JSON text differences and data meaning
A text comparison and a semantic comparison answer different questions.
Separate formatting from data changes
Two JSON documents can use different indentation and object member order while describing similar data. A text diff will show those presentation changes. Format copies consistently before comparison if that makes review easier. Do not sort arrays to suppress differences: order is part of an array’s meaning. Do not overwrite the originals merely to obtain a smaller diff.
Look for values that only appear equivalent
A string containing digits is not the same JSON type as a number. Likewise a missing key, null and an empty string can mean different things to an API. Use a sample that includes these cases when reviewing a conversion. A textual merge cannot decide which interpretation is required by your application’s schema.
{"id":"0012","note":null}
{"id":12,"note":""}Finish with validation in the real consumer
Inspect duplicate object keys and very large numbers before parsing with a tool that may discard or round them. After accepting edits, validate JSON syntax and test against your application’s schema and rules. DiffTidy highlights text changes; it is not a complete semantic equivalence checker. Keep a short record of why a changed field is intentional so the next reviewer does not have to infer the reason from color highlights alone.