Skip to content

Lightweight, decorator-based Python implementation of a Finite State Machine

License

Notifications You must be signed in to change notification settings

alysivji/finite-state-machine

Repository files navigation

Finite State Machine Banner

Finite State Machine

Latest Release Supports Python 3.6+ License: MIT Code Style: Black

Build Status codecov

Lightweight, decorator-based Python implementation of a Finite State Machine.

Table of Contents

Installation

pip install finite-state-machine

Usage

Subclass StateMachine and set the state instance variable:

from finite_state_machine import StateMachine, transition

class LightSwitch(StateMachine):
    def __init__(self):
        self.state = "off"
        super().__init__()

The transition decorator can be used to specify valid state transitions with an optional parameter for conditions. States can be of type: string, int, bool, Enum, or IntEnum. Can specify a single sate or a list of states for the source parameter; can only specify a single state as the target target. All condition functions need to return True for the transition to occur, else a ConditionsNotMet exception will be raised. Condition functions require the same positional position and keyword arguments present in the transition function.

    @transition(source="off", target="on", conditions=[light_is_off])
    def turn_on(self):
        # transition function
        # logic to define what happens to "complete a transition"
        # ex. update database record,
        ...

def light_is_off(machine):
    # condition function, first param will always be the state machine class
    # return a boolean to specify if a transition is valid
    return machine.state == "off"

Can also specify an on_error state to handle situations where the transition function raises an exception:

    @transition(source="off", target="on", on_error="failed")
    def turn_on(self):
        raise ValueError

Example

from finite_state_machine import StateMachine, transition

class Turnstile(StateMachine):
    initial_state = "close"

    def __init__(self):
        self.state = self.initial_state
        super().__init__()

    @transition(source=["close", "open"], target="open")
    def insert_coin(self):
        pass

    @transition(source="open", target="close")
    def pass_thru(self):
        pass

REPL

In [2]: turnstile = Turnstile()

In [3]: turnstile.state
Out[3]: 'close'

In [4]: turnstile.insert_coin()

In [5]: turnstile.state
Out[5]: 'open'

In [6]: turnstile.insert_coin()

In [7]: turnstile.state
Out[7]: 'open'

In [8]: turnstile.pass_thru()

In [9]: turnstile.state
Out[9]: 'close'

In [10]: turnstile.pass_thru()
---------------------------------------------------------------------------
InvalidStartState                         Traceback (most recent call last)
<ipython-input-10-6abc6f4be1cd> in <module>
----> 1 turnstile.pass_thru()

~/state_machine.py in _wrapper(*args, **kwargs)
     32
     33             if self.state not in source:
---> 34                 raise InvalidStartState
     35
     36             for condition in conditions:

InvalidStartState:

The examples folder contains additional workflows.

Asynchronous Support

finite-state-machine can be used to build both synchronous and asynchronous State Machines. The @transition decorator supports transition functions and condition functions as follows:

Synchronous transition function Asynchronous transition function
Synchronous condition function
Asynchronous condition function

State Diagram

State Machine workflows can be visualized using a state diagram.

finite-state-machine generates diagrams using Mermaid Markdown syntax, which can be viewed using the Mermaid Live Editor.

Use the fsm_draw_state_diagram command and point to State Machine workflow class that inherits from StateMachine.

# class parameter is required
$ fsm_draw_state_diagram --class examples.turnstile:Turnstile

# initial_state parameter is optional
$ fsm_draw_state_diagram --class examples.turnstile:Turnstile --initial_state close

Contributing

  1. Clone repo
  2. Create a virtual environment
  3. pip install -r requirements_dev.txt
  4. Install pre-commit
  5. Set up pre-commit hooks in repo: pre-commit install

To install a package locally for development, run:

flit install [--symlink] [--python path/to/python]

Running Tests

pytest

Inspiration

This project is inspired by django-fsm. I wanted a decorator-based state machine without having to use Django.