Another facet that is frequently applied to XSD string-based types is enumeration. This results in the generation of a C enum typedef that enumerates all of the identifiers that can be used in the type.

The general mapping is as follows:

XSD type:

   <xsd:simpleType name="TypeName">
      <restriction base="xsd:string">
         <xsd:enumeration value="enum1"/>
         <xsd:enumeration value="enum2"/>
         ...
         <xsd:enumeration value="enumN"/>
      </xsd:restriction>
   </xsd:simpleType>

Generated C code:

   typedef enum {
      TypeName_enum1 = 0,
      TypeName_enum2 = 1,
      ...
      TypeName_enumN = N - 1,
   } TypeName;

   typedef TypeName TypeName_ENUM;  //deprecated

Generated C++ code:

   class TypeName : public OSRTBaseType {
     public:
      enum Enum {
         enum1 = 0,
         enum2 = 1,
         ...
         enumN = N - 1,
      } ;

      Enum value;
      ...
   } ;

Note that for C, TypeNameis used on the enumerated identifiers as a namespace mechanism in order to prevent name clashes if two or more enumerated types use the same identifier names. In this case, the type name may only be a partial fragment of the full name to keep the names shorter. This is not a problem in C++ as the class provides a namespace for the enumeration constants defined within (for example, enum1would be referenced as TypeName::enum1outside the class).

In XSD, the rules for naming enumerated identifiers are more liberal than in the C/C++ programming language. For example, enumerated identifiers can start with numbers or punctuation marks. The logic to transform the XSD enumeration names to C/C++ form makes use of the following rules to ensure the names are valid C/C++ names: