adding precommit hook (#132)

undefined
DiamondLightSource · Jul 10, 2024 · 053e780 · 053e780
1 parent 83b05be
commit 053e780
Show file tree

Hide file tree

Showing 19 changed files with 152 additions and 100 deletions.
diff --git a/.github/workflows/conda_upload.yml b/.github/workflows/conda_upload.yml
@@ -29,7 +29,7 @@ jobs:
 
       # install dependencies with conda
       - name: Install dependencies with conda
-        run: |      
+        run: |
           $CONDA/bin/conda env create --name httomolibgpu --file conda/environment.yml
           $CONDA/bin/conda run -n httomolibgpu pip install -e .
           $CONDA/bin/conda list

diff --git a/.github/workflows/tests_run.yml b/.github/workflows/tests_run.yml
@@ -41,5 +41,5 @@ jobs:
           micromamba list
 
       - name: Run tests
-        run: |     
+        run: |
           pytest tests/
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
@@ -0,0 +1,34 @@
+repos:
+- repo: https://github.com/psf/black
+  rev: 23.12.1
+  hooks:
+    - id: black
+      language_version: python3
+      exclude: ^docs/source/examples/
+- repo: https://github.com/pre-commit/pre-commit-hooks
+  rev: v4.5.0
+  hooks:
+    - id: trailing-whitespace  # This hook trims trailing whitespace.
+    - id: check-docstring-first  # Checks a common error of defining a docstring after code.
+    - id: check-merge-conflict  # Check for files that contain merge conflict strings.
+    - id: check-yaml  # This hook checks yaml files for parseable syntax.
+    - id: detect-private-key  # Detects the presence of private keys.
+    - id: check-symlinks
+    - id: check-toml
+- repo: https://github.com/pre-commit/pygrep-hooks
+  rev: v1.10.0
+  hooks:
+    - id: python-use-type-annotations
+- repo: https://github.com/kynan/nbstripout
+  rev: 0.6.1
+  hooks:
+    - id: nbstripout
+      exclude: ^docs/source/examples/
+-   repo: https://github.com/asottile/blacken-docs
+    rev: 1.16.0
+    hooks:
+    - id: blacken-docs
+- repo: https://github.com/nbQA-dev/nbQA
+  rev: 1.7.1
+  hooks:
+    - id: nbqa-black
diff --git a/README.rst b/README.rst
@@ -2,17 +2,17 @@ HTTomolibGPU is a library of GPU accelerated methods for tomography
 --------------------------------------------------------------------
 
 **HTTomolibGPU** is a collection of image processing methods in Python for computed tomography.
-The methods are GPU-accelerated with the open-source Python library `CuPy <https://cupy.dev/>`_. Most of the 
+The methods are GPU-accelerated with the open-source Python library `CuPy <https://cupy.dev/>`_. Most of the
 methods migrated from `TomoPy <https://tomopy.readthedocs.io/en/stable/>`_ and `Savu <https://savu.readthedocs.io/en/latest/>`_ software packages.
 Some of the methods also have been optimised to ensure higher computational efficiency, before ported to CuPy.
 
 The purpose of HTTomolibGPU
 ===========================
 
-**HTTomolibGPU** can be used as a stand-alone library, see Examples section in `Documentation <https://diamondlightsource.github.io/httomolibgpu/>`_. 
-However, it has been specifically developed to work together with the `HTTomo <https://diamondlightsource.github.io/httomo/>`_ package as 
-its backend for data processing. HTTomo is a user interface (UI) written in Python for fast big tomographic data processing using 
-MPI protocols. 
+**HTTomolibGPU** can be used as a stand-alone library, see Examples section in `Documentation <https://diamondlightsource.github.io/httomolibgpu/>`_.
+However, it has been specifically developed to work together with the `HTTomo <https://diamondlightsource.github.io/httomo/>`_ package as
+its backend for data processing. HTTomo is a user interface (UI) written in Python for fast big tomographic data processing using
+MPI protocols.
 
