Contexts

Requirement: This guide expects that you have gone through the introductory guides and got a Phoenix application up and running.

Requirement: This guide expects that you have gone through the Request life-cycle guide.

Requirement: This guide expects that you have gone through the Ecto guide.

So far, we've built pages, wired up controller actions through our routers, and learned how Ecto allows data to be validated and persisted. Now it's time to tie it all together by writing web-facing features that interact with our greater Elixir application.

When building a Phoenix project, we are first and foremost building an Elixir application. Phoenix's job is to provide a web interface into our Elixir application. Naturally, we compose our applications with modules and functions, but simply defining a module with a few functions isn't enough when designing an application. We need to consider the boundaries between modules and how to group functionality. In other words, it's vital to think about application design when writing code.

Thinking about design

Contexts are dedicated modules that expose and group related functionality. For example, anytime you call Elixir's standard library, be it Logger.info/1 or Stream.map/2, you are accessing different contexts. Internally, Elixir's logger is made of multiple modules, but we never interact with those modules directly. We call the Logger module the context, exactly because it exposes and groups all of the logging functionality.

By giving modules that expose and group related functionality the name contexts, we help developers identify these patterns and talk about them. At the end of the day, contexts are just modules, as are your controllers, views, etc.

In Phoenix, contexts often encapsulate data access and data validation. They often talk to a database or APIs. Overall, think of them as boundaries to decouple and isolate our systems into manageable, independent parts. Let's use these ideas to build out our web application. Our goal is to build an ecommerce system where we can showcase products, allow users to add products to their cart, and complete their orders.

How to read this guide: Using the context generators is a great way for beginners and intermediate Elixir programmers alike to get up and running quickly while thoughtfully designing their applications. This guide focuses on those readers. On the other hand, experienced developers may get more mileage from nuanced discussions around application design. For those readers, we include a frequently asked questions (FAQ) section at the end of the guide which brings different perspectives to some design decisions made throughout the guide. Beginners can safely skip the FAQ sections and return later when they're ready to dig deeper.

Adding a Catalog Context

An ecommerce platform has wide-reaching coupling across a codebase so it's important to think upfront about writing well-defined interfaces. With that in mind, our goal is to build a product catalog API that handles creating, updating, and deleting the products available in our system. We'll start off with the basic features of showcasing our products, but as we add cart features later. We'll see how starting with a solid foundation with isolated boundaries allows us to grow our application naturally as we add functionality.

Phoenix includes the mix phx.gen.html, mix phx.gen.json, mix phx.gen.live, and mix phx.gen.context generators that apply the ideas of isolating functionality in our applications into contexts. These generators are a great way to hit the ground running while Phoenix nudges you in the right direction to grow your application. Let's put these tools to use for our new product catalog context.

In order to run the context generators, we need to come up with a module name that groups the related functionality that we're building. In the Ecto guide, we saw how we can use Changesets and Repos to validate and persist user schemas, but we didn't integrate this with our application at large. In fact, we didn't think about where a "user" in our application should live at all. Let's take a step back and think about the different parts of our system. We know that we'll have products to showcase on pages for sale, along with descriptions, pricing, etc. Along with selling products, we know we'll need to support carting, order checkout, and so on. While the products being purchased are related to the cart and checkout processes, showcasing a product and managing the exhibition of our products is distinctly different than tracking what a user has placed in their cart or how an order is placed. A Catalog context is a natural place for the management of our product details and the showcasing of those products we have for sale.

Naming things is hard. If you're stuck when trying to come up with a context name when the grouped functionality in your system isn't yet clear, you can simply use the plural form of the resource you're creating. For example, a Products context for managing products. As you grow your application and the parts of your system become clear, you can simply rename the context to a more refined name at a later time.

To jump-start our catalog context, we'll use mix phx.gen.html which creates a context module that wraps up Ecto access for creating, updating, and deleting products, along with web files like controllers and templates for the web interface into our context. Run the following command at your project root:

$ mix phx.gen.html Catalog Product products title:string \
description:string price:decimal views:integer

* creating lib/hello_web/controllers/product_controller.ex
* creating lib/hello_web/templates/product/edit.html.heex
* creating lib/hello_web/templates/product/form.html.heex
* creating lib/hello_web/templates/product/index.html.heex
* creating lib/hello_web/templates/product/new.html.heex
* creating lib/hello_web/templates/product/show.html.heex
* creating lib/hello_web/views/product_view.ex
* creating test/hello_web/controllers/product_controller_test.exs
* creating lib/hello/catalog/product.ex
* creating priv/repo/migrations/20210201185747_create_products.exs
* creating lib/hello/catalog.ex
* injecting lib/hello/catalog.ex
* creating test/hello/catalog_test.exs
* injecting test/hello/catalog_test.exs

Add the resource to your browser scope in lib/hello_web/router.ex:

    resources "/products", ProductController


Remember to update your repository by running migrations:

    $ mix ecto.migrate

Note: we are starting with the basics for modeling an ecommerce system. In practice, modeling such systems yields more complex relationships such as product variants, optional pricing, multiple currencies, etc. We'll keep things simple in this guide, but the foundations will give you a solid starting point to building such a complete system.

Phoenix generated the web files as expected in lib/hello_web/. We can also see our context files were generated inside a lib/hello/catalog.ex file and our product schema in the directory of the same name. Note the difference between lib/hello and lib/hello_web. We have a Catalog module to serve as the public API for product catalog functionality, as well as an Catalog.Product struct, which is an Ecto schema for casting and validating product data. Phoenix also provided web and context tests for us, which we'll look at later. For now, let's follow the instructions and add the route according to the console instructions, in lib/hello_web/router.ex:

  scope "/", HelloWeb do
    pipe_through :browser

    get "/", PageController, :index
+   resources "/products", ProductController
  end

With the new route in place, Phoenix reminds us to update our repo by running mix ecto.migrate, but first we need to make a few tweaks to the generated migration in priv/repo/migrations/*_create_products.exs:

  def change do
    create table(:products) do
      add :title, :string
      add :description, :string
-     add :price, :decimal
+     add :price, :decimal, precision: 15, scale: 6, null: false
-     add :views, :integer
+     add :views, :integer, default: 0, null: false

      timestamps()
    end

We modified our price column to a specific precision of 15, scale of 6, along with a not-null constraint. This ensures we store currency with proper precision for any mathematical operations we may perform. Next, we added a default value and not-null constraint to our views count. With our changes in place, we're ready to migrate up our database. Let's do that now:

$ mix ecto.migrate
14:09:02.260 [info]  == Running 20210201185747 Hello.Repo.Migrations.CreateProducts.change/0 forward

14:09:02.262 [info]  create table products

14:09:02.273 [info]  == Migrated 20210201185747 in 0.0s

Before we jump into the generated code, let's start the server with mix phx.server and visit http://localhost:4000/products. Let's follow the "New Product" link and click the "Save" button without providing any input. We should be greeted with the following output:

Oops, something went wrong! Please check the errors below.

When we submit the form, we can see all the validation errors inline with the inputs. Nice! Out of the box, the context generator included the schema fields in our form template and we can see our default validations for required inputs are in effect. Let's enter some example product data and resubmit the form:

Product created successfully.

Title: Metaprogramming Elixir
Description: Write Less Code, Get More Done (and Have Fun!)
Price: 15.000000
Views: 0

If we follow the "Back" link, we get a list of all products, which should contain the one we just created. Likewise, we can update this record or delete it. Now that we've seen how it works in the browser, it's time to take a look at the generated code.

Starting With Generators

That little mix phx.gen.html command packed a surprising punch. We got a lot of functionality out-of-the-box for creating, updating, and deleting products in our catalog. This is far from a full-featured app, but remember, generators are first and foremost learning tools and a starting point for you to begin building real features. Code generation can't solve all your problems, but it will teach you the ins and outs of Phoenix and nudge you towards the proper mindset when designing your application.

