SharpFsm

A flexible finite state machine (FSM) library for .NET written in C#, it is designed to create and manage Finite State Machines (FSMs) in a simple and efficient way. It allows developers to define states, transitions, and events, enabling the modeling of complex behaviors in a structured manner.

Quick Start

Install SharpFsm via NuGet Package Manager or dotnet.

Install-Package SharpFsm

dotnet add package SharpFsm

public enum SwitchState { Off, On }
public class SwitchContext { }
var builder = FiniteStateMachineBuilder<SwitchState, SwitchContext>.Create("Switch")
    .WithInitialState(SwitchState.Off)
    .AddTransition(SwitchState.Off, SwitchState.On).Done()
    .AddTransition(SwitchState.On, SwitchState.Off).Done();
var fsm = new FiniteStateMachine<SwitchState, SwitchContext>(builder.Build());
var context = new SwitchContext();
Console.WriteLine(fsm.Current); // Off
fsm.TryTransitionTo(SwitchState.On, context);
Console.WriteLine(fsm.Current); // On

Features

• State Management: Easily define and manage states in your FSM. Strongly-typed FSMs using enums for states
• Transition Handling: Transition conditions and side effects.
• Serialization: Serialization to/from JSON and YAML.
• Builder Pattern: Builder pattern for easy FSM construction.
• Extensible: Easily extend the library to fit specific needs or integrate with other systems.
• Cross-Platform: Multi-targeted for broad .NET compatibility

Concepts

State

A state represents a distinct condition or situation in the lifecycle of an object or process. In SharpFsm, states are typically defined using an enum (e.g., Open, Closed, InProgress).

Transition

A transition is a rule that defines how the FSM moves from one state to another. Each transition specifies:

• The source state (where the transition starts)
• The target state (where the transition ends)
• An optional condition (a function that must return true for the transition to occur)
• An optional side effect (an action to execute when the transition happens)

Condition

A condition is a predicate (function) that determines if a transition is allowed, based on the current context. Example: ctx => ctx.IsAgentAssigned

Side Effect

A side effect is an action that is executed when a transition occurs. Example: ctx => Console.WriteLine("Customer notified")

Context

The context is an object that holds data relevant to the FSM’s operation and is passed to conditions and side effects. Example: a TicketContext with properties like IsAgentAssigned.

State Machine Definition

A state machine definition describes all possible states, transitions, the initial state, and associated logic for an FSM.

Advantage of using FSM

Using a finite state machine (FSM) for managing state offers several advantages over ad-hoc approaches (like scattered conditionals, flags, or event-driven code) and even over some object-oriented state patterns. Here are the key benefits:

1. Clarity and Explicitness
   • All possible states and transitions are explicitly defined.
   • The system’s behavior is easy to visualize, reason about, and document.
   • Reduces ambiguity and hidden state changes.
2. Predictability and Robustness
   • Transitions are controlled and validated.
   • Only allowed transitions can occur, preventing invalid or unexpected state changes.
   • Makes it easier to handle edge cases and errors.
3. Maintainability
   • Adding or modifying states and transitions is straightforward.
   • Changes are localized to the FSM definition, not scattered across the codebase.
   • Reduces the risk of introducing bugs when requirements change.
4. Testability
   • FSMs are easy to test.
   • You can systematically test all states and transitions.
   • Makes it easier to write unit tests for state-dependent logic.
5. Separation of Concerns
   • State logic is separated from business logic.
   • Conditions and side effects are encapsulated, making the codebase cleaner and more modular.
6. Scalability
   • FSMs scale well as complexity grows.
   • Adding new states or transitions does not exponentially increase code complexity, unlike nested if/else or switch statements.
7. Visualization and Documentation
   • FSMs can be visualized as state diagrams.
   • This aids in communication with stakeholders and helps onboard new developers.

