Empty path gets upgraded to pathlib.Path(".") by FMModel

[veenstrajelmer]

Describe the bug
When initializing an FMModel with empty paths (""), these are upgraded to pathlib.Path("."), which results in a . in the mdu file instead of an empty value.

To Reproduce
The following code fails with AssertionError since mdu.output.outputdir=WindowsPath("."). When generating a mdu file from scratch (mdu_clean), the value is according to expectations.

from hydrolib.core.dflowfm import FMModel

# this succeeds
mdu_clean = FMModel()
assert mdu_clean.output.outputdir == ""

# this fails
mdu_clean = FMModel()
mdu_clean.output.outputdir = ""
assert mdu_clean.output.outputdir == "" # AssertionError, since it is a pathlib.Path(".")

# this fails (comparable as above, but the hydromt_delft3dfm approach for initializing an FMModel)
cf_dict = {"output": {"outputdir":""}}
mdu = FMModel(**cf_dict)
assert mdu.output.outputdir == "" # AssertionError, since it is a pathlib.Path(".")

Related issues:

• for -999: feat: Don't write keywords with None values #663
• for .: Don't write keywords with a None value to the files #661 (comment)
• overview of unexpected keywords: Revert incorrect hydrolib-core expected values hydromt_delft3dfm#148

Expected behavior
mdu.output.outputdir to be an emtpy string ("")

Version info (please complete the following information):

• OS: Windows 10
• Version 0.8.0

[veenstrajelmer]

Investigation by @tim-vd-aardweg:
Volgens mij is het met Path alléén niet mogelijk om onderscheid te maken tussen een lege string en een ., want Path("") == Path("."). Ik heb naar de internals gekeken van Path en kon ook niets vinden.. Dus ik denk dat er iets van een custom class gemaakt moet worden hiervoor. Iets als:

class CustomPath:
    def __init__(self, path_str):
        self.path_str = path_str
        if path_str and path_str != '.':
            self.path = Path(path_str)
        elif path_str == '.':
            self.path = Path('.')
        else:
            self.path = None  # Represents an empty path
    def __str__(self):
        return self.path_str

En daar moeten dan weer allemaal validators voor gemaakt worden voor pydantic en natuurlijk testjes toegevoegd worden.