Merge pull request #136 from Encamina/@mramos/update-skVisionImageDoc…

…umentConnector

Add resolution limit to `SkVisionImageDocumentConnector`

CHANGELOG.md

@@ -28,7 +28,7 @@ Previous classification is not required if changes are simple or all belong to t
 
 - Added the `IEnmarchaDocumentConnector` interface that extends the existing `IDocumentConnector`. This interface, by now, adds a `CompatibleFileFormats` property that returns the file formats supported by the connector. Existing document connectors have been updated to implement this interface.
 - Added `CsvTsvDocumentConnector` document connector that allows to read CSV and TSV files keeping the headers in different chunks.
-- Added `SkVisionImageDocumentConnector` which allows to read images and extract text from them. Using Semantic Kernel vision capabilities.
+- Added `SkVisionImageDocumentConnector` which allows to read images and extract text from them. Using Semantic Kernel vision capabilities. You can configure the connector with the `SkVisionImageDocumentConnectorOptions` class.
 - The `IDocumentConnectorProvider` interface now works with the `IEnframeDocumentConnector` interface instead of `IDocumentConnector`.
     - The `AddDocumentConnector` function has been modified by removing the `fileExtension` parameter, which will now come in the `CompatibleFileFormats` property of the document connector.
 - The `ParagraphPptxDocumentConnector` class is no longer sealed, allowing the creation of derived classes.

Directory.Build.props

@@ -17,7 +17,7 @@
 
   <PropertyGroup>
     <VersionPrefix>8.1.8</VersionPrefix>
-    <VersionSuffix>preview-05</VersionSuffix>
+    <VersionSuffix>preview-06</VersionSuffix>
   </PropertyGroup>
 
   <!--

src/Encamina.Enmarcha.SemanticKernel.Connectors.Document/Connectors/SkVisionImageDocumentConnector.cs

@@ -1,8 +1,11 @@
 using CommunityToolkit.Diagnostics;
 
 using Encamina.Enmarcha.SemanticKernel.Connectors.Document.Exceptions;
+using Encamina.Enmarcha.SemanticKernel.Connectors.Document.Options;
 using Encamina.Enmarcha.SemanticKernel.Connectors.Document.Utils;
 
+using Microsoft.Extensions.Options;
+
 using Microsoft.SemanticKernel;
 using Microsoft.SemanticKernel.ChatCompletion;
 
@@ -49,14 +52,17 @@ 4. Transcribe only in Markdown format.
         """;
 
     private readonly IChatCompletionService chatCompletionService;
+    private readonly SkVisionImageDocumentConnectorOptions options;
 
     /// <summary>
     /// Initializes a new instance of the <see cref="SkVisionImageDocumentConnector"/> class.
     /// </summary>
     /// <param name="kernel">A valid <see cref="Kernel"/> instance.</param>
-    public SkVisionImageDocumentConnector(Kernel kernel)
+    /// <param name="options">Configuration options for this connector.</param>
+    public SkVisionImageDocumentConnector(Kernel kernel, IOptions<SkVisionImageDocumentConnectorOptions> options)
     {
         chatCompletionService = kernel.GetRequiredService<IChatCompletionService>();
+        this.options = options.Value;
     }
 
     /// <inheritdoc/>
@@ -67,9 +73,15 @@ public virtual string ReadText(Stream stream)
     {
         Guard.IsNotNull(stream);
 
-        var mimeType = ImageHelper.GetMimeType(stream);
+        var (mimeType, width, height) = ImageHelper.GetImageInfo(stream);
         stream.Position = 0;
 
+        // Check image resolution
+        if (width > options.ResolutionLimit || height > options.ResolutionLimit)
+        {
+            throw new DocumentTooLargeException();
+        }
+
         var history = new ChatHistory(SystemPrompt);
 
         var message = new ChatMessageContentItemCollection()

src/Encamina.Enmarcha.SemanticKernel.Connectors.Document/Encamina.Enmarcha.SemanticKernel.Connectors.Document.csproj

@@ -20,6 +20,8 @@
 
   <ItemGroup>
     <PackageReference Include="ExcelNumberFormat" Version="1.1.0" />
+    <PackageReference Include="Microsoft.Extensions.Options.ConfigurationExtensions" Version="8.0.0" />
+    <PackageReference Include="Microsoft.Extensions.Options.DataAnnotations" Version="8.0.0" />
     <PackageReference Include="Microsoft.SemanticKernel.Plugins.Document" Version="1.15.0-alpha" />
     <PackageReference Include="PdfPig" Version="0.1.8" />
     <PackageReference Include="SixLabors.ImageSharp" Version="3.1.5" />

src/Encamina.Enmarcha.SemanticKernel.Connectors.Document/Extensions/IServiceCollectionExtensions.cs

