dataset-stats

#!/usr/bin/env python

import argparse
from dsf.data import lemnatec
from datetime import datetime


def options():
    parser = argparse.ArgumentParser(description='Output dataset statistics.',
                                     formatter_class=argparse.ArgumentDefaultsHelpFormatter)
    parser.add_argument("-d", "--dataset", help="Dataset directory.", required=True)
    args = parser.parse_args()

    return args


def main():
    # Read user options
    args = options()

    # Load the dataset metadata
    meta = lemnatec.load_dataset(dataset_dir=args.dataset)

    # Number of snapshots
    snapshot_count = len(meta["environment"])

    # Number of images
    image_count = len(meta["images"])

    # Find the experiment start and end dates
    dates = []
    for snapshot in meta["environment"]:
        timestamp = meta["environment"][snapshot]["timestamp"]
        dates.append(datetime.strptime(timestamp, "%Y-%m-%dT%H:%M:%S.%fZ"))
    dates.sort()
    start_date = dates[0].date().strftime("%Y-%m-%d")
    end_date = dates[-1].date().strftime("%Y-%m-%d")

    print("start_date end_date snapshots images")
    print(" ".join(map(str, [start_date, end_date, snapshot_count, image_count])))


if __name__ == "__main__":
    main()