Awesome

Building a FoodTracker Backend with Kitura

This tutorial teaches how to create a Kitura Swift backend for the FoodTracker iOS app tutorial from Apple. This project contains a version of the tutorial code that has been updated to use Codable rather than NSCoder.

Upon completion of this tutorial there are several next steps you can take to add further functionality to the application:

If you would like to jump straight to one of these next steps a completed version of this tutorial with instruction on starting the server and application is available on the CompletedFoodTracker branch.

Pre-Requisites:

Note: This workshop has been developed for Swift 4, Xcode 9.x and Kitura 2.x.

Install the Kitura CLI:
1. Configure the Kitura homebrew tap
  brew tap ibm-swift/kitura
2. Install the Kitura CLI from homebrew
  brew install kitura
Ensure you have CocoaPods installed
1. Install CocoaPods: sudo gem install cocoapods
Clone this project from GitHub to your machine (don't use the Download ZIP option):
```
cd ~
git clone http://github.com/IBM/FoodTrackerBackend
cd ~/FoodTrackerBackend
```

Getting Started

Run the Food Tracker App:

The Food Tracker application is taken from the Apple tutorial for building your first iOS application. It allows you to store names, photos and ratings for "meals". The meals are then stored onto the device using NSKeyedArchiver. The following shows you how to see the application running.

Change into the iOS app directory:

cd ~/FoodTrackerBackend/iOS/FoodTracker

Open the Xcode Project

open FoodTracker.xcodeproj

Run the project to ensure that it's working
1. Press the Run button or use the ⌘+R key shortcut.
2. Add a meal in the Simulator by clicking the '+' button, providing a name, selecting a photo and a rating and clicking "Save".
3. Check that you receive a “Meals successfully saved.” message in the console

Building a Kitura Backend

The Food Tracker application stores the meal data to the local device, which means it's not possible to share the data with other users, or to build an additional web interface for the application. The following steps show you how to create a Kitura Backend to allow you to store and share the data.

1. Initialize a Kitura Server Project

Create a directory for the server project

mkdir ~/FoodTrackerBackend/FoodServer
cd ~/FoodTrackerBackend/FoodServer

Create a Kitura starter project

kitura init

The Kitura CLI will now create and build an starter Kitura application for you. This includes adding best-practice implementations of capabilities such as configuration, health checking and monitoring to the application for you.

More information about the project structure is available on kitura.io.

2. Create an in-memory data store for Meals

The init command has created a fully running Kitura application, but one which has no application logic. In order to use it to store the data from the FoodTracker application, you need to create a datastore in the Kitura application for the Meal data from the FoodTracker iOS application. This tutorial uses a simple in-memory dictionary to store exactly the same Meal types that are used in the FoodTracker application.

Copy the Meal.swift file from the FoodTracker app to the Server

cd ~/FoodTrackerBackend
cp ./iOS/FoodTracker/FoodTracker/Meal.swift ./FoodServer/Sources/Application

Regenerate the FoodServer Xcode project to link "Meal.swift".

cd ~/FoodTrackerBackend/FoodServer
swift package generate-xcodeproj

Open the FoodServer project in Xcode
```
 open FoodServer.xcodeproj
```

Note: When you open xcode you may see the error "No such module 'Kitura'". This is an Xcode bug, it has found and linked Kitura so disregard the message.

Add a dictionary to the Application.swift file to store the Meal types
1. Open the Sources > Application > Application.swift file
2. Add a "mealstore" inside the app class, by inserting the following code:
```
private var mealStore: [String: Meal] = [:]
```
  On the line below let cloudEnv = CloudEnv()

This now provides a simple dictionary to store Meal data passed to the FoodServer from the FoodTracker app.

3. Create a REST API to allow FoodTracker to store Meals

REST APIs typically consist of a HTTP request using a verb such as POST, PUT, GET or DELETE along with a URL and an optional data payload. The server then handles the request and responds with an optional data payload.

