Language specific date formats

[mnikolop]

Hello,

In swedish there are 2 specific ways that dates are written very commonly that I have not seen support for. Specifically dates can be written like "den 24:e December" and "201226" (2020-12-26). The first one is not recognised due to the ":e" which is a very common way of writing dates in sweden and the second returns the wrong date datetime.datetime(2022, 2, 1, 6, 0). I am assuming the second can be rectified with some setting but short of removing the ":e" I cannot see a way of tackling the first one.
Any help is appreciated.
Thank you.

[noviluni]

Hi @mnikolop
You can add additional/custom formats by using the date_formats parameter. For example:

>>> dateparser.parse('201226', date_formats=['%y%m%d'])
datetime.datetime(2020, 12, 26, 0, 0)

This approach could be used for the other date format, but it will only work if the configured locale in your computer is sv_SE:

>>> dateparser.parse('den 24:e Januari', date_formats=['den %d:e %B'])
datetime.datetime(2020, 1, 24, 0, 0)

(I changed to "Januari" to check that it works as expected because "December" is the same word in English).

If it doesn't work for you, you can change temporarily the locale by doing:

>>> import locale
>>> locale.setlocale(locale.LC_TIME, "sv_SE")
'sv_SE'
>>> dateparser.parse('den 24:e Januari', date_formats=['den %d:e %B'])
datetime.datetime(2020, 1, 24, 0, 0)

I hope this will work for you. I'm not going to close this issue as I think that we could think about it and see if we should add these formats with ":e", or try the date_formats with different locales.

Let me know if you find any issues!

[mnikolop]

How could this be applied in the search_dates function without is raising a SettingValidationError?

[noviluni]

Hi @mnikolop the search_dates() function does not support date_formats. I have just opened a new issue to discuss this: #846

As stated in the docs, the support for searching dates needs a lot of improvement, so I really encourage you to add your ideas and PRs :)