token.c

MicroChip的12BIt和14Bit的PIC系列单片机的汇编器

    tp = 0;
    while(tp < TOKSIZE-1 && tok_char != '"' && tok_char != EOF)	{
      token_string[tp++] = tok_char;
      read_src_char();
    }
    if(tok_char != '\"' && !ifskip_mode)
      error(0, "String not terminated");
    token_string[tp] = '\0';
    read_src_char();
    token_type = TOK_STRCONST;
    return;
  }

/*
 * integer number
 */
  if(isdigit(tok_char)) {
    token_type = TOK_INTCONST;
    token_string[0] = tok_char;
    tp = 1;
    read_src_char();
    if(token_string[0] == '0') {
      if(tok_char == 'x' || tok_char == 'X') { /* hex number */
	token_string[tp++] = tok_char;
	read_src_char();
	while(tp < TOKSIZE-1 && isxdigit(tok_char)) {
	  token_string[tp++] = tok_char;
	  read_src_char();
	}
	token_string[tp] = '\0';
	token_int_val = strtoul(&token_string[2], NULL, 16);
	/* should put range check here */
	return;
      }
    }

    while(tp < TOKSIZE-2 && isxdigit(tok_char))	{
      token_string[tp++] = tok_char;
      read_src_char();
    }

    base = 10;
    switch(tok_char) {
      case 'H': /* hex */
      case 'h':
        base = 16; /* hex */
	token_string[tp++] = tok_char;
	read_src_char();
	break;

      case 'O': /* octal */
      case 'o':
	base = 8; /* octal */
	token_string[tp++] = tok_char;
	read_src_char();
	break;

      default:
	if(token_string[0] == '0' &&
	   (token_string[1] == 'b' || token_string[1] == 'B')) {
	  token_string[tp] = '\0';
	  token_int_val = strtoul(&token_string[2], &cp, 2);
	  if(cp != &token_string[tp] && !ifskip_mode)
	    error(0, "Invalid digit in a number");
	  /* should put range check here */
	  return;
	} else if(token_string[tp-1] == 'B' || token_string[tp-1] == 'b') {
	  base = 2;
	} else {
	  if(token_string[tp-1] != 'D' && token_string[tp-1] != 'd')
	    token_string[tp++] = '\0';
	}
	break;
    }

    token_string[tp] = '\0';
    token_int_val = strtoul(token_string, &cp, base);
    if(cp != &token_string[tp-1] && !ifskip_mode)
      error(0, "Invalid digit in a number");
    /* should put range check here */
    return;
  }

/*
 * Handle B'10010100' binary etc.
 */
  if((tok_char == 'b' || tok_char == 'B' ||
      tok_char == 'd' || tok_char == 'D' ||
      tok_char == 'h' || tok_char == 'H' ||
      tok_char == 'o' || tok_char == 'O') &&
     line_buf_ptr != NULL && *line_buf_ptr == '\'') {
    token_string[0] = tok_char;
    read_src_char();
    token_string[1] = tok_char;
    read_src_char();
    tp = 2;
    while(tp < TOKSIZE-1 && isxdigit(tok_char))	{
      token_string[tp++] = tok_char;
      read_src_char();
    }
    if(tok_char != '\'')
      goto invalid_token;
    token_string[tp++] = tok_char;
    read_src_char();
    token_string[tp] = '\0';

    switch(token_string[0]) {
      case 'b':
      case 'B':
        base = 2;
	break;

      case 'o':
      case 'O':
	base = 8;
	break;

      case 'h':
      case 'H':
	base = 16;
	break;

      case 'd':
      case 'D':
      default:
	base = 10;
	break;
    }

    token_int_val = strtoul(&token_string[2], &cp, base);
    if(cp != &token_string[tp-1] && !ifskip_mode)
      error(0, "Invalid digit in a number");
    /* should put range check here */
    token_type = TOK_INTCONST;
    return;
  }

/*
 * keyword or identifier
 */
  if(tok_char == '_' || tok_char == '.' || isalpha(tok_char)) {
    line_buf_off = (line_buf_ptr - &line_buffer[1]);

    token_string[0] = tok_char;
    tp = 1;
    read_src_char();

    if(token_string[0] == '.' &&
       tok_char != '_' && !isalnum(tok_char)) {
      token_string[1] = '\0';
      token_type = TOK_PERIOD;
      return;
    }

    while(tp < TOKSIZE-1 &&
	  (tok_char == '_' || tok_char == '.' || isalnum(tok_char))) {
      token_string[tp++] = tok_char;
      read_src_char();
    }
    token_string[tp] = '\0';

    token_type = FIRST_KW;
    cp = Keyword_Table;
    while(*cp) {
      if(strcasecmp(token_string, cp) == 0)
	return;
      while(*cp++)
	;
      token_type++;
    }
    token_type = TOK_IDENTIFIER;
    return;
  }

