/* shindou7.c - Don Yang (uguu.org)

   02/19/07
                                                                           */

#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#ifdef _WIN32
   #include<io.h>
   #include<fcntl.h>
#endif

static FILE *infile;
static unsigned char *tmp, *buffer,
   *const_str = " \0!\0 \xef\xbc\x81I\1\xff\1\xa1\xaa";
static int u, a,
           state,
           buffer_size, read_size,
           i, i0, last_char_offset,
           c,
           utf8_kana,
           sjis_kana,
           eucjp_kana,
           encoding;

static void Output(void *data, size_t size)
{
   fwrite(data, size, 1, stdout);
}

static void ExtendBuffer(void)
{
   if( read_size - buffer_size )
   {
      read_size +=
         fread(buffer + read_size, 1, buffer_size - read_size, infile);
   }
   else
   {
      tmp = buffer;
      buffer = malloc((buffer_size *= 2) + 3);
      memcpy(buffer, tmp, read_size);
      free(tmp);
   }
}

static void NextByte(void)
{
   c = (c << 8) | buffer[i++];
}

static void NextUCS2BEChar(void)
{
   c = 0;
   NextByte();
   NextByte();
}

static void NextUCS2LEChar(void)
{
   c = buffer[i] | (buffer[i + 1] << 8);
   i += 2;
}

static void NextUTF8Char(void)
{
   c = 0;
   if( (buffer[i] & 224) - 192 )
   {
      if( (buffer[i] & 240) - 224 )
      {
         if( (buffer[i] & 248) - 240 )
         {
            NextByte();
         }
         else
         {
            if( (buffer[i + 1] & 192) - 128 ||
                (buffer[i + 2] & 192) - 128 ||
                (buffer[i + 2] & 192) - 128 )
            {
               c--;
            }
            else
            {
               NextByte();
               NextByte();
               NextByte();
               NextByte();
            }
         }
      }
      else
      {
         if( (buffer[i + 1] & 192) - 128 ||
             (buffer[i + 2] & 192) - 128 )
         {
            c--;
         }
         else
         {
            NextByte();
            NextByte();
            NextByte();
         }
      }
   }
   else
   {
      if( (buffer[i + 1] & 192) - 128 )
      {
         c--;
      }
      else
      {
         NextByte();
         NextByte();
      }
   }
}

static void NextSJISChar(void)
{
   c = 0;
   if( (buffer[i] > 128 && buffer[i] < 160) ||
       (buffer[i] > 223 && buffer[i] < 240) )
   {
      if( buffer[i + 1] < 64 || buffer[i + 1] > 252 )
      {
         c--;
      }
      else
      {
         NextByte();
         NextByte();
      }
   }
   else
   {
      NextByte();
   }
}

static void NextEUCJPChar(void)
{
   c = 0;
   if( (buffer[i] > 160 && buffer[i] < 169) ||
       buffer[i] == 173 ||
       (buffer[i] > 175 && buffer[i] < 245) ||
       (buffer[i] > 248 && buffer[i] < 253) )
   {
      if( buffer[i + 1] < 161 || buffer[i + 1] > 254 )
      {
         c--;
      }
      else
      {
         NextByte();
         NextByte();
      }
   }
   else
   {
      NextByte();
   }
}

static void DetectFileEncoding(void)
{
   read_size =
   utf8_kana =
   sjis_kana =
   eucjp_kana = 0;
   encoding = 28;
   for(; !feof(infile) && encoding & (encoding - 1);)
   {
      ExtendBuffer();

      if( !(encoding & 3) )
      {
         if( buffer[0] - 255 || buffer[1] - 254 )
         {
            if( buffer[0] - 254 || buffer[1] - 255 )
            {
               for(i = 0; i < read_size - 1; i += 2)
               {
                  if( buffer[i] == 10 && !buffer[i + 1] )
                  {
                     encoding = 1;
                     goto finalize;
                  }
                  if( !buffer[i] && buffer[i + 1] == 10 )
                  {
                     encoding = 2;
                     goto finalize;
                  }
               }

               if( buffer[0] - 239 || buffer[1] - 187 || buffer[2] - 191 )
               {
                  for(i = 0; i < read_size - 3 && encoding & 4;)
                  {
                     NextUTF8Char();
                     if( c < 0 )
                        encoding &= ~4;
                     else if( (c > 0xe38180 && c < 0xe381c0) ||
                              (c > 0xe3827f && c < 0xe38294) )
                        utf8_kana++;
                  }

