.Net: Add Chroma to the vector store samples

dotnet/Directory.Packages.props

@@ -189,6 +189,7 @@
     <PackageVersion Include="CommunityToolkit.VectorData.AzureAISearch" Version="1.0.0" />
     <PackageVersion Include="CommunityToolkit.VectorData.AzureDocumentDB" Version="1.0.0" />
     <PackageVersion Include="CommunityToolkit.VectorData.AzureCosmosDB" Version="1.0.0" />
+    <PackageVersion Include="CommunityToolkit.VectorData.Chroma" Version="1.0.1-preview" />
     <PackageVersion Include="CommunityToolkit.VectorData.InMemory" Version="1.0.1" />
     <PackageVersion Include="CommunityToolkit.VectorData.PgVector" Version="1.0.0" />
     <PackageVersion Include="CommunityToolkit.VectorData.Qdrant" Version="1.0.0" />

dotnet/samples/Concepts/Concepts.csproj

@@ -53,6 +53,7 @@
     <PackageReference Include="Grpc.Net.ClientFactory" />
     <PackageReference Include="CommunityToolkit.VectorData.AzureAISearch" />
     <PackageReference Include="CommunityToolkit.VectorData.AzureDocumentDB" />
+    <PackageReference Include="CommunityToolkit.VectorData.Chroma" />
     <PackageReference Include="CommunityToolkit.VectorData.InMemory" />
     <PackageReference Include="CommunityToolkit.VectorData.Qdrant" />
     <PackageReference Include="CommunityToolkit.VectorData.PgVector" />

dotnet/samples/Concepts/Memory/VectorStoreFixtures/VectorStoreChromaContainerFixture.cs

@@ -0,0 +1,56 @@
+// Copyright (c) Microsoft. All rights reserved.
+
+using ChromaDB.Client;
+using Docker.DotNet;
+
+namespace Memory.VectorStoreFixtures;
+
+/// <summary>
+/// Fixture to use for creating a Chroma container before tests and delete it after tests.
+/// </summary>
+public class VectorStoreChromaContainerFixture : IAsyncLifetime
+{
+    private DockerClient? _dockerClient;
+    private string? _chromaContainerId;
+
+    public async Task InitializeAsync()
+    {
+    }
+
+    public async Task ManualInitializeAsync()
+    {
+        if (this._chromaContainerId == null)
+        {
+            // Connect to docker and start the docker container.
+            using var dockerClientConfiguration = new DockerClientConfiguration();
+            this._dockerClient = dockerClientConfiguration.CreateClient();
+            this._chromaContainerId = await VectorStoreInfra.SetupChromaContainerAsync(this._dockerClient);
+
+            // Delay until the Chroma server is ready.
+            var chromaClient = new ChromaClient("http://localhost:8000");
+            var succeeded = false;
+            var attemptCount = 0;
+            while (!succeeded && attemptCount++ < 10)
+            {
+                try
+                {
+                    await chromaClient.ListCollectionsAsync();
+                    succeeded = true;
+                }
+                catch (Exception)
+                {
+                    await Task.Delay(1000);
+                }
+            }
+        }
+    }
+
+    public async Task DisposeAsync()
+    {
+        if (this._dockerClient != null && this._chromaContainerId != null)
+        {
+            // Delete docker container.
+            await VectorStoreInfra.DeleteContainerAsync(this._dockerClient, this._chromaContainerId);
+        }
+    }
+}

dotnet/samples/Concepts/Memory/VectorStoreFixtures/VectorStoreInfra.cs

@@ -10,6 +10,46 @@ namespace Memory.VectorStoreFixtures;
 /// </summary>
 internal static class VectorStoreInfra
 {
+    /// <summary>
+    /// Setup the chroma container by pulling the image and running it.
+    /// </summary>
+    /// <param name="client">The docker client to create the container with.</param>
+    /// <returns>The id of the container.</returns>
+    public static async Task<string> SetupChromaContainerAsync(DockerClient client)
+    {
+        await client.Images.CreateImageAsync(
+            new ImagesCreateParameters
+            {
+                FromImage = "chromadb/chroma",
+                Tag = "latest",
+            },
+            null,
+            new Progress<JSONMessage>());
+
+        var container = await client.Containers.CreateContainerAsync(new CreateContainerParameters()
+        {
+            Image = "chromadb/chroma",
+            HostConfig = new HostConfig()
+            {
+                PortBindings = new Dictionary<string, IList<PortBinding>>
+                {
+                    {"8000", new List<PortBinding> {new() {HostPort = "8000" } }},
+                },
+                PublishAllPorts = true
+            },
+            ExposedPorts = new Dictionary<string, EmptyStruct>
+            {
+                { "8000", default },
+            },
+        });
+
+        await client.Containers.StartContainerAsync(
+            container.ID,
+            new ContainerStartParameters());
+
+        return container.ID;
+    }
+
     /// <summary>
     /// Setup the postgres pgvector container by pulling the image and running it.
     /// </summary>

dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_AzureAISearch.cs

@@ -22,6 +22,7 @@ namespace Memory;
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Redis"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_InMemory"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Postgres"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Chroma"/></para>
 ///
 /// To run this sample, you need an already existing Azure AI Search instance.
 /// To set your secrets use:

dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_Chroma.cs

@@ -0,0 +1,86 @@
+// Copyright (c) Microsoft. All rights reserved.
+
+using Azure.AI.OpenAI;
+using Azure.Identity;
+using ChromaDB.Client;
+using CommunityToolkit.VectorData.Chroma;
+using Memory.VectorStoreFixtures;
+using Microsoft.Extensions.AI;
+using Microsoft.Extensions.DependencyInjection;
+using Microsoft.SemanticKernel;
+
+namespace Memory;
+
+/// <summary>
+/// An example showing how to use common code, that can work with any vector database, with a Chroma database.
+/// The common code is in the <see cref="VectorStore_VectorSearch_MultiStore_Common"/> class.
+/// The common code ingests data into the vector store and then searches over that data.
+/// This example is part of a set of examples each showing a different vector database.
+///
+/// For other databases, see the following classes:
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_AzureAISearch"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Qdrant"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Redis"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_InMemory"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Postgres"/></para>
+///
+/// To run this sample, you need a local instance of Docker running, since the associated fixture will try and start a Chroma container in the local docker instance.
+/// </summary>
+public class VectorStore_VectorSearch_MultiStore_Chroma(ITestOutputHelper output, VectorStoreChromaContainerFixture chromaFixture) : BaseTest(output), IClassFixture<VectorStoreChromaContainerFixture>
+{
+    [Fact]
+    public async Task ExampleWithDIAsync()
+    {
+        // Use the kernel for DI purposes.
+        var kernelBuilder = Kernel
+            .CreateBuilder();
+
+        // Register an embedding generation service with the DI container.
+        kernelBuilder.AddAzureOpenAIEmbeddingGenerator(
+            deploymentName: TestConfiguration.AzureOpenAIEmbeddings.DeploymentName,
+            endpoint: TestConfiguration.AzureOpenAIEmbeddings.Endpoint,
+            credential: new AzureCliCredential(),
+            dimensions: 1536);
+
+        // Initialize the Chroma docker container via the fixtures and register the Chroma VectorStore.
+        await chromaFixture.ManualInitializeAsync();
+        kernelBuilder.Services.AddChromaVectorStore("http://localhost:8000");
+
+        // Register the test output helper common processor with the DI container.
+        kernelBuilder.Services.AddSingleton<ITestOutputHelper>(this.Output);
+        kernelBuilder.Services.AddTransient<VectorStore_VectorSearch_MultiStore_Common>();
+
+        // Build the kernel.
+        var kernel = kernelBuilder.Build();
+
+        // Build a common processor object using the DI container.
+        var processor = kernel.GetRequiredService<VectorStore_VectorSearch_MultiStore_Common>();
+
+        // Run the process and pass a key generator function to it, to generate unique record keys.
+        // The key generator function is required, since different vector stores may require different key types.
+        // E.g. Chroma supports Guid and string keys, but others may not support Guid.
+        await processor.IngestDataAndSearchAsync("skglossaryWithDI", () => Guid.NewGuid());
+    }
+
+    [Fact]
+    public async Task ExampleWithoutDIAsync()
+    {
+        // Create an embedding generation service.
+        var embeddingGenerator = new AzureOpenAIClient(new Uri(TestConfiguration.AzureOpenAIEmbeddings.Endpoint), new AzureCliCredential())
+            .GetEmbeddingClient(TestConfiguration.AzureOpenAIEmbeddings.DeploymentName)
+            .AsIEmbeddingGenerator(1536);
+
+        // Initialize the Chroma docker container via the fixtures and construct the Chroma VectorStore.
+        await chromaFixture.ManualInitializeAsync();
+        var chromaClient = new ChromaClient("http://localhost:8000");
+        var vectorStore = new ChromaVectorStore(chromaClient, ownsClient: true);
+
+        // Create the common processor that works for any vector store.
+        var processor = new VectorStore_VectorSearch_MultiStore_Common(vectorStore, embeddingGenerator, this.Output);
+
+        // Run the process and pass a key generator function to it, to generate unique record keys.
+        // The key generator function is required, since different vector stores may require different key types.
+        // E.g. Chroma supports Guid and string keys, but others may not support Guid.
+        await processor.IngestDataAndSearchAsync("skglossaryWithoutDI", () => Guid.NewGuid());
+    }
+}

dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_Common.cs

@@ -16,6 +16,7 @@ namespace Memory;
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Redis"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_InMemory"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Postgres"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Chroma"/></para>
 /// </summary>
 /// <param name="vectorStore">The vector store to ingest data into.</param>
 /// <param name="embeddingGenerator">The service to use for generating embeddings.</param>

dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_InMemory.cs

@@ -20,6 +20,7 @@ namespace Memory;
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Redis"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Qdrant"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Postgres"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Chroma"/></para>
 /// </summary>
 public class VectorStore_VectorSearch_MultiStore_InMemory(ITestOutputHelper output) : BaseTest(output)
 {

dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_Postgres.cs

@@ -21,6 +21,7 @@ namespace Memory;
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Qdrant"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Redis"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_InMemory"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Chroma"/></para>
 ///
 /// To run this sample, you need a local instance of Docker running, since the associated fixture will try and start a Postgres container in the local docker instance.
 /// </summary>

dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_Qdrant.cs

@@ -22,6 +22,7 @@ namespace Memory;
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Redis"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_InMemory"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Postgres"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Chroma"/></para>
 ///
 /// To run this sample, you need a local instance of Docker running, since the associated fixture will try and start a Qdrant container in the local docker instance.
 /// </summary>

dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_Redis.cs

@@ -22,6 +22,7 @@ namespace Memory;
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Qdrant"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_InMemory"/></para>
 /// <para><see cref="VectorStore_VectorSearch_MultiStore_Postgres"/></para>
+/// <para><see cref="VectorStore_VectorSearch_MultiStore_Chroma"/></para>
 ///
 /// Redis supports two record storage types: Json and HashSet.
 /// Note the use of the <see cref="RedisStorageType"/> enum to specify the preferred storage type.

dotnet/samples/Demos/VectorStoreRAG/Options/ApplicationConfig.cs

@@ -16,6 +16,7 @@ internal sealed class ApplicationConfig
     private readonly RagConfig _ragConfig = new();
     private readonly AzureAISearchConfig _azureAISearchConfig = new();
     private readonly CosmosConfig _azureDocumentDBConfig = new();
+    private readonly ChromaConfig _chromaConfig = new();
     private readonly CosmosConfig _cosmosNoSqlConfig = new();
     private readonly QdrantConfig _qdrantConfig = new();
     private readonly RedisConfig _redisConfig = new();
@@ -45,6 +46,9 @@ public ApplicationConfig(ConfigurationManager configurationManager)
         configurationManager
             .GetRequiredSection($"VectorStores:{CosmosConfig.DocumentDBConfigSectionName}")
             .Bind(this._azureDocumentDBConfig);
+        configurationManager
+            .GetRequiredSection($"VectorStores:{ChromaConfig.ConfigSectionName}")
+            .Bind(this._chromaConfig);
         configurationManager
             .GetRequiredSection($"VectorStores:{CosmosConfig.NoSqlConfigSectionName}")
             .Bind(this._cosmosNoSqlConfig);
@@ -73,6 +77,8 @@ public ApplicationConfig(ConfigurationManager configurationManager)
 
     public CosmosConfig AzureDocumentDBConfig => this._azureDocumentDBConfig;
 
+    public ChromaConfig ChromaConfig => this._chromaConfig;
+
     public CosmosConfig CosmosNoSqlConfig => this._cosmosNoSqlConfig;
 
     public QdrantConfig QdrantConfig => this._qdrantConfig;

dotnet/samples/Demos/VectorStoreRAG/Options/ChromaConfig.cs

@@ -0,0 +1,16 @@
+// Copyright (c) Microsoft. All rights reserved.
+
+using System.ComponentModel.DataAnnotations;
+
+namespace VectorStoreRAG.Options;
+
+/// <summary>
+/// Chroma service settings.
+/// </summary>
+internal sealed class ChromaConfig
+{
+    public const string ConfigSectionName = "Chroma";
+
+    [Required]
+    public string ConnectionString { get; set; } = string.Empty;
+}

dotnet/samples/Demos/VectorStoreRAG/Program.cs

@@ -82,6 +82,11 @@
             appConfig.AzureDocumentDBConfig.ConnectionString,
             appConfig.AzureDocumentDBConfig.DatabaseName);
         break;
+    case "Chroma":
+        kernelBuilder.Services.AddChromaCollection<string, TextSnippet<string>>(
+            appConfig.RagConfig.CollectionName,
+            appConfig.ChromaConfig.ConnectionString);
+        break;
     case "CosmosNoSql":
         kernelBuilder.Services.AddCosmosCollection<string, TextSnippet<string>>(
             appConfig.RagConfig.CollectionName,
@@ -121,6 +126,7 @@
 {
     case "AzureAISearch":
     case "AzureDocumentDB":
+    case "Chroma":
     case "CosmosNoSql":
     case "InMemory":
     case "Redis":

