Operation select after workflow loading

[haesleinhuepf]

Hi Ryan @Cryaaa ,

I'm sending this PR as attempt to fix #9 and would love to hear your opinion!

It's just 90% well-thought-through. I hope it doesn't break to many things. The one major change is that I removed a GUI-generating function from your code and call an older one instead. Thus, after loading a workflow, we see now a GUI that looks like before saving the workflow. Example:

Before (current main branch):

New (after this PR):

The only problem I have now is that the pre-selected layer is not correct:

closes #2 closes #9 According to the yaml-file (link below), it should select the result of Gaussian blur:

It may be related to some type-changes: I think the GUI-code I removed was somehow using LayerData, while the GUI I'm using now uses Layers. Thus, there must be code somewhere that compares Layers with LayerData and doesn't find a match. Maybe you immediately know what I mean. Otherwise, I'm happy to explain in a meeting. :-)

I'm also adding this test-workflow. After opening "blobs" from File > Open Samples > clEsperanto, it should be loadable. test_.zip

Also note: this will only work with this napari-workflows branch installed.

No hurry btw.

Best, Robert

[haesleinhuepf]

Codecov Report

Merging #11 (4be61af) into main (2f0d159) will decrease coverage by 0.97%. The diff coverage is 6.06%.

@@            Coverage Diff             @@
##             main      #11      +/-   ##
==========================================
- Coverage   30.15%   29.18%   -0.98%     
==========================================
  Files          12       12              
  Lines         892      932      +40     
==========================================
+ Hits          269      272       +3     
- Misses        623      660      +37     

Impacted Files	Coverage Δ
napari_assistant/_gui/_Assistant.py	37.38% <0.00%> (-0.72%)	:arrow_down:
napari_assistant/_workflow_io_utility.py	0.00% <0.00%> (ø)
napari_assistant/_gui/_category_widget.py	13.75% <10.34%> (-1.34%)	:arrow_down:
napari_assistant/_categories.py	52.94% <13.33%> (-3.46%)	:arrow_down:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 2f0d159...4be61af. Read the comment docs.

[Cryaaa]

Hey @haesleinhuepf, I unfortunately can't use the napari assistant at all when using this branch since everytime I try to add a processing step (like noise removal) I get this error: Traceback (most recent call last):

File "c:\users\ryans\onedrive\documents\bia pol stuff\napari workflow repositories\napari-assistant\napari_assistant_gui_Assistant.py", line 156, in _on_item_clicked self._activate(CATEGORIES.get(item.text())) File "c:\users\ryans\onedrive\documents\bia pol stuff\napari workflow repositories\napari-assistant\napari_assistant_gui_Assistant.py", line 173, in _activate gui = make_gui_for_category(category, self.seach_field.text(), self._viewer, button_size=self.button_size_spin_box.value()) File "c:\users\ryans\onedrive\documents\bia pol stuff\napari workflow repositories\napari-assistant\napari_assistant_gui_category_widget.py", line 404, in make_gui_for_category op_name_widget.value = operation_name File "C:\Users\ryans\anaconda3\envs\np_workflows_v1\lib\site-packages\magicgui\widgets_bases\categorical_widget.py", line 47, in value raise ValueError( ValueError: None is not a valid choice. must be in ('Gaussian (scikit-image, nsbatwm)', 'gaussian_blur (clesperanto)', 'mean_box (clesperanto)', 'mean_sphere (clesperanto)', 'Median (scipy, nsbatwm)', 'median_box (clesperanto)', 'median_sphere (clesperanto)', 'Percentile (scipy, nsbatwm)')

I'm guessing the added code seems to break some old functionality? I'm looking into it but am unsure of how to deal with it...

[Cryaaa]

@haesleinhuepf, nevermind, I found a fix for this, which was not too complicated (but added a function). What worries me most is tthat we both get this too many positional arguments error and I think that tackling this won't be such an easy fix if we want to do it properly instead of relying on a try except solution

[Cryaaa]

Hey @haesleinhuepf! I have done some cleaning up and simplification, but think that maybe now is the time for somebody else to look through the code. There are some notebooks and workflows which I introduced in the docs folder, which can be deleted after reviewing and testing is done. Maybe Johannes could look into the code (although I cannot tag him yet). Also I do not know how to make the codecov tests pass. Maybe I could also write some tests for loading workflows using the notebooks I used.... Let me know what you think and maybe we can schedule a meeting.

[Cryaaa]

@haesleinhuepf I have used this code sucessfully to implement undo and redo as well as just loading. I would suggest merging this branch before I make pull requests to implement undo and redo in the assistant (which will come in the next few days!).

[Cryaaa]

@haesleinhuepf, This PR is ready for mering from my side. Loading now works for all functions implemented in the napari assistant (including pyclesperanto functions).

[Cryaaa]

@haesleinhuepf, I would close this PR if it is alright with you as the changes suggested here were already implemented by merging #12 as undo and loading share some of the same functionality.

[haesleinhuepf]

Yes. Awesome, thanks for taking care of this @Cryaaa !