NFO agent

[MTI]

And the devs know that they are in a small way hindering their own project by not having documentation. But when your engineers are the very ones best suited to write said documentation, you have to make a priority call. Documentation only comes around at the last minute possible. That's the way of engineering.

Sure, I understand and respect that. (Although, it seems that every "complaint" can be refuted by the "lack of resources" comment)...

[hrcolb0]

Why not write a wiki about how to write metadata agents. You’ve learned enough by now to do a pretty good one. That could get your documentation ball rolling.

[MTI]

Look, I’m really not that much into the code… Nor do I have the time…



I’m a coding noob, but I like to try and learn along the way. I’m an amateur, but I come here with good intentions trying to fill some gaps (NFO files, EyeTV scanning/agent). I succeeded (with your and other’s great help).



Basically we were told that for things that were not the focus or the personal interests of the dev team: Do it yourselves. Okay, so we did - and succeeded.



Now I ask: If I have to go forward doing the code myself it would be nice to have some docs to lean against. Then again I’m told: Do it [the docs] yourself…



Well, I’m sorry, but I don’t have that much time…

[hrcolb0]

I’m saying do it yourself. Who the hell am I? I’m just some forum poster same as you. Sorry for flying off the handle. I get flustered easily if you couldn’t tell already. Yeah, it’s gotten to be a time suck for me, but I enjoy it. Like you said, you have the ability to keep dragging on your same talking points, just like I have the ability to bag on you when you do. I simply felt that we were making productive steps towards my Plex goals (i.e. getting more devs in the community) and I felt that you were coming in like a wet blanket.



I do want to point out that you are making the argument that YOU don’t have enough time, but you don’t seem to accept that same argument for the dev team, some of which I am pretty sure still have real world jobs, and little time too.

[ussalaska]

Glad to hear you two have calmed down. I believe that you two are the only that ever have responded to my posts in these forums (quite sure honestly) in which I am very thankful for the help. I know it is frustrating to figure this stuff out and we are all looking for some help.



The bad news about the sort title, I did alittle digging around under the Plex Media Server and either I missed it or did not recognize it but I could not find the code that generates the sort title from the title, which I think is the key to adding sort title to the metadata plugins.



Chris

[rwk]

I thought I would share my version of the NFO agent. It is based on the work by Harley, but I've tried to simplify the code a bit.

It is a secondary scanner so you need to use the Freebase agent as the primary agent. You can also use it in combination with the Personal Media agent - therefore, I don't see the need for an primary NFO agent...

It requires you to have a movie file with the name "moviename (year).ext" and a nfo file with the name "moviename (year).nfo" in the same folder. The nfo file should at least contain the following:

<?xml version="1.0" encoding="UTF-8"?><br />
<movie><br />
<title>Movie title</title><br />
</movie><br />

My nfo files also contain elements such as

<collection>Star Wars</collection><br />
<rating>8.1</rating><br />
<mpaa>PG-13</mpaa>

Also, if you included the IMDB id in the nfo like this

<id>tt1234567</id>

Plex will have an easier time matching your movies.

Looking at the code, it should be quite easy for anyone to modify it to their hearts desires. For example, because I keep my movies on separate external hard disks, which I attach when I want to watch a movie, I've included a

<disk>Disk 1</disk>

element in my nfo files. Then in the agent code I've changed this line

try: metadata.tagline=nfoXML.xpath('./tagline')[0].text

to this

try: metadata.tagline=nfoXML.xpath('./disk')[0].text

This will have Plex show "Disk 1" as the tagline, just below the title of the movie in the client (Info List view).

Put the attached bundle into this folder:

~/Library/Application Support/Plex Media Server/Plug-Ins

Have fun!

A question for the dev team:

Is it possible to define the Sort Title in this way? I.e. is it possible to update the metadata.sorttitle through agents? In general, it would be nice, if somewhere it could be learned, which metadata items (metadata.*) is exposed to the agents (a simple list of the items would do it)...

Hi and thank you for your NFO movies scanner, which I am testing!

