Fix entropy occasional error

* Update version and dependencies

* Fix conan and create new version (#11)

* First approach

* Fix debug conan build target

* Add viewcoverage and fix coverage generation

* Add more tests to cover new integrity checks

* Add tests to accomplish 100%

* Fix conan-create makefile target

* Update debug build

* Fix release build

* Update github build workflow

* Update github workflow

* Update github workflow

* Update github workflow

* Update github workflow remove coverage report

.github/workflows/build.yml

@@ -19,26 +19,29 @@ jobs:
           submodules: recursive
       - name: Install sonar-scanner and build-wrapper
         uses: SonarSource/sonarcloud-github-c-cpp@v2
+      - name: Install Python and Conan
+        run: |
+          sudo apt-get update
+          sudo apt-get -y install python3 python3-pip
+          pip3 install conan
       - name: Install lcov & gcovr
         run: |
           sudo apt-get -y install lcov
           sudo apt-get -y install gcovr
-      - name: Install Libtorch
+      - name: Setup Conan profileson
         run: |
-          wget https://download.pytorch.org/libtorch/cpu/libtorch-cxx11-abi-shared-with-deps-2.3.1%2Bcpu.zip
-          unzip libtorch-cxx11-abi-shared-with-deps-2.3.1+cpu.zip
+          conan profile detect --force
+          conan remote add cimmeria https://conan.rmontanana.es/artifactory/api/conan/Cimmeria
+      - name: Install dependencies with Conan
+        run: |
+          conan install . --build=missing -of build_debug -s build_type=Debug -o enable_testing=True
+      - name: Configure with CMake
+        run: |
+          cmake -S . -B build_debug -DCMAKE_TOOLCHAIN_FILE=build_debug/build/Debug/generators/conan_toolchain.cmake -DCMAKE_BUILD_TYPE=Debug -DENABLE_TESTING=ON
       - name: Tests & build-wrapper
         run: |
-          cmake -S . -B build -Wno-dev -DCMAKE_PREFIX_PATH=$(pwd)/libtorch -DCMAKE_BUILD_TYPE=Debug -DENABLE_TESTING=ON
-          build-wrapper-linux-x86-64 --out-dir ${{ env.BUILD_WRAPPER_OUT_DIR }} cmake --build build/ --config Debug
-          cmake --build build -j 4
-          cd build
-          ctest -C Debug --output-on-failure -j 4
-          gcovr -f ../src/CPPFImdlp.cpp -f ../src/Metrics.cpp -f ../src/BinDisc.cpp -f ../src/Discretizer.cpp --txt --sonarqube=coverage.xml
-      - name: Run sonar-scanner
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          SONAR_TOKEN: ${{ secrets.SONAR_TOKEN }}
-        run: |
-          sonar-scanner --define sonar.cfamily.compile-commands="${{ env.BUILD_WRAPPER_OUT_DIR }}" \
-                        --define sonar.coverageReportPaths=build/coverage.xml
+          build-wrapper-linux-x86-64 --out-dir ${{ env.BUILD_WRAPPER_OUT_DIR }} cmake --build build_debug --config Debug -j 4
+          cp -r tests/datasets build_debug/tests/datasets
+          cd build_debug/tests
+          ctest --output-on-failure -j 4
+          

CHANGELOG.md

@@ -5,6 +5,24 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [2.1.2] - 2025-08-22
+
+### Added
+
+- make info now gives info about the build status
+
+### Fix
+
+- Mistake in entropy computation
+
+## [2.1.1] - 2025-07-17
+
+### Internal Changes
+
+- Updated Libtorch to version 2.7.1
+- Updated ArffFiles library to version 1.2.1
+- Enhance CMake configuration for better compatibility
+
 ## [2.1.0] - 2025-06-28
 
 ### Added

CMakeLists.txt

@@ -1,10 +1,10 @@
 cmake_minimum_required(VERSION 3.20)
 
-project(fimdlp 
+project(fimdlp
+    VERSION 2.1.2
     LANGUAGES CXX
     DESCRIPTION "Discretization algorithm based on the paper by Fayyad & Irani Multi-Interval Discretization of Continuous-Valued Attributes for Classification Learning."
     HOMEPAGE_URL "https://github.com/rmontanana/mdlp"
-    VERSION 2.1.0
 )
 set(CMAKE_CXX_STANDARD 17)
 cmake_policy(SET CMP0135 NEW)

CONAN_README.md

@@ -7,9 +7,11 @@ This directory contains the Conan package configuration for the fimdlp library.
 The package manages the following dependencies:
 
