ENH: Add date_format and date_unit to to_dict similar to what exists in to_json

[lucasjamar]

Feature Type

• Adding new functionality to pandas

• Changing existing functionality in pandas

• Removing existing functionality in pandas

Problem Description

df.to_dict is often (maybe even mostly) to post data via REST API.
My usual workflow for posting data looks like this:

df = pd.DataFrame({"date": pd.to_daterange("2021-01-01", "2024-01-01")})
df["values"] = 1
df["date"] = df["date"].map(lambda x: x.isoformat())
df = df.to_dict("records")
df["some_key"] = "some_value"
response= requests.post(url, data=df)

Ideally, the datetime to string could happen inside of to_dict via an optional date_format and date_unit parameter similar to df.to_json.
However, df.to_json is not a suitable alternative because often I need to add an additional key as seen above.
If this seems like an appropriate new feature, I will dig into how to implement this.

Feature Description

Add 2 new parameters to to_dict:

1. date_format. Possible values: None, Epoch, isoformat. Default: None. If None, do nothing.
2. date_unit. One of ‘s’, ‘ms’, ‘us’, ‘ns’ . Default ms

Alternative Solutions

None really

Additional Context

Code for to_dict:

pandas/pandas/core/methods/to_dict.py

Line 56 in 1be2637

def to_dict(

to_json: https://github.com/pandas-dev/pandas/blob/v2.2.3/pandas/core/generic.py#L2428-L2717

[rhshadrach]

Thanks for the request. I'm negative here - this can be done with two lines of code:

columns = df.select_dtypes("datetime").columns
df[columns] = df[columns].map(lambda x: x.isoformat())

Adding it to the API would increase the maintenance burden on pandas while not providing users with any new functionality that wasn't readily achievable.