Function IsNumeric

source/util.cpp:317–419

SymbolType IsNumeric(LPCTSTR aBuf, BOOL aAllowNegative, BOOL aAllowAllWhitespace
	, BOOL aAllowFloat, BOOL aAllowImpure)  // BOOL vs. bool might squeeze a little more performance out of this frequently-called function.
// String can contain whitespace.
// If aBuf doesn't contain something purely numeric, PURE_NOT_NUMERIC is returned.  The same happens if
// aBuf contains a float but aAllowFloat is false.  (The aAllowFloat parameter isn't strictly necessary
// because the caller could just check whether the return value is/isn't PURE_FLOAT to get the same effect.
// However, supporting aAllowFloat seems to greatly improve maintainability because it saves many callers
// from having to compare the return value to PURE_INTEGER [they can just interpret the return value as BOOL].
// It also improves readability due to the "Is" part of the function name.  So it seems worth keeping.)
// Otherwise, PURE_INTEGER or PURE_FLOAT is returned.
// If aAllowAllWhitespace==true and the string is blank or all whitespace, PURE_INTEGER is returned.
// Obsolete comment: Making this non-inline reduces the size of the compressed EXE by only 2K.  Since this
// function is called so often, it seems preferable to keep it inline for performance.
{
	aBuf = omit_leading_whitespace(aBuf); // i.e. caller doesn't have to have ltrimmed, only rtrimmed.
	if (!*aBuf) // The string is empty or consists entirely of whitespace.
		return aAllowAllWhitespace ? PURE_INTEGER : PURE_NOT_NUMERIC;

	if (*aBuf == '-')
	{
		if (aAllowNegative)
			++aBuf;
		else
			return PURE_NOT_NUMERIC;
	}
	else if (*aBuf == '+')
		++aBuf;

	// Relies on short circuit boolean order to prevent reading beyond the end of the string:
	BOOL is_hex = IS_HEX(aBuf); // BOOL vs. bool might squeeze a little more performance out this frequently-called function.
	if (is_hex)
		aBuf += 2;  // Skip over the 0x prefix.

	// Set defaults:
	BOOL has_decimal_point = false;
	BOOL has_exponent = false;
	BOOL has_at_least_one_digit = false; // i.e. a string consisting of only "+", "-" or "." is not considered numeric.
	int c; // int vs. char might squeeze a little more performance out of it (it does reduce code size by 5 bytes). Probably must stay signed vs. unsigned for some of the uses below.

	for (;; ++aBuf)
	{
		c = *aBuf;
		if (IS_SPACE_OR_TAB(c))
		{
			if (*omit_leading_whitespace(aBuf)) // But that space or tab is followed by something other than whitespace.
				if (!aAllowImpure) // e.g. "123 456" is not a valid pure number.
					return PURE_NOT_NUMERIC;
				// else fall through to the bottom logic.
			// else since just whitespace at the end, the number qualifies as pure, so fall through to the bottom
			// logic (it would already have returned in the loop if it was impure)
			break;
		}
		if (!c) // End of string was encountered.
			break; // The number qualifies as pure, so fall through to the logic at the bottom. (It would already have returned elsewhere in the loop if the number is impure).
		if (c == '.')
		{
			if (!aAllowFloat || has_decimal_point || is_hex) // If aAllowFloat==false, a decimal point at the very end of the number is considered non-numeric even if aAllowImpure==true.  Some callers might rely on this.
				// i.e. if aBuf contains 2 decimal points, it can't be a valid number.