I’m using faust to process some data, the example model looks like this:

class MyModel(faust.Record):
    entry_id: uuid.uuid4
    metadata: json.loads
    timestamp: lambda x: pd.Timestamp(x, tz=None)
    _offset: int
    # ... some other int, str, bool fields

Note I’m using functions as field descriptors to process incoming fields.
Now, I need to pre-fetch some rows first from database:

# cols is a list of declared 'fields' for a type, cls = MyModel
cols = get_class_variables(cls)

# this returns a DataFrame with given columns
df = client.select(
    f'SELECT {", ".join(cols)} FROM ... WHERE ...', columns=cols)

# here I want values to be list[MyModel]
values = [??? for row in df.to_dict(orient='records')]

I tried the following for ???

• cls(**row)
• cls.loads(row)
• cls.from_data(row)

none of them seems to work, it just “casts” the dictionary into the type (i.e. metadata is still string, not parsed json). Any idea what function should I use for the dictionary fields to be processed according to the model? I want it to be as generic as possible since I’m using the same code for many models.