In my experience, capturing all possible variants of phone numbers in a single regex is impossible, so I would use a battery of expressions. I assume that if multiple phone numbers are found in your record, that they are separated by commas. This code should do the trick. Note that I used ^ and $ in the expressions, because the office phone pattern is a proper subset of the international pattern, and both numbers match the office pattern if I do not insist that the pattern matches the whole string. Sometimes you need to sort your patterns carefully according to priority if multiple patterns match the same string.

#!/usr/bin/perl

my @office_patterns = ('^\d{3}-\d{4}$');
my @international_patterns = ('^[+]\d \d{3} \d{3}-\d{4}$');

my $record = "OfficePhoneNumber: 662-5555,+1 102 892-1314";
my ($field, $data) = split(/:\s*/, $record);
my @office_phones;
my @international_phones;
NUMBER: foreach $number (split(/,\s*/, $data)) {
  # Trim spaces.
  $number =~ s/^\s+//;
  $number =~ s/\s+$//;
  
  foreach $office_pattern (@office_patterns) {
    if ($number =~ /$office_pattern/) {
      push @office_phones, $number;
      next NUMBER;
    }
  }
  foreach $international_pattern (@international_patterns) {
    if ($number =~ /$international_pattern/) {
      push @international_phones, $number;
      next NUMBER;
    }
  }
}

print "Office #: ", join(", ", @office_phones), "\n";
print "International #: ", join(", ", @international_phones), "\n";
[download]