Let's first check out the ProductController that was generated in lib/hello_web/controllers/product_controller.ex:

defmodule HelloWeb.ProductController do
  use HelloWeb, :controller

  alias Hello.Catalog
  alias Hello.Catalog.Product

  def index(conn, _params) do
    products = Catalog.list_products()
    render(conn, "index.html", products: products)
  end

  def new(conn, _params) do
    changeset = Catalog.change_product(%Product{})
    render(conn, "new.html", changeset: changeset)
  end

  def create(conn, %{"product" => product_params}) do
    case Catalog.create_product(product_params) do
      {:ok, product} ->
        conn
        |> put_flash(:info, "Product created successfully.")
        |> redirect(to: Routes.product_path(conn, :show, product))

      {:error, %Ecto.Changeset{} = changeset} ->
        render(conn, "new.html", changeset: changeset)
    end
  end

  def show(conn, %{"id" => id}) do
    product = Catalog.get_product!(id)
    render(conn, "show.html", product: product)
  end
  ...
end

We've seen how controllers work in our controller guide, so the code probably isn't too surprising. What is worth noticing is how our controller calls into the Catalog context. We can see that the index action fetches a list of products with Catalog.list_products/0, and how products are persisted in the create action with Catalog.create_product/1. We haven't yet looked at the catalog context, so we don't yet know how product fetching and creation is happening under the hood – but that's the point. Our Phoenix controller is the web interface into our greater application. It shouldn't be concerned with the details of how products are fetched from the database or persisted into storage. We only care about telling our application to perform some work for us. This is great because our business logic and storage details are decoupled from the web layer of our application. If we move to a full-text storage engine later for fetching products instead of a SQL query, our controller doesn't need to be changed. Likewise, we can reuse our context code from any other interface in our application, be it a channel, mix task, or long-running process importing CSV data.

In the case of our create action, when we successfully create a product, we use Phoenix.Controller.put_flash/3 to show a success message, and then we redirect to the router's product_path show page. Conversely, if Catalog.create_product/1 fails, we render our "new.html" template and pass along the Ecto changeset for the template to lift error messages from.

Next, let's dig deeper and check out our Catalog context in lib/hello/catalog.ex:

defmodule Hello.Catalog do
  @moduledoc """
  The Catalog context.
  """

  import Ecto.Query, warn: false
  alias Hello.Repo

  alias Hello.Catalog.Product

  @doc """
  Returns the list of products.

  ## Examples

      iex> list_products()
      [%Product{}, ...]

  """
  def list_products do
    Repo.all(Product)
  end
  ...
end

This module will be the public API for all product catalog functionality in our system. For example, in addition to product detail management, we may also handle product category classification and product variants for things like optional sizing, trims, etc. If we look at the list_products/0 function, we can see the private details of product fetching. And it's super simple. We have a call to Repo.all(Product). We saw how Ecto repo queries worked in the Ecto guide, so this call should look familiar. Our list_products function is a generalized function name specifying the intent of our code – namely to list products. The details of that intent where we use our Repo to fetch the products from our PostgreSQL database is hidden from our callers. This is a common theme we'll see re-iterated as we use the Phoenix generators. Phoenix will push us to think about where we have different responsibilities in our application, and then to wrap up those different areas behind well-named modules and functions that make the intent of our code clear, while encapsulating the details.

Now we know how data is fetched, but how are products persisted? Let's take a look at the Catalog.create_product/1 function:

  @doc """
  Creates a product.

  ## Examples

      iex> create_product(%{field: value})
      {:ok, %Product{}}

      iex> create_product(%{field: bad_value})
      {:error, %Ecto.Changeset{}}

  """
  def create_product(attrs \\ %{}) do
    %Product{}
    |> Product.changeset(attrs)
    |> Repo.insert()
  end

There's more documentation than code here, but a couple of things are important to highlight. First, we can see again that our Ecto Repo is used under the hood for database access. You probably also noticed the call to Product.changeset/2. We talked about changesets before, and now we see them in action in our context.

If we open up the Product schema in lib/hello/catalog/product.ex, it will look immediately familiar:

defmodule Hello.Catalog.Product do
  use Ecto.Schema
  import Ecto.Changeset

  schema "products" do
    field :description, :string
    field :price, :decimal
    field :title, :string
    field :views, :integer

    timestamps()
  end

  @doc false
  def changeset(product, attrs) do
    product
    |> cast(attrs, [:title, :description, :price, :views])
    |> validate_required([:title, :description, :price, :views])
  end
end

This is just what we saw before when we ran mix phx.gen.schema, except here we see a @doc false above our changeset/2 function. This tells us that while this function is publicly callable, it's not part of the public context API. Callers that build changesets do so via the context API. For example, Catalog.create_product/1 calls into our Product.changeset/2 to build the changeset from user input. Callers, such as our controller actions, do not access Product.changeset/2 directly. All interaction with our product changesets is done through the public Catalog context.

Adding Catalog functions

As we've seen, your context modules are dedicated modules that expose and group related functionality. Phoenix generates generic functions, such as list_products and update_product, but they only serve as a basis for you to grow your business logic and application from. Let's add one of the basic features of our catalog by tracking product page view count.

For any ecommerce system, the ability to track how many times a product page has been viewed is essential for marketing, suggestions, ranking, etc. While we could try to use the existing Catalog.update_product function, along the lines of Catalog.update_product(product, %{views: product.views + 1}), this would not only be prone to race conditions, but it would also require the caller to know too much about our Catalog system. To see why the race condition exists, let's walk through the possible execution of events:

Intuitively, you would assume the following events:

  1. User 1 loads the product page with count of 13
  2. User 1 saves the product page with count of 14
  3. User 2 loads the product page with count of 14
  4. User 2 loads the product page with count of 15

While in practice this would happen:

  1. User 1 loads the product page with count of 13
  2. User 2 loads the product page with count of 13
  3. User 1 saves the product page with count of 14
  4. User 2 saves the product page with count of 14

The race conditions would make this an unreliable way to update the existing table since multiple callers may be updating out of date view values. There's a better way.

Let's think of a function describes what we want to accomplish. Here's how we would like to use it:

product = Catalog.inc_page_views(product)

That looks great. Our callers will have no confusion over what this function does, and we can wrap up the increment in an atomic operation to prevent race conditions.

Open up your catalog context (lib/hello/catalog.ex), and add this new function:

  def inc_page_views(%Product{} = product) do
    {1, [%Product{views: views}]} =
      from(p in Product, where: p.id == ^product.id, select: [:views])
      |> Repo.update_all(inc: [views: 1])

    put_in(product.views, views)
  end

We built a query for fetching the current product given its ID which we pass to Repo.update_all. Ecto's Repo.update_all allows us to perform batch updates against the database, and is perfect for atomically updating values, such as incrementing our views count. The result of the repo operation returns the number of updated records, along with the selected schema values specified by the select option. When we receive the new product views, we use put_in(product.views, views) to place the new view count within the product struct.

With our context function in place, let's make use of it in our product controller. Update your show action in lib/hello_web/controllers/product_controller.ex to call our new function:

  def show(conn, %{"id" => id}) do
    product =
      id
      |> Catalog.get_product!()
      |> Catalog.inc_page_views()

    render(conn, "show.html", product: product)
  end

We modified our show action to pipe our fetched page into Catalog.inc_page_views/1, which will return the updated product. Then we rendered our template just as before. Let's try it out. Refresh one of your product pages a few times and watch the view count increase.

We can also see our atomic update in action in the ecto debug logs:

[debug] QUERY OK source="products" db=0.5ms idle=834.5ms
UPDATE "products" AS p0 SET "views" = p0."views" + $1 WHERE (p0."id" = $2) RETURNING p0."views" [1, 1]

Good work!

As we've seen, designing with contexts gives you a solid foundation to grow your application from. Using discrete, well-defined APIs that expose the intent of your system allows you to write more maintainable applications with reusable code. Now that we know how to start extending our context API, lets explore handling relationships within a context.

In-context Relationships

Our basic catalog features are nice, but let's take it up a notch by categorizing products. Many ecommerce solutions allow products to be categorized in different ways, such as a product being marked for fashion, power tools, and so on. Starting with a one-to-one relationship between product and categories will cause major code changes later if we need to start supporting multiple categories. Let's set up a category association that will allow us to start off tracking a single category per product, but easily support more later as we grow our features.

For now, categories will contain only textual information. Our first order of business is to decide where categories live in the application. We have our Catalog context, which manages the exhibition of our products. Product categorization is a natural fit here. Phoenix is also smart enough to generate code inside an existing context, which makes adding new resources to a context a breeze. Run the following command at your project root:

Sometimes it may be tricky to determine if two resources belong to the same context or not. In those cases, prefer distinct contexts per resource and refactor later if necessary. Otherwise you can easily end-up with large contexts of loosely related entities. Also keep in mind that the fact two resources are related does not necessarily mean they belong to the same context, otherwise you would quickly end-up with one large context, as the majority of resources in an application are connected to each other. To sum it up: if you are unsure, you should prefer explicit modules (contexts) between resources.

