from sys import argv import polars as pl import polars_hash as pl_hash file = argv[1] pl.scan_parquet(file).with_columns( pl.col("user_id").cast(pl.String).add("Lt2N5xgjJOqRsT5qVt7wWYw6SqOPZDI7") ).with_columns( pl_hash.col("user_id").chash.sha2_256().str.head(16) ).collect().write_parquet( file.replace(".parquet", "_pub.parquet"), compression="zstd", compression_level=15 )