bikoret/vav_pipeline/test/doorpost_test.exs

defmodule VavPipeline.DoorpostTest do
  @moduledoc """
  Tests for the doorpost-text identity verification (vehaya.md
  §3.3, §6 item 3). Devarim 4:2's lo tosif enforced as a
  computable property: any addition or subtraction across the
  category-grade specification files produces a hash mismatch
  detectable at /kodesh/specification_identity.

  These tests verify behavior at three layers:
    1. The pure functions (compute_identity, manifest, files)
    2. The verify_identity gate against a known-good hash
    3. Drift detection against a synthetic spec set in a temp dir
       (uses application env override for repo_root)
  """

  use ExUnit.Case, async: false

  alias VavPipeline.Doorpost

  describe "files/0" do
    test "returns the full list of category-grade spec files" do
      files = Doorpost.files()
      assert is_list(files)
      # Sanity: includes the load-bearing artifacts
      assert "BUILD_AZAZEL.md" in files
      assert "BUILD_KORBAN.md" in files
      assert "talmudic_engine/src/halt/avodah_zarah.pl" in files
      assert "talmudic_engine/src/halt/gilui_arayot.pl" in files
      assert "talmudic_engine/src/halt/shfichut_damim.pl" in files
      assert "specs/eish_zarah_reading.md" in files
    end

    test "explicitly excludes operations files (pekudei, terumah)" do
      files = Doorpost.files()
      refute "specs/bikoret_pekudei.md" in files
      refute "specs/bikoret_terumah.md" in files
    end

    test "list is sorted (canonical order independent of filesystem)" do
      files = Doorpost.files()
      assert files == Enum.sort(files)
    end

    test "no file appears twice" do
      files = Doorpost.files()
      assert length(files) == length(Enum.uniq(files))
    end
  end

  describe "manifest/0" do
    test "returns one entry per file in @spec_files" do
      manifest = Doorpost.manifest()
      assert length(manifest) == length(Doorpost.files())

      # Each entry is %{path, hash}
      Enum.each(manifest, fn entry ->
        assert Map.has_key?(entry, :path)
        assert Map.has_key?(entry, :hash)
      end)
    end

    test "manifest paths match @spec_files order" do
      manifest_paths = Enum.map(Doorpost.manifest(), & &1.path)
      assert manifest_paths == Doorpost.files()
    end

    test "each per-file hash is a 64-character hex SHA-256 (or :missing)" do
      Enum.each(Doorpost.manifest(), fn entry ->
        case entry.hash do
          :missing ->
            :ok

          hash when is_binary(hash) ->
            assert String.length(hash) == 64
            assert String.match?(hash, ~r/^[0-9a-f]{64}$/)
        end
      end)
    end
  end

  describe "compute_identity/0 against the live repo" do
    test "produces a stable 64-char hex hash" do
      hash = Doorpost.compute_identity()
      assert is_binary(hash), "compute_identity returned: #{inspect(hash)}"
      assert String.length(hash) == 64
      assert String.match?(hash, ~r/^[0-9a-f]{64}$/)
    end

    test "is deterministic — two calls return the same hash" do
      hash1 = Doorpost.compute_identity()
      hash2 = Doorpost.compute_identity()
      assert hash1 == hash2
    end
  end

  describe "verify_identity/0 against the live repo" do
    test "returns :ok when the live spec set matches the baked hash" do
      assert :ok = Doorpost.verify_identity()
    end
  end

  # ============================================================
  # Drift-detection tests against a synthetic spec set
  # ============================================================
  # These tests build a fake spec set in a temp dir, point Doorpost
  # at it via the application env, and verify drift detection works
  # end-to-end. We can't override @spec_files (compile-time list);
  # instead we set up a temp dir that has THE SAME files as the
  # live spec set and verify the drift behaviors against THAT.

  describe "drift detection (synthetic spec set in temp dir)" do
    setup do
      # Build a temp dir mirror of the spec files
      temp = System.tmp_dir!() |> Path.join("doorpost-test-#{:erlang.unique_integer([:positive])}")
      File.mkdir_p!(temp)

      original_root = Application.get_env(:vav_pipeline, :doorpost_repo_root)

