JSON Structure Inspector

Features

• Nesting depth, total node count, number of objects and arrays
• Type breakdown (string × 12, number × 5 …), sorted by frequency
• The longest array, plus every array length listed
• Longest string length and total string length
• ⚠️ Duplicate keys: scans the raw text for repeated keys in one object, with path and count, and explains that JSON.parse keeps only the last
• Shuffle / reverse / dedupe, applied recursively to arrays at every level
• Dedupe compares values by their JSON text — `1` and `"1"` are different, two identical objects are not
• Parsing reuses the formatter’s error locator (line and column, and it recognises Python literals)
• Everything runs in the browser; nothing is uploaded

How to use

1. Paste your JSON
2. Leave the action on structure stats — the numbers appear immediately
3. Look at the duplicate-keys row; it says “none” explicitly when there are none
4. For tidying up, pick shuffle, reverse or dedupe to get formatted JSON back

FAQ

Why do duplicate keys matter? The value my code reads looks right

Because you are reading the last one. JSON.parse keeps the later value and discards the earlier, and the discard happens at parse time, not read time — so the value you get is “correct”, just not necessarily from the place you assume. The real trouble is collaborative: the server says it returned A, and A is indeed in the JSON you paste (its first occurrence), but your code receives the B that came later. This tool points at both, so you can discuss the same line.

Why not just say JSON must not have duplicate keys?

Because the spec does not forbid it — RFC 8259 leaves the behaviour to the implementation. Parsers disagree: JavaScript keeps the last, Python’s json keeps the last, and some strict parsers reject the document outright. The same file yielding different results in different languages is exactly what makes it dangerous. This tool does not judge; it states the facts.

Where does the depth count start?

The root counts as level 1, so `{"a":1}` has depth 2 (the root object plus the number) and `{"a":{"b":1}}` has depth 3. That makes depth mean “how many nodes you pass through from the root to the deepest value”, which is closer to intuition than counting brace levels.

Is the shuffle cryptographically random?

It uses `Math.random` and a Fisher–Yates shuffle. ⚠️ It deliberately does not use `crypto.getRandomValues`, because the goal is an unpredictable ordering of your data, not cryptographic strength. For unpredictable values (passwords, tokens), use the password generator or random number tools.