ChromaDotNet.VectorData 0.4.0

README.md

@@ -17,7 +17,7 @@ Website: [chromadotnet.org](https://chromadotnet.org)
 | [VectorStore_VectorSearch_MultiStore_Chroma](./samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_Chroma.cs) | Ingests a glossary into Chroma and searches it, with and without a filter. It goes through the common vector store sample code, with and without dependency injection. Adapted from [VectorStore_VectorSearch_MultiStore_Qdrant](https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_Qdrant.cs). |
 | [VectorStore_DynamicDataModel_Interop_Chroma](./samples/Concepts/Memory/VectorStore_DynamicDataModel_Interop_Chroma.cs) | Writes records as dictionaries described by a record definition, then reads them back as a .NET data model. It also does the reverse. Adapted from [VectorStore_DynamicDataModel_Interop](https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/Concepts/Memory/VectorStore_DynamicDataModel_Interop.cs). |
 | [VectorStore_VectorSearch_Paging_Chroma](./samples/Concepts/Memory/VectorStore_VectorSearch_Paging_Chroma.cs) | Uses `Top` and `Skip` to page through vector search results over 1,000 records. Needs no model. Adapted from [VectorStore_VectorSearch_Paging](https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/Concepts/Memory/VectorStore_VectorSearch_Paging.cs), but with the Euclidean distance. Chroma searches with an approximate index, and with the cosine distance that index can miss some of the made-up vectors in the sample. |
-| [VectorStore_HybridSearch_Simple_Chroma](./samples/Concepts/Memory/VectorStore_HybridSearch_Simple_Chroma.cs) | Searches a glossary on Chroma Cloud with a vector and keywords, with and without a filter. The vector store creates the collection with a BM25 index for the full-text indexed property. Adapted from [VectorStore_HybridSearch_Simple_AzureAISearch](https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/Concepts/Memory/VectorStore_HybridSearch_Simple_AzureAISearch.cs). |
+| [VectorStore_HybridSearch_Simple_Chroma](./samples/Concepts/Memory/VectorStore_HybridSearch_Simple_Chroma.cs) | Searches a glossary on Chroma Cloud with a vector and keywords, with and without a filter. On Chroma Cloud, the collection gets a BM25 index for the full-text indexed property when it is created. Adapted from [VectorStore_HybridSearch_Simple_AzureAISearch](https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/Concepts/Memory/VectorStore_HybridSearch_Simple_AzureAISearch.cs). |
 | [ChatCompletion_Rag_Chroma](./samples/Concepts/Agents/ChatCompletion_Rag_Chroma.cs) | A `ChatCompletionAgent` that answers through a `TextSearchProvider`, from documents that a `TextSearchStore` keeps in Chroma. It also covers citations and a search namespace. Adapted from [ChatCompletion_Rag](https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/Concepts/Agents/ChatCompletion_Rag.cs). |
 | [ChatCompletion_Rag_Chroma_Ollama](./samples/Concepts/Agents/ChatCompletion_Rag_Chroma_Ollama.cs) | The same sample with models that run locally in Ollama. |
 | [Step5_Search_With_Chroma](./samples/GettingStartedWithTextSearch/Step5_Search_With_Chroma.cs) | Searches records that have their own data model in Chroma, through `VectorStoreTextSearch`. It then either passes the results to the model in a Handlebars prompt, or gives the model the search as a function to call. Adapted from [Step4_Search_With_VectorStore](https://github.com/microsoft/semantic-kernel/blob/main/dotnet/samples/GettingStartedWithTextSearch/Step4_Search_With_VectorStore.cs). |

samples/Concepts/Agents/ChatCompletion_Rag_Chroma.cs

@@ -51,6 +51,7 @@ public async Task UseChatCompletionAgentWithBasicRag()
         using var httpClient = new HttpClient();
         using var vectorStore = new ChromaVectorStore(
             new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient),
+            ownsClient: true,
             new() { EmbeddingGenerator = embeddingGenerator });
 
         // Create a store that uses a built in schema for storing text documents
@@ -114,6 +115,7 @@ public async Task RagWithCitationsAndFiltering()
         using var httpClient = new HttpClient();
         using var vectorStore = new ChromaVectorStore(
             new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient),
+            ownsClient: true,
             new() { EmbeddingGenerator = embeddingGenerator });
 
         // Create a store that uses a built in schema for storing text documents

samples/Concepts/Agents/ChatCompletion_Rag_Chroma_Ollama.cs

@@ -48,6 +48,7 @@ public async Task UseChatCompletionAgentWithBasicRag()
         using var httpClient = new HttpClient();
         using var vectorStore = new ChromaVectorStore(
             new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient),
+            ownsClient: true,
             new() { EmbeddingGenerator = embeddingGenerator });
 
