#!/usr/bin/perl

# renames files in the current directory the way I (steve) like it
# also makes sub directories if the artist has multiple songs
# does NOT currently fix up any m3u lists with the new song names

# if you use foreign characters in your file names, you need this
use locale;

use Getopt::Std;
use strict;

our( $opt_d, $opt_h, $opt_g, $opt_D, $opt_H );

getopts("DdhgH");

usage() if $opt_h;

my $dir = "./";
$dir = $ARGV[0] if $ARGV[0];

my $extentions = 'avi|mp\w|mpeg|asf|mov|ogg|rm|ram|wmv';

my $doit = $opt_g;
my $debug = $opt_D;
my $allow_dirs = $opt_d;
my $allow_hidden = $opt_H;

# extract the directory that the script lives in
my ($progdir) = $0 =~ /^(.+)\//;

print "loading hyphenated list..." if $debug;
 
my $hlist = "$progdir/hyphenated.txt";
open HLIST, $hlist or die "cannot open hyphenation list: $hlist: $!";

my @hlist = <HLIST>;

close HLIST;

print "done\n" if $debug;

print "loading lowercase list..." if $debug;

my $lclist = "$progdir/lowercase.txt";
open LCLIST, $lclist or die "cannot open lowercase list: $lclist: $!";

my @lclist = <LCLIST>;

close LCLIST;

# remove comments
@lclist = grep {!/^\#/} @lclist;

print "done\n" if $debug;

opendir(DIR, "$dir") or die $!;
print "opening directory: $dir\n" if $debug;

my %makedir;
my %newnames;

while( my $file = readdir(DIR)){
  if( -d $file ){ $makedir{$file}++; } # this should test for existing dirs

  # ignore "hidden" files
  if( $file =~ /^\./  && !$allow_hidden ) {
      next;
  }
  print "examining: $file\n" if $debug;

 # initial filter
  if( $file =~ /\.($extentions)$/i || ( $allow_dirs && $file !~ /^\.+/ 
					&& -d $file )){ 

      print "working on: $file\n" if $debug;
      my $is_dir = $allow_dirs && -d $file;
      my $changed = 1;
      
      $file = mark_hyphenated( $file );
      
      my $track;
      my $artist;
      my $song;
      my $extension;
      my $newfile;
      
      # first style: [03 -] Artist - songname.mp3
      if( $file =~ /^(?:(\b\d{1,2}\b)\s*\-\s*)?([:alpha:][^\-]{1,}|[^\-]{3,})\s*\-\s*(.{2,})(?:\(\d\))?\.($extentions)$/i ){

	  $track = $1;
	  $artist = $2; $song = $3;
	  $extension = $4;
	  
	  # another style: (artist) songname.mp3
      }elsif( $file =~ /^\((.+?)\)\s*(.+)\.($extentions)$/i ){ 
	  
	  $artist =  $1; $song = $2;
	  $extension = $3;
	  
	  # another style: songname.mp3
      }elsif( $file =~ /^(.+)\.($extentions)$/i  ){ 
	  
	  $song = $1;
	  $extension = $2;
	  $artist = "";
	  
      }elsif( $is_dir && $file =~ /^(.+)$/i ){
	  $song = "";
	  $artist = $1;
	  $extension = "";
      }else{
	  
	  $changed = 0;
	  print "unknown file pattern. skipping.\n" if $debug;
      }
      
      $file = unmark_hyphenated( $file );
      $song = unmark_hyphenated( $song );
      $artist = unmark_hyphenated( $artist );
      
      if($changed){
	  print "processing...\n" if $debug;

	  $artist = fix_escaped( $artist );
	  $song   = fix_escaped( $song );

	  $artist = pre_regexp( $artist );
	  $song   = pre_regexp( $song );

	  $artist =~ s/(?<!\_)\_(20)?/ /g;
	  $song   =~ s/(?<!\_)\_(20)?/ /g;

	  $artist = filterBadChars($artist) if $artist ne "";
	  $song   = filterBadChars($song)   if $song ne "";

	  # trailing and proceeding whitespace
	  $artist =~ s/^\s*(.+?)\s*$/$1/g;
	  $song   =~ s/^\s*(.+?)\s*$/$1/g;

	  # duplicate whitespace
	  $artist =~ s/\s{2,}/ /g;
	  $song   =~ s/\s{2,}/ /g;

	  $artist = fixCase($artist) if $artist ne "";
	  $song   = fixCase($song)   if $song ne "";

	  # lowercase extension
	  $extension = lc $extension;

	  print "Artist: $artist\n"  if $debug;
	  print "Song name: $song\n" if $debug;

	  $makedir{$artist}++ if $artist ne "";


