X-Git-Url: http://git.tdb.fi/?a=blobdiff_plain;f=source%2Fstringcodec%2Futf8.cpp;h=4c75d8b073e277c2adc2f38db5f3db806da7b0e5;hb=9a79b9ca04d934e8e5b865a1d2b6719f76ec9d15;hp=c7e1705f814e2d099d3b0180730c57e3aa210d85;hpb=b56eb5ec1da675da0c66abc53c1e4f6c4e4cccbd;p=libs%2Fcore.git

diff --git a/source/stringcodec/utf8.cpp b/source/stringcodec/utf8.cpp
index c7e1705..4c75d8b 100644
--- a/source/stringcodec/utf8.cpp
+++ b/source/stringcodec/utf8.cpp
@@ -1,18 +1,11 @@
-/* $Id$
-
-This file is part of libmspstrings
-Copyright © 2006-2007 Mikko Rasa
-Distributed under the LGPL
-*/
-
 #include "utf8.h"
 
 using namespace std;
 
 namespace Msp {
-namespace Codecs {
+namespace StringCodec {
 
-void Utf8::Encoder::encode_char(UnicodeChar ch, string &buf)
+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");
@@ -42,20 +35,20 @@ void Utf8::Encoder::encode_char(UnicodeChar ch, string &buf)
 	}
 }
 
-void Utf8::Encoder::transliterate(UnicodeChar, 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
 }
 
 
-UnicodeChar Utf8::Decoder::decode_char(const string &str, string::const_iterator &i)
+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)
 	{
-		UnicodeChar result = error("UTF-8 tail byte found when expecting head");
+		unichar result = error("UTF-8 tail byte found when expecting head");
 		++i;
 		return result;
 	}
@@ -68,7 +61,7 @@ UnicodeChar Utf8::Decoder::decode_char(const string &str, string::const_iterator
 
 		string::const_iterator j = i;
 
-		UnicodeChar result = (*j++)&(mask-1);
+		unichar result = (*j++)&(mask-1);
 
 		unsigned k;
 		for(k=1; (k<bytes && j!=str.end() && (*j&0xC0)==0x80); ++k)
@@ -78,7 +71,7 @@ UnicodeChar Utf8::Decoder::decode_char(const string &str, string::const_iterator
 			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;
@@ -88,5 +81,5 @@ UnicodeChar Utf8::Decoder::decode_char(const string &str, string::const_iterator
 		return *i++;
 }
 
-} // namespace Codecs
+} // namespace StringCodec
 } // namespace Msp