I'm trying to save a pandas.DataFrame object to disk, as a JSON. The slight complication is that I need to store some additional data alongside it. So my intended JSON would be

{
  "additionalData": ...,
  "table": {
    "schema": ...,
    "data": ...
  }
}

I can obtain the desired layout for table by doing df.to_json(orient='table'), however this a str and not a dict. So if I do something like

dict_to_write = {
    'additionalData': ...,
    'table': df.to_json(orient='table'),
}

when dict_to_write gets written to disk (which already involves a json.dumps step), the JSON looks like

{
  "additionalData": ...,
  "table": "{\"schema\": ...,\"data\": ...}"
}

where the table has been double-stringified.

I can fix this by doing

dict_to_write = {
    'additionalData': ...,
    'table': json.loads(df.to_json(orient='table')),
}

instead. This turns df first into a string, and then into a dict with the desired structure, stopping it from being double-stringified.

It seems like df.to_dict should make this possible in one call to me. Is there any particular reason df.to_dict(orient='table') is not supported?

(I'm not sure if this is best placed here, as it's not really a question and more of a discussion, please suggest an alternative place if not.)