orchestra/internal/herdr/tmux_test.go

package herdr

import (
	"context"
	"errors"
	"fmt"
	"os"
	"os/exec"
	"path/filepath"
	"strings"
	"testing"
	"time"
)

func TestTmuxBackendStartsCapturesPromptsAndKillsClaude(t *testing.T) {
	if testing.Short() {
		t.Skip("requires tmux")
	}
	dir := t.TempDir()
	harness := filepath.Join(dir, "fake-claude")
	script := "#!/bin/sh\nprintf '❯ ready\\n'\nwhile IFS= read -r line; do printf 'GOT:%s\\n' \"$line\"; done\n"
	if err := os.WriteFile(harness, []byte(script), 0o755); err != nil {
		t.Fatal(err)
	}
	b := NewTmuxBackend(filepath.Join(t.TempDir(), "tmux.sock"), harness)
	if err := b.Check(context.Background()); err != nil {
		t.Fatal(err)
	}
	s, err := b.StartAgent(context.Background(), dir, dir, "", "claude", "tmux-backend-test")
	if err != nil {
		t.Fatal(err)
	}
	t.Cleanup(func() { _ = b.Kill(context.Background(), s) })
	if s.Worktree != dir || s.Harness != "claude" || !strings.Contains(s.PaneID, ":") || !strings.Contains(s.PaneID, ".") {
		t.Fatalf("unexpected session: %+v", s)
	}
	if err := b.Prompt(context.Background(), s.PaneID, "hello from Orchestra", 0); err != nil {
		t.Fatal(err)
	}
	deadline := time.Now().Add(2 * time.Second)
	for {
		capture, err := b.PaneCapture(context.Background(), s, "recent")
		if err != nil {
			t.Fatal(err)
		}
		if strings.Contains(capture, "GOT:hello from Orchestra") {
			break
		}
		if time.Now().After(deadline) {
			t.Fatalf("prompt was not captured: %q", capture)
		}
		time.Sleep(20 * time.Millisecond)
	}
	for _, line := range []string{"/clear", "@HANDOFF.md"} {
		if err := b.SendText(context.Background(), s, line); err != nil {
			t.Fatal(err)
		}
		if err := b.SendKeys(context.Background(), s, []string{"ENTER"}); err != nil {
			t.Fatal(err)
		}
	}
	deadline = time.Now().Add(2 * time.Second)
	for {
		capture, err := b.PaneCapture(context.Background(), s, "recent")
		if err != nil {
			t.Fatal(err)
		}
		if strings.Contains(capture, "GOT:/clear") && strings.Contains(capture, "GOT:@HANDOFF.md") {
			break
		}
		if time.Now().After(deadline) {
			t.Fatalf("Claude rollover lines were not captured: %q", capture)
		}
		time.Sleep(20 * time.Millisecond)
	}
	if status, err := b.AgentStatus(context.Background(), s); err != nil || status != "idle" {
		t.Fatalf("status=%q err=%v", status, err)
	}
	if err := b.ReleaseAgent(context.Background(), s, "claude"); err != nil {
		t.Fatal(err)
	}
	if err := b.Kill(context.Background(), s); err != nil {
		t.Fatal(err)
	}
	if _, err := b.PaneCapture(context.Background(), s, "recent"); err == nil {
		t.Fatal("killed tmux session remained readable")
	}
}

func TestTmuxBackendRefusesUnverifiedHarnesses(t *testing.T) {
	b := NewTmuxBackend("test", "true")
	if _, err := b.StartAgent(context.Background(), "", t.TempDir(), "", "codex", "task"); err == nil {
		t.Fatal("tmux backend accepted Codex before its terminal behavior was implemented")
	}
}

func TestTmuxSessionNameKeepsCollisionResistantSuffix(t *testing.T) {
	a := tmuxSessionName(strings.Repeat("same-prefix", 10) + "-one")
	b := tmuxSessionName(strings.Repeat("same-prefix", 10) + "-two")
	if a == b || len(a) > 64 || len(b) > 64 {
		t.Fatalf("unsafe tmux session names %q %q", a, b)
	}
}

