Merge pull request #90 from paulsengroup/feature/modular-deps

Make pandas and scipy optional deps

.github/workflows/fuzzy-testing.yml

@@ -156,7 +156,7 @@ jobs:
           conan remove --confirm "*"
 
       - name: Build and install
-        run: pip install --verbose .
+        run: pip install --verbose '.[all]'
 
       - name: Clean Conan cache (post-build)
         if: steps.cache-conan.outputs.cache-hit != 'true'

.github/workflows/pip.yml

@@ -85,7 +85,7 @@ jobs:
           echo 'CXX=clang++' >> $GITHUB_ENV
 
       - name: Build and install
-        run: pip install --verbose '.[test]'
+        run: pip install --verbose '.[all,test]'
 
       - name: Save Conan cache
         uses: actions/cache/save@v4

pyproject.toml

@@ -5,10 +5,10 @@
 [build-system]
 requires = [
   "conan>=2.0.5",
-  "nanobind==2.2.0", # This is required in order to run stubgen
+  "nanobind>=2", # This is required in order to run stubgen
   "numpy",
-  "pandas>=2.1.0,!=2.2.0",
-  "pyarrow==17.0.0",
+  "pandas>=2.1,!=2.2.0",
+  "pyarrow>=14",
   "scikit-build-core>=0.10",
   "scipy",
   "typing_extensions",
@@ -33,13 +33,32 @@ classifiers = [
 
 dependencies = [
   "numpy",
-  "pandas>=2.1.0,!=2.2.0",
   "pyarrow>=14",
+]
+
+optional-dependencies.pandas = [
+  "pandas>=2.1.0,!=2.2.0",
+]
+
+optional-dependencies.scipy = [
   "scipy",
 ]
 
+optional-dependencies.all = [
+  "hictkpy[pandas,scipy]",
+]
+
 optional-dependencies.test = [
-  "pytest>=8.0"
+  "hictkpy[all]",
+  "pytest>=8.0",
+]
+
+optional-dependencies.dev = [
+  "hictkpy[all,test]",
+  "black>=24.10",
+  "clang-format>=19.1",
+  "isort>=5.13",
+  "gersemi>=0.15",
 ]
 
 [tool.scikit-build]

src/hictkpy.cpp

@@ -37,10 +37,6 @@ NB_MODULE(_hictkpy, m) {
     throw std::runtime_error("failed to initialize pyarrow runtime");
   }
 
-  [[maybe_unused]] auto np = nb::module_::import_("numpy");
-  [[maybe_unused]] auto pd = nb::module_::import_("pandas");
-  [[maybe_unused]] auto ss = nb::module_::import_("scipy.sparse");
-
   m.attr("__hictk_version__") = hictk::config::version::str();
 
   m.doc() = "Blazing fast toolkit to work with .hic and .cool files.";

src/include/hictkpy/bin_table.hpp

@@ -18,7 +18,7 @@ namespace hictkpy {
 
 template <typename File>
 inline nanobind::object get_bins_from_file(const File &f) {
-  auto pd = nanobind::module_::import_("pandas");
+  auto pd = import_module_checked("pandas");
 
   Dynamic1DA<std::int32_t> chrom_ids{};
   Dynamic1DA<std::int32_t> starts{};

src/include/hictkpy/nanobind.hpp

@@ -27,3 +27,16 @@ HICTKPY_DISABLE_WARNING_USELESS_CAST
 #include <nanobind/stl/vector.h>
 HICTKPY_DISABLE_WARNING_POP
 // clang-format on
+
+#include <string>
+
+inline nanobind::module_ import_module_checked(const std::string& module_name) {
+  try {
+    return nanobind::module_::import_(module_name.c_str());
+  } catch (nanobind::python_error& e) {
+    // NOLINTNEXTLINE(*-pro-type-vararg)
+    nanobind::raise_from(e, PyExc_ModuleNotFoundError,
+                         "To enable %s support, please install %s with: pip install 'hictkpy[%s]'",
+                         module_name.c_str(), module_name.c_str(), module_name.c_str());
+  }
+}

src/pixel_selector.cpp

@@ -206,6 +206,7 @@ nb::object PixelSelector::to_arrow(std::string_view span) const {
 }
 
 nb::object PixelSelector::to_pandas(std::string_view span) const {
+  import_module_checked("pandas");
   return to_arrow(span).attr("to_pandas")();
 }
 
