com.toasttab.protokt:protokt-codegen

Protobuf compiler and runtime for Kotlin


License
Apache-2.0

Documentation

protokt

CircleCI Maven Central Gradle Portal

Protocol Buffer compiler and runtime for Kotlin.

Supports only version 3 of the Protocol Buffers language.

Overview

Features

Not yet implemented

  • Kotlin/Native support
  • Protobuf JSON support

Compatibility

The Gradle plugin requires Java 8+ and Gradle 5.6+. It runs on recent versions of MacOS, Linux, and Windows.

The runtime and generated code are compatible with Kotlin 1.8+, Java 8+, and Android 4.4+.

Usage

See examples in testing.

Gradle

plugins {
    id("com.toasttab.protokt") version "<version>"
}

or

buildscript {
    dependencies {
        classpath("com.toasttab.protokt:protokt-gradle-plugin:<version>")
    }
}

apply(plugin = "com.toasttab.protokt")

This will automatically download and install protokt, apply the Google protobuf plugin, and configure all the necessary boilerplate. By default it will also add protokt-core to the api scope of the project. On the JVM you must explicitly choose to depend on protobuf-java or protobuf-javalite:

dependencies {
    implementation("com.google.protobuf:protobuf-java:<version>")
}

or

dependencies {
    implementation("com.google.protobuf:protobuf-javalite:<version>")
}

If your project has no Java code you may run into the following error:

Execution failed for task ':compileJava'.
> error: no source files

To work around it, disable all JavaCompile tasks in the project:

tasks.withType<JavaCompile> {
    enabled = false
}

Generated Code

Generated code is placed in <buildDir>/generated/<sourceSet.name>/protokt.

A simple example:

syntax = "proto3";

package toasttab.protokt.sample;

message Sample {
  string sample_field = 1;
}

will produce:

// Generated by protokt version 1.0.0-alpha.7-SNAPSHOT. Do not modify.
// Source: test_file.proto
@file:Suppress("DEPRECATION")

package protokt.v1.toasttab.protokt.sample

import protokt.v1.AbstractKtDeserializer
import protokt.v1.AbstractKtMessage
import protokt.v1.KtBuilderDsl
import protokt.v1.KtMessageDeserializer
import protokt.v1.KtMessageSerializer
import protokt.v1.SizeCodecs.sizeOf
import protokt.v1.UnknownFieldSet
import protokt.v1.google.protobuf.Descriptor
import protokt.v1.google.protobuf.FileDescriptor
import kotlin.Any
import kotlin.Boolean
import kotlin.Int
import kotlin.String
import kotlin.Suppress
import kotlin.Unit
import kotlin.jvm.JvmStatic
import com.toasttab.protokt.rt.KtGeneratedFileDescriptor as RtKtGeneratedFileDescriptor
import com.toasttab.protokt.rt.KtGeneratedMessage as RtKtGeneratedMessage
import protokt.v1.KtGeneratedFileDescriptor as V1KtGeneratedFileDescriptor
import protokt.v1.KtGeneratedMessage as V1KtGeneratedMessage

@V1KtGeneratedMessage("toasttab.protokt.sample.Sample")
@RtKtGeneratedMessage("toasttab.protokt.sample.Sample")
public class Sample private constructor(
  public val sampleField: String,
  public val unknownFields: UnknownFieldSet = UnknownFieldSet.empty()
) : AbstractKtMessage() {
  override val messageSize: Int by lazy { messageSize() }

  private fun messageSize(): Int {
    var result = 0
    if (sampleField.isNotEmpty()) {
      result += sizeOf(10u) + sizeOf(sampleField)
    }
    result += unknownFields.size()
    return result
  }

  override fun serialize(serializer: KtMessageSerializer) {
    if (sampleField.isNotEmpty()) {
      serializer.writeTag(10u).write(sampleField)
    }
    serializer.writeUnknown(unknownFields)
  }

  override fun equals(other: Any?): Boolean =
    other is Sample &&
      other.sampleField == sampleField &&
      other.unknownFields == unknownFields

  override fun hashCode(): Int {
    var result = unknownFields.hashCode()
    result = 31 * result + sampleField.hashCode()
    return result
  }

  override fun toString(): String =
    "Sample(" +
      "sampleField=$sampleField" +
      if (unknownFields.isEmpty()) ")" else ", unknownFields=$unknownFields)"

  public fun copy(builder: Builder.() -> Unit): Sample =
    Builder().apply {
      sampleField = this@Sample.sampleField
      unknownFields = this@Sample.unknownFields
      builder()
    }.build()

  @KtBuilderDsl
  public class Builder {
    public var sampleField: String = ""

    public var unknownFields: UnknownFieldSet = UnknownFieldSet.empty()

    public fun build(): Sample =
      Sample(
        sampleField,
        unknownFields
      )
  }

  public companion object Deserializer : AbstractKtDeserializer<Sample>() {
    @JvmStatic
    override fun deserialize(deserializer: KtMessageDeserializer): Sample {
      var sampleField = ""
      var unknownFields: UnknownFieldSet.Builder? = null

      while (true) {
        when (deserializer.readTag()) {
          0 -> return Sample(
            sampleField,
            UnknownFieldSet.from(unknownFields)
          )
          10 -> sampleField = deserializer.readString()
          else ->
            unknownFields =
              (unknownFields ?: UnknownFieldSet.Builder()).also {
                it.add(deserializer.readUnknown())
              }
        }
      }
    }

    @JvmStatic
    public operator fun invoke(dsl: Builder.() -> Unit): Sample = Builder().apply(dsl).build()
  }
}

@V1KtGeneratedFileDescriptor
@RtKtGeneratedFileDescriptor
public object test_file_file_descriptor {
  public val descriptor: FileDescriptor by lazy {
    val descriptorData =
      arrayOf(
        "\n�test_file.proto��toasttab.protokt.samp" +
          "le\"\n�Sample��\nsample_field�� �(\tb�prot" +
          "o3"
      )

    FileDescriptor.buildFrom(
      descriptorData,
      listOf()
    )
  }
}

public val Sample.Deserializer.descriptor: Descriptor
  get() = test_file_file_descriptor.descriptor.messageTypes[0]

Construct your protokt object like so:

Sample {
    sampleField = "some-string"
}

Why not expose a public constructor or use a data class? One of the design goals of protocol buffers is that protobuf definitions can be modified in backwards-compatible ways without breaking wire or API compatibility of existing code. Using a DSL to construct the object emulates named arguments and allows shuffling of protobuf fields within a definition without breaking code as would happen for a standard constructor or method call.

The canonical copy method on data classes is emulated via a generated copy method:

val sample = Sample { sampleField = "some-string" }

val sample2 = sample.copy { sampleField = "some-other-string" }

Assigning a Map or List in the DSL makes a copy of that collection to prevent any escaping mutability of the provided collection. The Java protobuf implementation takes a similar approach; it only exposes mutation methods on the builder and not assignment. Mutating the builder does a similar copy operation.

Runtime Notes

Package

The Kotlin package of a generated file is the protobuf package prefixed with protokt.v1. This scheme allows protokt-generated files to coexist on the classpath with files generated by other compilers.

Message

Each protokt message implements the KtMessage interface. KtMessage defines the serialize() method and its overloads which can serialize to a byte array, a KtMessageSerializer, or an OutputStream.

Each protokt message has a companion object Deserializer that implements the KtDeserializer interface, which provides the deserialize() method and its overloads to construct an instance of the message from a byte array, a Java InputStream, or others.

