Skip to content

A type safe, purely functional configuration library for Scala

License

Notifications You must be signed in to change notification settings

lambdista/config

Repository files navigation

Build Status codecov.io Javadocs Maven Central Latest version

config: a type safe, purely functional configuration library for Scala

Table of Contents

Not only another Typesafe's config wrapper

Right from the start I didn't want to depend on other config libraries when I started implementing this one so I wrote my own parser for a simple JSONish syntax. One of the advantages in using your own parser is you can add other custom types. For example this lib allows you to define a Scala Range while HOCON doesn't. Hence, this is not only another Typesafe's config wrapper. However, if you are already using Typesafe's config library and/or just prefer HOCON syntax for your configuration, there's an adapter that will convert a Typesafe Config object into this config's AST. See this example.

Configuration Syntax

The syntax expected by this library is a JSON-superset. This means that any JSON file should be a valid configuration. However, the null JSON values can only be converted to Option[A], where A is the type you expect because, of course, we don't fancy null in Scala code. The superset part means that:

  • You can optionally use = instead of :
  • You can use comments: you can start a comment line using both // and #
  • You can avoid putting the keys between quotes, unless your key contains white spaces
  • You can use a Scala Duration
  • You can use a Scala Range

Usage

As a first step you need to add the dependency to your build file:

libraryDependencies += "com.lambdista" %% "config" % "0.8.1"

Only Scala 2.13.x versions are supported.

Ok, let's see the typical usage scenarios. As a use case consider the following configuration, unless otherwise specified:

// comment 1
{
  bar = "hello",
  # comment 2
  baz = 42,
  list = [1, // comment 3
          2, 3],
  map = {
    alpha = "hello",
    beta = 42
  },
  mapList = [
    {
      alpha = "hello",
      beta = 42
    },
    {
      alpha = "world",
      beta = 24
    }
  ],
  range = 2 to 10 by 2,
  duration = 5 seconds
}

Suppose the previous configuration is at the relative path: core/src/test/resources/foo.conf.

First thing first, load and parse your config:

import scala.concurrent.duration.Duration

import java.nio.file.Paths

import com.lambdista.config._


val confPath = "core/src/test/resources/foo.conf"
// confPath: String = "core/src/test/resources/foo.conf"
val config: Result[Config] = Config.from(Paths.get(confPath))
// config: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = HashMap(
//         "duration" -> AbstractDuration(value = 5 seconds),
//         "range" -> AbstractRange(value = Range(2, 4, 6, 8, 10)),
//         "bar" -> AbstractString(value = "hello"),
//         "mapList" -> AbstractList(
//           values = List(
//             AbstractMap(
//               value = Map(
//                 "alpha" -> AbstractString(value = "hello"),
//                 "beta" -> AbstractNumber(value = 42.0)
//               )
//             ),
//             AbstractMap(
//               value = Map(
//                 "alpha" -> AbstractString(value = "world"),
//                 "beta" -> AbstractNumber(value = 24.0)
//               )
//             )
//           )
//         ),
//         "baz" -> AbstractNumber(value = 42.0),
//         "list" -> AbstractList(
//           values = List(
//             AbstractNumber(value = 1.0),
//             AbstractNumber(value = 2.0),
//             AbstractNumber(value = 3.0)
//           )
//         )
//       )
//     )
//   )
// )

Apart from java.nio.file.Path you can load your config from other resources using Config Loaders.

As you can see the result is a Result[Config] (where Result[A] is just a type alias for Either[Error, A]). Indeed you can get two types of error here:

  • The resource cannot be found.
  • The resource can be found but its parsing failed.

In both cases you would get a Left wrapping the appropriate Error (a subclass of Exception).

Once you have a Config object you can do two main things with it:

  • Convert it entirely into a case class representing the whole configuration.
  • Retrieve a single value and convert it to whatever it's convertible to.

Automatic conversion to case class

Here's how you would map the previous configuration to a case class (config is the value from the previous example):

case class Greek(alpha: String, beta: Int)


case class FooConfig(
    bar: String, 
    baz: Option[Int], 
    list: List[Int],
    map: Greek,
    mapList: List[Greek], 
    range: Range, 
    duration: Duration,
    missingValue: Option[String]
)


val fooConfig: Result[FooConfig] = for {
  conf <- config
  result <- conf.as[FooConfig]
} yield result
// fooConfig: Result[FooConfig] = Left(
//   value = com.lambdista.config.ConversionError: Could not convert {duration = 5 seconds, range = [2.0, 4.0, 6.0, 8.0, 10.0], bar = "hello", mapList = [{alpha = "hello", beta = 42.0}, {alpha = "world", beta = 24.0}], baz = 42.0, list = [1.0, 2.0, 3.0]} to the type requested
// )

The value of fooConfig will be:

Right(FooConfig(hello,Some(42),List(1, 2, 3),Greek(hello,42),List(Greek(hello,42), Greek(world,24)),Range(2, 4, 6, 8, 10),5 seconds))

Here you can already notice some interesting features of this library:

  • The conversion to a case class happens automatically, no boilerplate on the client side is required.
  • Since baz is declared as Option[Int] the library automatically wraps the Int value into a Some.
  • By the way, note also how missingValue is not present in the config but since it's declared as Option in the case class its value becomes None.
  • The automatic conversion works also for nested structures, see mapList for example.
  • Range and Duration work like a charm. Note that for both Range and Duration you can use the syntax you would use in regular Scala code. For example, you could have used 5 secs instead of 5 seconds in foo.conf and it would have worked smoothly.

Automatic conversion to sealed trait

Example:

sealed trait Foo

final case class Bar(a: Int, b: Option[String]) extends Foo

final case class Baz(z: Int)                    extends Foo


val barCfg: String = """
    {
      a: 42,
      b: "hello"
    }
"""
// barCfg: String = """
//     {
//       a: 42,
//       b: "hello"
//     }
// """

val bazCfg: String = """
    {
      z: 1
    }
"""
// bazCfg: String = """
//     {
//       z: 1
//     }
// """

val barFoo: Result[Foo] = for {
  cfg <- Config.from(barCfg)
  foo <- cfg.as[Foo]
} yield foo
// barFoo: Result[Foo] = Right(value = Bar(a = 42, b = Some(value = "hello")))

val bazFoo: Result[Foo] = for {
  cfg <- Config.from(bazCfg)
  foo <- cfg.as[Foo]
} yield foo
// bazFoo: Result[Foo] = Right(value = Baz(z = 1))

The value of barFoo will be:

Right(Bar(42,Some(hello)))

The value of bazFoo will be:

Right(Baz(1))

Automatic conversion to Map[String, A]

You can convert a configuration to a Map[String, A], provided that all the values in the Map have the same type, A, and there exists an implicit instance of ConcreteValue[A] in scope. Most of the times this instance is automatically generated as in the case class example so you don't need to worry about that. Here's an example:

val cfgStr = """
{
  foo = 0,
  bar = 1,
  baz = 42
}
"""
// cfgStr: String = """
// {
//   foo = 0,
//   bar = 1,
//   baz = 42
// }
// """

val config: Result[Config] = Config.from(cfgStr)
// config: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "foo" -> AbstractNumber(value = 0.0),
//         "bar" -> AbstractNumber(value = 1.0),
//         "baz" -> AbstractNumber(value = 42.0)
//       )
//     )
//   )
// )

val confAsMap: Result[Map[String, Int]] = for {
  conf <- config
  result <- conf.as[Map[String, Int]]
} yield result
// confAsMap: Result[Map[String, Int]] = Right(
//   value = Map("foo" -> 0, "bar" -> 1, "baz" -> 42)
// )

The value of confAsMap will be:

Right(Map(foo -> 0, bar -> 1, baz -> 42))

Value-by-value conversion

Instead of using a case class you may want to retrieve the single values and convert them as you go:

val bar: Result[String] = for {
  conf <- config
  result <- conf.getAs[String]("bar")
} yield result
// bar: Result[String] = Left(
//   value = com.lambdista.config.ConversionError: Could not convert 1.0 to the type requested
// )

