如何从pandas中的一系列字符串中提取小时和分钟

Question

I have been stuck on this seemingly simple problem for hours. I would like to convert the following strings to minutes. (Or hours and minutes if I could).

foo['stringtime'] = pd.Series(['1 hour and 59 minutes','2 hours', np.nan, '38 minutes', '4 hours and 31 minutes'])

#What I've tried:
foo['stringtime'] = foo['stringtime'].str.replace(r'hours?','').str.replace(' minutes','').str.split(' and ')

However this would create a situation where '2 hours' and '38 minutes' become ['2'] and ['38']

#What I would like to happen:
foo.head()
output:
119
120
NaN (or 0)
38
271

Is there any beautiful elegant pythonic way to do this?

Answer 1

Try Using Regex.

Ex:

import re

def p_time(val):
    try:
        t = 0
        h = re.search(r"(\d+) hour(s)?", val)
        if h:
            t += int(h.group(1)) * 60
        m = re.search(r"(\d+) minute(s)?", val)
        if m:
            t += int(m.group(1))
        return t
    except:
        pass
    return 0

s = pd.Series(['1 hour and 59 minutes','2 hours', np.nan, '38 minutes', '4 hours and 31 minute'])
print(s.apply(p_time).astype(int))

Output:

0    119
1    120
2      0
3     38
4    271
dtype: int32

Answer 2

Another way might be just to use numexpr to evaluate a numerical equation:

import numexpr

foo = pd.Series(['1 hour and 59 minutes','2 hours', np.nan, '38 minutes', '4 hours and 31 minutes'])

(foo.str.replace(r' hours?','*60').str.replace(' minutes','').str.replace(' and ', '+')
    .fillna('0').apply(numexpr.evaluate))

Output:

0    119
1    120
2      0
3     38
4    271