add functionality similar to Pendulum's Interval.in_words

[theavey]

Hi there:

Just came across this recently and it looks very interesting for my use cases. So far, the only thing that I have seen that I would like added is some method similar to Interval.in_words.

It seems like it would be a pretty similar implementation to methods like format_common_iso and in_hrs_mins_secs_nanos, but with the "niceness" of an easy-for-a-human-to-read format.

IMO, the hardest part of this implementation would be localization, which is indeed a challenge.

This is obviously not a requirement for a functional datetime library, but I wanted to suggest what I saw as one of the more useful aspects of Pendulum.

[ariebovenberg]

@theavey thanks for taking the time to post this issue. Good suggestion! I agree that localization would be both useful, and hard. I think it'd be a good addition to whenever

However, priority-wise, I'll probably only get to it after the 1.0 release. My reasons for doing this:

• It's hard to get right, and there are other more high-prio things taking up time
• Adding this feature won't be a breaking change, so no reason to get to this before 1.0
• Temporal for JS also skips localization in its first version
• localization is a bit lower prio in a language like Python, which isn't as often on the direct interface with users like JS or C# is.
• As a crude fallback, in_hrs_mins_secs_nanos exists for devs to write their own

[BurntSushi]

Temporal for JS also skips localization in its first version

I don't think this is quite the right characterization. It doesn't so much as skip localization as it does delegate it to Intl.DateTimeFormat. See: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Temporal/ZonedDateTime/toLocaleString

I would definitely advise caution on the localization front. My perception is that its complexity is comparable to the complexity of the datetime library itself. A huge chunk of ICU4X is devoted to localizing datetimes. (The calendar, datetime, locale, and time modules specifically.)

[theavey]

agree this is not (or should not be: not my place to say) a high priority

If you wanted to delegate localization to the user/developer, could do something like

def time_delta_in_words(
    time_delta,
    hours_word="hours",
    minutes_word="minutes",
    seconds_word="seconds",
    nanos_word="nanoseconds",
    join_word="and",
):
    hmsn = time_delta.in_hrs_mins_secs_nanos()
    list_output = [
        f"{quantity} {word}"
        for quantity, word in zip(
            hmsn, (hours_word, minutes_word, seconds_word, nanos_word)
        )
        if quantity
    ]
    if len(list_output) > 1:
        list_output[-1] = f"{join_word} {list_output[-1]}"
    return ", ".join(list_output)

[ariebovenberg]

Another interesting idea is to integrate with humanize, which already supports expressing date & time in various languages.

[kdeldycke]

Whenever is cool but the only feature left-over for me to switch from Arrow is that humanize-like function: https://arrow.readthedocs.io/en/latest/guide.html#humanize . Keeping track of progress in his discussion then! :)