Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[glynor]

While working through the couple of wonderful new features Matt added to Carnac to handle imports from SageTV, I discovered a little oddness.

Matt asked this:

If you have specific filenames Carnac goofs on, please post them.

I didn't have any of the season / episode Sage filenames when testing, for example.  I just used the one you had posted.

You got it, Matt.  Here goes...

[glynor]

I had six new recordings from Sage tonight.  One started recording right around the time that I actually did the upgrade to tonight's build of MC, though, so we're going to leave that one off the list for now.  I can't be sure that its "miss" wasn't because it was imported using the older build, so...

But, of the rest, I got two "misses" and one other one that I just want to mention (it was as good as could be expected, though).  For each of these, I'm just going to list out what the filename was, and then what Carnac gave me by itself, verbatim.  If I get any Media Sub Type misses in the future, I'll include the Duration tag from the sidecar file (and anything else you tell me to), but that didn't happen tonight.

Filename: T:\recordings\PersonofInterest-S01E07-Witness-6574312-0.mpg
[Series]=PersonofInterest
[Name]=Witness-6574312-0
[Media Sub Type], [Season], and [Episode] were correctly parsed.
Note: I'm not sure why, but it did not parse the [Series] with a vanilla FixSpacing() pass, eg. "Personof Interest".  There were no spaces at all included, even before the I.  And, it didn't remove the junk from the tail of [Name].

Filename: T:\recordings\TheColbertReport-S07E140-6507020-0.ts
[Season]=blank
[Episode]=blank
[Name]=S 07 E 140
[Media Sub Type] and [Series] were correctly parsed.
Note: It is obvious why this happened, but you should be aware that many shows, particularly things like The Colbert Report and other "daily" or "news" shows, show up like this with no Episode Title, but with Series and Episode numbers encoded.  I'm not sure what I'd do with the [Name] tag on these if I were you.  I, personally, tag them with "No Title" or leave them alone.  I usually watch-and-delete these anyway, so I don't care really, as long as I can find them.  [Series] and [Media Sub Type] are the important things to grab, and Carnac got those right.

Filename: T:\recordings\TheDailyShowWithJonStewart-6507003-0.ts
[Media Sub Type] was correctly parsed
[Name]=The Daily Show With Jon Stewart
Note: This one is the one that I figure is the best you can do.  I just wanted to let you know that sometimes you get this.  I think it depends on what the guide data Sage has at the time of recording gives them.  Obviously, they didn't know much about this airing from the Guide data.  Often these end up being repeats when I watch them, but sometimes they're new ones.  There's nothing you can do about this and I think what MC did was the right call.  Plus, Movies' file names look just like this one, and had it been a Movie, it would have been right (assuming [Media Sub Type] got detected properly).

And, for the record, these parsed perfectly in every way:
T:\recordings\TheOffice-S08E06-Doomsday-6502326-0.ts
T:\recordings\TheBigBangTheory-S05E08-TheIsolationPermutation-6502377-0.ts

And, just FYI, the last one was this:
Filename: T:\recordings\Community-S03E06-AdvancedGay-6496501-0.mpg
[Name]: AdvancedGay-6496501-0
[Media Sub Type], [Series], [Season], and [Episode] all parsed correctly.
Note: I'm not absolutely positive which build of MC actually imported this one.  I think it was probably imported by the new build (I'm fairly sure I finished before that one started recording), but I can't swear to it.  It was close.  Either way, again, simply another FixSpacing() and useless junk issue at the tail.

[Matt]

First, thanks for the samples.  I'll work through them in one of the next couple builds.

At the risk of going off topic, the complexity with splitting a Sage name with no spaces and mixed capitalization into something good is a nice example of why Carnac isn't just a list or regular expressions a user can edit.  This was requested by a few users, but some of the tasks Carnac performs aren't really possible with regular expressions alone (no, this isn't a challenge MrC!).

[Matt]

Filename: T:\recordings\PersonofInterest-S01E07-Witness-6574312-0.mpg

I didn't know Sage used .mpg.  Carnac was only looking for .ts.  Fixed next build.

