ENH: Add a Pandas display option and/or DataFrame.to_string parameter to control datetime formatting

[gwerbin]

Is your feature request related to a problem?

It would be useful to set a display option that controls how datetimes are formatted. This is useful for both "pretty printing" purposes in situations e.g. Styler doesn't work for some reason, and for easier visual debugging and inspection of the data.

The spirit of this enhancement request is similar to #46744, and perhaps there is some room for unification here, e.g. a display option that takes a dict of dtype_or_lambda -> formatter functions.

Describe the solution you'd like

One solution is to add a dispaly option, display.date_format, with behavior analogous to the date_format= parameter in DataFrame.to_csv.

Another option is to add an analogous date_format= parameter to DataFrame.to_string. In general, is there any reason why display options shouldn't all be mirrored to DataFrame.to_string?

You can also add a general-purpose display option and/or to_string parameter that controls output formatting in more detail. It could accepts either a function or a dict. If a function, the function would be applied to format all entries. The default value would be str, for backward compatibility; if you set this to repr, it would cover the use case in #46744. If the value is a dict, the keys would be any of a literal dtype object, a column label, or a function that can be used to dispatch on dtype and/or even column name. And the values would be callables that format individual entries in the series. For example:

text_output = my_data.to_string(
    formatters={
        pandas.api.types.is_datetime64_any_dtype: lambda x: x.dt.strftime('...'),  # Format the 'timestamp' column
        '': repr,   # Handle all other columns with a special dict key
    }
)

API breaking implications

None that I know of.

Describe alternatives you've considered

Manually applying .strftime is ad-hoc, and is undesirable in some cases (e.g. very large data frames).

Additional context

Inspired by a StackOverflow question, and of the same general category as #46744.

[mroeschke]

Thanks for the request, but it appears this feature request hasn't gain much traction in years so closing