google · gmlewis · May 1, 2024 · Apr 1, 2024 · Apr 2, 2024 · Apr 2, 2024
diff --git a/README.md b/README.md
@@ -180,6 +180,12 @@ if _, ok := err.(*github.AbuseRateLimitError); ok {
 }
 ```
 
+Alternatively, you can block until the rate limit is reset by using the `context.WithValue` method:
+
+````go
+repos, _, err := client.Repositories.List(context.WithValue(ctx, github.SleepUntilPrimaryRateLimitResetWhenRateLimited, true), "", nil)
+```
+
 You can use [go-github-ratelimit](https://github.com/gofri/go-github-ratelimit) to handle
 secondary rate limit sleep-and-retry for you.
 

diff --git a/github/github.go b/github/github.go
@@ -804,6 +804,7 @@ type requestContext uint8
 
 const (
 	bypassRateLimitCheck requestContext = iota
+	SleepUntilPrimaryRateLimitResetWhenRateLimited
 )
 
 // BareDo sends an API request and lets you handle the api response. If an error
@@ -889,6 +890,15 @@ func (c *Client) BareDo(ctx context.Context, req *http.Request) (*Response, erro
 			err = aerr
 		}
 
+		rateLimitError, ok := err.(*RateLimitError)
+		if ok && req.Context().Value(SleepUntilPrimaryRateLimitResetWhenRateLimited) != nil {
+			if err := sleepUntilResetWithBuffer(req.Context(), rateLimitError.Rate.Reset.Time); err != nil {
+				return response, err
+			}
+			// retry the request once when the rate limit has reset
+			return c.BareDo(context.WithValue(req.Context(), SleepUntilPrimaryRateLimitResetWhenRateLimited, nil), req)
+		}
+
 		// Update the secondary rate limit if we hit it.
 		rerr, ok := err.(*AbuseRateLimitError)
 		if ok && rerr.RetryAfter != nil {
@@ -950,6 +960,18 @@ func (c *Client) checkRateLimitBeforeDo(req *http.Request, rateLimitCategory Rat
 			Header:     make(http.Header),
 			Body:       io.NopCloser(strings.NewReader("")),
 		}
+
+		if req.Context().Value(SleepUntilPrimaryRateLimitResetWhenRateLimited) != nil {
+			if err := sleepUntilResetWithBuffer(req.Context(), rate.Reset.Time); err == nil {
+				return nil
+			}
+			return &RateLimitError{
+				Rate:     rate,
+				Response: resp,
+				Message:  fmt.Sprintf("Context cancelled while waiting for rate limit to reset until %v, not making remote request.", rate.Reset.Time),
+			}
+		}
+
 		return &RateLimitError{
 			Rate:     rate,
 			Response: resp,
@@ -1514,6 +1536,20 @@ func formatRateReset(d time.Duration) string {
 	return fmt.Sprintf("[rate reset in %v]", timeString)
 }
 
+func sleepUntilResetWithBuffer(ctx context.Context, reset time.Time) error {
+	buffer := time.Second
+	timer := time.NewTimer(time.Until(reset) + buffer)
+	select {
+	case <-ctx.Done():
+		if !timer.Stop() {
+			<-timer.C
+		}
+		return ctx.Err()
+	case <-timer.C:
+	}
+	return nil
+}
+
 // When using roundTripWithOptionalFollowRedirect, note that it
 // is the responsibility of the caller to close the response body.
 func (c *Client) roundTripWithOptionalFollowRedirect(ctx context.Context, u string, maxRedirects int, opts ...RequestOption) (*http.Response, error) {

diff --git a/github/github_test.go b/github/github_test.go
@@ -1381,6 +1381,176 @@ func TestDo_rateLimit_ignoredFromCache(t *testing.T) {
 	}
 }
 
+// Ensure sleeps until the rate limit is reset when the client is rate limited.
