Add Per-extractor Instructions

src/Elzik.Breef.Application/BreefGenerator.cs

@@ -10,7 +10,11 @@ public class BreefGenerator(
         public async Task<PublishedBreef> GenerateBreefAsync(string url)
         {
             var extract = await contentExtractor.ExtractAsync(url);
-            var summary = await contentSummariser.SummariseAsync(extract.Content);
+
+            var instructionsPath = Path.Combine(AppContext.BaseDirectory, "SummarisationInstructions", "HtmlContent.md");
+            var instructions = await File.ReadAllTextAsync(instructionsPath);
+
+            var summary = await contentSummariser.SummariseAsync(extract.Content, instructions);
 
             var breef = new Domain.Breef(url, extract.Title ,summary, extract.PreviewImageUrl);
 

src/Elzik.Breef.Domain/IContentSummariser.cs

@@ -2,6 +2,6 @@
 {
     public interface IContentSummariser
     {
-        Task<string> SummariseAsync(string content);
+        Task<string> SummariseAsync(string content, string instructions);
     }
 }

src/Elzik.Breef.Infrastructure/AI/AiContentSummariser.cs

@@ -7,25 +7,15 @@ namespace Elzik.Breef.Infrastructure.AI;
 
 public class AiContentSummariser(
     ILogger<AiContentSummariser> logger, 
-    IChatCompletionService Chat, 
-    IOptions<AiContentSummariserOptions> summariserOptions) : IContentSummariser
+    IChatCompletionService Chat) : IContentSummariser
 {
-    public async Task<string> SummariseAsync(string content)
+    public async Task<string> SummariseAsync(string content, string instructions)
     {
-        var systemPrompt = @$"
-You are an expert summarizer. Your task is to summarize the provided text:
-  - Summarise text, including HTML entities.
-  - Limit summaries to {summariserOptions.Value.TargetSummaryLengthPercentage}% of the original length but never more then {summariserOptions.Value.TargetSummaryMaxWordCount} words.
-  - Ensure accurate attribution of information to the correct entities.
-  - Do not include a link to the original articles.
-  - Do not include the title in the response.
-  - Do not include any metadata in the response.
-  - Do not include a code block in the response.";
-
+        ArgumentNullException.ThrowIfNullOrWhiteSpace(instructions);
         var formattingInstructions = "Summarise this content in an HTML format using paragraphs and " +
             "bullet points to enhance readability\n:";
 
-        var chatHistory = new ChatHistory(systemPrompt);
+        var chatHistory = new ChatHistory(instructions);
         chatHistory.AddMessage(AuthorRole.User, $"{formattingInstructions}{content}");
 
         var result = await Chat.GetChatMessageContentAsync(chatHistory);

src/Elzik.Breef.Infrastructure/Elzik.Breef.Infrastructure.csproj

@@ -27,4 +27,11 @@
     <ProjectReference Include="..\Elzik.Breef.Domain\Elzik.Breef.Domain.csproj" />
   </ItemGroup>
 
+  <ItemGroup>
+    <None Include="..\Elzik.Breef.Infrastructure\SummarisationInstructions\**\*.md" 
+          Link="SummarisationInstructions\%(RecursiveDir)%(Filename)%(Extension)">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </None>
+  </ItemGroup>
+
 </Project>

src/Elzik.Breef.Infrastructure/SummarisationInstructions/HtmlContent.md

@@ -0,0 +1,8 @@
+You are an expert summarizer. Your task is to summarize the provided text:
+ - Summarise text, including HTML entities.
+ - Limit summaries to 10% of the original length but never more then200 words.
+ - Ensure accurate attribution of information to the correct entities.
+ - Do not include a link to the original articles.
+ - Do not include the title in the response.
+ - Do not include any metadata in the response.
+ - Do not include a code block in the response.

tests/Elzik.Breef.Api.Tests.Integration/Elzik.Breef.Api.Tests.Integration.csproj

@@ -39,4 +39,11 @@
     <Using Include="Xunit" />
   </ItemGroup>
 
+  <ItemGroup>
+    <None Include="..\..\src\Elzik.Breef.Infrastructure\SummarisationInstructions\**\*.md" 
+		  Link="SummarisationInstructions\%(RecursiveDir)%(Filename)%(Extension)">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+  </None>
+  </ItemGroup>
+
 </Project>

tests/Elzik.Breef.Infrastructure.Tests.Integration/AiContentSummariserTests.cs

@@ -8,13 +8,14 @@
 using Microsoft.SemanticKernel.ChatCompletion;
 using Shouldly;
 using Xunit.Abstractions;
+using System.IO;
 
 namespace Elzik.Breef.Infrastructure.Tests.Integration;
 
 public class AiContentSummariserTests(ITestOutputHelper testOutputHelper)
 {
     private readonly ITestOutputHelper _testOutputHelper = testOutputHelper
-            ?? throw new ArgumentNullException(nameof(testOutputHelper));
+      ?? throw new ArgumentNullException(nameof(testOutputHelper));
     private readonly FakeLogger<AiContentSummariser> _fakeLogger = new();
 
     [Theory]
@@ -24,36 +25,32 @@ public async Task Summarise_WithValidContent_ReturnsSummary(string testExtracted
     {
         // Arrange
         var configuration = new ConfigurationBuilder()
-                .AddEnvironmentVariables("breef_")
-                .Build();
+          .AddEnvironmentVariables("breef_")
+       .Build();
         var aiServiceOptions = configuration.GetRequiredSection("AiService").Get<AiServiceOptions>()
-            ?? throw new InvalidOperationException("AI service options not found in AiService configuration section.");
+     ?? throw new InvalidOperationException("AI service options not found in AiService configuration section.");
 
         var testContent = await File.ReadAllTextAsync(Path.Join("../../../../TestData", testExtractedContentFile));
 
