src/queries/nflverse/github_data.py

import pandas as pd


def get_parquet_github(season_int: int, parquet_prefix: str):
    location = f"https://github.com/nflverse/nflverse-data/releases/download/{parquet_prefix}_{season_int}.parquet"
    df = pd.read_parquet(location)
    return df


FANTASY_POSITIONS = [
    "QB",
    "RB",
    "WR",
    "TE",
    "FB",
    "K",
]


def get_snap_counts(season_int: int) -> pd.DataFrame:
    df = get_parquet_github(season_int, "snap_counts/snap_counts")
    df["fantasy_position"] = df["position"].isin(FANTASY_POSITIONS)
    return df


def get_play_by_play(season_int: int) -> pd.DataFrame:
    return get_parquet_github(season_int, "pbp/play_by_play")


def get_player_stats(season_int: int) -> pd.DataFrame:
    return get_parquet_github(season_int, "player_stats/player_stats")


def get_ftn_charting(season_int: int) -> pd.DataFrame:
    return get_parquet_github(season_int, "ftn_charting/ftn_charting")


def get_pbp_participation(season_int: int) -> pd.DataFrame:
    return get_parquet_github(season_int, "pbp_participation/pbp_participation")