dotnet/samples/Demos/VectorStoreRAG/README.md

@@ -10,6 +10,7 @@ The sample can be configured in various ways:
 1. You can choose your preferred vector store by setting the `Rag:VectorStoreType` configuration setting in the `appsettings.json` file to one of the following values:
    1. AzureAISearch
    1. AzureDocumentDB
+   1. Chroma
    1. CosmosNoSql
    1. InMemory
    1. Qdrant
@@ -130,6 +131,22 @@ dotnet user-secrets set "VectorStores:CosmosNoSql:ConnectionString" "<yourconnec
 dotnet user-secrets set "VectorStores:CosmosNoSql:DatabaseName" "<yourdbname>"
 ```
 
+### Chroma
+
+If you want to use Chroma as your vector store, you will need to have an instance of Chroma available.
+
+You can use the following command to start a Chroma instance in docker, and this will work with the default configured settings:
+
+```cli
+docker run -d --name chroma -p 8000:8000 chromadb/chroma:latest
+```
+
+If you want to use a different instance of Chroma, you can update the appsettings.json file or add the following secret to reconfigure:
+
+```cli
+dotnet user-secrets set "VectorStores:Chroma:ConnectionString" "<yourchromaconnectionstring>"
+```
+
 ### Qdrant
 
 If you want to use Qdrant as your vector store, you will need to have an instance of Qdrant available.

dotnet/samples/Demos/VectorStoreRAG/VectorStoreRAG.csproj

@@ -20,6 +20,7 @@
     <PackageReference Include="CommunityToolkit.VectorData.AzureAISearch" />
     <PackageReference Include="CommunityToolkit.VectorData.AzureDocumentDB" />
     <PackageReference Include="CommunityToolkit.VectorData.AzureCosmosDB" />
+    <PackageReference Include="CommunityToolkit.VectorData.Chroma" />
     <PackageReference Include="CommunityToolkit.VectorData.InMemory" />
     <PackageReference Include="CommunityToolkit.VectorData.Qdrant" />
     <PackageReference Include="CommunityToolkit.VectorData.Redis" />

dotnet/samples/Demos/VectorStoreRAG/appsettings.json

@@ -33,6 +33,9 @@
       "ConnectionString": "",
       "DatabaseName": ""
     },
+    "Chroma": {
+      "ConnectionString": "http://localhost:8000"
+    },
     "CosmosNoSql": {
       "ConnectionString": "",
       "DatabaseName": ""

dotnet/src/InternalUtilities/samples/InternalUtilities/TestConfiguration.cs

@@ -41,7 +41,6 @@ public static void Initialize(IConfigurationRoot configRoot)
     public static PostgresConfig Postgres => LoadSection<PostgresConfig>();
     public static RedisConfig Redis => LoadSection<RedisConfig>();
     public static JiraConfig Jira => LoadSection<JiraConfig>();
-    public static ChromaConfig Chroma => LoadSection<ChromaConfig>();
     public static MongoDBConfig MongoDB => LoadSection<MongoDBConfig>();
     public static ChatGPTRetrievalPluginConfig ChatGPTRetrievalPlugin => LoadSection<ChatGPTRetrievalPluginConfig>();
     public static MsGraphConfiguration MSGraph => LoadSection<MsGraphConfiguration>();
@@ -219,11 +218,6 @@ public class JiraConfig
         public string Domain { get; set; }
     }
 
-    public class ChromaConfig
-    {
-        public string Endpoint { get; set; }
-    }
-
     public class MongoDBConfig
     {
         public string ConnectionString { get; set; }

dotnet/src/VectorData/Chroma/README.md

@@ -1 +1 @@
-Microsoft.SemanticKernel.Connectors.Chroma and the MemoryStore abstraction has been deprecated.  Please use the new Microsoft.Extensions.VectorData abstractions with the Chroma implementation that can be found in the [CommunityToolkit/AI repository](https://github.com/CommunityToolkit/AI/tree/main/MEVD/src/Chroma)
+Microsoft.SemanticKernel.Connectors.Chroma and the MemoryStore abstraction have been deprecated. Please use the new Microsoft.Extensions.VectorData abstractions with the Chroma implementation that can be found in the [CommunityToolkit/AI repository](https://github.com/CommunityToolkit/AI/tree/main/MEVD/src/Chroma), published as [CommunityToolkit.VectorData.Chroma](https://www.nuget.org/packages/CommunityToolkit.VectorData.Chroma).