Type Driven BERT Parser Generator

Supported Languages

• JavaScript
• Swift

Contest Description

As you may know, Erlang has its own binary encoding BERT inside its virtual machine, called BEAM. For enterprise RPC usually, you use protobuf or MessagePack or Thrift or ASN.1 binary parser generators. However, as you may know Erlang is not so fast in any tasks except moving binaries between sockets. So we at Synrc usually use native Erlang BERT encoding on all clients with zero encoding/decoding on server side.

The encoders/decoders could be of two types: strict (with checking the model for particular type signature with Sums and Products) and general which encode/decode anything that can be translated into correct encoding. For example, JavaScript encoder/decode generator presented in this repo is just like that (it didn't check types and constants, given in Erlang HRL files). However, Swift version has the ability to check encoded/decoded term to comply the Erlang Type Specification.

The aim of this contest is to create encoders/decoders for each language!

Rules

First, you can choose the language which is not presented in this repo and try to implement your own BERT enc/dec generator for this language using Swift (Type Spec precise) and JavaScript (open relay) generator as examples.

If you think you can deliver clean first class code and you should be paid for this task please inform me about your lowest price you'll be satisfied to maxim@synrc.com. You will be chosen!

Erlang Model

-record(error, { code=[] :: [] | binary() }).

-record(ok, { code=[] :: [] | binary() }).

-record(io, { code=[] :: [] | #ok{} | #error{},
              data=[] :: [] | <<>> | { atom(), binary() | integer() } }).

Swift Model

class Err {
    var code: AnyObject?
}

class Ok {
    var code: AnyObject?
}

class Io {
    var code: AnyObject?
    var data: AnyObject?
}

Specification

Chain(types: [
     Model(value:Tuple(name: "io", body: [
          Model(value:Chain(types: [
               Model(value: Tuple(name:"ok", body: [
                    Model(value:Atom())])),
               Model(value: Tuple(name:"error", body: [
                    Model(value:Atom())]))])),
          Model(value:Tuple(name:"", body:[
               Model(value:Atom()),
               Model(value:Chain(types: [
                    Model(value:Binary()),
                    Model(value:Number())]))]))])) ])

JavaScript

function check() {
    var res = true;
    //@TODO: MORE TEST DATA
    testData = [
        1,
        [1, 2, 3],
        "string",
        {tup: 'io', code: 'login', data: {tup: '$', 0: 'Auth', 1: 12}},
        {tup: 'io', code: 'login', data: {tup: 'Auth'}},
        {tup: 'io', code: 'login', data: {tup: '$', 0: 'doe', 1: 12}},
        {tup: 'Roster', userlist: [{tup: 'Contact'}], status: 'get'},
        {tup: 'p2p', from: 'john', to: 'doe'},
        {tup: 'Profile', accounts: [1], status: 'maxim'}
    ];
    testData.forEach(function (o) {
        var o = JSON.stringify(o);
        var d = JSON.stringify(decode(dec(enc(encode(o)).buffer))).replace(/\\/g, '');

        if (JSON.stringify(o) != JSON.stringify(decode(dec(enc(encode(o)).buffer)))) {
            console.log("Original: " + o + " <=> Decode: " + d + " %c [Error]", "color: red");
            res = false;
        } else {
            console.log("Data: " + o + " %c [OK]", "color: green");
        }
    });

    return res;
}

Run

$ rebar compile

Credits

• Anton Makarov
• Viacheslav Katsuba
• Maxim Sokhatsky

Let's build true native Erlang zero-overhead encoders/decoders for all languages with you!

OM A HUM