Add support for ISO-8859-1 as application encoding

New runtime configuration parameter, XSDE_ENCODING. New option,
--char-encoding. New test, tests/cxx/hybrid/iso8859-1.

6 files changed, 64 insertions, 2 deletions

diff --git a/xsde/cxx/parser/cli.hxx b/xsde/cxx/parser/cli.hxx
index 62a1891..a50c3ff 100644
--- a/xsde/cxx/parser/cli.hxx
+++ b/xsde/cxx/parser/cli.hxx
@@ -24,6 +24,7 @@ namespace CXX
       typedef Char const Key[];
 
       extern Key type_map;
+      extern Key char_encoding;
       extern Key no_stl;
       extern Key no_iostream;
       extern Key no_exceptions;
@@ -86,6 +87,7 @@ namespace CXX
 
       typedef Cult::CLI::Options<
         type_map,                 Cult::Containers::Vector<NarrowString>,
+        char_encoding,            NarrowString,
         no_stl,                   Boolean,
         no_iostream,              Boolean,
         no_exceptions,            Boolean,
diff --git a/xsde/cxx/parser/elements.cxx b/xsde/cxx/parser/elements.cxx
index 4b624ff..3c62b5b 100644
--- a/xsde/cxx/parser/elements.cxx
+++ b/xsde/cxx/parser/elements.cxx
@@ -20,6 +20,7 @@ namespace CXX
                         root,
                         "p:name",
                         "char",
+                        ops.value<CLI::char_encoding> (),
                         ops.value<CLI::include_with_brackets> (),
                         ops.value<CLI::include_prefix> (),
                         "", // export symbol
diff --git a/xsde/cxx/parser/generator.cxx b/xsde/cxx/parser/generator.cxx
index e59180c..1936286 100644
--- a/xsde/cxx/parser/generator.cxx
+++ b/xsde/cxx/parser/generator.cxx
@@ -113,6 +113,7 @@ namespace CXX
     namespace CLI
     {
       extern Key type_map                 = "type-map";
+      extern Key char_encoding            = "char-encoding";
       extern Key no_stl                   = "no-stl";
       extern Key no_iostream              = "no-iostream";
       extern Key no_exceptions            = "no-exceptions";
@@ -188,6 +189,12 @@ namespace CXX
       << " order of appearance and the first match is used."
       << endl;
 
+    e << "--char-encoding <enc>" << endl
+      << " Specify the character encoding that should be\n"
+      << " used for the extracted text data. Valid values\n"
+      << " are 'utf8' (default) and 'iso8859-1'."
+      << endl;
+
     e << "--no-stl" << endl
       << " Generate code that does not use STL."
       << endl;
@@ -514,6 +521,8 @@ namespace CXX
   {
     CLI::OptionsSpec spec;
 
+    spec.option<CLI::char_encoding> ().default_value ("utf8");
+
     spec.option<CLI::skel_file_suffix> ().default_value ("-pskel");
     spec.option<CLI::skel_type_suffix> ().default_value ("_pskel");
     spec.option<CLI::impl_file_suffix> ().default_value ("-pimpl");
@@ -1207,6 +1216,25 @@ namespace CXX
           hxx << "#include <xsde/cxx/config.hxx>" << endl
               << endl;
 
+          if (ops.value<CLI::char_encoding> () == "iso8859-1")
+          {
+            hxx << "#ifndef XSDE_ENCODING_ISO8859_1" << endl
+                << "#error the generated code uses the ISO-8859-1 encoding" <<
+              "while the XSD/e runtime does not (reconfigure the runtime " <<
+              "or change the --char-encoding value)" << endl
+                << "#endif" << endl
+                << endl;
+          }
+          else
+          {
+            hxx << "#ifndef XSDE_ENCODING_UTF8" << endl
+                << "#error the generated code uses the UTF-8 encoding" <<
+              "while the XSD/e runtime does not (reconfigure the runtime " <<
+              "or change the --char-encoding value)" << endl
+                << "#endif" << endl
+                << endl;
+          }
+
           if (ops.value<CLI::no_stl> ())
           {
             hxx << "#ifdef XSDE_STL" << endl
@@ -1585,6 +1613,14 @@ namespace CXX
 
       return sloc;
     }
+    catch (UnrepresentableCharacter const& e)
+    {
+      wcerr << "error: character at position " << e.position () << " "
+            << "in string '" << e.string () << "' is unrepresentable in "
+            << "the target encoding" << endl;
+
+      throw Failed ();
+    }
     catch (NoNamespaceMapping const& e)
     {
       wcerr << e.file () << ":" << e.line () << ":" << e.column ()
diff --git a/xsde/cxx/parser/name-processor.cxx b/xsde/cxx/parser/name-processor.cxx
index 658b70a..a168c28 100644
--- a/xsde/cxx/parser/name-processor.cxx
+++ b/xsde/cxx/parser/name-processor.cxx
@@ -34,6 +34,7 @@ namespace CXX
                             root,
                             "p:name",
                             "char",
+                            ops.value<CLI::char_encoding> (),
                             ops.value<CLI::include_with_brackets> (),
                             ops.value<CLI::include_prefix> (),
                             "", // export symbol
diff --git a/xsde/cxx/parser/parser-header.cxx b/xsde/cxx/parser/parser-header.cxx
index f3dd45c..6ef9578 100644
--- a/xsde/cxx/parser/parser-header.cxx
+++ b/xsde/cxx/parser/parser-header.cxx
@@ -1634,6 +1634,14 @@ namespace CXX
              << "typedef xsde::cxx::parser::context parser_context;"
              << endl;
 
+          if (char_encoding == L"iso8859-1")
+          {
+            os << "// ISO-8859-1 transcoder." << endl
+               << "//" << endl
+               << "using xsde::cxx::iso8859_1;"
+               << endl;
+          }
+
           post (ns);
         }
       };
@@ -1675,8 +1683,12 @@ namespace CXX
       else
       {
         ctx.os << "#include <xsde/config.h>" << endl
-               << "#include <xsde/cxx/ro-string.hxx>" << endl
-               << endl;
+               << "#include <xsde/cxx/ro-string.hxx>" << endl;
+
+        if (ctx.char_encoding == L"iso8859-1")
+          ctx.os << "#include <xsde/cxx/iso8859-1.hxx>" << endl;
+
+        ctx.os << endl;
 
         // Data types.
         //
diff --git a/xsde/cxx/parser/validator.cxx b/xsde/cxx/parser/validator.cxx
index 96170de..33fe2e2 100644
--- a/xsde/cxx/parser/validator.cxx
+++ b/xsde/cxx/parser/validator.cxx
@@ -601,6 +601,16 @@ namespace CXX
 
       //
       //
+      NarrowString enc (options.value<CLI::char_encoding> ());
+
+      if (enc != "utf8" && enc != "iso8859-1")
+      {
+        wcerr << "error: unknown encoding '" << enc.c_str () << "'" << endl;
+        return false;
+      }
+
+      //
+      //
       if (options.value<CLI::generate_noop_impl> () &&
           options.value<CLI::generate_print_impl> ())
       {