|
1
|
|
|
import datetime |
|
2
|
|
|
|
|
3
|
|
|
import pandas as pd |
|
4
|
|
|
|
|
5
|
|
|
|
|
6
|
|
|
def normalize_data_query_time(dt, time, tz): |
|
7
|
|
|
"""Apply the correct time and timezone to a date. |
|
8
|
|
|
|
|
9
|
|
|
Parameters |
|
10
|
|
|
---------- |
|
11
|
|
|
dt : pd.Timestamp |
|
12
|
|
|
The original datetime that represents the date. |
|
13
|
|
|
time : datetime.time |
|
14
|
|
|
The time to query before. |
|
15
|
|
|
tz : tzinfo |
|
16
|
|
|
The timezone the time applies to. |
|
17
|
|
|
|
|
18
|
|
|
Returns |
|
19
|
|
|
------- |
|
20
|
|
|
query_dt : pd.Timestamp |
|
21
|
|
|
The timestamp with the correct time and date in utc. |
|
22
|
|
|
""" |
|
23
|
|
|
# merge the correct date with the time in the given timezone then convert |
|
24
|
|
|
# back to utc |
|
25
|
|
|
return pd.Timestamp( |
|
26
|
|
|
datetime.datetime.combine(dt.date(), time), |
|
27
|
|
|
tz=tz, |
|
28
|
|
|
).tz_convert('utc') |
|
29
|
|
|
|
|
30
|
|
|
|
|
31
|
|
|
def normalize_timestamp_to_query_time(df, |
|
32
|
|
|
time, |
|
33
|
|
|
tz, |
|
34
|
|
|
inplace=False, |
|
35
|
|
|
ts_field='timestamp'): |
|
36
|
|
|
"""Update the timestamp field of a dataframe to normalize dates around |
|
37
|
|
|
some data query time/timezone. |
|
38
|
|
|
|
|
39
|
|
|
Parameters |
|
40
|
|
|
---------- |
|
41
|
|
|
df : pd.DataFrame |
|
42
|
|
|
The dataframe to update. This needs a column named ``ts_field``. |
|
43
|
|
|
time : datetime.time |
|
44
|
|
|
The time to query before. |
|
45
|
|
|
tz : tzinfo |
|
46
|
|
|
The timezone the time applies to. |
|
47
|
|
|
inplace : bool, optional |
|
48
|
|
|
Update the dataframe in place. |
|
49
|
|
|
ts_field : str, optional |
|
50
|
|
|
The name of the timestamp field in ``df``. |
|
51
|
|
|
|
|
52
|
|
|
Returns |
|
53
|
|
|
------- |
|
54
|
|
|
df : pd.DataFrame |
|
55
|
|
|
The dataframe with the timestamp field normalized. If ``inplace`` is |
|
56
|
|
|
true, then this will be the same object as ``df`` otherwise this will |
|
57
|
|
|
be a copy. |
|
58
|
|
|
""" |
|
59
|
|
|
if not inplace: |
|
60
|
|
|
# don't mutate the dataframe in place |
|
61
|
|
|
df = df.copy() |
|
62
|
|
|
|
|
63
|
|
|
dtidx = pd.DatetimeIndex(df.loc[:, ts_field], tz='utc') |
|
64
|
|
|
dtidx_local_time = dtidx.tz_convert(tz) |
|
65
|
|
|
to_roll_forward = dtidx_local_time.time > time |
|
66
|
|
|
# for all of the times that are greater than our query time add 1 |
|
67
|
|
|
# day and truncate to the date |
|
68
|
|
|
df.loc[to_roll_forward, ts_field] = ( |
|
69
|
|
|
dtidx_local_time[to_roll_forward] + datetime.timedelta(days=1) |
|
70
|
|
|
).normalize().tz_localize(None).tz_localize('utc') # cast back to utc |
|
71
|
|
|
df.loc[~to_roll_forward, ts_field] = dtidx[~to_roll_forward].normalize() |
|
72
|
|
|
return df |
|
73
|
|
|
|
quantopian /
zipline
