rename and move files to the import directory

[Gog]

I use unRaid as a NAS and run a torrent client on it. I used to manually rename and move the TV episodes but that got tiresome and I used the opportunity to learn some perl and regular expressions. I make no claim to be any good at either

It is customized to my quirks and setup but maybe it can be useful to others.

It should run fine on linux and windows, I used both to test it. If you find odd comments or variable names it's because I forgot to translate from french

usage:

perl tv_rename.pl /path/to/files/to/rename/*

tv_rename.pl:

#!/usr/bin/perl
#TV series file rename script, v0.9.1, 2011-01-22
#v0.9.1: added forced 2 digits in series and episode numbers, tweaked the handling of " - " for reruns
#v0.9.2: use regular expression for "csi" substitution, series name optional in path
#v0.9.3: deal with the 6x09 format, 
#		 added caps to series first letter of each word, 
#		 remove multiple spaces, 
#		 check if the dir exists, check if it's a caps issue then create a new one
#v0.9.4: replace "_" with " "
#v0.9.5: fancy up the ripper info remover a bit using a scalar and deal better with "-" in the ripper's signatures
#		 same / in win and linux, who'd a thunk
#v0.9.6: deal with episodes like "201".  The episode number MUST be 2 digits for the script to work.
#v0.9.7: reworked the seep grabber to limit the problems when a series name ends with numbers
#
# ToDo
# get the parameters from the command line

use feature ':5.10';
use strict;
use warnings;


#config section
my $dest_path_tv= "/mnt/disk7/tv";   
my $test_mode = 1;
my $include_episode_title =1;
my $add_series_in_path=1;

my $old_name="";

# windows section, deal with * in the arguments which are not expanded in windows
# and the "/" vs "\" problem
if ($^O eq "MSWin32") {
	use File::DosGlob;
	@ARGV = map {
	my @g = File::DosGlob::glob($_) if /[*?]/;
	@g ? @g : $_;
	} @ARGV;
	
	#different path under windows
	$dest_path_tv= "//tower/disk7/tv";   
}

sub it_is_video($$$)
	{

	my $filename = shift;
	my $extension = shift;
	$old_name = shift;
	my $new_name;
	my $serie;
	my $seep;
	my $episode;
	#print "var      <$1><$2><$3><$4><$5><$6><$7><$8><$9><$10>\n";

	#remove the path
	$filename =~ s/(.*\/)//;
	#replace "_" with " "
	$filename =~ s/_/ /g;
	
	#extract series name, season, episode and episode name.
	#pattern: s01e01  s1e01 01x01 1x01
	$filename =~ m/(.*?)( - )?(S|s)?(\d{1,2})(E|e|X|x)(\d{2})( - )?(.*)/;
	if (defined($4) and defined ($6)){
		print"this is TV pattern 1\n";
		$serie=lc($1);
		$seep= sprintf ('s%#.2ue%#.2u', $4, $6);
		$episode=$8;
	}
	else{
		#pattern: 201  separated it, it had the hapit of eating numbers at the end of a series's name ex: V 2009
		$filename =~ m/(.*?)( - )?(\d{1,2})(\d{2})( - )?(.*)/;
		if(defined($3) and defined ($4)){
			print"this is TV pattern 2\n";
			$serie=lc($1);
			$seep= sprintf ('s%#.2ue%#.2u', $3, $4);
			$episode=$6;			
		}
		else {
			print "$old_name is not a TV episode\n";
			return;
		}
	}
	


	#episode name cleanup
	#remove ripper's info
	my $CRAP = '(HDTV|-H|FEVER|xvid|INTERNAL|SYS|LOL|LOL|\[VTV\]|PROPER|FQM|2HD|notv|FQM|www|therealworld de|tcm[|]|\(|\)|{|})';
	#remove "-" between tags but not in "real" episode name
	$episode =~ s/($CRAP-$CRAP)|$CRAP//gi;
	#remove leading and lasting "." and " "
	$episode =~ s/[\. ]*$//g;
	$episode =~ s/^[\. ]*//g;
	#remove multiple "."
	$episode =~ s/\.{2,}/./g;
	#replace "." by " "
	$episode =~ s/\./ /g;
	#remove multiple " "
	$episode =~ s/\ {2,}/ /g;	
	
	#series name cleanup
	#remove leading and lasting "."  and " "
	$serie =~ s/[\. ]*$//g;
	$serie =~ s/^[\. ]*//g;
	#remove multiple "."
	$serie =~ s/\.{2,}/./g;
	#replace "." by " "
	$serie =~ s/\./ /g;
	#remove multiple " "
	$serie =~ s/\ {2,}/ /g;
	#rename CSI for BMT data scraper
	$serie =~ s/^csi$/crime scene investigation/;
	#capitalize the first letters  I wonder if this will help BMT...
	$serie =~ s/\b([a-z])(\w+)\b/\u$1$2/g;
	#keep NCIS all in caps because I'm an annoying bastard
	$serie =~ s/(Ncis)(.*)/\U$1\E$2/;
	#print "$serie\n";
	if ($add_series_in_path==1) {
		$new_name="$dest_path_tv/$serie/";
		my $dir="$dest_path_tv/$serie";
		unless(-d $dir){
			my $serielc=lc($serie);
			my $dirlc="$dest_path_tv/$serielc";
			#if the directory exists in lower caps, use that one
			if(-d $dirlc) {
				$new_name="$dest_path_tv/$serielc/";
			}
			#if not, create with caps
			else {
				if ($test_mode==0){
					use File::Path;
					mkpath($dir,1) or print "unable to create $dir ($!)";
				}
				else { print "test mode, will not ";}
				print "make dir *$dir* here ($!)\n";
			}
		}
	}
	else{
		$new_name="$dest_path_tv/";
	}
	
	if ($episode ne "" and $include_episode_title==1) {
		$new_name="$new_name$serie - $seep - $episode.$extension";
	}
	else{
		$new_name="$new_name$serie - $seep.$extension";
	}
	
	if ($old_name eq $new_name) {
		print "Name hasn't change, skipping $old_name\n";
	}
	elsif ($test_mode==1) { 	
		print "test mode, file is not renamed\n$old_name\n$new_name\n\n";
	}
	elsif (rename($old_name, $new_name)){
		print "file moved:\n$old_name to\n$new_name\n\n";
	}
	else {
		print "unable to rename $old_name to $new_name\nCheck that the destination folder exists and matches caps/now_caps";
	}
}

foreach (@ARGV) {
	my $old_name = $_;
	die $@ if $@;
	my @video  = ("avi", "mpg", "mpeg", "mp4");
	my $working = $old_name;
	
	#remove the file type and save it
	$working =~ m/(.*\.)(.*$)/;
	$working = $1;
	my $extension = $2;
	#check if the extension matches the known video file type
	foreach (@video) {
		if ($_ eq lc($extension)){
			it_is_video($working,$extension, $old_name);
		}
	}
}

exit(0);

I added this in my crontab so that it runs every 5 minutes. The output is emailed to me so that I can double-check where the files are sent.

*/5 * * * * perl /boot/custom/tv_rename.pl /mnt/disk7/torrent/copy_to_tv/*

Gog


edits:
#v0.9.1: added forced 2 digits in series and episode numbers, tweaked the handling of " - " for reruns
#v0.9.2: use regular expression for "csi" substitution, series name optional in path
#v0.9.3: deal with the 6x09 format,
# added caps to series first letter of each word,
# remove multiple spaces,
# check if the dir exists, check if it's a caps issue then create a new one
#v0.9.4: replace "_" with " "
#v0.9.5: fancy up the ripper info remover a bit using a scalar and deal better with "-" in the ripper's signatures
# same / in win and linux, who'd a thunk
#v0.9.6: deal with episodes like "201". The episode number MUST be 2 digits for the script to work.
#v0.9.7: reworked the seep grabber to limit the problems when a series name ends with numbers