"""
MLB Statcast enrichment using pybaseball.

Provides:
- Pitcher pitch-mix + zone heatmap data for K-prop grading
- Batter vs Pitcher historical matchup data

We avoid wide-net `statcast()` calls that pull every pitch league-wide —
those routinely time out. Pitcher-specific calls are scoped to a 30-day
trailing window which keeps payloads under a few hundred KB.
"""
from __future__ import annotations

import time
from datetime import datetime, timedelta
from typing import Optional

import pandas as pd
from pybaseball import statcast_pitcher

from app.utils.cache import cache_get, cache_set
from app.config import SPLITS_TTL


def _today_iso() -> str:
    return datetime.utcnow().strftime("%Y-%m-%d")


def _date_n_days_ago(n: int) -> str:
    return (datetime.utcnow() - timedelta(days=n)).strftime("%Y-%m-%d")


def get_pitcher_profile(pitcher_id: int, days_back: int = 30) -> dict:
    """
    Aggregate a pitcher's recent pitch-level data into pitch mix,
    velocity, whiff/chase, and zone heatmap counts.
    """
    if not isinstance(pitcher_id, int) or pitcher_id <= 0:
        return {"error": "invalid pitcher_id"}

    cache_key = f"mlb:pitcher:{pitcher_id}:d{days_back}"
    cached = cache_get(cache_key)
    if cached is not None:
        cached["source"] = "cache"
        return cached

    end = _today_iso()
    start = _date_n_days_ago(days_back)

    try:
        data = statcast_pitcher(start, end, pitcher_id)
    except Exception as exc:
        return {"error": f"statcast fetch failed: {exc!s}"}

    if data is None or data.empty:
        return {"pitcher_id": pitcher_id, "pitch_mix": [], "zone": [], "note": "no data"}

    # Pitch mix
    description_col = data["description"] if "description" in data.columns else pd.Series(dtype=str)
    pitch_mix_grouped = data.groupby("pitch_type") if "pitch_type" in data.columns else None
    pitch_mix: list[dict] = []
    if pitch_mix_grouped is not None:
        for ptype, g in pitch_mix_grouped:
            total = len(g)
            d = g["description"] if "description" in g.columns else pd.Series(dtype=str)
            swings = d.isin([
                "swinging_strike", "foul", "foul_tip", "hit_into_play",
                "swinging_strike_blocked",
            ]).sum() if not d.empty else 0
            whiffs = (d == "swinging_strike").sum() if not d.empty else 0
            pitch_mix.append({
                "pitch_type": str(ptype),
                "count": int(total),
                "share": float(total / len(data)) if len(data) else 0.0,
                "avg_velocity": float(g["release_speed"].mean()) if "release_speed" in g.columns else None,
                "whiff_rate": float(whiffs / swings) if swings else 0.0,
            })

    # Zone heatmap (the existing pybaseball 'zone' column is the 13-zone scheme)
    zone_data: list[dict] = []
    if "zone" in data.columns:
        for zone, g in data.groupby("zone"):
            d = g["description"] if "description" in g.columns else pd.Series(dtype=str)
            zone_data.append({
                "zone": int(zone) if pd.notna(zone) else None,
                "pitches": int(len(g)),
                "whiff_rate": float((d == "swinging_strike").mean()) if not d.empty else 0.0,
            })

    result = {
        "pitcher_id": pitcher_id,
        "window_days": days_back,
        "total_pitches": int(len(data)),
        "avg_velocity": float(data["release_speed"].mean()) if "release_speed" in data.columns else None,
        "k_rate_estimate": float((data["events"] == "strikeout").mean()) if "events" in data.columns else None,
        "pitch_mix": pitch_mix,
        "zone": zone_data,
        "source": "statcast",
    }
    cache_set(cache_key, result, SPLITS_TTL)
    return result


def get_batter_vs_pitcher(batter_id: int, pitcher_id: int, years_back: int = 3) -> dict:
    """
    Historical matchup. We scope to the pitcher because their pitch stream
    is small enough to fetch quickly; then filter to plate appearances by
    the batter.
    """
    if not isinstance(batter_id, int) or not isinstance(pitcher_id, int):
        return {"error": "invalid ids"}

    cache_key = f"mlb:bvp:{batter_id}:{pitcher_id}:y{years_back}"
    cached = cache_get(cache_key)
    if cached is not None:
        cached["source"] = "cache"
        return cached

    end = _today_iso()
    start = _date_n_days_ago(365 * years_back)

    try:
        pitcher_data = statcast_pitcher(start, end, pitcher_id)
    except Exception as exc:
        return {"error": f"statcast fetch failed: {exc!s}"}

    if pitcher_data is None or pitcher_data.empty or "batter" not in pitcher_data.columns:
        return {"batter_id": batter_id, "pitcher_id": pitcher_id, "matchup": "no data"}

    matchup = pitcher_data[pitcher_data["batter"] == batter_id]
    if matchup.empty:
        return {"batter_id": batter_id, "pitcher_id": pitcher_id, "matchup": "no history"}

    events = matchup["events"] if "events" in matchup.columns else pd.Series(dtype=str)
    result = {
        "batter_id": batter_id,
        "pitcher_id": pitcher_id,
        "plate_appearances": int(events.notna().sum()),
        "hits": int(events.isin(["single", "double", "triple", "home_run"]).sum()),
        "strikeouts": int((events == "strikeout").sum()),
        "home_runs": int((events == "home_run").sum()),
        "walks": int((events == "walk").sum()),
        "avg_exit_velocity": float(matchup["launch_speed"].mean()) if "launch_speed" in matchup.columns else None,
        "pitches_seen": int(len(matchup)),
        "pitch_types_faced": {
            str(k): int(v)
            for k, v in (matchup["pitch_type"].value_counts().to_dict().items() if "pitch_type" in matchup.columns else {}).items()
        },
        "source": "statcast",
    }
    # Cache aggressively — historical matchup data is stable.
    cache_set(cache_key, result, SPLITS_TTL * 2)
    return result