 ### Build Requirements
+
 - **libtorch/2.4.1** - PyTorch C++ library for tensor operations
 
 ### Test Requirements (when testing enabled)
+
 - **catch2/3.8.1** - Modern C++ testing framework
 - **arff-files** - ARFF file format support (included locally in tests/lib/Files/)
 
@@ -67,7 +69,7 @@ conan create . -o shared=True --profile:build=default --profile:host=default
 
 ```bash
 # Add Cimmeria remote
-conan remote add cimmeria <cimmeria-server-url>
+conan remote add cimmeria https://conan.rmontanana.es/artifactory/api/conan/Cimmeria
 
 # Login to Cimmeria
 conan remote login cimmeria <username>
@@ -150,4 +152,4 @@ conan create . -o enable_testing=True
 
 - C++17 compatible compiler
 - CMake 3.20 or later
-- Conan 2.0 or later
+- Conan 2.0 or later

Makefile

@@ -1,5 +1,5 @@
 SHELL := /bin/bash
-.DEFAULT_GOAL := release
+.DEFAULT_GOAL := help
 .PHONY: debug release install test conan-create viewcoverage
 lcov := lcov
 
@@ -8,21 +8,48 @@ f_release = build_release
 genhtml = genhtml
 docscdir = docs
 
+# Set the number of parallel jobs to the number of available processors minus 7
+CPUS := $(shell getconf _NPROCESSORS_ONLN 2>/dev/null \
+                 || nproc --all 2>/dev/null \
+                 || sysctl -n hw.ncpu)
+JOBS := $(shell n=$(CPUS); [ $${n} -gt 7 ] && echo $$((n-7)) || echo 1)
+
+# Colors for output
+GREEN = \033[0;32m
+YELLOW = \033[1;33m
+RED = \033[0;31m
+NC = \033[0m # No Color
+
 define build_target
 	@echo ">>> Building the project for $(1)..."
 	@if [ -d $(2) ]; then rm -fr $(2); fi
-	@conan install . --build=missing -of $(2) -s build_type=$(1)
+	@conan install . --build=missing -of $(2) -s build_type=$(1) $(4)
 	@cmake -S . -B $(2) -DCMAKE_TOOLCHAIN_FILE=$(2)/build/$(1)/generators/conan_toolchain.cmake -DCMAKE_BUILD_TYPE=$(1) -D$(3)
 	@cmake --build $(2) --config $(1) -j 8
 endef
 
+define status_file_folder
+	@if [ -d $(1) ]; then \
+		st1=" ✅ $(GREEN)"; \
+	else \
+		st1=" ❌ $(RED)"; \
+	fi; \
+	if [ -f $(1)/libfimdlp.a ]; then \
+		st2=" ✅ $(GREEN)"; \
+	else \
+		st2=" ❌ $(RED)"; \
+	fi; \
+	printf "  $(YELLOW)$(2):$(NC) $$st1 Folder $(NC)  $$st2 Library $(NC)\n"
+endef
+
+
 debug: ## Build Debug version of the library
-	@$(call build_target,"Debug","$(f_debug)", "ENABLE_TESTING=ON")
+	@$(call build_target,"Debug","$(f_debug)", "ENABLE_TESTING=ON", "-o enable_testing=True")
 
 release: ## Build Release version of the library
-	@$(call build_target,"Release","$(f_release)", "ENABLE_TESTING=OFF")
+	@$(call build_target,"Release","$(f_release)", "ENABLE_TESTING=OFF", "-o enable_testing=False")
 
-install: ## Install the project
+install: ## Install the library
 	@echo ">>> Installing the project..."
 	@cmake --build $(f_release) --target install -j 8		
 
@@ -31,7 +58,8 @@ test: ## Build Debug version and run tests
 	@$(MAKE) debug;
 	@cp -r tests/datasets $(f_debug)/tests/datasets
 	@cd $(f_debug)/tests && ctest --output-on-failure -j 8
-	@cd $(f_debug)/tests && $(lcov) --capture --directory ../ --demangle-cpp --ignore-errors source,source --ignore-errors mismatch --output-file coverage.info >/dev/null 2>&1; \
+	@echo ">>> Generating coverage report..."
+	@cd $(f_debug)/tests && $(lcov) --capture --directory ../ --demangle-cpp --ignore-errors source,source --ignore-errors mismatch --ignore-errors inconsistent --output-file coverage.info >/dev/null 2>&1; \
 	$(lcov) --remove coverage.info '/usr/*' --output-file coverage.info >/dev/null 2>&1; \
 	$(lcov) --remove coverage.info 'lib/*' --output-file coverage.info >/dev/null 2>&1; \
 	$(lcov) --remove coverage.info 'libtorch/*' --output-file coverage.info >/dev/null 2>&1; \
@@ -60,11 +88,41 @@ viewcoverage: ## View the html coverage report
 	@xdg-open $(docscdir)/coverage/index.html || open $(docscdir)/coverage/index.html 2>/dev/null
 	@echo ">>> Done";
 
+info: ## Show project information
+	@version=$$(grep -A1 "project(fimdlp" CMakeLists.txt | grep "VERSION" | sed 's/.*VERSION \([0-9.]*\).*/\1/'); \
