在SqlG中执行广度优先搜索以获取顶点与边信息

我正在使用SqlG（Apache TinkerPop针对PostgreSQL的扩展），希望从初始顶点执行广度优先搜索，获取所有顶点及对应的边信息。目前运行的代码仅能获取顶点，无法得到边信息。

创建图的代码 #

Map<String, String> conf = new HashMap<>();
conf.put("jdbc.url", "jdbc:postgresql://localhost:5432/marko");
conf.put("jdbc.username", "postgres");
conf.put("jdbc.password", "postgres");
Configuration configuration = new MapConfiguration(conf);
Graph graph = SqlgGraph.open(configuration);

//Creating a usual TinkerPop example Graph
graph.traversal().tx().open();
Vertex personMarko = graph.addVertex(T.label, "person", "name", "marko", "age", 29);
Vertex personJosh = graph.addVertex(T.label, "person", "name", "josh", "age", 32);
Vertex personVadas = graph.addVertex(T.label, "person", "name", "vadas", "age", 27);
Vertex personPeter = graph.addVertex(T.label, "person", "name", "peter", "age", 35);
Vertex softwareRipple = graph.addVertex(T.label, "software", "name", "ripple", "lang", "java");
Vertex softwareLop = graph.addVertex(T.label, "software", "name", "lop", "lang", "java");

personMarko.addEdge("knows", personJosh, "weight", 1.0);
personMarko.addEdge("knows", personVadas, "weight", 0.5);

personMarko.addEdge("created", softwareLop, "weight", 0.4);
personJosh.addEdge("created", softwareLop, "weight", 0.4);
personPeter.addEdge("created", softwareLop, "weight", 0.2);
personJosh.addEdge("created", softwareRipple, "weight", 1.0);
graph.tx().commit();

原遍历代码（无法获取边） #

Vertex start = graph.traversal().V().hasLabel("person").has("name", "marko").next();
graph.traversal().V(start).repeat(__.outE().inV()).emit().path().forEachRemaining(path -> {
    System.out.println("PATH DETAILS");
    for(Object object : path) {
        if(object instanceof Vertex) {
            Vertex vertex = (Vertex) object;
            String name = vertex.value("name") + "";
            String label = vertex.label();
            System.out.print(String.format("%s[%s] ", name, label));
        }
        if(object instanceof Edge) {
            //This block is never executed!!!
            System.out.println("EDGE FOUND");
            Edge edge = (Edge) object;
            String label = edge.label();
            System.out.print(String.format("%s ", label));
        }
    }
    System.out.println("\n");
});

解决方案 #

问题核心在于原遍历逻辑中，虽然路径经过了边，但元素判断逻辑或遍历写法需要调整，确保边元素被正确识别和捕获。以下是两种可行的修改方案：

方案一：保留完整路径并捕获边信息

调整遍历代码，添加simplePath()避免循环，同时优化元素判断逻辑：

Vertex start = graph.traversal().V().hasLabel("person").has("name", "marko").next();
graph.traversal().V(start)
    .repeat(__.outE().inV())
    .simplePath() // 防止循环遍历重复节点
    .emit()
    .path()
    .forEachRemaining(path -> {
        System.out.println("PATH DETAILS");
        for (Object object : path) {
            if (object instanceof Vertex) {
                Vertex vertex = (Vertex) object;
                String name = vertex.value("name").toString();
                String label = vertex.label();
                System.out.print(String.format("%s[%s] ", name, label));
            } else if (object instanceof Edge) {
                Edge edge = (Edge) object;
                String label = edge.label();
                double weight = edge.value("weight");
                System.out.print(String.format("---[%s, %.1f]--- ", label, weight));
            }
        }
        System.out.println("\n");
    });

方案二：结构化获取边与关联顶点

使用as和select显式记录每一步的边和顶点，适合需要单独处理边信息的场景：

Vertex start = graph.traversal().V().hasLabel("person").has("name", "marko").next();
graph.traversal().V(start)
    .repeat(__.outE().as("edge").inV().as("vertex"))
    .emit()
    .select("edge", "vertex")
    .forEachRemaining(map -> {
        Edge edge = map.get("edge");
        Vertex vertex = map.get("vertex");
        System.out.println("EDGE: " + edge.label() + " (weight: " + edge.value("weight") + ")");
        System.out.println("CONNECTED TO: " + vertex.value("name") + "[" + vertex.label() + "]\n");
    });

额外检查点 #

1. 确认SqlG版本为3.1.0，与对应的TinkerPop版本兼容
2. 确保事务已正确提交，边数据确实存在于图中

内容的提问来源于stack exchange，提问作者Juvenik