Fix LLM export on in-memory filesystems (#3656)

Author: theletterf

src/Elastic.Markdown/Exporters/LlmMarkdownExporter.cs

@@ -54,27 +54,37 @@ public class LlmMarkdownExporter(bool branded = false) : IMarkdownExporter
 
 	public async ValueTask<bool> FinishExportAsync(IDirectoryInfo outputFolder, Cancel ctx)
 	{
+		var fs = outputFolder.FileSystem;
+		if (!outputFolder.Exists)
+			outputFolder.Create();
+
 		var outputDirectory = outputFolder.FullName;
-		var zipPath = Path.Join(outputDirectory, "llm.zip");
+		var zipPath = fs.Path.Join(outputDirectory, "llm.zip");
 
 		// Create the llms.txt file; omit Elastic boilerplate for branded builds
-		var llmsTxt = Path.Join(outputDirectory, "llms.txt");
-		await outputFolder.FileSystem.File.WriteAllTextAsync(llmsTxt, branded ? string.Empty : LlmsTxtTemplate, ctx);
-
-		using var zip = ZipFile.Open(zipPath, ZipArchiveMode.Create);
-		var llmsTxtRelativePath = Path.GetRelativePath(outputDirectory, llmsTxt);
-		_ = zip.CreateEntryFromFile(llmsTxt, llmsTxtRelativePath);
+		var llmsTxt = fs.FileInfo.New(fs.Path.Join(outputDirectory, "llms.txt"));
+		await fs.File.WriteAllTextAsync(llmsTxt.FullName, branded ? string.Empty : LlmsTxtTemplate, ctx);
 
-		var markdownFiles = Directory.GetFiles(outputDirectory, "*.md", SearchOption.AllDirectories);
+		await using var zipStream = fs.File.Create(zipPath);
+		using var zip = new ZipArchive(zipStream, ZipArchiveMode.Create);
+		await AddFileAsync(zip, llmsTxt, llmsTxt.Name, ctx);
 
-		foreach (var file in markdownFiles)
+		foreach (var file in outputFolder.EnumerateFiles("*.md", SearchOption.AllDirectories))
 		{
-			var relativePath = Path.GetRelativePath(outputDirectory, file);
-			_ = zip.CreateEntryFromFile(file, relativePath);
+			var relativePath = fs.Path.GetRelativePath(outputDirectory, file.FullName).Replace('\\', '/');
+			await AddFileAsync(zip, file, relativePath, ctx);
 		}
 		return true;
 	}
 
+	private static async Task AddFileAsync(ZipArchive archive, IFileInfo file, string entryName, Cancel ctx)
+	{
+		var entry = archive.CreateEntry(entryName, CompressionLevel.Optimal);
+		await using var entryStream = entry.Open();
+		await using var sourceStream = file.FileSystem.File.OpenRead(file.FullName);
+		await sourceStream.CopyToAsync(entryStream, ctx);
+	}
+
 	public async ValueTask<bool> ExportAsync(MarkdownExportFileContext fileContext, Cancel ctx)
 	{
 		var fs = fileContext.BuildContext.WriteFileSystem;

tests/Elastic.Markdown.Tests/Exporters/LlmMarkdownExporterTests.cs

@@ -0,0 +1,33 @@
+// Licensed to Elasticsearch B.V under one or more agreements.
+// Elasticsearch B.V licenses this file to you under the Apache 2.0 License.
+// See the LICENSE file in the project root for more information
+
+using System.IO.Abstractions.TestingHelpers;
+using System.IO.Compression;
+using AwesomeAssertions;
+using Elastic.Markdown.Exporters;
+
+namespace Elastic.Markdown.Tests.Exporters;
+
+public class LlmMarkdownExporterTests
+{
+	[Fact]
+	public async Task FinishExportAsync_InMemoryFileSystem_CreatesArchiveFromInMemoryFiles()
+	{
+		const string outputPath = "/repo/.artifacts/docs/html";
+		var fileSystem = new MockFileSystem(new Dictionary<string, MockFileData>
+		{
+			[$"{outputPath}/guide/page.md"] = new("# Page")
+		});
+		var outputFolder = fileSystem.DirectoryInfo.New(outputPath);
+		var exporter = new LlmMarkdownExporter();
+
+		var result = await exporter.FinishExportAsync(outputFolder, TestContext.Current.CancellationToken);
+
+		result.Should().BeTrue();
+		fileSystem.File.Exists($"{outputPath}/llm.zip").Should().BeTrue();
+		await using var zipStream = fileSystem.File.OpenRead($"{outputPath}/llm.zip");
+		using var archive = new ZipArchive(zipStream, ZipArchiveMode.Read);
+		archive.Entries.Select(entry => entry.FullName).Should().BeEquivalentTo("llms.txt", "guide/page.md");
+	}
+}