Web search is returning correct result but not the API

[anidotnet]

I was searching subtitles for the following movie file
(download at superseeds.org) Blitz.2011.BRRip.XviD-EM0C0RE.avi (hash = 326356631aa59ff4)
When I use it as a search string in OpenSubtitle search box I am getting the correct movie info, but when I am trying to use the API I am getting the wrong results as follows:

CheckMovieHash2
-The Westing Game
==================================================================
SearchMoviesOnIMDB
-KabaDayi.DvdRip.DvdFilmindir.Org (2007)
-Matthias Rath - Movement of Life (Warsaw.Poland June/18/2011) (2011)
-Kung Fu Panda 2 (2011) BRRiP (2011)
-BBC.Wildlife.Dragonfly.Beauty.or.Beast.DVB.Xvid.mp3.www.mvgroup.org (2004)
==================================================================
SearchSubtitles (using moviehash + moviebytesize; query + language combinations)
-null

What it would take to identify the movie correctly via API as the web-search does? Is the .net library I am using is buggy or the OSDB API is returning the wrong results itself?

Note: I am using OSDB.net https://github.com/kappy/OSDB.net

[oss]

I see your point, when you using this:
http://www.opensubtitles.org/en/search/ ... 631aa59ff4

it seems it is right.

When you are using CheckMovieHash()/CheckMovieHash2() you get wrong results.

The background of getting right moviehash, is to summarize of how many times we "see" this hash in the past with imdb. It is actually wrong (poisoned database by BSPlayer), and we are trying to do something against that.

There might be hashes like this, but most of them should be OK. We have to clean database from this soon, it is quite complicated process but it is in TODO list.

To add more technical stuff: you might be asking, how is possible to get that results on website right and in api it is wrong, and not mentioned in CheckMovieHash2 the right IMDB at all. The problem is following:

Code: Select all

120495   161   
408060   1   
1297919   6   


first is IMDBID and then is number of Seen (how many time we encounter this hash with that idmbid. I not put in results, when the difference is too big, 161 vs 1 or 161 vs 6, => that must be some salt. In this case it is not.

[sarathkcm]

But if the wrong movie has large seen count, why come the correct movie appear as first result in website search....? As a suggestion, If some other factors are concerned in displaying movie results on webpage, why not use similar algorithm for delivering webpage results and xml results?

I encountered this problem for 4-5 movies out of tested 12-15 movies... And for one movie named "3 Idiots" I got no result at all through xmlRequest but through websearch I got movie as first result.