[router] Support history management using conversation (#11339)

sgl-router/tests/responses_api_test.rs

@@ -125,6 +125,7 @@ async fn test_non_streaming_mcp_minimal_e2e_with_persistence() {
         top_k: -1,
         min_p: 0.0,
         repetition_penalty: 1.0,
+        conversation: None,
     };
 
     let resp = router
@@ -371,6 +372,7 @@ fn test_responses_request_creation() {
         top_k: -1,
         min_p: 0.0,
         repetition_penalty: 1.0,
+        conversation: None,
     };
 
     assert!(!request.is_stream());
@@ -411,6 +413,7 @@ fn test_sampling_params_conversion() {
         top_k: 10,
         min_p: 0.05,
         repetition_penalty: 1.1,
+        conversation: None,
     };
 
     let params = request.to_sampling_params(1000, None);
@@ -524,6 +527,7 @@ fn test_json_serialization() {
         top_k: 50,
         min_p: 0.1,
         repetition_penalty: 1.2,
+        conversation: None,
     };
 
     let json = serde_json::to_string(&request).expect("Serialization should work");
@@ -651,6 +655,7 @@ async fn test_multi_turn_loop_with_mcp() {
         top_k: 50,
         min_p: 0.0,
         repetition_penalty: 1.0,
+        conversation: None,
     };
 
     // Execute the request (this should trigger the multi-turn loop)
@@ -828,6 +833,7 @@ async fn test_max_tool_calls_limit() {
         top_k: 50,
         min_p: 0.0,
         repetition_penalty: 1.0,
+        conversation: None,
     };
 
     let response = router.route_responses(None, &req, None).await;
@@ -1023,6 +1029,7 @@ async fn test_streaming_with_mcp_tool_calls() {
         top_k: 50,
         min_p: 0.0,
         repetition_penalty: 1.0,
+        conversation: None,
     };
 
     let response = router.route_responses(None, &req, None).await;
@@ -1301,6 +1308,7 @@ async fn test_streaming_multi_turn_with_mcp() {
         top_k: 50,
         min_p: 0.0,
         repetition_penalty: 1.0,
+        conversation: None,
     };
 
     let response = router.route_responses(None, &req, None).await;

sgl-router/tests/test_openai_routing.rs

@@ -9,6 +9,7 @@ use axum::{
     Json, Router,
 };
 use serde_json::json;
+use sglang_router_rs::data_connector::MemoryConversationItemStorage;
 use sglang_router_rs::{
     config::{
         ConfigError, ConfigValidator, HistoryBackend, OracleConfig, RouterConfig, RoutingMode,
@@ -95,6 +96,7 @@ async fn test_openai_router_creation() {
         None,
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await;
 
@@ -113,6 +115,7 @@ async fn test_openai_router_server_info() {
         None,
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -143,6 +146,7 @@ async fn test_openai_router_models() {
         None,
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -222,6 +226,7 @@ async fn test_openai_router_responses_with_mock() {
         None,
         storage.clone(),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -482,6 +487,7 @@ async fn test_openai_router_responses_streaming_with_mock() {
         None,
         storage.clone(),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -586,6 +592,7 @@ async fn test_unsupported_endpoints() {
         None,
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -627,6 +634,7 @@ async fn test_openai_router_chat_completion_with_mock() {
         None,
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -669,6 +677,7 @@ async fn test_openai_e2e_with_server() {
         None,
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -739,6 +748,7 @@ async fn test_openai_router_chat_streaming_with_mock() {
         None,
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -792,6 +802,7 @@ async fn test_openai_router_circuit_breaker() {
         Some(cb_config),
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();
@@ -820,6 +831,7 @@ async fn test_openai_router_models_auth_forwarding() {
         None,
         Arc::new(MemoryResponseStorage::new()),
         Arc::new(MemoryConversationStorage::new()),
+        Arc::new(sglang_router_rs::data_connector::MemoryConversationItemStorage::new()),
     )
     .await
     .unwrap();