Multilevel indexing with .to_latex() method merges two index columns

[jobingr]

The to_latex method produces wrong latex code for multi-indexed funds. looks like it are missing a few blank {} and &. Net result is that the current method merges the two indeces. when used with output from pivot_table with major index giving categories and other one details this produces totally garbled results.

The formatters= parameter of to_latex method also creates exceptions for multi-indexed data

[jbzdak]

Sometime before 0.16.2 this bug has transformed. Now there are no "missing {} and &" but data is subtly wrong.

Here is code example:

import pandas as pd 

df = pd.DataFrame([
        {'a': 'a', 'b': 'a', 'c': 'a', 'val': 1},
        {'a': 'a', 'b': 'a', 'c': 'b', 'val': 2},
        {'a': 'a', 'b': 'a', 'c': 'c', 'val': 3},
        {'a': 'c', 'b': 'd', 'c': 'd', 'val': 4},
    ], columns=['a', 'b', 'c', 'val'])

df.set_index(['a', 'b', 'c'], inplace=True)
print(df.to_latex())

Here is output:

\begin{tabular}{lllr}
\toprule
  &   &   &  val \\
\midrule
a & a & a &      \\
  &   & b &    1 \\
  &   & c &    2 \\
c & d & d &    3 \\
\bottomrule
\end{tabular

Notice that val collumn has a blank line in first row, is missing last val and all rows are moved one row down.

[srib]

I think this seems to work okay now.

This is the output I get:

\begin{tabular}{lllr}
\toprule
  &   &   &  val \\
a & b & c &      \\
\midrule
a & a & a &    1 \\
  &   & b &    2 \\
  &   & c &    3 \\
c & d & d &    4 \\
\bottomrule
\end{tabular}

I think this is expected. Maybe we can close this?

[TomAugspurger]

Yep, thanks. Haven't looked too closely, but it looks like #11233 closed it. Thanks @srib.