bblean 0.6.0b1__tar.gz → 0.6.0b2__tar.gz

{bblean-0.6.0b1 → bblean-0.6.0b2}/.github/workflows/ci-cpp.yaml

@@ -36,6 +36,29 @@ jobs:
       - name: run-pytest
         run: BITBIRCH_CANT_SKIP_CPP_TESTS=1 pytest -s ./tests/test_similarity.py
 
+  cpp-tests-win:
+    runs-on: windows-latest
+    steps:
+      - name: checkout-repo
+        uses: actions/checkout@v5
+
+      - name: setup-python
+        uses: actions/setup-python@v6
+        with:
+          python-version: ${{ env.PYTHON_VERSION }}
+
+      - name: install-repo
+        run: |
+          python -m pip install --upgrade pip
+          pip install pytest pytest-subtests inline-snapshot
+          $env:BITBIRCH_BUILD_CPP="1"
+          pip install --verbose .
+
+      - name: run-pytest
+        run: |
+          $env:BITBIRCH_CANT_SKIP_CPP_TESTS="1"
+          pytest -s ./tests/test_similarity.py
+
 
   # Ubuntu 24.04 for reproducibility in regression tests only
   # coverage makes tests slightly slower, so regression tests should not run with

{bblean-0.6.0b1 → bblean-0.6.0b2}/.github/workflows/upload-to-pypi.yaml

@@ -23,7 +23,7 @@ env:
   # Skip py 3.14, 32 bit and musllinux (Alpine) wheels
   CIBW_SKIP: "cp314-* cp314t-* *-manylinux_i686 *-win32 *-musllinux_*"
   CIBW_TEST_REQUIRES: pytest pytest-subtests inline-snapshot
-  CIBW_TEST_COMMAND: pytest -k 'not regression' {project}/tests
+  CIBW_TEST_COMMAND: pytest -k "not regression" {project}/tests
   CIBW_PRERELEASE_PYTHONS: False
   CIBW_ENVIRONMENT: "SETUPTOOLS_SCM_PRETEND_VERSION=${{ github.event.inputs.version }} BITBIRCH_BUILD_CPP=1"
   # Build wheels that support both aarch64 and x86_64 on macOS
@@ -63,7 +63,7 @@ jobs:
   build_wheels:
     strategy:
       matrix:
-        os: [ubuntu-latest, macos-latest]
+        os: [ubuntu-latest, macos-latest, windows-latest]
     name: build-${{ matrix.os }}-wheels
     runs-on: ${{ matrix.os }}
     steps:

{bblean-0.6.0b1 → bblean-0.6.0b2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bblean
-Version: 0.6.0b1
+Version: 0.6.0b2
 Summary: BitBirch-Lean Python package
 Author: The Miranda-Quintana Lab and other BitBirch developers
 Author-email: Ramon Alain Miranda Quintana <quintana@chem.ufl.edu>, Krisztina Zsigmond <kzsigmond@ufl.edu>, Ignacio Pickering <ipickering@ufl.edu>, Kenneth Lopez Perez <klopezperez@chem.ufl.edu>, Miroslav Lzicar <miroslav.lzicar@deepmedchem.com>
@@ -85,14 +85,19 @@ tuning](https://mqcomplab.github.io/bblean/devdocs/user-guide/parameters.html) g
 
 ## Installation
 
-BitBIRCH-Lean requires Python 3.11 or higher, and can be installed in Linux or macOS.
-Via pip, which automatically includes C++ extensions:
+BitBIRCH-Lean requires Python 3.11 or higher, and can be installed in Windows, Linux or
+macOS via pip, which automatically includes C++ extensions:
 
 ```bash
 pip install bblean
+bb --help
 ```
+
 We recommend installing `bblean` in a conda environment or a `venv`.
 