The value of bar will be:

Right("hello")

You can also use the dot syntax to retrieve a value. E.g.:

val cfgStr = """
{
  foo = {
    bar = 42
  }
}
"""
// cfgStr: String = """
// {
//   foo = {
//     bar = 42
//   }
// }
// """

val config: Result[Config] = Config.from(cfgStr)
// config: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "foo" -> AbstractMap(value = Map("bar" -> AbstractNumber(value = 42.0)))
//       )
//     )
//   )
// )

val bar: Result[Int] = for {
  c <- config
  bar <- c.getAs[Int]("foo.bar")
} yield bar
// bar: Result[Int] = Right(value = 42)

Note how the bar value was retrieved using the dot syntax.

Apart from converting the whole config into a case class, you can also convert a given value provided it's an object in the JSON-superset syntax:

val greekList: Result[List[Greek]] = for {
  conf <- config
  result <- conf.getAs[List[Greek]]("mapList")
} yield result
// greekList: Result[List[Greek]] = Left(
//   value = com.lambdista.config.KeyNotFoundError: No such key: mapList
// )

The value of greekList will be:

Right(List(Greek(hello,42), Greek(world,24)))

Sorry? You said you would have preferred a Vector[Greek] in place of List[Greek]? No problem:

val greekVector: Result[Vector[Greek]] = for {
  conf <- config
  result <- conf.getAs[Vector[Greek]]("mapList")
} yield result
// greekVector: Result[Vector[Greek]] = Left(
//   value = com.lambdista.config.KeyNotFoundError: No such key: mapList
// )

Here's the value of greekVector:

Right(Vector(Greek(hello,42), Greek(world,24)))

Oh, yes, Set[Greek] would have worked too:

val greekSet: Result[Set[Greek]] = for {
  conf <- config
  result <- conf.getAs[Set[Greek]]("mapList")
} yield result
// greekSet: Result[Set[Greek]] = Left(
//   value = com.lambdista.config.KeyNotFoundError: No such key: mapList
// )

Here's the value of greekSet:

Right(Set(Greek(hello,42), Greek(world,24)))

Analogously you can automatically convert a Range into a List, Vector or Set:

val rangeAsList: Result[List[Int]] = for {
  conf <- config
  result <- conf.getAs[List[Int]]("range")
} yield result
// rangeAsList: Result[List[Int]] = Left(
//   value = com.lambdista.config.KeyNotFoundError: No such key: range
// )

val rangeAsVector: Result[Vector[Int]] = for {
  conf <- config
  result <- conf.getAs[Vector[Int]]("range")
} yield result
// rangeAsVector: Result[Vector[Int]] = Left(
//   value = com.lambdista.config.KeyNotFoundError: No such key: range
// )

val rangeAsSet: Result[Set[Int]] = for {
  conf <- config
  result <- conf.getAs[Set[Int]]("range")
} yield result
// rangeAsSet: Result[Set[Int]] = Left(
//   value = com.lambdista.config.KeyNotFoundError: No such key: range
// )

Here are the results:

Right(List(2, 4, 6, 8, 10)) // rangeAsList

Right(Vector(2, 4, 6, 8, 10)) // rangeAsVector

Right(Set(4, 2, 8, 6, 10)) // rangeAsSet

Notice, however, that in case of Set the order is not guaranteed because of the very nature of sets.

Dynamic value-by-value conversion

You can also use a dynamic syntax to access the configuration values by pretending the Config object has those fields:

val alpha: Result[String] = for {
  conf <- config
  result <- conf.map.alpha.as[String] // equivalent to: conf.getAs[String]("map.alpha")
} yield result
// alpha: Result[String] = Left(
//   value = com.lambdista.config.KeyNotFoundError: No such key: map
// )

The value of alpha will be:

Right("hello")

Warning: Some IDEs could mark map.alpha as an error since they don't know about the dynamic nature of those fields. Nevertheless, your code will keep compiling and working like a charm.

