README.md in alba-0.13.1 vs README.md in alba-1.0.0
- old
+ new
@@ -17,12 +17,10 @@
### Easy to understand
DSL is great. It makes the coding experience natural and intuitive. However, remembering lots of DSL requires us a lot of effort. Unfortunately, most of the existing libraries have implemented their features via DSL and it's not easy to understand how they behave entirely. Alba's core DSL are only four (`attributes`, `attribute`, `one` and `many`) so it's easy to understand how to use.
-Alba is also understandable internally. The codebase is much smaller than the alternatives. In fact, it's about 330 lines of code. Look at the code on [GitHub](https://github.com/okuramasafumi/alba/tree/master/lib) and you'll be surprised how simple it is!
-
### Performance
Alba is faster than most of the alternatives. We have a [benchmark](https://github.com/okuramasafumi/alba/tree/master/benchmark).
## Installation
@@ -54,33 +52,34 @@
* Resource-based serialization
* Arbitrary attribute definition
* One and many association with the ability to define them inline
* Adding condition and filter to association
* Parameters can be injected and used in attributes and associations
-* Setting root key separately in Serializer
-* Adding metadata
+* Conditional attributes and associations
* Selectable backend
* Key transformation
+* Root key inference
+* Error handling
+* Resource name inflection based on association name
* No runtime dependencies
## Anti features
* Sorting keys
* Class level support of parameters
* Supporting all existing JSON encoder/decoder
* Cache
* [JSON:API](https://jsonapi.org) support
-* Association name inflection
* And many others
## Usage
### Configuration
Alba's configuration is fairly simple.
-#### Backend
+#### Backend configuration
Backend is the actual part serializing an object into JSON. Alba supports these backends.
* Oj, the fastest. Gem installation required.
* active_support, mostly for Rails. Gem installation required.
@@ -90,10 +89,30 @@
```ruby
Alba.backend = :oj
```
+#### Inference configuration
+
+You can enable inference feature using `enable_inference!` method.
+
+```ruby
+Alba.enable_inference!
+```
+
+You must install `ActiveSupport` to enable inference.
+
+#### Error handling configuration
+
+You can configure error handling with `on_error` method.
+
+```ruby
+Alba.on_error :ignore
+```
+
+For the details, see [Error handling section](#error-handling)
+
### Simple serialization with key
```ruby
class User
attr_accessor :id, :name, :email, :created_at, :updated_at
@@ -107,23 +126,19 @@
end
class UserResource
include Alba::Resource
+ key :user
+
attributes :id, :name
attribute :name_with_email do |resource|
"#{resource.name}: #{resource.email}"
end
end
-class SerializerWithKey
- include Alba::Serializer
-
- set key: :user
-end
-
user = User.new(1, 'Masafumi OKURA', 'masafumi@example.com')
UserResource.new(user).serialize
# => "{\"id\":1,\"name\":\"Masafumi OKURA\",\"name_with_email\":\"Masafumi OKURA: masafumi@example.com\"}"
```
@@ -179,20 +194,20 @@
### Inline definition with `Alba.serialize`
`Alba.serialize` method is a shortcut to define everything inline.
```ruby
-Alba.serialize(user, with: proc { set key: :foo }) do
+Alba.serialize(user, key: :foo) do
attributes :id
many :articles do
attributes :title, :body
end
end
# => '{"foo":{"id":1,"articles":[{"title":"Hello World!","body":"Hello World!!!"},{"title":"Super nice","body":"Really nice!"}]}}'
```
-Although this might be useful sometimes, it's generally recommended to define a class for both Resource and Serializer.
+Although this might be useful sometimes, it's generally recommended to define a class for Resource.
### Inheritance and Ignorance
You can `exclude` or `ignore` certain attributes using `ignoring`.
@@ -282,18 +297,151 @@
end
end
user = User.new(1, nil, nil)
UserResource.new(user).serialize # => '{"id":1}'
-
-
```
The key part is the use of `Proc#>>` since `Alba::Resource#converter` returns a `Proc` which contains the basic logic and it's impossible to change its behavior by just overriding the method.
It's not recommended to swap the whole conversion logic. It's recommended to always call `super` when you override `converter`.
+### Conditional attributes
+
+Filtering attributes with overriding `convert` works well for simple cases. However, It's cumbersome when we want to filter various attributes based on different conditions for keys.
+
+In these cases, conditional attributes works well. We can pass `if` option to `attributes`, `attribute`, `one` and `many`. Below is an example for the same effect as [filtering attributes section](#filtering-attributes).