I have installed the attachment bundle provided in your post and placed it in my Metadata Agent Setting as follows:

Rotten Tomatoes
NFO (Movie)
Freebase
Local Media Assets (Movies)
Wikipedia
...

Would you expect this to work?

The file name is: 7 Up Disc1 (1964)
It is in VIDEO_TS format.

The nfo file is in: 7 Up Disc1 (1964)/VIDEO_TS/7 Up Disc1 (1964).nfo

An example .nfo file follows:
<?xml version="1.0" encoding="UTF-8"?>

tt0058578
Seven Up / 7 Plus Seven
7.8
1964
68,797
92 minutes

Michael Apted

http://movies:81/covers/B000SAGGLO.jpg
Rated NR - Not rated. This movie has not been rated by the MPAA.
When Seven Up was broadcast in 1964, it broke new ground and became the first segment of a laborious and momentous labor of love: an engrossing chronicle of the lives of 14 students from varied social backgrounds, beginning when they were 7 years old. The first meeting with the group of wide-eyed, innocent children invites viewers to take a look at lives in the making; the follow-up film, 7 Plus Seven, checks in on them 7 years later.

Subtitles () Sound (English: Dolby Digital 2.0 Stereo) Full Screen 1.33:1
Documentary / Social & Cultural Documentaries


I am sure the nfo movie bundle is being called because it issues log messages to the PMS Plugin Logs log file.

My first and foremost priority is matching the correct movie (which is not happening). It is matching "A Fistful of Dollars".
I.e., it does not appear to be using the IMDB tt0058578 (which does exist and work when entered manually in the "Fix Incorrect Match" Search field.
(As I have hundreds of problems like this, and as I have the IMDB numbers in .nfo files for all my movies, I need to get this working.)

Can anyone see what I night be doing wrong? Can you confirm that tt-id's do in fact work for yourself or others?

Thanks for any help!

Dick

PS, I have noticed that the NFO (Movie) agent is only called on the initial scan of my movie files. This forces me to delete and re-add the source for every change I make during testing. Is there a way to get PMS re-run NFO (Movie) without deleting and re-adding the section every time? I tried the obvious things, Refresh Metadata, Fix Incorrect Match etc...

[MTI]

Hi and thank you for your NFO movies scanner, which I am testing!

I have installed the attachment bundle provided in your post and placed it in my Metadata Agent Setting as follows:
...

Would you expect this to work?

Well, if you set it like this, the data from RT takes precedence over your local NFO data. I personally set the NFO agent at highest priority. You can always include other agents further down to fill in any gaps.

The file name is: 7 Up Disc1 (1964)
It is in VIDEO_TS format.
The nfo file is in: 7 Up Disc1 (1964)/VIDEO_TS/7 Up Disc1 (1964).nfo
I am sure the nfo movie bundle is being called because it issues log messages to the PMS Plugin Logs log file.

I really have no experience with VIDEO_TS folders, but if you see in your logs that that agent is called, then you are probably fine.

My first and foremost priority is matching the correct movie (which is not happening). It is matching "A Fistful of Dollars".
I.e., it does not appear to be using the IMDB tt0058578 (which does exist and work when entered manually in the "Fix Incorrect Match" Search field.
(As I have hundreds of problems like this, and as I have the IMDB numbers in .nfo files for all my movies, I need to get this working.)

Can anyone see what I night be doing wrong? Can you confirm that tt-id's do in fact work for yourself or others?

I've seen this myself. I've quietly concluded that this functions erratically at best, and since we get absolutely no assistance here from the devs, I've given up on this part.
One thing that I've noticed is that in the PMS.log (or maybe in the IMDB agent log) it is printed on what basis the movie is matched. If I keep a NFO with a valid tt-id in it next to the movie file, the log will state that matching was based on ID. But the name still has to be exactly right. Try going to IMDB, and - using the ID - get the exact IMDB title for your movie. I did that for your movie, and it seem that you have to name it "Seven Up! (1964)". You can always use the element in the NFO file to get your custom title ("7 Up Disk1 (1964)") if you want.
So, in essence, filename is prioritized higher than IMDB-id when matching. I think that is unfortunate, since the IDs are unique and unambiguous - filenames are not...

