Changeset 1856

Timestamp:

2001-01-23T10:58:19+13:00 (23 years ago)

Author:

sjboddie

Message:

Added Encoding and Language options to main.cfg configuration file so
it's now hopefully a little easier to add new languages and encodings
to the interface. Each interface language also now has a default encoding
so that changing languages from the preferences page causes the encoding
to change to a value reasonable for the selected language.

Location:

trunk/gsdl

Files:

• etc/main.cfg (modified) (3 diffs)
• macros/english.dm (modified) (1 diff)
• macros/pref.dm (modified) (5 diffs)
• src/recpt/librarymain.cpp (modified) (1 diff)
• src/recpt/pageaction.cpp (modified) (3 diffs)
• src/recpt/receptionist.cpp (modified) (11 diffs)
• src/recpt/receptionist.h (modified) (3 diffs)

trunk/gsdl/etc/main.cfg

@@ -28 +28 @@
 # Set logcgiargs to true to keep a log of usage information in
 # $GSDLHOME/etc/usage.txt.
-logcgiargs      false
+logcgiargs      true
 
 # Set usecookies to true to use cookies to identify users (cookie 
 # information will be written to the usage log if logcgiargs is
 # true).
-usecookies      false
+usecookies      true
 
 # LogDateFormat sets the format that timestamps will be stored in the usage
@@ -75 +75 @@
         port.dm home.dm collect.dm
 
+
+# Define the interface languages and encodings supported by this receptionist
+
+# An "Encoding" line defines an encoding to be used by the receptionist
+# options are:
+# shortname -- Identifier for the given encoding. The shortname option is
+#              mandatory and must be unique for each "Encoding" line.
+# longname  -- The display name of the given encoding. If longname isn't set
+#              it will default to using shortname instead.
+
+# type      -- The type of encoding. Note that for most encodings this
+#              value is the directory name under which the map file for
+#              this encoding resides in the Greenstone unicode/MAPPINGS
+#              directory (e.g. 'WINDOWS', 'ISO_8859' etc.). It may also
+#              take the values 'GB' and 'UTF8'.
+# mapfile   -- The name of the map file for use when converting between
+#              utf8 and the given encoding. The mapfile option is mandatory
+#              for all encoding types with the exception of GB and UTF8.
+# label     -- The standard label to which you must set the value of
+#              "charset" within http headers or html meta tags to get a web
+#              browser to use the given encoding. The label option is
+#              mandatory.
+Encoding shortname=utf8 "longname=Unicode (UTF-8)" type=UTF8 label=UTF-8
+Encoding shortname=iso1 "longname=Western (ISO-8859-1)" type=ISO_8859 mapfile=1.TXT label=ISO-8859-1
+Encoding shortname=w1251 "longname=Cyrillic (Windows-1251)" type=WINDOWS mapfile=1251.TXT label=windows-1251
+Encoding shortname=w1256 "longname=Arabic (Windows-1256)" type=WINDOWS mapfile=1256.TXT label=windows-1256
+Encoding shortname=gb "longname=Simplified Chinese (GBK)" type=GB label=GBK
+Encoding shortname=koi8r "longname=Cyrillic (KOI8-R)" type=CYRILLIC mapfile=koi8_r.txt label=koi8-r
+
+
+# A "Language" line defines an interface language to be used by the
+# interface. Note that it is possible to display only a subset of the
+# specified languages on the preferences page for a given collection by
+# using the "PreferenceLanguages" format option in your collect.cfg
+# configuration file.
+# options are:
+# shortname        -- ISO 639 two letter language symbol. The shortname
+                      option is mandatory.
+# longname         -- The display name for the given language. If longname
+                      isn't set it will default to using shortname instead.
+# default_encoding -- The encoding to use by default when using the given
+#                     interface language. This should be set to the
+#                     "shortname" of a valid "Encoding" line
+Language shortname=en longname=English default_encoding=iso1
+Language shortname=fr longname=French default_encoding=iso1
+Language shortname=zh longname=Chinese default_encoding=gb
+Language shortname=de longname=German default_encoding=iso1
+Language shortname=es longname=Spanish default_encoding=iso1
+Language shortname=mi longname=Maori default_encoding=iso1
+Language shortname=ar longname=Arabic default_encoding=w1256
+Language shortname=pt longname=Portuguese default_encoding=iso1
+Language shortname=nl longname=Dutch default_encoding=iso1
+
+
 # Define any additional page parameters to be used by the above macro files
 # (the current default page parameters are c (collection) and l (language)
@@ -113 +167 @@
 # set a default value for the built-in "p" cgi argument
 cgiarg          shortname=p argdefault=home
+

