source: trunk/gsdl/src/recpt/receptionist.cpp@ 430

/**********************************************************************
 *
 * receptionist.cpp -- a web interface for the gsdl
 * Copyright (C) 1999  The New Zealand Digital Library Project
 *
 * PUT COPYRIGHT NOTICE HERE
 *
 * $Id: receptionist.cpp 418 1999-07-30 02:13:10Z sjboddie $
 *
 *********************************************************************/

/*
   $Log$
   Revision 1.20  1999/07/30 02:13:09  sjboddie
   -added collectinfo argument to some functions
   -made some function prototypes virtual

   Revision 1.19  1999/07/15 06:02:05  rjmcnab
   Moved the setting of argsinfo into the constructor. Added the configuration
   command argdefault (as used by the actions). Added code to output the
   correct charset based on the page encoding so that the user does not need
   to specify the encoding used for a particular page.

   Revision 1.18  1999/07/11 01:05:20  rjmcnab
   Stored origin of cgiarg with argument.

   Revision 1.17  1999/07/10 22:18:26  rjmcnab
   Added calls to define_external_cgiargs.

   Revision 1.16  1999/06/27 21:49:03  sjboddie
   fixed a couple of version conflicts - tidied up some small things

   Revision 1.15  1999/06/26 01:14:32  rjmcnab
   Made a couple of changes to handle different encodings.

   Revision 1.14  1999/06/09 00:08:36  sjboddie
   query string macro (_cgiargq_) is now made html safe before being set

   Revision 1.13  1999/06/08 04:29:31  sjboddie
   added argsinfo to the call to check_cgiargs to make it easy to set
   args to their default if they're found to be screwed up

   Revision 1.12  1999/04/30 01:59:42  sjboddie
   lots of stuff - getting documentaction working (documentaction replaces
   old browseaction)

   Revision 1.11  1999/03/25 03:06:43  sjboddie

   altered receptionist slightly so it now passes *collectproto to
   define_internal_macros and define_external_macros - need it
   for browseaction

   Revision 1.10  1999/03/05 03:53:54  sjboddie

   fixed some bugs

   Revision 1.9  1999/02/28 20:00:16  rjmcnab


   Fixed a few things.

   Revision 1.8  1999/02/25 21:58:59  rjmcnab

   Merged sources.

   Revision 1.7  1999/02/21 22:33:55  rjmcnab

   Lots of stuff :-)

   Revision 1.6  1999/02/11 01:24:05  rjmcnab

   Fixed a few compiler warnings.

   Revision 1.5  1999/02/08 01:28:02  rjmcnab

   Got the receptionist producing something using the statusaction.

   Revision 1.4  1999/02/05 10:42:46  rjmcnab

   Continued working on receptionist

   Revision 1.3  1999/02/04 10:00:56  rjmcnab

   Developed the idea of an "action" and having them define the cgi arguments
   which they need and how those cgi arguments function.

   Revision 1.2  1999/02/04 01:17:27  rjmcnab

   Got it outputing something.


 */


#include "receptionist.h"
#include "fileutil.h"
#include "cgiutils.h"
#include "htmlutils.h"
#include "OIDtools.h"
#include <assert.h>
#include <time.h>



receptionist::receptionist () {
  // create a list of cgi arguments
  // this must be done before the configuration

  cgiarginfo ainfo;

  ainfo.shortname = "e";
  ainfo.longname = "compressed arguments";
  ainfo.multiplechar = true;
  ainfo.defaultstatus = cgiarginfo::good;
  ainfo.argdefault = "";
  ainfo.savedarginfo = cgiarginfo::mustnot;
  argsinfo.addarginfo (NULL, ainfo);

  ainfo.shortname = "a";
  ainfo.longname = "action";
  ainfo.multiplechar = true;
  ainfo.defaultstatus = cgiarginfo::none;
  ainfo.argdefault = "";
  ainfo.savedarginfo = cgiarginfo::must;
  argsinfo.addarginfo (NULL, ainfo);

