A production-grade error creation and serialization library designed for Typescript
A production-grade error creation library designed for Typescript. Useful for direct printing of errors to a client or for internal development / logs.
Error with additional methods added on.
The basic Javascript Error type is extremely bare bones - you can only specify a message.
In a production-level application, I've experienced the following use-cases:
new-error was built with these use-cases in mind.
$ npm i new-error --save
The error registry is the fastest way to define and create errors.
// This is a working example
import { ErrorRegistry } from 'new-error'
// Define high level errors
// Do *not* assign a Typescript type to the object
// or IDE autocompletion will not work!
const errors = {
INTERNAL_SERVER_ERROR: {
/**
* The class name of the generated error
*/
className: 'InternalServerError',
/**
* A user-friendly code to show to a client.
*/
code: 'ERR_INT_500',
/**
* (optional) Protocol-specific status code, such as an HTTP status code. Used as the
* default if a Low Level Error status code is not defined.
*/
statusCode: 500,
/**
* (optional) Log level string / number to associate with this error.
* Useful if you want to use your logging system to log the error but
* assign a different log level for it. Used as the default if a
* Low Level log level is not defined.
*/
logLevel: 'error'
}
}
// Define low-level errors
// Do *not* assign a Typescript type to the object
// or IDE autocompletion will not work!
const errorCodes = {
// 'type' of error
DATABASE_FAILURE: {
/**
* Full description of the error. sprintf() flags can be applied
* to customize it.
* @see https://www.npmjs.com/package/sprintf-js
*/
message: 'There was a database failure, SQL err code %s',
/**
* (optional) A user-friendly code to show to a client.
*/
subCode: 'DB_0001',
/**
* (optional) Protocol-specific status code, such as an HTTP status code.
*/
statusCode: 500,
/**
* (optional) Log level string / number to associate with this error.
* Useful if you want to use your logging system to log the error but
* assign a different log level for it.
*/
logLevel: 'error'
}
}
// Create the error registry by registering your errors and codes
// you will want to memoize this as you will be using the
// reference throughout your application
const errRegistry = new ErrorRegistry(errors, errorCodes)
// Create an instance of InternalServerError
// Typescript autocomplete should show the available definitions as you type the error names
// and type check will ensure that the values are valid
const err = errRegistry.newError('INTERNAL_SERVER_ERROR', 'DATABASE_FAILURE')
.setErrorId('err-1234')
.formatMessage('SQL_1234')
console.log(err.toJSON())
Produces:
(You can omit fields you do not need - see usage section below.)
{
errId: 'err-1234',
name: 'InternalServerError',
code: 'ERR_INT_500',
message: 'There was a database failure, SQL err code SQL_1234',
type: 'DATABASE_FAILURE',
subCode: 'DB_0001',
statusCode: 500,
meta: {},
stack: 'InternalServerError: There was a database failure, SQL err code %s\n' +
' at ErrorRegistry.newError (new-error/src/ErrorRegistry.ts:128:12)\n' +
' at Object.<anonymous> (new-error/src/test.ts:55:25)\n' +
' at Module._compile (internal/modules/cjs/loader.js:1158:30)\n' +
' at Module._compile (new-error/node_modules/source-map-support/source-map-support.js:541:25)\n' +
' at Module.m._compile (/private/var/folders/mx/b54hc2lj3fbfsndkv4xmz8d80000gn/T/ts-node-dev-hook-20649714243977457.js:57:25)\n' +
' at Module._extensions..js (internal/modules/cjs/loader.js:1178:10)\n' +
' at require.extensions.<computed> (/private/var/folders/mx/b54hc2lj3fbfsndkv4xmz8d80000gn/T/ts-node-dev-hook-20649714243977457.js:59:14)\n' +
' at Object.nodeDevHook [as .ts] (new-error/node_modules/ts-node-dev/lib/hook.js:61:7)\n' +
' at Module.load (internal/modules/cjs/loader.js:1002:32)\n' +
' at Function.Module._load (internal/modules/cjs/loader.js:901:14)'
}
You can create concrete error classes by extending the BaseRegistryError class, which
extends the BaseError class.
