Feel the thrill and enjoyment of testing when using Factories instead of Fixtures. Factories simplify the process of testing, making you more efficient and your tests more readable.
NEW and Improved starting with v2.7.0
- Support for using your factories in development environment
- Change your scenarios, hit refresh and the development data changes
Older but still fun things
- Support for ember-data-model-fragment usage is baked in since v2.5.0
- Support for ember-django-adapter usage is fried in since v2.6.1
- Support for adding meta data to payloads for use with ember-infinity ie. => pagination
- Support for adding headers to payloads
Why is FactoryGuy so awesome
- Since you using ember data, you don't need to create any ORM like things.
- You don't need to add any files to re create all the relationships in your models and serializers
- You don't have to setup anything besides making factories.
- Everything just works.
Questions: Slack => factory-guy
Contents:
- Installation
- How This Works
- Setup
- Defining Factories
- Using Factories
- Sequences
- Inline Function
- Traits
- Associations
- Extending Other Definitions
- Callbacks / Transient Attributes
- Using in Development
- Ember Data Model Fragments
- Creating Factories in Addons
- Ember Django Adapter
- Custom API formats
- Testing - creating scenarios
- Testing models, controllers, components
- Acceptance Tests
- Tips and Tricks
ChangeLog: ( Notes about what has changed in each version )
ember install ember-data-factory-guy
( ember-data-1.13.5+ )ember install [email protected]
( ember-data-1.13.0 + )ember install [email protected]
( ember-data-1.0.0-beta.19.1 )ember install [email protected]
( ember-data-1.0.0-beta.16.1 )
- remove ember-data-factory-guy from package.json
npm prune
ember install ember-data-factory-guy ( for latest release )
- You create factories for you models.
- put them in tests/factories directory
- Then you use them to create models in your tests.
- You can make records that persist in the store
- Or you can build a json payload used for mocking an ajax call's payload
In the following examples, assume the models look like this:
// standard models
User = DS.Model.extend({
name: DS.attr('string'),
style: DS.attr('string'),
projects: DS.hasMany('project'),
hats: DS.hasMany('hat', {polymorphic: true})
});
Project = DS.Model.extend({
title: DS.attr('string'),
user: DS.belongsTo('user')
});
// polymorphic models
Hat = DS.Model.extend({
type: DS.attr('string'),
user: DS.belongsTo('user')
});
BigHat = Hat.extend();
SmallHat = Hat.extend();
- A factory has a name and a set of attributes.
- The name should match the model type name. So, for 'User' model, the name would be 'user'
- Create factory files in the tests/factories directory.
- Can use generator to create the outline of a factory file:
ember g factory user
This will create a file named user.js in the tests/factories directory.
-
Brief sample of a factory definition
// file tests/factories/user.js
import FactoryGuy from 'ember-data-factory-guy';
FactoryGuy.define('user', {
// Put default 'user' attributes in the default section
default: {
style: 'normal',
name: 'Dude'
},
// Create a named 'user' with custom attributes
admin: {
style: 'super',
name: 'Admin'
}
});
- If you are using an attribute named 'type' and this is not a polymorphic model, use the option
polymorphic: false
in your definition
// file: tests/factories/cat.js
FactoryGuy.define('cat', {
polymorphic: false, // manually flag this model as NOT polymorphic
default: {
// usually, an attribute named 'type' is for polymorphic models, but the defenition
// is set as NOT polymorphic, which allows this type to work as attibute
type: 'Cute',
name: (f)=> `Cat ${f.id}`
}
});
- Define each polymorphic model in it's own typed definition
- The attribute named 'type' is used to hold the model name
- May want to extend parent factory here
// file tests/factories/small-hat.js
import FactoryGuy from 'ember-data-factory-guy';
FactoryGuy.define('small-hat', {
default: {
type: 'SmallHat'
}
})
// file tests/factories/big-hat.js
import FactoryGuy from 'ember-data-factory-guy';
FactoryGuy.define('big-hat', {
default: {
type: 'BigHat'
}
})
In other words, don't do this:
// file tests/factories/hat.js
import FactoryGuy from 'ember-data-factory-guy';
FactoryGuy.define('hat', {
default: {},
small-hat: {
type: 'SmallHat'
},
big-hat: {
type: 'BigHat'
}
})
- FactoryGuy.make
- Loads model instance into the store
- FactoryGuy.makeList
- Loads zero to many model instances into the store
- FactoryGuy.build
- Builds json in accordance with the adapters specifications
- RESTAdapter (assume this adapter being used in most of the following examples)
- ActiveModelAdapter
- JSONAPIAdapter
- DrfAdapter (Ember Django Adapter)
- Builds json in accordance with the adapters specifications
- FactoryGuy.buildList
- Builds json with list of zero or more items in accordance with the adapters specifications
- Can override default attributes by passing in an object of options
- Can add attributes or relationships with traits
- Can compose relationships
- By passing in other objects you've made with build/buildList or make/makeList
- all instances loaded into the ember data store
Usage:
import { make } from 'ember-data-factory-guy';
// make basic user with the default attributes in user factory
let user = make('user');
user.toJSON({includeId: true}) // => {id: 1, name: 'User1', style: 'normal'}
// make user with default attributes plus those defined as 'admin' in user factory
let user = make('admin');
user.toJSON({includeId: true}) // => {id: 2, name: 'Admin', style: 'super'}
// make user with default attributes plus these extra attributes
let user = make('user', {name: 'Fred'});
user.toJSON({includeId: true}) // => {id: 3, name: 'Fred', style: 'normal'}
// make admin defined user with these extra attributes
let user = make('admin', {name: 'Fred'});
