[Misc] clang-format auto fix. (#4831)

* [Misc] clang-format auto fix.

* blabla

* nolint

* blabla
Co-authored-by: Steve <ubuntu@ip-172-31-34-29.ap-northeast-1.compute.internal>

src/array/cuda/bf16.cuh

-/*!
+/**
  *  Copyright (c) 2022 by Contributors
  *
  *  Licensed under the Apache License, Version 2.0 (the "License");
@@ -13,19 +13,19 @@
  *  See the License for the specific language governing permissions and
  *  limitations under the License.
  *
- * \file array/cuda/bf16.cuh
- * \brief bfloat16 related functions.
+ * @file array/cuda/bf16.cuh
+ * @brief bfloat16 related functions.
  */
 #ifndef DGL_ARRAY_CUDA_BF16_CUH_
 #define DGL_ARRAY_CUDA_BF16_CUH_
 
-
 #if BF16_ENABLED
 #include <cuda_bf16.h>
+
 #include <algorithm>
 
-static __device__ __forceinline__ __nv_bfloat16 max(__nv_bfloat16 a,
-    __nv_bfloat16 b) {
+static __device__ __forceinline__ __nv_bfloat16
+max(__nv_bfloat16 a, __nv_bfloat16 b) {
 #if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 800
   return __hmax(a, b);
 #else
@@ -33,8 +33,8 @@ static __device__ __forceinline__ __nv_bfloat16 max(__nv_bfloat16 a,
 #endif
 }
 
-static __device__ __forceinline__ __nv_bfloat16 min(__nv_bfloat16 a,
-    __nv_bfloat16 b) {
+static __device__ __forceinline__ __nv_bfloat16
+min(__nv_bfloat16 a, __nv_bfloat16 b) {
 #if defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= 800
   return __hmin(a, b);
 #else
@@ -43,53 +43,68 @@ static __device__ __forceinline__ __nv_bfloat16 min(__nv_bfloat16 a,
 }
 
 #ifdef __CUDACC__
-// Arithmetic BF16 operations for architecture >= 8.0 are already defined in cuda_bf16.h
+// Arithmetic BF16 operations for architecture >= 8.0 are already defined in
+// cuda_bf16.h
 #if defined(__CUDA_ARCH__) && (__CUDA_ARCH__ < 800)
-__device__ __forceinline__ __nv_bfloat16 operator+(
-    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
+__device__ __forceinline__ __nv_bfloat16
+operator+(const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return __nv_bfloat16(float(lh) + float(rh));  // NOLINT
 }
-__device__ __forceinline__ __nv_bfloat16 operator-(
-    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
+__device__ __forceinline__ __nv_bfloat16
+operator-(const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return __nv_bfloat16(float(lh) - float(rh));  // NOLINT
 }
-__device__ __forceinline__ __nv_bfloat16 operator*(
-    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
+__device__ __forceinline__ __nv_bfloat16
+operator*(const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return __nv_bfloat16(float(lh) * float(rh));  // NOLINT
 }
-__device__ __forceinline__ __nv_bfloat16 operator/(
-    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
+__device__ __forceinline__ __nv_bfloat16
+operator/(const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return __nv_bfloat16(float(lh) / float(rh));  // NOLINT
 }
 
 __device__ __forceinline__ __nv_bfloat16& operator+=(
     __nv_bfloat16& lh, const __nv_bfloat16& rh) {  // NOLINT
-  lh = __nv_bfloat16(float(lh) + float(rh)); return lh;  // NOLINT
+  lh = __nv_bfloat16(float(lh) + float(rh));       // NOLINT
+  return lh;
 }
 __device__ __forceinline__ __nv_bfloat16& operator-=(
     __nv_bfloat16& lh, const __nv_bfloat16& rh) {  // NOLINT
-  lh = __nv_bfloat16(float(lh) - float(rh)); return lh;  // NOLINT
+  lh = __nv_bfloat16(float(lh) - float(rh));       // NOLINT
+  return lh;
 }
 __device__ __forceinline__ __nv_bfloat16& operator*=(
     __nv_bfloat16& lh, const __nv_bfloat16& rh) {  // NOLINT
-  lh = __nv_bfloat16(float(lh) * float(rh)); return lh;  // NOLINT
+  lh = __nv_bfloat16(float(lh) * float(rh));       // NOLINT
+  return lh;
 }
 __device__ __forceinline__ __nv_bfloat16& operator/=(
     __nv_bfloat16& lh, const __nv_bfloat16& rh) {  // NOLINT
-  lh = __nv_bfloat16(float(lh) / float(rh)); return lh;  // NOLINT
+  lh = __nv_bfloat16(float(lh) / float(rh));       // NOLINT
+  return lh;
 }
 
