How efficiently processing the result from str.split?

[yohplala]

Hi,

This question is for my own information, i.e. it is a 'general' question.

I was recently answered in #1906 by Jovan that following approach to process results from str.split in vaex is not efficient (use of apply ).

max_date = vdf_max_date["fnames"].str.split('.').str.split('-',1).apply(lambda x: np.datetime64(x[0][1])).max().item()

(just for being complete, this methodology comes from this SO answer)

Meanwhile, I also stumbled on this Medium article, showcasing speed of vaex for operations on string, including split.

So, the underlying thought I have is that, if vaex shows amazing speed for split, surely, the way of efficiently processing results from split has also been thought about.
Hence my question, what kind of processings ('efficient' ones) can be done after a split?

split produces an expression which items are lists.
My need would be a way to select an item in these lists.
This might not be the 'usual' processing done after a split. If this is the case indeed, please, what processing after a split can be managed efficiently?

Thanks in advance for your feedbacks!
Bests

[maartenbreddels]

I think the only thing we currently support is joining back, note that you can operate on a list of string as if it is a flattened column of strings. so you can do e.g. split -> strip -> join