Last week I had some down time and manage to crank out a relatively basic mashup of the iTunes Libary XML file and Pandora’s favourites feed. Essentially my code reads in the iTunes XML file and converts it into a readable form using XSLT then grabs the Pandora feed and compares the two using XPath.

The application is written in VB.NET on the ASP.NET platform. Yes, I know, VB is the poor ugly stepchild language but we use it at work and it was far easier to pick up and go. C# is on the docket for the next project, I promise. That said, I’m hoping to get some Ruby and Python up here too. I just need to get a ticket in with my Linux hosting provider.

Why did you write it?

I wanted to write this application for awhile but I never had the time or the motivation. Thankfully boredom breeds innovation for me. I had started the code months ago and abandoned it after getting stuck on the WebRequest to Pandora. I managed to dig up the code last week and had a few eureka moments which lead to release worthy code.

The reason behind the construction of this application is that I use Pandora every single day. I practically need their radio service to work. One thing that I do when I’m listening is bookmark the songs I like, and mod them up with the thumbs up button. It’s intuitive and quick to do which is great. However, when you’re ready to plop down a bunch of money on new music how do you know which songs to buy? I always forget to update my favourites list to remove the songs I’ve bought, and I’m too lazy to search my iTunes every time I want to buy a song. Sure, iTunes sometimes prevents you from buying the same song twice, but having been bitten by that bug before I’d rather have a third part verify that for me.

Improvements, Ideas, Concerns

The application is very rudimentary and will be sitting on a very restricted shared hosting environment. The code is definitely not rocket science, but it works for the most basic use case. That said, it doesn’t handle advanced searches or merge results for remixes and the like. It isn’t using a special algorithm for matching. It just does a direct XPath contains on the artist and song title. If it doesn’t match, the application will falsely mark it as if it doesn’t exist in your library. Likewise, single quotes are currently not handled correctly because I was lazy and XPath parameterization looked like a pain. This is just a proof of concept.

iTunes XML files can get pretty big. I think my library file was almost 6mb, which is rather large for a web upload. I might look into letting users upload a zip file and decompress it on the fly. Not sure how you’d go about that though. My guess is this application will thrash the server with IO under load.

One of my coworkers/friends suggested implementing affiliate IDs for the various music stores and injecting those into the application in order to generate revenue. If I ever wanted to turn this into a real product that might be a way to monetize, but rest assured that isn’t implemented at the moment. If anyone wants to see this as an actual product, feel free to drop me a line.

Where is it?

It will be hosted on my new domain at once the kinks are worked out. GoDaddy’s Windows hosting is far from perfect, but it gets the job done. Unfortunately, it appears they are trying to proxy my WebRequest and XSLT discovery outgoing sockets which is throwing an error. I have a ticket in to find out what I can do to fix that.

If you are interested in seeing the code I’ll gladly post it. Right now it isn’t up to much and still needs some cleaning up so I’ll leave it off for now.