Move the file-system and generator-registry checks out of Schema to fix CA1506 [patch]

Schema/Models/Schema.Validation.cs

@@ -54,7 +54,7 @@
 	/// <summary>
 	/// Gets every element a generator names an output file after, with that name.
 	/// </summary>
-	private IEnumerable<(string Name, ISchemaElement Element)> GeneratedFileNames() =>
+	internal IEnumerable<(string Name, ISchemaElement Element)> GeneratedFileNames() =>
 		ClassesInternal.Select(c => (c.Name.ToString(), (ISchemaElement)c))
 			.Concat(EnumsInternal.Select(e => (e.Name.ToString(), (ISchemaElement)e)))
 			.Concat(InterfacesInternal.Select(i => (i.Name.ToString(), (ISchemaElement)i)))
@@ -614,7 +614,7 @@
 	/// A numeric default has to fit the member's type, or each component of its vector, and lie
 	/// inside its range.
 	/// </summary>
 	private static void ValidateNumberDefault(Collection<SchemaValidationIssue> issues, ISchemaMetadataCarrier carrier, NumberDefault number, BaseType type, string path, ISchemaElement element, string kind)
 	{
 		if (!type.IsNumeric && !IsVector(type))
 		{
@@ -1598,53 +1598,16 @@
 					Element = dataSource,
 				});
 			}
-			else
+			else if (CanResolvePaths)
 			{
-				ValidateDataSourceFileExists(issues, dataSource);
+				// Only possible once the schema knows its own location, since the path is relative to it.
+				// A schema built in memory has no anchor, and its data sources are left unchecked rather
+				// than resolved against whatever the working directory happens to be.
+				SchemaEnvironmentValidator.ReportMissingDataSourceFile(issues, dataSource);
 			}
 		}
 	}
 
-	/// <summary>
-	/// Reports a data source whose file is not there.
-	/// </summary>
-	/// <remarks>
-	/// Only possible once the schema knows its own location, since the path is relative to it.
-	/// A schema built in memory has no anchor, and its data sources are left unchecked rather
-	/// than resolved against whatever the working directory happens to be.
-	/// </remarks>
-	private void ValidateDataSourceFileExists(Collection<SchemaValidationIssue> issues, DataSource dataSource)
-	{
-		if (!CanResolvePaths || !dataSource.TryResolveFile(out Semantics.Paths.AbsoluteFilePath resolved))
-		{
-			return;
-		}
-
-		try
-		{
-			if (!File.Exists(resolved))
-			{
-				issues.Add(new()
-				{
-					Severity = SchemaValidationSeverity.Error,
-					Path = dataSource.Name,
-					Message = $"Data source file '{dataSource.File}' does not exist (resolved to '{resolved}').",
-					Element = dataSource,
-				});
-			}
-		}
-		catch (Exception ex) when (ex is IOException or UnauthorizedAccessException)
-		{
-			issues.Add(new()
-			{
-				Severity = SchemaValidationSeverity.Warning,
-				Path = dataSource.Name,
-				Message = $"Data source file '{dataSource.File}' could not be checked: {ex.Message}",
-				Element = dataSource,
-			});
-		}
-	}
-
 	private void ValidateCodeGenerators(Collection<SchemaValidationIssue> issues)
 	{
 		foreach (SchemaCodeGenerator codeGenerator in CodeGeneratorsInternal)
@@ -1682,40 +1645,7 @@
 					Element = codeGenerator,
 				});
 			}
-			ReportReservedNames(issues, codeGenerator);
-		}
-	}
-
-	/// <summary>
-	/// Reports elements named after a file the configured generator writes for itself.
-	/// </summary>
-	/// <remarks>
-	/// The element's file and the generator's own would be the same entry in the output, so one
-	/// replaces the other on every file system. Matched case-insensitively for the reason
-	/// <see cref="ReportNamesDifferingOnlyInCase"/> gives.
-	/// </remarks>
-	private void ReportReservedNames(Collection<SchemaValidationIssue> issues, SchemaCodeGenerator codeGenerator)
-	{
-		if (!Generation.SchemaGenerator.TryGetGenerator(codeGenerator.Language, out Generation.ISchemaCodeGenerator? generator))
-		{
-			return;
-		}
-
-		HashSet<string> reserved = new(generator.ReservedNames(codeGenerator), StringComparer.OrdinalIgnoreCase);
-		if (reserved.Count == 0)
-		{
-			return;
-		}
-
-		foreach ((string name, ISchemaElement element) in GeneratedFileNames().Where(entry => reserved.Contains(entry.Name)))
-		{
-			issues.Add(new()
-			{
-				Severity = SchemaValidationSeverity.Error,
-				Path = name,
-				Message = $"'{name}' is the name of a file code generator '{codeGenerator.Name}' writes for itself, so the two would overwrite each other.",
-				Element = element,
-			});
+			SchemaEnvironmentValidator.ReportReservedNames(issues, codeGenerator, GeneratedFileNames());
 		}
 	}
 }

