Closed will-moore closed 4 years ago
Tested on merge-ci user-3 (we have the idr0021 in full with ROIs imported there too).
All works fine, on the dataset which errored on ome-training-4 previously, as well as on the whole idr0021 project.
Ready to merge fmpov
The idea would be to roll this into the next release, correct? Is there anything else foreseen or shall we call this 5.6.1
?
@will-moore might come back from workshop with a fix for a script :-) in that case it might be included in a 5.6.1 minor comments on this PR that will be good to have it in
I could not reproduce the error on workshop using few images from idr21 (images 9539, 9540, 9541) using the older version of the script
The handling of the string is quite fragile in a way This PR handles potential cases we have seen and it does not affect existing shapes
No, I can't find the shapes that broke this script originally.
But you only need to have Points string with trailing whitespace "1,2 3,4 "
to break the script (without this PR).
Found this bug when trying Batch_ROI_Export from outreach server, using idr0022 data. Failed with
float(x.strip(", ")
since points string was not in the exact expected format.I now strip whitespace before split(' ') so don't end up with empty string from trailing whitespace. Also wrap with try/except in case some other unexpected format - Don't want a single invalid points list to break the whole script.
To test: