以前的文章寫了紅黑樹的實現，由於本身實現了插入和刪除的算法。爲了測試算法的性能，以及算法的正確性，又寫了幾個函數，用來檢查一棵樹是不是紅黑樹，並進行壓力測試，代碼以下：linux
1 import 'dart:math'; 2 import 'package:data_struct/tree/tree.dart'; 3 4 void run() { 5 var size = 100000, loops = 1024; 6 check(size, loops); 7 stress(size); 8 } 9 10 void check(int size, int loops) { 11 var rd = Random(); 12 for (var i = 0; i < loops; i++) { 13 print('the $i times check...'); 14 15 var arr = List.generate(size, (_) => rd.nextDouble() * size); 16 var tree = RBTree.from(arr); 17 var bh = _checkIsRBTree(tree); 18 print('black height: $bh\n\r'); 19 20 for (var d in arr) { 21 tree.delete(d); 22 _checkIsRBTree(tree); 23 } 24 25 tree = RBTree.from(arr); 26 for (var d in arr) { 27 tree.quickDelete(d); 28 _checkIsRBTree(tree); 29 } 30 } 31 } 32 33 void stress(int size) { 34 var rd = Random(); 35 var arr = List.generate(size, (_) => rd.nextDouble() * size); 36 37 var tree = RBTree.from(arr); 38 var st = DateTime.now(); 39 for (var d in arr) tree.delete(d); 40 var ft = DateTime.now(); 41 print('my delete implement cost:\t${ft.difference(st)}'); 42 43 tree = RBTree.from(arr); 44 st = DateTime.now(); 45 for (var d in arr) tree.quickDelete(d); 46 ft = DateTime.now(); 47 print(' linux implement cost:\t${ft.difference(st)}'); 48 } 49 50 void traverse(RBTNode r, void func(RBTNode r)) { 51 if (r != null) { 52 traverse(r.left, func); 53 func(r); 54 traverse(r.right, func); 55 } 56 } 57 58 int _checkIsRBTree(RBTree t) { 59 assert(t.isEmpty || (!t.isEmpty && t.root.isBlack)); 60 return _goThrough(t.root); 61 } 62 63 int _goThrough(RBTNode r) { 64 if (r == null) return 0; 65 assert(r.isBlack || (r.isRed && r.parent.isBlack)); 66 var lbh = _goThrough(r.left); 67 var rbh = _goThrough(r.right); 68 assert(lbh == rbh); 69 return lbh + (r.isRed ? 0 : 1); 70 }