I would probably suggest that you partition by format (though you can get
the file name from the build in function input_file_name()). You can load
multiple streams from different directories and union them together as long
as the schema is the same after parsing. Otherwise you can just run
Hi,
I have different files being dumped on S3, I want to ingest them and join them.
What does sound better to you? Have one " directory" for all or one per file
format?
If I have one directory for all, can you get some metadata about the file, like
its name?
If multiple directory, how can I