You can use a type predicate function in the .filter
to avoid opting out of strict type checking:
function notEmpty<TValue>(value: TValue | null | undefined): value is TValue {
return value !== null && value !== undefined;
}
const array: (string | null)[] = ['foo', 'bar', null, 'zoo', null];
const filteredArray: string[] = array.filter(notEmpty);
Alternatively you can use array.reduce<string[]>(...)
.
2021 update: stricter predicates
While this solution works in most scenarios, you can get a more rigorous type check in the predicate. As presented, the function notEmpty
does not actually guarantee that it identifies correctly whether the value is null
or undefined
at compile time. For example, try shortening its return statement down to return value !== null;
, and you'll see no compiler error, even though the function will incorrectly return true
on undefined
.
One way to mitigate this is to constrain the type first using control flow blocks, and then to use a dummy variable to give the compiler something to check. In the example below, the compiler is able to infer that the value
parameter cannot be a null
or undefined
by the time it gets to the assignment. However, if you remove || value === undefined
from the if condition, you will see a compiler error, informing you of the bug in the example above.
function notEmpty<TValue>(value: TValue | null | undefined): value is TValue {
if (value === null || value === undefined) return false;
const testDummy: TValue = value;
return true;
}
A word of caution: there exist situations where this method can still fail you. Be sure to be mindful of issues associated with contravariance.