build: migrate extension build to scikit-build-core

.github/actions/build-gk/action.yaml

@@ -24,7 +24,7 @@ runs:
       id: restore_gk_pkg
       uses: actions/cache/restore@v4
       with:
-        key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'src/**', 'genome_kit/**', 'setup.py', 'setup/**', 'tests/**') }}
+        key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**', 'tests/**') }}
         path: |
           ~/conda-bld
 
@@ -101,6 +101,6 @@ runs:
       id: save_gk_pkg
       uses: actions/cache/save@v4
       with:
-        key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'src/**', 'genome_kit/**', 'setup.py', 'setup/**', 'tests/**') }}
+        key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**', 'tests/**') }}
         path: |
           ~/conda-bld

.github/actions/curl-meta-yaml/action.yaml

@@ -21,7 +21,7 @@ runs:
           curl -s -L -o conda-recipe/meta.yaml https://raw.githubusercontent.com/conda-forge/genomekit-feedstock/main/recipe/meta.yaml
         fi
 
-        export GK_VERSION=$(grep "version = " setup.py | awk -F'"' '{print $2}')
+        export GK_VERSION=$(grep -E "^version = " pyproject.toml | head -1 | awk -F'"' '{print $2}')
 
         export OS_TYPE=$(uname)
         if [[ "$OS_TYPE" == "Darwin" ]]; then
@@ -38,5 +38,59 @@ runs:
         # set the version to the local release-please version (for docs and docker publish)
         $SED_CMD "s/{% set version = \"[^\"]*\" %}/{% set version = \"${GK_VERSION}\" %}/" conda-recipe/meta.yaml
 
+        python3 - <<'PY'
+        from pathlib import Path
+
+        meta = Path("conda-recipe/meta.yaml")
+        text = meta.read_text()
+        pip_install = "{{ PYTHON }} -m pip install --no-deps --ignore-installed ."
+        pip_install_no_isolation = "{{ PYTHON }} -m pip install --no-deps --ignore-installed --no-build-isolation ."
+        if pip_install in text:
+            text = text.replace(pip_install, pip_install_no_isolation)
+        elif pip_install_no_isolation not in text:
+            raise SystemExit("Unable to find GenomeKit pip install command in conda-recipe/meta.yaml")
+
+        def add_requirement(text, section, requirement):
+            lines = text.splitlines()
+            section_start = None
+            section_end = None
+
+            in_requirements = False
+            for i, line in enumerate(lines):
+                if line == "requirements:":
+                    in_requirements = True
+                    continue
+                if not in_requirements:
+                    continue
+                if line and not line.startswith(" "):
+                    break
+                if line == f"  {section}:":
+                    section_start = i
+                    for j in range(i + 1, len(lines)):
+                        next_line = lines[j]
+                        if next_line.startswith("  ") and not next_line.startswith("    "):
+                            section_end = j
+                            break
+                        if next_line and not next_line.startswith(" "):
+                            section_end = j
+                            break
+                    else:
+                        section_end = len(lines)
+                    break
+
+            if section_start is None or section_end is None:
+                raise SystemExit(f"Unable to find requirements.{section} in conda-recipe/meta.yaml")
+            if any(line.strip() == f"- {requirement}" for line in lines[section_start + 1:section_end]):
+                return text
+
+            lines.insert(section_end, f"    - {requirement}")
+            return "\n".join(lines) + ("\n" if text.endswith("\n") else "")
+
+        text = add_requirement(text, "build", "cmake")
+        text = add_requirement(text, "build", "ninja")
+        text = add_requirement(text, "host", "scikit-build-core")
+        meta.write_text(text)
+        PY
+
         head -10 conda-recipe/meta.yaml
         set +x

.github/workflows/build-wheels.yaml

@@ -24,7 +24,8 @@ jobs:
         id: restore_linux_wheels
         uses: actions/cache/restore@v4
         with:
-          key: linux-wheels-${{ matrix.arch }}-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+          # setup.py/setup/ were removed; wheel build config now lives in pyproject.toml and CMakeLists.txt.
+          key: linux-wheels-${{ matrix.arch }}-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
           path: wheelhouse/*.whl
 
       # skip remaining steps on cache hit
@@ -77,7 +78,7 @@ jobs:
         name: cache linux wheels
         uses: actions/cache/save@v4
         with:
-          # hashFiles() would produce a different key due to setup/*.pyc files created
+          # Reuse the restore primary key so saved wheels match the restore key.
           key: ${{ steps.restore_linux_wheels.outputs.cache-primary-key }}
           path: wheelhouse/*.whl
 