-        var builder = Kernel.CreateBuilder().AddAzureOpenAIChatCompletion(
-            aiServiceOptions.ModelId, aiServiceOptions.EndpointUrl, aiServiceOptions.ApiKey);
-        builder.Services.AddLogging(loggingBuilder =>
-        {
+      var builder = Kernel.CreateBuilder().AddAzureOpenAIChatCompletion(
+aiServiceOptions.ModelId, aiServiceOptions.EndpointUrl, aiServiceOptions.ApiKey);
+ builder.Services.AddLogging(loggingBuilder =>
+     {
             loggingBuilder.AddProvider(new TestOutputLoggerProvider(_testOutputHelper));
-            loggingBuilder.SetMinimumLevel(LogLevel.Information);
+        loggingBuilder.SetMinimumLevel(LogLevel.Information);
         });
 
         var kernel = builder.Build();
-        var chatCompletionService = kernel.GetRequiredService<IChatCompletionService>();
+     var chatCompletionService = kernel.GetRequiredService<IChatCompletionService>();
 
-        var summariserOptions = Options.Create(new AiContentSummariserOptions
-        {
-            TargetSummaryLengthPercentage = 10,
-            TargetSummaryMaxWordCount = 200
-        });
+        var systemPrompt = await File.ReadAllTextAsync(Path.Combine(AppContext.BaseDirectory, "SummarisationInstructions", "HtmlContent.md"));
 
         // Act
-        var contentSummariser = new AiContentSummariser(_fakeLogger, chatCompletionService, summariserOptions);
-        var summary = await contentSummariser.SummariseAsync(testContent);
+        var contentSummariser = new AiContentSummariser(_fakeLogger, chatCompletionService);
+        var summary = await contentSummariser.SummariseAsync(testContent, systemPrompt);
 
         // Assert
-        _testOutputHelper.WriteLine(summary);
-        summary.Length.ShouldBeGreaterThan(100, "because this is the minimum acceptable summary length");
+      _testOutputHelper.WriteLine(summary);
+      summary.Length.ShouldBeGreaterThan(100, "because this is the minimum acceptable summary length");
     }
 }

tests/Elzik.Breef.Infrastructure.Tests.Integration/Elzik.Breef.Infrastructure.Tests.Integration.csproj

@@ -53,4 +53,10 @@
     <Using Include="Xunit" />
   </ItemGroup>
 
+  <ItemGroup>
+    <None Include="..\..\src\Elzik.Breef.Infrastructure\SummarisationInstructions\**\*.md" Link="SummarisationInstructions\%(RecursiveDir)%(Filename)%(Extension)">
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </None>
+  </ItemGroup>
+
 </Project>

tests/Elzik.Breef.Infrastructure.Tests.Unit/ContentSummariserTests.cs

@@ -22,12 +22,7 @@ public ContentSummariserTests()
     {
         _logger = new FakeLogger<AiContentSummariser>();
         _chatCompletionService = Substitute.For<IChatCompletionService>();
-        var summariserOptions = Options.Create(new AiContentSummariserOptions
-        {
-            TargetSummaryLengthPercentage = 10,
-            TargetSummaryMaxWordCount = 200
-        });
-        _contentSummariser = new AiContentSummariser(_logger, _chatCompletionService, summariserOptions);
+        _contentSummariser = new AiContentSummariser(_logger, _chatCompletionService);
 
         _testContent = "This is a test content.";
         _testSummary = "Test summary.";
@@ -52,7 +47,7 @@ public ContentSummariserTests()
     public async Task SummariseAsync_ValidContent_ReturnsSummary()
     {
         // Act
-        var result = await _contentSummariser.SummariseAsync(_testContent);
+        var result = await _contentSummariser.SummariseAsync(_testContent, "Test instructions");
 
         // Assert
         result.ShouldBe("Test summary.");
@@ -61,22 +56,16 @@ public async Task SummariseAsync_ValidContent_ReturnsSummary()
     [Fact]
     public async Task SummariseAsync_ValidContent_ProvidesModelInstructions()
     {
+        // Arrange
+        var instructions = "Test instructions";
+
         // Act
-        _ = await _contentSummariser.SummariseAsync(_testContent);
+        _ = await _contentSummariser.SummariseAsync(_testContent, instructions);
 
         // Assert
-        var systemPrompt = @$"
-You are an expert summarizer. Your task is to summarize the provided text:
-  - Summarise text, including HTML entities.
-  - Limit summaries to 10% of the original length but never more then 200 words.
-  - Ensure accurate attribution of information to the correct entities.
-  - Do not include a link to the original articles.
-  - Do not include the title in the response.
-  - Do not include any metadata in the response.
-  - Do not include a code block in the response.";
         await _chatCompletionService.Received(1).GetChatMessageContentsAsync(
             Arg.Is<ChatHistory>(ch => ch.Any(m =>
-                m.Content == systemPrompt && m.Role == AuthorRole.System)),
+                m.Content == instructions && m.Role == AuthorRole.System)),
             Arg.Any<PromptExecutionSettings>(),
             Arg.Any<Kernel>(),
             Arg.Any<CancellationToken>());
@@ -86,7 +75,7 @@ await _chatCompletionService.Received(1).GetChatMessageContentsAsync(
     public async Task SummariseAsync_ValidContent_Logs()
     {
         // Act
-        await _contentSummariser.SummariseAsync(_testContent);
+        await _contentSummariser.SummariseAsync(_testContent, "Test instructions");
 
         // Assert
         _logger.Collector.Count.ShouldBe(1);