You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

166 lines
5.1 KiB
Python

import hashlib
import logging
from dataclasses import dataclass
from enum import Enum
from typing import List, Optional
from ... import redis
log = logging.getLogger(__name__)
import importlib # noqa: E402
import pkgutil # noqa: E402
class MigrationError(Exception):
pass
def import_submodules(root_module_name: str):
"""Import all submodules of a module, recursively."""
# TODO: Call this without specifying a module name, to import everything?
root_module = importlib.import_module(root_module_name)
if not hasattr(root_module, "__path__"):
raise MigrationError(
"The root module must be a Python package. "
f"You specified: {root_module_name}"
)
for loader, module_name, is_pkg in pkgutil.walk_packages(
root_module.__path__, root_module.__name__ + "." # type: ignore
):
importlib.import_module(module_name)
def schema_hash_key(index_name):
return f"{index_name}:hash"
async def create_index(conn: redis.Redis, index_name, schema, current_hash):
db_number = conn.connection_pool.connection_kwargs.get("db")
if db_number and db_number > 0:
raise MigrationError(
"Creating search indexes is only supported in database 0. "
f"You attempted to create an index in database {db_number}"
)
try:
await conn.execute_command(f"ft.info {index_name}")
except redis.ResponseError:
await conn.execute_command(f"ft.create {index_name} {schema}")
# TODO: remove "type: ignore" when type stubs will be fixed
await conn.set(schema_hash_key(index_name), current_hash) # type: ignore
else:
log.info("Index already exists, skipping. Index hash: %s", index_name)
class MigrationAction(Enum):
CREATE = 2
DROP = 1
@dataclass
class IndexMigration:
model_name: str
index_name: str
schema: str
hash: str
action: MigrationAction
conn: redis.Redis
previous_hash: Optional[str] = None
async def run(self):
if self.action is MigrationAction.CREATE:
await self.create()
elif self.action is MigrationAction.DROP:
await self.drop()
async def create(self):
try:
await create_index(self.conn, self.index_name, self.schema, self.hash)
except redis.ResponseError:
log.info("Index already exists: %s", self.index_name)
async def drop(self):
try:
await self.conn.execute_command(f"FT.DROPINDEX {self.index_name}")
except redis.ResponseError:
log.info("Index does not exist: %s", self.index_name)
class Migrator:
def __init__(self, module=None):
self.module = module
self.migrations: List[IndexMigration] = []
async def detect_migrations(self):
# Try to load any modules found under the given path or module name.
if self.module:
import_submodules(self.module)
# Import this at run-time to avoid triggering import-time side effects,
# e.g. checks for RedisJSON, etc.
from aredis_om.model.model import model_registry
for name, cls in model_registry.items():
hash_key = schema_hash_key(cls.Meta.index_name)
conn = cls.db()
try:
schema = cls.redisearch_schema()
except NotImplementedError:
log.info("Skipping migrations for %s", name)
continue
current_hash = hashlib.sha1(schema.encode("utf-8")).hexdigest() # nosec
try:
await conn.execute_command("ft.info", cls.Meta.index_name)
except redis.ResponseError:
self.migrations.append(
IndexMigration(
name,
cls.Meta.index_name,
schema,
current_hash,
MigrationAction.CREATE,
conn,
)
)
continue
stored_hash = await conn.get(hash_key)
schema_out_of_date = current_hash != stored_hash
if schema_out_of_date:
# TODO: Switch out schema with an alias to avoid downtime -- separate migration?
self.migrations.append(
IndexMigration(
name,
cls.Meta.index_name,
schema,
current_hash,
MigrationAction.DROP,
conn,
stored_hash,
)
)
self.migrations.append(
IndexMigration(
name,
cls.Meta.index_name,
schema,
current_hash,
MigrationAction.CREATE,
conn,
stored_hash,
)
)
async def run(self):
# TODO: Migration history
# TODO: Dry run with output
await self.detect_migrations()
for migration in self.migrations:
await migration.run()