Session authentication with Django channels - python

Trying to get authentication working with Django channels with a very simple websockets app that echoes back whatever the user sends over with a prefix "You said: ".
My processes:
web: gunicorn myproject.wsgi --log-file=- --pythonpath ./myproject
realtime: daphne myproject.asgi:channel_layer --port 9090 --bind 0.0.0.0 -v 2
reatime_worker: python manage.py runworker -v 2
I run all processes when testing locally with heroku local -e .env -p 8080, but you could also run them all separately.
Note I have WSGI on localhost:8080 and ASGI on localhost:9090.
Routing and consumers:
### routing.py ###
from . import consumers
channel_routing = {
'websocket.connect': consumers.ws_connect,
'websocket.receive': consumers.ws_receive,
'websocket.disconnect': consumers.ws_disconnect,
}
and
### consumers.py ###
import traceback
from django.http import HttpResponse
from channels.handler import AsgiHandler
from channels import Group
from channels.sessions import channel_session
from channels.auth import channel_session_user, channel_session_user_from_http
from myproject import CustomLogger
logger = CustomLogger(__name__)
#channel_session_user_from_http
def ws_connect(message):
logger.info("ws_connect: %s" % message.user.email)
message.reply_channel.send({"accept": True})
message.channel_session['prefix'] = "You said"
# message.channel_session['django_user'] = message.user # tried doing this but it doesn't work...
#channel_session_user_from_http
def ws_receive(message, http_user=True):
try:
logger.info("1) User: %s" % message.user)
logger.info("2) Channel session fields: %s" % message.channel_session.__dict__)
logger.info("3) Anything at 'django_user' key? => %s" % (
'django_user' in message.channel_session,))
user = User.objects.get(pk=message.channel_session['_auth_user_id'])
logger.info(None, "4) ws_receive: %s" % user.email)
prefix = message.channel_session['prefix']
message.reply_channel.send({
'text' : "%s: %s" % (prefix, message['text']),
})
except Exception:
logger.info("ERROR: %s" % traceback.format_exc())
#channel_session_user_from_http
def ws_disconnect(message):
logger.info("ws_disconnect: %s" % message.__dict__)
message.reply_channel.send({
'text' : "%s" % "Sad to see you go :(",
})
And then to test, I go into Javascript console on the same domain as my HTTP site, and type in:
> var socket = new WebSocket('ws://localhost:9090/')
> socket.onmessage = function(e) {console.log(e.data);}
> socket.send("Testing testing 123")
VM481:2 You said: Testing testing 123
And my local server log shows:
ws_connect: test#test.com
1) User: AnonymousUser
2) Channel session fields: {'_SessionBase__session_key': 'chnb79d91b43c6c9e1ca9a29856e00ab', 'modified': False, '_session_cache': {u'prefix': u'You said', u'_auth_user_hash': u'ca4cf77d8158689b2b6febf569244198b70d5531', u'_auth_user_backend': u'django.contrib.auth.backends.ModelBackend', u'_auth_user_id': u'1'}, 'accessed': True, 'model': <class 'django.contrib.sessions.models.Session'>, 'serializer': <class 'django.core.signing.JSONSerializer'>}
3) Anything at 'django_user' key? => False
4) ws_receive: test#test.com
Which, of course, makes no sense. Few questions:
Why would Django see message.user as an AnonymousUser but have the actual user id _auth_user_id=1 (this is my correct user ID) in the session?
I am running my local server (WSGI) on 8080 and daphne (ASGI) on 9090 (different ports). And I didn't include session_key=xxxx in my WebSocket connection - yet Django was able to read my browser's cookie for the correct user, test#test.com? According to Channels docs, this shouldn't be possible.
Under my setup, what is the best / simplest way to carry out authentication with Django channels?

Note: This answer is explicit to channels 1.x, channels 2.x uses a different auth mechanism.
I had a hard time with django channels too, i had to dig into the source code to better understand the docs ...
Question 1:
The docs mention this kind of long trail of decorators relying on each other (http_session, http_session_user ...) that you can use to wrap your message consumers, in the middle of that trail it states this:
Now, one thing to note is that you only get the detailed HTTP information during the connect message of a WebSocket connection (you can read more about that in the ASGI spec) - this means we’re not wasting bandwidth sending the same information over the wire needlessly.
This also means we’ll have to grab the user in the connection handler and then store it in the session;....
Its easy to get lost in all that, at least we both did ...
You just have to remember that this happens when you use channel_session_user_from_http:
It calls http_session_user
a. calls http_session which will parse the message and give us a message.http_session attribute.
b. Upon returning from the call, it initiates a message.user based on the information it got in message.http_session ( this will bite you later)
It calls channel_session which will initiate a dummy session in message.channel_session and ties it to the message reply channel.
Now it calls transfer_user which will move the http_session into the channel_session
This happens during the connection handling of a websocket, so on subsequent messages you won't have acces to detailed HTTP information, so what's happening after the connect is that you're calling channel_session_user_from_http again, which in this situation (post-connect messages) calls http_session_user which will attempt reading the Http information but fails resulting in setting message.http_session to None and overriding message.user to AnonymousUser.
That's why you need to use channel_session_user in this case.
Question 2:
Channels can use Django sessions either from cookies (if you’re running your websocket server on the same port as your main site, using something like Daphne), or from a session_key GET parameter, which works if you want to keep running your HTTP requests through a WSGI server and offload WebSockets to a second server process on another port.
Remember http_session, that decorator that gets us the message.http_session data? it appears that if it doesn't find a session_key GET parameter it fails to settings.SESSION_COOKIE_NAME, which is the regular sessionid cookie, so whether you provide session_key or not, you'll still get connected if you're logged in, of course that happens only when your ASGI and WSGI servers are on the same domain (127.0.0.1 in this case), the port difference doesn't matter.
I think the difference that the docs are trying to communicate but didn't expand on is that you need to setup session_key GET parameter when having your ASGI and WSGI servers on different domains since cookies are restricted by domain not port.
Due to that lack of explanation i had to test running ASGI and WSGI on same port and different port and the result was the same, i was still getting authenticated, changed one server domain to 127.0.0.2 instead of 127.0.0.1 and the authentication was gone, set the session_key get parameter and the authentication was back again.
Update: a rectification of the docs paragraph was just pushed to the channels repo, it was meant to mention domain instead of port like i mentioned.
Question 3:
my answer is the same as turbotux's but longer, you should use #channel_session_user_from_http on ws_connect and #channel_session_user on ws_receive and ws_disconnect, nothing from what you showed tells that it won't work if you do that change, maybe try removing http_user=True from your receive consumer? even thou i suspect it has no effect since its undocumented and intended only to be used by Generic Consumers...
Hope this helps!

To answer your first question you need to use the:
channel_session_user
decorator in the receive and disconnect calls.
channel_session_user_from_http
calls the transfer_user session during the connect method to transfer the http session to the channel session. This way all future calls may access the channel session to retrieve user information.
To your second question I believe what you are seeing is that default web socket library passes the browser cookies over the connection.
Third, I think your setup will be working quite well once have changed the decorators.

I ran into this problem and I found that it was due to a couple of issues that might be the cause. I'm not suggesting this will solve your issue, but might give you some insight. Keep in mind I am using rest framework. First I was overriding the User model. Second when I defined the application variable in my root routing.py I didn't use my own AuthMiddleware. I was using the docs suggested AuthMiddlewareStack. So, per the Channels docs, I defined my own custom authentication middleware, which takes my JWT value from the cookies, authenticates it and assigns it to the scope["user"] like so:
routing.py
from channels.routing import ProtocolTypeRouter, URLRouter
import app.routing
from .middleware import JsonTokenAuthMiddleware
application = ProtocolTypeRouter(
{
"websocket": JsonTokenAuthMiddleware(
(URLRouter(app.routing.websocket_urlpatterns))
)
}
middleware.py
from http import cookies
from django.contrib.auth.models import AnonymousUser
from django.db import close_old_connections
from rest_framework.authtoken.models import Token
from rest_framework_jwt.authentication import BaseJSONWebTokenAuthentication
class JsonWebTokenAuthenticationFromScope(BaseJSONWebTokenAuthentication):
def get_jwt_value(self, scope):
try:
cookie = next(x for x in scope["headers"] if x[0].decode("utf-8")
== "cookie")[1].decode("utf-8")
return cookies.SimpleCookie(cookie)["JWT"].value
except:
return None
class JsonTokenAuthMiddleware(BaseJSONWebTokenAuthentication):
def __init__(self, inner):
self.inner = inner
def __call__(self, scope):
try:
close_old_connections()
user, jwt_value =
JsonWebTokenAuthenticationFromScope().authenticate(scope)
scope["user"] = user
except:
scope["user"] = AnonymousUser()
return self.inner(scope)
Hope this helps this helps!

