Optimize Memory Usage and Add Cancellation Support in TesseractOcrEngine and ServiceConfiguration

shared/Ocr/Tesseract/TesseractOcrEngine.cs

@@ -13,7 +13,7 @@ public sealed class TesseractOcrEngine : IOcrEngine, IDisposable
     private readonly TesseractEngine _engine;
 
     /// <summary>
-    /// Creates a new instance of the TesseractEngineWrapper passing in a valid TesseractEngine.
+    /// Creates a new instance of the TesseractOcrEngine passing in a valid TesseractEngine.
     /// </summary>
     public TesseractOcrEngine(TesseractConfig tesseractConfig)
     {
@@ -28,15 +28,25 @@ public TesseractOcrEngine(TesseractConfig tesseractConfig)
     ///<inheritdoc/>
     public async Task<string> ExtractTextFromImageAsync(Stream imageContent, CancellationToken cancellationToken = default)
     {
-        await using (var imgStream = new MemoryStream())
+        try
         {
-            await imageContent.CopyToAsync(imgStream, cancellationToken);
-            imgStream.Position = 0;
-
-            using var img = Pix.LoadFromMemory(imgStream.ToArray());
-
-            using var page = this._engine.Process(img);
-            return page.GetText();
+            // Use a buffer for CopyToAsync to reduce memory usage for large images
+            await using (var imgStream = new MemoryStream())
+            {
+                await imageContent.CopyToAsync(imgStream, 81920, cancellationToken).ConfigureAwait(false); // Buffered copy with 80 KB buffer size
+                imgStream.Position = 0; // Reset position for reading
+
+                // Load image from memory and process with Tesseract
+                using var img = Pix.LoadFromMemory(imgStream.ToArray());
+                using var page = this._engine.Process(img);
+
+                return page.GetText(); // Return the extracted text
+            }
+        }
+        catch (OperationCanceledException)
+        {
+            // If operation is canceled, return an empty string or handle accordingly
+            return string.Empty;
         }
     }
 

shared/ServiceConfiguration.cs

@@ -496,23 +496,28 @@ private void SetupForOpenAI()
     /// <param name="builder">KM builder</param>
     /// <param name="addCustomService">Action used to configure the service collection</param>
     /// <typeparam name="T">Target type/interface</typeparam>
-    private T GetServiceInstance<T>(IKernelMemoryBuilder builder, Action<IServiceCollection> addCustomService)
+    private T GetServiceInstance<T>(IKernelMemoryBuilder builder, Action<IServiceCollection> addCustomService) where T : class
     {
-        // Clone the list of service descriptors, skipping T descriptor
-        IServiceCollection services = new ServiceCollection();
-        foreach (ServiceDescriptor d in builder.Services)
+        // Temporarily register the service with a scoped lifecycle to ensure it’s not duplicated or retained in memory unnecessarily
+        var serviceProvider = builder.Services.BuildServiceProvider();
+
+        // Add the custom service configuration directly in a new scope
+        using (var scope = serviceProvider.CreateScope())
         {
-            if (d.ServiceType == typeof(T)) { continue; }
-
-            services.Add(d);
+            // Apply custom service configuration within the scope
+            addCustomService(scope.ServiceProvider.GetRequiredService<IServiceCollection>());
+
+            // Attempt to resolve the service
+            T instance = scope.ServiceProvider.GetService<T>();
+
+            // Check if the instance was successfully created
+            if (instance == null)
+            {
+                throw new ConfigurationException($"Unable to build {typeof(T).Name}");
+            }
+
+            return instance;
         }
-
-        // Add the custom T descriptor
-        addCustomService.Invoke(services);
-
-        // Build and return an instance of T, as defined by `addCustomService`
-        return services.BuildServiceProvider().GetService<T>()
-               ?? throw new ConfigurationException($"Unable to build {nameof(T)}");
     }
 
     /// <summary>