Filename: T:\recordings\TheColbertReport-S07E140-6507020-0.ts

Fixed next build.

Filename: T:\recordings\Community-S03E06-AdvancedGay-6496501-0.mpg

That's the same .mpg issue.

Just let me know if you find anything else.

Thanks.

[glynor]

I thought the MPG thing might be it.

You pretty much have to assume that Sage might write a wide variety of file types.  Mine are basically all TS files (from my HD-PVR) or MPG files (from my "regular" tuners).  But, since Sage has an Auto-Compress When Finished Recording feature that can take those TS files and recompress them to smaller H.264 MKV files, you could have other users with a variety of different File Types coming from Sage.  I don't use it (it did a crappy job, unsynced audio and video a bunch of times, and used poor quality presets for the recompression), but a bunch of people might have used it out there.

I'd guess you'd need to look for these at a minimum: TS, MPG, AVI, MKV, and MP4.  That should cover most of it, I think, but someone else might want to chime in on that.

[nwboater]

My SageTV files from the HDPVR are almost entirely .ts. Out of a few hundred recordings only five are .mpg and they are from the same UnderseaWorldofJacquesCousteau Series. No idea why they are mpgs though.

Rod

[glynor]

These all seemed fixed.  Amazing job.  The spellcheck addition to the FixSpacing() command in Carnac is fantastic.  I haven't seen it fail once yet tonight.

I'm going to test a movie recording shortly to test the [Media Sub Type] setting, but that worked pretty well before, so unless something messed up...

[glynor]

I'm going to test a movie recording shortly to test the [Media Sub Type] setting, but that worked pretty well before, so unless something messed up...

Unfortunately, this seems to be broken.

I recorded a few movies last night and today.  All came in marked as TV Shows.  All were fairly clearly movies from the duration and whatnot.

[MrC]

... a nice example of why Carnac isn't just a list or regular expressions a user can edit.  This was requested by a few users, but some of the tasks Carnac performs aren't really possible with regular expressions alone (no, this isn't a challenge MrC!).

No challenge taken, but thanks for the nice ribbing. :-)  The suggestion was just one component of a set of tools.

I tried using Carnac with Fill Properties for some downloaded/recorded streams whose names are as follows:

  Program Name - DJ Name (some_digit)

It doesn't find anything.  The existing Template solution doesn't work, due to the trailing parenthesized digit (or for that matter, any other junk anywhere that should be ignored).  There's no way to specify via the Template patterns to be ignored.

Would it be possible to add something like the following picture, which allows entering an RE and a list of ordered properties which are assigned from specified captures in the RE?

[Matt]

Would it be possible to add something like the following picture, which allows entering an RE and a list of ordered properties which are assigned from specified captures in the RE?

I think that's a pretty good idea.

I wonder, should the expression be run against the entire filename, or just the name part (without path / folders)?

[Matt]

I recorded a few movies last night and today.  All came in marked as TV Shows.  All were fairly clearly movies from the duration and whatnot.

Could you post a full filename?  Also, did we get the duration correct?  Finally, what's in the 'Compression' column?

Thanks.

[nwboater]

All my old Sage Recorded movies are listed as 'Shows'. Let me know if you want any more info here.

Rod

[MrC]

I think that's a pretty good idea.

I wonder, should the expression be run against the entire filename, or just the name part (without path / folders)?

I'm working my way up to "Great Idea".

Perhaps a checkbox that allows either, defaulting to filename only.  I have cases where the path contains useful info, not contained in the filename.

If the dialog becomes too cluttered, perhaps a pulldown could allow selection of either Template or Regular Expression mode (this is how Directory Opus reduces clutter in its Rename dialog).

[glynor]

I think that's a pretty good idea.

I wonder, should the expression be run against the entire filename, or just the name part (without path / folders)?

I'd like to be able to use it against the entire file path.

[glynor]

Unfortunately, this seems to be broken.

I recorded a few movies last night and today.  All came in marked as TV Shows.  All were fairly clearly movies from the duration and whatnot.

This is still true.

Seems like ANY file of sufficient length that goes through Carnac's SageTV filename decode logic, gets marked as a TV Show.  I've recorded five different movies over the past 48 hours, and all came in as [Media Sub Type]=TV Show.