The registry example can be also written as:
import { BaseRegistryError, LowLevelError } from 'new-error'
class InternalServerError extends BaseRegistryError {
constructor (errDef: LowLevelError) {
super({
code: 'ERR_INT_500',
statusCode: 500
}, errDef)
}
}
const err = new InternalServerError({
type: 'DATABASE_FAILURE',
message: 'There was a database failure, SQL err code %s',
subCode: 'DB_0001',
statusCode: 500,
logLevel: 'error'
})
console.log(err.formatMessage('SQL_1234').toJSON())
If you want a native-style Error, you can use BaseError.
The registry example can be written as:
import { BaseError } from 'new-error'
class InternalServerError extends BaseError {}
const err = new InternalServerError('There was a database failure, SQL err code %s')
// calling these methods are optional
.withErrorType('DATABASE_FAILURE')
.withErrorCode('ERR_INT_500')
.withErrorSubCode('DB_0001')
.withStatusCode(500)
.withLogLevel('error')
console.log(err.formatMessage('SQL_1234').toJSON())
import express from 'express'
import { ErrorRegistry, BaseError } from 'new-error'
const app = express()
const port = 3000
const errors = {
INTERNAL_SERVER_ERROR: {
className: 'InternalServerError',
code: 'ERR_INT_500',
statusCode: 500
}
}
const errorCodes = {
DATABASE_FAILURE: {
message: 'There was a database failure.',
subCode: 'DB_0001',
statusCode: 500
}
}
const errRegistry = new ErrorRegistry(errors, errorCodes)
// middleware definition
app.get('/', async (req, res, next) => {
try {
// simulate a failure
throw new Error('SQL issue')
} catch (e) {
const err = errRegistry.newError('INTERNAL_SERVER_ERROR', 'DATABASE_FAILURE')
err.causedBy(err)
// errors must be passed to next()
// to be caught when using an async middleware
return next(err)
}
})
// catch errors
app.use((err, req, res, next) => {
// error was sent from middleware
if (err) {
// check if the error is a generated one
if (err instanceof BaseError) {
// generate an error id
// you'll want to use a library like 'nanoid' instead
// this is just an example
err.withErrorId(Math.random().toString(36).slice(2))
// log the error
// the "null, 2" options formats the error into a readable structure
console.error(JSON.stringify(err.toJSON(), null, 2))
// get the status code, if the status code is not defined, default to 500
res.status(err.getStatusCode() ?? 500)
// spit out the error to the client
return res.json({
err: err.toJSONSafe()
})
}
// You'll need to modify code below to best fit your use-case
// err.message could potentially expose system internals
return res.json({
err: {
message: err.message
}
})
}
// no error, proceed
next()
})
app.listen(port, () => console.log(`Example app listening at http://localhost:${port}`))
If you visit http://localhost:3000, you'll get a 500 status code, and the following response:
{"err": {"errId": "xd0v1szkziq", code":"ERR_INT_500","subCode":"DB_0001","statusCode":500,"meta":{}}}
You might want to use a different log level when logging common errors, such as validation errors.
import { ErrorRegistry } from 'new-error'
const errors = {
VALIDATION_ERROR: {
className: 'ValidationError',
code: 'VALIDATION_ERROR',
statusCode: 400,
// probably don't want to log every validation error
// in production since these errors tend to happen frequently
// and would pollute the logs
logLevel: 'debug'
}
}
const errorCodes = {
MISSING_FORM_FIELDS: {
message: 'Form submission data is missing fields',
subCode: 'MISSING_FORM_FIELDS',
statusCode: 400
}
}
const errRegistry = new ErrorRegistry(errors, errorCodes)
// some part of the application throws the error
const err = errRegistry.newError('VALIDATION_ERROR', 'MISSING_FORM_FIELDS')
// another part of the application catches the error
if (err.getLogLevel() === 'debug') {
console.debug(JSON.stringify(err.toJSON(), null, 2))
} else {
console.error(JSON.stringify(err.toJSON(), null, 2))
}
The ErrorRegistry is responsible for the registration and creation of errors.
new ErrorRegistry(highLvErrors, lowLvErrors, config = {})
interface IErrorRegistryConfig {
/**
* Options when creating a new BaseError
*/
baseErrorConfig?: IBaseErrorConfig
}
Example:
const errRegistry = new ErrorRegistry(errors, errorCodes, {
// Config for all BaseErrors created from the registry
baseErrorConfig: {
// Remove the `meta` field if there is no data present for `toJSON` / `toJSONSafe`
omitEmptyMetadata: true
}
})
Errors generated by the registry extends BaseError.