trunk/gsdl/macros/english.dm

@@ -627 +627 @@
 _textdigitlib_ {the digital library}
 _textweb_ {the web}
-_textlangeng_ {English}
-_textlanggerman_ {German}
-_textlangfrench_ {French}
-_textlangspanish_ {Spanish}
-_textlangmaori_ {M_amn_ori}
-_textlangchinese_ {Chinese}
-_textlangarabic_ {Arabic}
-_textlangportuguese_ {Portuguese}
-_textlangdutch_ {Dutch}
 _textgraphical_ {Graphical}
 _texttextual_ {Textual}

trunk/gsdl/macros/pref.dm

@@ -200 +200 @@
 function updatel () \{
   value = document.PrefForm.l.options[document.PrefForm.l.selectedIndex].value;
-  location.href = location.href + "&l=" + value + getargs();
+  location.href = location.href + "&nl=1&l=" + value + getargs();
 \}
 
@@ -210 +210 @@
 function updatew () \{
   value = document.PrefForm.w.options[document.PrefForm.w.selectedIndex].value;
-  location.href = location.href + "&nw=" + value + getargs();
+  location.href = location.href + "&nl=0&nw=" + value + getargs();
 \}
 }
@@ -330 +330 @@
 }
 
+# set from within pageaction
 _languageoption_ {}
-
-_enlanguageoption_ {<option value="en"_If_("_cgiargl_",, selected)_If_("_cgiargl_" eq "en", selected)>_textlangeng_}
-_delanguageoption_ {<option value="de"_If_("_cgiargl_" eq "de", selected)>_textlanggerman_}
-_frlanguageoption_ {<option value="fr"_If_("_cgiargl_" eq "fr", selected)>_textlangfrench_}
-_eslanguageoption_ {<option value="es"_If_("_cgiargl_" eq "es", selected)>_textlangspanish_}
-_milanguageoption_ {<option value="mi"_If_("_cgiargl_" eq "mi", selected)>_textlangmaori_}
-_zhlanguageoption_ {<option value="zh"_If_("_cgiargl_" eq "zh", selected)>_textlangchinese_}
-_arlanguageoption_ {<option value="ar"_If_("_cgiargl_" eq "ar", selected)>_textlangarabic_}
-_ptlanguageoption_ {<option value="pt"_If_("_cgiargl_" eq "pt", selected)>_textlangportuguese_}
-_nllanguageoption_ {<option value="nl"_If_("_cgiargl_" eq "nl", selected)>_textlangdutch_}
-
-_encodingoption_ {
-_textencoding_
-<select name="w" onChange="updatew();">
-  <option value="w" _If_("_cgiargw_" eq "w",selected)>Western (ISO-8859-1)
-  <option value="u" _If_("_cgiargw_" eq "u",selected)>UTF-8
-  <option value="g" _If_("_cgiargw_" eq "g",selected)>GBK
-  <option value="a" _If_("_cgiargw_" eq "a",selected)>Arabic (windows 1256)
-  <option value="c" _If_("_cgiargw_" eq "c",selected)>Cyrillic (windows 1251)
-</select>
-}
+_encodingoption_ {}
 
 _formatoption_ {
@@ -365 +346 @@
 _navigationbar_
 </center>
+
 <blockquote>
 
@@ -409 +391 @@
 
 <p>
-_languageoption_
-
-<p>
-_encodingoption_
+_If_(_languageoption_,_textlanguage__languageoption_)
+
+<p>
+_If_(_encodingoption_,_textencoding__encodingoption_)
 
