WizMetadataGrabber - can it be told to try to ignore '.'

[psycik]

I keep a lot of files names with dots between them, which obviously doesn't bode well for searching for a name. Just wondering if wizmetadatagrabber can be told to ignore dots, and maybe ignore other thinks like encoding information (xvid, x264, ac3, dts) etc found in file names?

I like the dump to xml files to that the metadata can sit with the movie.

Keep up the good work.

[JavaWiz]

I keep a lot of files names with dots between them, which obviously doesn't bode well for searching for a name. Just wondering if wizmetadatagrabber can be told to ignore dots, and maybe ignore other thinks like encoding information (xvid, x264, ac3, dts) etc found in file names?

I like the dump to xml files to that the metadata can sit with the movie.

Keep up the good work.

Actually, I think it can do that. There is an undocumented filter feature that may be able to transform those titles and remove those strings (See WizRegExFilterTester.exe in plugins/common). Can you run a few problem files thru WizMetadataGrabber and send me the logfile? I'll take a look and see if I can come up with something.

[pBS]

cool, didn't know about that one...regex is badly needed for this util...[silly web names]
is there a manual of the regex standards used?
once ya get the syntax down, they can be quite powerful..

[psycik]

Here's a log. Basically all of them throw up the dialog to ask me what to do.

I imagine ignoring [_ .] type stuff and ignoring a bunch of keywords (Xvid, dts, ac3 etc) would work the best.

[JavaWiz]

Ok, in your config.xml there is a <WizUtilities> section. This section contains the regular expression patterns used when the Scrapers try to manipulate the filenames into something resembling a movie or tv show title.

Back up your current config.xml, and replace the following section:

Code:

<WizUtilities>
      <TitleFilterPattern1>\.DivX|\.XVid|\.720p|\.x264|\.AC3|\.DTS</TitleFilterPattern1>
      <TitleFilterReplace1>
      </TitleFilterReplace1>
      <TitleFilterPattern2>\.720p|\.1080p</TitleFilterPattern2>
      <TitleFilterReplace2>
      </TitleFilterReplace2>
      <TitleFilterPattern3>[-|_]$|^[-|_]</TitleFilterPattern3>
      <TitleFilterReplace3>
      </TitleFilterReplace3>
      <TitleFilterPattern4>\[.+\]</TitleFilterPattern4>
      <TitleFilterReplace4>
      </TitleFilterReplace4>
      <TitleFilterPattern5>[\.|_]</TitleFilterPattern5>
      <TitleFilterReplace5>[space]</TitleFilterReplace5>
      <TitleFilterPattern6>\s+</TitleFilterPattern6>
      <TitleFilterReplace6>[space]</TitleFilterReplace6>
      <TitleFilterPattern7>^The Office</TitleFilterPattern7>
      <TitleFilterReplace7>The[space]Office[space](US)</TitleFilterReplace7>
      <TitleFilterPattern8>^Castle</TitleFilterPattern8>
      <TitleFilterReplace8>Castle[space](2009)</TitleFilterReplace8>
    </WizUtilities>

You can also edit these via the GUI in WizRegExFilterTester (found in plugins/commons directory), click on the workbench tab. The utility allows you to manage the patterns and supply test strings (i.e. titles) to see how they resolve.





I won't explain all of the above, but here is a short explanation of a few of them:

• Pattern1 - This filter will match all of the .DivX, .XVid, .720p, .x264, .AC3 and .DTS strings found in the filename. The pipe '|' symbol denotes an OR condition: this or that or this or ... The replace is an empty string, so this rule will remove the string if it finds a match. The '\' indicates to take the next character literally. (The '.' is a special pattern matching character).
  
• Pattern2 - Same as above, just a different set of strings.
  
• Pattern5 - This fillter will match any . or _ character and replace with a space character.
  
• Pattern7 - (Pattern8 similar) This filter will match any title starting with "The Office". The ^ indicates the pattern MUST start at the beginning of the string. If it finds the pattern, it will be replaced by "The Office (US)". This will allow all of the "The Office" Tv series to be forced to find the US episodes in TheTvDb. Feel free to remove 7 & 8 if they don't apply to you.
  

Hopefully this gives you some idea on how these Filter and Replace patterns can be used to improve your hit ratio on IMDB, TheTvDB (and soon TheMovieDB) title lookups.

A pretty good website to learn about these RegEx patterns can be found at: http://www.regular-expressions.info/reference.html

[JavaWiz]

Did the config settings below improve you lookup hit ratio?

[psycik]

Hi

Running the wizmetagrabber it still seemed to stop on each title and still seemed to default to the full name. So I couldn't see any difference. Sorry

[JavaWiz]

In the current version, you will always be prompted for a custom title on lookup when you click the Retrieve. You will only be prompted for custom title on Retrieve All, if the Prompt for custom title check box is checked on the Options tab.

I've made a slight change to this logic in the next version. You will only be prompted for custom title if the check box is selected.

If selected, you will be prompted every time on a Retrieve click, but only if no hits occur on the Retrieve All click. This is done to allow you to override if a false positive hit occurs.

Additionally, I found a bug in the scraper logic where the filtered title did not always get used. I need to do a little more testing and I will release a new version.