A request to store data typically consists of a POST request with the data to be stored, which the server then handles and responds with a copy of the data that has just been stored.

Register a handler for a POST request on /meals that stores the data
Add the following into the postInit() function in the App class:

router.post("/meals", handler: storeHandler)

Implement the storeHandler that receives a Meal, and returns the stored Meal
Add the following code as a function in the App class, beneath the postInit() function:

    func storeHandler(meal: Meal, completion: (Meal?, RequestError?) -> Void ) {
        mealStore[meal.name] = meal
        completion(mealStore[meal.name], nil)
    }

As well as being able to store Meals on the FoodServer, the FoodTracker app will also need to be able to access the stored meals. A request to load all of the stored data typically consists of a GET request with no data, which the server then handles and responds with an array of the data that has just been stored.

	router.get("/meals", handler: loadHandler)

Implement the loadHandler that returns the stored Meals as an array.
Add the following as another function in the App class:

    func loadHandler(completion: ([Meal]?, RequestError?) -> Void ) {
	    let meals: [Meal] = self.mealStore.map({ $0.value })
        completion(meals, nil)
    }

A GET request to this route will display pages of data representing the photo, which is not very human readable. To solve this, we will make a new route which only returns the meal name and rating.

Register a handler for a GET request on /summary that loads the data.
Add the following into the postInit() function:

	router.get("/summary", handler: summaryHandler)

Implement the summaryHandler that returns the names and rating of the stored Meals as an array.
Add the following as another function in the App class:

    func summaryHandler(completion: (Summary?, RequestError?) -> Void ) {
       let summary: Summary = Summary(self.mealStore)
       completion(summary, nil)
    }

4. Test the newly created REST API

Run the server project in Xcode
1. In the top left corner of Xcode you should see a small toolbox icon with the text "FoodServer-Package" next to it. Click this icon and then click "FoodServer" from the dropdown menu.
2. Press the Run button or use the ⌘+R key shortcut.
3. Select "Allow incoming network connections" if you are prompted.
Check that some of the standard Kitura URLs are running:
- Kitura Monitoring: http://localhost:8080/swiftmetrics-dash/
- Kitura Health check: http://localhost:8080/health
Test the GET REST API is running correctly
There are many utilities for testing REST APIs, such as Postman. Here we'll use "curl", which is a simple command line utility:

curl -X GET \
  http://localhost:8080/meals \
  -H 'content-type: application/json'

If the GET endpoint is working correctly, this should return an array of JSON data representing the stored Meals. As no data is yet stored, this should return an empty array, ie:

[]

Test the POST REST API is running correctly
In order to test the POST API, we make a similar call, but also sending in a JSON object that matches the Meal data:

curl -X POST \
  http://localhost:8080/meals \
  -H 'content-type: application/json' \
  -d '{
    "name": "test",
    "photo": "0e430e3a",
    "rating": 1
}'

If the POST endpoint is working correctly, this should return the same JSON that was passed in, eg:

{"name":"test","photo":"0e430e3a","rating":1}

Test the GET REST API is returning the stored data correctly
In order to check that the data is being stored correctly, re-run the GET check:

curl -X GET \
  http://localhost:8080/meals \
  -H 'content-type: application/json'

This should now return a single entry array containing the Meal that was stored by the POST request, eg:

[{"name":"test","photo":"0e430e3a","rating":1}]

Test the GET REST API for the summary route is returning just the name and rating. View the summary route by going to http://localhost:8080/summary. This will perform a GET request to your server and should display:

{"summary":[{"name":"test","rating":1}]}

Connect FoodTracker to the Kitura FoodServer

Any package that can make REST calls from an iOS app is sufficient to make the connection to the Kitura FoodServer to store and retrieve the Meals. Kitura itself provides a client connector called KituraKit which makes it easy to connect to Kitura using shared data types, in our case Meals, using an API that is almost identical on the client and the server. In this example we'll use KituraKit to make the connection.

