Add !!pb database convert_hash_method command to convert database

no need to pin on a hash method forever

lang/en_us.yml

@@ -141,6 +141,17 @@ prime_backup:
       pause.hover: Click to pause job {}
       resume: resume
       resume.hover: Click to resume job {}
+    db_convert_hash_method:
+      name: convert hash method
+      hash_method_unchanged: The hash method is already {}
+      missing_library:
+        Failed to import the target hasher, please make sure you have installed the required python library for {}.
+        Search hash_method in the document {} for more help
+      show_conversion: Prepare for hash method conversion, from {} to {}
+      confirm_target: convert
+      no_confirm: No choice has been made, convert hash method task aborted
+      aborted: Convert hash method task aborted
+      done: Converted the hash method from {} to {}
     db_overview:
       name: overview database
       title: Database overview
@@ -238,8 +249,10 @@ prime_backup:
           §7{prefix} database overview§r: Report an overview of the database
           §7{prefix} database validate §a<part>§r: Validate the correctness of contents in the database. Might take a long time
           §7{prefix} database vacuum§r: Compact the SQLite database manually, to reduce the size of the database file
+          §7{prefix} database convert_hash_method <hash_method>§r: Convert the currently used hash method to another. This will affect all data
           {scheduled_compact_notes}
           §d[Arguments]§r
+          §d<hash_method>§r: Available options: {hash_methods}
           §a<part>§r: 
           - §ablobs§r: Validate the correctness of blobs, e.g. data size, hash value
           - §afiles§r: Validate the correctness of file objects, e.g. the association between files and blobs

lang/zh_cn.yml

@@ -141,6 +141,17 @@ prime_backup:
       pause.hover: 点击以暂停运行作业{}
       resume: 继续
       resume.hover: 点击以继续运行作业{}
+    db_convert_hash_method:
+      name: 哈希算法转换
+      hash_method_unchanged: 哈希算法已经是{}了
+      missing_library:
+        无法导入目标哈希算法, 请确保你已经安装了算法{}所需的Python依赖库。
+        在文档{}中搜索hash_method以获得更多帮助
+      show_conversion: 准备把哈希算法从{}转换为{}
+      confirm_target: 转换
+      no_confirm: 未做出选择, 哈希算法转换任务中止
+      aborted: 哈希算法转换任务中止
+      done: 已将哈希算法从{}转换为{}
     db_overview:
       name: 概览数据库
       title: 数据库概览
@@ -238,8 +249,10 @@ prime_backup:
           §7{prefix} database overview§r: 查看数据库信息概览
           §7{prefix} database validate §a<组件>§r: 验证数据库内容的正确性。耗时可能较长
           §7{prefix} database vacuum§r: 手动执行SQLite数据库的精简操作，减少数据库文件的体积
+          §7{prefix} database convert_hash_method <哈希算法>§r: 转换当前使用的哈希算法为另一种算法。这将影响所有的数据
           {scheduled_compact_notes}
           §d【参数帮助】§r
+          §d<哈希算法>§r: 可以选项: {hash_methods}
           §a<组件>§r: 
           - §ablobs§r: 验证数据对象的正确性，如数据大小、哈希值
           - §afiles§r: 验证文件对象的正确性，如文件与数据的关联

