Just added a

levi.skipped_stats(delta_table, filters=[('a_float', '=', 4.5)])

function that returns this

{'num_files': 2, 'num_files_skipped': 1, 'num_bytes_skipped': 996}

. This allows users to see how many files / how much data gets skipped for different predicates. It’ll help them figure out when they should Z ORDER, etc. The new

get_add_actions

API is opening up all sorts of new possibilities to get insights on Delta tables. Here’s the code if you’re interested.