user.toJSON({includeId: true}) // => {id: 4, name: 'Fred', style: 'super'}
// make default user with traits and with extra attributes
let user = make('user', 'silly', {name: 'Fred'});
user.toJSON({includeId: true}) // => {id: 5, name: 'Fred', style: 'silly'}
// make user with hats relationship ( hasMany ) composed of a few pre 'made' hats
let hat1 = make('big-hat');
let hat2 = make('big-hat');
let user = make('user', {hats: [hat1, hat2]});
user.toJSON({includeId: true})
// => {id: 6, name: 'User2', style: 'normal', hats: [{id:1, type:"big_hat"},{id:1, type:"big_hat"}]}
// note that hats are polymorphic. if they weren't, the hats array would be a list of ids: [1,2]
// make user with company relationship ( belongsTo ) composed of a pre 'made' company
let company = make('company');
let user = make('user', {company: company});
user.toJSON({includeId: true}) // => {id: 7, name: 'User3', style: 'normal', company: 1}
- check out (user factory): to see 'bob' user and 'with_car' trait
Usage:
import { make, makeList } from 'ember-data-factory-guy';
// Let's say bob is a named type in the user factory
makeList('bob') // makes 0 bob's
makeList('bob', 2) // makes 2 bob's
makeList('bob', 2, 'with_car' , {name: "Dude"})
// makes 2 bob's that have 'with_car' trait and name of "Dude"
// In other words, applies the traits and options to every bob made
makeList('bob', 'with_car', ['with_car',{name: "Dude"}])
// 2 User models with bob attributes, where the first also has 'with_car' trait
// the last has 'with_car' trait and name of "Dude", so you get 2 different bob's
- for building json that you can pass as json payload in acceptance tests
- takes same arguments as make
- can compose relationships with other build/buildList payloads
- to inspect the json use the get() method
- use add method
- to include extra sideloaded data to the payload
- to include meta data
Usage:
import { build, buildList } from 'ember-data-factory-guy';
// build basic user with the default attributes in user factory
let json = build('user');
json.get() // => {id: 1, name: 'User1', style: 'normal'}
// build user with default attributes plus those defined as 'admin' in user factory
let json = build('admin');
json.get() // => {id: 2, name: 'Admin', style: 'super'}
// build user with default attributes plus these extra attributes
let json = build('user', {name: 'Fred'});
json.get() // => {id: 3, name: 'Fred', style: 'normal'}
// build admin defined user with extra attributes
let json = build('admin', {name: 'Fred'});
json.get() // => {id: 4, name: 'Fred', style: 'super'}
// build default user with traits and with extra attributes
let json = build('user', 'silly', {name: 'Fred'});
json.get() // => {id: 5, name: 'Fred', style: 'silly'}
// build user with hats relationship ( hasMany ) composed of a few pre 'built' hats
let hat1 = build('big-hat');
let hat2 = build('big-hat');
let json = build('user', {hats: [hat1, hat2]});
// note that hats are polymorphic. if they weren't, the hats array would be a list of ids: [1,2]
json.get() // => {id: 6, name: 'User2', style: 'normal', hats: [{id:1, type:"big_hat"},{id:1, type:"big_hat"}]}
// build user with company relationship ( belongsTo ) composed of a pre 'built' company
let company = build('company');
let json = build('user', {company: company});
json.get() // => {id: 7, name: 'User3', style: 'normal', company: 1}
// build and compose relationships to unlimited degree
let company1 = build('company', {name: 'A Corp'});
let company2 = build('company', {name: 'B Corp'});
let owners = buildList('user', { company:company1 }, { company:company2 });
let buildJson = build('property', { owners });
- Example of what json payload from build looks like
- Although the RESTAdapter is being used, this works the same with ActiveModel or JSONAPI adapters
let json = build('user', 'with_company', 'with_hats');
json // =>
{
user: {
id: 1,
name: 'User1',
company: 1,
hats: [
{type: 'big_hat', id:1},
{type: 'big_hat', id:2}
]
},
companies: [
{id: 1, name: 'Silly corp'}
],
'big-hats': [
{id: 1, type: "BigHat" },
{id: 2, type: "BigHat" }
]
}
- for building json that you can pass as json payload in acceptance tests
- takes the same arguments as makeList
- can compose relationships with other build/buildList payloads
- to inspect the json use the get() method
- can use get(index) to get to items in the list
- use add method
- to add extra sideloaded data to the payload =>
.add({json})
- to add meta data =>
.add({meta})
- to add extra sideloaded data to the payload =>
Usage:
import { build, buildList } from 'ember-data-factory-guy';
let bobs = buildList('bob', 2); // builds 2 Bob's
let bobs = buildList('bob', 2, {name: 'Rob'); // builds 2 Bob's with name of 'Rob'
// builds 2 users, one with name 'Bob' , the next with name 'Rob'
let users = buildList('user', { name:'Bob' }, { name:'Rob' });
// builds 2 users, one with 'boblike' and the next with name 'adminlike' features
// NOTE: you don't say how many to make, because each trait is making new user
let users = buildList('user', 'boblike', 'adminlike');
// builds 2 users:
// one 'boblike' with stoner style
// and the next 'adminlike' with square style
// NOTE: how you are grouping traits and attributes for each one by wrapping them in array
let users = buildList('user', ['boblike',{ style: 'stoner' }], ['adminlike', {style: 'square'}]);
- when you need to add more json to payload
- will be sideloaded
- only JSONAPI, and REST based serializers can do sideloading
- so DRFSerializer and JSONSerializer users can not user this feature
Usage:
let batMan = build('bat_man');
let userPayload = build('user').add({json:batMan});
userPayload = {
user: {
id: 1,
name: 'User1',
style: "normal"
},
'super-heros': [
{
id: 1,
name: "BatMan",
type: "SuperHero"
}
]
};
- when you want to add meta data to payload
- only JSONAPI, and REST based and serializers and DRFSerializer can handle meta data
