rio is a library for temporarily patching python objects so they are executed by a remote server. It's built using mock and zerorpc and has a goal of simply wrapping existing code in a context manager to alter behavior.

Build

$ git clone https://github.com/sambvfx/rio.git
$ cd rio
$ pip install .

Example

Lets pick a host machine to serve our remote methods. For the sake of this example, this server's cname is sv-rio01.

Host [sv-rio01]

First lets just touch a path on disk.

$ touch /tmp/example

Then start up a server on port 4242. For now we'll only bother hosting/patching os.path.exists.

import os
import rio.server

methods = {
    'os.path.exists': os.path.exists,
}

rio.server.start(methods=methods, port=4242)

TIP: Check out rio.collections.fs to see helpers to patch all file system methods.

Next we can test it's working by running some python from a different machine.

Client

import os
from rio import rio


# connection string to connect Client->Server
remotefs = 'tcp://sv-rio01:4242'

path = '/tmp/example'

# local call
assert not os.path.exists(path)

# remote call
with rio(remotefs):
    assert os.path.exists(path)

What just happened here? Within the rio context manager, the os.path.exists method was patched to execute via rpc.

NOTE: The methods that get patched are determined by the server connected to.

If you're familiar with the testing library mock, it's similar to doing:

with mock.patch('os.path.exists', side_effect=rpc_os_path_exists):
    ...

Contribute

Contributions welcome! This is in a prototype stage and could use more robust testing and error handling.

pip install in editable mode with the tests bundle.

$ git clone https://github.com/sambvfx/rio.git
$ cd rio
$ virtualenv venv
$ source venv/bin/activate
$ pip install -e ".[tests]"