// The launch dump is Orchestra's scratch space, not the session's work. A
// worktree that starts dirty pollutes the gate, the review diff, and the
// agent's own reading of `git status`.
func TestWriteLaunchContextLeavesTheWorktreeClean(t *testing.T) {
	dir := t.TempDir()
	for _, args := range [][]string{{"init"}, {"config", "user.email", "t@t"}, {"config", "user.name", "t"}, {"commit", "--allow-empty", "-m", "init"}} {
		if out, err := exec.Command("git", append([]string{"-C", dir}, args...)...).CombinedOutput(); err != nil {
			t.Fatalf("git %v: %v: %s", args, err, out)
		}
	}
	if err := WriteLaunchContext(dir, "the instruction"); err != nil {
		t.Fatal(err)
	}
	b, err := os.ReadFile(filepath.Join(dir, LaunchContextFile))
	if err != nil || string(b) != "the instruction" {
		t.Fatalf("launch context = %q, err %v", b, err)
	}
	out, err := exec.Command("git", "-C", dir, "status", "--short").CombinedOutput()
	if err != nil {
		t.Fatal(err)
	}
	if strings.TrimSpace(string(out)) != "" {
		t.Fatalf("worktree is dirty after a launch dump:\n%s", out)
	}
}

func TestLaunchTransportIsHarnessSpecific(t *testing.T) {
	b := &TmuxBackend{}
	if got := b.LaunchTransport("claude"); got != LaunchFileRef {
		t.Fatalf("claude transport=%q, want %q", got, LaunchFileRef)
	}
	for _, harness := range []string{"opencode", "codex", ""} {
		if got := b.LaunchTransport(harness); got != LaunchInline {
			t.Fatalf("%s transport=%q, want %q", harness, got, LaunchInline)
		}
	}
}

// F17, found live during burn-in run 3: a queued or already-accepted prompt
// renders with the same "❯" prefix as an unsubmitted one, so only cursor
// ownership can tell them apart.
func TestInputStateReadsTheEditorOwningTheCursor(t *testing.T) {
	wrapped := "\u2500\u2500\u2500\u2500\n\u276f @.orchestra/launch.md is your complete Orchestra launch instruction. Read\n  .orchestra/launch.md now and follow it.\n\u2500\u2500\u2500\u2500\n"
	b := paneTmux(t, wrapped, 2)
	state, err := b.inputState(context.Background(), Session{PaneID: "s:1.0"})
	if err != nil {
		t.Fatal(err)
	}
	if !state.Active || !sameInput(state.Text, LaunchReference) {
		t.Fatalf("state=%+v, want the wrapped launch reference in the active editor", state)
	}

	// The same text, but the cursor sits in the empty editor below it: the
	// harness took the prompt and queued it.
	queued := "  \u276f @.orchestra/launch.md is your complete Orchestra launch instruction. Read\n    .orchestra/launch.md now and follow it.\n\u2500\u2500\u2500\u2500\n\u276f Press up to edit queued messages\n"
	b = paneTmux(t, queued, 3)
	state, err = b.inputState(context.Background(), Session{PaneID: "s:1.0"})
	if err != nil {
		t.Fatal(err)
	}
	if !state.Active || sameInput(state.Text, LaunchReference) {
		t.Fatalf("state=%+v, want the queued marker rather than the launch text", state)
	}
	if !queuedInput(state.Text) {
		t.Fatalf("state=%+v, want queued input recognized", state)
	}
}