 Install HTTomolibGPU as a pre-built conda Python package
 =========================================================
@@ -28,7 +28,7 @@ Setup the development environment:
 ==================================
 
 .. code-block:: console
-    
+
    $ git clone [email protected]:DiamondLightSource/httomolibgpu.git # clone the repo
    $ conda env create --name httomolibgpu --file conda/environment.yml # install dependencies
    $ conda activate httomolibgpu # activate the environment

diff --git a/docs/source/examples/Cor_largesino.ipynb b/docs/source/examples/Cor_largesino.ipynb
@@ -43,10 +43,12 @@
     "\n",
     "# Load the sinogram data\n",
     "path_lib = os.path.dirname(httomolibgpu.__file__)\n",
-    "in_file = os.path.abspath(os.path.join(path_lib, '..', \"tests/test_data/\",\"3600proj_sino.npz\"))\n",
+    "in_file = os.path.abspath(\n",
+    "    os.path.join(path_lib, \"..\", \"tests/test_data/\", \"3600proj_sino.npz\")\n",
+    ")\n",
     "l_infile = np.load(in_file)\n",
-    "sinogram = l_infile['sinogram']\n",
-    "angles = l_infile['angles']\n",
+    "sinogram = l_infile[\"sinogram\"]\n",
+    "angles = l_infile[\"angles\"]\n",
     "sinogram = cp.asarray(sinogram)\n",
     "\n",
     "print(\"The shape of the sinogram is {}\".format(cp.shape(sinogram)))"
@@ -95,7 +97,8 @@
    ],
    "source": [
     "from httomolibgpu.recon.rotation import find_center_vo\n",
-    "print (\"Finding the Center of Rotation for the reconstruction\")\n",
+    "\n",
+    "print(\"Finding the Center of Rotation for the reconstruction\")\n",
     "cor = find_center_vo(sinogram)\n",
     "print(\"The found Center of Rotation is {}\".format(cor))"
    ]
@@ -126,15 +129,18 @@
    ],
    "source": [
     "from httomolibgpu.prep.stripe import remove_all_stripe\n",
-    "print (\"Remove stripes using remove_all_stripe\")\n",
     "\n",
-    "sino3d = cp.ascontiguousarray(cp.ones((sinogram.shape[0], 3, sinogram.shape[1])), dtype=cp.float32)\n",
-    "sino3d[:,1,:] = sinogram\n",
+    "print(\"Remove stripes using remove_all_stripe\")\n",
+    "\n",
+    "sino3d = cp.ascontiguousarray(\n",
+    "    cp.ones((sinogram.shape[0], 3, sinogram.shape[1])), dtype=cp.float32\n",
+    ")\n",
+    "sino3d[:, 1, :] = sinogram\n",
     "\n",
     "sino3d = remove_all_stripe(sino3d)\n",
     "\n",
     "plt.figure()\n",
-    "plt.imshow(sino3d[:,1,:].get())\n",
+    "plt.imshow(sino3d[:, 1, :].get())\n",
     "plt.title(\"Sinogram after stripes removal\")\n",
     "plt.show()"
    ]
@@ -164,14 +170,14 @@
     }
    ],
    "source": [
-    "print (\"Perform Reconstruction using FBP\")\n",
+    "print(\"Perform Reconstruction using FBP\")\n",
     "from httomolibgpu.recon.algorithm import FBP\n",
     "\n",
     "reconFBP = FBP(sino3d, angles=angles, center=cor)\n",
     "reconFBP_np = reconFBP.get()\n",
     "\n",
     "plt.figure()\n",
-    "plt.imshow(reconFBP_np[500:2000,1,500:2000],vmin=0,vmax=0.01)\n",
+    "plt.imshow(reconFBP_np[500:2000, 1, 500:2000], vmin=0, vmax=0.01)\n",
     "plt.title(\"Reconstruction with FBP\")\n",
     "plt.show()"
    ]

