ts-runtime-typecheck

ts-runtime-typecheck

Simple functions for validating complex data.

JavaScript is a very flexible language. Meaning it's easy to get tripped up by a value with an unexpected type. Even in TypeScript you occasionally have to deal with values which cannot be safely typed at compile time. This library provides a comprehensive selection of functions to simplify type checking code with clear and concise function calls. Ensuring strict type safety throughout your program, no matter the input.

Installation

Releases are available on the npm repository and our GitHub releases page. ESM and CJS formats are both included, as well as TypeScript type definition files. Both formats work without TypeScript if you prefer plain JS.

npm install ts-runtime-typecheck

Contents

• Installation
• Type Casts
• Type Checks
• Type Coerce
• Type Asserts
• JSON Types
• Ensuring an optional value is defined
• Array/Object of Type Casts
• Validating interfaces
• Union types
• Class instances
• Guarded functions
• Reference
• Changelog

Type Casts

Type Casts take an unknown value as an argument, and return a typed value as the result. These functions take the form as{TYPE}, for example asNumber. If the input value does not match the required type the function will throw. This does not perform any coercion on the value, passing a string of a number to asNumber will cause it to throw.

import { asNumber } from 'ts-runtime-typecheck';

function square (input: unknown): number {
  const value: number = asNumber(input);
  return value * value;
}

square(10)
// 100
square()
// Error: Unable to cast undefined to number
square('10')
// Error: Unable to cast string to number

Type Casts are meant to primarily validate questionable values that are expected to be in a well defined structure. Such as network responses, interfacing with untyped JavaScript or reading data back from a file. If you are looking to validate a type, without throwing an error then take a look at Type Checks.

Optional Type Casts

In the situation you want to check a value meets an Optional type there exists an alternate function for each type cast. These take the form asOpt{TYPE}. Unlike the standard functions when a Nullish value is passed in they will emit undefined instead of throwing. If the input is not Nullish, then it behaves the same as the standard type casts. If the type condition is met then it emits the value, otherwise it will throw.

Type Checks

Type Checks take an unknown value as an argument, and return a boolean indicating if the given value matches the required type. These functions take the form is{TYPE}. In the correct situation TypeScript is capable of refining the type of a value through the use of these functions and flow analysis, like the below example.

import { isNumber } from 'ts-runtime-typecheck';

export function printSq (value: unknown) {
  if (isNumber(value)) {
    // inside this block `value` is a `number`
    console.log(`${value} * ${value} = ${value * value}`);
  }
  else {
    // inside this block `value` is `unknown`
    console.log('Invalid input', value);
  }
}

In addition all relevant Type Checks have an alternate variant that take the form isOpt{TYPE}. These variants return true if the value meets the given type or Nullish.

import { isOptNumber } from 'ts-runtime-typecheck';

export function printSq (input: unknown) {
  if (isOptNumber(input)) {
    // inside this block `input` is `number | undefined | null`
    const value = input ?? 1; // use nullish coalescing operator to ensure value is number
    console.log(`${value} * ${value} = ${value * value}`);
  }
  else {
    // inside this block `input` is `unknown`
    console.log('Invalid input', value);
  }
}

Type Coerce

Type Coercion functions take an unknown value as an argument, and convert it into a specific type. These functions take the format make{TYPE}. Unlike the other functions this only works for small subset of types: number, string and boolean. They make a best effort to convert the type, but if the input is not suitable then they will throw an error. For instance passing a non-numeric string to makeNumber will cause it to throw, as will passing a string that is not "true" | "false" to makeBoolean. While these functions will take any input value, this is to allow the input of values that have not been validated. The only valid input types for all 3 functions are number | string | boolean. The intention here is to allow useful conversion, but prevent accidentally passing complex types.

There is an argument that makeString could support using the toString method of an object, but the default toString method returns the useless [object Object] string. It is possible to detect if an object has implemented it's own toString method, but is it correct to use it in this situation? That depends on the intention of the programmer. In the absence of a clear answer the line has been drawn at only accepting primitives.

import { makeNumber } from 'ts-runtime-typecheck';

makeNumber('80') // 80
makeNumber(80) // 80
makeNumber(true) // 1
makeNumber(false) // 0
makeNumber('hello') // Error: Unable to cast string to Number
makeNumber({
  toString () { return 'hello' }
}) // Error: Unable to cast object to Number

Type Asserts

Type Assert functions accept an unknown value and throw if the value does not meet the type requirement, they do not return a value. While this may seem very similar to Type Casts they are capable of providing a hint to the TypeScript compiler without needing to reassign the value. As such they are very helpful for validating function arguments before using them.

