Type-safe error handling without exception


Keywords
aws, serverless, lambda, error, error-handling, functional-programming, nodejs, type-safety, typescript
License
MIT
Install
npm install lambda-res@1.0.24

Documentation

Type-safe error handling without exceptions

Flexible and explicit error handling with a small flavor of functional languages.

TypeScript Test Coverage Status

Result type allows you go from this:

const user = getUser(email);

if (!user) {
  throw new Error("Can't get user"); 
  // but what happens, why email is invalid?
  // and who will catch this error?
}

to this:

const userResult = getUser(email);

if (userResult.isErr()) {
  const err = userResult.err();

  switch (err.type) {
    case 'InvalidEmail':
      console.log('User email is invalid');
      break;
    case 'UserNotExists':
      console.log("Can't find the user");
      break;
    case 'UserBannded':
      console.log(`User was bannded: ${err.ban}`);
      break;
  }
}

Usage:

Creating result

There are two main function for creating results ok and fail.

import { Err, Result, ok, fail } from 'lambda-res';

const success = ok('good'); // inferred as Success<string>

const error = fail<Err>('bad'); // inferred as Failure<string>

const result: Result<string, Err> = Math.random() ? success : error;

The Err and Errs type

Err used for creating errors of given type.

import { Failure, Err, fail, nope } from 'lambda-res';

// same as: { type: 'AppDatabaseError' } & Err
type DatabaseError = Err<'AppDatabaseError'>; 
type ConnectionError = Err<'AppConnectionError'>;

type AppErrors = DatabaseError | ConnectionError; // | Err<'NotHandled'>;

const check = (err: AppErrors): string => {
  if (err.type === 'AppDatabaseError') {
    return 'Database error';
  }

  if (err.type === 'AppConnectionError') {
    return 'Connection error';
  }

  // try unncoment NotHandled error and typescript will show a problem here!
  // that ensures that all error types are checked exhaustively
  nope(err);
};

// must match the types or Typescript will show an error
const failure: Failure<AppErrors> = fail<DatabaseError>('AppDatabaseError');

check(failure.err()); // get en error from failure

Errs type can be used to create multiple errors of a given kind, which may be useful in order not to pollute your exports with multiple individual errors (like: DatabaseError, ConnectionError).

import { Failure, Errs, fail } from 'lambda-res';

// only AppError needs to be exported
type AppError = Errs<{
  name: 'App';
  DatabaseError: string;
  ConnectionError: string;
}>;

// note an Errs here (inside Failure): 
// it transform AppError into Err<'AppDatabaseError'> | Err<'AppConnectionError'>
const failure: Failure<Errs<AppError>> = fail<AppError['DatabaseError']>('AppDatabaseError');

Using result

Once you created result it have a bunch of useful methods:

isOk, isErr, ok, err

import { Err, Result, ok, fail } from 'lambda-res';

const result: Result<string, Err> = Math.random() ? ok('ok') : fail<Err<'Error1'>>('Error1');

// isOk method allows to narrow result type to Success<string>
if (result.isOk()) {
  // once type is narrowed you can safely access the data
  console.log(result.ok()); // will outputs 'ok'
}

// isErr method allows to narrow result type to Failure<Err<'Error1'>>
if (result.isErr()) {
  console.log(result.err().type);
}

// result.err();
// ^
// this will throw an exception because result is not a Failure
// that is why isErr check used first

Mapping result with onOk and onErr (onFail)

import { Response, Errs, ok, fail, nope } from 'lambda-res';

type GetUserEmailError = Errs<{
  name: 'GetUserEmail';
  NotFound: string;
  Deleted: string;
}>;

const getUserName = async (id: string): Response<string, Errs<GetUserEmailError>> => {
  if (id === 'id1') {
    return ok('email@example.com');
  } else if (id === 'id2') {
    return fail('GetUserEmailDeleted');
  }
  return fail('GetUserEmailNotFound');
};

type SendEmailError = Errs<{
  name: 'SendEmail';
  DeliveryError: string;
}>;

const sendEmail = async (email: string): Response<{ sentCount: number }, Errs<SendEmailError>> => {
  if (email === 'email@example.com') {
    return ok({ sentCount: 3 })
  }
  return fail('SendEmailDeliveryError')
}

const sendEmailToUser = async () /*: Response<{ sentCount: number }, Err<'DeliveryError'> | Err<'LoadingUserError'>> */ => {
  const userNameRes = await getUserName('id1');

  return userNameRes.onOk(async (email) => {
    const sendRes = await sendEmail(email); 

    return sendRes;
  }).onErr((err) => {
    if (err.type === 'SendEmailDeliveryError') {
      return { type: 'DeliveryError' }
    } else if (err.type === 'GetUserEmailDeleted' || err.type === 'GetUserEmailNotFound') {
      return { type: 'LoadingUserError'  }
    }
    nope(err);
  }).res();
}

Result<Data, Fail>.onOk

On success return new Data2 and collects all errors. Used to sequentially run multiple functions which returns result. In the end onErr can map errors to desired error type.

Params:

  • cb: callback which receives

    • data: Data
    • res: Result<Data, never>
  • name (optional) : common error name, it will prepend all errors types

Returns:

  • Result<Data2, Fail | Fail2> or Response<Data2, Fail | Fail2> (for async functions)

Result<Data, Fail>.onErr, Result<Data, Fail>.onFail

Doesn't change Data and returns new Fail2 — used to map one type of errors to another.

Params:

  • cb: callback which receives
    • err: Fail
    • res: Result<never, Fail>

Returns:

  • onErr: string, { type: string } or Promise<string>, Promise<{ type: string }>
  • onFail: Result<never, Fail2> or Response<never, Fail2>