PureConfig

A boilerplate-free Scala library for loading configuration files.

Why

Loading configurations has always been a tedious and error-prone procedure. A common way to do it consists in writing code to deserialize each fields of the configuration. The more fields there are, the more code must be written (and tested and maintained...) and this must be replicated for each project.

This kind of code is boilerplate because most of the times the code can be automatically generated by the compiler based on what must be loaded. For instance, if you are going to load an Int for a field named foo, then probably you want some code that gets the values associated with the key foo in the configuration and assigns it to the proper field after converting it to Int.

The goal of this library is to create at compile-time the boilerplate necessary to load a configuration of a certain type. In other words, you define what to load and PureConfig provides how to load it.

Not yet another configuration library

PureConfig is not a configuration library in the sense that it doesn't search for files or parse them. It can be seen as a better front-end for the existing libraries. It uses the Typesafe Config library for loading raw configurations and then uses the raw configurations to do its magic.

Add PureConfig to your project

In the sbt configuration file use Scala 2.10, 2.11 or 2.12:

scalaVersion := "2.12.1" // or "2.11.8", "2.10.5"

Add PureConfig to your library dependencies. For Scala 2.11 and 2.12:

libraryDependencies ++= Seq(
  "com.github.pureconfig" %% "pureconfig" % "0.7.2"
)

For Scala 2.10 you need also the Macro Paradise plugin:

libraryDependencies ++= Seq(
  "com.github.pureconfig" %% "pureconfig" % "0.7.2",
  compilerPlugin("org.scalamacros" % "paradise" % "2.0.1" cross CrossVersion.patch)
)

For a full example of build.sbt you can have a look at this build.sbt used for the example.

Use PureConfig

Import the library package and use one of the loadConfig methods:

import pureconfig._
import pureconfig.error.ConfigReaderFailures

case class YourConfClass(name: String, quantity: Int)

val config: Either[pureconfig.error.ConfigReaderFailures,YourConfClass] = loadConfig[YourConfClass]

Supported Types

Currently supported types for fields are:

String, Boolean, Double (standard and percentage format ending with %), Float (also supporting percentage), Int, Long, Short, URL, URI, Duration, FiniteDuration;
java.lang.Enum
all collections implementing the TraversableOnce trait where the type of the elements is in this list;
Option for optional values, i.e. values that can or cannot be in the configuration;
Map with String keys and any value type that is in this list;
everything in java.time (must be configured first - see Configurable converters);
java.io.File;
java.util.UUID;
java.nio.file.Path;
java.util.regex.Pattern and scala.util.matching.Regex;
Typesafe ConfigValue, ConfigObject and ConfigList;
value classes for which readers and writers of the inner type are used;
case classes;
sealed families of case classes (ADTs).

Example

First, import the library, define data types, and a case class to hold the configuration:

import com.typesafe.config.ConfigFactory.parseString
import pureconfig.loadConfig

sealed trait MyAdt
case class AdtA(a: String) extends MyAdt
case class AdtB(b: Int) extends MyAdt
final case class Port(value: Int) extends AnyVal
case class MyClass(
  boolean: Boolean,
  port: Port,
  adt: MyAdt,
  list: List[Double],
  map: Map[String, String],
  option: Option[String])

Then, load the configuration (in this case from a hard-coded string):

val conf = parseString("""{ 
  "boolean": true,
  "port": 8080, 
  "adt": { 
    "type": "adtb", 
    "b": 1 
  }, 
  "list": ["1", "20%"], 
  "map": { "key": "value" } 
}""")
// conf: com.typesafe.config.Config = Config(SimpleConfigObject({"adt":{"b":1,"type":"adtb"},"boolean":true,"list":["1","20%"],"map":{"key":"value"},"port":8080}))

loadConfig[MyClass](conf)
// res3: Either[pureconfig.error.ConfigReaderFailures,MyClass] = Right(MyClass(true,Port(8080),AdtB(1),List(1.0, 0.2),Map(key -> value),None))

Integrating with other libraries

Internal Modules

The core of PureConfig eschews unnecessary dependencies. Separate modules exist to support types which are not part of the standard Scala and Java libraries.

pureconfig-enum provides converters for enums generated by julienrf's enum library.
pureconfig-enumeratum provides converters for enums generated by Enumeratum.
pureconfig-javax provides converters for value classes in the javax packages.
pureconfig-joda provides configurable converters for Joda Time types.
pureconfig-squants provides converters for Squants's beautiful types representing units of measure.

External Integrations

A non-comprehensive list of other libraries which have integrated with PureConfig to provide a richer experience include:

refined-pureconfig allows PureConfig to play nicely with refined's type refinements. Viktor Lövgren's blog post gleefully explains how PureConfig and refined work together.

FAQ

How can I use PureConfig with Spark 2.1.0 (problematic Shapeless dependency)?

Apache Spark (specifically version 2.1.0) has a transitive dependency on Shapeless 2.0.0. This version is too old to be used by PureConfig, making your Spark project fail when using spark-submit.

If you are using the sbt-assembly plugin to create your JARs you can shade this dependency by adding

assemblyShadeRules in assembly := Seq(ShadeRule.rename("shapeless.**" -> "new_shapeless.@1").inAll)

to your assembly.sbt file.

Contribute

PureConfig is a free library developed by several people around the world. Contributions are welcomed and encouraged. If you want to contribute, we suggest to have a look at the available issues and to talk with us on the pureconfig gitter channel.

If you'd like to add support for types which are not part the standard Java or Scala libraries, please consider submitting a pull request to create a module. Pull Request #108 created a very simple module. It should provide a good template for the pieces you'll need to add.

The steps to create a new module, called nexttopmod, are:

Define a new project in the root build.sbt. There are other examples near the top of the file.
Create a new modules/nexttopmod/ subdirectory.
Add a modules/nexttopmod/build.sbt defining the module's name and special dependencies.
Implement converters. Typically they're in a package object in modules/nexttopmod/src/main/scala/pureconfig/module/nexttopmod/package.scala.
Test the converters. Usually tests would be in modules/nexttopmod/src/test/scala/pureconfig/module/nexttopmod/NextTopModSuite.scala.
Optionally explain a little bit about how it works in modules/nexttopmod/README.md.

PureConfig supports the Typelevel code of conduct and wants all of its channels (Gitter, GitHub, etc.) to be welcoming environments for everyone.

License

Mozilla Public License, version 2.0

Special Thanks

To the Shapeless and to the Typesafe Config developers.

er-rishi/pureconfig