+Memory usage and C++ extensions are most optimized for Linux / macOS. We support windows
+on a best-effort basis, some releases may not have Windows support.
+
 ### From source
 
 To build from source instead (editable mode):

{bblean-0.6.0b1 → bblean-0.6.0b2}/README.md

@@ -42,14 +42,19 @@ tuning](https://mqcomplab.github.io/bblean/devdocs/user-guide/parameters.html) g
 
 ## Installation
 
-BitBIRCH-Lean requires Python 3.11 or higher, and can be installed in Linux or macOS.
-Via pip, which automatically includes C++ extensions:
+BitBIRCH-Lean requires Python 3.11 or higher, and can be installed in Windows, Linux or
+macOS via pip, which automatically includes C++ extensions:
 
 ```bash
 pip install bblean
+bb --help
 ```
+
 We recommend installing `bblean` in a conda environment or a `venv`.
 
+Memory usage and C++ extensions are most optimized for Linux / macOS. We support windows
+on a best-effort basis, some releases may not have Windows support.
+
 ### From source
 
 To build from source instead (editable mode):

{bblean-0.6.0b1 → bblean-0.6.0b2}/bblean/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.6.0b1'
-__version_tuple__ = version_tuple = (0, 6, 0, 'b1')
+__version__ = version = '0.6.0b2'
+__version_tuple__ = version_tuple = (0, 6, 0, 'b2')
 
 __commit_id__ = commit_id = None

{bblean-0.6.0b1 → bblean-0.6.0b2}/bblean/cli.py

@@ -957,7 +957,6 @@ def _run(
         bool,
         Option(
             "--monitor-mem/--no-monitor-mem",
-            "--monitor-rss/--no-monitor-rss",
             help="Monitor RAM used by all processes",
             rich_help_panel="Advanced",
         ),
@@ -966,7 +965,6 @@ def _run(
         float,
         Option(
             "--monitor-mem-seconds",
-            "--monitor-rss-seconds",
             help="Interval in seconds for RAM monitoring",
             rich_help_panel="Debug",
             hidden=True,
@@ -1284,8 +1282,7 @@ def _multiround(
     monitor_rss: Annotated[
         bool,
         Option(
-            "--monitor-mem",
-            "--monitor-rss",
+            "--monitor-mem/--no-monitor-mem",
             help="Monitor RAM used by all processes",
             rich_help_panel="Advanced",
         ),
@@ -1294,7 +1291,6 @@ def _multiround(
         float,
         Option(
             "--monitor-mem-seconds",
-            "--monitor-rss-seconds",
             help="Interval in seconds for RAM monitoring",
             rich_help_panel="Debug",
             hidden=True,

{bblean-0.6.0b1 → bblean-0.6.0b2}/bblean/csrc/similarity.cpp

@@ -69,7 +69,7 @@ uint32_t _popcount_1d(const py::array_t<uint8_t>& arr) {
 #endif
     uint32_t count{0};  // Output scalar
     py::ssize_t steps = arr.shape(0);
-    if (is_8byte_aligned(arr) and (steps % 64 == 0)) {
+    if (is_8byte_aligned(arr) && (steps % 64 == 0)) {
 #ifdef DEBUG_LOGS
         py::print("DEBUG: _popcount_1d fn triggered uint64 + popcount 64");
 #endif
@@ -110,7 +110,7 @@ py::array_t<uint32_t> _popcount_2d(const CArrayForcecast<uint8_t>& arr) {
     print_8byte_alignment_check(arr);
 #endif
     py::ssize_t steps = arr.shape(1);
-    if (is_8byte_aligned(arr) and (steps % 64 == 0)) {
+    if (is_8byte_aligned(arr) && (steps % 64 == 0)) {
 #ifdef DEBUG_LOGS
         py::print("DEBUG: _popcount_2d fn triggered uint64 + popcount 64");
 #endif
@@ -243,7 +243,7 @@ py::array_t<uint8_t> centroid_from_sum(const CArrayForcecast<T>& linear_sum,
         }
     }
 
-    if (not pack) {
+    if (!pack) {
         return centroid_unpacked;
     }
 
@@ -351,7 +351,7 @@ py::array_t<double> jt_sim_packed_precalc_cardinalities(
     }
     auto out = py::array_t<double>(n_samples);
 
