Recursive Type Extraction | Dataford Interview Questions

Globant's frontend teams pass nested configuration objects to design-system components. Implement infer_type_properties(obj) to recursively infer a normalized schema for a JSON-like object.

Return a descriptor with type and, when applicable, additional fields:

Primitive values return {"type": "string"}, {"type": "number"}, {"type": "boolean"}, or {"type": "null"}.
An object returns {"type": "object", "properties": {...}}, where each property maps to its own descriptor.
An array returns {"type": "array", "items": descriptor}. Infer the item descriptor by merging all element descriptors.
An empty array returns {"type": "array", "items": {"type": "unknown"}}.
If values cannot be represented by one descriptor, return {"type": "mixed"}. Arrays containing objects should merge their property keys recursively.
The input itself is always a dictionary. Do not mutate it.

Constraints

obj is a dictionary.
Values are limited to dictionaries, lists, strings, numbers, booleans, and None.
1 <= total number of input nodes <= 10^5.
Dictionary keys are strings.
Nesting depth is at most 1,000.
Array elements may have different types.

Function Signature

def infer_type_properties(obj):

Your question is Recursive Type Extraction. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem