src/stats.py

from dataclasses import dataclass
import json
import pandas as pd
import requests
import streamlit as st

from domain.constants import SEASON
from domain.playoffs import (
    SHORT_TEAM_NAMES_TO_DEFENSE_PLAYER_ID,
    DEFENSE_PLAYER_ID_TO_ROSTER_TEAM_NAMES,
    ROSTER_TEAM_NAMES_TO_DEFENSE_PLAYER_ID,
    PLAYOFF_TEAM_DEF_PLAYER,
)
from login import get_stat_overrides
from queries.nflverse.github_data import get_player_kicking_stats, get_player_stats, get_team_defense_stats
from queries.pfr.league_schedule import get_season_game_map

STAT_CACHE_SECONDS = 60


@dataclass
class StatType:
    key: str
    score: float

    def __post_init__(self):
        STAT_KEY_MAP[self.key] = self


STAT_KEY_MAP: dict[str, StatType] = {}

RUSH_TD = StatType(key="RUSH TD", score=6.0)
REC_TD = StatType(key="REC TD", score=6.0)
OFF_FUM_TD = StatType(key="OFF FUM TD", score=6.0)
PASS_TD = StatType(key="PASS TD", score=4.0)
FG_0_49 = StatType(key="FG 0-49", score=3.0)
FG_50_ = StatType(key="FG 50+", score=5.0)
TWO_PT = StatType(key="2 PT", score=2.0)
RECEPTION = StatType(key="REC", score=1.0)
RUSH_YD = StatType(key="RUSH YD", score=0.1)
REC_YD = StatType(key="REC YD", score=0.1)
PASS_YD = StatType(key="PASS YD", score=0.04)
XP = StatType(key="XP", score=1.0)
FUM_LOST = StatType(key="FUM LOST", score=-2.0)
PASS_INT = StatType(key="PASS INT", score=-2.0)
RET_TD = StatType(key="RET TD", score=6.0)
DEF_TD = StatType(key="DEF TD", score=6.0)
DEF_INT = StatType(key="DEF INT", score=2.0)
FUM_REC = StatType(key="FUM REC", score=2.0)
SAFETY = StatType(key="SAFETY", score=2.0)
SACK = StatType(key="SACK", score=1.0)
PTS_ALLOW_0 = StatType(key="PTS 0", score=10.0)
PTS_ALLOW_1_6 = StatType(key="PTS 1-6", score=7.0)
PTS_ALLOW_7_13 = StatType(key="PTS 7-13", score=4.0)
PTS_ALLOW_14_20 = StatType(key="PTS 14-20", score=1.0)
PTS_ALLOW_21_27 = StatType(key="PTS 21-27", score=0.0)
PTS_ALLOW_28_34 = StatType(key="PTS 28-34", score=-1.0)
PTS_ALLOW_35_ = StatType(key="PTS 35+", score=-4.0)
TEAM_WIN = StatType(key="TEAM WIN", score=5.0)
ST_TD = StatType(key="ST TD", score=6.0)


NFLVERSE_STAT_COL_TO_ID: dict[str, str] = {
    "passing_tds": PASS_TD.key,
    "passing_yards": PASS_YD.key,
    "passing_2pt_conversions": TWO_PT.key,
    "sack_fumbles_lost": FUM_LOST.key,
    "interceptions": PASS_INT.key,
    "rushing_tds": RUSH_TD.key,
    "rushing_yards": RUSH_YD.key,
    "rushing_2pt_conversions": TWO_PT.key,
    "rushing_fumbles_lost": FUM_LOST.key,
    "receptions": RECEPTION.key,
    "receiving_tds": REC_TD.key,
    "receiving_yards": REC_YD.key,
    "receiving_2pt_conversions": TWO_PT.key,
    "receiving_fumbles_lost": FUM_LOST.key,
    "special_teams_tds": ST_TD.key,
    "pat_made": XP.key,
    "fg_made_0_19": FG_0_49.key,
    "fg_made_20_29": FG_0_49.key,
    "fg_made_30_39": FG_0_49.key,
    "fg_made_40_49": FG_0_49.key,
    "fg_made_50_59": FG_50_.key,
    "fg_made_60_": FG_50_.key,
    "def_sacks": SACK.key,
    "def_interceptions": DEF_INT.key,
    "def_tds": DEF_TD.key,
    "def_fumble_recovery_opp": FUM_REC.key,
    "def_safety": SAFETY.key,
}

NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK = {
    19: 1,
    20: 2,
    21: 3,
    22: 4,
}


def add_stats_from_player_df_to_stat_map(df: pd.DataFrame, stat_map):
    df_playoffs = df[df.week.isin(NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK.keys())]
    df_playoffs.week = df_playoffs.week.apply(lambda x: NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK[x])
    for week_player_id_tuple, row in df_playoffs.set_index(["week", "player_id"]).iterrows():
        if isinstance(week_player_id_tuple, tuple):
            week, player_id = week_player_id_tuple
        else:
            # this won't happen but makes mypy happy
            continue
        player_stats: dict[str, float] = {}
        for k, v in row.to_dict().items():
            if k in NFLVERSE_STAT_COL_TO_ID:
                if (mapped_k := NFLVERSE_STAT_COL_TO_ID[k]) in player_stats:
                    player_stats[mapped_k] += v
                else:
                    player_stats[mapped_k] = v

        if player_id not in stat_map[week]:
            stat_map[week][player_id] = player_stats
        else:
            stat_map[week][player_id].update(player_stats)


def add_stats_from_team_def_df_to_stat_map(df: pd.DataFrame, stat_map):
    df_playoffs = df[
        (
            df.week.isin(NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK.keys())
            & df.team.isin(ROSTER_TEAM_NAMES_TO_DEFENSE_PLAYER_ID.keys())
        )
    ]
    df_playoffs.week = df_playoffs.week.apply(lambda x: NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK[x])

    for week_team_tuple, row in df_playoffs.set_index(["week", "team"]).iterrows():
        if isinstance(week_team_tuple, tuple):
            week, team = week_team_tuple
        else:
            # this won't happen but makes mypy happy
            continue
        player_stats: dict[str, float] = {}
        player_id = ROSTER_TEAM_NAMES_TO_DEFENSE_PLAYER_ID[team]
        for k, v in row.to_dict().items():
            if k in NFLVERSE_STAT_COL_TO_ID:
                if (mapped_k := NFLVERSE_STAT_COL_TO_ID[k]) in player_stats:
                    player_stats[mapped_k] += v
                else:
                    player_stats[mapped_k] = v

        if player_id not in stat_map[week]:
            stat_map[week][player_id] = player_stats
        else:
            stat_map[week][player_id].update(player_stats)


def add_st_stats_to_defense(df: pd.DataFrame, stat_map):
    df_playoffs = df[
        (
            df.week.isin(NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK.keys())
            & df.team.isin(ROSTER_TEAM_NAMES_TO_DEFENSE_PLAYER_ID.keys())
        )
    ]
    df_playoffs.week = df_playoffs.week.apply(lambda x: NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK[x])

    for week_team_tuple, row in df_playoffs.set_index(["week", "team"]).iterrows():
        if isinstance(week_team_tuple, tuple):
            week, team = week_team_tuple
        else:
            # this won't happen but makes mypy happy
            continue
        player_id = ROSTER_TEAM_NAMES_TO_DEFENSE_PLAYER_ID[team]
        player_stats: dict[str, float] = stat_map[week].get(player_id, {})