Custom concrete value decoders

Sometimes you may want to provide a custom concrete value decoder for some configuration parameter. For example you may want to decode a UUID as such instead of using the provided String concrete value decoder, you know, for a better type safety.

import java.util.UUID


val confStr: String = """
  {
    uuid = "238dfdf4-850d-4643-b4f3-019252515ed8"
  }
"""
// confStr: String = """
//   {
//     uuid = "238dfdf4-850d-4643-b4f3-019252515ed8"
//   }
// """
final case class Foo(uuid: UUID)

implicit val uuidCv: ConcreteValue[UUID] = new ConcreteValue[UUID] {
  override def apply(abstractValue: AbstractValue): Option[UUID] = abstractValue match {
    case AbstractString(x) => Result.attempt(UUID.fromString(x)).toOption
    case _                 => None
  }
}
// uuidCv: ConcreteValue[UUID] = repl.MdocSession$App$$anon$11@5c8322a6

val foo: Result[Foo] = for {
  conf <- Config.from(confStr)
  result <- conf.as[Foo]
} yield result
// foo: Result[Foo] = Right(
//   value = Foo(uuid = 238dfdf4-850d-4643-b4f3-019252515ed8)
// )

The value of foo will be:

Right(Foo(238dfdf4-850d-4643-b4f3-019252515ed8))

Config loaders

Apart from loading your config through a java.nio.file.Path you can also use the following resources:

  • String
  • java.io.File
  • scala.io.Source
  • java.net.URI
  • java.net.URL
  • com.typesafe.config.Config

If that's not enough it's not so difficult provide your implementation of the ConfigLoader type class and make it available in scope. Here's how the ConfigLoader looks like:

trait ConfigLoader[R] {
  def load(resource: R): Result[Config]
}

Actually all you need to do is find a way to read your resource into a String and your done. Have a look at the ConfigLoader companion object for some examples.

Loading config from a simple String

What follows is an example of loading the config from a simple String. In this example you can also appreciate two other features of the library: how it deals with null values and its ability to convert char ranges too.

val confStr: String = "{age = null, charRange = 'a' to 'z'}"
// confStr: String = "{age = null, charRange = 'a' to 'z'}"
    
val config: Result[Config] = Config.from(confStr)
// config: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "age" -> AbstractNone,
//         "charRange" -> AbstractRange(
//           value = Range(
//             97,
//             98,
//             99,
//             100,
//             101,
//             102,
//             103,
//             104,
//             105,
//             106,
//             107,
//             108,
//             109,
//             110,
//             111,
//             112,
//             113,
//             114,
//             115,
//             116,
//             117,
//             118,
//             119,
//             120,
//             121,
//             122
//           )
//         )
//       )
//     )
//   )
// )

val age: Result[Option[Int]] = for {
  conf <- config
  result <- conf.getAs[Option[Int]]("age")
} yield result
// age: Result[Option[Int]] = Right(value = None)

val charRange: Result[List[Char]] = for {
  conf <- config
  result <- conf.getAs[List[Char]]("charRange")
} yield result
// charRange: Result[List[Char]] = Right(
//   value = List(
//     'a',
//     'b',
//     'c',
//     'd',
//     'e',
//     'f',
//     'g',
//     'h',
//     'i',
//     'j',
//     'k',
//     'l',
//     'm',
//     'n',
//     'o',
//     'p',
//     'q',
//     'r',
//     's',
//     't',
//     'u',
//     'v',
//     'w',
//     'x',
//     'y',
//     'z'
//   )
// )

As you may expect the values of age and charRange will be:

Right(None) // age

Right(List(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t, u, v, w, x, y, z)) // charRange

Loading a config from Typesafe Config

Here's how simple is loading a configuration passing through Typesafe config library. First thing first, you need to add the dependency for the Typesafe config adapter:

libraryDependencies += "com.lambdista" %% "config-typesafe" % "0.8.1"

The example configuration is the following:

string = "hello"
int = 42
double = 1.414
boolean = true
list = [1, 2, 3]
mapList = [
  {
    firstName = "John"
    lastName = "Doe"
  }
  {
    firstName = "Jane"
    lastName = "Doe"
  }
]

