defmodule Contexted do @moduledoc ~S""" Contexted helps you structure Phoenix contexts by enforcing separation, allowing modularization, and generating common CRUD operations. ## Features - `Contexted.Tracer` - trace and enforce definite separation between specific context modules. - `Contexted.Delegator` - divide the big context module into smaller parts and use delegations to build the final context. - `Contexted.CRUD` - auto-generate the most common CRUD operations whenever needed.
## Installation Add the following to your `mix.exs` file: ```elixir defp deps do [ {:contexted, "~> 0.3.4"} ] end ``` Then run `mix deps.get`.
## Step by step overview To describe a sample usage of this library, let's assume that your project has three contexts: - `Account` - `Subscription` - `Blog` Our goal, as the project grows, is to: 1. Keep contexts separate and not create any cross-references. For this to work, we'll raise errors during compilation whenever such a cross-reference happens. 2. Divide each context into smaller parts so that it is easier to maintain. In this case, we'll refer to each of these parts as **Subcontext**. It's not a new term added to the Phoenix framework but rather a term proposed to emphasize that it's a subset of Context. For this to work, we'll use delegates. 3. Not repeat ourselves with common business logic operations. For this to work, we'll be using CRUD functions generator, since these are the most common.
### Keep contexts separate It's very easy to monitor cross-references between context modules with the `contexted` library. First, add `contexted` as one of the compilers in _mix.exs_: ```elixir def project do [ ... compilers: [:contexted] ++ Mix.compilers(), ... ] end ``` Next, define a list of contexts available in the app inside config file: ```elixir config :contexted, contexts: [ # list of context modules goes here, for instance: # [App.Account, App.Subscription, App.Blog] ] ``` And that's it. From now on, whenever you will cross-reference one context with another, you will see an error raised during compilation. Here is an example of such an error: ``` == Compilation error in file lib/app/accounts.ex == ** (RuntimeError) You can't reference App.Blog context within App.Accounts context. ``` See more in [Contexted.Tracer](https://hexdocs.pm/contexted/Contexted.Tracer.html).
#### Exclude files and folders from cross-references context check In special cases, you may need to exclude certain folders or files from cross-reference checks due to project structure or naming conventions. To do this, add a list of exclusions in config `exclude_paths` option: ```elixir config :contexted, exclude_paths: ["app/test"] ```
### Dividing each context into smaller parts To divide big Context into smaller Subcontexts, we can use `delegate_all/1` macro from `Contexted.Delegator` module. Let's assume that the `Account` context has `User`, `UserToken` and `Admin` resources. Here is how we can split the context module: ```elixir # Users subcontext defmodule App.Account.Users do def get_user(id) do ... end end # UserTokens subcontext defmodule App.Account.UserTokens do def get_user_token(id) do ... end end # Admins subcontext defmodule App.Account.Admins do def get_admin(id) do ... end end # Account context defmodule App.Account do import Contexted.Delegator delegate_all App.Account.Users delegate_all App.Account.UserTokens delegate_all App.Account.Admins end ``` From now on, you can treat the `Account` context module as the API for the "outside" world. Instead of calling: ```elixir App.Account.Users.find_user(1) ``` You will simply do: ```elixir App.Account.find_user(1) ```
#### Being able to access docs and specs in auto-delegated functions Both docs and specs are attached as metadata of module once it's compiled and saved as `.beam`. In reference to the example of `App.Account` context, it's possible that `App.Account.Users` will not be saved in `.beam` file before the `delegate_all` macro is executed. Therefore, first, all of the modules have to be compiled, and saved to `.beam` and only then we can create `@doc` and `@spec` of each delegated function. As a workaround, in `Contexted.Tracer.after_compiler/1` all of the contexts `.beam` files are first deleted and then recompiled. This is an opt-in functionality, as it extends compilation time. If you want to enable it, set the following config values: ```elixir config :contexted, app: :your_app_name, # replace 'your_app_name' with your real app name enable_recompilation: true ``` You may also want to enable it only for certain environments, like `dev`. *Please also note that when this functionality is enabled, during the recompilation process, warnings are temporarily silenced to avoid logging conflict warnings. It will still log warnings as intended, during the first compilation, therefore it won't have any affect on normal compilation flow.* See more in [Contexted.Delegator](https://hexdocs.pm/contexted/Contexted.Delegator.html).
### Don't repeat yourself with CRUD operations In most web apps CRUD operations are very common. Most of these, have the same pattern. Why not autogenerate them? Here is how you can generate common CRUD operations for `App.Account.Users`: ```elixir defmodule App.Account.Users do use Contexted.CRUD, repo: App.Repo, schema: App.Accounts.User end ``` This will generate the following functions: ``` [ change_user: 1, change_user: 2, create_user: 0, create_user: 1, create_user!: 0, create_user!: 1, delete_user: 1, delete_user!: 1, get_user: 1, get_user!: 1, list_users: 0, update_user: 1, update_user: 2, update_user!: 1, update_user!: 2 ] ``` See more in [Contexted.CRUD](https://hexdocs.pm/contexted/Contexted.CRUD.html). """ end