feat: log structured reconnect diagnostics for kiosk display layers (#2)

internal/display/diagnostics_test.go

@@ -0,0 +1,182 @@
+package display
+
+import (
+	"context"
+	"errors"
+	"sync"
+	"testing"
+	"time"
+)
+
+// diagnosticCapture collects structured reconnect records from Run.
+type diagnosticCapture struct {
+	mu      sync.Mutex
+	records []Diagnostic
+}
+
+func (c *diagnosticCapture) record(d Diagnostic) {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+	c.records = append(c.records, d)
+}
+
+func (c *diagnosticCapture) all() []Diagnostic {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+	return append([]Diagnostic(nil), c.records...)
+}
+
+func TestRunDiagnosesEachFailedOpenWithStructuredFields(t *testing.T) {
+	session := newFakeSession()
+	var opens int
+	opener := fakeOpener{open: func(int) (Session, error) {
+		opens++
+		if opens <= 2 {
+			return nil, errors.New("x11 down")
+		}
+		return session, nil
+	}}
+
+	capture := &diagnosticCapture{}
+	ctx, cancel := context.WithCancel(context.Background())
+	done := make(chan error, 1)
+	go func() {
+		done <- Run(ctx, opener, NewStatusTemplate("hello"), Options{
+			Target:     "x11",
+			Layer:      "fallback-canvas",
+			Diagnostic: capture.record,
+			RetryDelay: time.Millisecond,
+			MaxDelay:   2 * time.Millisecond,
+		})
+	}()
+
+	waitFor(t, func() bool { return session.canvas.lineCount() > 0 }, "connect after retries")
+	cancel()
+
+	select {
+	case err := <-done:
+		if err != nil {
+			t.Fatalf("run: %v", err)
+		}
+	case <-time.After(time.Second):
+		t.Fatal("run did not stop after cancel")
+	}
+
+	records := capture.all()
+	if len(records) != 2 {
+		t.Fatalf("records=%d want exactly one per failed open (2)", len(records))
+	}
+	first, second := records[0], records[1]
+	if first.Target != "x11" {
+		t.Fatalf("target=%q want x11", first.Target)
+	}
+	if first.Layer != "fallback-canvas" {
+		t.Fatalf("layer=%q want fallback-canvas", first.Layer)
+	}
+	if first.Event != EventOpenFailed {
+		t.Fatalf("event=%q want %q", first.Event, EventOpenFailed)
+	}
+	if first.Err == nil || first.Err.Error() != "x11 down" {
+		t.Fatalf("underlying error %v must be carried in the record", first.Err)
+	}
+	if first.RetryDelay != time.Millisecond {
+		t.Fatalf("first retry delay=%v want %v", first.RetryDelay, time.Millisecond)
+	}
+	if second.RetryDelay != 2*time.Millisecond {
+		t.Fatalf("backed-off retry delay=%v want %v", second.RetryDelay, 2*time.Millisecond)
+	}
+}
+
+func TestRunDiagnosesBrokenSessionOnceAndRecovers(t *testing.T) {
+	first := newFakeSession()
+	second := newFakeSession()
+	var opens int
+	opener := fakeOpener{open: func(int) (Session, error) {
+		opens++
+		if opens == 1 {
+			return first, nil
+		}
+		return second, nil
+	}}
+
+	capture := &diagnosticCapture{}
+	ctx, cancel := context.WithCancel(context.Background())
+	done := make(chan error, 1)
+	go func() {
+		done <- Run(ctx, opener, NewStatusTemplate("hello"), Options{
+			Target:     "x11",
+			Diagnostic: capture.record,
+			RetryDelay: time.Millisecond,
+			MaxDelay:   2 * time.Millisecond,
+		})
+	}()
+
+	waitFor(t, func() bool { return first.canvas.lineCount() == 1 }, "draw on first session")
+	first.events <- eventResult{err: errors.New("connection lost")}
+	waitFor(t, func() bool { return second.canvas.lineCount() == 1 }, "recover on reconnected session")
+	cancel()
+
+	select {
+	case err := <-done:
+		if err != nil {
+			t.Fatalf("run: %v", err)
+		}
+	case <-time.After(time.Second):
+		t.Fatal("run did not stop after cancel")
+	}
+
+	records := capture.all()
+	if len(records) != 1 {
+		t.Fatalf("records=%d want exactly one per broken session (1)", len(records))
+	}
+	record := records[0]
+	if record.Target != "x11" {
+		t.Fatalf("target=%q want x11", record.Target)
+	}
+	if record.Event != EventSessionEnded {
+		t.Fatalf("event=%q want %q", record.Event, EventSessionEnded)
+	}
+	if record.Err == nil || record.Err.Error() != "connection lost" {
+		t.Fatalf("underlying error %v must be carried in the record", record.Err)
+	}
+	if record.RetryDelay != time.Millisecond {
+		t.Fatalf("retry delay=%v want the reset %v", record.RetryDelay, time.Millisecond)
+	}
+}
+
+// blockingOpener fails only after the context is done, so cancellation is
+// the sole reason Run stops: no diagnostic may fire for shutdown.
+type blockingOpener struct{}
+
+func (blockingOpener) Open(ctx context.Context) (Session, error) {
+	<-ctx.Done()
+	return nil, ctx.Err()
+}
+
+func TestRunStaysSilentWhenCancellationEndsTheLoop(t *testing.T) {
+	capture := &diagnosticCapture{}
+	ctx, cancel := context.WithCancel(context.Background())
+	done := make(chan error, 1)
+	go func() {
+		done <- Run(ctx, blockingOpener{}, NewStatusTemplate("hello"), Options{
+			Target:     "x11",
+			Diagnostic: capture.record,
+			RetryDelay: time.Millisecond,
+			MaxDelay:   2 * time.Millisecond,
+		})
+	}()
+
+	cancel()
+
+	select {
+	case err := <-done:
+		if err != nil {
+			t.Fatalf("run: %v", err)
+		}
+	case <-time.After(time.Second):
+		t.Fatal("run did not stop after cancel")
+	}
+	if records := capture.all(); len(records) != 0 {
+		t.Fatalf("records=%v must stay empty: shutdown is not a reconnect boundary", records)
+	}
+}