Swift-Kuery-ORM is an ORM (Object Relational Mapping) library built for Swift. Using it allows you to simplify persistence of model objects with your server.
Swift-Kuery-ORM is built on top of Swift-Kuery, which means that its possible to use Swift-Kuery to customize SQL queries made to the database, if the functionality of the ORM is insufficient.
The key component of Swift-Kuery-ORM is the protocol Model
.
Let's propose a struct to use as an example. We can declare an object that looks like so:
struct Grade: Codable {
var course: String
var grade: Int
}
Thanks to Codable Routing in Kitura 2.0, we declare our struct to be Codable
to simplify our RESTful routes for these objects on our server. The Model
protocol extends what Codable
does to work with the ORM. In your server application, you would extend your object like so:
extension Grade: Model { }
Now that your Grade
struct conforms to Model
, after you have set up your database connection pool and created a database table, you automatically have access to a slew of convenience functions for your object.
Need to retrieve all instances of Grade
? You can implement:
Grade.findAll()
Need to add a new instance of Grade
? Here's how:
grade.save()
The Model
protocol is the key to using the ORM. Let's walk through how to fully set up an application to make use of the ORM.
Follow Getting Started to create a Kitura server. In this example you'll be using the Swift Kuery PostgreSQL plugin, so you will need PostgreSQL running on your local machine, which you can install with brew install postgresql
. The default port for PostgreSQL is 5432.
Add Swift-Kuery-ORM and Swift-Kuery-PostgreSQL to your application's Package.swift
. Substitute "x.x.x"
with the latest Swift-Kuery-ORM
release and the latest Swift-Kuery-PostgreSQL
release.
dependencies: [
...
// Add these two lines
.package(url: "https://github.com/IBM-Swift/Swift-Kuery-ORM.git", from: "x.x.x"),
.package(url: "https://github.com/IBM-Swift/Swift-Kuery-PostgreSQL.git", from: "x.x.x"),
],
targets: [
.target(
name: ...
// Add these two modules to your target(s)
dependencies: [..., "SwiftKueryORM", "SwiftKueryPostgreSQL"]),
]
Let's assume you want to add ORM functionality to a file called Application.swift
. You'll need to add the following import statements at the top of the file:
import SwiftKueryORM
import SwiftKueryPostgreSQL
As mentioned before, we recommend you use Homebrew to set up PostgreSQL on your machine. You can install PostgreSQL and set up your table like so:
brew install postgresql
brew services start postgresql
createdb school
Initialize your database in your Application.swift
file:
let pool = PostgreSQLConnection.createPool(host: "localhost", port: 5432, options: [.databaseName("school")], poolOptions: ConnectionPoolOptions(initialCapacity: 10, maxCapacity: 50, timeout: 10000))
Database.default = Database(pool)
Like before, assume you will work with a struct that looks like so:
struct Grade : Codable {
var course: String
var grade: Int
}
In your Application.swift
file, extend Grade
to conform to Model
extension Grade : Model {
// here, you can add any server-side specific logic to your object
}
Now, you need to create your table. If you are configuring your database while you start up your server, you can use createTableSync()
, which runs synchronously. If you want to use an asynchronous function, you can use createTable()
elsewhere. You can implement either of these functions like so:
do {
try Grade.createTableSync()
} catch let error {
// Error
}
It's important to point out that if you've already created your table, this will throw an error here.
Your application is now ready to make use of all the functions available in the Model
protocol. If you'd like to see a fully working example of the ORM using Codable Routing, visit our FoodTracker example.
Let's cover all the functionality you have available to you now.
If you'd like to save a new object to your database, you have to create the object and use the save()
function:
let grade = Grade(course: "physics", grade: 80)
grade.save { grade, error in
...
}
You also optionally have the ability to pass the ID of the newly saved object into your closure. Add it to the collection of parameters like so:
grade.save { (id: Int?, grade: Grade?, error: RequestError?) in
...
}
If you have the id for an existing record of your object, and you'd like to update the record with an object, you can use the update()
function to do so:
let grade = Grade(course: "physics", grade: 80)
grade.course = "maths"
grade.update(id: 1) { grade, error in
...
}
If you'd like to find a specific object, and you have its id, you can use the find()
function to retrieve it:
Grade.find(id: 1) { result, error in
...
}
If you'd like to retrieve all instances of a particular object, you can make use of findAll()
as a static function on the type you are trying to retrieve:
Grade.findAll { (result: [Grade]?, error: RequestError?) in
...
}
You also have the ability to form your results in different ways and formats, like so:
Grade.findAll { (result: [(Int, Grade)]?, error: RequestError?) in
...
}
Grade.findAll { (result: [Int: Grade]?, error: RequestError?) in
...
}
If you'd like to delete an object, and you have its id, you can use the delete()
function like so:
Grade.delete(id: 1) { error in
...
}
If you're feeling bold, and you'd like to remove all instances of an object from your database, you can use the static function deleteAll()
with your type:
Grade.deleteAll { error in
...