+func TestDo_rateLimit_sleepUntilResponseResetLimit(t *testing.T) {
+	client, mux, _, teardown := setup()
+	defer teardown()
+
+	reset := time.Now().UTC().Add(time.Second)
+
+	var firstRequest = true
+	mux.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
+		if firstRequest {
+			firstRequest = false
+			w.Header().Set(headerRateLimit, "60")
+			w.Header().Set(headerRateRemaining, "0")
+			w.Header().Set(headerRateReset, fmt.Sprint(reset.Unix()))
+			w.Header().Set("Content-Type", "application/json; charset=utf-8")
+			w.WriteHeader(http.StatusForbidden)
+			fmt.Fprintln(w, `{
+   "message": "API rate limit exceeded for xxx.xxx.xxx.xxx. (But here's the good news: Authenticated requests get a higher rate limit. Check out the documentation for more details.)",
+   "documentation_url": "https://docs.github.com/en/rest/overview/resources-in-the-rest-api#abuse-rate-limits"
+}`)
+			return
+		}
+		w.Header().Set(headerRateLimit, "5000")
+		w.Header().Set(headerRateRemaining, "5000")
+		w.Header().Set(headerRateReset, fmt.Sprint(reset.Add(time.Hour).Unix()))
+		w.Header().Set("Content-Type", "application/json; charset=utf-8")
+		w.WriteHeader(http.StatusOK)
+		fmt.Fprintln(w, `{}`)
+	})
+
+	req, _ := client.NewRequest("GET", ".", nil)
+	ctx := context.Background()
+	resp, err := client.Do(context.WithValue(ctx, SleepUntilPrimaryRateLimitResetWhenRateLimited, true), req, nil)
+	if err != nil {
+		t.Errorf("Do returned unexpected error: %v", err)
+	}
+	if got, want := resp.StatusCode, http.StatusOK; got != want {
+		t.Errorf("Response status code = %v, want %v", got, want)
+	}
+}
+
+// Ensure tries to sleep until the rate limit is reset when the client is rate limited, but only once.
+func TestDo_rateLimit_sleepUntilResponseResetLimitRetryOnce(t *testing.T) {
+	client, mux, _, teardown := setup()
+	defer teardown()
+
+	reset := time.Now().UTC().Add(time.Second)
+
+	requestCount := 0
+	mux.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
+		requestCount++
+		w.Header().Set(headerRateLimit, "60")
+		w.Header().Set(headerRateRemaining, "0")
+		w.Header().Set(headerRateReset, fmt.Sprint(reset.Unix()))
+		w.Header().Set("Content-Type", "application/json; charset=utf-8")
+		w.WriteHeader(http.StatusForbidden)
+		fmt.Fprintln(w, `{
+   "message": "API rate limit exceeded for xxx.xxx.xxx.xxx. (But here's the good news: Authenticated requests get a higher rate limit. Check out the documentation for more details.)",
+   "documentation_url": "https://docs.github.com/en/rest/overview/resources-in-the-rest-api#abuse-rate-limits"
+}`)
+	})
+
+	req, _ := client.NewRequest("GET", ".", nil)
+	ctx := context.Background()
+	_, err := client.Do(context.WithValue(ctx, SleepUntilPrimaryRateLimitResetWhenRateLimited, true), req, nil)
+	if err == nil {
+		t.Error("Expected error to be returned.")
+	}
+	if got, want := requestCount, 2; got != want {
+		t.Errorf("Expected 2 requests, got %d", got)
+	}
+}
+
+// Ensure a network call is not made when it's known that API rate limit is still exceeded.
+func TestDo_rateLimit_sleepUntilClientResetLimit(t *testing.T) {
+	client, mux, _, teardown := setup()
+	defer teardown()
+
+	reset := time.Now().UTC().Add(time.Second)
+	client.rateLimits[CoreCategory] = Rate{Limit: 5000, Remaining: 0, Reset: Timestamp{reset}}
+	requestCount := 0
+	mux.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
+		requestCount++
+		w.Header().Set(headerRateLimit, "5000")
+		w.Header().Set(headerRateRemaining, "5000")
+		w.Header().Set(headerRateReset, fmt.Sprint(reset.Add(time.Hour).Unix()))
+		w.Header().Set("Content-Type", "application/json; charset=utf-8")
+		w.WriteHeader(http.StatusOK)
+		fmt.Fprintln(w, `{}`)
+	})
+	req, _ := client.NewRequest("GET", ".", nil)
+	ctx := context.Background()
+	resp, err := client.Do(context.WithValue(ctx, SleepUntilPrimaryRateLimitResetWhenRateLimited, true), req, nil)
+	if err != nil {
+		t.Errorf("Do returned unexpected error: %v", err)
+	}
+	if got, want := resp.StatusCode, http.StatusOK; got != want {
+		t.Errorf("Response status code = %v, want %v", got, want)
+	}
+	if got, want := requestCount, 1; got != want {
+		t.Errorf("Expected 1 request, got %d", got)
+	}
+}
+
+// Ensure sleep is aborted when the context is cancelled.
