Reply to thread

Message

<blockquote data-quote="gamejester" data-source="post: 296362" data-attributes="member: 50634">Rubbish, turned out to be toss!Still it led me to another method which is sweet, so far it has even managed to match mispelt movies first time!What it now does is manipulate your filename and the returned grabber list before calcualting levenshtein distance and if it gets a zero score (exact match) does not bother looking through the rest of the list. Also if the first match in the returned list is the best match it will not ask you to confirm, even if there are other movies with the same match score.Finally it formats the year information the same to both matching strings; so if you have no year stored in your movie names it strips them off both sides; but if you do use movie year in the name it makes sure it is formated the same both sides for the compare; ie in brackets (2000) + it also removes the grabbers tag like (moviexml) from the string allowing for a perfect match.This new method leads itself to the possibility of finding a year in a long string.If we look at the example previously givenBlack.Hawk.Down.2001.BluRay.1080p.DTS.x264.dx va-EuReKA.mkvWhat we could do is search the string in reverse order for a 4 digit number.Then assume this is the year, dump everything after it and grab what is before it as the movie name.Then we just need to create some logic to decide when to do this, maybe some regex to find strings where the majority of words are seperated with . rather than ' '.........hummm, will require some thought.</blockquote>

[QUOTE="gamejester, post: 296362, member: 50634"] Rubbish, turned out to be toss! Still it led me to another method which is sweet, so far it has even managed to match mispelt movies first time! What it now does is manipulate your filename and the returned grabber list before calcualting levenshtein distance and if it gets a zero score (exact match) does not bother looking through the rest of the list. Also if the first match in the returned list is the best match it will not ask you to confirm, even if there are other movies with the same match score. Finally it formats the year information the same to both matching strings; so if you have no year stored in your movie names it strips them off both sides; but if you do use movie year in the name it makes sure it is formated the same both sides for the compare; ie in brackets (2000) + it also removes the grabbers tag like (moviexml) from the string allowing for a perfect match. This new method leads itself to the possibility of finding a year in a long string. If we look at the example previously given Black.Hawk.Down.2001.BluRay.1080p.DTS.x264.dx va-EuReKA.mkv What we could do is search the string in reverse order for a 4 digit number. Then assume this is the year, dump everything after it and grab what is before it as the movie name. Then we just need to create some logic to decide when to do this, maybe some regex to find strings where the majority of words are seperated with . rather than ' '.........hummm, will require some thought. [/QUOTE]

Verification