w4c-workflows-api/Services/PerTenantWorkflowSource.cs

using System.Diagnostics;
using Npgsql;
using NpgsqlTypes;

namespace w4c_workflows.Services;

/// <summary>
/// Factory for creating <see cref="IWorkflowSource"/> instances scoped to a
/// specific tenant. The caller provides the tenant ID explicitly (from auth
/// middleware); the factory resolves directories and returns a source bound
/// to that tenant's workflow directory.
///
/// <b>Two modes</b> (automatic, config-driven):
/// <list type="number">
///   <item><b>Forgejo-backed</b> (preferred): when <c>Forgejo:AdminToken</c> is
///   configured, each tenant's workflow files live in a private Forgejo repo
///   (<c>workflows-{tenantId}</c>). The factory clones/pulls via
///   <see cref="ForgejoWorkflowRepoService"/> and returns a
///   <see cref="PerTenantWorkflowSource"/> backed by the local clone.
///   This gives tenants real git history, branching, and remote sync.</item>
///   <item><b>Filesystem-only</b> (fallback): when Forgejo is not configured,
///   uses plain directory copies seeded from the shared template dir.
///   Preserves the original behavior for dev/self-hosted setups.</item>
/// </list>
///
/// Tenant isolation is <b>mandatory</b>: <c>WorkflowSource:CopiesRoot</c> must
/// be set in configuration. The factory throws at construction time if it is
/// missing, preventing any request from serving cross-tenant workflows.
///
/// Registered as a <b>Singleton</b> (shared config, no per-request state);
/// the created sources are lightweight and backed by filesystem operations.
/// </summary>
public sealed class WorkflowSourceFactory
{
    private readonly ILoggerFactory _loggers;
    private readonly string _copiesRoot;
    private readonly ForgejoWorkflowRepoService? _forgejo;
    private readonly NpgsqlDataSource? _ds;

    public WorkflowSourceFactory(IConfiguration config, IWebHostEnvironment env, ILoggerFactory loggers,
        NpgsqlDataSource? ds = null)
    {
        _loggers = loggers;
        _ds = ds;

        _copiesRoot = Path.GetFullPath(
            config["WorkflowSource:CopiesRoot"]
            ?? throw new InvalidOperationException(
                "WorkflowSource:CopiesRoot is not configured. " +
                "Per-tenant filesystem isolation is required — set it to a writable " +
                "directory path (e.g. \"/data/workflow-tenants\" or \".data/workflow-tenants\")."));

        // Create the Forgejo service when admin provisioning is configured.
        var forgejoLogger = loggers.CreateLogger<ForgejoWorkflowRepoService>();
        var forgejoSvc = new ForgejoWorkflowRepoService(config, forgejoLogger);
        _forgejo = forgejoSvc.IsConfigured ? forgejoSvc : null;

        if (_forgejo != null)
            _loggers.CreateLogger<WorkflowSourceFactory>()
                .LogInformation("Workflow source: Forgejo-backed mode (owner={Owner})",
                    config["Forgejo:Owner"] ?? config["Forgejo:WorkflowRepoOwner"]);
        else
            _loggers.CreateLogger<WorkflowSourceFactory>()
                .LogInformation("Workflow source: filesystem-only mode (no Forgejo admin token)");
    }

    /// <summary>True when Forgejo-backed mode is active.</summary>
    public bool IsForgejoBacked => _forgejo != null;

    /// <summary>The Forgejo repo service (null when in filesystem-only mode).</summary>
    public ForgejoWorkflowRepoService? Forgejo => _forgejo;

    /// <summary>
    /// Creates a per-tenant workflow source.
    ///
    /// In Forgejo-backed mode: ensures the local clone exists (creating the
    /// Forgejo repo if needed), then returns a source backed by the clone.
    ///
    /// In filesystem-only mode: the tenant directory
    /// <c>{CopiesRoot}/{sanitizedTenantId}/</c> is created empty on first access
    /// (never seeded from shared templates — workflows are strictly per-tenant).
    /// </summary>
    public IWorkflowSource Create(string tenantId)
    {
        var tenantDir = TenantSourceDir(tenantId);
        var logger = _loggers.CreateLogger<PerTenantWorkflowSource>();
        return new PerTenantWorkflowSource(tenantId, tenantDir, logger);
    }