Each type assert takes an optional second argument that is a label for the passed value, this will be included in the thrown TypeAssertion error if the value does not meet the type requirement, making it easier to isolate the type violation.

import { assertDefined } from 'ts-runtime-typecheck';

function main (meaningOfLife: Optional<number>) {
  meaningOfLife // number | null | undefined
  assertDefined(meaningOfLife, 'Meaning of Life');
  meaningOfLife // number
  return 'but what is the question?';
}

main(42); // 'but what is the question?'
main(); // TypeAssertion: Meaning of Life is not defined

TypeAsserts cannot be based on generic types, due to limits in the TypeScript type system. Hence there is no analogous function to isStruct and similar TypeCheck. However, there is an alternative. It's possible to utilise an invariant ( or assert) function with a TypeCheck to get the same effect, and an implementation of invariant is provided for this purpose.

import { invariant, isLiteral } from 'ts-runtime-typecheck';

function main (meaningOfLife: unknown) {
  meaningOfLife // unknown
  invariant(isLiteral(42)(meaningOfLife), "Universe is broken, meaning of life isn't 42!");
  meaningOfLife // 42
  return 'but what is the question?';
}

main(42); // 'but what is the question?'
main(); // TypeAssertion: Universe is broken, meaning of life isn't 42!

JSON Types

Dealing with validating JSON values can often be frustrating, so to make it a little easier JSON specific types and checks are provided. Using the JSONValue type in your code will ensure that TS statically analyses any literal values as serializable to JSON.

import type { JSONArray, JSONObject, JSONValue } from 'ts-runtime-typecheck';

// JSONArray is an Array of JSONValues
const a: JSONArray = [12, 'hello'];
// JSONObject is a Dictionary of JSONValues
const b: JSONObject = {
  num: 12,
  str: 'hello'
};
// JSONValue can be any of the following: JSONObject, JSONArray, string, number, boolean or null
const c: JSONValue = 12;

const d: JSONValue = new Error('hi'); // Type 'Error' is not assignable to type 'JSONValue'

For dynamic data isJSONValue and asJSONValue provide recursive type validation on a value.

Type Check and Type Casts are provided for JSONArrays and JSONObjects, with the caveat that they only accept JSONValues. This is to avoid needing to recursively validate values which have already been validated.

import { asJSONValue, isJSONObject, isJSONArray } from 'ts-runtime-typecheck';
import type { JSONValue } from 'ts-runtime-typecheck';

function main (a: unknown) {
  const obj: JSONValue =  asJSONValue(a);
  // obj: JSONValue
  if (isJSONArray(obj)) {
    // obj: JSONArray
  }
  else if (isJSONObject(obj)) {
    // obj: JSONObject
  }
  else {
    // obj: number | string | boolean | null
  }
}

One other caveat of JSONValue is that it does not guarantee that the value is not cyclic. It is not possible to serialize cyclic object with JSON.stringify, but they are otherwise valid. Using isJSONValue or asJSONValue on a cyclic object will fail.

import { asJSONValue } from 'ts-runtime-typecheck';
import type { Dictionary } from 'ts-runtime-typecheck';

const almost_right: Dictionary = {};
almost_right.self = almost_right;

// BANG! this will fail, it recurses endlessly
const obj = asJSONValue(almost_right);

Ensuring an optional value is defined

A common situation is that you have an Optional value, with a well defined type. At a specific time it should be defined, but the type system is not aware of this. TypeScript will allow you to cast the value to a non-optional type using !, but this is often discouraged in style guides. A safer alternative is to use asDefined and friends, which can be used to subtract the optionality from the type.

import { asDefined, assertDefined, isDefined } from 'ts-runtime-typecheck';

function doThing (value: number) {
  // [...]
}

function branching (value?: number) {
  // used for conditions, allows for an alternative to throwing or custom error behavior
  if (isDefined(value)) {
    doThing(value)
  } else {
    console.log('bad things')
  }
}

function inline (value?: number) {
  // used inline, throws if the value is Nullish otherwise returns the value
  doThing(asDefined(value))
}

function assertion (value?: number) {
  // used to ensure execution doesn't progress if the value isn't defined, throws if the value is Nullish
  assertDefined(value)
  doThing(value)
}

Array/Object of Type Casts

Validating that a value is an array or dictionary is easy enough, but how about the type of the contents? asArrayOf and asDictionaryOf allow you to cast the elements of a collection using a user defined Type Check. For example, to cast to Array<string>:

import { isString, asArrayOf } from 'ts-runtime-typecheck';

function main (obj: unknown) {
  const asStringArray = asArrayOf(isString);

  const arr: string[] = asArrayOfString(obj);
}

Or Array<Dictionary<number>>:

import { isNumber, isDictionaryOf, asArrayOf } from 'ts-runtime-typecheck';

function main () {
  const isDictionaryOfNumber = isDictionaryOf(isNumber);
  const asArrayOfDictionaryOfNumber = asArrayOf(isDictionaryOfNumber);

  const arr = asArrayOfDictionaryOfNumber([
    {
      a: 12,
      b: 42
    },
    {
      n: 90
    }
  ]);
}

Validating interfaces

Validating the shape of an object using a combination of asDictionary and other Type Casts specific to property types can be a bit verbose. To simplify this scenario you can use asStruct. This function takes an InterfacePattern that defines a specific structure and returns a new function that will cast an unknown value to that structure. An InterfacePattern is a fancy name for a Dictionary of Type Checks.

import { asStruct, isString, isNumber } from 'ts-runtime-typecheck';

interface Item {
  name: string;
  value: number;
}

const asItem = asStruct({ name: isString, value: isNumber })

function main (obj: unknown) {
  const item: Item = asItem(obj);

  console.log(`${item.name} = ${item.value}`);
}

There is also a Type Check variant of the this function called isStruct which works in a very similar way. As an InterfacePattern is composed of Type Check functions it's possible to compose nested interface Type Checks.

import { asStruct, isStruct, isString, isOptString, isNumber } from 'ts-runtime-typecheck';
import type { Optional } from 'ts-runtime-typecheck';

interface Declaration {
  item: Item;
  description: Optional<string>
}

interface Item {
  name: string;
  value: number;
}

const isItem = isStruct({ name: isString, value: isNumber });
const asDeclaration = asStruct({ item: isItem, description: isOptString });

function main (obj: unknown) {
  const { item, description } = asDeclaration(obj);

  const comment: string = description ? `// ${description}` : '';
  console.log(`${item.name} = ${item.value} ${comment}`);
}

Union types

When a value can be 2 or more types it is relatively easy to do Type Check.

import { isString, isArray } from 'ts-runtime-typecheck';

if (isString(a) || isArray(a)) {
  // a: string | unknown[]
}

But you can't cast to that type, or pass it into a function like asArrayOf or isStruct which require a Type Check for their input. To do this you can use isUnion or asUnion. These functions take a variable number of Type Checks and produce a union of them.

import {
  isString,
  isArray,
  isUnion,
  asArrayOf
} from 'ts-runtime-typecheck';

const check = asArrayOf(isUnion(isString, isArray));
const b = check(['hello', [0, 1, 2], 'world']);

Class instances

Under most scenarios you will know if a value is an instance of a given class. However, there are scenarios where this is not the case. For these situations you can use isInstance or asInstance to ensure you have the correct type.

import { isInstance } from 'ts-runtime-typecheck';

function print_error (err) {
  if (isInstance(Error)(err)) {
    print_string(err.message);
  } else {
    print_unknown(err)
  }
}

When validating a value matches an interface it may be desirable to instead use isInstance instead of isStruct. While it doesn't provide the same guarantees it will often be significantly faster, as it does not perform a Type Check on each member to see that they exist and contain the right type of value.

Literal types

TypeScript supports value types for Primitives. For example

let a: 'hello world' = 'hello world';

a = 'goodbye world'; // Type '"goodbye world"' is not assignable to type '"hello world"'

You can use asLiteral and isLiteral to construct TypeCasts and TypeChecks respectively for these value types. These checks can be particularly useful for discriminated unions.

interface A {
  type: 'a';
  value: number;
}

interface B {
  type: 'b';
  value: number;
}

const isA = isStruct({
  type: isLiteral('a'),
  value: isNumber,
});

function main (n: A | B) {
  n // A | B
  if (isA(n)) {
    n // A
  } else {
    n // B
  }
}

Guarded Functions

TypeScript provides excellent compile time protections against misusing functions. The number of parameters, the types of parameters and the return type of the function are all statically type checked. But it provides no runtime protections. In a pure TS project this isn't normally required, but sometimes you need to pass around abstract functions and have them behave reliably. Alternatively you could be authoring a library and want to ensure that developers using your library without TS use your functions correctly. In both these scenarios you can wrap the function using asGuardedFunction, specifying the return and parameter types. This will return a new function that when called will validate the incoming arguments, and the outgoing return value at runtime. If the types don't match up it will throw an error. The resulting function will have accurate parameter and return types, irrespective of the original function.

import { asGuardedFunction } from 'ts-runtime-typecheck';

export const publicFunction = asGuardedFunction(
  (a: number, b: number) => `${(a / b) * 100}%`, // function to wrap
  isString, // return value
  isNumber, isNumber // parameters
);

Reference

Reference: Type Casts

• asString

  Cast unknown to string.

