Merge pull request #632 from betalgo/629-how-to-use-chunking-strategy…

…-in-vector-store

chunking strategy in vector store

OpenAI.Playground/TestHelpers/AssistantHelpers/VectorTestHelper.cs

@@ -73,6 +73,34 @@ public static async Task CreateVector(IOpenAIService openAI)
             }
         }
 
+        public static async Task CreateVectorWithChunkingStrategy(IOpenAIService openAI)
+        {
+            ConsoleExtensions.WriteLine("Create Vector Testing is starting:", ConsoleColor.Cyan);
+            var result = await openAI.Beta.VectorStores.CreateVectorStore(new()
+            {
+                Name = "Support FAQ",
+                ChunkingStrategy = new()
+                {
+                    Type = StaticValues.VectorStoreStatics.ChunkingStrategyType.Static,
+                    StaticParameters = new()
+                    {
+                        ChunkOverlapTokens = 400,
+                        MaxChunkSizeTokens = 800
+                    }
+                }
+            });
+
+            if (result.Successful)
+            {
+                CreatedVectorId = result.Id;
+                ConsoleExtensions.WriteLine($"Vector Created Successfully with ID: {result.Id}", ConsoleColor.Green);
+            }
+            else
+            {
+                ConsoleExtensions.WriteError(result.Error);
+            }
+        }
+
         public static async Task ListVectors(IOpenAIService openAI)
         {
             ConsoleExtensions.WriteLine("List Vectors Testing is starting:", ConsoleColor.Cyan);

OpenAI.SDK/ObjectModels/RequestModels/CreateVectorStoreFileBatchRequest.cs

@@ -10,4 +10,11 @@ public class CreateVectorStoreFileBatchRequest
     /// </summary>
     [JsonPropertyName("file_ids")]
     public List<string> FileIds { get; set; }
+
+    /// <summary>
+    ///     The chunking strategy used to chunk the file(s). If not set, will use the auto strategy.
+    ///     Only applicable if file_ids is non-empty.
+    /// </summary>
+    [JsonPropertyName("chunking_strategy")]
+    public ChunkingStrategy? ChunkingStrategy { get; set; }
 }

OpenAI.SDK/ObjectModels/RequestModels/CreateVectorStoreRequest.cs

@@ -31,4 +31,43 @@ public class CreateVectorStoreRequest
     /// </summary>
     [JsonPropertyName("metadata")]
     public Dictionary<string, string>? Metadata { get; set; }
+
+    /// <summary>
+    ///     The chunking strategy used to chunk the file(s). If not set, will use the auto strategy.
+    ///     Only applicable if file_ids is non-empty.
+    /// </summary>
+    [JsonPropertyName("chunking_strategy")]
+    public ChunkingStrategy? ChunkingStrategy { get; set; }
+}
+
+public class ChunkingStrategy
+{
+    /// <summary>
+    ///     The type of chunking strategy. Must be either "auto" or "static".
+    /// </summary>
+    [JsonPropertyName("type")]
+    public string Type { get; set; }
+
+    /// <summary>
+    ///     The static chunking parameters. Required if type is "static".
+    /// </summary>
+    [JsonPropertyName("static")]
+    public StaticChunkingParameters? StaticParameters { get; set; }
+}
+
+public class StaticChunkingParameters
+{
+    /// <summary>
+    ///     The maximum number of tokens in each chunk. The default value is 800.
+    ///     The minimum value is 100 and the maximum value is 4096.
+    /// </summary>
+    [JsonPropertyName("max_chunk_size_tokens")]
+    public int MaxChunkSizeTokens { get; set; }
+
+    /// <summary>
+    ///     The number of tokens that overlap between chunks. The default value is 400.
+    ///     Note that the overlap must not exceed half of max_chunk_size_tokens.
+    /// </summary>
+    [JsonPropertyName("chunk_overlap_tokens")]
+    public int ChunkOverlapTokens { get; set; }
 }

OpenAI.SDK/ObjectModels/RequestModels/ToolDefinition.cs

@@ -15,12 +15,18 @@ public class ToolDefinition
     public string Type { get; set; }
 
     /// <summary>
-    /// Structured Outputs for function calling can be enabled with a single parameter, just by supplying strict: true.
-    /// Please note: This field is not mentioned in the API documentation but is referenced in other documents.
+    ///     Structured Outputs for function calling can be enabled with a single parameter, just by supplying strict: true.
+    ///     Please note: This field is not mentioned in the API documentation but is referenced in other documents.
     /// </summary>
     [JsonPropertyName("strict")]
     public bool? Strict { get; set; }
 
