Login at Kodi Home

ccMatrix · 2009-02-12, 21:02

I noticed that that there are many movies where the TMDB scraper uses a different movie even if there is a movie with a perfectly matching name. For example I have a file called "M:\Movies\I Am Legend (2007).avi" but the scraper stores this as "Anchorman: The Legend of Ron Burgundy" even though there is a matching "I am Legend" entry in the XML. I had this for several movies now and it is frustrating to check and re-scan every movie manually (there are about 1000 movies in my library).

RDHoworth · 2009-02-12, 22:59

I have given up on XBMC for scraping and use Media Companion, which still uses IMDB and is also much faster. http://www.billyad2000.co.uk/

CrashX · 2009-02-12, 23:11

Please post debug log..

ccMatrix · 2009-02-12, 23:16

CrashX Wrote:Please post debug log..

How would that help? There is no crash. It just gets the wrong movie information for several files. I can give some more examples with "filename" -> "found movie information" if needed.

I guess that the most useful change would be in the API of themoviedb.org and improving their result ranking. If they return the perfect match first then XBMC will pick that up automatically. From my understanding of the scraper system of XBMC the sorting relies on the result of the webpage.

tslayer · 2009-02-12, 23:57

A debug log would help for sure. It shows what is being passed to tmdb website.

ccMatrix · 2009-02-13, 00:28

tslayer Wrote:A debug log would help for sure. It shows what is being passed to tmdb website.

I don't have the original debug log with the automatic scanning anymore. I did some tests when refreshing movies. Here a few examples:

Code:
File: M:\Movies\X-Men (2000).avi

Extracted name: x-men

TMDB Url: http://api.themoviedb.org/2.0/Movie.search?title=x%2dmen&api_key=57983e31fb435df4df77afb854740ea9

Automatically used name for movie: X-Men: The Last Stand

Actual name: X-Men

File: M:\Movies\The Kingdom (2007).avi

Extracted name: the kingdom

TMDB Url: http://api.themoviedb.org/2.0/Movie.search?title=the%20kingdom&api_key=57983e31fb435df4df77afb854740ea9

Automatically used name for movie: Indiana Jones and the Kingdom of the Crystal Skull

Actual name: The Kingdom

File: M:\Movies\You Kill Me (2007).avi

Extracted name: you kill me

TMDB Url: http://api.themoviedb.org/2.0/Movie.search?title=you%20kill%20me&api_key=57983e31fb435df4df77afb854740ea9

Automatically used name for movie: Kill Bill Vol. 1

Actual name: You Kill Me

It is always using the first result in the XML file returned by the TheMovieDB API.

I could clear the entire db and run a rescan but I hope this is enough to reproduce the issue.

tslayer · 2009-02-13, 00:31

Ya, it depends on the results returned by themoviedb. Not sure why they don't return exact matches first.

Maybe something to take up with them? Post to their forums?

ccMatrix · 2009-02-13, 00:33

tslayer Wrote:Ya, it depends on the results returned by themoviedb. Not sure why they don't return exact matches first.

Maybe something to take up with them? Post to their forums?

Ok, I'll post there. It should help all developers using the API so I hope they are open for the changes.

tslayer · 2009-02-13, 00:34

Even searching manually (without api) on their website produces same results.

jmarshall · 2009-02-13, 00:38

Ideally, though, XBMC shouldn't depend on the order of the results - we should be smarter than that and get the best match from the returned results.

afail · 2009-02-13, 02:59

im still using IMDB - i respect what TMDB is trying to do and for fanart it is excellent but its hard to go against IMDB, its a legend

Dudu.exe · 2009-02-13, 04:10

too bad i cant use IMDB in SVN ( it simply do not workanymore )

IMDB allways get the name wright! TMBD allmost every time get the wrong info!

ccMatrix · 2009-02-13, 04:33

Dudu.exe Wrote:too bad i cant use IMDB in SVN ( it simply do not workanymore )

IMDB allways get the name wright! TMBD allmost every time get the wrong info!

Well not almost but in about 99% of the cases. But sine TMDB is very new I think that people who are unsatisfied with it (like me in this case) should let the developers know so they can look into it.

I found that the issue is already listed in their issue tracker here: http://meticulo.lighthouseapp.com/projec...icket-82-2

This means the developer should be aware of it. Let's hope he takes action and has enough time to implement a great search algorithm.

cewan · 2009-02-13, 09:20

RDHoworth Wrote:I have given up on XBMC for scraping and use Media Companion, which still uses IMDB and is also much faster. http://www.billyad2000.co.uk/

Are you integrating the results from Media Companion in XBMC. If so, how?

johnny2 · 2009-02-13, 10:44

tslayer Wrote:Ya, it depends on the results returned by themoviedb. Not sure why they don't return exact matches first.

Maybe something to take up with them? Post to their forums?

Why? XBMC already had a perfectly working scraper. When Team XBMC replaced IMDB, they should have replaced it with a scraper that actually works. themoviedb is not ready for prime time. Seriously. It can't even tell the difference between You Kill Me and Kill Bill and you guys expect the end users to put up with this? It's almost as if the devs are intentionally trying to drive as many people as possible away from the project and to other alternatives.