def wash_data(self, data_to_wash):
# replace any non-word character with a forward slash
data_to_wash = Wa.replace_x_with_y("\W+", "/", data_to_wash)
# remove st, nd and rd from date, e.g. 21st, 22nd, 23rd
data_to_wash = Wa.replace_x_with_y("st|nd|rd", "", data_to_wash)
data_to_wash = self.month_string_to_number(data_to_wash)
return data_to_wash
def determine_month_format(self, data):
# default to number format like 09 for September
result = "%m"
date_to_check = Wa.replace_x_with_y("\W+", "/", data)
split_date = self.split_string("/", date_to_check)
for value in split_date:
if value.isalpha():
if len(value) > 3:
result = "%B"
else:
result = "%b"
return result
def is_valid(self, data_to_validate):
result = False
data_to_validate = data_to_validate.lstrip(' ')
# If there's no numbers in string, just return as is, it's bad data
if Va.has_this_many_numbers(0, data_to_validate):
date_output = data_to_validate
else:
washed_data = self.wash_data(data_to_validate)
# add zeros if needed
washed_data = self.add_zeros(washed_data)
# remove all spaces
washed_data = washed_data.strip()
date_format = self.determine_date_format(washed_data)
date_to_check = Wa.replace_x_with_y('\W+', " ", washed_data)
result = self.is_real_date(date_to_check, date_format)
date_output = Wa.replace_x_with_y(" ", "/", date_to_check)
return date_output, result
def month_string_to_number(self, data):
date_to_check = Wa.replace_x_with_y("\W+", "/", data)
split_date = self.split_string("/", date_to_check)
for value in split_date:
if value.isalpha():
if len(value) > 3:
# result = "%B"
text_month = datetime.strptime(value, '%B')
split_date[1] = text_month.strftime('%m')
else:
# result = "%b"
text_month = datetime.strptime(value, '%b')
split_date[1] = text_month.strftime('%m')
output = ""
try:
output = split_date[0] + "/" + split_date[1] + "/" + split_date[2]
except IndexError:
pass
return output