diff --git a/docs/source/examples/DistortionCorr.ipynb b/docs/source/examples/DistortionCorr.ipynb
diff --git a/docs/source/examples/pipeline1_FBP.ipynb b/docs/source/examples/pipeline1_FBP.ipynb
@@ -51,19 +51,25 @@
     "\n",
     "# Load the projection data\n",
     "path_lib = os.path.dirname(httomolibgpu.__file__)\n",
-    "in_file = os.path.abspath(os.path.join(path_lib, '..', \"tests/test_data/\",'tomo_standard.npz'))\n",
+    "in_file = os.path.abspath(\n",
+    "    os.path.join(path_lib, \"..\", \"tests/test_data/\", \"tomo_standard.npz\")\n",
+    ")\n",
     "datafile = np.load(in_file)\n",
-    "host_data = datafile['data']\n",
-    "host_flats = datafile['flats']\n",
-    "host_darks = datafile['darks']\n",
+    "host_data = datafile[\"data\"]\n",
+    "host_flats = datafile[\"flats\"]\n",
+    "host_darks = datafile[\"darks\"]\n",
     "\n",
-    "print(\"The shape of the data is {} as (projections, detector Y, detector X)\".format(np.shape(host_data)))\n",
+    "print(\n",
+    "    \"The shape of the data is {} as (projections, detector Y, detector X)\".format(\n",
+    "        np.shape(host_data)\n",
+    "    )\n",
+    ")\n",
     "\n",
     "print(\"Normalising the data\")\n",
     "data = cp.asarray(host_data)\n",
     "flats = cp.asarray(host_flats)\n",
     "darks = cp.asarray(host_darks)\n",
-    "data_normalised = normalize(data, flats, darks, cutoff = 10, minus_log = False)\n",
+    "data_normalised = normalize(data, flats, darks, cutoff=10, minus_log=False)\n",
     "\n",
     "sliceSel = 64\n",
     "data_normalised_np = data_normalised.get()\n",
@@ -100,7 +106,7 @@
    "source": [
     "from httomolibgpu.recon.rotation import find_center_vo\n",
     "\n",
-    "print (\"Finding the Center of Rotation for the reconstruction\")\n",
+    "print(\"Finding the Center of Rotation for the reconstruction\")\n",
     "cor = find_center_vo(data_normalised, ind=64)\n",
     "print(\"The found Center of Rotation is {}\".format(cor))"
    ]
@@ -131,8 +137,10 @@
    "source": [
     "from httomolibgpu.prep.phase import paganin_filter_tomopy\n",
     "\n",
-    "print (\"Applying Paganin filter\")\n",
-    "phase_contrast_data = paganin_filter_tomopy(data_normalised, pixel_size=0.1, dist=50, energy=53, alpha=1e-5)\n",
+    "print(\"Applying Paganin filter\")\n",
+    "phase_contrast_data = paganin_filter_tomopy(\n",
+    "    data_normalised, pixel_size=0.1, dist=50, energy=53, alpha=1e-5\n",
+    ")\n",
     "\n",
     "sliceSel = 64\n",
     "phase_contrast_data_np = phase_contrast_data.get()\n",
@@ -176,12 +184,12 @@
     }
    ],
    "source": [
-    "print (\"Perform Reconstruction using FBP\")\n",
+    "print(\"Perform Reconstruction using FBP\")\n",
     "\n",
     "from httomolibgpu.recon.algorithm import FBP\n",
     "\n",
     "angles = np.linspace(0.0 * np.pi / 180.0, 180.0 * np.pi / 180.0, data.shape[0])\n",
-    "        \n",
+    "\n",
     "reconFBP = FBP(phase_contrast_data, angles=angles, center=cor)\n",
     "\n",
     "reconFBP_np = reconFBP.get()\n",
@@ -226,12 +234,12 @@
     }
    ],
    "source": [
-    "print (\"Perform Reconstruction using Fourier (LPRec)\")\n",
+    "print(\"Perform Reconstruction using Fourier (LPRec)\")\n",
     "\n",
     "from httomolibgpu.recon.algorithm import LPRec\n",
     "\n",
     "angles = np.linspace(0.0 * np.pi / 180.0, 180.0 * np.pi / 180.0, data.shape[0])\n",
-    "        \n",
+    "\n",
     "reconLPRec = LPRec(phase_contrast_data, angles=angles, center=cor)\n",
     "\n",
     "reconLPRec_np = reconLPRec.get()\n",

diff --git a/docs/source/examples/pipeline2_iterative.ipynb b/docs/source/examples/pipeline2_iterative.ipynb
diff --git a/docs/source/index.rst b/docs/source/index.rst
@@ -19,7 +19,7 @@
     examples/pipeline1_FBP
     examples/pipeline2_iterative
     examples/Cor_largesino
-    examples/DistortionCorr    
+    examples/DistortionCorr
 
