Spoolman/spoolman/database/database.py

"""SQLAlchemy database setup."""

import datetime
import logging
import shutil
import sqlite3
from collections.abc import AsyncGenerator
from os import PathLike
from pathlib import Path

from scheduler.asyncio.scheduler import Scheduler
from sqlalchemy import URL
from sqlalchemy.ext.asyncio import AsyncEngine, AsyncSession, async_sessionmaker, create_async_engine

from spoolman import env
from spoolman.prometheus.metrics import filament_metrics, spool_metrics

logger = logging.getLogger(__name__)


def get_connection_url() -> URL:
    """Construct the connection URL for the database based on environment variables."""
    db_type = env.get_database_type()
    host = env.get_host()
    port = env.get_port()
    database = env.get_database()
    query = env.get_query()
    username = env.get_username()
    password = env.get_password()

    if db_type is None:
        db_type = env.DatabaseType.SQLITE

        database = str(env.get_data_dir().joinpath("spoolman.db"))
        logger.info('No database type specified, using a default SQLite database located at "%s"', database)
    elif db_type is env.DatabaseType.SQLITE:
        if database is not None:
            raise ValueError("Cannot specify a database name when using SQLite.")

        database = str(env.get_data_dir().joinpath("spoolman.db"))
        logger.info('Using SQLite database located at "%s"', database)
    else:
        logger.info('Connecting to database of type "%s" at "%s:%s"', db_type, host, port)

    return URL.create(
        drivername=db_type.to_drivername(),
        host=host,
        port=port,
        database=database,
        query=query or {},
        username=username,
        password=password,
    )


class Database:
    connection_url: URL
    engine: AsyncEngine | None
    session_maker: async_sessionmaker[AsyncSession] | None

    def __init__(self, connection_url: URL) -> None:
        """Construct the Database wrapper and set config parameters."""
        self.connection_url = connection_url

    def is_file_based_sqlite(self) -> bool:
        """Return True if the database is file based."""
        return (
            self.connection_url.drivername[:6] == "sqlite"
            and self.connection_url.database is not None
            and self.connection_url.database != ":memory:"
        )

    def connect(self) -> None:
        """Connect to the database."""
        if env.get_logging_level() == logging.DEBUG:
            logging.getLogger("sqlalchemy.engine").setLevel(logging.INFO)

        connect_args = {}
        if self.connection_url.drivername == "sqlite+aiosqlite":
            connect_args["timeout"] = 60
        connection_options = {}
        if self.connection_url.drivername == "mysql+aiomysql":
            connection_options["pool_recycle"] = 3600
        self.engine = create_async_engine(
            self.connection_url,
            connect_args=connect_args,
            pool_pre_ping=True,
            **connection_options,
        )
        self.session_maker = async_sessionmaker(self.engine, autocommit=False, autoflush=True, expire_on_commit=False)

    def backup(self, target_path: str | PathLike[str]) -> None:
        """Backup the database."""
        if not self.is_file_based_sqlite() or self.connection_url.database is None:
            return

        logger.info("Backing up SQLite database to %s", target_path)

        def progress(_: int, remaining: int, total: int) -> None:
            logger.info("Copied %d of %d pages.", total - remaining, total)

        if self.connection_url.database == target_path:
            raise ValueError("Cannot backup database to itself.")
        if Path(target_path).exists():
            raise ValueError("Backup target file already exists.")

        with sqlite3.connect(self.connection_url.database) as src, sqlite3.connect(target_path) as dst:
            src.backup(dst, pages=1, progress=progress)

        logger.info("Backup complete.")

    def backup_and_rotate(
        self,
        backup_folder: str | PathLike[str],
        num_backups: int = 5,
    ) -> Path | None:
        """Backup the database and rotate existing backups.

        Args:
            backup_folder: The folder to store the backups in.
            num_backups: The number of backups to keep.

        Returns:
            The path to the created backup or None if no backup was created.

        """
        if not self.is_file_based_sqlite() or self.connection_url.database is None:
            logger.info("Skipping backup as the database is not SQLite.")
            return None

        backup_folder = Path(backup_folder)
        backup_folder.mkdir(parents=True, exist_ok=True)

        # Delete oldest backup
        backup_path = backup_folder.joinpath(f"spoolman.db.{num_backups}")
        if backup_path.exists():
            logger.info("Deleting oldest backup %s", backup_path)
            backup_path.unlink()

        # Rotate existing backups
        for i in range(num_backups - 1, -1, -1):
            if i == 0:
                backup_path = backup_folder.joinpath("spoolman.db")
            else:
                backup_path = backup_folder.joinpath(f"spoolman.db.{i}")
            if backup_path.exists():
                logger.debug("Rotating backup %s to %s", backup_path, backup_folder.joinpath(f"spoolman.db.{i + 1}"))
                shutil.move(backup_path, backup_folder.joinpath(f"spoolman.db.{i + 1}"))

        # Create new backup
        backup_path = backup_folder.joinpath("spoolman.db")
        self.backup(backup_path)

        return backup_path


__db: Database | None = None


def setup_db(connection_url: URL) -> None:
    """Connect to the database.

    Args:
        connection_url: The URL to connect to the database.

    """
    global __db  # noqa: PLW0603
    __db = Database(connection_url)
    __db.connect()


def get_engine() -> AsyncEngine:
    """Get the connected engine.

    Raises:
        RuntimeError: If the database has not been set up.

    Returns:
        AsyncEngine: The engine.

    """
    if __db is None or __db.engine is None:
        raise RuntimeError("DB is not setup.")
    return __db.engine


async def backup_global_db(num_backups: int = 5) -> Path | None:
    """Backup the database and rotate existing backups.

    Returns:
        The path to the created backup or None if no backup was created.

    """
    if __db is None:
        raise RuntimeError("DB is not setup.")
    return __db.backup_and_rotate(env.get_backups_dir(), num_backups=num_backups)


async def _backup_task() -> Path | None:
    """Perform scheduled backup of the database."""
    logger.info("Performing scheduled database backup.")
    if __db is None:
        raise RuntimeError("DB is not setup.")
    return __db.backup_and_rotate(env.get_backups_dir(), num_backups=5)


async def _metrics() -> None:
    """Create some useful prometheus metrics."""
    logger.debug("Start metrics collection")
    async for session in get_db_session():
        await filament_metrics(session)
        await spool_metrics(session)
    logger.debug("End metrics collection")


def schedule_tasks(scheduler: Scheduler) -> None:
    """Schedule tasks to be executed by the provided scheduler.

    Args:
        scheduler: The scheduler to use for scheduling tasks.

    """
    if __db is None:
        raise RuntimeError("DB is not setup.")
    if env.is_metrics_enabled():
        logger.info("Scheduling automatic metric collection.")
        # Run every minute, may be needs specify timer
        scheduler.minutely(datetime.time(second=0), _metrics)  # type: ignore[arg-type]
    if not env.is_automatic_backup_enabled():
        return
    if "sqlite" in __db.connection_url.drivername:
        logger.info("Scheduling automatic database backup for midnight.")
        # Schedule for midnight
        scheduler.daily(datetime.time(hour=0, minute=0, second=0), _backup_task)  # type: ignore[arg-type]


async def get_db_session() -> AsyncGenerator[AsyncSession]:
    """Get a DB session to be used with FastAPI's dependency system.

    Yields:
        The database session.

    """
    if __db is None or __db.session_maker is None:
        raise RuntimeError("DB is not setup.")
    async with __db.session_maker() as session:
        try:
            yield session
            await session.commit()
        except Exception as exc:
            await session.rollback()
            raise exc
        finally:
            await session.close()