Yeah mostly regex. Plus some heuristics. Works most of the times. Pretty efficient for the simple amount of heuristics it uses. You can see the code here https://github.com/legalosLOTR/mdb/blob/master/MDB/DBbuilder... Check out the function get_movie_name!

Wouldnt this break a movie like 2012?

    #2 remove year and stuff following it
    filename = re.sub('\d\d\d\d.*', ' ', filename)

Yeah it totally would! Thanks for pointing this out. Btw, I am planning to use another library that has better detection rates - https://github.com/wackou/guessit