    /// <summary>
    /// Resolves the absolute on-disk directory that holds a tenant's workflow
    /// files. In Forgejo-backed mode this is the per-user clone dir; in
    /// filesystem-only mode it is <c>{CopiesRoot}/{sanitizedTenantId}/</c>.
    /// Used by the workflow-file CRUD controller so edits always land in the same
    /// place <see cref="IWorkflowSource"/> reads from.
    /// </summary>
    public string ResolveTenantSourceDir(string tenantId, string? forgejoLogin = null, string? repoName = null)
    {
        if (_forgejo != null && !string.IsNullOrWhiteSpace(forgejoLogin))
            return _forgejo.TenantCloneDir(tenantId, forgejoLogin, repoName);
        return TenantSourceDir(tenantId);
    }

    /// <summary>The per-tenant filesystem source directory (not Forgejo-backed).</summary>
    public string TenantSourceDir(string tenantId)
        => Path.GetFullPath(Path.Combine(_copiesRoot, Sanitize(tenantId)));

    /// <summary>
    /// Creates a per-tenant workflow source asynchronously. In Forgejo-backed
    /// mode this ensures the local clone exists (cloning from Forgejo if needed).
    /// In filesystem-only mode, delegates to <see cref="Create"/>.
    /// </summary>
    public async Task<IWorkflowSource> CreateAsync(string tenantId, string? repoName = null, CancellationToken ct = default)
    {
        if (_forgejo == null)
            return Create(tenantId);

        // Resolve the user's Forgejo login from the tenant ID (forgejo_id).
        // The repo lives under the user's own account: {login}/{workflowRepoName}.
        var login = await ResolveForgejoLoginAsync(tenantId, ct);
        if (string.IsNullOrEmpty(login))
        {
            _loggers.CreateLogger<WorkflowSourceFactory>()
                .LogWarning("Could not resolve Forgejo login for tenant {TenantId}, falling back to filesystem",
                    tenantId);
            return Create(tenantId);
        }

        // Ensure the Forgejo repo exists and is cloned locally.
        var cloneDir = await _forgejo.EnsureCloneAsync(tenantId, login, repoName, ct);

        // Pull latest changes before reading.
        await _forgejo.PullAsync(tenantId, login, ct);

        var logger = _loggers.CreateLogger<PerTenantWorkflowSource>();
        return new PerTenantWorkflowSource(tenantId, cloneDir, logger);
    }

    /// <summary>
    /// Resolves the Forgejo login for a tenant (forgejo_id) from the auth_users table.
    /// </summary>
    public async Task<string?> ResolveForgejoLoginAsync(string tenantId, CancellationToken ct)
    {
        if (_ds == null || !long.TryParse(tenantId, out var forgejoId))
            return null;

        try
        {
            await using var conn = await _ds.OpenConnectionAsync(ct);
            await using var cmd = conn.CreateCommand();
            cmd.CommandTimeout = 3;
            cmd.CommandText = "SELECT login FROM auth_users WHERE forgejo_id = @fid LIMIT 1";
            cmd.Parameters.Add(new NpgsqlParameter("@fid", NpgsqlDbType.Bigint) { Value = forgejoId });
            var result = await cmd.ExecuteScalarAsync(ct);
            return result as string;
        }
        catch (Exception ex)
        {
            _loggers.CreateLogger<WorkflowSourceFactory>()
                .LogWarning(ex, "Could not resolve Forgejo login for tenant {TenantId}", tenantId);
            return null;
        }
    }

    private static string Sanitize(string id)
    {
        if (string.IsNullOrWhiteSpace(id))
            return "_";
        var sb = new System.Text.StringBuilder(id.Length);
        foreach (var ch in id)
            sb.Append(char.IsLetterOrDigit(ch) || ch == '-' || ch == '_' ? ch : '_');
        var result = sb.ToString();
        return result.Length > 120 ? result[..120] : result;
    }
}

