raise more explanatory error when failing initialising pd.Timestamp

[giuliobeseghi]

It's not really an issue, just a suggestion. I noticed this behavior when initialising a timestamp:

# this doesn't fail
pd.Timestamp('2020')
# this does
pd.Timestamp(year=2020)

I guess that the latter fails because of args/kwargs specification issues (when an integer is passed it is read as a posix timestamp in the first place), and I'm OK with it. However, the error it raises is:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-9-e2faf77534f3> in <module>
----> 1 pd.Timestamp(year=2020)

pandas/_libs/tslibs/timestamps.pyx in pandas._libs.tslibs.timestamps.Timestamp.__new__()

TypeError: an integer is required (got type NoneType)

This is a bit misleading for beginners - it doesn't say anything about why it failed, but just about what failed.

Is there a chance it can be changed to something more explicit in the long term? Something like "provide at least year, month, day"?

[mroeschke]

Since Timestamp inherits from datetime.datetime, we should just ideally dispatch to the same error message datetime.datetime would raise.

In [2]: datetime.datetime(year=2020)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-2-0cf88a91ae91> in <module>()
----> 1 datetime.datetime(year=2020)

TypeError: Required argument 'month' (pos 2) not found

[fujiaxiang]

I will take a look