This is a pretty straightforward thing to do. I'd break up the construction of $pattern in to steps:

[trwww@www misc]$ cat 618115.pl
use warnings;
use strict;

my @delimiters = (
  ',',
  '|',
  ':',
  '>',
  '][',
  '_|_',
);
my $pattern = join '|', map quotemeta, @delimiters;

my $text = "jojo,has|some:big>balls][nuts,sometimes_|_he,scratches";

foreach my $particle (split /$pattern/, $text) {
  print $particle."\n";
}
[download]

That gives the following output:

[trwww@www misc]$ perl 618115.pl
jojo
has
some
big
balls
nuts
sometimes
he
scratches
[download]

Hope this helps,

trwww

I just discovered quotemeta as a result of the recent functional functions node, so I'm pleased to see it in use here.

I have a question about your solution, though. When I print "$pattern\n", I get

\,|\||\:|\>|\]\[|_\|_
[download]

To me, it's odd that this works correctly with the comma, colon, and chevron escaped. I'll go look at perlre, but why doesn't it matter that these are escaped?

~dewey

In a regex, "\," is exactly equivalent to "," -- and likewise for colon and angle brackets. Those characters do not have any "magical" force in the regex syntax when used without escapes (in contrast to period, asterisk, square brackets and so on), nor do they have any special meaning when preceded by backslash (in contrast to "n", "t", "b", "d" and so on).

Meanwhile, quotemeta is a more-or-less general-purpose function -- according to the manual, it 'Returns the value of EXPR with all non-"word" characters backslashed. (That is, all characters not matching "/[A-Za-z_0-9]/" will be preceded by a backslash in the returned string, regardless of any locale settings.)'

(updated to fix display of square brackets in last paragraph)

Your description is such a mess due to bad HTML that I'm not sure what you really want to do, but most likely your problem is that \Q quotes meta characters so that the contents of $pattern are treated as a string to match. Omit \Q and in this case your life may be happier.

You also need to quote all the meta characters in $pattern and you need to "double quote" in a double quoted string: "(,|\\||:|>|\\]\\[|_\\|_)" so the quote character is available to be parsed by the regex engine. The two changed lines then become:

my $pattern= "(,|\\||:|>|\\]\\[|_\\|_)";
my @splitted= split /$pattern/,$text;
[download]

The two changed lines then become:

my $pattern= "(,|\\||:|>|\\]\\[|_\\|_)";
my @splitted= split /$pattern/,$text;
[download]

I know you're trying to stay as close as possible to the OP's code, but as an additional recommendation, for maximum clarity in such a situation one should really follow a cleaner approach like trwww's++ or if using a single hardcoded regex, then taking advantage of the /x modifier and throw in suitable whitespace.

At some point adding whitespace

leads

to

reduced

clarity

and

slower

comprehension

.

Two things would clean those particular lines up in my view - adding /x as you suggest, and using a character set for the single character delimiters:

my $pattern= "([,|:>] | \\]\\[ | _\\|_)";
my @splitted= split /$pattern/x, $text;
[download]

---

DWIM is Perl's answer to Gödel

my $text="jojo,has|some:big>balls][nuts,sometimes_|_he,scratches";
my @splitted= split /,|\||:|>|\]\[|_\|_/, $text;
foreach $particle (@splitted) {
    print $particle."\n";
}
[download]