Merge pull request #38 from OkGoDoIt/metjuperry/master

Merging of @metjuperry: Added embedding endpoint and tests

OpenAI_API/Embedding/EmbeddingEndpoint.cs

@@ -0,0 +1,59 @@
+using System.Threading.Tasks;
+
+namespace OpenAI_API.Embedding
+{
+	/// <summary>
+	/// OpenAI’s text embeddings measure the relatedness of text strings by generating an embedding, which is a vector (list) of floating point numbers. The distance between two vectors measures their relatedness. Small distances suggest high relatedness and large distances suggest low relatedness.
+	/// </summary>
+	public class EmbeddingEndpoint : EndpointBase
+	{
+		/// <summary>
+		/// This allows you to send request to the recommended model without needing to specify. Every request uses the <see cref="Model.AdaTextEmbedding"/> model
+		/// </summary>
+		public EmbeddingRequest DefaultEmbeddingRequestArgs { get; set; } = new EmbeddingRequest() { Model = Model.AdaTextEmbedding };
+
+		/// <summary>
+		/// The name of the endpoint, which is the final path segment in the API URL.  For example, "embeddings".
+		/// </summary>
+		protected override string Endpoint { get { return "embeddings"; } }
+
+		/// <summary>
+		/// Constructor of the api endpoint.  Rather than instantiating this yourself, access it through an instance of <see cref="OpenAIAPI"/> as <see cref="OpenAIAPI.Embeddings"/>.
+		/// </summary>
+		/// <param name="api"></param>
+		internal EmbeddingEndpoint(OpenAIAPI api) : base(api) { }
+
+		/// <summary>
+		/// Ask the API to embedd text using the default embedding model <see cref="Model.AdaTextEmbedding"/>
+		/// </summary>
+		/// <param name="input">Text to be embedded</param>
+		/// <returns>Asynchronously returns the embedding result. Look in its <see cref="Data.Embedding"/> property of <see cref="EmbeddingResult.Data"/> to find the vector of floating point numbers</returns>
+		public async Task<EmbeddingResult> CreateEmbeddingAsync(string input)
+		{
+			EmbeddingRequest req = new EmbeddingRequest(DefaultEmbeddingRequestArgs.Model, input);
+			return await CreateEmbeddingAsync(req);
+		}
+
+		/// <summary>
+		/// Ask the API to embedd text using a custom request
+		/// </summary>
+		/// <param name="request">Request to be send</param>
+		/// <returns>Asynchronously returns the embedding result. Look in its <see cref="Data.Embedding"/> property of <see cref="EmbeddingResult.Data"/> to find the vector of floating point numbers</returns>
+		public async Task<EmbeddingResult> CreateEmbeddingAsync(EmbeddingRequest request)
+		{
+			return await HttpPost<EmbeddingResult>(postData: request);
+		}
+
+		/// <summary>
+		/// Ask the API to embedd text using the default embedding model <see cref="Model.AdaTextEmbedding"/>
+		/// </summary>
+		/// <param name="input">Text to be embedded</param>
+		/// <returns>Asynchronously returns the first embedding result as an array of floats.</returns>
+		public async Task<float[]> GetEmbeddingsAsync(string input)
+		{
+			EmbeddingRequest req = new EmbeddingRequest(DefaultEmbeddingRequestArgs.Model, input);
+			var embeddingResult = await CreateEmbeddingAsync(req);
+			return embeddingResult?.Data?[0]?.Embedding;
+		}
+	}
+}

OpenAI_API/Embedding/EmbeddingRequest.cs

@@ -0,0 +1,51 @@
+using Newtonsoft.Json;
+
+namespace OpenAI_API.Embedding
+{
+	/// <summary>
+	/// Represents a request to the Completions API. Matches with the docs at <see href="https://platform.openai.com/docs/api-reference/embeddings">the OpenAI docs</see>
+	/// </summary>
+	public class EmbeddingRequest
+	{
+		/// <summary>
+		/// ID of the model to use. You can use <see cref="ModelsEndpoint.GetModelsAsync()"/> to see all of your available models, or use a standard model like <see cref="Model.AdaTextEmbedding"/>.
+		/// </summary>
+		[JsonProperty("model")]
+		public string Model { get; set; }
+
+		/// <summary>
+		/// Main text to be embedded
+		/// </summary>
+		[JsonProperty("input")]
+		public string Input { get; set; }
+
+		/// <summary>
+		/// Cretes a new, empty <see cref="EmbeddingRequest"/>
+		/// </summary>
+		public EmbeddingRequest()
+		{
+
+		}
+
+		/// <summary>
+		/// Creates a new <see cref="EmbeddingRequest"/> with the specified parameters
+		/// </summary>
+		/// <param name="model">The model to use. You can use <see cref="ModelsEndpoint.GetModelsAsync()"/> to see all of your available models, or use a standard model like <see cref="Model.AdaTextEmbedding"/>.</param>
+		/// <param name="input">The prompt to transform</param>
+		public EmbeddingRequest(Model model, string input)
+		{
+			Model = model;
+			this.Input = input;
+		}
+
+		/// <summary>
+		/// Creates a new <see cref="EmbeddingRequest"/> with the specified input and the <see cref="Model.AdaTextEmbedding"/> model.
+		/// </summary>
+		/// <param name="input">The prompt to transform</param>
+		public EmbeddingRequest(string input)
+		{
+			Model = OpenAI_API.Model.AdaTextEmbedding;
+			this.Input = input;
+		}
+	}
+}

