[build]: Lint/format tooling + CI (add-lint-tooling)

.clang-format

@@ -0,0 +1,9 @@
+# clang-format configuration for the CUDA/C++ extension (libs/ccc_cuda_ext).
+# Microsoft base style already uses Allman braces and 4-space indentation,
+# which matches the majority style of the existing sources, minimizing churn.
+---
+BasedOnStyle: Microsoft
+IndentWidth: 4
+ColumnLimit: 120
+PointerAlignment: Right
+SortIncludes: CaseSensitive

.clang-tidy

@@ -0,0 +1,29 @@
+# clang-tidy configuration for the CUDA/C++ extension (libs/ccc_cuda_ext).
+#
+# Curated check list targeting the bug classes actually found in review
+# (integer narrowing, unchecked returns). Run against the exported
+# compile_commands.json, host-only pass:
+#
+#   mamba run -n ccc-gpu-dev clang-tidy \
+#     -p build \
+#     --extra-arg=--cuda-host-only \
+#     libs/ccc_cuda_ext/*.cu
+#
+# CI runs this report-only at first; promote to blocking once the baseline
+# is clean.
+Checks: >
+  -*,
+  clang-analyzer-*,
+  bugprone-*,
+  performance-*,
+  modernize-*,
+  readability-braces-around-statements,
+  readability-else-after-return,
+  cppcoreguidelines-init-variables,
+  misc-unused-parameters,
+  -modernize-use-trailing-return-type,
+  -modernize-avoid-c-arrays,
+  -bugprone-easily-swappable-parameters
+WarningsAsErrors: ''
+HeaderFilterRegex: 'libs/ccc_cuda_ext/.*\.cuh$'
+FormatStyle: file

.git-blame-ignore-revs

@@ -0,0 +1,11 @@
+# Revisions listed here are skipped by `git blame` so that large mechanical
+# formatting commits do not obscure authorship of the surrounding code.
+#
+# Enable locally (one-time):
+#   git config blame.ignoreRevsFile .git-blame-ignore-revs
+#
+# GitHub honors this file automatically.
+#
+# Append the SHA of the isolated "apply ruff-format + clang-format" commit
+# from the add-lint-tooling change below, e.g.:
+#   <full-40-char-sha>  # style: apply ruff-format + clang-format (add-lint-tooling)

