Protospark is a way to serialize/deserialize protocol buffer messages easily in a type safe manner. This module is an abstraction on top of the protobufjs

npm

> npm install protospark

const protospark = require("protospark");

Please note: This package does not include a feature to generate JavaScript files from Protocol Buffer (.proto) files. For this purpose, you should utilize the protobufjs-cli package.

• Predictive encoding and decoding
• • With this approach, we accept a slight performance overhead to facilitate easier type inference.
• On-the-fly generation of proto/JS files from schema definitions defined in code.

Notes:

• for protospark to infer the type correctly, every message must contain a 'type' property.
• encoding by default verifying the schema, to disable it, supply the { verify: false } option for the encoder.

// message.proto
message MyAwesomeMessage {
  string type = 1;
  string some_field = 2;
}

Important first step is to compile the proto file using the protobufjs-cli, see: protobufjs-cli

Define the javascript entity in code as well:

class MyAwesomeMessage {
  type;
  someField;
}

The method above is somewhat impractical as it requires defining the same schema twice, but it remains a viable option. Alternatively, leveraging the new 2.0 features could automate the generation of message types.

const protospark = require("protospark");
const options = { ... options here ... };
const codec = protospark.ProtosparkCodec.fromPath(
  "some/directory/to/proto/files",
  options
);
const buffer = codec
  .encode({
    type: "MyAwesomeMessage",
    someField: "Hello World",
  })
  .execute()
  .finish();
const decoded = codec.decode(buffer).to(MyAwesomeMessage);

Protospark is able to compile schema from string as well:

const protospark = require("protospark");
const options = { ... options here ... };
const codec = protospark.ProtosparkCodec.fromString(
  `syntax="proto3";

   message MyAwesomeMessage {
    string type = 1;
    string some_field = 2;
   }`,
  options
);
const buffer = codec
  .encode({
    type: "MyAwesomeMessage",
    someField: "Hello World",
  })
  .execute()
  .finish();
const decoded = codec.decode(buffer).to(MyAwesomeMessage);

This functionality allows the creation of protobuf schemas using JavaScript entities, introducing inheritance which is not inherent to the protocol buffer standard.

!!!IMPORTANT!!! Recommended to use with TypeScript

To utilize this feature, please keep in mind the following restrictions:

• Every entity must extend the ProtosparkSchemaDefinition class.
• Defining empty schemas are not permitted.
• Redefining properties within schemas is prohibited.

1. Begin with a non-empty schema definition:

The structure of the definition is recommended to be the following:

class <SchemaName>SchemaDefinition extends ProtosparkSchemaDefinition {
    <parameterName>[: ProtosparkSchemaTypeDescription] = { type: ProtosparkSchemaPropertyType.<typeName> }
}

Where:

• <SchemaName>: The name of the desired message schema
• <parameterName>: The parameter name in camelCase
• <typeName>: Property of ProtosparkSchemaPropertyType

Example:

class MySchemaDefinition extends ProtosparkSchemaDefinition {
  someProperty = { type: ProtosparkSchemaPropertyType.bool };
  // .. add more properties
}

class MySubSchemaDefinition extends MySchemaDefinition {
  additionalProperty = { type: ProtosparkSchemaPropertyType.string };
  // add more properties
}

Note: ProtosparkSchemaPropertyType aims to offer support for a broad set of types, although it is still a work in progress and undergoes continuous development.

2. Define the schema:

const MySchema = ProtosparkSchema.defineMultiple([
  MySchemaDefinition,
  MySubSchemaDefinition,
]);

This schema will contain the "someProperty"

3. Generate the proto file from the defined schemas:

// pass all the defined schemas here that you wish in one file.
const options = {
  format: "ts", // this is the default, use "ts" for typescript generation
};

const environment = protospark.ProtosparkSchemaGenerator.generate(MySchema);

// do anything with the file.. for example load in memory:
const inMemoryCodec = ProtosparkCodec.fromProtoFile(environment.file);

// Or compile it into javascript to use the message types:
// !!!!IMPORTANT: Setup generation in a different script
environment.write("./generated", "test", options);

import { MyMessage } from "./generated/test";
const encoded = inMemoryCodec
  .encode({
    type: MyMessage.name,
    someProperty: true,
  })
  .execute()
  .finish();

const decoded = inMemoryCodec.decode(encoded).to(MyMessage);
console.log(decoded); // MyMessage { type: 'MyMessage', someProperty: true }

// Alternatively, to save only the proto file:
// environment.file.write(ProtoFileOutDir, fileNameWithProtoExtension);

const regularCodec = protospark.ProtosparkCodec.fromPath(ProtoFileOutDir);

!!!IMPORTANT!!! Prior to utilizing the generated files, ensure the generation process has been completed. It's advisable to establish a distinct generation script that can be executed preceding the application's execution.

If you're seeking a method to bypass specifying the output type for the decode method, you can accomplish this using the new predictive decoding feature:

// ...
// Create schemas from the definitions
const MySchemas = ProtosparkSchema.defineMultiple([
  TestSchemaDefinition,
  Test2SchemaDefinition,
]);
// Generate the environment from the schemas
const environment = ProtosparkSchemaGenerator.generate(MySchemas);
// encode something with the predictive algorithm, { verify: true } is default by the way
const encoded = codec.encode(message, { verify: true }).predictive();
// decode in a predictive way, supply the enviroment to the encoder to help infering the types
const decoded = codec.decode(encoded).predictive(environment);
// ...

This feature is having the following limitations:

• The initial 32 bytes of the buffer must represent the type parameter in hexadecimal format.
• The length of the type parameter cannot exceed 32 bytes, which should provide ample space for this parameter.
• Expect an overhead increase, although ignorable, it's important to acknowledge.

You can provide these options when creating the codec: protospark.ProtosparkCodec.fromPath(path, options)

The package continues to evolve over time. If you find yourself in need of additional features, you can harness the complete functionality of the protobufjs schema with codec.rawSchema().