prime_backup/action/convert_hash_method_action.py

@@ -0,0 +1,98 @@
+import shutil
+import time
+from pathlib import Path
+from typing import List, Dict, Set
+
+from prime_backup.action import Action
+from prime_backup.compressors import Compressor
+from prime_backup.db.access import DbAccess
+from prime_backup.db.session import DbSession
+from prime_backup.exceptions import PrimeBackupError
+from prime_backup.types.hash_method import HashMethod
+from prime_backup.utils import blob_utils, hash_utils, collection_utils
+
+
+class HashCollisionError(PrimeBackupError):
+	"""
+	Same hash value, between 2 hash methods
+	"""
+	pass
+
+
+class ConvertHashMethodAction(Action[None]):
+	def __init__(self, new_hash_method: HashMethod):
+		super().__init__()
+		self.new_hash_method = new_hash_method
+
+	def __convert_blobs(self, session: DbSession, blob_hashes: List[str], old_hashes: Set[str], processed_hash_mapping: Dict[str, str]):
+		hash_mapping: Dict[str, str] = {}
+		blobs = list(session.get_blobs(blob_hashes).values())
+
+		# calc blob hashes
+		for blob in blobs:
+			blob_path = blob_utils.get_blob_path(blob.hash)
+			with Compressor.create(blob.compress).open_decompressed(blob_path) as f:
+				sah = hash_utils.calc_reader_size_and_hash(f, hash_method=self.new_hash_method)
+			hash_mapping[blob.hash] = sah.hash
+			if sah.hash in old_hashes:
+				raise HashCollisionError(sah.hash)
+
+		# update the objects
+		for blob in blobs:
+			old_hash, new_hash = blob.hash, hash_mapping[blob.hash]
+			old_path = blob_utils.get_blob_path(old_hash)
+			new_path = blob_utils.get_blob_path(new_hash)
+			old_path.rename(new_path)
+
+			processed_hash_mapping[old_hash] = new_hash
+			blob.hash = new_hash
+
+		for file in session.get_file_by_blob_hashes(list(hash_mapping.keys())):
+			file.blob_hash = hash_mapping[file.blob_hash]
+
+	def __replace_blob_store(self, old_store: Path, new_store: Path):
+		trash_bin = self.config.storage_path / 'temp' / 'old_blobs'
+		trash_bin.parent.mkdir(parents=True, exist_ok=True)
+
+		old_store.rename(trash_bin)
+		new_store.rename(old_store)
+		shutil.rmtree(trash_bin)
+
+	def run(self):
+		processed_hash_mapping: Dict[str, str] = {}  # old -> new
+		try:
+			t = time.time()
+			with DbAccess.open_session() as session:
+				meta = session.get_db_meta()
+				if meta.hash_method == self.new_hash_method.name:
+					self.logger.info('Hash method of the database is already {}, no need to convert'.format(self.new_hash_method.name))
+					return
+
+				total_blob_count = session.get_blob_count()
+				all_hashes = session.get_all_blob_hashes()
+				all_hash_set = set(all_hashes)
+				cnt = 0
+				for blob_hashes in collection_utils.slicing_iterate(all_hashes, 1000):
+					blob_hashes: List[str] = list(blob_hashes)
+					cnt += len(blob_hashes)
+					self.logger.info('Converting blobs {} / {}'.format(cnt, total_blob_count))
+
+					self.__convert_blobs(session, blob_hashes, all_hash_set, processed_hash_mapping)
+					session.flush_and_expunge_all()
+
+				meta = session.get_db_meta()  # get the meta again, cuz expunge_all() was called
+				meta.hash_method = self.new_hash_method.name
+
+			self.logger.info('Syncing config and variables')
+			DbAccess.sync_hash_method()
+			self.config.backup.hash_method = self.new_hash_method.name
+
+			self.logger.info('Conversion done, cost {}s'.format(round(time.time() - t, 2)))
+
+		except Exception:
+			self.logger.info('Error occurs during convertion, applying rollback')
+			for old_hash, new_hash in processed_hash_mapping.items():
+				old_path = blob_utils.get_blob_path(old_hash)
+				new_path = blob_utils.get_blob_path(new_hash)
+				new_path.rename(old_path)
+			raise

prime_backup/action/export_backup_action.py

@@ -17,18 +17,14 @@
 from prime_backup.db import schema
 from prime_backup.db.access import DbAccess
 from prime_backup.db.session import DbSession
-from prime_backup.exceptions import PrimeBackupError
+from prime_backup.exceptions import PrimeBackupError, VerificationError
 from prime_backup.types.backup_meta import BackupMeta
 from prime_backup.types.export_failure import ExportFailures
 from prime_backup.types.tar_format import TarFormat
 from prime_backup.utils import file_utils, blob_utils, misc_utils, hash_utils
 from prime_backup.utils.bypass_io import BypassReader
 
 
-class VerificationError(PrimeBackupError):
-	pass
-
-
 class _ExportInterrupted(PrimeBackupError):
 	pass
 

prime_backup/config/config.py

@@ -35,11 +35,11 @@ def get(cls) -> 'Config':
 			return cls.__get_default()
 		return _config
 
-	@functools.cached_property
+	@property
 	def storage_path(self) -> Path:
 		return Path(self.storage_root)
 
-	@functools.cached_property
+	@property
 	def source_path(self) -> Path:
 		return Path(self.backup.source_root)
 

prime_backup/db/access.py

@@ -39,19 +39,23 @@ def init(cls, auto_migrate: bool = True):
 		else:
 			migration.ensure_version()
 
-		with cls.open_session() as session:
-			hash_method_str = str(session.get_db_meta().hash_method)
-		try:
-			cls.__hash_method = HashMethod[hash_method_str]
-		except KeyError:
-			raise ValueError('invalid hash method {!r} in db meta'.format(hash_method_str)) from None
+		cls.sync_hash_method()
 
