Changed ND2 metadata string handling to work properly with empty strings

[csachs]

Hello,

on demand by @ctrueden in reference to mailing list problem 'Bioformats opening ND2 error' by Peter Saxon, here's a patch fixing the file opening problem, which seems due to the reader fetching too many bytes on empty string fields (in metadata), then reading garbage and getting stuck in an infinite loop.

As a positive side effect, this seems to clear away a lot of nonsensical metadata normally present in ND2 files opened with bioformats. I diff'd the results of Fiji's 'Show Info' for a file opened with a DEVELOP version and a patched version, and some nonsensical fields got removed, and some fields got filled with more meaningful values.

I guess there would be a way to get the findString function to work properly, but this 'easy' solution was just too tempting ;)

I only tested it with two files; please check it with your collection of different ND2 files ...

Best regards,
Christian Sachs

[ctrueden]

Thanks, @csachs!

[ctrueden]

For users wanting to test this change (in ImageJ or otherwise), I uploaded a build here:

• http://curtis.imagej.net/2014-12-05-bio-formats-pr-1461/

[melissalinkert]

Thanks again, @csachs. This looks fine on review; we're just waiting for confirmation that our CI builds continue to pass with this change included. I would expect to be able to merge this in the next day or two.

[melissalinkert]

--no-rebase