 <p>

trunk/gsdl/src/recpt/librarymain.cpp

@@ -111 +111 @@
 #endif
 
+  // Read main.cfg to get all the "Encoding" lines and add corresponding converters.
+  // It might be possible to move this to somewhere like receptionist::configure, depending
+  // on whether we need the converters before then (I don't think we do).
+  text_tarray cfgline;
+  text_t maincfg = filename_cat (gsdlhome, "etc", "main.cfg");
+  if (file_exists (maincfg)) {
+    char *maincfgc = maincfg.getcstr();
+#ifdef GSDL_USE_IOS_H
+    ifstream confin (maincfgc, ios::in | ios::nocreate);
+#else
+    ifstream confin (maincfgc, ios::in);
+#endif
+    delete maincfgc;
+  
+    if (confin) {
+      text_t subkey, subvalue, shortname;
+      text_tset saved;
+      text_tmap tmp;
+      text_t::const_iterator cfglinesub_here;
+      text_tarray::const_iterator cfgline_here;
+      text_tarray::const_iterator cfgline_end;
+      while (read_cfg_line(confin, cfgline) >= 0) {
+    if (cfgline.size () >= 4 && cfgline[0] == "Encoding") {
+      tmp.erase(tmp.begin(), tmp.end());
+      cfgline_here = cfgline.begin();
+      cfgline_end = cfgline.end();
+      while (cfgline_here != cfgline_end) {
+        cfglinesub_here = getdelimitstr((*cfgline_here).begin(),
+                        (*cfgline_here).end(), '=', subkey);
+        if (subkey == "shortname") {
+          shortname = substr (cfglinesub_here, (*cfgline_here).end());
+        } else {
+          tmp[subkey] = substr (cfglinesub_here, (*cfgline_here).end());
+        }
+        cfgline_here++;
+      }
+      // we just use the saved set to prevent multiple encodings being added
+      // that use the same shortname (i.e. any encodings after the first with
+      // the same name will be ignored).
+      if (!shortname.empty() && saved.find(shortname) == saved.end()) {
+        saved.insert(shortname);
+
+        if (tmp["type"] == "UTF8") {
+          utf8inconvertclass *utf8inconvert = new utf8inconvertclass();
+          utf8outconvertclass *utf8outconvert = new utf8outconvertclass();
+          recpt.add_converter (shortname, utf8inconvert, utf8outconvert);
+
+        } else if (tmp["type"] == "GB") {
+          mapinconvertclass *gbinconvert = new mapinconvertclass();
+          gbinconvert->setmapfile (gsdlhome, "gbku", 0x25a1);
+          mapoutconvertclass *gboutconvert = new mapoutconvertclass();
+          gboutconvert->setmapfile (gsdlhome, "ugbk", 0xa1f5);
+          recpt.add_converter (shortname, gbinconvert, gboutconvert);
+
+        } else {
+          if (!tmp["mapfile"].empty()) {
+
+        if (tmp["type"] == "ISO_8859" && tmp["mapfile"] == "1.TXT") {
+          // iso-8859-1 is a special case as it'll always be
+          // supported by the standard converter class and
+          // therefore doesn't need to use its mapping file
+          inconvertclass *inconvert = new inconvertclass();
+          rzwsoutconvertclass *outconvert = new rzwsoutconvertclass();
+          recpt.add_converter (shortname, inconvert, outconvert);  
+
+        } else {
+          text_t mapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS", tmp["type"], tmp["mapfile"]);
+          if (file_exists (mapfile)) {
+            simplemapinconvertclass *inconvert = new simplemapinconvertclass();
+            inconvert->setmapfile (mapfile);
+            simplemapoutconvertclass *outconvert = new simplemapoutconvertclass();
+            outconvert->setmapfile (mapfile);
+            recpt.add_converter (shortname, inconvert, outconvert);  
+          }
+        }
+          }
+        }
+      }
+    }
+      }
+      confin.close ();
+    }
+  }
+
+
   // add other converters
-  utf8inconvertclass utf8inconvert;
-  utf8outconvertclass utf8outconvert;
-  recpt.add_converter ("u", &utf8inconvert, &utf8outconvert);
-
-  mapinconvertclass gbinconvert;
-  gbinconvert.setmapfile (gsdlhome, "gbku", 0x25a1);
-  mapoutconvertclass gboutconvert;
-  gboutconvert.setmapfile (gsdlhome, "ugbk", 0xa1f5);
-  recpt.add_converter ("g", &gbinconvert, &gboutconvert);
+  //  utf8inconvertclass utf8inconvert;
+  //  utf8outconvertclass utf8outconvert;
+  //  recpt.add_converter ("u", &utf8inconvert, &utf8outconvert);
+
+  //  mapinconvertclass gbinconvert;
+  //  gbinconvert.setmapfile (gsdlhome, "gbku", 0x25a1);
+  //  mapoutconvertclass gboutconvert;
+  //  gboutconvert.setmapfile (gsdlhome, "ugbk", 0xa1f5);
+  //  recpt.add_converter ("g", &gbinconvert, &gboutconvert);
 
