Config helps you easily manage environment specific settings in an easy and usable manner.
- simple YAML config files
- config files support ERB
- config files support inheritance and multiple environments
- access config information via convenient object member notation
- support for multi-level settings (
Settings.group.subgroup.setting
) - local developer settings ignored when committing the code
Current version supports and is tested for the following interpreters and frameworks:
- Interpreters
- Ruby
>= 2.4
- JRuby
>= 9.2
- TruffleRuby
>= 19.3
- Ruby
- Application frameworks
- Rails
>= 4.2
,5
and6
- Padrino
- Sinatra
- Rails
For Ruby 2.0
to 2.3
or Rails 3
to 4.1
use version 1.x
of this gem. For older versions of Rails or Ruby use AppConfig.
Add gem 'config'
to your Gemfile
and run bundle install
to install it. Then run
rails g config:install
which will generate customizable config file config/initializers/config.rb
and set of default settings files:
config/settings.yml
config/settings.local.yml
config/settings/development.yml
config/settings/production.yml
config/settings/test.yml
You can now edit them to adjust to your needs.
Add the gem to your Gemfile
and run bundle install
to install it. Then edit app.rb
and register Config
register Config
Add the gem to your Gemfile
and run bundle install
to install it. Afterwards in need to register Config
in your app and give it a root so it can find the config files.
set :root, File.dirname(__FILE__)
register Config
Add the gem to your Gemfile
and run bundle install
to install it. Then initialize Config
manually within your configure block.
Config.load_and_set_settings(Config.setting_files("/path/to/config_root", "your_project_environment"))
It's also possible to initialize Config
manually within your configure block if you want to just give it some yml paths to load from.
Config.load_and_set_settings("/path/to/yaml1", "/path/to/yaml2", ...)
After installing the gem, Settings
object will become available globally and by default will be compiled from the files listed below. Settings defined in files that are lower in the list override settings higher.
config/settings.yml
config/settings/#{environment}.yml
config/environments/#{environment}.yml
config/settings.local.yml
config/settings/#{environment}.local.yml
config/environments/#{environment}.local.yml
Entries can be accessed via object member notation:
Settings.my_config_entry
Nested entries are supported:
Settings.my_section.some_entry
Alternatively, you can also use the []
operator if you don't know which exact setting you need to access ahead of time.
# All the following are equivalent to Settings.my_section.some_entry
Settings.my_section[:some_entry]
Settings.my_section['some_entry']
Settings[:my_section][:some_entry]
You can reload the Settings object at any time by running Settings.reload!
.
You can also reload the Settings
object from different config files at runtime.
For example, in your tests if you want to test the production settings, you can:
Rails.env = "production"
Settings.reload_from_files(
Rails.root.join("config", "settings.yml").to_s,
Rails.root.join("config", "settings", "#{Rails.env}.yml").to_s,
Rails.root.join("config", "environments", "#{Rails.env}.yml").to_s
)
You can have environment specific config files. Environment specific config entries take precedence over common config entries.
Example development environment config file:
#{Rails.root}/config/environments/development.yml
Example production environment config file:
#{Rails.root}/config/environments/production.yml
If you want to have local settings, specific to your machine or development environment, you can use the following files, which are automatically .gitignore
:
Rails.root.join("config", "settings.local.yml").to_s,
Rails.root.join("config", "settings", "#{Rails.env}.local.yml").to_s,
Rails.root.join("config", "environments", "#{Rails.env}.local.yml").to_s
NOTE: The file settings.local.yml
will not be loaded in tests to prevent local configuration from causing flaky or non-deterministic tests. Environment-specific files (e.g. settings/test.local.yml
) will still be loaded to allow test-specific credentials.
You can add new YAML config files at runtime. Just use:
Settings.add_source!("/path/to/source.yml")
Settings.reload!
This will use the given source.yml file and use its settings to overwrite any previous ones.
On the other hand, you can prepend a YML file to the list of configuration files:
Settings.prepend_source!("/path/to/source.yml")
Settings.reload!
This will do the same as add_source
, but the given YML file will be loaded first (instead of last) and its settings will be overwritten by any other configuration file. This is especially useful if you want to define defaults.
One thing I like to do for my Rails projects is provide a local.yml config file that is .gitignored (so its independent per developer). Then I create a new initializer in config/initializers/add_local_config.rb
with the contents
Settings.add_source!("#{Rails.root}/config/settings/local.yml")
Settings.reload!
