Pass none in physical value converter's convert_back

[MattRoyle]

Fixes #46
Passes the 'pass_none' parameter into the float to string converter used in the convert back function.
Searching for usages currently no PhysicalValueToStringConverter's use the pass_none or fuzzy parameters so there shouldn't be any behavior changes

[Tiomat85]

The PR is fine for what it identifies to fix - pass_none with convert_back.

This converter looks to have other, unrelated, pretty significant flaws.

[cmeyer]

Is this used anywhere? It's always worth extra precaution to ensure this doesn't change existing behavior unintentionally.

[MattRoyle]

Is this used anywhere? It's always worth extra precaution to ensure this doesn't change existing behavior unintentionally.

Currently the pass_none and fuzzy parameters are not used anywhere for PhysicalValueToStringConverter, however something I am working on needed to use this converter but with pass_none and I noticed it was not passing None as expected so updated it to properly handle the parameter.

[cmeyer]

I'm actually ok with the reformatting - I do that as minor cleanup. Technically @KRLango is correct, but it's minor.

I'm also ok with this "as is".

However, there is an incongruity - the __float_to_string_converter is used in convert_back but not in convert. This highlights that one direction goes from a raw float (e.g. 3.1) to a string scaled with units (e.g. "31 nm") and then the user may type units on the way back (e.g. "42 nm") resulting in a new underlying value (e.g. 4.2).

Does this convert_back function handle "42 nm"?

Overall, this class is kind of a hack (by me, years ago). Whatever fixes it for your needs is ok with me, but we might consider thinking about the bigger picture of how we want a proper unit converting converter to work.

[MattRoyle]

Does this convert_back function handle "42 nm"?

Yes the convert back by default can handle the units being there.

>>> from nion.utils import Converter
>>> con = Converter.PhysicalValueToStringConverter(units="nm")
>>> print(con.convert_back("42 nm"))
42.0

However if you set fuzzy to false then there will be a traceback.

>>> con2 = Converter.PhysicalValueToStringConverter(units="nm", fuzzy=False)
>>> print(con2.convert_back("42 nm"))
Traceback (most recent call last):
  File "<python-input-5>", line 1, in <module>
    print(con2.convert_back("42 nm"))
          ~~~~~~~~~~~~~~~~~^^^^^^^^^
  File "C:\NionApps\Developer\main\nionutils\nion\utils\Converter.py", line 154, in convert_back
    value = self.__float_to_string_converter.convert_back(formatted_value)
  File "C:\NionApps\Developer\main\nionutils\nion\utils\Converter.py", line 91, in convert_back
    return locale.atof(formatted_value) if formatted_value else 0.0
           ~~~~~~~~~~~^^^^^^^^^^^^^^^^^
  File "C:\Users\matthew.royle\AppData\Roaming\uv\python\cpython-3.14.2-windows-x86_64-none\Lib\locale.py", line 328, in atof
    return func(delocalize(string))

Maybe I should change this PR to only pass the pass_none parameter and not pass fuzzy since for a physical value it should always be true to not hit the traceback.

[cmeyer]

Now with unexpected behavior exposed during this conversation, I think we should add a test case or two for this. They will be useful for catching regressions if we migrate this to something with more features.

[KRLango]

LGTM