Add GitHub action to format and lint code (#71)

* Add and run pre-commit hooks

* Restore clang-format

* Fix yaml spacing

* Normalize spacing

* Fix indentation of pre-commit-config.yaml

* Clang to enforce 80 chars, pre-commit all PRs

* Update copyrights

* Remove extra line

Author: dyastremsky

.clang-format

@@ -2,6 +2,7 @@
 BasedOnStyle: Google
 
 IndentWidth: 2
+ColumnLimit: 80
 ContinuationIndentWidth: 4
 UseTab: Never
 MaxEmptyLinesToKeep: 2

.github/workflows/pre-commit.yml

@@ -0,0 +1,39 @@
+# Copyright 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#  * Redistributions of source code must retain the above copyright
+#    notice, this list of conditions and the following disclaimer.
+#  * Redistributions in binary form must reproduce the above copyright
+#    notice, this list of conditions and the following disclaimer in the
+#    documentation and/or other materials provided with the distribution.
+#  * Neither the name of NVIDIA CORPORATION nor the names of its
+#    contributors may be used to endorse or promote products derived
+#    from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+name: pre-commit
+
+on:
+  pull_request:
+
+jobs:
+  pre-commit:
+    runs-on: ubuntu-22.04
+    steps:
+    - uses: actions/checkout@v3
+    - uses: actions/setup-python@v3
+    - uses: pre-commit/[email protected]
+

.pre-commit-config.yaml

@@ -0,0 +1,73 @@
+# Copyright 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#  * Redistributions of source code must retain the above copyright
+#    notice, this list of conditions and the following disclaimer.
+#  * Redistributions in binary form must reproduce the above copyright
+#    notice, this list of conditions and the following disclaimer in the
+#    documentation and/or other materials provided with the distribution.
+#  * Neither the name of NVIDIA CORPORATION nor the names of its
+#    contributors may be used to endorse or promote products derived
+#    from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+repos:
+- repo: https://github.com/timothycrosley/isort
+  rev: 5.12.0
+  hooks:
+  - id: isort
+    additional_dependencies: [toml]
+- repo: https://github.com/psf/black
+  rev: 23.1.0
+  hooks:
+  - id: black
+    types_or: [python, cython]
+- repo: https://github.com/PyCQA/flake8
+  rev: 5.0.4
+  hooks:
+  - id: flake8
+    args: [--max-line-length=88, --select=C,E,F,W,B,B950, --extend-ignore = E203,E501]
+    types_or: [python, cython]
+- repo: https://github.com/pre-commit/mirrors-clang-format
+  rev: v16.0.5
+  hooks:
+  - id: clang-format
+    types_or: [c, c++, cuda, proto, textproto, java]
+    args: ["-fallback-style=none", "-style=file", "-i"]
+- repo: https://github.com/codespell-project/codespell
+  rev: v2.2.4
+  hooks:
+  - id: codespell
+    additional_dependencies: [tomli]
+    args: ["--toml", "pyproject.toml"]
+    exclude: (?x)^(.*stemmer.*|.*stop_words.*|^CHANGELOG.md$)
+# More details about these pre-commit hooks here:
+# https://pre-commit.com/hooks.html
+- repo: https://github.com/pre-commit/pre-commit-hooks
+  rev: v4.4.0
+  hooks:
+  - id: check-case-conflict
+  - id: check-executables-have-shebangs
+  - id: check-merge-conflict
+  - id: check-json
+  - id: check-toml
+  - id: check-yaml
+  - id: check-shebang-scripts-are-executable
+  - id: end-of-file-fixer
+    types_or: [c, c++, cuda, proto, textproto, java, python]
+  - id: mixed-line-ending
+  - id: requirements-txt-fixer
+  - id: trailing-whitespace

CMakeLists.txt

@@ -1,4 +1,4 @@
-# Copyright 2021-2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# Copyright 2021-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
 # modification, are permitted provided that the following conditions
@@ -51,7 +51,7 @@ set(TRITON_TENSORRT_BACKEND_INSTALLDIR ${CMAKE_INSTALL_PREFIX}/backends/tensorrt
 #
 # Dependencies
 #
-# FetchContent's composibility isn't very good. We must include the
+# FetchContent's composability isn't very good. We must include the
 # transitive closure of all repos so that we can override the tag.
 #
 include(FetchContent)

