X-Git-Url: http://git.tdb.fi/?p=libs%2Fcore.git;a=blobdiff_plain;f=source%2Fstringcodec%2Fcodec.cpp;h=abec134b1a0fd00a25f936952368beba22c62742;hp=0014847a67cebd30a12585980f466cdf7e5ea98c;hb=HEAD;hpb=b56eb5ec1da675da0c66abc53c1e4f6c4e4cccbd

diff --git a/source/stringcodec/codec.cpp b/source/stringcodec/codec.cpp
index 0014847..ff2bd63 100644
--- a/source/stringcodec/codec.cpp
+++ b/source/stringcodec/codec.cpp
@@ -1,10 +1,3 @@
-/* $Id$
-
-This file is part of libmspstrings
-Copyright © 2006-2007 Mikko Rasa
-Distributed under the LGPL
-*/
-
 #include "ascii.h"
 #include "codec.h"
 #include "iso2022jp.h"
@@ -13,27 +6,22 @@ Distributed under the LGPL
 #include "iso885915.h"
 #include "jisx0201.h"
 #include "jisx0208.h"
+#include "utf16.h"
 #include "utf8.h"
 #include "windows1252.h"
 
 using namespace std;
 
 namespace Msp {
-namespace Codecs {
+namespace StringCodec {
 
 bool Codec::detect(const string &str) const
 {
-	Decoder *dec = create_decoder();
+	Decoder *dec = create_decoder(IGNORE_ERRORS);
+
 	bool result = true;
-	try
-	{
-		for(string::const_iterator i=str.begin(); i!=str.end(); )
-			dec->decode_char(str, i);
-	}
-	catch(const CodecError &)
-	{
-		result = false;
-	}
+	for(auto i=str.begin(); (result && i!=str.end()); )
+		result = (dec->decode_char(str, i)!=-1);
 
 	delete dec;
 
@@ -42,8 +30,8 @@ bool Codec::detect(const string &str) const
 
 void Codec::Encoder::encode(const ustring &str, string &buf)
 {
-	for(ustring::const_iterator i=str.begin(); i!=str.end(); ++i)
-		encode_char(*i, buf);
+	for(unichar c: str)
+		encode_char(c, buf);
 }
 
 string Codec::Encoder::encode(const ustring &str)
@@ -54,25 +42,13 @@ string Codec::Encoder::encode(const ustring &str)
 	return buf;
 }
 
-void Codec::Encoder::error(UnicodeChar ch, string &buf, const string &msg)
-{
-	switch(err_mode)
-	{
-	case TRANSLITERATE:
-		transliterate(ch, buf);
-	case IGNORE_ERRORS:
-		break;
-	default:
-		throw CodecError(msg);
-	}
-}
 
 
 void Codec::Decoder::decode(const string &str, ustring &buf)
 {
-	for(string::const_iterator i=str.begin(); i!=str.end();)
+	for(auto i=str.begin(); i!=str.end();)
 	{
-		UnicodeChar c = decode_char(str, i);
+		unichar c = decode_char(str, i);
 		if(c!=-1)
 			buf += c;
 	}
@@ -85,40 +61,47 @@ ustring Codec::Decoder::decode(const string &str)
 	return buf;
 }
 
-UnicodeChar Codec::Decoder::error(const string &msg)
-{
-	switch(err_mode)
-	{
-	case TRANSLITERATE:
-		return 0xFFFE;
-	case IGNORE_ERRORS:
-		return -1;
-	default:
-		throw CodecError(msg);
-	}
-}
-
 Codec *create_codec(const string &n)
 {
 	string name;
-	for(string::const_iterator i=n.begin(); i!=n.end(); ++i)
+	string::const_iterator i;
+	for(i=n.begin(); i!=n.end(); ++i)
 	{
-		if(isupper(*i))
+		if(*i==':')
+			break;
+		else if(isupper(*i))
 			name += tolower(*i);
 		else if(islower(*i) || isdigit(*i))
 			name += *i;
 	}
 
-	if(name=="ascii") return new Ascii;
-	if(name=="iso2022jp") return new Iso2022Jp;
-	if(name=="iso646fi") return new Iso646Fi;
-	if(name=="iso88591" || name=="latin1") return new Iso88591;
-	if(name=="iso885915" || name=="latin9") return new Iso885915;
-	if(name=="jisx0201") return new JisX0201;
-	if(name=="jisx0208") return new JisX0208;
-	if(name=="utf8") return new Utf8;
-	if(name=="windows1252" || name=="cp1252") return new Windows1252;
-	throw InvalidParameterValue("Unknown string codec");
+	ErrorMode em = DEFAULT;
+	if(i!=n.end() && *i==':')
+	{
+		string em_str(i+1, n.end());
+		if(em_str=="throw")
+			em = THROW_ON_ERROR;
+		else if(em_str=="ignore")
+			em = IGNORE_ERRORS;
+		else if(em_str=="trans" || em_str=="transliterate")
+			em = TRANSLITERATE;
+		else
+			throw invalid_argument("StringCodec::create_codec");
+	}
+
+	if(name=="ascii") return new Ascii(em);
+	if(name=="iso2022jp") return new Iso2022Jp(em);
+	if(name=="iso646fi") return new Iso646Fi(em);
+	if(name=="iso88591" || name=="latin1") return new Iso88591(em);
+	if(name=="iso885915" || name=="latin9") return new Iso885915(em);
+	if(name=="jisx0201") return new JisX0201(em);
+	if(name=="jisx0208") return new JisX0208(em);
+	if(name=="utf8") return new Utf8(em);
+	if(name=="utf16") return new Utf16(em, Utf16::AUTO);
+	if(name=="utf16be") return new Utf16(em, Utf16::BIG);
+	if(name=="utf16le") return new Utf16(em, Utf16::LITTLE);
+	if(name=="windows1252" || name=="cp1252") return new Windows1252(em);
+	throw invalid_argument("StringCodec::create_codec");
 }
 
 Codec *detect_codec(const string &str)
@@ -128,9 +111,8 @@ Codec *detect_codec(const string &str)
 	bool is_latin1 = true;
 	unsigned utf8_mb = 0;
 
-	for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
+	for(char c: str)
 	{
-		unsigned char c = *i;
 		if(c&0x80)
 		{
 			is_ascii = false;
@@ -173,5 +155,5 @@ Codec *detect_codec(const string &str)
 		return new Windows1252;
 }
 
-} // namespace Codecs
+} // namespace StringCodec
 } // namespace Msp