$ mix phx.gen.context Catalog Category categories \
title:string:unique

You are generating into an existing context.
...
Would you like to proceed? [Yn] y
* creating lib/hello/catalog/category.ex
* creating priv/repo/migrations/20210203192325_create_categories.exs
* injecting lib/hello/catalog.ex
* injecting test/hello/catalog_test.exs

Remember to update your repository by running migrations:

    $ mix ecto.migrate

This time around, we used mix phx.gen.context, which is just like mix phx.gen.html, except it doesn't generate the web files for us. Since we already have controllers and templates for managing products, we can integrate the new category features into our existing web form and product show page. We can see we now have a new Category schema alongside our product schema at lib/hello/catalog/category.ex, and Phoenix told us it was injecting new functions in our existing Catalog context for the category functionality. The injected functions will look very familiar to our product functions, with new functions like create_category, list_categories, and so on. Before we migrate up, we need to do a second bit of code generation. Our category schema is great for representing an individual category in the system, but we need to support a many-to-many relationship between products and categories. Fortunately, ecto allows us to do this simply with a join table, so let's generate that now with the ecto.gen.migration command:

$ mix ecto.gen.migration create_product_categories

* creating priv/repo/migrations/20210203192958_create_product_categories.exs

Next, let's open up the new migration file and add the following code to the change function:


defmodule Hello.Repo.Migrations.CreateProductCategories do
  use Ecto.Migration

  def change do
    create table(:product_categories, primary_key: false) do
      add :product_id, references(:products, on_delete: :delete_all)
      add :category_id, references(:categories, on_delete: :delete_all)
    end

    create index(:product_categories, [:product_id])
    create index(:product_categories, [:category_id])
    create unique_index(:product_categories, [:product_id, :category_id])
  end
end

We created a product_categories table and used the primary_key: false option since our join table does not need a primary key. Next we defined our :product_id and :category_id foreign key fields, and passed on_delete: :delete_all to ensure the database prunes our join table records if a linked product or category is deleted. By using a database constraint, we enforce data integrity at the database level, rather than relying on ad-hoc and error-prone application logic.

Next, we created indexes for our foreign keys, and a unique index to ensure a product cannot have duplicate categories. With our migrations in place, we can migrate up.

$ mix ecto.migrate

18:20:36.489 [info]  == Running 20210222231834 Hello.Repo.Migrations.CreateCategories.change/0 forward

18:20:36.493 [info]  create table categories

18:20:36.508 [info]  create index categories_title_index

18:20:36.512 [info]  == Migrated 20210222231834 in 0.0s

18:20:36.547 [info]  == Running 20210222231930 Hello.Repo.Migrations.CreateProductCategories.change/0 forward

18:20:36.547 [info]  create table product_categories

18:20:36.557 [info]  create index product_categories_product_id_index

18:20:36.558 [info]  create index product_categories_category_id_index

18:20:36.560 [info]  create index product_categories_product_id_category_id_index

18:20:36.562 [info]  == Migrated 20210222231930 in 0.0s

Now that we have a Catalog.Product schema and a join table to associate products and categories, we're nearly ready to start wiring up our new features. Before we dive in, we first need real categories to select in our web UI. Let's quickly seed some new categories in the application. Add the following code to your seeds file in priv/repo/seeds.exs:

for title <- ["Home Improvement", "Power Tools", "Gardening", "Books"] do
  {:ok, _} = Hello.Catalog.create_category(%{title: title})
end

We simply enumerate over a list of category titles and use the generated create_category/1 function of our catalog context to persist the new records. We can run the seeds with mix run:

$ mix run priv/repo/seeds.exs

[debug] QUERY OK db=3.1ms decode=1.1ms queue=0.7ms idle=2.2ms
INSERT INTO "categories" ("title","inserted_at","updated_at") VALUES ($1,$2,$3) RETURNING "id" ["Home Improvement", ~N[2021-02-03 19:39:53], ~N[2021-02-03 19:39:53]]
[debug] QUERY OK db=1.2ms queue=1.3ms idle=12.3ms
INSERT INTO "categories" ("title","inserted_at","updated_at") VALUES ($1,$2,$3) RETURNING "id" ["Power Tools", ~N[2021-02-03 19:39:53], ~N[2021-02-03 19:39:53]]
[debug] QUERY OK db=1.1ms queue=1.1ms idle=15.1ms
INSERT INTO "categories" ("title","inserted_at","updated_at") VALUES ($1,$2,$3) RETURNING "id" ["Gardening", ~N[2021-02-03 19:39:53], ~N[2021-02-03 19:39:53]]
[debug] QUERY OK db=2.4ms queue=1.0ms idle=17.6ms
INSERT INTO "categories" ("title","inserted_at","updated_at") VALUES ($1,$2,$3) RETURNING "id" ["Books", ~N[2021-02-03 19:39:53], ~N[2021-02-03 19:39:53]]

Perfect. Before we integrate categories in the web layer, we need to let our context know how to associate products and categories. First, open up lib/hello/catalog/product.ex and add the following association:

+ alias Hello.Catalog.Category

  schema "products" do
    field :description, :string
    field :price, :decimal
    field :title, :string
    field :views, :integer

+   many_to_many :categories, Category, join_through: "product_categories", on_replace: :delete

    timestamps()
  end

We used Ecto.Schema's many_to_many macro to let Ecto know how to associate our product to multiple categories thru the "product_categories" join table. We also used the on_replace: :delete option to declare that any existing join records should be deleted when we are changing our categories.

With our schema associations set up, we can implement the selection of categories in our product form. To do so, we need to translate the user input of catalog IDs from the front-end to our many-to-many association. Fortunately Ecto makes this a breeze now that our schema is set up. Open up your catalog context and make the following changes:

- def get_product!(id), do: Repo.get!(Product, id)
+ def get_product!(id) do
+   Product |> Repo.get(id) |> Repo.preload(:categories)
+ end

  def create_product(attrs \\ %{}) do
    %Product{}
-   |> Product.changeset(attrs)
+   |> change_product(attrs)
    |> Repo.insert()
  end

  def update_product(%Product{} = product, attrs) do
    product
-   |> Product.changeset(attrs)
+   |> change_product(attrs)
    |> Repo.update()
  end

  def change_product(%Product{} = product, attrs \\ %{}) do
-   Product.changeset(product, attrs)
+   categories = list_categories_by_id(attrs["category_ids"])

+   product
+   |> Repo.preload(:categories)
+   |> Product.changeset(attrs)
+   |> Ecto.Changeset.put_assoc(:categories, categories)
  end

+ def list_categories_by_id(nil), do: []
+ def list_categories_by_id(category_ids) do
+   Repo.all(from c in Category, where: c.id in ^category_ids)
+ end

First, we added Repo.preload to preload our categories when we fetch a product. This will allow us to reference product.categories in our controllers, templates, and anywhere else we want to make use of category information. Next, we modified our create_product and update_product functions to call into our existing change_product function to produce a changeset. Within change_product we added a lookup to find all categories if the "category_ids" attribute is present. Then we preloaded categories and called Ecto.Changeset.put_assoc to place the fetched categories into the changeset. Finally, we implemented the list_categories_by_id/1 function to query the categories matching the category IDs, or return an empty list if no "category_ids" attribute is present. Now our create_product and update_product functions receive a changeset with the category associations all ready to go once we attempt an insert or update against our repo.

Next, let's expose our new feature to the web by adding the category input to our product form. To keep our form template tidy, let's write a new function to wrap up the details of rendering a category select input for our product. Open up your ProductView in lib/hello_web/views/product_view.ex and key this in:

defmodule HelloWeb.ProductView do
  use HelloWeb, :view

  def category_select(f, changeset) do
    existing_ids = changeset |> Ecto.Changeset.get_field(:categories) |> Enum.map(& &1.id)

    category_opts =
      for cat <- Hello.Catalog.list_categories(),
          do: [key: cat.title, value: cat.id, selected: cat.id in existing_ids]

    multiple_select(f, :category_ids, category_opts)
  end
end

We added a new category_select/2 function which uses Phoenix.HTML's multiple_select/3 to generate a multiple select tag. We calculated the existing category IDs from our changeset, then used those values when generate the select options for the input tag. We did this by enumerating over all of our categories and returning the appropriate key, value, and selected values. We marked an option as selected if the category ID was found in those category IDs in our changeset.

