memory fix, global_skip fix, upload to pyi

kjgm · kjgm · commit 11dbb6f143c1 · 2025-08-06T11:07:51.000+02:00
diff --git a/.github/workflows/pypi_release.yml b/.github/workflows/pypi_release.yml
@@ -0,0 +1,84 @@
+name: Publish distributions to PyPI
+
+on:
+  workflow_dispatch:
+  push:
+    tags:
+     - '*'
+     
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+
+  build_sdist:
+    name: Build SDist
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v4
+
+    - name: Build SDist0
+      run: pipx run build --sdist
+
+    - name: Check metadata
+      run: pipx run twine check dist/*
+
+    - uses: actions/upload-artifact@v4
+      with:
+        name: dist-source
+        path: dist/*.tar.gz
+
+
+  build_wheels:
+    name: Wheels on ${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-latest, windows-latest, macos-latest ]
+
+    steps:
+    - uses: actions/checkout@v4
+
+    - uses: pypa/cibuildwheel@v2.16.5
+      env:
+        CIBW_SKIP: "pp*"
+        CIBW_ARCHS_MACOS: auto universal2
+        CIBW_PRERELEASE_PYTHONS: true
+
+    - name: Verify clean directory
+      run: git diff --exit-code
+      shell: bash
+
+    - name: Upload wheels
+      uses: actions/upload-artifact@v4
+      with:
+        name: dist-wheel-${{ matrix.os }}
+        path: wheelhouse/*.whl
+
+  upload_all:
+    name: Upload if release
+    needs: [build_wheels, build_sdist]
+    runs-on: ubuntu-latest
+    #if: github.event_name == 'release'
+
+    steps:
+    - uses: actions/setup-python@v5
+      name: Set up Python 3.x
+      with:
+        python-version: "3.8"
+
+    - uses: actions/download-artifact@v4
+      name: Download wheels
+      with:
+        pattern: dist-*
+        merge-multiple: true
+        path: dist
+
+    - uses: pypa/gh-action-pypi-publish@release/v1
+      name: Publish to PyPI
+      with:
+        user: __token__
+        password: ${{ secrets.PYPI_API_TOKEN }}
+    
diff --git a/MANIFEST.in b/MANIFEST.in
@@ -0,0 +1,2 @@
+# Include everything in the code folder
+recursive-include code *
diff --git a/README.md b/README.md
@@ -12,6 +12,13 @@ If you use ConTree, please cite our paper:
 
 ## Python usage
 
+### Install from PyPi
+The `pycontree` python package can be installed from PyPi using `pip`:
+
+```sh
+pip install pycontree
+```
+
 ### Install from source using pip
 The `pycontree` python package can be installed from source as follows:
 
diff --git a/code/Engine/src/specialized_solver.cpp b/code/Engine/src/specialized_solver.cpp
@@ -108,8 +108,8 @@ void SpecializedSolver::get_best_left_right_scores(const Dataview& dataview, int
         left_optimal_dt->make_leaf(left_tree.max_label, left_tree.size - left_tree.classification_score);
     } else {
         left_optimal_dt->update_split(left_tree.best_feature_index, left_tree.best_threshold, std::make_shared<Tree>(left_tree.best_left_label, -1), std::make_shared<Tree>(left_tree.best_right_label, -1));
-        RUNTIME_ASSERT(left_tree.best_left_label != -1, "Left tree left label should be initialized.");
-        RUNTIME_ASSERT(left_tree.best_right_label != -1, "Left tree right label should be initialized.");
+        //RUNTIME_ASSERT(left_tree.best_left_label != -1, "Left tree left label should be initialized.");
+        //RUNTIME_ASSERT(left_tree.best_right_label != -1, "Left tree right label should be initialized.");
     }
     left_optimal_dt->misclassification_score = left_tree.size - left_tree.classification_score;
     RUNTIME_ASSERT(left_optimal_dt->misclassification_score >= 0, "LR - Left tree misclassification score should be non-negative.");
@@ -118,8 +118,8 @@ void SpecializedSolver::get_best_left_right_scores(const Dataview& dataview, int
         right_optimal_dt->make_leaf(right_tree.max_label, right_tree.size - right_tree.classification_score);
     } else {
         right_optimal_dt->update_split(right_tree.best_feature_index, right_tree.best_threshold, std::make_shared<Tree>(right_tree.best_left_label, -1), std::make_shared<Tree>(right_tree.best_right_label, -1));
-        RUNTIME_ASSERT(right_tree.best_left_label != -1, "Right tree left label should be initialized.");
-        RUNTIME_ASSERT(right_tree.best_right_label != -1, "Right tree right label should be initialized.");
+        //RUNTIME_ASSERT(right_tree.best_left_label != -1, "Right tree left label should be initialized.");
+        //RUNTIME_ASSERT(right_tree.best_right_label != -1, "Right tree right label should be initialized.");
     }
     right_optimal_dt->misclassification_score = right_tree.size - right_tree.classification_score;
     RUNTIME_ASSERT(right_optimal_dt->misclassification_score >= 0, "LR - Right tree misclassification score should be non-negative.");
@@ -133,7 +133,6 @@ void SpecializedSolver::process_depth_one_feature(const Dataview& dataview,
     const std::vector<Dataset::FeatureElement>& current_feature = dataview.get_sorted_dataset_feature(current_feature_index);
     const int class_number = dataview.get_class_number();
     const int dataset_size = dataview.get_dataset_size();
-    int global_skip = 0;
 
     left_tree.reset_label_frequency();
     right_tree.reset_label_frequency();
@@ -156,10 +155,9 @@ void SpecializedSolver::process_depth_one_feature(const Dataview& dataview,
             continue;
         }
 
-        global_skip--;
         tree.can_skip--;
 
-        if (current_feature_data.unique_value_index == tree.previous_unique_value_index || tree.can_skip > 0 || global_skip > 0) {
+        if (current_feature_data.unique_value_index == tree.previous_unique_value_index || tree.can_skip > 0){
             tree.current_element_count++;
             tree.current_label_frequency[current_feature_data.label]++;
             tree.previous_value = current_feature_data.value;
@@ -196,7 +194,6 @@ void SpecializedSolver::process_depth_one_feature(const Dataview& dataview,
             tree.best_right_label = right_label;
         } else {
             tree.can_skip = tree.classification_score - (left_classification_score + right_classification_score);
-            global_skip = std::max(global_skip, upper_bound - (dataset_size - left_tree.classification_score - right_tree.classification_score));
         }
 
         int remaining_size = tree.size - tree.current_element_count;
diff --git a/code/Utilities/include/dynamic_bitset.h b/code/Utilities/include/dynamic_bitset.h
@@ -10,15 +10,19 @@
 #include <algorithm>
 #include <cstddef>
 #include <cstring>
+#include <climits>
 
 struct dynamic_bitset {
 	using base_type = unsigned long;
+	static constexpr size_t BITS_PER_ELEMENT = sizeof(base_type) * CHAR_BIT;
+	static constexpr size_t BYTES_PER_ELEMENT = sizeof(base_type);
 
 	base_type* bitset;
 	size_t elements;
 
 	dynamic_bitset(size_t size) {
-		elements = (size - 1) / sizeof(base_type) + 1;
+		RUNTIME_ASSERT(size > 0, "Cannot create an empty bitset");
+		elements = (size - 1) / BITS_PER_ELEMENT + 1;
 		bitset = new base_type[elements];
 		std::fill(bitset, bitset + elements, 0);
 	}
@@ -31,7 +35,7 @@ struct dynamic_bitset {
 
 	dynamic_bitset(const dynamic_bitset& other) : elements(other.elements) {
 		bitset = new base_type[elements];
-		std::memcpy(bitset, other.bitset, elements * sizeof(base_type));
+		std::memcpy(bitset, other.bitset, elements * BYTES_PER_ELEMENT);
 	}
 
 	dynamic_bitset& operator=(const dynamic_bitset& other) {
@@ -40,7 +44,7 @@ struct dynamic_bitset {
 
