Incremental Roslyn Source Generators: Using Additional Files – Part 7

In the previous article the Source Generator itself needed a 3rd-party library Newtonsoft.Json in order to generate new source code. The JSON-strings were hard-coded inside the Source Generator for simplicity reasons. In this article we will see how to process not just .NET code, but also other files, like JSON or XML.

In this article:

Pawel Gerr is architect consultant at Thinktecture. He focuses on backends with .NET Core and knows Entity Framework inside out.

More information about the Smart Enums and the source code can be found on GitHub:

I recommend reading part 6 of the series because this article is based on the results of the previous one.

In the previous article the translations were rendered out of hard-coded JSON. This time, the JSON content should come from a file residing along with the production code.

Creation of an "Additional File"

The (content) files of a project are not automatically available in Source Generators. The files must be tagged as an AdditionalFile to be provided to a Source Generator. An additional file can be anywhere on the file system accessible by the compiler.

The content of previously hard-coded JSON is moved to a new file Translations.json into the project DemoConsoleApplication.

  "ProductCategory": { 
    "en": "Product category", 
    "de": "Produktkategorie" 

The JSON file is included into AdditionalFiles to contribute to code generation.

					<Project Sdk="Microsoft.NET.Sdk"> 
      <AdditionalFiles Include="Translations.json" /> 


Accessing Additional Files

The additional files are accessible by AdditionalTextsProvider and provide the same capabilities as the SyntaxProvider. All performance considerations described in part 4 are applicable here as well. After filtering the files by name, the file contents are combined with enumTypes and generators which leads to a change of the signature of the method GenerateCode.

public class DemoSourceGenerator : IIncrementalGenerator 
   public void Initialize(IncrementalGeneratorInitializationContext context) 
      var enumTypes = context.SyntaxProvider 
                             .CreateSyntaxProvider(CouldBeEnumerationAsync, GetEnumInfoOrNull) 
                             .Where(type => type is not null)! 
                             .SelectMany((enumInfos, _) => enumInfos.Distinct()); 
      var translations = context.AdditionalTextsProvider 
                                .Where(text => text.Path.EndsWith("translations.json",
                                .Select((text, token) => text.GetText(token)?.ToString()) 
                                .Where(text => text is not null)! 
      var generators = context.GetMetadataReferencesProvider() 
            .SelectMany(static (reference, _) => TryGetCodeGenerator(reference, out var factory) 
                                                  ? ImmutableArray.Create(factory) 
                                                  : ImmutableArray<ICodeGenerator>.Empty) 
   private static void GenerateCode( 
      SourceProductionContext context, 
      ((DemoEnumInfo, ImmutableArray<string>), ImmutableArray<ICodeGenerator>) args) 
      var ((enumInfo, translationsAsJson), generators) = args; 
      if (generators.IsDefaultOrEmpty) 
      var translationsByClassName = GetTranslationsByClassName(context, translationsAsJson); 
      foreach (var generator in generators.Distinct()) 
         if (translationsByClassName is null
            || !translationsByClassName.TryGetValue(enumInfo.Name, out var translations))
            translations = _noTranslations; 
         var ns = enumInfo.Namespace is null ? null : $"{enumInfo.Namespace}."; 
         var code = generator.Generate(enumInfo, translations); 
         if (!String.IsNullOrWhiteSpace(code)) 
            context.AddSource($"{ns}{enumInfo.Name}{generator.FileHintSuffix}.g.cs", code); 

Parsing of the JSON contents is delegated to a new method GetTranslationsByClassName. This method is not just parsing the JSON, but makes a few sanity checks and catches potential parsing errors.

					   private static Dictionary<string, IReadOnlyDictionary<string, string>>? 
        GetTranslationsByClassName(SourceProductionContext context, 
                                   ImmutableArray<string> translationsAsJson) 
      if (translationsAsJson.Length <= 0) 
         return null; 
      if (translationsAsJson.Length > 1) 
         var error = Diagnostic.Create(DemoDiagnosticsDescriptors.MultipleTranslationsFound,
         return JsonConvert.DeserializeObject<
            Dictionary<string, IReadOnlyDictionary<string, string>>>(translationsAsJson[0]); 
      catch (Exception ex) 
         var error = Diagnostic.Create(DemoDiagnosticsDescriptors.TranslationDeserializationError, 
         return null; 
The diagnostic descriptors MultipleTranslationsFound and TranslationDeserializationError are defined in DemoDiagnosticsDescriptors, which is in the same project as the Source Generator.
					public static class DemoDiagnosticsDescriptors 
   public static readonly DiagnosticDescriptor MultipleTranslationsFound 
      = new("DEMO002", 
            "Multiple translations found", 
            "Multiple translations found", 
   public static readonly DiagnosticDescriptor TranslationDeserializationError 
      = new("DEMO003", 
            "Translations could not be deserialized", 
            "Translations could not be deserialized: {0}", 



Additional files are an excellent way to generate code out of (rather static) files, like JSON or XML.

One of the use cases is translation management. Whether the translations come from an external company or are maintained by the developers themselves, a Source Generator can generate constants instead of using magic strings. That way, the translations become virtually type-safe, and there will be no issue with missing translations anymore.


Current articles, screencasts and interviews by our experts

Don’t miss any content on Angular, .NET Core, Blazor, Azure, and Kubernetes and sign up for our free monthly dev newsletter.

EN Newsletter Anmeldung (#7)
Related Articles
One of the more pragmatic ways to get going on the current AI hype, and to get some value out of it, is by leveraging semantic search. This is, in itself, a relatively simple concept: You have a bunch of documents and want to find the correct one based on a given query. The semantic part now allows you to find the correct document based on the meaning of its contents, in contrast to simply finding words or parts of words in it like we usually do with lexical search. In our last projects, we gathered some experience with search bots, and with this article, I'd love to share our insights with you.
If you previously wanted to integrate view transitions into your Angular application, this was only possible in a very cumbersome way that needed a lot of detailed knowledge about Angular internals. Now, Angular 17 introduced a feature to integrate the View Transition API with the router. In this two-part series, we will look at how to leverage the feature for route transitions and how we could use it for single-page animations.
.NET 8 brings Native AOT to ASP.NET Core, but many frameworks and libraries rely on unbound reflection internally and thus cannot support this scenario yet. This is true for ORMs, too: EF Core and Dapper will only bring full support for Native AOT in later releases. In this post, we will implement a database access layer with Sessions using the Humble Object pattern to get a similar developer experience. We will use Npgsql as a plain ADO.NET provider targeting PostgreSQL.