[![Build Status] (https://travis-ci.org/numberoverzero/pyservice.svg?branch=master)] (https://travis-ci.org/numberoverzero/pyservice)[![Coverage Status] (https://coveralls.io/repos/numberoverzero/pyservice/badge.png?branch=master)] (https://coveralls.io/r/numberoverzero/pyservice?branch=master)
Downloads https://pypi.python.org/pypi/pyservice
Source https://github.com/numberoverzero/pyservice
Microservice framework for high throughput, designed for readability and code re-use
pip install pyservice
pyservice was designed from the ground up to minimize request overhead, while still exposing the relevant pieces of the request chain for extension.
First, we'll define a small api. These are just nested dictonaries - feel free to load them from a json file.
# Service/Client just use a dict for specifying an api
api = {
"debug": True,
"endpoint": {
"scheme": "http",
"host": "localhost",
"port": 8080,
"path": "/api/{version}/{operation}"
},
"operations": ["get_item", "put_item"],
"exceptions": ["IDRequired", "DoesNotExist", "ItemRequired"]
}
Next we'll set up the service, and define the get/put operations:
import uuid
import pyservice
service = pyservice.Service(**api)
items = {}
@service.operation(name="put_item")
def put_item(request, response, context):
if item not in request:
raise service.exceptions.ItemRequired("Need an item to put")
id = uuid.uuid4()
items[id] = request.item
response.id = id
@service.operation(name="get_item")
def get_item(request, response, context):
if id not in request:
raise service.exceptions.IDRequired("Can't get an item without an ID")
try:
item = items[request.id]
except KeyError:
raise service.exceptions.DoesNotExist("No item with id " + response.id)
else:
response.item = item
Finally, to get a server running, we'll use the wsgiref reference server:
from wsgiref.simple_server import make_server
def run_server():
print("Starting Server...")
host, port = api["endpoint"]["host"], api["endpoint"]["port"]
httpd = make_server(host, port, service.wsgi_application)
httpd.serve_forever()
if __name__ == "__main__":
run_server()
To make a call from a client, we'll use the same api
defined above. The
client calls are even simpler:
import pyservice
# ... Same api definition above
client = pyservice.Client(**api)
# put
item = "some string"
id = client.put_item(item=item)
# get
same_item = client.get_item(id=id)
assert item == same_item
We can plug into calls in two scopes:
request
, which is before the request and response bodies have been created and after they've been consumedoperation
, which is after the request and response bodies have been created and before they've been consumed.
The difference is important for things like sqlalchemy, where serialization should occur before the connection is closed.
@service.plugin(scope="request")
def some_plugin(context):
print("Before request '{}'".format(context.operation))
context.process_request()
print("After request '{}'".format(context.operation))
@service.plugin(scope="operation")
def some_plugin(request, response, context):
print("Before operation '{}'".format(context.operation))
print("Request: {}".format(request))
context.process_request()
print("Response: {}".format(response))
print("After operation '{}'".format(context.operation))
Contributions welcome! Please make sure tox
passes (including flake8) before submitting a PR.
pyservice uses tox
, pytest
and flake8
. To get everything set up:
# RECOMMENDED: create a virtualenv with:
# mkvirtualenv pyservice
git clone https://github.com/numberoverzero/pyservice.git
pip install tox
tox
- docs (0.9.0)
- README
- docstrings
- examples
- plugins
- metadata
- api versions
- plugins (1.0.0)
- sqlalchemy
- structures
- chain logging