Support clang format and format almost all C++ code (#423)

* refactor: format almost all code

* fix: update IncludeCategories

* fix: make sure structmember.h is before Python.h

.clang-format

@@ -0,0 +1,3 @@
+---
+Language: Cpp
+BasedOnStyle: Google

.github/workflows/format.yml

@@ -0,0 +1,22 @@
+name: code format
+
+on:
+  workflow_dispatch:
+  pull_request:
+  push:
+    branches:
+      - main
+
+jobs:
+  cpp-linter:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - uses: cpp-linter/cpp-linter-action@v2
+        id: linter
+        with:
+          style: file
+          tidy-checks: '-*' # disable clang tidy at this stage
+      - name: Fail test
+        if: steps.linter.outputs.checks-failed > 0
+        run: echo "Some files failed the linting checks!" && exit 1

dicp/.clang-format

@@ -0,0 +1,3 @@
+---
+DisableFormat: true
+SortIncludes: Never

dipu/.clang-format

@@ -0,0 +1,36 @@
+---
+BasedOnStyle: InheritParentConfig
+IncludeCategories:
+  - Regex:         '^("|<)csrc_dipu/'
+    Priority:      90
+    CaseSensitive: false
+  - Regex:         '^("|<)diopi/'
+    Priority:      80
+    CaseSensitive: false
+  - Regex:         '^("|<)(c10|aten|torch)/'
+    Priority:      40
+    CaseSensitive: false
+  - Regex:         '^("|<)Python\.h'
+    Priority:      50
+    CaseSensitive: false
+  - Regex:         '^("|<)(frameobject|structmember)\.h'
+    Priority:      50
+    SortPriority:  51
+    CaseSensitive: false
+  - Regex:         '^("|<)(pybind11)'
+    Priority:      50
+    SortPriority:  52
+    CaseSensitive: false
+  - Regex:         '^<((ext/.*)|pthread)\.h'
+    Priority:      20
+    SortPriority:  21
+    CaseSensitive: false
+  - Regex:         '^("|<)(cuda|su|cn|(..?ccl)|(.*_runtime)).*\.h'
+    Priority:      30
+    CaseSensitive: false
+  - Regex:         '^<.*'
+    Priority:      20
+    CaseSensitive: false
+  - Regex:         '.*'
+    Priority:      100
+    CaseSensitive: false

dipu/tests/cpp/test_relu.cpp

@@ -1,21 +1,22 @@
 // Copyright (c) 2023, DeepLink.
-#include <torch/torch.h>
 #include <iostream>
 
-void testTensorRelu(at::Tensor& self) {
-    std::cout << self << std::endl;
-    std::cout << torch::relu(self) << std::endl;
-    std::cout << self << std::endl;
+#include <torch/torch.h>
+
+void testTensorRelu(at::Tensor &self) {
+  std::cout << self << std::endl;
+  std::cout << torch::relu(self) << std::endl;
+  std::cout << self << std::endl;
 
-    std::cout << torch::relu_(self) << std::endl;
-    std::cout << self << std::endl;
+  std::cout << torch::relu_(self) << std::endl;
+  std::cout << self << std::endl;
 }
 
 int main() {
-    torch::Tensor tensor = torch::randn(10).cuda();
-    testTensorRelu(tensor);
+  torch::Tensor tensor = torch::randn(10).cuda();
+  testTensorRelu(tensor);
 
-    torch::Tensor tensor_cpu = torch::randn(10);
-    testTensorRelu(tensor_cpu);
-    return 0;
+  torch::Tensor tensor_cpu = torch::randn(10);
+  testTensorRelu(tensor_cpu);
+  return 0;
 }

dipu/tests/cpp/test_tensor_add.cpp

@@ -1,20 +1,20 @@
 // Copyright (c) 2023, DeepLink.
 #include <iostream>
 
-#include <ATen/Tensor.h>
 #include <ATen/ATen.h>
+#include <ATen/Tensor.h>
 
-void testTensorAdd(const at::Tensor& lhs, const at::Tensor& rhs) {
-    at::Tensor result = lhs + rhs;
-    std::cout << lhs << std::endl;
-    std::cout << rhs << std::endl;
-    std::cout << result << std::endl;
+void testTensorAdd(const at::Tensor &lhs, const at::Tensor &rhs) {
+  at::Tensor result = lhs + rhs;
+  std::cout << lhs << std::endl;
+  std::cout << rhs << std::endl;
+  std::cout << result << std::endl;
 }
 
 int main() {
-    at::Tensor t0 = at::randn({2, 2}).cuda();
-    at::Tensor t1 = at::randn({2}).cuda();
-    testTensorAdd(t0, t1);
-    testTensorAdd(t0.cpu(), t1.cpu());
-    return 0;
+  at::Tensor t0 = at::randn({2, 2}).cuda();
+  at::Tensor t1 = at::randn({2}).cuda();
+  testTensorAdd(t0, t1);
+  testTensorAdd(t0.cpu(), t1.cpu());
+  return 0;
 }

dipu/tests/cpp/testrt.cpp

@@ -1,8 +1,10 @@
 // Copyright (c) 2023, DeepLink.
-#include <torch/torch.h>
 #include <iostream>
-#include <csrc_dipu/runtime/devproxy/deviceproxy.h>
+
+#include <torch/torch.h>
+
 #include <csrc_dipu/runtime/core/DIPUStream.h>
+#include <csrc_dipu/runtime/devproxy/deviceproxy.h>
 
 using namespace dipu;
 void testcopy() {
@@ -53,12 +55,11 @@ void testStream1() {
   auto stream3 = getCurrentDIPUStream();
   rawStream = stream3.rawstream();
   std::cout << "current stream =" << rawStream << std::endl;
-
 }
 
 // need change to use gtest.
 int main() {
-  for(int i=0; i<3; i++) {
+  for (int i = 0; i < 3; i++) {
     // testcopy();
     testDeviceSwitch();
     // testStream1();

dipu/third_party/.clang-format

@@ -0,0 +1,3 @@
+---
+DisableFormat: true
+SortIncludes: Never