-    if (is_8byte_aligned(arr) and is_8byte_aligned(vec) and
+    if (is_8byte_aligned(arr) && is_8byte_aligned(vec) &&
         (n_features % 64 == 0)) {
 #ifdef DEBUG_LOGS
         py::print("DEBUG: jt_sim_packed fn triggered uint64 + popcount 64");

{bblean-0.6.0b1 → bblean-0.6.0b2}/bblean.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: bblean
-Version: 0.6.0b1
+Version: 0.6.0b2
 Summary: BitBirch-Lean Python package
 Author: The Miranda-Quintana Lab and other BitBirch developers
 Author-email: Ramon Alain Miranda Quintana <quintana@chem.ufl.edu>, Krisztina Zsigmond <kzsigmond@ufl.edu>, Ignacio Pickering <ipickering@ufl.edu>, Kenneth Lopez Perez <klopezperez@chem.ufl.edu>, Miroslav Lzicar <miroslav.lzicar@deepmedchem.com>
@@ -85,14 +85,19 @@ tuning](https://mqcomplab.github.io/bblean/devdocs/user-guide/parameters.html) g
 
 ## Installation
 
-BitBIRCH-Lean requires Python 3.11 or higher, and can be installed in Linux or macOS.
-Via pip, which automatically includes C++ extensions:
+BitBIRCH-Lean requires Python 3.11 or higher, and can be installed in Windows, Linux or
+macOS via pip, which automatically includes C++ extensions:
 
 ```bash
 pip install bblean
+bb --help
 ```
+
 We recommend installing `bblean` in a conda environment or a `venv`.
 
+Memory usage and C++ extensions are most optimized for Linux / macOS. We support windows
+on a best-effort basis, some releases may not have Windows support.
+
 ### From source
 
 To build from source instead (editable mode):

{bblean-0.6.0b1 → bblean-0.6.0b2}/setup.py

@@ -23,13 +23,15 @@ APPLE_SILICON = platform.machine().lower() in ["arm64", "aarch64"]
 ARM = platform.machine().lower().startswith("arm") and not APPLE_SILICON
 
 # Build C++ extensions (recommended)
+extra_compile_args = []
 if os.getenv("BITBIRCH_BUILD_CPP"):
     import pybind11
     from pybind11.setup_helpers import Pybind11Extension, WIN
 
     # setuptools paths must be relative
     ext_sources = [str((Path(name) / "csrc" / "similarity.cpp"))]
-    extra_compile_args = ["-O3"]  # -O3 includes -ftree-vectorize
+    if not WIN:
+        extra_compile_args.append("-O3")  # -O3 includes -ftree-vectorize
     if not WIN:
         if X86:
             if os.getenv("BITBIRCH_BUILD_NATIVE"):

{bblean-0.6.0b1 → bblean-0.6.0b2}/tests/test_cli.py

@@ -254,6 +254,7 @@ def test_multiround() -> None:
                 "--no-verbose",
                 "--set-mid-merge",
                 "tolerance-legacy",
+                "--no-monitor-mem",
             ],
         )
         with open(out_dir / "clusters.pkl", mode="rb") as f:
@@ -288,7 +289,18 @@ def test_run() -> None:
         np.save(dir / "fingerprints.npy", fps)
         out_dir = dir / "output"
         result = runner.invoke(
-            app, ["run", str(dir), "-o", str(out_dir), "-b", "50", "-t", "0.65"]
+            app,
+            [
+                "run",
+                str(dir),
+                "-o",
+                str(out_dir),
+                "-b",
+                "50",
+                "-t",
+                "0.65",
+                "--no-monitor-mem",
+            ],
         )
         with open(out_dir / "clusters.pkl", mode="rb") as f:
             obj = pickle.load(f)

