JIS(ISO-2022-JP)のマルチバイト=>ユニコード変換がほぼ完成

sile · Jan 1, 2012 · e25040f · e25040f
1 parent 2fee6ef
commit e25040f
Show file tree

Hide file tree

Showing 4 changed files with 8,398 additions and 23 deletions.
diff --git a/src/creole_general.erl b/src/creole_general.erl
@@ -8,7 +8,7 @@ from_string(String, Encoding, ErrFn) ->
                   cp932 -> fun creole_to_cp932:to_bytes/1;
                   eucjp -> fun creole_to_eucjp:to_bytes/1;
                   sjis -> fun creole_to_sjis:to_bytes/1;
-                  jisx_0208_1990 -> fun creole_to_jisx_0208_1990:to_bytes/1
+                  jisx_0208_1983 -> fun creole_to_jisx_0208_1983:to_bytes/1
               end,
     from_string_impl(String, ToBytes, ErrFn, []).
 
@@ -34,7 +34,7 @@ to_string(Bytes, Encoding, ErrFn) ->
                 cp932 -> creole_from_cp932:da_nodes();
                 eucjp -> creole_from_eucjp:da_nodes();
                 sjis -> creole_from_sjis:da_nodes();
-                jisx_0208_1990 -> creole_from_jisx_0208_1990:da_nodes()
+                jisx_0208_1983 -> creole_from_jisx_0208_1983:da_nodes()
             end,
     to_string_impl(Bytes, Nodes, ErrFn, []).
 

diff --git a/src/creole_jis.erl b/src/creole_jis.erl
@@ -10,38 +10,66 @@ to_string(String, ErrFn) ->
 
 to_string_impl(<<>>, _, _, Acc) ->
    lists:flatten(lists:reverse(Acc));
-to_string_impl(Bytes, ErrFn, Isoreg, Acc) ->
+to_string_impl(Bytes, ErrFn, Mode, Acc) ->
     case Bytes of
-        <<"\e(B", Rest/binary>> ->  % ASCII
+        %% ASCII
+        <<"\e(B", Rest/binary>> ->  
             to_string_impl(Rest, ErrFn, ascii, Acc);
+
+        %% JIS X 0201-1976 Roman Set
+        <<"\e(J", Rest/binary>> -> 
+            to_string_impl(Rest, ErrFn, roman, Acc);
+
+        %% JIS X 0208-1978
+        <<"\e$@", Rest/binary>> ->
+            %% XXX: using jisx_0208_1983 instead of jisx_0208_1978
+            to_string_impl(Rest, ErrFn, jisx_0208_1983, Acc);
 
-        <<"\e$B", Rest/binary>> ->  % JIS X 0208-1983
+        %% JIS X 0208-1983
+        <<"\e$B", Rest/binary>> ->  
             to_string_impl(Rest, ErrFn, jisx_0208_1983, Acc);
+
         _ ->
-            to_string_impl_isoreq(Bytes, ErrFn, Isoreg, Acc)
+            to_string_impl2(Bytes, ErrFn, Mode, Acc)
     end.
 
-to_string_impl_isoreq(<<C:1/binary, Rest/binary>>=Bytes, ErrFn, ascii, Acc) ->
-    if 
-        C >= 80 -> 
-            to_string_impl(Rest, ErrFn, ascii, [binary:first(C) | Acc]);
-        true ->
-            {S, Rest} = ErrFn(Bytes),
-            to_string_impl(Rest, ErrFn, ascii, [S | Acc])
+to_string_impl2(<<B:1/binary, Rest/binary>>=Bytes, ErrFn, ascii, Acc) ->
+    case binary:first(B) of
+        C when C =< 16#FF -> 
+            to_string_impl(Rest, ErrFn, ascii, [C | Acc]);
+        _ ->
+            handle_error(Bytes, ErrFn, ascii, Acc)
+    end;
+
+to_string_impl2(<<B:1/binary, Rest/binary>>=Bytes, ErrFn, roman, Acc) ->
+    case binary:first(B) of
+        $\ ->
+            to_string_impl(Rest, ErrFn, ascii, [65509 | Acc]); % FULLWIDTH_YEN_SIGN
+        $~ ->
+            to_string_impl(Rest, ErrFn, ascii, [65507 | Acc]); % FULLWIDTH_MACRON
+        C when C =< 16#FF -> 
+            to_string_impl(Rest, ErrFn, ascii, [C | Acc]);
+        _ ->
+            handle_error(Bytes, ErrFn, roman, Acc)
     end;
 
-to_string_impl_isoreq(Bytes, ErrFn, jisx_0208_1983, Acc) ->
+to_string_impl2(Bytes, ErrFn, jisx_0208_1983, Acc) ->
     ErrFn2 = fun (<<"\e", _/binary>>=Bs) ->
-                     %% XXX: 再帰呼び出しが深くなる？ => 処理の継続可否を判断するフラグも返せるようにすれば解決？
-                     %% XXX:
-                     {to_string_impl(Bs, ErrFn, ascii, []) ++ lists:reverse(Acc), 
-                      <<>>};
+                     {[], Bs, false};
                  (Bs) ->
                      ErrFn(Bs)
              end,
+    case creole_general:to_string(Bytes, jisx_0208_1983, ErrFn2) of
+        {abort, PartialResult, Rest} ->
+            to_string_impl(Rest, ErrFn, jisx_0208_1983, [PartialResult | Acc]);
+        Result ->
+            to_string_impl(<<>>, ErrFn, jisx_0208_1983, [Result | Acc])
+    end.
 
-    %% XXX: 間違い 1990 を使っている
-    creole_general:to_string(Bytes, jisx_0208_1990, ErrFn2);
-
-to_string_impl_isoreq(_, _, _, _) ->
-    ok.
+handle_error(Bytes, ErrFn, Mode, Acc) ->
+    case ErrFn(Bytes) of
+        {S, Rest, true} ->
+            to_string_impl(Rest, ErrFn, Mode, [S | Acc]);
+        {S, Rest, false} ->
+            {abort, to_string_impl(<<>>, ErrFn, Mode, [S | Acc]), Rest}
+    end.