Implement deep copy option in Series.copy

[hughesadam87]

Not quite sure if this is intentional behavior, but noticed that df.copy() accepts argument deep and series.copy() does not.

from pandas import Series, DataFrame
s=Series([1,2,3])
df=DataFrame(s)
df.copy(deep=True)
0
0 1
1 2
2 3
s.copy(deep=True)
Traceback (most recent call last):
File "", line 1, in
TypeError: copy() got an unexpected keyword argument 'deep'

[wesm]

Just hasn't been implemented

[cpcloud]

i think this might be simple after series inherits ndframe

[jreback]

closing in favor of #4324, side note, IIRC deep copy should also do a deep copy of the axes (currently it does not), or is that a separate option / action?

[jtratner]

@jreback probably should. (and index probably needs to take an option to deepcopy as well - since generally just copying metadata is enough)

[jreback]

@jtratner

df.copy(deep=True,axes=True) ?

or if deep=True then always copy axes, otherwise don't?

[jtratner]

After the PR about multi index naming (which tweaks how copying works for Index) is merged it should:

• deep copy axis when deep copying
• shallow copy axis (ie call view) on copy without deep=True

I guess you could pass kwarg for axes but it should default to True, and at the minimum should always shallow copy (which just copies metadata) because contract for objects with axes is/should bw that they can manipulate them in place without any checks.

[jreback]

@jtrater just hit this with a Series.copy(deep=True) (which is part of rename!)

[jtratner]

@jreback what? confused.

[jreback]

no, what I mean is I hit your error message! (was trying to do something funky)....doesn't matter...fixed