@@ -93,7 +94,7 @@ jobs:
         id: restore_macos_wheels
         uses: actions/cache/restore@v4
         with:
-          key: macos-wheels-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+          key: macos-wheels-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
           path: wheelhouse/*.whl
 
       # skip remaining steps on cache hit
@@ -120,7 +121,7 @@ jobs:
         name: cache macos wheels
         uses: actions/cache/save@v4
         with:
-          # hashFiles() would produce a different key due to setup/*.pyc files created
+          # Reuse the restore primary key so saved wheels match the restore key.
           key: ${{ steps.restore_macos_wheels.outputs.cache-primary-key }}
           path: wheelhouse/*.whl
 
@@ -151,7 +152,7 @@ jobs:
       - name: Download built wheels
         uses: actions/cache/restore@v4
         with:
-          key: macos-wheels-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+          key: macos-wheels-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
           path: wheelhouse/*.whl
           fail-on-cache-miss: true
 
@@ -193,23 +194,23 @@ jobs:
 #        id: restore_linux_intel_wheels
 #        uses: actions/cache/restore@v4
 #        with:
-#          key: linux-wheels-x86_64-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+#          key: linux-wheels-x86_64-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
 #          path: wheelhouse/*.whl
 #          fail-on-cache-miss: true
 #
 #      - name: Restore the linux arm wheels cache
 #        id: restore_linux_arm_wheels
 #        uses: actions/cache/restore@v4
 #        with:
-#          key: linux-wheels-aarch64-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+#          key: linux-wheels-aarch64-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
 #          path: wheelhouse/*.whl
 #          fail-on-cache-miss: true
 #
 #      - name: Restore the macos wheels cache
 #        id: restore_macos_wheels
 #        uses: actions/cache/restore@v4
 #        with:
-#          key: macos-wheels-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+#          key: macos-wheels-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
 #          path: wheelhouse/*.whl
 #          fail-on-cache-miss: true
 #
@@ -246,23 +247,23 @@ jobs:
         id: restore_linux_intel_wheels
         uses: actions/cache/restore@v4
         with:
-          key: linux-wheels-x86_64-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+          key: linux-wheels-x86_64-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
           path: wheelhouse/*.whl
           fail-on-cache-miss: true
 
       - name: Restore the linux arm wheels cache
         id: restore_linux_arm_wheels
         uses: actions/cache/restore@v4
         with:
-          key: linux-wheels-aarch64-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+          key: linux-wheels-aarch64-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
           path: wheelhouse/*.whl
           fail-on-cache-miss: true
 
       - name: Restore the macos wheels cache
         id: restore_macos_wheels
         uses: actions/cache/restore@v4
         with:
-          key: macos-wheels-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'src/**', 'genome_kit/**', 'pyproject.toml', 'setup.py', 'setup/**') }}
+          key: macos-wheels-${{ hashFiles('.github/workflows/build-wheels.yaml', '.github/actions/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**') }}
           path: wheelhouse/*.whl
           fail-on-cache-miss: true
 

.github/workflows/dockerize.yaml

@@ -30,7 +30,7 @@ jobs:
         id: restore_gk_pkg
         uses: actions/cache/restore@v4
         with:
-          key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'src/**', 'genome_kit/**', 'setup.py', 'setup/**', 'tests/**') }}
+          key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**', 'tests/**') }}
           path: |
             ~/conda-bld
           fail-on-cache-miss: true

.github/workflows/publish-docs.yaml

@@ -35,7 +35,7 @@ jobs:
         id: restore_gk_pkg
         uses: actions/cache/restore@v4
         with:
-          key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'src/**', 'genome_kit/**', 'setup.py', 'setup/**', 'tests/**') }}
+          key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**', 'tests/**') }}
           path: |
             ~/conda-bld
           fail-on-cache-miss: true

.github/workflows/run-tests.yaml

@@ -7,8 +7,8 @@ on:
     - ".github/workflows/**"
     - "conda-recipe/**"
     - "genome_kit/**"
-    - "setup.py"
-    - "setup/**"
+    - "CMakeLists.txt"
+    - "pyproject.toml"
     - "src/**"
     - "tests/**"
 