README.md

@@ -1,5 +1,5 @@
 <!--
-# Copyright 2021, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# Copyright 2021-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
 # modification, are permitted provided that the following conditions
@@ -30,7 +30,7 @@
 
 # TensorRT Backend
 
-The Triton backend for [TensorRT](https://github.com/NVIDIA/TensorRT). 
+The Triton backend for [TensorRT](https://github.com/NVIDIA/TensorRT).
 You can learn more about Triton backends in the [backend
 repo](https://github.com/triton-inference-server/backend). Ask
 questions or report problems on the [issues

pyproject.toml

@@ -0,0 +1,49 @@
+# Copyright 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#  * Redistributions of source code must retain the above copyright
+#    notice, this list of conditions and the following disclaimer.
+#  * Redistributions in binary form must reproduce the above copyright
+#    notice, this list of conditions and the following disclaimer in the
+#    documentation and/or other materials provided with the distribution.
+#  * Neither the name of NVIDIA CORPORATION nor the names of its
+#    contributors may be used to endorse or promote products derived
+#    from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+[tool.codespell]
+# note: pre-commit passes explicit lists of files here, which this skip file list doesn't override -
+# this is only to allow you to run codespell interactively
+skip = "./.git,./.github"
+# ignore short words, and typename parameters like OffsetT
+ignore-regex = "\\b(.{1,4}|[A-Z]\\w*T)\\b"
+# use the 'clear' dictionary for unambiguous spelling mistakes
+builtin = "clear"
+# disable warnings about binary files and wrong encoding
+quiet-level = 3
+
+[tool.isort]
+profile = "black"
+use_parentheses = true
+multi_line_output = 3
+include_trailing_comma = true
+force_grid_wrap = 0
+ensure_newline_before_comments = true
+line_length = 88
+balanced_wrapping = true
+indent = "    "
+skip = ["build"]
+

src/instance_state.cc

@@ -1071,7 +1071,7 @@ ModelInstanceState::Run(
       io_binding_info.batch_output_ = model_state_->FindBatchOutput(name);
 
       // Process the output tensors with pinned memory address if zero-copy is
-      // supported, otherwise use device memory. Peform memory copies
+      // supported, otherwise use device memory. Perform memory copies
       // asynchronously and wait for model execution.
       payload_->responder_->ProcessBatchOutput(
           name, *(io_binding_info.batch_output_),
@@ -1116,7 +1116,7 @@ ModelInstanceState::Run(
 
       if (io_binding_info.is_requested_output_tensor_) {
         // Process the output tensors with pinned memory address if zero-copy is
-        // supported, otherwise use device memory. Peform memory copies
+        // supported, otherwise use device memory. Perform memory copies
         // asynchronously and wait for model execution.
         payload_->responder_->ProcessTensor(
             name, dt, batchn_shape,
@@ -2244,7 +2244,7 @@ ModelInstanceState::InitializeBatchInputBindings(
         }
       } else {
         // For most type 'dims' will be empty as the full shape
-        // of the batch input is [-1] which will be coverred by
+        // of the batch input is [-1] which will be covered by
         // batch dimension.
         switch (batch_input.BatchInputKind()) {
           case BatchInput::Kind::BATCH_ELEMENT_COUNT:
@@ -2481,7 +2481,7 @@ ModelInstanceState::InitializeConfigShapeOutputBindings(
       // [DLIS-4283] review below comment
       // Allocate CUDA memory. Use cudaHostAlloc if zero copy
       // supported. We rely on buffer_bindings_ being non-nullptr to
-      // indicate that the buffer has been correctly initalized so
+      // indicate that the buffer has been correctly initialized so
       // even for zero-sized tensors always allocate something.
       void* buffer = nullptr;
       cudaError_t err = cudaSuccess;
@@ -2727,7 +2727,7 @@ ModelInstanceState::InitializeExecuteInputBinding(
 
   // Allocate CUDA memory. Use cudaHostAlloc if zero copy supported.
   // We rely on buffer_bindings_ being non-nullptr to indicate that
-  // the buffer has been correctly initalized so even for zero-sized
+  // the buffer has been correctly initialized so even for zero-sized
   // tensors always allocate something.
   void* buffer = nullptr;
   cudaError_t err = cudaSuccess;
@@ -2901,7 +2901,7 @@ ModelInstanceState::InitializeExecuteOutputBinding(
 