 		elements = other.elements;
 		base_type* new_bitset = new base_type[elements];
-		std::memcpy(new_bitset, other.bitset, elements * sizeof(base_type));
+		std::memcpy(new_bitset, other.bitset, elements * BYTES_PER_ELEMENT);
 		delete[] bitset;
 		bitset = new_bitset;
 		return *this;
@@ -58,20 +62,23 @@ struct dynamic_bitset {
 	}
 
 	void set_bit(size_t index) {
-		size_t element = index / sizeof(base_type);
-		size_t bit_index = index % sizeof(base_type);
+		size_t element = index / BITS_PER_ELEMENT;
+		RUNTIME_ASSERT(element <= elements, "set_bit - Writing beyond the bitset size: byte index " << element << ", while max is " << elements);
+		size_t bit_index = index % BITS_PER_ELEMENT;
 		bitset[element] |= 1UL << bit_index;
 	}
 
 	void clear_bit(size_t index) {
-		size_t element = index / sizeof(base_type);
-		size_t bit_index = index % sizeof(base_type);
+		size_t element = index / BITS_PER_ELEMENT;
+		RUNTIME_ASSERT(element <= elements, "clear_bit - Writing beyond the bitset size: byte index " << element << ", while max is " << elements);
+		size_t bit_index = index % BITS_PER_ELEMENT;
 		bitset[element] &= 1UL << bit_index;
 	}
 
 	void toggle_bit(size_t index) {
-		size_t element = index / sizeof(base_type);
-		size_t bit_index = index % sizeof(base_type);
+		size_t element = index / BITS_PER_ELEMENT;
+		RUNTIME_ASSERT(element <= elements, "toggle_bit - Writing beyond the bitset size: byte index " << element << ", while max is " << elements);
+		size_t bit_index = index % BITS_PER_ELEMENT;
 		bitset[element] ^= 1UL << bit_index;
 	}
 
diff --git a/pyproject.toml b/pyproject.toml
@@ -1,16 +1,16 @@
 [build-system]
 requires = [
-    "setuptools>=42",
+    "setuptools>=42, <72.2.0",
     "pybind11>=2.12.0"
 ]
 build-backend = "setuptools.build_meta"
-target-version = "py38"
 
 [project]
 name = "pycontree"
-version = "1.0"
+version = "1.0.4"
 description = "Python Wrapper ConTree: Optimal Decision Trees for Continuous Feature Data"
 license=  {file = "LICENSE"}
+readme = "README.md"
 authors = [
     {name = "Cătălin E. Briţa", email="C.E.Brita@student.tudelft.nl"},
     {name = "Jacobus G. M. van der Linden", email="J.G.M.vanderLinden@tudelft.nl"},
@@ -32,13 +32,8 @@ classifiers = [
     "License :: OSI Approved :: MIT License",
     "Operating System :: OS Independent", ]
 
-[project.optional-dependencies]
-dev = []
+[tool.setuptools.packages.find]
+exclude = ["datasets*", "train-datasets*", "examples*"]
 
-[tool.pytest.ini_options]
-pythonpath = [
-  "."
-]
-
-[tool.setuptools.packages]
-find = {}  # Scanning implicit namespaces is active by default
+[project.urls]
+Repository = "https://github.com/consol-lab/contree"
diff --git a/setup.py b/setup.py
@@ -6,7 +6,7 @@
 # Define package metadata
 package_name = 'pycontree'
 extension_name = 'ccontree'
-__version__ = "1.0"
+__version__ = "1.0.4"
 
 ext_modules = [
     Pybind11Extension(package_name + '.' + extension_name,
@@ -23,5 +23,7 @@
     version=__version__,
     ext_modules=ext_modules,
     dev_requires=[],
-    install_requires=['pandas', 'numpy']
+    install_requires=['pandas', 'numpy'],
+    long_description=open('README.md').read(),
+    long_description_content_type='text/markdown'
 )

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+# Include everything in the code folder`
	`2`	`+recursive-include code *`