Sanitize molecules when SMILES needs to be produced in PandasTools

[mwojcikowski]

When loading SDF to a Pandas dataframe and discarding of molecules and retaining SMILES, the SDMolSupplier was not sanitizing the molecules before producing of SMILES. This tiny PR addresses this issue.

Additional change to improve the performance by getting property dict in one call instead of a loop over all properties.

[greglandrum]

@mwojcikowski : the CI test failures here seem real.
It looks like you need to take a look at the tests for the PandasTools module and update the expected results.

[mwojcikowski]

@greglandrum Yes, I've seen them. I simply didn't manage to address them yesterday yet. The issue is that mol.GetPropsAsDict() returns typed properties instead of strings. I will handle that and ping you as soon as this is sorted out.

[mwojcikowski]

@greglandrum I removed the changes improving performance, and left only the bugfix part and now tests pass.

[greglandrum]

@greglandrum I removed the changes improving performance, and left only the bugfix part and now tests pass.

Thanks.
I'm not convinced the use of GetPropsAsDict() is a good idea here anyway: there's no guarantee that the given property will have the same type for every molecule and I'm not sure if pandas will do the right thing.
It would be nice (though not 100% backwards compatible) to have real types there when it can work though... maybe something for the future.

[mwojcikowski]

Indeed, I'd expect the sanitization to be the bottleneck anyways.

[greglandrum]

Indeed, I'd expect the sanitization to be the bottleneck anyways.

most definitely