added kernel boilerplate

.gitignore

@@ -2,5 +2,7 @@ __pycache__/
 _ext/
 build/
 dist/
+.cache/
 .eggs/
 *.egg-info/
+.coverage

.travis.yml

+language: python
+sudo: required
+dist: trusty
+matrix:
+  include:
+    - python: 2.7
+    - python: 3.5
+    - python: 3.6
+install:
+  - if [[ $TRAVIS_PYTHON_VERSION == 2.7 ]]; then pip install --quiet http://download.pytorch.org/whl/cu80/torch-0.3.0.post4-cp27-cp27mu-linux_x86_64.whl; fi
+  - if [[ $TRAVIS_PYTHON_VERSION == 3.5 ]]; then pip install --quiet http://download.pytorch.org/whl/cu80/torch-0.3.0.post4-cp35-cp35m-linux_x86_64.whl; fi
+  - if [[ $TRAVIS_PYTHON_VERSION == 3.6 ]]; then pip install --quiet http://download.pytorch.org/whl/cu80/torch-0.3.0.post4-cp36-cp36m-linux_x86_64.whl; fi
+  - pip install pycodestyle
+  - pip install flake8
+  - pip install cffi
+  - pip install codecov
+script:
+  - pycodestyle .
+  - flake8 .
+  - python setup.py install
+  - python setup.py test
+after_success:
+  - codecov
+notifications:
+  email: false

MANIFEST.in

 include LICENSE
+include build.py
+include build.sh
+
+recursive-include torch_cluster/src *
+recursive-include torch_cluster/kernel *
+
+recursive-exclude torch_cluster/_ext *

README.md

+[pypi-image]: https://badge.fury.io/py/torch-cluster.svg
+[pypi-url]: https://pypi.python.org/pypi/torch-cluster
+[build-image]: https://travis-ci.org/rusty1s/pytorch_cluster.svg?branch=master
+[build-url]: https://travis-ci.org/rusty1s/pytorch_cluster
+[coverage-image]: https://codecov.io/gh/rusty1s/pytorch_cluster/branch/master/graph/badge.svg
+[coverage-url]: https://codecov.io/github/rusty1s/pytorch_cluster?branch=master
+
 # PyTorch Cluster
 
+[![PyPI Version][pypi-image]][pypi-url]
+[![Build Status][build-image]][build-url]
+[![Code Coverage][coverage-image]][coverage-url]
+
 --------------------------------------------------------------------------------
 
 This package consists of a small extension library of highly optimised graph cluster algorithms for the use in [PyTorch](http://pytorch.org/).
@@ -12,7 +23,7 @@ If not, add cuda (`/usr/local/cuda/bin`) to your `$PATH`.
 Then run:
 
 ```
-pip install cffi
+pip install cffi torch-cluster
 python setup.py install
 ```
 

build.py

@@ -11,7 +11,7 @@ extra_objects = []
 with_cuda = False
 
 if torch.cuda.is_available():
-    subprocess.call('./build.sh')  # Compile kernel.
+    subprocess.call('./build.sh')
 
     headers += ['torch_cluster/src/cuda.h']
     sources += ['torch_cluster/src/cuda.c']

build.sh

 #!/bin/sh
 
+echo "Compiling kernel..."
+
 TORCH=$(python -c "import os; import torch; print(os.path.dirname(torch.__file__))")
-SRC_DIR=torch_unique/kernel
-BUILD_DIR=torch_unique/build
+SRC_DIR=torch_cluster/kernel
+BUILD_DIR=torch_cluster/build
 
 mkdir -p $BUILD_DIR
 nvcc -c -o $BUILD_DIR/kernel.so $SRC_DIR/kernel.cu -arch=sm_35 -Xcompiler -fPIC -shared -I$TORCH/lib/include/TH -I$TORCH/lib/include/THC -I$SRC_DIR

setup.py

@@ -2,8 +2,6 @@ from os import path as osp
 
 from setuptools import setup, find_packages
 
-import build  # noqa
-
 __version__ = '0.1.0'
 url = 'https://github.com/rusty1s/pytorch_cluster'
 