        # special teams td update only
        for k, v in row.to_dict().items():
            if k == "special_teams_tds":
                if (mapped_k := NFLVERSE_STAT_COL_TO_ID[k]) in player_stats:
                    player_stats[mapped_k] += v
                else:
                    player_stats[mapped_k] = v

        stat_map[week][player_id] = player_stats


# 24 hour cache
@st.cache_data(ttl=60 * 60 * 24)
def assemble_nflverse_stats() -> dict[int, dict[str, dict[str, float]]]:
    # map week -> player_id -> stat_key -> stat value
    stat_map: dict[int, dict[str, dict[str, float]]] = {w: {} for w in NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK.values()}

    df_player_stats = get_player_stats()
    df_kicking_stats = get_player_kicking_stats()
    df_def_stats = get_team_defense_stats()

    add_stats_from_player_df_to_stat_map(df_player_stats, stat_map)
    add_stats_from_player_df_to_stat_map(df_kicking_stats, stat_map)
    add_stats_from_team_def_df_to_stat_map(df_def_stats, stat_map)
    add_st_stats_to_defense(df_player_stats, stat_map)

    return stat_map


def get_live_stats() -> dict[int, dict[str, dict[str, float]]]:
    try:
        return get_yahoo_stats()
    except Exception as e:
        print(f"Failed to get yahoo live stats: {str(e)}")
        return {w: {} for w in NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK.values()}


YAHOO_TO_STAT_MAP: dict[str, dict[str, str]] = {
    "PASSING": {
        "PASSING_YARDS": PASS_YD.key,
        "PASSING_TOUCHDOWNS": PASS_TD.key,
        "PASSING_INTERCEPTIONS": PASS_INT.key,
        "FUMBLES_LOST": FUM_LOST.key,
    },
    "RUSHING": {
        "RUSHING_TOUCHDOWNS": RUSH_TD.key,
        "FUMBLES_LOST": FUM_LOST.key,
        "RUSHING_YARDS": RUSH_YD.key,
    },
    "RECEIVING": {
        "RECEPTIONS": RECEPTION.key,
        "RECEIVING_YARDS": REC_YD.key,
        "RECEIVING_TOUCHDOWNS": REC_TD.key,
        "FUMBLES_LOST": FUM_LOST.key,
    },
    "KICKING": {
        "FIELD_GOALS_MADE_0_19": FG_0_49.key,
        "FIELD_GOALS_MADE_20_29": FG_0_49.key,
        "FIELD_GOALS_MADE_30_39": FG_0_49.key,
        "FIELD_GOALS_MADE_40_49": FG_0_49.key,
        "FIELD_GOALS_MADE_50_PLUS": FG_50_.key,
        "EXTRA_POINTS_MADE": XP.key,
    },
    "DEFENSE": {
        "SACKS": SACK.key,
        "INTERCEPTIONS_FORCED": DEF_INT.key,
        "INTERCEPTION_RETURN_TOUCHDOWNS": DEF_TD.key,
        "FORCED_FUMBLES": FUM_REC.key,
        "FUMBLE_RETURN_TOUCHDOWNS": DEF_TD.key,
        "SAFETIES": SAFETY.key,
    },
    "RETURNING": {
        "KICKOFF_RETURN_TOUCHDOWNS": ST_TD.key,
        "PUNT_RETURN_TOUCHDOWNS": ST_TD.key,
    },
}


def get_yahoo_id_map() -> dict[str, str]:
    try:
        teams_included = [x.id_map_short_name for x, _ in PLAYOFF_TEAM_DEF_PLAYER]
        df = pd.read_csv(r"https://raw.githubusercontent.com/dynastyprocess/data/master/files/db_playerids.csv")
        df = df[(df["yahoo_id"].notna() & df["gsis_id"].notna() & df["team"].isin(teams_included))]
        df["yahoo_id"] = df["yahoo_id"].astype(int).astype(str)
        return df.set_index("yahoo_id")["gsis_id"].to_dict()
    except Exception as e:
        print(f"Failed to get yahoo id map: {str(e)}")
        return {}


YAHOO_PLAYER_ID_MAP = get_yahoo_id_map()

