#Genres by Dale Swanson June 3rd 2010
#requires:
# wget http://users.ugent.be/~bpuype/wget/
# tag http://www.synthetic-soul.co.uk/tag/
#
#pulls genre data from last.fm and tags mp3s
#goes through all directories wherever it's ran (nonrecursive) and uses each as a band name
#Uses the last.fm artist top tags API http://www.last.fm/api/show?service=288
#takes top 5 genres and adds them to a batch file which actually adds the tags (doesn't run the batch file)
#This batch file uses the tag command line to add the tags
#You should go over the batch file manually, and confirm the genres, then run it manually.
#That CLI tagger seems to have a problem with everything except APE2 Mp3 tags.

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

use Cwd;
use URI::Escape;

my $debug = 1; #debug mode, set to 1 to get lot's of output
my $maxtags = 5; #how many genre tags you wanted for each band
my $tagseparator = ", "; #what to put between seperate genres.  ex: ", ", "\0", or "/" Note you can't pass NULLs.  Instead pass something unique then use Foobar2000's mass tagger plugin to split Genre feild by that separator.

my $dir; #the directory the program is running from
my @allfiles; #stores all the files in the directory
my $file; #used to access each file in the allfiles array.  Will be name of directory, which should be band's name.
my $band; #separate variable for band's name so that we can escape it for URI.
my $tagcount; #counter of how many tags we have for that band
my $taglist; #list of all genres as they will appear in tag
my @filearray; #will store the contents of the file
my $fileline; #will store each line of the file

my $testcount; #testing stuff remove later (i.e. never)

$dir = cwd(); #starting directory

(print "\ndir - $dir") if ($debug);
opendir THISDIR, $dir or die "Can't open directory: $!";
@allfiles = grep !/^\.\.?$/, readdir THISDIR; #gets all the files in this directory, strips . and ..
closedir THISDIR;
(print "\nALL FILES:\n@allfiles\n") if ($debug);

foreach $file (@allfiles)
{# go through each file, check to see if it's a directory
	(print "\nTest File - $file") if ($debug);
	$taglist = ""; #reset tag list for new band
	$tagcount=0; #new band reset tag count
	if (-d $dir . "/" . $file )
	{# if it's a directory we assume it's a band name.  Thus $file is our band's name.
		(print "\nList DIR - $file") if ($debug);
		$band = $file; #file will be unescaped band's name
		(print "\nBand - $band") if ($debug);
		$band = uri_escape($file); #band will be escaped band's name
		(print "\nBand - $band") if ($debug);
		$band =~ s/%/%%/g; #escape % to %% for batch files to work
		(print "\nBand - $band") if ($debug);
		
		open(ofile, ">downxml.bat"); #output the line
		print ofile ("wget \"http://ws.audioscrobbler.com/2.0/?method=artist.gettoptags&artist=$band&api_key=b25b959554ed76058ac220b7b2e0a026\" -O tags.xml"); 
		close(ofile);
		system "downxml.bat";
		
		open(ifile, "tags.xml"); #opens the results of our API search
		@filearray = <ifile>; #loads the contents of the API search into an array
		close(ifile);
		#unlink("tags.xml"); #delete file once we are done with it so that it doesn't show up for next band
		foreach $fileline (@filearray)
		{#go through each line in XML file searching for genre tags
			chomp($fileline);
			if ($fileline =~ m/\<name\>/)
			{#a tag is here
				(print "\nPossible Tag - $fileline") if ($debug);
				$tag = $fileline;
				chomp($tag);
				$tag =~ s/\s+/ /; #turns multiple whitespace into single space
				$tag =~ s/\<name\>//g; #strips <name> tag
				$tag =~ s/\<\/name\>//g; #strips </name> tag
				$tag =~ s/^\s+//;#strip leading spaces
				$tag =~ s/\s+$//;#strip trailing spaces
				
				#here we filter stupid tags, some ideas, band's name, curses, common tags, words count > 3
				if ($tag !~ m/$file/ && # no need to tag band's name
				$tag !~ m/fuck/i && #probably not a descriptive tag
				$tag !~ m/albums/i && #albums i own
				$tag !~ m/seen/i && #seen em live
				$tagcount < $maxtags) 
				{
					$testcount = ($tag =~ tr/ //); #remove all this later, or don't
					if ($testcount >= 2)
					{#quick test to see if long tags tend to be valid
						open(ofile, ">>longtags.txt"); #output the line
						print ofile ("\n$tag"); #testing
						close(ofile);
					}
					
					(print "\nProbable Tag - $tag") if ($debug);
					$tagcount++; #count this tag
					if ($tagcount == 1)
					{#first tag, add without delimiter
						$taglist = $tag;
					}
					else
					{#adds this tag plus the delimiter to the list
						$taglist .= $tagseparator . $tag;
					}
				}
			}
		}
		(print "\nTaglist - $taglist") if ($debug);
		open(ofile, ">>addtags.bat"); #output the line
		print ofile ("\ntag --ape2 --recursive --caps --genre \"$taglist\" \"$file\\*.*\""); #hopefully $taglist is a list of genres, and $file is the bands name/directory
		close(ofile);
	}
}

system ("PAUSE");
print "\nDone!\n";