         // Create a store that uses a built in schema for storing text documents
@@ -113,6 +114,7 @@ public async Task RagWithCitationsAndFiltering()
         using var httpClient = new HttpClient();
         using var vectorStore = new ChromaVectorStore(
             new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient),
+            ownsClient: true,
             new() { EmbeddingGenerator = embeddingGenerator });
 
         // Create a store that uses a built in schema for storing text documents

samples/Concepts/Concepts.csproj

@@ -16,7 +16,7 @@
 
   <ItemGroup>
     <PackageReference Include="Azure.Identity" Version="1.21.0" />
-    <PackageReference Include="ChromaDotNet.VectorData" Version="0.3.7" />
+    <PackageReference Include="ChromaDotNet.VectorData" Version="0.4.0" />
     <PackageReference Include="Docker.DotNet" Version="3.125.15" />
     <PackageReference Include="Microsoft.Extensions.Configuration" Version="10.0.2" />
     <PackageReference Include="Microsoft.Extensions.Configuration.Binder" Version="10.0.2" />

samples/Concepts/Memory/VectorStore_DynamicDataModel_Interop_Chroma.cs

@@ -49,7 +49,7 @@ public async Task UpsertWithDynamicRetrieveWithCustomAsync()
         // Initiate the docker container and construct the vector store.
         await chromaFixture.ManualInitializeAsync();
         using var httpClient = new HttpClient();
-        using var vectorStore = new ChromaVectorStore(new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient));
+        using var vectorStore = new ChromaVectorStore(new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient), ownsClient: true);
 
         // Get and create collection if it doesn't exist using the dynamic data model and record definition that defines the schema.
         var dynamicDataModelCollection = vectorStore.GetDynamicCollection("skglossary", s_definition);
@@ -87,7 +87,7 @@ public async Task UpsertWithCustomRetrieveWithDynamicAsync()
         // Initiate the docker container and construct the vector store.
         await chromaFixture.ManualInitializeAsync();
         using var httpClient = new HttpClient();
-        using var vectorStore = new ChromaVectorStore(new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient));
+        using var vectorStore = new ChromaVectorStore(new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient), ownsClient: true);
 
         // Get and create collection if it doesn't exist using the custom data model.
         var customDataModelCollection = vectorStore.GetCollection<string, Glossary>("skglossary");

samples/Concepts/Memory/VectorStore_HybridSearch_Simple_Chroma.cs

@@ -14,7 +14,7 @@ namespace Memory;
 ///
 /// The example shows the following steps:
 /// 1. Create an embedding generator.
-/// 2. Create a Chroma Vector Store that creates a BM25 index for the full-text indexed properties of its collections.
+/// 2. Create a Chroma Vector Store: on Chroma Cloud, its collections get a BM25 index for each full-text indexed property.
 /// 3. Ingest some data into the vector store.
 /// 4. Do a hybrid search on the vector store with various text+keyword and filtering options.
 ///
@@ -30,7 +30,7 @@ public async Task IngestDataAndUseHybridSearch()
             .GetEmbeddingClient(TestConfiguration.AzureOpenAIEmbeddings.DeploymentName)
             .AsIEmbeddingGenerator(1536);
 
-        // Construct the Chroma VectorStore, which creates the collections with a BM25 index for each full-text indexed property.
+        // Construct the Chroma VectorStore. On Chroma Cloud, creating a collection creates a BM25 index for each full-text indexed property, which the hybrid search uses.
         using var httpClient = new HttpClient();
         var chromaClient = new ChromaClient(
             new ChromaConfigurationOptions(
@@ -39,7 +39,7 @@ public async Task IngestDataAndUseHybridSearch()
                 database: TestConfiguration.Chroma.Database,
                 chromaToken: TestConfiguration.Chroma.ApiKey),
             httpClient);
-        using var vectorStore = new ChromaVectorStore(chromaClient, new() { CreateBm25Indexes = true });
+        using var vectorStore = new ChromaVectorStore(chromaClient, ownsClient: true);
 
         // Get and create collection if it doesn't exist.
         var collection = vectorStore.GetCollection<string, Glossary>("skglossary");

samples/Concepts/Memory/VectorStore_VectorSearch_MultiStore_Chroma.cs

@@ -67,7 +67,7 @@ public async Task ExampleWithoutDIAsync()
         await chromaFixture.ManualInitializeAsync();
         using var httpClient = new HttpClient();
         var chromaClient = new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient);
-        using var vectorStore = new ChromaVectorStore(chromaClient);
+        using 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);

samples/Concepts/Memory/VectorStore_VectorSearch_Paging_Chroma.cs

@@ -25,7 +25,7 @@ public async Task VectorSearchWithPagingAsync()
         // Initiate the docker container and construct the Chroma vector store.
         await chromaFixture.ManualInitializeAsync();
         using var httpClient = new HttpClient();
-        using var vectorStore = new ChromaVectorStore(new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient));
+        using var vectorStore = new ChromaVectorStore(new ChromaClient(new ChromaConfigurationOptions("http://localhost:8000"), httpClient), ownsClient: true);
 
