IMDB.com (default info grabber)

[swe_rush]

true maybe to use moviexml only then.. will try that

[xanadu1979]

Sorry, I'm a bit new to this thread and it's a little confusing. Could I please get a summary? Which script or scripts is currently giving the best results at the moment?

Thanks for all your hard work, guys.

[gamejester]

For English language

movieXML.csscript will give the best resutls - if the movie is on the site (which most are now) and it has high res covers - you need to add this script yourself, the download is on page 3 of this thread.


imdb_txt+moviexml_gfx.csscript will be the want you to add as second in your search list (page 5 of this thread), it gets info from IMDB but a high res cover from movie.xml. This is what your seach will fall back to if the movie is not found on movie-xml

I usual tick 'automaticlly select the nearest match' on teh first pass then have a look through the results and delete any that are wrong, then I un-tick this for the second pass as this will give me a list from both searches of movies that are a clse match and i can manually select them.

[s7oneyuk]

i'm with gamrjester on this one. I'll do the imdb script that i promised with the exact matches 1st though - and will then probably then have this as the second script in my list with the moviexml one 1st. The data in MovieXML is getting more rich, the artwork rocks and the xml format that MovieXML is so much easier than the html scraping thats involved in the imdb script. I've also seen the work that gamejester has done on the artwork grabbing for MyMovies - and this looks like the way to go too. The old IMPA routine was hit and miss and the amazon pictures were just a waste of time. Looking forward to getting this sorted finally

[gamejester]

I have just found out that the way the fuzzylogic works on 'best match auto choose' is NOT as i thought.
It does not auto match the best film from the first search, but in fact what it thinks is the best match from all your searches. So if you have moviexml and IMDB in you list it will auto pick from either.
Usually it will pick the moviexml one, but I noticed on Pirates of the Caribbean: Dead Man's Chest it used IMDB.

[s7oneyuk]

bummer

[gamejester]

OK, just posted up my fixes and enhancements to the Movies core, please downloaded in and test it out

http://forum.team-mediaportal.com/ge...trieval-43405/

Basically it is abstracting the covers searches from the movie details retrival with a lot more cover search options and now 100% of the covers returned are relevant (if you do not se the Amazon cover search which is hopeless and only has low quality covers anyway)

[s7oneyuk]

Please find attached a revised IMDB script that will searh imdb are return a list of films. The changes in this version are that the "exact name" matches are listed 1st now - with the "popular titles" returned next. Also - anything with (VG) in the title is now dropped (VG = Video Game!). It will also attempt to get the coverart from the MovieXML site as these tend to be of a higher quality. Also fixed a bug that could prevent an accurate list of titles being returned. (incorrect logic in the counter incrementer)

I'm still siding with gamejester that the moviexml script is better/cleaner - however this might make a nice backup script if the movie you are searching for is not listed in moviexml. Also - the plot details can be a little off by using the imdb website as per my earlier post.

have fun!

[Paranoid Delusion]

Have just run this, using gamegeesters modified config, with your csscript as default, with download nearest match selected in config.

This was running 215 files (drive E), all with exact names taken from imdb, started scanning at 2117 ended 2138, 61 files saying no imdb info found at all, even on the manual find option. Original imdb_com scan had all bar 3, considering names of my files are actually taken from imdb including year in brackets, downside mainly crap covers, missing summary.

Cleared database and rescanned with moviexml, start 2144 end 2159 and got 43 queries, which allowed you to manually download your choice fine.

Then went back to imdb_com script, started scanning at 2210 end 2235, usual errors (arthur et les minimoys, should read Arthur and the Invisibles etc) mainly covers though.

This is turning into a real twat of a problem to get that perfect match, glad you are looking into it, any testing like this, let me know.

ps. "Charlottes Web" only imdb found, just disappointed my daughter LOL

[gamejester]

The more test data we can get the closer we will be to a 'perfect solution'

How are you finding the quality of the data frmo movie-xml? I personally find it very high now, but there is still a bit of a stigma from users who tried it 9 - 12 months ago and refuse to believe it has got better!

If I fix the problem (well might be by design have yet to check the code) in the core where it will ably _fuzzy matching logic to ALL your selected search engine results rather than pick the best match from your top listed search then that will ensure we get most files from move-xml.
The ones it cannot find it can then use IMDB, which has a larger database so should have these movies, it is these tricky movies we need a list of so the IMDB script can be tunned to pick them up and the fuzzy logic tunned to indetify them, can you supply a list of films that are NOT found and put by it the seach engine which fails to find it.

Can people perhaps post in a standard format, somthing like this (I am guessing a bit here on what your results are):

Engine: moviexml
movies: 215
Auto matched correct: 172
Auto matched wrong: 0
Manually duplicate available: 43
Not Found: 0
Time Take: 15 mins

Engine: IMDB_movieXML
movies: 215
Auto matched correct:
Auto matched wrong:
Manually duplicate available:
Not Found: 61
Time Taken: 21 mins

Engine: IMDB_com
movies: 215
Auto matched correct:
Auto matched wrong:
Manually duplicate available:
Not Found:
Time Taken:

Engine: IMDB
movies: 215
Auto matched correct:
Auto matched wrong:
Manually duplicate available:
Not Found:
Time Taken:

This will show us if we are getting closer, + advertise to others which the good search engines are; with examples of movies that fail we might be able to tweak things further.

Once we have acurate searches which can retrive the correct movie list in manual mode every time then maybe the fuzzy logic auto chooser can be tunned to find it better.