perf(workflow): add tests, adapt some LLM node output formats, optimize sandbox return format

sandbox/app/controllers/sandbox_controller.py

@@ -36,7 +36,7 @@ async def run_code(request: RunCodeRequest):
     elif request.language == "javascript":
         return await run_nodejs_code(request.code, request.preload, request.options)
     else:
-        return error_response(-400, "unsupported language")
+        return error_response(400, "unsupported language")
 
 
 @router.get("/dependencies", response_model=ApiResponse)
@@ -45,7 +45,7 @@ async def get_dependencies(language: str):
     if language == "python3":
         return await list_python_dependencies()
     else:
-        return error_response(-400, "unsupported language")
+        return error_response(400, "unsupported language")
 
 
 @router.post("/dependencies/update", response_model=ApiResponse)
@@ -54,4 +54,4 @@ async def update_dependencies(request: UpdateDependencyRequest):
     if request.language == "python3":
         return await update_python_dependencies()
     else:
-        return error_response(-400, "unsupported language")
+        return error_response(400, "unsupported language")

sandbox/app/models.py

@@ -75,6 +75,4 @@ def success_response(data: Any) -> ApiResponse:
 
 def error_response(code: int, message: str) -> ApiResponse:
     """Create error response"""
-    if code >= 0:
-        code = -1
     return ApiResponse(code=code, message=message, data=None)

sandbox/app/services/nodejs_service.py

@@ -27,11 +27,11 @@ async def run_nodejs_code(code: str, preload: str, options: RunnerOptions):
     try:
         runner = NodejsRunner()
         result = await runner.run(code, options, preload)
-        if result.exit_code == signal.SIGSYS + 0x80:
+        if result.exit_code in [signal.SIGSYS + 0x80, -signal.SIGSYS]:
             return error_response(31, "sandbox security policy violation")
 
         if result.exit_code != 0:
-            return error_response(500, result.stderr)
+            return error_response(result.exit_code, result.stderr)
 
         return success_response(RunCodeResponse(
             stdout=result.stdout,
@@ -39,5 +39,5 @@ async def run_nodejs_code(code: str, preload: str, options: RunnerOptions):
         ))
 
     except Exception as e:
-        logger.error(f"Python execution failed: {e}", exc_info=True)
-        return error_response(-500, str(e))
+        logger.error(f"JavaScript execution failed: {e}", exc_info=True)
+        return error_response(500, str(e))

sandbox/app/services/python_service.py

@@ -47,7 +47,7 @@ async def run_python_code(code: str, preload: str, options: RunnerOptions):
         
     except Exception as e:
         logger.error(f"Python execution failed: {e}", exc_info=True)
-        return error_response(-500, str(e))
+        return error_response(500, str(e))
 
 
 async def list_python_dependencies():