[ruby] use WCHAR_MAX to determine the encoding of std::wstring.

swig · Feb 19, 2017 · 5722d81 · 5722d81
1 parent bfe7204
commit 5722d81
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 25 deletions.
diff --git a/Lib/ruby/rubywstrings.swg b/Lib/ruby/rubywstrings.swg
@@ -1,31 +1,18 @@
 /* -----------------------------------------------------------------------------
  * rubywstrings.swg
  *
- * Currently, Ruby does not support Unicode or WChar properly, so these
- * are still treated as char arrays for now.
- * There are other libraries available that add support to this in
- * ruby including WString, FXString, etc.
- * ----------------------------------------------------------------------------- */
-
-/* ------------------------------------------------------------
  *  utility methods for wchar_t strings 
  * ------------------------------------------------------------ */
 
 %fragment("SWIG_AsWCharPtrAndSize","header",fragment="<wchar.h>",fragment="SWIG_pwchar_descriptor",fragment="SWIG_AsCharPtrAndSize") {
 SWIGINTERN int
 SWIG_AsWCharPtrAndSize(VALUE obj, wchar_t **cptr, size_t *psize, int *alloc)
 {
-  VALUE tmp;
+  VALUE tmp = rb_str_conv_enc(obj, rb_enc_get(obj),
+                                   rb_to_encoding(rb_str_new_cstr( SWIG_RUBY_WSTRING_ENCODING )) );
   *alloc = SWIG_NEWOBJ;
-  if(sizeof(wchar_t) == 4) {
-    tmp = rb_str_conv_enc(obj, rb_enc_get(obj), rb_to_encoding(rb_str_new_cstr("UTF-32")) );
-  } else if (sizeof(wchar_t) == 2) {
-    tmp = rb_str_conv_enc(obj, rb_enc_get(obj), rb_to_encoding(rb_str_new_cstr("UTF-16")) );
-  } else {
-    rb_raise(rb_eRuntimeError, "unsupported wchar_t size");
-  }
-  return SWIG_AsCharPtrAndSize( tmp, (char**)cptr, psize, alloc);
 
+  return SWIG_AsCharPtrAndSize(tmp, (char**)cptr, psize, alloc);
 }
 }
 
@@ -35,18 +22,13 @@ SWIG_FromWCharPtrAndSize(const wchar_t * carray, size_t size)
 {
   VALUE ret = SWIG_FromCharPtrAndSize( (const char*)carray, size);
   rb_encoding* enc = rb_default_internal_encoding();
-  if(sizeof(wchar_t) == 4) {
-    rb_enc_associate(ret, rb_to_encoding(rb_str_new_cstr("UTF-32")));
-  } else if (sizeof(wchar_t) == 2) {
-    rb_enc_associate(ret, rb_to_encoding(rb_str_new_cstr("UTF-16")));
-  } else {
-    rb_raise(rb_eRuntimeError, "unsupported wchar_t size");
-  }
+
+  rb_enc_associate(ret, rb_to_encoding(rb_str_new_cstr( SWIG_RUBY_WSTRING_ENCODING )) );
 
   if( !enc ) {
-    enc = rb_to_encoding(rb_str_new_cstr("UTF-8"));
+    enc = rb_to_encoding(rb_str_new_cstr( SWIG_RUBY_INTERNAL_ENCODING ));
   }
-  return rb_str_conv_enc(ret, rb_enc_get(ret), enc );
+  return rb_str_conv_enc(ret, rb_enc_get(ret), enc);
 }
 }
 

diff --git a/Lib/ruby/std_wstring.i b/Lib/ruby/std_wstring.i
@@ -7,6 +7,31 @@ extern "C" {
 #include "ruby/encoding.h"
 #endif
 
+/**
+ *  The internal encoding of std::wstring is defined based on
+ *  the size of wchar_t. If it is not appropriate for your library,
+ *  SWIG_RUBY_WSTRING_ENCODING must be given when compiling.
+ */
+#ifndef SWIG_RUBY_WSTRING_ENCODING
+
+#if WCHAR_MAX == 0x7fff || WCHAR_MAX == 0xffff
+#define SWIG_RUBY_WSTRING_ENCODING "UTF-16"
+#elif WCHAR_MAX == 0x7fffffff || WCHAR_MAX == 0xffffffff
+#define SWIG_RUBY_WSTRING_ENCODING "UTF-32"
+#else
+#error unsupported wchar_t size. SWIG_RUBY_WSTRING_ENCODING must be given.
+#endif
+
+#endif
+
+/**
+ *  If Encoding.default_internal is nil, this encoding will be used
+ *  when coverting from std::wstring to String object in Ruby.
+ */
+#ifndef SWIG_RUBY_INTERNAL_ENCODING
+#define SWIG_RUBY_INTERNAL_ENCODING "UTF-8"
+#endif
+
 #ifdef __cplusplus
 }
 #endif