codecvt workarounds -

some implementations don't provide function bodies for the codecvt template
clase - I need to check spec on this
some implementations have an opaque definition of std::mbstate_t which means
it's actually exceeding difficult to use :)

SAX/ArabicaConfig.S

@@ -23,6 +23,8 @@ define pasty(ARABICA, _NO_WCHAR_T) 1
 #if (_MSC_VER < 1300)
 define ARABICA_VS6_WORKAROUND
 #endif
+
+define ARABICA_NO_CODECVT_SPECIALISATIONS
 #endif
 
 #ifdef _WIN32

SAX/Makefile

@@ -32,18 +32,22 @@ LDFLAGS += $(DYNAMIC_LIBS)
 # SOURCE FILES
 SRCS =  saxlib.cpp \
 	helpers/InputSourceResolver.cpp \
+	../Utils/rot13codecvt.cpp \
 	../Utils/utf8ucs2codecvt.cpp \
 	../Utils/impl/ucs2_utf8.cpp \
+	../Utils/utf16beucs2codecvt.cpp \
+	../Utils/utf16leucs2codecvt.cpp \
+	../Utils/impl/ucs2_utf16.cpp \
+	../Utils/iso88591utf8codecvt.cpp \
+	../Utils/utf8iso88591codecvt.cpp \
+	../Utils/impl/iso88591_utf8.cpp \
+	../Utils/utf16utf8codecvt.cpp \
+	../Utils/base64codecvt.cpp \
 	../XML/XMLCharacterClasses.cpp
-#	../Utils/base64codecvt.cpp \
+
-#	../Utils/iso88591utf8codecvt.cpp \
+
-#	../Utils/rot13codecvt.cpp \
+
-#	../Utils/utf16beucs2codecvt.cpp \
+
-#	../Utils/utf16leucs2codecvt.cpp \
-#	../Utils/utf16utf8codecvt.cpp \
-#	../Utils/utf8iso88591codecvt.cpp \
-#	../Utils/impl/iso88591_utf8.cpp \
-#	../Utils/impl/ucs2_utf16.cpp \
 
 
 HDRS = $(patsubst %.c,%.h,$(patsubst %.cpp,%.h,$(SRCS)))

Utils/base64codecvt.cpp

@@ -9,7 +9,7 @@
 static const std::string base64_charset("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/");
 static const int NO_MORE = 256;
 
-std::codecvt_base::result base64codecvt::do_out(state_t& state,
+std::codecvt_base::result base64codecvt::do_out(std::mbstate_t& state,
                         const char* from,
                         const char* from_end,
                         const char*& from_next,
@@ -67,7 +67,7 @@ std::codecvt_base::result base64codecvt::do_out(state_t& state,
   return ((getState() == 0) && (getCurrentOutChar() == NO_MORE) && (getPreviousChar() == 0)) ? std::codecvt_base::ok : std::codecvt_base::partial;
 } // do_out
 
-std::codecvt_base::result base64codecvt::do_in(state_t& state,
+std::codecvt_base::result base64codecvt::do_in(std::mbstate_t& state,
                        const char* from,
                        const char* from_end,
                        const char*& from_next,
@@ -109,7 +109,7 @@ std::codecvt_base::result base64codecvt::do_in(state_t& state,
   return (from_next == from_end) ? std::codecvt_base::ok : std::codecvt_base::partial; 
 } // do_in
 
-std::codecvt_base::result base64codecvt::do_unshift(state_t& state,
+std::codecvt_base::result base64codecvt::do_unshift(std::mbstate_t& state,
                             char* to,
                             char* to_limit,
                             char*& to_next) const
@@ -126,7 +126,7 @@ std::codecvt_base::result base64codecvt::do_unshift(state_t& state,
   return (to_next == to_limit) ? codecvt_base::ok : codecvt_base::partial;
 } // do_unshift
 
-int base64codecvt::do_length(const state_t&,
+int base64codecvt::do_length(const std::mbstate_t&,
                         const char* from,
                         const char* end,
                         size_t max) const
@@ -158,7 +158,7 @@ int base64codecvt::getState() const
 
 void base64codecvt::nextState() const
 {
-  state_t s = getState();
+  int s = getState();
   s = s + 1;
   if(s == 4)
     s = 0;

Utils/base64codecvt.h

@@ -8,15 +8,18 @@
 
 #include <locale>
 
-class base64codecvt : public std::codecvt<char, char, int>
+#ifdef ARABICA_VS6_WORKAROUND
-{
+namespace std {
-public:
+  typedef ::mbstate_t mbstate_t;
-  typedef int state_t;
+}
+#endif
 
+class base64codecvt : public std::codecvt<char, char, std::mbstate_t>
+{
 protected:
   virtual ~base64codecvt() { }
 
-  virtual result do_out(state_t& state,
+  virtual result do_out(std::mbstate_t& state,
                         const char* from,
                         const char* from_end,
                         const char*& from_next,
@@ -24,7 +27,7 @@ protected:
                         char* to_limit,
                         char*& to_next) const;
 
-  virtual result do_in(state_t& state,
+  virtual result do_in(std::mbstate_t& state,
                        const char* from,
                        const char* from_end,
                        const char*& from_next,
@@ -32,7 +35,7 @@ protected:
                        char* to_limit,
                        char*& to_next) const;
 
-  virtual result do_unshift(state_t&  state,
+  virtual result do_unshift(std::mbstate_t&  state,
                        char* to,
                        char* to_limit,
                        char*& to_next) const;
@@ -41,7 +44,7 @@ protected:
 
   virtual bool do_always_noconv() const throw() { return false; }
 
-  virtual int do_length(const state_t&,
+  virtual int do_length(const std::mbstate_t&,
                         const char* from,
                         const char* end,
                         size_t max) const;
@@ -53,10 +56,10 @@ private:
   // the state counter, and in some case we need to "pad" the input
   // strings.  I use these helper functions to mungle them
   // together and keep the details neater (or try to anyway)
-  mutable state_t* state_;
+  mutable int* state_;
   mutable const char** from_next_;
   mutable const char** from_end_;
-  void grabState(state_t& state) const { state_ = &state; }
+  void grabState(std::mbstate_t& state) const { state_ = reinterpret_cast<int*>(&state); }
   void grabFromNext(const char*& from_next) const { from_next_ = &from_next; }
   void grabFromEnd(const char*& from_end) const { from_end_ = &from_end; }
 

Utils/utf16utf8codecvt.h

@@ -14,7 +14,10 @@
 //---------------------------------------------------------------------------
 #include <locale>
 
-#if(_MSC_VER < 1300)
+#ifndef ARABICA_NO_CODECVT_SPECIALISATIONS
+#include <Utils/impl/codecvt_specialisations.h>
+#endif
+#ifdef ARABICA_VS6_WORKAROUND
 namespace std {
   typedef ::mbstate_t mbstate_t;
 }
@@ -59,3 +62,4 @@ protected:
 }; // class utf16utf8codecvt
 
 #endif
+