numeric comparison

[hotshot]

hello guys!

How can I check that a variable I have is numeric before comparison in order not to get the 'Argument "f" isn't numeric in numeric lt (<) at ..' warning, as in:

my $foo = 'xxx';

if ($foo < 4.556) {   # will throw the warning
  # do something
}

Trying to match to a regexp /^\d$ is not enough when dealing with fractions (because of th period).

Thanks

Hotshot

[gjb]

Have a look at Regexp::Common, it has patterns for real numbers (and many other common things).

Just my 2 cents, -gjb-

[Abigail-II]

If it's ok if 'xxx' compares as 0, you could also turn the appropriate warning off (locally). Something like:

if (do {no warnings 'numeric'; $foo < 4.556}) {
    ...
}

Abigail

[dbp]

This topic is covered nicely in the Perl Cookbook. Here's the applicable code:

warn "has nondigits"        if     /\D/;
warn "not a natural number" unless /^\d+$/;             # rejects -3
warn "not an integer"       unless /^-?\d+$/;           # rejects +3
warn "not an integer"       unless /^[+-]?\d+$/;
warn "not a decimal number" unless /^-?\d+\.?\d*$/;     # rejects .2
warn "not a decimal number" unless /^-?(?:\d+(?:\.\d*)?|\.\d+)$/;
warn "not a C float"
       unless /^([+-]?)(?=\d|\.\d)\d*(\.\d*)?([Ee]([+-]?\d+))?$/;

[BrowserUk]

This is the function I used to use for this.

sub is_numeric{ use POSIX (); !(POSIX::strtod($_[0]))[1]; }

However, if you download and build the latest cpan version of Scalar::Util (the version that comes with 5.8.0 doesn't have it), it provides access to perl's internal looks_like_number(); which ought to be quicker than either strtod() or a regex.

NOTE: If your not set up to build XS modules, then List::Util and Scalar::Util will silently fall back to using Perl implementations of the functions they provide. In the case of Looks_like_number(), this is a regex.

---

Examine what is said, not who speaks.
"Efficiency is intelligent laziness." -David Dunham
"When I'm working on a problem, I never think about beauty. I think only how to solve the problem. But when I have finished, if the solution is not beautiful, I know it is wrong." -Richard Buckminster Fuller

[Zaxo]

Perl 5.8.1's stock Scalar::Util has looks_like_number(). Here's a little workout to show what it can and can't do,

use Scalar::Util 'looks_like_number';
my @candidates = qw/ 1 1.1 1,1 0E0 0e0 0377 100_000 0xdeadbeef/;

print $_,
      looks_like_number($_)? ' is' : ' is not',
      ' a number.',
      $/
    for @candidates;

__END__
1 is a number.
1.1 is a number.
1,1 is not a number.
0E0 is a number.
0e0 is a number.
0377 is a number.
100_000 is not a number.
0xdeadbeef not is a number.

After Compline,
Zaxo

[BrowserUk]

I was a bit surprised by 100_000 not being recognised as a number but then I tried this

print 3 + '0xdeadbeef';
Argument "0xdeadbeef" isn't numeric in addition (+) at ...
3

print 3 + 0xdeadbeef;
3735928562

print 3+ '100_000';
Argument "100_000" isn't numeric in addition (+) at ...
103

print 3+ 100_000;
100003

Which I guess means that the code used to extract a number from a string at runtime is a different routine to that used for parsing numeric constants at compile time. Which is slightly disappointing, but is probably done that way for very good reasons.

I just wish that the version of Scalar-List-Utils on the AS 5.8 list included the compiled XS code. That is that possible isn't it?

---

Examine what is said, not who speaks.
"Efficiency is intelligent laziness." -David Dunham
"When I'm working on a problem, I never think about beauty. I think only how to solve the problem. But when I have finished, if the solution is not beautiful, I know it is wrong." -Richard Buckminster Fuller

[derby]

dbp gave you the faq supported answer and it can be found in perlfaq4 under "How do I determine whether a scalar is a number/whole/integer/float?". Then check the Floating-point Arithmetic section of perlop as to why your comparison of a float is going to lead you down a subtle path of bugs.

-derby

[Tanalis]

Why not use a regexp? Something like ...

unless ($foo =~ /^\-?\d*\.?\d*$/) {
   # do something
}

should do the trick.

Hope that helps.

-- Foxcub
#include www.liquidfusion.org.uk

Update: Doh, updated regexp to something that stands a chance in hell of actually working .. (thanks, antirice)

[antirice]

So "..." and "-124.-8.5-" are valid numbers?

antirice    
The first rule of Perl club is - use Perl
The ith rule of Perl club is - follow rule i - 1 for i > 1

[antirice]

I don't know if you want it to work with octal or hex but this will work for decimal.

$word =~ /^-?(\d*)\.?(\d*)$/ && (length($1) || length($2));

What's with the length checks? ".807" and "30." are valid numbers whereas "." is not. Hope this helps.

antirice    
The first rule of Perl club is - use Perl
The ith rule of Perl club is - follow rule i - 1 for i > 1

[edan]

I recently wrote a regex to match a floating-point number, including (optional) scientific notatation. This looks like a good chance to get some critique, as well as (possibly) help you. I think it's a bit more complete than the regexen I've seen so far in this post, though I'm sure it's far from perfect.

/^[+-]?(?=\.?\d)\d*(\.\d*)?([Ee][+-]?\d+)?$/

HTH

Update: Well, at least my regex is validated by the Perl Cookbook, as shown in dbp's post. Though I'm curious to know why they used (?=\d|\.\d) instead of (?=\.?\d), when it seems to me the latter is more concise... anyone know of a good reason?

--
3dan