# Bens amazing 'folder structure of mp3 files -> folder of book cover images' convertor.
# March, 2010
#
# Freeware. Use it for the benefit of mankind (or not, it's up to you).
# Credit Ben Dunham [www.toasta.com] if you feel like it.

require 'ftools'
require 'rubygems'
require "mp3info"
require 'amazon/aws/search'
require 'open-uri'
include Amazon::AWS
include Amazon::AWS::Search

# Initialise the cache of previously extracted Albums
found = []

# Initialise the image file name counter
i = 0

# A dirty, but easy deep scan of all the Music folders on the NAS. Negates the need for recursion.
Dir.glob('/Volumes/media/Music/**/*.mp3').each do |mp3|
  
  # Use mp3info to read the ID3 tag from each MP3 file
  Mp3Info.open(mp3) do |mp3|
    
    # Don't bother checking if we have already dealt with an MP3 from this Album (doesn't 
    # cater for the case where multiple Albums from different artists have the same name!)
    unless found.include?(mp3.tag2.TALB)
      
      # Store the Album name in the cache
      found << mp3.tag2.TALB
      
      # If the MP3 has an embedded Artwork field...
      if mp3.tag2.APIC
        
        # Increment the counter
        i += 1
        
        # Extract the Image data
        text_encoding, mime_type, picture_type, description, picture_data = mp3.tag2["APIC"].unpack("c Z* c Z* a*")
        
        # Write it to the image folder
        File.open('../music/' + i.to_s + '.' + mime_type.split('/').last,'w'){|f| f.write(picture_data) }
      
      # Otherwise, look up the item in Amazon
      else
        begin
          # Initialise the Search objects (use a variety of MP3 ID3 tags to deduce the band name)
          is = ItemSearch.new('Music', { 'Artist' => (mp3.tag2.TPE1 || mp3.tag2.TCOM || mp3.tag2.TPE2).to_s, 'Title' => mp3.tag2.TALB.to_s } )
          rg = ResponseGroup.new( 'Large' )
          req = Request.new
          
          # Restrict to Amazon UK
          req.locale = 'uk'
          
          # Do the search
          resp = req.search( is, rg )
          
          # If any items were matched...
          if resp.item_search_response[0].items.any?
            
            # Increment the counter
            i += 1
            
            # Download the file
            image = open(resp.item_search_response[0].items[0].item[0].small_image.url)
            
            # And write it to disk
            File.open('../music/' + i.to_s + '.jpg','w'){|f| f.write(image.read) }
            image.close
            
          end
        rescue
          # No image found. Oh well!
        end
      end
    end
  end
end