[SYCLomatic] Update two the YMAL migration rules in pytorch_api.yaml on stream migration(#2772)

Signed-off-by: Jiang, Zhiwei <zhiwei.jiang@intel.com>

Author: zhiweij1

clang/test/dpct/pytorch/ATen.cu

@@ -55,7 +55,7 @@ void test_CUDAStream_as_arg() {
 }
 
 int main() {
-  // CHECK: dpct::queue_ptr st = &static_cast<sycl::queue&>(c10::xpu::getCurrentXPUStream());
+  // CHECK: dpct::queue_ptr st = &c10::xpu::getCurrentXPUStream().queue();
   cudaStream_t st = 0;
 
   // stream APIs
@@ -66,9 +66,9 @@ int main() {
   // CHECK: auto deviceStream = c10::xpu::getCurrentXPUStream(devInd);
   auto deviceStream = at::cuda::getCurrentCUDAStream(devInd);
 
-  // CHECK: dpct::queue_ptr curr_cuda_st = &static_cast<sycl::queue &>(c10::xpu::getCurrentXPUStream(). queue());
+  // CHECK: dpct::queue_ptr curr_cuda_st = &(c10::xpu::getCurrentXPUStream(). queue());
   cudaStream_t curr_cuda_st = at::cuda::getCurrentCUDAStream().stream();
-  // CHECK: dpct::queue_ptr dev_cuda_st = &static_cast<sycl::queue &>(c10::xpu::getCurrentXPUStream(devInd). queue());
+  // CHECK: dpct::queue_ptr dev_cuda_st = &(c10::xpu::getCurrentXPUStream(devInd). queue());
   cudaStream_t dev_cuda_st = at::cuda::getCurrentCUDAStream(devInd).stream();
 
   test_CUDAStream_as_arg();
@@ -77,7 +77,7 @@ int main() {
 }
 
 // CHECK: void foo2(c10::DeviceGuard device_guard, float *f)  try {
-// CHECK-NEXT: (DPCT_CHECK_ERROR(f = (float *)sycl::malloc_device(4, static_cast<sycl::queue&>(c10::xpu::getCurrentXPUStream()))));
+// CHECK-NEXT: (DPCT_CHECK_ERROR(f = (float *)sycl::malloc_device(4, c10::xpu::getCurrentXPUStream().queue())));
 void foo2(at::cuda::CUDAGuard device_guard, float *f) {
   C10_CUDA_CHECK(cudaMalloc(&f, 4));
 }

clang/test/dpct/pytorch/c10.cu

@@ -63,11 +63,9 @@ int main() {
   // CHECK: auto currentStream = c10::xpu::getCurrentXPUStream();
   auto currentStream = c10::cuda::getCurrentCUDAStream();
 
-  // CHECK: dpct::queue_ptr curr_cuda_st =
-  // CHECK-NEXT: &static_cast<sycl::queue &>(currentStream.queue());
+  // CHECK: dpct::queue_ptr curr_cuda_st = &(currentStream.queue());
   cudaStream_t curr_cuda_st = currentStream.stream();
-  // CHECK: curr_cuda_st =
-  // CHECK-NEXT: &static_cast<sycl::queue &>(c10::xpu::getCurrentXPUStream().queue());
+  // CHECK: curr_cuda_st = &(c10::xpu::getCurrentXPUStream().queue());
   curr_cuda_st = c10::cuda::getCurrentCUDAStream().stream();
 
   // CHECK: auto deviceStream = c10::xpu::getCurrentXPUStream(0);

clang/tools/dpct/extensions/pytorch_api_rules/pytorch_api.yaml

@@ -87,7 +87,7 @@
   Out: c10::xpu::XPUStream
   Methods:
     - In: stream
-      Out: \&static_cast<sycl::queue &>($method_base queue())
+      Out: \&($method_base queue())
 
 - Rule: rule_c10_cuda_getCurrentCUDAStream
   Kind: API
@@ -165,7 +165,7 @@
   Kind: HelperFunction
   Priority: Takeover
   In: get_in_order_queue
-  Out: static_cast<sycl::queue&>(c10::xpu::getCurrentXPUStream())
+  Out: c10::xpu::getCurrentXPUStream().queue()
   Includes: [<c10/xpu/XPUStream.h>]
 
 - Rule: rule_THC_THCAtomics_cuh