Enums

Representation

Protokt represents enum fields as sealed classes with an integer value and name. Protobuf enums cannot be represented as Kotlin enum classes since Kotlin enum classes are closed and cannot represent unknown values. The Protocol Buffers specification requires that unknown enum values are preserved for reserialization, so this compromise enables exhaustive case switching while allowing representation of unknown values.

public sealed class PhoneType(
  override val `value`: Int,
  override val name: String
) : KtEnum() {
  public object MOBILE : PhoneType(0, "MOBILE")

  public object HOME : PhoneType(1, "HOME")

  public object WORK : PhoneType(2, "WORK")

  public class UNRECOGNIZED(
    `value`: Int
  ) : PhoneType(value, "UNRECOGNIZED")

  public companion object Deserializer : KtEnumDeserializer<PhoneType> {
    override fun from(`value`: Int): PhoneType =
      when (value) {
        0 -> MOBILE
        1 -> HOME
        2 -> WORK
        else -> UNRECOGNIZED(value)
      }
  }
}

Naming

To keep enums ergonomic while promoting protobuf best practices, enums that have all values prefixed with the enum type name will have that prefix stripped in their Kotlin representations.

Other Notes

  • optimize_for is ignored.
  • repeated fields are represented as Lists.
  • map fields are represented as Maps.
  • oneof fields are represented as subtypes of a sealed base class with a single property.
  • bytes fields are wrapped in the protokt Bytes class to ensure immutability akin to protobuf-java's ByteString.
  • Protokt implements proto3's optional.

Extensions

See extension options defined in protokt.proto.

See examples of each option in the options project. All protokt-specific options require importing protokt/v1/protokt.proto in the protocol file.

Wrapper Types

Sometimes a field on a protobuf message corresponds to a concrete nonprimitive type. In standard protobuf the user would be responsible for this extra transformation, but the protokt wrapper type option allows specification of a converter that will automatically encode and decode custom types to protobuf types. Some standard types are implemented in extensions.

Wrap a field by invoking the (protokt.v1.property).wrap option:

message WrapperMessage {
  google.protobuf.Timestamp instant = 1 [
    (protokt.v1.property).wrap = "java.time.Instant"
  ];
}

Converters implement the Converter interface:

interface Converter<S : Any, T : Any> {
    val wrapper: KClass<S>

    val wrapped: KClass<T>

    fun wrap(unwrapped: T): S

    fun unwrap(wrapped: S): T
}

and protokt will reference the converter's methods to wrap and unwrap from protobuf primitives:

object InstantConverter : Converter<Instant, Timestamp> {
    override val wrapper = Instant::class

    override val wrapped = Timestamp::class

    override fun wrap(unwrapped: Timestamp): Instant =
        Instant.ofEpochSecond(unwrapped.seconds, unwrapped.nanos.toLong())

    override fun unwrap(wrapped: Instant) =
        Timestamp {
            seconds = wrapped.epochSecond
            nanos = wrapped.nano
        }
}
@V1KtGeneratedMessage("protokt.v1.testing.Wrappers")
@RtKtGeneratedMessage("protokt.v1.testing.Wrappers")
public class Wrappers private constructor(
  public val instant: Instant,
  public val unknownFields: UnknownFieldSet = UnknownFieldSet.empty()
) : AbstractKtMessage() {
  override val messageSize: Int by lazy { messageSize() }

  private fun messageSize(): Int {
    var result = 0
    result += sizeOf(50u) + sizeOf(InstantConverter.unwrap(instant))
    result += unknownFields.size()
    return result
  }

  override fun serialize(serializer: KtMessageSerializer) {
    serializer.writeTag(50u).write(InstantConverter.unwrap(instant))
    serializer.writeUnknown(unknownFields)
  }

  override fun equals(other: Any?): Boolean =
    other is Wrappers &&
            other.instant == instant &&
            other.unknownFields == unknownFields

  override fun hashCode(): Int {
    var result = unknownFields.hashCode()
    result = 31 * result + instant.hashCode()
    return result
  }

  override fun toString(): String =
    "Wrappers(" +
            "instant=$instant" +
            if (unknownFields.isEmpty()) ")" else ", unknownFields=$unknownFields)"

  public fun copy(builder: Builder.() -> Unit): Wrappers =
    Builder().apply {
      instant = this@Wrappers.instant
      unknownFields = this@Wrappers.unknownFields
      builder()
    }.build()

