Question

我有以下点文件：

digraph finite_state_machine {
    {
        rank=same;
        node [shape = doublecircle]; q_5;
        node [shape = circle];
        q_1 -> q_2 [ label = "." ];
        q_1 -> q_2 [ label = "\epsilon" ];
        q_2 -> q_1 [ label = "\epsilon" ];
        q_2 -> q_3 [ label = "a" ];
        q_3 -> q_4 [ label = "^\wedge a" ];
        q_3 -> q_4 [ label = "\epsilon" ];
        q_4 -> q_3 [ label = "\epsilon" ];
        q_4 -> q_5 [ label = "b" ];
    }
}

我的印象是节点会按照提到的顺序出现。但顺序对我来说似乎是随机的。我希望订单从左到右依次为下标。此外，边缘穿过标签，我该如何避免？

这是当前图片： current output

Answer 1

Graphviz会根据彼此之间的边缘布置节点，而不是按照外观的顺序排列。如果您希望边缘不影响任何节点的位置，可以通过添加constraint=false来完成此操作。

因此，这张图：

digraph finite_state_machine {
  pad=0.2;
    {
        rank=same;
        node [shape = doublecircle]; q_5;
        node [shape = circle];
        q_1 -> q_2 [ label = "." ];
        q_1 -> q_2 [ label = "\epsilon", constraint=false ];
        q_2 -> q_1 [ label = "\epsilon", constraint=false ];
        q_2 -> q_3 [ label = "a" ];
        q_3 -> q_4 [ label = "^\wedge a" ];
        q_3 -> q_4 [ label = "\epsilon", constraint=false ];
        q_4 -> q_3 [ label = "\epsilon", constraint=false ];
        q_4 -> q_5 [ label = "b" ];
    }
}

会得到你：

graphviz output finite state machine

我必须添加pad，以便不会切断某些标签。

Answer 2

这是其中一个想法。

1）使用以下命令查找放置节点时点所使用的位置。

dot filename.dot

2）注意生成的输出中每个节点的 pos 属性。然后根据您的代码中的喜好更改其值。以下是您的示例的修改点代码。

digraph finite_state_machine {
{
    rank=same;
    node [shape = circle];
    q_1[pos="30"];
    q_2[pos="130"];
    q_3[pos="230"];
    q_4[pos="330"];
    node [shape = doublecircle]; 
    q_5 [pos="430"];
    q_1 -> q_2 [ label = "." ];
    q_1 -> q_2 [ label = "\epsilon" ];
    q_2 -> q_1 [ label = "\epsilon" ];
    q_2 -> q_3 [ label = "a" ];
    q_3 -> q_4 [ label = "^\wedge a" ];
    q_3 -> q_4 [ label = "\epsilon" ];
    q_4 -> q_3 [ label = "\epsilon" ];
    q_4 -> q_5 [ label = "b" ];
}

}

3）再次运行点，这次生成输出文件。

结果可以在这里看到：

Graphviz output http://i46.tinypic.com/2duxz88.png

Graphviz ---随机节点顺序和通过标签的边

2 个答案: