to_json should make separators configurable (similar to json.dump)

[bersbersbers]

Code Sample, a copy-pastable example if possible

import numpy as np
import pandas as pd

print(pd.DataFrame([1]).to_json(indent=4))

outputs

{
    "0":{
        "0":1
    }
}

Problem description

Any JSON prettyfier (including bare VS Code) wants to replace all : by : . So output of to_json should behave as json.dump does, at least when indent is not None (in which case the user clearly does not care about file size):

If specified, separators should be an (item_separator, key_separator) tuple.
The default is (', ', ': ') if indent is None and (',', ': ') otherwise.
To get the most compact JSON representation, you should specify (',', ':') to eliminate whitespace.

https://docs.python.org/3/library/json.html#:~:text=To%20get%20the%20most%20compact%20JSON%20representation

Expected Output

{
    "0": {
        "0": 1
    }
}

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.8.2.final.0
python-bits : 64
OS : Linux
OS-release : 4.12.14-lp151.28.32-default
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.0.3
numpy : 1.18.2
pytz : 2019.3
dateutil : 2.8.1
pip : 20.0.2
setuptools : 46.1.1
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : 1.2.8
lxml.etree : 4.5.0
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : None
IPython : None
pandas_datareader: None
bs4 : 4.8.0
bottleneck : None
fastparquet : None
gcsfs : None
lxml.etree : 4.5.0
matplotlib : 3.2.1
numexpr : 2.7.1
odfpy : None
openpyxl : 3.0.3
pandas_gbq : None
pyarrow : None
pytables : None
pytest : None
pyxlsb : None
s3fs : None
scipy : 1.4.1
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : 1.2.0
xlwt : None
xlsxwriter : 1.2.8
numba : None

[ss-is-master-chief]

take

[TomAugspurger]

Let's confirm that we want to suppor tthis first, and what the API will look like (cc @WillAyd). We only recently added support for indent.

[WillAyd]

I think this is fine (actually thought indent provides spacing here already) but don’t think we need a new keyword.

[WillAyd]

Actually indifferent in the keyword. I don’t see why it needs to accept an open ended array of arguments but I suppose it does match the stdlib

[WillAyd]

@ss-is-master-chief if you or whomever wants to work on this, I would suggest looking at the ujson source for the JSON_NO_EXTRA_WHITESPACE checks

pandas/pandas/_libs/src/ujson/lib/ultrajsonenc.c

Line 962 in 25d893b

#ifndef JSON_NO_EXTRA_WHITESPACE

Right now this is a compile-time constant defined in ultrajson.h, but I think we can make a runtime check and just replace with if (enc->indent) > 0 checks.

This would just add spaces when indent is provided, which is a little different than the request but I still think solves the same problem

[ss-is-master-chief]

@WillAyd I'll start off with it right away. Thanks for the information.

[bersbersbers]

@ss-is-master-chief how is it going? :)

[jpmckinney]

I'd be happy to write a patch that passes through separators to json.dumps, the same as is done for indent.

Personally, it feels simpler to just pass any extra keyword arguments to json.dumps, but no problem with the explicit approach.

Edit: Oh, dang, ujson doesn't support separators. I guess it'd have to be done like in #33014 (comment)