from __future__ import annotations

from typing import Any

import pandas as pd


def build_batter_prop_outcome_rows_from_audit(
    audit_df: pd.DataFrame,
    graded_at: str,
) -> pd.DataFrame:
    """
    Safe scaffold for batter-prop outcome grading.

    This does NOT yet infer real hit/HR/TB outcomes from play-by-play.
    It creates a persistent batter-level grading table structure from
    recommendation audit rows so later phases can fill realized values.
    """
    if audit_df is None or audit_df.empty:
        return pd.DataFrame()

    rows: list[dict[str, Any]] = []

    for _, row in audit_df.iterrows():
        rows.append(
            {
                "created_at": row.get("created_at"),
                "graded_at": graded_at,
                "game_pk": str(row.get("game_pk", "") or "").strip(),
                "away_team": str(row.get("away_team", "") or "").strip(),
                "home_team": str(row.get("home_team", "") or "").strip(),
                "slot": str(row.get("slot", "") or "").strip(),
                "batter_name": str(row.get("batter_name", "") or "").strip(),
                "pitcher_name": str(row.get("pitcher_name", "") or "").strip(),
                "market": "hr",
                "fair_hr_odds": row.get("fair_hr_odds"),
                "book_hr_odds": row.get("book_hr_odds"),
                "adjusted_edge": row.get("adjusted_edge"),
                "confidence": row.get("confidence"),
                "recommendation_tier": row.get("recommendation_tier"),
                "realized_hit": None,
                "realized_hr": None,
                "realized_tb2p": None,
                "grade_status": "pending",
                "outcome_source": "pending_batter_prop_grade",
            }
        )

    df = pd.DataFrame(rows)
    if df.empty:
        return df
    # One outcome row per (game_pk, batter_name, market) — keep latest by created_at
    df = (
        df.sort_values("created_at", ascending=False, na_position="last")
          .drop_duplicates(subset=["game_pk", "batter_name", "market"], keep="first")
          .reset_index(drop=True)
    )
    return df