Approach	Pros of FSM over this approach
If/else, switch	Avoids spaghetti code, centralizes state logic
Flags/booleans	Prevents invalid state combinations
Event-driven	Makes allowed transitions explicit and predictable
State pattern	FSM is more declarative and easier to visualize

Use Cases

Here are some common use cases for implementing a finite state machine (FSM) in software development:

1. Workflow and Process Management
   • Example: Ticketing systems, order processing, approval workflows.
   • Why: Each item moves through a series of well-defined states (e.g., Open → In Progress → Resolved).
2. User Interface (UI) Navigation
   • Example: Wizard dialogs, multi-step forms, menu navigation.
   • Why: UI components often have distinct states and transitions based on user actions.
3. Game Development
   • Example: Character states (Idle, Walking, Jumping, Attacking), enemy AI behaviors.
   • Why: Game entities often have clear, rule-based state transitions.
4. Protocol and Communication Handling
   • Example: Network protocol implementations (TCP handshake, HTTP request/response), parsers.
   • Why: Protocols are defined by sequences of states and transitions based on received data.
5. Device and Hardware Control
   • Example: Embedded systems, robotics, IoT devices (e.g., a washing machine: Idle → Washing → Rinsing → Spinning → Done).
   • Why: Devices operate in modes with strict rules for moving between them.
6. Authentication and Authorization Flows
   • Example: Login processes, multi-factor authentication, session management.
   • Why: Security flows require strict control over allowed transitions.
7. Error Handling and Recovery
   • Example: Retry logic, circuit breakers, transaction management.
   • Why: Systems need to move between normal, error, and recovery states in a controlled way.
8. Text Parsing and Lexical Analysis
   • Example: Tokenizers, interpreters, compilers.
   • Why: Parsing often involves moving through states based on input characters or tokens.

Example

Let's define a state machine for order management.

stateDiagram-v2
    [*] --> Created
    Created --> Paid: PaymentReceived
    Paid --> Packed: PackingComplete
    Packed --> Shipped: Shipped
    Shipped --> Delivered: Delivered
    Delivered --> Returned: ReturnRequested

    Created --> Cancelled: CancelRequested
    Paid --> Cancelled: CancelRequested
    Packed --> Cancelled: CancelRequested
    Shipped --> Cancelled: CancelRequested

    Cancelled --> [*]
    Delivered --> [*]
    Returned --> [*]

Loading

1. Define States and Context

public enum OrderState
{
    Created,
    Paid,
    Packed,
    Shipped,
    Delivered,
    Cancelled,
    Returned
}

public class OrderContext
{
    // This class can contain direct access to your data stores 
    // ex: connect to your database to do complex calculations
    public bool PaymentReceived { get; set; }
    public bool PackingComplete { get; set; }
    public bool Shipped { get; set; }
    public bool Delivered { get; set; }
    public bool CancelRequested { get; set; }
    public bool ReturnRequested { get; set; }
}

2. Register Conditions and Side Effects

using SharpFsm;

var registry = new TransitionRegistry<OrderState, OrderContext>();
registry.RegisterCondition("PaymentReceived", ctx => ctx.PaymentReceived);
registry.RegisterCondition("PackingComplete", ctx => ctx.PackingComplete);
registry.RegisterCondition("Shipped", ctx => ctx.Shipped);
registry.RegisterCondition("Delivered", ctx => ctx.Delivered);
registry.RegisterCondition("CancelRequested", ctx => ctx.CancelRequested);
registry.RegisterCondition("ReturnRequested", ctx => ctx.ReturnRequested);

registry.RegisterSideEffect("NotifyShipment", (ctx, _, _) => Console.WriteLine("Customer notified: Order shipped"));
registry.RegisterSideEffect("NotifyDelivery", (ctx, _, _) => Console.WriteLine("Customer notified: Order delivered"));
registry.RegisterSideEffect("NotifyCancel", (ctx, _, _) => Console.WriteLine("Customer notified: Order cancelled"));
registry.RegisterSideEffect("NotifyReturn", (ctx, _, _) => Console.WriteLine("Customer notified: Order returned"));

3. Build a State Machine

var builder = FiniteStateMachineBuilder<OrderState, OrderContext>.Create("Order")
    .WithInitialState(OrderState.Created)
    .WithRegistry(registry)
    .AddTransition(OrderState.Created, OrderState.Paid)
        .When("PaymentReceived")
        .Done()
    .AddTransition(OrderState.Paid, OrderState.Packed)
        .When("PackingComplete")
        .Done()
    .AddTransition(OrderState.Packed, OrderState.Shipped)
        .When("Shipped")
        .WithSideEffect("NotifyShipment")
        .Done()
    .AddTransition(OrderState.Shipped, OrderState.Delivered)
        .When("Delivered")
        .WithSideEffect("NotifyDelivery")
        .Done()
    .AddTransition(OrderState.Created, OrderState.Cancelled)
        .When("CancelRequested")
        .WithSideEffect("NotifyCancel")
        .Done()
    .AddTransition(OrderState.Paid, OrderState.Cancelled)
        .When("CancelRequested")
        .WithSideEffect("NotifyCancel")
        .Done()
    .AddTransition(OrderState.Packed, OrderState.Cancelled)
        .When("CancelRequested")
        .WithSideEffect("NotifyCancel")
        .Done()
    .AddTransition(OrderState.Shipped, OrderState.Cancelled)
        .When("CancelRequested")
        .WithSideEffect("NotifyCancel")
        .Done()
    .AddTransition(OrderState.Delivered, OrderState.Returned)
        .When("ReturnRequested")
        .WithSideEffect("NotifyReturn")
        .Done();

var fsm = new FiniteStateMachine<OrderState, OrderContext>(builder.Build());

3. Use the State Machine

var context = new OrderContext { PaymentReceived = true };
fsm.TryTransitionTo(OrderState.Paid, context); // Moves to Paid

context.PackingComplete = true;
fsm.TryTransitionTo(OrderState.Packed, context); // Moves to Packed

context.Shipped = true;
fsm.TryTransitionTo(OrderState.Shipped, context); // Moves to Shipped, notifies shipment

context.Delivered = true;
fsm.TryTransitionTo(OrderState.Delivered, context); // Moves to Delivered, notifies delivery

context.ReturnRequested = true;
fsm.TryTransitionTo(OrderState.Returned, context); // Moves to Returned, notifies return

Serialization/Deserialization Example

1. JSON

using System.Text.Json;

// Serialization
var definition = builder.Build(); // Ensure build is called before calling ToSerializable
var serializedObject = builder.ToSerializable(); // returns SerializableStateMachine object
string json = JsonSerializer.Serialize(serializedObject);

// Deserialization
var jsonDto = JsonSerializer.Deserialize<SerializableStateMachine>(json);
var loadedBuilder = FiniteStateMachineBuilder<TicketState, TicketContext>.Create(jsonDto.EntityType)
    .LoadFrom(jsonDto, registry);
var loadedFsm = loadedBuilder.Build();
var sm = new FiniteStateMachine<TicketState, TicketContext>(loadedFsm);

2. YAML

using YamlDotNet.Serialization;
using YamlDotNet.Serialization.NamingConventions;

// Serialization
var definition = builder.Build(); // Ensure build is called before calling ToSerializable
var serializedObject = builder.ToSerializable(); // returns SerializableStateMachine object
var yamlSerializer = new SerializerBuilder()
	.WithNamingConvention(CamelCaseNamingConvention.Instance)
	.Build();
var yaml = yamlSerializer.Serialize(serializedObject);

// Deserialization
var yamlDeserializer = new DeserializerBuilder()
	.WithNamingConvention(CamelCaseNamingConvention.Instance)
	.IgnoreUnmatchedProperties()
	.Build();
var deserializedDto = yamlDeserializer.Deserialize<SerializableStateMachine>(new StringReader(yaml));
var loadedBuilder = FiniteStateMachineBuilder<TicketState, TicketContext>.Create("Ticket")
    .LoadFrom(deserializedDto!, registry);
var sm = new FiniteStateMachine<TicketState, TicketContext>(loadedBuilder.Build());

API Overview

Namespaces

• SharpFsm
• SharpFsm.Serialization

Core Types

1. FiniteStateMachine<TState, TContext>: Represents a runtime finite state machine.

   • Type Parameters:
     • TState: Enum type representing states.
     • TContext: Type for contextual data passed to conditions and side effects.
   • Properties:
     • TState Current The current state of the FSM.
   • Methods:
     • bool CanTransitionTo(TState target, TContext context) Checks if a transition to the target state is possible given the context.
     • bool TryTransitionTo(TState target, TContext context) Attempts to transition to the target state. Returns true if successful.