   // arabic
-  text_t mapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS");
-  mapfile = filename_cat (mapfile, "WINDOWS", "1256.TXT");
-  simplemapinconvertclass arinconvert;
-  arinconvert.setmapfile (mapfile);
-  simplemapoutconvertclass aroutconvert;
-  aroutconvert.setmapfile (mapfile);
-  recpt.add_converter ("a", &arinconvert, &aroutconvert);  
+  //  text_t mapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS");
+  //  mapfile = filename_cat (mapfile, "WINDOWS", "1256.TXT");
+  //  simplemapinconvertclass arinconvert;
+  //  arinconvert.setmapfile (mapfile);
+  //  simplemapoutconvertclass aroutconvert;
+  //  aroutconvert.setmapfile (mapfile);
+  //  recpt.add_converter ("a", &arinconvert, &aroutconvert);  
 
   // cyrillic
-  mapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS");
-  mapfile = filename_cat (mapfile, "WINDOWS", "1251.TXT");
-  simplemapinconvertclass cyinconvert;
-  cyinconvert.setmapfile (mapfile);
-  simplemapoutconvertclass cyoutconvert;
-  cyoutconvert.setmapfile (mapfile);
-  recpt.add_converter ("c", &cyinconvert, &cyoutconvert);  
+  //  mapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS");
+  //  mapfile = filename_cat (mapfile, "WINDOWS", "1251.TXT");
+  //  simplemapinconvertclass cyinconvert;
+  //  cyinconvert.setmapfile (mapfile);
+  //  simplemapoutconvertclass cyoutconvert;
+  //  cyoutconvert.setmapfile (mapfile);
+  //  recpt.add_converter ("c", &cyinconvert, &cyoutconvert);  
   
   // hindi

trunk/gsdl/src/recpt/pageaction.cpp

@@ -187 +187 @@
   // _languageoption_      interface languages to select from (dependant on PreferenceLanguages)
 
-  // _encodingoption_      will be overridden to "" if chinese language interface isn't specified
+  // _encodingoption_      encodings to select from
 
