Reputation: 39
Looking to improve my python and coding skills. I have a function that adds a particular timeframe to a time. I pass in:
1M, 7D, 6M, 2H, M
etc .. and return the value. I feel like I am repeating myself. Is there a more pythonic approach to this?
def add_timeframe(time, timeframe):
if 'H' in timeframe:
┆ try:
┆ ┆ period = int(re.sub('\D', '', timeframe))
┆ ┆ return convert_datetime(time + datetime.timedelta(hours=period))
┆ except ValueError:
┆ ┆ return convert_datetime(time + datetime.timedelta(hours=1))
if 'D' in timeframe:
┆ try:
┆ ┆ period = int(re.sub('\D', '', timeframe))
┆ ┆ return convert_datetime(time + datetime.timedelta(days=period))
┆ except ValueError:
┆ ┆ return convert_datetime(time + datetime.timedelta(days=1))
if 'W' in timeframe:
┆ try:
┆ ┆ period = int(re.sub('\D', '', timeframe))
┆ ┆ return convert_datetime(time + datetime.timedelta(weeks=period))
┆ except ValueError:
┆ ┆ return convert_datetime(time + datetime.timedelta(weeks=period))
if 'M' in timeframe:
┆ try:
┆ ┆ period = int(re.sub('\D', '', timeframe))
┆ ┆ return convert_datetime(time + datetime.timedelta(days=365/12*period))
┆ except ValueError:
┆ ┆ return convert_datetime(time + datetime.timedelta(days=365/12))
Upvotes: 0
Views: 85
Reputation: 15568
I usually avoid lots of ifs by using dictionaries. I map each condition to a dictionary and execute. Here is my first take:
I created a function for adding months as timedelta does not have it. I then use re
to get digit and letter as tuples. So '4M' would be ('4','M'). Then I would map M to month addition, so 4 *(add months function), W for add weeks etc
import calendar
import datetime
import re
# add month hack
def add_month(num_months, date=None):
'''Add N months'''
assert num_months > 0, 'Positive N only'
if date is None:
date = datetime.datetime.now()
for num in range(num_months):
month_days = calendar.monthrange(date.year, date.month)[1]
dt = date + datetime.timedelta(days=month_days)
if dt.day != date.day:
dt.replace(day=1) - datetime.timedelta(days=1)
else:
dt
date = dt
return dt
def delta(data, pattern):
# dict instead of lots of ifs
time_convert = {'M': lambda x : add_month(x),
'W': lambda x :datetime.timedelta(weeks=x),
'D': lambda x: datetime.timedelta(days=x),
'H': lambda x: datetime.timedelta(hours=x),
}
_ = [re.match(pattern, item).groups() for item in data]
return [time_convert.get(letter)(int(number))for number, letter in _]
# test 1
data = ['1M', '7D', '4M', '2H']
pattern = '(\d+)(\w+)'
s = delta(data, pattern)
print(s)
If we are expecting unclean data, we need to create a data preparation function that will ensure that our data is in a format we want digitsletter(s). As our code will fail if we receive just letter data. Instead of try-catch, If we have just a letter, we could add a 1. This would be my take 2:
import calendar
import datetime
import re
# add month hack
def add_month(num_months, date=None):
'''Add N months'''
assert num_months > 0, 'Positive N only'
if date is None:
date = datetime.datetime.now()
for num in range(num_months):
month_days = calendar.monthrange(date.year, date.month)[1]
dt = date + datetime.timedelta(days=month_days)
if dt.day != date.day:
dt.replace(day=1) - datetime.timedelta(days=1)
else:
dt
date = dt
return dt
def data_prep(data, check_patter='\d+'):
'''Our data preparation happens here'''
_ = [bool(re.search(check_patter,item)) for item in data]
for index, truth in enumerate(_):
if not truth:
data[index] = '1'+data[index]
return data
def delta(data, pattern):
time_convert = {'M': lambda x : add_month(x),
'W': lambda x :datetime.timedelta(weeks=x),
'D': lambda x: datetime.timedelta(days=x),
'H': lambda x: datetime.timedelta(hours=x),
}
# clean data. if M --> 1M
data = data_prep(data)
_ = [re.match(pattern, item).groups() for item in data]
return [time_convert.get(letter)(int(number))for number, letter in _]
data = ['1M', '7D', '4M', '2H','H']
pattern = '(\d+)(\w+)'
s = delta(data, pattern)
print(s)
It is in data_prep function where you will deal with all possible uncleanliness:)
Upvotes: 1
Reputation: 44444
You can use re
to extract both the number portion and all the period-chars in a single go..
>>> import re
>>> inp = ["1M", "7D", "6M", "2H", "M"]
>>> [re.findall('(\d)?(M|D|H)', x) for x in inp]
[[('1', 'M')], [('7', 'D')], [('6', 'M')], [('2', 'H')], [('', 'M')]]
>>> extracted = [re.findall('(\d)?(M|D|H)', x) for x in inp]
>>> [(int(x[0][0] or '1'), x[0][1]) for x in extracted if x] # Filter out invalids.
[(1, 'M'), (7, 'D'), (6, 'M'), (2, 'H'), (1, 'M')]
You can then use convert_datetime(..)
and other stuff you are doing in your original code.
PS: I'd perform more error-checking -- the code above is to only suggest a slightly more pythonic way of doing the same thing.
Upvotes: 0