+	printf "$(GREEN)FImdlp Library: $(YELLOW)ver. $$version$(NC)\n"
+	@echo ""
+	@printf "$(GREEN)Project folders:$(NC)\n"
+	$(call status_file_folder, $(f_release), "Build\ Release")
+	$(call status_file_folder, $(f_debug), "Build\ Debug\ \ ")
+	@echo ""
+	@printf "$(GREEN)Build commands:$(NC)\n"
+	@printf "   $(YELLOW)make release$(NC) - Build library for release\n"
+	@printf "   $(YELLOW)make debug$(NC)   - Build library for debug\n"
+	@printf "   $(YELLOW)make test$(NC)    - Run tests\n"
+	@printf "   $(YELLOW)Usage:$(NC) make help\n"
+	@echo ""
+	@printf "   $(YELLOW)Parallel Jobs:   $(GREEN)$(JOBS)$(NC)\n"
+
 conan-create: ## Create the conan package
 	@echo ">>> Creating the conan package..."
 	conan create . --build=missing -tf "" -s:a build_type=Release 
 	conan create . --build=missing -tf "" -s:a build_type=Debug -o "&:enable_testing=False"
 	@echo ">>> Done"
 
-
-
+help: ## Show help message
+	@IFS=$$'\n' ; \
+	help_lines=(`grep -Fh "##" $(MAKEFILE_LIST) | grep -Fv fgrep | sed -e 's/\\$$//' | sed -e 's/##/:/'`); \
+	printf "%s\n\n" "Usage: make [task]"; \
+	printf "%-20s %s\n" "task" "help" ; \
+	printf "%-20s %s\n" "------" "----" ; \
+	for help_line in $${help_lines[@]}; do \
+		IFS=$$':' ; \
+		help_split=($$help_line) ; \
+		help_command=`echo $${help_split[0]} | sed -e 's/^ *//' -e 's/ *$$//'` ; \
+		help_info=`echo $${help_split[2]} | sed -e 's/^ *//' -e 's/ *$$//'` ; \
+		printf '\033[36m'; \
+		printf "%-20s %s" $$help_command ; \
+		printf '\033[0m'; \
+		printf "%s\n" $$help_info; \
+	done

README.md

@@ -3,7 +3,7 @@
 [![Reliability Rating](https://sonarcloud.io/api/project_badges/measure?project=rmontanana_mdlp&metric=reliability_rating)](https://sonarcloud.io/summary/new_code?id=rmontanana_mdlp)
 [![Coverage Badge](https://img.shields.io/badge/Coverage-100,0%25-green)](html/index.html)
 [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/rmontanana/mdlp)
-[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.14245443.svg)](https://doi.org/10.5281/zenodo.14245443)
+[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.16025501.svg)](https://doi.org/10.5281/zenodo.16025501)
 
 # <img src="logo.png" alt="logo" width="50"/> mdlp
 
@@ -18,9 +18,7 @@ Other features:
 
 - Intervals with the same value of the variable are not taken into account for cutpoints.
 - Intervals have to have more than two examples to be evaluated (mdlp).
-
 - The algorithm returns the cut points for the variable.
-
 - The transform method uses the cut points returning its index in the following way:
 
         cut[i - 1] <= x < cut[i]

conanfile.py

@@ -54,7 +54,7 @@ class FimdlpConan(ConanFile):
         self.requires("libtorch/2.7.1")
         
     def build_requirements(self):
-        self.requires("arff-files/1.2.0") # for tests and sample
+        self.requires("arff-files/1.2.1") # for tests and sample
         if self.options.enable_testing: 
             self.test_requires("gtest/1.16.0")
     

src/BinDisc.cpp

@@ -98,7 +98,7 @@ namespace mdlp {
             const precision_t fraction =
                 (percentile / 100.0 - percentI) /
                 (static_cast<precision_t>(indexLower + 1) / static_cast<precision_t>(data.size() - 1) - percentI);
-            if (const auto value = data[indexLower] + (data[indexLower + 1] - data[indexLower]) * fraction; value != results.back() || first) // first needed as results.back() return is undefined for empty vectors
+            if (const auto value = data[indexLower] + (data[indexLower + 1] - data[indexLower]) * fraction; first || results.empty() || value != results.back()) // Check empty before calling back()
                 results.push_back(value);
             first = false;
         }

src/CPPFImdlp.h

@@ -41,6 +41,9 @@ namespace mdlp {
         pair<precision_t, size_t> valueCutPoint(size_t, size_t, size_t);