/// <summary>
/// Per-tenant workflow YAML source with directory isolation. Each tenant gets
/// its own directory at <c>{CopiesRoot}/{tenantId}/</c>, initialized from the
/// shared template directory on first access.
/// </summary>
public sealed class PerTenantWorkflowSource : IWorkflowSource
{
    private readonly string _tenantDir;
    private readonly ILogger<PerTenantWorkflowSource> _logger;

    public PerTenantWorkflowSource(string tenantId, string tenantDir, ILogger<PerTenantWorkflowSource> logger)
    {
        _tenantDir = tenantDir;
        _logger = logger;
        _logger.LogInformation(
            "Per-tenant workflow source: tenant={TenantId} dir={Dir}",
            tenantId, _tenantDir);
    }

    public Task<IReadOnlyList<WorkflowFile>> ListAsync(CancellationToken ct)
    {
        ct.ThrowIfCancellationRequested();
        EnsureInitialized();
        return Task.FromResult(ListYamlFiles(_tenantDir));
    }

    public async Task<string> ReadAsync(string path, CancellationToken ct)
    {
        EnsureInitialized();
        return await File.ReadAllTextAsync(
            Path.Combine(_tenantDir, path.Replace('/', Path.DirectorySeparatorChar)), ct);
    }

    public WorkflowSourceState GetState()
    {
        EnsureInitialized();
        var head = RunGit(_tenantDir, "rev-parse", "HEAD");
        var status = RunGit(_tenantDir, "status", "--porcelain");
        return new WorkflowSourceState(
            string.IsNullOrWhiteSpace(head) ? null : head,
            !string.IsNullOrEmpty(status));
    }

    /// <summary>
    /// Ensures the tenant directory exists. Workflows are strictly per-tenant:
    /// we never copy shared example templates into a tenant's directory, so a
    /// tenant sees only the definitions that belong to them.
    /// </summary>
    private void EnsureInitialized()
    {
        if (Directory.Exists(_tenantDir))
            return;

        _logger.LogInformation("Initializing (empty) tenant workflow directory {Dir}", _tenantDir);
        try
        {
            Directory.CreateDirectory(_tenantDir);
        }
        catch (Exception ex)
        {
            throw new InvalidOperationException(
                $"Cannot create tenant workflow directory '{_tenantDir}'. " +
                "Check that WorkflowSource:CopiesRoot points to a writable location " +
                $"and that the process has filesystem permissions. {ex.Message}", ex);
        }
    }

    private static IReadOnlyList<WorkflowFile> ListYamlFiles(string dir)
    {
        var files = new List<WorkflowFile>();
        if (!Directory.Exists(dir))
            return files;

        foreach (var full in Directory.EnumerateFiles(dir, "*", SearchOption.AllDirectories))
        {
            var ext = Path.GetExtension(full);
            if (ext is not (".yaml" or ".yml"))
                continue;
            // Skip .git
            var rel = Path.GetRelativePath(dir, full);
            if (rel.StartsWith(".git", StringComparison.OrdinalIgnoreCase))
                continue;
            files.Add(new WorkflowFile(rel.Replace(Path.DirectorySeparatorChar, '/')));
        }

        return files.OrderBy(f => f.Path, StringComparer.Ordinal).ToList();
    }

    private string? RunGit(string workDir, params string[] args)
    {
        try
        {
            var psi = new ProcessStartInfo("git")
            {
                WorkingDirectory = workDir,
                RedirectStandardOutput = true,
                RedirectStandardError = true,
                UseShellExecute = false,
            };
            foreach (var arg in args)
                psi.ArgumentList.Add(arg);

            using var process = Process.Start(psi);
            if (process == null)
                return null;

            var output = process.StandardOutput.ReadToEnd();
            if (!process.WaitForExit(5000))
            {
                process.Kill(entireProcessTree: true);
                return null;
            }

            return process.ExitCode == 0 ? output.Trim() : null;
        }
        catch (Exception ex)
        {
            _logger.LogDebug(ex, "git invocation failed in {Dir}", workDir);
            return null;
        }
    }
}