• asNumber

  Cast unknown to number.

• asIndex

  Cast unknown to Index.

• asPrimitive

  Cast unknown to Primitive.

• asIndexable

  Cast unknown to Indexable.

• asBoolean

  Cast unknown to boolean.

• asArray

  Cast unknown to Array<unknown>.

• asDictionary

  Cast unknown to Dictionary<unknown>.

• asFunction

  Cast unknown to UnknownFunction.

• asDefined

  Cast Type | Nullish to Type, where Type is a generic parameter.

• asJSONValue

  Cast unknown to JSONValue. This function recursively validates the value, and hence will fail if given a cyclic value.

• asJSONObject

  Cast JSONValue to JSONObject. Unlike asJSONValue this does not perform recursive validation, hence it only accepts a JSONValue so that the sub-elements are of a known type.

• asJSONArray

  Cast JSONValue to JSONArray. Unlike asJSONValue this does not perform recursive validation, hence it only accepts a JSONValue so that the sub-elements are of a known type.

• asArrayOf

  Takes a Type Cast function for Type and returns a new Type Cast function for Array<Type> where type is a generic parameter. Refer to Array/Object of Type Casts for examples.

• asDictionaryOf

  Takes a Type Cast function for Type and returns a new Type Cast function for Dictionary<Type> where type is a generic parameter. Refer to Array/Object of Type Casts for examples.

• asStruct

  Takes an InterfacePattern which is equivalent to Type and returns a new Type Cast function for Type, where Type is an interface defined by the TypeChecks specified in the pattern. Refer to Validating interfaces for examples.

• asInstance

  Takes a class (not a instance of a class) and returns a new Type Cast for an instance of that class.

• asLiteral

  Takes a Primitive value and returns a new Type Cast for that specific value.

• asGuardedFunction

  Takes a unknown value and a series of TypeChecks and returns a strongly typed function. It will be a new function that wraps the original with runtime type validation, ensuring that the arguments and return value are the expected types. The first TypeCheck is for the return value, and subsequent variadic TypeChecks are used for validating arguments.

Reference: Optional Type Casts

• asOptString

  Cast unknown value to string | undefined. If value is Nullish then return undefined.

• asOptNumber

  Cast unknown value to number | undefined. If value is Nullish then return undefined.

• asOptIndex

  Cast unknown value to Index | undefined. If value is Nullish then return undefined.

• asOptPrimitive

  Cast unknown value to Primitive | undefined. If value is Nullish then return undefined.

• asOptIndexable

  Cast unknown value to Indexable | undefined. If value is Nullish then return undefined.

• asOptBoolean

  Cast unknown value to boolean | undefined. If value is Nullish then return undefined.

• asOptArray

  Cast unknown value to Array<unknown> | undefined. If value is Nullish then return undefined.

• asOptDictionary

  Cast unknown value to Dictionary<unknown> | undefined. If value is Nullish then return undefined.

• asOptFunction

  Cast unknown value to UnknownFunction | undefined. If value is Nullish then return undefined.

• asUnion

  Takes a variable number of type checks as parameters <A>(...checks: TypeCheck<A>[]) and returns a new type cast that composes them into type cast for the union A. This allows creating a cast for a type union by composing any existing type checks.

• asOptUnion

  Identical to asUnion but it the resulting cast returns A | null | undefined.

• asOptJSONValue

  Cast unknown value to JSONValue | undefined. If value is Nullish then return undefined.

• asOptJSONObject

  Cast JSONValue | undefined value to JSONObject | undefined. If value is Nullish then return undefined.

• asOptJSONArray

  Cast JSONValue | undefined value to JSONArray | undefined. If value is Nullish then return undefined.

• asOptArrayOf

  Takes a Type Cast function for Type and returns a new Type Cast function for Array<Type> | undefined where type is a generic parameter. Refer to Array/Object of Type Casts for examples.