 .. _reference_content:
 

diff --git a/httomolibgpu/cuda_kernels/__init__.py b/httomolibgpu/cuda_kernels/__init__.py
@@ -2,8 +2,10 @@
 from typing import List, Optional, Tuple
 
 from httomolibgpu import cupywrapper
+
 cp = cupywrapper.cp
 
+
 def load_cuda_module(
     file: str,
     name_expressions: Optional[List[str]] = None,

diff --git a/httomolibgpu/cuda_kernels/calc_metrics.cu b/httomolibgpu/cuda_kernels/calc_metrics.cu
@@ -10,17 +10,17 @@
  *
  * The key is the formula to calculate the Peason correlation coefficient.
  * This is calculated manually for every shifted matrix position in the same kernel.
- * 
+ *
  * The correlation coefficient between two vectors (we flatten the matrices) is:
  *
  * m1_norm = m1 - mean(m1)
  * m2_norm = m2 - mean(m2)
- * m1_sqr = dot(m1_norm, m1_norm) 
+ * m1_sqr = dot(m1_norm, m1_norm)
  * m2_sqr = dot(m2_norm, m2_norm)
  * m1_m2  = dot(m1_norm, m2_norm)
  * r = m1_m2 / sqrt(m1_sqr * m2_sqr)
  *
- * The kernels in the following compute these directly pretty much, taking into 
+ * The kernels in the following compute these directly pretty much, taking into
  * consideration normalisation, overlaps, and position offsets. Also note that the
  * version with overlap requries 3 correlation coefficients (between 3 matrices).
  */
@@ -73,7 +73,7 @@ float clip(float x, float min, float max) {
 }
 
 __device__ inline
-float sum_abs_row(const float* row, int win_width) 
+float sum_abs_row(const float* row, int win_width)
 {
     float sum_abs = 0.0;
     for (int x = 0; x < win_width; ++x) {
@@ -116,7 +116,7 @@ __device__ void _calc_metrics_no_overlap(const float *mat1, int mat1_nx,
     {
         float norm_factor = 1.0f;
         if (norm) {
-            norm_factor = sum_abs_row(&mat2_roi[y * mat2_nx], win_width) / 
+            norm_factor = sum_abs_row(&mat2_roi[y * mat2_nx], win_width) /
                 sum_abs_row(&mat1_roi[y * mat1_nx], win_width);
         }
         for (int x = 0; x < win_width; ++x)
@@ -142,7 +142,7 @@ __device__ void _calc_metrics_no_overlap(const float *mat1, int mat1_nx,
     {
         float norm_factor = 1.0f;
         if (norm) {
-            norm_factor = sum_abs_row(&mat2_roi[y * mat2_nx], win_width) / 
+            norm_factor = sum_abs_row(&mat2_roi[y * mat2_nx], win_width) /
                 sum_abs_row(&mat1_roi[y * mat1_nx], win_width);
         }
         for (int x = 0; x < win_width; ++x)
@@ -200,10 +200,10 @@ __device__ void _calc_metrics_overlap(const float *mat1, int mat1_nx,
     extern __shared__ float smem[];
 