  @KtBuilderDsl
  public class Builder {
    public var instant: Instant? = null

    public var unknownFields: UnknownFieldSet = UnknownFieldSet.empty()

    public fun build(): Wrappers =
      Wrappers(
        requireNotNull(instant) {
          StringBuilder("instant")
            .append(" specified nonnull with (protokt.property).non_null but was null")
        },
        unknownFields
      )
  }

  public companion object Deserializer : AbstractKtDeserializer<Wrappers>() {
    @JvmStatic
    override fun deserialize(deserializer: KtMessageDeserializer): Wrappers {
      var instant: Instant? = null
      var unknownFields: UnknownFieldSet.Builder? = null

      while (true) {
        when (deserializer.readTag()) {
          0 -> return Wrappers(
            requireNotNull(instant) {
              StringBuilder("instant")
                .append(" specified nonnull with (protokt.property).non_null but was null")
            },
            UnknownFieldSet.from(unknownFields)
          )
          50 -> instant = InstantConverter.wrap(deserializer.readMessage(Timestamp))
          else ->
            unknownFields =
              (unknownFields ?: UnknownFieldSet.Builder()).also {
                it.add(deserializer.readUnknown())
              }
        }
      }
    }

    @JvmStatic
    public operator fun invoke(dsl: Builder.() -> Unit): Wrappers = Builder().apply(dsl).build()
  }
}

Each converter must be registered in a META-INF/services/protokt.v1.Converter classpath resource following the standard ServiceLoader convention. For example, Google's AutoService can register converters with an annotation:

@AutoService(Converter::class)
object InstantConverter : Converter<Instant, Timestamp> { ... }

Converters can also implement the OptimizedSizeofConverter interface adding sizeof(), which allows them to optimize the calculation of the wrapper's size rather than unwrap the object twice. For example, a UUID is always 16 bytes:

object UuidBytesConverter : OptimizedSizeOfConverter<UUID, Bytes> {
    override val wrapper = UUID::class

    override val wrapped = Bytes::class

    private val sizeOfProxy = ByteArray(16)

    override fun sizeOf(wrapped: UUID) =
        sizeOf(sizeOfProxy)

    override fun wrap(unwrapped: Bytes): UUID {
        val buf = unwrapped.asReadOnlyBuffer()

        require(buf.remaining() == 16) {
            "UUID source must have size 16; had ${buf.remaining()}"
        }

        return buf.run { UUID(long, long) }
    }

    override fun unwrap(wrapped: UUID): Bytes =
        Bytes.from(
            ByteBuffer.allocate(16)
                .putLong(wrapped.mostSignificantBits)
                .putLong(wrapped.leastSignificantBits)
               .array()
        )

Rather than convert a UUID to a byte array both for size calculation and for serialization (which is what a naïve implementation would do), UuidConverter always returns the size of a constant 16-byte array.

If the wrapper type is in the same package as the generated protobuf message, then it does not need a fully-qualified name. Custom wrapper type converters can be in the same project as protobuf types that reference them. In order to use any wrapper type defined in extensions, the project must be included as a dependency:

dependencies {
    protoktExtensions("com.toasttab.protokt:protokt-extensions:<version>")
}

Wrapper types that wrap protobuf messages are nullable. For example, java.time.Instant wraps the well-known type google.protobuf.Timestamp. They can be made non-nullable by using the non-null option described below.

Wrapper types that wrap protobuf primitives, for example java.util.UUID which wraps bytes, are not nullable and may present malformed inputs to converters when absent in deserialization. It is up to the converter to determine what behavior should be in these cases. To represent nullable primitive wrappers use well-known types or Proto3's optional. For example for a nullable UUID:

google.protobuf.BytesValue uuid = 1 [
  (protokt.v1.property).wrap = "java.util.UUID"
];

// or:
optional bytes optional_uuid = 2 [
  (protokt.v1.property).wrap = "java.util.UUID"
];

Wrapper types can be repeated:

repeated bytes uuid = 1 [
  (protokt.v1.property).wrap = "java.util.UUID"
];

And they can also be used for map keys and values:

map<string, protokt.ext.InetSocketAddress> map_string_socket_address = 1 [
  (protokt.v1.property).key_wrap = "StringBox",
  (protokt.v1.property).value_wrap = "java.net.InetSocketAddress"
];

Wrapper types should be immutable. If a wrapper type is defined in the same package as generated protobuf message that uses it, then it does not need to be referenced by its fully-qualified name and instead can be referenced by its simple name, as done with StringBox in the map example above.

