butterrobot/internal/observability/scrub_test.go

package observability

import (
	"context"
	"io"
	"log/slog"
	"net/http"
	"net/http/httptest"
	"strings"
	"sync"
	"testing"
	"time"

	"github.com/getsentry/sentry-go"

	"git.nakama.town/fmartingr/butterrobot/internal/config"
)

// testToken mimics the shape of a Telegram bot token, which the webhook URL
// embeds in its path.
const testToken = "123456:AAFakeTelegramBotTokenValue"

// captureTransport records events locally instead of shipping them to Sentry.
type captureTransport struct {
	mu     sync.Mutex
	events []*sentry.Event
}

func (t *captureTransport) Configure(sentry.ClientOptions) {}

func (t *captureTransport) SendEvent(event *sentry.Event) {
	t.mu.Lock()
	defer t.mu.Unlock()
	t.events = append(t.events, event)
}

func (t *captureTransport) Flush(time.Duration) bool { return true }

func (t *captureTransport) FlushWithContext(context.Context) bool { return true }

func (t *captureTransport) Close() {}

func (t *captureTransport) captured() []*sentry.Event {
	t.mu.Lock()
	defer t.mu.Unlock()
	return append([]*sentry.Event(nil), t.events...)
}

// enableSentry binds a client that captures events in memory to the current
// hub, and unbinds it when the test ends.
func enableSentry(t *testing.T, secrets []string) *captureTransport {
	t.Helper()

	transport := &captureTransport{}

	options := clientOptions(config.SentryConfig{
		DSN:              "https://key@example.invalid/1",
		Environment:      "test",
		TracesSampleRate: 1.0,
	}, false, secrets)
	options.Transport = transport

	client, err := sentry.NewClient(options)
	if err != nil {
		t.Fatalf("failed to create client: %v", err)
	}

	hub := sentry.CurrentHub()
	hub.BindClient(client)
	t.Cleanup(func() { hub.BindClient(nil) })

	return transport
}

func TestWebhookTokenIsScrubbedFromRequestContext(t *testing.T) {
	transport := enableSentry(t, []string{testToken})

	handler := HTTPMiddleware(http.HandlerFunc(func(_ http.ResponseWriter, r *http.Request) {
		sentry.GetHubFromContext(r.Context()).CaptureMessage("handler failure")
	}))

	handler.ServeHTTP(httptest.NewRecorder(),
		httptest.NewRequest(http.MethodPost, "/telegram/incoming/"+testToken, nil))

	events := transport.captured()
	if len(events) < 2 {
		t.Fatalf("expected an error event and a transaction, got %d", len(events))
	}

	for _, event := range events {
		if event.Request != nil && strings.Contains(event.Request.URL, testToken) {
			t.Errorf("request URL leaks the bot token: %s", event.Request.URL)
		}
		if strings.Contains(event.Transaction, testToken) {
			t.Errorf("transaction name leaks the bot token: %s", event.Transaction)
		}
	}
}

func TestErrorLogIsForwardedTaggedAndScrubbed(t *testing.T) {
	transport := enableSentry(t, []string{testToken})

	logger := WrapLogger(slog.New(slog.NewTextHandler(io.Discard, nil)), config.SentryConfig{})
	ctx := WithTags(context.Background(), map[string]string{"platform": "telegram"})

	logger.ErrorContext(ctx, "Error sending message",
		"error", `Post "https://api.telegram.org/bot`+testToken+`/sendMessage": timeout`)

	events := transport.captured()
	if len(events) != 1 {
		t.Fatalf("expected exactly one event, got %d", len(events))
	}

	event := events[0]
	if event.Tags["platform"] != "telegram" {
		t.Errorf("expected the platform tag to reach Sentry, got %v", event.Tags)
	}

	for _, exception := range event.Exception {
		if strings.Contains(exception.Value, testToken) {
			t.Errorf("exception value leaks the bot token: %s", exception.Value)
		}
	}

	for name, context := range event.Contexts {
		for key, value := range context {
			if text, ok := value.(string); ok && strings.Contains(text, testToken) {
				t.Errorf("context %s.%s leaks the bot token: %s", name, key, text)
			}
		}
	}
}

func TestNewScrubberIgnoresShortSecrets(t *testing.T) {
	if newScrubber([]string{"", "1234"}) != nil {
		t.Fatal("expected short secrets to be ignored")
	}

	if newScrubber([]string{testToken}) == nil {
		t.Fatal("expected a scrubber for a usable secret")
	}
}

func TestScrubberRedactsNestedValues(t *testing.T) {
	s := newScrubber([]string{testToken})

	event := sentry.NewEvent()
	event.Message = "failed calling " + testToken
	event.Contexts[logContextKey] = sentry.Context{
		"nested": map[string]any{"url": "https://api.telegram.org/bot" + testToken},
		"count":  3,
	}

	scrubbed := s.event(event)

	if strings.Contains(scrubbed.Message, testToken) {
		t.Errorf("message leaks the token: %s", scrubbed.Message)
	}

	data := scrubbed.Contexts[logContextKey]
	nested := data["nested"].(map[string]any)
	if strings.Contains(nested["url"].(string), testToken) {
		t.Errorf("nested value leaks the token: %v", nested["url"])
	}

	if data["count"] != 3 {
		t.Errorf("expected non-string values to be untouched, got %v", data["count"])
	}
}