X-Git-Url: http://git.tdb.fi/?p=libs%2Fcore.git;a=blobdiff_plain;f=source%2Fstringcodec%2Futf8.cpp;h=b75b39780edfa6269ad09fa780ca9a565f574872;hp=f56ba546064ee0c234f8d47af5a2d4fc25d0fb04;hb=8245b8036c8bdc51625616ca6248b0f2b0271dc0;hpb=02794ef3620d0d9cc3b8f1c0d8f2995c825fdf4f

diff --git a/source/stringcodec/utf8.cpp b/source/stringcodec/utf8.cpp
index f56ba54..b75b397 100644
--- a/source/stringcodec/utf8.cpp
+++ b/source/stringcodec/utf8.cpp
@@ -7,8 +7,8 @@ namespace StringCodec {
 
 void Utf8::Encoder::encode_char(unichar ch, string &buf)
 {
-	if(ch<0 || ch>0x10FFFF)
-		return error(ch, buf, "Can't express character in UTF-8");
+	if(!is_valid_unichar(ch))
+		return error(ch, buf, invalid_character(ch, "UTF-8"));
 
 	unsigned bytes = 1;
 	if(ch>0xFFFF)
@@ -37,18 +37,18 @@ 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
 }
 
 
 unichar Utf8::Decoder::decode_char(const string &str, string::const_iterator &i)
 {
 	if(i==str.end())
-		return error("No input");
+		return -1;
 
 	if((*i&0xC0)==0x80)
 	{
-		unichar result = error("UTF-8 tail byte found when expecting head");
+		unichar result = error(invalid_sequence(i, i+1, "stray UTF-8 head byte"));
 		++i;
 		return result;
 	}
@@ -68,11 +68,11 @@ unichar Utf8::Decoder::decode_char(const string &str, string::const_iterator &i)
 			result = (result<<6) | ((*j++)&0x3F);
 
 		if(k<bytes)
-			result = error("Incomplete UTF-8 character");
+			result = error(invalid_sequence(i, j, "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))
-			result = error("Invalid Unicode code point");
+			result = error(invalid_sequence(i, j, "denormalized UTF-8 sequence"));
+		else if(!is_valid_unichar(result))
+			result = error(invalid_sequence(i, j, "undefined UTF-8 character"));
 
 		i = j;
 		return result;