src/queries/pfr/league_schedule.py

import pandas as pd


def get_full_schedule(season_int: str | int) -> pd.DataFrame:
    url = f"https://www.pro-football-reference.com/years/{season_int}/games.htm#games"
    df = pd.read_html(url)[0]

    # remove extra header rows in table
    df = df[df.Week != "Week"]
    return df


def get_week_team_time_map(df_schedule: pd.DataFrame) -> dict[int, dict[str, pd.Timestamp]]:
    min_week = 1
    max_week = 23
    week_team_time_map: dict[int, dict[str, pd.Timestamp]] = {k: {} for k in range(min_week, max_week + 1)}
    for _, row in df_schedule.iterrows():
        game_time = pd.to_datetime(row.Date + " " + row.Time, yearfirst=True)
        week_team_time_map[int(row.Week)][row["Winner/tie"]] = game_time
        week_team_time_map[int(row.Week)][row["Loser/tie"]] = game_time
    return week_team_time_map


def get_season_time_map(season_int: str | int) -> dict[int, dict[str, pd.Timestamp]]:
    df_schedule = get_full_schedule(season_int)
    return get_week_team_time_map(df_schedule)