With our category_select function in place, we can open up lib/hello_web/templates/product/form.html.heex and add:

  ...
  <%= label f, :views %>
  <%= number_input f, :views %>
  <%= error_tag f, :views %>

+ <%= category_select f, @changeset %>

  <div>
    <%= submit "Save" %>
  </div>

We added a category_select above our save button. Now let's try it out. Next, let's show the product's categories in the product show template. Add the following code to lib/hello_web/templates/product/show.html.heex:

  ...
+ <li>
+   <strong>Categories:</strong>
+   <%= for cat <- @product.categories do %>
+     <%= cat.title %>
+     <br>
+   <% end %>
+ </li>
</ul>

Now if we start the server with mix phx.server and visit http://localhost:4000/products/new, we'll see the new category multiple select input. Enter some valid product details, select a category or two, and click save.

Title: Elixir Flashcards
Description: Flash card set for the Elixir programming language
Price: 5.000000
Views: 0
Categories:
Education
Books

It's not much to look at yet, but it works! We added relationships within our context complete with data integrity enforced by the database. Not bad. Let's keep building!

Cross-context dependencies

Now that we have the beginnings of our product catalog features, let's begin to work on the other main features of our application – carting products from the catalog. In order to properly track products that have been added to a user's cart, we'll need a new place to persist this information, along with point-in-time product information like the price at time of carting. This is necessary so we can detect product price changes in the future. We know what we need to build, but now we need to decide where the cart functionality lives in our application.

If we take a step back and think about the isolation of our application, the exhibition of products in our catalog distinctly differs from the responsibilities of managing a user's cart. A product catalog shouldn't care about the rules of a our shopping cart system, and vice-versa. There's a clear need here for a separate context to handle the new cart responsibilities. Let's call it ShoppingCart.

Let's create a ShoppingCart context to handle basic cart duties. Before we write code, let's imagine we have the following feature requirements:

  1. Add products to a user's cart from the product show page
  2. Store point-in-time product price information at time of carting
  3. Store and update quantities in cart
  4. Calculate and display sum of cart prices

From the description, it's clear we need a Cart resource for storing the user's cart, along with a CartItem to track products in the cart. With our plan set, let's get to work. Run the following command to generate our new context:

$ mix phx.gen.context ShoppingCart Cart carts user_uuid:uuid:unique

* creating lib/hello/shopping_cart/cart.ex
* creating priv/repo/migrations/20210205203128_create_carts.exs
* creating lib/hello/shopping_cart.ex
* injecting lib/hello/shopping_cart.ex
* creating test/hello/shopping_cart_test.exs
* injecting test/hello/shopping_cart_test.exs

Remember to update your repository by running migrations:

    $ mix ecto.migrate

We generated our new context ShoppingCart, with a new ShoppingCart.Cart schema to tie a user to their cart which holds cart items. We don't have real users yet, so for now our cart will be tracked by an anonymous user UUID that we'll add to our plug session in a moment. With our cart in place, let's generate our cart items:

$ mix phx.gen.context ShoppingCart CartItem cart_items \
cart_id:references:carts product_id:references:products \
price_when_carted:decimal quantity:integer

You are generating into an existing context.
...
Would you like to proceed? [Yn] y
* creating lib/hello/shopping_cart/cart_item.ex
* creating priv/repo/migrations/20210205213410_create_cart_items.exs
* injecting lib/hello/shopping_cart.ex
* injecting test/hello/shopping_cart_test.exs

Remember to update your repository by running migrations:

    $ mix ecto.migrate

We generated a new resource inside our ShoppingCart named CartItem. This schema and table will hold references to a cart and product, along with the price at the time we added the item to our cart, and the quantity the user wishes to purchase. Let's touch up the generated migration file in priv/repo/migrations/*_create_cart_items.ex:

    create table(:cart_items) do
-     add :price_when_carted, :decimal
+     add :price_when_carted, :decimal, precision: 15, scale: 6, null: false
      add :quantity, :integer
-     add :cart_id, references(:carts, on_delete: :nothing)
+     add :cart_id, references(:carts, on_delete: :delete_all)
-     add :product_id, references(:products, on_delete: :nothing)
+     add :product_id, references(:products, on_delete: :delete_all)

      timestamps()
    end

    create index(:cart_items, [:cart_id])
    create index(:cart_items, [:product_id])
+   create unique_index(:cart_items, [:cart_id, :product_id])

We used the :delete_all strategy again to enforce data integrity. This way, when a cart or product is deleted from the application, we don't have to rely on application code in our ShoppingCart or Catalog contexts to worry about cleaning up the records. This keeps our application code decoupled and the data integrity enforcement where it belongs – in the database. We also added a unique constraint to ensure a duplicate product is not allowed to be added to a cart. With our database tables in place, we can now migrate up:

$ mix ecto.migrate

16:59:51.941 [info]  == Running 20210205203342 Hello.Repo.Migrations.CreateCarts.change/0 forward

16:59:51.945 [info]  create table carts

16:59:51.949 [info]  create index carts_user_uuid_index

16:59:51.952 [info]  == Migrated 20210205203342 in 0.0s

16:59:51.988 [info]  == Running 20210205213410 Hello.Repo.Migrations.CreateCartItems.change/0 forward

16:59:51.988 [info]  create table cart_items

16:59:51.998 [info]  create index cart_items_cart_id_index

16:59:52.000 [info]  create index cart_items_product_id_index

16:59:52.001 [info]  create index cart_items_cart_id_product_id_index

16:59:52.002 [info]  == Migrated 20210205213410 in 0.0s

Our database is ready to go with new carts and cart_items tables, but now we need to map that back into application code. You may be wondering how we can mix database foreign keys across different tables and how that relates to the context pattern of isolated, grouped functionality. Let's jump in and discuss the approaches and their tradeoffs.

Cross-context data

Dependencies in your software are often unavoidable, but we can do our best to limit them where possible and lessen the maintenance burden when a dependency is necessary. So far, we've done a great job isolating the two main contexts of our application from each other, but now we have a necessary dependency to handle.

Our Catalog.Product resource serves to keep the responsibilities of representing a product inside the catalog, but ultimately for an item to exist in the cart, a product from the catalog must be present. Given this, our ShoppingCart context will have a data dependency on the Catalog context. With that in mind, we have two options. One is to expose APIs on the Catalog context that allows us to efficiently fetch product data for use in the ShoppingCart system, or we can use database joins to fetch the dependent data. Both are valid options given your tradeoffs and application size, but joining data from the database when you have a hard data dependency is just fine for a large class of applications. Additionally, joining data yourself results in increased application code complexity, worse performance, and potential data inconsistencies. Also, if you decide to break out coupled contexts into entirely separate applications and databases at a later time, you still gain the benefits of isolation. This is because your public context APIs will likely remain unchanged.

Now that we know where our data dependencies exist, let's add our schema associations so we can tie shopping cart items to products. First, let's make a quick change to our cart schema in lib/hello/shopping_cart/cart.ex to associate a cart to its items:

  schema "carts" do
    field :user_uuid, Ecto.UUID

+   has_many :items, Hello.ShoppingCart.CartItem

    timestamps()
  end

Now that our cart is associated to the items we place in it, let's set up the cart item associations inside lib/hello/shopping_cart/cart_item.ex:

  schema "cart_items" do
-   field :cart_id, :id
-   field :product_id, :id
    field :price_when_carted, :decimal
    field :quantity, :integer

+   belongs_to :cart, Hello.ShoppingCart.Cart
+   belongs_to :product, Hello.Catalog.Product

    timestamps()
  end

  @doc false
  def changeset(cart_item, attrs) do
    cart_item
    |> cast(attrs, [:price_when_carted, :quantity])
    |> validate_required([:price_when_carted, :quantity])
+   |> validate_number(:quantity, greater_than_or_equal_to: 0, less_than: 100)
  end

First, we replaced the cart_id field with a standard belongs_to pointing at our ShoppingCart.Cart schema. Next, we replaced our product_id field by adding our first cross-context data dependency with a belongs_to for the Catalog.Product schema. Here, we intentionally coupled the data boundaries because it provides exactly what we need. An isolated context API with the bare minimum knowledge necessary to reference a product in our system. Next, we added a new validation to our changeset. With validate_number/3, we ensure any quantity provided by user input is between 0 and 100.

