Python - How to use Conch to create a Virtual SSH server - python

I'm looking at creating a server in python that I can run, and will work as an SSH server. This will then let different users login, and act as if they'd logged in normally, but only had access to one command.
I want to do this so that I can have a system where I can add users to without having to create a system wide account, so that they can then, for example, commit to a VCS branch, or similar.
While I can work out how to do this with conch to get it to a "custom" shell... I can't figure out how to make it so that the SSH stream works as if it were a real one (I'm preferably wanting to limit to /bin/bzr so that bzr+ssh will work.
It needs to be in python (which i can get to do the authorisation) but don't know how to do the linking to the app.
This needs to be in python to work within the app its designed for, and to be able to be used for those without access to add new users

When you write a Conch server, you can control what happens when the client makes a shell request by implementing ISession.openShell. The Conch server will request IConchUser from your realm and then adapt the resulting avatar to ISession to call openShell on it if necessary.
ISession.openShell's job is to take the transport object passed to it and associate it with a protocol to interpret the bytes received from it and, if desired, to write bytes to it to be sent to the client.
In an unfortunate twist, the object passed to openShell which represents the transport is actually an IProcessProtocol provider. This means that you need to call makeConnection on it, passing an IProcessTransport provider. When data is received from the client, the IProcessProtocol will call writeToChild on the transport you pass to makeConnection. When you want to send data to the client, you should call childDataReceived on it.
To see the exact behavior, I suggest reading the implementation of the IProcessProtocol that is passed in. Don't depend on anything that's not part of IProcessProtocol, but seeing the implementation can make it easier to understand what's going on.
You may also want to look at the implementation of the normal shell-creation to get a sense of what you're aiming for. This will give you a clue about how to associate the stdio of the bzr child process you launch with the SSH channel.

While Python really is my favorite language, I think you need not create you own server for this. When you look at the OpenSSH Manualpage for sshd you'll find the "command" options for the authorized keys file that lets you define a specific command to run on login.
Using keys, you can use one system account to allow many user to log in, just put their public keys in the account's authorized keys file.
We are using this to create SSH tunnels for SVN and it works just great.

Related

How can you Interact and log in to the physical steam client using Python?

In summary; I want to be able to interact with a non-logged in steam client and log it in, and I dont want to use some type of Windows only interaction. I don't mean the SteamCLI, or logging into steam using the steam library for python. I mean directly interacting with the steam client in some way and physically logging in.
When using SteamCLI and other modules I've noticed it just logs you into their session instead of the client session that you get from physically logging into steam.
for example:
from steam.client import SteamClient
x = SteamClient()
x.login(username=, password=)
Doesn't actually log you in, since it is its own client.
I need this because I have made a script that can connect me to servers and figure out who's on it, and it relies on you being logged in to said client.
Is there any modules/libraries that will allow me to do this? and if pywinauto is the one I should use, are there any guides you know of so I can interact with said application in a good way, even on linux.
So here is what I ended up doing:
First you get a dummy display. I used this https://askubuntu.com/questions/453109/add-fake-display-when-no-monitor-is-plugged-in
After that you create the config (https://askubuntu.com/a/463000) and start up the display.
Next I created a systemd unit to launch steam using command line
[Unit]
Description=example systemd service unit file.
[Service]
User=root
Environment=XAUTHORITY=/var/run/lightdm/root/:0
Environment=DISPLAY=:0
ExecStart=/bin/bash /root/start.sh
[Install]
WantedBy=multi-user.target
(you probably shouldn't be using root for this, but its the easiest way I found to always have perms to the XAUTHORITY env, so you can login to the display)
Here is what /root/start.sh looks like
#!/bin/sh
export DISPLAY=:0
export XAUTHORITY=/var/run/lightdm/root/:0
export PATH="$PATH:/usr/games"
steam -login username_goes_here password_goes_here steam_guard_code
Note that you may not need to do the export DISPLAY and export XAUTHORITY twice. I haven't tested it.
This should just work™, it will log in, but it will ask for a steam guard code, so what you should do is attempt to start the service, receive the code through email, then append it after password.
As for doing this in python, as I said originally. you can probably use it to call start.sh, implementing command line arguments for steam code. You need these arguments because I haven't found a way for it to trust the computer you use, so it might be beneficial to implement python IMAP for reading the codes sent to the email which owns the account and using said arguments.
Theoretical Section
Please note this whole section is untested, but intended to answer my original question.
Heres what a python file might look like (very roughly)
def start_steam(code=""):
# Assuming start.sh includes reading variables from cmd line (my example didnt)
proc = subprocess.Popen(["/bin/bash", "start.sh", code])
# This function would probably just interact iwth an IMAP server to retrieve an email sent by steam support with the code. If its not found within a certain range of time, it would theoretically just stop looking
steam_guard_code = do_something_to_find_steam_guard_code_email(timeout=30)
if code:
proc.terminate()
start_steam(code=code)
else:
# proc should continue to run in the background, until restart, that means this script should ALSO run on startup, or until there isnt a PID detected for steam
with open("/var/run/steam.pid", "w") as f:
f.write(process.pid)
exit()
this python script would have to be run by systemd, it would be a forking process, and PIDfile would have to be specified by systemd (I think) so that it can be monitored.
edit: this solution has become even more theoretical, since it is not possible to pass steam guard code to steam -login username password. You would need to disable steam guard.

