From ce20587591620213dd768171cbbbe6f3c8324064 Mon Sep 17 00:00:00 2001 From: Erick Ahmed Date: Thu, 9 Jul 2026 23:29:25 +0200 Subject: [PATCH] Move to /src --- parser.py => src/parser.py | 16 ++-------------- 1 file changed, 2 insertions(+), 14 deletions(-) rename parser.py => src/parser.py (88%) diff --git a/parser.py b/src/parser.py similarity index 88% rename from parser.py rename to src/parser.py index aab46a3..a4e76d9 100644 --- a/parser.py +++ b/src/parser.py @@ -88,18 +88,6 @@ def parse_csv(csv_path: PathLike) -> pl.LazyFrame: pl.col("Timestamp").cast(pl.Float64) ]) -def parse_parquet(parquet_path: PathLike) -> pl.LazyFrame: - """ - Loads a pre-optimized Parquet file directly into a Polars LazyFrame. - """ - return pl.scan_parquet(parquet_path) - -def save_parquet(lf: pl.LazyFrame, output_path: PathLike) -> None: - """ - Materializes the LazyFrame query plan and saves it to a Parquet file. - """ - lf.sink_parquet(output_path) - if __name__ == '__main__': import argparse @@ -119,10 +107,10 @@ if __name__ == '__main__': if args.command == "csv": parse_log(args.input, args.out_bus1, args.out_bus2) - print(f"[+] Successfully csved logs to {args.out_bus1} and {args.out_bus2}") + print(f"[+] Saved csv file to {args.out_bus1} and {args.out_bus2}") elif args.command == "parquet": print(f"[*] Processing {args.input_csv}...") lf = parse_csv(args.input_csv) save_parquet(lf, args.output_parquet) - print(f"[+] Successfully optimized and saved to {args.output_parquet}") + print(f"[+] Saved parquet file to {args.output_parquet}")