         inline precision_t safe_X_access(size_t idx) const
         {
+            if (indices.empty()) {
+                throw std::out_of_range("Indices array is empty");
+            }
             if (idx >= indices.size()) {
                 throw std::out_of_range("Index out of bounds for indices array");
             }
@@ -52,6 +55,9 @@ namespace mdlp {
         }
         inline label_t safe_y_access(size_t idx) const
         {
+            if (indices.empty()) {
+                throw std::out_of_range("Indices array is empty");
+            }
             if (idx >= indices.size()) {
                 throw std::out_of_range("Index out of bounds for indices array");
             }

src/Metrics.cpp

@@ -7,6 +7,7 @@
 #include "Metrics.h"
 #include <set>
 #include <cmath>
+#include <iostream>
 
 using namespace std;
 namespace mdlp {
@@ -18,8 +19,13 @@ namespace mdlp {
     int Metrics::computeNumClasses(size_t start, size_t end)
     {
         set<int> nClasses;
+        if (indices.empty() || start >= indices.size() || end > indices.size()) {
+            return 0;
+        }
         for (auto i = start; i < end; ++i) {
-            nClasses.insert(y[indices[i]]);
+            if (i < indices.size() && indices[i] < y.size()) {
+                nClasses.insert(y[indices[i]]);
+            }
         }
         return static_cast<int>(nClasses.size());
     }
@@ -38,7 +44,7 @@ namespace mdlp {
     {
         if (end - start < 2)
             return 0;
-            
+
         // Check cache first with read lock
         {
             std::lock_guard<std::mutex> lock(cache_mutex);
@@ -46,15 +52,37 @@ namespace mdlp {
                 return entropyCache[{start, end}];
             }
         }
-        
+
         // Compute entropy outside of lock
         precision_t p;
         precision_t ventropy = 0;
         int nElements = 0;
-        labels_t counts(numClasses + 1, 0);
         
-        for (auto i = &indices[start]; i != &indices[end]; ++i) {
-            counts[y[*i]]++;
+        if (indices.empty() || start >= indices.size() || end > indices.size()) {
+            return 0;
+        }
+        
+        // First pass: find max label to size counts array properly
+        size_t max_label = 0;
+        for (size_t i = start; i < end; ++i) {
+            if (i >= indices.size()) break;
+            size_t idx = indices[i];
+            if (idx >= y.size()) continue;
+            size_t label = y[idx];
+            if (label > max_label) {
+                max_label = label;
+            }
+        }
+        
+        labels_t counts(max_label + 1, 0);
+        
+        // Second pass: count occurrences
+        for (size_t i = start; i < end; ++i) {
+            if (i >= indices.size()) break;
+            size_t idx = indices[i];
+            if (idx >= y.size()) continue;
+            size_t label = y[idx];
+            counts[label]++;
             nElements++;
         }
         for (auto count : counts) {
@@ -63,13 +91,13 @@ namespace mdlp {
                 ventropy -= p * log2(p);
             }
         }
-        
+
         // Update cache with write lock
         {
             std::lock_guard<std::mutex> lock(cache_mutex);
             entropyCache[{start, end}] = ventropy;
         }
-        
+
         return ventropy;
     }
 
@@ -82,7 +110,7 @@ namespace mdlp {
                 return igCache[make_tuple(start, cut, end)];
             }
         }
-        
+
         // Compute information gain outside of lock
         precision_t iGain;
         precision_t entropyInterval;
@@ -91,7 +119,7 @@ namespace mdlp {
         size_t nElementsLeft = cut - start;
         size_t nElementsRight = end - cut;
         size_t nElements = end - start;
-        
+
         entropyInterval = entropy(start, end);
         entropyLeft = entropy(start, cut);
         entropyRight = entropy(cut, end);
