[BUG] adorn_pct_formatting, adorn_ns and adorn_rounding overwrite a numeric first column

[dylanpulver]

Brief Description

adorn_pct_formatting, adorn_ns and adorn_rounding treat a numeric first column as data, so the row labels of a tabyl get overwritten.

Using the cyl/gear shape from tests/functions/test_tabyl.py:

basic_df.tabyl("cyl", "gear").adorn_percentages("row").adorn_pct_formatting()

returns cyl as ['400.0%', '600.0%', '800.0%'] instead of [4, 6, 8].

With a counts frame whose first column is year = [2020, 2021]:

call	year becomes	expected
adorn_pct_formatting()	['202000.0%', '202100.0%']	[2020, 2021]
adorn_ns()	['202000.0% (2,020)', '202100.0% (2,021)']	[2020, 2021]
adorn_rounding(digits=2)	[2020.0, 2021.0], int64 becomes float64	[2020, 2021]

adorn_totals and adorn_percentages exclude the first column, and the docstrings of test_adorn_totals_numeric_first_column and test_adorn_percentages_numeric_first_column state that the first column is "always treated as a row identifier (consistent with R janitor tabyl behavior)". The three functions above disagree with that.

A string first column is unaffected.

System Information

• Operating system: macOS
• OS details (optional): 26.5.1, arm64
• Python version (required): 3.12.13

pandas 3.0.5, pyjanitor dev at 6edcbdc.

Minimally Reproducible Code

import pandas as pd
import janitor  # noqa: F401

df = pd.DataFrame(
    {"cyl": [4, 6, 8, 4, 6, 8, 4, 6], "gear": [3, 3, 3, 4, 4, 4, 5, 5]}
)
print(df.tabyl("cyl", "gear").adorn_percentages("row").adorn_pct_formatting())

counts = pd.DataFrame({"year": [2020, 2021], "yes": [10, 20], "no": [5, 15]})
pct = counts.adorn_percentages("row")  # year is correctly left alone here
print(pct.adorn_pct_formatting())
print(pct.adorn_pct_formatting().adorn_ns())
print(pct.adorn_rounding(digits=2))

Error Messages

No exception is raised. The output is silently wrong:

      cyl      3      4      5
0  400.0%  33.3%  33.3%  33.3%
1  600.0%  33.3%  33.3%  33.3%
2  800.0%  50.0%  50.0%   0.0%

        year    yes     no
0  202000.0%  66.7%  33.3%
1  202100.0%  57.1%  42.9%

                year         yes          no
0  202000.0% (2,020)  66.7% (10)   33.3% (5)
1  202100.0% (2,021)  57.1% (20)  42.9% (15)

     year   yes    no
0  2020.0  0.67  0.33
1  2021.0  0.57  0.43