- so JSONSerializer users can not user this feature ( though this might be a bug on my part )
Usage:
let json1 = buildList('profile', 2).add({ meta: { previous: '/profiles?page=1', next: '/profiles?page=3' } });
let json2 = buildList('profile', 2).add({ meta: { previous: '/profiles?page=2', next: '/profiles?page=4' } });
mockQuery('profile', {page: 2}).returns({ json: json1 });
mockQuery('profile', {page: 3}).returns({ json: json2 });
store.query('profile', {page: 2}).then((records)=> // first 2 from json1
store.query('profile', {page: 3}).then((records)=> // second 2 from json2
- for inspecting contents of json payload
- get() returns all attributes of top level model
- get(attribute) gives you attribute in top level model
- get(index) gives you the info for hasMany relationship at that index
- get(relationships) gives you just id or type ( if polymorphic )
- better to compose the build relationships by hand if you need more info
- check out user factory: to see 'boblike' and 'adminlike' user traits
let json = build('user');
json.get() //=> {id: 1, name: 'User1', style: 'normal'}
json.get('id') // => 1
let json = buildList('user', 2);
json.get(0) //=> {id: 1, name: 'User1', style: 'normal'}
json.get(1) //=> {id: 2, name: 'User2', style: 'normal'}
let json = buildList('user', 'boblike', 'adminlike');
json.get(0) //=> {id: 1, name: 'Bob', style: 'boblike'}
json.get(1) //=> {id: 2, name: 'Admin', style: 'super'}
- building relationships inline
let json = build('user', 'with_company', 'with_hats');
json.get() //=> {id: 1, name: 'User1', style: 'normal'}
// to get hats (hasMany relationship) info
json.get('hats') //=> [{id: 1, type: "big_hat"},{id: 1, type: "big_hat"}]
// to get company ( belongsTo relationship ) info
json.get('company') //=> {id: 1, type: "company"}
- by composing the relationships you can get the full attributes of those associations
let company = build('company');
let hats = buildList('big-hats');
let user = build('user', {company , hats});
user.get() //=> {id: 1, name: 'User1', style: 'normal'}
// to get hats info from hats json
hats.get(0) //=> {id: 1, type: "BigHat", plus .. any other attributes}
hats.get(1) //=> {id: 2, type: "BigHat", plus .. any other attributes}
// to get company info
company.get() //=> {id: 1, type: "Company", name: "Silly corp"}
- For generating unique attribute values.
- Can be defined:
- In the model definition's sequences hash
- Inline on the attribute
- Values are generated by calling FactoryGuy.generate
FactoryGuy.define('user', {
sequences: {
userName: (num)=> `User${num}`
},
default: {
// use the 'userName' sequence for this attribute
name: FactoryGuy.generate('userName')
}
});
let json = FactoryGuy.build('user');
json.get('name') // => 'User1'
let user = FactoryGuy.make('user');
user.get('name') // => 'User2'
FactoryGuy.define('project', {
special_project: {
title: FactoryGuy.generate((num)=> `Project #${num}`)
},
});
let json = FactoryGuy.build('special_project');
json.get('title') // => 'Project #1'
let project = FactoryGuy.make('special_project');
project.get('title') // => 'Project #2'
- Declare a function for an attribute
- The fixture is passed as parameter so you can reference all other attributes, even id
FactoryGuy.define('user', {
default: {
// Don't need the userName sequence, since the id is almost
// always a sequential number, and you can use that.
// f is the fixture being built as the moment for this factory
// definition, which has the id available
name: (f)=> `User${f.id}`
},
traits: {
boringStyle: {
style: (f)=> `${f.id} boring `
}
funnyUser: {
style: (f)=> `funny ${f.name}`
}
}
});
let json = FactoryGuy.build('user', 'funny');
json.get('name') // => 'User1'
json.get('style') // => 'funny User1'
let user = FactoryGuy.make('user', 'boring');
user.get('id') // => 2
user.get('style') // => '2 boring'
Note the style attribute was built from a function which depends on the name and the name is a generated attribute from a sequence function
- Used with build, buildList, make, or makeList
- For grouping attributes together
- Can use one or more traits in a row
- The last trait included overrides any values in traits before it
FactoryGuy.define('user', {
traits: {
big: { name: 'Big Guy' }
friendly: { style: 'Friendly' }
}
});
let json = FactoryGuy.build('user', 'big', 'friendly');
json.get('name') // => 'Big Guy'
json.get('style') // => 'Friendly'
let user = FactoryGuy.make('user', 'big', 'friendly');
user.get('name') // => 'Big Guy'
user.get('style') // => 'Friendly'
You can still pass in a hash of options when using traits. This hash of attributes will override any trait attributes or default attributes
let user = FactoryGuy.make('user', 'big', 'friendly', {name: 'Dave'});
user.get('name') // => 'Dave'
user.get('style') // => 'Friendly'
- Can setup belongsTo or hasMany associations in factory definitions
- As inline attribute definition
- With traits
- Can setup belongsTo or hasMany associations manually
// Recall ( from above setup ) that there is a user belongsTo on the Project model
// Also, assume 'user' factory is same as from 'user' factory definition above in
// 'Defining Factories' section
FactoryGuy.define('project', {
project_with_user: {
// create user model with default attributes
user: {}
},
project_with_bob: {
// create user model with custom attributes
user: {name: 'Bob'}
},
project_with_admin: {
// create a named user model with the FactoryGuy.belongsTo helper method
user: FactoryGuy.belongsTo('admin')
}
});
let project = FactoryGuy.make('project_with_admin');
project.get('user.name') // => 'Admin'
project.get('user.style') // => 'super'
You could also accomplish the above with traits:
FactoryGuy.define('project', {
traits: {
with_user: { user: {} },
with_admin: { user: FactoryGuy.belongsTo('admin') }
}
});
let user = FactoryGuy.make('project', 'with_user');
