fix ansi2latex filtering

[t-makaro]

Alternative to #1073. Fixes #1071

This produces:

As opposed to what #1073 produces:

@MSeal I'll leave it to you to decide which option to choose.

[mgeier]

FWIW, I've solved this problem in nbsphinx by stripping '\n' from the end of all stream outputs:

https://github.com/spatialaudio/nbsphinx/blob/f09f4490525300d5c45213aab3d50d4da68ec40a/src/nbsphinx.py#L1247-L1250

I assume this would still need #1073 (which is a good idea anyway).

I think there is no perfect solution though, because the Jupyter notebook format has some inconsistencies regarding trailing '\n'.

[njapke]

The results here look much nicer than mine in #1073, but I feel like trimming all trailing \n might be a better solution. The verbatim environment should be produced correctly without needing to trim whitespace in the templates.

[t-makaro]

Alright, I adjusted to just remove that whitespace. I agree this is a cleaner solution, and it does appear to render as expected with my test pdfs.

Right now this technically removes trailing '\n' from output blocks too, so if someone did print('test\n') in a block then the '\n' would not be rendered in the pdf. I would consider this as fine since it leaves the rest of the whitespacing up to latex. I could create a separate filter for removing this newline and only apply it too the stream block.

[mgeier]

@t-makaro When you think about it, it doesn't really make sense to add this functionality to a function named ansi2latex(). I think the '\n' should be removed before passing the string to ansi2latex().

I could create a separate filter for removing this newline and only apply it too the stream block.

Yes, this sounds more reasonable.

This will not change the result, but I think in the long run it is better to have functions that do what their name suggests and not arbitrary other things on top of that.

[t-makaro]

Now implemented using a filter called: strip_trailing_newline.

[willingc]

Looks good to me @t-makaro. I'll leave it up to you which PR to go with. Though sentiment seems that this one is preferred.

[meeseeksmachine]

This pull request has been mentioned on Jupyter Community Forum. There might be relevant details there:

https://discourse.jupyter.org/t/nbconvert-5-6-0-release/1867/1