Convert Representation of datetime.date to Representation of pandas.Timestamp

Question

I've converted a bunch of code from using datetime.date objects to using Timestamps . The code is covered by a ton of unit tests, which means that I need to convert all the instances like "datetime.date(2016, 12, 20)" to `"Timestamp(2016-12-20)".
The easy solution is:

re.sub(r"datetime.date\((\d{4}), (\d{1,2}), (\d{1,2})\)", r"Timestamp(\1-\2-\3)", string)

Which works fine in some cases. The problem is that date uses one or two digits to display the month and day, whereas Timestamp always uses two. So if the date was datetime.date(2016, 1, 1) I'd get back "Timestamp(2016-1-1) " but the correct representation should be "Timestamp(2016-01-01)" .

Some of the string instances also contain multiple substring matches.

Is there a way that I can use re.sub() to do this conversion?

Answer 1

string = "datetime.date(2016, 2, 20)"
def repl(matchobj):
    return "Timestamp(%s-%s-%s)"%(matchobj.group(1), matchobj.group(2).zfill(2), matchobj.group(3).zfill(2))

print re.sub(r"datetime.date\((\d{4}), (\d{1,2}), (\d{1,2})\)", repl, string)

Output:

Timestamp(2016-02-20)

Use zfill with width 2 .

Answer 2

you can use a combination of pd.to_datetime and eval
make sure you import datetime to get eval to work.

import datetime
import pandas as pd

pd.to_datetime(eval("datetime.date(2016, 3, 31)"))

Timestamp('2016-03-31 00:00:00')

Answer 3

why can't you simply replace datetime.date( with pd.Timestamp( :

In [26]: datetime.date(2000,1,30)
Out[26]: datetime.date(2000, 1, 30)

In [27]: pd.Timestamp(2000,1,30)
Out[27]: Timestamp('2000-01-30 00:00:00')

In [28]: datetime.date(2000,1,3)
Out[28]: datetime.date(2000, 1, 3)

In [29]: pd.Timestamp(2000,1,3)
Out[29]: Timestamp('2000-01-03 00:00:00')

RegEx:

re.sub(r'datetime.date\s*\(', r'pd.Timestamp(', string)

pd.Timestamp docstring :

TimeStamp is the pandas equivalent of python's Datetime and is interchangable with it in most cases . It's the type used for the entries that make up a DatetimeIndex, and other timeseries oriented data structures in pandas.

There are essentially three calling conventions for the constructor. The primary form accepts four parameters. They can be passed by position or keyword.

Parameters ---------- ts_input : datetime-like, str, int, float Value to be converted to Timestamp freq : str, DateOffset Offset which Timestamp will have tz : string, pytz.timezone, dateutil.tz.tzfile or None Time zone for time which Timestamp will have. unit : string numpy unit used for conversion, if ts_input is int or float offset : str, DateOffset Deprecated, use freq

The other two forms mimic the parameters from datetime.datetime . They can be passed by either position or keyword, but not both mixed together.

:func: datetime.datetime Parameters

.. versionadded:: 0.19.0

year : int month : int day : int hour : int, optional, default is 0 minute : int, optional, default is 0 second : int, optional, default is 0 microsecond : int, optional, default is 0 tzinfo : datetime.tzinfo, optional, default is None