project.get('user').toJSON({includeId: true}) // => {id:1, name: 'Dude', style: 'normal'}
let user = FactoryGuy.make('user');
let project = FactoryGuy.make('project', {user: user});
project.get('user').toJSON({includeId: true}) // => {id:1, name: 'Dude', style: 'normal'}
Note that though you are setting the 'user' belongsTo association on a project, the reverse user hasMany 'projects' association is being setup for you on the user ( for both manual and factory defined belongsTo associations ) as well
user.get('projects.length') // => 1
FactoryGuy.define('user', {
user_with_projects: { projects: FactoryGuy.hasMany('project', 2) }
});
let user = FactoryGuy.make('user_with_projects');
user.get('projects.length') // => 2
You could also accomplish the above with traits:
FactoryGuy.define('project', {
traits: {
with_projects: {
projects: FactoryGuy.hasMany('project', 2)
}
}
});
let user = FactoryGuy.make('user', 'with_projects');
user.get('projects.length') // => 2
let project1 = FactoryGuy.make('project');
let project2 = FactoryGuy.make('project');
let user = FactoryGuy.make('user', {projects: [project1,project2]});
user.get('projects.length') // => 2
// or
let projects = FactoryGuy.makeList('project', 2);
let user = FactoryGuy.make('user', {projects: projects});
user.get('projects.length') // => 2
Note that though you are setting the 'projects' hasMany association on a user, the reverse 'user' belongsTo association is being setup for you on the project ( for both manual and factory defined hasMany associations ) as well
projects.get('firstObject.user') // => user
-
Extending another definition will inherit these sections:
- sequences
- traits
- default attributes
-
Inheritance is fine grained, so in each section, any attribute that is local will take precedence over an inherited one. So you can override some attributes in the default section ( for example ), and inherit the rest
- afterMake
- Uses transient attributes
Assuming the factory-guy model definition defines afterMake function:
FactoryGuy.define('property', {
default: {
name: 'Silly property'
},
// optionally set transient attributes, that will be passed in to afterMake function
transient: {
for_sale: true
},
// The attributes passed to after make will include any optional attributes you
// passed in to make, and the transient attributes defined in this definition
afterMake: function(model, attributes) {
if (attributes.for_sale) {
model.set('name', model.get('name') + '(FOR SALE)');
}
}
}
You would use this to make models like:
Ember.run(function () {
let property = FactoryGuy.make('property');
property.get('name'); // => 'Silly property(FOR SALE)')
let property = FactoryGuy.make('property', {for_sale: false});
property.get('name'); // => 'Silly property')
});
-
You can set up scenarios for you app that use all your factories from tests
-
In config/environment.js place a flag => factoryGuy: true
// file: config/environment.js if (environment === 'development') { ENV.factoryGuy = true; ENV.locationType = 'auto'; ENV.rootURL = '/'; }
-
Place your scenarios in app/scenarios directory
- Start by creating at least a scenarios/main.js file since this is the starting point
- Your scenario classes should inherit from Scenario class
- A Scenario class should declare a run method where you do things like:
- include other scenarios
- Meaning you can compose scenarios like a symphony of notes
- make your data or mock your requests using the typical FactoryGuy methods
- these methods are all built into Scenario classes so you don't have to import them
- include other scenarios
// file: app/scenarios/main.js import {Scenario} from 'ember-data-factory-guy'; import Users from './users'; // Just for fun, set the log level ( to 1 ) and see all FactoryGuy response info in console Scenario.settings({ logLevel: 1, // 1 is the max for now, default is 0 }); export default class extends Scenario { run() { this.include([Users]); // include other scenarios this.mockFindAll('products', 3); // mock some finds } }
// file: app/scenarios/users.js import {Scenario} from 'ember-data-factory-guy'; export default class extends Scenario { run() { this.mockFindAll('user', 'boblike', 'normal'); this.mockDelete('user'); } }
As of 2.5.2 you can create factories which contain ember-data-model-fragments. Setting up your fragments is easy and follows the same process as setting up regular factories. The mapping between fragment types and their associations are like so:
Fragment Type | Association |
---|---|
fragment |
FactoryGuy.belongsTo |
fragmentArray |
FactoryGuy.hasMany |
array |
[] |
For example, say we have the following Employee
model which makes use of the fragment
, fragmentArray
and array
fragment types.
//Employee model
export default Model.extend({
name : fragment('name'),
phoneNumbers: fragmentArray('phone-number')
})
//Name fragment
export default Fragment.extend({
titles: array('string'),
firstName : attr('string'),
lastName : attr('string')
});
//Phone Number fragment
export default Fragment.extend({
number: attr('string')
type: attr('string')
});
A factory for this model and its fragments would look like so:
// Employee factory
FactoryGuy.define('employee', {
default: {
name: FactoryGuy.belongsTo('name'), //fragment
phoneNumbers: FactoryGuy.hasMany('phone-number') //fragmentArray
}
});
// Name fragment factory
FactoryGuy.define('employee', {
default: {
titles: ['Mr.', 'Dr.'], //array
firstName: 'Jon',
lastName: 'Snow'
}
});
// Phone number fragment factory
FactoryGuy.define('phone-number', {
default: {
number: '123-456-789',
type: 'home'
}
});
// TIP: You can set up associations manually ( and not necessarily in factory )
// To set up an employee ( hasMany ) phone numbers manually, try this:
let phoneNumbers = buildList('phone-numbers', 2).get();
let employee = make('employee', { phoneNumbers });
For a more detailed example of setting up fragments have a look at:
- model test employee test.
