EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[router] Support multiple worker URLs for OpenAI router (#11723)

Browse Source
This commit is contained in:
Keyang Ru
2025-10-22 09:27:58 -07:00
committed by GitHub
parent 1d097aac87
commit 77258ce039
9 changed files with 426 additions and 150 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
sgl-router/tests/responses_api_test.rs
View File

@@ -100,7 +100,7 @@ async fn test_non_streaming_mcp_minimal_e2e_with_persistence() {
max_output_tokens: Some(64),
max_tool_calls: None,
metadata: None,
model: Some("mock-model".to_string()),
model: "mock-model".to_string(),
parallel_tool_calls: Some(true),
previous_response_id: None,
reasoning: None,
@@ -134,7 +134,7 @@ async fn test_non_streaming_mcp_minimal_e2e_with_persistence() {
};
let resp = router
.route_responses(None, &req, req.model.as_deref())
.route_responses(None, &req, Some(req.model.as_str()))
.await;
assert_eq!(resp.status(), StatusCode::OK);
@@ -349,7 +349,7 @@ fn test_responses_request_creation() {
max_output_tokens: Some(100),
max_tool_calls: None,
metadata: None,
model: Some("test-model".to_string()),
model: "test-model".to_string(),
parallel_tool_calls: Some(true),
previous_response_id: None,
reasoning: Some(ResponseReasoningParam {
@@ -397,7 +397,7 @@ fn test_responses_request_sglang_extensions() {
max_output_tokens: Some(50),
max_tool_calls: None,
metadata: None,
model: Some("test-model".to_string()),
model: "test-model".to_string(),
parallel_tool_calls: Some(true),
previous_response_id: None,
reasoning: None,
@@ -506,7 +506,7 @@ fn test_json_serialization() {
max_output_tokens: Some(200),
max_tool_calls: Some(5),
metadata: None,
model: Some("gpt-4".to_string()),
model: "gpt-4".to_string(),
parallel_tool_calls: Some(false),
previous_response_id: None,
reasoning: Some(ResponseReasoningParam {
@@ -545,7 +545,7 @@ fn test_json_serialization() {
parsed.request_id,
Some("resp_comprehensive_test".to_string())
);
assert_eq!(parsed.model, Some("gpt-4".to_string()));
assert_eq!(parsed.model, "gpt-4");
assert_eq!(parsed.background, Some(true));
assert_eq!(parsed.stream, Some(true));
assert_eq!(parsed.tools.as_ref().map(|t| t.len()), Some(1));
@@ -636,7 +636,7 @@ async fn test_multi_turn_loop_with_mcp() {
max_output_tokens: Some(128),
max_tool_calls: None, // No limit - test unlimited
metadata: None,
model: Some("mock-model".to_string()),
model: "mock-model".to_string(),
parallel_tool_calls: Some(true),
previous_response_id: None,
reasoning: None,
@@ -812,7 +812,7 @@ async fn test_max_tool_calls_limit() {
max_output_tokens: Some(128),
max_tool_calls: Some(1), // Limit to 1 call
metadata: None,
model: Some("mock-model".to_string()),
model: "mock-model".to_string(),
parallel_tool_calls: Some(true),
previous_response_id: None,
reasoning: None,
@@ -1006,7 +1006,7 @@ async fn test_streaming_with_mcp_tool_calls() {
max_output_tokens: Some(256),
max_tool_calls: Some(3),
metadata: None,
model: Some("mock-model".to_string()),
model: "mock-model".to_string(),
parallel_tool_calls: Some(true),
previous_response_id: None,
reasoning: None,
@@ -1287,7 +1287,7 @@ async fn test_streaming_multi_turn_with_mcp() {
max_output_tokens: Some(512),
max_tool_calls: Some(5), // Allow multiple rounds
metadata: None,
model: Some("mock-model".to_string()),
model: "mock-model".to_string(),
parallel_tool_calls: Some(true),
previous_response_id: None,
reasoning: None,