Put unichar and ustring in their own file

[libs/core.git] / source / stringcodec / utf8.cpp

diff --git a/source/stringcodec/utf8.cpp b/source/stringcodec/utf8.cpp
index b0431eb4f49e3d595ddefd44e5dfca57e421d060..4c75d8b073e277c2adc2f38db5f3db806da7b0e5 100644 (file)
--- a/source/stringcodec/utf8.cpp
+++ b/source/stringcodec/utf8.cpp
@@ -37,7 +37,7 @@ void Utf8::Encoder::encode_char(unichar ch, string &buf)
 
 void Utf8::Encoder::transliterate(unichar, string &buf)
 {
-       buf.append("\357\277\275", 3);  // � U+FFFE Replacement Character
+       buf.append("\357\277\275", 3);  // � U+FFFD Replacement Character
 }
 
 
@@ -71,7 +71,7 @@ unichar Utf8::Decoder::decode_char(const string &str, string::const_iterator &i)
                        result = error("Incomplete UTF-8 character");
                else if(!(result>>(bytes*5-4)) || !(result>>7))
                        result = error("Denormalized UTF-8 multibyte sequence");
-               else if(result>0x10FFFF || (result>=0xD800 && result<=0xDFFF))
+               else if(!is_valid_unichar(result))
                        result = error("Invalid Unicode code point");
 
                i = j;