Why do int keys of a python dict turn into strings when using json.dumps?

Question

According to this conversion table, Python ints get written as JSON numbers when serialized using the JSON module--as I would expect and desire.

I have a dictionary with an integer key and integer value:

>>> d = {1:2}
>>> type(d.items()[0][0])
<type 'int'>
>>> type(d.items()[0][1])
<type 'int'>

When I use the json module to serialize this to a JSON string, the value is written as a number, but the key is written as a string:

>>> json.dumps(d)
'{"1": 2}'

This isn't the behavior I want, and it seems particularly broken since it breaks json.dumps/json.loads round-tripping:

>>> d == json.loads(json.dumps(d))
False

Why does this happen, and is there a way I can force the key to be written as a number?

...and JSON uses strings because they're byte-order independent. — martineau
– martineau, Commented Jun 14, 2013 at 1:01
If JSON is not strictly required, you could go for YAML using PyYAML instead : d = {1:2, 3:4}; import yaml; yaml.safe_load(yaml.safe_dump(d)) == d returns True. In general the 'inline' style of YAML looks like more flexible json (keys can be numbers, strings usually don't have to be quoted). I use safe_load here because normal yaml.load has features (construction of classes,etc) that it's hard to secure correctly; safe_load|dump restrict the set of supported input/outputs to primitive types (bool,int,float,string,list,set,dict) and so are safe to use on arbitrary inputs. — kampu
– kampu, Commented Jun 14, 2013 at 2:15
Unfortunately the round tripping does not hold with yaml.safe_dump because it converts strings of things looking like ints into ints. E.g. try it with d = {'1':'2'} — Timothy Corbett-Clark
– Timothy Corbett-Clark, Commented Oct 24, 2013 at 10:21
Bug for above round tripping issue: bitbucket.org/xi/pyyaml/issue/21 — Timothy Corbett-Clark
– Timothy Corbett-Clark, Commented Oct 24, 2013 at 12:28

Sean Vieira · Accepted Answer · 2013-06-14 00:58:18Z

51

The simple reason is that JSON does not allow integer keys.

object
    {}
    { members } 
members
    pair
    pair , members
pair
    string : value  # Keys *must* be strings.

As to how to get around this limitation - you will first need to ensure that the receiving implementation can handle the technically-invalid JSON. Then you can either replace all of the quote marks or use a custom serializer.

answered Jun 14, 2013 at 0:58

Sean Vieira

161k34 gold badges321 silver badges297 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Paulo Scardine Over a year ago

And unlike plain javascript. all keys should be quoted.

JLT · Accepted Answer · 2018-09-20 11:54:27Z

1

If you really want to, you can check keys for being convertable to integers again using:

def pythonify(json_data):
    for key, value in json_data.iteritems():
        if isinstance(value, list):
            value = [ pythonify(item) if isinstance(item, dict) else item for item in value ]
        elif isinstance(value, dict):
            value = pythonify(value)
        try:
            newkey = int(key)
            del json_data[key]
            key = newkey
        except TypeError:
            pass
        json_data[key] = value
    return json_data

answered Sep 20, 2018 at 11:54

JLT

76010 silver badges16 bronze badges

Comments

Jonas Melin · Accepted Answer · 2019-04-28 06:53:51Z

1

This function will recursively cast all string-keys to int-keys, if possible. If not possible the key-type will remain unchanged.

I adjusted JLT's example below slightly. With some of my huge nested dictionaries that code made the size of the dictionary change, ending with an exception. Anyhow, credit goes to JLT!

def pythonify(json_data):

    correctedDict = {}

    for key, value in json_data.items():
        if isinstance(value, list):
            value = [pythonify(item) if isinstance(item, dict) else item for item in value]
        elif isinstance(value, dict):
            value = pythonify(value)
        try:
            key = int(key)
        except Exception as ex:
            pass
        correctedDict[key] = value

    return correctedDict

edited Apr 28, 2019 at 6:53

answered Apr 25, 2019 at 5:53

Jonas Melin

436 bronze badges

Collectives™ on Stack Overflow

Why do int keys of a python dict turn into strings when using json.dumps?

3 Answers 3

1 Comment

Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related