   // Allocate CUDA memory. Use cudaHostAlloc if zero copy supported.
   // We rely on buffer_bindings_ being non-nullptr to indicate that
-  // the buffer has been correctly initalized so even for zero-sized
+  // the buffer has been correctly initialized so even for zero-sized
   // tensors always allocate something.
   void* buffer = nullptr;
   cudaError_t err = cudaSuccess;
@@ -3111,7 +3111,7 @@ ModelInstanceState::InitializeShapeInputBinding(
   if (max_byte_size != 0) {
     // Allocate CUDA memory. Use cudaHostAlloc if zero copy supported.
     // We rely on buffer_bindings_ being non-nullptr to indicate that
-    // the buffer has been correctly initalized so even for zero-sized
+    // the buffer has been correctly initialized so even for zero-sized
     // tensors always allocate something.
     void* buffer = nullptr;
     cudaError_t err = cudaSuccess;
@@ -4131,7 +4131,7 @@ TRTv3Interface::SetBindingDimensions(
   }
 
   if (!trt_context.is_dynamic_per_binding_[io_index]) {
-    // No need to set dimension for the binding that does not inlcude
+    // No need to set dimension for the binding that does not include
     // dynamic shape.
     return nullptr;
   }

src/instance_state.h

@@ -1,4 +1,4 @@
-// Copyright 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// Copyright 2022-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 //
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions
@@ -111,7 +111,7 @@ struct TensorRTContext {
 
   // The key is packed input dimensions prepended by batch size, so
   // that uniqueness is guaranteed and the CUDA graphs are sorted to
-  // provide convinence to find the closest CUDA graph in the
+  // provide convenience to find the closest CUDA graph in the
   // future.
   //
   // vector is used to map index of event sets to corresponding
@@ -181,7 +181,7 @@ class TRTInterface {
   virtual bool Enqueue(nvinfer1::IExecutionContext* context) = 0;
 
   // This function will be called to specify the runtime shape of the input and
-  // adding metadata into exisiting 'cuda_graph_key' for graph lookup.
+  // adding metadata into existing 'cuda_graph_key' for graph lookup.
   virtual TRITONSERVER_Error* SetBindingDimensions(
       const std::string& input_name, const std::vector<int64_t>& shape,
       const TensorRTContext& trt_context, const size_t io_index,

src/model_state.cc

@@ -240,7 +240,7 @@ ModelState::CreateEngine(
             .c_str());
 
     if (IsEngineSharingEnabled()) {
-      // This logic runs atleast once to validate whether the engine
+      // This logic runs at least once to validate whether the engine
       // can be shared.
       bool is_dynamic = false;
       for (int idx = 0; idx < eit->second.second->getNbBindings(); idx++) {
@@ -929,4 +929,4 @@ ModelState::FixIO(
   return nullptr;
 }
 
-}}}  // namespace triton::backend::tensorrt
+}}}  // namespace triton::backend::tensorrt

src/model_state.h

@@ -75,7 +75,7 @@ class ModelState : public TensorRTModel {
 
   // Query the execution arbitrator to return the instance for the execution and
   // the semaphore to check whether the next execution should be initiated.
-  // 'device_id', 'instance' are the metadata assoicated with the
+  // 'device_id', 'instance' are the metadata associated with the
   // TRITONBACKEND_ModelInstance.
   std::pair<ModelInstanceState*, Semaphore*> ExecutionState(
       const int device_id, ModelInstanceState* instance)

src/tensorrt.cc

@@ -111,7 +111,7 @@ TRITONBACKEND_Initialize(TRITONBACKEND_Backend* backend)
     RETURN_IF_ERROR(backend_config.Parse(buffer, byte_size));
   }
 
-  // Default execution policy, may be overriden by backend config
+  // Default execution policy, may be overridden by backend config
   auto execution_policy = TRITONBACKEND_EXECUTION_DEVICE_BLOCKING;
 
   std::unique_ptr<BackendConfiguration> lconfig(new BackendConfiguration());