Python & pandas: Batching data where difference between timestamps < set value

I'm trying to create a data set in python (preferably pandas) that groups together all rows where the amount of time between the end_time of the last entry and the start_time of the subsequent one is < 10 minutes.

Example data:

In the above, the solution would batch together foo/bar rows as one output, and baz/bay rows for another.

Some traits of the data:

• No times overlap (aka there is at most one entry with start_time before and end_time after any given time)
• There may be hundreds/thousands of rows per "batch"
• A batch may go through midnight

I realize this may well be a common problem, but I can't figure out quite how to (elegantly) solve it, or frankly, quite how to elegantly google it. Any thoughts appreciated

Try this:

start_time = pd.to_datetime(df["start_time"])
end_time = pd.to_datetime(df["end_time"])

# Since your data does not have date, we have to assign a dummy date. A row is
# considered to be in the next day if its start_time is before the previous
# row's start_time. This requires your data to be sorted already.
is_next_day = start_time.diff().dt.total_seconds() < 0

# Increment the dummy date to handle mid-night cross-over
delta = pd.to_timedelta(is_next_day.cumsum(), "D")
start_time += delta
end_time += delta

# Assign each row to a batch number based on gap to the previous row
gap = pd.to_timedelta(10, "T")
exceed_gap = (start_time - end_time.shift()) > gap
df["batch_number"] = exceed_gap.cumsum()