Accept de-structured elements in type predicates

Search Terms

• type predicate
• reference binding pattern
• type predicate cannot reference
• destructured

Suggestion

The possibility to use destructured parameters in type predicates.

Use Cases

Destructuring is heavily used in functional/reactive programming, notably with rxjs where various contextual properties tend to be passed between each operator.

Having the ability to succinctly test for types would make the code more readable, e.g.:

type Example = {
  a: number;
  b: string | undefined;
};

const example: Example = {
  a: 42,
  b: 'hello';
};

of(example).pipe(
  guard(({ b }): b is string => b !== undefined, 'b cannot be undefined'),
  tap({ b }) => { /* b is now a string rather than a string | undefined })
);

Right now the alternative is

of(example).pipe(
  guard((x): x is Omit<typeof x, 'b'> & { b: string } => x.b !== undefined, 'b cannot be undefined'),
  tap({ b }) => { /* b is now a string rather than a string | undefined })
);

Or, without a predicate

of(example).pipe(
  map(x => {
    if (x.b === undefined) {
      throw new Error();
    }

    return x;
  }),
  tap({ b }) => { /* b is now a string rather than a string | undefined })
);

Examples

function assertSomething(
  { property }: T
): property is AssertionHere {
  return true;
}

This would roughly translate to something like:

function assertSomething(
  obj: T
): obj is Omit<T, 'property'> & { property: AssertionHere } {
  return true;
}

Checklist

My suggestion meets these guidelines:

• This wouldn’t be a breaking change in existing TypeScript/JavaScript code
• This wouldn’t change the runtime behavior of existing JavaScript code
• This could be implemented without emitting different JS based on the types of the expressions
• This isn’t a runtime feature (e.g. library functionality, non-ECMAScript syntax with JavaScript output, etc.)
• This feature would agree with the rest of TypeScript’s Design Goals.

[manbearwiz]

I run into this when filtering the output of Object.entries. I find it much more readable to be able to reference key and value instead of pair[0] and pair[1]. Simplified example but demonstrates a use case outside of rxjs.

If I want all pairs from the query params that are arrays, I currently have to do:

const queryParams = {text: 'foo', statuses: ['status1', 'status2'], regions: []}

Object.entries(queryParams)
 .filter((pair): pair is [string, string[]] => Array.isArray(pair[1]))

or

Object.entries(queryParams)
  .filter(([_, value]) => Array.isArray(value))
  .map(pair => pair as [string, string[]])

I would prefer to do:

Object.entries(queryParams )
  .filter(([_, value]): value is string[] => Array.isArray(value))

[artu-ole]

I agree that this is a useful feature and an unfortunate oversight on typescript’s part. I would, however, note that @rraziel’s workaround example can be slightly better versed by not causing an additional loop with map and using destructuring in the filter body to preserve readability.

const filtered = xs
  .filter((x): x is { value: number } => {
    const { value } = x;
    return typeof value === 'number';
  });

Same goes for array destructuring(my use case which involved rxjs’s combineLatest)

type X = [ number | string ];
const xs: Array<X> = [[ 42 ], [ 'hello' ]];

// without the feature
const filtered = xs
  .filter((x) => {
    const [ value ] = x;
    return typeof value === 'number';
  })
;

// with the feature
const filtered = xs
  .filter(([ value ]): value is number => typeof value === 'number')
;