Skip to content

Scala Library for Data Consolidation and Validation

Notifications You must be signed in to change notification settings

AdrielC/Promethean

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Promethean

Promethean is a type class based generic programming library using Scala's reflection API The primary uses for this library are data consolidation between conceptually similar case classes and for data validation.

Using Promethean

Promethean is an abstract class that can be extended by any case class to enrich it with generic programming methods. Any case class that extends Promethean must define the identifyingFields method, which is the method used by Promethean to determine equatability between Promethean classes.

Consolidating Data with absorb

case class Person(name: String, age: Int = -1, address: String = "") extends Promethean[Person] {
  def identifyingFields = Seq(field("name"))
}

case class PhoneBookRecord(name: String, phoneNumber: Int, address: String) extends Promethean[Person] {
  def identifyingFields = Seq(field("name"))
}

Let's say we don't know Layton's address when we instantiate his Person class. We can omit his address since we defined the default to be a blank string

val layton = Person(name = "Layton", age = 23)

// Perhaps we have another datasource of phone book records and we find one with Layton's address. 

val laytonRecord = PhoneBookRecord(name = "Layton", phoneNumber = 5555555555, address = "6 Columbia Lane")

We can use Promethean to consolidate data for us since both case classes represent the same concept and share field names. Both layton and laytonRecord have the same value in their identifyingFields, making them equatable.

layton === laytonRecord // results in true

// consolidating data is done with `absorb`

val laytonFull = layton.absorb(laytonRecord)
// returns Person("Layton", 23, "6 Columbia Lane")

Creating Reactive Data with reactiveFields

We can also define how case classes behave when exposed to certain scenarios. Defining reactiveFields in the case class definition allows the user to specify additional behavior of an absorbtion.

For example, if we find out that Layton has a phone number, we can assume that he must have a phone. Instead of definig that behavior elsewhere in our project, we can define that behavior by overriding the reactiveFields method.

// redefine case classes with reactiveFields


case class Person(name: String, age: Int -1, address: String = "", hasPhone: Boolean = false) extends Promethean[Person] {
  def identifyingFields = Seq(field("name"))
  def reactiveFields = Seq(
    AbsorbReaction(field = field("hasPhone"), precipitate = a => true, trigger = a => a.isInstanceOf[PhoneBookRecord])
  )
}

case class PhoneBookRecord(name: String, phoneNumber: Int, address: String) extends Promethean[Person] {
  def identifyingFields = Seq(field("name"))
}

Within the reactiveFields definition for Person, we create an AbsorbReaction which defines the rules for absorbtion. This reaction recipe is evaluated each time and instance of Person absorbs another Promethean type class. The three components to a PrometheanReaction type, of which include AbsorbReaction and AutoReaction, are the following:

  1. field This defines which field in the parent class will be subject to change

  2. precipitate This is the outcome of the reaction that will replace the current value of the specified field. This value is an anonymous function that must evaluate to a value of the same type expected by the field. The value can be static, as seen in this example, or derived from the class being absorbed (by using the LHS variable)

  3. trigger This is an anonymous function that is evaluated to determine whether a new value will be precipitated by the reaction. In this example, anytime an instance of Person absorbs an instance of PhoneBookRecord, we then can say that the person has a phone, thus we precipitate a value of true into the hasPhone of the absorbing Person instance.

AutoReactions behave in a similar manner, although they are self reactions which are catalyzed with a call to the Promethean method autocatalyze.

Case Class Conversions with hardenAs[_] and Autocatalysis

case class Person(name: String, age: Int -1, address: String = "", phoneNumber = -1, hasPhone: Boolean = false) extends Promethean[Person] {
  def identifyingFields = Seq(field("name"))
  def reactiveFields = Seq(
    AbsorbReaction(field = field("hasPhone"), precipitate = a => true, trigger = a => a.isInstanceOf[PhoneBookRecord]),
    AutoReaction(field = field("hasPhone"), precipitate = true, trigger = !clayFields.contains("phoneNumber"))
  )
}

case class PhoneBookRecord(name: String, phoneNumber: Int, address: String) extends Promethean[Person] {
  def identifyingFields = Seq(field("name"))
}

val layton = Person(name = "Layton", age = 23)
val laytonRecord = PhoneBookRecord(name = "Layton", phoneNumber = 5555555555, address = "6 Columbia Lane")

val personFromRecord = laytonRecord.hardenAs[Person].autocatalyze
// personFromRecord: Person = Person(name = "Layton", 
//                                   age = -1, 
//                                   address = "6 Columbia Lane", 
//                                   phoneNumber = 5555555555, 
//                                   hasPhone = true)
                                     
val fullPersonFromRecord = personFromRecord.absorb(layton)
// fullPersonFromRecord: Person = Person(name = "Layton", 
//                                       age = 23, 
//                                       address = "6 Columbia Lane", 
//                                       phoneNumber = 5555555555, 
//                                       hasPhone = true)

In this example we introduce the following

  1. case class conversions using the hardenAs[_] method
  2. clayFields and the importance of defaults
  3. Autoreactions

The hardenAs[Person] call converts the instance of PhoneBookRecord to an instance of Person using the shared field names and their corresponding values, using defualt values to fill in the rest. This will fail at runtime however if there was no default value for age since PhoneBookRecord does not define that field. However since the default for Person.age is defined as -1 it is able to complete the conversion. The method canHardenAs[_] can be used to check if a conversion is possible.

clayFields returns a Set[String] that contains all of the field names that are still equal to their default values. In this example, we are converting from a PhoneBookRecord to a person, the phoneNumber field differs from the default of -1, thus it will not be found in the set returned by clayFields.

Since we converted from a PhoneBookRecord to a Person, we now have access to the AutoReaction defined in Persons reactiveFields. The AutoReaction specifies that if the value for the phoneNumber field differs from -1---which means the field name will no longer be returned by clayFields and will now be found as a key in the Map returned by clayValues---then we can flip the hasPhone value to true when we call the autocatalyze method on the newly hardened person.

About

Scala Library for Data Consolidation and Validation

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages