nameko-worker-postmortem
Have you ever been frustrated with the behaviour of a Nameko service in pytest using the --pdb flag?
If your Nameko service throws a worker exception, --pdb won't interrupt with a breakpoint until the exception is serialised back to the caller of the entrypoint.
nameko-worker-postmortem is a pytest plugin that inserts the tracepoint earlier, so you can see a more useful stack trace.
Example
Here's an example test that will fail:
# test.py
from nameko.web.handlers import http
class BadRequest(Exception):
pass
class Service:
name = "service"
@http("GET", "/resource")
def resource(self, request):
param = request.args.get('param')
if param == "good":
return 200, "OK"
raise BadRequest()
def test_service(container_factory, web_config, web_session):
container = container_factory(Service, web_config)
container.start()
res = web_session.get('/resource?param=bad')
assert res.status_code == 200
With --pdb
Using the --pdb flag, the breakpoint is set where the assertion fails, which isn't very helpful if you want to see the exception that the worker raised instead:
$ py.test test.py
============================= test session starts ==============================
platform darwin -- Python 3.4.6, pytest-3.7.4, py-1.6.0, pluggy-0.7.1
rootdir: /private/tmp, inifile:
plugins: nameko-2.11.0, nameko-worker-postmortem-0.0.1
collected 1 item
test.py F
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> captured log >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
containers.py 399 ERROR error handling worker <WorkerContext [service.resource] at 0x1104a77f0>:
Traceback (most recent call last):
File "/Users/mattbennett/.virtualenvs/tmp-302905cac73c0a2/lib/python3.4/site-packages/nameko/containers.py", line 391, in _run_worker
result = method(*worker_ctx.args, **worker_ctx.kwargs)
File "/private/tmp/test.py", line 14, in resource
raise BadRequest()
test.BadRequest
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> traceback >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
container_factory = <function container_factory.<locals>.make_container at 0x11022f620>
web_config = {'WEB_SERVER_ADDRESS': '127.0.0.1:55934'}
web_session = <nameko.testing.pytest.web_session.<locals>.WebSession object at 0x11047ac18>
def test_service(container_factory, web_config, web_session):
container = container_factory(Service, web_config)
container.start()
res = web_session.get('/resource?param=bad')
> assert res.status_code == 200
E assert 500 == 200
E + where 500 = <Response [500]>.status_code
test.py:22: AssertionError
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> entering PDB >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
[11] > /private/tmp/test.py(22)test_service()
-> assert res.status_code == 200
6 frames hidden (try 'help hidden_frames')
> /private/tmp/test.py(22)
17 def test_service(container_factory, web_config, web_session):
18 container = container_factory(Service, web_config)
19 container.start()
20
21 res = web_session.get('/resource?param=bad')
22 -> assert res.status_code == 200
With --worker-postmortem
Replacing --pdb with --worker-postmortem (or --worker-pdb) the breakpoint is inserted at where the worker exception is raised instead.
Note that -s must also be passed to disable output capturing.
$ py.test test.py --worker-postmortem -s
============================= test session starts ==============================
platform darwin -- Python 3.4.6, pytest-3.7.4, py-1.6.0, pluggy-0.7.1
rootdir: /private/tmp, inifile:
plugins: nameko-2.11.0, nameko-worker-postmortem-0.0.1
collected 1 item
test.py [1] > /private/tmp/test.py(14)resource()
-> raise BadRequest()
> /private/tmp/test.py(14)
9 @http("GET", "/resource")
10 def resource(self, request):
11 param = request.args.get('param')
12 if param == "good":
13 return 200, "OK"
14 -> raise BadRequest()
(Pdb++)
TODO
- Fix coverage collection
- Add pre-commit hooks
- Automatically disable capturing so you don't have to specify
-severy time - Attempt to make navigation
upanddownthe stack trace work as they would with apdb.set_trace()(at the moment,updoesn't take you up the stack but into thenameko-worker-postmortemcode that inserts the breakpoint.)