• asOptDictionaryOf

  Takes a Type Cast function for Type and returns a new Type Cast function for Dictionary<Type> | undefined where type is a generic parameter. Refer to Array/Object of Type Casts for examples.

• asOptStruct

  Takes an InterfacePattern which is equivalent to Type and returns a new Type Cast function for Type | undefined, where Type is an interface defined by the TypeAsserts specified in the pattern. Refer to Validating interfaces for examples.

• asOptInstance

  Takes a class (not a instance of a class) and returns a new Type Cast for a Optional instance of that class.

• asOptLiteral

  Takes a Primitive value and returns a new optional Type Cast for that specific value.

Reference: Type Checks

• isDictionary

  Takes an unknown value and returns a boolean indicating if the value is of the type Dictionary<unknown>.

• isFunction

  Takes an unknown value and returns a boolean indicating if the value is of the type UnknownFunction.

• isBoolean

  Takes an unknown value and returns a boolean indicating if the value is of the type boolean.

• isString

  Takes an unknown value and returns a boolean indicating if the value is of the type string.

• isNumber

  Takes an unknown value and returns a boolean indicating if the value is of the type number.

• isIndex

  Takes an unknown value and returns a boolean indicating if the value is of the type Index.

• isPrimitive

  Takes an unknown value and returns a boolean indicating if the value is of the type Primitive.

• isIndexable

  Takes an unknown value and returns a boolean indicating if the value is of the type Indexable.

• isArray

  Takes an unknown value and returns a boolean indicating if the value is of the type Array<unknown>.

• isUndefined

  Takes an unknown value and returns a boolean indicating if the value is of the type undefined.

• isNullish

  Takes an unknown value and returns a boolean indicating if the value is of the type Nullish.

• isDefined

  Takes an unknown value and returns a boolean indicating if the value is not of the type Nullish.

• isUnion

  Takes a variable number of type checks as parameters <A>(...checks: TypeCheck<A>[]) and returns a new type check that composes them into union type check TypeCheck<A>. This allows creating a test for a type union by composing any existing type checks. For inline code it will generally make sense to use logical OR operators instead of this, for example if ( isNumber(n) || isArray(n) ) {}. This particular function is intended for when you wish to compose a type check or cast that contains a union, or create a library type check for a common union type.

• isJSONValue

  Takes an unknown value and returns a boolean indicating if the value is of the type JSONValue.

• isJSONArray

  Takes an JSONValue value and returns a boolean indicating if the value is of the type JSONArray.

• isJSONObject

  Takes an JSONValue value and returns a boolean indicating if the value is of the type JSONObject.

• isArrayOf

  Takes a Type Check function for Type and returns a new Type Check function for Array<Type> where Type is a generic parameter.

• isDictionaryOf

  Takes a Type Check function for Type and returns a new Type Check function for Dictionary<Type> where Type is a generic parameter.

• isStruct

  Takes an InterfacePattern which is equivalent to Type and returns a new TypeAssert function for Type, where Type is an interface defined by the TypeAsserts specified in the pattern. Refer to Validating interfaces for examples.

• isInstance

  Takes a class (not a instance of a class) and returns a new Type Check for an instance of that class.

• isLiteral

  Takes a Primitive value and returns a new Type Check for that specific value.

Reference: Optional Type Checks

• isOptDictionary

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<Dictionary<unknown>>.

• isOptFunction

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<UnknownFunction>.

• isOptBoolean

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<boolean>.

• isOptString

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<string>.

• isOptNumber

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<number>.

• isOptPrimitive

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<Primitive>.

• isOptIndex

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<Index>.

• isOptIndexable

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<Indexable>.

• isOptArray

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<Array<unknown>>.

• isOptUnion

  Identical to isUnion but it the resulting typecheck is TypeCheck<A | null | undefined>.

• isOptJSONValue

  Takes an unknown value and returns a boolean indicating if the value is of the type Optional<JSONValue>.

• isOptJSONArray

  Takes an Optional<JSONValue> value and returns a boolean indicating if the value is of the type Optional<JSONArray>.

• isOptJSONObject

  Takes an Optional<JSONValue> value and returns a boolean indicating if the value is of the type Optional<JSONObject>.

• isOptStruct

  Takes an InterfacePattern which is equivalent to Type and returns a new TypeAssert function for Optional<Type>, where Type is an interface defined by the TypeAsserts specified in the pattern. Refer to Validating interfaces for examples.