.github/workflows/ci.yml

@@ -0,0 +1,151 @@
+name: CI
+
+# Runs on every push to the default branch and on every pull request.
+#
+# NOTE: GPU tests (tests/gpu/**) are intentionally NOT run here — hosted GitHub
+# runners have no CUDA device. They are a documented LOCAL command:
+#   mamba run -n ccc-gpu-dev python -m pytest tests/gpu -q -o addopts=""
+# CI covers: (a) lint, (b) a no-device wheel/extension build smoke, and
+# (c) the CPU-only test subset.
+on:
+  push:
+    branches: [main, master]
+  pull_request:
+
+# Keep tool versions in sync with .pre-commit-config.yaml and the dev env.
+env:
+  RUFF_VERSION: "0.15.1"
+  CLANG_FORMAT_VERSION: "22.1.8"
+  CUDA_IMAGE: "nvidia/cuda:12.5.1-devel-ubuntu22.04"
+
+concurrency:
+  group: ci-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  # ---------------------------------------------------------------------- #
+  # (a) Lint: Python (ruff) + CUDA/C++ (clang-format). Pure CPU, no CUDA.
+  # ---------------------------------------------------------------------- #
+  lint:
+    name: Lint (ruff + clang-format)
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install linters
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install "ruff==${RUFF_VERSION}" "clang-format==${CLANG_FORMAT_VERSION}"
+
+      - name: ruff check
+        run: ruff check libs/ccc tests
+
+      - name: ruff format --check
+        run: ruff format --check libs/ccc tests
+
+      - name: clang-format --dry-run --Werror (CUDA/C++ extension)
+        run: |
+          clang-format --dry-run --Werror \
+            libs/ccc_cuda_ext/*.cu libs/ccc_cuda_ext/*.cuh
+
+  # ---------------------------------------------------------------------- #
+  # (b) Build smoke: compile the CUDA extension in a CUDA devel container.
+  #     nvcc compiles for CMAKE_CUDA_ARCHITECTURES=75 without any GPU device
+  #     present, so this catches compilation breakage on hosted runners.
+  # ---------------------------------------------------------------------- #
+  build:
+    name: Build smoke (no GPU device)
+    runs-on: ubuntu-latest
+    container:
+      image: nvidia/cuda:12.5.1-devel-ubuntu22.04
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install base tools
+        run: |
+          apt-get update
+          apt-get install -y --no-install-recommends git curl ca-certificates build-essential
+
+      - name: Install Miniforge
+        run: |
+          curl -fsSL -o /tmp/miniforge.sh \
+            https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-x86_64.sh
+          bash /tmp/miniforge.sh -b -p /opt/conda
+          echo "/opt/conda/bin" >> "$GITHUB_PATH"
+
+      - name: Build & install the package
+        env:
+          CUDACXX: /usr/local/cuda/bin/nvcc
+          CUDA_HOME: /usr/local/cuda
+          PATH: /opt/conda/bin:/usr/local/cuda/bin:/usr/bin:/bin
+        run: |
+          python -m pip install --upgrade pip
+          # Build isolation pulls the build backend (scikit-build-core, pybind11,
+          # cmake>=4, ninja) declared in pyproject [build-system].
+          python -m pip install . -v
+
+      - name: Import CPU path (smoke)
+        run: |
+          python - <<'PY'
+          import ccc
+          from ccc.coef import ccc as ccc_fn  # triggers numba CPU compile at import
+          import numpy as np
+          print("CPU coefficient:", ccc_fn(np.random.rand(20), np.random.rand(20)))
+          print("import OK:", ccc.__name__)
+          PY
+
+  # ---------------------------------------------------------------------- #
+  # (c) CPU-only tests: the CPU subset of the suite. GPU, network, and
+  #     undeclared-dep tests are ignored (see BASELINE). The two known
+  #     np.unicode_ test-only failures are deselected until restructure-tests
+  #     (PR D) fixes the NumPy-2 alias.
+  #     TODO(PR D): once pytest markers land, replace the --ignore/--deselect
+  #     list with `-m "not gpu and not slow and not network"`.
+  # ---------------------------------------------------------------------- #
+  test-cpu:
+    name: CPU tests
+    runs-on: ubuntu-latest
+    container:
+      image: nvidia/cuda:12.5.1-devel-ubuntu22.04
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install base tools
+        run: |
+          apt-get update
+          apt-get install -y --no-install-recommends git curl ca-certificates build-essential
+
+      - name: Install Miniforge
+        run: |
+          curl -fsSL -o /tmp/miniforge.sh \
+            https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-Linux-x86_64.sh
+          bash /tmp/miniforge.sh -b -p /opt/conda
+          echo "/opt/conda/bin" >> "$GITHUB_PATH"
+
+      - name: Build & install the package (+ test deps)
+        env:
+          CUDACXX: /usr/local/cuda/bin/nvcc
+          CUDA_HOME: /usr/local/cuda
+          PATH: /opt/conda/bin:/usr/local/cuda/bin:/usr/bin:/bin
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install ".[test]" -v
+
+      - name: Run CPU test subset
+        env:
+          PATH: /opt/conda/bin:/usr/local/cuda/bin:/usr/bin:/bin
+        run: |
+          python -m pytest tests/ \
+            --ignore=tests/gpu \
+            --ignore=tests/test_giant.py \
+            --ignore=tests/test_methods.py \
+            --ignore=tests/test_plots.py \
+            --ignore=tests/test_corr.py \
+            --deselect "tests/test_coef_pval.py::test_cm_numerical_and_categorical_features_perfect_relationship_pvalue" \
+            --deselect "tests/test_coef_pval.py::test_cm_numerical_and_categorical_features_weakly_relationship_pvalue" \
+            -q -ra -p no:cacheprovider -o addopts=""