PS, I have noticed that the NFO (Movie) agent is only called on the initial scan of my movie files. This forces me to delete and re-add the source for every change I make during testing. Is there a way to get PMS re-run NFO (Movie) without deleting and re-adding the section every time? I tried the obvious things, Refresh Metadata, Fix Incorrect Match etc...

To fully refresh a single movie, right/option-click it and select Refresh Metadata. To fully refresh an entire section, shift-click the update button in the lower right corner of the media manager.

Good luck! :)

[rwk]

Well, if you set it like this, the data from RT takes precedence over your local NFO data. I personally set the NFO agent at highest priority. You can always include other agents further down to fill in any gaps.



I really have no experience with VIDEO_TS folders, but if you see in your logs that that agent is called, then you are probably fine.



I've seen this myself. I've quietly concluded that this functions erratically at best, and since we get absolutely no assistance here from the devs, I've given up on this part.
One thing that I've noticed is that in the PMS.log (or maybe in the IMDB agent log) it is printed on what basis the movie is matched. If I keep a NFO with a valid tt-id in it next to the movie file, the log will state that matching was based on ID. But the name still has to be exactly right. Try going to IMDB, and - using the ID - get the exact IMDB title for your movie. I did that for your movie, and it seem that you have to name it "Seven Up! (1964)". You can always use the element in the NFO file to get your custom title ("7 Up Disk1 (1964)") if you want.
So, in essence, filename is prioritized higher than IMDB-id when matching. I think that is unfortunate, since the IDs are unique and unambiguous - filenames are not...




To fully refresh a single movie, right/option-click it and select Refresh Metadata. To fully refresh an entire section, shift-click the update button in the lower right corner of the media manager.

Good luck! :)

Thanks for your reply!

I put RottenTomatoes first because I want to use the rotten tomatoes ratings over any others. Using the Preferences for the rotten tomatoes scanner, I tell it to ignore everything but the rating. My intent was then to have the rest filled in from my nfo files (or subsequent scanners). I did try it with nfo scanner first in the list but it makes no difference to the imdb id issue, which, as I said, my primary goal here.

It would make no sense at all, and I would classify it as a bug, if movie title was used before tt-id to do the match. If you are certain of this, I will report it as a bug. Would anyone else be able to confirm this?

How do you know that tt-id is in fact being parsed in the first place?

It would defeat the whole purpose if I had to name my movies the way IMDB names them to get tt-id to work. Ie, that is precisely what I am trying to avoid. I would assume if exact IMDB movie names and year were used, that I would not need my .nfo files (with their included tags). It is because I have hundreds of mis-matches that I am trying to get tt-id to solve the problem. Am I misunderstanding your suggestion?

Thanks for the tip on refreshing the whole section!

Dick

[MTI]

It would make no sense at all, and I would classify it as a bug, if movie title was used before tt-id to do the match. If you are certain of this, I will report it as a bug. Would anyone else be able to confirm this?


How do you know that tt-id is in fact being parsed in the first place?


It would defeat the whole purpose if I had to name my movies the way IMDB names them to get tt-id to work. Ie, that is precisely what I am trying to avoid. I would assume if exact IMDB movie names and year were used, that I would not need my .nfo files (with their included tags). It is because I have hundreds of mis-matches that I am trying to get tt-id to solve the problem. Am I misunderstanding your suggestion?

Let's try to drag elan in here to take a look. Maybe he can clarify the issue. I think it is a bug, but maybe it is by design.