• isOptArrayOf

  Takes a Type Check function for Type and returns a new Type Check function for Optional<Array<Type>> where Type is a generic parameter.

• isOptDictionaryOf

  Takes a Type Check function for Type and returns a new Type Check function for Optional<Dictionary<Type>> where Type is a generic parameter.

• isOptInstance

  Takes a class (not a instance of a class) and returns a new Type Check for a Optional instance of that class.

• isOptLiteral

  Takes a Primitive value and returns a new optional Type Check for that specific value.

Reference: Type Coerce

• makeString

  Takes an unknown value and converts it to it's textual representation. A value that cannot be cleanly converted will trigger an error.

• makeNumber

  Takes an unknown value and converts it to it's numerical representation. A value that cannot be cleanly converted will trigger an error.

• makeBoolean

  Takes an unknown value and converts it to it's boolean representation. A value that cannot be cleanly converted will trigger an error.

• makeStrictPartial

  Takes a value of the generic type T and returns a copy of the object excluding any members that were Nullish. The returned object meets the type StrictPartial<T>.

Reference: Type Assert

• assertDefined

Assert value of type Type | Nullish is Type, where Type is a generic parameter. Accepts an optional name for the value that is included in the error if the value is nullish.

• assertNumber

Assert value of type unknown is number. Accepts an optional name for the value that is included in the error if the value is not a number.

• assertString

Assert value of type unknown is string. Accepts an optional name for the value that is included in the error if the value is not a string.

• assertBoolean

Assert value of type unknown is boolean. Accepts an optional name for the value that is included in the error if the value is not a boolean.

• assertIndex

Assert value of type unknown is Index. Accepts an optional name for the value that is included in the error if the value is not a Index.

• assertPrimitive

Assert value of type unknown is Primitive. Accepts an optional name for the value that is included in the error if the value is not a Primitive.

• assertArray

Assert value of type unknown is unknown[]. Accepts an optional name for the value that is included in the error if the value is not a unknown[].

• assertDictionary

Assert value of type unknown is Dictionary. Accepts an optional name for the value that is included in the error if the value is not a Dictionary.

• assertIndexable

Assert value of type unknown is Indexable. Accepts an optional name for the value that is included in the error if the value is not a Indexable.

• assertFunction

Assert value of type unknown is UnknownFunction. Accepts an optional name for the value that is included in the error if the value is not a UnknownFunction.

• assertOptNumber

Assert value of type unknown is Optional<number>. Accepts an optional name for the value that is included in the error if the value is not a Optional<number>.

• assertOptString

Assert value of type unknown is Optional<string>. Accepts an optional name for the value that is included in the error if the value is not a Optional<string>.

• assertOptBoolean

Assert value of type unknown is Optional<boolean>. Accepts an optional name for the value that is included in the error if the value is not a Optional<boolean>.

• assertOptIndex

Assert value of type unknown is Optional<Index>. Accepts an optional name for the value that is included in the error if the value is not a Optional<Index>.

• assertOptPrimitive

Assert value of type unknown is Optional<Primitive>. Accepts an optional name for the value that is included in the error if the value is not a Optional<Primitive>.

• assertOptArray

Assert value of type unknown is Optional<unknown[]>. Accepts an optional name for the value that is included in the error if the value is not a Optional<unknown[]>.

• assertOptDictionary

Assert value of type unknown is Optional<Dictionary>. Accepts an optional name for the value that is included in the error if the value is not a Optional<Dictionary>.

• assertOptIndexable

Assert value of type unknown is Optional<Indexable>. Accepts an optional name for the value that is included in the error if the value is not a Optional<Indexable>.

• assertOptFunction

Assert value of type unknown is Optional<UnknownFunction>. Accepts an optional name for the value that is included in the error if the value is not a Optional<UnknownFunction>.

• assertJSONValue

Assert value of type unknown is JSONValue. Accepts an optional name for the value that is included in the error if the value is not a JSONValue.

• assertJSONArray

Assert value of type JSONValue is JSONArray. Accepts an optional name for the value that is included in the error if the value is not a JSONArray.

• assertJSONObject

Assert value of type JSONValue is JSONObject. Accepts an optional name for the value that is included in the error if the value is not a JSONObject.

• assertOptJSONValue

Assert value of type unknown is JSONValue | undefined. Accepts an optional name for the value that is included in the error if the value is not a JSONValue | undefined.

