practical-dotnet-aspire Documentation

The CoffeeShop app demonstrates intelligent application development through two primary AI patterns:

1. Seeding data via Chat Completion: Using LLMs to generate initial datasets for the application.
2. Semantic Searching via Vector Embeddings: Implementing search capabilities based on the meaning of queries rather than exact keyword matches.

These features are implemented using Microsoft.Extensions.AI and can be configured to use Ollama for local development or Azure OpenAI for cloud-based services.

To run the CoffeeShop application locally, you can use Visual Studio or the .NET CLI. The application is built on .NET 9.0 and .NET Aspire 9, utilizing a microservices architecture.

Using Visual Studio

1. Open the project in Visual Studio.
2. Press F5 to build and run the application.

Using .NET CLI

Run the following commands from the root directory:

> dotnet build coffeeshop-aspire.sln
> dotnet run --project app-host/CoffeeShop.AppHost.csproj

Once running, the application dashboard/entry point is available at http://localhost:5019.

When using the provided Docker Compose configuration, you can control the container images and registry using the following environment variables:

• DOCKER_REGISTRY: The registry URL. Defaults to ghcr.io/thangchung/coffeeshop-aspire.
• IMAGE_TAG: The specific image tag to pull. Defaults to latest.

# Example usage
export DOCKER_REGISTRY=my-private-registry.io
export IMAGE_TAG=v1.0.0
docker-compose up

The PlaceOrderCommand is a MediatR request used to initiate an order within the CounterApi. It encapsulates all necessary details for an order, including the source, location, and specific items for both the Barista (drinks) and Kitchen (food).

var command = new PlaceOrderCommand
{
    OrderId = Guid.NewGuid(), // Or use GuidHelper.NewGuid()
    CommandType = CommandType.PLACE_ORDER,
    OrderSource = OrderSource.COUNTER,
    Location = Location.ATLANTA,
    LoyaltyMemberId = Guid.NewGuid(),
    BaristaItems = new List<CommandItem> 
    {
        new CommandItem { ItemType = ItemType.COFFEE } 
    },
    KitchenItems = new List<CommandItem> 
    {
        new CommandItem { ItemType = ItemType.PASTRY } 
    },
    Timestamp = DateTime.UtcNow
};

// Sent via MediatR
var result = await _mediator.Send(command);

The PlaceOrderCommand utilizes the following supporting types to categorize items and command intent:

public class CommandItem
{
    public ItemType ItemType { get; set; }
}

public enum CommandType
{
    PLACE_ORDER
}

The RabbitMQ service is used for messaging and exposes the following ports:

• 5672: RabbitMQ broker port.
• 15672: RabbitMQ Management UI port.

It uses the masstransit/rabbitmq:latest image and includes a healthcheck using rabbitmq-diagnostics -q ping.

rabbitmq:
    image: masstransit/rabbitmq:latest
    healthcheck:
      test: rabbitmq-diagnostics -q ping
      interval: 30s
      timeout: 30s
      retries: 3
    ports:
      - "5672:5672"
      - "15672:15672"
    networks:
      - coffeeshop-network

Each microservice exposes a specific port for external access:

The following properties define the structure of a PlaceOrderCommand request:

public class PlaceOrderCommand : IRequest<IResult>
{
    public Guid OrderId { get; set; } // Defaults to GuidHelper.NewGuid()
    public CommandType CommandType { get; set; } // Defaults to CommandType.PLACE_ORDER
    public OrderSource OrderSource { get; set; } // Defaults to OrderSource.COUNTER
    public Location Location { get; set; } // Defaults to Location.ATLANTA
    public Guid LoyaltyMemberId { get; set; } // Defaults to GuidHelper.NewGuid()
    public List<CommandItem> BaristaItems { get; set; } // List of items for the barista
    public List<CommandItem> KitchenItems { get; set; } // List of items for the kitchen
    public DateTime Timestamp { get; set; } // Defaults to DateTimeHelper.NewDateTime()
}