macros

Type Members

1. trait Cuber[I] extends AnyRef

   

   "Cubes" a case class or tuple, i.e.

   "Cubes" a case class or tuple, i.e. for a tuple of type (T1, T2, ... , TN) generates all 2^N possible combinations of type (Option[T1], Option[T2], ... , Option[TN]).

   This is useful for comparing some metric across all possible subsets. For example, suppose we have a set of people represented as case class Person(gender: String, age: Int, height: Double) and we want to know the average height of

   • people, grouped by gender and age
   • people, grouped by only gender
   • people, grouped by only age
   • all people

   Then we could do > import com.twitter.algebird.macros.Cuber.cuber > val people: List[People] > val averageHeights: Map[(Option[String], Option[Int]), Double] = > people.flatMap { p => cuber((p.gender, p.age)).map((_,p)) } > .groupBy(_._1) > .mapValues { xs => val heights = xs.map(_.height); heights.sum / heights.length }

2. trait Roller[I] extends AnyRef

   

   Given a TupleN, produces a sequence of (N + 1) tuples each of arity N such that, for all k from 0 to N, there is a tuple with k Somes followed by (N - k) Nones.

   Given a TupleN, produces a sequence of (N + 1) tuples each of arity N such that, for all k from 0 to N, there is a tuple with k Somes followed by (N - k) Nones.

   This is useful for comparing some metric across multiple layers of some hierarchy. For example, suppose we have some climate data represented as case class Data(continent: String, country: String, city: String, temperature: Double) and we want to know the average temperatures of

   • each continent
   • each (continent, country) pair
   • each (continent, country, city) triple

   Here we desire the (continent, country) and (continent, country, city) pair because, for example, if we grouped by city instead of by (continent, country, city), we would accidentally combine the results for Paris, Texas and Paris, France.

   Then we could do > import com.twitter.algebird.macros.Roller.roller > val data: List[Data] > val averageTemps: Map[(Option[String], Option[String], Option[String]), Double] = > data.flatMap { d => roller((d.continent, d.country, d.city)).map((_, d)) } > .groupBy(_._1) > .mapValues { xs => val temps = xs.map(_.temperature); temps.sum / temps.length }