python convert str to timedelta

Question

I want to convert time offsets formatted as str to datetime.timedelta objects, then plot the curve of observations versus time.

'5 days, 4:23:52.480000'  ;      17.647166
'5 days, 5:56:09.166000' ;        22.916071
'5 days, 8:21:40.261000'  ;      18.922329
'5 days, 9:53:18.070000'    ;    21.392157
'6 days, 0:07:54.487000'  ;      20.275597

Welcome to SO, could you show your efforts as SO is not a code writing service, please show what the desired plot looks like — EdChum, Jun 03 '16 at 10:16
This looks like you actually want `timedelta`, not `datetime`, correct? — Paul, Jun 03 '16 at 10:17

score 0 · Answer 1 · answered Jun 03 '16 at 10:23

There is no parser for timedeltas. So you have to build one yourself:

import re
delta = '5 days, 4:23:52.480000' 
days, hours, minutes, seconds = re.match('(?:(\d+) days, )?(\d+):(\d+):([.\d+]+)', delta).groups()
total_seconds = ((int(days or 0) * 24 + int(hours)) * 60 + int(minutes)) * 60 + float(seconds)

score 0 · Answer 2 · answered Jun 03 '16 at 14:23

While I don't know of anything that parses text into timedelta, the format you have specified is understood by the pandas.Timedelta constructor, so if you want, you can use that to construct your timedeltas:

import pandas as pd
import matplotlib.pyplot as plt

x = [['5 days, 4:23:52.480000', 17.647166],
     ['5 days, 5:56:09.166000', 22.916071],
     ['5 days, 8:21:40.261000', 18.922329],
     ['5 days, 9:53:18.070000', 21.392157],
     ['6 days, 0:07:54.487000', 20.275597]]

df = pd.DataFrame(x, columns=['TimeOffset', 'Value'])

df['TimeOffset'] = df['TimeOffset'].apply(pd.Timedelta)

plt.plot(df['TimeOffset'], df['Value'])
plt.show()

People who know pandas better, feel free to edit this answer (and remove this message) - I feel like the way I constructed the pd.Timedelta column may be non-idiomatic.

python convert str to timedelta

2 Answers2