[Python] Parsing dates from strings that contain unrelated text while avoiding parsing...

What I want to Achieve
My goal is to extract a date that has at least a day and a Month, but could also have a minute, hour, and year.

I want to avoid the parser finding integers and thinking that implies a day of the current month.

Furthermore, I also want the parser to find a date that is only a small part of a larger string.

Think:

'Today is the most wonderful 27th of March 2025' = datetime.datetime(2025, 3, 27, 0, 0)
'2gb of ram' != datetime.datetime(2025, 3, 2, 0, 0) (Assuming we are currently in March)

What I tried so far

Using the fuzzy=True argument in dateutils

from dateutil import parser

#OK: Correct Datetime is returned: datetime.datetime(2025, 3, 30, 0, 0)
parser.parse('Today is the most wonderful 30th March 2025', fuzzy=True)

#NOT OK: Integer is not ignored, Datetime is returned: datetime.datetime(2025, 3, 2, 0, 0)
parser.parse('2 is my lucky number', fuzzy=True)


Using the REQUIRED_PARTS setting in dateparser

frome dateparser import parse

# OK: Returns correct datetime: datetime.datetime(2025, 3, 30, 0, 0
parse('30 March', settings={'REQUIRE_PARTS': ['month', 'day']})

#OK: Integer is Ignored, no datetime returned
parse('30', settings={'REQUIRE_PARTS': ['month', 'day']})

#NOT OK: Datetime Should be Found
parse('Today is the most wonderful 30th of March', settings={'REQUIRE_PARTS': ['month', 'day']})


It would be great if I could combine the fuzzy=True argument from the dateutils module with the settings argument from the dateparser module, but seeing as they are separate modules, that is not feasible.

Is there another way to achieve the same functionality?

Continue reading...