Interestingly, one of my actual "TV Shows" came in with [Media Sub Type] blank.  This was a recording of the BBC World News, that happened to be only 17 minutes long (those aren't always 30-ish minutes minimum, and they are sometimes longer and sometimes shorter).

[Matt]

I'm still hoping for full filenames for the misses:
http://yabb.jriver.com/interact/index.php?topic=67469.msg453728#msg453728

Thanks.

[glynor]

Yep.  I'll get them for you.  Sorry, missed that post.

[glynor]

Here's your requested filenames:

T:\recordings\Legend-6618033-0.ts
T:\recordings\Cyrus-6601813-0.ts
T:\recordings\BacktotheFuture-6531793-0.ts
T:\recordings\BacktotheFuturePartII-6601476-0.ts
T:\recordings\BacktotheFuturePartIII-6531798-0.ts
T:\recordings\BatteriesnotIncluded-6601475-0.ts
T:\recordings\WheretheWildThingsAre-6584059-0.ts

All were detected as TV Shows instead of Movies.  Also, this one:
T:\recordings\BatteriesnotIncluded-6601475-0.ts
Was parsed as [Name]=Batteriesnot Included

No huge deal.  All the other names parsed perfectly.  All were movie-length recordings, 90 min+, recorded from my HD-PVR just like anything else I record.  Well, all except for one of the Back to the Futures, but more on that in a bit.  MC used to detect these types of recordings as Movies correctly.

Hmmm... If it isn't just the Carnac SageTV filename parsing stuff, I wonder if it might have to do with the "don't add them to the bad files list unless the date modified is old" change?  Maybe AutoImport is seeing them when they are only 25 minutes long or something?

One last thing though... My one recording, of Back to the Future Part II actually got spit into two pieces.  I'm not sure why, it doesn't happen often.  Perhaps my cable went out for a bit, or my HD-PVR momentarily coughed up a lung?  Anyway, MC imported both files.  They were:
T:\recordings\BacktotheFuturePartII-6601476-0.ts
T:\recordings\BacktotheFuturePartII-6601476-1.ts
(and, if there was ever a third or fourth hunk, they'd continue to be named sequentially like that.)

MC imported both of these with an identical (and correct) [Name] tag.  In an absolutely perfect implementation, it would import any "-0" copy just like it does now, but import any -1+ version as (1), (2), (3), (4), etc (or maybe use (2) as the base to make it sensical to people who don't understand counting from 0).

As I mentioned, I just delete them.  But in some cases, you might want to use it anyway.  If, for example, you know you only missed a split second.  Or maybe you know it happened during a commercial break or something (from flipping channels).  So, both of these two "pieces" were shorter than you'd expect for a movie (42 min for one piece, and just over an hour for the other)... So, it is reasonable that they came in as TV Shows, but not any of the rest of them above.

In any case, you don't want all the -0 ones (which is most of them) to show anything extra, but if "-1" versions could be differentiated somehow, that'd be good.

[glynor]

Also, this one:
T:\recordings\BBCWorldNews-6591954-0.mpg

Imported as [Media Sub Type]=blank.
The Duration was only 9:05 though, so I'm not sure what you should have done.  I think maybe blank is the best choice, really.

[Matt]

Please try again with tonight's build:
Changed: Carnac assumes that a file with a television recording extension that's over 70 minutes is a movie instead of a television show.
Changed: Carnac deals with the uncapitalized word "not" from SageTV naming better.

I'm a little nervous about the first one, because some shows are two hours.  If you record with MC, a recorded movie has always been a show and not a movie.  I'm not sure what's best here.

[glynor]

Please try again with tonight's build:
Changed: Carnac assumes that a file with a television recording extension that's over 70 minutes is a movie instead of a television show.
Changed: Carnac deals with the uncapitalized word "not" from SageTV naming better.

I'm a little nervous about the first one, because some shows are two hours.  If you record with MC, a recorded movie has always been a show and not a movie.  I'm not sure what's best here.

Ahh...  My imports always worked that way with the old auto-detection system in MC16.

