Any way to know the current object being parsed?

[TheSauceMan]

Hi,
I am dealing with very large objects and potentially large arrays of stuff. I'm not sure if I can tell from start_object or end_object the name of the object that was being parsed. For instance, if start_object told me the name I could choose to ignore that name (by passing a dummy dictionary) rather than loading it into memory.

Take this for example:

{
  "hello": "world",
  "obj": {
    "obj_val": "first_obj_val",
    "obj_val2": "second_obj_val",
    "obj_val3": 3
  },
  "arr": [
    2, 4, 6, 8
  ]
}

I only get callbacks on start of the object (twice, but nothing telling me it is unnamed or called "obj"), the end of the "obj" and main object, the end of the "arr" array, and strings for "first_obj_val" and "second_obj_val". I wish I knew the keys / names for any of these.

For example, there's no way of knowing that "first_obj_val" refers to "obj_val" (or better yet, "obj.obj_val") until the very end.

Am I doing it wrong?

[lelit]

No, there's no way of knowing that, indeed.

Maybe the decoder class could expose that information, possibly as a property returning something like a JSON Pointer?

[lelit]

Looking at the code, it does not seem very difficult to do that: internally, the decoding logic uses an handler that has access the decoder instance: the handler maintains a stack of context structures, one for each level, so that property could basically be implemented by traversing the stack and building a pointer value.