This is from my imdb agent log:

2011-01-20 16:45:53,413 (-4faed000) :  DEBUG (core) - Requesting http://plexmovie.plexapp.com/movie.12.unicode/guid/54/54a4aa2d27937313397e9302309278064b0e7c5f.xml<br />
2011-01-20 16:45:53,692 (-4faed000) :  INFO (core) - distance: 0<br />
2011-01-20 16:45:53,692 (-4faed000) :  INFO (core) - score penalty (used to determine if google is needed) = 0<br />
2011-01-20 16:45:53,692 (-4faed000) :  INFO (core) - PLEXMOVIE INFO RETRIEVAL: FINDBYID: True CACHE: True SEARCH_ENGINE: False<br />
2011-01-20 16:45:53,693 (-4faed000) :  INFO (core) - id=tt1608108 score=100 -> Best name being changed from Baby Einstein: World Animal Adventure to Baby Einstein: World Animal Adventure

This was what I was referring to previously. Note that it says "FINDBYID: True" - this, I guess, indicate that it found the tt-id in the nfo file. If you take a look in the IMDB/Freebase agent code, there is a specific function that scans for a tt-id in the a nfo file. However, it is not certain that this actually means that it in fact uses the id going forward or that the ID used was in fact derived from the nfo file and not from the (correct) file name.

But again, since we lack any kind of documentation, this is really just guestimates... :)

[bliblablo]

Great to see that this thread is not dead.

I have a problem with my movies. I’m german speaking and so i do have some files that have some special characters in them. These are äüö.

The nfo scanner seems not to be able to handle these files well. It’s not picking up informations on these movies.

Harley told me to use the .encode(“utf-8”) funktion but i can’t get it working with this.

So maybe someone here could help me out on this?

I have a sample nfo file here: Sample NFO



Thanks for any help

[MTI]

I have a sample nfo file here: [Sample NFO](http://dl.dropbox.com/u/809469/Die%20Fälscher.nfo)

OK, so I set up a test section in PMS. The section uses the stock Plex Movie Scanner media scanner, Freebase as the Primary media agent and Deutsch as language. The Freebase agent is set to only be supplemented by my NFO scanner. I add the source dir containing a file called Die Fälscher (2007).m4v and Die Fälscher (2007).nfo (your NFO file).

(To be sure that data was in fact coming from the nfo file, I edited the nfo file so that the title element is "Die Fälscher NFO").

This is what I get (attached image). It works fine for me, including correct unicode characters (ich bin aus Dänemark und meine Æ/Ø/Å characters work nicely as well).

BTW, your nfo file doesn't contain a tagline, so that is why you get that in English...



[bliblablo]

And you are using the latest version of the NFO agent from GitHub?

Hm… seems like I’m doing something wrong here… I guess I’ll try deleting all the scanned media and then replace my NFO agent again.

maybe that’s gonna help



Thanks for helping

[MTI]

I’m using my version of the agent, the one I posted recently

[ussalaska]

So has anyone done anymore digging into adding sort title to the nfo agent. I cannot find where the sort title is generated anywhere. In fact when looking at the metadata xml files after combining it is not stored there either. The only reference I can find to anything is to a titleSort string in the GUI building. I would love to add ‘sort title’ to the metadata agents, any developers with a hint would be greatly appreciated.



Chris

[MTI]

OK, so the questions are:

1. Is it possible to use metadata agents to insert the Sort title?
   
2. If not possible at the moment, can you change the framework to allow for it, please?
   
   
   
   

[elan]

1. Is it possible to use metadata agents to insert the Sort title?
2. If not possible at the moment, can you change the framework to allow for it, please?

1. No.

2. Possibly, in the future. Right now we're trying to wrap up the next release :)

[MTI]

2. Possibly, in the future. Right now we're trying to wrap up the next release :)

OK, thanks a lot for the update :)

[swordsandtequila]

I haven't looked at the framework in a little while so I don't know if a certain change has been implemente yet. But there is a change coming where I say in XBMCnfo what gents it pulls info from. Once this is in place, everything will work. For now though, you can put a thumb tag in your nfo and put a link to a poster you like, and XBMCnfo will find it.

Harley, any update on this? Just started using your nfo agent as a primary for a music/concert section, works well. Just don't have any option to choose local media assets as a secondary to pick up posters and thumbs. Can do it manually, but with a large library would rather PMM did the work. Appreciate all your effort, makes plex very versatile. Thanks.

[hrcolb0]

I’ll take a look at it this week. Haven’t really thought about it much to be honest.

[swordsandtequila]

Great, thank you. Cheers.