- acceptance test employee-view-test.
If you are making an addon with factories and you want the factories available to Ember apps using your addon, place the factories in test-support/factories
instead of tests/factories
. They should be available both within your addon and in Ember apps using your addon.
- As of 2.6.1 you can use ember-django-adapter
- Everything is setup automatically
- Remember that sideloading is not supported in DRFSerializer so all relationships should either
- be set as embedded with DS.EmbeddedRecordsMixin if you want to use build/buildList
- or user make/makeList and in you mocks, and return models instead of json:
let projects = makeList('projects', 2); // put projects in the store
let user = make('user', { projects }); // attatch them to user
mockFindRecord('user').returns({model: user}); // now the mock will return a user that has projects
- using
fails()
with errors hash is not working reliably- so you can always just
mockWhatever(args).fails()
- so you can always just
FactoryGuy handles JSON-API / RESTSerializer / JSONSerializer out of the box.
In case your API doesn't follow either of these conventions, you can still make a custom fixture builder, or modify the FixtureConverters and JSONPayload classes that exist.
- For now, before I launch into the details, let me know if you need this hookup and I can guide you to a solution, since the use cases will be rare and varied.
- Easy to create complex scenarios involving multi layered relationships.
- Can use model instances to create relationships for making other models.
Example:
- Setup a scenario where a user has two projects and belongs to a company
let company = make('company');
let user = make('user', {company: company});
let projects = makeList('project', 2, {user: user});
You can use traits to help create the relationships as well, but this strategy allows you to build up complex scenarios in a different way that has it's own benefits.
####cacheOnlyMode
- FactoryGuy.cacheOnlyMode
- Allows you to setup the adapters to prevent them from fetching data with ajax call
- for single models ( find ) you have to put something in the store
- for collections ( findAll ) you don't have to put anything in the store.
- Takes
except
parameter as a list of models you don't want to cache.- These model requests will go to server with ajax call and need to be mocked.
- Allows you to setup the adapters to prevent them from fetching data with ajax call
This is helpful, when:
- you want to set up the test data with make/makeList, and then prevent calls like store.find or findAll from fetching more data, since you have already setup the store with make/makeList data.
- you have an application that starts up and loads data that is not relevant to the test page you are working on.
Usage:
import FactoryGuy, { makeList } from 'ember-data-factory-guy';
import moduleForAcceptance from '../helpers/module-for-acceptance';
moduleForAcceptance('Acceptance | Profiles View');
test("Using FactoryGuy.cacheOnlyMode", function() {
FactoryGuy.cacheOnlyMode();
// the store.find call for the user will go out unless there is a user
// in the store
make('user', {name: 'current'});
// the application starts up and makes calls to findAll a few things, but
// those can be ignored because of the cacheOnlyMode
// for this test I care about just testing profiles
makeList("profile", 2);
visit('/profiles');
andThen(()=> {
// test stuff
});
});
test("Using FactoryGuy.cacheOnlyMode with except", function() {
FactoryGuy.cacheOnlyMode({except: ['profile']});
make('user', {name: 'current'});
// this time I want to allow the ajax call so I can return built json payload
mockFindAll("profile", 2);
visit('/profiles');
andThen(()=> {
// test stuff
});
});
-
FactoryGuy needs to setup the factories before the test run.
- By default, you only need to call
manualSetup(this.container)
in unit/component tests - If you are using mock methods, such as mockQuery, use
mockSetup()
andmockTeardown()
- By default, you only need to call
-
Sample model test (profile-test.js):
- Use 'moduleForModel' ( ember-qunit ), or describeModel ( ember-mocha ) test helper
- manually set up Factory guy
-
Sample component test #1 (single-user-manual-setup-test.js):
- Using 'moduleForComponent' ( ember-qunit ), or describeComponent ( ember-mocha ) helper
- Manually sets up Factory guy ( so it's faster )
import { make, manualSetup } from 'ember-data-factory-guy'; import hbs from 'htmlbars-inline-precompile'; import { test, moduleForComponent } from 'ember-qunit'; moduleForComponent('single-user', 'Integration | Component | single-user (manual setup)', { integration: true, beforeEach: function () { manualSetup(this.container); } }); test("shows user information", function () { let user = make('user', {name: 'Rob'}); this.render(hbs`{{single-user user=user}}`); this.set('user', user); ok(this.$('.name').text().match(user.get('name'))); ok(this.$('.funny-name').text().match(user.get('funnyName'))); });
-
Sample component test #2 (single-user-test.js):
- Using 'moduleForComponent' ( ember-qunit ), or describeComponent ( ember-mocha ) helper
- Starts a new application with startApp() before each test ( slower )
-
Uses mockjax
- for mocking the ajax calls made by ember-data.
- this library is installed with ember-data-factory-guy.
-
http GET mocks
- mockFindRecord
- mockFindAll
- mockReload
- mockQuery
- mockQueryRecord
- takes modifier method
returns()
for setting the payload responsereturns()
accepts parameters like: json, model, models, id, ids, headers- headers are cumulative so you can add as many as you like
- Example:
let mock = mockFindAll('user').returns({headers: {'X-Man': "Wolverine"}); mock.returns({headers: {'X-Weapon': "Claws"});
- these mocks are are reusable
- so you can simulate making the same ajax call ( url ) and return a different payload
-
http POST/PUT/DELETE
-
Use method
fails()
to simulate failure -
Use method
succeeds()
to simulate success -
Only used if the mock was set to fail with
fails()
and you want to set the mock to succeed to simulate a successful retry -
Use property
timesCalled
to verify how many times the ajax call was mocked- works when you are using mockQuery, mockQueryRecord, mockFindAll, or mockUpdate
- mockFindRecord will always be at most 1 since it will only make ajax call the first time, and then the store will use cache the second time
- Example:
const mock = mockQueryRecord('company', {}).returns({ json: build('company') }); FactoryGuy.store.queryRecord('company', {}).then(()=> { FactoryGuy.store.queryRecord('company', {}).then(()=> { mock.timesCalled //=> 2 }); });
-
Use method
disable()
to temporarily disable the mock. You can re-enable the disabled mock usingenable()
. -
Use method
destroy()
to completely remove the mockjax handler for the mock. TheisDestroyed
property is set totrue
when the mock is destroyed.
