def test_no_spaces_strict_parsing(date_string, expected_result):
parser = DateDataParser(settings={
'PARSERS': ['no-spaces-time'],
'STRICT_PARSING': False
})
assert parser.get_date_data(date_string)['date_obj'] == expected_result
parser = DateDataParser(settings={
'PARSERS': ['no-spaces-time'],
'STRICT_PARSING': True
})
assert parser.get_date_data(date_string)['date_obj'] is None
def get_mount_number(name: str) -> int:
"""
Get month number by name
Use `dateparser` for cross-platform solution. Develop on Mac os, use on Windows
Because native solution have different name in module calendar, for example
on Windows `Январь` on Mac Os `января`
"""
ddp = DateDataParser(languages=['ru'])
date_data = ddp.get_date_data(f'1 {name}')
return date_data.date_obj.month
class InputParser:
LANGUAGES = ["en"]
DATE_PARSER_SETTINGS = {
"STRICT_PARSING": False,
"NORMALIZE": True,
"RETURN_AS_TIMEZONE_AWARE": True,
"PREFER_DATES_FROM": "past",
}
def __init__(self):
self._date_parser = DateDataParser(languages=self.LANGUAGES,
settings=self.DATE_PARSER_SETTINGS)
def _as_datetime(self, date_time):
dt_obj = self._date_parser.get_date_data(date_time)
if dt_obj:
return dt_obj["date_obj"]
def parse_datetime(self, date_time):
dt = date_time.strip()
if not dt:
return None
parsed_dt = self._as_datetime(dt)
if not parsed_dt:
raise InvalidDateTimeError
if parsed_dt > self._as_datetime("now"):
raise DateTimeInFutureError
return parsed_dt
def parse(self, work_desc):
if "@" in work_desc:
work, _, date_time = work_desc.rpartition("@")
else:
work, date_time = (work_desc, "")
work = work.strip()
if not work:
raise InvalidWorkError
date_time = self.parse_datetime(date_time)
return work, date_time
def test_confidence_threshold_setting_is_applied():
ddp = DateDataParser(detect_languages_function=detect_languages, settings={'LANGUAGE_DETECTION_CONFIDENCE_THRESHOLD': 0.6})
assert ddp.get_date_data('21/06/2020').locale == 'en'
ddp2 = DateDataParser(detect_languages_function=detect_languages, settings={'LANGUAGE_DETECTION_CONFIDENCE_THRESHOLD': 0.4})
assert ddp2.get_date_data('21/06/2020').locale == 'fr'