@rocket.chat/apps-engine
Version:
The engine code for the Rocket.Chat Apps which manages, runs, translates, coordinates and all of that.
42 lines • 1.76 kB
JavaScript
;
Object.defineProperty(exports, "__esModule", { value: true });
exports.newDecoder = exports.newEncoder = void 0;
const msgpack_1 = require("@msgpack/msgpack");
const extensionCodec = new msgpack_1.ExtensionCodec();
extensionCodec.register({
type: 0,
encode: (object) => {
// We don't care about functions, but also don't want to throw an error
if (typeof object === 'function') {
return new Uint8Array([0]);
}
},
decode: (_data) => undefined,
});
// We need to handle Buffers because Deno needs its own decoding
extensionCodec.register({
type: 1,
encode: (object) => {
if (object instanceof Buffer) {
return new Uint8Array(object.buffer, object.byteOffset, object.byteLength);
}
},
// msgpack will reuse the Uint8Array instance, so WE NEED to copy it instead of simply creating a view
decode: (data) => Buffer.from(data),
});
/**
* The Encoder and Decoder classes perform "stateful" operations, i.e. they read from a
* stream, store the data locally and decode it from its buffer.
*
* In practice, this affects the decoder when there is decode error. After an error, the decoder
* keeps the malformed data in its buffer, and even if we try to decode from another source (e.g. different stream)
* it will fail again as there's still data in the buffer.
*
* For that reason, we can't have a singleton instance of Encoder and Decoder, but rather one
* instance for each time we create a new subprocess
*/
const newEncoder = () => new msgpack_1.Encoder({ extensionCodec });
exports.newEncoder = newEncoder;
const newDecoder = () => new msgpack_1.Decoder({ extensionCodec });
exports.newDecoder = newDecoder;
//# sourceMappingURL=codec.js.map