   // if page is "about"
@@ -283 +283 @@
 
     // _languageoption_
-    // this is kind of a hack (another one!!). note that there're potential 
-    // problems here as no checking is done anywhere to make sure language
-    // is set to a valid value. this does allow us to set which languages 
-    // should go in the preference selection box (from collect.cfg) however
-    // and that's sufficient for current requirements
-    
-    // currently supported languages (_languageoption_ will default to all
-    // these languages unless PreferenceLanguages is set
-    text_tarray languages;
-    languages.push_back ("en");
-    languages.push_back ("fr");
-    languages.push_back ("de");
-    languages.push_back ("es");
-    languages.push_back ("mi");
-    languages.push_back ("zh");
-    languages.push_back ("ar");
-    languages.push_back ("pt");
-    languages.push_back ("nl");
-    text_tarray::const_iterator this_lang = languages.begin();
-    text_tarray::const_iterator end_lang = languages.end();
-
-    text_t languageoption = "_textlanguage_\n<select name=\"l\" onChange=\"updatel();\">\n";
+    // Create the "interface language" selection box for the preferences page.
+    // You can use something like "format PreferenceLanguages en|fr|zn" from within
+    // a collect.cfg file to use only a subset of the available languages for
+    // any given collection. This facility is kind of ugly though and should be
+    // replaced by something better when the configuration files are tidied up (as
+    // should all the other "format Preference..." options).
+    
+    text_t &arg_l = args["l"];
+    const recptconf &configinfo = recpt->get_configinfo();
+    // put languages in another map to sort them by longname
+    text_tmap languages;
+    languageinfo_tmap::const_iterator thislang = configinfo.languages.begin();
+    languageinfo_tmap::const_iterator endlang = configinfo.languages.end();
+    while (thislang != endlang) {
+      languages[(*thislang).second.longname] = (*thislang).first;
+      thislang++;
+    }
+    text_tmap::iterator tlang = languages.begin();
+    text_tmap::iterator elang = languages.end();
+
+    text_t languageoption;
     it = cinfo->format.find ("PreferenceLanguages");
     if ((it != cinfo->format.end()) && (!(*it).second.empty())) {
@@ -310 +309 @@
       splitchar ((*it).second.begin(), (*it).second.end(), '|', pref_langs);
       if (pref_langs.size() > 1) {
-    while (this_lang != end_lang) {
-      if (pref_langs.find(*this_lang) != pref_langs.end())
-        languageoption += "_" + *this_lang + "languageoption_\n";
-      this_lang ++;
+    while (tlang != elang) {
+      if (pref_langs.find((*tlang).second) != pref_langs.end()) {
+        languageoption += "<option value=\"" + (*tlang).second + "\"";
+        if ((*tlang).second == arg_l) languageoption += " selected";
+        languageoption += ">" + (*tlang).first + "\n";
+      }
+      tlang ++;
     }
-    languageoption += "</select>\n";
-    disp.setmacro ("languageoption", "preferences", languageoption);
-      }
-    
-      // _encodingoption_
-
-      if (pref_langs.find("zh") == pref_langs.end())
-    disp.setmacro ("encodingoption", "preferences", "");
+      }
 
     } else {
-      while (this_lang != end_lang) {
-    languageoption += "_" + *this_lang + "languageoption_\n";
-    this_lang ++;
-      }
+      while (tlang != elang) {
+    languageoption += "<option value=\"" + (*tlang).second + "\"";
+    if ((*tlang).second == arg_l) languageoption += " selected";
+    languageoption += ">" + (*tlang).first + "\n";
+    tlang ++;
+      }
+    }
+    if (!languageoption.empty()) {
+      languageoption = "<select name=\"l\" onChange=\"updatel();\">\n" + languageoption;
       languageoption += "</select>\n";
       disp.setmacro ("languageoption", "preferences", languageoption);
+    }
+
+    // _encodingoption_
+    // create the "encoding" selection box for the preferences page
+    text_t &arg_w = args["w"];
+    // put encodings in another map to sort them by longname
+    text_tmap encodings;
+    encodinginfo_tmap::const_iterator thisenc = configinfo.encodings.begin();
+    encodinginfo_tmap::const_iterator endenc = configinfo.encodings.end();
+    while (thisenc != endenc) {
+      encodings[(*thisenc).second.longname] = (*thisenc).first;
+      thisenc++;
+    }
+    text_tmap::iterator tenc = encodings.begin();
+    text_tmap::iterator eenc = encodings.end();
+
+    text_t encodingoption;
+    while (tenc != eenc) {
+      encodingoption += "<option value=\"" + (*tenc).second + "\"";
+      if ((*tenc).second == arg_w) encodingoption += " selected";
+      encodingoption += ">" + (*tenc).first + "\n";
+      tenc ++;
+    }
+
+    if (!encodingoption.empty()) {
+      encodingoption = "<select name=\"w\" onChange=\"updatew();\">\n" + encodingoption;
+      encodingoption += "</select>\n";
+      disp.setmacro ("encodingoption", "preferences", encodingoption);
     }
     

trunk/gsdl/src/recpt/receptionist.cpp

@@ -65 +65 @@
   EmailUserEvents = false;
 
