DAG拓扑排序
2018-03-21 本文已影响0人
张前峰
/**
* Java: 无回路有向图(Directed Acyclic Graph)的拓扑排序
* 该DAG图是通过邻接表实现的。
*
* @author skywang
* @date 2014/04/22
*/
import java.io.IOException;
import java.util.Scanner;
import java.util.Stack;
import java.util.List;
import java.util.ArrayList;
import java.util.Queue;
import java.util.LinkedList;
public class ListDG {
// 邻接表中表对应的链表的顶点
private class ENode {
int ivex; // 该边所指向的顶点的位置
ENode nextEdge; // 指向下一条弧的指针
}
// 邻接表中表的顶点
private class VNode {
String data; // 顶点信息
ENode firstEdge; // 指向第一条依附该顶点的弧
};
private List<VNode> mVexs; // 顶点数组
/*
* 创建图(用已提供的矩阵)
*
* 参数说明: vexs -- 顶点数组 edges -- 边数组
*/
public ListDG(String[] vexs, String[][] edges) {
// 初始化"顶点数"和"边数"
int vlen = vexs.length;
int elen = edges.length;
// 初始化"顶点"
mVexs = new ArrayList<VNode>();
for (int i = 0; i < vlen; i++) {
// 新建VNode
VNode vnode = new VNode();
vnode.data = vexs[i];
vnode.firstEdge = null;
// 将vnode添加到数组mVexs中
mVexs.add(vnode);
}
// 初始化"边"
for (int i = 0; i < elen; i++) {
// 读取边的起始顶点和结束顶点
String c1 = edges[i][0];
String c2 = edges[i][1];
// 读取边的起始顶点和结束顶点
int p1 = getPosition(edges[i][0]);
int p2 = getPosition(edges[i][1]);
// 初始化node1
ENode node1 = new ENode();
node1.ivex = p2;
// 将node1链接到"p1所在链表的末尾"
if (mVexs.get(p1).firstEdge == null)
mVexs.get(p1).firstEdge = node1;
else
linkLast(mVexs.get(p1).firstEdge, node1);
}
}
/*
* 将node节点链接到list的最后
*/
private void linkLast(ENode list, ENode node) {
ENode p = list;
while (p.nextEdge != null)
p = p.nextEdge;
p.nextEdge = node;
}
/*
* 返回ch位置
*/
private int getPosition(String ch) {
for (int i = 0; i < mVexs.size(); i++)
if (mVexs.get(i).data == ch)
return i;
return -1;
}
/*
* 读取一个输入字符
*/
private char readChar() {
char ch = '0';
do {
try {
ch = (char) System.in.read();
} catch (IOException e) {
e.printStackTrace();
}
} while (!((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z')));
return ch;
}
/*
* 读取一个输入字符
*/
private int readInt() {
Scanner scanner = new Scanner(System.in);
return scanner.nextInt();
}
/*
* 深度优先搜索遍历图的递归实现
*/
private void DFS(int i, boolean[] visited) {
ENode node;
visited[i] = true;
System.out.printf("%s ", mVexs.get(i).data);
node = mVexs.get(i).firstEdge;
while (node != null) {
if (!visited[node.ivex])
DFS(node.ivex, visited);
node = node.nextEdge;
}
}
/*
* 深度优先搜索遍历图
*/
public void DFS() {
boolean[] visited = new boolean[mVexs.size()]; // 顶点访问标记
// 初始化所有顶点都没有被访问
for (int i = 0; i < mVexs.size(); i++)
visited[i] = false;
System.out.printf("== DFS: ");
for (int i = 0; i < mVexs.size(); i++) {
if (!visited[i])
DFS(i, visited);
}
System.out.printf("\n");
}
/*
* 广度优先搜索(类似于树的层次遍历)
*/
public void BFS() {
int head = 0;
int rear = 0;
int[] queue = new int[mVexs.size()]; // 辅组队列