Note: this is an example usage, it is easier to just use the default local files
settings.local.yml
,settings/#{Rails.env}.local.yml
andenvironments/#{Rails.env}.local.yml
for your developer specific settings.
You also have the option to add a raw hash as a source. One use case might be storing settings in the database or in environment variables that overwrite what is in the YML files.
Settings.add_source!({some_secret: ENV['some_secret']})
Settings.reload!
You may pass a hash to prepend_source!
as well.
Embedded Ruby is allowed in the configuration files. Consider the two following config files.
#{Rails.root}/config/settings.yml
size: 1
server: google.com
#{Rails.root}/config/environments/development.yml
size: 2
computed: <%= 1 + 2 + 3 %>
section:
size: 3
servers: [ {name: yahoo.com}, {name: amazon.com} ]
Notice that the environment specific config entries overwrite the common entries.
Settings.size # => 2
Settings.server # => google.com
Notice the embedded Ruby.
Settings.computed # => 6
Notice that object member notation is maintained even in nested entries.
Settings.section.size # => 3
Notice array notation and object member notation is maintained.
Settings.section.servers[0].name # => yahoo.com
Settings.section.servers[1].name # => amazon.com
There are multiple configuration options available, however you can customize Config
only once, preferably during application initialization phase:
Config.setup do |config|
config.const_name = 'Settings'
# ...
end
After installing Config
in Rails, you will find automatically generated file that contains default configuration located at config/initializers/config.rb
.
const_name
- name of the object holing you settings. Default:'Settings'
overwrite_arrays
- overwrite arrays found in previously loaded settings file. Default:true
merge_hash_arrays
- merge hashes inside of arrays from previously loaded settings files. Makes sense only whenoverwrite_arrays = false
. Default:false
knockout_prefix
- ability to remove elements of the array set in earlier loaded settings file. Makes sense only whenoverwrite_arrays = false
, otherwise array settings would be overwritten by default. Default:nil
merge_nil_values
-nil
values will overwrite an existing value when merging configs. Default:true
.
# merge_nil_values is true by default
c = Config.load_files("./spec/fixtures/development.yml") # => #<Config::Options size=2, ...>
c.size # => 2
c.merge!(size: nil) => #<Config::Options size=nil, ...>
c.size # => nil
# To reject nil values when merging settings:
Config.setup do |config|
config.merge_nil_values = false
end
c = Config.load_files("./spec/fixtures/development.yml") # => #<Config::Options size=2, ...>
c.size # => 2
c.merge!(size: nil) => #<Config::Options size=nil, ...>
c.size # => 2
Check Deep Merge for more details.
With Ruby 2.1 or newer, you can optionally define a schema or contract (added in config-2.1
) using dry-rb to validate presence (and type) of specific config values. Generally speaking contracts allow to describe more complex validations with depencecies between fields.
If you provide either validation option (or both) it will automatically be used to validate your config. If validation fails it will raise a Config::Validation::Error
containing information about all the mismatches between the schema and your config.
Both examples below demonstrates how to ensure that the configuration has an optional email
and the youtube
structure with the api_key
field filled. The contract adds an additional rule.
Leverage dry-validation, you can create a contract with a params schema and rules:
class ConfigContract < Dry::Validation::Contract
params do
optional(:email).maybe(:str?)
required(:youtube).schema do
required(:api_key).filled
end
end
rule(:email) do
unless /\A[\w+\-.]+@[a-z\d\-]+(\.[a-z\d\-]+)*\.[a-z]+\z/i.match?(value)
key.failure('has invalid format')
end
end
end
Config.setup do |config|
config.validation_contract = ConfigContract.new
end
The above example adds a rule to ensure the email
is valid by matching it against the provided regular expression.
Check dry-validation for more details.
You may also specify a schema using dry-schema:
Config.setup do |config|
# ...
config.schema do
optional(:email).maybe(:str?)
required(:youtube).schema do
required(:api_key).filled
end
end
end
Check dry-schema for more details.
For an example settings file:
size: 1
server: google.com
You can test if a value was set for a given key using key?
and its alias has_key?
:
Settings.key?(:path)
# => false
Settings.key?(:server)
# => true
By default, accessing to a missing key returns nil
:
Settings.key?(:path)
# => false
Settings.path
# => nil
This is not "typo-safe". To solve this problem, you can configure the fail_on_missing
option:
Config.setup do |config|
config.fail_on_missing = true
# ...
end
So it will raise a KeyError
when accessing a non-existing key (similar to Hash#fetch
behaviour):
Settings.path
# => raises KeyError: key not found: :path
See section below for more details.
