cexp.c

编译工具

      if (wide_flag)
	{
	  width = MAX_WCHAR_TYPE_SIZE;
#ifdef MULTIBYTE_CHARS
	  max_chars = MB_CUR_MAX;
#else
	  max_chars = 1;
#endif
	}
      else
	max_chars = MAX_LONG_TYPE_SIZE / width;

      token_buffer = (char *) alloca (max_chars + 1);

      while (1)
	{
	  c = *lexptr++;

	  if (c == '\'' || c == EOF)
	    break;

	  if (c == '\\')
	    {
	      c = parse_escape (&lexptr, mask);
	    }

	  num_chars++;

	  /* Merge character into result; ignore excess chars.  */
	  if (num_chars <= max_chars)
	    {
	      if (width < HOST_BITS_PER_WIDE_INT)
		result = (result << width) | c;
	      else
		result = c;
	      token_buffer[num_chars - 1] = c;
	    }
	}

      token_buffer[num_chars] = 0;

      if (c != '\'')
	error ("malformatted character constant");
      else if (num_chars == 0)
	error ("empty character constant");
      else if (num_chars > max_chars)
	{
	  num_chars = max_chars;
	  error ("character constant too long");
	}
      else if (num_chars != 1 && ! traditional)
	warning ("multi-character character constant");

      /* If char type is signed, sign-extend the constant.  */
      if (! wide_flag)
	{
	  int num_bits = num_chars * width;

	  if (lookup ((U_CHAR *) "__CHAR_UNSIGNED__",
		      sizeof ("__CHAR_UNSIGNED__") - 1, -1)
	      || ((result >> (num_bits - 1)) & 1) == 0)
	    yylval.integer.value
	      = result & (~ (unsigned HOST_WIDE_INT) 0
			  >> (HOST_BITS_PER_WIDE_INT - num_bits));
	  else
	    yylval.integer.value
	      = result | ~(~ (unsigned HOST_WIDE_INT) 0
			   >> (HOST_BITS_PER_WIDE_INT - num_bits));
	}
      else
	{
#ifdef MULTIBYTE_CHARS
	  /* Set the initial shift state and convert the next sequence.  */
	  result = 0;
	  /* In all locales L'\0' is zero and mbtowc will return zero,
	     so don't use it.  */
	  if (num_chars > 1
	      || (num_chars == 1 && token_buffer[0] != '\0'))
	    {
	      wchar_t wc;
	      (void) mbtowc (NULL_PTR, NULL_PTR, 0);
	      if (mbtowc (& wc, token_buffer, num_chars) == num_chars)
		result = wc;
	      else
		pedwarn ("Ignoring invalid multibyte character");
	    }
#endif
	  yylval.integer.value = result;
	}
    }

    /* This is always a signed type.  */
    yylval.integer.signedp = SIGNED;
    
    return CHAR;

    /* some of these chars are invalid in constant expressions;
       maybe do something about them later */
  case '/':
  case '+':
  case '-':
  case '*':
  case '%':
  case '|':
  case '&':
  case '^':
  case '~':
  case '!':
  case '@':
  case '<':
  case '>':
  case '[':
  case ']':
  case '.':
  case '?':
  case ':':
  case '=':
  case '{':
  case '}':
  case ',':
  case '#':
    if (keyword_parsing)
      break;
  case '(':
  case ')':
    lexptr++;
    return c;

  case '"':
    mask = MAX_CHAR_TYPE_MASK;
  string_constant:
    if (keyword_parsing) {
      char *start_ptr = lexptr;
      lexptr++;
      while (1) {
	c = *lexptr++;
	if (c == '\\')
	  c = parse_escape (&lexptr, mask);
	else if (c == '"')
	  break;
      }
      yylval.name.address = tokstart;
      yylval.name.length = lexptr - start_ptr;
      return NAME;
    }
    yyerror ("string constants not allowed in #if expressions");
    return ERROR;
  }

  if (c >= '0' && c <= '9' && !keyword_parsing) {
    /* It's a number */
    for (namelen = 1; ; namelen++) {
      int d = tokstart[namelen];
      if (! ((is_idchar[d] || d == '.')
	     || ((d == '-' || d == '+')
		 && (c == 'e' || c == 'E'
		     || ((c == 'p' || c == 'P') && ! c89))
		 && ! traditional)))
	break;
      c = d;
    }
    return parse_number (namelen);
  }