Install KituraKit into the FoodTracker app

KituraKit is designed to be used both in iOS apps and in server projects. Currently the easiest way to install KituraKit into an iOS app it to download a bundling containing KituraKit and its dependencies, and to install it into the app as a CocoaPod.

If the "FoodTracker" Xcode project it is open, close it. Installing the KituraKit bundle as a CocoaPod will edit the project and create a workspace, so it is best if the project is closed.
Create a Podfile in the FoodTracker iOS application directory:

cd ~/FoodTrackerBackend/iOS/FoodTracker/
pod init

Edit the Podfile to use KituraKit:
1. Open the Podfile for editing
```
open Podfile
```
1. Set a global platform of iOS 11 for your project
  Uncomment # platform :ios, '9.0' and set the value to 11.0
```
platform :ios, '11.0'
```
1. Under the "# Pods for FoodTracker" line add:
```
# Pods for FoodTracker
pod 'KituraKit'
```
1. Save and close the file
Install KituraKit:

pod install

Open the Xcode workspace (not project!)

cd ~/FoodTrackerBackend/iOS/FoodTracker/
open FoodTracker.xcworkspace

KituraKit should now be installed, and you should be able to build and run the FoodTracker project as before. Note that from now on you should open the Xcode workspace ('FoodTracker.xcworkspace') not project.

Update FoodTracker to call the Kitura FoodServer

Now that KituraKit is installed into the FoodTracker application, it needs to be updated to use it to call the Kitura FoodServer. The code to do that is already provided. As a result, you only need to uncomment the code that invokes those APIs. The code to uncomment is marked with UNCOMMENT.

Edit the FoodTracker > MealTableViewController.swift file:

Uncomment the import of KituraKit at the top of the file.

import KituraKit

Uncomment the following at the start of the saveMeals() function:

        for meal in meals {
              saveToServer(meal: meal)
        }

Uncomment the following saveToServer(meal:) function towards the end of the file:

private func saveToServer(meal: Meal) {
    guard let client = KituraKit(baseURL: "http://localhost:8080") else {
        print("Error creating KituraKit client")
        return
    }
    client.post("/meals", data: meal) { (meal: Meal?, error: Error?) in
        guard error == nil else {
            print("Error saving meal to Kitura: \(error!)")
            return
        }
        print("Saving meal to Kitura succeeded")
    }
}

Update the FoodTracker app to allow interaction with a Server

The final step is to update the FoodTracker application to allow loads from a server.

Update the FoodTracker > Info.plist file to allow loads from a server:
Note: This step has been completed for you already:

    <key>NSAppTransportSecurity</key>
	<dict>
	    <key>NSAllowsArbitraryLoads</key>
	<true/>
	</dict>

Run the FoodTracker app, storing data to the Kitura server

Make sure the Kitura server is still running and you have the Kitura monitoring dashboard open in your browser (http://localhost:8080/swiftmetrics-dash)
Build and run the FoodTracker app in the iOS simulator and add or remove a Meal entry You should see the following messages in the Xcode console:
```
Saving meal to Kitura succeeded
Saving meal to Kitura succeeded
```
View the monitoring panel to see the responsiveness of the API call
Check the data has been persisted by the Kitura server

curl -X GET \
  http://localhost:8080/summary \
  -H 'content-type: application/json'

This should now return an array containing the meals' names and ratings stored by the POST request.

Congratulations, you have successfully build a Kitura Backend for an iOS app!

Connecting A PostgreSQL Database

Creating a PostgreSQL Database

We created a server and connected it to the iOS application. This means created meals are posted to the server and a user can then view these meals on localhost:8080/summary. Since the meals are stored on the server, if the server is restarted the meal data is lost. To solve this problem, we will start by creating a PostgreSQL database where the meals will be stored.

Install PostgreSQL:

brew install postgresql
brew services start postgresql

You should receive a message that either PostgreSQL has been started or the service is already running. This installation should have installed two applications we need, namely createdb and psql, which will be used as clients to your locally running PostgreSQL.