YAHOO_WEEK_MAP = {
    19: 1,
    20: 2,
    21: 3,
    23: 4,
}


def add_yahoo_stat_type_to_stat_map(
    stats_object, yahoo_stat_type: str, stat_map: dict[int, dict[str, dict[str, float]]]
):
    assert yahoo_stat_type in YAHOO_TO_STAT_MAP

    nfl_object = stats_object["nfl"]["200"][f"{SEASON}"]

    for raw_week, week_dict in nfl_object.items():
        week = YAHOO_WEEK_MAP[int(raw_week)]
        if week not in stat_map:
            stat_map[week] = {}

        if yahoo_stat_type == "KICKING":
            week_leaders = week_dict["POSTSEASON"][""]["FIELD_GOALS_MADE"]["leagues"][0]["leagueWeeks"][0]["leaders"]
        elif yahoo_stat_type == "DEFENSE":
            week_leaders = week_dict["POSTSEASON"][""]["TOTAL_TACKLES"]["leagues"][0]["leagueWeeks"][0]["leaders"]
        elif yahoo_stat_type == "RETURNING":
            week_leaders = week_dict["POSTSEASON"][""]["RETURN_YARDS_PER_KICKOFF"]["leagues"][0]["leagueWeeks"][0][
                "leaders"
            ]
        else:
            week_leaders = week_dict["POSTSEASON"][""][f"{yahoo_stat_type}_YARDS"]["leagues"][0]["leagueWeeks"][0][
                "leaders"
            ]

        for player in week_leaders:
            def_player_id = ""
            player_id = ""
            if yahoo_stat_type == "DEFENSE":
                def_player_id = SHORT_TEAM_NAMES_TO_DEFENSE_PLAYER_ID[player["player"]["team"]["abbreviation"]]
            elif yahoo_stat_type == "RETURNING":
                raw_player_id = player["player"]["playerId"].split(".")[-1]
                player_id = YAHOO_PLAYER_ID_MAP.get(raw_player_id, "")
                def_player_id = SHORT_TEAM_NAMES_TO_DEFENSE_PLAYER_ID[player["player"]["team"]["abbreviation"]]
            else:
                raw_player_id = player["player"]["playerId"].split(".")[-1]
                player_id = YAHOO_PLAYER_ID_MAP.get(raw_player_id, "")

            map_stats_to_week_player_id(player_id, week, player, stat_map, yahoo_stat_type)
            map_stats_to_week_player_id(def_player_id, week, player, stat_map, yahoo_stat_type)


def map_stats_to_week_player_id(player_id: str, week: int, player, stat_map, yahoo_stat_type):
    if not player_id:
        return

    if player_id not in stat_map[week]:
        stat_map[week][player_id] = {}
    stats = player["stats"]
    for stat in stats:
        if stat_key := YAHOO_TO_STAT_MAP[yahoo_stat_type].get(stat["statId"]):
            if stat_key in stat_map[week][player_id]:
                stat_map[week][player_id][stat_key] += float(stat["value"] or 0.0)
            else:
                stat_map[week][player_id][stat_key] = float(stat["value"] or 0.0)
        # else:
        #     # remove after mapping all intended
        #     stat_map[week][player_id][stat["statId"]] = stat["value"]


def get_yahoo_stat_json_obj():
    url = "https://sports.yahoo.com/nfl/stats/weekly/?selectedTable=0"
    request = requests.get(url)
    request_content_str = request.text

    start_str = """root.App.main = """
    end_str = """;\n}(this));"""

    start_slice_pos = request_content_str.find(start_str) + len(start_str)
    first_slice = request_content_str[start_slice_pos:]
    end_slice_pos = first_slice.find(end_str)
    dom_str = first_slice[:end_slice_pos]
    dom_json = json.loads(dom_str)
    return dom_json


