refactor importClass and valgrind

Makefile

@@ -4,7 +4,7 @@ SHELL := /bin/bash
 
 f_release = build_release
 f_debug = build_debug
-app_targets = example
+app_targets = example test
 test_targets = unit_tests_pywrap
 n_procs = -j 16
 

example/CMakeLists.txt

@@ -5,4 +5,7 @@ include_directories(${PyWrap_SOURCE_DIR}/src)
 include_directories(${TORCH_INCLUDE_DIRS})
 
 add_executable(example example.cc)
-target_link_libraries(example PyWrap)
+target_link_libraries(example PyWrap)
+
+add_executable(test test.cpp)
+target_link_libraries(test ${Python3_LIBRARIES})

example/example.cc

@@ -1,10 +1,10 @@
 #include <torch/torch.h>
 #include "ArffFiles.h"
+#include <random> 
+#include <algorithm>
 #include <vector>
 #include <string>
 #include <iostream>
-#include <map>
-#include <tuple>
 #include "STree.h"
 #include "SVC.h"
 #include "RandomForest.h"
@@ -36,21 +36,42 @@ tuple<Tensor, Tensor> loadDataset(const string& name, bool class_last)
     return { Xd, torch::tensor(y, torch::kInt32) };
 }
 
+pair<torch::Tensor, torch::Tensor> get_train_test_indices(int size)
+{
+    int seed = 17;
+    float train_size_p = 0.7;
+    int train_size = static_cast<int>(size * train_size_p);
+    int test_size = size - train_size;
+    std::vector<int> indices(size);
+    std::iota(indices.begin(), indices.end(), 0);
+    shuffle(indices.begin(), indices.end(), std::default_random_engine(seed));
+    auto train_indices = torch::zeros({ train_size }, torch::kInt32);
+    auto test_indices = torch::zeros({ test_size }, torch::kInt32);
+    for (auto i = 0; i < size; ++i) {
+        if (i < train_size) {
+            train_indices[i] = indices[i];
+        } else if (i < size) {
+            test_indices[i - train_size] = indices[i];
+        }
+    }
+    return { train_indices, test_indices };
+}
+
 int main(int argc, char* argv[])
 {
     using json = nlohmann::json;
     cout << "* Begin." << endl;
     {
         using namespace torch::indexing;
-        auto datasetName = "iris";
+        auto datasetName = "wine";
         bool class_last = true;
         auto [X, y] = loadDataset(datasetName, class_last);
-        auto m = y.size(0);
-        int train_split = m * .7;
-        auto Xtrain = X.index({ "...", Slice(0, train_split) });
-        auto ytrain = y.index({ Slice(0, train_split) });
-        auto Xtest = X.index({ "...", Slice(train_split, m) });
-        auto ytest = y.index({ Slice(train_split, m) });
+        // Split train/test
+        auto [train_indices, test_indices] = get_train_test_indices(X.size(1));
+        auto Xtrain = X.index({ "...", train_indices });
+        auto ytrain = y.index({ train_indices });
+        auto Xtest = X.index({ "...", test_indices });
+        auto ytest = y.index({ test_indices });
         cout << "Dataset: " << datasetName << endl;
         cout << "X: " << X.sizes() << endl;
         cout << "y: " << y.sizes() << endl;
@@ -62,36 +83,36 @@ int main(int argc, char* argv[])
         // STree
         //
         auto clf = pywrap::STree();
-        clf.fit(X, y);
-        double clf_score = clf.score(X, y);
-        auto stree = pywrap::STree();