With our schemas in place, we can start integrating the new data structures and ShoppingCart context APIs into our web-facing features.

Adding Shopping Cart functions

As we mentioned before, the context generators are only a starting point for our application. We can and should write well-named, purpose built functions to accomplish the goals of our context. We have a few new features to implement. First, we need to ensure every user of our application is granted a cart if one does not yet exist. From there, we can then allow users to add items to their cart, update item quantities, and calculate cart totals. Let's get started!

We won't focus on a real user authentication system at this point, but by the time we're done, you'll be able to naturally integrate one with what we've written here. To simulate a current user session, open up your lib/hello_web/router.ex and key this in:

  pipeline :browser do
    plug :accepts, ["html"]
    plug :fetch_session
    plug :fetch_flash
    plug :protect_from_forgery
    plug :put_secure_browser_headers
+   plug :fetch_current_user
+   plug :fetch_current_cart
  end

+ defp fetch_current_user(conn, _) do
+   if user_uuid = get_session(conn, :current_uuid) do
+     assign(conn, :current_uuid, user_uuid)
+   else
+     new_uuid = Ecto.UUID.generate()
+
+     conn
+     |> assign(:current_uuid, new_uuid)
+     |> put_session(:current_uuid, new_uuid)
+   end
+ end

+ alias Hello.ShoppingCart
+
+ def fetch_current_cart(conn, _opts) do
+   if cart = ShoppingCart.get_cart_by_user_uuid(conn.assigns.current_uuid) do
+     assign(conn, :cart, cart)
+   else
+     {:ok, new_cart} = ShoppingCart.create_cart(conn.assigns.current_uuid)
+     assign(conn, :cart, new_cart)
+   end
+ end

We added a new :fetch_current_user and :fetch_current_cart plug to our browser pipeline to run on all browser-based requests. Next, we implemented the fetch_current_user plug which simply checks the session for a user UUID that was previously added. If we find one, we add a current_uuid assign to the connection and we're done. In the case we haven't yet identified this visitor, we generate a unique UUID with Ecto.UUID.generate(), then we place that value in the current_uuid assign, along with a new session value to identify this visitor on future requests. A random, unique ID isn't much to represent a user, but it's enough for us to track and identify a visitor across requests, which is all we need for now. Later as our application becomes more complete, you'll be ready to migrate to a complete user authentication solution. With a guaranteed current user, we then implemented the fetch_current_cart plug which either finds a cart for the user UUID or creates a cart for the current user and assigns the result in the connection assigns. We'll need to implement our ShoppingCart.get_cart_by_user_uuid/1 and modify the create cart function to accept a UUID, but let's add our routes first.

We'll need to implement a cart controller for handling cart operations like viewing a cart, updating quantities, and initiating the checkout process, as well as a cart items controller for adding and removing individual items to and from the cart. Add the following routes to your router in lib/hello_web/router.ex:

  scope "/", HelloWeb do
    pipe_through :browser

    get "/", PageController, :index
    resources "/products", ProductController

+   resources "/cart_items", CartItemController, only: [:create, :delete]

+   get "/cart", CartController, :show
+   put "/cart", CartController, :update
  end

We added a resources declaration for a CartItemController, which will wire up the routes for a create and delete action for adding and remove individual cart items. Next, we added two new routes pointing at a CartController. The first route, a GET request, will map to our show action, to show the cart contents. The second route, a PUT request, will handle the submission of a form for updating our cart quantities.

With our routes in place, let's add the ability to add an item to our cart from the product show page. Create a new file at lib/hello_web/cart_item_controller.ex and key this in:

defmodule HelloWeb.CartItemController do
  use HelloWeb, :controller

  alias Hello.{ShoppingCart, Catalog}

  def create(conn, %{"product_id" => product_id}) do
    product = Catalog.get_product!(product_id)

    case ShoppingCart.add_item_to_cart(conn.assigns.cart, product) do
      {:ok, _item} ->
        conn
        |> put_flash(:info, "Item added to your cart")
        |> redirect(to: Routes.cart_path(conn, :show))

      {:error, _changeset} ->
        conn
        |> put_flash(:error, "There was an error adding the item to your cart")
        |> redirect(to: Routes.cart_path(conn, :show))
    end
  end

  def delete(conn, %{"id" => product_id}) do
    {:ok, _cart} = ShoppingCart.remove_item_from_cart(conn.assigns.cart, product_id)
    redirect(conn, to: Routes.cart_path(conn, :show))
  end
end

We defined a new CartItemController with the create and delete actions that we declared in our router. For create, we first lookup the product in the catalog with Catalog.get_product!/1, then we call a ShoppingCart.add_item_to_cart/2 function which we'll implement in a moment. If successful, we show a flash successful message and redirect to the cart show page; else, we show a flash error message and redirect to the cart show page. For delete, we'll call a remove_item_from_cart function which we'll implement on our ShoppingCart context and then redirect back to the cart show page. We haven't implemented these two shopping cart functions yet, but notice how their names scream their intent: add_item_to_cart and remove_item_from_cart make it obvious what we are accomplishing here. It also allows us to spec our our web layer and context APIs without thinking about all the implementation details at once.

Let's implement our new interface of ShoppingCart context API in lib/hello/shopping_cart.ex:

  alias Hello.Catalog
  alias Hello.ShoppingCart.{Cart, CartItem}

  def get_cart_by_user_uuid(user_uuid) do
    Repo.one(
      from(c in Cart,
        where: c.user_uuid == ^user_uuid,
        left_join: i in assoc(c, :items),
        left_join: p in assoc(i, :product),
        order_by: [asc: i.inserted_at],
        preload: [items: {i, product: p}]
      )
    )
  end

- def create_cart(attrs \\ %{}) do
-   %Cart{}
+ def create_cart(user_uuid) do
+   %Cart{user_uuid: user_uuid}
    |> Cart.changeset(%{})
    |> Repo.insert()
+   |> case do
+     {:ok, cart} -> {:ok, reload_cart(cart)}
+     {:error, changeset} -> {:error, changeset}
+   end
  end

  defp reload_cart(%Cart{} = cart), do: get_cart_by_user_uuid(cart.user_uuid)

  def add_item_to_cart(%Cart{} = cart, %Catalog.Product{} = product) do
    %CartItem{quantity: 1, price_when_carted: product.price}
    |> CartItem.changeset(%{})
    |> Ecto.Changeset.put_assoc(:cart, cart)
    |> Ecto.Changeset.put_assoc(:product, product)
    |> Repo.insert(
      on_conflict: [inc: [quantity: 1]],
      conflict_target: [:cart_id, :product_id]
    )
  end

  def remove_item_from_cart(%Cart{} = cart, product_id) do
    {1, _} =
      Repo.delete_all(
        from(i in CartItem,
          where: i.cart_id == ^cart.id,
          where: i.product_id == ^product_id
        )
      )

    {:ok, reload_cart(cart)}
  end

We started by implementing get_cart_by_user_uuid/1 which fetches our cart and joins the cart items, and their products so that we have the full cart populated with all preloaded data. Next, we modified our create_cart function to accept a user UUID instead of attributes, which we used to populate the user_uuid field. If the insert is successful, we reload the cart contents by calling a private reload_cart/1 function, which simply calls get_cart_by_user_uuid/1 to refetch data. Next, we wrote our new add_item_to_cart/2 function which accepts a cart struct and a product struct from the catalog. We used an upsert operation against our repo to either insert a new cart item into the database, or increase the quantity by one if it already exists in the cart. This is accomplished via the on_conflict and conflict_target options, which tells our repo how to handle an insert conflict. Next, we implemented remove_item_from_cart/2 where we simply issue a Repo.delete_all call with a query to delete the cart item in our cart that matches the product ID. Finally, we reload the cart contents by calling reload_cart/1.

With our new cart functions in place, we can now expose the "Add to cart" button on the product catalog show page. Open up your template in lib/hello_web/templates/product/show.html.heex and make the following changes:

<h1>Show Product</h1>

+<%= link "Add to cart",
+  to: Routes.cart_item_path(@conn, :create, product_id: @product.id),
+  method: :post %>
...

