947961

Author: Yaminisrisf4389

Smart_Summarize/AIAssitViewModel.cs

@@ -2,6 +2,7 @@
 using Syncfusion.Windows.PdfViewer;
 using System.Collections.ObjectModel;
 using System.ComponentModel;
+using System.Text;
 
 namespace Smart_Summarize
 {
@@ -13,6 +14,7 @@ internal class AIAssitViewModel : INotifyPropertyChanged
         private Author currentUser;
         private PdfViewerControl pdfViewer;
         MicrosoftAIExtension microsoftAIExtension;
+        StringBuilder processedText = new StringBuilder();
         public event PropertyChangedEventHandler PropertyChanged;
         #endregion
 
@@ -129,27 +131,51 @@ private async Task AddSuggestions(String text)
         /// <summary>
         /// Extracts text from each page of the PDF document.
         /// </summary>
-        private async Task ExtractDetailsFromPDF()
+        private async Task<string> ExtractDetailsFromPDF()
         {
-            List<string> extractedText = new List<string>();
+            StringBuilder extractedText = new StringBuilder();
             Syncfusion.Pdf.TextLines textLines = new Syncfusion.Pdf.TextLines();
             //Extract the text from the PDF document
             for (int pageIndex = 0; pageIndex < pdfViewer.PageCount; pageIndex++)
             {
                 string text = $"... Page {pageIndex + 1} ...\n";
                 text += pdfViewer.ExtractText(pageIndex, out textLines);
-                extractedText.Add(text);
+                extractedText.AppendLine(text);
             }
 
-            await microsoftAIExtension.CreateEmbeddedPage(extractedText.ToArray());
+            return ProcessExtractedText(extractedText.ToString());
+        }
+        /// <summary>
+        /// Processes the extracted full text from a document by splitting it into pages.
+        /// </summary>
+        /// <param name="fullText">The complete extracted text from the document.</param>
+        /// <returns>A formatted string containing the processed text.</returns>
+        private string ProcessExtractedText(string fullText)
+        {
+            string[] pages = fullText.Split(new string[] { "\f", "\n\nPage " }, StringSplitOptions.RemoveEmptyEntries);
+
+            for (int i = 0; i < pages.Length; i++)
+            {
+                processedText.AppendLine($"... Page {i + 1} ...");
+                string[] lines = pages[i].Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries);
+                int maxLines = Math.Min(1000, lines.Length);
+
+                for (int j = 0; j < maxLines; j++)
+                {
+                    processedText.AppendLine(lines[j]);
+                }
+                processedText.AppendLine();
+            }
+
+            return processedText.ToString();
         }
         /// <summary>
         /// Summarizes the extracted text from the PDF using Extension AI.
         /// </summary>
         private async Task<string> SummarizePDF()
         {
             //Summarize the text using the Semantic Kernel AI
-            string summary = await microsoftAIExtension.GetAnswerFromGPT("You are a helpful assistant. Your task is to analyze the provided text and generate short summary as a plain text.");
+            string summary = await microsoftAIExtension.GetAnswerFromGPT(processedText.ToString());
             return summary;
         }
         /// <summary>

Smart_Summarize/MicrosoftAIExtension.cs

@@ -1,7 +1,6 @@
 using Microsoft.Extensions.AI;
 using Azure.AI.OpenAI;
 using System.Text;
-using SmartComponents.LocalEmbeddings;
 namespace Smart_Summarize
 {
     internal class MicrosoftAIExtension
@@ -12,7 +11,6 @@ internal class MicrosoftAIExtension
         internal string key = string.Empty;
 
         private IChatClient clientAI;
-        public Dictionary<string, EmbeddingF32>? PageEmbeddings { get; set; }
 
         /// <summary>
         /// Initializes a new instance of the <see cref="MicrosoftAIExtension"/> class.
@@ -23,30 +21,20 @@ public MicrosoftAIExtension(string key)
             clientAI = new AzureOpenAIClient(new System.Uri(endpoint), new System.ClientModel.ApiKeyCredential(key)).AsChatClient(deploymentName);
         }
 
-        /// <summary>
-        /// Create the embedded page from the extracted chunks in the PDF
-        /// </summary>
-        /// <param name="chunks">Extracted text from pdfViewer</param>
-        /// <returns></returns>
-        public async Task CreateEmbeddedPage(string[] chunks)
-        {
-            var embedder = new LocalEmbedder();
-            PageEmbeddings = chunks.Select(x => KeyValuePair.Create(x, embedder.Embed(x))).ToDictionary(k => k.Key, v => v.Value);
-        }
         /// <summary>
         /// Method to get the answer from GPT using the semantic kernel
         /// </summary>
-        /// <param name="systemPrompt">Prompt for the system message</param>
+        /// <param name="extractedText">Extracted text from the document</param>
         /// <returns>Returns the form data as a string</returns>
-        public async Task<string> GetAnswerFromGPT(string systemPrompt)
+        public async Task<string> GetAnswerFromGPT(string extractedText)
         {
             if (clientAI != null)
             {
-                List<string> message = PageEmbeddings.Keys.Take(10).ToList();
+                string systemPrompt = "You are a helpful assistant. Your task is to analyze the provided text and generate short summary as a plain text";
                 var chatMessages = new List<ChatMessage>
                 {
                     new ChatMessage(ChatRole.System, systemPrompt),
-                    new ChatMessage(ChatRole.User, string.Join(" ", message))
+                    new ChatMessage(ChatRole.User, extractedText)
                 };
                 var response = await clientAI.GetResponseAsync(chatMessages);
                 return response.ToString();
@@ -76,21 +64,5 @@ public async Task<string> GetAnswerFromGPT(string systemPrompt, string userText)
             }
             return string.Empty;
         }
-
-        public async Task<string> AnswerQuestion(string question)
-        {
-            var embedder = new LocalEmbedder();
-            var questionEmbedding = embedder.Embed(question);
-            var results = LocalEmbedder.FindClosestWithScore(questionEmbedding, PageEmbeddings.Select(x => (x.Key, x.Value)), 5, 0.5f);
-            StringBuilder builder = new StringBuilder();
-            foreach (var result in results)
-            {
-                builder.AppendLine(result.Item);
-            }
-            string message = builder.ToString();
-            var answer = await GetAnswerFromGPT("You are a helpful assistant. Use the provided PDF document pages and pick a precise page to answer the user question. Provide the answer in plain text without any special formatting or Markdown syntax. Pages: " + message, question);
-
-            return answer;
-        }
     }
 }

Smart_Summarize/Smart_Summarize.csproj

@@ -14,7 +14,6 @@
 
   <ItemGroup>
 
-   <PackageReference Include="SmartComponents.LocalEmbeddings" Version="0.1.0-preview10148" />
 	  <PackageReference Include="Syncfusion.PdfViewer.WPF" Version="*" />
 	  <PackageReference Include="Syncfusion.SfBusyIndicator.WPF" Version="*" />
 	  <PackageReference Include="Syncfusion.SfChat.WPF" Version="*" />