@@ -99,13 +127,13 @@ namespace mdlp {
             (static_cast<precision_t>(nElementsLeft) * entropyLeft +
                 static_cast<precision_t>(nElementsRight) * entropyRight) /
             static_cast<precision_t>(nElements);
-            
+
         // Update cache with write lock
         {
             std::lock_guard<std::mutex> lock(cache_mutex);
             igCache[make_tuple(start, cut, end)] = iGain;
         }
-        
+
         return iGain;
     }
 

tests/Discretizer_unittest.cpp

@@ -41,7 +41,7 @@ namespace mdlp {
         Discretizer* disc = new BinDisc(4, strategy_t::UNIFORM);
         auto version = disc->version();
         delete disc;
-        EXPECT_EQ("2.1.0", version);
+        EXPECT_EQ("2.1.2", version);
     }
     TEST(Discretizer, BinIrisUniform)
     {

tests/FImdlp_unittest.cpp

@@ -417,6 +417,28 @@ namespace mdlp {
         EXPECT_THROW_WITH_MESSAGE(safe_y_access(2), std::out_of_range, "Index out of bounds for y array");
     }
 
+    TEST_F(TestFImdlp, SafeXAccessEmptyIndices)
+    {
+        // Test safe_X_access with empty indices array
+        X = { 1.0f, 2.0f, 3.0f };
+        y = { 1, 2, 3 };
+        indices = {}; // empty indices array
+
+        // This should trigger the indices.empty() exception in safe_X_access
+        EXPECT_THROW_WITH_MESSAGE(safe_X_access(0), std::out_of_range, "Indices array is empty");
+    }
+
+    TEST_F(TestFImdlp, SafeYAccessEmptyIndices)
+    {
+        // Test safe_y_access with empty indices array
+        X = { 1.0f, 2.0f, 3.0f };
+        y = { 1, 2, 3 };
+        indices = {}; // empty indices array
+
+        // This should trigger the indices.empty() exception in safe_y_access
+        EXPECT_THROW_WITH_MESSAGE(safe_y_access(0), std::out_of_range, "Indices array is empty");
+    }
+
     TEST_F(TestFImdlp, SafeSubtractUnderflow)
     {
         // Test safe_subtract with underflow condition (b > a)

tests/Metrics_unittest.cpp

@@ -37,12 +37,15 @@ namespace mdlp {
         y = { 1, 1, 1, 1, 1, 1, 1, 1, 2, 1 };
         setData(y, indices);
         ASSERT_NEAR(0.468996f, entropy(0, 10), precision);
+        y = { 0, 0, 1, 2, 3 };
+        ASSERT_NEAR(1.5f, entropy(0, 4), precision);
+        ASSERT_NEAR(1.921928f, entropy(0, 5), precision);
     }
 
     TEST_F(TestMetrics, EntropyDouble)
     {
         y = { 0, 0, 1, 2, 3 };
-        samples_t expected_entropies = { 0.0, 0.0, 0.91829583, 1.5, 1.4575424759098898 };
+        samples_t expected_entropies = { 0.0, 0.0, 0.91829583, 1.5, 1.9219280948873623 };
         for (auto idx = 0; idx < y.size(); ++idx) {
             ASSERT_NEAR(expected_entropies[idx], entropy(0, idx + 1), precision);
         }
@@ -56,4 +59,28 @@ namespace mdlp {
         setData(y, indices);
         ASSERT_NEAR(0.108032f, informationGain(0, 5, 10), precision);
     }
+
+    TEST_F(TestMetrics, EntropyBoundsChecking)
+    {
+        // Test the conditions that cause entropy to return 0
+        
+        // Test 1: Empty indices array
+        indices_t empty_indices = {};
+        labels_t test_y = { 1, 2, 3 };
+        setData(test_y, empty_indices);
+        EXPECT_EQ(0, entropy(0, 1)) << "Should return 0 when indices is empty";
+        
+        // Test 2: start >= indices.size()
+        indices_t small_indices = { 0, 1 };
+        setData(test_y, small_indices);
+        EXPECT_EQ(0, entropy(2, 3)) << "Should return 0 when start >= indices.size()";
+        EXPECT_EQ(0, entropy(5, 6)) << "Should return 0 when start >> indices.size()";
+        
+        // Test 3: end > indices.size()
+        EXPECT_EQ(0, entropy(0, 3)) << "Should return 0 when end > indices.size()";
+        EXPECT_EQ(0, entropy(1, 5)) << "Should return 0 when end >> indices.size()";
+        
+        // Test edge case: start == indices.size()
+        EXPECT_EQ(0, entropy(2, 2)) << "Should return 0 when start == indices.size()";
+    }
 }