run.py

import pandas as pd
from algo import Algo
from db.db_utils import get_connection

if __name__ == "__main__":
    db_conn = get_connection()
    algo = Algo(db_conn, True)
    raw_file_name = 'MFB-2023-raw-data.csv'
    input_file_path = f'./raw/{raw_file_name}'
    df_input = pd.read_csv(input_file_path)

    input_data = [(desc, donor, date, weight) for desc, donor, date, weight in zip(df_input['description'].astype(str).tolist(), df_input['donor'].astype(str).tolist(), df_input['date'].astype(str).tolist(), df_input['weight'].astype(str).tolist())]

    algo.match_words(input_data)
    # algo.match_words([['bananas']])

    db_conn.close()