#!/usr/bin/env python3

import argparse
from pathlib import Path
from pprint import pprint

import utils

argparser = argparse.ArgumentParser(description="OCR-Logfile evaluation")
argparser.add_argument("vp_dir", help="VP Directory")

args = argparser.parse_args()

data_path = Path(args.vp_dir)
video_path = next(data_path.glob("*.mkv"))
ocr_path = data_path / "analysis_results.csv"
log_path = data_path / f"{data_path.stem}.csv"

df = utils.combine_ocr_logs(video_path, ocr_path, log_path)
df = df.fillna('')

df = utils.calc_levenshtein_distance(df)


url_groups = utils.group_urls(list(df["url"].values))
pprint(len(url_groups))

df.to_csv(f"{data_path}/metrics.csv")
utils.write_grouped_metrics(df, url_groups, data_path)