jadecoder: remove charset check; now kakasi.py handle unknown chars better

src/calibre/ebooks/unihandecode/jadecoder.py

@@ -34,7 +34,6 @@ class Jadecoder(Unidecoder):
 
     def decode(self, text):
         try:
-            dummy = text.encode("eucjp") # test if text contains only Japanese and ASCII characters.
             result=self.kakasi.do(text)
             return re.sub('[^\x00-\x7f]', lambda x: self.replace_point(x.group()),result)
         except:

src/calibre/ebooks/unihandecode/pykakasi/kakasi.py

@@ -53,7 +53,9 @@ class kakasi(object):
             if self.j2h.isKanji(text[i]):
                 (t, l) = self.j2h.convert(text[i:])
                 if l <= 0:
-                    break
+                    otext  = otext + text[i]
+                    i = i + 1
+                    continue
                 i = i + l
                 m = 0
                 tmptext = ""