 	@classmethod
 	def shutdown(cls):
 		if (logger := db_logger.get()) is not None:
 			for hdr in list(logger.handlers):
 				logger.removeHandler(hdr)
 
+	@classmethod
+	def sync_hash_method(cls):
+		with cls.open_session() as session:
+			hash_method_str = str(session.get_db_meta().hash_method)
+		try:
+			cls.__hash_method = HashMethod[hash_method_str]
+		except KeyError:
+			raise ValueError('invalid hash method {!r} in db meta'.format(hash_method_str)) from None
+
 	@classmethod
 	def __ensure_not_none(cls, value):
 		if value is None:

prime_backup/db/session.py

@@ -1,6 +1,6 @@
 import contextlib
 import time
-from typing import Optional, Sequence, Dict, ContextManager
+from typing import Optional, Sequence, Dict, ContextManager, Iterator
 from typing import TypeVar, List
 
 from sqlalchemy import select, delete, desc, func, Select, JSON, text
@@ -40,9 +40,22 @@ def __init__(self, session: Session):
 	def add(self, obj: schema.Base):
 		self.session.add(obj)
 
+	def expunge(self, obj: schema.Base):
+		self.session.expunge(obj)
+
+	def expunge_all(self):
+		self.session.expunge_all()
+
 	def flush(self):
 		self.session.flush()
 
+	def flush_and_expunge_all(self):
+		self.flush()
+		self.expunge_all()
+
+	def commit(self):
+		self.session.commit()
+
 	@contextlib.contextmanager
 	def no_auto_flush(self) -> ContextManager[None]:
 		with self.session.no_autoflush:
@@ -94,6 +107,14 @@ def list_blobs(self, limit: Optional[int] = None, offset: Optional[int] = None)
 			s = s.offset(offset)
 		return _list_it(self.session.execute(s).scalars().all())
 
+	def iterate_blob_batch(self, *, batch_size: int = 3000) -> Iterator[List[schema.Blob]]:
+		limit, offset = batch_size, 0
+		while True:
+			blobs = self.list_blobs(limit=limit, offset=offset)
+			if len(blobs) == 0:
+				break
+			yield blobs
+
 	def get_all_blob_hashes(self) -> List[str]:
 		return _list_it(self.session.execute(select(schema.Blob.hash)).scalars().all())
 
@@ -161,6 +182,16 @@ def get_file(self, backup_id: int, path: str) -> schema.File:
 	def get_file_raw_size_sum(self) -> int:
 		return _int_or_0(self.session.execute(func.sum(schema.File.blob_raw_size).select()).scalar_one())
 
+	def get_file_by_blob_hashes(self, hashes: List[str]) -> List[schema.File]:
+		hashes = collection_utils.deduplicated_list(hashes)
+		result = []
+		for view in collection_utils.slicing_iterate(hashes, self.__safe_var_limit):
+			result.extend(self.session.execute(
+				select(schema.File).
+				where(schema.File.blob_hash.in_(view))
+			).scalars().all())
+		return result
+
 	def get_file_count_by_blob_hashes(self, hashes: List[str]) -> int:
 		cnt = 0
 		for view in collection_utils.slicing_iterate(hashes, self.__safe_var_limit):
@@ -179,6 +210,14 @@ def list_files(self, limit: Optional[int] = None, offset: Optional[int] = None)
 			s = s.offset(offset)
 		return _list_it(self.session.execute(s).scalars().all())
 
+	def iterate_file_batch(self, *, batch_size: int = 3000) -> Iterator[List[schema.File]]:
+		limit, offset = batch_size, 0
+		while True:
+			files = self.list_files(limit=limit, offset=offset)
+			if len(files) == 0:
+				break
+			yield files
+
 	def delete_file(self, file: schema.File):
 		self.session.delete(file)
 

prime_backup/exceptions.py

@@ -18,3 +18,7 @@ def __init__(self, backup_id: int, path: str):
 class UnsupportedFileFormat(PrimeBackupError):
 	def __init__(self, mode: int):
 		self.mode = mode
+
+
+class VerificationError(PrimeBackupError):
+	pass