-__device__ __forceinline__ __nv_bfloat16& operator++(__nv_bfloat16& h) {  // NOLINT
-  h = __nv_bfloat16(float(h) + 1.0f); return h;  // NOLINT
+__device__ __forceinline__ __nv_bfloat16& operator++(
+    __nv_bfloat16& h) {                // NOLINT
+  h = __nv_bfloat16(float(h) + 1.0f);  // NOLINT
+  return h;
 }
-__device__ __forceinline__ __nv_bfloat16& operator--(__nv_bfloat16& h) {  // NOLINT
-  h = __nv_bfloat16(float(h) - 1.0f); return h;  // NOLINT
+__device__ __forceinline__ __nv_bfloat16& operator--(
+    __nv_bfloat16& h) {                // NOLINT
+  h = __nv_bfloat16(float(h) - 1.0f);  // NOLINT
+  return h;
 }
-__device__ __forceinline__ __nv_bfloat16  operator++(__nv_bfloat16& h, int) {  // NOLINT
-  __nv_bfloat16 ret = h; h = __nv_bfloat16(float(h) + 1.0f); return ret;  // NOLINT
+__device__ __forceinline__ __nv_bfloat16
+operator++(__nv_bfloat16& h, int) {  // NOLINT
+  __nv_bfloat16 ret = h;
+  h = __nv_bfloat16(float(h) + 1.0f);  // NOLINT
+  return ret;
 }
-__device__ __forceinline__ __nv_bfloat16  operator--(__nv_bfloat16& h, int) {  // NOLINT
-  __nv_bfloat16 ret = h; h = __nv_bfloat16(float(h) - 1.0f); return ret;  // NOLINT
+__device__ __forceinline__ __nv_bfloat16
+operator--(__nv_bfloat16& h, int) {  // NOLINT
+  __nv_bfloat16 ret = h;
+  h = __nv_bfloat16(float(h) - 1.0f);  // NOLINT
+  return ret;
 }
 
 __device__ __forceinline__ __nv_bfloat16 operator+(const __nv_bfloat16& h) {
@@ -99,28 +114,28 @@ __device__ __forceinline__ __nv_bfloat16 operator-(const __nv_bfloat16& h) {
   return __nv_bfloat16(-float(h));  // NOLINT
 }
 
-__device__ __forceinline__ bool operator==(const __nv_bfloat16& lh,
-    const __nv_bfloat16& rh) {
+__device__ __forceinline__ bool operator==(
+    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return float(lh) == float(rh);  // NOLINT
 }
-__device__ __forceinline__ bool operator!=(const __nv_bfloat16& lh,
-    const __nv_bfloat16& rh) {
+__device__ __forceinline__ bool operator!=(
+    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return float(lh) != float(rh);  // NOLINT
 }
-__device__ __forceinline__ bool operator> (const __nv_bfloat16& lh,
-    const __nv_bfloat16& rh) {
+__device__ __forceinline__ bool operator>(
+    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return float(lh) > float(rh);  // NOLINT
 }
-__device__ __forceinline__ bool operator< (const __nv_bfloat16& lh,
-    const __nv_bfloat16& rh) {
+__device__ __forceinline__ bool operator<(
+    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return float(lh) < float(rh);  // NOLINT
 }
-__device__ __forceinline__ bool operator>=(const __nv_bfloat16& lh,
-    const __nv_bfloat16& rh) {
+__device__ __forceinline__ bool operator>=(
+    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return float(lh) >= float(rh);  // NOLINT
 }
-__device__ __forceinline__ bool operator<=(const __nv_bfloat16& lh,
-    const __nv_bfloat16& rh) {
+__device__ __forceinline__ bool operator<=(
+    const __nv_bfloat16& lh, const __nv_bfloat16& rh) {
   return float(lh) <= float(rh);  // NOLINT
 }
 #endif  // defined(__CUDA_ARCH__) && (__CUDA_ARCH__ < 800)

src/graph/serialize/dglstream.h

-/*!
+/**
  *  Copyright (c) 2019 by Contributors
- * \file graph/serialize/dglstream.h
- * \brief Graph serialization header
+ * @file graph/serialize/dglstream.h
+ * @brief Graph serialization header
  */
 #ifndef DGL_GRAPH_SERIALIZE_DGLSTREAM_H_
 #define DGL_GRAPH_SERIALIZE_DGLSTREAM_H_
@@ -9,20 +9,22 @@
 #include <dgl/aten/spmat.h>
 #include <dmlc/io.h>
 #include <dmlc/type_traits.h>
+
 #include <memory>
 
 namespace dgl {
 namespace serialize {
 
-/*!
- * \brief DGLStream counts the bytes that already written into the
+/**
+ * @brief DGLStream counts the bytes that already written into the
  * underlying stream.
  */
 class DGLStream : public dmlc::Stream {
  public:
-  /*! \brief create a new DGLStream instance */
-  static DGLStream *Create(const char *uri, const char *const flag,
-                                  bool allow_null, dgl_format_code_t formats) {
+  /** @brief create a new DGLStream instance */
+  static DGLStream *Create(
+      const char *uri, const char *const flag, bool allow_null,
+      dgl_format_code_t formats) {
     return new DGLStream(uri, flag, allow_null, formats);
   }
 
@@ -45,10 +47,11 @@ class DGLStream : public dmlc::Stream {
   uint64_t FormatsToSave() const { return formats_to_save_; }
 
  private:
-  DGLStream(const char *uri, const char *const flag, bool allow_null,
+  DGLStream(
+      const char *uri, const char *const flag, bool allow_null,
       dgl_format_code_t formats)
-      : strm_(dmlc::Stream::Create(uri, flag, allow_null)), formats_to_save_(formats) {
-  }
+      : strm_(dmlc::Stream::Create(uri, flag, allow_null)),
+        formats_to_save_(formats) {}
   // stream for serialization
   std::unique_ptr<dmlc::Stream> strm_;
   // size of already written to stream

src/graph/traversal.cc

@@ -99,7 +99,9 @@ struct Frontiers {
   /** @brief a vector store for the nodes/edges in all the frontiers */
   std::vector<dgl_id_t> ids;
 
-  /** @brief a vector store for node/edge tags. Empty if no tags are requested */
+  /**
+   * @brief a vector store for node/edge tags. Empty if no tags are requested
+   */
   std::vector<int64_t> tags;
 
   /** @brief a section vector to indicate each frontier */