cpias

Cell Profiling Image Analysis Server

Install

git clone https://github.com/CellProfiling/cpias.git
cd cpias
pip install .

Run

• Open a terminal, we call it terminal 1. In terminal 1, start the server.

cpias --help
# Start the server
cpias start-server

• Open another terminal, we call it terminal 2. In terminal 2 run the client.

# Run the client
cpias run-client

Add new commands

New commands should preferably be added in a standalone package, by using a setup.py file and the entry_points interface. cpias will look for entry points registered under "cpias.commands".

# setup.py
...
entry_points={
    "cpias.commands": ["hello = cpias.commands.hello"],
},
...

See the setup.py file of this package for a real example. See the packaging docs for details.

This will load all the command modules specified in the list, eg cpias.commands.hello. Inside each module there should be a function defined named register_command. It should accept one positional argument, server. This is the cpias server instance.

def register_command(server: "CPIAServer") -> None:
    """Register the hello command."""
    server.register_command("hello", hello)

See the hello.py command included in this package for examples of different types of commands.

Message structure

cpias uses a json serialized format for the messages sent over the socket. Each message should contain three items in the json object, cli, cmd and dta.

• Here's an example message as json.

{"cli": "client-1", "cmd": "hello", "dta": {"param1": "world"}}

• Here's the same message serialized and with line break to mark message end. It's the serialized version of the message that should be sent over the socket.

'{"cli": "client-1", "cmd": "hello", "dta": {"planet": "world"}}\n'

• The cli item should mark the client id.
• The cmd item should mark the command id.
• The dta item should hold another json object with arbitrary data items. Only requirement is that the message can be serialized. Each item in dta will be passed to the command function as a named argument.

Development

pip install -r requirements_dev.txt
# Use the makefile for common dev tasks
make

• Here's a list of development tools we use.
  • black
  • flake8
  • pylint
  • pydocstyle
  • mypy
  • pytest
  • tox