fix(feishu): break infinite typing-indicator retry loop on rate-limit / quota errors (openclaw#28494) thanks @guoqunabc

Verified:
- pnpm build
- pnpm check
- pnpm test:macmini

Co-authored-by: guoqunabc <9532020+guoqunabc@users.noreply.github.com>
Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>

extensions/feishu/src/typing.ts

@@ -7,13 +7,97 @@ import { createFeishuClient } from "./client.js";
 // Full list: https://github.com/go-lark/lark/blob/main/emoji.go
 const TYPING_EMOJI = "Typing"; // Typing indicator emoji
 
+/**
+ * Feishu API error codes that indicate the caller should back off.
+ * These must propagate to the typing circuit breaker so the keepalive loop
+ * can trip and stop retrying.
+ *
+ * - 99991400: Rate limit (too many requests per second)
+ * - 99991403: Monthly API call quota exceeded
+ * - 429: Standard HTTP 429 returned as a Feishu SDK error code
+ *
+ * @see https://open.feishu.cn/document/server-docs/api-call-guide/generic-error-code
+ */
+const FEISHU_BACKOFF_CODES = new Set([99991400, 99991403, 429]);
+
+/**
+ * Custom error class for Feishu backoff conditions detected from non-throwing
+ * SDK responses. Carries a numeric `.code` so that `isFeishuBackoffError()`
+ * recognises it when the error is caught downstream.
+ */
+export class FeishuBackoffError extends Error {
+  code: number;
+  constructor(code: number) {
+    super(`Feishu API backoff: code ${code}`);
+    this.name = "FeishuBackoffError";
+    this.code = code;
+  }
+}
+
 export type TypingIndicatorState = {
   messageId: string;
   reactionId: string | null;
 };
 
 /**
- * Add a typing indicator (reaction) to a message
+ * Check whether an error represents a rate-limit or quota-exceeded condition
+ * from the Feishu API that should stop the typing keepalive loop.
+ *
+ * Handles two shapes:
+ * 1. AxiosError with `response.status` and `response.data.code`
+ * 2. Feishu SDK error with a top-level `code` property
+ */
+export function isFeishuBackoffError(err: unknown): boolean {
+  if (typeof err !== "object" || err === null) {
+    return false;
+  }
+
+  // AxiosError shape: err.response.status / err.response.data.code
+  const response = (err as { response?: { status?: number; data?: { code?: number } } }).response;
+  if (response) {
+    if (response.status === 429) {
+      return true;
+    }
+    if (typeof response.data?.code === "number" && FEISHU_BACKOFF_CODES.has(response.data.code)) {
+      return true;
+    }
+  }
+
+  // Feishu SDK error shape: err.code
+  const code = (err as { code?: number }).code;
+  if (typeof code === "number" && FEISHU_BACKOFF_CODES.has(code)) {
+    return true;
+  }
+
+  return false;
+}
+
+/**
+ * Check whether a Feishu SDK response object contains a backoff error code.
+ *
+ * The Feishu SDK sometimes returns a normal response (no throw) with an
+ * API-level error code in the response body. This must be detected so the
+ * circuit breaker can trip. See codex review on #28157.
+ */
+export function getBackoffCodeFromResponse(response: unknown): number | undefined {
+  if (typeof response !== "object" || response === null) {
+    return undefined;
+  }
+  const code = (response as { code?: number }).code;
+  if (typeof code === "number" && FEISHU_BACKOFF_CODES.has(code)) {
+    return code;
+  }
+  return undefined;
+}
+
+/**
+ * Add a typing indicator (reaction) to a message.
+ *
+ * Rate-limit and quota errors are re-thrown so the circuit breaker in
+ * `createTypingCallbacks` (typing-start-guard) can trip and stop the
+ * keepalive loop. See #28062.
+ *
+ * Also checks for backoff codes in non-throwing SDK responses (#28157).
  */
 export async function addTypingIndicator(params: {
   cfg: ClawdbotConfig;
@@ -36,18 +120,34 @@ export async function addTypingIndicator(params: {
       },
     });
 
+    // Feishu SDK may return a normal response with an API-level error code
+    // instead of throwing. Detect backoff codes and throw to trip the breaker.
+    const backoffCode = getBackoffCodeFromResponse(response);
+    if (backoffCode !== undefined) {
+      console.log(
+        `[feishu] typing indicator response contains backoff code ${backoffCode}, stopping keepalive`,
+      );
+      throw new FeishuBackoffError(backoffCode);
+    }
+
     // eslint-disable-next-line @typescript-eslint/no-explicit-any -- SDK response type
     const reactionId = (response as any)?.data?.reaction_id ?? null;
     return { messageId, reactionId };
   } catch (err) {
-    // Silently fail - typing indicator is not critical
+    if (isFeishuBackoffError(err)) {
+      console.log(`[feishu] typing indicator hit rate-limit/quota, stopping keepalive`);
+      throw err;
+    }
+    // Silently fail for other non-critical errors (e.g. message deleted, permission issues)
     console.log(`[feishu] failed to add typing indicator: ${err}`);
     return { messageId, reactionId: null };
   }
 }
 
 /**
- * Remove a typing indicator (reaction) from a message
+ * Remove a typing indicator (reaction) from a message.
+ *
+ * Rate-limit and quota errors are re-thrown for the same reason as above.
  */
 export async function removeTypingIndicator(params: {
   cfg: ClawdbotConfig;
@@ -67,14 +167,27 @@ export async function removeTypingIndicator(params: {
   const client = createFeishuClient(account);
 
   try {
-    await client.im.messageReaction.delete({
+    const result = await client.im.messageReaction.delete({
       path: {
         message_id: state.messageId,
         reaction_id: state.reactionId,
       },
     });
+
+    // Check for backoff codes in non-throwing SDK responses
+    const backoffCode = getBackoffCodeFromResponse(result);
+    if (backoffCode !== undefined) {
+      console.log(
+        `[feishu] typing indicator removal response contains backoff code ${backoffCode}, stopping keepalive`,
+      );
+      throw new FeishuBackoffError(backoffCode);
+    }
   } catch (err) {
-    // Silently fail - cleanup is not critical
+    if (isFeishuBackoffError(err)) {
+      console.log(`[feishu] typing indicator removal hit rate-limit/quota, stopping keepalive`);
+      throw err;
+    }
+    // Silently fail for other non-critical errors
     console.log(`[feishu] failed to remove typing indicator: ${err}`);
   }
 }