Reply to thread

Message

There is a function in IMDBFetcher.cs called RefreshIMDB() that uses the filename to build an IMDB search string. It wouldn't be too hard to change this function to parse out the directory name as search criteria instead of using the file name.

I have changed my version to look for files of the form <movie name><date><crap>. The movie name and the date are used as the search criteria and the crap after the date is ignored. The only issue now is that the MP comparison function uses fuzzy logic implemented as a Levenshtein comparison routine. It compares the IMDB title to the original filename instead of the search criteria. I'm certain this is a bug and will attempt to fix. I will change the FuzzyMatch() function to skew results towards IMDB results that match the date and then use the Levenshtein comparison to only check the title.

A funny thing with MP implementation is that it will match the IMDB "aka" title, but only record the original title. The consequence is, for example, that "The Good, The Bad, and The Ugly" is recorded in the database as "Buono, il brutto, il cattivo, Il ". This is funny.

There is also an O^2 optimization problem as the directory is being scanned for each search. This makes the scan slower as the number of files increases. I don't know much about windows programming so this may be required because of threading / context issues.

<blockquote data-quote="rjburke377" data-source="post: 280461" data-attributes="member: 70884">There is a function in IMDBFetcher.cs called RefreshIMDB() that uses the filename to build an IMDB search string. It wouldn't be too hard to change this function to parse out the directory name as search criteria instead of using the file name.I have changed my version to look for files of the form &lt;movie name&gt;&lt;date&gt;&lt;crap&gt;. The movie name and the date are used as the search criteria and the crap after the date is ignored. The only issue now is that the MP comparison function uses fuzzy logic implemented as a Levenshtein comparison routine. It compares the IMDB title to the original filename instead of the search criteria. I'm certain this is a bug and will attempt to fix. I will change the FuzzyMatch() function to skew results towards IMDB results that match the date and then use the Levenshtein comparison to only check the title.A funny thing with MP implementation is that it will match the IMDB &quot;aka&quot; title, but only record the original title. The consequence is, for example, that &quot;The Good, The Bad, and The Ugly&quot; is recorded in the database as &quot;Buono, il brutto, il cattivo, Il &quot;. This is funny.There is also an O^2 optimization problem as the directory is being scanned for each search. This makes the scan slower as the number of files increases. I don't know much about windows programming so this may be required because of threading / context issues.</blockquote>

[QUOTE="rjburke377, post: 280461, member: 70884"] There is a function in IMDBFetcher.cs called RefreshIMDB() that uses the filename to build an IMDB search string. It wouldn't be too hard to change this function to parse out the directory name as search criteria instead of using the file name. I have changed my version to look for files of the form <movie name><date><crap>. The movie name and the date are used as the search criteria and the crap after the date is ignored. The only issue now is that the MP comparison function uses fuzzy logic implemented as a Levenshtein comparison routine. It compares the IMDB title to the original filename instead of the search criteria. I'm certain this is a bug and will attempt to fix. I will change the FuzzyMatch() function to skew results towards IMDB results that match the date and then use the Levenshtein comparison to only check the title. A funny thing with MP implementation is that it will match the IMDB "aka" title, but only record the original title. The consequence is, for example, that "The Good, The Bad, and The Ugly" is recorded in the database as "Buono, il brutto, il cattivo, Il ". This is funny. There is also an O^2 optimization problem as the directory is being scanned for each search. This makes the scan slower as the number of files increases. I don't know much about windows programming so this may be required because of threading / context issues. [/QUOTE]