PyCharm debugging leads to TypeError if add_xxx() is called

[LucaMarconato]

🐛 Bug Report

I used to be able to use the debugger in PyCharm for code using napari. Since around 2 weeks if I try to run a script in debug mode in PyCharm and the script calls one of the add_xxx functions, for instance add_points(), I receive a TypeError exception.

Running the script in Run mode works. Also the debugger works everywhere else. I am using the latest released napari.

The backtrace for the error I get is the following:

/Users/macbook/miniconda3/envs/ome/bin/python /Users/macbook/Applications/PyCharm Professional Edition.app/Contents/plugins/python/helpers/pydev/pydevd.py --multiprocess --qt-support=auto --client 127.0.0.1 --port 58587 --file /Users/macbook/embl/projects/basel/spatialdata/napari-spatialdata/temp/bug_points_scaling.py 
Connected to pydev debugger (build 231.9225.15)
Traceback (most recent call last):
  File "/Users/macbook/Applications/PyCharm Professional Edition.app/Contents/plugins/python/helpers/pydev/pydevd.py", line 1496, in _exec
    pydev_imports.execfile(file, globals, locals)  # execute the script
  File "/Users/macbook/Applications/PyCharm Professional Edition.app/Contents/plugins/python/helpers/pydev/_pydev_imps/_pydev_execfile.py", line 18, in execfile
    exec(compile(contents+"\n", file, 'exec'), glob, loc)
  File "/Users/macbook/embl/projects/basel/spatialdata/napari-spatialdata/temp/bug_points_scaling.py", line 4, in <module>
    viewer.add_points([[10, 10]], size=[10], name="points_layer")
  File "/Users/macbook/miniconda3/envs/ome/lib/python3.10/site-packages/napari/components/viewer_model.py", line 7, in add_points
    from functools import lru_cache
TypeError: PostInit.__call__() got multiple values for argument 'self'
python-BaseExceptio

💡 Steps to Reproduce

1. in PyCharm run the following script in Debug Mode

import napari

viewer = napari.Viewer()
viewer.add_points([[10, 10]], size=[10], name="points_layer")
napari.run()

💡 Expected Behavior

The script should execute normally.

🌎 Environment

napari: 0.4.19.post1
Platform: macOS-13.4.1-arm64-arm-64bit
System: MacOS 13.4.1
Python: 3.10.13 | packaged by conda-forge | (main, Oct 26 2023, 18:09:17) [Clang 16.0.6 ]
Qt: 5.15.8
PyQt5: 5.15.9
NumPy: 1.26.3
SciPy: 1.11.4
Dask: 2023.12.1
VisPy: 0.14.1
magicgui: 0.8.1
superqt: 0.6.1
in-n-out: 0.1.9
app-model: 0.2.4
npe2: 0.7.3

OpenGL:

• GL version: 2.1 Metal - 83.1
• MAX_TEXTURE_SIZE: 16384

Screens:

• screen 1: resolution 3840x2160, scale 1.0
• screen 2: resolution 3840x2160, scale 1.0

Settings path:

• /Users/macbook/Library/Application Support/napari/ome_e544a94aa34edcbdceb29c82a6079cdcab084ea5/settings.yaml
  Plugins:
• napari: 0.4.19.post1 (77 contributions)
• napari-console: 0.0.9 (0 contributions)
• napari-matplotlib: 1.2.0 (10 contributions)
• napari-spatialdata: 0.3.2.dev10+g3c8dfd5 (6 contributions)
• napari-svg: 0.1.10 (2 contributions)

💡 Additional Context

No response

[Czaki]

Hi @LucaMarconato. It looks really strange as self is popped form kwargs here:

napari/napari/utils/_register.py

Line 9 in daf7205

kwargs.pop('self', None)

And a line referred by this error is

napari/napari/utils/_register.py

Line 13 in daf7205

layer = {cls_name}(**kwargs)

(we need to fix this generation).

As I cannot reproduce this on my machine so I could only ask you to try to use debugger with steep by steep and see where is a problem, why self is not popped. With such information we could try to debug this.

[LucaMarconato]

Thanks for the answer. I tried using the "Step Over" option in the PyCharm debugger. After 2 steps I get the error:

[LucaMarconato]

I tried adding a print(kwargs) before and after the kwargs.pop('self', None) and the self is removed correctly; still I am keeping getting the error.

[LucaMarconato]

I haven't tried updating/reinstalling PyCharm or installing napari from main. I could try.

[LucaMarconato]

Reinstalling napari from main didn't help, working on a fresh conda env neither. I ended up completely removing conda from my machine and reinstalling it, which worked. So now I am finally not experiencing this bug anymore.

[LucaMarconato]

I celebrated too early, the problem was not fixed, but now I managed to implement a workaround.

In details:

• the problem actually still occurred, I just mistakenly didn't call add_xxxx when testing the new conda installation
• I switched to main (note, I had to do a full install and not an editable install otherwise I was getting a different error upon opening napari)
• Now that improve stack trace of autogenerated ViewerModel.add_xxxx methods #6791 is merged, I could see a debugger at add_layer.py_tmpl when calling add_xxxx(); I verified and kwargs contained self even after pop() was called.
• I modified add_layer.py_tmpl to match the code below, and this workaround now the problem disappeared for me. I am reopening the issue.

def {name}{signature}:
    kwargs = locals()
    kwargs.pop('self', None)
    pos_kwargs = dict()
    for name in getattr({cls_name}.__init__, "_deprecated_constructor_args", []):
        pos_kwargs[name] = kwargs.pop(name, None)
    kwargs_workaround = dict()
    for key in kwargs.copy():
        if key not in ['self', 'kwargs_workaround', 'kwargs', 'pos_kwargs']:
            kwargs_workaround[key] = kwargs[key]
    layer = {cls_name}(**kwargs_workaround)
    for name, value in pos_kwargs.items():
        if value is not None:
            setattr(layer, name, value)
    self.layers.append(layer)
    return layer

# This file is created to improve user bug reports when
# user use `viewer.add_xxxx` and some error occurs.
# this allow to have more readable stacktrace.

[Czaki]

What happens if you change kwargs = locals() to kwargs = dict(locals()) without your workaround?

[Czaki]

@LucaMarconato ?

[LucaMarconato]

Good point! Yes, now it works, by applying the change you proposed in napari.utils._register.py.

[jni]

@LucaMarconato would you like to open a PR? 😃