Create best-in-class factories for your synthetic data in Kotlin.
⭐ Test fixtures
⭐ DB seeding
⭐ Feature demos
⭐ Pre-production environments
Chances are that synthetic data are going to be a major painpoint for your project sooner than later.
- Local development
- Unit tests
- Exploratory testing
- Showcase a feature
- Pre-production environments
Are you going to need synthetic data for any of the above?
Yes! - Probably for all of them
There are 3 major ways to create synthetic data that we are aware of:
- Static fixtures
- Factories
- Production copies
💙 Static fixtures usually live in plain YAML or JSON files. They are typically fed directly to the underlying database, skipping standard validation/integrity checks for simplicity and performance. Since data integrity is not a priority certain things like exploratory testing and refactorings become much harder. Finally, due to their hardcoded nature, generating large volumes of data is out of question. That's why we typically recommend static fixtures only for small apps.
💚 Production copies provide a great way to populate a system with data. You typically get valid, large and versatile enough data to cover most of use cases. BUT if you think about it... usually production data contains sensitive data and you'll need some sort of obfuscation/anonymization before you load them into a develoment/test system. That's 🤯 - especially when there are multiple databases involved.
Some other times, production copies can be too small - during the early days of a product - or too large to be of practical use. And certainly, you cannot really write unit or functional tests on production copies cause they are dynamic and unpredictable. We typically recommend production copies for populating pre-production environments.
❤️ Factories or dynamic fixtures if you like, directly produce models from within our domain. Models that can be validated and stored in the underlying database with all integrity checks in place. A well made set of factories, is one that captures all the necessary abstractions that let you create data for a certain business scenario in a few lines of code. We typically recommend factories for most use cases, but primarily for creating test fixtures and populating local development environments.
Other ecosystems have robust synthetic data solutions for some time now, mostly inspired from Ruby's amazing FactoryBot.
KFactory is also inspired by FactoryBot - in a Kotlin idiomatic way.
- Built-in helpers
- Composable factories
- Traits
- Lazy sequence builds
KFactory is published on mavenCentral. In order to use it just add the following dependency:
implementation("io.github.bluegroundltd:kfactory:1.0.0")- API reference
- Check the examples for some common use cases
When you have a domain entity that you want to create a Factory for you can start
by doing the following:
class AddressFactory : Factory<Address> {
override fun produce() : Address = Address()
}A lot of times we want to produce fixtures from factories, but we only need to change only a few of their attributes/characteristics.
For example:
class AddressFactory(
private var city: String = "city",
private var state: String = "state",
) : Factory<Address> {
fun withCity(city: String) = apply {
this.city = city
}
fun withState(state: String) = apply {
this.state = state
}
override fun produce() : Address = Address(
city = city,
state = state
)
}Now if we want to produce several instances of Address that will retain city
but will have another specific value for state, we can create a new FactoryTrait
that we will later apply to that Factory.
object CaliforniaTrait : FactoryTrait<AddressFactory> {
override fun modifyWithTrait(factory: AddressFactory): AddressFactory = factory
.withState(state = "California")
}In order to enhance our Factory with a FactoryTrait like we previously saw,
we need to use the TraitEnhancedFactory marker interface.
For example consider the following:
class AddressFactory(
private var city: String = "city",
private var state: String = "state",
) : Factory<Address>, TraitEnhancedFactory {
fun withCity(city: String) = apply {
this.city = city
}
fun withState(state: String) = apply {
this.state = state
}
override fun produce() : Address = Address(
city = city,
state = state
)
}This immediately adds two new extension functions on our Factory:
fun withTraits(vararg traits: FactoryTrait)
fun withTrait(trait: FactoryTrait)We can now start building factories with distinctive characteristics:
val californiaFactory: AddressFactory = AddressFactory()
.withTraits(CaliforniaTrait)
As described above we utilize factories in order to produce fixture data.
This can be done by invoking the following function on a Factory:
val address: Address = californiaFactory.produce()If we need to generate more than on instance of our fixture data, we can utilize the
following function of a Factory that returns a Sequence of objects:
val addresses: List<Address> = californiaFactory.produceMany()
.take(5)
.toList()Most of the time in our fixture data we might need to produce random values, or have
a new value generated every time we invoke .produce() on one of our factories.
For that purpose, we include a typealias in our library, named Yielded and our
proposed usage is the following:
class AddressFactory(
private var city: String = "city",
private var state: String = "state",
private var streetNum: Yielded<Int> = { Random.nextint(1,5) }
) : Factory<Address>, TraitEnhancedFactory {
fun withCity(city: String) = apply {
this.city = city
}
fun withState(state: String) = apply {
this.state = state
}
fun withStreetNum(streetNum: Int) = apply {
this.streetNum = { streetNum }
}
fun withStreetNum(streetNum: Yielded<Int>) = apply {
this.streetNum = streetNum
}
override fun produce() : Address = Address(
city = city,
state = state,
streetNum = streetNum()
)
}From the above example, we can see that we have two new functions in our Factory.
These functions allow us to override the value generated for streetNum to have either
a static value every time we invoke .produce(), or a dynamic one. By default, the value
of it will be a lambda function which delegates to Random.nextInt() each time.
- Bump version in
gradle.propertiesofkfactorymodule. - Execute the following to upload artifact:
$ ./gradlew :kfactory:publish \
--no-daemon --no-parallel \
-Psigning.secretKeyRingFile=<keyring_file_path> \
-Psigning.password=<keyring_password> \
-Psigning.keyId=<keyring_id> \
-PmavenCentralUsername=<nexus_username> \
-PmavenCentralPassword=<nexus_password>After this operation finishes, you can promote the artifact to be released with:
$ ./gradlew closeAndReleaseRepository \
-PmavenCentralUsername=<nexus_username> \
-PmavenCentralPassword=<nexus_password>The core maintainer of this project, is the Platform Team of Blueground!