                  for(i = 0; i < read_size - 1 && encoding & 8;)
                  {
                     NextSJISChar();
                     if( c < 0 )
                        encoding &= ~8;
                     else if( c > 33438 && c < 33522 )
                        sjis_kana++;
                  }

                  for(i = 0; i < read_size - 1 && encoding & 16;)
                  {
                     NextEUCJPChar();
                     if( c < 0 )
                        encoding &= ~16;
                     else if( c > 42144 && c < 42228 )
                        eucjp_kana++;
                  }
               }
               else
               {
                  encoding = 4;
               }
            }
            else
            {
               encoding = 2;
            }
         }
         else
         {
            encoding = 1;
         }
      }
   }

finalize:
   if( !(encoding & 3) )
   {
      if( !(encoding & 4) ) utf8_kana = 0;
      if( !(encoding & 8) ) sjis_kana = 0;
      if( !(encoding & 16) ) eucjp_kana = 0;

      encoding =
         utf8_kana < sjis_kana
            ?  sjis_kana < eucjp_kana ? 16 : 8
            :  utf8_kana < eucjp_kana ? 16 :
               encoding & 4 ? 4 : 8;
   }
}

static int IsAlpha(void)
{
   return
      (c > 96 && c < 123) ||
      (c > 64 && c < 91) ||
      (c > 47 && c < 58) ||
      c == 45 || c == 95 ||
      c == 32 || c == 9;
}

static int IsExclaimationPunct(void)
{
   return
      c == 33 ||
      (encoding & 3 && c == 65281) ||
      (encoding & 4 && c == 0xefbc81) ||
      (encoding & 8 && c == 33097) ||
      (encoding & 16 && c == 41386);
}

static int IsEOSPunct(void)
{
   return
      c == 46 || c == 63 ||
      (encoding & 3 &&
       (c == 65311 || c == 65294 || c == 12290 || c == 65377)) ||
      (encoding & 4 &&
       (c == 0xefbc9f || c == 0xefbc8e || c == 0xe38082 || c == 0xefbda1)) ||
      (encoding & 8 &&
       (c == 33096 || c == 33092 || c == 33090)) ||
      (encoding & 16 &&
       (c == 41385 || c == 41381 || c == 41379)) ||
      IsExclaimationPunct();
}

static int IsPunct(void)
{
   return
      c == 44 || c == 10 ||
      (encoding & 3 &&
       (c == 65292 || c == 12289 || c == 65380)) ||
      (encoding & 4 &&
       (c == 0xefbc8c || c == 0xe38081 || c == 0xefbda4)) ||
      (encoding & 8 &&
       (c == 33091 || c == 33089)) ||
      (encoding & 16 &&
       (c == 41380 || c == 41378)) ||
      IsEOSPunct();
}

static void ToUpperChar(void)
{
   buffer[u] -= buffer[u] > 96 && buffer[u] < 123 ? 32 : 0;
   u++;
}

static void ToUpperUCS2LE(void)
{
   for(u = i0; u < i; u++)
      ToUpperChar();
}

static void ToUpperUCS2BE(void)
{
   i0++;
   ToUpperUCS2LE();
   i0--;
}

static void ToUpperUTF8(void)
{
   for(u = i0; u < i;)
   {
      if( (buffer[u] & 248) - 240 )
      {
         if( (buffer[u] & 240) - 224 )
         {
            if( (buffer[u] & 224) - 192 )
               ToUpperChar();
            else
               u += 2;
         }
         else
         {
            u += 3;
         }
      }
      else
      {
         u += 4;
      }
   }
}

static void ToUpperJIS(void)
{
   for(u = i0; u < i;)
   {
      if( (buffer[u] > 128 && buffer[i] < 160) ||
          (buffer[u] > 223 && buffer[i] < 240) )
         u++;
      else
         ToUpperChar();
   }
}

static void WriteExclamationMark(void)
{
   Output(const_str +
          (c > 255
            ?  encoding & 1 ? 9 :
               encoding & 2 ? 10 :
               encoding & 4 ? 5 :
               encoding & 8 ? 7 : 12
            :  encoding & 2 ? 1 : 2),
          (c > 255
            ? encoding & 4 ? 3 : 2
            : encoding & 3 ? 2 : 1));
}

