Update: (Tue Jan 8 16:03:52 UTC 2002) Removed spurious left paren (thanks, Petruchio).

#!/usr/bin/perl
     # check for numbers with punctuation

     use Strict;

     # examples for "common-written" numbers 
     foreach my $var (

             # should be false
             '','foo 30','.20','20.',',20','20,',    

             # should be numbers
         0,100,'100.23','1,100.23' 
         
                      ) {

             if (IsANumber($var)) {
                     print "True:    $var\n";
             } else {
                     print "False:   $var\n";
             }

     }

             
     sub IsANumber {

             my $var = $_[0];
             
             if (

               # contains digits,commas and 1 period
               ($var =~ /(^[0-9]{1,}?$)|(\,*?)|(\.{1})/) &&

               # does not contain alpha's, more than 1 period
               # commas or periods at the beggining and ends of
               # each line
               !($var =~ /([a-zA-Z])|(^[\.\,]|[\.\,]$)/) &&

               # is not null
               ($var ne '')

               ) { return(1) } 
               else { return(0) }
     }
[download]

USe an eval, it's correct (by definition) and avoids possible errors

my @numbers = (       '1',     '2',       '10',
                    '1.2',   '1e2',   '1.02e2',
                '1.02e-2',   'e-2',       'er',
                     'et',   '0er'              );

foreach my $number (@numbers) {
    my $test = $number;
    eval { $test +=0; };
    if ( $@ ) {
        print "$number isn't a number \n";
    }
}
[download]

{QA Editors note: This solution doesn't work as described. See Re: Answer: How to identify a number datatype in a string? for more explanation.}

Edited by davido: Repaired <code> tags, reformatted code. Added editorial note to clarify why this example is broken.

--

Anthony Staines

Uh, this doesn't seem to work at all...

laptop:~> perl -T
my @numbers = ('1','2','10','1.2','1e2','1.02e2',
               '1.02e-2','e-2','er','et','0er',
               'total and complete non number stuff');
foreach my $number (@numbers) {
   my $test = $number;
   eval { $test +=0; }; 
   if ( $@ ) { print "$number isn't a number \n"; }
}
laptop:~>
[download]

Here are my particulars...
Read more... (4 kB)

Add the following after if(@){}, and you will see...

  else { print "number: ", $test , "\n"; }
[download]

...Apparently, $test is assigned zero (perl 5.8.0) when $number is not a number. Ergo, no errors in eval, no print.

#!/usr/bin/perl
     # check for numbers with punctuation

     use strict;

     # examples for "common-written" numbers 
     foreach my $var (

  '','foo 30','.20','20.',',20','20,','1.3E2', '2..3','++3','2.3.',
 ,'..3',  0,100,'100.23','1,100.23','-3E-2'
         
                      ) {
             print "String: $var\n";
             if (IsANumber($var)) {
                     print "vroom says True, ";
             } else {
                     print "vroom says False, ";
             }
             if (IsANumberCookbook($var)) {
                     print "Cookbook says True, ";
             } else {
                     print "Cookbook says False, ";
             }
             if (IsANumberMine($var)) {
                     my $out = $var * $var;
                     print "I say True and square is $out\n\n";
             } else {
                     print "I say False\n\n";
             }

     }

             
     sub IsANumber {

# vroom version

             my $var = $_[0];
             
             if (

               # contains digits,commas and 1 period
               ($var =~ /(^[0-9]{1,}?$)|(\,*?)|(\.{1})/) &&

               # does not contain alpha's, more than 1 period
               # commas or periods at the beggining and ends of
               # each line
               !($var =~ /([a-zA-Z])|(^[\.\,]|[\.\,]$)/) &&

               # is not null
               ($var ne '')

               ) { return(1) } 
               else { return(0) }
     }


     sub IsANumberCookbook {

# Perl Cookbook version

             my $var = $_[0];
             
             if (

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

               ) { return(1) } 
               else { return(0) }
     }

     sub IsANumberMine {

# my version

             my $var = $_[0];
             
             if (

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

               ) { return(1) } 
               else { return(0) }
     }
[download]

You can use looks_like_number() in conjunction with int() to tell what type of number you have.

#!/usr/bin/perl

use strict;
use warnings;
use Scalar::Util qw(looks_like_number);

my $x = $ARGV[0];

if (looks_like_number($x)) {
    if (int($x) == $x) {
        print("$x is an integer\n");
    } else {
        print("$x is numeric\n");
    }
} else {
    print("Your input was not numeric\n");
}
[download]

if (($m == 0) && !($m eq '0')) {print "string\n";} else {print "number
+\n";}
[download]

It works since PERL treats strings as 0 in numeric tests. (It also pulls numbers out of strings if the numbers come first)

Edited by davido: Added code tags. Reformatted.

"1.233e+23",
"1.233"
[download]

"1.2e+233"
[download]

Appears to be some sort of problem with multiple digits after a decimal point.

The regex is:

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

A slightly modified version which does work for the given cases:

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

The only difference is the extra plus's after each \.

Edited by davido: Added <code> tags.

The mehod suggested by astaines works but the script he gives does not because perl only warns about adding to non-numeric variables, so this is not caught by the eval. It is therefore necessary to make warnings fatal within the eval block:

sub is_number {
    my $test = shift;

    eval {
        local $SIG{__WARN__} = sub {die $_[0]};
        $test += 0;
    };
    if ($@) {return 0;} else {return 1;} 
}
[download]