-        auto hyperparameters = json::parse("{\"C\": 0.7, \"max_iter\": 10000, \"kernel\": \"rbf\", \"random_state\": 17}");
-        stree.setHyperparameters(hyperparameters);
-        cout << "STree Version: " << clf.version() << endl;
-        auto prediction = clf.predict(X);
-        cout << "Prediction: " << endl << "{";
-        for (int i = 0; i < prediction.size(0); ++i) {
-            cout << prediction[i].item<int>() << ", ";
-        }
-        cout << "}" << endl;
+        clf.fit(Xtrain, ytest);
+        double clf_score = clf.score(Xtest, ytest);
+        // auto stree = pywrap::STree();
+        // auto hyperparameters = json::parse("{\"C\": 0.7, \"max_iter\": 10000, \"kernel\": \"rbf\", \"random_state\": 17}");
+        // stree.setHyperparameters(hyperparameters);
+        // cout << "STree Version: " << clf.version() << endl;
+        // auto prediction = clf.predict(X);
+        // cout << "Prediction: " << endl << "{";
+        // for (int i = 0; i < prediction.size(0); ++i) {
+        //     cout << prediction[i].item<int>() << ", ";
+        // }
+        // cout << "}" << endl;
         //
         // SVC
         //
-        auto svc = pywrap::SVC();
-        cout << "SVC with hyperparameters" << endl;
-        svc.fit(X, y);
+        // auto svc = pywrap::SVC();
+        // cout << "SVC with hyperparameters" << endl;
+        // svc.fit(Xtrain, ytrain);
         //
         // Random Forest
         //
-        cout << "Building Random Forest" << endl;
-        auto rf = pywrap::RandomForest();
-        rf.fit(Xtrain, ytrain);
+        // cout << "Building Random Forest" << endl;
+        // auto rf = pywrap::RandomForest();
+        // rf.fit(Xtrain, ytrain);
         //
         // XGBoost
         //
-        cout << "Building XGBoost" << endl;
-        auto xg = pywrap::XGBoost();
-        cout << "Fitting XGBoost" << endl;
+        // cout << "Building XGBoost" << endl;
+        // auto xg = pywrap::XGBoost();
+        // cout << "Fitting XGBoost" << endl;
         // xg.fit(Xtrain, ytrain);
         // double xg_score = xg.score(Xtest, ytest);
         //
@@ -100,10 +121,10 @@ int main(int argc, char* argv[])
         cout << "Scoring dataset: " << datasetName << endl;
         cout << "Scores:" << endl;
         cout << "STree Score ......: " << clf_score << endl;
-        cout << "STree train/test .: " << clf.fit(Xtrain, ytrain).score(Xtest, ytest) << endl;
-        cout << "STree hyper score : " << stree.fit(Xtrain, ytrain).score(Xtest, ytest) << endl;
-        cout << "RandomForest Score: " << rf.score(Xtest, ytest) << endl;
-        cout << "SVC Score ........: " << svc.score(X, y) << endl;
+        // cout << "STree train/test .: " << clf.fit(Xtrain, ytrain).score(Xtest, ytest) << endl;
+        // cout << "STree hyper score : " << stree.fit(Xtrain, ytrain).score(Xtest, ytest) << endl;
+        // cout << "RandomForest Score: " << rf.score(Xtest, ytest) << endl;
+        // cout << "SVC Score ........: " << svc.score(Xtest, ytest) << endl;
         // cout << "XGBoost Score ....: " << xg_score << endl;
     }
     cout << "* End." << endl;

