Stricter filename validations

src/Elastic.Markdown/DocumentationGenerator.cs

@@ -4,6 +4,7 @@
 
 using System.IO.Abstractions;
 using System.Text.Json;
+using System.Text.RegularExpressions;
 using Elastic.Documentation;
 using Elastic.Documentation.Configuration;
 using Elastic.Documentation.Configuration.LegacyUrlMappings;
@@ -37,7 +38,7 @@ public record GenerationResult
 	public IReadOnlyDictionary<string, LinkRedirect> Redirects { get; set; } = new Dictionary<string, LinkRedirect>();
 }
 
-public class DocumentationGenerator
+public partial class DocumentationGenerator
 {
 	private readonly IDocumentationFileOutputProvider? _documentationFileOutputProvider;
 	private readonly IConversionCollector? _conversionCollector;
@@ -231,6 +232,29 @@ private async Task ExtractEmbeddedStaticResources(Cancel ctx)
 		}
 	}
 
+	[GeneratedRegex(@"^[a-z0-9\s\-_\.\/\\]*[a-z0-9_\-]\.([a-z]+)$")]
+	private static partial Regex FilePathRegex();
+
+	[GeneratedRegex(@"^[a-z0-9_][a-z0-9_\-\s\.]*?\.([a-z]+)$")]
+	private static partial Regex FileNameRegex();
+
+	public static bool IsValidFileName(string strToCheck) =>
+		strToCheck switch
+		{
+			//prior art
+			_ when strToCheck.StartsWith("release-notes/elastic-agent/_snippets/") => true,
+			_ when strToCheck.StartsWith("reference/query-languages/esql/_snippets/") => true,
+			_ when strToCheck.EndsWith(".svg") => true,
+			_ when strToCheck.EndsWith(".gif") => true,
+			_ when strToCheck.EndsWith(".png") => true,
+			_ when strToCheck.EndsWith(".png") => true,
+			"reference/security/prebuilt-rules/audit_policies/windows/README.md" => true,
+			"extend/integrations/developer-workflow-fleet-UI.md" => true,
+			"reference/elasticsearch/clients/ruby/Helpers.md" => true,
+			"explore-analyze/ai-features/llm-guides/connect-to-vLLM.md" => true,
+			_ => FilePathRegex().IsMatch(strToCheck) && FileNameRegex().IsMatch(Path.GetFileName(strToCheck))
+		};
+
 	private async Task ProcessFile(HashSet<string> offendingFiles, DocumentationFile file, DateTimeOffset outputSeenChanges, Cancel ctx)
 	{
 		if (!Context.Force)
@@ -243,8 +267,16 @@ private async Task ProcessFile(HashSet<string> offendingFiles, DocumentationFile
 
 		_logger.LogTrace("--> {FileFullPath}", file.SourceFile.FullName);
 		var outputFile = OutputFile(file.RelativePath);
+
 		if (outputFile is not null)
 		{
+			var relative = Path.GetRelativePath(Context.OutputDirectory.FullName, outputFile.FullName);
+			if (!IsValidFileName(relative))
+			{
+				Context.Collector.EmitError(file.SourceFile.FullName, $"File name {relative} is not valid needs to be lowercase and contain only alphanumeric characters, spaces, dashes, dots and underscores");
+				return;
+			}
+
 			var context = new ProcessingFileContext
 			{
 				BuildContext = Context,
@@ -368,4 +400,5 @@ public async Task<RenderResult> RenderLayout(MarkdownFile markdown, Cancel ctx)
 		await DocumentationSet.ResolveDirectoryTree(ctx);
 		return await HtmlWriter.RenderLayout(markdown, ctx);
 	}
+
 }

tests/Elastic.Markdown.Tests/OutputDirectoryTests.cs

@@ -41,6 +41,276 @@ public async Task CreatesDefaultOutputDirectory()
 		await collector.StopAsync(TestContext.Current.CancellationToken);
 
 		fileSystem.Directory.Exists(".artifacts").Should().BeTrue();
+	}
+
+	[Theory]
+	[MemberData(nameof(ValidFileNames))]
+	public void OutputFileValidationValidNames(string fileName)
+	{
+		var valid = DocumentationGenerator.IsValidFileName(fileName);
+		valid.Should().BeTrue($"'{fileName}' should be a valid filename");
+	}
 