	  $newnames{$file} = { 'artist' => $artist, 
			       'song' => $song, 
			       'ext' => $extension,
			       'track' => $track 
			       };
      }

      print "------\n\n" if $debug;

  }
}

closedir DIR;

#-------------------------------------------------------#
#      actual files are changed below this point        #
#-------------------------------------------------------#

foreach $dir (keys %makedir){
  if($makedir{$dir} > 1 && !-d $dir){
    my $command = "mkdir \"$dir\"";

    if($doit){

      print "executing: $command\n";      
      `$command`;

    }else{

      print "$command\n";
    }
  }
}


foreach my $file (keys %newnames){
  my $newfile;
  my $command;

  my %song = %{$newnames{$file}};

  my $track = "";
  $track = "$song{'track'} - " if $song{'track'};

  if($song{'artist'} eq ""){
    $newfile = "$track$song{'song'}.$song{'ext'}";
  }else{
    if($makedir{$song{'artist'}} > 1){

	if( $song{'ext'} && $song{'song'} ){
	    $newfile = "$song{'artist'}/$track$song{'song'}.$song{'ext'}";
	}else{
	    $newfile = "$song{'artist'}";
	}
    }else{
	if( $song{'ext'} && $song{'song'} ){
	    $newfile = "$song{'artist'} - $track$song{'song'}.$song{'ext'}";
	}else{
	    $newfile = "$song{'artist'}";
	}
    }
  }
  if($file ne $newfile){
    $command =  "mv -i \"$file\" \"$newfile\"";

    if($doit){

      print "Renaming $file to $newfile\n";
      rename( $file, $newfile );
#      `$command`;
    }else{
      print "$command\n";
    }
  }
}

# gets rid of characters like '!' and the like from a filename
sub filterBadChars { # input raw scalar; output scalar with bad chars removed

  my($name) = @_;

  $name =~ s/\!|\~|\`//g;
  return $name;

}

sub fixCase { # input filename; output case fixed
    
    my($file)=@_;
    $file =~ s/(?<!^)\b([\w|\']+)\b/upperSome($1)/gei;
    $file = ucfirst($file);
    return $file;

}

sub upperSome { # upper case words that should be

  my($word) = @_;
  if(grep /^$word$/i, @lclist){
    $word = lc $word;
  }else{
    $word = ucfirst $word;
  }
  return $word;
}

sub mark_hyphenated { # $filename -- will look and mark any hyphenated
                      #              words with %2D

  my ($fname) = @_;
  foreach my $pattern (@hlist){
    if( $pattern =~ /^\-(.+)/ ){
      if($fname =~ s/\-($1)/%2D$1/ig){
        print "hyphenated: $fname\n" if $debug;
      }
    }

    if( $pattern =~ /(.+)\-$/ ){
      if($fname =~ s/($1)\-/$1%2D/ig){
        print "hyphenated: $fname\n" if $debug;

      }
    }

    if( $pattern =~ /(.+)\-(.+)/ ){
      if($fname =~ s/($1)\-($2)/$1%2D$2/ig){
        print "hyphenated: $fname\n" if $debug;
      }
    }     
      
  }
  return $fname;
}

sub unmark_hyphenated { #undoes the marking

  my( $fname ) = @_;

  $fname =~ s/\%2D/-/g;
  return $fname;

}

# searches and replaces the list beforehand
sub pre_regexp($){
    my ($fname) = @_;

    my @regexps;
    push @regexps, [qr/(?<=\w{2})n[ _]t\b/i, "n't"];
    push @regexps, [qr/(?<=\w{2})[ _]re\b/i, "'re"];
    push @regexps, [qr/\bl[ _]+\b/i, "L'"];
    push @regexps, [qr/\b[ _]+s\b/i, "'s"];

    push @regexps, [qr/\byoure\b/i, "you're"];
    push @regexps, [qr/\btheyre\b/i, "they're"];

    foreach my $reg (@regexps){
	$fname =~ s/@$reg[0]/@$reg[1]/g;
    }

    return $fname;
}

sub fix_escaped { #turns stuff like %20 into " "
  
  my( $fname ) = @_;

  $fname =~ s/%([\dA-F]{2})/pack('c',hex($1))/egi;

  return $fname;
}

sub usage(){
    die<<USAGE;
usage: $0 [-g] [-d] [-D] [-h] [directory]

options:
    -d      process directories as well
    -H      process hidden files (files whose names start with a ".")
    -g      Go! nothing will be done until this flag is set, allowing you to 
            preview changes
    -D      debug
    -h      this help

this program goes through a directory of media files and attempts to clean
up the file names according to the convention:

   Artist/Album/Song.mp3

or:
   Artist - Album - Song.mp3

for single files.

Copyright(C) 2002-2003 Steve Pomeroy <steve\@staticfree.info>
USAGE

}