// Run 3 proved one Enter is not enough. A launch still owning the cursor must
// be resubmitted, and the evidence must say how many submits it took.
func TestConfirmInputResubmitsUntilTheEditorClears(t *testing.T) {
	b, keys := submitTmux(t, 2)
	evidence, err := b.ConfirmInput(context.Background(), Session{PaneID: "s:1.0"}, LaunchReference)
	if err != nil {
		t.Fatalf("confirm: %v", err)
	}
	if !strings.Contains(evidence, "submit_attempts=2") {
		t.Fatalf("evidence=%q, want two submits recorded", evidence)
	}
	if got := enterCount(t, keys); got != 1 {
		t.Fatalf("resent Enter %d times, want 1", got)
	}
}

// The intervention is bounded independently of the observation deadline: a TUI
// that already accepted the prompt must not receive a fortieth Enter.
func TestConfirmInputBoundsResubmission(t *testing.T) {
	b, keys := submitTmux(t, 99)
	// Generous deadline on purpose: the cap must be what stops the resends.
	b.LaunchConfirmTimeout = 3 * time.Second
	if _, err := b.ConfirmInput(context.Background(), Session{PaneID: "s:1.0"}, LaunchReference); !errors.Is(err, ErrPromptNotSubmitted) {
		t.Fatalf("err=%v, want ErrPromptNotSubmitted", err)
	}
	if got := enterCount(t, keys); got != launchResubmitLimit {
		t.Fatalf("resent Enter %d times, want the %d-resubmit cap", got, launchResubmitLimit)
	}
}

// Queued is submitted. Treating it as a stalled editor would kill a pane whose
// harness had already accepted the instruction.
func TestConfirmInputAcceptsQueuedInput(t *testing.T) {
	pane := "\u2500\u2500\u2500\u2500\n\u276f Press up to edit queued messages\n"
	b := paneTmux(t, pane, 1)
	evidence, err := b.ConfirmInput(context.Background(), Session{PaneID: "s:1.0"}, LaunchReference)
	if err != nil {
		t.Fatalf("confirm: %v", err)
	}
	if !strings.Contains(evidence, "confirmation=queued") {
		t.Fatalf("evidence=%q, want queued input confirmed", evidence)
	}
}

// paneTmux serves one fixed pane with the cursor on the given row.
func paneTmux(t *testing.T, pane string, cursor int) *TmuxBackend {
	t.Helper()
	dir := t.TempDir()
	paneFile := filepath.Join(dir, "pane")
	if err := os.WriteFile(paneFile, []byte(pane), 0o644); err != nil {
		t.Fatal(err)
	}
	bin := filepath.Join(dir, "tmux")
	script := `#!/bin/sh
cmd=""
for a in "$@"; do
  case "$a" in
    capture-pane|has-session|display-message|send-keys) cmd=$a; break;;
  esac
done
case "$cmd" in
  has-session) exit 0 ;;
  capture-pane) cat ` + paneFile + ` ; exit 0 ;;
  display-message)
    case "$*" in
      *cursor_y*) printf '` + fmt.Sprint(cursor) + `\n' ;;
      *) printf '0\tclaude\n' ;;
    esac
    exit 0 ;;
esac
exit 0
`
	if err := os.WriteFile(bin, []byte(script), 0o755); err != nil {
		t.Fatal(err)
	}
	return &TmuxBackend{Binary: bin, LaunchConfirmTimeout: time.Second, LaunchConfirmPoll: 5 * time.Millisecond}
}