+  languages.erase(languages.begin(), languages.end());
+  encodings.erase(encodings.begin(), encodings.end());
+
   // these default page parameters can always be overriden
   // in the configuration file
@@ -84 +87 @@
   info_loaded = false;
   info.clear();
+}
+
+void languageinfo_t::clear () {
+  longname.clear();
+  defaultencoding.clear();
+}
+
+void encodinginfo_t::clear () {
+  longname.clear();
+  label.clear();
 }
 
@@ -112 +125 @@
   ainfo.longname = "encoding";
   ainfo.multiplechar = true;
-  ainfo.defaultstatus = cgiarginfo::weak;
-  ainfo.argdefault = "w";
+  ainfo.defaultstatus = cgiarginfo::none;
+  ainfo.argdefault = "";
   ainfo.savedarginfo = cgiarginfo::must;
   argsinfo.addarginfo (NULL, ainfo);
@@ -142 +155 @@
   ainfo.savedarginfo = cgiarginfo::must;
   argsinfo.addarginfo (NULL, ainfo);
-
+  
+  ainfo.shortname = "nl";
+  ainfo.longname = "new language";
+  ainfo.multiplechar = false;
+  ainfo.defaultstatus = cgiarginfo::none;
+  ainfo.argdefault = "0";
+  ainfo.savedarginfo = cgiarginfo::mustnot;
+  argsinfo.addarginfo (NULL, ainfo);
+  
   // the GSDL_UID (cookie)
   ainfo.shortname = "z";
@@ -293 +314 @@
       cfgline_here++;
     }
+      }
+
+    } else if (key == "Encoding") {
+      text_t subkey, subvalue;
+      text_t shortname, longname, label;
+      text_t::const_iterator cfglinesub_here;
+      text_tarray::const_iterator cfgline_here = cfgline.begin();
+      text_tarray::const_iterator cfgline_end = cfgline.end();
+      while (cfgline_here != cfgline_end) {
+    cfglinesub_here = getdelimitstr((*cfgline_here).begin(),
+                    (*cfgline_here).end(), '=', subkey);
+    if (subkey == "shortname") {
+      shortname = substr (cfglinesub_here, (*cfgline_here).end());
+    } else if (subkey == "longname") {
+      longname = substr (cfglinesub_here, (*cfgline_here).end());
+    } else if (subkey == "label") {
+      label = substr (cfglinesub_here, (*cfgline_here).end());
+    }
+    cfgline_here++;
+      }
+      if (!shortname.empty() && !label.empty()) {
+    encodinginfo_t enc;
+    if (longname.empty()) enc.longname = shortname;
+    else enc.longname = longname;
+    enc.label = label;
+    configinfo.encodings[shortname] = enc;
+      }
+
+    } else if (key == "Language") {
+      text_t subkey, subvalue;
+      text_t shortname, longname, defaultencoding;
+      text_t::const_iterator cfglinesub_here;
+      text_tarray::const_iterator cfgline_here = cfgline.begin();
+      text_tarray::const_iterator cfgline_end = cfgline.end();
+      while (cfgline_here != cfgline_end) {
+    cfglinesub_here = getdelimitstr((*cfgline_here).begin(),
+                    (*cfgline_here).end(), '=', subkey);
+    if (subkey == "shortname") {
+      shortname = substr (cfglinesub_here, (*cfgline_here).end());
+    } else if (subkey == "longname") {
+      longname = substr (cfglinesub_here, (*cfgline_here).end());
+    } else if (subkey == "default_encoding") {
+      defaultencoding = substr (cfglinesub_here, (*cfgline_here).end());
+    }
+    cfgline_here++;
+      }
+      if (!shortname.empty()) {
+    languageinfo_t lang;
+    if (longname.empty()) lang.longname = shortname;
+    else lang.longname = longname;
+    lang.defaultencoding = defaultencoding;
+    configinfo.languages[shortname] = lang;
       }
     }
@@ -400 +473 @@
   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->argdefault != "w") {
-      if ((ainfo != NULL) && (converters.get_outconverter(ainfo->argdefault) == NULL)) {
-    ainfo->defaultstatus = cgiarginfo::good;
-    ainfo->argdefault = defaultconvertname;
-      }
-    }
   }
 
