#include <stdio.h>
#include <stdint.h>
#include <stdlib.h>
#include <unistd.h>
#include <stdbool.h>
#include <string.h>

int verbose1 = false; // show gross parsing state changes
int verbose2 = false; // show legitimate non-cp0 UTF-8 characters
int verbose3 = false; // show cp0 characters (ASCII)
int verbose5 = false; // show UTF-8 problems 

unsigned long long location = 0; // offset from beginning of stream

enum utf8_state
{
  cp0,    // code point 0 --> U+0000..U+007F, also known as ASCII
  cp1,    // code point 1 --> U+0080..U+07FF, also known as two-byte UTF8
  cp2,    // code point 2 --> U+0800..U+0FFF, also known as three-byte UTF8
  cp3,    // code point 3 --> U+1000..U+CFFF, also known as three-byte UTF8
  cp4,    // code point 4 --> U+D000..U+D7FF, also known as three-byte UTF8
  cp5,    // code point 5 --> U+E000..U+FFFF, also known as three-byte UTF8
  cp6,    // code point 6 --> U+10000..U+3FFFF, also known as four-byte UTF8
  cp7,    // code point 7 --> U+40000..U+FFFFF, also known as four-byte UTF8
  cp8,    // code point 8 --> U+100000..U+10FFFF, also known as four-byte UTF8
  cp2_1,  // cp2 second byte state
  cp3_1,  // cp3 second byte state
  cp4_1,  // cp4 second byte state
  cp5_1,  // cp5 second byte state
  cp6_1,  // cp6 second byte state
  cp7_1,  // cp7 second byte state
  cp8_1,  // cp8 second byte state
  cp6_2,  // cp6 third byte state
  cp7_2,  // cp7 third byte state
  cp8_2   // cp8 third byte state
};


char *state_name[cp8_2+1];


// returns false if state machine fails; also, the new_state is reset back to cp0
bool fsm_utf8(uint8_t c, int current_state, int *new_state)
{
  static uint8_t byte_0;
  static uint8_t byte_1;
  static uint8_t byte_2;
  static uint8_t byte_3;
  
  switch(current_state)
    {
    case cp0:
      byte_0 = c;
      if(c >= 0x0 && c <= 0x7f)
	{
	  // stay in same cp0 state
	  *new_state = cp0;
	  if(verbose3)
	    write(1,&byte_0,1);
	}
      else if(c >= 0xc2 && c <= 0xdf)
	{
	  // go to cp1 state, need another byte
	  *new_state = cp1;
	}
      else if(c == 0xe0)
	{
	}
      /*
	.... YOUR CODE GOES HERE
       */
    case cp8_2:
      byte_3 = c;
      if(c >= 0x80 && c <= 0xbf)
	{
	  // recognized! back to cp0
	  *new_state = cp0;
	  if(verbose2)
	    {
	      write(1,&byte_0,1);
	      write(1,&byte_1,1);
	      write(1,&byte_2,1);
	      write(1,&byte_3,1);
	    }
	  return(true);
	}

      // error! back to cp0

      if(verbose5)
	{
	  char s[512];
	  sprintf(s,"\nThere are four ill-formed bytes starting at offset %u: %x %x\n",location-3,byte_0,byte_1,byte_2,byte_3);
	  write(1,s,strlen(s));
	}
      
      *new_state = cp0;
      return(false);
      break;
    default:
      printf("This is impossible!\n");
      return(false);
      
    }
}

void do_options(int argc, char *argv[])
{
  int opt;
  while((opt = getopt(argc, argv, "uap")) != -1)
    {
      switch(opt)
	{
	case 'u':
	  verbose2 = true;
	  break;
	case 'a':
	  verbose3 = true;
	  break;
	case 'p':
	  verbose5 = true;
	  break;
	default:
	  fprintf(stderr,"Usage: %s [-u] [-a] [-p]\n",argv[0]);
	  exit(1);
	  break;
	}
    }
}

int main(int argc, char *argv[])
{

  do_options(argc, argv);

  state_name[cp0] = "cp0";
  state_name[cp1] = "cp1";
  state_name[cp2] = "cp2";
  state_name[cp3] = "cp3";
  state_name[cp4] = "cp4";
  state_name[cp5] = "cp5";
  state_name[cp6] = "cp6";
  state_name[cp7] = "cp7";
  state_name[cp8] = "cp8";
  state_name[cp2_1] = "cp2_1";
  state_name[cp3_1] = "cp3_1";
  state_name[cp4_1] = "cp4_1";
  state_name[cp5_1] = "cp5_1";
  state_name[cp6_1] = "cp6_1";
  state_name[cp7_1] = "cp7_1";
  state_name[cp8_1] = "cp8-1";
  state_name[cp6_2] = "cp6_2";
  state_name[cp7_2] = "cp7_2";
  state_name[cp8_2] = "cp8_2";
  
  uint8_t c;
  int current_state = cp0;
  int new_state = cp0;
  while(read(0,&c,1) == 1)
    {
      bool okay = fsm_utf8(c,current_state,&new_state);
      if(verbose1)
	{
	  write(1,&c,1);
	  printf(" : current_state = %s, new_state = %s \n",state_name[current_state],state_name[new_state]);
	  fflush(stdout);
	}
      current_state = new_state;
      location++;
    }
}