The link functions from Phoenix.HTML accepts a :method option to issue an HTTP verb when clicked, instead of the default GET request. With this link in place, the "Add to cart" link will issue a POST request, which will be matched by the route we defined in router which dispatches to the CartItemController.create/2 function.

Let's try it out. Start your server with mix phx.server and visit a product page. If we try clicking the add to cart link, we'll be greeted by an error page with the following logs in the console:

[info] POST /cart_items
[debug] Processing with HelloWeb.CartItemController.create/2
  Parameters: %{"_method" => "post", "product_id" => "1", ...}
  Pipelines: [:browser]
INSERT INTO "cart_items" ...
[info] Sent 302 in 24ms
[info] GET /cart
[debug] Processing with HelloWeb.CartController.show/2
  Parameters: %{}
  Pipelines: [:browser]
[debug] QUERY OK source="carts" db=1.9ms idle=1798.5ms

[error] #PID<0.856.0> running HelloWeb.Endpoint (connection #PID<0.841.0>, stream id 5) terminated
Server: localhost:4000 (http)
Request: GET /cart
** (exit) an exception was raised:
    ** (UndefinedFunctionError) function HelloWeb.CartController.init/1 is undefined
       (module HelloWeb.CartController is not available)
       ...

It's working! Kind of. If we follow the logs, we see our POST to the /cart_items path. Next, we can see our ShoppingCart.add_item_to_cart function successfully inserted a row into the cart_items table, and then we issued a redirect to /cart. Before our error, we also see a query to the carts table, which means we're fetching the current user's cart. So far so good. We know our CartItem controller and new ShoppingCart context functions are doing their jobs, but we've hit our next unimplemented feature when the router attempts to dispatch to a non-existent cart controller. Let's create the cart controller, view, and template to display and manage user carts.

Create a new file at lib/hello_web/controllers/cart_controller.ex and key this in:

defmodule HelloWeb.CartController do
  use HelloWeb, :controller

  alias Hello.ShoppingCart

  def show(conn, _params) do
    render(conn, "show.html", changeset: ShoppingCart.change_cart(conn.assigns.cart))
  end
end

We defined a new cart controller to handle the get "/cart" route. For showing a cart, we render a "show.html" template which we'll create in moment. We know we need to allow the cart items to be changed by quantity updates, so right away we know we'll need a cart changeset. Fortunately, the context generator included a ShoppingChart.change_cart/1 function, which we'll use. We pass it our cart struct which is already in the connection assigns thanks to the fetch_current_cart plug we defined in the router.

Next, we can implement the view and template. Create a new view file at lib/hello_web/views/cart_view.ex with the following content:

defmodule HelloWeb.CartView do
  use HelloWeb, :view

  alias Hello.ShoppingCart

  def currency_to_str(%Decimal{} = val), do: "$#{Decimal.round(val, 2)}"
end

We created a view to render our show.html template and aliased our ShoppingCart context so it will be in scope for our template. We'll need to display the cart prices like product item price, cart total, etc, so we defined a currency_to_str/1 which takes our decimal struct, rounds it properly for display, and prepends a USD dollar sign.

Next we can create the template at lib/hello_web/templates/cart/show.html.heex:

<h1>My Cart</h1>

<%= if @cart.items == [] do %>
  Your cart is empty
<% else %>
  <%= form_for @changeset, Routes.cart_path(@conn, :update), fn f -> %>
    <ul>
      <%= for item_form <- inputs_for(f, :items), item = item_form.data do %>
        <li>
          <%= hidden_inputs_for(item_form) %>
          <%= item.product.title %>
          <%= number_input item_form, :quantity %>
          <%= currency_to_str(ShoppingCart.total_item_price(item)) %>
        </li>
      <% end %>
    </ul>

    <%= submit "update cart" %>
  <% end %>

  <b>Total</b>: <%= currency_to_str(ShoppingCart.total_cart_price(@cart)) %>
<% end %>

We started by showing the empty cart message if our preloaded cart.items is empty. If we have items, we use form_for to take our cart changeset that we assigned in the CartController.show/2 action and create a form which maps to our cart controller update/2 action. Within the form, we use Phoenix.HTML.Form.inputs_for to render inputs for the nested cart items. For each item form input, we use hidden_inputs_for which will render out the item ID as a hidden input tag. This will allow us to map item inputs back together when the form is submitted. Next, we display the product title for the item in the cart, followed by a number input for the item quantity. We finish the item form by converting the item price to string. We haven't written the ShoppingCart.total_item_price/1 function yet, but again we employed the idea of clear, descriptive public interfaces for our contexts. After rendering inputs for all the cart items, we show an "update cart" submit button, along with the total price of the entire cart. This is accomplished with another new ShoppingCart.total_cart_price/1 function which we'll implement in a moment.

We're almost ready to try out our cart page, but first we need to implement our new currency calculation functions. Open up your shopping cart context at lib/hello/shopping_cart.ex and add these new functions:

  def total_item_price(%CartItem{} = item) do
    Decimal.mult(item.product.price, item.quantity)
  end

  def total_cart_price(%Cart{} = cart) do
    Enum.reduce(cart.items, 0, fn item, acc ->
      item
      |> total_item_price()
      |> Decimal.add(acc)
    end)
  end

We implemented total_item_price/1 which accepts a %CartItem{} struct. To calculate the total price, we simply take the preloaded product's price and multiply it by the item's quantity. We used Decimal.mult/2 to take our decimal currency struct and multiply it with proper precision. Similarly for calculating the total cart price, we implemented a total_cart_price/1 function which accepts the cart and sums the preloaded product prices for items in the cart. We again make use of the Decimal functions to add our decimal structs together.

Now that we can calculate price totals, let's try it out! Visit http://localhost:4000/cart and you should already see your first item in the cart. Going back to the same product and clicking "add to cart" will show our upsert in action. Your quantity should now show be two. Nice work!

Our cart page is almost complete, but submitting the form will yield yet another error.

Request: POST /cart
** (exit) an exception was raised:
    ** (UndefinedFunctionError) function HelloWeb.CartController.update/2 is undefined or private

Let's head back to our CartController at lib/hello_web/controllers/cart_controller.ex and implement the update action:

  def update(conn, %{"cart" => cart_params}) do
    case ShoppingCart.update_cart(conn.assigns.cart, cart_params) do
      {:ok, cart} ->
        redirect(conn, to: Routes.cart_path(conn, :show))

      {:error, _changeset} ->
        conn
        |> put_flash(:error, "There was an error updating your cart")
        |> redirect(to: Routes.cart_path(conn, :show))
    end
  end

We started by plucking out the cart params from the form submit. Next, we call our existing ShoppingCart.update_cart/2 function which was added by the context generator. We'll need to make some changes to this function, but the interface is good as is. If the update is successful, we redirect back to the cart page, otherwise we show a flash error message and send the user back to the cart page to fix any mistakes. Out-of-the-box, our ShoppingCart.update_cart/2 function only concerned itself with casting the cart params into a changeset and updates it against our repo. For our purposes, we now need it to handled nested cart item associations, and most importantly, business logic for how to handle quantity updates like zero-quantity items being removed from the cart.

Head back over to your shopping cart context in lib/hello/shopping_cart.ex and replace your update_cart/2 function with the following implementation:

  def update_cart(%Cart{} = cart, attrs) do
    changeset =
      cart
      |> Cart.changeset(attrs)
      |> Ecto.Changeset.cast_assoc(:items, with: &CartItem.changeset/2)

    Ecto.Multi.new()
    |> Ecto.Multi.update(:cart, changeset)
    |> Ecto.Multi.delete_all(:discarded_items, fn %{cart: cart} ->
      from(i in CartItem, where: i.cart_id == ^cart.id and i.quantity == 0)
    end)
    |> Repo.transaction()
    |> case do
      {:ok, %{cart: cart}} -> {:ok, cart}
      {:error, :cart, changeset, _changes_so_far} -> {:error, changeset}
    end
  end

We started much like how our out-of-the-box code started – we take the cart struct and cast the user input to a cart changeset, except this time we use Ecto.Changeset.cast_assoc/3 to cast the nested item data into CartItem changesets. Remember the hidden_inputs_for/1 call in our cart form template? That hidden ID data is what allows Ecto's cast_assoc to map item data back to existing item associations in the cart. Next we use Ecto.Multi.new(), which you may not have seen before. Ecto's Multi is a feature that allows lazily defining a chain of named operations to eventually execute inside a database repo transaction. Each operation in the multi chain receives the values from the previous steps and executes until a failed step is encountered. When an operation fails, the transaction is rolled back and an error is returned, otherwise the transaction is committed.