• assertOptJSONArray

Assert value of type JSONValue | undefined is JSONArray | undefined. Accepts an optional name for the value that is included in the error if the value is not a JSONArray | undefined.

• assertOptJSONObject

Assert value of type JSONValue | undefined is JSONObject | undefined. Accepts an optional name for the value that is included in the error if the value is not a JSONObject | undefined.

Reference: Helper functions

• invariant

  An invariant is a logical declaration about a condition which the programmer knows to be true, but the compiler cannot. Many of the patterns in ts-runtime-typecheck are based around this concept, albeit to encourage a stricter and safer environment. This helper accepts a logical condition and a message. If the condition is true nothing happens, if it's false then a TypeAssertion is thrown with the given message. If the condition is the result of a TypeCheck then the type predicate is enforced by the invariant.

  import { invariant, isString } from 'ts-runtime-typecheck';
  
  function main (username: unknown) {
    invariant(isString(username), 'Expected username to be a string');
  
    username // string
  }
  

• inspectType

Inspects the type of a given value and returns a description of it as a string. Useful for constructing debug messages from unknown values. Instances of classes are described using the name of their class. Abstract objects are traversed recursively so that their elements can be described as well. Recursion uses a depth limit to prevent overlarge descriptions. Once the limit has been reached abstract objects will be described as Dictionary.

An optional second argument can be passed to modify the default behaviour. For example you can change the maxDepth limit from it's default of 3; a value of 0 would prevent recursion whereas Infinity would remove the limit.

Arrays currently do not feature recursive type description, but this might be introduced in future.

Internally this is used to describe values in type error messages.

import { inspectType } from 'ts-runtime-typecheck';

class Example {}

inspectType('hello world'); // string
inspectType(null) // null
inspectType([]) // Array
inspectType(new Example) // Example
inspectType({}) // {}
inspectType({
  foo: 'hello',
  bar: 'world'
}) // { foo: string, bar: string }
inspectType({ foo: 'bar' }, { maxDepth: 0 }); // Dictionary

Reference: Types

• JSONValue

  A union of all the JSON compatible types: JSONArray, JSONObject, number, string, boolean, null.

• JSONObject

  An alias to Dictionary<JSONValue> which can represent any JSON Object value.

• JSONArray

  An alias to Array<JSONValue> which can represent any JSON Array value.

• Dictionary

  An alias to Record<string, Type> where Type is a generic parameter that default to unknown. This type can be used to represent a typical key-value map constructed from a JS Object. Where possible use Map instead, as it is specifically designed for this purpose and has better protection against null errors in TS.

• Index

  A union of the number and string types that represent a value that could be used to index an element within a JS Object.

• Primitive

  A union of the number, string and boolean types that are the key primitive values in JS.

• Indexable

  An alias to Record<Index, Type> where Type is a generic parameter that default to unknown. This type can be used to represent an unknown key-value object that can be indexed using a number or string. It is intended to be used to ease the transition of JS project to TS. Where possible use Dictionary or preferably Map instead, as it is specifically designed for this purpose and has better protection against null errors in TS.

• Nullish

  A union of undefined and null. Generally preferable to either null or undefined on non-validated input. However, be aware of varying behavior between these 2 types in JS around optional members, default parameters and equality.

• Optional

  A union of Type and Nullish where Type is a generic parameter.

• UnknownFunction

  A stricter alternative to the type Function. It accepts any number of unknown parameters, and returns an unknown value. Allowing you to reference an untyped function in a slightly safer manner. This does not provide any arity or type checks for the parameters.

• UnknownAsyncFunction

  Identical to UnknownFunction in all ways but 1, it returns Promise<unknown> instead.

• TypeCheck

  An alias for a function that meets the requirements of TypeScript Type Guards. They take the format (value: unknown) => value is TYPE. With the exception of specialist JSON checks all TypeChecks conform to this type.

• InterfacePattern

  An alias for a Dictionary of TypeAssert functions. When used in conjunction with isStruct or asStruct they can validate an object against the equivalent interface to the pattern.

• StrictRequired

  A variant of the inbuilt Required<T>, which is the opposite of Optional<T> in that it subtracts the type undefined from each member of the type T. StrictRequired varies in that it also subtracts the type null from each member. Ensuring that all members meet the constraint NonNullable.

• StrictPartial

  A variant of the Partial<T> inbuilt, and closely related to FuzzyPartial. Partial makes no guarantees about the members of the type T, as such they can be unions of null. This can introduce inconsistency for users of the type; expecting that members can be specified using either null or undefined, where only some can also use null. StrictPartial resolves this by specifying that no members of the type T can be null, ensuring a consistent interface.