For me, that works better.  The list of new Movies is WAY SMALLER than the list of new TV episodes.  So, if a Movie is mislabeled as a TV Show, it can be difficult to spot.  If a two hour episode of a TV Show gets mislabeled as a Movie that:

1. Makes sense.  A two-hour finale episode of a modern TV Show really is a movie.
2. Is easier to spot and fix in a smaller list.

But, if people complain, feel free to switch it back.  So long as I know what the behavior is, this is small potatoes to fix manually.  I could even fix it automatically with a Tag On Import rule.  Of course, if someone doesn't like the behavior, they could also fix it easily with a Tag On Import rule (and a simpler one at that, since you'd need no [Duration] is greater than calculation).

[nwboater]

Matt,

I was about to give you a list of some of the Movies that Carnac called TV Shows then I saw this info on tonight's build. It's way better now. The previous build caught a lot of mpg files and re-labelled them as Movies. But I don't think it had any of the .ts files from Sage correctly.

We do have a few Shows that are longer than 1 hour now under movies, but as glynor said that's a small amount compared to the mis-labelled Movies.

I do wonder about the sports nuts though. They could have a lot of recorded games over 1 hour. Should you perhaps give an option on this >70 minutes being called a Movie? Or are you trying to keep Carnac with 0 user settings?

Thanks for all the improvements Matt - It's really getting there now!

Rod

[JimH]

Thanks for all the improvements Matt - It's really getting there now!

Thanks, Rod.  You deserve a lot of the credit.  You've been extremely patient and reported the problems you found very well.  Thanks Mrs. Rod, too.  Tell her we plan on exceeding her expectations.

[nwboater]

......... Thanks Mrs. Rod, too.  Tell her we plan on exceeding her expectations.

This is Sandra (Mrs. Rod). Just wanted to thank all of you for the good programming that's developing. I reap all the benefits, and being a movie-lover, really appreciate your efforts. Rod & I cocoon most evenings and enjoy the latest accomplishments. Your WAF is very high!

[JimH]

Hi Sandra!  Thanks for letting us know!

[bspachman]

Some recordings that aren't coming through perfectly for me:

V:\Television\Fringe.S04E06.And.Those.We.Left.Behind.avi
V:\Television\Chuck.S05E03.Chuck.Versus.The.Frosted.Tips.avi

Series, Season, Episode & Media SubType were all correctly identified, however, NAME came through as:
And.Those.We.Left.Behind
Chuck.Versus.the.Frosted.Tips

Is there a way to have Carnac switch the punctuation for spaces? It would save a trip to the "Clean File Properties" dialog...

Also (perhaps related), running the new FixSpacing() function on the NAMEs with periods simply adds a space after each period. I'm not sure that FixSpacing is designed to solve the above issue, but since it was new, I would try it out

Thanks,
brad

[Hendrik]

I also had one weird case of Carnac misses today.

Stargate.SG-1.S08E03.Lockdown.720p.DD.5.1.h.264.mkv

Carnac filled the fields as following (for all eps in that season)
Season: 8
Episode: 1
Name: Stargate.SG-1.S08E1

Funny enough, following file (and the whole season) worked 100% fine:
Stargate.SG-1.S09E03.Origin.720p.DD5.1.h.264.mkv

I manually fixed it since it was only a couple of files, but it was still rather odd.
The only difference seems to be "DD.5.1" vs "DD5.1"

While i was manually fixing this, i noticed that "Clean File Properties" would greatly benefit from an option to clean leading zeros of the Episode/Season numbers. Would make a great addition, since i couldn't figure out how to do this automated otherwise.

[Matt]

I really appreciate the examples.

Next build:
Changed: Carnac improvements when dealing with values that use periods instead of spaces to delimit words.
Changed: Numeric season and episode numbers will be padded to two digits (using leading zeros as necessary) when building a filename with 'Rename, Move, & Copy Files...'.
NEW: Added option to 'Remove leading zeros' from the Clean File Properties tool (useful for cleaning up season and episode numbers).

