Charfield choice with blank=True fix

[legios89]

Hi,
I just found this problem during my work, and also found the solution in the following issue: #503
I also modified tests, what I forget to do in the previous pull request.

[coveralls]

Coverage increased (+0.02%) to 94.537% when pulling dd1e169 on vertisfinance:charfield_choice_blank_fix into f76f38e on graphql-python:master.

[legios89]

I am a little bit confused about what this error means? For me, it looks like everything went fine.

[patrick91]

Hi @legios89 why did you close this? I think it is quite good, my only doubt would be that we are returning an empty string where it might make more sense to return None (just thinking from a frontend perspective).

Anyway, I've attached a test case that you can put in the PR, might be worth merging this and closing #474 :)

@spockNinja @syrusakbary do you have any additional feedback on this?

# https://github.com/graphql-python/graphene-django/issues/474

import graphene
from graphene_django import DjangoObjectType
from django.db import models

class Car(models.Model):
    class Meta:
        app_label = 'app'

    choice = models.CharField(
        choices=[('blue', 'Blue'), ('green', 'Green')],
        max_length=20,
        blank=True
    )

class CarNode(DjangoObjectType):
    class Meta:
        model = Car

def test_blank_choice():

    class Query(graphene.ObjectType):
        car = graphene.Field(CarNode)

        def resolve_car(self, info):
            return Car(choice="")

    schema = graphene.Schema(query=Query)
    result = schema.execute('{ car { choice } }')

    print(result.data, result.errors)

    assert not result.errors

[zbyte64]

I too am confused as to why this is closed. This is still an issue and the PR seemed reasonable. The only thing I would change is the choice label for the null/empty value. With this PR it will generate "A_" for the enum label when it should likely be "BLANK" or "NULL"

I want to reopen this but only if the @legios89 agrees. Otherwise we need to make our own PR or merge with #654