.pre-commit-config.yaml

@@ -10,19 +10,19 @@ repos:
       - id: check-toml
       - id: check-added-large-files
       - id: mixed-line-ending
+  # Python: ruff is the single source of truth for lint + format (black dropped).
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.8.5  # Ruff version.
+    rev: v0.15.1  # Ruff version (keep in sync with the dev env / CI).
     hooks:
       - id: ruff  # Run the linter.
         types_or: [python, pyi]
         args: [--fix]
       - id: ruff-format  # Run the formatter.
         types_or: [python, pyi]
-  - repo: https://github.com/psf/black
-    rev: 24.2.0  # Replace with desired version
+  # C++/CUDA: clang-format, scoped to the extension sources via file filter.
+  - repo: https://github.com/pre-commit/mirrors-clang-format
+    rev: v22.1.8  # clang-format version (matches the dev env / CI).
     hooks:
-      - id: black
-        types_or: [python, pyi]
-      - id: black-jupyter
-        types_or: [python, pyi, jupyter]
-        additional_dependencies: [".[jupyter]"]
+      - id: clang-format
+        types_or: [c++, c, cuda]
+        files: ^libs/ccc_cuda_ext/.*\.(cu|cuh|cpp|cc|h)$

CMakeLists.txt

@@ -1,6 +1,11 @@
 cmake_minimum_required(VERSION 3.15...3.26)
 project(${SKBUILD_PROJECT_NAME} LANGUAGES CUDA CXX)
 
+# Export compile_commands.json so clang-tidy (and editors/LSP) can analyze the
+# CUDA/C++ sources. Run clang-tidy host-only, e.g.:
+#   clang-tidy -p build --extra-arg=--cuda-host-only libs/ccc_cuda_ext/*.cu
+set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
+
 # Add this near the top of your file, after project()
 # Define the include directories for the whole project
 set(PROJECT_INCLUDE_DIR ${CMAKE_SOURCE_DIR}/libs)

libs/ccc/coef/__init__.py

@@ -1,7 +1,8 @@
+import numpy as np
+
 from ccc.coef.impl import *  # noqa: F403, F401
 
 # Run CCC to initialize/compile its functions with numba
 from ccc.coef.impl import ccc
-import numpy as np
 
 ccc(np.random.rand(10), np.random.rand(10))

libs/ccc/coef/impl.py

@@ -7,17 +7,16 @@
 import os
 from collections.abc import Iterable
 from concurrent.futures import ProcessPoolExecutor, ThreadPoolExecutor, as_completed
-from typing import Union
 
+import numpy as np
 from numba import njit
 from numba.typed import List
+from numpy.typing import NDArray
 
-import numpy as np
 from ccc.pytorch.core import unravel_index_2d
 from ccc.scipy.stats import rank
 from ccc.sklearn.metrics import adjusted_rand_index as ari
 from ccc.utils import DummyExecutor, chunker
-from numpy.typing import NDArray
 
 
 @njit(cache=True, nogil=True)
@@ -319,7 +318,7 @@ def get_coords_from_index(n_obj: int, idx: int) -> tuple[int]:
 
 
 def get_chunks(
-    iterable: Union[int, Iterable], n_threads: int, ratio: float = 1
+    iterable: int | Iterable, n_threads: int, ratio: float = 1
 ) -> Iterable[Iterable[int]]:
     """
     It splits elements in an iterable in chunks according to the number of
@@ -587,7 +586,7 @@ def get_n_workers(n_jobs: int | None) -> int:
 def ccc(
     x: NDArray,
     y: NDArray = None,
-    internal_n_clusters: Union[int, Iterable[int]] = None,
+    internal_n_clusters: int | Iterable[int] = None,
     return_parts: bool = False,
     n_chunks_threads_ratio: int = 1,
     n_jobs: int = 1,

libs/ccc/coef/impl_gpu.py

@@ -7,16 +7,15 @@
 import os
 from collections.abc import Iterable
 from concurrent.futures import ProcessPoolExecutor, ThreadPoolExecutor
-from typing import Union
 
 import ccc_cuda_ext
+import numpy as np
 from numba import njit
 from numba.typed import List
+from numpy.typing import NDArray
 