Create a database called FoodDatabase to store the data:

createdb FoodDatabase

Adding Swift-Kuery-ORM dependencies to your server

Swift-Kuery-ORM is an ORM that works alongside a specific database library, such as Swift-Kuery-PostgreSQL, to allow a user to easily interact with database in Swift. These two libraries are added to our Package.swift file, so the server can access them.

If the "FoodServer" Xcode project is open, close it. Installing Swift-Kuery-ORM and Swift-Kuery-PostgreSQL will modify the Xcode project.
Open a new terminal window and go to your Package.swift file.

cd ~/FoodTrackerBackend/FoodServer
open Package.swift

Add the Swift-Kuery-ORM and Swift-Kuery-PostgreSQL packages.

.package(url: "https://github.com/IBM-Swift/Swift-Kuery-ORM.git", .upToNextMinor(from: "0.3.0")),
.package(url: "https://github.com/IBM-Swift/Swift-Kuery-PostgreSQL.git", from: "1.1.0"),

below the line .package(url: "https://github.com/IBM-Swift/Health.git", from: "1.0.0"),

Change the target for Application to include SwiftKueryORM and SwiftKueryPostgreSQL after Health

.target(name: "Application", dependencies: [ "Kitura","CloudEnvironment","SwiftMetrics","Health", "SwiftKueryORM", "SwiftKueryPostgreSQL"]),

Regenerate the server Xcode project: Now we have added the dependencies to our Package.swift file we need to regenerate our FoodServer Xcode project to link the Swift package changes in Xcode.

swift package generate-xcodeproj
open FoodServer.xcodeproj/

Making Meal a Model

To work with the ORM, the struct Meal needs to implement the Model.

Open your Sources > Application > Application.swift file
Add two libraries to the import statements:

import SwiftKueryORM
import SwiftKueryPostgreSQL

Below the line that reads public let health = Health(), extend Meal to conform to Model like so:

extension Meal: Model {
    static var idColumnName = "name"
}

Deleting the server mealStore

Since we will be storing the meal data in a database, we no longer need a local meal store on the server.

Open your Sources > Application > Application.swift file
Delete the mealStore initialiser:

private var mealStore: [String: Meal] = [:]

Delete the mealStore references in storeHandler:

mealStore[meal.name] = meal
completion(mealStore[meal.name], nil)

Delete the mealStore references in loadHandler:

let meals: [Meal] = self.mealStore.map({ $0.value })
completion(meals, nil)

Delete the mealStore references in summaryHandler:

let summary: [Summary] = Summary(self.mealStore)
completion(summary, nil)

Connecting to the PostgreSQL database

We will now connect to our server to the PostgreSQL database. This will allow us to send and receive information from the database.

In the same Application.swift file, go underneath your extension of Meal and add a new class:

class Persistence {

}

Inside this class, create a static function that will set up a connection pool and assign it to a default database:

static func setUp() {
    let pool = PostgreSQLConnection.createPool(host: "localhost", port: 5432, options: [.databaseName("FoodDatabase")], poolOptions: ConnectionPoolOptions(initialCapacity: 10, maxCapacity: 50, timeout: 10000))
    Database.default = Database(pool)
}

Note We use a connection pool since we have concurrent requests.

Go to the postInit function below the line router.get("/summary", handler: summaryHandler) and call your setup function, and create a table sync for your Meal object:

Persistence.setUp()
do {
    try Meal.createTableSync()
} catch let error {
    print(error)
}

Before we start using the PostgreSQL Database, we need to create a table in the database. Add the following:

Add the @escaping keyword to the completion closure in the storeHandler signatures.

func storeHandler(meal: Meal, completion: @escaping (Meal?, RequestError?) -> Void ) {

Add the @escaping keyword to the completion closure in the loadHandler signatures.

func loadHandler(completion: @escaping ([Meal]?, RequestError?) -> Void ) {

Add the @escaping keyword to the completion closure in the summaryHandler signatures.

func summaryHandler(completion: @escaping (Summary?, RequestError?) -> Void ) {

Allowing the completion closure to be escaping means the database queries can be asynchronous.