[darichman]

Next build:
Changed: Carnac improvements when dealing with values that use periods instead of spaces to delimit words.
Changed: Numeric season and episode numbers will be padded to two digits (using leading zeros as necessary) when building a filename with 'Rename, Move, & Copy Files...'.
NEW: Added option to 'Remove leading zeros' from the Clean File Properties tool (useful for cleaning up season and episode numbers).

Brilliant. I like where this is going.

[glynor]

Agreed.  Both changes are very nice.

[aaronshaw]

I like the upcoming changes, solves my previous question

"I was just wondering if there's an update coming (or a workaround) so that Carnac will pad series and episode numbers ie: s9 to s09. Or maybe allow default Library fields to be customised with functions?

I currently have to run through a convoluted series of steps to import and rename my files so that these fields are padded."

[MrC]

"I was just wondering if there's an update coming (or a workaround) so that Carnac will pad series and episode numbers ie: s9 to s09. Or maybe allow default Library fields to be customised with functions?

I currently have to run through a convoluted series of steps to import and rename my files so that these fields are padded."

Version 17.0.34 provides:

4. Changed: Numeric season and episode numbers will be padded to two digits (using leading zeros as necessary) when building a filename with 'Rename, Move, & Copy Files...'.

[aaronshaw]

Version 17.0.34 provides:

4. Changed: Numeric season and episode numbers will be padded to two digits (using leading zeros as necessary) when building a filename with 'Rename, Move, & Copy Files...'.

That doesn't fill the library field Season/Episode with the padded numbers though, that's what I would love to see.

[MrC]

And it won't.  They are integer values, and you can use PadNumber() if you want views to pad an integer.

Can you describe your end goal?

[aaronshaw]

And it won't.  They are integer values, and you can use PadNumber() if you want views to pad an integer.

Can you describe your end goal?

ok, so I prefer my Series and Episode feilds to be 01 02 03 etc, instead of 1 2 3, for continuity with the file names.
I currently use a custom function field called EpisodePad with PadNumber([Episode],2) to get around this. Not a big deal, I was just curious if Carnac would be getting an option to write padded numbers.

[MrC]

Customize your theater view to display these fields as you'd like:

For the File Info panel (Options > Theater View > Appearance > Customize file info panel...) and go to the Video section, and locate the Season Field.  Now, change the Type in the right to Expression, and paste this in the Value area:

   PadNumber([Season],2)

This gives you 2-digit padding for your Season.  Do likewise for other fields you want padded.

You can do similarly to the categories displayed too (Items To Show).  Select the item you want to customize, and in the right area, use an expression to output what you want.

[glynor]

ok, so I prefer my Series and Episode feilds to be 01 02 03 etc, instead of 1 2 3, for continuity with the file names.
I currently use a custom function field called EpisodePad with PadNumber([Episode],2) to get around this. Not a big deal, I was just curious if Carnac would be getting an option to write padded numbers.

Also, FYI: You don't need to use the PadNumber() Function when you do a Rename, Move, and Copy on files.  It pads those fields out to two digits for you automatically.

Since the filesystem is the part that is dumb and displays 12 before 9, that's where I really care about padding, and it does it for you.

[MrC]

Also, FYI: You don't need to use the PadNumber() Function when you do a Rename, Move, and Copy on files.  It pads those fields out to two digits for you automatically.

See reply #31.

You had a rough yesterday...

[glynor]

See reply #31.

You had a rough yesterday...

That was this morning at 8:30.

There was too much blood in my caffeine system. 

[JustinChase]

That doesn't fill the library field Season/Episode with the padded numbers though, that's what I would love to see.

I also (thought I) wanted this a while back, but I really just need/want it for systems that don't understand single-digit 1-9 (windows explorer).  I have changed all of my seasons and episodes to single digit numbers for 1-9 in MC, and just use move, rename... to rename the filenames with leading zeros for the benifit of other programs that need the leading zeros to sort properly.

This has been a wonderful change for me, since I can now stop fighting MC, which will only use 1-9 internally.  It works perfectly in MC, and also the filenames are now 'proper' outside MC, which is all I could (and did) ask for.  {Thanks again Matt }

I suggest you may wish reconsider your willingness to just accept 1-9 inside MC, since it works flawlessly, and automatically

There was too much blood in my caffeine system. 