// submitTmux holds the launch reference in the active editor until the given
// number of submits has been received, then clears it. The first submit is the
// one Prompt already sent, so submitOn=2 means one resend is required.
func submitTmux(t *testing.T, submitOn int) (*TmuxBackend, string) {
	t.Helper()
	dir := t.TempDir()
	keys := filepath.Join(dir, "keys")
	bin := filepath.Join(dir, "tmux")
	script := `#!/bin/sh
dir=` + dir + `
cmd=""
for a in "$@"; do
  case "$a" in
    capture-pane|has-session|display-message|send-keys) cmd=$a; break;;
  esac
done
submits=$(cat $dir/submits 2>/dev/null || echo 1)
case "$cmd" in
  has-session) exit 0 ;;
  send-keys)
    case "$*" in
      *Enter*)
        echo Enter >> ` + keys + `
        echo $((submits+1)) > $dir/submits ;;
    esac
    exit 0 ;;
  capture-pane)
    if [ "$submits" -ge "` + fmt.Sprint(submitOn) + `" ]; then
      printf 'esc to interrupt\n\xe2\x9d\xaf \n'
    else
      printf '\xe2\x94\x80\x0a\xe2\x9d\xaf ` + LaunchReference + `\n'
    fi
    exit 0 ;;
  display-message)
    case "$*" in
      *cursor_y*)
        if [ "$submits" -ge "` + fmt.Sprint(submitOn) + `" ]; then printf '1\n'; else printf '1\n'; fi ;;
      *) printf '0\tclaude\n' ;;
    esac
    exit 0 ;;
esac
exit 0
`
	if err := os.WriteFile(bin, []byte(script), 0o755); err != nil {
		t.Fatal(err)
	}
	return &TmuxBackend{Binary: bin, LaunchConfirmTimeout: time.Second, LaunchConfirmPoll: 5 * time.Millisecond}, keys
}

func enterCount(t *testing.T, path string) int {
	t.Helper()
	b, err := os.ReadFile(path)
	if os.IsNotExist(err) {
		return 0
	}
	if err != nil {
		t.Fatal(err)
	}
	return len(strings.Fields(string(b)))
}

// fakeTmux scripts capture-pane so confirmation can be driven without waiting
// on a real harness. clearAfter is how many captures still show the stalled
// editor before it clears.
func fakeTmux(t *testing.T, clearAfter int, failCapture bool) *TmuxBackend {
	t.Helper()
	dir := t.TempDir()
	bin := filepath.Join(dir, "tmux")
	fail := "0"
	if failCapture {
		fail = "1"
	}
	script := `#!/bin/sh
state=` + filepath.Join(dir, "count") + `
cmd=""
for a in "$@"; do
  case "$a" in
    capture-pane|has-session|display-message) cmd=$a; break;;
  esac
done
case "$cmd" in
  has-session) exit 0 ;;
  display-message)
    case "$*" in
      *cursor_y*) printf '0\n' ;;
      *) printf '0\tclaude\n' ;;
    esac
    exit 0 ;;
  capture-pane)
    if [ "` + fail + `" = "1" ]; then echo "no server running" >&2; exit 1; fi
    n=$(cat $state 2>/dev/null || echo 0); n=$((n+1)); echo $n > $state
    if [ "$n" -le "` + fmt.Sprint(clearAfter) + `" ]; then
      printf '\xe2\x9d\xaf [Pasted text #1 +66 lines]\n'
    else
      printf 'esc to interrupt\n\xe2\x9d\xaf \n'
    fi
    exit 0 ;;
esac
exit 0
`
	if err := os.WriteFile(bin, []byte(script), 0o755); err != nil {
		t.Fatal(err)
	}
	return &TmuxBackend{Binary: bin, LaunchConfirmTimeout: 2 * time.Second, LaunchConfirmPoll: 5 * time.Millisecond}
}

func TestConfirmInputAcceptsObservedActivity(t *testing.T) {
	b := fakeTmux(t, 1, false)
	evidence, err := b.ConfirmInput(context.Background(), Session{PaneID: "s:1.0"}, LaunchReference)
	if err != nil {
		t.Fatalf("confirm: %v", err)
	}
	if !strings.Contains(evidence, "busy") {
		t.Fatalf("evidence=%q, want the observed activity named", evidence)
	}
}

