Handle JSONDecodeError while processing request data (#5599)

python/sglang/srt/openai_api/adapter.py

@@ -715,7 +715,10 @@ def v1_generate_response(
 
 
 async def v1_completions(tokenizer_manager, raw_request: Request):
-    request_json = await raw_request.json()
+    try:
+        request_json = await raw_request.json()
+    except Exception as e:
+        return create_error_response("Invalid request body, error: ", str(e))
     all_requests = [CompletionRequest(**request_json)]
     created = int(time.time())
     adapted_request, request = v1_generate_request(all_requests)
@@ -1379,7 +1382,10 @@ def v1_chat_generate_response(
 async def v1_chat_completions(
     tokenizer_manager, raw_request: Request, cache_report=False
 ):
-    request_json = await raw_request.json()
+    try:
+        request_json = await raw_request.json()
+    except Exception as e:
+        return create_error_response("Invalid request body, error: ", str(e))
     all_requests = [ChatCompletionRequest(**request_json)]
     created = int(time.time())
     adapted_request, request = v1_chat_generate_request(all_requests, tokenizer_manager)
@@ -1800,7 +1806,10 @@ def v1_embedding_response(ret, model_path, to_file=False):
 
 
 async def v1_embeddings(tokenizer_manager, raw_request: Request):
-    request_json = await raw_request.json()
+    try:
+        request_json = await raw_request.json()
+    except Exception as e:
+        return create_error_response("Invalid request body, error: ", str(e))
     all_requests = [EmbeddingRequest(**request_json)]
     adapted_request, request = v1_embedding_request(all_requests, tokenizer_manager)