dry-validation
Data validation library based on predicate logic and rule composition.
Overview
Unlike other, well known, validation solutions in Ruby, dry-validation
takes
a different approach and focuses a lot on explicitness, clarity and preciseness
of validation logic. It is designed to work with any data input, whether it's a
simple hash, an array or a complex object with deeply nested data.
It is based on a simple idea that each validation is encapsulated by a simple,
stateless predicate, that receives some input and returns either true
or false
.
Those predicates are encapsulated by rules
which can be composed together using
predicate logic
. This means you can use the common logic operators to build up
a validation schema
.
It's very explicit, powerful and extendible.
Validations can be described with great precision, dry-validation
eliminates
ambigious concepts like presence
validation where we can't really say whether
some attribute or key is missing or it's just that the value is nil
.
There's also the concept of type-safety, completely missing in other validation libraries, which is quite important and useful. It means you can compose a validation that does rely on the type of a given value. In example it makes no sense to validate each element of an array when it turns out to be an empty string.
The DSL
The core of dry-validation
is rules composition and predicate logic. The DSL
is a simple front-end for that. It only allows you to define the rules by using
predicate identifiers. There are no magical options, conditionals and custom
validation blocks known from other libraries. The focus is on pure validation
logic.
Examples
Basic
Here's a basic example where we validate following things:
- The input must have a key called
:email
- Provided the email key is present, its value must be filled
- The input must have a key called
:age
- Provided the age key is present, its value must be an integer and it must be greater than 18
This can be easily expressed through the DSL:
require 'dry-validation'
class Schema < Dry::Validation::Schema
key(:email) { |email| email.filled? }
key(:age) do |age|
age.int? & age.gt?(18)
end
end
schema = Schema.new
errors = schema.(email: '[email protected]', age: 19)
puts errors.inspect
# []
errors = schema.(email: nil, age: 19)
puts errors.inspect
# [[:email, ["email must be filled"]]]
A couple of remarks:
key
assumes that we want to use the:key?
predicate to check the existance of that keyage.gt?(18)
translates to calling a predicate like this:schema[:gt?].(18, age)
age.int? & age.gt?(18)
is a conjunction, so we don't bother aboutgt?
unlessint?
returnstrue
- You can also use
|
for disjunction - Schema object does not carry the input as its state, nor does it know how to access the input values, we
pass the input to
call
and get error set as the response
Nested Hash
We are free to define validations for anything, including deeply nested structures:
require 'dry-validation'
class Schema < Dry::Validation::Schema
key(:address) do |address|
address.key(:city) do |city|
city.min_size?(3)
end
address.key(:street) do |street|
street.filled?
end
address.key(:country) do |country|
country.key(:name, &:filled?)
country.key(:code, &:filled?)
end
end
end
schema = Schema.new
errors = schema.({})
puts errors.inspect
# [[:address, ["address is missing"]]]
errors = schema.(address: { city: 'NYC' })
puts errors.inspect
# [[:address, [[:street, ["street is missing"]], [:country, ["country is missing"]]]]]
Defining Custom Predicates
You can simply define predicate methods on your schema object:
class Schema < Dry::Validation::Schema
key(:email) { |value| value.str? & value.email? }
def email?(value)
! /magical-regex-that-matches-emails/.match(value).nil?
end
end
You can also re-use a predicate container across multiple schemas:
module MyPredicates
include Dry::Validation::Predicates
predicate(:email?) do |input|
! /magical-regex-that-matches-emails/.match(value).nil?
end
end
class Schema < Dry::Validation::Schema
configure do |config|
config.predicates = MyPredicates
end
key(:email) { |value| value.str? & value.email? }
end
List of Built-In Predicates
empty?
eql?
exclusion?
filled?
format?
gt?
gteq?
inclusion?
int?
key?
lt?
lteq?
max_size?
min_size?
nil?
size?
str?
Error Messages
By default dry-validation
comes with a set of pre-defined error messages for
every built-in predicate. They are defined in a yaml file
which is shipped with the gem.
You can provide your own messages and configure your schemas to use it like that:
class Schema < Dry::Validation::Schema
configure { |config| config. = '/path/to/my/errors.yml' }
end
You can also provide a namespace per-schema that will be used by default:
class Schema < Dry::Validation::Schema
configure { |config| config.namespace = :user }
end
Lookup rules:
filled?: "%{name} must be filled"
attributes:
email:
filled?: "the email is missing"
user:
filled?: "%{name} name cannot be blank"
attributes:
address:
filled?: "You gotta tell us where you live"
Given the yaml file above, messages lookup works as follows:
= Dry::Validation::Messages.load('/path/to/our/errors.yml')
.lookup(:filled?, :age) # => "age must be filled"
.lookup(:filled?, :address) # => "address must be filled"
.lookup(:filled?, :email) # => "the email is missing"
# with namespaced messages
= .namespaced(:user)
.lookup(:filled?, :age) # "age cannot be blank"
.lookup(:filled?, :address) # "You gotta tell us where you live"
By configuring messages_file
and/or namespace
in a schema, default messages
are going to be automatically merged with your overrides and/or namespaced.
I18n Integration
Coming (very) soon...
Rule AST
Internally, dry-validation
uses a simple AST representation of rules and errors
to produce rule objects and error messages. If you would like to programatically
generate rules, it is a very simple process:
ast = [
[
:and,
[
[:key, [:age, [:predicate, [:key?, []]]]],
[
:and,
[
[:val, [:age, [:predicate, [:filled?, []]]]],
[:val, [:age, [:predicate, [:gt?, [18]]]]]
]
]
]
]
]
compiler = Dry::Validation::RuleCompiler.new(Dry::Validation::Predicates)
# compile an array of rule objects
rules = compiler.call(ast)
puts rules.inspect
# [
# #<Dry::Validation::Rule::Conjunction
# left=#<Dry::Validation::Rule::Key name=:age predicate=#<Dry::Validation::Predicate id=:key?>>
# right=#<Dry::Validation::Rule::Conjunction
# left=#<Dry::Validation::Rule::Value name=:age predicate=#<Dry::Validation::Predicate id=:filled?>>
# right=#<Dry::Validation::Rule::Value name=:age predicate=#<Dry::Validation::Predicate id=:gt?>>>>
# ]
# dump it back to ast
puts rules.map(&:to_ary).inspect
# [[:and, [:key, [:age, [:predicate, [:key?, [:age]]]]], [[:and, [:val, [:age, [:predicate, [:filled?, []]]]], [[:val, [:age, [:predicate, [:gt?, [18]]]]]]]]]]
Complete docs for the AST format are coming soon, for now please refer to this spec.
Status and Roadmap
This library is in a very early stage of development but you are encauraged to try it out and provide feedback.
For planned features check out the issues.
License
See LICENSE
file.