Nested Cluster with neato layout

[bandoos]

Hi!

Thanks for the lib!

I'm encountering an issue with nested clusters while using the neato layout to manually specify position.

With default layout nesting works as charm:

from diagrams import Diagram, Cluster
from diagrams.onprem.container import Docker

with Diagram('Working cluster nesting'):
    with Cluster("A"):
        for b in range(3):
            with Cluster("B-"+str(b)):
                Docker('Docker-'+str(b))

When using neato:

from diagrams import Diagram, Cluster
from diagrams.onprem.container import Docker

graph_attr = {
    "layout":"neato",
    }

with Diagram('Borken cluster nesting 2',
             graph_attr=graph_attr):
    with Cluster("A"):
        Docker('Docker-A',pin="true",
               pos="0,-2")
        for b in range(3):
            with Cluster("B-"+str(b)):
                Docker('Docker-'+str(b),
                       pin="true",
                       pos="0,"+str(b*2))

Am I doing something wrong or is it not possible to do such nesting with neato?

Thanks in advace!

[clayms]

You don't need to nest the clusters with neato, because you can force them to be drawn where ever you want by using the coordinates.

For more fine grained control over the size and position of each Cluster, use some blank "plaintext" Nodes to pin the opposite corners.

from diagrams import Diagram, Cluster, Node
from diagrams.onprem.container import Docker

graph_attr = {
    "layout":"neato",
    }

clusB_graph_attr = {
    "bgcolor": "azure",
}

with Diagram('Neato "nested" Clusters', show=False,
            graph_attr=graph_attr) as diag:
    with Cluster("A"):
        Node('', shape="plaintext", pin="true", pos="-1,-0.5")
        Node('', shape="plaintext", pin="true", pos="1,5")
        Docker('Docker-A',pin="true",
               pos="0,-2")
    for b in range(3):
        with Cluster("B-"+str(b), graph_attr=clusB_graph_attr, ):
            Docker('Docker-'+str(b),
                    pin="true",
                    pos="0,"+str(b*2))

diag

[clayms]

slight variation to use neato's node separation algorithms.

This can be faster, as you will not have to specify each coordinate. However, it can get more tricky to get the final plot laid out the way you want.

from diagrams import Diagram, Cluster, Node
from diagrams.onprem.container import Docker

graph_attr = {
    "layout": "neato",
    "mode": "major",
    "overlap": "voronoi",
    "sep": "0.0", 
    "splines": "spline",
}

clusB_graph_attr = {
    "bgcolor": "azure",
}

with Diagram('Neato "nested" Clusters',show=False,
            graph_attr=graph_attr) as diag:
    with Cluster("A"):
        Node('', shape="plaintext", width="0.0", height="0.0", pin="true", pos="0,0")
        Node('', shape="plaintext", width="0.0", height="0.0", pin="true", pos="0,5")
        A1 = Docker('Docker-A', )
    for b in range(3):
        with Cluster(f"B-{str(b)}", graph_attr=clusB_graph_attr, ):
            exec(f"B{str(b)} = Docker('Docker-{str(b)}', )")

    (
        A1 - Edge(penwidth="0.0") - 
        B0 - Edge(penwidth="0.0") - 
        B1 - Edge(penwidth="0.0") - 
        B2 
    )

diag