Version 1.2.8736.39976 with DataExtractorTool enhancements

nuget/GenAI.NET.nuspec

@@ -2,7 +2,7 @@
 <package xmlns="http://schemas.microsoft.com/packaging/2012/06/nuspec.xsd">
     <metadata>
         <id>GenAI.NET</id>
-        <version>1.2.8734.98</version>
+        <version>1.2.8736.39976</version>
         <authors>Automation Agent</authors>
         <owners>Automation Agent</owners>
         <projectUrl>https://github.com/automaze1/GenAI.NET</projectUrl>

src/GenAIFramework.Test/ToolsTest.cs

@@ -474,5 +474,47 @@ public async Task SemanticSearchWithContext()
             Assert.AreEqual(3, searchresults.Length);
             Assert.IsTrue(searchresults.Any(r => r.content.Contains("three quarter")));
         }
+
+        [TestMethod]
+        public async Task DataExtractorPipeline()
+        {
+            Logger.WriteLog(LogLevel.Info, LogOps.Test, "DataExtractorPipeline");
+
+            var textExtractor = TextExtractorTool.Create();
+            Assert.IsNotNull(textExtractor);
+
+            var parameters = new Dictionary<string, string>()
+            {
+                { "Revenue Growth", "What is the revenue growth in dollars in Q1 2023?" },
+                { "Revenue Growth Percent", "What is the revenue growth in percentage in Q1 2023?" },
+                { "Revenue Forecast", "What is the revenue forecast for Q2 2023?" },
+                { "Net Income", "What is the net income in dollars in Q1 2023?" },
+                { "Stock Price Change", "What is the change is stock price after the quarterly results of Q1?" },
+                { "Market Sentiment", "Based on market response, what is the market sentiment (positive or negative) after the results?" },
+                { "Challenges", "What are the key challenges mentioned in this report?" },
+                { "Initiatives", "What is the key initiatives to mitigate challenges are mentioned in this report?" },
+            };
+
+            //var llm = new OpenAILanguageModel("gpt-3.5-turbo");
+            //var prametersJson = FunctionTool.ToJsonString(parameters);
+            var dataExtractor = DataExtractorTool.Create()
+                                                 .WithParameters(parameters);
+                                                 //.WithLanguageModel(llm);
+
+            Assert.IsNotNull(dataExtractor);
+
+            var pipeline = Pipeline.WithTools(new IFunctionTool[]  {textExtractor, dataExtractor});
+            Assert.IsNotNull(pipeline);
+
+            var file = Path.Combine(RootPath, @"..\..\..\..\..\tests\input\article.txt");
+            var context = new ExecutionContext();
+            context["input"] = file;
+
+            var result = await pipeline.ExecuteAsync(context);
+            Assert.IsTrue(!string.IsNullOrEmpty(result));
+
+            var results = FunctionTool.Deserialize<Dictionary<string, string>>(result);
+            Assert.AreEqual(8, results.Count);
+        }
     }
 }

src/GenerativeAI/Configuration.cs

@@ -5,13 +5,36 @@
 
 namespace Automation.GenerativeAI
 {
+    /// <summary>
+    /// Implements the configuration for OpenAI/AzureOpenAI
+    /// </summary>
     internal class OpenAIConfig
     {
+        /// <summary>
+        /// Endpoint URL which contains resource name for azure
+        /// </summary>
         public string EndPointUrl { get; set; }
+
+        /// <summary>
+        /// Deployment ID for GPT Model in Azure
+        /// </summary>
         public string GPTDeployment { get; set; }
+
+        /// <summary>
+        /// Deployment ID for Text Embedding model in Azure
+        /// </summary>
         public string EmbeddingDeployment { get; set; }
+
+        /// <summary>
+        /// ApiVersion, applicable to Azure
+        /// </summary>
         public string ApiVersion { get; set; }
-        private string apiKey;
+
+        private string apiKey; //stores api key
+
+        /// <summary>
+        /// API Key
+        /// </summary>
         public string ApiKey 
         {
             get
@@ -34,7 +57,20 @@ public string ApiKey
                 apiKey = value;
             } 
         }
-        public string Model { get; set; }
+
+        /// <summary>
+        /// GPT Model Name
+        /// </summary>
+        public string Model { get; set; } = "gpt-3.5-turbo"; //default model for OpenAI
+
+        /// <summary>
+        /// Provides token limit for the given model
+        /// </summary>
+        public int TokenLimit { get; set; } = 4000; //Default is 4K model
+
+        /// <summary>
+        /// Gets full URL for the Text Embedding API
+        /// </summary>
         public string EmbeddingUrl 
         { 
             get 
@@ -49,6 +85,9 @@ public string EmbeddingUrl
             } 
         }
 
+        /// <summary>
+        /// Gets full URL for the chat completion API
+        /// </summary>
         public string CompletionsUrl
         {
             get
@@ -63,6 +102,9 @@ public string CompletionsUrl
             }
         }
 
