BUG: Fix MI repr with long names

[TomAugspurger]

Closes #21180

In [4]:         try:
   ...:             from unittest import mock
   ...:         except ImportError:
   ...:             mock = pytest.importorskip("mock")
   ...:
   ...:         terminal_size = os.terminal_size((118, 96))
   ...:         p1 = mock.patch('pandas.io.formats.console.get_terminal_size',
   ...:                         return_value=terminal_size)
   ...:         p2 = mock.patch('pandas.io.formats.format.get_terminal_size',
   ...:                         return_value=terminal_size)
   ...:
   ...:         index = range(5)
   ...:         columns = pd.MultiIndex.from_tuples([
   ...:             ('This is a long title with > 37 chars.', 'cat'),
   ...:             ('This is a loooooonger title with > 43 chars.', 'dog'),
   ...:         ])
   ...:         df = pd.DataFrame(1, index=index, columns=columns)
   ...:
   ...:

In [5]: with p1, p2:
   ...:     print('-' * 80)
   ...:     print(repr(df))
   ...:     print('-' * 80)
   ...:

output:

--------------------------------------------------------------------------------
  ...
  ...
0 ...
1 ...
2 ...
3 ...
4 ...

[5 rows x 2 columns]
--------------------------------------------------------------------------------

This matches the repr for non-hierarchical

In [6]: s = pd.DataFrame({"A" * 41: [1, 2], 'B' * 41: [1, 2]})

In [7]: with p1, p2:
   ...:     print('-' * 80)
   ...:     print(repr(s))
   ...:     print('-' * 80)
   ...:

output:

--------------------------------------------------------------------------------
  ...
0 ...
1 ...

[2 rows x 2 columns]
--------------------------------------------------------------------------------

These can certainly be improved, though I'm not sure we'll (I'll) get to it for 0.23.2.

[TomAugspurger]

from @jorisvandenbossche

Overflowing the line is what is happening in my console if I make it smaller

Yeah, agreed that would be best, but I can't tell whose doing the wrapping, if that makes sense.

[TomAugspurger]

Well, I guess it has to be the terminal... So we should just always print two columns? And let the terminal wrap if needed?

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@1cc5471). Click here to learn what that means.
The diff coverage is 100%.

@@            Coverage Diff            @@
##             master   #21655   +/-   ##
=========================================
  Coverage          ?    91.9%           
=========================================
  Files             ?      154           
  Lines             ?    49657           
  Branches          ?        0           
=========================================
  Hits              ?    45638           
  Misses            ?     4019           
  Partials          ?        0

Flag	Coverage Δ
#multiple	90.28% <100%> (?)
#single	42.05% <75%> (?)

Impacted Files	Coverage Δ
pandas/io/formats/format.py	98.25% <100%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1cc5471...4391c24. Read the comment docs.

[TomAugspurger]

Running

import pandas as pd
import numpy as np

s = pd.DataFrame({"A" * 41: 1, "B" * 41: 1}, index=[0, 1, 2])
print("Regular Columns:")
print(repr(s))
print('\n\n')

index = range(5)
columns = pd.MultiIndex.from_tuples([
    ('This is a long title with > 37 chars.', 'cat'),
    ('This is a loooooonger title with > 43 chars.', 'dog'),
])
df = pd.DataFrame(1, index=index, columns=columns)

print("MultiIndex")
print(repr(df))

output

Regular Columns:
   AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA  BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
0                                          1                                          1
1                                          1                                          1
2                                          1                                          1



MultiIndex
  This is a long title with > 37 chars. This is a loooooonger title with > 43 chars.
                                    cat                                          dog
0                                     1                                            1
1                                     1                                            1
2                                     1                                            1
3                                     1                                            1
4                                     1                                            1

[jorisvandenbossche]

The output in the top-post is not longer correct with the latest changes?

[TomAugspurger]

Correct, #21655 (comment) has the current output.

[jreback]

should PR in 0.24.0 (to move this to test_decorators)

[TomAugspurger]

Forgot to lay out my plan.

I'd like to merge this with the try / except, backport to 0.23.2, and then make a PR removing this and using the mock fixture from #20729

#20729 isn't being backported, so that seems easiset.

[jreback]

maybe make make the comments consistent here

# subtract index column
max_cols_adj = ...

# esnure we print at least two
....

[jorisvandenbossche]

I can confirm that this fixes #21327 (the one that I could reproduce in my local console without any hacks)