   I couldn't agree more; I'm off for another cup myself

[glynor]

I found a new one today!

Filename: M:\Incoming\Star Trek Enterprise.Season 3.HDTV 720p\Star.Trek.Enterprise.-.S03E01.(HDTV.720p.x264.English.AC3-5.1).HDNet.mkv
[Series]=Star Trek Enterprise - S03E01 (HDTV English AC3
[Season]=5
[Episode]=1
[Name]=) HDNet
[Media Sub Type] was correctly parsed.

This same thing applied for all of them (so M:\Incoming\Star Trek Enterprise.Season 3.HDTV 720p\Star.Trek.Enterprise.-.S03E02.(HDTV.720p.x264.English.AC3-5.1).HDNet.mkv suffered the same fate).  Seems like it is probably the () in the filename that tripped it up, though I'm not sure.

No comment on if that poor excuse for a Star Trek franchise deserves to be imported at all, properly or not.

[struct]

Not really a carnac miss, more to do with importing process....

When doing a move/rename of movies, it would be good if subs directories were also moved by default.

thanks

[lboregard]

hi, i imported my collection of movies residing in a nas and have these comments

- i dont like the fact that cover art for movies and series episodes are kept in the same folder as the media file. i would prefer an option as the one existing for cover art audio mode, which actually also works for series covers. is it possible to have this changed ?

- scene releases sometimes contain samples of the media, carnac is importing these ... is it possible that carnac ignores files with "sample" in its filename ?

- carnac incorrectly imported some of my movies
\\nas\films\blurip\9\9.2009.Blu-ray.DTS.720p.x264.HDBRiSe.mkv ... was imported with star trek 2009 description and cover art
\\nas\films\blurip\The Chronicles Of Narnia The Lion The Witch And The Wardrobe\1920-narnia.mkv ... is picked up as 1920-narnia.mkv, no metadata downloaded
\\nas\films\dvd\A Nos Amours\VIDEO_TS\VIDEO_TS.ifo ... is correctly labeled as A Nos Amours (1956) but gets the description of Akira Kurosawa's Seven Samurai ?

is there a suggested folder structure so that movies above are correctly scanned and corresponding metadata downloaded ?

- i deleted a library entry, renamed the folder on disk but auto-import would not pick it up again ... what could be causing this

i like the playback features of mc17 and carnac looks promising, although some improvements are still needed

 

[lboregard]

any comments ? anyone ?

[MrC]

I can't help much with the how-Carnac-works area.

But to your immediate question re: importing the titles, are all your films housed as:

  \\nas\films\{media}\{title}

where {media} is blurip, dvd, etc. and {title} is the title of the movie?  If so, rules can be created so auto-import will assign the title.

[lboregard]

  \\nas\films\{media}\{title}

where {media} is blurip, dvd, etc. and {title} is the title of the movie?  If so, rules can be created so auto-import will assign the title.

thanks MrC .. that's right ... that's the exact structure.
where may i find info on how to set the rules so that auto-import assigns the correct title ?
thanks in advance

[glynor]

- carnac incorrectly imported some of my movies
\\nas\films\blurip\9\9.2009.Blu-ray.DTS.720p.x264.HDBRiSe.mkv ... was imported with star trek 2009 description and cover art

I can't confirm this here.

I just made a copy of that same movie from my collection on my desktop.  Renamed it to match yours, and put it in a folder called "9".  Then I copied it into my main "incoming" watched folder.  It imported as:

Filename: M:\Incoming\9\9.2009.Blu-ray.DTS.720p.x264.HDBRiSe.mkv
[Name]=9 2009 Blu
[Media Sub Type] and [Year] were correctly parsed.


In my case, that's how it came in.  It did not do automatic cloud metadata lookup, so there was no additional metadata pulled in (description, actors, etc).  That's a pretty tough filename to parse, though, so I don't think that's an unreasonable result.  It would be nice if it knew the terms "BluRay" and "Blu-Ray" I suppose, but it didn't mess it up for me the way you're seeing.