+        /// <summary>
+        /// Checks if it is Azure config
+        /// </summary>
         public bool AzureConfig
         {
             get
@@ -76,11 +118,21 @@ public bool AzureConfig
         }
     }
 
+    /// <summary>
+    /// Implements config for Bing Search API
+    /// </summary>
     internal class BingAPIConfig
     {
+        /// <summary>
+        /// Bing API endpoint url
+        /// </summary>
         public string EndPointUrl { get; set; } = "https://api.bing.microsoft.com/v7.0/";
 
-        private string apiKey;
+        private string apiKey; //stores api key
+
+        /// <summary>
+        /// Api Key
+        /// </summary>
         public string ApiKey
         {
             get
@@ -98,11 +150,20 @@ public string ApiKey
         }
     }
 
+    /// <summary>
+    /// Implements a configuration
+    /// </summary>
     internal class Configuration
     {
+        /// <summary>
+        /// Default constructor
+        /// </summary>
         public Configuration() { }
         private static Configuration instance = null;
 
+        /// <summary>
+        /// Global instance of the configuration
+        /// </summary>
         public static Configuration Instance
         {
             get
@@ -117,12 +178,25 @@ public static Configuration Instance
             }
         }
 
+        /// <summary>
+        /// Configuration for OpenAI or AzureOpenAI APIs
+        /// </summary>
         public OpenAIConfig OpenAIConfig { get; set; } = new OpenAIConfig();
 
+        /// <summary>
+        /// Configuration for Bing API
+        /// </summary>
         public BingAPIConfig BingAPIConfig { get; set; } = new BingAPIConfig();
 
+        /// <summary>
+        /// Log file path
+        /// </summary>
         public string LogFile { get; set; }
 
+        /// <summary>
+        /// Gets executing DLL Path
+        /// </summary>
+        /// <returns></returns>
         internal static string GetDLLPath()
         {
             var asm = Assembly.GetExecutingAssembly();
@@ -133,6 +207,10 @@ internal static string GetDLLPath()
             return path;
         }
 
+        /// <summary>
+        /// Saves the config as json to a given file
+        /// </summary>
+        /// <param name="filePath">Full file path to save the config as json</param>
         public void Save(string filePath)
         {
             var jsonfile = Path.ChangeExtension(filePath, "json");
@@ -141,6 +219,11 @@ public void Save(string filePath)
             File.WriteAllText(jsonfile, jsontxt);
         }
 
+        /// <summary>
+        /// Loads the configuration from the given json file
+        /// </summary>
+        /// <param name="filePath">Full path of config json file</param>
+        /// <returns>Configuration</returns>
         public static Configuration Load(string filePath)
         {
             if(!File.Exists(filePath)) return new Configuration();

src/GenerativeAI/GenerativeAI.csproj

@@ -110,7 +110,11 @@
     <None Include="ClassDiagram.cd" />
     <None Include="packages.config" />
   </ItemGroup>
-  <ItemGroup />
+  <ItemGroup>
+    <EmbeddedResource Include="Prompts\DataExtractorPrompt.txt">
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+    </EmbeddedResource>
+  </ItemGroup>
   <ItemGroup>
     <EmbeddedResource Include="Prompts\MRKLPrompt.txt">
       <CopyToOutputDirectory>Always</CopyToOutputDirectory>

src/GenerativeAI/Interfaces/Interfaces.cs

@@ -41,6 +41,21 @@ public interface ILanguageModel
         /// Higher temperature will lead to more randomness. Lower temperature will be more deterministic.</param>
         /// <returns>An LLMResponse response object</returns>
         Task<LLMResponse> GetResponseAsync(IEnumerable<ChatMessage> messages, IEnumerable<FunctionDescriptor> functions, double temperature);
+
+        /// <summary>
+        /// Returns number of Prompt tokens used so far.
+        /// </summary>
+        int PromptTokensUsed { get; }
+
+        /// <summary>
+        /// Returns number of Completion tokens used so far.
+        /// </summary>
+        int CompletionTokensUsed { get; }
+
+        /// <summary>
+        /// Returns max token limit of the LLM for prompt and completion.
+        /// </summary>
+        int MaxTokenLimit { get; }
     }
 
     /// <summary>

src/GenerativeAI/LLM/AzureOpenAILanguageModel.cs

@@ -23,6 +23,12 @@ public AzureOpenAILanguageModel(OpenAIConfig config)
 
         public IVectorTransformer VectorTransformer => openAIClient.VectorTransformer;
 