• FuzzyPartial

  A variant of the Partial<T> inbuilt, and closely related to StrictPartial. Partial makes no guarantees about the members of the type T, as such they can be unions of null. This can introduce inconsistency for users of the type; expecting that members can be specified using either null or undefined, where only some can also use null. FuzzyPartial resolves this by specifying that all members of the type T can ALSO be null, ensuring a consistent interface.

• TypeAssertion

  A custom error with the name TypeAssertion. This type is exported as a value so that Errors of this type can be isolated from other errors using instance checks. It is possible to use the constructor to create and throw your own Errors if you wish, but this may change in future.

Changelog

1.0.0

• Initial release

1.1.0

• Documentation update.
• Fix: asDefined was returning unknown.
• Breaking change: rename ObjectDict to Dictionary.
• Add: Nullish type ( null | undefined ).
• Change: Dictionary no longer contains T | undefined union.
• Change: Optional type now also includes null in the type union.

1.1.1

• Change: return type of asOpt{TYPE} is now TYPE | undefined instead of Optional<TYPE> ( removes null from union )
• Documentation corrections.

1.2.0

• Add: Introduce isStruct and asStruct that allow the inspection of a object to see if it meets a specific interface.
• Add: Optional variants of Type Checks with form isOpt{TYPE}.
• Breaking Change: asDefined can longer accept null as a fallback parameter.
• Change: asIndexable now accepts arrays.
• Add: isIndexable type check.
• Change: Expose the TypeAssert type publicly.
• Add: InterfacePattern type.
• Change: modify the type names in errors to be closer to the TypeScript names.

2.0.0

• Add: isUnion and isOptUnion to allow checking if a value matches any type of a type union.
• Add: asUnion and asOptUnion to allow casting a value to a type union.
• Add: isArrayOf and isOptArrayOf to allow checking if a value is an array of a given type.
• Add: isDictionaryOf and isOptDictionaryOf to allow checking if a value is a Dictionary of a given type.
• Breaking Change: Recursive Type Casts now take a Type Check as an argument instead of a Type Cast, and no longer emit a copy of the input. As a side effect if you are upgrading from asArrayRecursive(asOptString) to asArraryOf(isOptString) or (similar) the output array may contain null as the elements are no longer transformed by an inner cast ( optional cast methods normalize output to undefined ).
• Add: isInstance and isOptInstance to allow checking if a value is an instance of a given class.
• Add: asInstance and asOptInstance to allow casting a value to an instance of a given class.
• Breaking Change: asRecord, asOptRecord, asRecordRecursive and asOptRecordRecursive have been renamed to asDictionary, asOptDictionary, asDictionaryOf and asOptDictionaryOf respectively.
• Breaking Change: asArrayRecursive and asOptArrayRecursive have been renamed to asArrayOf and asOptArrayOf respectively.
• Breaking Change: rename TypeAssert to TypeCheck.

2.1.0

• Add: makeStrictPartial for converting Partial<T> to StrictPartial.
• Add: types StrictPartial and FuzzyPartial, variants of the inbuilt Partial type.
• Add: type StrictRequired, variant of Required.

2.1.1

• Fix: incorrect constraint on makeStrictPartial prevented passing in non-indexable instances.

2.2.0

• Add: assertDefined throws if the passed value is Nullish.
• Add: TypeAssertion error class thrown by TypeAsserts.

2.2.1

• Fix: update sub-dependency to resolve npm advisory 1654

2.2.2

• Fix: asInstance, asOptInstance, isInstance and isOptInstance were not exported from the package.

2.3.0

• Change: build target to ES2018 instead of ES3.

2.4.0

• Add: invariant function to assist type assertion
• Add: JSON assertion functions
• Add: Basic type assertion functions
• Add: Literal type casts and checks
• Add: Primitive type, casts and checks
• Documentation: Correct some typos in the isStruct/asStruct examples

2.5.0

• Add: inspectType function to describe the type of a value
• Fix: makeNumber no longer returns a number strings prefixed with a number
• Change: Type error messages now use the more descriptive inspectType instead of typeof for erroneous values

2.6.0

• Add: asGuardedFunction to wrap a function with parameter/return value type checks
• Change: Depreciate the fallback parameter for all TypeCasts
• Documentation: Simplify the contents into primary headings only
• Change: Instead of being a custom subclass of Error TypeAssertion is now an alias to TypeError