However, I can see how it could make that mistake.  When I then did that metadata lookup manually, I got this as the top result in the search box:



That's just a web search though, and TMDB's engine is generating that result.  Not surprising considering the strange search string.

But, in either case, Carnac/Auto-Metadata Lookup isn't supposed to change the [Name] tag automatically for files where [Media Sub Type]=Movie (that was worked out in the initial builds).  Are you sure you didn't do a manual lookup on that file (maybe as part of a batch group)?  If you select more than one file and do the Get Movie & TV Info command, it accepts the first result, so you have to check the files for tagging sanity before you submit them or you can get unpredictable results.  A manual lookup will overwrite [Name] with the top matching result.  The automatic system isn't supposed to, and like I said, it did nothing for mine.

In my case anyway, it worked as designed (except it didn't parse the filename, but like I said, that's a tough one).  A single-digit number is, in aggregate, not likely to be the only significant part of the name of the movie given that filename.

\\nas\films\blurip\The Chronicles Of Narnia The Lion The Witch And The Wardrobe\1920-narnia.mkv ... is picked up as 1920-narnia.mkv, no metadata downloaded

Confirmed (well, in my case it did capitalize it as 1920-Narnia but maybe you just typed lazily).  I'm not sure what it should have done.  Of course, it would be nice if it could get that one... But how does it know that in this particular case the filename isn't useful and that the folder name is?  What if it was the other way around?

I think filename has to trump folder name if the filename could possibly be valid data.  Now, if the filename is a known quantity (you can identify all of the parts from the "usual suspects" that might be tagged), then you might be able to deduce that the folder name is the real value part.  But that one could be anything.  It is unpredictable and only obvious to a human's eye.  A failure to "nothing" is better than a false positive that gets renamed as something else (like your top example).

\\nas\films\dvd\A Nos Amours\VIDEO_TS\VIDEO_TS.ifo ... is correctly labeled as A Nos Amours (1956) but gets the description of Akira Kurosawa's Seven Samurai ?

I don't have any IFOs to test this.

But, I reused my narnia test MKV and renamed it using your naming convention.  It parsed perfectly and got the right metadata (or at least I assume it is the right one), here's the description it got:

A veteran samurai, who has fallen on hard times, answers a village's request for protection from bandits. He gathers 6 other samurai to help him, and they teach the townspeople how to defend themselves, and they supply the samurai with three small meals a day. The film culminates in a giant battle when 40 bandits attack the village.

That matches the top hit in tmdb, so that's what it got.  Not sure if having an actual ISO would have made any difference.  Perhaps TMDB was down for a bit when you happened to import that one?  If so, it fails back to Rotten Tomatoes and Wikipedia which are less reliably indexed (IMHO).  

The top Rotten Tomatoes hit in the manual search list, for me, is a different (french) movie called "A nos amours (To Our Loves) (Suzanne)".  There are also about a bazillion other TMDB hits for that search, including a bunch of general "love" movies further down on the list, and a bunch of other Samurai-related movies.

I used this filename:
M:\Incoming\dvd\A Nos Amours\VIDEO_TS\VIDEO_TS.dvd

EDIT: That filename was a copy-pasta error because of MC's new IFO parsing scheme.  I just repeated it to confirm.  I tested with this actual filename:
M:\Incoming\films\dvd\A Nos Amours\VIDEO_TS\VIDEO_TS.ifo

MC got:
[Name]=A Nos Amours
[Media Sub Type]=Movie
[Year]=1956
[Genre]=Action; Drama; Foreign; Indie
[Keywords], [Actors], [Director], and [Description] also filled.
[Filename]=M:\Incoming\films\dvd\A Nos Amours\VIDEO_TS\VIDEO_TS.dvd;1 (hence the copy-pasta error).

Not sure what happened there on your end.  Maybe using a real IFO would be different?  But, again, was it your automatic metadata lookup that was wrong, or did you do a batch lookup?

[glynor]

Regarding the first one... Just to make absolutely sure, I repeated the test with this filename:
M:\Incoming\films\blurip\9\9.2009.Blu-ray.DTS.720p.x264.HDBRiSe.mkv

In case the rest of your directory structure could be affecting the lookup somehow, but it didn't.  Same results as before.

