Add an action that validates if assembler.yml content-source mapping is fully published to the link registry (#1300)

Author: Mpdreamz

actions/assembler-config-validate/action.yml

@@ -0,0 +1,12 @@
+name: 'Documentation Assembler Configuration Validator'
+description: 'Ensures the assembler configuration is valid'
+
+branding:
+  icon: 'filter'
+  color: 'blue'
+
+runs:
+  using: 'docker'
+  image: "docker://ghcr.io/elastic/docs-assembler:edge"
+  env:
+    INPUT_COMMAND: "content-source validate"

docs-builder.sln

@@ -94,6 +94,11 @@ Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "assembler-match", "assemble
 		actions\assembler-match\action.yml = actions\assembler-match\action.yml
 	EndProjectSection
 EndProject
+Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "assembler-config-validate", "assembler-config-validate", "{E20FEEF9-1D1A-4CDA-A546-7FDC573BE399}"
+	ProjectSection(SolutionItems) = preProject
+		actions\assembler-config-validate\action.yml = actions\assembler-config-validate\action.yml
+	EndProjectSection
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU
@@ -178,5 +183,6 @@ Global
 		{059E787F-85C1-43BE-9DD6-CE319E106383} = {BE6011CC-1200-4957-B01F-FCCA10C5CF5A}
 		{7D36DDDA-9E0B-4D2C-8033-5D62FF8B6166} = {059E787F-85C1-43BE-9DD6-CE319E106383}
 		{FB1C1954-D8E2-4745-BA62-04DD82FB4792} = {245023D2-D3CA-47B9-831D-DAB91A2FFDC7}
+		{E20FEEF9-1D1A-4CDA-A546-7FDC573BE399} = {245023D2-D3CA-47B9-831D-DAB91A2FFDC7}
 	EndGlobalSection
 EndGlobal

src/Elastic.Markdown/Links/CrossLinks/CrossLinkFetcher.cs

@@ -33,6 +33,7 @@ public record FetchedCrossLinks
 
 public abstract class CrossLinkFetcher(ILoggerFactory logger) : IDisposable
 {
+	public const string RegistryUrl = $"https://elastic-docs-link-index.s3.us-east-2.amazonaws.com/link-index.json";
 	private readonly ILogger _logger = logger.CreateLogger(nameof(CrossLinkFetcher));
 	private readonly HttpClient _client = new();
 	private LinkReferenceRegistry? _linkIndex;
@@ -42,16 +43,16 @@ public static LinkReference Deserialize(string json) =>
 
 	public abstract Task<FetchedCrossLinks> Fetch(Cancel ctx);
 
-	protected async Task<LinkReferenceRegistry> FetchLinkIndex(Cancel ctx)
+	public async Task<LinkReferenceRegistry> FetchLinkIndex(Cancel ctx)
 	{
 		if (_linkIndex is not null)
 		{
 			_logger.LogTrace("Using cached link index");
 			return _linkIndex;
 		}
-		var url = $"https://elastic-docs-link-index.s3.us-east-2.amazonaws.com/link-index.json";
-		_logger.LogInformation("Fetching {Url}", url);
-		var json = await _client.GetStringAsync(url, ctx);
+
+		_logger.LogInformation("Fetching {Url}", RegistryUrl);
+		var json = await _client.GetStringAsync(RegistryUrl, ctx);
 		_linkIndex = LinkReferenceRegistry.Deserialize(json);
 		return _linkIndex;
 	}

src/tooling/docs-assembler/Building/AssemblerCrossLinkFetcher.cs

@@ -11,7 +11,8 @@
 
 namespace Documentation.Assembler.Building;
 
-public class AssemblerCrossLinkFetcher(ILoggerFactory logger, AssemblyConfiguration configuration, PublishEnvironment publishEnvironment) : CrossLinkFetcher(logger)
+public class AssemblerCrossLinkFetcher(ILoggerFactory logger, AssemblyConfiguration configuration, PublishEnvironment publishEnvironment)
+	: CrossLinkFetcher(logger)
 {
 	public override async Task<FetchedCrossLinks> Fetch(Cancel ctx)
 	{

src/tooling/docs-assembler/Cli/ContentSourceCommands.cs

@@ -6,8 +6,10 @@
 using System.IO.Abstractions;
 using Actions.Core.Services;
 using ConsoleAppFramework;
+using Documentation.Assembler.Building;
 using Elastic.Documentation.Configuration.Assembler;
 using Elastic.Documentation.Tooling.Diagnostics.Console;
+using Elastic.Markdown.Links.CrossLinks;
 using Microsoft.Extensions.Logging;
 
 namespace Documentation.Assembler.Cli;
@@ -22,6 +24,54 @@ private void AssignOutputLogger()
 		ConsoleApp.LogError = msg => log.LogError(msg);
 	}
 
+	[Command("validate")]
+	public async Task<int> Validate(Cancel ctx = default)
+	{
+		AssignOutputLogger();
+		await using var collector = new ConsoleDiagnosticsCollector(logFactory, githubActionsService)
+		{
+			NoHints = true
+		};
+
+		_ = collector.StartAsync(ctx);
+
+		// environment does not matter to check the configuration, defaulting to dev
+		var context = new AssembleContext("dev", collector, new FileSystem(), new FileSystem(), null, null)
+		{
+			Force = false,
+			AllowIndexing = false
+		};
+		var fetcher = new AssemblerCrossLinkFetcher(logFactory, context.Configuration, context.Environment);
+		var links = await fetcher.FetchLinkIndex(ctx);
+		var repositories = context.Configuration.ReferenceRepositories.Values.Concat<Repository>([context.Configuration.Narrative]).ToList();
+
+		foreach (var repository in repositories)
+		{
+			if (!links.Repositories.TryGetValue(repository.Name, out var registryMapping))
+			{
+				collector.EmitError(context.ConfigurationPath, $"'{repository}' does not exist in {CrossLinkFetcher.RegistryUrl}");
+				continue;
+			}
+
+			var current = repository.GetBranch(ContentSource.Current);
+			var next = repository.GetBranch(ContentSource.Next);
+			if (!registryMapping.TryGetValue(next, out _))
+			{
+				collector.EmitError(context.ConfigurationPath,
+					$"'{repository.Name}' has not yet published links.json for configured 'next' content source: '{next}' see  {CrossLinkFetcher.RegistryUrl}");
+			}
+			if (!registryMapping.TryGetValue(current, out _))
+			{
+				collector.EmitError(context.ConfigurationPath,
+					$"'{repository.Name}' has not yet published links.json for configured 'current' content source: '{current}' see  {CrossLinkFetcher.RegistryUrl}");
+			}
+		}
+
+
+		await collector.StopAsync(ctx);
+		return collector.Errors == 0 ? 0 : 1;
+	}
+
 	/// <summary>  </summary>
 	/// <param name="repository"></param>
 	/// <param name="branchOrTag"></param>