N.b. Well-known type nullability is implemented with predefined wrapper types for each message defined in wrappers.proto.

Non-null fields

If a message has no meaning whatsoever when a particular non-scalar field is missing, you can emulate proto2's required key word by using the (protokt.v1.property).non_null option:

message Sample {}

message NonNullSampleMessage {
  Sample non_null_sample = 1 [
    (protokt.v1.property).non_null = true
  ];
}

Generated code will not have a nullable type, so the field can be referenced without using Kotlin's !!.

Oneof fields can also be declared non-null:

message NonNullSampleMessage {
  oneof non_null_oneof {
    option (protokt.v1.oneof).non_null = true;

    string message = 1;
  }
}

Note that deserialization of a message with a non-nullable field will fail if the message being decoded does not contain an instance of the required field.

Interface implementation

Messages

To avoid the need to create domain-specific objects from protobuf messages you can declare that a protobuf message implements a custom interface with properties and default methods.

package com.protokt.sample

interface Model {
    val id: String
}
package protokt.sample;

message ImplementsSampleMessage {
  option (protokt.v1.class).implements = "Model";

  string id = 1;
}

Like wrapper types, if the implemented interface is in the same package as the generated protobuf message that uses it, then it does not need to be referenced by its fully-qualified name. Implemented interfaces cannot be used by protobuf messages in the same project that defines them; the dependency must be declared with protoktExtensions in build.gradle:

dependencies {
    protoktExtensions(project(":api-project"))
}

Messages can also implement interfaces by delegation to one of their fields; in this case the delegated interface need not live in a separate project, as protokt requires no inspection of it:

message ImplementsWithDelegate {
  option (protokt.v1.class).implements = "Model2 by modelTwo";

  ImplementsModel2 model_two = 1 [
    (protokt.v1.property).non_null = true
  ];
}

Note that the by clause references the field by its lower camel case name.

Oneof Fields

Oneof fields can declare that they implement an interface with the (protokt.v1.oneof).implements option. Each possible field type of the oneof must also implement the interface. This allows access of common properties without a when statement that always ultimately extracts the same property.

Suppose you have a domain object MyObjectWithConfig that has a non-null configuration that specifies a third-party server for communication. For flexibility, this configuration will be modifiable by the server and versioned by a simple integer. To hasten subsequent loading of the configuration, a client may save a resolved version of the configuration with the same version and an additional field storing an InetAddress representing the location of the server. Since the server address may change over time, the client-resolved version of the config will retain a copy of the original server copy. We can model this domain with protokt:

Given the Config interface:

package com.toasttab.example

interface Config {
    val version: Int
}

And protobuf definitions:

syntax = "proto3";

package toasttab.example;

import "protokt/v1/protokt.proto";

message MyObjectWithConfig {
  bytes id = 1 [
    (protokt.v1.property).wrap = "java.util.UUID"
  ];

  oneof Config {
    option (protokt.v1.oneof).non_null = true;
    option (protokt.v1.oneof).implements = "com.toasttab.example.Config";

    ServerSpecified server_specified = 2;
    ClientResolved client_resolved = 3;
  }
}

message ServerSpecified {
  option (protokt.v1.class).implements = "Config";

  int32 version = 1;

  string server_registry = 2;
  string server_name = 3;
}

message ClientResolved {
  option (protokt.v1.class).implements = "Config by config";