For our multi operations, we start by issuing an update of our cart, which we named :cart. After the cart update is issued, we perform a multi delete_all operation, which takes the updated cart and applies our zero-quantity logic. We prune any items in the cart with zero quantity by returning an ecto query that finds all cart items for this cart with an empty quantity. Calling Repo.transaction/1 with our multi will execute the operations in a new transaction and we return the success or failure result to the caller just like the original function.

Let's head back to the browser and try it out. Add a few products to your cart, update the quantities, and watch the values changes along with the price calculations. Setting any quantity to 0 will also remove the item. Pretty neat!

Adding an Orders context

With our Catalog and ShoppingCart contexts, we're seeing first-hand how our well-considered modules and function names are yielding clear and maintainable code. Our last order of business is to allow the user to initiate the checkout process. We won't go as far as integrating payment processing or order fulfillment, but we'll get you started in that direction. Like before, we need to decide where code for completing an order should live. Is it part of the catalog? Clearly not, but what about the shopping cart? Shopping carts are related to orders – after all the user has to add items in order to purchase any products, but should the order checkout process be grouped here?

If we stop and consider the order process, we'll see that orders involve related, but distinctly different data from the cart contents. Also, business rules around the checkout process are much different than carting. For example, we may allow a user to add a back-ordered item to their cart, but we could not allow an order with no inventory to be completed. Additionally, we need to capture point-in-time product information when an order is completed, such as the price of the items at payment transaction time. This is essential because a product price may change in the future, but the line items in our order must always record and display what we charged at time of purchase. For these reasons, we can start to see ordering can reasonably stand on its own with its own data concerns and business rules.

Naming wise, Orders clearly defines the scope of our context, so let's get started by again taking advantage of the context generators. Run the following command in your console:

$ mix phx.gen.html Orders Order orders user_uuid:uuid total_price:decimal

* creating lib/hello_web/controllers/order_controller.ex
* creating lib/hello_web/templates/order/edit.html.heex
* creating lib/hello_web/templates/order/form.html.heex
* creating lib/hello_web/templates/order/index.html.heex
* creating lib/hello_web/templates/order/new.html.heex
* creating lib/hello_web/templates/order/show.html.heex
* creating lib/hello_web/views/order_view.ex
* creating test/hello_web/controllers/order_controller_test.exs
* creating lib/hello/orders/order.ex
* creating priv/repo/migrations/20210209214612_create_orders.exs
* creating lib/hello/orders.ex
* injecting lib/hello/orders.ex
* creating test/hello/orders_test.exs
* injecting test/hello/orders_test.exs

Add the resource to your browser scope in lib/hello_web/router.ex:

    resources "/orders", OrderController


Remember to update your repository by running migrations:

    $ mix ecto.migrate

We generated a Orders context along with HTML controllers, views, etc. We added a user_uuid field to associate our placeholder current user to an order, along with a total_price column. With our starting point in place, let's open up the newly created migration in priv/repo/migrations/*_create_orders.exs and make the following changes:

  def change do
    create table(:orders) do
      add :user_uuid, :uuid
-     add :total_price, :decimal
+     add :total_price, :decimal, precision: 15, scale: 6, null: false

      timestamps()
    end
  end

Like we did previously, we gave appropriate precision and scale options for our decimal column which will allow us to store currency without precision loss. We also added a not-null constraint to enforce all orders to have a price.

The orders table alone doesn't hold much information, but we know we'll need to store point-in-time product price information of all the items in the order. For that, we'll add additional struct for this context named LineItem. Line items will capture the price of the product at payment transaction time. Please run the following command:

$ mix phx.gen.context Orders LineItem order_line_items \
price:decimal quantity:integer \
order_id:references:orders product_id:references:products

You are generating into an existing context.
Would you like to proceed? [Yn] y
* creating lib/hello/orders/line_item.ex
* creating priv/repo/migrations/20210209215050_create_order_line_items.exs
* injecting lib/hello/orders.ex
* injecting test/hello/orders_test.exs

Remember to update your repository by running migrations:

    $ mix ecto.migrate

We used the phx.gen.context command to generate the LineItem Ecto schema and inject supporting functions into our orders context. Like before, let's modify the migration in priv/repo/migrations/*_create_order_line_items.exs and make the following decimal field changes:

  def change do
    create table(:order_line_items) do
-     add :price, :decimal
+     add :price, :decimal, precision: 15, scale: 6, null: false
      add :quantity, :integer
      add :order_id, references(:orders, on_delete: :nothing)
      add :product_id, references(:products, on_delete: :nothing)

      timestamps()
    end

    create index(:order_line_items, [:order_id])
    create index(:order_line_items, [:product_id])
  end

With our migration in place, let's wire up our orders and line items associations in lib/hello/orders/order.ex:

  schema "orders" do
    field :total_price, :decimal
    field :user_uuid, Ecto.UUID

+   has_many :line_items, Hello.Orders.LineItem
+   has_many :products, through: [:line_items, :product]

    timestamps()
  end

We used has_many :line_items to associate orders and line items, just like we've seen before. Next, we used the :through feature of has_many, which allows us to instruct ecto how to associate resources across another relationship. In this case, we can associate products of an order by finding all products through associated line items. Next, let's wire up the association in the other direction in lib/hello/orders/line_item.ex:

  schema "order_line_items" do
    field :price, :decimal
    field :quantity, :integer
-   field :order_id, :id
-   field :product_id, :id

+   belongs_to :order, Hello.Orders.Order
+   belongs_to :product, Hello.Catalog.Product

    timestamps()
  end

We used belongs_to to associate line items to orders and products. With our associations in place, we can start integrating the web interface into our order process. Open up your router lib/hello_web/router.ex and add the following line:

  scope "/", HelloWeb do
    pipe_through :browser

    ...
+   resources "/orders", OrderController, only: [:create, :show]
  end

We wired up create and show routes for our generated OrderController, since these are the only actions we need at the moment. With our routes in place, we can now migrate up:

$ mix ecto.migrate

17:14:37.715 [info]  == Running 20210209214612 Hello.Repo.Migrations.CreateOrders.change/0 forward

17:14:37.720 [info]  create table orders

17:14:37.755 [info]  == Migrated 20210209214612 in 0.0s

17:14:37.784 [info]  == Running 20210209215050 Hello.Repo.Migrations.CreateOrderLineItems.change/0 forward

17:14:37.785 [info]  create table order_line_items

17:14:37.795 [info]  create index order_line_items_order_id_index

17:14:37.796 [info]  create index order_line_items_product_id_index

17:14:37.798 [info]  == Migrated 20210209215050 in 0.0s

Before we render information about our orders, we need to ensure our order data is fully populated and can be looked up by a current user. Open up your orders context in lib/hello/orders.ex and replace your get_order!/1 function by a new get_order!/2 definition:

  def get_order!(user_uuid, id) do
    Order
    |> Repo.get_by!(id: id, user_uuid: user_uuid)
    |> Repo.preload([line_items: [:product]])
  end

We rewrote the function to accept a user UUID and query our repo for an order matching the user's ID for a given order ID. Then we populated the order by preloading our line item and product associations.

To complete an order, our cart page can issue a POST to the OrderController.create action, but we need to implement the operations and logic to actually complete an order. Like before, we'll start at the web interface by rewriting the create function in lib/hello_web/controllers/order_controller.ex:

  def create(conn, _) do
    case Orders.complete_order(conn.assigns.cart) do
      {:ok, order} ->
        conn
        |> put_flash(:info, "Order created successfully.")
        |> redirect(to: Routes.order_path(conn, :show, order))

      {:error, _reason} ->
        conn
        |> put_flash(:error, "There was an error processing your order")
        |> redirect(to: Routes.cart_path(conn, :show))
    end
  end

We rewrote the create action to call an as-yet-implemented Orders.complete_order/1 function. The code that phoenix generated had a generic Orders.create_order/1 call. Our code is technically "creating" an order, but it's important to step back and consider the naming of your interfaces. The act of completing an order is extremely important in our system. Money changes hands in a transaction, physical goods could be automatically shipped, etc. Such an operation deserves a better, more obvious function name, such as complete_order. If the order is completed successfully we redirect to the show page, otherwise a flash error is shown as we redirect back to the cart page.

