to_latex with MI column and index names

[hayd]

I introduced/uncovered a bug here with #8219.

If you pass in DataFrame with MI index and col names it doesn't output correctly (and never did). Should be able to fix pretty easily, since it's my fault am assigning myself!

[jorisvandenbossche]

I also just stumbled upon this. And it does seem to be a regression, as it worked in 0.14:

In [1]: idx = pd.MultiIndex.from_product([['A0', 'A1', 'A2'],['B0','B1']])

In [2]: df = pd.DataFrame(np.random.random((6, 2)), index=idx, columns=['col1','
col2'])

In [3]: print df.to_latex()
\begin{tabular}{lrr}
\toprule
{} &      col1 &      col2 \\
A0 B0 &  0.465015 &  0.706859 \\
\midrule
   B1 &  0.768717 &  0.528846 \\
A1 B0 &  0.762849 &  0.801075 \\
   B1 &  0.391274 &  0.751734 \\
A2 B0 &  0.647738 &  0.039428 \\
   B1 &  0.437429 &  0.512462 \\
\bottomrule
\end{tabular}

and now in 0.16.0 the same gives:

In [53]: print df.to_latex()
\begin{tabular}{llrr}
\toprule
   &    &      col1 &      col2 \\
\midrule
A0 & B0 &  0.512408 &  0.441161 \\
A1 & B1 &  0.456337 &  0.472716 \\
\bottomrule
\end{tabular}

[toobaz]

I think this was fixed (in current git at least):

In [3]: idx = pd.MultiIndex.from_product([['A0', 'A1', 'A2'],['B0','B1']])

In [4]: df = pd.DataFrame(np.random.random((6, 2)), index=idx, columns=['col1','col2'])

In [5]: print(df.to_latex())
\begin{tabular}{llrr}
\toprule
   &    &      col1 &      col2 \\
\midrule
A0 & B0 &  0.184294 &  0.850768 \\
   & B1 &  0.762865 &  0.842426 \\
A1 & B0 &  0.137573 &  0.266622 \\
   & B1 &  0.365801 &  0.512295 \\
A2 & B0 &  0.449834 &  0.631251 \\
   & B1 &  0.374689 &  0.269184 \\
\bottomrule
\end{tabular}

[jorisvandenbossche]

Yes, this was actually already fixed in 0.16.1: closed by #9908