shapeless-datatype

Shapeless utilities for common data types. Also see Magnolify for a simpler and faster alternative based on Magnolia.

Modules

This library includes the following modules.

• shapeless-datatype-core
• shapeless-datatype-avro
• shapeless-datatype-bigquery
• shapeless-datatype-datastore
• shapeless-datatype-tensorflow

Core

Core includes the following components.

• A MappableType for generic conversion between case class and other data types, used by BigQuery and Datastore modules.
• A RecordMapper for generic conversion between case class types.
• A RecordMatcher for generic type-based equality check bewteen case classes.
• A LensMatcher for generic lens-based equality check between case classes.

RecordMapper

RecordMapper[A, B] maps instances of case class A and B with different field types.

import shapeless._
import shapeless.datatype.record._
import scala.language.implicitConversions

// records with same field names but different types
case class Point1(x: Double, y: Double, label: String)
case class Point2(x: Float, y: Float, label: String)

// implicit conversion bewteen fields of different types
implicit def f2d(x: Float) = x.toDouble
implicit def d2f(x: Double) = x.toFloat

val m = RecordMapper[Point1, Point2]
m.to(Point1(0.5, -0.5, "a"))  // Point2(0.5,-0.5,a)
m.from(Point2(0.5, -0.5, "a")) // Point1(0.5,-0.5,a)

RecordMatcher

RecordMatcher[T] performs equality check of instances of case class T with custom logic based on field types.

import shapeless.datatype.record._

case class Record(id: String, name: String, value: Int)

// custom comparator for String type
implicit def compareStrings(x: String, y: String) = x.toLowerCase == y.toLowerCase

val m = RecordMatcher[Record]
Record("a", "RecordA", 10) == Record("A", "RECORDA", 10)  // false

// compareStrings is applied to all String fields
m(Record("a", "RecordA", 10), Record("A", "RECORDA", 10))  // true

LensMatcher

LensMatcher[T] performs equality check of instances of case class T with custom logic based on Lenses.

import shapeless.datatype.record._

case class Record(id: String, name: String, value: Int)

// compare String fields id and name with different logic
val m = LensMatcher[Record]
  .on(_ >> 'id)(_.toLowerCase == _.toLowerCase)
  .on(_ >> 'name)(_.length == _.length)

Record("a", "foo", 10) == Record("A", "bar", 10)  // false
m(Record("a", "foo", 10), Record("A", "bar", 10))  // true

AvroType

AvroType[T] maps bewteen case class T and Avro GenericRecord. AvroSchema[T] generates schema for case class T.

import shapeless.datatype.avro._

case class City(name: String, code: String, lat: Double, long: Double)

val t = AvroType[City]
val r = t.toGenericRecord(City("New York", "NYC", 40.730610, -73.935242))
val c = t.fromGenericRecord(r)

AvroSchema[City]

Custom types are also supported.

import shapeless.datatype.avro._
import java.net.URI
import org.apache.avro.Schema

implicit val uriAvroType = AvroType.at[URI](Schema.Type.STRING)(v => URI.create(v.toString), _.toString)

case class Page(uri: URI, rank: Int)

val t = AvroType[Page]
val r = t.toGenericRecord(Page(URI.create("www.google.com"), 42))
val c = t.fromGenericRecord(r)

AvroSchema[Page]

BigQueryType

BigQueryType[T] maps bewteen case class T and BigQuery TableRow. BigQuerySchema[T] generates schema for case class T.

import shapeless.datatype.bigquery._

case class City(name: String, code: String, lat: Double, long: Double)

val t = BigQueryType[City]
val r = t.toTableRow(City("New York", "NYC", 40.730610, -73.935242))
val c = t.fromTableRow(r)

BigQuerySchema[City]

Custom types are also supported.

import shapeless.datatype.bigquery._
import java.net.URI

implicit val uriBigQueryType = BigQueryType.at[URI]("STRING")(v => URI.create(v.toString), _.toString)

case class Page(uri: URI, rank: Int)

val t = BigQueryType[Page]
val r = t.toTableRow(Page(URI.create("www.google.com"), 42))
val c = t.fromTableRow(r)

BigQuerySchema[Page]

DatastoreType

DatastoreType[T] maps between case class T and Cloud Datastore Entity or Entity.Builder Protobuf types.

import shapeless.datatype.datastore._

case class City(name: String, code: String, lat: Double, long: Double)

val t = DatastoreType[City]
val r = t.toEntity(City("New York", "NYC", 40.730610, -73.935242))
val c = t.fromEntity(r)
val b = t.toEntityBuilder(City("New York", "NYC", 40.730610, -73.935242))
val d = t.fromEntityBuilder(b)

Custom types are also supported.

import shapeless.datatype.datastore._
import com.google.datastore.v1.client.DatastoreHelper._
import java.net.URI

implicit val uriDatastoreType = DatastoreType.at[URI](
  v => URI.create(v.getStringValue),
  u => makeValue(u.toString).build())

case class Page(uri: URI, rank: Int)

val t = DatastoreType[Page]
val r = t.toEntity(Page(URI.create("www.google.com"), 42))
val c = t.fromEntity(r)
val b = t.toEntityBuilder(Page(URI.create("www.google.com"), 42))
val d = t.fromEntityBuilder(b)

TensorFlowType

TensorFlowType[T] maps between case class T and TensorFlow Example or Example.Builder Protobuf types.

import shapeless.datatype.tensorflow._

case class Data(floats: Array[Float], longs: Array[Long], strings: List[String], label: String)

val t = TensorFlowType[Data]
val r = t.toExample(Data(Array(1.5f, 2.5f), Array(1L, 2L), List("a", "b"), "x"))
val c = t.fromExample(r)
val b = t.toExampleBuilder(Data(Array(1.5f, 2.5f), Array(1L, 2L), List("a", "b"), "x"))
val d = t.fromExampleBuilder(b)

Custom types are also supported.

import shapeless.datatype.tensorflow._
import java.net.URI

implicit val uriTensorFlowType = TensorFlowType.at[URI](
  TensorFlowType.toStrings(_).map(URI.create),
  xs => TensorFlowType.fromStrings(xs.map(_.toString)))

case class Page(uri: URI, rank: Int)

val t = TensorFlowType[Page]
val r = t.toExample(Page(URI.create("www.google.com"), 42))
val c = t.fromExample(r)
val b = t.toExampleBuilder(Page(URI.create("www.google.com"), 42))
val d = t.fromExampleBuilder(b)

License

Copyright 2016 Neville Li.

Licensed under the Apache License, Version 2.0: http://www.apache.org/licenses/LICENSE-2.0