1

This is a near duplicate of How to reliably hash JavaScript objects?, where someone wants to reliably hash javascript objects ;

Now that the json-ld specification has been validated, I saw that there is a normalization procedure that they advertise as a potential way to normalize a json object :

normalize the data using the RDF Dataset normalization algorithm, and then dump the output to normalized NQuads format. The NQuads can then be processed via SHA-256, or similar algorithm, to get a deterministic hash of the contents of the Dataset.

Building a hash of a json object has always been a pain because something like 

sha1(JSON.stringify(object))

does not work or is not guaranteed to work the same across implementations (the order of the keys is not defined of example).

Does json-ld work as advertized ? Is it safe to use it as universal json normalization procedure for hashing objects ? Can those objects be standard json objects or do they need some json-ld decorations (@context,..) to be normalized ?

Community
  • 1
  • 1
Jerome WAGNER
  • 21,986
  • 8
  • 62
  • 77

1 Answers1

4

Yes, normalization works with JSON-LD, but the objects do need to be given context (via the @context property) in order for them to produce any RDF. It is the RDF that is deterministically output in NQuads format (and that can then be hashed, for example).

If a property in a JSON-LD document is not defined via @context, then it will be dropped during processing. JSON-LD requires that you provide global meaning (semantics) to the properties in your document by associating them with URLs. These URLs may provide further machine-readable information about the meaning of the properties, their range, domain, etc. In this way data becomes "linked" -- you can both understand the meaning of a JSON document from one API in the context of another and you can traverse documents (via HTTP) to find more information.

So the short answer to the main question is "Yes, you can use JSON-LD normalization to build a unique hash for a JSON object", however, the caveat is that the JSON object must be a JSON-LD object, which really constitutes a subset of JSON. One of the main reasons for the invention of the normalization algorithm was for hashing and digitally-signing graphs (JSON-LD documents) for comparison.

dlongley
  • 2,078
  • 14
  • 17