s/(?<=\d)(?=\D)/:/g;
[download]

   print "A3:xyz45:M98:AB7:Z9";  # :)
[download]

Using $,

my $s = 'A3xyz45M98AB7Z9';
local $, = ":";
print split /(?<=\d)(?=\D)/, $s;
[download]

Using $"

my $s = 'A3xyz45M98AB7Z9';
local $" = ":";
print "@{[split /(?<=\d)(?=\D)/, $s]}";
[download]

Using $;

my $s = 'A3xyz45M98AB7Z9';
my $i = 0;
%_ = map { (++$i, $_) } split /(?<=\d)(?=\D)/, $s;
local $; = ":";
$_{ (),  map { delete $_{$_} } 1..$i  }=0;
print keys %_;
[download]

Couldn't help myself O:-) Sorry :-)

{
  local ($,, $\) = (":", "\n");
  print unpack 'A2A5A3A3A2', 'A3xyz45M98AB7Z9';
}
[download]

Works for me ;)

I thought about doing it using fixed-length fields, but he seems to want to put a colon after each number followed by a letter (i.e. \d:[a-zA-Z]).

Yeah, what did not surprise me was the fact that people jumped on to find "patterns" in the string etc. The question simply asked "Take string X and transform it into string Y". I have learned, as a life lesson, that second-guessing people is not always profitable, in fact it's not most of the times. Just give 'em what they want, with minimum complications. My post was intended for contrast.

On a lighter side, I guess one could come up with an even more braindead solution (I'd investigate substr, index, for-a-la-C etc.)

TIMTOWTDI

my $s = 'A3xyz45M98AB7Z9';
$s =~ s/[A-Za-z]+\d(?!$)/$&:/gsi;
print "s='$s'"
[download]

Minor correction:

my $s = 'A3xyz45M98AB7Z9';
$s =~ s/([A-Za-z]+\d+)(?!$)/$1:/g;
print "s='$s'\n"
[download]

Here's one solution.

local $_ = "A3xyz45M98AB7Z9"; 
s/(?<=\d)\B(?=\D)/:/g;
print $_, $/;
[download]

Update: I see this is the same as the solution of Roy Johnson.

To be honest, a description of what your cow-orker wants would be helpful, too. I prefer to write my solution in the language of the problem, rather than the language of the solution. It makes things more obvious later (and possibly, more robust when strange things happen).

For example, if you were to say that "Each set of letters followed by set of numbers is a token, and we want colons to separate our tokens," then, IMO, splitting the string into said tokens, and joining with a colon, as you did, seems to be the best way to me to do it.

But, if the requirement is, "I don't want numbers and letters to touch when the number is prior to the letter, so separate them with a colon," then I would suggest an alternate approach, such as s/(?<\d)(?>[[:alpha:]])/:/g.

The point is to really pay attention to what you really need, and to write the solution that matches.

perl -e \
'$_="A3xyz45M98AB7Z9",s/(\d)([a-zA-Z])/$1:$2/gsi,print'
[download]

use strict;

my $s = "A3xyz45M98AB7Z9";
my @stuff = split(//,$s);
for (my $j=0;$j<@stuff;$j++) {
  if ($stuff[$j] =~ /\d/ && $stuff[$j+1] && $stuff[$j+1] =~ /[a-zA-Z]/
+) {
    print "$stuff[$j]:";
  } else {
    print "$stuff[$j]";
  }
}
[download]

I'm working on a way to make it as complicated as possible. Just because.

piroufreek

my $s = 'A3xyz45M98AB7Z9';

($t=join'',map{/\d+/g?$_.=':':$_}split//,$s)=~s/:(\d+|$)/$1/g;

print $t;
[download]