X-Git-Url: http://git.tdb.fi/?p=libs%2Fcore.git;a=blobdiff_plain;f=source%2Fcodec.cpp;h=42315e52e9f7179540bcb5836021328f470cee2b;hp=865176edf2a8cb5223e721cdb6040a0511b4acd3;hb=0ac97daad8d24f6762e1b870fec782c7aace695a;hpb=c88a239ff3b218ad0226e8c53356350ccc1f7014

diff --git a/source/codec.cpp b/source/codec.cpp
index 865176e..42315e5 100644
--- a/source/codec.cpp
+++ b/source/codec.cpp
@@ -46,6 +46,14 @@ void Codec::Encoder::encode(const ustring &str, string &buf)
 		encode_char(*i, buf);
 }
 
+string Codec::Encoder::encode(const ustring &str)
+{
+	string buf;
+	encode(str, buf);
+	sync(buf);
+	return buf;
+}
+
 void Codec::Encoder::error(UnicodeChar ch, string &buf, const string &msg)
 {
 	switch(err_mode)
@@ -70,6 +78,13 @@ void Codec::Decoder::decode(const string &str, ustring &buf)
 	}
 }
 
+ustring Codec::Decoder::decode(const string &str)
+{
+	ustring buf;
+	decode(str, buf);
+	return buf;
+}
+
 UnicodeChar Codec::Decoder::error(const string &msg)
 {
 	switch(err_mode)
@@ -102,9 +117,61 @@ Codec *create_codec(const string &n)
 	if(name=="jisx0201") return new JisX0201;
 	if(name=="jisx0208") return new JisX0208;
 	if(name=="utf8") return new Utf8;
-	if(name=="windows1252") return new Windows1252;
+	if(name=="windows1252" || name=="cp1252") return new Windows1252;
 	throw InvalidParameterValue("Unknown string codec");
 }
 
+Codec *detect_codec(const string &str)
+{
+	bool is_utf8=true;
+	bool is_ascii=true;
+	bool is_latin1=true;
+	unsigned utf8_mb=0;
+
+	for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
+	{
+		unsigned char c=*i;
+		if(c&0x80)
+		{
+			is_ascii=false;
+			if((c&0xC0)==0x80)
+			{
+				if((c&0xE0)==0x80)
+					is_latin1=false;
+				if(utf8_mb)
+					--utf8_mb;
+				else
+					is_utf8=false;
+			}
+			else if((c&0xC0)==0xC0)
+			{
+				if(utf8_mb)
+				{
+					is_utf8=false;
+					utf8_mb=0;
+				}
+				else
+				{
+					for(utf8_mb=1; (c>>(6-utf8_mb))&1; ++utf8_mb) ;
+				}
+			}
+		}
+		else if(utf8_mb)
+		{
+			is_utf8=false;
+			utf8_mb=0;
+		}
+	}
+
+	if(is_ascii)
+		return new Ascii;
+	else if(is_utf8)
+		return new Utf8;
+	else if(is_latin1)
+		return new Iso88591;
+	else
+		return new Windows1252;
+}
+
 } // namespace Codecs
 } // namespace Msp