paper-dynasty-card-creation/pd_cards/commands/retrosheet.py

"""
Retrosheet historical data processing commands.

Commands for generating cards from historical Retrosheet play-by-play data.
"""

import asyncio
from pathlib import Path
from typing import Optional

import typer
from rich.console import Console

app = typer.Typer(no_args_is_help=True)
console = Console()


@app.command()
def process(
    year: int = typer.Argument(..., help="Season year to process (e.g., 2005)"),
    cardset_id: int = typer.Option(..., "--cardset-id", "-c", help="Target cardset ID"),
    description: str = typer.Option("Live", "--description", "-d", help="Player description ('Live' or 'Month PotM')"),
    start_date: Optional[str] = typer.Option(None, "--start", help="Start date YYYYMMDD (defaults to March 1)"),
    end_date: Optional[str] = typer.Option(None, "--end", help="End date YYYYMMDD (defaults to Oct 2)"),
    events_file: Optional[str] = typer.Option(None, "--events", "-e", help="Retrosheet events CSV filename"),
    data_input: Optional[str] = typer.Option(None, "--input", "-i", help="Data input directory path"),
    season_pct: float = typer.Option(1.0, "--season-pct", help="Season percentage (0.0-1.0)"),
    min_pa_vl: int = typer.Option(None, "--min-pa-vl", help="Minimum PA vs LHP (default: 20 Live, 1 PotM)"),
    min_pa_vr: int = typer.Option(None, "--min-pa-vr", help="Minimum PA vs RHP (default: 40 Live, 1 PotM)"),
    post_data: bool = typer.Option(True, "--post/--no-post", help="Post data to database"),
    dry_run: bool = typer.Option(False, "--dry-run", "-n", help="Preview without saving to database"),
    last_week_ratio: float = typer.Option(0.0, "--last-week-ratio", help="Recency bias: weight for last week's stats (0.0-1.0)"),
    last_twoweeks_ratio: Optional[float] = typer.Option(None, "--last-twoweeks-ratio", help="Recency bias: weight for last 2 weeks' stats (0.0-1.0, default: 0.2 for Live after May 30)"),
    last_month_ratio: float = typer.Option(0.0, "--last-month-ratio", help="Recency bias: weight for last month's stats (0.0-1.0)"),
):
    """
    Process Retrosheet data and create player cards.

    Generates batting and pitching cards from historical play-by-play data.

    Example:
        pd-cards retrosheet process 2005 --cardset-id 27 --description Live
    """
    console.print()
    console.print("=" * 70)
    console.print(f"[bold]RETROSHEET PROCESSING - {year}[/bold]")
    console.print("=" * 70)

    # Calculate defaults based on description
    is_live = 'live' in description.lower()

    if min_pa_vl is None:
        min_pa_vl = 20 if is_live else 1
    if min_pa_vr is None:
        min_pa_vr = 40 if is_live else 1

    if start_date is None:
        start_date = f"{year}0301"
    if end_date is None:
        end_date = f"{year}1002"

    if events_file is None:
        events_file = f"retrosheets_events_{year}.csv"

    if data_input is None:
        data_input = f"data-input/{year} Live Cardset/"

    # Auto-enable recency bias for Live series after May 30
    if last_twoweeks_ratio is None:
        end_mmdd = int(end_date[4:])  # Extract MMDD from YYYYMMDD
        if is_live and end_mmdd > 530:
            last_twoweeks_ratio = 0.2
            console.print(f"[cyan]Auto-enabled recency bias (end date > May 30)[/cyan]")
        else:
            last_twoweeks_ratio = 0.0

    console.print(f"Cardset ID: {cardset_id}")
    console.print(f"Description: {description}")
    console.print(f"Date Range: {start_date} - {end_date}")
    console.print(f"Season %: {season_pct:.0%}")
    console.print(f"Min PA: vL={min_pa_vl}, vR={min_pa_vr}")
    console.print(f"Events: {events_file}")
    console.print(f"Input: {data_input}")
    if last_week_ratio > 0 or last_twoweeks_ratio > 0 or last_month_ratio > 0:
        console.print(f"Recency Bias: week={last_week_ratio}, 2weeks={last_twoweeks_ratio}, month={last_month_ratio}")
    console.print()

    if dry_run:
        console.print("[yellow]DRY RUN - no changes will be made[/yellow]")
        console.print()
        console.print("[green]Validation passed - ready to run[/green]")
        console.print()
        console.print("To run for real, remove --dry-run flag")
        raise typer.Exit(0)

