Deep Omit

In this medium-level challenge, you'll implement DeepOmit<T, S> which works like the built-in Omit utility type but supports dot-separated paths to omit deeply nested properties from an object type.

type obj = {
person: {
name: string;
age: {
value: number
}
}
}
 
type test1 = DeepOmit<obj, 'person'>    // {}
type test2 = DeepOmit<obj, 'person.name'> // { person: { age: { value: number } } }
type test3 = DeepOmit<obj, 'name'> // { person: { name: string; age: { value: number } } }
type test4 = DeepOmit<obj, 'person.age.value'> // { person: { name: string; age: {} } }

Detailed Explanation

type DeepOmit<T, S extends string> = {
  [K in keyof T as K extends (S extends `${infer First}.${string}` ? First : S) ? never : K]:
    S extends `${infer First}.${infer Rest}`
      ? K extends First
        ? DeepOmit<T[K], Rest>
        : T[K]
      : T[K]
}

How it works:

• The mapped type iterates over all keys K of T
• The as clause filters out keys: if the path S contains a dot (e.g., 'person.name'), we extract the first segment; if there is no dot, S itself is the key to omit. If K matches, it is mapped to never (removed)
• For the value side, if S is a dotted path like 'person.name', we check whether K matches the first segment (person). If so, we recursively apply DeepOmit<T[K], Rest> to dig deeper into the nested object
• If K does not match the first segment, or if S has no dots, we keep the value as T[K]
• This handles cases like DeepOmit<obj, 'person'> (removes the top-level key), DeepOmit<obj, 'person.name'> (removes a nested key), and DeepOmit<obj, 'name'> (no match at top level, so the object is unchanged)

This challenge helps you understand recursive mapped types with key filtering and dot-path string parsing, and how to apply these concepts in real-world scenarios.