  // w=western
  ainfo.shortname = "w";
  ainfo.longname = "encoding";
  ainfo.multiplechar = true;
  ainfo.defaultstatus = cgiarginfo::weak;
  ainfo.argdefault = "w";
  ainfo.savedarginfo = cgiarginfo::must;
  argsinfo.addarginfo (NULL, ainfo);
  
  ainfo.shortname = "nw";
  ainfo.longname = "new encoding";
  ainfo.multiplechar = true;
  ainfo.defaultstatus = cgiarginfo::none;
  ainfo.argdefault = "";
  ainfo.savedarginfo = cgiarginfo::mustnot;
  argsinfo.addarginfo (NULL, ainfo);
  
  ainfo.shortname = "c";
  ainfo.longname = "collection";
  ainfo.multiplechar = true;
  ainfo.defaultstatus = cgiarginfo::none;
  ainfo.argdefault = "";
  ainfo.savedarginfo = cgiarginfo::must;
  argsinfo.addarginfo (NULL, ainfo);
  
  // 0=text+graphics, 1=text
  ainfo.shortname = "v";
  ainfo.longname = "version";
  ainfo.multiplechar = false;
  ainfo.defaultstatus = cgiarginfo::weak;
  ainfo.argdefault = "0";
  ainfo.savedarginfo = cgiarginfo::can;
  argsinfo.addarginfo (NULL, ainfo);
  
  // 0=normal, 1=big
  ainfo.shortname = "f";
  ainfo.longname = "query box size";
  ainfo.multiplechar = false;
  ainfo.defaultstatus = cgiarginfo::weak;
  ainfo.argdefault = "0";
  ainfo.savedarginfo = cgiarginfo::can;
  argsinfo.addarginfo (NULL, ainfo);
  
  // the interface language name should use the ISO 639
  // standard
  ainfo.shortname = "l";
  ainfo.longname = "interface language";
  ainfo.multiplechar = true;
  ainfo.defaultstatus = cgiarginfo::weak;
  ainfo.argdefault = "en";
  ainfo.savedarginfo = cgiarginfo::must;
  argsinfo.addarginfo (NULL, ainfo);
}


// configure should be called for each line in the
// configuration files to configure the receptionist and everything
// it contains. The configuration should take place after everything
// has been added but before the initialisation.
void receptionist::configure (const text_t &key, const text_tarray &cfgline) {
  // configure the receptionist
  if (cfgline.size() >= 1) {
    cgiarginfo *info = NULL;
    if (key == "gsdlhome") configinfo.gsdlhome = cfgline[0];
    else if (key == "collection") {
      configinfo.collection = cfgline[0];
      // also need to set the default arg to this collection
      if ((info = argsinfo.getarginfo("c")) != NULL) {
    info->defaultstatus = cgiarginfo::good;
    info->argdefault = cfgline[0];
      }
      
    } else if (key == "collectdir") configinfo.collectdir = cfgline[0];
    else if (key == "httpprefix") configinfo.httpprefix = cfgline[0];
    else if (key == "httpimg") configinfo.httpimg = cfgline[0];
    else if (key == "gwcgi") configinfo.gwcgi = cfgline[0];
    else if (key == "macrofiles") configinfo.macrofiles = cfgline;
    else if (key == "saveconf") configinfo.saveconf = cfgline[0];

    else if ((key == "argdefault") && (cfgline.size() == 2) && 
         ((info = argsinfo.getarginfo(cfgline[0])) != NULL)) {
      if (info->defaultstatus <= cgiarginfo::config) {
    info->defaultstatus = cgiarginfo::config;
    info->argdefault = cfgline[1];
      }
    }
  }

  // configure the actions
  actionptrmap::iterator actionhere = actions.begin ();
  actionptrmap::iterator actionend = actions.end ();

  while (actionhere != actionend) {
    assert ((*actionhere).second.a != NULL);
    if ((*actionhere).second.a != NULL) 
      (*actionhere).second.a->configure(key, cfgline);

    actionhere++;
  }

  // configure the protocols
  recptprotolistclass::iterator protohere = protocols.begin ();
  recptprotolistclass::iterator protoend = protocols.end ();

  while (protohere != protoend) {
    assert ((*protohere).p != NULL);
    if ((*protohere).p != NULL) 
      (*protohere).p->configure(key, cfgline);
    
    protohere++;
  }
}

void receptionist::configure (const text_t &key, const text_t &value) {
  text_tarray cfgline;
  cfgline.push_back (value);
  configure(key, cfgline);
}