}
The ORM defines an extension to Model
which provides a number of public static executeQuery(…)
functions. These can be used to create custom functions within your model that perform more complex database operations. The example below defines a Person model and with a custom function that will retrieve all records which have age > 20:
// define the Person struct
struct Person: Codable {
var firstname: String
var surname: String
var age: Int
}
// extend Person to conform to model and add overTwenties function
extension Person: Model {
// Define a synchronous function to retrieve all records of Person with age > 20
public static func getOverTwenties() -> [Person]? {
let wait = DispatchSemaphore(value: 0)
// First get the table
var table: Table
do {
table = try Person.getTable()
} catch {
// Handle error
}
// Define result, query and execute
var overTwenties: [Person]? = nil
let query = Select(from: table).where("age > 20")
Person.executeQuery(query: query, parameters: nil) { results, error in
guard let results = results else {
// Handle error
}
overTwenties = results
wait.signal()
return
}
wait.wait()
return overTwenties
}
}
Alternatively you can define and asynchronous getOverTwenties function:
public static func getOverTwenties(oncompletion: @escaping ([Person]?, RequestError?)-> Void) {
var table: Table
do {
table = try Person.getTable()
} catch {
// Handle error
}
let query = Select(from: table).where("age > 20")
Person.executeQuery(query: query, parameters: nil, oncompletion)
}
which can be called in a fashion similar to the following:
Person.getOverTwenties() { result, error in
guard let result = result else {
// Handle error
}
// Use result
}
If you'd like to learn more about how you can customize queries, check out the Swift-Kuery repository for more information.
The ORM has several options available for identifying an instance of a model.
If you define your Model
without specifying an ID property, either by using the idColumnName
property or the default name of id
, then the ORM will create an auto-incrementing column named id
in the database table for the model, eg.
struct Person: Model {
var firstname: String
var surname: String
var age: Int
}
The model does not contain a property for the ID. The ORM provides a specific save
API that will return the ID that was assigned. It is important to note the ORM will not link the returned ID to the instance of the Model in any way; you are responsible for maintaining this relationship if necessary. Below is an example of retrieving an ID for an instance of the Person
model defined above:
let person = Person(firstname: "example", surname: "person", age: 21)
person.save() { (id: Int?, person, error) in
guard let id = id, let person = person else{
// Handle error
return
}
// Use person and id
}
The compiler requires you to declare the type of the ID received by your completion handler; the type should be Int?
for an ID that has been automatically assigned.
You can manage the assignment of IDs yourself by adding an id
property to your model. You may customise the name of this property by defining idColumnName
. For example:
struct Person: Model {
var myIDField: Int
var firstname: String
var surname: String
var age: Int
static var idColumnName = "myIDField"
static var idColumnType = Int.self
}
When using a Model
defined in this way, you are responsible for the assignment and management of IDs. Below is an example of saving an instance of the Person
model defined above:
let person = Person(myIDField: 1, firstname: "example", surname: "person", age: 21)
person.save() { (person, error) in
guard let person = person else {
// Handle error
return
}
// Use newly saved person
}
Declaring your ID property as optional allows the ORM to assign the ID automatically when the model is saved. If the value of ID is nil
, the database will assign an auto-incremented value. At present this is only support for an Int?
type.
You may instead provide an explicit value, which will be used instead of automatic assignment.
Optional IDs must be identified by defining the idKeypath: IDKeyPath
property, as in the example below:
struct Person: Model {
var id: Int?
var firstname: String
var surname: String
var age: Int
static var idKeypath: IDKeyPath = \Person.id
}
In the example above, the Model
is defined with an ID property matching the default idColumnName
value, but should you wish to use an alternative name you must define idColumnName
accordingly.
Below is an example of saving an instance of the Person
defined above, both with an explicitly defined ID and without:
let person = Person(id: nil, firstname: “Banana”, surname: “Man”, age: 21)
let specificPerson = Person(id: 5, firstname: “Super”, surname: “Ted”, age: 26)
person.save() { (savedPerson, error) in
guard let newPerson = savedPerson else {
// Handle error
}
print(newPerson.id) // Prints the next value in the databases identifier sequence, eg. 1
}
specificPerson.save() { (savedPerson, error) in
guard let newPerson = savedPerson else {
// Handle error
}
print(newPerson.id) // Prints 5
}
NOTE - When using manual or optional ID properties, you should be prepared to handle violation of unique identifier constraints. These can occur if you attempt to save a model with an ID that already exists, or in the case of Postgres, if the auto-incremented value collides with an ID that was previously inserted explicitly.
By default any property on your Model that is declared as a Date
will be encoded and decoded as a Double
.
You can change this behaviour by overriding the default value of the property dateEncodingStrategy
. The dateEncodingStrategy will apply to all Date properties on your Model.
The example below defines a model which will have its Date properties encoded and decoded as a timestamp:
struct Person: Model {
static var dateEncodingStrategy: DateEncodingFormat = .timestamp
var firstname: String
var surname: String
var age: Int
var dob: Date
}
For more information visit our API reference.
We love to talk server-side Swift, and Kitura. Join our Slack to meet the team!
This library is licensed under Apache 2.0. Full license text is available in LICENSE.