Method: ErrorRegistry#newError(highLevelErrorName, LowLevelErrorName)
This is the method you should generally use as you are forced to use your well-defined high and low level error definitions. This allows for consistency in how errors are defined and thrown.
// Creates an InternalServerError error with a DATABASE_FAILURE code and corresponding
// message and status code
const err = errRegistry.newError('INTERNAL_SERVER_ERROR', 'DATABASE_FAILURE')
Method: ErrorRegistry#newBareError(highLevelErrorName, message)
This method does not include a low level error code, and allows direct specification of an error message.
// Creates an InternalServerError error with a custom message
const err = errRegistry.newBareError('INTERNAL_SERVER_ERROR', 'An internal server error has occured.')
instanceOf / comparisonsMethod: ErrorRegistry#instanceOf(classInstance, highLevelErrorName)
Performs an instanceof operation against a custom error.
// creates an InternalServerError error instance
const err = errRegistry.newError('INTERNAL_SERVER_ERROR', 'DATABASE_FAILURE')
if (errRegistry.instanceOf(err, 'INTERNAL_SERVER_ERROR')) {
// resolves to true since err is an InternalServerError instance
}
instanceofYou can also check if the error is custom-built using this check:
import { BaseError } from 'new-error'
function handleError(err) {
if (err instanceof BaseError) {
// err is a custom error
}
}
Except for the getter and serialization methods, all other methods are chainable.
Generated errors extend the BaseError class, which supplies the manipulation methods.
new BaseError(message: string, config: IBaseErrorConfig: IBaseErrorConfig = {})
message: The error message you would use in new Error(message)interface IBaseErrorConfig {
/**
* A list of fields to always omit when calling toJSON
*/
toJSONFieldsToOmit?: string[]
/**
* A list of fields to always omit when calling toJSONSafe
*/
toJSONSafeFieldsToOmit?: string[]
/**
* If the metadata has no data defined, remove the `meta` property on `toJSON` / `toJSONSafe`.
*/
omitEmptyMetadata?: boolean
/**
* A function to run against the computed data when calling `toJSON`. This is called prior
* to field omission. If defined, must return the data back.
*/
onPreToJSONData?: (data: Partial<SerializedError>) => Partial<SerializedError>
/**
* A function to run against the computed safe data when calling `toJSONSafe`. This is called
* prior to field omission. If defined, must return the data back.
*/
onPreToJSONSafeData?: (data: Partial<SerializedErrorSafe>) => Partial<SerializedErrorSafe>
}
BaseError#getErrorId()BaseError#getErrorName()BaseError#getCode()BaseError#getErrorType()BaseError#getSubCode()BaseError#getStatusCode()BaseError#getCausedBy()BaseError#getMetadata()BaseError#getSafeMetadata()BaseError#getLogLevel()BaseError#getConfig()If you use the registry, you should not need to us these setters as the registry sets the values already.
BaseError#withErrorType(type: string): thisBaseError#withErrorCode(code: string | number): thisBaseError#withErrorSubCode(code: string | number): thisBaseError#withLogLevel(level: string | number): thisBaseError#setConfig(config: IBaseErrorConfig): voidstatic BaseError#fromJSON(data: object, options?: object): BaseErrorMethod: BaseError#withErrorId(errId: string)
Attaches an id to the error. Useful if you want to display an error id to a client / end-user and want to cross-reference that id in an internal logging system for easier troubleshooting.
For example, you might want to use nanoid to generate ids for errors.
import { nanoid } from 'nanoid'
err.withErrorId(nanoid())
// In your logging system, log the error, which will include the error id
logger.error(err.toJSON())
// expose the error to the client via err.toJSONSafe() or err.getErrorId(), which
// will also include the error id - an end-user can reference this id to
// support for troubleshooting
Method: BaseError#causedBy(err: any)
You can attach another error to the error.
const externalError = new Error('Some thrown error')
err.causedBy(externalError)
Method: BaseError#formatMessage(...formatParams)
See the sprintf-js package for usage.
