Design questions about json-schema-validator

[fge]

You probably mean JsonTree?

This class is meant to be used for JSON instances to validate. For reporting purposes, I want the path into the instance to be available.

When I .append(), I navigate into the node, and push the current context, and node, into the queues. When I .pop(), I get back to the previous context by pop()ing from the queues.

For instance, when I validate an array, I do:

• validate the array itself;
• if it is valid, go through its elements:
  • grab the schema matching an index;
  • append that index using the tree;
  • validate the node at this index;
  • pop.

Recursively so.

[fge]

As to building a JSON Schema, I don't really build one: it is provided to me as a JsonNode. I extract information out of it and process with ref resolving, syntax checking and instance validation in that order.

But that is the part which I am currently refactoring!

[fge]

Oh, and finally, there is also JsonSchemaTree: it does the same as JsonTree, except that it also pushes/pops resolution context information (necessary for $ref).

[reinert]

Who is responsible for navigating into the tree and validante it's nodes recursively?

[fge]

This is not commited yet. It will be a series of (for now) SchemaProcessors.

My intent is to make it a sufficiently generic interface to be used for ref resolution, syntax checking and instance validation.

And maybe even more in the future.

[fge]

Closing this one, will put word in mouth and plug JJSchema.