Perl script versus one-liner - differences in functionality with regex

Question

I have a perl program that takes the STDIN (piped from another bash command). The output from the bash command is quite large, about 200 lines. I want to take the entire input (multiple lines) and feed that to a one-liner perl script, but so far nothing i've tried has worked. Conversely, if I use the following perl (.pl file):

#!/usr/bin/perl
use strict;

my $regex = qr/{(?:\n|.)*}(?:\n)/p;

if ( <> =~ /$regex/g ) {
  print "${^MATCH}\n";
}

And execute my bash command like this:

<bash command> | perl -0777 try_m_1.pl

It works. But as a one-liner, it doesn't work with the same regex/bash command. The result of the print command is nothing. I've tried it like this:

<bash command> | perl -0777 -e '/{(?:\n|.)*}(?:\n)/pg && print "$^MATCH";'

and this:

<bash command> | perl -0777 -e '/{(?:\n|.)*}(?:\n)/g; print "$1\n";'

And a bunch of other things, too many to list them all. I'm new to perl and only want to use it to get regex output from the text. If there's something better than perl to do this (I understand from reading around that sed wouldn't work for this?) feel free to suggest.

Update: based on @zdim answer, I tried the following, which worked:

<bash command> |  perl -0777 -ne '/(\{(?:\n|.)*\}(?:\n))/s and print "$1\n"'

I guess my regex needed to be wrapped in () and the { curly braces needed to be escaped.

${^MATCH} vs $^MATCH might be an issue. And you never capture anything for $1 to have a value... Plus your one liners need a -n... — Shawn
– Shawn, Commented Jun 7, 2019 at 1:33

zdim · Accepted Answer · 2019-06-07 22:08:05Z

3

A one-liner needs -n (or -p) to process input, so that files are opened, streams attached, and a loop set up. It still needs that even as the -0777 unsets the input record separator, so the file is read at once; see Why use the -p|-n in slurp mode in perl one liner?

That regex matches either a newline or any character other than a newline, and there is a modifier for that, /s, with which . matches newline as well. Then that need be inside curly braces, which you need to escape in newer Perls. The newline that follows doesn't need grouping.

So altogether you'd have

<bash command> | perl -0777 -ne'/(\{(.*)\}\n)/s and print "$1\n"'

edited Jun 7, 2019 at 22:08

answered Jun 7, 2019 at 2:01

zdim

67.2k5 gold badges59 silver badges87 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

Nena Over a year ago

The regex and syntax wasn't quite correct (there is a single quote character ' after the /s that should not be there for example, but based on your suggestion @zdim, I tried ` perl -0777 -ne '/(\{(?:\n|.)*\}(?:\n))/s and print "$1\n"'` and that worked. Thanks.

zdim Over a year ago

@Nena Great, glad it worked and thank you for attribution. Thank you for notice about the typo! Fixed

Léo Natan Over a year ago

Thank you for this explanation here. It was better than the duplicate in my closed question. :)

zdim Over a year ago

@LeoNatan Great, and thank you for saying this :). For reference here are then a couple more of related posts, a short one and one with more stuff.

Collectives™ on Stack Overflow

Perl script versus one-liner - differences in functionality with regex

1 Answer 1

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related