  ServerSpecified config = 1 [
    (protokt.v1.property).non_null = true
  ];

  bytes last_known_address = 2 [
    (protokt.v1.property).wrap = "java.net.InetAddress"
  ];
}

Protokt will generate:

@V1KtGeneratedMessage("toasttab.example.MyObjectWithConfig")
@RtKtGeneratedMessage("toasttab.example.MyObjectWithConfig")
public class MyObjectWithConfig private constructor(
  public val id: UUID,
  public val config: Config,
  public val unknownFields: UnknownFieldSet = UnknownFieldSet.empty()
) : AbstractKtMessage() {
  // methods omitted

  public sealed class Config : ExampleConfig {
    public data class ServerSpecified(
      public val serverSpecified: protokt.v1.toasttab.example.ServerSpecified
    ) : Config(), ExampleConfig by serverSpecified

    public data class ClientResolved(
      public val clientResolved: protokt.v1.toasttab.example.ClientResolved
    ) : Config(), ExampleConfig by clientResolved
  }

  @V1KtGeneratedMessage("toasttab.example.ServerSpecified")
  @RtKtGeneratedMessage("toasttab.example.ServerSpecified")
  public class ServerSpecified private constructor(
    override val version: Int,
    public val serverRegistry: String,
    public val serverName: String,
    public val unknownFields: UnknownFieldSet = UnknownFieldSet.empty()
  ) : AbstractKtMessage(), ExampleConfig {
    // methods omitted
  }


  @V1KtGeneratedMessage("toasttab.example.ClientResolved")
  @RtKtGeneratedMessage("toasttab.example.ClientResolved")
  public class ClientResolved private constructor(
    public val config: ServerSpecified,
    public val lastKnownAddress: InetAddress,
    public val unknownFields: UnknownFieldSet = UnknownFieldSet.empty()
  ) : AbstractKtMessage(), Config by config {
    // methods omitted
  }

A MyObjectWithConfig.Config instance can be queried for its version without accessing the property via a when expression:

fun printVersion(config: MyObjectWithConfig.Config) {
    println(config.version)
}

This pattern can be dangerous: since the oneof must be marked non-nullable, you cannot compatibly add new implementing fields to a producer before a consumer is updated with the new generated code. The old consumer will attempt to deserialize the new field as an unknown field and the non-null assertion on the oneof field during the constructor call will fail.

BytesSlice

When reading messages that contain other serialized messages as bytes fields, protokt can keep a reference to the originating byte array to prevent a large copy operation on deserialization. This can be desirable when the wrapping message is short-lived or a thin metadata shim and doesn't include much memory overhead:

message SliceModel {
  int64 version = 1;

  bytes encoded_message = 2 [
    (protokt.v1.property).bytes_slice = true
  ];
}

gRPC code generation

Protokt will generate variations of code for gRPC method and service descriptors when the gRPC generation options are enabled:

protokt {
  generate {
      grpcDescriptors = true
      grpcKotlinStubs = true
  } 
}

The options can be enabled independently of each other.

Generated gRPC code

grpcDescriptors

Consider gRPC's canonical Health service:

syntax = "proto3";

package grpc.health.v1;

message HealthCheckRequest {
  string service = 1;
}

message HealthCheckResponse {
  enum ServingStatus {
    UNKNOWN = 0;
    SERVING = 1;
    NOT_SERVING = 2;
  }
  ServingStatus status = 1;
}

service Health {
  rpc Check(HealthCheckRequest) returns (HealthCheckResponse);
}

