C++ backend for Red Black Trees (#560)

Author: Kishan-Ved

pydatastructs/trees/_backend/cpp/BinarySearchTree.hpp

@@ -135,7 +135,7 @@ static PyObject* BinarySearchTree_insert(BinarySearchTree* self, PyObject* args)
     PyObject* key = Py_None;
     Py_INCREF(Py_None);
     PyObject* data = Py_None;
-    if (!PyArg_ParseTuple(args, "O|O", &key, &data)) { // ret_parent is optional
+    if (!PyArg_ParseTuple(args, "O|O", &key, &data)) { // data is optional
         return NULL;
     }
 

pydatastructs/trees/_backend/cpp/BinaryTreeTraversal.hpp

@@ -14,6 +14,7 @@
 #include "BinaryTree.hpp"
 #include "BinarySearchTree.hpp"
 #include "SelfBalancingBinaryTree.hpp"
+#include "RedBlackTree.hpp"
 
 typedef struct {
     PyObject_HEAD
@@ -36,7 +37,14 @@ static PyObject* BinaryTreeTraversal___new__(PyTypeObject* type, PyObject *args,
     if (PyType_Ready(&SelfBalancingBinaryTreeType) < 0) { // This has to be present to finalize a type object. This should be called on all type objects to finish their initialization.
         return NULL;
     }
-    if (PyObject_IsInstance(tree, (PyObject *)&SelfBalancingBinaryTreeType)) {
+    if (PyType_Ready(&RedBlackTreeType) < 0) { // This has to be present to finalize a type object. This should be called on all type objects to finish their initialization.
+        return NULL;
+    }
+
+    if (PyObject_IsInstance(tree, (PyObject *)&RedBlackTreeType)) {
+        self->tree = reinterpret_cast<RedBlackTree*>(tree)->sbbt->bst->binary_tree;
+    }
+    else if (PyObject_IsInstance(tree, (PyObject *)&SelfBalancingBinaryTreeType)) {
         self->tree = reinterpret_cast<SelfBalancingBinaryTree*>(tree)->bst->binary_tree;
     }
     else if (PyObject_IsInstance(tree, (PyObject *)&BinarySearchTreeType)) {

pydatastructs/trees/_backend/cpp/RedBlackTree.hpp

@@ -1,5 +1,5 @@
-#ifndef TREES_SELFBALANCINGSelfBalancingBinaryTree_HPP
-#define TREES_SELFBALANCINGSelfBalancingBinaryTree_HPP
+#ifndef TREES_SELFBALANCINGBINARYTREE_HPP
+#define TREES_SELFBALANCINGBINARYTREE_HPP
 
 #define PY_SSIZE_T_CLEAN
 #include <Python.h>

pydatastructs/trees/_backend/cpp/SelfBalancingBinaryTree.hpp

@@ -3,6 +3,7 @@
 #include "BinarySearchTree.hpp"
 #include "BinaryTreeTraversal.hpp"
 #include "SelfBalancingBinaryTree.hpp"
+#include "RedBlackTree.hpp"
 
 static struct PyModuleDef trees_struct = {
     PyModuleDef_HEAD_INIT,
@@ -40,5 +41,11 @@ PyMODINIT_FUNC PyInit__trees(void) {
     Py_INCREF(&SelfBalancingBinaryTreeType);
     PyModule_AddObject(trees, "SelfBalancingBinaryTree", reinterpret_cast<PyObject*>(&SelfBalancingBinaryTreeType));
 
+    if (PyType_Ready(&RedBlackTreeType) < 0) {
+        return NULL;
+    }
+    Py_INCREF(&RedBlackTreeType);
+    PyModule_AddObject(trees, "RedBlackTree", reinterpret_cast<PyObject*>(&RedBlackTreeType));
+
     return trees;
 }

pydatastructs/trees/_backend/cpp/trees.cpp

@@ -1199,9 +1199,18 @@ class RedBlackTree(SelfBalancingBinaryTree):
     pydatastructs.trees.binary_trees.SelfBalancingBinaryTree
     """
 
+    def __new__(cls, key=None, root_data=None, comp=None,
+                is_order_statistic=False, **kwargs):
+        backend = kwargs.get('backend', Backend.PYTHON)
+        if backend == Backend.CPP:
+            if comp is None:
+                comp = lambda key1, key2: key1 < key2
+            return _trees.RedBlackTree(key, root_data, comp, is_order_statistic, **kwargs) # If any argument is not given, then it is passed as None, except for comp
+        return super().__new__(cls, key, root_data, comp, is_order_statistic, **kwargs)
+
     @classmethod
     def methods(cls):
-        return ['insert', 'delete']
+        return ['__new__', 'insert', 'delete']
 
     def _get_parent(self, node_idx):
         return self.tree[node_idx].parent

pydatastructs/trees/binary_trees.py

@@ -1,5 +1,5 @@
 import timeit, functools, os, pytest
-from pydatastructs.trees.binary_trees import (BinarySearchTree)
+from pydatastructs.trees.binary_trees import (BinarySearchTree, RedBlackTree)
 from pydatastructs.utils.misc_util import Backend
 