Schema/Models/SchemaEnvironmentValidator.cs

@@ -0,0 +1,82 @@
+// Copyright (c) 2023-2026 ktsu-dev contributors
+
+namespace ktsu.Schema.Models;
+
+using System.Collections.ObjectModel;
+using ktsu.Schema.Generation;
+using ktsu.Semantics.Paths;
+
+/// <summary>
+/// The schema checks that consult something outside the schema: the file system and the generator registry.
+/// </summary>
+/// <remarks>
+/// Kept out of <see cref="Schema"/> because every type the schema reaches counts against its
+/// class-coupling budget (CA1506), and these checks reach types nothing else in it needs. It also keeps
+/// the schema model from depending on the generators, which already depend on it.
+/// </remarks>
+internal static class SchemaEnvironmentValidator
+{
+	/// <summary>
+	/// Reports a data source whose file is not there.
+	/// </summary>
+	/// <param name="issues">Where to add the issues.</param>
+	/// <param name="dataSource">A data source of a schema that knows its own location.</param>
+	internal static void ReportMissingDataSourceFile(Collection<SchemaValidationIssue> issues, DataSource dataSource)
+	{
+		if (!dataSource.TryResolveFile(out AbsoluteFilePath resolved))
+		{
+			return;
+		}
+
+		// File.Exists answers false rather than throwing for a path it can't read, so there is nothing to catch.
+		if (!File.Exists(resolved))
+		{
+			issues.Add(new()
+			{
+				Severity = SchemaValidationSeverity.Error,
+				Path = dataSource.Name,
+				Message = $"Data source file '{dataSource.File}' does not exist (resolved to '{resolved}').",
+				Element = dataSource,
+			});
+		}
+	}
+
+	/// <summary>
+	/// Reports each element whose generated file would collide with one the generator writes for itself.
+	/// </summary>
+	/// <remarks>
+	/// The element's file and the generator's own would be the same entry in the output, so one
+	/// replaces the other on every file system. Matched case-insensitively, because Windows and macOS
+	/// file systems ignore case.
+	/// </remarks>
+	/// <param name="issues">Where to add the issues.</param>
+	/// <param name="codeGenerator">The configured generator.</param>
+	/// <param name="generatedFileNames">Every element a generator names an output file after, with that name.</param>
+	internal static void ReportReservedNames(
+		Collection<SchemaValidationIssue> issues,
+		SchemaCodeGenerator codeGenerator,
+		IEnumerable<(string Name, ISchemaElement Element)> generatedFileNames)
+	{
+		if (!SchemaGenerator.TryGetGenerator(codeGenerator.Language, out ISchemaCodeGenerator? generator))
+		{
+			return;
+		}
+
+		HashSet<string> reserved = new(generator.ReservedNames(codeGenerator), StringComparer.OrdinalIgnoreCase);
+		if (reserved.Count == 0)
+		{
+			return;
+		}
+
+		foreach ((string name, ISchemaElement element) in generatedFileNames.Where(entry => reserved.Contains(entry.Name)))
+		{
+			issues.Add(new()
+			{
+				Severity = SchemaValidationSeverity.Error,
+				Path = name,
+				Message = $"'{name}' is the name of a file code generator '{codeGenerator.Name}' writes for itself, so the two would overwrite each other.",
+				Element = element,
+			});
+		}
+	}
+}