respect Encoding.default_internal in BSON decoder

ext/cbson/cbson.c

@@ -88,7 +88,15 @@ static VALUE DigestMD5;
 
 #if HAVE_RUBY_ENCODING_H
 #include "ruby/encoding.h"
-#define STR_NEW(p,n) rb_enc_str_new((p), (n), rb_utf8_encoding())
+#define STR_NEW(p,n)                                                    \
+    ({                                                                  \
+        VALUE _str = rb_enc_str_new((p), (n), rb_utf8_encoding());      \
+        rb_encoding* internal_encoding = rb_default_internal_encoding(); \
+        if (internal_encoding) {                                        \
+            _str = rb_str_export_to_enc(_str, internal_encoding);       \
+        }                                                               \
+        _str;                                                           \
+    })
 /* MUST call TO_UTF8 before calling write_utf8. */
 #define TO_UTF8(string) rb_str_export_to_enc((string), rb_utf8_encoding())
 static void write_utf8(buffer_t buffer, VALUE string, char check_null) {

lib/bson/bson_ruby.rb

@@ -325,6 +325,16 @@ module BSON
       Regexp.new(str, options)
     end
 
+    def encoded_str(str)
+      if RUBY_VERSION >= '1.9'
+        str.force_encoding("utf-8")
+        if Encoding.default_internal
+          str.encode!(Encoding.default_internal)
+        end
+      end
+      str
+    end
+
     def deserialize_string_data(buf)
       len = buf.get_int
       bytes = buf.get(len)
@@ -332,10 +342,7 @@ module BSON
       if str.respond_to? "pack"
         str = str.pack("C*")
       end
-      if RUBY_VERSION >= '1.9'
-        str.force_encoding("utf-8")
-      end
-      str
+      encoded_str(str)
     end
 
     def deserialize_code_w_scope_data(buf)
@@ -345,15 +352,12 @@ module BSON
       if code.respond_to? "pack"
         code = code.pack("C*")
       end
-      if RUBY_VERSION >= '1.9'
-        code.force_encoding("utf-8")
-      end
 
       scope_size = buf.get_int
       buf.position -= 4
       scope = BSON_CODER.new().deserialize(buf.get(scope_size))
 
-      Code.new(code, scope)
+      Code.new(encoded_str(code), scope)
     end
 
     def deserialize_oid_data(buf)
@@ -536,10 +540,7 @@ module BSON
         break if b == 0
         chars << b.chr
       end
-      if RUBY_VERSION >= '1.9'
-        chars.force_encoding("utf-8") # Mongo stores UTF-8
-      end
-      chars
+      encoded_str(chars)
     end
 
     def bson_type(o)

test/mongo_bson/bson_test.rb

@@ -90,6 +90,23 @@ class BSONTest < Test::Unit::TestCase
       bson = BSON::BSON_CODER.serialize({'aé'.encode('iso-8859-1') => 'hello'})
       assert_equal 'hello', BSON::BSON_CODER.deserialize(bson)['aé']
     end
+
+    # Based on a test from sqlite3-ruby
+    def test_default_internal_is_honored
+      before_enc = Encoding.default_internal
+
+      str = "壁に耳あり、障子に目あり"
+      bson = BSON::BSON_CODER.serialize("x" => str)
+
+      Encoding.default_internal = 'EUC-JP'
+      out = BSON::BSON_CODER.deserialize(bson)["x"]
+
+      assert_equal Encoding.default_internal, out.encoding
+      assert_equal str.encode('EUC-JP'), out
+      assert_equal str, out.encode(str.encoding)
+    ensure
+      Encoding.default_internal = before_enc
+    end
   end
 
   def test_code