# Database Module - Async PostgreSQL Persistence

## Purpose

Async PostgreSQL persistence layer using SQLAlchemy 2.0. Handles all database operations with connection pooling and proper transaction management.

## Structure

```
app/database/
├── __init__.py      # Package exports
├── session.py       # Async session management, Base declarative
└── operations.py    # DatabaseOperations class
```

## Session Management

### Async Session Pattern
```python
from app.database.session import get_session

async def some_function():
    async with get_session() as session:
        result = await session.execute(query)
        # Auto-commits on success, rolls back on exception
```

### Connection Pool
- **Driver**: asyncpg
- **Pool**: SQLAlchemy async pool
- Configurable pool size via env vars

## DatabaseOperations

Singleton class with all database operations.

### Game Operations
```python
from app.database.operations import DatabaseOperations
db_ops = DatabaseOperations()

await db_ops.create_game(game_id, league_id, home_team_id, away_team_id, ...)
await db_ops.update_game_state(game_id, inning, half, home_score, away_score)
game_data = await db_ops.load_game_state(game_id)
```

### Play Operations
```python
play_id = await db_ops.save_play(game_id, play_data, stats_data)
plays = await db_ops.get_plays(game_id, limit=100)
```

### Lineup Operations
```python
lineup_id = await db_ops.add_pd_lineup_card(game_id, team_id, card_id, position, batting_order)
lineup_id = await db_ops.add_sba_lineup_player(game_id, team_id, player_id, position, batting_order)
lineup = await db_ops.get_active_lineup(game_id, team_id)
await db_ops.deactivate_lineup_player(lineup_id)
```

### Roster Operations
```python
await db_ops.add_pd_roster_card(game_id, card_id, team_id)
await db_ops.add_sba_roster_player(game_id, player_id, team_id)
roster = await db_ops.get_pd_roster(game_id, team_id)
```

### Session Operations
```python
await db_ops.create_game_session(game_id)
await db_ops.update_session_snapshot(game_id, state_snapshot)
```

## Common Patterns

### Transaction Handling
```python
async with AsyncSessionLocal() as session:
    try:
        # Multiple operations
        await session.commit()
    except Exception:
        await session.rollback()
        raise
```

### Efficient Queries
```python
# Use joinedload for relationships needed immediately
from sqlalchemy.orm import joinedload

query = select(Play).options(
    joinedload(Play.batter),
    joinedload(Play.pitcher)
)
```

### Direct UPDATE
```python
# More efficient than SELECT + modify + commit
stmt = update(Game).where(Game.id == game_id).values(inning=5)
await session.execute(stmt)
```

## Database Schema

See `../models/CLAUDE.md` for model details.

| Table | Purpose |
|-------|---------|
| games | Game container, scores, status |
| plays | At-bat records with 25+ stats |
| lineups | Player assignments, substitutions |
| game_sessions | WebSocket state |
| roster_links | Eligible cards/players |

## Environment

```bash
DATABASE_URL=postgresql+asyncpg://user:pass@10.10.0.42:5432/paperdynasty_dev
```

## References

- **Database Schema**: See `../../.claude/DATABASE_SCHEMA.md` for complete table details
- **Models**: See `../models/CLAUDE.md`
- **State Recovery**: See `../core/state_manager.py`

---

**Tests**: `tests/integration/database/` | **Updated**: 2025-01-19