Rosie is a factory for building JavaScript objects, mostly useful for setting up test data. It is inspired by factory_girl.
To use Rosie you first define a factory. The factory is defined in terms of attributes, sequences, options, callbacks, and can inherit from other factories. Once the factory is defined you use it to build objects.
There are two phases of use:
- Factory definition
- Object building
Factory Definition: Define your factory, giving it a name and optionally a constructor function (game
in this example):
Factory.define('game')
.sequence('id')
.attr('is_over', false)
.attr('created_at', () => new Date())
.attr('random_seed', () => Math.random())
// Default to two players. If players were given, fill in
// whatever attributes might be missing.
.attr('players', ['players'], (players) => {
if (!players) {
players = [{}, {}];
}
return players.map((data) => Factory.attributes('player', data));
});
Factory.define('player')
.sequence('id')
.sequence('name', (i) => {
return 'player' + i;
})
// Define `position` to depend on `id`.
.attr('position', ['id'], (id) => {
const positions = ['pitcher', '1st base', '2nd base', '3rd base'];
return positions[id % positions.length];
});
Factory.define('disabled-player').extend('player').attr('state', 'disabled');
Object Building: Build an object, passing in attributes that you want to override:
const game = Factory.build('game', { is_over: true });
// Built object (note scores are random):
//{
// id: 1,
// is_over: true, // overriden when building
// created_at: Fri Apr 15 2011 12:02:25 GMT-0400 (EDT),
// random_seed: 0.8999513240996748,
// players: [
// {id: 1, name:'Player 1'},
// {id: 2, name:'Player 2'}
// ]
//}
For a factory with a constructor, if you want just the attributes:
Factory.attributes('game'); // return just the attributes
You can specify options that are used to programmatically generate the attributes:
const moment = require('moment');
Factory.define('matches')
.attr('seasonStart', '2016-01-01')
.option('numMatches', 2)
.attr('matches', ['numMatches', 'seasonStart'], (numMatches, seasonStart) => {
const matches = [];
for (const i = 1; i <= numMatches; i++) {
matches.push({
matchDate: moment(seasonStart).add(i, 'week').format('YYYY-MM-DD'),
homeScore: Math.floor(Math.random() * 5),
awayScore: Math.floor(Math.random() * 5),
});
}
return matches;
});
Factory.build('matches', { seasonStart: '2016-03-12' }, { numMatches: 3 });
// Built object (note scores are random):
//{
// seasonStart: '2016-03-12',
// matches: [
// { matchDate: '2016-03-19', homeScore: 3, awayScore: 1 },
// { matchDate: '2016-03-26', homeScore: 0, awayScore: 4 },
// { matchDate: '2016-04-02', homeScore: 1, awayScore: 0 }
// ]
//}
In the example numMatches
is defined as an option
, not as an attribute
. Therefore numMatches
is not part of the output, it is only used to generate the matches
array.
In the same example seasonStart
is defined as an attribute
, therefore it appears in the output, and can also be used in the generator function that creates the matches
array.
The convenience function attrs
simplifies the common case of specifying multiple attributes in a batch. Rewriting the game
example from above:
Factory.define('game')
.sequence('id')
.attrs({
is_over: false,
created_at: () => new Date(),
random_seed: () => Math.random(),
})
.attr('players', ['players'], (players) => {
/* etc. */
});
You can also define a callback function to be run after building an object:
Factory.define('coach')
.option('buildPlayer', false)
.sequence('id')
.attr('players', ['id', 'buildPlayer'], (id, buildPlayer) => {
if (buildPlayer) {
return [Factory.build('player', { coach_id: id })];
}
})
.after((coach, options) => {
if (options.buildPlayer) {
console.log('built player:', coach.players[0]);
}
});
Factory.build('coach', {}, { buildPlayer: true });
Multiple callbacks can be registered, and they will be executed in the order they are registered. The callbacks can manipulate the built object before it is returned to the callee.
If the callback doesn't return anything, rosie will return build object as final result. If the callback returns a value, rosie will use that as final result instead.
This is an advanced use case that you can probably happily ignore, but store this away in case you need it.
When you define a factory you can optionally provide a class definition, and anything built by the factory will be passed through the constructor of the provided class.
Specifically, the output of .build
is used as the input to the constructor function, so the returned object is an instance of the specified class:
class SimpleClass {
constructor(args) {
this.moops = 'correct';
this.args = args;
}
isMoopsCorrect() {
return this.moops;
}
}
testFactory = Factory.define('test', SimpleClass).attr('some_var', 4);
testInstance = testFactory.build({ stuff: 2 });
console.log(JSON.stringify(testInstance, {}, 2));
// Output:
// {
// "moops": "correct",
// "args": {
// "stuff": 2,
// "some_var": 4
// }
// }
console.log(testInstance.isMoopsCorrect());
// Output:
// correct
Mind. Blown.
