Fix ~ normalization in Path:normalize()

[fenuks]

This fixes nvim-telescope/telescope.nvim#1521. The problem is that if $HOME directory contains trailing slash, then Path:normalize() will convert /home/user/my_file into ~my_file.

[l-kershaw]

I'm happy to merge this once the changes pointed out by stylua are made (we use double quotes for strings were possible)

[fenuks]

Thanks, I (hopefully) fixed stylua problems.

[l-kershaw]

Looks good to me 🙂
Thanks!

[fenuks]

Thank you for your help. ;)

[steve-lorimer]

Note that PR has introduced the arguably incorrect behaviour that you can now end up with double slashes in your normalized path.

eg: if path.home is /home/user and filename is /home/user/foo, this change results in the normalized path being ~//foo instead of the arguably correct ~/foo

I think a more correct approach will be to check if path.home has a trailing slash, and selectively apply it in the case it doesn't

I'll submit a new PR

[fenuks]

It is expected in sense there is even test for that. Personally I think this is not a problem at all, it's a valid, doesn't cause any problems, and makes code simpler.

[Conni2461]

assert.are.same("~//./test_file", p:normalize())

This is not normalized. This shouldnt have been merged.

Ref: nvim-telescope/telescope.nvim#1662 nvim-telescope/telescope.nvim#1683

[steve-lorimer]

@Conni2461 I have a change in a local branch which just checks for trailing slash in path.home, which will "fix" the above issue which is to indiscriminantly add path.sep in normalize

-- remove trailing slash from homedir if any
if path.home:sub(-1) == path.sep then
  path.home = path.home:sub(1,-2)
end

It will result in the following behaviour:

assert.are.same("~/./test_file", p:normalize())

So at least the double slash is removed, but arguably this isn't normalized either, as the ./ should be removed too, would you agree?

In any event, I can submit the above change as a PR if you wish?

[fenuks]

It's not normalized, and I wouldn't have been looking for what caused double slashes in the first place if it "is not a problem at all"

Well, so I was wrong about it not being a problem…

[steve-lorimer]

@Conni2461 I've created a PR (#308) with the above mentioned change - if you feel it's not the correct approach more than happy to alter it...