Changes at discogs.com

Since mine is not the only discogs script, I decided to open an extra topic for that.

I noted yesterday, that discogs has changed a lot more than I thought a few weeks ago.

Obvious was the change of the Catalog# from an extra line into the line of the Label, because that was on all sites and stopt all the non-api script from working.
Now, I see two more new things:
They have now something callled "Series" on a some releases, which is kind of a second label:
http://www.discogs.com/Don-McLean-American...release/2122666

And they have a new section called "Companies etc" under the tracklist before the "Notes", "Credits" and "Barcodes and Other Identifiers" sections:
http://www.discogs.com/tobias-Leaning-Over...release/2958926

http://blog.discogs.com/2011/08/label-and-...ugust-2011.html
http://www.discogs.com/help/submission-gui...any_Definitions

I read somewhere in their forum they are planning to do more updates on the Companies etc section, and something of a completly change for the way catalog#s are treated. Whatever that means, looks their is coming some work coming for us scripters to keep our scripts working.

And they seem to have introduced a new API (version 2.0) some time ago.
http://blog.discogs.com/2011/06/api-v20.html
The old one will stop working on December 17.
The old and new are basically the same if you change "www" to "api" in the URL and leave away the "&api_key=..." ending. The only thing I noticed is that now the youtube links are at the end of the xml document.
But you can also leave away the "?f=xml" in the URL now (change only "www" to "api") and you will get an output in JSON instead of API. This shows me the same output in my browser (Firefox), but a completly different formated text in the debug file. I can't make up much of that JSON output, but maybe it's helpful for others.
The strange thing is, the new things, "series" and "companies etc" don't show up in both api outputs. So I think there could be another change for it soon.

Now I wonder what to do with these things in the scripts. I think, for my script I will treat Series as an additional Label and make a new option to seperate it in an extra tag-field. Similar like I have done with "Style", which can be written as addtional genre or seperated into an extra field.
And I will treat the "Companies etc" like "Notes", "Credits" and "Barcodes and Other Identifiers": There I have the option to combine them all in one tag-field or writting them all or partially in extra fields. Another update for it will be comming soon. Another option would be to split this up into different tag-fields, as there is only a closed list of 26 entries which are allowed in that section.
Stupid thing with the API. I was planning to switch my script complety to the xml documents for more stability. But now that they have less information than the html documents, that doesn't seem to be good thing to do.

The old API (1.0) will be shut down on December 15!
It looks like the API-List of all releases by one artist, which dano's and my scripts use, will not work with the new API (2.0).
The new version lists only the master releases when more then one versions of a release exist. And this master release has unfortunately only one release id, the other versions are missing.
Releases with only one version are still all displayed. But these come now without the album artist.

Have a look a these links:

Normal Artist page:
http://www.discogs.com/artist/Lester+Jones

Old API Artist page:
http://www.discogs.com/artist/Lester+Jones..._key=41e338cd4e

New API Artist page:
http://api.discogs.com/artist/Lester+Jones...&releases=1

Notice the second release "Kurbel Compilation 4". In the old API it had two entries for every version. In the new one, there is only one entry for the master release with a id of the main release.

In lack of all releases we can sort out the issue by local PHP web server script.
Have a look at this topic Discogs: Parse Album (Release)

Discogs introduced translations into different languages:
http://www.discogs.com/help/forums/topic/333307

That makes parsing the html page more difficult.

Does anybody know some solution how to parse the always the English Version, matter from what country the scripts are used?

I've tried to use more html elements to locate the lines.
Don't know how reliable that is but maybe the change isn't so bad.