();
840
841  while (here != end) {
842    if ((*here).second.value.getencoding() > 0) {
843      // Call reset() before converting each argument, to prevent problems when converting the last
844      //   argument left the converter in a bad state
845      inconvert.reset();
846      (*here).second.value = inconvert.convert((*here).second.value);
847    }
848   
849    ++here;
850  }
851}
852
853// fcgienv will be loaded with environment name-value pairs
854// if using fastcgi (had to do this as getenv doesn't work
855// with our implementation of fastcgi). if fcgienv is empty
856// we'll simply use getenv
857text_t gsdl_getenv (const text_t &name, text_tmap &fcgienv) {
858  if (fcgienv.empty()) {
859    char *n = name.getcstr();
860    char *v = getenv(n);
861    delete []n;
862    if (v != NULL) return v;
863    return g_EmptyText;
864
865  } else return fcgienv[name];
866}
Note: See TracBrowser for help on using the browser.