// init should be called after all the actions, protocols, and
// converters have been added to the receptionist and after everything
// has been configured but before any pages are created.
// It returns true on success and false on failure. If false is
// returned getpage should not be called (without producing
// meaningless output), instead an error page should be
// produced by the calling code.
bool receptionist::init (ostream &logout) {
  // first configure collectdir
  text_t thecollectdir = configinfo.gsdlhome;
  if (!configinfo.collection.empty()) {
    // collection specific mode
    if (!configinfo.collectdir.empty()) {
      // has already been configured
      thecollectdir = configinfo.collectdir;
    } else {
      // decide where collectdir is by searching for collect.cfg
      // look in $GSDLHOME/collect/collection-name/etc/collect.cfg and
      // then $GSDLHOME/etc/collect.cfg
      thecollectdir = filename_cat (configinfo.gsdlhome, "collect");
      thecollectdir = filename_cat (thecollectdir, configinfo.collection);
      text_t filename = filename_cat (thecollectdir, "etc");
      filename = filename_cat (filename, "collect.cfg");

      if (!file_exists(filename)) thecollectdir = configinfo.gsdlhome;
    }
  }
  configure("collectdir", thecollectdir);

  // read in the macro files
  if (!read_macrofiles (logout)) return false;

  // there must be at least one action defined
  if (actions.empty()) {
    logout << "Error: no actions have been added to the receptionist\n";
    return false;
  }

  // add the cgi arguments from the actions
  actionptrmap::iterator here = actions.begin ();
  actionptrmap::iterator end = actions.end ();
  while (here != end) {
    assert ((*here).second.a != NULL);
    if ((*here).second.a != NULL) {
      if (!argsinfo.addarginfo (&logout, (*here).second.a->getargsinfo()))
    return false;
    }
    here++;
  }

  // create a saveconf string if there isn't one already
  if (configinfo.saveconf.empty())
    configinfo.saveconf = create_save_conf_str (argsinfo, logout);

  // check the saveconf string
  if (!check_save_conf_str (configinfo.saveconf, argsinfo, logout))
    return false;

  // set a random seed
  srand (time(NULL));

  // make the output converters remove all the zero-width spaces
  convertinfoclass::iterator converthere = converters.begin ();
  convertinfoclass::iterator convertend = converters.end ();
  text_t defaultconvertname;
  while (converthere != convertend) {
    assert ((*converthere).second.outconverter != NULL);
    if ((*converthere).second.outconverter != NULL) {
      (*converthere).second.outconverter->set_rzws(1);
      if (defaultconvertname.empty())
    defaultconvertname = (*converthere).second.name;
    }
    converthere++;
  }

  // set default converter if no good one has been defined
  if (!defaultconvertname.empty()) {
    cgiarginfo *ainfo = argsinfo.getarginfo ("w");
    if ((ainfo != NULL) && (ainfo->defaultstatus < cgiarginfo::config)) {
      ainfo->defaultstatus = cgiarginfo::good;
      ainfo->argdefault = defaultconvertname;
    }
  }
  
  // init the actions
  actionptrmap::iterator actionhere = actions.begin ();
  actionptrmap::iterator actionend = actions.end ();
  while (actionhere != actionend) {
    if (((*actionhere).second.a == NULL) || 
    !(*actionhere).second.a->init(logout)) return false;
    actionhere++;
  }

  // init the protocols
  recptprotolistclass::iterator protohere = protocols.begin ();
  recptprotolistclass::iterator protoend = protocols.end ();
  while (protohere != protoend) {
    if (((*protohere).p == NULL) ||
    !(*protohere).p->init(logout)) return false;
    protohere++;
  }

  return true;
}


// parse_cgi_args parses cgi arguments into an argument class.
// This function should be called for each page request. It returns false
// if there was a major problem with the cgi arguments.
bool receptionist::parse_cgi_args (const text_t &argstr, cgiargsclass &args, 
                   ostream &logout) {
  outconvertclass text_t2ascii;

