Rework the codec API completely to remove the internal buffering

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

diff --git a/source/jisx0208.cpp b/source/jisx0208.cpp
index 4d46f709f89df5f739aa7c11a6a298505f41ed45..0dd0d211743c5f3680376e069760189e26c2313c 100644 (file)
--- a/source/jisx0208.cpp
+++ b/source/jisx0208.cpp
@@ -1,3 +1,10 @@
+/* $Id$
+
+This file is part of libmspstrings
+Copyright © 2006-2007 Mikko Rasa
+Distributed under the LGPL
+*/
+
 #include "jisx0208.h"
 
 #include "jisx0208.table"
@@ -5,59 +12,63 @@
 using namespace std;
 
 namespace Msp {
+namespace Codecs {
 
-void JisX0208::Encoder::encode_char(wchar_t ucs)
+void JisX0208::Encoder::encode_char(UnicodeChar ucs, string &buf)
 {
        unsigned short jis=ucs_to_jisx0208(ucs);
-       if(!jis) throw CodecError("Can't express character in JIS X 0208");
-
-       char buf[2]={jis>>8, jis};
-       append(buf, 2);
+       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::Decoder::decode_char(const string &str, string::const_iterator &i)
+void JisX0208::Encoder::transliterate(UnicodeChar, string &buf)
 {
-       if(i==str.end())
-               return;
+       buf.append("!)", 2);
+}
 
-       if(!high)
-               high=*i++;
 
+UnicodeChar JisX0208::Decoder::decode_char(const string &str, string::const_iterator &i)
+{
        if(i==str.end())
-               return;
-
-       wchar_t ucs=jisx0208_to_ucs(high<<8 | *i++);
-       high=0;
+               return error("No input");
 
-       if(!ucs)
-               throw CodecError("Invalid JIS X 0208 string (undefined character)");
+       string::const_iterator j=i;
+       Kuten jis;
+       jis.ku=*j++-0x20;
 
-       append(ucs);
-}
+       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");
+       }
 
-void JisX0208::Decoder::sync()
-{
-       if(high)
-               throw CodecError("Sync in middle of JIS X 0208 character");
+       i=j;
+       return result;
 }
 
-void JisX0208::Decoder::reset()
-{
-       high=0;
-}
 
-wchar_t jisx0208_to_ucs(unsigned short jis)
+UnicodeChar jisx0208_to_ucs(Kuten jis)
 {
-       if((jis&0xFF)<0x21 || (jis&0xFF)>0x7E || (jis&0xFF00)<0x2100 || (jis&0xFF00)>0x7E00)
+       if(jis.ku==0 || jis.ku>0x5E || jis.ten==0 || jis.ten>0x5E)
                return 0;
 
-       return jisx0208_to_ucs_table[((jis>>8)&0xFF)*94 + (jis&0xFF)-0x21*95];
+       return jisx0208_to_ucs_table[jis.ku*94 + jis.ten - 95];
 }
 
-unsigned short ucs_to_jisx0208(wchar_t c)
+Kuten ucs_to_jisx0208(UnicodeChar c)
 {
-       if(c&0xFFFF0000) return 0;
+       if(c<0 || c>0xFFFF)
+               return Kuten();
 
        unsigned i=0;
        for(unsigned bit=0x1000; bit; bit>>=1)
@@ -68,9 +79,15 @@ unsigned short ucs_to_jisx0208(wchar_t c)
                        i+=bit;
        }
 
+       Kuten result;
        if(ucs_to_jisx0208_table[i].ucs==static_cast<unsigned short>(c))
-               return ucs_to_jisx0208_table[i].jis;
-       return 0;
+       {
+               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