+	[Theory]
+	[MemberData(nameof(InvalidFileNames))]
+	public void OutputFileValidationInvalidNames(string fileName)
+	{
+		var valid = DocumentationGenerator.IsValidFileName(fileName);
+		valid.Should().BeFalse($"'{fileName}' should be an invalid filename");
 	}
+
+	public static TheoryData<string> ValidFileNames =>
+	[
+		"test.md",
+		"file.txt",
+		"index.html",
+		"readme.rst",
+
+		// With numbers
+		"test123.md",
+		"123test.md",
+		"file2.md",
+		"99bottles.md",
+
+		// With underscores
+		"test_file.md",
+		"my_long_file_name.md",
+		"_leading_underscore.md",
+		"trailing_underscore_.md",
+
+		// With hyphens
+		"test-file.md",
+		"my-long-file-name.md",
+		"trailing-hyphen-.md",
+
+		// Combined underscores and hyphens
+		"test_file-name.md",
+		"my-file_name.md",
+
+		// With dots in filename (before extension)
+		"test.config.md",
+		"file.test.backup.md",
+		"v1.0.0.md",
+
+		// With spaces (allowed per regex)
+		"test file.md",
+		"my document.md",
+
+		// Paths with all lowercase directories
+		"path/to/file.md",
+		"deep/nested/path/to/file.md",
+		"folder/subfolder/document.md",
+
+		// Paths with numbers
+		"path123/file.md",
+		"v1/docs/guide.md",
+
+		// Paths with underscores and hyphens
+		"my_folder/file.md",
+		"my-folder/file.md",
+		"path_to/sub-folder/file.md",
+
+		// SVG files exception (even with uppercase - per the .EndsWith checks)
+		"image.svg",
+		"Icon.svg",
+		"LOGO.svg",
+		"path/to/Image.svg",
+
+		// PNG files exception
+		"image.png",
+		"Screenshot.png",
+		"IMAGE.png",
+		"path/to/Logo.png",
+
+		// GIF files exception
+		"animation.gif",
+		"Loading.gif",
+		"SPINNER.gif",
+
+		// ESQL snippets exception (prior art)
+		"reference/query-languages/esql/_snippets/functions/examples/cbrt.md",
+		"reference/query-languages/esql/_snippets/anything/here/File.md",
+		"reference/query-languages/esql/_snippets/UPPERCASE.md",
+
+		// Hardcoded exceptions
+		"reference/security/prebuilt-rules/audit_policies/windows/README.md",
+		"extend/integrations/developer-workflow-fleet-UI.md",
+		"reference/elasticsearch/clients/ruby/Helpers.md",
+		"explore-analyze/ai-features/llm-guides/connect-to-vLLM.md"
+	];
+
+	public static TheoryData<string> InvalidFileNames =>
+	[
+		"Test.md",
+		"FILE.md",
+		"MyFile.md",
+		"testFile.md",
+		"README.md",
+
+		// Uppercase in extension
+		"test.MD",
+		"test.Md",
+		"file.TXT",
+		"document.Html",
+
+		// Uppercase in directory path
+		"Path/file.md",
+		"path/To/file.md",
+		"FOLDER/file.md",
+		"docs/MyFolder/file.md",
+
+		// Filenames starting with invalid characters (must start with [a-z0-9_])
+		"-leading-hyphen.md",
+		"-file.md",
+		".hidden.md",
+		" leading-space.md",
+		"path/to/-invalid.md",
+		"path/to/.hidden.md",
+		"path/to/ space.md",
+
+		// Special characters - parentheses
+		"test(1).md",
+		"file (copy).md",
+		"document(v2).md",
+
+		// Special characters - square brackets
+		"test[1].md",
+		"file[copy].md",
+
+		// Special characters - curly braces
+		"test{1}.md",
+
+		// Special characters - exclamation mark
+		"test!.md",
+		"important!file.md",
+
+		// Special characters - at sign
+		"[email protected]",
+		"[email protected]",
+
+		// Special characters - hash
+		"test#1.md",
+		"file#.md",
+
+		// Special characters - dollar sign
+		"test$file.md",
+		"price$.md",
+
+		// Special characters - percent
+		"test%file.md",
+		"100%done.md",
+
+		// Special characters - caret
+		"test^file.md",
+
+		// Special characters - ampersand
+		"test&file.md",
+		"this&that.md",
+
+		// Special characters - asterisk
+		"test*file.md",
+		"*.md",
+
+		// Special characters - plus sign
+		"test+file.md",
+		"c++.md",
+
+		// Special characters - equals sign
+		"test=file.md",
+
+		// Special characters - pipe
+		"test|file.md",
+
+		// Special characters - less than / greater than
+		"test<file>.md",
+
+		// Special characters - colon
+		"test:file.md",
+
+		// Special characters - semicolon
+		"test;file.md",
+
+		// Special characters - single quote
+		"test'file.md",
+		"it's.md",
+
+		// Special characters - double quote
+		"test\"file.md",
+
+		// Special characters - backtick
+		"test`file.md",
+
+		// Special characters - tilde
+		"test~file.md",
+		"~temp.md",
+
+		// Special characters - comma
+		"test,file.md",
+		"a,b,c.md",
+
+		// Special characters - question mark
+		"test?.md",
+		"what?.md",
+
+		// No extension
+		"testfile",
+		"README",
+		"Makefile",
+
+		// Just extension
+		".md",
+		".txt",
+
+		// Empty extension
+		"test.",
+
+		// Double extension edge cases with uppercase
+		"test.Config.md",
+		"file.Test.md",
+
+		// Non-ASCII characters - accented
+		"tëst.md",
+		"café.md",
+		"naïve.md",
+		"résumé.md",
+
+		// Non-ASCII characters - other alphabets
+		"тест.md",
+		"测试.md",
+		"テスト.md",
+
+		// Non-ASCII characters - symbols
+		"test™.md",
+		"file©.md",
+
+		// Empty string
+		"",
+
+		// Whitespace only
+		"   ",
+
+		// Extension only variations
+		"..md",
+
+		// Numbers in extension (if we expect only letters)
+		"test.md5",
+		"file.mp3",
+		"video.mp4",
+
+		// CamelCase variations
+		"camelCase.md",
+		"PascalCase.md",
+		"mixedCASE.md",
+
+		// Acronyms
+		"API.md",
+		"HTTP.md",
+		"XMLParser.md",
+
+		// Common problematic filenames
+		"CHANGELOG.md",
+		"LICENSE.md",
+		"CONTRIBUTING.md",
+		"TODO.md"
+	];
 }