@@ -2,6 +2,9 @@
 
 using Encamina.Enmarcha.SemanticKernel.Connectors.Document;
 using Encamina.Enmarcha.SemanticKernel.Connectors.Document.Connectors;
+using Encamina.Enmarcha.SemanticKernel.Connectors.Document.Options;
+
+using Microsoft.Extensions.Configuration;
 
 namespace Microsoft.Extensions.DependencyInjection;
 
@@ -196,12 +199,18 @@ public static IServiceCollection AddCsvTsvDocumentConnector(this IServiceCollect
     }
 
     /// <summary>
-    /// Adds the <see cref="SkVisionImageDocumentConnector"/> implementation of <see cref="IEnmarchaDocumentConnector"/> to the specified <see cref="IServiceCollection"/> as a singleton service.
+    /// Adds and configures the <see cref="SkVisionImageDocumentConnector"/> implementation of <see cref="IEnmarchaDocumentConnector"/> to the specified <see cref="IServiceCollection"/> as a scoped service.
     /// </summary>
     /// <param name="services">The <see cref="IServiceCollection"/> to add services to.</param>
+    /// <param name="configuration">The current set of key-value application configuration parameters.</param>
     /// <returns>The <see cref="IServiceCollection"/> so that additional calls can be chained.</returns>
-    public static IServiceCollection AddSkVisionImageDocumentConnector(this IServiceCollection services)
+    public static IServiceCollection AddSkVisionImageDocumentConnector(this IServiceCollection services, IConfiguration configuration)
     {
+        services.AddOptions<SkVisionImageDocumentConnectorOptions>()
+                .Bind(configuration.GetSection(nameof(SkVisionImageDocumentConnectorOptions)))
+                .ValidateDataAnnotations()
+                .ValidateOnStart();
+
         return services.AddScoped<IEnmarchaDocumentConnector, SkVisionImageDocumentConnector>();
     }
 

src/Encamina.Enmarcha.SemanticKernel.Connectors.Document/Options/SkVisionImageDocumentConnectorOptions.cs

@@ -0,0 +1,14 @@
+using Encamina.Enmarcha.SemanticKernel.Connectors.Document.Connectors;
+
+namespace Encamina.Enmarcha.SemanticKernel.Connectors.Document.Options;
+
+/// <summary>
+/// Configuration options for <see cref="SkVisionImageDocumentConnector"/>.
+/// </summary>
+public sealed class SkVisionImageDocumentConnectorOptions
+{
+    /// <summary>
+    /// Gets the resolution limit (in pixels) for images.
+    /// </summary>
+    public int ResolutionLimit { get; init; } = 8192;
+}

src/Encamina.Enmarcha.SemanticKernel.Connectors.Document/Utils/ImageHelper.cs

@@ -10,18 +10,42 @@ namespace Encamina.Enmarcha.SemanticKernel.Connectors.Document.Utils;
 public static class ImageHelper
 {
     /// <summary>
-    /// Gets the MIME type of an image from a stream.
+    /// Gets both the MIME type and resolution (width and height) of an image from a stream.
     /// </summary>
     /// <param name="stream">The stream containing the image.</param>
-    /// <returns>The MIME type of the image, or "application/octet-stream" if the type is unknown.</returns>
+    /// <returns>A tuple containing the MIME type, width, and height of the image.</returns>
     /// <exception cref="ArgumentNullException">Thrown when the stream is null.</exception>
     /// <exception cref="ArgumentException">Thrown when the stream does not contain a valid image.</exception>
-    public static string GetMimeType(Stream stream)
+    public static (string MimeType, int Width, int Height) GetImageInfo(Stream stream)
     {
         Guard.IsNotNull(stream);
 
-        var image = Image.Load(stream);
+        using var image = Image.Load(stream);
+
+        var mimeType = image.Metadata.DecodedImageFormat?.DefaultMimeType ?? System.Net.Mime.MediaTypeNames.Application.Octet;
+
+        return (mimeType, image.Width, image.Height);
+    }
 
-        return image.Metadata.DecodedImageFormat?.DefaultMimeType ?? System.Net.Mime.MediaTypeNames.Application.Octet;
+    /// <summary>
+    /// Gets the MIME type of an image from a stream.
+    /// </summary>
+    /// <param name="stream">The stream containing the image.</param>
+    /// <returns>The MIME type of the image, or "application/octet-stream" if the type is unknown.</returns>
+    public static string GetMimeType(Stream stream)
+    {
+        var (mimeType, _, _) = GetImageInfo(stream);
+        return mimeType;
+    }
+
+    /// <summary>
+    /// Gets the resolution (width and height) of an image from a stream.
+    /// </summary>
+    /// <param name="stream">The stream containing the image.</param>
+    /// <returns>A tuple containing the width and height of the image.</returns>
+    public static (int Width, int Height) GetResolution(Stream stream)
+    {
+        var (_, width, height) = GetImageInfo(stream);
+        return (width, height);
     }
 }