protocol-buffers

protocol-buffers

Protocol Buffers for Node.js

npm install protocol-buffers

Usage

Assuming the following test.proto file exists

enum FOO {
	BAR = 1;
}

message Test {
  required float num  = 1;
  required string payload = 2;
}

message AnotherOne {
	repeated FOO list = 1;
}

Use the above proto file to encode/decode messages by doing

var protobuf = require('protocol-buffers')

// pass a proto file as a buffer/string or pass a parsed protobuf-schema object
var messages = protobuf(fs.readFileSync('test.proto'))

var buf = messages.Test.encode({
	num: 42,
	payload: 'hello world'
});

console.log(buf); // should print a buffer

To decode a message use Test.decode

var obj = messages.Test.decode(buf);
console.log(obj); // should print an object similar to above

You can also use protocol-buffers/require to require .proto files from disk instead of passing them as buffers.

var protobuf = require('protocol-buffers/require')
var messages = protobuf('test.proto') // will load and parse __dirname/test.proto

Enums are accessed in the same way as messages

var buf = messages.AnotherOne.encode({
	list: [
		messages.FOO.BAR
	]
})

See the Google Protocol Buffers docs for more information about the available types etc.

Performance

This module is fast.

It uses code generation to build as fast as possible encoders/decoders for the protobuf schema. You can run the benchmarks yourself by doing npm run bench.

On my Macbook Air it gives the following results

Benchmarking JSON (baseline)
  Running object encoding benchmark...
  Encoded 1000000 objects in 2142 ms (466853 enc/s)

  Running object decoding benchmark...
  Decoded 1000000 objects in 970 ms (1030928 dec/s)

  Running object encoding+decoding benchmark...
  Encoded+decoded 1000000 objects in 3131 ms (319387 enc+dec/s)

Benchmarking protocol-buffers
  Running object encoding benchmark...
  Encoded 1000000 objects in 2089 ms (478698 enc/s)

  Running object decoding benchmark...
  Decoded 1000000 objects in 735 ms (1360544 dec/s)

  Running object encoding+decoding benchmark...
  Encoded+decoded 1000000 objects in 2826 ms (353857 enc+dec/s)

Note that JSON parsing/serialization in node is a native function that is really fast.

License

MIT