2. FiniteStateMachineBuilder<TState, TContext>: Fluent builder for defining FSMs

   • Type Parameters:
     • TState: Enum type representing states.
     • TContext: Type for contextual data passed to conditions and side effects.
   • Static Methods:
     • FiniteStateMachineBuilder<TState, TContext> Create(string entityType): Creates a new instance of the FiniteStateMachineBuilder for the specified entity type.
   • Instance Methods:
     • FiniteStateMachineBuilder<TState, TContext> WithInitialState(TState state): Sets the initial state of the finite state machine. This method also adds the state to the list of states if it is not already present.
     • TransitionBuilder AddTransition(TState from, TState to): Adds a transition between two states in the finite state machine. This method also adds both states to the list of states if they are not already present.
     • EnumStateMachineDefinition<TState, TContext> Build(): Builds the finite state machine definition. This method will throw an exception if the initial state has not been set.
     • SerializableStateMachine ToSerializable(): Converts the current state machine builder to a serializable state machine definition.
     • FiniteStateMachineBuilder<TState, TContext> LoadFrom(SerializableStateMachine dto, TransitionRegistry<TContext> registry): Loads a finite state machine from a serializable state machine definition.
   • Nested Type:
     • TransitionBuilder: Builder for defining transitions in the finite state machine.

3. EnumStateMachineDefinition<TState, TContext>: Describes the structure of an FSM (states, transitions, initial state).

   • Type Parameters:
     • TState: Enum type representing states.
     • TContext: Type for contextual data passed to conditions and side effects.
   • Properties:
     • string EntityType: The type of work item this state machine is associated with.
     • IEnumerable<IState> States: A collection of all states in the state machine, represented as IState instances.
     • IEnumerable<ITransition<TContext>> Transitions: A collection of transitions between states, represented as ITransition instances.
     • IState InitialState: The initial state of the state machine, represented as an IState instance.

4. TransitionRegistry<TContext>: Registry for reusable conditions and side effects.

   • Type Parameters:
     • TContext: Type for contextual data passed to conditions and side effects.
   • Properties:
     • Dictionary<string, Func<TContext, bool>> Conditions: Stores conditions that can be used to evaluate whether a transition can occur.
     • Dictionary<string, Action<TContext>> SideEffects: Stores side effects that can be executed when a transition occurs.
   • Methods:
     • void RegisterCondition(string name, Func<TContext, bool> fn): Registers a condition with a name that can be used to evaluate transitions.
     • void RegisterSideEffect(string name, Action<TContext> fn): Registers a side effect with a name that can be executed when a transition occurs.

5. ITransition<TContext>: Represents a transition between two states.

   • Type Parameters:
     • TContext: Type for contextual data passed to conditions and side effects.
   • Properties:
     • IState From: Gets the source state of the transition.
     • IState To: Gets the target state of the transition.
     • Func<TContext, bool> Condition: Gets the condition that must be met for this transition to occur.
     • Action<TContext> SideEffect: Gets an optional side effect that occurs when the transition is taken.
     • string ConditionName: Gets the name of the condition that triggers this transition, if any.
     • string SideEffectName: Gets the name of the side effect that occurs when this transition is taken, if any.

6. IState: Represents a state in the FSM.

   • Property:
     • string Name: Gets the name of the state.

7. SerializableStateMachine:

   • Properties:
     • string EntityType: The type of entity this state machine is associated with.
     • string InitialState: The initial state of the state machine.
     • List<string> States: A list of all states in the state machine.
     • List<SerializableTransition> Transitions: A list of transitions between states.

8. SerializableTransition:

   • Properties:
     • string From: The state this transition is from.
     • string To: The state this transition is to.
     • string ConditionName: The name of the condition that triggers this transition.
     • string SideEffectName: The name of the side effect that occurs when this transition is taken.