     // we need to space for 6 sum reductions for calculating the correlation coefficient
-    float v[6]; 
+    float v[6];
 
     float d_ramp = 1.0f / (win_width - 1);
-    
+
     ////////////////////////
     // 1. We  need the mean of the 3 matrices (flattend)
     v[0] = 0.0f;
@@ -213,7 +213,7 @@ __device__ void _calc_metrics_overlap(const float *mat1, int mat1_nx,
     {
         float norm_factor = 1.0f;
         if (norm) {
-            norm_factor = sum_abs_row(&mat2_roi[y * mat2_nx], win_width) / 
+            norm_factor = sum_abs_row(&mat2_roi[y * mat2_nx], win_width) /
                 sum_abs_row(&mat1_roi[y * mat1_nx], win_width);
         }
         for (int x = 0; x < win_width; ++x)
@@ -222,8 +222,8 @@ __device__ void _calc_metrics_overlap(const float *mat1, int mat1_nx,
             float ramp_up = 1.0f - ramp_down;
             float mat1_roi_val = mat1_roi[y * mat1_nx + x] * norm_factor;
             float mat2_roi_val = mat2_roi[y * mat2_nx + x];
-            float mat_comb_val = side == 1 ? 
-                (mat1_roi_val * ramp_down + mat2_roi_val * ramp_up) : 
+            float mat_comb_val = side == 1 ?
+                (mat1_roi_val * ramp_down + mat2_roi_val * ramp_up) :
                 (mat1_roi_val * ramp_up + mat2_roi_val * ramp_down);
 
             v[0] += mat1_roi_val;
@@ -251,7 +251,7 @@ __device__ void _calc_metrics_overlap(const float *mat1, int mat1_nx,
     {
         float norm_factor = 1.0f;
         if (norm) {
-            norm_factor = sum_abs_row(&mat2_roi[y * mat2_nx], win_width) / 
+            norm_factor = sum_abs_row(&mat2_roi[y * mat2_nx], win_width) /
                 sum_abs_row(&mat1_roi[y * mat1_nx], win_width);
         }
         for (int x = 0; x < win_width; ++x)
@@ -260,8 +260,8 @@ __device__ void _calc_metrics_overlap(const float *mat1, int mat1_nx,
             float ramp_up = 1.0f - ramp_down;
             float mat1_roi_val = mat1_roi[y * mat1_nx + x] * norm_factor;
             float mat2_roi_val = mat2_roi[y * mat2_nx + x];
-            float mat_comb_val = side == 1 ? 
-                (mat1_roi_val * ramp_down + mat2_roi_val * ramp_up) : 
+            float mat_comb_val = side == 1 ?
+                (mat1_roi_val * ramp_down + mat2_roi_val * ramp_up) :
                 (mat1_roi_val * ramp_up + mat2_roi_val * ramp_down);
 
             // for covariance matrix, we need to remove the mean first
@@ -279,7 +279,7 @@ __device__ void _calc_metrics_overlap(const float *mat1, int mat1_nx,
         }
 
     }
-    
+
     // 6 smem reductions
     sum_reduction_n<6>(smem, v);
 

diff --git a/httomolibgpu/misc/corr.py b/httomolibgpu/misc/corr.py
@@ -79,7 +79,6 @@ def __median_filter(
     axis: int = 0,
     dif: float = 0.0,
 ) -> cp.ndarray:
-
     try:
         from cucim.skimage.filters import median
         from cucim.skimage.morphology import disk

diff --git a/httomolibgpu/misc/morph.py b/httomolibgpu/misc/morph.py
@@ -67,7 +67,6 @@ def sino_360_to_180(
 def __sino_360_to_180(
     data: cp.ndarray, overlap: int = 0, rotation: Literal["left", "right"] = "left"
 ) -> cp.ndarray:
-
     if data.ndim != 3:
         raise ValueError("only 3D data is supported")
 
@@ -136,7 +135,6 @@ def data_resampler(
 def __data_resampler(
     data: cp.ndarray, newshape: list, axis: int = 1, interpolation: str = "linear"
 ) -> cp.ndarray:
-
     from cupyx.scipy.interpolate import interpn
 
     if data.ndim != 3:
@@ -206,7 +204,6 @@ def __data_resampler(
                 res, [newshape[0], newshape[1]], order="C"
             )
     elif axis == 1:
-
         for j in range(M):
             res = interpn(
                 points,