Prettify Series repr

[Julian-J-S]

Problem description

Problem

• cell output in plain mode notebook is not nice / inconsistent

Example

DATA = {'text': ['nospace', '     before', 'after     ', '     both     ', 'bet     ween',]}
df = pl.DataFrame(data=DATA)
ser = df.get_column('text')

Implementation

• I would suggest extraction the DataFrame Display/fmt function and using that for DataFrame and Series. It can receive a "shape" string to still make Series and DataFrame distinction for 1 col DataFrames

[stevenlis]

I prefer the vertical option you proposed above instead of a horizontal one.

[Julian-J-S]

I have no strong feeling abount vertical/horizontal but would actually prefer horizontal like @mcrumiller suggested here: https://github.com/pola-rs/polars/issues/10597#issuecomment-1687073987 Reasons:

• easy to differentiate from DataFrame
  • for one column Dataframe you have to look very close to spot difference: shape (123, 1) vs shape (123, )
  • if you hide the shape (set_tbl_hide_dataframe_shape) there is no way to know (problematic for from_repr?)
• flexibility
  • if you still want vertical you can just call to_frame
• space efficiency
  • vertical series take so much space in notebooks which is actually annoying.
  • with horizontal you can see the data in a single line and much more of the other cells around

[stevenlis]

I think space efficiency is contingent upon your screen ratio. I think what you have above makes more sense, but another option might be to simply put the values in a table.

# from this
shape: (3,)
Series: 'col' [i64]
[
    1
    2
    3
]

# to this?
shape: (3,)
Series: 'col' [i64]
╞═════╡
│ 1   │
│ 2   │
│ 3   │
└─────┘