add log atoms

[zffocussss]

• http request header
• http request body size

[zffocussss]

@gi0baro can you help have a check?

[gi0baro]

@zffocussss thank you for your contribution.

Unfortunately, I can't merge this due to the following reasons:

• it's missing the RSGI protocol implementation. Any change in Granian regarding shared interfaces should be performed on all protocols
• I find the WSGI implementation to be very hacky (eg: manually patching the scope to store data), and I'm confused on why you del stuff from objects
• The ASGI implementation is incomplete
• The response size wouldn't be accurate
• The overall implementation might have severe performance implications (parsing all headers just to print one or few, usage of re.match, SafeAtoms wrapper, double iteration of the response, etc)

So in general, I think this should be designed quite differently.

[zffocussss]

@zffocussss thank you for your contribution.

Unfortunately, I can't merge this due to the following reasons:

• it's missing the RSGI protocol implementation. Any change in Granian regarding shared interfaces should be performed on all protocols
• I find the WSGI implementation to be very hacky (eg: manually patching the scope to store data), and I'm confused on why you del stuff from objects
• The ASGI implementation is incomplete
• The response size wouldn't be accurate
• The overall implementation might have severe performance implications (parsing all headers just to print one or few, usage of re.match, SafeAtoms wrapper, double iteration of the response, etc)

So in general, I think this should be designed quite differently.

I will consider your comments.I am not a user of rsgi at all.My goal is to make granian work more flexibly than gunicorn + uvicorn any way.
The response size is calculated differently indeed between asgi and wsgi.
Regarding to the performance issues,I agree with you,trying to find a better way to make it.