  // get an initial list of cgi arguments
  args.clear();
  split_cgi_args (argstr, args);

  // expand the compressed argument (if there was one)
  if (!expand_save_args (argsinfo, configinfo.saveconf, args, logout)) return false;

  // add the defaults
  add_default_args (argsinfo, args, logout);

  
  // get the input encoding
  text_t &arg_w = args["w"];
  inconvertclass defaultinconvert;
  inconvertclass *inconvert = converters.get_inconverter (arg_w);
  if (inconvert == NULL) inconvert = &defaultinconvert;

  // see if the next page will have a different encoding
  if (args.getarg("nw") != NULL) arg_w = args["nw"];

  // convert arguments which aren't in unicode to unicode
  args_tounicode (args, *inconvert);


  // decide on the output conversion class (needed for checking the external
  // cgi arguments)
  rzwsoutconvertclass defaultoutconverter;
  rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
  if (outconverter == NULL) outconverter = &defaultoutconverter;
  outconverter->reset();


  // check the main cgi arguments
  if (!check_mainargs (args, logout)) return false;

  // check the arguments for the action
  action *a = actions.getaction (args["a"]);
  if (a != NULL) {
    if (!a->check_cgiargs (argsinfo, args, logout)) return false;
  } else {
    // the action was not found!!
    logout << text_t2ascii << "Error: the action \"" << args["a"] 
       << "\" could not be found.\n";
    return false;
  }

  // check external cgi arguments for each action
  actionptrmap::iterator actionhere = actions.begin ();
  actionptrmap::iterator actionend = actions.end ();
  while (actionhere != actionend) {
    assert ((*actionhere).second.a != NULL);
    if ((*actionhere).second.a != NULL) {
      if (!(*actionhere).second.a->check_external_cgiargs (argsinfo, args, *outconverter,
                               configinfo.saveconf, logout))
    return false;
    }
    actionhere++;
  }

  // the action might have changed but we will assume that
  // the cgiargs were checked properly when the change was made

  return true;
}


// produce_cgi_page will call get_cgihead_info and 
// produce_content in the appropriate way to output a cgi header and
// the page content (if needed). If a page could not be created it
// will return false
bool receptionist::produce_cgi_page (cgiargsclass &args, ostream &contentout, 
                     ostream &logout) {
  outconvertclass text_t2ascii;

  response_t response;
  text_t response_data;

  // produce cgi header
  get_cgihead_info (args, response, response_data, logout);
  if (response == location) {
    // I've forgotten how to do this :-/
    return true;
  } else if (response == content) {
    // content response
    contentout << text_t2ascii << "Content-type: " << response_data << "\n\n";
  } else {
    // unknown response
    logout << "Error: get_cgihead_info returned an unknown response type.\n";
    return false;
  }

  // produce cgi page
  if (!produce_content (args, contentout, logout)) return false;

  // flush contentout
  contentout << flush;
  return true;
}


// get_cgihead_info determines the cgi header information for
// a set of cgi arguments. If response contains location then
// response_data contains the redirect address. If reponse
// contains content then reponse_data contains the content-type.
// Note that images can now be produced by the receptionist.
void receptionist::get_cgihead_info (cgiargsclass &args, response_t &response,
                     text_t &response_data, ostream &logout) {
  outconvertclass text_t2ascii;

  // get the action
  action *a = actions.getaction (args["a"]);
  if (a != NULL) {
    a->get_cgihead_info (args, response, response_data, logout);

  } else {
    // the action was not found!!
    logout << text_t2ascii << "Error receptionist::get_cgihead_info: the action \"" 
       << args["a"] << "\" could not be found.\n";
    response = content;
    response_data = "text/html";
  }

  // add the encoding information
  if (response == content) {
    if (args["w"] == "u") {
      response_data += "; charset=UTF-8";
    } else {
      response_data += "; charset=ISO-8859-1";      
    }
  }
}


