[BUG] Misaligned column headers of pandas DataFrame in IPython when using rich.pretty

[dlukes]

Compare the vanilla Out[2] with Out[4] prettified by rich: the column headers are no longer right-aligned:

The fix should be easy -- just make rich put the column headers on a separate line, like the vanilla representation. But maybe there's a catch? One thing that comes to mind is that maybe this can't be fully fixed on rich's side -- maybe it's an unfortunate interaction between rich, ipython and pandas? FWIW, the problem doesn't occur in the regular Python REPL:

[willmcgugan]

The problem is that IPython prints the Out[2] text and leaves the cursor at the next character, where Rich starts printing from. This is fine if you are printing something on a single line, but it breaks formatted columns as you've demonstrated.

I'm assuming that IPython has a special case for output that has newlines. If there is a newline character it will start the output on the following line and not the line containing Out[X]. Rich could probably replicate this behavior.

[dlukes]

I'm assuming that IPython has a special case for output that has newlines. If there is a newline character it will start the output on the following line and not the line containing Out[X]. Rich could probably replicate this behavior.

That sounds great as far as I'm concerned :)

[willmcgugan]

Can you give 10.5.0 a try and let me know if that fixes it?

[dlukes]

Works like a charm, thank you very much!