Validation fails on tomlkit parsed data

[Frost Ming]

When I try validating data parsed by tomlkit, it always fails. After some digging, the causes are as follows:

In cerberus/validator.py:

def __init_processing(self, document, schema=None):
        ...
        self.document = copy(document)

Calling copy on document, which is a tomlkit.container.Container instance, breaks the iterability for unknown reason:

>>> list(document)
['name', 'url', 'verify_ssl']
>>> list(copy(document))
[]

Therefore, all fields defined in the data will be recognized as missing.

I am not sure which of cerberus/tomlkit/plette is responsible to fix this. But it will be easy to fix it here: just calls data.copy() to pass to the validator.

[Tzu-ping Chung]

Oops, I also saw this a couple of weeks ago, but forgot to raise it. You can work around the problem by changing line 25 below to explicitly copy to dict before validating:

plette/src/plette/models/base.py

Lines 16 to 27 in fad7956

 def validate(cls, data): 
     if not cerberus:    # Skip validation if Cerberus is not available. 
         return 
     schema = cls.__SCHEMA__ 
     key = id(schema) 
     try: 
         v = VALIDATORS[key] 
     except KeyError: 
         v = VALIDATORS[key] = cerberus.Validator(schema, allow_unknown=True) 
     if v.validate(data, normalize=False): 
         return 
     raise ValidationError(data, v) 

if v.validate(dict(data), normalize=False):
    ...

This should be raised to TOMLkit to fix the copy behaviour (it should work). The above patch should probably also be applied as well.

---

Edit: You also mentioned a similar workaround; sorry I missed that.