def get_yahoo_schedule() -> dict[int, dict[str, dict[str, str | int | pd.Timestamp]]]:
    schedule_map: dict[int, dict[str, dict[str, str | int | pd.Timestamp]]] = {}
    dom_json = get_yahoo_stat_json_obj()
    team_id_to_abbr = {}
    teams_json = dom_json["context"]["dispatcher"]["stores"]["TeamsStore"]["teams"]
    for team_key, team_dict in teams_json.items():
        if not team_key.split(".")[0] == "nfl":
            continue
        team_id_to_abbr[team_dict["team_id"]] = team_dict["abbr"]

    games_json = dom_json["context"]["dispatcher"]["stores"]["GamesStore"]["games"]
    for game_id, game in games_json.items():
        if not game_id.split(".")[0] == "nfl":
            continue
        away_team = team_id_to_abbr[game["away_team_id"]]
        home_team = team_id_to_abbr[game["home_team_id"]]

        if "week_number" not in game:
            continue

        week = YAHOO_WEEK_MAP[int(game["week_number"])]

        if week not in schedule_map:
            schedule_map[week] = {}

        home_team_map: dict[str, str | int | pd.Timestamp] = {}
        away_team_map: dict[str, str | int | pd.Timestamp] = {}

        if game["status_type"] != "pregame":
            away_score = int(game["total_away_points"] or 0)
            home_score = int(game["total_home_points"] or 0)
            home_team_map.update(
                {
                    "score": home_score,
                    "opponent_score": away_score,
                }
            )
            away_team_map.update(
                {
                    "score": away_score,
                    "opponent_score": home_score,
                }
            )

            if game["status_type"] == "in_progress":
                clock_status = game["status_display_name"]
                if clock_status:
                    home_team_map.update({"status": clock_status})
                    away_team_map.update({"status": clock_status})
            elif game["status_type"] == "final":
                home_team_win = home_score > away_score
                home_status = "Win" if home_team_win else "Loss"
                away_status = "Loss" if home_team_win else "Win"
                home_team_map.update({"status": home_status})
                away_team_map.update({"status": away_status})

        schedule_map[week][home_team] = home_team_map
        schedule_map[week][away_team] = away_team_map

    return schedule_map


def get_yahoo_stats() -> dict[int, dict[str, dict[str, float]]]:
    dom_json = get_yahoo_stat_json_obj()
    stat_map: dict[int, dict[str, dict[str, float]]] = {w: {} for w in NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK.values()}

    stats_json = dom_json["context"]["dispatcher"]["stores"]["GraphStatsStore"]

    add_yahoo_stat_type_to_stat_map(stats_json["weeklyStatsFootballPassing"], "PASSING", stat_map)
    add_yahoo_stat_type_to_stat_map(stats_json["weeklyStatsFootballRushing"], "RUSHING", stat_map)
    add_yahoo_stat_type_to_stat_map(stats_json["weeklyStatsFootballReceiving"], "RECEIVING", stat_map)
    add_yahoo_stat_type_to_stat_map(stats_json["weeklyStatsFootballKicking"], "KICKING", stat_map)
    add_yahoo_stat_type_to_stat_map(stats_json["weeklyStatsFootballReturns"], "RETURNING", stat_map)
    add_yahoo_stat_type_to_stat_map(stats_json["weeklyStatsFootballDefense"], "DEFENSE", stat_map)
    return stat_map


def get_live_schedule() -> dict[int, dict[str, dict[str, str | int | pd.Timestamp]]]:
    try:
        return get_yahoo_schedule()
    except Exception as e:
        print(f"Failed to get yahoo schedule: {str(e)}")
        return {}


@st.cache_data(ttl=60 * 60 * 24)
def get_daily_updated_schedule() -> dict[int, dict[str, dict[str, str | int | pd.Timestamp]]]:
    schedule, _ = get_season_game_map(SEASON)
    return schedule


