Is this available yet or going to be made available.
One use case is where the date/timestamp of the file is in the filename, but there is no record within the file. (e.g. a structure used by AACT . Logically we want to hold entities in a single folder and be able to use the filename to add a date column to allow time-series analaysis.
It looks like the files from AACT can be pipe delimited or tab separated. If that’s accurate and if you are able to “partition” the files into separate directories, you should be able to see the directory names (as dir0, dir1… etc) when you promote/format the upper level directory. Like so:
This does not work in all cases (like parquet or Iceberg datasets, i.e. post “unlimited splits”) and YMMV. But the method I posted above will work for all datasets.