Suppose it's in a file at the relative path typesafe/src/test/resources/typesafe.conf:

import java.io.File

import com.typesafe.config.{Config => TSConfig, ConfigFactory}

import com.lambdista.config.typesafe._ // important to bring into scope the ConfigLoader for Typesafe's Config
 // important to bring into scope the ConfigLoader for Typesafe's Config

case class Person(firstName: String, lastName: String)


case class TypesafeConfig(string: String, int: Int, double: Double, boolean: Boolean, list: List[Int], mapList: List[Person])


val confPath = "typesafe/src/test/resources/typesafe.conf"
// confPath: String = "typesafe/src/test/resources/typesafe.conf"

val tsConfig: TSConfig = ConfigFactory.parseFile(new File(confPath))
// tsConfig: com.typesafe.config.Config = Config(SimpleConfigObject({"boolean":true,"double":1.414,"int":42,"list":[1,2,3],"mapList":[{"firstName":"John","lastName":"Doe"},{"firstName":"Jane","lastName":"Doe"}],"string":"hello"}))

val configTry: Result[Config] = Config.from(tsConfig)
// configTry: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = HashMap(
//         "string" -> AbstractString(value = "hello"),
//         "double" -> AbstractNumber(value = 1.414),
//         "boolean" -> AbstractBool(value = true),
//         "int" -> AbstractNumber(value = 42.0),
//         "mapList" -> AbstractList(
//           values = List(
//             AbstractMap(
//               value = Map(
//                 "firstName" -> AbstractString(value = "John"),
//                 "lastName" -> AbstractString(value = "Doe")
//               )
//             ),
//             AbstractMap(
//               value = Map(
//                 "lastName" -> AbstractString(value = "Doe"),
//                 "firstName" -> AbstractString(value = "Jane")
//               )
//             )
//           )
//         ),
//         "list" -> AbstractList(
//           values = List(
//             AbstractNumber(value = 1.0),
//             AbstractNumber(value = 2.0),
//             AbstractNumber(value = 3.0)
//           )
//         )
//       )
//     )
//   )
// )

val typesafeConfig: Result[TypesafeConfig] = config.flatMap(_.as[TypesafeConfig])
// typesafeConfig: Result[TypesafeConfig] = Left(
//   value = com.lambdista.config.ConversionError: Could not convert {age = None, charRange = [97.0, 98.0, 99.0, 100.0, 101.0, 102.0, 103.0, 104.0, 105.0, 106.0, 107.0, 108.0, 109.0, 110.0, 111.0, 112.0, 113.0, 114.0, 115.0, 116.0, 117.0, 118.0, 119.0, 120.0, 121.0, 122.0]} to the type requested
// )

The value of typesafeConfig will be:

Right(TypesafeConfig(hello,42,1.414,true,List(1, 2, 3),List(Person(John,Doe), Person(Jane,Doe))))

Merging two configurations

You can also merge two configurations using either the recursivelyMerge or merge method of Config, as in config.recursivelyMerge(thatConfig) or config.merge(thatConfig). The behaviour of the former is that, given a key, if the correspondent value is a map then thatConfig's value is recursively merged to this config's value otherwise thatConfig's value replaces this config's value. An example should clarify the difference between the two approaches:

val confStr1 = """
{
  foo = {
    alpha = 1,
    bar = "hello"
  },
  baz = 42
}
"""
// confStr1: String = """
// {
//   foo = {
//     alpha = 1,
//     bar = "hello"
//   },
//   baz = 42
// }
// """

val confStr2 = """
{
  foo = {
    baz = 15,
    bar = "goodbye"
  },
  baz = 1,
  zoo = "hi"
}
"""
// confStr2: String = """
// {
//   foo = {
//     baz = 15,
//     bar = "goodbye"
//   },
//   baz = 1,
//   zoo = "hi"
// }
// """

