Sweet. Someone has gotten around to writing a Perl module to search the All Music Guide. Right now it seems to choke on album searches that return multiple results (I emailed the author) but otherwise seems to work pretty well. Lots of potential for those of us who want to get year and genre metadata done right...

Keep an eye on it. Right now the bug I mentioned keeps it from being useful for my scripts (without some *severe* hacking around it) but it's a great start.

Check it out @ http://mirrors.develooper.com/perl/CPAN/modules/by-module/WWW/YOHAMED/

Talking of AMG, what's the best alternative now that they've broken all their links by replacing perfectly good "http" URLs with useless "javascript" ones. I don't want to have to resort to using a Javascript browser, because of all the irritating things I see javascript do to other users, and I don't know of any browsers in Debian that let you control it on a per-site basis, for any sensible definition of "site"...

jEmplode 45 has a java interface to searching AllMusic also (it was needed to do song recommendations). Probably not as robust as the Perl module that's designed to do nothing but this, but it's probably a pretty good starting point for someone to add other searches in... If people are particularly interested in this API, let me know. It will show up in the next release...

Cool... I personally have more use for a Perl module since I already have an existing array of Perl scripts that do a lot of my mass-tagging and maintenance of MP3's... But I think someone out there would be interested in a Java API. I do Java and C by day, so Perl is kinda my "get stuff done at home" language.

Did you write your own code for parsing Allmusic? And does it correctly handle what seems to be a new interface when multiple albums are returned on an album search? If so I might be interested in it just to port the logic to the Perl module.

I know! It sucks that they moved to javascript. This means you can't open in new window (or new tab for mozilla) and screws my browsing style.

I e-mailed them about it but they didn't respond. Big suprise.

BTW, all the javascript does is redirect the page to a link. All it frickin does is build a URL. Rediculous.

yeah -- basically that javascript method just takes the ID of the page to redirect to .. I'm sure they did this just to screw with people who try to just steal those links ... it stopped me for all of 10 seconds though and just made their users annoyed.

Yeah .. i did the parsing myself ... It's pretty nasty code since you just have to hunt around for keywords in the page. I haven't done the by-album search, but i handle artist searches that return multiple artists, so I'm guessing it would be pretty similar.... It would be interesting to compare notes on how the Perl module does the parsing.

The JavaScript is done intentionally so they can "protect their information" from modules like the Perl module I'm speaking of. It doesn't prevent anyone from getting the info, just makes it a little more of an exercise. Of course they're not going to respond to your "petty requests." Because if we use a Perl module or anything else to get the information easier, they don't get web hits, ad revenue, etc.

I don't care how hard they try to obfuscate and hide their data, someone will figure out a way to get it. Hell, I would pay a good sum for a better search interface from them that I could write my own programs to query against.

Well, they can make it compleetly useless by obfuscating the information. I'm sure I'm not the only one who remembers lyics.ch and their wonderfull java applet interface which made it compleetly impossible to use...

Matthew

Well, if you wanted the lyrics archive, it was easy to find some major corporation to keep a mirror for you, but it appears the mirrors.aol.com ftp site finally no longer has the old uwp.edu lyrics archive mirror.

What a shame

I emailed them a while ago and talked with their sales guys about a better interface to the data ... They weren't too keen on the idea for exactly the reasons you listed. Their entire business model is predicated on people being required to see their ads. All these purely-data-providers are scared to death to provide a nice web-services-style interface to their data because they think one person will buy it and just give it away ... what they're not realizing is that people who want to steal their data will already do it so they might as well make _some_ money off of it. Oh well. I had this same discussion with MoodLogic and the guys who provide TV Guide information to companies like TiVo -- they all say the same thing. The head of sales for the company that does TV Guide data was actually really responsive and open to any suggestions I had for allowing them to provide a nice interface but also a mechanism to protect the data. Tough problem ...

Looks like this module has been updated, and artist and album searches are both working. It even has an option to save album covers! I like it.