    # Import and configure the retrosheet module
    try:
        import sys
        sys.path.insert(0, str(Path(__file__).parent.parent.parent))

        import retrosheet_data as rd

        # Configure the module's globals
        rd.CARDSET_ID = cardset_id
        rd.PLAYER_DESCRIPTION = description
        rd.START_DATE = int(start_date)
        rd.END_DATE = int(end_date)
        rd.SEASON_PCT = season_pct
        rd.MIN_PA_VL = min_pa_vl
        rd.MIN_PA_VR = min_pa_vr
        rd.MIN_TBF_VL = min_pa_vl
        rd.MIN_TBF_VR = min_pa_vr
        rd.POST_DATA = post_data
        rd.EVENTS_FILENAME = events_file
        rd.DATA_INPUT_FILE_PATH = data_input
        rd.LAST_WEEK_RATIO = last_week_ratio
        rd.LAST_TWOWEEKS_RATIO = last_twoweeks_ratio
        rd.LAST_MONTH_RATIO = last_month_ratio

        console.print("[bold]Starting Retrosheet processing...[/bold]")
        console.print()

        # Run the main function (args is legacy, pass empty list)
        asyncio.run(rd.main([]))

        console.print()
        console.print("=" * 70)
        console.print(f"[bold green]✓ RETROSHEET PROCESSING COMPLETE[/bold green]")
        console.print("=" * 70)

    except ImportError as e:
        console.print(f"[red]Error importing modules: {e}[/red]")
        console.print("Make sure you're running from the card-creation directory")
        raise typer.Exit(1)
    except Exception as e:
        console.print(f"[red]Error: {e}[/red]")
        import traceback
        traceback.print_exc()
        raise typer.Exit(1)


@app.command()
def arms(
    year: int = typer.Argument(..., help="Season year"),
    events_file: Path = typer.Option(..., "--events", "-e", help="Retrosheet events CSV file"),
    output: Optional[Path] = typer.Option(None, "--output", "-o", help="Output CSV file"),
    season_pct: float = typer.Option(1.0, "--season-pct", help="Season percentage for min sample"),
):
    """
    Generate outfield arm ratings from Retrosheet data.

    Analyzes play-by-play events to calculate OF arm strength ratings.

    Example:
        pd-cards retrosheet arms 2005 --events data-input/retrosheet/retrosheets_events_2005.csv
    """
    console.print()
    console.print("=" * 70)
    console.print(f"[bold]OUTFIELD ARM RATINGS - {year}[/bold]")
    console.print("=" * 70)

    if not events_file.exists():
        console.print(f"[red]Events file not found: {events_file}[/red]")
        raise typer.Exit(1)

    if output is None:
        output = Path(f"data-output/retrosheet_arm_ratings_{year}.csv")

    console.print(f"Events file: {events_file}")
    console.print(f"Output: {output}")
    console.print(f"Season %: {season_pct:.0%}")
    console.print()

    try:
        import sys
        sys.path.insert(0, str(Path(__file__).parent.parent.parent))

        import pandas as pd
        from defenders.retrosheet_arm_calculator import calculate_of_arms_from_retrosheet

        console.print("Loading Retrosheet events...")
        df_events = pd.read_csv(events_file)
        console.print(f"  Loaded {len(df_events)} events")

        console.print("Calculating arm ratings...")
        arm_ratings = calculate_of_arms_from_retrosheet(df_events, season_pct=season_pct)

        # Convert to DataFrame for CSV output
        rows = []
        for key, rating in arm_ratings.items():
            player_id, position = key.rsplit('_', 1)
            rows.append({
                'player_id': player_id,
                'position': position.upper(),
                'season': year,
                'arm_rating': rating.get('arm_rating', 0),
                'balls_fielded': rating.get('balls_fielded', 0),
                'total_assists': rating.get('total_assists', 0),
                'assist_rate': rating.get('assist_rate', 0),
                'z_score': rating.get('z_score', 0),
            })

        df_output = pd.DataFrame(rows)
        df_output = df_output.sort_values(['position', 'arm_rating'])

        # Ensure output directory exists
        output.parent.mkdir(parents=True, exist_ok=True)

        df_output.to_csv(output, index=False)
        console.print()
        console.print(f"[green]✓ Saved {len(df_output)} arm ratings to {output}[/green]")

        # Show distribution summary
        from rich.table import Table
        table = Table(title="Arm Rating Distribution")
        table.add_column("Position")
        table.add_column("Count", justify="right")
        table.add_column("Avg Rating", justify="right")
        table.add_column("Min", justify="right")
        table.add_column("Max", justify="right")

        for pos in ['LF', 'CF', 'RF']:
            pos_df = df_output[df_output['position'] == pos]
            if len(pos_df) > 0:
                table.add_row(
                    pos,
                    str(len(pos_df)),
                    f"{pos_df['arm_rating'].mean():.1f}",
                    str(int(pos_df['arm_rating'].min())),
                    str(int(pos_df['arm_rating'].max()))
                )

        console.print()
        console.print(table)

    except ImportError as e:
        console.print(f"[red]Error importing modules: {e}[/red]")
        raise typer.Exit(1)
    except Exception as e:
        console.print(f"[red]Error: {e}[/red]")
        import traceback
        traceback.print_exc()
        raise typer.Exit(1)


@app.command()
def validate(
    cardset_id: int = typer.Argument(..., help="Cardset ID to validate"),
    api_url: str = typer.Option("https://pd.manticorum.com/api", "--api", help="API URL"),
):
    """
    Validate positions for a cardset.

