Make rel_dfs of from_dfs optional

[adamnsch]

Thank you for your contribution to the Graph Visualization for Python project by Neo4j.

Before submitting this PR, please read Contributing to the Neo4j Ecosystem.

Make sure:

• You signed the Neo4j CLA (Contributor License Agreement) so that we are allowed to ship your code in our library
• Your contribution is covered by tests

Based on #189

[FlorentinD]

NIT could move the dropna logic into the k,v loop below (line 83).

[adamnsch]

Not sure I understand :) You mean we check each value if it's NaN? If so this feels a bit cleaner I think

[FlorentinD]

yes :)

with this you mean the row.dropna?

[adamnsch]

Yes, I think only making one call to row.dropna feels nicer than checking every value of the row

[FlorentinD]

Great improvement :)

We could also allow node_dfs to have the shape of dict[str, DataFrame].
This might make the drop_na option not as relevant? (gds could pass a df per label)

[adamnsch]

Great improvement :)

We could also allow node_dfs to have the shape of dict[str, DataFrame]. This might make the drop_na option not as relevant? (gds could pass a df per label)

I don't think that matters, since GDS can already pass a list of dataframes, one for each label. We could also remove dropna from the API, as long as we have in the internal _from_dfs that's enough for us