// Prompt returning nil is not proof. An editor that still holds the text at
// the deadline is a launch failure, and it must be the one class that releases
// the lease instead of holding it as uncertain.
func TestConfirmInputFailsWhileTheEditorStillHoldsThePrompt(t *testing.T) {
	b := fakeTmux(t, 1000, false)
	b.LaunchConfirmTimeout = 60 * time.Millisecond
	_, err := b.ConfirmInput(context.Background(), Session{PaneID: "s:1.0"}, LaunchReference)
	if !errors.Is(err, ErrPromptNotSubmitted) {
		t.Fatalf("err=%v, want ErrPromptNotSubmitted", err)
	}
	if !strings.Contains(err.Error(), "Pasted text") {
		t.Fatalf("err=%v, want the stalled editor content quoted", err)
	}
}

// A confirmation that cannot read the pane must not become an acknowledgement.
func TestConfirmInputSurfacesTransportErrors(t *testing.T) {
	b := fakeTmux(t, 0, true)
	if _, err := b.ConfirmInput(context.Background(), Session{PaneID: "s:1.0"}, LaunchReference); err == nil {
		t.Fatal("a failed pane capture must not confirm a launch")
	}
}

// F16 and F20 are separate concerns and this is the line between them:
// Orchestra confirms delivery of what it originates, but nothing typed at a
// pane counts as the agent doing work. A capture whose only difference is the
// input line must produce the same progress digest.
func TestPaneProgressIgnoresInputLines(t *testing.T) {
	body := "────\n  ran the checks, nothing to change\n────\n"
	idle := paneTmux(t, body+"❯ \n", 3)
	typed := paneTmux(t, body+"❯ go ahead and implement it\n", 3)
	a, err := idle.PaneProgress(context.Background(), Session{PaneID: "s:1.0"})
	if err != nil {
		t.Fatal(err)
	}
	b, err := typed.PaneProgress(context.Background(), Session{PaneID: "s:1.0"})
	if err != nil {
		t.Fatal(err)
	}
	if a != b {
		t.Fatalf("typing changed the progress digest:\n%q\n%q", a, b)
	}
	if !strings.Contains(a, "ran the checks") {
		t.Fatalf("progress digest dropped harness output: %q", a)
	}

	// Harness output still moves it.
	worked := paneTmux(t, "────\n  edited scripts/orchestra_e2e_healthcheck.sh\n────\n❯ \n", 3)
	c, err := worked.PaneProgress(context.Background(), Session{PaneID: "s:1.0"})
	if err != nil {
		t.Fatal(err)
	}
	if c == a {
		t.Fatal("real harness output left the progress digest unchanged")
	}
}

// TestMissingSessionIsRecognisedOnAnEmptyRunningServer guards F32. A running
// but empty tmux server answers "no current target", not "no server running".
// The runtime only stays alive past its last pane since it became its own unit
// running `tmux -D`, so this reply had never been seen before. Treating it as
// a real error made Kill fail for a pane that was already gone, which left the
// quarantine set and pinned the worker's only session slot.
func TestMissingSessionIsRecognisedOnAnEmptyRunningServer(t *testing.T) {
	for _, message := range []string{"no current target", "can't find session: x", "no server running"} {
		bin := filepath.Join(t.TempDir(), "tmux")
		script := "#!/bin/sh\necho \"" + message + "\" >&2\nexit 1\n"
		if err := os.WriteFile(bin, []byte(script), 0o755); err != nil {
			t.Fatal(err)
		}
		b := &TmuxBackend{Binary: bin}
		if err := b.Kill(context.Background(), Session{PaneID: "s:1.0"}); err != nil {
			t.Fatalf("%q: killing an already-gone session failed: %v", message, err)
		}
		status, err := b.AgentStatus(context.Background(), Session{PaneID: "s:1.0"})
		if err != nil || status != "exited" {
			t.Fatalf("%q: status=%q err=%v, want exited", message, status, err)
		}
	}
}