- Use
mockSetup()
in test setup/beforeEach - set logging options here:
- logLevel ( 0 - off , 1 - on ) for seeing the factory guy responses
- responseTime ( in millis ) for simulating slower repsonses
- mockjaxLogLevel ( 1 - 4 max ) for seeing mockjax logging
- Example:
javascript mockSetup({logLevel: 1, responseTime: 1000, mockjaxLogLevel: 4});
- Use
mockTeardown()
in test teardown/afterEach - Easiest is to set them up in module-for-acceptance.js:
-
Usable on all mocks
-
Use optional object arguments status and response and convertErrors to customize
- status : must be number in the range of 3XX, 4XX, or 5XX ( default is 500 )
- response : must be object with errors key ( default is null )
- convertErrors : set to false and object will be left untouched ( default is true )
- errors must be in particular format for ember-data to accept them
- factory guy allows you to use a simple style:
{errors: {name: "Name too short"}}
- Behind the scenes converts to another format for ember-data to consume
- factory guy allows you to use a simple style:
- errors must be in particular format for ember-data to accept them
-
Examples:
let errors401 = {errors: {description: "Unauthorized"}};
let mock = mockFindAll('user').fails({status: 401, response: errors401});
let errors422 = {errors: {name: "Name too short"}};
let mock = mockFindRecord('profile').fails({status: 422, response: errors422});
let errorsMine = {errors: [{detail: "Name too short", title: "I am short"}]};
let mock = mockFindRecord('profile').fails({status: 422, response: errorsMine, convertErrors: false});
- For dealing with finding one record of a model type =>
store.find('modelType')
- Can pass in arguments just like you would for make or build
- mockFindRecord( fixture or model name, optional traits, optional attributes object)
- Takes modifier method
returns()
for controlling the response payload- returns( model / json / id )
- Sample acceptance tests using mockFindRecord: user-view-test.js:
Usage:
import { build, make, mockFindRecord } from 'ember-data-factory-guy';
- To return default factory model type ( 'user' in this case )
// mockFindRecord automatically returns json for the modelType ( in this case 'user' )
let mock = mockFindRecord('user');
let userId = mock.get('id');
- Using
returns({json})
to return json object
let user = build('user', 'whacky', {isDude: true});
let mock = mockFindRecord('user').returns({ json: user });
// user.get('id') => 1
// user.get('style') => 'whacky'
// or to acccomplish the same thing with less code
let mock = mockFindRecord('user', 'whacky', {isDude: true});
// mock.get('id') => 1
// mock.get('style') => 'whacky'
let user = mock.get();
// user.id => 1
// user.style => 'whacky'
- Using
returns({model})
to return model instance
let user = make('user', 'whacky', {isDude: false});
let mock = mockFindRecord('user').returns({ model: user });
// user.get('id') => 1
// you can now also user.get('any-computed-property')
// since you have a real model instance
- Simper way to return a model instance
let user = make('user', 'whacky', {isDude: false});
let mock = mockFindRecord(user);
// user.get('id') === mock.get('id')
// basically a shortcut to the above .returns({ model: user })
// as this sets up the returns for you
- To reuse the mock
let user2 = build('user', {style: "boring"});
mock.returns({ json: user2 });
// mock.get('id') => 2
- To mock failure case use
fails
method
mockFindRecord('user').fails();
- To mock failure when you have a model already
let profile = make('profile');
mockFindRecord(profile).fails();
// mock.get('id') === profile.id
- For dealing with finding all records for a model type =>
store.findAll(modelType)
- Takes same parameters as makeList
- mockFindAll( fixture or model name, optional number, optional traits, optional attributes object)
- Takes modifier method
returns()
for controlling the response payload- returns( models / json / ids )
- Sample acceptance tests using mockFindAll: users-view-test.js
Usage:
import { buildList, makeList, mockFindAll } from 'ember-data-factory-guy';
- To mock and return no results
let mock = mockFindAll('user');
- Using
returns({json})
to return json object
// that has 2 different users:
let users = buildList('user', 'whacky', 'silly');
let mock = mockFindAll('user').returns({ json: users });
let user1 = users.get(0);
let user2 = users.get(1);
// user1.style => 'whacky'
// user2.style => 'silly'
// or to acccomplish the same thing with less code
let mock = mockFindAll('user', 'whacky', 'silly');
let user1 = mock.get(0);
let user2 = mock.get(1);
// user1.style => 'whacky'
// user2.style => 'silly'
- Using
returns({models})
to return model instances
let users = makeList('user', 'whacky', 'silly');
let mock = mockFindAll('user').returns({ models: users });
let user1 = users[0];
// you can now also user1.get('any-computed-property')
// since you have a real model instance
- To reuse the mock and return different payload
let users2 = buildList('user', 3);
mock.returns({ json: user2 });
- To mock failure case use
fails()
method
mockFindAll('user').fails();
- To handle reloading a model
- Pass in a record ( or a typeName and id )
Usage:
- Passing in a record / model instance
let profile = make('profile');
mockReload(profile);
// will stub a call to reload that profile
profile.reload()
- Using
returns({attrs})
to return new attributes
let profile = make('profile', { description: "whatever" });