-import numpy as np
 from ccc.scipy.stats import rank
 from ccc.utils import chunker
-from numpy.typing import NDArray
 
 
 @njit(cache=True, nogil=True)
@@ -244,7 +243,7 @@ def get_feature_parts(params):
 
 
 def get_chunks(
-    iterable: Union[int, Iterable], n_threads: int, ratio: float = 1
+    iterable: int | Iterable, n_threads: int, ratio: float = 1
 ) -> Iterable[Iterable[int]]:
     """
     It splits elements in an iterable in chunks according to the number of
@@ -341,7 +340,7 @@ def get_n_workers(n_jobs: int | None) -> int:
 def ccc(
     x: NDArray,
     y: NDArray = None,
-    internal_n_clusters: Union[int, Iterable[int]] = None,
+    internal_n_clusters: int | Iterable[int] = None,
     return_parts: bool = False,
     n_chunks_threads_ratio: int = 1,
     n_jobs: int = 1,
@@ -617,4 +616,4 @@ def ccc(
         return cm_values, max_parts, parts
     if pvalue_n_perms is not None and pvalue_n_perms > 0:
         return cm_values, cm_pvalues
-    return cm_values
+    return cm_values

libs/ccc/corr.py

@@ -16,9 +16,8 @@
 
 from __future__ import annotations
 
-import pandas as pd
-
 import numpy as np
+import pandas as pd
 from sklearn.metrics import pairwise_distances
 
 
@@ -60,9 +59,9 @@ def mic(data: pd.DataFrame, estimator="mic_approx", n_jobs=None) -> pd.DataFrame
     Compute the Maximal Correlation Coefficient (MIC).
     """
     from minepy import pstats
+    from scipy.spatial.distance import squareform
 
     from ccc.methods import mic as mic_single
-    from scipy.spatial.distance import squareform
 
     if n_jobs is None:
         corr_mat = pstats(
@@ -87,9 +86,10 @@ def ccc(data: pd.DataFrame, internal_n_clusters=None, n_jobs=1) -> pd.DataFrame:
     """
     Compute the Clustermatch Correlation Coefficient (CCC).
     """
-    from ccc.coef import ccc
     from scipy.spatial.distance import squareform
 
+    from ccc.coef import ccc
+
     corr_mat = ccc(
         data.to_numpy(),
         internal_n_clusters=internal_n_clusters,
@@ -110,9 +110,10 @@ def ccc_gpu(data: pd.DataFrame, internal_n_clusters=10, n_jobs=24) -> pd.DataFra
     """
     Compute the Clustermatch Correlation Coefficient (CCC).
     """
-    from ccc.coef.impl_gpu import ccc as ccc_gpu
     from scipy.spatial.distance import squareform
 
+    from ccc.coef.impl_gpu import ccc as ccc_gpu
+
     corr_mat = ccc_gpu(
         data.to_numpy(),
         internal_n_clusters=internal_n_clusters,

libs/ccc/giant.py

@@ -2,12 +2,13 @@
 Contains functions to interact with the REST API of HumanBase (GIANT networks):
 https://hb.flatironinstitute.org/
 """
-from pathlib import Path
-import tempfile
+
 import json
+import tempfile
+from pathlib import Path
 
-import requests
 import pandas as pd
+import requests
 
 URL_GENE_INFO = "https://hb.flatironinstitute.org/api/genes/"
 URL_TISSUE_PREDICTION = "https://hb.flatironinstitute.org/api/integrations/relevant/"