Visualization graph is empty in certain cases

[GCHQDeveloper314]

Describe the bug
If relationships are specified where the source or target is missing from the nodes DataFrame, no error is raised or printed, instead the visualisation produced is entirely empty.

Edit: This also happens when creating Relationship objects directly, as described here. This is not specific to using DataFrames.

To Reproduce

from pandas import DataFrame
from neo4j_viz.pandas import from_dfs

nodes = DataFrame({
    "id": ["a", "b", "c"],
    "caption": ["Alice", "Bob", "Charlie"],
    "size": [20, 10, 10],
})

relationships = DataFrame({
    "source": ["a", "b", "d"], # The node 'd' does not exist
    "target": ["b", "c", "e"], # The node 'e' does not exist
    "caption": ["KNOWS", "KNOWS", "KNOWS"],
})

VG = from_dfs(nodes, relationships)

VG.render()

graph-viz library version: 1.0.0
Python version: 3.12.9
Operating system: Ubuntu 22.04.5 LTS

Steps to reproduce the behaviour:
Create a new notebook with a cell containing the above code. After running it the visualisation canvas is empty. I cannot see any error messages or warnings in the notebook.

Expected behaviour
Either an exception is raised explaining that a row in relationships references nodes which are missing, or a warning printed and any rows with missing node references ignored. The latter would be preferable as it would make using samples from a DataFrame easier.

[FlorentinD]

Thanks for raising the issue. We will have a look into providing a better user-experience. I like the idea of throwing by default and having an opt-in parameter to ignore dangling relationships.