mjhjmtu

Based on How do I prevent sed -i from destroying symlinks?, but I use Perl.

I tried by middle of all these questions:

• Is there a way to make perl -i not clobber symlinks?


• perl symlink on gz file


• How do I prevent sed from destroying hardinks?

It was unsucessfull. Here is the little and simple sample:

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

while (<>) 


 ## For all lines, replace all occurrences of #5c616c with another colour
 s/#5c616c/#8bbac9/g $(readlink -e -- "<>")

 ## $(readlink -e -- "<>") is similar to --in-place --follow-symlinks

 ## Print the line
 print;

The readlink -e command is not portable so should not be relied upon.

$ cat input
Like quills upon the fretful porpentine.
$ ln -s input alink
$ readlink -e alink
readlink: illegal option -- e
usage: readlink [-n] [file ...]

Within the Perl code, instead replace the links with the filename it points to using Perl's readlink function then loop over the input as usual.

$ perl -i -ple 'BEGINfor(@ARGV) $_=readlink if -l tr/A-Z/a-z/' alink

alink is still a symbolic link and the contents of input have been modified:

$ perl -E 'say readlink "alink"'
input
$ cat input
LIKE QUILLS UPON THE FRETFUL PORPENTINE.

Within a Perl script this might look something like

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

for my $arg (@ARGV) 
 $arg = readlink $arg if -l $arg;


# in-place edit with backup filename, perldoc -v '$^I'
$^I = ".whoops";

while (readline) 
 s/#5c616c/#8bbac9/g;
 print;

though may need List::Util::uniq or similar to avoid modifying the same filename two or more times, if the input contains duplicate filenames.