Related

New Relic on AWS lambda not reading its config file

I'm using zappa to deploy a python/django wsgi app to AWS API Gateway and Lambda.
I have all of these in my environment:
NEW_RELIC_CONFIG_FILE: /var/task/newrelic.ini
NEW_RELIC_LICENSE_KEY: redacted
NEW_RELIC_ENVIRONMENT: dev-zappa
NEW_RELIC_STARTUP_DEBUG: "on"
NEW_RELIC_ENABLED: "on"
I'm doing "manual agent start" in my wsgi.py as documented:
import newrelic.agent
# Will collect NEW_RELIC_CONFIG_FILE and NEW_RELIC_ENVIRONMENT from the environment
# Dear god why??!?!
# NB: Looks like this IS what makes it go
newrelic.agent.global_settings().enabled = True
newrelic.agent.initialize('/var/task/newrelic.ini', 'dev-zappa', log_file='stderr', log_level=logging.DEBBUG)
I'm not using #newrelic.agent.wsgi_application since django should be auto-magically detected
I've added a middleware to shutdown the agent before the lambda gets frozen, but the logging suggests that only the first request is being sent to New Relic. Without the shutdown, I get no logging from the New Relic agent, and there are no events in APM.
class NewRelicShutdownMiddleware(MiddlewareMixin):
"""Simple middleware that shutsdown the NR agent at the end of a request"""
def process_request(self, request):
pass
# really wait for the agent to register with collector
# Enabling this causes more log messages about starting data samplers, but only on the first request
# newrelic.agent.register_application(timeout=10)
def process_response(self, request, response):
newrelic.agent.shutdown_agent(timeout=2.5)
return response
def process_exception(self, request, exception):
pass
newrelic.agent.shutdown_agent(timeout=2.5)
In my newrelic.ini I have the following, but when I log newrelic.agent.global_settings() it contains the default App name (which did get created in APM) and enabled = False, which led to some of the hacks above (environment var, and just editing newrelic.agent.global_settings() before initialize :
[newrelic:dev-zappa]
app_name = DEV APP zappa
monitor_mode = true
TL;DR - Two questions:
How to get New Relic to read it's ini file when it doesn't want to?
How to get New Relic to record data for all requests in AWS lambda?
Zappa does not use your wsgi.py file (currently), so the hooks there aren't happening. Take a look at this PR which allows for it: https://github.com/Miserlou/Zappa/pull/1251

TurboGears and backlash: How to add extra, per-request context to Raven for more informative Sentry reports?

Through backlash, TurboGears supports error reporting to Sentry via Raven. Enabling the error reporting is quite easy, just add the appropriate setting in the .ini configuration file, for example:
[DEFAULT]
debug = false
trace_errors.sentry_dsn = https://[…]
trace_slowreqs.enable = true
trace_slowreqs.sentry_dsn = https://[…]
set debug = false
According to Raven's documentation, adding more context to what gets reported should be as simple as
def handle_request(request): # In TurboGears, this would be a controller instead.
client.context.merge({'user': {
'email': request.user.email
}})
try:
...
finally:
client.context.clear()
However, now I wonder what is the easiest, or most correct, way to get hold of the client instance that backlash will use for reporting? I would like to add per-request information, typically from within the request handlers, or Controller methods.
Editing the raven context is currently quite hard as the error reporters are not registered anywhere, so you cannot say "hey give me the error reporters" and look for the Sentry one in that list.
Currently the only way is to register an after_config hook, gather the Raven Client during the configuration process and store it somewhere accessible.
Changing backlash middlewares to store the reporters somewhere accessible should be fairly easy (e.g. the environ) but currently it's not available.
By the way here is a short example of the after_config solution that should make the client available as tg.app_globals.sentry_clients, copy it in your app_cfg.py and it should do what you expect (didn't have time to try it, sorry if you find errors), then you can get the context from the client whenever is needed:
def gather_sentry_client(app):
from backlash import TraceErrorsMiddleware, TraceSlowRequestsMiddleware
try:
trace_errors_app = app.app.application
except:
return app
if not isinstance(trace_errors_app, TraceErrorsMiddleware):
return app
trace_errors_client = None
for reporter in trace_errors_app.reporters:
if hasattr(reporter, 'client'):
trace_errors_client = reporter.client
slow_reqs_app = trace_errors_app.app
slow_reqs_client = None
if isinstance(slow_reqs_app, TraceSlowRequestsMiddleware):
for reporter in slow_reqs_app.reporters:
if hasattr(reporter, 'client'):
slow_reqs_client = reporter.client
from tg import config
app_globals = config['tg.app_globals']
app_globals.sentry_clients = {
'errors': trace_errors_client,
'slowreqs': slow_reqs_client
}
return app
from tg import hooks
hooks.register('after_config', gather_sentry_client)