+        public int PromptTokensUsed => openAIClient.PromptTokensUsed;
+
+        public int CompletionTokensUsed => openAIClient.CompletionTokensUsed;
+
+        public int MaxTokenLimit => openAIClient.MaxTokenLimit;
+
         public Task<LLMResponse> GetResponseAsync(IEnumerable<ChatMessage> messages, double temperature)
         {
             return openAIClient.GetResponseAsync(messages, temperature);

src/GenerativeAI/LLM/MockLanguageModel.cs

@@ -43,6 +43,12 @@ public MockLanguageModel(string model, Dictionary<string, string> responses)
 
         public IVectorTransformer VectorTransformer => client.VectorTransformer;
 
+        public int PromptTokensUsed => throw new NotImplementedException();
+
+        public int CompletionTokensUsed => throw new NotImplementedException();
+
+        public int MaxTokenLimit => 4000; //default is 4K
+
         /// <summary>
         /// Gets the response for given list of chat messages
         /// </summary>

src/GenerativeAI/LLM/OpenAIClient.cs

@@ -5,6 +5,7 @@
 using System;
 using System.Collections.Generic;
 using System.Linq;
+using System.Threading;
 using System.Threading.Tasks;
 using System.Web.Script.Serialization;
 
@@ -17,10 +18,16 @@ internal class OpenAIClient : ILanguageModel
         /// </summary>
         public string ModelName => config.Model;
 
-        public int VectorLength => throw new NotImplementedException();
-
         public IVectorTransformer VectorTransformer => transformer;
 
+        private int promptTokens = 0;
+        public int PromptTokensUsed => promptTokens;
+
+        private int completionTokens = 0;
+        public int CompletionTokensUsed => completionTokens;
+
+        public int MaxTokenLimit => config.TokenLimit;
+
         private readonly IVectorTransformer transformer;
         private readonly OpenAIConfig config;
         private readonly HttpTool httpTool;
@@ -126,8 +133,17 @@ public async Task<LLMResponse> GetResponseAsync(IEnumerable<ChatMessage> message
                 string json = await httpTool.PostAsync(config.CompletionsUrl, jsonPayload);
 
                 var response = serializer.Deserialize<ChatResponse>(json);
+
+                //update usage tokens
+                Interlocked.Add(ref promptTokens, response.usage.prompt_tokens);
+                Interlocked.Add(ref completionTokens, response.usage.completion_tokens);
+
                 var llmResponse = ToLLMResponse(response);
+
+                //Log info
                 Logger.WriteLog(LogLevel.Info, LogOps.Response, llmResponse.Response);
+                Logger.WriteLog(LogLevel.Info, LogOps.Response, $"Prompts Tokens: {response.usage.prompt_tokens}, Completion Tokens: {response.usage.completion_tokens}");
+
                 return llmResponse; 
             }
             catch (Exception ex)

src/GenerativeAI/LLM/OpenAILanguageModel.cs

@@ -30,6 +30,14 @@ internal class ChatResponse
         public string id { get; set; }
         public int created { get; set; }
         public Choice[] choices { get; set; }
+        public Usage usage { get; set; }
+    }
+
+    internal class Usage
+    {
+        public int prompt_tokens { get; set; }
+        public int completion_tokens { get; set; }
+        public int total_tokens { get; set; }
     }
 
     /// <summary>
@@ -47,18 +55,20 @@ public class OpenAILanguageModel : ILanguageModel
         /// to get the API key using OPENAI_API_KEY environment variable.</param>
         public OpenAILanguageModel(string model, string apikey = "") 
         {
-            if(string.IsNullOrEmpty(apikey))
-            {
-                apikey = Configuration.Instance.OpenAIConfig.ApiKey;
-            }
-
-            openAIClient = new OpenAIClient(new OpenAIConfig() { ApiKey = apikey, Model = model });
+            var config = new OpenAIConfig() { ApiKey = apikey, Model = model };
+            openAIClient = new OpenAIClient(config);
         }
 
         public string ModelName => openAIClient.ModelName;
 
         public IVectorTransformer VectorTransformer => openAIClient.VectorTransformer;
 
+        public int PromptTokensUsed => openAIClient.PromptTokensUsed;
+
+        public int CompletionTokensUsed => openAIClient.CompletionTokensUsed;
+
+        public int MaxTokenLimit => openAIClient.MaxTokenLimit;
+
         public Task<LLMResponse> GetResponseAsync(IEnumerable<ChatMessage> messages, double temperature)
         {
             return openAIClient.GetResponseAsync(messages, temperature);

src/GenerativeAI/Prompts/DataExtractorPrompt.txt

@@ -0,0 +1,7 @@
+Extract arguments and values from the below CONTEXT only based on function specification provided, 
+do not include extra parameter and always return your answers in a valid JSON format. 
+
+CONTEXT:
+Today's Date: {{$Today}}
+
+{{$text}}

src/GenerativeAI/Stores/MemoryStore.cs

@@ -81,7 +81,7 @@ private static int GetMessageLength(ChatMessage message)
         {
             if(message == null) return 0;
             if(message.content !=null) return message.content.Length;
-            return 0;
+            return 100; //assume 100 characters for function call messages
         }
 
         /// <summary>