Only allow connections from custom clients

I'm writing a Socket Server in Python, and also a Socket Client to connect to the Server.
The Client interacts with the Server in a way that the Client sends information when an action is invoked, and the Server processes the information.
The problem I'm having, is that I am able to connect to my Server with Telnet, and probably other things that I haven't tried yet. I want to disable connection from these other Clients, and only allow connections from Python Clients. (Preferably my custom-made client, as it sends information to communicate)
Is there a way I could set up authentication on connection to differentiate Python Clients from others?
Currently there is no code, as this is a problem I want to be able to solve before getting my hands dirty.
When a new connection is made to your server, your protocol will have to specify some way for the client to authenticate. Ultimately there is nothing that the network infrastructure can do to determine what sort of process initiated the connection, so you will have to specify some exchange that allows the server to be sure that it really is talking to a valid client process.
#holdenweb has already given a good answer with basic info.
If a (terminal) software sends the bytes that your application expects as a valid identification, your app will never know whether it talks to an original client or anything else.
A possible way to test for valid clients could be, that your server sends an encrypted and authenticated question (should be different at each test!), e.g. something like "what is 18:37:12 (current date and time) plus 2 (random) hours?"
Encryption/Authentication would be another issue then.
If you keep this algorithm secret, only your clients can answer it and validate themselves successfully. It can be hacked/reverse engineered, but it is safe against basic attackers.

Running web.py as a service on linux

I've used web.py to create a web service that returns results in json.
I run it on my local box as python scriptname.py 8888
However, I now want to run it on a linux box.
How can I run it as a service on the linux box?
update
After the answers it seems like the question isn't right. I am aware of the deployment process, frameworks, and the webserver. Maybe the following back story will help:
I had a small python script that takes as input a file and based on some logic splits that file up. I wanted to use this script with a web front end I already have in place (Grails). I wanted to call this from the grails application but did not want to do it by executing a command line. So I wrapped the python script as a webservice. which takes in two parameters and returns, in json, the number of split files. This webservice will ONLY be used by my grails front end and nothing else.
So, I simply wish to run this little web.py service so that it can respond to my grails front end.
Please correct me if I'm wrong, but would I still need ngix and the like after the above? This script sounds trivial but eventually i will be adding more logic to it so I wanted it as a webservice which can be consumed by a web front end.
In general, there are two parts of this.
The "remote and event-based" part: Service used remotely over network needs certain set of skills: to be able to accept (multiple) connections, read requests, process, reply, speak at least basic TCP/HTTP, handle dead connections, and if it's more than small private LAN, it needs to be robust (think DoS) and maybe also perform some kind of authentication.
If your script is willing to take care of all of this, then it's ready to open its own port and listen. I'm not sure if web.py provides all of these facilities.
Then there's the other part, "daemonization", when you want to run the server unattended: running at boot, running under the right user, not blocking your parent (ssh, init script or whatever), not having ttys open but maybe logging somewhere...
Servers like nginx and Apache are built for this, and provide interfaces like mod_python or WSGI, so that much simpler applications can give up as much of the above as possible.
So the answer would be: yes, you still need Nginx or the likes, unless:
you can implement it yourself in Python,
or you are using the script on localhost only and are willing to take some
risks of instability.
Then probably you can do on your own.
try this
python scriptname.py 8888 2>/dev/null
it will run as daemon

Open root owned system files for reading with python

The business case...
The app server (Ubuntu/nginx/postgresql/python) that I use writes gzipped system log files as root to /var/log
I need to present data from these log files to users' browsers
My approach
I need to do a fair bit of searching and string manipulation server side so I have a python script that deals with the opening and processing and then returns a nicely formatted JSON result set. The python (cgi) script is then called using ajax from the web page.
My problem
The script works perfectly when called from the command line as SU but (...obviously) the file opening method I'm using ( gzip.open(filename) ) is failing when invoked as user www-data by the webserver.
Other useful info
The app server concerned is (contractually rather than physically) a bit of a black box - I have SU access, I can write scripts, I can read anything but I can't change file permissions, add additional python libs or or mess with config.
The subset of users who can would use this log extract also have the SU password so could be presented with a login dialog that I could pass to the script.
Given the restrictions I have, how would you go about it?
One option would be to do this somewhat sensitive "su" work in a background process that is disconnected from the web.
Likely running via cron, this script would take the root owned log files, possibly change them to a format that the web-side code could deal with easily like loading them into a database, or merely unzipping them and placing them into a different location with slightly more laxed permissions.
Then the web-side code could easily have access to the data without having to jump through the "su" hoops.
From my perspective this plan does not seem to violate your contractual rules. The web server config, permissions, etc remain intact.
My two cents. You should give a try to paramiko, allowing you to access a host (even "localhost") through SSH:
import paramiko
ssh = paramiko.SSHClient()
ssh.connect('127.0.0.1', username='jesse', password='lol')
As you have the opportunity to ask for a login/password, those would be the one provided by the user querying the log. Accessing the files is then just a matter or reading a file under SSH. And you have the opportunity to close the connection as soon as you have finished that "sensitive" work.

I need to send multiple clients a python file and have them all execute it, then download the result

So here's the deal. I have a server that has a ton of clients. I need a way to send them all a python script, and once they receive it they must immediately execute the script.
Said script will create a file which I then need to download back to the server.
The only thing I have to start with is a file with a list of client IP addresses (though with not too much effort I can change that to be client "names" if that would make the code easier).
As of right now it does not matter whether this is accomplished through POST or FTP or any other file transfer service you can think of, the only goal is that it is fast.
The script that is being executed on all the clients is a simple key generator, which I can provide if need be.
As said before the main goal of this needs to be speed, any help that can be provided would be appreciated.
Im part of a project trying to "map" the internet. I was picked, not because of any sort of networking skills (of which i have none) but because i was the only candidate that knew any python. Right now im just trying to establish a connection with all the (1000+) clients we are using and get the ssh (rsa) keys from them. Later i will be telling the clients to send traceroutes.
EDIT ---- provided additional information to make the question clearer
Have a look at Fabric. It is a tool that exactly fits what you need, though I don't know how fast it is.

Categories

Resources