If I do a manual lookup on it and just click OK (or part of a batch, which is the same thing), then it does get renamed Star Trek.  But I have to do that manually.

[lboregard]

... But, in either case, Carnac/Auto-Metadata Lookup isn't supposed to change the [Name] tag automatically for files where [Media Sub Type]=Movie (that was worked out in the initial builds).  Are you sure you didn't do a manual lookup on that file (maybe as part of a batch group)?  If you select more than one file and do the Get Movie & TV Info command, it accepts the first result, so you have to check the files for tagging sanity before you submit them or you can get unpredictable results.  A manual lookup will overwrite [Name] with the top matching result.  The automatic system isn't supposed to, and like I said, it did nothing for mine.

In my case anyway, it worked as designed (except it didn't parse the filename, but like I said, that's a tough one).  A single-digit number is, in aggregate, not likely to be the only significant part of the name of the movie given that filename.

agree ... "9" is an uncommon name for a movie .. the filename structure should exist more frequently though (as in {title}.{year}.{rip}.... and such).

... But, I reused my narnia test MKV and renamed it using your naming convention.  It parsed perfectly and got the right metadata (or at least I assume it is the right one), here's the description it got:

"A veteran samurai, who has fallen on hard times, answers a village's request for protection from bandits. He gathers 6 other samurai to help him, and they teach the townspeople how to defend themselves, and they supply the samurai with three small meals a day. The film culminates in a giant battle when 40 bandits attack the village."

glynor, that's the wrong description for the movie ... actually, "A Nos Amours" is from 1982 and has nothing to do with samurais ... but, i got the same top hit from tmdb, so essentially the metadata lookup is working "correctly".

Not sure what happened there on your end.  Maybe using a real IFO would be different?  But, again, was it your automatic metadata lookup that was wrong, or did you do a batch lookup?

what i did was ... configure folders to auto-import (tagged them with media sub-type) and the let it run ... i started from zero items in the library to approx 2000 movies .. not sure if that's what you mean by batch lookup, but i never did any manual lookup.

there are two key components at play here:
1-how to infer the movie name from file/folder name
granted, this is the hardest part ... some other htpc applications will lookup an nfo file located in the same folder as the main movie, in order to get an exact match. or allow for a custom regex (i assume that's what you're using internally) to parse file/folder

2-metadata lookup
from what i understand, carnac is querying tmdb, rotten and the wikipedia ... are all of them api-based ? i dont see why imdb isn't used as the main service, licensing issues and/or not wanting to do html scraping perhaps ?

i hope you could address some of these issue to make importing more reliable ... i like the way carnac works, maybe sometimes it needs a little help

[glynor]

Carnac isn't AI.  It is a pretty darn good parlor trick based on some pattern recognition and a spell checking dictionary, but it isn't a miracle worker.  There are tons of possible wacky filenaming patterns, and there's only so much it can do if there are multiple movies with the exact same name (or a potentially ambiguous name).  Again, I'd rather it fail and get no information occasionally than have a bunch of false positives.

I'll have to check further, but I did test using those exact filenames, and you saw my results.  Not sure how yours ended up renamed to Star Trek in that once instance unless you did a manual lookup (by which I mean: right click on the file(s) and choose Get Movie and TV Info, not just run Auto-Import).

However, MC does have the ability to also use RegEx to change and modify Carnac's results.  Do a search on the forum for "Tag on Import".  MrC, who has commented here, is our resident RegEx expert.  If he can't figure out how to parse something, then it almost certainly can't be done.

Also....

2-metadata lookup
from what i understand, carnac is querying tmdb, rotten and the wikipedia ... are all of them api-based ? i dont see why imdb isn't used as the main service, licensing issues and/or not wanting to do html scraping perhaps ?

IMDB wants a ton of money.  Jim asked.
Most open source applications that use it just don't follow the rules and scrape the site "illegitimately".  JRiver isn't going to do that because they're a company with a checkbook that can be sued.  TMDB is really pretty good though.  As good as IMDB?  No.  Not that good, but they provide open API access with reasonable terms.