+
+```ruby
+class User
+ attr_accessor :id, :name, :email, :created_at, :updated_at
+
+ def initialize(id, name, email)
+ @id = id
+ @name = name
+ @email = email
+ end
+end
+
+class UserResource
+ include Alba::Resource
+
+ attributes :id, :name, :email, if: proc { |user, attribute| !attribute.nil? }
+end
+
+user = User.new(1, nil, nil)
+UserResource.new(user).serialize # => '{"id":1}'
+```
+
+### Inference
+
+After `Alba.enable_inference!` called, Alba tries to infer root key and association resource name.
+
+```ruby
+Alba.enable_inference!
+
+class User
+ attr_reader :id
+ attr_accessor :articles
+
+ def initialize(id)
+ @id = id
+ @articles = []
+ end
+end
+
+class Article
+ attr_accessor :id, :title
+
+ def initialize(id, title)
+ @id = id
+ @title = title
+ end
+end
+
+class ArticleResource
+ include Alba::Resource
+
+ attributes :title
+end
+
+class UserResource
+ include Alba::Resource
+
+ key!
+
+ attributes :id
+
+ many :articles
+end
+
+user = User.new(1)
+user.articles << Article.new(1, 'The title')
+
+UserResource.new(user).serialize # => '{"user":{"id":1,"articles":[{"title":"The title"}]}}'
+UserResource.new([user]).serialize # => '{"users":[{"id":1,"articles":[{"title":"The title"}]}]}'
+```
+
+This resource automatically sets its root key to either "users" or "user", depending on the given object is collection or not.
+
+Also, you don't have to specify which resource class to use with `many`. Alba infers it from association name.
+
+Note that to enable this feature you must install `ActiveSupport` gem.
+
+### Error handling
+
+You can set error handler globally or per resource using `on_error`.
+
+```ruby
+class User
+ attr_accessor :id, :name
+
+ def initialize(id, name, email)
+ @id = id
+ @name = name
+ @email = email
+ end
+
+ def email
+ raise RuntimeError, 'Error!'
+ end
+end
+
+class UserResource
+ include Alba::Resource
+
+ attributes :id, :name, :email
+
+ on_error :ignore
+end
+
+user = User.new(1, 'Test', 'email@example.com')
+UserResource.new(user).serialize # => '{"id":1,"name":"Test"}'
+```
+
+This way you can exclude an entry when fetching an attribute gives an exception.
+
+There are four possible arguments `on_error` method accepts.
+
+* `:raise` re-raises an error. This is the default behavior.
+* `:ignore` ignores the entry with the error.
+* `:nullify` sets the attribute with the error to `nil`.
+* Block gives you more control over what to be returned.
+
+The block receives five arguments, `error`, `object`, `key`, `attribute` and `resource class` and must return a two-element array. Below is an example.
+
+```ruby
+# Global error handling
+Alba.on_error do |error, object, key, attribute, resource_class|
+ if resource_class == MyResource
+ ['error_fallback', object.error_fallback]
+ else
+ [key, error.message]
+ end
+end
+```
+
## Comparison
Alba is faster than alternatives.
For a performance benchmark, see https://gist.github.com/okuramasafumi/4e375525bd3a28e4ca812d2a3b3e5829.
@@ -306,21 +454,9 @@
```
## Why named "Alba"?
The name "Alba" comes from "albatross", a kind of birds. In Japanese, this bird is called "Aho-dori", which means "stupid bird". I find it funny because in fact albatrosses fly really fast. I hope Alba looks stupid but in fact it does its job quick.
-
-## Alba internals
-
-Alba has three component, `Serializer`, `Resource` and `Value` (`Value` is conceptual and not implemented directly).
-
-`Serializer` is a component responsible for rendering JSON output with `Resource`. `Serializer` can add more data to `Resource` such as `metadata`. Users can define one single `Serializer` and reuse it for all `Resource`s. The main interface is `#serialize`.
-
-`Resource` is a component responsible for defining how an object (or a collection of objects) is converted into JSON. The difference between `Serializer` and `Resource` is that while `Serializer` can add arbitrary data into JSON, `Resource` can get data only from the object under it. The main interface is `#serializable_hash`.
-
-`One` and `Many` are the special object fetching other resources and converting them into Hash.
-
-The main `Alba` module holds config values and one convenience method, `.serialize`.
## Development
After checking out the repo, run `bin/setup` to install dependencies. Then, run `rake test` to run the tests. You can also run `bin/console` for an interactive prompt that will allow you to experiment.