In addition to the request and response types, protokt will generate a service descriptor and method descriptors for each method on the service:

public object HealthGrpc {
  public const val SERVICE_NAME: String = "grpc.health.v1.Health"

  private val _serviceDescriptor: GrpcServiceDescriptor by lazy {
    serviceDescriptorNewBuilder(SERVICE_NAME)
      .addMethod(_checkMethod)
      .setSchemaDescriptor(
        SchemaDescriptor(
          className = "protokt.v1.grpc.health.v1.Health",
          fileDescriptorClassName = "protokt.v1.grpc.health.v1.health_file_descriptor"
        )
      )
      .build()
  }

  private val _checkMethod: MethodDescriptor<HealthCheckRequest, HealthCheckResponse> by lazy {
    methodDescriptorNewBuilder<HealthCheckRequest, HealthCheckResponse>()
      .setType(UNARY)
      .setFullMethodName(generateFullMethodName(SERVICE_NAME, "Check"))
      .setRequestMarshaller(KtMarshaller(HealthCheckRequest))
      .setResponseMarshaller(KtMarshaller(HealthCheckResponse))
      .build()
  }

  @JvmStatic
  public fun getServiceDescriptor(): GrpcServiceDescriptor = _serviceDescriptor

  @JvmStatic
  public fun getCheckMethod(): MethodDescriptor<HealthCheckRequest, HealthCheckResponse> = _checkMethod
}

Both grpc-java and grpc-kotlin expose server stubs for implementation via abstract classes.

grpcKotlinStubs and gRPC's Kotlin API

Protokt uses grpc-kotlin to generate Kotlin coroutine-based stubs that compile against protokt's generated types.

Integrating with gRPC's Java API

A gRPC service using grpc-java (and therefore using StreamObservers for asynchronous communication):

abstract class HealthCheckService : BindableService {
    override fun bindService() =
        ServerServiceDefinition.builder(serviceDescriptor)
            .addMethod(checkMethod, asyncUnaryCall(::check))
            .build()

    open fun check(
        request: HealthCheckRequest,
        responseObserver: StreamObserver<HealthCheckResponse>
    ): Unit =
        throw UNIMPLEMENTED.asException()
}

Calling methods from a client:

fun checkHealth(): HealthCheckResponse =
    ClientCalls.blockingUnaryCall(
        channel.newCall(HealthGrpc.checkMethod, CallOptions.DEFAULT),
        HealthCheckRequest { service = "foo" }
    )

Integrating with gRPC's NodeJS API

Protokt generates complete server and client stub implementations for use with NodeJS. The generated implementations are nearly the same as those generated by grpc-kotlin and are supported by an analogous runtime library in ServerCalls and ClientCalls objects.

These implementations are alpha-quality and for demonstration only. External contributions to harden the implementation are welcome. They use the same grpcDescriptors and grpcKotlinStubs plugin options to control code generation.

IntelliJ integration

If IntelliJ doesn't automatically detect the generated files as source files, you may be missing the idea plugin. Apply the idea plugin to your Gradle project:

plugins {
  id 'idea'
}

Command line code generation

protokt % ./gradlew assemble

protokt % protoc \
    --plugin=protoc-gen-custom=protokt-codegen/build/install/protoc-gen-protokt/bin/protoc-gen-protokt \
    --custom_out=<output-directory> \
    -I<path-to-proto-file-containing-directory> \
    -Iprotokt-runtime/src/main/resources \
    <path-to-proto-file>.proto

For example, to generate files in protokt/foo from a file called test.proto located at protokt/test.proto:

protokt % protoc \
    --plugin=protoc-gen-custom=protokt-codegen/build/install/protoc-gen-protokt/bin/protoc-gen-protokt \
    --custom_out=foo \
    -I. \
    -Iprotokt-runtime/src/main/resources \
    test.proto

Contribution

Community contributions are welcome. See the contribution guidelines and the project code of conduct.

To enable rapid development of the code generator, the protobuf conformance tests have been compiled and included in the testing project. They run on Mac OS 10.14+ and Ubuntu 16.04 x86-64 as part of normal Gradle builds.

When integration testing the Gradle plugin, note that after changing the plugin and republishing it to the integration repository, ./gradlew clean is needed to trigger regeneration of the protobuf files with the fresh plugin.

Acknowledgements

Authors

Ben Gordon, Andrew Parmet, Oleg Golberg, Frank Moda, Romey Sklar, and everyone in the commit history.

Thanks to the Google Kotlin team for their Kotlin API Design which inspired the DSL builder implemented in this library.