Fix list keys verb

[mikaelarguedas]

Fixes #183
setup:
ros2 security generate_artifacts -k keystore -e /foo /foobar/baz /bar /a/b/c/deep

before:

root@5ab52a51093c:/ws# ros2 security list_keys keystore/
foobar
bar
foo
a

after:

root@5ab52a51093c:/ws# ros2 security list_keys keystore/
/a/b/c/deep
/bar
/foo
/foobar/baz

Note: a similar fix could be backported to Eloquent and Dashing

[codecov]

Codecov Report

Merging #219 into master will increase coverage by 0.07%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #219      +/-   ##
==========================================
+ Coverage   77.95%   78.02%   +0.07%     
==========================================
  Files          16       16              
  Lines         576      578       +2     
  Branches       52       51       -1     
==========================================
+ Hits          449      451       +2     
  Misses        107      107              
  Partials       20       20              

Flag	Coverage Δ
#unittests	78.02% <100.00%> (+0.07%)	⬆️

Impacted Files	Coverage Δ
sros2/sros2/api/_key.py	90.41% <100.00%> (+0.27%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 54e68e8...23e41bf. Read the comment docs.

[mikaelarguedas]

@kyrofa CI on this PR is another occurrence of test_security tests being flaky https://github.com/ros2/sros2/pull/219/checks?check_run_id=722938561

[kyrofa]

Should this use os.sep?

[ruffsl]

If it uses the same / used in designating the enclave name path, then the stdout could be piped for more commands involving the enclave name.

[kyrofa]

That's a good idea, @ruffsl. I actually think my initial comment is wrong; I believe pathlib will use forward slashes regardless of the OS (I don't have a Windows box to confirm-- @Arnatious help), but your suggestion requires a consistency in path handling that I don't think we have right now. We should take a pass on converting things to pathlib and then we can use either slash without issue. That's outside this PR, of course.

[Arnatious]

Running the code as-is on windows, with a file .\foo\bar\key.pem, gave a printout /foo\bar

[Arnatious]

Pathlib can read ./foo\\bar just fine and resolve it to the same path

[kyrofa]

Haha, yeah I'm not sure about that part either, but it's still path-based, regardless of the slashes, so we might as well do the same thing here. Otherwise it could be pretty confusing to print the contents of the files but then have RCL not actually match those up on the backend, you know?

[mikaelarguedas]

fair enough 👍

[ruffsl]

I'm not sure where you're going, but I always saw the list command as a numerating a list of fully qualified enclave names. If the certificates were generated by SROS, then we could also look at the certificate common name, but those aren't guaranteed to match it's location as it can be moved in the key store. But I think we should stick with extrapolating the enclave name from its placement in the key store, as that's how RCL will resolve the enclave name within file system type key stores.

[kyrofa]

But I think we should stick with extrapolating the enclave name from its placement in the key store, as that's how RCL will resolve the enclave name within file system type key stores.

Sorry if I was unclear, that ^ is what I meant.

[mikaelarguedas]

added as_posix in 23e41bf

[mikaelarguedas]

With some non-Linux CI to see the effects:

• Linux
• Linux-aarch64
• macOS
• Windows

[kyrofa]

Thanks @mikaelarguedas!