CODEWITHSUNDEEP
pythonmarshmallow
brillenheini
brillenheini

Reputation: 983

Allowing empty dates with Marshmallow

I try to get data from a webpage. This page contains several release information, but allow values not to be set. I.e. the date for testing from/to might be an empty string.

Now I try to deserialize all my data sucked from the page to insert it to a database and face problems handling empty dates.

from marshmallow import fields, Schema, ValidationError

class TestSchema(Schema):
    training_necessary = fields.Function(deserialize=lambda x: True if x == 'Yes' else False)
    test_from = fields.Date()
    test_to = fields.Date()

data = dict(training_necessary='Yes', test_from='', test_to='')

try:
    validated = TestSchema().load(data)
except ValidationError as err:
    print(f"{err}")

Result:

{'test_to': ['Not a valid date.'], 'test_from': ['Not a valid date.']}

I already tried several combinations of allow_none=True or default='' but none of them helped my to get through. So, how to manage to allow empty dates? Setting a default to somewhat like 1970-01-01 won't help in that case.

Any hints?

Regards, Thomas

+++ EDIT: SOLUTION +++ Here's the working code I ended up after Jérômes helpful tipp:

from marshmallow import fields, Schema, ValidationError, pre_load

class TestSchema(Schema):
    training_necessary = fields.Function(deserialize=lambda x: True if x == 'Yes' else False)
    test_from = fields.Date(allow_none=True)
    test_to = fields.Date(allow_none=True)

    @pre_load(pass_many=False)
    def string_to_none(self, data, many, **kwargs):
        turn_to_none = lambda x: None if x == '' else x
        for k, v in data.items():
            data[k] = turn_to_none(v)
        return data

data = dict(training_necessary='Yes', test_from='', test_to='')

try:
    validated = TestSchema().load(data)
except ValidationError as err:
    print(f"{err}")

Upvotes: 1

Views: 2011

Answers (1)

Jérôme
Jérôme

Reputation: 14714

I would pass no value at all.

data = dict(training_necessary='Yes')

Or I'd make the date fields allow_none and I'd pass None, not an empty string.

data = dict(training_necessary='Yes', test_from=None, test_to=None)

If the issue is that your input contains empty strings, I'd say this is a client issue, but you can add a pre_load method to delete empty strings from the input before deserializing. This is more or less equivalent to modifying the values you scrape from the page before feeding them to marshmallow.

Upvotes: 1

Related Questions