@@ -14,7 +12,8 @@ tests_require = ['pytest', 'pytest-cov']
 setup(
     name='torch_cluster',
     version=__version__,
-    description='PyTorch Geometric Deep Learning Graph Cluster Algorithms',
+    description='PyTorch Extension Library of Optimised Graph Cluster '
+    'Algorithms',
     author='Matthias Fey',
     author_email='matthias.fey@tu-dortmund.de',
     url=url,

test/utils.py

+import torch
+from torch._tensor_docs import tensor_classes
+
+tensors = [t[:-4] for t in tensor_classes]
+
+
+def Tensor(str, x):
+    tensor = getattr(torch, str)
+    return tensor(x)

torch_cluster/kernel/generic/kernel.cu

+#ifndef THC_GENERIC_FILE
+#define THC_GENERIC_FILE "generic/kernel.cu"
+#else
+
+void cluster_(grid)(THCState *state, int C, THCudaLongTensor *output, THCTensor *position, THCTensor *size, THCudaLongTensor *count) {
+}
+
+#endif

torch_cluster/kernel/kernel.cu

+#include <THC.h>
+
+#include "kernel.h"
+
+#define cluster_(NAME) TH_CONCAT_4(cluster, NAME, _kernel_, Real)
+
+#include "generic/kernel.cu"
+#include "THCGenerateAllTypes.h"

torch_cluster/kernel/kernel.h

+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void cluster_grid_kernel_Float (THCState *state, int C, THCudaLongTensor *output, THCudaTensor       *position, THCudaTensor       *size, THCudaLongTensor *count);
+void cluster_grid_kernel_Double(THCState *state, int C, THCudaLongTensor *output, THCudaDoubleTensor *position, THCudaDoubleTensor *size, THCudaLongTensor *count);
+void cluster_grid_kernel_Byte  (THCState *state, int C, THCudaLongTensor *output, THCudaByteTensor   *position, THCudaByteTensor   *size, THCudaLongTensor *count);
+void cluster_grid_kernel_Char  (THCState *state, int C, THCudaLongTensor *output, THCudaCharTensor   *position, THCudaCharTensor   *size, THCudaLongTensor *count);
+void cluster_grid_kernel_Short (THCState *state, int C, THCudaLongTensor *output, THCudaShortTensor  *position, THCudaShortTensor  *size, THCudaLongTensor *count);
+void cluster_grid_kernel_Int   (THCState *state, int C, THCudaLongTensor *output, THCudaIntTensor    *position, THCudaIntTensor    *size, THCudaLongTensor *count);
+void cluster_grid_kernel_Long  (THCState *state, int C, THCudaLongTensor *output, THCudaLongTensor   *position, THCudaLongTensor   *size, THCudaLongTensor *count);
+
+#ifdef __cplusplus
+}
+#endif

torch_cluster/src/cuda.c

+#include <THC/THC.h>
+
+#include "kernel.h"
+
+#define cluster_(NAME) TH_CONCAT_4(cluster_, NAME, _cuda_, Real)
+#define cluster_kernel_(NAME) TH_CONCAT_4(cluster_, NAME, _kernel_, Real)
+
+extern THCState *state;
+
+#include "generic/cuda.c"
+#include "THCGenerateFloatType.h"
+#include "generic/cuda.c"
+#include "THCGenerateDoubleType.h"
+#include "generic/cuda.c"
+#include "THCGenerateByteType.h"
+#include "generic/cuda.c"
+#include "THCGenerateCharType.h"
+#include "generic/cuda.c"
+#include "THCGenerateShortType.h"
+#include "generic/cuda.c"
+#include "THCGenerateIntType.h"
+#include "generic/cuda.c"
+#include "THCGenerateLongType.h"

torch_cluster/src/cuda.h

+void cluster_grid_cuda_Float (int C, THCudaLongTensor *output, THCudaTensor       *position, THCudaTensor       *size, THCudaLongTensor *count);
+void cluster_grid_cuda_Double(int C, THCudaLongTensor *output, THCudaDoubleTensor *position, THCudaDoubleTensor *size, THCudaLongTensor *count);
+void cluster_grid_cuda_Byte  (int C, THCudaLongTensor *output, THCudaByteTensor   *position, THCudaByteTensor   *size, THCudaLongTensor *count);
+void cluster_grid_cuda_Char  (int C, THCudaLongTensor *output, THCudaCharTensor   *position, THCudaCharTensor   *size, THCudaLongTensor *count);
+void cluster_grid_cuda_Short (int C, THCudaLongTensor *output, THCudaShortTensor  *position, THCudaShortTensor  *size, THCudaLongTensor *count);
+void cluster_grid_cuda_Int   (int C, THCudaLongTensor *output, THCudaIntTensor    *position, THCudaIntTensor    *size, THCudaLongTensor *count);
+void cluster_grid_cuda_Long  (int C, THCudaLongTensor *output, THCudaLongTensor   *position, THCudaLongTensor   *size, THCudaLongTensor *count);

torch_cluster/src/generic/cuda.c

+#ifndef THC_GENERIC_FILE
+#define THC_GENERIC_FILE "generic/cuda.c"
+#else
+
+void cluster_(grid)(int C, THCudaLongTensor *output, THCTensor *position, THCTensor *size, THCudaLongTensor *count) {
+  return cluster_kernel_(grid)(state, C, output, position, size, count);
+}
+
+#endif