How to make code smaller--saving steps and writing one liners

[Petras]

Hello! I wrote something recently, and then had to go through it and actually seperate things into a few lines to get it to work (not that I did it right the first time...) It's a simple code to make a list of files dropping out the first few levels of file paths. The code in question is here:

use File::Find;

open (OUT, ">filelist.txt");

find( sub {
    $temp = $File::Find::name;
    $temp =~ s/C:\/users\/User\/Music\/(.*)/$1/gi;
    print OUT "$temp\n";
}, "C:/users/User/Music/");

close OUT;

If you can't tell just by looking at it that I'm running Windows, you probably can't help me here :~)

It's Strawberry Perl 5.8

I recall reading once that someone wrote a web spider in one line...I'm not really good at one liners. Is there a way to squish

    $temp = $File::Find::name;
    $temp =~ s/C:\/users\/User\/Music\/(.*)/$1/gi;
    print OUT "$temp\n";

into two or even one lines? And does anyone know a good place to start looking for something like a brevity tutorial? I recently watched a podcast where the speaker said, "If statements are obsolete if you know how to use the space-ship operator!" I'm not so much looking to obfuscate, but I'd like to save a few keystrokes.

Thanks and Cheers!
~Petras

[perrin]

In general, making things easy to read is more important than making them short. In this case though, I'd agree that you're taking the long way around without a good reason. I'd do something like this (untested!):

print ($File::Find::name =~ m/^C:\/users\/User\/Music\/(.*)$/i);

P.S. Please use strict and warnings!

[ikegami]

I wasn't happy with the fragile redundancy present in the OP and my earlier solution, so what follows are solutions without that problem:

perl -MFile::Find -le"chdir($ARGV[0]); find { wanted=>sub{ -d || print
+ /..(.*)/ }, no_chdir=>1 }, '.'" C:\users\User\Music >filelist.txt

perl -MFile::Find::Rule -le"chdir($ARGV[0]); print for File::Find::Rul
+e->file->in('.')" C:\users\User\Music >filelist.txt

The key is to use chdir to select the perspective you want (and preserve that perspective using no_chdir=>1 with File::Find)

In both cases, I filtered out directories.

Ref: File::Find, File::Find::Rule

[afoken]

My ideas:

• If you invoke File::Find::find() with the no_chdir=>1 option, $_ and $File::Find::name have the same value. Use local $_=$_; in the "wanted" callback and you can modify it as you like.
• Just remove the unwanted prefix instead of matching the entire name (unless you want to untaint), and use different RE quotes to avoid the leaning toothpick syndrom: s|^C:/users/User/Music/||i;
• Save the path in a variable, pass that to File::Find::find(), and use the variable in the substitution using quotemeta: s|^\Q$path\E||i;
• Your code lacks a check for a successful open, and it uses the "ancient" two-argument open: open OUT,'>','filename.txt' or die "Can't open filename.txt: $!";
• If you would redirect STDOUT when invoking the script, you don't have to work with OUT and open() at all. That would also allow to pipe the output into another program.
• If you want to list only files, test for files inside the callback, before modifying $_: return unless -f $_;

Alexander

[Anonymous Monk]

Use local $_=$_; in the "wanted" callback and you can modify it as you like.
Or just modify it as you like :) File::Find doesn't mind

[afoken]

You are right, I was too careful here. The written contract (the File::Find documentation) explicitly states:

The above variables [$File::Find::dir, $File::Find::name, and $_] have all been localized and may be changed without effecting data outside of the wanted function.

Alexander

[ikegami]

dir /s/b C:\users\User\Music | perl -pe"s/^C:\\users\\User\\Music\\//"
+ >filelist.txt

You probably want to add /a-d to avoid listing directory names.

[ig]

I agree with perrin that making code easy to read and understand is more important than making them short.

Sometimes making code short does make it easier to read and understand, but not always. In some cases, making code shorter may make it easier to read while you are actively working on it but it may be harder for others, or yourself when you return to it after some time.

Perl is a free-form language, so one can put anything and everything on one line. I sometimes put multiple statements on one line and sometimes split a long statement across multiple lines, both to improve readability.

I would make only a slight change to perrin's recommended solution - changing the delimiters on the re to avoid the escapes (tested (on linux)):

print ($File::Find::name =~ m!^/some/path/prefix/(.*)$!i);

[Marshall]

"If statements are obsolete if you know how to use the space-ship operator!"

That is just NONSENSE!

1) The space-ship operator, <=>, is a numeric comparison.
2) eq is an alpha-numeric comparison.
There are very good reasons for using <=> or eq in conjunction with an if statement! DUH!!

[Anonymous Monk]

Add autodie since you're not error checking, all one line

use File::Find; use autodie; open (OUT, ">filelist.txt");  find( sub {
+     $temp = $File::Find::name;     $temp =~ s/C:\/users\/User\/Music
+\/(.*)/$1/gi;     print OUT "$temp\n"; }, "C:/users/User/Music/");  c
+lose OUT;

[BrowserUk]

Not perl, but simple, effective and fast:

for /r "C:\users\User\Music\" %d in (.) do @dir /a-d /b %d 2>nul >>fil
+elist.txt

[afoken]

I would consider that line noise. ;-)

(Not exactly your code, but that sick syntax extension of DOS batches based on implementation errors in ancient DOS versions.)

Alexander