@@ -84,7 +84,7 @@ jobs:
         id: restore_gk_pkg
         uses: actions/cache/restore@v4
         with:
-          key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'src/**', 'genome_kit/**', 'setup.py', 'setup/**', 'tests/**') }}
+          key: gk-tarballs-${{ runner.os }}-${{ runner.arch }}-${{ hashFiles('.github/workflows/run-tests.yaml', '.github/actions/**', 'conda-recipe/**', 'CMakeLists.txt', 'pyproject.toml', 'src/**', 'genome_kit/**', 'tests/**') }}
           path: |
             ~/conda-bld
           fail-on-cache-miss: true
@@ -131,4 +131,4 @@ jobs:
           fi
           conda mambabuild --croot /tmp/conda-bld -t "${files[@]}" --extra-deps "${extra_deps[@]}"
           conda clean -it
-          set +x
+          set +x

CMakeLists.txt

@@ -1,23 +1,120 @@
 cmake_minimum_required(VERSION 3.21)
-project(GenomeKit)
+project(GenomeKit LANGUAGES CXX)
 
 set(CMAKE_CXX_STANDARD 20)
+set(CMAKE_CXX_STANDARD_REQUIRED ON)
+set(CMAKE_CXX_EXTENSIONS OFF)
+
+file(GLOB SOURCES CONFIGURE_DEPENDS src/*.cpp src/*.h)
+file(GLOB GK_PY_EXTENSION_SOURCES CONFIGURE_DEPENDS src/*.cpp)
+file(GLOB GK_HEADERS CONFIGURE_DEPENDS src/*.h)
+
+find_package(Python COMPONENTS Interpreter Development.Module NumPy REQUIRED)
+find_package(ZLIB REQUIRED)
+
+if(NOT MSVC)
+    find_program(CCACHE_PROGRAM ccache)
+    if(CCACHE_PROGRAM)
+        set(CMAKE_CXX_COMPILER_LAUNCHER "${CCACHE_PROGRAM}")
+    endif()
+endif()
+
+Python_add_library(_cxx MODULE WITH_SOABI ${GK_PY_EXTENSION_SOURCES} ${GK_HEADERS})
+target_include_directories(_cxx PRIVATE src "${CMAKE_CURRENT_BINARY_DIR}/gen")
+target_link_libraries(_cxx PRIVATE Python::Module Python::NumPy ZLIB::ZLIB)
+target_compile_definitions(_cxx PRIVATE GKPY_LIBNAME=genome_kit)
+set_target_properties(_cxx PROPERTIES
+    CXX_VISIBILITY_PRESET hidden
+    VISIBILITY_INLINES_HIDDEN ON
+)
+
+if(MSVC)
+    target_compile_definitions(_cxx PRIVATE _CRT_SECURE_NO_WARNINGS)
+    target_compile_options(_cxx PRIVATE
+        /permissive-
+        /Zc:__cplusplus
+        /Zc:strictStrings-
+        /Zc:preprocessor
+        /W3
+        /EHsc
+        /wd5033
+        $<$<CONFIG:Debug>:/GS>
+        $<$<CONFIG:Debug>:/Zi>
+        $<$<CONFIG:Debug>:/Od>
+        $<$<NOT:$<CONFIG:Debug>>:/GL>
+        $<$<NOT:$<CONFIG:Debug>>:/Gy>
+        $<$<NOT:$<CONFIG:Debug>>:/Oy>
+        $<$<NOT:$<CONFIG:Debug>>:/Oi>
+    )
+    target_link_options(_cxx PRIVATE
+        $<$<CONFIG:Debug>:/DEBUG>
+        $<$<NOT:$<CONFIG:Debug>>:/LTCG>
+        "/PDB:$<TARGET_FILE_DIR:_cxx>/_cxx.pdb"
+    )
+else()
+    target_compile_definitions(_cxx PRIVATE _FILE_OFFSET_BITS=64)
+    target_compile_options(_cxx PRIVATE
+        -Wall
+        -Wno-write-strings
+        -Wno-invalid-offsetof
+    )
+    target_link_options(_cxx PRIVATE
+        $<$<NOT:$<CONFIG:Debug>>:-Wl,-S>
+        $<$<NOT:$<CONFIG:Debug>>:-Wl,-x>
+    )
+    if(APPLE)
+        if(DEFINED ENV{MACOSX_DEPLOYMENT_TARGET})
+            set(GENOMEKIT_MACOSX_DEPLOYMENT_TARGET "$ENV{MACOSX_DEPLOYMENT_TARGET}")
+        else()
+            set(GENOMEKIT_MACOSX_DEPLOYMENT_TARGET "10.15")
+        endif()
+        if(DEFINED ENV{CONDA_BUILD_SYSROOT})
+            set(GENOMEKIT_MACOS_SYSROOT "$ENV{CONDA_BUILD_SYSROOT}")
+        elseif(CMAKE_OSX_SYSROOT)
+            set(GENOMEKIT_MACOS_SYSROOT "${CMAKE_OSX_SYSROOT}")
+        else()
+            execute_process(
+                COMMAND xcrun --sdk macosx --show-sdk-path
+                OUTPUT_VARIABLE GENOMEKIT_MACOS_SYSROOT
+                OUTPUT_STRIP_TRAILING_WHITESPACE
+                ERROR_QUIET
+            )
+        endif()
+        target_compile_definitions(_cxx PRIVATE _LIBCPP_DISABLE_AVAILABILITY)
+        target_compile_options(_cxx PRIVATE
+            "-mmacosx-version-min=${GENOMEKIT_MACOSX_DEPLOYMENT_TARGET}"
+            -stdlib=libc++
+            -Wshorten-64-to-32
+            -Wsign-compare
+            -Wconditional-uninitialized
+            -Wuninitialized
+            -Wno-unknown-warning-option
+        )
+        target_link_options(_cxx PRIVATE
+            "-mmacosx-version-min=${GENOMEKIT_MACOSX_DEPLOYMENT_TARGET}"
+        )
+        if(GENOMEKIT_MACOS_SYSROOT)
+            target_compile_options(_cxx PRIVATE "-isysroot${GENOMEKIT_MACOS_SYSROOT}")
+            target_link_options(_cxx PRIVATE "-isysroot${GENOMEKIT_MACOS_SYSROOT}")
+        endif()
+    endif()
+endif()
+
+install(TARGETS _cxx DESTINATION genome_kit)
 
