In Python json.dumps fails to process dictionary of { integer : 'string' ,}

Question

I have looked for an answer on SO, however no answer was particulary useful. I have a dictionary of values that I have to convert to JSON string. The data looks like this:

In [127]: ddd
Out[127]: 
{'ID': {'condition': '',
  'data': {1: '2',
   2: '3',
   3: '4',
   4: '5',
   5: '6',
   6: '7',
   7: '8',
   8: '9',
   9: '10'},
  'desc_long': 'Id szko\xc5\x82y',
  'desc_short': 'Id szko\xc5\x82y',
  'df_name': 'pierszytest',
}}

When I try to execute json.dumps(ddd) I get :

TypeError: keys must be a string

I created a test variable "what"

In [126]: what
Out[126]: {1: '2', 2: '3', 3: '4', 4: '5', 5: '6', 6: '7', 7: '8', 8: '9', 9: '10'}

json.dumps(what) returns:

Out[129]: '{"1": "2", "2": "3", "3": "4", "4": "5", "5": "6", "6": "7", "7": "8", "8": "9", "9": "10"}'

Again, I tried to convert only ddd['ID']['data']:

 In [131]: ddd['ID']['data']
 Out[131]: {1: '2', 2: '3', 3: '4', 4: '5', 5: '6', 6: '7', 7: '8', 8: '9', 9: '10'}

 In [130]: json.dumps(ddd['ID']['data'])

 TypeError: keys must be a string

So, this are basically the same variables and still json.dumps fails to deal with the latter. This is a big surpise for me. I have done some research, but it returned nothing useful for my case. If it's a duplicate, please let me know.

* EDIT * I attach result of type() on both variables:

In [132]: type(ddd['ID']['data'])
Out[132]: dict

In [133]: type(what)
Out[133]: dict

What is type(ddd['ID']['data'])? And can you include the definition of that type in the question, or at least link to it? — phihag
– phihag, Commented Jul 5, 2013 at 11:01
I have edited the question to attach result of running "type" on both variables — mhnatiuk
– mhnatiuk, Commented Jul 5, 2013 at 11:22
The problem is that the keys are numpy objects. See my updated answer for a solution. — phihag
– phihag, Commented Jul 5, 2013 at 12:50

Community · Accepted Answer · 2021-10-07 06:15:07Z

1

The error message is correct; keys in JSON objects must be strings, according to the JSON standard:

object = begin-object [ member *( value-separator member ) ]
         end-object
member = string name-separator value

Therefore, either choose another format than JSON, or let your keys be strings. In your case, you want to convert numpy.int64 variables to strings; try the following:

# This paragraph is just for testing, replace it with your actual code
import pandas  as pd
import numpy as np
d = pd.Series([1,3,5,np.nan,6,8]).to_dict()
ddd = {'ID': {'DATA': d}}


import json
import copy
def _convert(dct):
    return dict((str(k),v) for k, v in dct.items())
ddd_json = copy.deepcopy(ddd)
ddd_json['ID']['data'] = _convert(ddd_json['ID']['data'])

print(json.dumps(ddd_json, indent=2))

edited Oct 7, 2021 at 6:15

CommunityBot

11 silver badge

answered Jul 5, 2013 at 10:56

phihag

289k75 gold badges475 silver badges489 bronze badges

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

Comments

E1000i · Accepted Answer · 2013-07-05 11:23:42Z

1

In fact Json needs as key a string. Is mandatory that they keys must be an integer? Just take this into account when you tray to deserialize the json an manage type conversion.

When deserializing: int(key) to get agin the integer. I supose is posible.

edited Jul 5, 2013 at 11:23

answered Jul 5, 2013 at 11:03

E1000i

1,5492 gold badges11 silver badges10 bronze badges

5 Comments

mhnatiuk Over a year ago

They don't have to be integers, however look at the comparison between ddd['ID']['data'] and what. They're both dicts, so the issue must be laying deeper than that. I don't have to actually use integer type for the keys, that's the return format of pandas.DataFrame.to_dict() function. I can convert it of course, however, I want to understand that and also I don't want to do the conversion unless it's the only way to deal with this problem. I don't understand why it works with one var and doesn't with the other.

E1000i Over a year ago

well, actually a copied and pasted your code, and it serialised correctly on my python shell. json.dumps(ddd['ID']['data']) '{"1": "2", "2": "3", "3": "4", "4": "5", "5": "6", "6": "7", "7": "8", "8": "9", "9": "10"}'

mhnatiuk Over a year ago

yes, I have copied and pasted it too and it works, however, when I retrieve data from pandas.DataFrame.to_dict() it fails to serialize. How strange is that?

E1000i Over a year ago

As @phihag sugested, check the types that pandas.DataFrame.to_dict() is returning. The only thing come tu my mind is that actually what you will use as key, is not even an integer... in that case you probably will need to write the custom method he suggested.

mhnatiuk Over a year ago

Thanks, I will investigate this further. As of now, I just converted keys to string as a temporay workaround. However, I am beware of a constant nature of temporary workarounds. Thanks for helping me out here.

Collectives™ on Stack Overflow

In Python json.dumps fails to process dictionary of { integer : 'string' ,}

2 Answers 2

Comments

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related