// TestEmptyEditorIsNotProofBeforeItEverHeldTheText guards F33. The submit
// races the TUI's render: a poll 7ms after Enter found an empty editor, called
// it confirmation=editor_cleared, and the launch text then sat unsent for the
// whole lease. Run 5 died on exactly that. An editor that never held the text
// has to keep being observed, and once the text appears the resubmit path can
// do its job.
func TestEmptyEditorIsNotProofBeforeItEverHeldTheText(t *testing.T) {
	dir := t.TempDir()
	bin := filepath.Join(dir, "tmux")
	calls := filepath.Join(dir, "calls")
	// capture-pane reports an empty editor on the first two polls, exactly as a
	// TUI that has not rendered the paste yet, then shows the submitted text.
	script := `#!/bin/sh
cmd=""
for a in "$@"; do
  case "$a" in capture-pane|has-session|display-message|send-keys) cmd=$a; break;; esac
done
case "$cmd" in
  has-session) exit 0 ;;
  send-keys) exit 0 ;;
  display-message)
    case "$*" in
      *cursor_y*) printf '1\n' ;;
      *) printf '0\tclaude\n' ;;
    esac
    exit 0 ;;
  capture-pane)
    n=$(cat ` + calls + ` 2>/dev/null || echo 0); n=$((n+1)); echo $n > ` + calls + `
    if [ "$n" -le 2 ]; then printf '\xe2\x9d\xaf \n\n'; else printf '\xe2\x9d\xaf hello world\n\n'; fi
    exit 0 ;;
esac
exit 0
`
	if err := os.WriteFile(bin, []byte(script), 0o755); err != nil {
		t.Fatal(err)
	}
	b := &TmuxBackend{Binary: bin, LaunchConfirmTimeout: 300 * time.Millisecond, LaunchConfirmPoll: 5 * time.Millisecond}
	_, err := b.ConfirmInput(context.Background(), Session{PaneID: "s:1.0"}, "hello world")
	if err == nil {
		t.Fatal("an editor still holding the submitted text was reported as confirmed")
	}
	if !errors.Is(err, ErrPromptNotSubmitted) {
		t.Fatalf("err=%v, want ErrPromptNotSubmitted", err)
	}
}

// F41, live on run 5: the review agent produced nothing after 02:12 and the
// 02:46 renewal was granted, because the digest covered the harness footer and
// one of its fields ticked. The layout below is the real pane, with the usage
// percentage and the version banner moved on.
func TestPaneProgressIgnoresHarnessFooter(t *testing.T) {
	body := "────\n  reviewed the diff, no blocking findings\n────\n❯ \n"
	footer := func(pct, banner string) string {
		return body +
			"  [Opus 5] 📁 06G4A4F0TFXKZHJE48N05XN1HG | 7d: " + pct + "\n" +
			"  cf474986 - Orchestra launch instructions | 57.9…\n" +
			"  ⏵⏵ auto mode on (shift+tab to cycle) · " + banner + "\n"
	}
	before := paneTmux(t, footer("51%", "current: 2.1.247 · latest: 2.1.248"), 3)
	after := paneTmux(t, footer("52%", "✔ Update installed · Restart to update"), 3)
	a, err := before.PaneProgress(context.Background(), Session{PaneID: "s:1.0"})
	if err != nil {
		t.Fatal(err)
	}
	b, err := after.PaneProgress(context.Background(), Session{PaneID: "s:1.0"})
	if err != nil {
		t.Fatal(err)
	}
	if a != b {
		t.Fatalf("footer churn changed the progress digest:\n%q\n%q", a, b)
	}
	if !strings.Contains(a, "reviewed the diff") {
		t.Fatalf("progress digest dropped harness output: %q", a)
	}
	worked := paneTmux(t, "────\n  reviewed the diff, wrote .orchestra/done\n────\n❯ \n"+
		"  [Opus 5] 📁 06G4A4F0TFXKZHJE48N05XN1HG | 7d: 51%\n", 3)
	c, err := worked.PaneProgress(context.Background(), Session{PaneID: "s:1.0"})
	if err != nil {
		t.Fatal(err)
	}
	if c == a {
		t.Fatal("real agent output left the progress digest unchanged")
	}
}