Initial MyVideo Scan... 40% not found. (1 Viewer)

[sdalberg]

This isn't a super big deal, but I have about 200 movies that I have ripped to divx, and when I got to do the initial IMDB scan into the DB, its recognizing maybe 60% of them, but the 40% that its unable to find a perfect match for, a majority I'm just scratching my head. In a few cases that I understood, there was a sequal, or it was a remake. Those ones I understand.

Its the ones like 40 Year Old Virgin... When I scanned "40 year old virgin", which is the name of the file, it found 0 results, but when I typed "40 year" it found it right away, and with the title of "40 year old virgin". I'm guessing its just IMDB being crappy, but I just want someone else to verify that its sketchy.

TV-Server Version:
MediaPortal Version: 0.2.2.0
MediaPortal Skin: Blue-wide
Windows Version: XP SP2
CPU Type: P4 2.8Ghz
HDD: 80G WD
Memory: 512M
Motherboard: Intel
Motherboard Chipset: Intel
Motherboard Bios:
Video Card: Intel 845G
Video Card Driver:
Sound Card:
Sound Card AC3:
Sound Card Driver:
1. TV Card: HDHomeRun
1. TV Card Type: ATSC/QAM
1. TV Card Driver:
2. TV Card: TBC
2. TV Card Type:
2. TV Card Driver:
3. TV Card:
3. TV Card Type:
3. TV Card Driver:
4. TV Card:
4. TV Card Type:
4. TV Card Driver:
MPEG2 Video Codec:
MPEG2 Audio Codec:
Satelite/CableTV Provider:
HTPC Case:
Cooling:
Power Supply:
Remote: MS MCE v2
TV: Pioneer Elite HD-910
TV - HTPC Connection: Component (ideally)

 

[Paranoid Delusion]

IMDB seems to have changed it's database\search a few weeks ago, which caused problems as you described.

Unfortunately then had to do a limited name search as yourself, to get the right results.

As this does seem to be down to IMDB, not a lot we can do if the keywords do not bring up the film details.

If you do manually search\update, really worthwhile backing up the video database and the thumbs\video folder to save having to do again.

Happy Easter

 

[W_Z]

I noticed the same problem when scanning my collection of movies. I ran a packet sniffer and it appears that imdb is returning the proper results so I'd say there is a flaw in the parser or perhaps some imdb change broke it for certain movies.

 

[DalaNorth]

Hmm... Could you please give me some examples of movies that don't work and I'll try to fix it during the day.

Regards
DalaNorth

Btw, are you using 0.2.2.0 release version or an SVN version? I recently fixed the IMDB search in SVN, so if you are still using the release version, I suggest updating to the latest SVN (or waiting for the upcoming 0.2.3.0 release) and then trying again...

 

[DalaNorth]

Nevermind, I think I found the problem. IMDB previously listed directors as "Directed by:" but have changed it to "Director:". If an exact match for a movie was found, this messed up our parser. Even for movies you did find, director will probably be missing, right?

Regards
DalaNorth

 

[W_Z]

It scans much better now although it still seems to miss a fair chunk. It seems like if there are multiple direct matches (ie. Envy, Glory, Hercules, Hero, etc) then it doesn't take the first hit even though it is still right about 90% of the time in my library anyway. I have 'automatically select the nearest match' checked.

And yes... I did notice my previous scans were missing the director info.

Thanks for the fix!

 

[DalaNorth]

If I understand you correctly, the problem is that even though you have selected 'automatically select the nearest match', you still have to select which movie you want?

For the movies you have listed, this is probably because there is no single 'nearest match'. For instance, 'Envy'. IMDB finds the following:

Envy (2004)
Envy (1999)
Envy (2002)
Envy (1930)
Envy (2002) (V)
Envy (1917)
Envy (2005)

Which is the nearest match?

This is just the way the fuzzy matching works today. If two or more hits are 'equally close' to the title, the user must step in. Maybe we should add an option to automatically select the first hit, but that would belong in the improvment suggestion forum.

Regards
DalaNorth

 

[Maiklas]

Here are some problems with the fuzzy matching...

For "All or Nothing (2002)", which of the following does MediaPortal find as the nearest match?

1. All or Nothing (2002/I)
2. All or Nothing (2001)
3. All or Nothing (1998)
4. All or Nothing (2002) (TV)

Answer: 2. All or Nothing (2001).

For "All or Nothing (2002 I)", which is the nearest match?

Answer: 2. All or Nothing (2001).

I'm trying to add years to the titles to avoid having to do the disambiguation, so this is kind of annoying. Another frustrating case is "Jamón, jamón". That really should be enough, as if you type it into IMDB it goes straight to the movie page, but MediaPortal comes up with these two choices (only):

Freddy vs. Jason (2003)
Date Movie (2006)

MediaPortal does this even if the name is "Jamón, jamón (1992)", with the date.

If you type "Jamon, jamon", without the accents, into IMBD, then you can see where this is coming from:

Popular Titles (Displaying 2 Results)

1. Freddy vs. Jason (2003)

2. Date Movie (2006)
aka "Untitled Aaron Seltzer/Jason Friedberg Project" - USA (working title)

Titles (Exact Matches) (Displaying 1 Result)

1. Jamón, jamón (1992)

So, it looks like MediaPortal is stripping the accents, and if possible this should be changed to preserve the accents before the query to IMDB. I realize that WWW conventions pose a problem here.

Also note that when MediaPortal did the query for "Jamon, jamon" without the accents, it no doubt got back an "exact match" (listed above)... and totally ignored it! That is, MediaPortal found an exact match to the movie name, but then did not even include it in the list of candidates, much less call it a match without bothering the user.