Rendering a tree in python using anytree and graphviz, without merging common nodes

Question

I am creating a tree from a list ["abc", "abd", "aec", "add", "adcf"] using anytree package of python3. In this tree first character of each list element - a is a root, and subsequently, other characters are added as their children. When I render tree it looks like:

a
├── b
│   ├── c
│   └── d
├── e
│   └── c
└── d
    ├── d
    └── c
        └── f

But when I render the tree to picture using to_picture method, the image is -

I don't want the common nodes to be merged, as it is adding unwanted paths to my tree.

Answer 1

The node are arranged in graphviz using their id. In your case the graph is generated with just node names and then graphviz created the loop edge as you get it.

What you really want is different id for each node and a label associated with the same. The DotExporter class has a attribute named nodeattrfunc to which we can pass a function or a lambda and generate attributes for the nodes

Below is how you would do it based on your array

import anytree
from anytree import Node, RenderTree

data = ["abc", "abd", "aec", "add", "adcf"]
from anytree.exporter import DotExporter

nodes = {}
first_node = None

for elem in data:
    parent_node = None
    parent_node_name = ""
    for i, val in enumerate(elem):
        if i not in nodes:
            nodes[i] = {}
        key = parent_node_name + val
        if key not in nodes[i]:
            print("Creating new node for ", key)
            nodes[i][key] = Node(key, parent=parent_node, display_name=val)

        if first_node is None:
            first_node = nodes[i][key]
        parent_node = nodes[i][key]
        parent_node_name = val

print(nodes)
DotExporter(nodes[0]["a"],
            nodeattrfunc=lambda node: 'label="{}"'.format(node.display_name)).to_dotfile("graph.txt")
DotExporter(nodes[0]["a"],
            nodeattrfunc=lambda node: 'label="{}"'.format(node.display_name)).to_picture("graph.png")

This will generate the below dot file

digraph tree {
    "a" [label="a"];
    "ab" [label="b"];
    "bc" [label="c"];
    "bd" [label="d"];
    "ae" [label="e"];
    "ec" [label="c"];
    "ad" [label="d"];
    "dd" [label="d"];
    "dc" [label="c"];
    "cf" [label="f"];
    "a" -> "ab";
    "a" -> "ae";
    "a" -> "ad";
    "ab" -> "bc";
    "ab" -> "bd";
    "ae" -> "ec";
    "ad" -> "dd";
    "ad" -> "dc";
    "dc" -> "cf";
}

And the below graph

Answer 2

As anytree to_picture() function use graphviz to generate the file.

From here = similar question, you can read :

graphviz uses the node id as label. If distinct nodes need to have the same label, the label has to be defined explicitly.

Answer 3

I was about to implement the recommended fix but found that using UniqueDotExporter dealt with the repeated ids perfectly!