About NaN and NoneType

[njmmatthieu]

OntoWeaver adapters tranform NaN into false (type boolean). So if we want to pass a column of type String for a property, the property will not appear for the corresponding nodes where there is NaN in the Graph DataBase but will appear for the other nodes where is not NaN.

In the same way, OntoWeaver adapters transform NoneType into String "'None'".

Perhaps it would be better if the properties appeared for all nodes with ‘NaN’ or ‘None’?

[mbaric758]

Hi @njmmatthieu , thanks for posting the issue.

Do you mean it would be more optimal if empty and NoneType / NaN values encountered in tables were still mapped with the properties added, or that none of these values should appear in the graph? Or the choice of inclusion / exclusion of these values should be left to the user?

Currently, the checking of the validity of extracted values is handled by the base.valid() function, and all NaN values and empty strings are returned as false and therefore not transformed into nodes. We should perhaps add the NoneType for validity checking, to avid the creation of nodes with meaningless values (which would then also solve the bug of the property addition to meaningless nodes)?