git.stella-ops.org/tests/reachability/StellaOps.Reachability.FixtureTests/CorpusFixtureTests.cs

using System.Security.Cryptography;
using System.Text.Json;
using FluentAssertions;
using Xunit;

namespace StellaOps.Reachability.FixtureTests;

public class CorpusFixtureTests
{
    private static readonly string RepoRoot = ReachbenchFixtureTests.LocateRepoRoot();
    private static readonly string CorpusRoot = Path.Combine(RepoRoot, "tests", "reachability", "corpus");

    [Fact]
    public void ManifestExistsAndIsDeterministic()
    {
        var manifestPath = Path.Combine(CorpusRoot, "manifest.json");
        File.Exists(manifestPath).Should().BeTrue("corpus manifest should exist");

        using var stream = File.OpenRead(manifestPath);
        using var doc = JsonDocument.Parse(stream);
        doc.RootElement.ValueKind.Should().Be(JsonValueKind.Array);
    }

    [Fact]
    public void CorpusEntriesMatchManifestHashes()
    {
        var manifestPath = Path.Combine(CorpusRoot, "manifest.json");
        var manifest = JsonDocument.Parse(File.ReadAllBytes(manifestPath)).RootElement.EnumerateArray().ToArray();

        manifest.Should().NotBeEmpty("corpus manifest must have entries");

        foreach (var entry in manifest)
        {
            var id = entry.GetProperty("id").GetString();
            var language = entry.GetProperty("language").GetString();
            var files = entry.GetProperty("files");

            id.Should().NotBeNullOrEmpty();
            language.Should().NotBeNullOrEmpty();

            var caseDir = Path.Combine(CorpusRoot, language!, id!);
            Directory.Exists(caseDir).Should().BeTrue($"case folder missing: {caseDir}");

            foreach (var fileProp in files.EnumerateObject())
            {
                var filename = fileProp.Name;
                var expectedHash = fileProp.Value.GetString();
                File.Exists(Path.Combine(caseDir, filename)).Should().BeTrue($"{id} missing {filename}");

                var actualHash = BitConverter.ToString(SHA256.HashData(File.ReadAllBytes(Path.Combine(caseDir, filename)))).Replace("-", "").ToLowerInvariant();
                actualHash.Should().Be(expectedHash, $"{id} hash mismatch for {filename}");
            }
        }
    }

    [Fact]
    public void GroundTruthFilesContainRequiredFields()
    {
        var manifestPath = Path.Combine(CorpusRoot, "manifest.json");
        var manifest = JsonDocument.Parse(File.ReadAllBytes(manifestPath)).RootElement.EnumerateArray().ToArray();
        const string expectedSchemaVersion = "reachbench.reachgraph.truth/v1";
        var allowedVariants = new[] { "reachable", "unreachable" };

        foreach (var entry in manifest)
        {
            var id = entry.GetProperty("id").GetString()!;
            var language = entry.GetProperty("language").GetString()!;
            var truthPath = Path.Combine(CorpusRoot, language, id, "ground-truth.json");
            File.Exists(truthPath).Should().BeTrue($"{id} missing ground-truth.json");

            using var truthDoc = JsonDocument.Parse(File.ReadAllBytes(truthPath));
            truthDoc.RootElement.GetProperty("schema_version").GetString().Should().Be(expectedSchemaVersion, $"{id} ground-truth schema_version mismatch");
            truthDoc.RootElement.GetProperty("case_id").GetString().Should().Be(id, $"{id} ground-truth case_id must match manifest id");

            var variant = truthDoc.RootElement.GetProperty("variant").GetString();
            variant.Should().NotBeNullOrWhiteSpace($"{id} ground-truth must set variant");
            allowedVariants.Should().Contain(variant!, $"{id} variant must be reachable|unreachable");

            truthDoc.RootElement.TryGetProperty("paths", out var pathsProp).Should().BeTrue($"{id} ground-truth must include paths");
            pathsProp.ValueKind.Should().Be(JsonValueKind.Array, $"{id} paths must be an array");

            if (string.Equals(variant, "reachable", StringComparison.Ordinal))
            {
                pathsProp.GetArrayLength().Should().BeGreaterThan(0, $"{id} reachable ground-truth should include at least one path");
            }

            foreach (var path in pathsProp.EnumerateArray())
            {
                path.ValueKind.Should().Be(JsonValueKind.Array, $"{id} each path must be an array");
                path.GetArrayLength().Should().BeGreaterThan(0, $"{id} each path must contain at least one symbol");

                foreach (var segment in path.EnumerateArray())
                {
                    segment.ValueKind.Should().Be(JsonValueKind.String, $"{id} path segments must be strings");
                    segment.GetString().Should().NotBeNullOrWhiteSpace($"{id} path segments must be non-empty strings");
                }
            }
        }
    }
}