Using the PostgreSQL Database

Handling an HTTP POST request

We are now going to save a meal in our storeHandler. This will mean that when our server receives an HTTP POST request, it will take the Meal instance received and save it to the database.

Inside the storeHandler function add the following line:

meal.save(completion)

Your completed storeHandler function should now look as follows:

func storeHandler(meal: Meal, completion: @escaping (Meal?, RequestError?) -> Void ) {
      meal.save(completion)
}

You can verify this by:

Starting the FoodTracker application in Xcode.
Creating a meal in the application.
Accessing your database: psql FoodDatabase
Viewing your meals table: SELECT name, rating FROM "Meals"; This should produce a table with the name and the rating of your newly added meal.

NOTE We do not print out the photo because it is too large 5. Close psql by entering \q Now when you create a meal in the application, the server will save it to the PostgreSQL database.

Handling an HTTP GET request

We are going to get our meals in our loadHandler function. This will mean that when the server receives an HTTP GET request, it will get the meals from the database. This means the data the server returns to the client is taken from the database and will persist, even if the server is restarted.

Inside the loadHander function add the following line:

Meal.findAll(completion)

Your completed loadHandler function should now look as follows:

func loadHandler(completion: @escaping ([Meal]?, RequestError?) -> Void ) {
      Meal.findAll(completion)
}

Now when you perform a GET call to your server it will retrieve the meals from your database.
Update the summaryHandler function to get the meals from the database:

Inside the summaryHandler function add the following line:

Meal.findAll { meals, error in
    guard let meals = meals else {
        completion(nil, .internalServerError)
        return
    }
    completion(Summary(meals), nil)
}

Your completed summaryHandler function should now look as follows:

func summaryHandler(completion: @escaping ([Summary]?, RequestError?) -> Void ) {
    Meal.findAll { meals, error in
        guard let meals = meals else {
            completion(nil, .internalServerError)
            return
        }
        completion(Summary(meals), nil)
    }
}

You can verify this by going to http://localhost:8080/summary, where you should see your meals. You can now restart your server and this data will persist, since it is stored within the database!

Congratulations, you have successfully built a Kitura backend and stored the data in a PostgreSQL database!

Next Steps

If you have sufficient time, the following tasks can be completed to update your application.

Add a Website Frontend using the Stencil Template Engine

The current implementation of the Kitura FoodServer returns a JSON array of the meals. To create a website, you would want to use html to structure the page. The following contains steps to embed the meal data into html and add a Website Frontend using the Stencil Template Engine

Add Support for Retrieving and Deleting Meals from the FoodServer

The current implementation of the Kitura FoodServer has support for retrieving all of the stored Meals using a GET request on /meals, but the FoodTracker app is currently only saving the Meals to the FoodServer. The following contains the steps to add Retrieving and Deleting Meals from the FoodServer.

Add a Web Application to the Kitura server

Now that the Meals from the FoodTracker app are being stored on a server, it becomes possible to start building a web application that also provides users with access to the stored Meal data. The following steps describe how to start to Build a FoodTracker Web Application.

Add HTTP Basic authentication to the Kitura server

The current implementation of the Kitura FoodServer will allow anyone to request all the meals. We can add HTTP Basic authentication to make the routes require a username and password. This allows you to authenticate a user and have the server respond accordingly.

Deploy and host the Kitura FoodServer in the Cloud

In order for a real iOS app to connect to a Kitura Server, it needs to be hosted at a public URL that the iOS app can reach.

Kitura is deployable to any cloud, but the project created with kitura init provides additonal files so that it is pre-configured for clouds that support any of Cloud Foundry, Docker or Kubernetes. The follow contains the steps to take the Kitura FoodServer and Deploy to the IBM Cloud using Cloud Foundry.

View a sample Todo list application using Kitura

This tutorial takes you through the basics of creating a Kitura server. To see a completed Todo list application with demonstrations of all HTTP requests go to iOSSampleKituraKit