+    /// <summary>
+    ///     Overrides for the file search tool.
+    /// </summary>
+    [JsonPropertyName("file_search")]
+    public FileSearchTool? FileSearchTool { get; set; }
+
     /// <summary>
     ///     A list of functions the model may generate JSON inputs for.
     /// </summary>
@@ -73,11 +79,56 @@ public static ToolDefinition DefineRetrieval()
         };
     }
 
-    public static ToolDefinition DefineFileSearch()
+    public static ToolDefinition DefineFileSearch(FileSearchTool? fileSearchTool = null)
     {
         return new()
         {
-            Type = StaticValues.AssistantsStatics.ToolCallTypes.FileSearch
+            Type = StaticValues.AssistantsStatics.ToolCallTypes.FileSearch,
+            FileSearchTool = fileSearchTool
         };
     }
+}
+
+public class FileSearchTool
+{
+    /// <summary>
+    ///     The maximum number of results the file search tool should output. The default is 20 for gpt-4* models and 5 for
+    ///     gpt-3.5-turbo. This number should be between 1 and 50 inclusive.
+    ///     Note that the file search tool may output fewer than max_num_results results.
+    ///     <a href="https://platform.openai.com/docs/assistants/tools/file-search/customizing-file-search-settings">
+    ///         See the
+    ///         file search tool documentation
+    ///     </a>
+    ///     for more information.
+    /// </summary>
+    [JsonPropertyName("max_num_results")]
+    public int? MaxNumberResults { get; set; }
+
+    /// <summary>
+    ///     The ranking options for the file search. If not specified, the file search tool will use the auto ranker and a
+    ///     score_threshold of 0.
+    ///     See the
+    ///     <a href="https://platform.openai.com/docs/assistants/tools/file-search/customizing-file-search-settings">
+    ///         file
+    ///         search tool documentation
+    ///     </a>
+    ///     for more information.
+    /// </summary>
+    [JsonPropertyName("ranking_options")]
+    public RankingOptions? RankingOptions { get; set; }
+}
+
+public class RankingOptions
+{
+    /// <summary>
+    ///     The ranker to use for the file search. If not specified will use the auto ranker.
+    /// </summary>
+    [JsonPropertyName("ranker")]
+    public string? Ranker { get; set; }
+
+    /// <summary>
+    ///     The score threshold for the file search. All values must be a floating point number between 0 and 1.
+    /// </summary>
+    [JsonPropertyName("score_threshold")]
+    public int ScoreThreshold { get; set; }
 }

OpenAI.SDK/ObjectModels/ResponseModels/VectorStoreResponseModels/VectorStoreFileObject.cs

@@ -1,4 +1,5 @@
 using System.Text.Json.Serialization;
+using OpenAI.ObjectModels.RequestModels;
 
 namespace OpenAI.ObjectModels.ResponseModels.VectorStoreResponseModels;
 
@@ -46,4 +47,10 @@ public record VectorStoreFileObject : BaseResponse
     /// </summary>
     [JsonPropertyName("last_error")]
     public Error? LastError { get; set; }
+
+    /// <summary>
+    ///     The strategy used to chunk the file.
+    /// </summary>
+    [JsonPropertyName("chunking_strategy")]
+    public ChunkingStrategy? ChunkingStrategy { get; set; }
 }

OpenAI.SDK/ObjectModels/ResponseModels/VectorStoreResponseModels/VectorStoreObjectResponse.cs

@@ -1,4 +1,5 @@
 using System.Text.Json.Serialization;
+using OpenAI.ObjectModels.RequestModels;
 
 namespace OpenAI.ObjectModels.ResponseModels.VectorStoreResponseModels;
 
@@ -63,4 +64,11 @@ public record VectorStoreObjectResponse : BaseResponse
     /// </summary>
     [JsonPropertyName("metadata")]
     public Dictionary<string, string>? Metadata { get; set; }
+
+    /// <summary>
+    ///     The chunking strategy used to chunk the file(s). If not set, will use the auto strategy.
+    ///     Only applicable if file_ids is non-empty.
+    /// </summary>
+    [JsonPropertyName("chunking_strategy")]
+    public ChunkingStrategy? ChunkingStrategy { get; set; }
 }

OpenAI.SDK/ObjectModels/StaticValueHelper.cs

@@ -174,4 +174,13 @@ public static class RequiredActionTypes
             public static string SubmitToolOutputs => "submit_tool_outputs";
         }
     }
+
+    public static class VectorStoreStatics
+    {
+        public static class ChunkingStrategyType
+        {
+            public static string Auto => "auto";
+            public static string Static => "static";
+        }
+    }
 }