Python - How to check if Redis server is available

I'm developing a Python Service(Class) for accessing Redis Server. I want to know how to check if Redis Server is running or not. And also if somehow I'm not able to connect to it.
Here is a part of my code
import redis
rs = redis.Redis("localhost")
print rs
It prints the following
<redis.client.Redis object at 0x120ba50>
even if my Redis Server is not running.
As I found that my Python Code connects to the Server only when I do a set() or get() with my redis instance.
So I dont want other services using my class to get an Exception saying
redis.exceptions.ConnectionError: Error 111 connecting localhost:6379. Connection refused.
I want to return proper message/Error code. How can I do that??
If you want to test redis connection once at startup, use the ping() command.
from redis import Redis
redis_host = '127.0.0.1'
r = Redis(redis_host, socket_connect_timeout=1) # short timeout for the test
r.ping()
print('connected to redis "{}"'.format(redis_host))
The command ping() checks the connection and if invalid will raise an exception.
Note - the connection may still fail after you perform the test so this is not going to cover up later timeout exceptions.
The official way to check if redis server availability is ping ( http://redis.io/topics/quickstart ).
One solution is to subclass redis and do 2 things:
check for a connection at instantiation
write an exception handler in the case of no connectivity when making requests
As you said, the connection to the Redis Server is only established when you try to execute a command on the server. If you do not want to go head forward without checking that the server is available, you can just send a random query to the server and check the response. Something like :
try:
response = rs.client_list()
except redis.ConnectionError:
#your error handlig code here
There are already good solutions here, but here's my quick and dirty for django_redis which doesn't seem to include a ping function (though I'm using an older version of django and can't use the newest django_redis).
# assuming rs is your redis connection
def is_redis_available():
# ... get redis connection here, or pass it in. up to you.
try:
rs.get(None) # getting None returns None or throws an exception
except (redis.exceptions.ConnectionError,
redis.exceptions.BusyLoadingError):
return False
return True
This seems to work just fine. Note that if redis is restarting and still loading the .rdb file that holds the cache entries on disk, then it will throw the BusyLoadingError, though it's base class is ConnectionError so it's fine to just catch that.
You can also simply except on redis.exceptions.RedisError which is the base class of all redis exceptions.
Another option, depending on your needs, is to create get and set functions that catch the ConnectionError exceptions when setting/getting values. Then you can continue or wait or whatever you need to do (raise a new exception or just throw out a more useful error message).
This might not work well if you absolutely depend on setting/getting the cache values (for my purposes, if cache is offline for whatever we generally have to "keep going") in which case it might make sense to have the exceptions and let the program/script die and get the redis server/service back to a reachable state.
I have also come across a ConnectionRefusedError from the sockets library, when redis was not running, therefore I had to add that to the availability check.
r = redis.Redis(host='localhost',port=6379,db=0)
def is_redis_available(r):
try:
r.ping()
print("Successfully connected to redis")
except (redis.exceptions.ConnectionError, ConnectionRefusedError):
print("Redis connection error!")
return False
return True
if is_redis_available(r):
print("Yay!")
Redis server connection can be checked by executing ping command to the server.
>>> import redis
>>> r = redis.Redis(host="127.0.0.1", port="6379")
>>> r.ping()
True
using the ping method, we can handle reconnection etc. For knowing the reason for error in connecting, exception handling can be used as suggested in other answers.
try:
is_connected = r.ping()
except redis.ConnectionError:
# handle error
Use ping()
from redis import Redis
conn_pool = Redis(redis_host)
# Connection=Redis<ConnectionPool<Connection<host=localhost,port=6379,db=0>>>
try:
conn_pool.ping()
print('Successfully connected to redis')
except redis.exceptions.ConnectionError as r_con_error:
print('Redis connection error')
# handle exception

