我应该如何在Java中表示依赖图?

时间:2011-04-09 10:31:32

标签: java object graph

我正在研究JavaScript依赖管理的一些代码,我发现有人已经解决了Java中的依赖图问题。

我的第一次尝试是在我的JSResource对象上实现可比较,但是当有多个叶子节点没有依赖关系,因此没有合理的顺序,除非受到他们的家属的影响,它就会失败。

所以我认为我需要一个图表然后一种迭代图表的方法。这不是一个不可能的问题,但我想我会在重新发明轮子之前发布这里。

干杯, 皮特

3 个答案:

答案 0 :(得分:6)

我发布的内容可能会回答您的问题。链接在这里: http://nicolaecaralicea.blogspot.com/2010/11/dependency-graphs-generic-approach-in.html

以下是代码:



    package org.madeforall.graph.test;

import java.util.ArrayList;
import java.util.List;

import org.madeforall.graph.Graph;
import org.madeforall.graph.NodeValueListener;

public class TestDependecyGraph {
    public static void main(String[] args) {
        testWithGenericInt();
        testWithGenericString();
    }

    public static void testWithGenericInt() {
        final List<Integer> nodeValueList = new ArrayList<Integer>();
        Graph<Integer> graph = new Graph<Integer>(new NodeValueListener<Integer>() {
            public void evaluating(Integer nodeValue) {
                nodeValueList.add(nodeValue);
            }
        });
        graph.addDependency(1, 2);
        graph.addDependency(1, 3);
        graph.addDependency(3, 4);
        graph.addDependency(3, 5);
        graph.addDependency(5, 8);
        graph.addDependency(2, 7);
        graph.addDependency(2, 9);
        graph.addDependency(2, 8);
        graph.addDependency(9, 10);
        graph.generateDependencies();

        System.out.println(nodeValueList);

    }

    public static void testWithGenericString() {
        final List<String> nodeValueList = new ArrayList<String>();
        Graph<String> graph = new Graph<String>(new NodeValueListener<String>() {
            public void evaluating(String nodeValue) {
                nodeValueList.add(nodeValue);
            }
        });
        graph.addDependency("a", "b");
        graph.addDependency("a", "c");
        graph.addDependency("a", "f");
        graph.addDependency("c", "d");
        graph.addDependency("d", "g");
        graph.addDependency("f", "d");
        graph.addDependency("h", "e");
        graph.generateDependencies();
        System.out.println(nodeValueList);

    }
}

The first and the second argument of the addDependency method of the Graph class are some two arbitrarily chosen nodes of the oriented graph. Watch out for circular dependencies, because I did not take care of them yet.


Here are the classes. 

package org.madeforall.graph;

import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Set;

/**
 *
 * Represents a graph of nodes. Every node is of GraphNode type and it has set a
 * value of the generic type <T>. It basically derives an evaluation order out
 * of its nodes. A node gets the chance to be evaluated when all the incoming
 * nodes were previously evaluated. The evaluating method of the
 * NodeValueListener is used to notify the outside of the fact that a node just
 * got the chance to be evaluated. A value of the node that is of the generic
 * type <T> is passed as argument to the evaluating method.
 *
 *
 * @author nicolae caralicea
 *
 * @param <T>
 */
final public class Graph<T> {
    /**
     * These are basically the nodes of the graph
     */
    private HashMap<T, GraphNode<T>> nodes = new HashMap<T, GraphNode<T>>();
    /**
     * The callback interface used to notify of the fact that a node just got
     * the evaluation
     */
    private NodeValueListener<T> listener;
    /**
     * It holds a list of the already evaluated nodes
     */
    private List<GraphNode<T>> evaluatedNodes = new ArrayList<GraphNode<T>>();

    /**
     * The main constructor that has one parameter representing the callback
     * mechanism used by this class to notify when a node gets the evaluation.
     *
     * @param listener
     *            The callback interface implemented by the user classes
     */
    public Graph(NodeValueListener<T> listener) {
        this.listener = listener;
    }

    /**
     * Allows adding of new dependicies to the graph. "evalFirstValue" needs to
     * be evaluated before "evalAfterValue"
     *
     * @param evalFirstValue
     *            The parameter that needs to be evaluated first
     * @param evalAfterValue
     *            The parameter that needs to be evaluated after
     */
    public void addDependency(T evalFirstValue, T evalAfterValue) {
        GraphNode<T> firstNode = null;
        GraphNode<T> afterNode = null;
        if (nodes.containsKey(evalFirstValue)) {
            firstNode = nodes.get(evalFirstValue);
        } else {
            firstNode = createNode(evalFirstValue);
            nodes.put(evalFirstValue, firstNode);
        }
        if (nodes.containsKey(evalAfterValue)) {
            afterNode = nodes.get(evalAfterValue);
        } else {
            afterNode = createNode(evalAfterValue);
            nodes.put(evalAfterValue, afterNode);
        }
        firstNode.addGoingOutNode(afterNode);
        afterNode.addComingInNode(firstNode);
    }

