I’ve changed the latest beta to kick off a library scan when the request is received. Before it was only refreshing the individual items which currently doesn’t provide feedback in the web interface (coming soon though). So once we have the visible refresh queue it will be changed back to just re-scan that one series.
Good to know, thanks @luke.
Hey guys,
I was just playing with NzbDrone myself and I noticed the Media Browser metadata support is very limited.
Since we fully support Xbmc nfo, why not just use that and remove whatever xml writing you’ve added? It’s actually our default now anyway so it won’t be long before our nfo users outnumber our xml users (if they haven’t already).
Then you can keep the Media Browser option but just write nfo’s. If the user selects both Xbmc and Media Browser, there’s only one set of files to write.
By the way guys, I took a look at your episode file name parsing engine. I thought perhaps we might take some inspiration from yours if yours is a little better.
I made the switch and some of our unit tests started failing. Anyway, that’s just the background. I actually prefer yours despite the failing tests.
Then I thought, since we’re both .NET based projects, and since we both have to support the same file naming, it would be neat to extract this to a dependency-free standalone nuget package that is well supported and thoroughly unit tested. I’m sure there are developers on both your side and ours who would have some interest in helping maintain and improve it.
@luke thanks for dropping by, we are actually thinking about reworking the way parsing is done. and by the end of it it’ll actually be a more integrated engine.
What i mean is, right now our parsing is fairly static, we parse a string based on the string alone, what we are trying to get to is to parse the string using the string as well as the data we have in the database about the show/episode etc. once we make the switch I’m not sure if the parser would be something that could still be used in isolation.
also, I’m wondering if could share some of your failing tests. maybe we can fix our parser to support those.
Hey @luke.
At the time it was added, XBMC metadata wasn’t yet supported. As for switching it, that becomes a bit tricky, since we track what metadata is used its a bit messy if we have two different providers that produce the same file and attempt to claim handling of those files. I’m not sure how used it is at the moment, but we tend to shy away from breaking changes, which this could be in the user hasn’t updated yet.
To further @kayone’s comment, we have a collision between 101 matching season 1 episode 1 and absolute number 101 when the file name doesn’t have any anime markers (hash or release group in ), which is a big factor in why we want to make changes. This would mean we’d be breaking up the regex, while it would still be the same regex, how it gets applied would be changed when we know we’re dealing with files of a certain type, or when we’re dealing with a file of unknown type (like from an RSS feed).
Thanks guys. Here are our unit tests:
https://github.com/MediaBrowser/MediaBrowser/blob/master/MediaBrowser.Tests/Resolvers/TvUtilTests.cs
I would have to run it again to recall which ones failed. I think you can probably easily convert them into your own tests. They were mostly taken from Sickbeard. Keep in mind I am not trying to say our parsing is more accurate than yours, in fact, if we were to run your tests against our engine there would probably be just as many issues. That’s why i thought I would propose the idea of a library that can be shared by multiple tools.
But I hear your points about using internet data to resolve file names. I didn’t think of that. Although I think that could always be handled via dependency injection. Well, it was just an idea.
Hey guys,
Just thought I’d let you know we are going to move forward with a stand-alone, zero-dependency file naming project.
I’ve found a couple contributors to help with it, and we’re shooting pretty high as we’ll be supporting Kodi’s entire feature set, and the consuming application will be able to configure it such that there will be no hard-coded algorithms.
http://mediabrowser.tv/community/index.php?/topic/13206-contribute-to-the-new-file-naming-project/
I know it sounded like you’ll be proceeding differently, but if you decide you would like to collaborate in some fashion we’ll always be receptive to that.