ENH: tilde expansion for write functions, #11438

[aechase]

closes #11438
Added calls to _expand_user() in multiple Formatter classes so that writer functions, such as to_csv() and to_html(), can perform tilde expansion on file paths beginning ~/. This makes the writer functions consistent with the reader functions, such as read_csv(). I also changed the __init__() for CSVFormatter to conform to the pep8 width limit.

[jreback]

tests pls

[jreback]

doesn't belong here

[jreback]

actually on 2nd thought.

I think you these should ONLY be used in core/frame.py/DataFrame.to_csv and to_html

I don't really like this here, as all of the other IO functions are localized in pandas/io/* but that's how it is for now.

[aechase]

I had the same thought, but then I assumed that would cause problems for Series.to_csv(). But, having now checked the code, I see that they are the same function, with each Series being temporarily converted to a DataFrame.

Should I put the _expand_user() calls in frame.py?

Regarding the tests: let me start out by saying that I am oh-so-close to being out of my depth here. The current suite seems to run OK (various DeprecationWarnings and RuntimeWarnings notwithstanding), but I'm a bit unclear as to how (or whether) I should create a new test function. That function would need to write and read some temporary files, but it looks like it would be difficult, if not impossible, to use testing.ensure_clean() to generate expanded and non-expanded file paths.

[jreback]

you might have a look at these: https://github.com/pydata/pandas/blob/master/pandas/io/tests/test_common.py

yeh this might be quite tricky to test

[jreback]

can you add a release note and we'll stick in 0.17.1

[aechase]

@jreback looks like I might have caused some trouble editing the most recent version of the release notes file...

[jreback]

merged via ea758cd

thanks!