@@ -470 +529 @@
 }
 
+// get the default encoding for the given language - if it fails for any
+// reason return ""
+text_t receptionist::get_default_encoding (const text_t &language) {
+  
+  // make sure language is valid
+  if (configinfo.languages.find(language) == configinfo.languages.end()) return "";
+
+  text_t default_encoding = configinfo.languages[language].defaultencoding;
+
+  // make sure the encoding is valid
+  if (configinfo.encodings.find(default_encoding) == configinfo.encodings.end()) return "";
+
+  return default_encoding;
+}
 
 // parse_cgi_args parses cgi arguments into an argument class.
@@ -476 +549 @@
 bool receptionist::parse_cgi_args (const text_t &argstr, cgiargsclass &args, 
                    ostream &logout, text_tmap &fcgienv) {
-  outconvertclass text_t2ascii;
 
   // get an initial list of cgi arguments
@@ -491 +563 @@
   if (configinfo.usecookies) get_cookie(args["z"], fcgienv);
   
+  // if we're changing languages, set the encoding to the default for the new language
+  if (args["nl"] == "1") {
+    args["nw"] = get_default_encoding(args["l"]);
+  }
+
   // get the input encoding
+  // if encoding isn't set, set it to the default for the current language
+  if ((args.getarg("w") == NULL) || args["w"].empty()) {
+    args["w"] = get_default_encoding(args["l"]);
+  }
+
   text_t &arg_w = args["w"];
+
   inconvertclass defaultinconvert;
   inconvertclass *inconvert = converters.get_inconverter (arg_w);
@@ -520 +603 @@
   } else {
     // the action was not found!!
+    outconvertclass text_t2ascii;
     logout << text_t2ascii << "Error: the action \"" << args["a"] 
        << "\" could not be found.\n";
@@ -744 +828 @@
   // add the encoding information
   if (response == content) {
-    if (args["w"] == "u") {
-      response_data += "; charset=UTF-8";
-    } else if (args["w"] == "g") {
-      response_data += "; charset=GBK";
-    } else if (args["w"] == "a") {
-      response_data += "; charset=windows-1256";
-    } else if (args["w"] == "c") {
-      response_data += "; charset=windows-1251";
+    if (configinfo.encodings.find(args["w"]) != configinfo.encodings.end()) {
+      response_data += "; charset=" + configinfo.encodings[args["w"]].label;
     } else {
-      response_data += "; charset=ISO-8859-1";      
+      // default to latin 1
+      response_data += "; charset=ISO-8859-1";
     }
 

trunk/gsdl/src/recpt/receptionist.h

@@ -60 +60 @@
 };
 
+struct languageinfo_t {
+  void clear();
+  languageinfo_t () {clear();}
+
+  text_t longname;
+  text_t defaultencoding;
+};
+
+struct encodinginfo_t {
+  void clear();
+  encodinginfo_t () {clear();}
+  
+  text_t longname;
+  text_t label;
+};
+
 typedef map<text_t, collectioninfo_t, lttext_t> colinfo_tmap;
+typedef map<text_t, languageinfo_t, lttext_t> languageinfo_tmap;
+typedef map<text_t, encodinginfo_t, lttext_t> encodinginfo_tmap;
 
 enum events_t {Disabled, CollectorEvents, AllEvents};
@@ -91 +109 @@
   text_t macroprecedence;
   
+  languageinfo_tmap languages;
+  encodinginfo_tmap encodings;
+
   void clear ();
   recptconf () {clear();}
@@ -118 +139 @@
   bool get_cookie (text_t &cookie, text_tmap &fcgienv);
   bool get_cookie (text_tmap &fcgienv);
+
+  // get the default encoding for the given language - if it fails for any
+  // reason, return ""
+  text_t get_default_encoding (const text_t &language);
 
   virtual void define_general_macros (cgiargsclass &args, outconvertclass &outconvert,