Flatten Nested JSON / HashMap to String

Given a nested JSON object (or equivalent nested HashMap), flatten it into a single-level representation. The natural representation is dotted key paths (`a.b.c -> value`) — recursion on values that are themselves objects, base case on scalars. Requirements Input is a parsed JSON / nested map whose leaf values are scalars (string / number / boolean / null) and whose interior values are nested objects. Output is a flat representation. Two common shapes interviewers accept: A map { dotted.path: scalar, ... }. A string serialisation of that map (sorted or unsorted as specified). Array values: not always required. Clarify with the interviewer; if needed, use index-bracket notation (a.b[0].c). Empty object value: clarify whether to emit the path with null / "{}" or to drop it. Examples Input: { "a": 1, "b": { "c": 2, "d": { "e": 3 } } } Output (dotted path map): { "a": 1, "b.c": 2, "b.d.e": 3 } Notes The standard solution is a recursive DFS: flatten(node, prefix, out): for k, v in node.items(): path = prefix + "." + k if prefix else k if v is a dict: flatten(v, path, out) else: out[path] = v Time O(n) where n is the total number of map entries / visited nodes; auxiliary space O(d) for recursion depth d, plus O(number of leaves) for the required output. An iterative version uses an explicit stack of (prefix, node) pairs — useful when interviewers ask about extremely deep inputs (recursion blows the call stack at d ≈ 10⁴). Edge cases to surface: keys containing the path separator (a.b as a key collides with nested form — usually resolved by escaping or by clarifying that keys are alphanumeric); duplicate paths (impossible if input is a valid map, but worth stating); empty top-level object (return {}). For the string-output variant, decide separator semantics up-front: "a.b.c=2;b.d.e=3" is one common form; another is JSON-Pointer (/b/c). Common interviewer follow-up: write the inverse (unflatten) — take a dotted-key map and rebuild the nested structure. Same recursion in reverse. Preparation Write both the recursive and the iterative version. Verify on a deeply nested example (depth 5+) and a single-key example. Practise the unflatten follow-up — the trick is to split the path at the first separator and recurse on the rest. Have an answer for: "what if values can be arrays?" — extend the recursion to call flatten on each element with prefix[i]. Walk through correctness out loud — interviewers will pause the candidate at the recursion line to confirm the base case is clearly stated.