mockReload(profile).returns({ attrs: { description: "moo" } });
profile.reload(); // description is now "moo"
- Using
returns({json})
to return new all new attributes
let profile = make('profile', { description: "tomatoes" });
// all new values EXCEPT the profile id ( you should keep that id the same )
let profileAllNew = build('profile', { id: profile.get('id'), description: "potatoes" }
mockReload(profile).returns({ json: profileAllNew });
profile.reload(); // description = "potatoes"
- Mocking a failed reload
mockReload('profile', 1).fails();
- For dealing with querying for all records for a model type =>
store.query(modelType, params)
- Takes modifier method
returns()
for controlling the response payload - returns( models / json / ids )
- Takes modifier method
- Takes modifier methods for matching the query params - withParams( object ) - withSomeParams( object )
- Sample acceptance tests using mockQuery: user-search-test.js
Usage:
import FactoryGuy, { make, build, buildList, mockQuery } from 'ember-data-factory-guy';
let store = FactoryGuy.store;
// This simulates a query that returns no results
mockQuery('user', {age: 10});
store.query('user', {age: 10}}).then((userInstances) => {
/// userInstances will be empty
})
- with returns( models )
// Create model instances
let users = makeList('user', 2, 'with_hats');
mockQuery('user', {name:'Bob', age: 10}).returns({models: users});
store.query('user', {name:'Bob', age: 10}}).then((models)=> {
// models are the same as the users array
});
- with returns ( json )
// Create json with buildList
let users = buildList('user', 2, 'with_hats');
mockQuery('user', {name:'Bob', age: 10}).returns({json: users});
store.query('user', {name:'Bob', age: 10}}).then((models)=> {
// these models were created from the users json
});
- with returns( ids )
// Create list of models
let users = buildList('user', 2, 'with_hats');
let user1 = users.get(0);
mockQuery('user', {name:'Bob', age: 10}).returns({ids: [user1.id]});
store.query('user', {name:'Bob', age: 10}}).then(function(models) {
// models will be one model and it will be user1
});
- withParams() / withSomeParams()
// Create list of models
let users = buildList('user', 2, 'with_hats');
let user1 = users.get(0);
mock = mockQuery('user').returns({ids: [user1.id]});
mock.withParams({name:'Bob', age: 10})
// When using 'withParams' modifier, params hash must match exactly
store.query('user', {name:'Bob', age: 10}}).then(function(models) {
// models will be one model and it will be user1
});
// The following call will not be caught by the mock
store.query('user', {name:'Bob', age: 10, hair: 'brown'}})
// 'withSomeParams' is designed to catch requests by partial match
// It has precedence over strict params matching once applied
mock.withSomeParams({name:'Bob'})
// Now both requests will be intercepted
store.query('user', {name:'Bob', age: 10}})
store.query('user', {name:'Bob', age: 10, hair: 'brown'}})
- For dealing with querying for one record for a model type =>
store.queryRecord(modelType, params)
- Takes modifier method
returns()
for controlling the response payload - returns( model / json / id )
- Takes modifier method
- Takes modifier methods for matching the query params - withParams( object )
Usage:
import FactoryGuy, { make, build, mockQueryRecord } from 'ember-data-factory-guy';
let store = FactoryGuy.store;
// This simulates a query that returns no results
mockQueryRecord('user', {age: 10});
store.queryRecord('user', {age: 10}}).then((userInstance) => {
/// userInstance will be empty
})
- with returns( models )
// Create model instances
let user = make('user');
mockQueryRecord('user', {name:'Bob', age: 10}).returns({model: user});
store.queryRecord('user', {name:'Bob', age: 10}}).then((model)=> {
// model is the same as the user you made
});
- with returns ( json )
// Create json with buildList
let user = build('user');
mockQueryRecord('user', {name:'Bob', age: 10}).returns({json: user});
store.queryRecord('user', {name:'Bob', age: 10}}).then((model)=> {
// user model created from the user json
});
- with returns( ids )
// Create list of models
let user = build('user', 'with_hats');
mockQueryRecord('user', {name:'Bob', age: 10}).returns({id: user.get('id')});
store.queryRecord('user', {name:'Bob', age: 10}}).then(function(model) {
// model will be one model and it will be user1
});
- Use chainable methods to build the response
- match
- Attributes that must be in request json
- These will be added to the response json automatically, so you don't need to include them in the returns hash.
- If you match on a belongsTo association, you don't have to include that in the returns hash either ( same idea ).
- Attributes that must be in request json
- returns
- Attributes ( including relationships ) to include in response json
- match
- Need to wrap tests using mockCreate with: Ember.run(function() { 'your test' })
Realistically, you will have code in a view action or controller action that will create the record, and setup any associations.
// most actions that create a record look something like this:
action: {
addProject: function (user) {
let name = this.$('button.project-name').val();
this.store.createRecord('project', {name: name, user: user}).save();
}
}
In this case, you are are creating a 'project' record with a specific name, and belonging to a particular user. To mock this createRecord call here are a few ways to do this using chainable methods.
Usage:
import { mockCreate } from 'ember-data-factory-guy';
// Simplest case
// Don't care about a match just handle createRecord for any project
mockCreate('project');
// Matching some attributes
mockCreate('project').match({name: "Moo"});
// Match all attributes
mockCreate('project').match({name: "Moo", user: user});
// Exactly matching attributes, and returning extra attributes
mockCreate('project')
.match({name: "Moo", user: user})
.returns({created_at: new Date()});
// Returning belongsTo relationship. Assume outfit belongsTo 'person'
let person = build('super-hero'); // it's polymorphic
mockCreate('outfit').returns({ person });
// Returning hasMany relationship. Assume super-hero hasMany 'outfits'
let outfits = buildList('outfit', 2);
mockCreate('super-hero').returns({ outfits });
- mocking a failed create
// Mocking failure case is easy with chainable methods, just use #fails
mockCreate('project').match({name: "Moo"}).fails();
// Can optionally add a status code and/or errors to the response
mockCreate('project').fails({status: 422, response: {errors: {name: ['Moo bad, Bahh better']}}});
store.createRecord('project', {name: "Moo"}).save(); //=> fails
- mockUpdate(model)
- Single argument ( the model instance that will be updated )
- mockUpdate(modelType, id)
- Two arguments: modelType ( like 'profile' ) , and the profile id that will updated
- Use chainable methods to help build response:
- match
- Attributes with values that must be present on the model you are updating
- returns
- Attributes ( including relationships ) to include in response json
- match
- Need to wrap tests using mockUpdate with: Ember.run(function() { 'your test' })
Usage:
import { make, mockUpdate } from 'ember-data-factory-guy';
let profile = make('profile');
// Pass in the model that will be updated ( if you have it available )
mockUpdate(profile);
// If the model is not available, pass in the modelType and the id of
// the model that will be updated
mockUpdate('profile', 1);
profile.set('description', 'good value');
profile.save() //=> will succeed
// Returning belongsTo relationship. Assume outfit belongsTo 'person'
let outfit = make('outfit');
let person = build('super-hero'); // it's polymorphic
outfit.set('name','outrageous');
mockUpdate(outfit).returns({ person });
outfit.save(); //=> saves and returns superhero
// Returning hasMany relationship. Assume super-hero hasMany 'outfits'
let superHero = make('super-hero');
let outfits = buildList('outfit', 2, {name:'bell bottoms'});
superHero.set('style','laid back');
mockUpdate(superHero).returns({ outfits });
superHero.save(); // => saves and returns outfits
// using match() method to specify attribute values
let profile = make('profile');
profile.set('name', "woo");
let mock = mockUpdate(profile).match({name: "moo"});
profile.save(); // will not be mocked since the mock you set says the name must be "woo"
// either set the name to "moo" which will now be mocked correctly
profile.set('name', "moo");
profile.save(); // succeeds
// or
// keep the profile name as "woo"
// but change the mock to match the name "woo"
mock.match({name: "woo"});
profile.save(); // succeeds
- mocking a failed update
let profile = make('profile');
// set the succeed flag to 'false'
mockUpdate('profile', profile.id).fails({status: 422, response: 'Invalid data'});
// or
mockUpdate(profile).fails({status: 422, response: 'Invalid data'});
profile.set('description', 'bad value');
profile.save() //=> will fail
mocking a failed update and retry with succees
let profile = make('profile');
let mockUpdate = mockUpdate(profile);
mockUpdate.fails({status: 422, response: 'Invalid data'});
profile.set('description', 'bad value');
profile.save() //=> will fail
// After setting valid value
profile.set('description', 'good value');
// Now expecting success
mockUpdate.succeeds();
// Try that update again
profile.save() //=> will succeed!
- Need to wrap tests using mockDelete with: Ember.run(function() { 'your test' })
- To handle deleteing a model
- Pass in a record ( or a typeName and id )
Usage:
- Passing in a record / model instance
import { make, mockDelete } from 'ember-data-factory-guy';
let profile = make('profile');
mockDelete(profile);
profile.destroyRecord() // => will succeed
- Passing in a model typeName and id
import { make, mockDelete } from 'ember-data-factory-guy';
let profile = make('profile');
mockDelete('profile', profile.id);
profile.destroyRecord() // => will succeed
- Passing in a model typeName
import { make, mockDelete } from 'ember-data-factory-guy';
let profile1 = make('profile');
let profile2 = make('profile');
mockDelete('profile');
profile1.destroyRecord() // => will succeed
profile2.destroyRecord() // => will succeed
- Mocking a failed delete
mockDelete(profile).fails();
- This is probably the funnest thing in FactoryGuy, if your not using this syntax yet, your truely missing out.
let json = buildList('widget', 'square', 'round', ['round','broken']);
let widgets = makeList('widget', 'square', 'round', ['round','broken']);
- you just built/made 3 different widgets from traits ('square','round','broken')
- first will have square trait
- second will have round trait
- third will have round and broken trait
- The fact that you can match on attributes in mockUpdate and mockCreate means
that you can test a custom
serialize
method in a model serializer
// app/serializers/person.js
export default DS.RESTSerializer.extend({
// let's say your modifying all names to be Japanese honorific style
serialize: function(snapshot, options) {
var json = this._super(snapshot, options);
let honorificName = [snapshot.record.get('name'), 'san'].join('-');
json.name = honorificName;
return json;
}
});
// somewhere in your tests
let person = make('person', {name: "Daniel"});
mockUpdate(person).match({name: "Daniel-san"});
person.save(); // will succeed
// and voila, you have just tested the serializer is converting the name properly
- Can pass in random attributes to help build the fixture
- sorta like transient attributes but these don't get passed to afterMake
- hence can be used in build/buildList as well
Let's say you have a model and a factory like this:
// app/models/dog.js
import Model from 'ember-data/model';
import attr from 'ember-data/attr';
export default Model.extend({
dogNumber: attr('string'),
sound: attr('string')
});
// tests/factories/dog.js
import FactoryGuy from 'ember-data-factory-guy';
const defaultVolume = "Normal";
FactoryGuy.define('dog', {
default: {
dogNumber: (f)=> `Dog${f.id}`,
sound: (f) => `${f.volume||defaultVolume} Woof`
}
});
Then to build the fixture:
let volume = 'Soft';
let dog2 = build('dog', { volume });
dog2.get('sound'); //=> `Soft Woof`