feat: delete old db backups in scheduled db backup task

keep last 100 by default

Author: Fallen-Breath

TODO.md

@@ -10,7 +10,7 @@
 ## QOL
 
 - [x] More restrictive `!!pb confirm` and `!!pb abort`
-- [ ] Prune DB backups
+- [x] Prune DB backups
 
 ## Documents
 

docs/config.md

@@ -839,7 +839,8 @@ Configurations for the SQLite database, used by Prime Backup
         "enabled": true,
         "interval": null,
         "crontab": "0 6 * * 0",
-        "jitter": "1m"
+        "jitter": "1m",
+        "max_amount": 100
     },
     "compact_pack": {
         "enabled": true,
@@ -867,6 +868,17 @@ within the [storage root](#storage_root) periodically, just in case something wr
 
 Database backups are stored with the `.tar.xz` format, and won't take up much space
 
+#### backup.max_amount
+
+The maximum number of database backup files to keep
+
+The backup job will delete old database backup files whose names match `db_backup_YYYYMMDD_HHMMSS.tar.xz`.
+It keeps the latest `max_amount` files by the timestamp parsed from the file name.
+Set `max_amount` to `0` or a negative value to keep all database backup files
+
+- Type: `int`
+- Default: `100`
+
 #### compact_pack
 
 The pack file compaction job

docs/config.zh.md

@@ -841,7 +841,8 @@ Prime Backup 所使用的 SQLite 数据库的相关配置
         "enabled": true,
         "interval": null,
         "crontab": "0 6 * * 0",
-        "jitter": "1m"
+        "jitter": "1m",
+        "max_amount": 100
     },
     "compact_pack": {
         "enabled": true,
@@ -869,6 +870,17 @@ Prime Backup 所使用的 SQLite 数据库的相关配置
 
 数据库备份将以 `.tar.xz` 格式存储，不会占用太多空间
 
+#### backup.max_amount
+
+要保留的数据库备份文件数量上限
+
+数据库备份作业会清理文件名匹配 `db_backup_YYYYMMDD_HHMMSS.tar.xz` 的旧数据库备份文件。
+它会根据文件名中解析出的时间保留最新的 `max_amount` 个文件。
+将 `max_amount` 设置为 `0` 或负数表示保留全部数据库备份文件
+
+- 类型：`int`
+- 默认值：`100`
+
 #### compact_pack
 
 打包文件整理作业

prime_backup/config/database_config.py

@@ -16,6 +16,7 @@ class BackUpDatabaseConfig(CrontabJobSetting):
 	interval = None
 	crontab = '0 6 * * 0'
 	jitter = Duration('1m')
+	max_amount: int = 100
 
 
 class CompactPackDatabaseConfig(CrontabJobSetting):

prime_backup/mcdr/task/db/create_db_backup_task.py

@@ -1,9 +1,12 @@
 import contextlib
+import dataclasses
+import datetime
+import re
 import tarfile
 import threading
 import time
 from pathlib import Path
-from typing import Optional
+from typing import Optional, List
 
 from typing_extensions import override
 
@@ -15,8 +18,15 @@
 from prime_backup.utils.run_once import RunOnceFunc
 
 
+@dataclasses.dataclass(frozen=True)
+class DbBackupFile:
+	path: Path
+	date: datetime.datetime
+
+
 class CreateDbBackupTask(HeavyTask[Optional[threading.Thread]]):
 	__task_sem = threading.Semaphore(1)
+	_db_backup_file_regex = re.compile(r'^db_backup_(?P<date>\d{8})_(?P<time>\d{6})\.tar\.xz$')
 
 	@property
 	@override
@@ -56,6 +66,11 @@ def tar_thread():
 					))
 				except Exception:
 					self.logger.exception('db backup: Compress database backup to {} failed'.format(db_backup_file))
+				else:
+					try:
+						self._delete_old_db_backup_files(db_backup_root)
+					except Exception:
+						self.logger.exception('db backup: Delete old database backups failed')
 				finally:
 					sem_releaser()
 					temp_db_path.unlink(missing_ok=True)
@@ -77,3 +92,44 @@ def tar_thread():
 		except Exception:
 			sem_releaser()
 			raise
+
+	@classmethod
+	def __parse_db_backup_file(cls, path: Path) -> Optional[DbBackupFile]:
+		if not path.is_file():
+			return None
+		if (match := cls._db_backup_file_regex.fullmatch(path.name)) is None:
+			return None
+
+		try:
+			date = datetime.datetime.strptime(match['date'] + match['time'], '%Y%m%d%H%M%S')
+		except ValueError:
+			return None
+		return DbBackupFile(path=path, date=date)
+
+	@classmethod
+	def _get_db_backup_files(cls, db_backup_root: Path) -> List[DbBackupFile]:
+		files: List[DbBackupFile] = []
+		for path in db_backup_root.iterdir():
+			if (backup_file := cls.__parse_db_backup_file(path)) is not None:
+				files.append(backup_file)
+		files.sort(key=lambda f: (f.date, f.path.name), reverse=True)
+		return files
+
+	def _delete_old_db_backup_files(self, db_backup_root: Path):
+		max_amount = self.config.database.backup.max_amount
+		if max_amount <= 0:
+			return
+
+		db_backup_files = self._get_db_backup_files(db_backup_root)
+		files_to_delete = db_backup_files[max_amount:]
+		if not files_to_delete:
+			return
+
+		self.logger.info('db backup: Deleting {} old database backup(s), keeping latest {}'.format(len(files_to_delete), max_amount))
+		for backup_file in files_to_delete:
+			try:
+				backup_file.path.unlink()
+			except Exception:
+				self.logger.exception('db backup: Failed to delete old database backup {}'.format(backup_file.path.as_posix()))
+			else:
+				self.logger.info('db backup: Deleted old database backup {}'.format(backup_file.path.as_posix()))