    /**
     * Creates a graph node of the <T> generic type
     *
     * @param value
     *            The value that is hosted by the node
     * @return a generic GraphNode object
     */
    private GraphNode<T> createNode(T value) {
        GraphNode<T> node = new GraphNode<T>();
        node.value = value;
        return node;
    }

    /**
     *
     * Takes all the nodes and calculates the dependency order for them.
     *
     */
    public void generateDependencies() {
        List<GraphNode<T>> orphanNodes = getOrphanNodes();
        List<GraphNode<T>> nextNodesToDisplay = new ArrayList<GraphNode<T>>();
        for (GraphNode<T> node : orphanNodes) {
            listener.evaluating(node.value);
            evaluatedNodes.add(node);
            nextNodesToDisplay.addAll(node.getGoingOutNodes());
        }
        generateDependencies(nextNodesToDisplay);
    }

    /**
     * Generates the dependency order of the nodes passed in as parameter
     *
     * @param nodes
     *            The nodes for which the dependency order order is executed
     */
    private void generateDependencies(List<GraphNode<T>> nodes) {
        List<GraphNode<T>> nextNodesToDisplay = null;
        for (GraphNode<T> node : nodes) {
            if (!isAlreadyEvaluated(node)) {
                List<GraphNode<T>> comingInNodes = node.getComingInNodes();
                if (areAlreadyEvaluated(comingInNodes)) {
                    listener.evaluating(node.value);
                    evaluatedNodes.add(node);
                    List<GraphNode<T>> goingOutNodes = node.getGoingOutNodes();
                    if (goingOutNodes != null) {
                        if (nextNodesToDisplay == null)
                            nextNodesToDisplay = new ArrayList<GraphNode<T>>();
                        // add these too, so they get a chance to be displayed
                        // as well
                        nextNodesToDisplay.addAll(goingOutNodes);
                    }
                } else {
                    if (nextNodesToDisplay == null)
                        nextNodesToDisplay = new ArrayList<GraphNode<T>>();
                    // the checked node should be carried
                    nextNodesToDisplay.add(node);
                }
            }
        }
        if (nextNodesToDisplay != null) {
            generateDependencies(nextNodesToDisplay);
        }
        // here the recursive call ends
    }

    /**
     * Checks to see if the passed in node was aready evaluated A node defined
     * as already evaluated means that its incoming nodes were already evaluated
     * as well
     *
     * @param node
     *            The Node to be checked
     * @return The return value represents the node evaluation status
     */
    private boolean isAlreadyEvaluated(GraphNode<T> node) {
        return evaluatedNodes.contains(node);
    }

    /**
     * Check to see if all the passed nodes were already evaluated. This could
     * be thought as an and logic between every node evaluation status
     *
     * @param nodes
     *            The nodes to be checked
     * @return The return value represents the evaluation status for all the
     *         nodes
     */
    private boolean areAlreadyEvaluated(List<GraphNode<T>> nodes) {
        return evaluatedNodes.containsAll(nodes);
    }

    /**
     *
     * These nodes represent the starting nodes. They are firstly evaluated.
     * They have no incoming nodes. The order they are evaluated does not
     * matter.
     *
     * @return It returns a list of graph nodes
     */
    private List<GraphNode<T>> getOrphanNodes() {
        List<GraphNode<T>> orphanNodes = null;
        Set<T> keys = nodes.keySet();
        for (T key : keys) {
            GraphNode<T> node = nodes.get(key);
            if (node.getComingInNodes() == null) {
                if (orphanNodes == null)
                    orphanNodes = new ArrayList<GraphNode<T>>();
                orphanNodes.add(node);
            }
        }
        return orphanNodes;
    }
}

package org.madeforall.graph;

import java.util.ArrayList;
import java.util.List;

/**
 *
 * It represents the node of the graph. It holds a user value that is passed
 * back to the user when a node gets the chance to be evaluated.
 *
 * @author nicolae caralicea
 *
 * @param <T>
 */
final class GraphNode<T> {
    public T value;
    private List<GraphNode<T>> comingInNodes;
    private List<GraphNode<T>> goingOutNodes;

    /**
     * Adds an incoming node to the current node
     *
     * @param node
     *            The incoming node
     */
    public void addComingInNode(GraphNode<T> node) {
        if (comingInNodes == null)
            comingInNodes = new ArrayList<GraphNode<T>>();
        comingInNodes.add(node);
    }

