Usually in this situation, anchor the regex to the end of the string and report however many digits are before .htm or html. Unless there is some specific reason to disallow 1,2,3 digit numbers, I wouldn't code it that way. Go with something simple that works for 1,2,3,4,5,6,7,8 or more digits.

#!/usr/bin/perl
use strict;
use warnings;

foreach my $url ("abgc100.html", "xyz1000.htMl", 
                 "qwer10.html",  "abc123.htm", "qrz12345.htm",
                 "something-12341234.html")
{
   my ($number) = $url =~ /(\d+)\.htm(l)?$/i;
   print "$number\n";
}

__END__
100
1000
10
123
12345
12341234
[download]

You have to decide about something-12341234.html. The code above captures 12341234 which is usually what is desired. Do you really only want to have 41234, 5 digits in that situation? I suspect not.