Justification is broken with to_string(index=False)

[evanpw]

With 0.17.1, justification is handled correctly:

In [3]: df = pd.DataFrame(np.random.randn(3, 3), columns=['one', 'two', 'three'])

In [4]: print df.to_string(index=False)
      one       two     three
 0.096065 -0.529437 -2.058535
 0.588013  0.221911 -2.400129
 0.848234 -0.093931  1.221708

But in 0.18.0, it's broken:

In [4]: print df.to_string(index=False)
one       two     three
1.164170 -0.027236 -0.114962
-1.405931  0.019395  0.743320
0.598321  0.655802 -1.061834

This is caused by PR #11942. The issue it was meant to fix (#11833) is actually caused by using the format string "% d" for formatting integers, which produces a leading space if the integer is positive (where the minus sign would go for a negative integer). My preferred solution would be to change the formatter to "%d" (and to revert #11942), which would have the added bonus of eliminating double spaces between some columns (see the first to_string output in #11833, for example). A bunch of formatting tests would have to be adjusted, though. Is there anyone who prefers the extra space?

[jreback]

yeah looks like the prior fix only dealt with integers. a more comprehensive fix would be ok.

[hawkeyej]

The formatting issue seems more broad to me. This example demonstrates:

import pandas as pd

NAMES = ['Short', 'Longer', 'Much Longer name to the Max -----------']
VALUES = [1, 9374518, 32432]

d = pd.DataFrame({'Name': NAMES, 'Value': VALUES})

print("With index")
print(d.to_string())

print("\n\nWithout index")
print(d.to_string(index=False))

Produces this:

With index
                                      Name    Value
0                                    Short        1
1                                   Longer  9374518
2  Much Longer name to the Max -----------    32432

Without index
Name    Value
                                  Short        1
                                 Longer  9374518
Much Longer name to the Max -----------    32432

[jreback]

Might affect this.

display.max_colwidth : int
    The maximum width in characters of a column in the repr of
    a pandas data structure. When the column overflows, a "..."
    placeholder is embedded in the output.
    [default: 50] [currently: 50]

[hawkeyej]

Note that only the header line is shifted incorrectly.

My current workaround is this (data is the DataFrame):

# Find longest index string
longest = max([len(str(s)) for s in data.index]) + 1

# Print to lines and manually remove 'index' from each line
lines = data.to_string().split('\n')
lines = [l[longest:] for l in lines]
print('\n'.join(lines))

[jreback]

@hawkeyej certainly love for a PR to fix! just write up a test of the correct results, and use your fix; see what else breaks :)

[jorisvandenbossche]

Initial work in #14196, can possibly be used as a basis (it needs tests)

[kuraga]

Ping. Annoying bug, can I help? Though I'm not a pandas expert...