-FILE(GLOB SOURCES src/*.cpp src/*.h)
 if(DEFINED ENV{IN_CLION})
     execute_process(
             COMMAND $ENV{CONDA_PREFIX}/bin/python -c "import sys; print('{}.{}'.format(sys.version_info.major, sys.version_info.minor), end='')"
             OUTPUT_VARIABLE PYTHON_VERSION
     )
 
-    # Don't really need this lib (handled by setup.py), but required for CLion's search etc.
+    # Don't really need this lib for packaging, but required for CLion's search etc.
     # NOTE: Set env vars IN_CLION=1;CONDA_PREFIX=$HOME/conda/envs/genomekit_dev in CLion CMake preferences.
     # include numpy, Python.h
     include_directories(
             $ENV{CONDA_PREFIX}/lib/python${PYTHON_VERSION}/site-packages/numpy/core/include/
             $ENV{CONDA_PREFIX}/include/python${PYTHON_VERSION}/
     )
-    find_package(ZLIB)
     add_library(gkdev ${SOURCES})
     target_link_libraries(gkdev ZLIB::ZLIB)
     target_compile_definitions(gkdev PRIVATE -D_FILE_OFFSET_BITS=64 -DGKPY_LIBNAME=$ENV{CONDA_PREFIX}/lib -D_WANT_MAIN=1)
@@ -27,12 +124,11 @@ endif()
 set(NON_PY_SOURCES ${SOURCES})
 list(FILTER NON_PY_SOURCES EXCLUDE REGEX "py_.*")
 
-add_executable(main ${NON_PY_SOURCES})
+add_executable(main EXCLUDE_FROM_ALL ${NON_PY_SOURCES})
 target_compile_definitions(main PRIVATE -D_FILE_OFFSET_BITS=64 -D_FILE_ABS_PATH='\"${CMAKE_CURRENT_SOURCE_DIR}/src/main.cpp\"' -D_WANT_MAIN=1)
 set_target_properties(main PROPERTIES LINKER_LANGUAGE CXX)
-find_package(ZLIB)
 target_link_libraries(main ZLIB::ZLIB)
 
 include(CTest)
-add_test(main unittestbuild/main)
+add_test(NAME main COMMAND main)
 set_tests_properties(main PROPERTIES WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR})

docs-src/develop.rst

@@ -31,10 +31,9 @@ Build the package in development mode::
 
     pip install -e .
 
-This builds the C++ extension and copies it into
-your source tree (``genome_kit/_cxx.so``).
-It also ensures that ``import genome_kit`` works from any directory
-by linking your source tree from python's ``site-packages``.
+This builds the C++ extension with CMake through scikit-build-core and installs
+an editable wheel. Python files are loaded from your source tree, while the
+compiled extension is loaded from the editable build output.
 
 .. note:: Windows Prerequisites
 
@@ -63,7 +62,8 @@ In the CMake settings, set the following environment variables::
 Making changes
 --------------
 
-If the C/C++ code changed, you must re-run the ``develop`` command::
+If the C/C++ code or build configuration changed, you must re-run the editable
+install command::
 
     pip install -e .