// produce the page content
bool receptionist::produce_content (cgiargsclass &args, ostream &contentout,
                    ostream &logout) {
  // decide on the output conversion class
  text_t &arg_w = args["w"];
  rzwsoutconvertclass defaultoutconverter;
  rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
  if (outconverter == NULL) outconverter = &defaultoutconverter;
  outconverter->reset();

  // decide on the protocol used for communicating with
  // the collection server
  recptproto *collectproto = NULL;
  if (!args["c"].empty()) {
    collectproto = protocols.getrecptproto (args["c"], logout);
  }

  // produce the page using the desired action
  action *a = actions.getaction (args["a"]);
  if (a != NULL) {
    if (a->uses_display(args)) prepare_page (a, args, collectproto, (*outconverter), logout);
    if (!a->do_action (args, collectinfo, collectproto, disp, 
               (*outconverter), contentout, logout))
      return false;

  } else {
    // the action was not found!!
    outconvertclass text_t2ascii;

    logout << text_t2ascii << "Error receptionist::produce_content: the action \"" 
       << args["a"] << "\" could not be found.\n";
    
    contentout << (*outconverter) 
           << "<html>\n"
           << "<head>\n"
           << "<title>Error</title>\n"
           << "</head>\n"
           << "<body>\n"
           << "<h2>Oops!</h2>\n"
           << "Undefined Page. The action \"" 
           << args["a"] << "\" could not be found.\n"
           << "</body>\n"
           << "</html>\n";
  }

  return true;
}


// returns the compressed argument ("e") corresponding to the argument
// list. This can be used to save preferences between sessions.
text_t receptionist::get_compressed_arg (cgiargsclass &args, ostream &logout) {
  // decide on the output conversion class
  text_t &arg_w = args["w"];
  rzwsoutconvertclass defaultoutconverter;
  rzwsoutconvertclass *outconverter = converters.get_outconverter (arg_w);
  if (outconverter == NULL) outconverter = &defaultoutconverter;
  outconverter->reset();

  text_t compressed_args;
  if (compress_save_args (argsinfo, configinfo.saveconf, args, 
              compressed_args, *outconverter, logout))
    return compressed_args;

  return "";
}


// will read in all the macro files. If one is not found an
// error message will be written to logout and the method will
// return false.
bool receptionist::read_macrofiles (ostream &logout) {
  outconvertclass text_t2ascii;

  // redirect the error output to logout
  disp.setlogout (&logout);

  // load up the default macro files, the collection directory
  // is searched first for the file (if this is being used in
  // collection specific mode) and then the main directory
  text_t colmacrodir = filename_cat (configinfo.collectdir, "macros");
  text_t gsdlmacrodir = filename_cat (configinfo.gsdlhome, "macros");
  text_tarray::iterator arrhere = configinfo.macrofiles.begin();
  text_tarray::iterator arrend = configinfo.macrofiles.end();
  text_t filename;
  while (arrhere != arrend) {
    // filename is used as a flag to indicate whether
    // the macro file has been found
    filename.clear();

    // try in the collection directory if this is being
    // run in collection specific mode
    if (!configinfo.collection.empty()) {
      filename = filename_cat (colmacrodir, *arrhere);
      if (!file_exists (filename)) filename.clear ();
    }

    // if we haven't found the macro file yet try in
    // the main macro directory
    if (filename.empty()) {
      filename = filename_cat (gsdlmacrodir, *arrhere);
      if (!file_exists (filename)) filename.clear ();
    }

    // see if we found the file or not
    if (filename.empty()) {
      logout << text_t2ascii 
         << "Error: the macro file \"" << *arrhere << "\" could not be found.\n";
      if (configinfo.collection.empty()) {
    logout << text_t2ascii
           << "It should be in " << gsdlmacrodir << ".\n\n";
      } else {
    logout << text_t2ascii
           << "It should be in either " << colmacrodir << " or in " 
           << gsdlmacrodir << ".\n\n";
      }
      return false;

