[ome-devel] ST FilenamePattern & TheT

[Ilya Goldberg]

On Sep 7, 2004, at 9:35 AM, Joshua Moore wrote:

> Interesting problem--
>
>    the EMBL folks have in their file names the exact time of imaging  
> rather than t1, t2, t3,.... e.g.
>
> 0001--Hp1gamma_si--W00001--P00001--T00000m--SL00000--Q00--F03-- 
> I0025.tif
> 0001--Hp1gamma_si--W00001--P00001--T00030m--SL00000--Q00--F03-- 
> I0025.tif
> 0001--Hp1gamma_si--W00001--P00001--T00060m--SL00000--Q00--F03-- 
> I0025.tif
> 0001--Hp1gamma_si--W00001--P00001--T00090m--SL00000--Q00--F03-- 
> I0025.tif
> 0001--Hp1gamma_si--W00001--P00001--T00120m--SL00000--Q00--F03-- 
> I0025.tif
> 0001--Hp1gamma_si--W00001--P00001--T00150m--SL00000--Q00--F03-- 
> I0025.tif
>
> I can certainly get a regex around this, but what is going to happen  
> to TheT's?

Unfortunately, they won't be interpreted correctly.  This is a case  
where you're encoding meta-data in the file name, and it won't wind up  
where its supposed to in the DB.  The FilenamePattern ST really only  
works when the filename contains indexes in the 5-d space - I can't  
think of a good way for it to decode meta-data.  If there is some other  
way of preserving the meta-data, or annotating it later, then you could  
put in a mechanism (using a flag?) to get the dimension indexes through  
a lexical sort of whatever is in your regex.  But this kind of throwing  
away of important stuff is not very good.

The Import module is declared with a type "any" output, so it can  
already put out any kind of ST.  If you had a way to generate the  
indexes (using a flag to get them from a sort), you could also (maybe)  
specify that the regex should populate a certain ST.  This is much,  
much more than we had in mind for FilenamePattern.

With an hour or so's work, you can probably cut and paste your way to a  
customized tiff importer that does exactly what you want.

-Ilya


>
>    -Josh
>
> _______________________________________________
> ome-devel mailing list
> ome-devel at lists.openmicroscopy.org.uk
> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-devel
>