 @pytest.mark.xfail
@@ -22,7 +22,55 @@ def g(backend, tree):
         for node in range(-1000, 1000):
             tree.search(node)
     def h(backend, tree):
-        for node in range(2000):
+        for node in range(-1000, 1000):
+            tree.delete(node)
+
+    kwds_dict_PY = {"backend": Backend.PYTHON, "tree":b1}
+    kwds_dict_CPP = {"backend": Backend.CPP, "tree":b2}
+
+    timer_python = timeit.Timer(functools.partial(f, **kwds_dict_PY))
+    python_insert = min(timer_python.repeat(repeat, number))
+
+    timer_cpp = timeit.Timer(functools.partial(f, **kwds_dict_CPP))
+    cpp_insert = min(timer_cpp.repeat(repeat, number))
+    assert cpp_insert < python_insert
+
+    timer_python = timeit.Timer(functools.partial(g, **kwds_dict_PY))
+    python_search = min(timer_python.repeat(repeat, number))
+
+    timer_cpp = timeit.Timer(functools.partial(g, **kwds_dict_CPP))
+    cpp_search = min(timer_cpp.repeat(repeat, number))
+    assert cpp_search < python_search
+
+    timer_python = timeit.Timer(functools.partial(h, **kwds_dict_PY))
+    python_delete = min(timer_python.repeat(repeat, number))
+
+    timer_cpp = timeit.Timer(functools.partial(h, **kwds_dict_CPP))
+    cpp_delete = min(timer_cpp.repeat(repeat, number))
+    assert cpp_delete < python_delete
+
+@pytest.mark.xfail
+def test_RedBlackTree(**kwargs):
+    cpp = Backend.CPP
+    repeat = 1
+    number = 1
+
+    size = int(os.environ.get("PYDATASTRUCTS_BENCHMARK_SIZE", "1000"))
+    size = kwargs.get("size", size)
+
+    RBT = RedBlackTree
+    b1 = RBT(backend=Backend.PYTHON)
+    b2 = RBT(backend=Backend.CPP)
+
+    def f(backend, tree):
+        for node in range(-1000,1000):
+            tree.insert(node, node)
+
+    def g(backend, tree):
+        for node in range(-1000, 1000):
+            tree.search(node)
+    def h(backend, tree):
+        for node in range(-1000, 1000):
             tree.delete(node)
 
     kwds_dict_PY = {"backend": Backend.PYTHON, "tree":b1}

pydatastructs/trees/tests/benchmarks/test_binary_trees.py

@@ -542,8 +542,8 @@ def test_SplayTree():
     assert [node.key for node in in_order] == [20, 30, 50, 55, 100, 200]
     assert [node.key for node in pre_order] == [200, 55, 50, 30, 20, 100]
 
-def test_RedBlackTree():
-    tree = RedBlackTree()
+def _test_RedBlackTree(backend):
+    tree = RedBlackTree(backend=backend)
     tree.insert(10, 10)
     tree.insert(18, 18)
     tree.insert(7, 7)
@@ -556,8 +556,9 @@ def test_RedBlackTree():
     tree.insert(2, 2)
     tree.insert(17, 17)
     tree.insert(6, 6)
+    assert str(tree) == "[(11, 10, 10, 3), (10, 18, 18, None), (None, 7, 7, None), (None, 15, 15, None), (0, 16, 16, 6), (None, 30, 30, None), (1, 25, 25, 7), (5, 40, 40, 8), (None, 60, 60, None), (None, 2, 2, None), (None, 17, 17, None), (9, 6, 6, 2)]"
 
-    trav = BinaryTreeTraversal(tree)
+    trav = BinaryTreeTraversal(tree, backend=backend)
     in_order = trav.depth_first_search(order='in_order')
     pre_order = trav.depth_first_search(order='pre_order')
     assert [node.key for node in in_order] == [2, 6, 7, 10, 15, 16, 17, 18, 25, 30, 40, 60]
@@ -583,7 +584,7 @@ def test_RedBlackTree():
     assert tree.upper_bound(60) is None
     assert tree.upper_bound(61) is None
 