    } else { // found the file
      disp.loaddefaultmacros(filename);
    }

    arrhere++;
  }

  // success
  return true;
}


// check_mainargs will check all the main arguments. If a major
// error is found it will return false and no cgi page should
// be created using the arguments.
bool receptionist::check_mainargs (cgiargsclass &args, ostream &/*logout*/) {
  // if this receptionist is running in collection dependant mode
  // then it should always set the collection argument to the
  // collection
  if (!configinfo.collection.empty()) args["c"] = configinfo.collection;

  // argument "v" can only be 0 or 1. Use the default value
  // if it is out of range
  int arg_v = args.getintarg ("v");
  if (arg_v != 0 && arg_v != 1) {
    cgiarginfo *vinfo = argsinfo.getarginfo ("v");
    if (vinfo != NULL) args["v"] = vinfo->argdefault;
  }

  // argument "f" can only be 0 or 1. Use the default value
  // if it is out of range
  int arg_f = args.getintarg ("f");
  if (arg_f != 0 && arg_f != 1) {
    cgiarginfo *finfo = argsinfo.getarginfo ("f");
    if (finfo != NULL) args["f"] = finfo->argdefault;
  }

  return true;
}

// prepare_page sets up page parameters, sets display macros,
// gets format information and opens the page ready for output 
void receptionist::prepare_page (action *a, cgiargsclass &args, recptproto *collectproto,
                 outconvertclass &outconvert, ostream &logout) {
  // set up page parameters
  text_t pageparams;
  text_t &collection = args["c"];

  bool first = true;
  if (!collection.empty()) {
    pageparams +=  "collection=" + collection; first = false;}
  if (args.getintarg("u") == 1)
    if (first) {pageparams += "style=htmlonly"; first = false;}
    else pageparams += ",style=htmlonly";
  if (args.getintarg("v") == 1)
    if (first) {pageparams += "version=text"; first = false;}
    else pageparams += ",version=text";
  if (args.getintarg("f") == 1) 
    if (first) {pageparams += ",queryversion=big"; first = false;}
    else pageparams += ",queryversion=big";
  if (args["l"] != "en") 
    if (first) pageparams += ",language=" + args["l"];
    else pageparams += ",language=" + args["l"];
  
  // get collection information 
  collectinfo.clear();
  if (!collection.empty()) {
    comerror_t err;
    collectproto->get_collectinfo (collection, collectinfo, err, logout);
    if (err != noError) collectinfo.clear();
  }

  // open the page
  disp.openpage(pageparams, MACROPRECEDENCE);


  // define general macros
  define_general_macros (args, outconvert, logout);

  
  // define external macros for each action
  actionptrmap::iterator actionhere = actions.begin ();
  actionptrmap::iterator actionend = actions.end ();

  while (actionhere != actionend) {
    assert ((*actionhere).second.a != NULL);
    if ((*actionhere).second.a != NULL) 
      (*actionhere).second.a->define_external_macros (collectinfo, disp, args, collectproto, logout);
    actionhere++;
  }

  // define internal macros for the current action
  a->define_internal_macros (collectinfo, disp, args, collectproto, logout);
}

void receptionist::define_general_macros (cgiargsclass &args, outconvertclass &/*outconvert*/,
                      ostream &logout) {
  disp.setmacro ("gwcgi", "Global", configinfo.gwcgi);
  disp.setmacro ("httpimg", "Global", configinfo.httpimg);
  disp.setmacro ("httpprefix", "Global", configinfo.httpprefix);
  disp.setmacro("compressedoptions", "Global", get_compressed_arg(args, logout));

  // set _cgiargX_ macros for each cgi argument
  cgiargsclass::const_iterator argshere = args.begin();
  cgiargsclass::const_iterator argsend = args.end();
  while (argshere != argsend) {
    if ((*argshere).first == "q")
      // need to escape special characters from query string
      disp.setmacro ("cgiargq", "Global", html_safe((*argshere).second.value));
    else 
      disp.setmacro ("cgiarg" + (*argshere).first, "Global", (*argshere).second.value);
    argshere ++;
  }
}