Add request.id

[ahopkins]

This adds a convenience property on the Request instance for tracking: request.id. The property will default to the value of the X-REQUEST-ID header if it is set. If not, it will fallback to a UUID.

There is a class method that will allow you to override the default (Request.generate_id).

[codecov]

Codecov Report

Merging #2005 (d84d2e9) into master (6c03dd8) will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #2005      +/-   ##
==========================================
+ Coverage   91.48%   91.52%   +0.04%     
==========================================
  Files          29       29              
  Lines        3323     3341      +18     
  Branches      583      585       +2     
==========================================
+ Hits         3040     3058      +18     
  Misses        201      201              
  Partials       82       82              

Impacted Files	Coverage Δ
sanic/config.py	100.00% <ø> (ø)
sanic/request.py	99.25% <100.00%> (+0.05%)	⬆️

---

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 6c03dd8...d84d2e9. Read the comment docs.

[ashleysommer]

Is this in response to a Feature request? I don't really see the benefit of doing this, but since the UUID is lazy and only generated on the first access to request.id, its probably not a problem to merge in.

In the past, when I need to have a unique ID for a request, I use the built-in python identity fn id(request).

[ahopkins]

I had seen something in a somewhat older issue which triggered the idea. And, I had seen some related questions recently on SO. I thought about using id or some variant on itertools.count, but I think using a UUID is a well understood pattern.

As for the lazy, that was 100% the point. I did not want this to start throwing IDs where they are not needed. It feels like a somewhat nice compliment to request.token, which also could be made cacheable like request.id.

[alextsil]

I've been looking for a way to scope my sqlalchemy sessions to a per-request basis (docs here).
I've used request.__hash__() but its hashes are not 100% unique and could cause trouble.

This new ID property solves the issue neatly 👍