Explicit proxy support in RequestsHttpConnection

[fopina]

Description

This PR allows specifying an explicit proxy to the underlying requests.Session, overriding environment configuration.
This is useful when a specific proxy is required to access OpenSearch but nothing else (such as local tunnels).

Workaround

I've been using a very small/clean subclass across different projects for this, however I think it's clean and useful enough to be pushed to the main upstream class 😄

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 70.40%. Comparing base (ba715b9) to head (e31ace0).
Report is 90 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #908      +/-   ##
==========================================
- Coverage   71.95%   70.40%   -1.56%     
==========================================
  Files          91      125      +34     
  Lines        8001     9291    +1290     
==========================================
+ Hits         5757     6541     +784     
- Misses       2244     2750     +506     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[fopina]

Sorry for the push force spam, I did not expect so many linting complains off a single lint 😓

Ready for review now

[dblock]

Thanks. This needs tests, please.

Integ test CI is failing, needs to be looked at. Likely the 3.0 upgrade in CI to be done first in a separate PR (please help).

[fopina]

@dblock I looked at the integration tests but they are failing for the same reason as every other PR I guess. I assume #905 is trying to fix them?

I'll look into writing a unit test sure

[fopina]

Unit test I added. I suppose it's ok like this (as http_auth) as testing further would be testing requests proxy implementation.

codecov is doing something funky:

• it passed in the previous commit - this last one only adds a test, impossible to decrease coverage...
• seems it compared to main on ba715b9 (very old main) for some reason...

Thanks, looking forward to see this merged and stop using an external package 🙏

[dblock]

We can ignore codecov for now.

@nathaliellenaa are you working on #905? @fopina feel free to take it over

[fopina]

@dblock I have little to no context in this project setup, already took forever just to get tests to run locally, so I’ll have to pass.
I’ll rebase once it’s fixed, if it cannot be merged as is

[nathaliellenaa]

@dblock @fopina I'm waiting for this PR to get merged, then I will rebase and rerun the integ tests to identify and resolve any remaining issues.

[dblock]

Looks like it's 2 weeks old, afaik nobody is working on it.

[fopina]

Hi @dblock just noticed #905 was merged and fixed the tests so I rebased et voila, all green 👍

[dblock]

I can merge, but we have a base class with the intent to supporting these features across all connection types if I am not mistaken. Before I do mind checking if all the connection class parameters align and whether this needs to be added to the others? It also seems like we may need tests to ensure that's the case.

[fopina]

It could be added to base class but then isn't it expected for all connections to actually support specifying proxies?
Or would the parameter just be ignored?
I just assumed it'd be better to have it at subclass level and, if at some point, all (or majority) do make any use of similar parameter, to move to base?

[dblock]

Ok, I'll merge. Someone should refactor this whole interface and instead of passing a million different parameters it should take separate session options that allow passing anything to self.session that it supports.

[fopina]

That's a good point, "trust_env" is, sadly, something that has to be used quite often with requests.Session as he refused to fix a clear bug
There might be other edge cases / attributes.
I'll try to find the time for a proposal
Thanks for merging!