Now we can implement our Orders.complete_order/1 function. To complete an order, our job will require a few operations:

  1. A new order record must be persisted with the total price of the order
  2. All items in the cart must be transformed into new order line items records with quantity and point-in-time product price information
  3. After successful order insert (and eventual payment), items must be pruned from the cart

From our requirements alone, we can start to see why a generic create_order function doesn't cut it. Let's implement this new function in lib/hello/orders.ex:

  alias Hello.ShoppingCart
  alias Hello.Orders.LineItem

  def complete_order(%ShoppingCart.Cart{} = cart) do
    line_items =
      Enum.map(cart.items, fn item ->
        %{product_id: item.product_id, price: item.product.price, quantity: item.quantity}
      end)

    order =
      Ecto.Changeset.change(%Order{},
        user_uuid: cart.user_uuid,
        total_price: ShoppingCart.total_cart_price(cart),
        line_items: line_items
      )

    Ecto.Multi.new()
    |> Ecto.Multi.insert(:order, order)
    |> Ecto.Multi.run(:prune_cart, fn _repo, _changes ->
      ShoppingCart.prune_cart_items(cart)
    end)
    |> Repo.transaction()
    |> case do
      {:ok, %{order: order}} -> {:ok, order}
      {:error, name, value, _changes_so_far} -> {:error, {name, value}}
    end
  end

We started by mapping the %ShoppingCart.CartItem{}'s in our shopping cart into a map of order line items structs. The job of order line item record is to capture the price of the product at payment transaction time, so we reference the product's price here. Next, we create a bare order changeset with Ecto.Changeset.change/2 and associate our user UUID, set our total price calculation, and place our order line items in the changeset. With a fresh order changeset ready to be inserted, we can again make use of Ecto.Multi to execute our operations in a database transaction. We start by inserting the order, followed by a run operation. The Ecto.Multi.run function allows us to run any code in the function which must either succeed with {:ok, result} or error, which halts and rolls back the transaction. Here, we simply can call into our shopping cart context and ask it to prune all items in a cart. Running the transaction will execute the multi as before and we return the result to the caller.

To close out our order completion, we need to implement the ShoppingCart.prune_cart_items/1 function in lib/hello/shopping_cart.ex:

  def prune_cart_items(%Cart{} = cart) do
    {_, _} = Repo.delete_all(from(i in CartItem, where: i.cart_id == ^cart.id))
    {:ok, reload_cart(cart)}
  end

Our new function accepts the cart struct and issues a Repo.delete_all which accepts a query of all items for the provided cart. We return a success result by simply reloading the pruned cart to the caller. With our context complete, we now need to show the user their completed order. Head back to your order controller and modify the show/2 action:

  def show(conn, %{"id" => id}) do
-   order = Orders.get_order!(id)
+   order = Orders.get_order!(conn.assigns.current_uuid, id)
    render(conn, "show.html", order: order)
  end

We tweaked the show action to pass our conn.assigns.current_uuid to get_order! which authorizes orders to be viewable only by the owner of the order. Next, we can replace the order show template in lib/hello_web/templates/order/show.html.heex:

<h1>Thank you for your order!</h1>

<ul>
  <li>
    <strong>User uuid:</strong>
    <%= @order.user_uuid %>
  </li>

  <%= for item <- @order.line_items do %>
    <li>
      <%= item.product.title %>
      (<%= item.quantity %>) - <%= HelloWeb.CartView.currency_to_str(item.price) %>
    </li>
  <% end %>

  <li>
    <strong>Total price:</strong>
    <%= HelloWeb.CartView.currency_to_str(@order.total_price) %>
  </li>

</ul>

<span><%= link "Back", to: Routes.cart_path(@conn, :show) %></span>

To show our completed order, we displayed the order's user, followed by the line item listing with product title, quantity, and the price we "transacted" when completing the order, along with the total price.

Our last addition will be to add the "complete order" button to our cart page to allow completing an order. Add the following button to the bottom of the cart show template in lib/hello_web/templates/cart/show.html.heex:

  <b>Total</b>: <%= currency_to_str(ShoppingCart.total_cart_price(@cart)) %>

+ <%= link "complete order", to: Routes.order_path(@conn, :create), method: :post %>
<% end %>

We added a link with method: :post to send a POST request to our OrderController.create action. If we head back to our cart page at http://localhost:4000/cart and complete an order, we'll be greeted by our rendered template:

Thank you for your order!

User uuid: 08964c7c-908c-4a55-bcd3-9811ad8b0b9d
Metaprogramming Elixir (2) - $15.00
Total price: $30.00

Nice work! We haven't added payments, but we can already see how our ShoppingCart and Orders context splitting is driving us towards a maintainable solution. With our cart items separated from our order line items, we are well equipped in the future to add payment transactions, cart price detection, and more.

Great work!

FAQ

Returning Ecto structures from context APIs

As we explored the context API, you might have wondered:

If one of the goals of our context is to encapsulate Ecto Repo access, why does create_user/1 return an Ecto.Changeset struct when we fail to create a user?

The answer is we've decided to expose %Ecto.Changeset{} as a public data-structure in our application. We saw before how changesets allow us to track field changes, perform validations, and generate error messages. Its use here is decoupled from the private Repo access and Ecto changeset API internals. We're exposing a data structure that the caller understands which contains the rich information like field errors. Conveniently for us, the phoenix_ecto project implements the necessary Phoenix.Param and Phoenix.HTML.FormData protocols which know how to handle %Ecto.Changeset{}'s for things like form generation and error messages. You can also think about it as being as if you had defined your own %Accounts.Changes{} struct for the same purpose and implemented the Phoenix protocols for the web-layer integration.

Strategies for cross-context workflows

Our CMS context supports lazily creating authors in the system when a user decides to publish page content. This makes sense for our use case because not all users of our system will be CMS authors. But what if our use case were for when all users of our app are indeed authors?

If we require a CMS.Author to exist every time an Accounts.User is created, we have to think carefully where to place this dependency. We know our CMS context depends on the Accounts context, but it's important to avoid cyclic dependencies across our contexts. For example, imagine we changed our Accounts.create_user function to:

def create_user(attrs) do
  %User{}
  |> User.changeset(attrs)
  |> Ecto.Changeset.cast_assoc(:credential, with: &Credential.changeset/2)
  |> Ecto.Changeset.put_assoc(:author, %Author{...})
  |> Repo.insert()
end

This may accomplish what we want, but now we need to wire up the schema relationships in the Accounts context to the CMS author. Worse, we have now taken our isolated Accounts context and required it to know about a content management system, which in turn knows about Accounts too. With time, cyclic dependencies would blurry the lines between contexts, which is not what we want for isolated responsibilities in our application. There's a better way to handle these requirements.

If you find yourself in similar situations where you feel your use case is requiring you to create circular dependencies across contexts, it's a sign you need a new context in the system to handle these application requirements. In our case, what we really want is an interface that handles all requirements when a user is created or registers in our application. To handle this, we could create a UserRegistration context, which calls into both the Accounts and CMS APIs to create a user, then associate a CMS author. Not only would this allow our Accounts to remain as isolated as possible, it gives us a clear, obvious API to handle UserRegistration needs in the system. If you take this approach, you can also use tools like Ecto.Multi to handle transactions across different context operations without deeply coupling the internal database calls. Part of our UserRegistration API could look something like this:

defmodule Hello.UserRegistration do
  alias Ecto.Multi
  alias Hello.{Accounts, CMS, Repo}

  def register_user(params) do
    Multi.new()
    |> Multi.run(:user, fn _repo, _changes_so_far -> Accounts.create_user(params) end)
    |> Multi.run(:author, fn _repo, %{user: user} ->
      {:ok, CMS.ensure_author_exists(user)}
    end)
    |> Repo.transaction()
  end
end

We can take advantage of Ecto.Multi to create a pipeline of operations that can be run inside a transaction of our Repo. If any given operation fails, the transaction will be rolled back and an error will be returned containing which operation failed, as well as the changes up to that point. In our register_user/1 example, we specified two operations, one that calls into Accounts.create_user/1 and another that passes the newly created user to CMS.ensure_author_exists/1. The final step of our function is to invoke the operations with Repo.transaction/1.

The UserRegistration setup is likely simpler to implement than the dynamic author system we built – we decided to take the harder path exactly because those are decisions developers take on their applications every day.