static void Flush(void)
{
   if( i0 - last_char_offset )
      Output(buffer + i0, last_char_offset - i0);
   if( IsExclaimationPunct() )
   {
      WriteExclamationMark();
      WriteExclamationMark();
   }
   else if( IsPunct() && c - 10 )
   {
      WriteExclamationMark();
   }
   else
   {
      Output(buffer + last_char_offset, i - last_char_offset);
   }
}

static void FlushUCS2WithSpace(void)
{
   for(; i0 < last_char_offset; i0 += 2)
   {
      Output(buffer + i0, 2);
      Output(const_str + (encoding & 1 ? 0 : 3), 2);
   }
}

static void FlushUTF8WithSpace(void)
{
   for(; i0 < last_char_offset; Output(const_str, 1))
   {
      u =
         (buffer[i0] & 248) - 240
            ?  (buffer[i0] & 240) - 224
               ?  (buffer[i0] & 224) - 192
                  ? 1
                  : 2
               : 3
            : 4;
      Output(buffer + i0, u);
      i0 += u;
   }
}

static void FlushJISWithSpace(void)
{
   for(; i0 < last_char_offset; i0 += u)
   {
      u = buffer[i0] > 127 ? 2 : 1;
      Output(buffer + i0, u);
      Output(const_str, 1);
   }
}

static void Filter(void (*NextChar)(void),
                   void (*ToUpper)(void),
                   void (*FlushWithSpace)(void))
{
   i = i0 =
   state = 0;
   for(; !feof(infile) || i < read_size;)
   {
      last_char_offset = i;
      NextChar();
      if( i < read_size - 3 )
      {
         memmove(buffer, buffer + i0, read_size - i0);
         read_size -= i0;
         i -= i0;
         last_char_offset -= i0;
         i0 = 0;

         ExtendBuffer();
      }

      if( state - 1 )
      {
         if( state - 2 )
         {
            if( c > 127 )
            {
               state = 2;
            }
            else if( IsAlpha() )
            {
               state = 1;
            }
            else
            {
               Output(buffer + last_char_offset, i - last_char_offset);
            }
            i0 = last_char_offset;
         }
         else
         {
            if( IsAlpha() || IsPunct() )
            {
               if( IsEOSPunct() )
                  FlushWithSpace();
               Flush();
               state = 0;
            }
         }
      }
      else
      {
         if( !IsAlpha() || IsPunct() )
         {
            if( IsEOSPunct() )
               ToUpper();
            Flush();
            state = 0;
         }
      }
   }
}

int main(int argc, char **argv)
{
   #ifdef _WIN32
      setmode(fileno(stdin), O_BINARY);
      setmode(fileno(stdout), O_BINARY);
   #endif

   buffer_size = 4096;
   buffer = malloc(buffer_size);

   if( argc > 1 )
   {
      for(a = 1; a < argc; a++)
      {
         if( (infile = fopen(argv[a], "rb")) )
         {
            DetectFileEncoding();
            printf("%s: %s\n", argv[a],
                   encoding - 1
                     ? encoding - 2
                        ? encoding - 4
                           ? encoding - 8
                              ? "EUC-JP"
                              : "Shift_JIS"
                           : "UTF-8"
                        : "UCS-2BE"
                     : "UCS-2LE");

            fclose(infile);
         }
         else
         {
            printf("%s: can not open file\n", argv[a]);
         }
      }
   }
   else
   {
      infile = stdin;
      DetectFileEncoding();
      if( encoding & 3 )
      {
         if( encoding & 1 )
            Filter(NextUCS2LEChar, ToUpperUCS2LE, FlushUCS2WithSpace);
         else
            Filter(NextUCS2BEChar, ToUpperUCS2BE, FlushUCS2WithSpace);
      }
      else if( encoding & 4 )
      {
         Filter(NextUTF8Char, ToUpperUTF8, FlushUTF8WithSpace);
      }
      else
      {
         if( encoding & 8 )
            Filter(NextSJISChar, ToUpperJIS, FlushJISWithSpace);
         else
            Filter(NextEUCJPChar, ToUpperJIS, FlushJISWithSpace);
      }
   }

   free(buffer);
   return 0;
}