{bblean-0.6.0b1 → bblean-0.6.0b2}/tests/test_global_clustering.py

@@ -1,11 +1,19 @@
 import sys
+
+import pytest
 import numpy as np
+
 from bblean.bitbirch import BitBirch
 from bblean.fingerprints import make_fake_fingerprints, unpack_fingerprints
 from inline_snapshot import snapshot
 
 
 def test_random_fps_consistency() -> None:
+    # TODO For some strage reason this test *fails on macOS and Windows*
+    # The kmeans implementation of sklearn seems to work different in linux and macOS
+    if sys.platform != "linux":
+        pytest.skip("Currently global clustering is non-deterministic on mac / windows")
+
     fps = make_fake_fingerprints(3000, n_features=2048, seed=126205095409235, pack=True)
     tree = BitBirch(branching_factor=50, threshold=0.65, merge_criterion="diameter")
     tree.fit(fps, n_features=2048)
@@ -29,33 +37,31 @@ def test_random_fps_consistency() -> None:
             [235, 255, 123, 255, 255],
         ]
     )
-    # TODO For some strage reason this test *fails on macOS*
-    # The kmeans implementation of sklearn seems to work different in linux and macOS
-    if sys.platform != "darwin":
-        tree.global_clustering(
-            20,
-            method="kmeans",
-            n_init=1,
-            init=unpack_fingerprints(np.vstack(output_cent))[::2][:20],
-            max_iter=10,
-        )
-        output_mol_ids = tree.get_cluster_mol_ids(global_clusters=True, sort=False)
-        output_med = tree.get_medoids(fps, global_clusters=True, sort=False)
-        assert [o[:5] for o in output_mol_ids[:5]] == snapshot(
-            [
-                [16, 1023, 1793, 2, 15],
-                [1873, 1882, 1912, 1954, 1970],
-                [12, 1877, 1861, 2068, 2012],
-                [1560, 1901, 2065, 2037, 2396],
-                [62, 73, 75, 87, 121],
-            ]
-        )
-        assert output_med[:5, :5].tolist() == snapshot(
-            [
-                [255, 127, 252, 111, 223],
-                [255, 255, 95, 255, 239],
-                [123, 239, 238, 135, 126],
-                [223, 14, 207, 187, 104],
-                [255, 255, 255, 247, 255],
-            ]
-        )
+
+    tree.global_clustering(
+        20,
+        method="kmeans",
+        n_init=1,
+        init=unpack_fingerprints(np.vstack(output_cent))[::2][:20],
+        max_iter=10,
+    )
+    output_mol_ids = tree.get_cluster_mol_ids(global_clusters=True, sort=False)
+    output_med = tree.get_medoids(fps, global_clusters=True, sort=False)
+    assert [o[:5] for o in output_mol_ids[:5]] == snapshot(
+        [
+            [16, 1023, 1793, 2, 15],
+            [1873, 1882, 1912, 1954, 1970],
+            [12, 1877, 1861, 2068, 2012],
+            [1560, 1901, 2065, 2037, 2396],
+            [62, 73, 75, 87, 121],
+        ]
+    )
+    assert output_med[:5, :5].tolist() == snapshot(
+        [
+            [255, 127, 252, 111, 223],
+            [255, 255, 95, 255, 239],
+            [123, 239, 238, 135, 126],
+            [223, 14, 207, 187, 104],
+            [255, 255, 255, 247, 255],
+        ]
+    )

{bblean-0.6.0b1 → bblean-0.6.0b2}/tests/test_regression.py

@@ -77,7 +77,7 @@ def test_speed_regression(subtests) -> None:
     # all_max_allowed_ns = [1_200_000_000, 1_900_000_000, 2_500_000_000]
     # For the ubuntu-24.04 in gh CI the following are required:
     if CSIM_AVAIL:
-        all_max_allowed_ns = [900_000_000, 1_400_000_000, 2_000_000_000]
+        all_max_allowed_ns = [900_000_000, 1_500_000_000, 2_000_000_000]
     else:
         all_max_allowed_ns = [1_700_000_000, 2_600_000_000, 3_600_000_000]
     for fps_num, max_allowed_ns in zip(all_fps_nums, all_max_allowed_ns):