/*
 * non-numeric & non-alpha tokens
 */
  switch(tok_char) {
    case '\n':
    case '\0':
      token_type = TOK_NEWLINE;
      strcpy(token_string, "\\n");
      skip_eol();
      return;

    case '<':
      token_string[0] = tok_char;
      read_src_char();
      if(tok_char == '<') {
	token_string[1] = tok_char;
	token_string[2] = '\0';
	token_type = TOK_LSHIFT;
	read_src_char();
	return;
      }

      if(tok_char == '=') {
	token_string[1] = tok_char;
	token_string[2] = '\0';
	token_type = TOK_LESS_EQ;
	read_src_char();
	return;
      }

      if(tok_char == '>') {
	token_string[1] = tok_char;
	token_string[2] = '\0';
	token_type = TOK_NOT_EQ;
	read_src_char();
	return;
      }

      token_type = TOK_LESS;
      token_string[1] = '\0';
      return;

    case '>':
      token_string[0] = tok_char;
      read_src_char();
      if(tok_char == '>') {
	token_string[1] = tok_char;
	token_string[2] = '\0';
	token_type = TOK_RSHIFT;
	read_src_char();
	return;
      }

      if(tok_char == '=') {
	token_string[1] = tok_char;
	token_string[2] = '\0';
	token_type = TOK_GT_EQ;
	read_src_char();
	return;
      }

      token_string[1] = '\0';
      token_type = TOK_GREATER;
      return;

    case '!':
      token_string[0] = tok_char;
      read_src_char();
      if(tok_char != '=')
	goto invalid_token;
      token_string[1] = tok_char;
      token_string[2] = '\0';
      read_src_char();
      token_type = TOK_NOT_EQ;
      return;

    case '=':
      token_string[0] = tok_char;
      read_src_char();
      if(tok_char == '=') {
	token_string[1] = tok_char;
	read_src_char();
	token_string[2] = '\0';
	token_type = TOK_EQ;
	return;
      }

      if(tok_char == '<') {
	token_string[1] = tok_char;
	read_src_char();
	token_string[2] = '\0';
	token_type = TOK_LESS_EQ;
	return;
      }

      if(tok_char == '>') {
	token_string[1] = tok_char;
	read_src_char();
	token_string[2] = '\0';
	token_type = TOK_GT_EQ;
	return;
      }

      if(tok_char == '_' || tok_char == '.' ||
	 isalpha(tok_char)) { /* local symbol */
	line_buf_off = (line_buf_ptr - &line_buffer[2]);

	token_string[0] = tok_char;
	tp = 1;
	read_src_char();

	while(tp < TOKSIZE-1 &&
	      (tok_char == '_' || tok_char == '.' || isalnum(tok_char))) {
	  token_string[tp++] = tok_char;
	  read_src_char();
	}
	token_string[tp] = '\0';

	token_type = TOK_LOCAL_ID;
	return;
      }

      token_string[1] = '\0';
      token_type = TOK_EQUAL;
      return;

    case '$':
      read_src_char();
      if(!isxdigit(tok_char))
	{
	  token_string[0] = '$';
	  token_string[1] = '\0';
	  token_type = TOK_DOLLAR;
	  return;
	}

      tp = 0;
      do
	{
	  token_string[tp++] = tok_char;
	  read_src_char();
	} while(tp < TOKSIZE-1 && isxdigit(tok_char));

      token_string[tp] = '\0';
      token_int_val = strtoul(&token_string[1], NULL, 16);
      token_type = TOK_INTCONST;
      /* should put range check here */
      return;

    case '\\':
      token_type = TOK_BACKSLASH;
      break;

    case ',':
      token_type = TOK_COMMA;
      break;

    case '(':
      token_type = TOK_LEFTPAR;
      break;

    case ')':
      token_type = TOK_RIGHTPAR;
      break;

    case '+':
      token_type = TOK_PLUS;
      break;

    case '-':
      token_type = TOK_MINUS;
      break;

    case '&':
      token_type = TOK_BITAND;
      break;

    case '|':
      token_type = TOK_BITOR;
      break;

    case '^':
      token_type = TOK_BITXOR;
      break;

    case '~':
      token_type = TOK_BITNOT;
      break;

    case '*':
      token_type = TOK_ASTERISK;
      break;

    case '/':
      token_type = TOK_SLASH;
      break;

    case '%':
      token_type = TOK_PERCENT;
      break;

    case ':':
      token_type = TOK_COLON;
      break;

    case '[':
      token_type = TOK_LEFTBRAK;
      break;

    case ']':
      token_type = TOK_RIGHTBRAK;
      break;

    default:
      goto invalid_token;
  }

  token_string[0] = tok_char;
  token_string[1] = '\0';
  read_src_char();
  return;

invalid_token:
  if(!ifskip_mode)
    error(0, "Invalid token");
  token_string[0] = '\0';
  token_type = TOK_INVALID;
}

/* skip to the next line */
void
skip_eol(void)
{
  line_buf_ptr = NULL;
  tok_char = ' ';
}