// specify the database specific error code
// Transforms the message to:
// 'There was a database failure, SQL err code %s' ->
// 'There was a database failure, SQL err code SQL_ERR_1234',
err.formatMessage('SQL_ERR_1234')
Method: BaseError#withSafeMetadata(data = {})
Safe metadata would be any kind of data that you would be ok with exposing to a client, like an HTTP response.
err.withSafeMetadata({
errorId: 'err-12345',
moreData: 1234
})
// can be chained to append more data
.withSafeMetadata({
requestId: 'req-12345'
})
This can also be written as:
err.withSafeMetadata({
errorId: 'err-12345',
moreData: 1234
})
// This will append requestId to the metadata
err.withSafeMetadata({
requestId: 'req-12345'
})
Method: BaseError#withMetadata(data = {})
Internal metadata would be any kind of data that you would not be ok with exposing to a client, but would be useful for internal development / logging purposes.
err.withMetadata({
email: 'test@test.com'
})
// can be chained to append more data
.withMetadata({
userId: 'user-abcd'
})
Method: BaseError#toJSONSafe(fieldsToOmit = [])
Generates output that would be safe for client consumption.
namemessagecausedBytypelogLevelBaseError#withMetadata()err.withSafeMetadata({
requestId: 'req-12345'
})
// you can remove additional fields by specifying property names in an array
//.toJSONSafe(['code']) removes the code field from output
.toJSONSafe()
Produces:
{
code: 'ERR_INT_500',
subCode: 'DB_0001',
statusCode: 500,
meta: { requestId: 'req-12345' }
}
Method: BaseError#toJSON(fieldsToOmit = [])
Generates output that would be suitable for internal use.
nametypemessagecausedByBaseError#withMetadata() and BaseError#withSafeMetadata() is includederr.withSafeMetadata({
reqId: 'req-12345',
}).withMetadata({
email: 'test@test.com'
})
// you can remove additional fields by specifying property names in an array
//.toJSON(['code', 'statusCode']) removes the code and statusCode field from output
.toJSON()
Produces:
{
name: 'InternalServerError',
code: 'ERR_INT_500',
message: 'There was a database failure, SQL err code %s',
type: 'DATABASE_FAILURE',
subCode: 'DB_0001',
statusCode: 500,
meta: { errorId: 'err-12345', requestId: 'req-12345' },
stack: 'InternalServerError: There was a database failure, SQL err code %s\n' +
' at ErrorRegistry.newError (new-error/src/ErrorRegistry.ts:128:12)\n' +
' at Object.<anonymous> (new-error/src/test.ts:55:25)\n' +
' at Module._compile (internal/modules/cjs/loader.js:1158:30)\n' +
' at Module._compile (new-error/node_modules/source-map-support/source-map-support.js:541:25)\n' +
' at Module.m._compile (/private/var/folders/mx/b54hc2lj3fbfsndkv4xmz8d80000gn/T/ts-node-dev-hook-17091160954051898.js:57:25)\n' +
' at Module._extensions..js (internal/modules/cjs/loader.js:1178:10)\n' +
' at require.extensions.<computed> (/private/var/folders/mx/b54hc2lj3fbfsndkv4xmz8d80000gn/T/ts-node-dev-hook-17091160954051898.js:59:14)\n' +
' at Object.nodeDevHook [as .ts] (new-error/node_modules/ts-node-dev/lib/hook.js:61:7)\n' +
' at Module.load (internal/modules/cjs/loader.js:1002:32)\n' +
' at Function.Module._load (internal/modules/cjs/loader.js:901:14)'
}
The BaseError config onPreToJSONData / onPreToJSONSafeData options allow post-processing of the data. This is useful if you want to decorate your data for all new
errors created.
const errRegistry = new ErrorRegistry(errors, errorCodes, {
baseErrorConfig: {
// called when toJSON is called
onPreToJSONData: (data) => {
// we want all new errors to contain a date field
data.date = new Date().tostring()
// add some additional metadata
// data.meta might be empty if omitEmptyMetadata is enabled
if (data.meta) {
data.meta.moreData = 'test'
}
return data
}
}
})
const err = errRegistry.newError('INTERNAL_SERVER_ERROR', 'DATABASE_FAILURE')
.setErrorId('err-1234')
.formatMessage('SQL_1234')
// should produce the standard error structure, but with the new fields added
console.log(err.toJSON())
name property (not present when using toJSONSafe()), then only a BaseError instance can be returned.causedBy data. As a result, it will be copied as-is.JSON.parse() as JSON.parse() in its raw form is susceptible to
prototype pollution
if the parse function does not have a proper sanitization function. It is up to the developer to properly
trust / sanitize / parse the data.ErrorRegistry#fromJSON() methodThis method will attempt to deserialize into a registered error type via the name property. If it is unable to, a BaseError instance is
returned instead.