-    tree = RedBlackTree()
+    tree = RedBlackTree(backend=backend)
 
     assert tree.lower_bound(1) is None
     assert tree.upper_bound(0) is None
@@ -606,10 +607,11 @@ def test_RedBlackTree():
     tree.insert(160)
     tree.insert(170)
     tree.insert(180)
+    assert str(tree) == "[(None, 10, None, None), (0, 20, None, 2), (None, 30, None, None), (1, 40, None, 5), (None, 50, None, None), (4, 60, None, 6), (None, 70, None, None), (3, 80, None, 11), (None, 90, None, None), (8, 100, None, 10), (None, 110, None, None), (9, 120, None, 13), (None, 130, None, None), (12, 140, None, 15), (None, 150, None, None), (14, 160, None, 16), (None, 170, None, 17), (None, 180, None, None)]"
 
     assert tree._get_sibling(7) is None
 
-    trav = BinaryTreeTraversal(tree)
+    trav = BinaryTreeTraversal(tree, backend=backend)
     in_order = trav.depth_first_search(order='in_order')
     pre_order = trav.depth_first_search(order='pre_order')
     assert [node.key for node in in_order] == [10, 20, 30, 40, 50, 60, 70, 80, 90,
@@ -684,36 +686,42 @@ def test_RedBlackTree():
     assert [node.key for node in in_order if node.key is not None] == []
     assert [node.key for node in pre_order if node.key is not None] == []
 
-    tree = RedBlackTree()
+    tree = RedBlackTree(backend=backend)
     tree.insert(50)
     tree.insert(40)
     tree.insert(30)
     tree.insert(20)
     tree.insert(10)
     tree.insert(5)
 
-    trav = BinaryTreeTraversal(tree)
+    trav = BinaryTreeTraversal(tree, backend=backend)
     in_order = trav.depth_first_search(order='in_order')
     pre_order = trav.depth_first_search(order='pre_order')
     assert [node.key for node in in_order] == [5, 10, 20, 30, 40, 50]
     assert [node.key for node in pre_order] == [40, 20, 10, 5, 30, 50]
 
+    assert tree.search(50) == 0
+    assert tree.search(20) == 3
+    assert tree.search(30) == 2
     tree.delete(50)
     tree.delete(20)
     tree.delete(30)
+    assert tree.search(50) is None
+    assert tree.search(20) is None
+    assert tree.search(30) is None
 
     in_order = trav.depth_first_search(order='in_order')
     pre_order = trav.depth_first_search(order='pre_order')
     assert [node.key for node in in_order] == [5, 10, 40]
     assert [node.key for node in pre_order] == [10, 5, 40]
 
-    tree = RedBlackTree()
+    tree = RedBlackTree(backend=backend)
     tree.insert(10)
     tree.insert(5)
     tree.insert(20)
     tree.insert(15)
 
-    trav = BinaryTreeTraversal(tree)
+    trav = BinaryTreeTraversal(tree, backend=backend)
     in_order = trav.depth_first_search(order='in_order')
     pre_order = trav.depth_first_search(order='pre_order')
     assert [node.key for node in in_order] == [5, 10, 15, 20]
@@ -726,15 +734,15 @@ def test_RedBlackTree():
     assert [node.key for node in in_order] == [10, 15, 20]
     assert [node.key for node in pre_order] == [15, 10, 20]
 
-    tree = RedBlackTree()
+    tree = RedBlackTree(backend=backend)
     tree.insert(10)
     tree.insert(5)
     tree.insert(20)
     tree.insert(15)
     tree.insert(2)
     tree.insert(6)
 
-    trav = BinaryTreeTraversal(tree)
+    trav = BinaryTreeTraversal(tree,backend=backend)
     in_order = trav.depth_first_search(order='in_order')
     pre_order = trav.depth_first_search(order='pre_order')
     assert [node.key for node in in_order] == [2, 5, 6, 10, 15, 20]
@@ -746,3 +754,9 @@ def test_RedBlackTree():
     pre_order = trav.depth_first_search(order='pre_order')
     assert [node.key for node in in_order] == [2, 5, 6, 15, 20]
     assert [node.key for node in pre_order] == [6, 5, 2, 20, 15]
+
+def test_RedBlackTree():
+    _test_RedBlackTree(Backend.PYTHON)
+
+def test_cpp_RedBlackTree():
+    _test_RedBlackTree(Backend.CPP)

pydatastructs/trees/tests/test_binary_trees.py

@@ -17,6 +17,7 @@ typedef struct {
     long height;
     PyObject* parent;
     long size;
+    long color;
 } TreeNode;
 
 static void TreeNode_dealloc(TreeNode *self) {
@@ -40,6 +41,7 @@ static PyObject* TreeNode___new__(PyTypeObject* type, PyObject *args, PyObject *
     self->height = 0;
     self->size = 1;
     self->is_root = false;
+    self->color = 1;
 
     return reinterpret_cast<PyObject*>(self);
 }
@@ -59,6 +61,7 @@ static struct PyMemberDef TreeNode_PyMemberDef[] = {
     {"left", T_OBJECT, offsetof(TreeNode, left), 0, "TreeNode left"},
     {"right", T_OBJECT, offsetof(TreeNode, right), 0, "TreeNode right"},
     {"parent", T_OBJECT, offsetof(TreeNode, parent), 0, "TreeNode parent"},
+    {"color", T_LONG, offsetof(TreeNode, size), 0, "RedBlackTreeNode color"},
     {NULL},
 };