I would like some help converting a tuple in format [(X, Y, Z),(..) of type (string, string, int)] to a JSON file in the format:

{"name": "X","children": [{"name": "Y","value": Z}]
}

I have at least 1M values to convert and at the moment I have attempted using a key for the dictionary:

b = (dict(zip(keys,row)) for row in tuples)  

using the JSON library

print (json.dumps(list(b)))  

however this yields a JSON in the format

[{"type": "y", "name": "z", "count": z},...  

Preferably I would like the Y and Z values to be nested under children and the X value to be used once per unique string.

   {"name": "X","children": [{"name": "Y","value": Z},{"name": "Y2","value": Z2}]}

Assuming you want a list of dicts as the output of the json with each dict be the form in your question:

The following one liner will put it into the data structure your looking for with each tuple generating it's own complete structure:

[{'name':i[0], 'children': [{'name': i[1], 'value': i[2]}]}  for i in tuples]

But I suspect you want the outer name to be unique with inner children being built from multiple tuples, such that there is only one such structure generated with 'name' == X.

To do it in a memory efficient manner start by sorting your tuples by X:

# should work as long as you aren't doing any fancy sorting
stuples = sorted(tuples) name = None
dat = None
for t in stuples:if name != t[0]:if dat is not None:writeDat(json.dumps(dat))name = t[0]dat = {'name': t[0], 'children': [{'name': t[1], 'value': t[2]}]}else:dat['children'].append({'name': t1, 'value': t[2]})

I assume you have a function to write one of these out such as writeDat() that takes the json so you aren't storing them all in ram.