[
https://issues.apache.org/jira/browse/ARROW-15883?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17503678#comment-17503678
]
Joris Van den Bossche commented on ARROW-15883:
-----------------------------------------------
The main problem is that there is no clear "standard" for strptime. Python for
example has the "%f" field for fractional
seconds(https://docs.python.org/3/library/datetime.html#strftime-and-strptime-format-codes).
The R / lubridate parser seems to use "%OS"
(https://rdrr.io/r/base/strptime.html,
https://lubridate.tidyverse.org/reference/parse_date_time.html). But the C
strptime does not support either of those
(https://man7.org/linux/man-pages/man3/strptime.3.html; it uses "%OS", but
apparently for something different?)
So as long as we use existing {{strptime}} implementation, I don't it is
possible or easy to "fix" this.
One option I was thinking about is when the users specifies an ISO-like format
string, that we could use our own fast ISO parser, instead of using
{{strptime}}. But that would of course also make the support for fractional
seconds dependent on the exact format you specified.
> [C++] Support for fractional seconds in strptime() for ISO format?
> ------------------------------------------------------------------
>
> Key: ARROW-15883
> URL: https://issues.apache.org/jira/browse/ARROW-15883
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
> Reporter: Joris Van den Bossche
> Priority: Major
> Labels: kernel
>
> Currently, we can't parse "our own" string representation of a timestamp
> array with the timestamp parser {{strptime}}:
> {code:python}
> import datetime
> import pyarrow as pa
> import pyarrow.compute as pc
> >>> pa.array([datetime.datetime(2022, 3, 5, 9)])
> <pyarrow.lib.TimestampArray object at 0x7f00c1d53dc0>
> [
> 2022-03-05 09:00:00.000000
> ]
> # trying to parse the above representation as string
> >>> pc.strptime(["2022-03-05 09:00:00.000000"], format="%Y-%m-%d %H:%M:%S",
> >>> unit="us")
> ...
> ArrowInvalid: Failed to parse string: '2022-03-05 09:00:00.000000' as a
> scalar of type timestamp[us]
> {code}
> The reason for this is the fractional second part, so the following works:
> {code:python}
> >>> pc.strptime(["2022-03-05 09:00:00"], format="%Y-%m-%d %H:%M:%S",
> >>> unit="us")
> <pyarrow.lib.TimestampArray object at 0x7f00c1d6f940>
> [
> 2022-03-05 09:00:00.000000
> ]
> {code}
> Now, I think the reason that this fails is because {{strptime}} only supports
> parsing seconds as an integer
> (https://man7.org/linux/man-pages/man3/strptime.3.html).
> But, it creates a strange situation where the timestamp parser cannot parse
> the representation we use for timestamps.
> In addition, for CSV we have a custom ISO parser (used by default), so when
> parsing the strings while reading a CSV file, the same string with fractional
> seconds does work:
> {code:python}
> s = b"""a
> 2022-03-05 09:00:00.000000"""
> from pyarrow import csv
> >>> csv.read_csv(io.BytesIO(s))
> pyarrow.Table
> a: timestamp[ns]
> ----
> a: [[2022-03-05 09:00:00.000000000]]
> {code}
> I realize that you can use the generic "cast" for doing this string parsing:
> {code:python}
> >>> pc.cast(["2022-03-05 09:00:00.000000"], pa.timestamp("us"))
> <pyarrow.lib.TimestampArray object at 0x7f00c1d53d60>
> [
> 2022-03-05 09:00:00.000000
> ]
> {code}
> But this was not the first way I thought about (I think it is quite typical
> to first think of {{strptime}}, and it is confusing that that doesn't work;
> the error message is also not helpful)
> cc [~apitrou] [~rokm]
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