To use Rosie in node, you'll need to import it first:
import { Factory } from 'rosie';
// or with `require`
const Factory = require('rosie').Factory;
You might also choose to use unregistered factories, as it fits better with node's module pattern:
// factories/game.js
import { Factory } from 'rosie';
export default new Factory().sequence('id').attr('is_over', false);
// etc
To use the unregistered Game
factory defined above:
import Game from './factories/game';
const game = Game.build({ is_over: true });
A tool like babel is currently required to use this syntax.
You can also extend an existing unregistered factory:
// factories/scored-game.js
import { Factory } from 'rosie';
import Game from './game';
export default new Factory().extend(Game).attrs({
score: 10,
});
As stated above the rosie factory signatures can be broken into factory definition and object creation.
Additionally factories can be defined and accessed via the Factory singleton, or they can be created and maintained by the callee.
Once you have an instance returned from a Factory.define
or a new Factory()
call, you do the actual of work of defining the objects. This is done using the methods below (note these are typically chained together as in the examples above):
- Factory.define(
factory_name
) - Defines a factory by name. Return an instance of a Factory that you call.attr
,.option
,.sequence
, and.after
on the result to define the properties of this factory. - Factory.define(
factory_name
,constructor
) - Optionally pass a constuctor function, and the objects produced by.build
will be passed through theconstructor
function.
Use this to define attributes of your objects
- instance.attr(
attribute_name
,default_value
) -attribute_name
is required and is a string,default_value
is the value to use by default for the attribute - instance.attr(
attribute_name
,generator_function
) -generator_function
is called to generate the value of the attribute - instance.attr(
attribute_name
,dependencies
,generator_function
) -dependencies
is an array of strings, each string is the name of an attribute or option that is required by thegenerator_function
to generate the value of the attribute. This list ofdependencies
will match the parameters that are passed to thegenerator_function
Use this as a convenience function instead of calling instance.attr
multiple times
- instance.attrs(
{attribute_1: value_1, attribute_2: value_2, ...}
) -attribute_i
is a string,value_i
is either an object or generator function.
See instance.attr
above for details. Note: there is no way to specify dependencies using this method, so if you need that, you should use instance.attr
instead.
Use this to define options. Options do not appear in the generated object, but they can be used in a generator_function
that is used to configure an attribute or sequence that appears in the generated object. See the Programmatic Generation Of Attributes section for examples.
- instance.option(
option_name
,default_value
) -option_name
is required and is a string,default_value
is the value to use by default for the option - instance.option(
option_name
,generator_function
) -generator_function
is called to generate the value of the option - instance.option(
option_name
,dependencies
,generator_function
) -dependencies
is an array of strings, each string is the name of an option that is required by thegenerator_function
to generate the value of the option. This list ofdependencies
will match the parameters that are passed to thegenerator_function
Use this to define an auto incrementing sequence field in your object
- instance.sequence(
sequence_name
) - define a sequence calledsequence_name
, set the start value to 1 - instance.sequence(
sequence_name
,generator_function
) -generator_function
is called to generate the value of the sequence. When thegenerator_function
is called the pre-incremented sequence number will be passed as the first parameter, followed by any dependencies that have been specified. - instance.sequence(
sequence_name
,dependencies
,generator_function
) -dependencies
is an array of strings, each string is the name of an attribute or option that is required by thegenerator_function
to generate the value of the option. The value of each specified dependency will be passed as parameters 2..N to thegenerator_function
, noting again that the pre-incremented sequence number is passed as the first parameter.
- instance.after(
callback
) - register acallback
function that will be called at the end of the object build process. Thecallback
is invoked with two params: (build_object
,object_options
). See the Post Build Callback section for examples.
Returns an object that is generated by the named factory. attributes
and options
are optional parameters. The factory_name
is required when calling against the rosie Factory singleton.
- Factory.build(
factory_name
,attributes
,options
) - when build is called against the rosie Factory singleton, the first param is the name of the factory to use to build the object. The second is an object containing attribute override key value pairs, and the third is a object containing option key value pairs - instance.build(
attributes
,options
) - when build is called on a factory instance only theattributes
andoptions
objects are necessary
Identical to .build
except it returns an array of built objects. size
is required, attributes
and options
are optional
- Factory.buildList(
factory_name
, size,attributes
,options
) - when buildList is called against the rosie Factory singleton, the first param is the name of the factory to use to build the object. Theattributes
andoptions
behave the same as the call to.build
. - instance.buildList(size,
attributes
,options
) - when buildList is called on a factory instance only the size,attributes
andoptions
objects are necessary (strictly speaking only the size is necessary)
You may find resetAll
useful when working with testing frameworks such as Jest. It resets any build state, such as sequences, to their original values:
import Factory from 'rosie';
beforeEach(() => {
Factory.resetAll();
});
Or call reset
on a specific factory:
import Game from './game';
beforeEach(() => {
Game.reset();
});
- Fork it
- Create your feature branch (
git checkout -b my-new-feature
) - Install the test dependencies (
yarn install
- requires NodeJS and yarn) - Make your changes and make sure the tests pass (
yarn test
) - Commit your changes (
git commit -am 'Added some feature'
) - Push to the branch (
git push origin my-new-feature
) - Create new Pull Request
Thanks to Daniel Morrison for the name and Jon Hoyt for inspiration and brainstorming the idea.