val config1: Result[Config] = Config.from(confStr1)
// config1: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "foo" -> AbstractMap(
//           value = Map(
//             "alpha" -> AbstractNumber(value = 1.0),
//             "bar" -> AbstractString(value = "hello")
//           )
//         ),
//         "baz" -> AbstractNumber(value = 42.0)
//       )
//     )
//   )
// )

val config2: Result[Config] = Config.from(confStr2)
// config2: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "foo" -> AbstractMap(
//           value = Map(
//             "baz" -> AbstractNumber(value = 15.0),
//             "bar" -> AbstractString(value = "goodbye")
//           )
//         ),
//         "baz" -> AbstractNumber(value = 1.0),
//         "zoo" -> AbstractString(value = "hi")
//       )
//     )
//   )
// )

val mergedConfig: Result[Config] = for {
  conf1 <- config1
  conf2 <- config2
} yield conf1.recursivelyMerge(conf2)
// mergedConfig: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "foo" -> AbstractMap(
//           value = Map(
//             "alpha" -> AbstractNumber(value = 1.0),
//             "bar" -> AbstractString(value = "goodbye"),
//             "baz" -> AbstractNumber(value = 15.0)
//           )
//         ),
//         "baz" -> AbstractNumber(value = 1.0),
//         "zoo" -> AbstractString(value = "hi")
//       )
//     )
//   )
// )

mergedConfig will represent a config such as the following:

{
  foo = {
    alpha = 1,
    baz = 15,
    bar = "goodbye"
  },
  baz = 1,
  zoo = "hi"
}

As you can see the value of config2's foo did not replace entirely the value of config1's foo, but they were recursively merged.

On the other hand merge's behaviour is more like Scala's default behaviour when using ++ between two Maps. Indeed, config2's values replace entirely config1's values with the same key. E.g.:

val confStr1 = """
{
  foo = {
    alpha = 1,
    bar = "hello"
  },
  zoo = "hi",
  baz = 42
}
"""
// confStr1: String = """
// {
//   foo = {
//     alpha = 1,
//     bar = "hello"
//   },
//   zoo = "hi",
//   baz = 42
// }
// """

val confStr2 = """
{
  foo = {
    baz = 15,
    bar = "goodbye"
  },
  baz = 1
}
"""
// confStr2: String = """
// {
//   foo = {
//     baz = 15,
//     bar = "goodbye"
//   },
//   baz = 1
// }
// """

val config1: Result[Config] = Config.from(confStr1)
// config1: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "foo" -> AbstractMap(
//           value = Map(
//             "alpha" -> AbstractNumber(value = 1.0),
//             "bar" -> AbstractString(value = "hello")
//           )
//         ),
//         "zoo" -> AbstractString(value = "hi"),
//         "baz" -> AbstractNumber(value = 42.0)
//       )
//     )
//   )
// )

val config2: Result[Config] = Config.from(confStr2)
// config2: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "foo" -> AbstractMap(
//           value = Map(
//             "baz" -> AbstractNumber(value = 15.0),
//             "bar" -> AbstractString(value = "goodbye")
//           )
//         ),
//         "baz" -> AbstractNumber(value = 1.0)
//       )
//     )
//   )
// )

val mergedConfig: Result[Config] = for {
  conf1 <- config1
  conf2 <- config2
} yield conf1.merge(conf2)
// mergedConfig: Result[Config] = Right(
//   value = Config(
//     abstractMap = AbstractMap(
//       value = Map(
//         "foo" -> AbstractMap(
//           value = Map(
//             "baz" -> AbstractNumber(value = 15.0),
//             "bar" -> AbstractString(value = "goodbye")
//           )
//         ),
//         "zoo" -> AbstractString(value = "hi"),
//         "baz" -> AbstractNumber(value = 1.0)
//       )
//     )
//   )
// )

mergedConfig will represent a config such as the following:

{
  foo = {
    baz = 15,
    bar = "goodbye"
  },
  baz = 1,
  zoo = "hi"
}

Look at the tests for this library to see the examples in practise.

Scaladoc API

config API

Bugs and Feedback

For bugs, questions and discussions please use Github Issues.

License

Copyright 2016-2021 Alessandro Lacava.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.