Can somebody please explain me the behaviour of the following pydantic model.
from datetime import datetime
from pydantic import BaseModel
first_format = {'time': '2018-01-05T16:59:33+00:00',}
second_format = {'time': '2021-03-05T08:21:00.000Z',}
class TimeModel(BaseModel):
time: datetime
class Config:
json_encoders = {
datetime: lambda v: v.isoformat(),
}
json_decoders = {
datetime: lambda v: datetime.fromisoformat(v),
}
print(TimeModel.parse_obj(first_format))
print("first_format successfull")
print(TimeModel.parse_obj(second_format))
print("second_format successfull")
Output:
time=datetime.datetime(2018, 1, 5, 16, 59, 33, tzinfo=datetime.timezone.utc)
first_format successfull
time=datetime.datetime(2021, 3, 5, 8, 21, tzinfo=datetime.timezone.utc)
second_format successfull
Expected behaviour:
Works on first_format
as it does now, but fails at second_format
as datetime.fromisoformat(second_format)
throws an error for the wrong format
What I want to do:
Parse the time only if it has the format of first_format
, else raise an exception.
Thanks for your help
Edit:
I later realized that there is no such thing as json_decoders
, so please do not get confused by this
Problem resolved
You need to use pre
validator
here, in order to parse and validate incoming datetime string:
from datetime import datetime
from pydantic import BaseModel, validator
first_format = {'time': '2018-01-05T16:59:33+00:00',}
second_format = {'time': '2021-03-05T08:21:00.000Z',}
class TimeModel(BaseModel):
time: datetime
class Config:
json_encoders = {
datetime: lambda v: v.isoformat(),
}
@validator('time', pre=True)
def time_validate(cls, v):
return datetime.fromisoformat(v)
print(TimeModel.parse_obj(first_format).json())
print("first_format successfull")
print(TimeModel.parse_obj(second_format))
print("second_format successfull")
Update:
If you are using pydantic v2 you need to use the field_validator
instead since the validator
was deprecated. The keyword argument is now mode='before'
.