      Enum.each(Doorpost.files(), fn rel ->
        src = Path.join(Doorpost.repo_root(), rel)
        dst = Path.join(temp, rel)
        File.mkdir_p!(Path.dirname(dst))
        if File.exists?(src), do: File.cp!(src, dst)
      end)

      on_exit(fn ->
        File.rm_rf!(temp)
        # NOTE: cannot restore Application env override because
        # @repo_root is read at compile time, not runtime. We
        # work around by computing identity directly via a
        # helper fn pointed at the temp dir — see below.
        if original_root do
          Application.put_env(:vav_pipeline, :doorpost_repo_root, original_root)
        end
      end)

      %{temp: temp}
    end

    # Direct-compute helper that mirrors Doorpost.compute_identity
    # but takes an explicit root. Used because @repo_root is baked
    # at compile time.
    defp compute_at(temp_root) do
      items =
        Enum.map(Doorpost.files(), fn rel ->
          full_path = Path.join(temp_root, rel)

          case File.read(full_path) do
            {:ok, contents} ->
              {rel, :crypto.hash(:sha256, contents) |> Base.encode16(case: :lower)}

            {:error, _} ->
              {rel, :missing}
          end
        end)

      case Enum.find(items, fn {_, h} -> h == :missing end) do
        nil ->
          combined =
            items
            |> Enum.map(fn {p, h} -> "#{p}\t#{h}" end)
            |> Enum.join("\n")

          :crypto.hash(:sha256, combined) |> Base.encode16(case: :lower)

        {missing, _} ->
          {:error, "missing spec file: #{missing}"}
      end
    end

    test "fresh mirror of the spec set produces the same hash as live", %{temp: temp} do
      assert compute_at(temp) == Doorpost.compute_identity()
    end

    test "modifying ANY one file changes the hash", %{temp: temp} do
      original = compute_at(temp)

      # Append a single byte to one file
      target = Path.join(temp, "talmudic_engine/src/halt/avodah_zarah.pl")
      File.write!(target, File.read!(target) <> "\n")

      drifted = compute_at(temp)
      refute original == drifted
    end

    test "removing any one file fails compute_identity", %{temp: temp} do
      target = Path.join(temp, "talmudic_engine/src/halt/avodah_zarah.pl")
      File.rm!(target)

      assert {:error, "missing spec file: " <> _} = compute_at(temp)
    end

    test "modifying a different file produces a different hash than modifying the first", %{
      temp: temp
    } do
      File.write!(
        Path.join(temp, "talmudic_engine/src/halt/avodah_zarah.pl"),
        File.read!(Path.join(temp, "talmudic_engine/src/halt/avodah_zarah.pl")) <> "\n"
      )

      hash_a = compute_at(temp)

      # Restore, then drift a different file
      target_b = Path.join(temp, "specs/eish_zarah_reading.md")
      File.write!(target_b, File.read!(target_b) <> "\n")

      hash_b = compute_at(temp)

      # Two distinct drifts → two distinct hashes
      refute hash_a == hash_b
    end
  end

  describe "/kodesh/specification_identity HTTP endpoint" do
    # Plug.Test integration. The endpoint is under /kodesh/* but is
    # currently NOT auth-gated (the kodesh tier is mTLS in production
    # via an upstream gateway, not in the application).

    import Plug.Test
    alias VavPipeline.BigdeiKodesh.Router

    @opts Router.init([])

    test "returns 200 when verify_identity is :ok" do
      conn = conn(:get, "/kodesh/specification_identity") |> Router.call(@opts)

      assert conn.status == 200
      body = Jason.decode!(conn.resp_body)
      assert body["status"] == "verified"
      assert is_binary(body["computed_hash"])
      assert is_binary(body["expected_hash"])
      assert body["computed_hash"] == body["expected_hash"]
      assert is_integer(body["file_count"])
      assert body["file_count"] > 0
      assert is_list(body["manifest"])
      assert is_list(body["torah_basis"])
      assert "Devarim 4:2 (lo tosif)" in body["torah_basis"]
    end

    test "manifest in response carries per-file hashes" do
      conn = conn(:get, "/kodesh/specification_identity") |> Router.call(@opts)
      body = Jason.decode!(conn.resp_body)

      manifest = body["manifest"]
      assert length(manifest) == length(Doorpost.files())

      Enum.each(manifest, fn entry ->
        assert Map.has_key?(entry, "path")
        assert Map.has_key?(entry, "hash")
      end)
    end
  end
end