`datetime.strptime` accepts incompatible arguments

[pganssle]

Bug report

Bug description:

If you over-specify a datetime strptime string, it seems that datetime.strptime when encountering a component that has been specified twice, will overwrite earlier values with more recent ones.

>>> datetime.strptime("200099", "%Y%y")
datetime.datetime(1999, 1, 1, 0, 0)

One approach would be to ban over-specified formats, but that would be a problem for someone who wants to, use a format like "%M %B" (e.g. something sortable by month number while also showing month name), so I think we don't want to do that.

Alternatively we could start raising an exception if you get different values from these two things, or possibly a warning.

Or we can just accept that if you use a weird format you should expect weird behavior.

CPython versions tested on:

CPython main branch

Operating systems tested on:

Linux

Linked PRs

• GH-124549: Add a note that when over-specifying datetime format, only the last directive gets used. #125009
• gh-124549: Warn on conflicting directives in datetime.*.strptime #132524

[Mariatta]

I'd be ok with raising exception, seems like it's a user error for passing these kind of conflicting format specifiers.

We could also start documenting this case, saying that the latest format specifiers that appears will prevail.

[Mariatta]

@pganssle I created a PR to document this weird behavior. I think it's still worth raising a warning though. #125009

[alliesw]

To avoid the issue where the last directive overwrites earlier values:
since more than one directive is needed, maybe split the string and parse the components manually (handling the split and parsing based on the length of the string)?
Maybe also apply custom logic to handle dynamic formats, especially if you encounter different scenarios with varying date formats.... (idk if your format is dynamic?)

[alliesw]

This is an example of what I'm getting at:

def parse_date(input_str):
    if len(input_str) == 6:  # 4-digit year followed by 2-digit year
        full_year = input_str[:4]
        short_year = input_str[4:]
        year_full = datetime.strptime(full_year, "%Y")
        year_short = datetime.strptime(short_year, "%y")
        return year_short.replace(year=year_full.year)  # Use full year from the first part
    elif len(input_str) == 4:  # Only 4-digit year
        return datetime.strptime(input_str, "%Y")
    else:
        raise ValueError("Invalid date format")

Example usage:

input_str = "200099"
parsed_date = parse_date(input_str)
print(parsed_date)  # datetime.datetime(1999, 1, 1, 0, 0)

[StanFromIreland]

This is a stdlib issue (strptime.py) the extension-modules label can be removed.