[Python] Computing time delta between two columns in Polars

I am struggling to find a good way to compute time delta in Polars. I have the following string data set and would like to compute the duration in hours between start and end. The tricky part is working around midnight.

df = pl.DataFrame(
{
'start_time': [
'23:00',
'00:00'
],
'end_time': [
'24:00',
'01:00'
]
}
)

(
df
.with_columns(
start_time = pl.col('start_time').str.to_time('%H:%M'),
end_time = pl.col('end_time').replace('24:00', '00:00').str.to_time('%H:%M')
)
.with_columns(
duration = (
pl.when(pl.col('end_time') == pl.time(0, 0, 0))
.then(86400000000000)
.otherwise(pl.col('end_time').cast(pl.Int64))
- pl.col('start_time').cast(pl.Int64)) / 3600000000000
)

)


I came up with a solution, but I am very unhappy with it, because it feels unnecessarily complicated. Basically I am substituting end times equal to 24:00 with 00:00, because Polars can't work with the value 24:00. Because of this I have to add a when/then/otherwise to again substitute the 00:00 in the end times with the time in nanoseconds which equal 24:00. Then I am able to cast the times to int, compute the difference in nanoseconds and convert back to hours by dividing through nanoseconds per hour.

Continue reading...