ErrorRegistry#fromJSON(data: object, [options]: DeserializeOpts): IBaseError
data: Data that is the output of BaseError#toJSON(). The data must be an object, not a string.options: Optional deserialization options.interface DeserializeOpts {
/**
* Fields from meta to pluck as a safe metadata field
*/
safeMetadataFields?: {
// the value must be set to true.
[key: string]: true
}
}
Returns a BaseError instance or an instance of a registered error type.
import { ErrorRegistry } from 'new-error'
const errors = {
INTERNAL_SERVER_ERROR: {
className: 'InternalServerError',
code: 'ERR_INT_500',
statusCode: 500,
logLevel: 'error'
}
}
const errorCodes = {
DATABASE_FAILURE: {
message: 'There was a database failure, SQL err code %s',
subCode: 'DB_0001',
statusCode: 500,
logLevel: 'error'
}
}
const errRegistry = new ErrorRegistry(errors, errorCodes)
const data = {
'errId': 'err-123',
'code': 'ERR_INT_500',
'subCode': 'DB_0001',
'message': 'test message',
'meta': { 'safeData': 'test454', 'test': 'test123' },
// maps to className in the high level error def
'name': 'InternalServerError',
'statusCode': 500,
'causedBy': 'test',
'stack': 'abcd'
}
// err should be an instance of InternalServerError
const err = errRegistry.fromJSON(data, {
safeMetadataFields: {
safeData: true
}
})
static BaseError#fromJSON() methodIf you are not using the registry, you can deserialize using this method. This also applies to any class that extends
BaseError.
static BaseError#fromJSON(data: object, [options]: DeserializeOpts): IBaseError
data: Data that is the output of BaseError#toJSON(). The data must be an object, not a string.options: Optional deserialization options.Returns a BaseError instance or an instance of the class that extends it.
import { BaseError } from 'new-error'
// assume we have serialized error data
const data = {
code: 'ERR_INT_500',
subCode: 'DB_0001',
statusCode: 500,
errId: 'err-1234',
meta: { requestId: 'req-12345', safeData: '123' }
}
// deserialize
// specify meta field assignment - fields that are not assigned will be assumed as withMetadata() type data
const err = BaseError.fromJSON(data, {
// (optional) Fields to pluck from 'meta' to be sent to BaseError#safeMetadataFields()
// value must be set to 'true'
safeMetadataFields: {
safeData: true
}
})
If the name property is present in the serialized data if it was serialized with toJson(), you can use a switch
to map to an instance:
const data = {
// be sure that you trust the source of the deserialized data!
// anyone can modify the 'name' property to whatever
name: 'InternalServerError',
code: 'ERR_INT_500',
subCode: 'DB_0001',
statusCode: 500,
errId: 'err-1234',
meta: { requestId: 'req-12345', safeData: '123' }
}
let err = null
switch (data.name) {
case 'InternalServerError':
// assume InternalServerError extends BaseError
return InternalServerError.fromJSON(data)
default:
return BaseError.fromJSON(data)
}
1.2.6 - Tue Mar 09 2021 00:23:04
Contributor: Theo Gravity
If you are calling ErrorRegistry#newError or related functions to create errors, the stack trace includes an ErrorRegistry entry. This change removes that entry for easier readability.
FAQs
A production-grade error creation and serialization library designed for Typescript
The npm package new-error receives a total of 66 weekly downloads. As such, new-error popularity was classified as not popular.
We found that new-error demonstrated a not healthy version release cadence and project activity because the last version was released a year ago. It has 2 open source maintainers collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Research
Security News
Socket researchers uncover a malicious npm package posing as a tool for detecting vulnerabilities in Etherium smart contracts.
Security News
Research
A supply chain attack on Rspack's npm packages injected cryptomining malware, potentially impacting thousands of developers.
Research
Security News
Socket researchers discovered a malware campaign on npm delivering the Skuld infostealer via typosquatted packages, exposing sensitive data.