Edge case when testing against local database

Hi,

Firstly, I love the work done on this, it’s a really powerful tool.

I’m running into a problem when testing a salesforce database locally. I believe the tables in the local database has been generated with an integer type for the ID field

For example:

a = Account(id="abc123abc123abc", ...)
a.save()

Would fail, as id has a type mismatch, and is expecting an integer

I believe this is because the SalesforceAutoField, as given to the id field in the SalesforceModel has not overriden the get_internal_type() method, and is defaulting to the integer type, of the parent class AutoField

Doing the following fixed my errors, but I wanted to check with you all first to see if I’m doing something wrong here. or if this should be pull requested?

class SalesforceAutoField(fields.AutoField):
    """
    An AutoField that works with Salesforce primary keys.

    It is used by SalesforceModel as a custom primary key. It doesn't convert
    its value to int.
    """
    description = _("Text")

    def get_internal_type(self):
        return "TextField"

    default_error_messages = {
        'invalid': _('This value must be a valid Salesforce ID.'),
    }

    def to_python(self, value):
        ...

Thanks Elliot

[ElliotSalisbury]

@thatguysimon Thanks for your input! 😃

In the end, I came up with a similar solution. Detecting when a test is running, and modifying the ID field type. I use the pytest_configure function in order to do this.

So within my tests folder, I have a conftest.py that looks like this:

import django

def pytest_configure():
    django.setup()

    # override a function in the ID field of the django-salesforce models
    # we only want to do this in a test runner, to avoid any issues with accessing the live salesforce database
    # the problem is that the autofield creates an 'int' id column, but salesforce id's should be a 'char' field
    # this isnt a problem in the live database, because its maintained by SF, but local copies used for testing
    # are generated incorrectly.
    from salesforce.fields import SalesforceAutoField
    SalesforceAutoField.get_internal_type = lambda s: "TextField"

In your solution, although this seems incredibly unlikely, and you might never run into it, it might be possible in your solution to generate 2 Account objects with the same ID, given that the ID is generated randomly. To avoid something like that, I generate my Accounts with a factory using a sequence:

def get_standardized_sf_id(sf_id):
    # ensure ID is salesforce 18 char case insensitive id
    # https://help.salesforce.com/articleView?id=000274977&type=1&language=en_US

    suffix = ""
    for i in range(3):
        loop = 0
        for position in range(5):
            current = sf_id[i * 5 + position]
            if current.isupper():
                loop += 1 << position
        suffix += "ABCDEFGHIJKLMNOPQRSTUVWXYZ012345"[loop]
    sf_id = sf_id[:15] + suffix
    return sf_id


class SFBaseFactory(factory.django.DjangoModelFactory):
    class Meta:
        abstract = True

    # creates id's that are incrementing integers, padded with 15 0's, and concatenated with the salesforce case insentive code.
    # e.g., "000000000000001AAA"
    id = factory.Sequence(lambda n: get_standardized_sf_id('%015d' % n))
    last_modified_date = factory.LazyFunction(timezone.now)
    ...

[hynekcer]

That solution is in PR #237

Changelog:

• Added: salesforce.models_extend module with SalesforceModel with varchar primary key that works also with default databases. Fixed methods for it: save() and bulk_create() to can create a new pk or to copy an object exactly. Fixed #231 (4d042e5, 79603b7)

The generator function for salesforce alternate primary key (default is uuid.uuid4().hex) is configurable by settings.SF_ALT_PK. This can be useful for your case if you want everything 15+3 chars.

Please try the new code.