example/test.cpp

@@ -0,0 +1,85 @@
+#include <Python.h>
+#include <iostream>
+#include <string>
+
+class Test {
+public:
+    Test() = default;
+    ~Test()
+    {
+        Py_DECREF(test);
+        Py_DECREF(object);
+    }
+    void create()
+    {
+        test = Py_BuildValue("s", "Hello World from a method in Embedded Python!!!");
+        std::cout << "test " << test->ob_refcnt << std::endl;
+        std::string moduleName = "sklearn.svm.SVC";
+        PyObject* pyModuleName = PyUnicode_FromString(moduleName.c_str());
+        std::cout << "py_module_name " << pyModuleName->ob_refcnt << std::endl;
+        object = PyImport_AddModuleObject(pyModuleName);
+        if (PyErr_Occurred()) {
+            std::cerr << "Couldn't import module " + moduleName;
+            Py_XDECREF(test);
+            Py_XDECREF(object);
+            Py_XDECREF(pyModuleName);
+            exit(1);
+        }
+        std::cout << "module " << object->ob_refcnt << std::endl;
+        Py_XDECREF(pyModuleName);
+    }
+    PyObject* test;
+    PyObject* object;
+
+};
+
+void print(PyObject* object, std::string name)
+{
+    std::cout << name << " " << object->ob_refcnt << std::endl;
+}
+
+int main(int argc, char* argv[])
+{
+    Py_Initialize();
+    std::string moduleName = "sklearn.svm";
+    std::cout << "1a" << std::endl;
+    PyObject* module = PyImport_ImportModule(moduleName.c_str());
+    if (PyErr_Occurred()) {
+        std::cerr << "Couldn't import module " << moduleName;
+        exit(1);
+    }
+    print(module, "module");
+    std::cout << "1b" << std::endl;
+    std::string className = "SVC";
+    PyObject* classObject = PyObject_GetAttrString(module, className.c_str());
+    if (PyErr_Occurred()) {
+        std::cerr << "Couldn't find class " << className;
+        exit(1);
+    }
+    print(classObject, "classObject");
+    std::cout << "1c" << std::endl;
+    PyObject* instance = PyObject_CallObject(classObject, NULL);
+    if (PyErr_Occurred()) {
+        std::cerr << "Couldn't create instance of class " << className;
+        exit(1);
+    }
+    print(instance, "instance");
+    std::cout << std::string(40, '-') << std::endl;
+    print(module, "module");
+    print(classObject, "classObject");
+    print(instance, "instance");
+    std::cout << "decref instance" << std::endl;
+    Py_DECREF(instance);
+    std::cout << "decref classobject" << std::endl;
+    Py_DECREF(classObject);
+    std::cout << "decref module" << std::endl;
+    Py_DECREF(module);
+    std::cout << std::string(40, '-') << std::endl;
+    print(module, "after decref module");
+    print(classObject, "after decref classObject");
+    print(instance, "after decref instance");
+    std::cout << "**call finalize" << std::endl;
+    Py_Finalize();
+    std::cout << "**after finalize" << std::endl;
+    return 0;
+}

src/PyWrap.cc

@@ -37,27 +37,25 @@ namespace pywrap {
     }
     void PyWrap::importClass(const clfId_t id, const std::string& moduleName, const std::string& className)
     {
+        std::lock_guard<std::mutex> lock(mutex);
         auto result = moduleClassMap.find(id);
         if (result != moduleClassMap.end()) {
             return;
         }
-        CPyObject module = PyImport_ImportModule(moduleName.c_str());
+        std::cout << "1a" << std::endl;
+        PyObject* module = PyImport_ImportModule(moduleName.c_str());
         if (PyErr_Occurred()) {
             errorAbort("Couldn't import module " + moduleName);
         }
-        CPyObject classObject = PyObject_GetAttrString(module, className.c_str());
+        PyObject* classObject = PyObject_GetAttrString(module, className.c_str());
         if (PyErr_Occurred()) {
             errorAbort("Couldn't find class " + className);
         }
-        CPyObject instance = PyObject_CallObject(classObject, NULL);
+        PyObject* instance = PyObject_CallObject(classObject, NULL);
         if (PyErr_Occurred()) {
             errorAbort("Couldn't create instance of class " + className);
         }
-        std::lock_guard<std::mutex> lock(mutex);
-        module.AddRef();
-        classObject.AddRef();
-        instance.AddRef();
-        moduleClassMap.insert({ id, { module.getObject(), classObject.getObject(), instance.getObject() } });
+        moduleClassMap.insert({ id, { module, classObject, instance } });
     }
     void PyWrap::clean(const clfId_t id)
     {

src/XGBoost.cc

@@ -1,5 +1,15 @@
 #include "XGBoost.h"
 
+
+
+
+See https ://stackoverflow.com/questions/36071672/using-xgboost-in-c
+
+
+
+
+
+
 namespace pywrap {
     std::string XGBoost::version()
     {