Bottle equivalent of engine.restart()

I am trying to transfer from Cherrypy to Bottle & Gevent(server).
After I run:
application=bottle.default_app() #bottle
WSGIServer(('', port), application, spawn=None).serve_forever() #gevent
I want to restart the server just as if the reloader reloaded the server (but only when I tell the server to).
So I want to access a page with credential request and only after correct authentication will it restart.
Here is my functional example in Cherrypy:
#expose
def reloadMe(self, u=None, p=None):
if u=="username" and p=="password":
engine.restart()
raise HTTPRedirect('/')
More simply I am asking how do I reload this script so that my edits to the source file are implemented but only when I retrieve a "restart" page.
I literally only need the Bottlepy equivalent of
engine.restart() #cherrypy
Does no one know how to do this?
You can write a small shell script to restart gevent wsgi server.
then using this code, you can call the script.
#get('/restartmyserver')
def handler():
http_auth_data = bottle.request.auth() # returns a tuple (username,password) only basic auth.
if http_auth_data[0] == user and http_auth_data[1] == password:
os.system("your_shell_script_to_restart_gevent_wsgi")
bottle.redirect('/')
let me know if you need more info.

Python - Twisted, Proxy and modifying content

So i've looked around at a few things involving writting an HTTP Proxy using python and the Twisted framework.
Essentially, like some other questions, I'd like to be able to modify the data that will be sent back to the browser. That is, the browser requests a resource and the proxy will fetch it. Before the resource is returned to the browser, i'd like to be able to modify ANY (HTTP headers AND content) content.
This ( Need help writing a twisted proxy ) was what I initially found. I tried it out, but it didn't work for me. I also found this ( Python Twisted proxy - how to intercept packets ) which i thought would work, however I can only see the HTTP requests from the browser.
I am looking for any advice. Some thoughts I have are to use the ProxyClient and ProxyRequest classes and override the functions, but I read that the Proxy class itself is a combination of the both.
For those who may ask to see some code, it should be noted that I have worked with only the above two examples. Any help is great.
Thanks.
To create ProxyFactory that can modify server response headers, content you could override ProxyClient.handle*() methods:
from twisted.python import log
from twisted.web import http, proxy
class ProxyClient(proxy.ProxyClient):
"""Mangle returned header, content here.
Use `self.father` methods to modify request directly.
"""
def handleHeader(self, key, value):
# change response header here
log.msg("Header: %s: %s" % (key, value))
proxy.ProxyClient.handleHeader(self, key, value)
def handleResponsePart(self, buffer):
# change response part here
log.msg("Content: %s" % (buffer[:50],))
# make all content upper case
proxy.ProxyClient.handleResponsePart(self, buffer.upper())
class ProxyClientFactory(proxy.ProxyClientFactory):
protocol = ProxyClient
class ProxyRequest(proxy.ProxyRequest):
protocols = dict(http=ProxyClientFactory)
class Proxy(proxy.Proxy):
requestFactory = ProxyRequest
class ProxyFactory(http.HTTPFactory):
protocol = Proxy
I've got this solution by looking at the source of twisted.web.proxy. I don't know how idiomatic it is.
To run it as a script or via twistd, add at the end:
portstr = "tcp:8080:interface=localhost" # serve on localhost:8080
if __name__ == '__main__': # $ python proxy_modify_request.py
import sys
from twisted.internet import endpoints, reactor
def shutdown(reason, reactor, stopping=[]):
"""Stop the reactor."""
if stopping: return
stopping.append(True)
if reason:
log.msg(reason.value)
reactor.callWhenRunning(reactor.stop)
log.startLogging(sys.stdout)
endpoint = endpoints.serverFromString(reactor, portstr)
d = endpoint.listen(ProxyFactory())
d.addErrback(shutdown, reactor)
reactor.run()
else: # $ twistd -ny proxy_modify_request.py
from twisted.application import service, strports
application = service.Application("proxy_modify_request")
strports.service(portstr, ProxyFactory()).setServiceParent(application)
Usage
$ twistd -ny proxy_modify_request.py
In another terminal:
$ curl -x localhost:8080 http://example.com
For two-way proxy using twisted see the article:
http://sujitpal.blogspot.com/2010/03/http-debug-proxy-with-twisted.html

Categories