+func TestDo_rateLimit_abortSleepContextCancelled(t *testing.T) {
+	client, mux, _, teardown := setup()
+	defer teardown()
+
+	// We use a 1 minute reset time to ensure the sleep is not completed.
+	reset := time.Now().UTC().Add(time.Minute)
+	requestCount := 0
+	mux.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
+		requestCount++
+		w.Header().Set(headerRateLimit, "60")
+		w.Header().Set(headerRateRemaining, "0")
+		w.Header().Set(headerRateReset, fmt.Sprint(reset.Unix()))
+		w.Header().Set("Content-Type", "application/json; charset=utf-8")
+		w.WriteHeader(http.StatusForbidden)
+		fmt.Fprintln(w, `{
+   "message": "API rate limit exceeded for xxx.xxx.xxx.xxx. (But here's the good news: Authenticated requests get a higher rate limit. Check out the documentation for more details.)",
+   "documentation_url": "https://docs.github.com/en/rest/overview/resources-in-the-rest-api#abuse-rate-limits"
+}`)
+	})
+
+	req, _ := client.NewRequest("GET", ".", nil)
+	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Millisecond)
+	defer cancel()
+	_, err := client.Do(context.WithValue(ctx, SleepUntilPrimaryRateLimitResetWhenRateLimited, true), req, nil)
+	if !errors.Is(err, context.DeadlineExceeded) {
+		t.Error("Expected context deadline exceeded error.")
+	}
+	if got, want := requestCount, 1; got != want {
+		t.Errorf("Expected 1 requests, got %d", got)
+	}
+}
+
+// Ensure sleep is aborted when the context is cancelled on initial request.
+func TestDo_rateLimit_abortSleepContextCancelledClientLimit(t *testing.T) {
+	client, mux, _, teardown := setup()
+	defer teardown()
+
+	reset := time.Now().UTC().Add(time.Minute)
+	client.rateLimits[CoreCategory] = Rate{Limit: 5000, Remaining: 0, Reset: Timestamp{reset}}
+	requestCount := 0
+	mux.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
+		requestCount++
+		w.Header().Set(headerRateLimit, "5000")
+		w.Header().Set(headerRateRemaining, "5000")
+		w.Header().Set(headerRateReset, fmt.Sprint(reset.Add(time.Hour).Unix()))
+		w.Header().Set("Content-Type", "application/json; charset=utf-8")
+		w.WriteHeader(http.StatusOK)
+		fmt.Fprintln(w, `{}`)
+	})
+	req, _ := client.NewRequest("GET", ".", nil)
+	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Millisecond)
+	defer cancel()
+	_, err := client.Do(context.WithValue(ctx, SleepUntilPrimaryRateLimitResetWhenRateLimited, true), req, nil)
+	rateLimitError, ok := err.(*RateLimitError)
+	if !ok {
+		t.Fatalf("Expected a *rateLimitError error; got %#v.", err)
+	}
+	if got, wantSuffix := rateLimitError.Message, "Context cancelled while waiting for rate limit to reset until"; !strings.HasPrefix(got, wantSuffix) {
+		t.Errorf("Expected request to be prevented because context cancellation, got: %v.", got)
+	}
+	if got, want := requestCount, 0; got != want {
+		t.Errorf("Expected 1 requests, got %d", got)
+	}
+}
+
 // Ensure *AbuseRateLimitError is returned when the response indicates that
 // the client has triggered an abuse detection mechanism.
 func TestDo_rateLimit_abuseRateLimitError(t *testing.T) {