/* Last edited on 2024-12-25 10:17:34 by stolfi */
/*
  Reads a GB-encoded Chinese text, insertes spaces around each Chinese 
  character (a pair of bytes in 0xA0 .. 0xFF).
*/

#include <stdio.h>
#include <stdlib.h>

#include <spread-chinese-text_doc.h>

#define LOCHA (0x80)
#define HICHA (0xFF)
#define LOCHB (0x21)
#define HICHB (0xFF)
#define SPACE (0x20)

void doputchar(int c);
void error(char *msg);

static int nerrors = 0;
static int nread = 0;

int main(int argc, char **argv)
{
  int q, r, s;
  q = 0;
  while(1)
    {
      r = getchar();
      if (r == EOF)
        { fclose(stdout); exit(0); }
      else
        { nread++; 
          if ((r < LOCHA) || (r > HICHA))
            { doputchar(r); q = r; } 
          else
            { s = getchar(); 
              if (s == EOF)
                { doputchar(r); q = r; error("unexpected EOF in Chinese character"); }
              else 
                { nread++;
                  if ((s < LOCHB) || (s > HICHB))
                    { doputchar(r); doputchar(s); q = s; error("incomplete Chinese character"); }
                  else
                    { if (q != SPACE) { doputchar(SPACE); }
                      doputchar(r); doputchar(s);
                      doputchar(SPACE);
                      q = SPACE;
                    }
                }
            }
        }
    }
}

void doputchar(int c)
{
  int res = putchar(c);
  if (res != c) { error("write error"); }
}

#define MAXERRORS (10)

void error(char *msg)
{
  fprintf(stdout, "«ERR»");
  if (nerrors < MAXERRORS) 
    { fprintf(stderr, "byte %d: %s\n", nread, msg); }
  else if (nerrors == MAXERRORS) 
    { fprintf(stderr, "(more...)\n"); } 
  nerrors++;
}