boolean[] visited = new boolean[mVexs.size()]; // 顶点访问标记
for (int i = 0; i < mVexs.size(); i++)
visited[i] = false;
System.out.printf("== BFS: ");
for (int i = 0; i < mVexs.size(); i++) {
if (!visited[i]) {
visited[i] = true;
System.out.printf("%s ", mVexs.get(i).data);
queue[rear++] = i; // 入队列
}
while (head != rear) {
int j = queue[head++]; // 出队列
ENode node = mVexs.get(j).firstEdge;
while (node != null) {
int k = node.ivex;
if (!visited[k]) {
visited[k] = true;
System.out.printf("%s ", mVexs.get(k).data);
queue[rear++] = k;
}
node = node.nextEdge;
}
}
}
System.out.printf("\n");
}
/*
* 打印矩阵队列图
*/
public void print() {
System.out.printf("== List Graph:\n");
for (int i = 0; i < mVexs.size(); i++) {
System.out.printf("%d(%s): ", i, mVexs.get(i).data);
ENode node = mVexs.get(i).firstEdge;
while (node != null) {
System.out.printf("%d(%s) ", node.ivex, mVexs.get(node.ivex).data);
node = node.nextEdge;
}
System.out.printf("\n");
}
}
/*
* 拓扑排序
*
* 返回值: -1 -- 失败(由于内存不足等原因导致) 0 -- 成功排序,并输入结果 1 -- 失败(该有向图是有环的)
*/
public int topologicalSort() {
Stack<String> stack = new Stack<>();
int index = 0;
int num = mVexs.size();
int[] ins; // 入度数组
List<String> tops; // 拓扑排序结果数组,记录每个节点的排序后的序号。
Queue<Integer> queue; // 辅组队列
ins = new int[num];
tops = new Stack<String>();
queue = new LinkedList<Integer>();
// 统计每个顶点的入度数
for (int i = 0; i < num; i++) {
ENode node = mVexs.get(i).firstEdge;
while (node != null) {
ins[node.ivex]++;
node = node.nextEdge;
}
}
// 将所有入度为0的顶点入队列
for (int i = 0; i < num; i++)
if (ins[i] == 0)
queue.offer(i); // 入队列
while (!queue.isEmpty()) { // 队列非空
int j = queue.poll().intValue(); // 出队列。j是顶点的序号
tops.add(mVexs.get(j).data); // 将该顶点添加到tops中,tops是排序结果
index++;
ENode node = mVexs.get(j).firstEdge;// 获取以该顶点为起点的出边队列
// 将与"node"关联的节点的入度减1;
// 若减1之后,该节点的入度为0;则将该节点添加到队列中。
while (node != null) {
// 将节点(序号为node.ivex)的入度减1。
ins[node.ivex]--;
// 若节点的入度为0,则将其"入队列"
if (ins[node.ivex] == 0)
queue.offer(node.ivex); // 入队列
node = node.nextEdge;
}
}
if (index != num) {
System.out.printf("Graph has a cycle\n");
return 1;
}
// 打印拓扑排序结果
System.out.printf("== TopSort: ");
for (int i = 0; i < num; i++) {
System.out.printf("%s ", tops.get(i));
}
System.out.printf("\n");
return 0;
}
public static void main(String[] args) {
String[] vexs = { "F1", "F2", "F3", "T1", "F4", "C1", "F6", "C2", "T2", "F7", "S1" };
String[][] edges = new String[][] { { "F1", "T1" }, { "T1", "F4" }, { "F4", "C1" }, { "C1", "S1" },
{ "F2", "C2" }, { "F3", "C2" }, { "C2", "C1" }, { "F6", "T2" }, { "T2", "F7" }, { "F7", "S1" } };
ListDG pG;
// 自定义"图"(输入矩阵队列)
// pG = new ListDG();
// 采用已有的"图"
pG = new ListDG(vexs, edges);
pG.print(); // 打印图
//pG.DFS(); // 深度优先遍历
// pG.BFS(); // 广度优先遍历
pG.topologicalSort(); // 拓扑排序
}
}