Move files around to prepare for assimilation into core

[libs/core.git] / source / stringcodec / jisx0208.cpp

diff --git a/source/stringcodec/jisx0208.cpp b/source/stringcodec/jisx0208.cpp
new file mode 100644 (file)
index 0000000..6b46b1b
--- /dev/null
+++ b/source/stringcodec/jisx0208.cpp
@@ -0,0 +1,93 @@
+/* $Id$
+
+This file is part of libmspstrings
+Copyright © 2006-2007 Mikko Rasa
+Distributed under the LGPL
+*/
+
+#include "jisx0208.h"
+
+#include "jisx0208.table"
+
+using namespace std;
+
+namespace Msp {
+namespace Codecs {
+
+void JisX0208::Encoder::encode_char(UnicodeChar ucs, string &buf)
+{
+       unsigned short jis = ucs_to_jisx0208(ucs);
+       if(jis)
+       {
+               char jbuf[2] = {jis>>8, jis};
+               buf.append(jbuf, 2);
+       }
+       else
+               error(ucs, buf, "Can't express character in JIS X 0208");
+}
+
+void JisX0208::Encoder::transliterate(UnicodeChar, string &buf)
+{
+       buf.append("!)", 2);
+}
+
+
+UnicodeChar JisX0208::Decoder::decode_char(const string &str, string::const_iterator &i)
+{
+       if(i==str.end())
+               return error("No input");
+
+       string::const_iterator j = i;
+       Kuten jis;
+       jis.ku = *j++-0x20;
+
+       UnicodeChar result;
+       if(j==str.end())
+               result = error("Incomplete JIS X 0208 character");
+       else
+       {
+               jis.ten = *j++-0x20;
+               result = jisx0208_to_ucs(jis);
+               if(result==0)
+                       result = error("Undefined JIS X 0208 character");
+       }
+
+       i = j;
+       return result;
+}
+
+
+UnicodeChar jisx0208_to_ucs(Kuten jis)
+{
+       if(jis.ku==0 || jis.ku>0x5E || jis.ten==0 || jis.ten>0x5E)
+               return 0;
+
+       return jisx0208_to_ucs_table[jis.ku*94 + jis.ten - 95];
+}
+
+Kuten ucs_to_jisx0208(UnicodeChar c)
+{
+       if(c<0 || c>0xFFFF)
+               return Kuten();
+
+       unsigned i = 0;
+       for(unsigned bit=0x1000; bit; bit>>=1)
+       {
+               if(i+bit>=ucs_to_jisx0208_table_size)
+                       continue;
+               if(ucs_to_jisx0208_table[i+bit].ucs<=static_cast<unsigned short>(c))
+                       i += bit;
+       }
+
+       Kuten result;
+       if(ucs_to_jisx0208_table[i].ucs==static_cast<unsigned short>(c))
+       {
+               result.ku = (ucs_to_jisx0208_table[i].jis>>8)+1;
+               result.ten = ucs_to_jisx0208_table[i].jis+1;
+       }
+
+       return result;
+}
+
+} // namespace Codecs
+} // namespace Msp