OpenAI_API/Embedding/EmbeddingResult.cs

@@ -0,0 +1,79 @@
+using Newtonsoft.Json;
+using System.Collections.Generic;
+using System.Linq;
+
+namespace OpenAI_API.Embedding
+{
+	/// <summary>
+	/// Represents an embedding result returned by the Embedding API.  
+	/// </summary>
+	public class EmbeddingResult : ApiResultBase
+	{
+		/// <summary>
+		/// List of results of the embedding
+		/// </summary>
+		[JsonProperty("data")]
+		public List<Data> Data { get; set; }
+
+		/// <summary>
+		/// Usage statistics of how many tokens have been used for this request
+		/// </summary>
+		[JsonProperty("usage")]
+		public Usage Usage { get; set; }
+
+		/// <summary>
+		/// Allows an EmbeddingResult to be implicitly cast to the array of floats repsresenting the first ebmedding result
+		/// </summary>
+		/// <param name="embeddingResult">The <see cref="EmbeddingResult"/> to cast to an array of floats.</param>
+		public static implicit operator float[](EmbeddingResult embeddingResult)
+		{
+			return embeddingResult.Data.FirstOrDefault()?.Embedding;
+		}
+	}
+
+	/// <summary>
+	/// Data returned from the Embedding API.
+	/// </summary>
+	public class Data
+	{
+		/// <summary>
+		/// Type of the response. In case of Data, this will be "embedding"  
+		/// </summary>
+		[JsonProperty("object")]
+
+		public string Object { get; set; }
+
+		/// <summary>
+		/// The input text represented as a vector (list) of floating point numbers
+		/// </summary>
+		[JsonProperty("embedding")]
+		public float[] Embedding { get; set; }
+
+		/// <summary>
+		/// Index
+		/// </summary>
+		[JsonProperty("index")]
+		public int Index { get; set; }
+
+	}
+
+	/// <summary>
+	/// Usage statistics of how many tokens have been used for this request.
+	/// </summary>
+	public class Usage
+	{
+		/// <summary>
+		/// How many tokens did the prompt consist of
+		/// </summary>
+		[JsonProperty("prompt_tokens")]
+		public int PromptTokens { get; set; }
+
+		/// <summary>
+		/// How many tokens did the request consume total
+		/// </summary>
+		[JsonProperty("total_tokens")]
+		public int TotalTokens { get; set; }
+
+	}
+
+}

OpenAI_API/OpenAIAPI.cs

@@ -1,4 +1,6 @@
 using OpenAI_API.Files;
+using Newtonsoft.Json;
+using OpenAI_API.Embedding;
 using System;
 
 namespace OpenAI_API
@@ -12,7 +14,7 @@ public class OpenAIAPI
 		/// Base url for OpenAI
 		/// </summary>
 		public string ApiUrlBase = "https://api.openai.com/v1/";
-
+		
 		/// <summary>
 		/// The API authentication information to use for API calls
 		/// </summary>
@@ -28,14 +30,19 @@ public OpenAIAPI(APIAuthentication apiKeys = null)
 			Completions = new CompletionEndpoint(this);
 			Models = new ModelsEndpoint(this);
 			//Search = new SearchEndpoint(this);
-			Files = new FilesEndpoint(this);
+			Embeddings = new EmbeddingEndpoint(this);
 		}
 
 		/// <summary>
 		/// Text generation is the core function of the API. You give the API a prompt, and it generates a completion. The way you “program” the API to do a task is by simply describing the task in plain english or providing a few written examples. This simple approach works for a wide range of use cases, including summarization, translation, grammar correction, question answering, chatbots, composing emails, and much more (see the prompt library for inspiration).
 		/// </summary>
 		public CompletionEndpoint Completions { get; }
 
+		/// <summary>
+		/// The API lets you transform text into a vector (list) of floating point numbers. The distance between two vectors measures their relatedness. Small distances suggest high relatedness and large distances suggest low relatedness.
+		/// </summary>
+		public EmbeddingEndpoint Embeddings { get; }
+
 		/// <summary>
 		/// The API endpoint for querying available Engines/models
 		/// </summary>

OpenAI_Tests/EmbeddingEndpointTests.cs

@@ -0,0 +1,43 @@
+using NUnit.Framework;
+using OpenAI_API;
+using OpenAI_API.Embedding;
+using System;
+using System.Linq;
+
+namespace OpenAI_Tests
+{
+    public class EmbeddingEndpointTests
+    {
+        [SetUp]
+        public void Setup()
+        {
+            OpenAI_API.APIAuthentication.Default = new OpenAI_API.APIAuthentication(Environment.GetEnvironmentVariable("TEST_OPENAI_SECRET_KEY"));
+        }
+
+        [Test]
+        public void GetBasicEmbedding()
+        {
+            var api = new OpenAI_API.OpenAIAPI();
+
+            Assert.IsNotNull(api.Embeddings);
+
+            var results = api.Embeddings.CreateEmbeddingAsync(new EmbeddingRequest(Model.AdaTextEmbedding, "A test text for embedding")).Result;
+            Assert.IsNotNull(results);
+            Assert.NotNull(results.Object);
+            Assert.NotZero(results.Data.Count);
+            Assert.That(results.Data.First().Embedding.Length == 1536);
+        }
+
+        [Test]
+        public void GetSimpleEmbedding()
+        {
+            var api = new OpenAI_API.OpenAIAPI();
+
+            Assert.IsNotNull(api.Embeddings);
+
+            var results = api.Embeddings.GetEmbeddingsAsync("A test text for embedding").Result;
+            Assert.IsNotNull(results);
+            Assert.That(results.Length == 1536);
+        }
+    }
+}