To load environment variables from the ENV
object, that will override any settings defined in files, set the use_env
to true in your config/initializers/config.rb
file:
Config.setup do |config|
config.const_name = 'Settings'
config.use_env = true
end
Now config would read values from the ENV object to the settings. For the example above it would look for keys starting with Settings
:
ENV['Settings.section.size'] = 1
ENV['Settings.section.server'] = 'google.com'
It won't work with arrays, though.
It is considered an error to use environment variables to simutaneously assign a "flat" value and a multi-level value to a key.
# Raises an error when settings are loaded
ENV['BACKEND_DATABASE'] = 'development'
ENV['BACKEND_DATABASE_USER'] = 'postgres'
Instead, specify keys of equal depth in the environment variable names:
ENV['BACKEND_DATABASE_NAME'] = 'development'
ENV['BACKEND_DATABASE_USER'] = 'postgres'
Heroku uses ENV object to store sensitive settings. You cannot upload such files to Heroku because it's ephemeral filesystem gets recreated from the git sources on each instance refresh. To use config with Heroku just set the use_env
var to true
as mentioned above.
To upload your local values to Heroku you could ran bundle exec rake config:heroku
.
You can customize how environment variables are processed:
env_prefix
(default:const_name
) - load only ENV variables starting with this prefix (case-sensitive)env_separator
(default:'.'
) - what string to use as level separator - default value of.
works well with Heroku, but you might want to change it for example for__
to easy override settings from command line, where using dots in variable names might not be allowed (eg. Bash)env_converter
(default::downcase
) - how to process variables names:nil
- no change:downcase
- convert to lower case
env_parse_values
(default:true
) - try to parse values to a correct type (Boolean
,Integer
,Float
,String
)
For instance, given the following environment:
SETTINGS__SECTION__SERVER_SIZE=1
SETTINGS__SECTION__SERVER=google.com
SETTINGS__SECTION__SSL_ENABLED=false
And the following configuration:
Config.setup do |config|
config.use_env = true
config.env_prefix = 'SETTINGS'
config.env_separator = '__'
config.env_converter = :downcase
config.env_parse_values = true
end
The following settings will be available:
Settings.section.server_size # => 1
Settings.section.server # => 'google.com'
Settings.section.ssl_enabled # => false
It is possible to parse variables stored in an AWS Secrets Manager Secret as if they were environment variables by using Config::Sources::EnvSource
.
For example, the plaintext secret might look like this:
{
"Settings.foo": "hello",
"Settings.bar": "world",
}
In order to load those settings, fetch the settings from AWS Secrets Manager, parse the plaintext as JSON, pass the resulting Hash
into a new EnvSource
, load the new source, and reload.
# fetch secrets from AWS
client = Aws::SecretsManager::Client.new
response = client.get_secret_value(secret_id: "#{ENV['ENVIRONMENT']}/my_application")
secrets = JSON.parse(response.secret_string)
# load secrets into config
secret_source = Config::Sources::EnvSource.new(secrets)
Settings.add_source!(secret_source)
Settings.reload!
In this case, the following settings will be available:
Settings.foo # => "hello"
Settings.bar # => "world"
By default, EnvSource
will use configuration for env_prefix
, env_separator
, env_converter
, and env_parse_values
, but any of these can be overridden in the constructor.
secret_source = Config::Sources::EnvSource.new(secrets,
prefix: 'MyConfig',
separator: '__',
converter: nil,
parse_values: false)
You are very warmly welcome to help. Please follow our contribution guidelines
Any and all contributions offered in any form, past present or future are understood to be in complete agreement and acceptance with MIT license.
- Piotr Kuczynski
- Fred Wu
- Jacques Crocker
- Inherited from AppConfig by Christopher J. Bottaro
This project exists thanks to all the people who contribute and you are very warmly welcome to help. Please follow our contribution guidelines.
Any and all contributions offered in any form, past present or future are understood to be in complete agreement and acceptance with the MIT license.
Become a backer and support us with a small monthly donation to help us continue our activities. Thank you if you already one! 🙏
Support this project by becoming a sponsor. Your logo will show up here with a link to your website.
Copyright (C) Piotr Kuczynski. Released under the MIT License.