Calculating total compressed size of a parquet column

parquet_column_size.py

import pyarrow.parquet

def calculate_column_size(column, file):
    m = pyarrow.parquet.read_metadata(file)
    sz = sum([c["total_compressed_size"] for x in m.to_dict()["row_groups"] for c in x["columns"] if c["path_in_schema"].startswith(column)])
    totalsize = sum([c["total_compressed_size"] for x in m.to_dict()["row_groups"] for c in x["columns"]])

    return sz, sz/totalsize