How to add a factory for a portal model#

FairDM’s SampleFactory and MeasurementFactory are abstract, because the base Sample and Measurement cannot be created. You add a factory for each sample or measurement type your portal defines, so tests and development data can create records of it with one call.

Before you start#

The type exists and is registered. See How to add a sample type or How to add a measurement type.
The examples use a sample type PlantSample with the fields species, collected_on and dry_weight_g, and a measurement type ChlorophyllMeasurement with a reading field. Both are in an app called plants. Use your own names.

Steps#

Create plants/factories.py. Keep it outside tests/ if you want the generate_fake_data command to use it.

Subclass SampleFactory, name the model in Meta, and give a value to each field the model requires. The factory already fills in name, local_id, status and dataset.

import factory

from fairdm.factories import MeasurementFactory, SampleFactory

from .models import ChlorophyllMeasurement, PlantSample


class PlantSampleFactory(SampleFactory):
    class Meta:
        model = PlantSample

    species = factory.Faker(
        "random_element",
        elements=["Quercus robur", "Fagus sylvatica", "Pinus sylvestris"],
    )
    collected_on = factory.Faker("date_between", start_date="-2y", end_date="today")
    dry_weight_g = factory.Faker(
        "pyfloat", left_digits=2, right_digits=1, min_value=1, max_value=99
    )

A field that has a default, or may be blank or null, can be left out. A required field you leave out makes create() fail with the database’s error.

Subclass MeasurementFactory for a measurement type. It gives sample no default, so declare one. Build the sample in the same dataset as the measurement.

class ChlorophyllMeasurementFactory(MeasurementFactory):
    class Meta:
        model = ChlorophyllMeasurement

    sample = factory.SubFactory(
        PlantSampleFactory, dataset=factory.SelfAttribute("..dataset")
    )
    reading = factory.Faker("pydecimal", left_digits=2, right_digits=2, positive=True)

SelfAttribute("..dataset") reads dataset from the measurement being built. A test that passes its own dataset or sample replaces the default.

Optionally, add a trait for a scenario you test often. A trait switches on with a keyword argument.

class PlantSampleFactory(SampleFactory):
    ...

    class Params:
        heavy = factory.Trait(dry_weight_g=95.0)

Use the factory in a test.

import pytest

from plants.factories import ChlorophyllMeasurementFactory, PlantSampleFactory


@pytest.mark.django_db
def test_a_measurement_is_made_on_a_sample_of_its_dataset():
    measurement = ChlorophyllMeasurementFactory()

    assert measurement.sample.dataset == measurement.dataset


@pytest.mark.django_db
def test_a_sample_with_descriptions():
    sample = PlantSampleFactory(heavy=True, descriptions=2)

    assert sample.dry_weight_g == 95.0
    assert sample.descriptions.count() == 2

Optionally, list the factories in the FAIRDM_FACTORIES setting so generate_fake_data creates records of the type. The setting maps a model label to the dotted path of its factory, and is described in Settings.
```
FAIRDM_FACTORIES = {
    "plants.PlantSample": "plants.factories.PlantSampleFactory",
    "plants.ChlorophyllMeasurement": "plants.factories.ChlorophyllMeasurementFactory",
}
```

The result#

PlantSampleFactory() creates a saved sample in a new private dataset, and ChlorophyllMeasurementFactory() creates a measurement on a sample of its own dataset. Both take the arguments every record factory takes, such as descriptions=2 and dates=1. They are listed in the factories reference.

How to test a portal with FairDM factories shows the factories in use. The generate_fake_data command picks one of the listed factories at random for each sample or measurement it creates.