  /* It is a name.  See how long it is.  */

  if (keyword_parsing) {
    for (namelen = 0;; namelen++) {
      if (is_space[tokstart[namelen]])
	break;
      if (tokstart[namelen] == '(' || tokstart[namelen] == ')')
	break;
      if (tokstart[namelen] == '"' || tokstart[namelen] == '\'')
	break;
    }
  } else {
    if (!is_idstart[c]) {
      yyerror ("Invalid token in expression");
      return ERROR;
    }

    for (namelen = 0; is_idchar[tokstart[namelen]]; namelen++)
      ;
  }
  
  lexptr += namelen;
  yylval.name.address = tokstart;
  yylval.name.length = namelen;
  return NAME;
}


/* Parse a C escape sequence.  STRING_PTR points to a variable
   containing a pointer to the string to parse.  That pointer
   is updated past the characters we use.  The value of the
   escape sequence is returned.

   RESULT_MASK is used to mask out the result;
   an error is reported if bits are lost thereby.

   A negative value means the sequence \ newline was seen,
   which is supposed to be equivalent to nothing at all.

   If \ is followed by a null character, we return a negative
   value and leave the string pointer pointing at the null character.

   If \ is followed by 000, we return 0 and leave the string pointer
   after the zeros.  A value of 0 does not mean end of string.  */

HOST_WIDE_INT
parse_escape (string_ptr, result_mask)
     char **string_ptr;
     HOST_WIDE_INT result_mask;
{
  register int c = *(*string_ptr)++;
  switch (c)
    {
    case 'a':
      return TARGET_BELL;
    case 'b':
      return TARGET_BS;
    case 'e':
    case 'E':
      if (pedantic)
	pedwarn ("non-ANSI-standard escape sequence, `\\%c'", c);
      return 033;
    case 'f':
      return TARGET_FF;
    case 'n':
      return TARGET_NEWLINE;
    case 'r':
      return TARGET_CR;
    case 't':
      return TARGET_TAB;
    case 'v':
      return TARGET_VT;
    case '\n':
      return -2;
    case 0:
      (*string_ptr)--;
      return 0;
      
    case '0':
    case '1':
    case '2':
    case '3':
    case '4':
    case '5':
    case '6':
    case '7':
      {
	register HOST_WIDE_INT i = c - '0';
	register int count = 0;
	while (++count < 3)
	  {
	    c = *(*string_ptr)++;
	    if (c >= '0' && c <= '7')
	      i = (i << 3) + c - '0';
	    else
	      {
		(*string_ptr)--;
		break;
	      }
	  }
	if (i != (i & result_mask))
	  {
	    i &= result_mask;
	    pedwarn ("octal escape sequence out of range");
	  }
	return i;
      }
    case 'x':
      {
	register unsigned HOST_WIDE_INT i = 0, overflow = 0;
	register int digits_found = 0, digit;
	for (;;)
	  {
	    c = *(*string_ptr)++;
	    if (c >= '0' && c <= '9')
	      digit = c - '0';
	    else if (c >= 'a' && c <= 'f')
	      digit = c - 'a' + 10;
	    else if (c >= 'A' && c <= 'F')
	      digit = c - 'A' + 10;
	    else
	      {
		(*string_ptr)--;
		break;
	      }
	    overflow |= i ^ (i << 4 >> 4);
	    i = (i << 4) + digit;
	    digits_found = 1;
	  }
	if (!digits_found)
	  yyerror ("\\x used with no following hex digits");
	if (overflow | (i != (i & result_mask)))
	  {
	    i &= result_mask;
	    pedwarn ("hex escape sequence out of range");
	  }
	return i;
      }
    default:
      return c;
    }
}

static void
yyerror (s)
     char *s;
{
  error ("%s", s);
  skip_evaluation = 0;
  longjmp (parse_return_error, 1);
}

static void
integer_overflow ()
{
  if (!skip_evaluation && pedantic)
    pedwarn ("integer overflow in preprocessor expression");
}

static HOST_WIDE_INT
left_shift (a, b)
     struct constant *a;
     unsigned HOST_WIDE_INT b;
{
   /* It's unclear from the C standard whether shifts can overflow.
      The following code ignores overflow; perhaps a C standard
      interpretation ruling is needed.  */
  if (b >= HOST_BITS_PER_WIDE_INT)
    return 0;
  else
    return (unsigned HOST_WIDE_INT) a->value << b;
}

static HOST_WIDE_INT
right_shift (a, b)
     struct constant *a;
     unsigned HOST_WIDE_INT b;
{
  if (b >= HOST_BITS_PER_WIDE_INT)
    return a->signedp ? a->value >> (HOST_BITS_PER_WIDE_INT - 1) : 0;
  else if (a->signedp)
    return a->value >> b;
  else
    return (unsigned HOST_WIDE_INT) a->value >> b;
}