    /**
     * Adds an outgoing node from the current node
     *
     * @param node
     *            The outgoing node
     */
    public void addGoingOutNode(GraphNode<T> node) {
        if (goingOutNodes == null)
            goingOutNodes = new ArrayList<GraphNode<T>>();
        goingOutNodes.add(node);
    }

    /**
     * Provides all the coming in nodes
     *
     * @return The coming in nodes
     */
    public List<GraphNode<T>> getComingInNodes() {
        return comingInNodes;
    }

    /**
     * Provides all the going out nodes
     *
     * @return The going out nodes
     */
    public List<GraphNode<T>> getGoingOutNodes() {
        return goingOutNodes;
    }
}


package org.madeforall.graph;

/**
 * The main mechanism used for notifying the outside of the fact that a node
 * just got its evaluation
 *
 * @author nicolae caralicea
 *
 * @param <T>
 */
public interface NodeValueListener<T> {
    /**
     *
     * The callback method used to notify the fact that a node that has assigned
     * the nodeValue value just got its evaluation
     *
     * @param nodeValue
     *            The user set value of the node that just got the evaluation
     */
    void evaluating(T nodeValue);
}

答案 1 :(得分:1)

讨论了java图形处理库。也许它会帮助你。

Good Java graph algorithm library?

答案 2 :(得分:0)

多次将任务表示为依赖图是不够的,您可能希望并行执行任务(表示为节点)以满足依赖条件,并执行一些内务管理工作,例如检查周期等。为此目的是你可以帮助this framework,这是一个非常轻量级的框架,以可靠的方式运行相关的任务。这是一个测试用例,为您提供一个简短的想法:

@Test
public void testDependentTaskExecution() {

    DefaultDependentTasksExecutor<Integer> executor = newTaskExecutor();

    executor.addDependency(1, 2);
    executor.addDependency(1, 3);
    executor.addDependency(3, 4);
    executor.addDependency(3, 5);
    executor.addDependency(3, 6);
    //executor.addDependency(10, 2); // cycle
    executor.addDependency(2, 7);
    executor.addDependency(2, 9);
    executor.addDependency(2, 8);
    executor.addDependency(9, 10);
    executor.addDependency(12, 13);
    executor.addDependency(13, 4);
    executor.addDependency(13, 14);
    executor.addIndependent(11);

    executor.execute();
}

private DefaultDependentTasksExecutor<Integer> newTaskExecutor() {
    return new DefaultDependentTasksExecutor<Integer>(newExecutor(), new SleepyTaskProvider<Integer>());
}

private ExecutorService newExecutor() {
    return Executors.newCachedThreadPool();
}

private static class SleepyTaskProvider<T> implements TaskProvider<T> {

    public Task provid(T id) {

        return new Task() {

            public void execute() {
                try {
                    Thread.sleep(500);
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
            }
        };
    }       
}

这是控制台输出,让您了解幕后的内容

19:57:43.705 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 1 node
19:57:43.710 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 11 node
19:57:43.710 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 12 node
19:57:44.212 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 1 done
19:57:44.213 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 2 node
19:57:44.214 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 3 node
19:57:44.215 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 11 done
19:57:44.216 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 12 done
19:57:44.217 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 13 node
19:57:44.717 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 2 done
19:57:44.718 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 7 node
19:57:44.719 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 9 node
19:57:44.720 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 8 node
19:57:44.721 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 3 done
19:57:44.722 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 83 - node 4 depends on [3, 13]
19:57:44.724 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 5 node
19:57:44.726 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 6 node
19:57:44.728 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 13 done
19:57:44.729 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 4 node
19:57:44.730 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 14 node
19:57:45.219 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 7 done
19:57:45.221 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 9 done
19:57:45.223 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doExecute 80 - Going to schedule 10 node
19:57:45.225 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 8 done
19:57:45.227 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 5 done
19:57:45.228 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 6 done
19:57:45.234 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 4 done
19:57:45.235 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 14 done
19:57:45.725 [main] DEBUG c.n.e.DefaultDependentTasksExecutor.doWaitForExecution 105 - Processing of node 10 done
19:57:45.726 [main] INFO  c.n.e.DefaultDependentTasksExecutor.execute 69 - Total Time taken to process 14 jobs each taking 500 ms is 2022 ms instead of 7000 ms
19:57:45.728 [main] INFO  c.n.e.DefaultDependentTasksExecutor.execute 70 - [1, 11, 12, 2, 3, 13, 7, 9, 8, 5, 6, 4, 14, 10]