@@ -222,6 +223,7 @@ template <typename N, typename PixelSelector>
 }
 
 nb::object PixelSelector::to_csr(std::string_view span) const {
+  import_module_checked("scipy");
   const auto query_span = parse_span(span);
 
   return std::visit(
@@ -237,6 +239,7 @@ nb::object PixelSelector::to_csr(std::string_view span) const {
 }
 
 nb::object PixelSelector::to_coo(std::string_view span) const {
+  import_module_checked("scipy");
   return to_csr(span).attr("tocoo")(false);
 }
 

test/test_fetch_dense.py

@@ -5,8 +5,8 @@
 import pathlib
 
 import numpy as np
+import numpy.typing as npt
 import pytest
-from scipy.linalg import issymmetric
 
 import hictkpy
 
@@ -21,6 +21,16 @@
 )
 
 
+def issymmetric(m: npt.NDArray) -> bool:
+    assert m.ndim == 2
+    if m.size == 0:
+        return True
+    if m.shape[0] != m.shape[1]:
+        return False
+
+    return np.allclose(m, m.T, atol=0.0, rtol=0.0)
+
+
 class TestClass:
     def test_genome_wide(self, file, resolution):
         f = hictkpy.File(file, resolution)

test/test_fetch_df.py

@@ -20,6 +20,16 @@
 )
 
 
+def pandas_avail() -> bool:
+    try:
+        import pandas
+    except ModuleNotFoundError:
+        return False
+
+    return True
+
+
+@pytest.mark.skipif(not pandas_avail(), reason="pandas is not available")
 class TestClass:
     def test_genome_wide(self, file, resolution):
         f = hictkpy.File(file, resolution)

test/test_fetch_sparse.py

@@ -20,6 +20,16 @@
 )
 
 
+def scipy_avail() -> bool:
+    try:
+        import scipy
+    except ModuleNotFoundError:
+        return False
+
+    return True
+
+
+@pytest.mark.skipif(not scipy_avail(), reason="scipy is not available")
 class TestClass:
     def test_genome_wide(self, file, resolution):
         f = hictkpy.File(file, resolution)

test/test_file_accessors.py

@@ -20,13 +20,25 @@
 )
 
 
+def pandas_avail() -> bool:
+    try:
+        import pandas
+    except ModuleNotFoundError:
+        return False
+
+    return True
+
+
 class TestClass:
+    @pytest.mark.skipif(not pandas_avail(), reason="pandas is not available")
     def test_attributes(self, file, resolution):
         f = hictkpy.File(file, resolution)
         assert f.resolution() == 100_000
+        # assert f.nchroms() == 8  # TODO enable after merging https://github.com/paulsengroup/hictk/pull/294
         assert f.nbins() == 1380
 
         assert "chr2L" in f.chromosomes()
+
         assert len(f.bins()) == 1380
         assert len(f.chromosomes()) == 8
 
@@ -47,4 +59,4 @@ def test_normalizations(self, file, resolution):
 
         name = "weight" if f.is_cooler() else "ICE"
         weights = f.weights(name)
-        assert len(weights) == len(f.bins())
+        assert len(weights) == f.nbins()

test/test_file_creation_cool.py

@@ -21,6 +21,16 @@
 )
 
 
+def pandas_avail() -> bool:
+    try:
+        import pandas
+    except ModuleNotFoundError:
+        return False
+
+    return True
+
+
+@pytest.mark.skipif(not pandas_avail(), reason="pandas is not available")
 class TestClass:
     @staticmethod
     def setup_method():

test/test_file_creation_hic.py

@@ -21,6 +21,16 @@
 )
 
 
+def pandas_avail() -> bool:
+    try:
+        import pandas
+    except ModuleNotFoundError:
+        return False
+
+    return True
+
+
+@pytest.mark.skipif(not pandas_avail(), reason="pandas is not available")
 class TestClass:
     @staticmethod
     def setup_method():