feat: implement backup restoration from old db schemas (#2213)

* Remove some implicit lazy-loads from user serialization

* implement full backup restore across different database versions
This commit is contained in:
Sören
2023-03-12 21:39:51 +01:00
committed by GitHub
parent 3118b0e423
commit ccb0b43cef
5 changed files with 71 additions and 90 deletions

View File

@@ -1,15 +1,19 @@
import datetime
import json
from os import path
from pathlib import Path
from fastapi.encoders import jsonable_encoder
from pydantic import BaseModel
from sqlalchemy import MetaData, create_engine, insert, text
from sqlalchemy import ForeignKeyConstraint, MetaData, create_engine, insert, text
from sqlalchemy.engine import base
from sqlalchemy.orm import sessionmaker
from alembic import command
from alembic.config import Config
from mealie.services._base_service import BaseService
PROJECT_DIR = Path(__file__).parent.parent.parent.parent
class AlchemyExporter(BaseService):
connection_str: str
@@ -53,33 +57,6 @@ class AlchemyExporter(BaseService):
data[key] = AlchemyExporter.DateTimeParser(time=value).time
return data
@staticmethod
def _compare_schemas(schema1: dict, schema2: dict) -> bool:
try:
# validate alembic version(s) are the same
return schema1["alembic_version"] == schema2["alembic_version"]
except KeyError:
return False
@staticmethod
def validate_schemas(schema1: Path | dict, schema2: Path | dict) -> bool:
"""
Validates that the schema of the database matches the schema of the database. In practice,
this means validating that the alembic version is the same
"""
def extract_json(file: Path) -> dict:
with open(file) as f:
return json.loads(f.read())
if isinstance(schema1, Path):
schema1 = extract_json(schema1)
if isinstance(schema2, Path):
schema2 = extract_json(schema2)
return AlchemyExporter._compare_schemas(schema1, schema2)
def dump_schema(self) -> dict:
"""
Returns the schema of the SQLAlchemy database as a python dictionary. This dictionary is wrapped by
@@ -115,6 +92,17 @@ class AlchemyExporter(BaseService):
return jsonable_encoder(result)
def restore(self, db_dump: dict) -> None:
# setup alembic to run migrations up the version of the backup
alembic_data = db_dump["alembic_version"]
alembic_version = alembic_data[0]["version_num"]
alembic_cfg = Config(str(PROJECT_DIR / "alembic.ini"))
# alembic's file resolver wants to use the "mealie" subdirectory when called from within the server package
# Just override this to use the correct migrations path
alembic_cfg.set_main_option("script_location", path.join(PROJECT_DIR, "alembic"))
command.upgrade(alembic_cfg, alembic_version)
del db_dump["alembic_version"]
"""Restores all data from dictionary into the database"""
with self.engine.begin() as connection:
data = AlchemyExporter.convert_to_datetime(db_dump)
@@ -123,8 +111,8 @@ class AlchemyExporter(BaseService):
for table_name, rows in data.items():
if not rows:
continue
table = self.meta.tables[table_name]
connection.execute(table.delete())
connection.execute(insert(table), rows)
if self.engine.dialect.name == "postgresql":
@@ -151,19 +139,45 @@ SELECT SETVAL('shopping_list_item_extras_id_seq', (SELECT MAX(id) FROM shopping_
)
)
# Run all migrations up to current version
command.upgrade(alembic_cfg, "head")
def drop_all(self) -> None:
"""Drops all data from the database"""
self.meta.reflect(bind=self.engine)
with self.session_maker() as session:
is_postgres = self.settings.DB_ENGINE == "postgres"
from sqlalchemy.engine.reflection import Inspector
from sqlalchemy.schema import DropConstraint, DropTable, MetaData, Table
try:
if is_postgres:
session.execute(text("SET session_replication_role = 'replica'"))
with self.engine.begin() as connection:
inspector = Inspector.from_engine(self.engine)
for table in self.meta.sorted_tables:
session.execute(text(f"DELETE FROM {table.name}"))
finally:
if is_postgres:
session.execute(text("SET session_replication_role = 'origin'"))
session.commit()
# We need to re-create a minimal metadata with only the required things to
# successfully emit drop constraints and tables commands for postgres (based
# on the actual schema of the running instance)
meta = MetaData()
tables = []
all_fkeys = []
for table_name in inspector.get_table_names():
fkeys = []
for fkey in inspector.get_foreign_keys(table_name):
if not fkey["name"]:
continue
fkeys.append(ForeignKeyConstraint((), (), name=fkey["name"]))
tables.append(Table(table_name, meta, *fkeys))
all_fkeys.extend(fkeys)
if self.engine.dialect.name == "postgresql":
# Only pg needs foreign key dropping
for fkey in all_fkeys:
connection.execute(DropConstraint(fkey))
for table in tables:
connection.execute(DropTable(table))
# I have no idea how to drop all custom types with sqlalchemy
# Since we only have one, this will have to do for now
connection.execute(text("DROP TYPE authmethod"))
else:
for table in tables:
connection.execute(DropTable(table))

View File

@@ -89,10 +89,6 @@ class BackupV2(BaseService):
database_json = contents.read_tables()
if not AlchemyExporter.validate_schemas(database_json, self.db_exporter.dump()):
self.logger.error("Invalid backup file. Database schemas do not match")
raise BackupSchemaMismatch("Invalid backup file. Database schemas do not match")
# ================================
# Purge Database