prime_backup/mcdr/command/commands.py

@@ -23,6 +23,7 @@
 from prime_backup.mcdr.task.crontab.list_crontab_task import ListCrontabJobTask
 from prime_backup.mcdr.task.crontab.operate_crontab_task import OperateCrontabJobTask
 from prime_backup.mcdr.task.crontab.show_crontab_task import ShowCrontabJobTask
+from prime_backup.mcdr.task.db.convert_hash_method_task import ConvertHashMethodTask
 from prime_backup.mcdr.task.db.show_db_overview_task import ShowDbOverviewTask
 from prime_backup.mcdr.task.db.vacuum_sqlite_task import VacuumSqliteTask
 from prime_backup.mcdr.task.db.validate_db_task import ValidateDbTask, ValidateParts
@@ -31,6 +32,7 @@
 from prime_backup.mcdr.task_manager import TaskManager
 from prime_backup.types.backup_filter import BackupFilter
 from prime_backup.types.backup_tags import BackupTagName
+from prime_backup.types.hash_method import HashMethod
 from prime_backup.types.operator import Operator
 from prime_backup.types.standalone_backup_format import StandaloneBackupFormat
 from prime_backup.utils import misc_utils
@@ -72,6 +74,10 @@ def cmd_db_validate(self, source: CommandSource, _: CommandContext, parts: Valid
 	def cmd_db_vacuum(self, source: CommandSource, _: CommandContext):
 		self.task_manager.add_task(VacuumSqliteTask(source))
 
+	def cmd_db_convert_hash_method(self, source: CommandSource, context: CommandContext):
+		new_hash_method = context['hash_method']
+		self.task_manager.add_task(ConvertHashMethodTask(source, new_hash_method))
+
 	def cmd_make(self, source: CommandSource, context: CommandContext):
 		def callback(_, err):
 			if err is None:
@@ -217,9 +223,13 @@ def create_backup_id(arg: str = 'backup_id', clazz: Type[Integer] = Integer) ->
 
 		builder = SimpleCommandBuilder()
 
+		# help
+
 		builder.command('help', self.cmd_help)
 		builder.command('help <what>', self.cmd_help)
 
+		builder.arg('what', Text).suggests(lambda: ShowHelpTask.COMMANDS_WITH_DETAILED_HELP)
+
 		# backup
 		builder.command('make', self.cmd_make)
 		builder.command('make <comment>', self.cmd_make)
@@ -228,32 +238,33 @@ def create_backup_id(arg: str = 'backup_id', clazz: Type[Integer] = Integer) ->
 		builder.command('delete_range <backup_id_range>', self.cmd_delete_range)
 		builder.command('prune', self.cmd_prune)
 
+		builder.arg('backup_id', create_backup_id)
+		builder.arg('backup_id_range', IdRangeNode)
+		builder.arg('comment', GreedyText)
+
 		# crontab
 		builder.command('crontab', self.cmd_crontab_show)
 		builder.command('crontab <job_id>', self.cmd_crontab_show)
 		builder.command('crontab <job_id> pause', self.cmd_crontab_pause)
 		builder.command('crontab <job_id> resume', self.cmd_crontab_resume)
 
+		builder.arg('job_id', lambda n: Enumeration(n, CrontabJobId))
+
 		# db
 		builder.command('database overview', self.cmd_db_overview)
 		builder.command('database validate all', functools.partial(self.cmd_db_validate, parts=ValidateParts.all()))
 		builder.command('database validate blobs', functools.partial(self.cmd_db_validate, parts=ValidateParts.blobs))
 		builder.command('database validate files', functools.partial(self.cmd_db_validate, parts=ValidateParts.files))
 		builder.command('database vacuum', self.cmd_db_vacuum)
+		builder.command('database convert_hash_method <hash_method>', self.cmd_db_convert_hash_method)
+
+		builder.arg('hash_method', lambda n: Enumeration(n, HashMethod))
 
 		# operations
 		builder.command('confirm', self.cmd_confirm)
 		builder.command('abort', self.cmd_abort)
 
-		# node defs
-		builder.arg('backup_id', create_backup_id)
-		builder.arg('backup_id_range', IdRangeNode)
-		builder.arg('comment', GreedyText)
-		builder.arg('job_id', lambda n: Enumeration(n, CrontabJobId))
-		builder.arg('page', lambda n: Integer(n).at_min(1))
-		builder.arg('per_page', lambda n: Integer(n).at_min(1))
-		builder.arg('what', Text).suggests(lambda: ShowHelpTask.COMMANDS_WITH_DETAILED_HELP)
-
+		# subcommand permissions
 		for name, level in permissions.items():
 			builder.literal(name).requires(get_permission_checker(name), get_permission_denied_text)