sweet-delights / delightful-anonymization   0.1.0

delightful-anonymization is a library for anonymizing case classes on-the-fly.

This library is built for Scala 2.12 and 2.13.

libraryDependencies += "org.sweet-delights" %% "delightful-anonymization" % "0.1.1"

<dependency>
  <groupId>org.sweet-delights</groupId>
  <artifactId>delightful-anonymization_2.12</artifactId>
  <version>0.0.1</version>
</dependency>

All files in delightful-anonymization are under the GNU Lesser General Public License version 3. Please read files COPYING and COPYING.LESSER for details.

Step 1: decorate a case class with @PII annotations. Example:

import sweet.delights.anonymization.{Hash, PII}

case class Foo(
  opt: Option[String] @PII(Hash.MD5),
  str: String         @PII(Hash.SHA512),
  integer: Int
)

Step 2: apply the anonymize function on an instance of Foo:

val foo = Foo(
  Some("opt"),
  "str",
  1
)

val anonymized == Foo(
opt = Some("@-A9WeZjwa+awzqZSdEZNQWg==")
,
str = "@-Ms3snktf//qQkCS0pxCFDuLhtNPxn/2PJImMPoQBmZes+h+d3Q39yiEojcksp2agyxDgzXstaSbe/+zMWSOVAg=="
,
integer = 1
)
//> true

By default, Anonymizer hashes arrays of bytes. But any type T - other than products and co-products - that can be transformed into an array of bytes can be hashed.

The support for additional types is done via Injections, a mechanism borrowed from the frameless library.

For example, support for strings is added with the following:

import org.apache.commons.codec.binary.Base64
import sweet.delights.anonymization.Injection

lazy val anonymizedPrefix = "@:"

implicit lazy val stringInjection: Injection[String, Array[Byte]] = new Injection[String, Array[Byte]] {
  override def isAnonymized(t: String): Boolean = t.startsWith(anonymizedPrefix)

  override def apply(t: String): Array[Byte] = t.getBytes("UTF-8")

  override def invert(u: Array[Byte]): String = anonymizedPrefix + Base64.encodeBase64String(u)
}

Comments:

• idempotence is achieved by calling the isAnonymized function. If it returns true then the value t is not re-hashed. Otherwise the specified hashing algorithm is applied.
• it is up to the user to decide which injections are to be idempotent or not
• the default hashing implementation of strings is idempotent

The hashing algorithms are those supported by Java 8:

• MD5
• SHA-1
• SHA-256
• SHA-384
• SHA-512

Other algorithms, not necessarly hashing algorithms, could be implemented. For instance, the anonymization method FirstLetter-of-a-string could be added. Contributions welcome!

• the shapeless library
• the frameless library for the Injection mechanism
• the Apache Commons Codec library
• the The Type Astronaut's Guide to Shapeless book