/* This page contains the entry point to this file.  */

/* Parse STRING as an expression, and complain if this fails
   to use up all of the contents of STRING.  */
/* STRING may contain '\0' bytes; it is terminated by the first '\n'
   outside a string constant, so that we can diagnose '\0' properly.  */
/* We do not support C comments.  They should be removed before
   this function is called.  */

HOST_WIDE_INT
parse_c_expression (string)
     char *string;
{
  lexptr = string;

  /* if there is some sort of scanning error, just return 0 and assume
     the parsing routine has printed an error message somewhere.
     there is surely a better thing to do than this.     */
  if (setjmp (parse_return_error))
    return 0;

  if (yyparse () != 0)
    abort ();

  if (*lexptr != '\n')
    error ("Junk after end of expression.");

  return expression_value;	/* set by yyparse () */
}

#ifdef TEST_EXP_READER

#if YYDEBUG
extern int yydebug;
#endif

int pedantic;
int traditional;

int main PROTO((int, char **));
static void initialize_random_junk PROTO((void));

/* Main program for testing purposes.  */
int
main (argc, argv)
     int argc;
     char **argv;
{
  int n, c;
  char buf[1024];

  pedantic = 1 < argc;
  traditional = 2 < argc;
#if YYDEBUG
  yydebug = 3 < argc;
#endif
  initialize_random_junk ();

  for (;;) {
    printf ("enter expression: ");
    n = 0;
    while ((buf[n] = c = getchar ()) != '\n' && c != EOF)
      n++;
    if (c == EOF)
      break;
    printf ("parser returned %ld\n", (long) parse_c_expression (buf));
  }

  return 0;
}

/* table to tell if char can be part of a C identifier. */
unsigned char is_idchar[256];
/* table to tell if char can be first char of a c identifier. */
unsigned char is_idstart[256];
/* table to tell if c is horizontal or vertical space.  */
unsigned char is_space[256];

/*
 * initialize random junk in the hash table and maybe other places
 */
static void
initialize_random_junk ()
{
  register int i;

  /*
   * Set up is_idchar and is_idstart tables.  These should be
   * faster than saying (is_alpha (c) || c == '_'), etc.
   * Must do set up these things before calling any routines tthat
   * refer to them.
   */
  for (i = 'a'; i <= 'z'; i++) {
    ++is_idchar[i - 'a' + 'A'];
    ++is_idchar[i];
    ++is_idstart[i - 'a' + 'A'];
    ++is_idstart[i];
  }
  for (i = '0'; i <= '9'; i++)
    ++is_idchar[i];
  ++is_idchar['_'];
  ++is_idstart['_'];
  ++is_idchar['$'];
  ++is_idstart['$'];

  ++is_space[' '];
  ++is_space['\t'];
  ++is_space['\v'];
  ++is_space['\f'];
  ++is_space['\n'];
  ++is_space['\r'];
}

void
error (PRINTF_ALIST (msg))
     PRINTF_DCL (msg)
{
  va_list args;

  VA_START (args, msg);
  fprintf (stderr, "error: ");
  vfprintf (stderr, msg, args);
  fprintf (stderr, "\n");
  va_end (args);
}

void
pedwarn (PRINTF_ALIST (msg))
     PRINTF_DCL (msg)
{
  va_list args;

  VA_START (args, msg);
  fprintf (stderr, "pedwarn: ");
  vfprintf (stderr, msg, args);
  fprintf (stderr, "\n");
  va_end (args);
}

void
warning (PRINTF_ALIST (msg))
     PRINTF_DCL (msg)
{
  va_list args;

  VA_START (args, msg);
  fprintf (stderr, "warning: ");
  vfprintf (stderr, msg, args);
  fprintf (stderr, "\n");
  va_end (args);
}

int
check_assertion (name, sym_length, tokens_specified, tokens)
     U_CHAR *name;
     int sym_length;
     int tokens_specified;
     struct arglist *tokens;
{
  return 0;
}

struct hashnode *
lookup (name, len, hash)
     U_CHAR *name;
     int len;
     int hash;
{
  return (DEFAULT_SIGNED_CHAR) ? 0 : ((struct hashnode *) -1);
}

GENERIC_PTR
xmalloc (size)
     size_t size;
{
  return (GENERIC_PTR) malloc (size);
}
#endif