mirror of
https://github.com/sprockets/sprockets.clients.http.git
synced 2024-11-15 03:00:26 +00:00
No description
f80f3dd16c
When running the LoggingTests on pypy3 (PyPy 2.4.0), both the server and
client error tests result in infinite recursion in
tornado.httpclient.HTTPResponse.__repr__ when logging the error with a
%r. The initializer for the response creates a HTTPError object with a
referential loop:
def __init__(self, ... error=None):
# ....
if error is None:
self.error = HTTPError(self.code, message=self.reason,
response=self)
When printing a HTTPError with %r it prints to response object's repr
which in turn prints the error's repr.
|
||
|---|---|---|
| docs | ||
| examples | ||
| requires | ||
| sprockets | ||
| tests | ||
| .gitignore | ||
| LICENSE | ||
| MANIFEST.in | ||
| README.rst | ||
| setup.cfg | ||
| setup.py | ||
| tox.ini | ||
sprockets.clients.http
======================
Simplifies calling HTTP APIs from Tornado-based applications.
This library implements a mix-in class that adds a single method which
makes a HTTP request and calls a callback when an error occurs. The
request is made using Tornado's asynchronous HTTP client so it does
not block the active IO loop.
.. code-block:: python
from sprockets.clients import http
from tornado import gen, web
class MyHandler(http.ClientMixin, web.RequestHandler):
@gen.coroutine
def get(self):
response = yield self.make_http_request(
some_server_url, on_error=self.handle_api_error)
if self._finished:
yield gen.Return()
# handle response as you wish
# ...
self.finish()
def handle_api_error(self, request, error):
self.send_error(error.code)
That's it. What you do not see is asynchronous client usage and logging
that happens inside of the library. There is some setup code in
``ClientMixin.initialize`` so make sure to call the super implementation
if you implement ``initialize`` in your request handler.