@st.cache_data(ttl=STAT_CACHE_SECONDS)
def get_schedule_with_live() -> dict[int, dict[str, dict[str, str | int | pd.Timestamp]]]:
    schedule = get_live_schedule()

    for week, week_live in get_daily_updated_schedule().items():
        if week not in schedule:
            schedule[week] = {}
        for team, team_live in week_live.items():
            if team not in schedule[week]:
                schedule[week][team] = {}
            schedule[week][team].update(team_live)

    return schedule


def add_points_against_team_win_stat(stat_map: dict[int, dict[str, dict[str, float]]]):
    schedule = get_schedule_with_live()

    for week, week_map in stat_map.items():
        for player_id in week_map.keys():
            if team_short_name := DEFENSE_PLAYER_ID_TO_ROSTER_TEAM_NAMES.get(player_id):
                try:
                    team_game = schedule[week][team_short_name]
                    opponent_team = str(team_game["opponent"])
                    opponent_player_id = ROSTER_TEAM_NAMES_TO_DEFENSE_PLAYER_ID[opponent_team]
                    opponent_def_stats = week_map[opponent_player_id]

                    if isinstance((opponent_score_str := team_game["opponent_score"]), pd.Timestamp):
                        # another case not possible but makes mypy happy
                        continue
                    opponent_score = float(opponent_score_str)  # noqa

                    opponent_def_points_scored = (
                        opponent_def_stats.get(SAFETY.key, 0.0) * 2.0 + opponent_def_stats.get(DEF_TD.key, 0.0) * 6.0
                    )
                    points_allowed = opponent_score - opponent_def_points_scored

                    if points_allowed == 0:
                        stat_map[week][player_id].update({PTS_ALLOW_0.key: 1})
                    elif points_allowed < 7:
                        stat_map[week][player_id].update({PTS_ALLOW_1_6.key: 1})
                    elif points_allowed < 14:
                        stat_map[week][player_id].update({PTS_ALLOW_7_13.key: 1})
                    elif points_allowed < 21:
                        stat_map[week][player_id].update({PTS_ALLOW_14_20.key: 1})
                    elif points_allowed < 28:
                        stat_map[week][player_id].update({PTS_ALLOW_21_27.key: 1})
                    elif points_allowed < 35:
                        stat_map[week][player_id].update({PTS_ALLOW_28_34.key: 1})
                    else:
                        stat_map[week][player_id].update({PTS_ALLOW_35_.key: 1})

                    # check for win
                    if team_game["status"] == "Win":
                        stat_map[week][player_id].update({TEAM_WIN.key: 1})

                except Exception:
                    continue


@st.cache_data(ttl=STAT_CACHE_SECONDS)
def get_stats_map() -> dict[int, dict[str, dict[str, float]]]:
    # use live stats if available
    stat_map = get_live_stats()

    # use more permanent nflverse stats over live
    nflverse_stats = assemble_nflverse_stats()

    # we overwrite the live stats with nflverse stats if they exist for the same player
    for week, week_stats in nflverse_stats.items():
        for player_id, player_stats in week_stats.items():
            stat_map[week][player_id] = player_stats

    add_points_against_team_win_stat(stat_map)
    stat_overrides = get_stat_overrides()
    # for stat overrides, override at the stat level
    for week, week_stats in stat_overrides.items():
        for player_id, player_stats in week_stats.items():
            for stat_key, stat_value in player_stats.items():
                if player_id not in stat_map[week]:
                    stat_map[week][player_id] = {}
                stat_map[week][player_id][stat_key] = stat_value

    return stat_map


@st.cache_data(ttl=STAT_CACHE_SECONDS)
def get_scores_map() -> dict[int, dict[str, float]]:
    scores_map: dict[int, dict[str, float]] = {w: {} for w in NFLVERSE_STAT_WEEK_TO_PLAYOFF_WEEK.values()}

    stat_map = get_stats_map()

    for week, week_stats in stat_map.items():
        for player_id, player_stats in week_stats.items():
            score = 0.0
            for stat_key, stat_value in player_stats.items():
                stat_type = STAT_KEY_MAP[stat_key]
                score += stat_type.score * stat_value
            scores_map[week][player_id] = score

    return scores_map