improve injection code, make backwards compat explicit, make ssl-api explicit

[betaboon]

This PR introduces the following changes:

• make the injection-code more readable
• make the injection-code more explicit with regards to version-specific backwards-compat
• introduce dedicated ssl.context.mock_wrap_socket to mirror stdlib api
• convert MocketSSLContext.wrap_socket and .wrap_bio to instance-methods to mirror stdlib api
• introduce dedicated urllib3.mock_ssl_wrap_socket methods mirror urllib3 api
• use a real ssl.SSLContext in MocketSSLSocket instead of urllib3.ssl_wrap_socket

[coveralls]

coverage: 99.025% (-0.2%) from 99.255%
when pulling a155190 on betaboon:refactor-inject
into 0da2722 on mindflayer:main.

[coveralls]

coverage: 98.876% (-0.4%) from 99.255%
when pulling 199e903 on betaboon:refactor-inject
into 0da2722 on mindflayer:main.

[coveralls]

coverage: 98.876% (-0.4%) from 99.255%
when pulling 199e903 on betaboon:refactor-inject
into 0da2722 on mindflayer:main.

[coveralls]

coverage: 98.876% (-0.4%) from 99.255%
when pulling 199e903 on betaboon:refactor-inject
into 0da2722 on mindflayer:main.

[coveralls]

coverage: 98.876% (-0.4%) from 99.255%
when pulling 199e903 on betaboon:refactor-inject
into 0da2722 on mindflayer:main.

[mindflayer]

I like the function _replace, it makes the code cleaner. I am not sure I see the need for dividing the injecting code into sub-functions. I feet it's a touch of over-engineering with no real benefit. The module we are patching is clearly there already (socket/ssl/etc), and dividing them in blocks should be enough to keep them tidy.
I will sleep on it, but I am not 100% sure I like this level of abstraction for something that simple as patching Python modules.

[mindflayer]

Also, have a look at code coverage, not sure what happened but it went down.

[betaboon]

sure. take your time.

i my mind having a dedicated place for each injection logic is more readable, especially considering the python- or urllib3-version specific things.

[mindflayer]

That's the other thing, I don't think we need to add a new dependency for something as simple as trying to import something from a module.
I'd keep the with contextlib.suppress(ImportError) instead. It's more robust and easier to maintain, and it does not need external modules.

[betaboon]

sleep on it and ping me tomorrow and I'll take care of whatever decision you come to :)

[mindflayer]

I've just had an idea. It would be nice if we had a dictionary of everything we patched, with the real and the mocked version, and use it to bring the status back, instead of using variables to know if I patched or not that thing like it is currently done.
If could be something that the _replace functions does for free. We just need to keep track of the changes.

[betaboon]

I'll give it a try

[mindflayer]

It could be literally used both ways. As a way to define the monkey-patches and to bring back their counterparts.

[betaboon]

ok.
i just pushed the following changes:

• removed packaging
• removed the individual functions from inject.py
• introduced something that might be close to your idea

[sonarqubecloud]

Quality Gate passed

Issues
1 New issue
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[mindflayer]

LGTM!