         // Get and create collection if it doesn't exist.
         // Chroma supports string and Guid keys.

samples/GettingStartedWithTextSearch/ChromaVectorStoreFixtureBase.cs

@@ -64,6 +64,7 @@ public async Task InitializeAsync()
         await this._chromaContainer.StartAsync();
         this.ChromaVectorStore = new ChromaVectorStore(
             new ChromaClient(new ChromaConfigurationOptions(this._chromaContainer.GetConnectionString()), this._httpClient),
+            ownsClient: true,
             new() { EmbeddingGenerator = this.EmbeddingGenerator });
 
         this.VectorStoreRecordCollection = await this.InitializeRecordCollectionAsync();

samples/GettingStartedWithTextSearch/GettingStartedWithTextSearch.csproj

@@ -17,8 +17,8 @@
 
   <ItemGroup>
     <PackageReference Include="Azure.Identity" Version="1.21.0" />
-    <PackageReference Include="ChromaDotNet.Testcontainers" Version="0.1.3" />
-    <PackageReference Include="ChromaDotNet.VectorData" Version="0.3.7" />
+    <PackageReference Include="ChromaDotNet.Testcontainers" Version="0.1.6" />
+    <PackageReference Include="ChromaDotNet.VectorData" Version="0.4.0" />
     <PackageReference Include="Microsoft.Extensions.Configuration" Version="10.0.2" />
     <PackageReference Include="Microsoft.Extensions.Configuration.Binder" Version="10.0.2" />
     <PackageReference Include="Microsoft.Extensions.Configuration.EnvironmentVariables" Version="10.0.2" />

test/ChromaDB.SemanticKernel.Sample.Tests/ChatCompletionRagTests.cs

@@ -57,7 +57,7 @@ public async Task Only_the_documents_of_the_search_namespace_reach_the_model_wit
     }
 
     private ChromaVectorStore CreateVectorStore(HttpClient httpClient)
-        => new(new ChromaClient(new ChromaConfigurationOptions(fixture.Endpoint), httpClient), new() { EmbeddingGenerator = new WordEmbeddingGenerator(EmbeddingDimensions) });
+        => new(new ChromaClient(new ChromaConfigurationOptions(fixture.Endpoint), httpClient), ownsClient: true, new() { EmbeddingGenerator = new WordEmbeddingGenerator(EmbeddingDimensions) });
 
     private static (ChatCompletionAgent Agent, RecordingChatClient Model) CreateAgent()
     {

test/ChromaDB.SemanticKernel.Sample.Tests/ChromaDB.SemanticKernel.Sample.Tests.csproj

@@ -9,7 +9,7 @@
   </PropertyGroup>
 
   <ItemGroup>
-    <PackageReference Include="ChromaDotNet.Testcontainers" Version="0.1.3" />
+    <PackageReference Include="ChromaDotNet.Testcontainers" Version="0.1.6" />
     <PackageReference Include="Microsoft.NET.Test.Sdk" Version="18.7.0" />
     <PackageReference Include="xunit" Version="2.9.3" />
     <PackageReference Include="xunit.abstractions" Version="2.0.3" />

test/ChromaDB.SemanticKernel.Sample.Tests/DataModelAndPagingTests.cs

@@ -105,7 +105,7 @@ public async Task The_pages_go_through_every_record_once_in_the_order_of_the_dis
     }
 
     private ChromaVectorStore CreateVectorStore(HttpClient httpClient)
-        => new(new ChromaClient(new ChromaConfigurationOptions(fixture.Endpoint), httpClient));
+        => new(new ChromaClient(new ChromaConfigurationOptions(fixture.Endpoint), httpClient), ownsClient: true);
 
     private sealed class Glossary
     {

test/ChromaDB.SemanticKernel.Sample.Tests/MultiStoreTests.cs

@@ -37,7 +37,7 @@ public async Task Without_DI_each_search_finds_its_glossary_entry()
     {
         var output = new RecordingOutput(testOutput);
         using var httpClient = new HttpClient();
-        using var vectorStore = new ChromaVectorStore(new ChromaClient(new ChromaConfigurationOptions(fixture.Endpoint), httpClient));
+        using var vectorStore = new ChromaVectorStore(new ChromaClient(new ChromaConfigurationOptions(fixture.Endpoint), httpClient), ownsClient: true);
 
         var processor = new VectorStore_VectorSearch_MultiStore_Common(vectorStore, new WordEmbeddingGenerator(EmbeddingDimensions), output);
         await processor.IngestDataAndSearchAsync("skglossaryWithoutDI", () => Guid.NewGuid());