Iterate through TransientObject items?

[mattdm]

This is somewhat related to #5. Although it is probably safe to assume that the JSON dictionary I'm trying to parse will always be in the same order... it might not.

Therefore, I'd like to do something like:

hourly_file = io.open("hourly.json",'rb')
json_data = adafruit_json_stream.load(hourly_file)
periods = json_data['properties']['periods']

for period in periods:
    for (key,value) in period.items():
        print(f"{key} = {value}")

But, there is no items() in TransientObject. How hard would this be to implement? (The above code works with full json_stream, as does something like:

    for (key,value) in period.items():
        if key == 'number':
            print(f"Number: {value}")
        elif key == 'temperature':
            print(f"Temp:   {value}")
        elif key == 'probabilityOfPrecipitation':
            print(f"Rain%:  {value['value']}")

which is basically exacatly what I want to do (except more than just printing of course).

[mattdm]

I tried briefly to implement this, but ran into problems (not surprisingly) nested dictionaries -- _IterToStream.next_value() returns those as transient objects, which presents a conundrum. If the iterator skips past that, that item can never be read. But if it doesn't, things are kind of in a weird state. (I think this is what the active_child value is for?)

I think maybe the solution is to "realize" any nested dictionaries or lists. That is, if the buffer expected to be a value starts with { or [ , buffer everything until the matching } or ] (including any further nested dictionaries or lists, so it might not necessarily be the next end-char encountered!) and then json.loads() that buffer and return the result as value. And then, document this as an implementation detail, warning people not to iterate over dictionaries with large nested objects unless there's plenty of memory.

Anyway, this turned into a longer lunch break than I intended :) so I am going to stop the yak-shaving here.

[justmobilize]

@mattdm on this the added .as_object() could help (if the object isn't too big) and there is active work on this that may add items() in the future.

[Neradoc]

Ah forgot to say, but this is implemented in #9