I'm working on an RPN calculator (don't ask why...) and I'm having
trouble getting a number from gets - how am I supposed to know whether
it's a number?

At the moment I'm doing [eval(oper) == oper.to_f], which is far from
ideal. Any ideas? You all get to win my calculator if you know of a
better way .
--
Posted via http://www.ruby-forum.com/.

Alle gioved=C3=AC 22 novembre 2007, Peter Bunyan ha scritto:
> I'm working on an RPN calculator (don't ask why...) and I'm having
> trouble getting a number from gets - how am I supposed to know whether
> it's a number?
>
> At the moment I'm doing [eval(oper) =3D=3D oper.to_f], which is far from
> ideal. Any ideas? You all get to win my calculator if you know of a
> better way .

Do you mean you need to find out whether a string contains a number or not?=
In=20
this case, you can either use Kernel#Float, which returns str.to_f if the=20
string contains a number and raises an exception owtherwise, or try to craf=
t=20
a regexp which only matches strings containing numbers. For example, the=20
(untested) regexp

/[+-]?\d+(.\d+)?(e[+-]?\d+)?/

should match numbers with an optional + or - in front, followed by at least=
=20
one digit, with an optional decimal part (if there's a dot, there should be=
=20
at least one digit following it; if you want to allow something like 11. th=
en=20
replace (.\d+) with (.\d*) ). The number can also be in exponential form,=20
with a downcase e and an optional + or - in front of the exponent.

I hope this s

Stefano

Is that the only way to do it? There's no built-in method for telling if
a string is a number?

--
Posted via http://www.ruby-forum.com/.

Alle gioved=C3=AC 22 novembre 2007, Peter Bunyan ha scritto:
> Is that the only way to do it? There's no built-in method for telling if
> a string is a number?

As far as I know, no.

Stefano

2007/11/22, Peter Bunyan <peter.bunyan@gmail.com>:
> I'm working on an RPN calculator (don't ask why...) and I'm having
> trouble getting a number from gets - how am I supposed to know whether
> it's a number?
>
> At the moment I'm doing [eval(oper) == oper.to_f], which is far from
> ideal. Any ideas? You all get to win my calculator if you know of a
> better way .

str = gets # chomp is not needed for Float
num = Float(str) rescue nil

if num
puts "I'm a number: #{num}"
end

Cheers

robert

--
use.inject do |as, often| as.you_can - without end

OK, this seems to work:

class String
def is_number?
Float(self) == self.to_f
rescue false
end
end
--
Posted via http://www.ruby-forum.com/.

On Nov 22, 2007, at 10:05 , Peter Bunyan wrote:

> Is that the only way to do it? There's no built-in method for
> telling if
> a string is a number?

Strings are never numbers in ruby. Ever. That is the beauty of working
in a clean typesafe language.

What you're doing, possibly without realizing it, is writing a parser
for RPN but it sounds like you've not addressed this project like a
parser. I suggest you go look at the screencast for treetop (a parser
generator). In this screencast they build up an infix parser, which is
much more than you need for RPN, but the tool is clean, powerful, and
worth learning.

http://www.pivotalblabs.com/files/tr...ic-example.mov

On 22.11.2007 20:11, Peter Bunyan wrote:
> OK, this seems to work:
>
> class String
> def is_number?
> Float(self) == self.to_f
> rescue false
> end
> end

That's an error prone approach because it relies on comparing floats
(which is also superfluous here because the real work is done by Float
and rescue). Why don't you just do

def float?
Float(self) rescue nil
end

Cheers

robert

Stefano Crocco wrote:
> Alle giovedì 22 novembre 2007...
> In this case, you can either use Kernel#Float, which returns str.to_f
> .. or try to craft a regexp which only matches strings containing
> numbers.
> For example, the (untested) regexp
>
> /[+-]?\d+(.\d+)?(e[+-]?\d+)?/

Two small additions:
a) the 'dot' should be literally a dot (else it matches any character)
b) the boundaries to the whole expresssion are needed (else "ab23.35de"
would be
validated).

So, the regexp should be something like:

/^ [+-]? \d+ (?: [.]\d+)? (?: e[+-]? \d+)? $ /x

(If capture of components is desired, parenthesis should enclose all
components, and the ?: removed).

Raul
--
Posted via http://www.ruby-forum.com/.

I do not think you can unless you can assume that it is base 10 or less.
--
Posted via http://www.ruby-forum.com/.