treetop grammar problem

[Robert Schaaf]

I have devised this grammar for roman numerals for treetop.

grammar RomanNumeral

rule roman_numeral
('MMM' / 'MM' / 'M')?
(('C' [DM]) / ('D'? ('CCC' / 'CC' / 'C')?))?
(('X' [LC]) / ('L'? ('XXX' / 'XX' / 'X')?))?
(('I' [VX]) / ('V'? ('III' / 'II' / 'I')?))?
end

end

It recognizes numbers in the range 1-3999, and rejects all bad input
as far as I can tell. T

But the empty string passes!

The problem is that any part is optional, but not ALL parts. At a
minimum, there must be one digit. Duh.

Treetop is ingenious, and easy to use, except when it's hard.

Does anyone have a suggestion? I'm new at this, so I beg your patience.

Also, is there a better way of translating "M{0,3}" than "('MMM' /
'MM' / 'M')?" ?

Bob Schaaf

 

[Clifford Heath]

But the empty string passes!

It's not a general solution, but you can require that *something*
follows, using &.

rule roman_numeral
&. ( stuff )
end

There's no better way to write M{0,3}

Clifford Heath.

 

[Robert Schaaf]

Of course! Lookahead non-gobbling. A lesson learned.

Thanks mightily,

Bob Schaaf