    Checks for anomalous DH counts and missing outfield positions.

    Example:
        pd-cards retrosheet validate 27
    """
    console.print()
    console.print("=" * 70)
    console.print(f"[bold]POSITION VALIDATION - Cardset {cardset_id}[/bold]")
    console.print("=" * 70)

    try:
        import sys
        sys.path.insert(0, str(Path(__file__).parent.parent.parent))
        from db_calls import db_get

        async def run_validation():
            # Get all players in cardset
            console.print(f"Fetching players from cardset {cardset_id}...")
            players = await db_get('players', params=[('cardset_id', cardset_id)])

            if players is None or players['count'] == 0:
                console.print(f"[yellow]No players found in cardset {cardset_id}[/yellow]")
                return

            console.print(f"Found {players['count']} players")
            console.print()

            # Count positions
            pos_counts = {}
            dh_players = []

            for player in players['players']:
                pos_1 = player.get('pos_1', '')
                if pos_1:
                    pos_counts[pos_1] = pos_counts.get(pos_1, 0) + 1
                    if pos_1 == 'DH':
                        dh_players.append(f"{player.get('p_name', 'Unknown')} (ID: {player['player_id']})")

            # Display position counts
            from rich.table import Table
            table = Table(title="Position Distribution")
            table.add_column("Position")
            table.add_column("Count", justify="right")

            for pos in sorted(pos_counts.keys()):
                count = pos_counts[pos]
                style = "red" if pos == 'DH' and count > 5 else None
                table.add_row(pos, str(count), style=style)

            console.print(table)
            console.print()

            # Check for anomalies
            issues = []

            dh_count = pos_counts.get('DH', 0)
            if dh_count > 5:
                issues.append(f"[red]⚠ Anomalous DH count: {dh_count} (should be <5 for full-season cards)[/red]")
                console.print("DH Players:")
                for p in dh_players[:10]:
                    console.print(f"  - {p}")
                if len(dh_players) > 10:
                    console.print(f"  ... and {len(dh_players) - 10} more")
                console.print()

            for pos in ['LF', 'CF', 'RF']:
                if pos_counts.get(pos, 0) == 0:
                    issues.append(f"[red]⚠ Missing {pos} positions (indicates defensive calculation failures)[/red]")

            if issues:
                console.print("[bold]Issues Found:[/bold]")
                for issue in issues:
                    console.print(f"  {issue}")
            else:
                console.print("[green]✓ No position anomalies detected[/green]")

        asyncio.run(run_validation())

    except ImportError as e:
        console.print(f"[red]Error importing modules: {e}[/red]")
        raise typer.Exit(1)
    except Exception as e:
        console.print(f"[red]Error: {e}[/red]")
        raise typer.Exit(1)


@app.command()
def defense(
    year: int = typer.Argument(..., help="Season year to fetch defense stats for"),
    output_dir: Path = typer.Option(None, "--output", "-o", help="Output directory for CSV files"),
):
    """
    Fetch and store defensive statistics from Baseball Reference.

    Downloads fielding stats for all positions and saves to CSV files.

    Example:
        pd-cards retrosheet defense 2005 --output "data-input/2005 Live Cardset/"
    """
    console.print()
    console.print("=" * 70)
    console.print(f"[bold]FETCH DEFENSIVE STATS - {year}[/bold]")
    console.print("=" * 70)

    if output_dir is None:
        output_dir = Path(f"data-input/{year} Live Cardset/")

    console.print(f"Output directory: {output_dir}")
    console.print()

    try:
        import sys
        sys.path.insert(0, str(Path(__file__).parent.parent.parent))

        import defenders.calcs_defense as cde

        positions = ['c', '1b', '2b', '3b', 'ss', 'lf', 'cf', 'rf', 'of', 'p']

        output_dir.mkdir(parents=True, exist_ok=True)

        async def fetch_defense():
            for position in positions:
                console.print(f"Fetching {position.upper()} defensive stats...")
                pos_df = cde.get_bbref_fielding_df(position, year)
                output_file = output_dir / f"defense_{position}.csv"
                pos_df.to_csv(output_file)
                console.print(f"  [green]✓ Saved {len(pos_df)} records to {output_file}[/green]")
                await asyncio.sleep(8)  # Rate limiting

            console.print()
            console.print(f"[green]✓ All defensive stats saved to {output_dir}[/green]")

        asyncio.run(fetch_defense())

    except ImportError as e:
        console.print(f"[red]Error importing modules: {e}[/red]")
        raise typer.Exit(1)
    except Exception as e:
        console.print(f"[red]Error: {e}[/red]")
        import traceback
        traceback.print_exc()
        raise typer.Exit(1)