X-Git-Url: http://git.tdb.fi/?p=libs%2Fcore.git;a=blobdiff_plain;f=source%2Fstringcodec%2Fiso2022jp.cpp;fp=source%2Fstringcodec%2Fiso2022jp.cpp;h=d9c7ceac147b0e08ccd3e8c3c684dce6cb947722;hp=0000000000000000000000000000000000000000;hb=b42ed73a1b241c0e93ee03c43c4584b41c549bac;hpb=5b1368cb791cab043f0435628cacbaff36e39b7b

diff --git a/source/stringcodec/iso2022jp.cpp b/source/stringcodec/iso2022jp.cpp
new file mode 100644
index 0000000..d9c7cea
--- /dev/null
+++ b/source/stringcodec/iso2022jp.cpp
@@ -0,0 +1,160 @@
+/* $Id$
+
+This file is part of libmspstrings
+Copyright © 2006-2007 Mikko Rasa
+Distributed under the LGPL
+*/
+#include "ascii.h"
+#include "iso2022jp.h"
+#include "jisx0201.h"
+#include "jisx0208.h"
+
+using namespace std;
+
+namespace Msp {
+namespace Codecs {
+
+void Iso2022Jp::Encoder::encode_char(UnicodeChar ch, string &buf)
+{
+	if(ch>=0 && ch<=0x7F && ch!=0x5C && ch!=0x7E)
+	{
+		if(mode!=ASCII && mode!=JISX0201)
+			switch_mode(ASCII, buf);
+		buf += ch;
+	}
+	else if(ch==0x5C || ch==0x7E)
+	{
+		if(mode!=ASCII)
+			switch_mode(ASCII, buf);
+		buf += ch;
+	}
+	else if(ch==0xA5 || ch==0x203E)
+	{
+		if(mode!=JISX0201)
+			switch_mode(JISX0201, buf);
+		if(ch==0xA5)
+			buf += 0x5C;
+		else if(ch==0x203E)
+			buf += 0x7E;
+	}
+	else
+	{
+		Kuten jis = ucs_to_jisx0208(ch);
+		if(!jis)
+			return error(ch, buf, "Can't express character in ISO-2022-JP");
+
+		if(mode!=JISX0208)
+			switch_mode(JISX0208, buf);
+
+		char jbuf[2] = {jis.ku+0x20, jis.ten+0x20};
+		buf.append(jbuf, 2);
+	}
+}
+
+void Iso2022Jp::Encoder::sync(string &buf)
+{
+	if(mode!=ASCII)
+		switch_mode(ASCII, buf);
+}
+
+void Iso2022Jp::Encoder::reset()
+{
+	mode = ASCII;
+}
+
+void Iso2022Jp::Encoder::switch_mode(Mode m, string &buf)
+{
+	mode = m;
+	switch(mode)
+	{
+	case ASCII:    buf.append("\033(B", 3); break;
+	case JISX0201: buf.append("\033(J", 3); break;
+	case JISX0208: buf.append("\033$B", 3); break;
+	default: throw CodecError("WTF?  Invalid mode in Iso2022Jp::Encoder::switch_mode");
+	}
+}
+
+void Iso2022Jp::Encoder::transliterate(UnicodeChar, string &buf)
+{
+	if(mode!=ASCII)
+		switch_mode(ASCII, buf);
+	buf += '?';
+}
+
+
+Iso2022Jp::Decoder::Decoder(ErrorMode em):
+	Codec::Decoder(em),
+	mode(ASCII),
+	dec(new Ascii::Decoder)
+{ }
+
+UnicodeChar Iso2022Jp::Decoder::decode_char(const string &str, string::const_iterator &i)
+{
+	if(i==str.end())
+		return error("No input");
+
+	while(i!=str.end())
+	{
+		string::const_iterator j = i;
+
+		UnicodeChar result = -1;
+		if(*j==033)
+		{
+			unsigned escape = 0;
+			for(++j; j!=str.end(); ++j)
+			{
+				escape = escape<<8 | static_cast<unsigned char>(*j);
+				if(*j>='@' && *j<='Z')
+					break;
+			}
+
+			bool ok = true;
+			switch(escape)
+			{
+			case 0x2842: switch_mode(ASCII); break;    // ESC ( B
+			case 0x284A: switch_mode(JISX0201); break; // ESC ( J
+			case 0x2440:                               // ESC $ @
+			case 0x2442: switch_mode(JISX0208); break; // ESC $ B
+			default: ok = false;
+			}
+
+			if(ok)
+				i = j;
+			else
+				result = *i;
+			++i;
+		}
+		else if(dec)
+			return dec->decode_char(str, i);
+		else
+			throw CodecError("WTF?  No sub-decoder for Iso2022Jp::Decoder");
+
+		if(result>=0)
+			return result;
+	}
+
+	return -1;
+}
+
+void Iso2022Jp::Decoder::reset()
+{
+	delete dec;
+	mode = ASCII;
+	dec = new Ascii::Decoder;
+}
+
+void Iso2022Jp::Decoder::switch_mode(Mode m)
+{
+	delete dec;
+
+	mode = m;
+	switch(mode)
+	{
+	case ASCII: dec = new Ascii::Decoder; break;
+	case JISX0201: dec = new JisX0201::Decoder; break;
+	case JISX0208: dec = new JisX0208::Decoder; break;
+	}
+}
+
+} // namespace Codecs
+} // namespace Msp