PDA

View Full Version : A Kludge for the DOES NOT START WITH "xyz" problem



Lifferds
08-27-2013, 05:49 AM
I was putzin' around trying to figure how to solve this natty little issue and came up with the following. It is not elegant and requires some active participation and sacrifice on your part, but it does work.

For English the popular beginning is THE which clogs up the STARTS WITH 'T' playlist. Le in French, Die in German, Ha in Bulgarian might cause the same problem for those playlists (L, D, H). The following pertains to English.

1. Create a Smart Playlist for Titles starting with THE.
2. When populated, select all files, <ctl>A on a PC works fine. Right Click and bring up the properties dialog.
3. Change the Comment Field to some unique set of characters, e.g., qwerty12345 (with all files selected this performs a batch replacement). Sorry about messing up your comment but that looks like the least damaging field.
4. Now go to your STARTS WITH 'T' playlist and add a filter for - Comment Does Not Contain qwerty12345 - or whatever unique string you chose.

Now all the 'THE' titles should be gone from the 'T' playlist and show up in the 'THE' playlist.

Note: I tried changing the comment to 'The' and use COMMENT IS NOT 'The', but that failed and the 'The' files still showed in the 'T' playlist. IS NOT 'qwerty12345' does work, however.

Tom

Paul
08-27-2013, 11:23 AM
Thanks for sharing this, Tom. We're still planning to have an ignore feature for leading 'noise' words like 'The'. We are discussing this here as to the best way to implement it. Your solution seems a valid workaround for the current version of Mezzmo (v3.3.0.0).

Coises
08-27-2013, 12:10 PM
We're still planning to have an ignore feature for leading 'noise' words like 'The'. We are discussing this here as to the best way to implement it.

I have to throw in my two cents once again... whatever you do, the most important thing is to make sure the user can override it. I suggest that the Properties dialog should include a Sort As field. When that field is empty, show the auto-derived sort key in gray; but allow the user to enter any text there.

Best would be to let the user configure the auto-derivation rules, too. For example, a user with many German movies might want to include "Die" in the list of initial words to bypass when auto-deriving the sort key (but would have to be able to override that for "Die Hard").

Paul
08-27-2013, 01:35 PM
Thanks, Coises. All good thoughts.

Lifferds
08-29-2013, 09:33 PM
There might also be consideration to noise words being followed by a space, else titles starting with Theater, Then, Them, etc. would also be excluded. Because some languages, French for one, have two words for 'the' (le, les) makes the problem even more complex. Would it be possible to have an ALL OF THESE level and an ANY OF THESE sublevel within the ALL level; condition A AND condition B AND (condition C OR condition D). And then there are languages reading right to left. A tough nut to crack.

Thanks for the best support in the business,
Tom

Paul
08-30-2013, 08:34 AM
Thanks for your thoughts, Tom. We'll keep this thread open for more ideas from users. We are not quite ready to implement anything at this stage (we are very busy on other features at the moment & still do not have a firm opinion on noise word implementation), but I imagine we can get into this feature later in the year.

jbinkley60
08-30-2013, 10:43 AM
Thanks for your thoughts, Tom. We'll keep this thread open for more ideas from users. We are not quite ready to implement anything at this stage (we are very busy on other features at the moment & still do not have a firm opinion on noise word implementation), but I imagine we can get into this feature later in the year.

Paul,

SInce I was the one who first brought this up, could it be as simple as allowing us to select a noise word at a rule level within a smart playlisty ?

Example:

Does not begin with "noise word A"
Does not begin with "noise word B"

RIght now you have does not contain as an option, just add does not begin with and allow us to add words on a rule by rule basis. That seems to be the most flexible option. Please don't try to pre-define them.

Coises
08-30-2013, 02:39 PM
SInce I was the one who first brought this up, could it be as simple as allowing us to select a noise word at a rule level within a smart playlist?


I’m curious now...

If the programmers gave us the ability to specify a list of leading words (e.g., A, An, The) that would be ignored when sorting—so that The Crucible would appear under C and The Pawnbroker under P—would that resolve this issue? Or do you, Lifferds and/or others specifically want all the The titles together, just broken out separately from the other T titles?

That is, is this (as it seems to me) all about sorting titles like humans normally sort them instead of the way computers normally sort? Or is there some other dimension to it?

jbinkley60
08-31-2013, 04:26 AM
I think it is that simple. Just allow us on a playlist/rule level to remove entries which start with specific noisy words. Making it a global filter or pre-defining the words would limit the flexibility.

Lifferds
09-03-2013, 03:55 AM
I think playlist specific is the way to go. Having looked into the db I was thinking of adding a field (ValueOneLength) to capture the length of ValueOneText (the noise word). You could, of course, have multiple records for multiple noise words to accommodate languages with noise words for singular, plural and male and female gender. This would lead to a query with conditions such as:
Condition A
AND
Condition B
AND
Left(Title,ValueOneLength+1) != ValueOneText & <space>
AND
Left(Title,ValueTwoLength+1) != ValueTwoText & <space>
AND
Left(Title,ValueThreeLength+1) != ValueThreeText & <space>
AND
Left(Title,ValueFourLength+1) != ValueFourText & <space>
...

To answer Coises and the Die (German) vs. Die Hard problem add an extra piece of Metadata for 'Noise Word Filter' (Boolean) to the media properties to include or exclude that piece of media from filtering. In the setup the user could select True or False to be the default and all new properties would automatically start out with that default and existing ones could be batch changed with a dialog "Do you want to change existing media properties?" Yes|No.
This would add an auto Condition of:

AND
NoiseWordFilter == True

To the query when a DOES NOT START WITH operand is selected.

With regards to Coises other remark about having The Crucible end up in the C playlist, one would need the ability to add a
.
.
.
AND
(Condition A
OR
Condition B)

to the Match All Following Rules where Condition A is "STARTS WITH 'C'" and Condition B is "STARTS WITH 'The C'". That might be a bit more work, nesting an OR inside an AND, but certainly not impossible.

That's probably more than you wanted or different than you were expecting, but just some thoughts.

Tom