mediagoblin/mediagoblin/db/migration_tools.py
Jessica Tallon 2ddebb97bf Fix #5391 - Alembic migrations would only work for SQLite
The database connection was being set in a alembic.ini config file, if the user
had specified postgres the "sqlite" connection URL in alembic.ini would override
that. We probably should look into this more so i've opened #5395.
2016-01-18 09:39:34 +00:00

400 lines
14 KiB
Python

# GNU MediaGoblin -- federated, autonomous media hosting
# Copyright (C) 2011, 2012 MediaGoblin contributors. See AUTHORS.
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU Affero General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
from __future__ import unicode_literals
import logging
import os
from alembic import command
from alembic.config import Config
from alembic.migration import MigrationContext
from mediagoblin.db.base import Base
from mediagoblin.tools.common import simple_printer
from sqlalchemy import Table
from sqlalchemy.sql import select
log = logging.getLogger(__name__)
class TableAlreadyExists(Exception):
pass
class AlembicMigrationManager(object):
def __init__(self, session):
root_dir = os.path.abspath(os.path.dirname(os.path.dirname(
os.path.dirname(__file__))))
alembic_cfg_path = os.path.join(root_dir, 'alembic.ini')
self.alembic_cfg = Config(alembic_cfg_path)
self.alembic_cfg.set_main_option("sqlalchemy.url", str(session.get_bind().url))
self.session = session
def get_current_revision(self):
context = MigrationContext.configure(self.session.bind)
return context.get_current_revision()
def upgrade(self, version):
return command.upgrade(self.alembic_cfg, version or 'head')
def downgrade(self, version):
if isinstance(version, int) or version is None or version.isdigit():
version = 'base'
return command.downgrade(self.alembic_cfg, version)
def stamp(self, revision):
return command.stamp(self.alembic_cfg, revision=revision)
def init_tables(self):
Base.metadata.create_all(self.session.bind)
# load the Alembic configuration and generate the
# version table, "stamping" it with the most recent rev:
# XXX: we need to find a better way to detect current installations
# using sqlalchemy-migrate because we don't have to create all table
# for them
command.stamp(self.alembic_cfg, 'head')
def init_or_migrate(self, version=None):
# XXX: we need to call this method when we ditch
# sqlalchemy-migrate entirely
# if self.get_current_revision() is None:
# self.init_tables()
self.upgrade(version)
class MigrationManager(object):
"""
Migration handling tool.
Takes information about a database, lets you update the database
to the latest migrations, etc.
"""
def __init__(self, name, models, foundations, migration_registry, session,
printer=simple_printer):
"""
Args:
- name: identifier of this section of the database
- session: session we're going to migrate
- migration_registry: where we should find all migrations to
run
"""
self.name = name
self.models = models
self.foundations = foundations
self.session = session
self.migration_registry = migration_registry
self._sorted_migrations = None
self.printer = printer
# For convenience
from mediagoblin.db.models import MigrationData
self.migration_model = MigrationData
self.migration_table = MigrationData.__table__
@property
def sorted_migrations(self):
"""
Sort migrations if necessary and store in self._sorted_migrations
"""
if not self._sorted_migrations:
self._sorted_migrations = sorted(
self.migration_registry.items(),
# sort on the key... the migration number
key=lambda migration_tuple: migration_tuple[0])
return self._sorted_migrations
@property
def migration_data(self):
"""
Get the migration row associated with this object, if any.
"""
return self.session.query(
self.migration_model).filter_by(name=self.name).first()
@property
def latest_migration(self):
"""
Return a migration number for the latest migration, or 0 if
there are no migrations.
"""
if self.sorted_migrations:
return self.sorted_migrations[-1][0]
else:
# If no migrations have been set, we start at 0.
return 0
@property
def database_current_migration(self):
"""
Return the current migration in the database.
"""
# If the table doesn't even exist, return None.
if not self.migration_table.exists(self.session.bind):
return None
# Also return None if self.migration_data is None.
if self.migration_data is None:
return None
return self.migration_data.version
def set_current_migration(self, migration_number=None):
"""
Set the migration in the database to migration_number
(or, the latest available)
"""
self.migration_data.version = migration_number or self.latest_migration
self.session.commit()
def migrations_to_run(self):
"""
Get a list of migrations to run still, if any.
Note that this will fail if there's no migration record for
this class!
"""
assert self.database_current_migration is not None
db_current_migration = self.database_current_migration
return [
(migration_number, migration_func)
for migration_number, migration_func in self.sorted_migrations
if migration_number > db_current_migration]
def init_tables(self):
"""
Create all tables relative to this package
"""
# sanity check before we proceed, none of these should be created
for model in self.models:
# Maybe in the future just print out a "Yikes!" or something?
if model.__table__.exists(self.session.bind):
raise TableAlreadyExists(
u"Intended to create table '%s' but it already exists" %
model.__table__.name)
self.migration_model.metadata.create_all(
self.session.bind,
tables=[model.__table__ for model in self.models])
def populate_table_foundations(self):
"""
Create the table foundations (default rows) as layed out in FOUNDATIONS
in mediagoblin.db.models
"""
for Model, rows in self.foundations.items():
self.printer(u' + Laying foundations for %s table\n' %
(Model.__name__))
for parameters in rows:
new_row = Model(**parameters)
self.session.add(new_row)
def create_new_migration_record(self):
"""
Create a new migration record for this migration set
"""
migration_record = self.migration_model(
name=self.name,
version=self.latest_migration)
self.session.add(migration_record)
self.session.commit()
def dry_run(self):
"""
Print out a dry run of what we would have upgraded.
"""
if self.database_current_migration is None:
self.printer(
u'~> Woulda initialized: %s\n' % self.name_for_printing())
return u'inited'
migrations_to_run = self.migrations_to_run()
if migrations_to_run:
self.printer(
u'~> Woulda updated %s:\n' % self.name_for_printing())
for migration_number, migration_func in migrations_to_run():
self.printer(
u' + Would update %s, "%s"\n' % (
migration_number, migration_func.func_name))
return u'migrated'
def name_for_printing(self):
if self.name == u'__main__':
return u"main mediagoblin tables"
else:
return u'plugin "%s"' % self.name
def init_or_migrate(self):
"""
Initialize the database or migrate if appropriate.
Returns information about whether or not we initialized
('inited'), migrated ('migrated'), or did nothing (None)
"""
assure_migrations_table_setup(self.session)
# Find out what migration number, if any, this database data is at,
# and what the latest is.
migration_number = self.database_current_migration
# Is this our first time? Is there even a table entry for
# this identifier?
# If so:
# - create all tables
# - create record in migrations registry
# - print / inform the user
# - return 'inited'
if migration_number is None:
self.printer(u"-> Initializing %s... " % self.name_for_printing())
self.init_tables()
# auto-set at latest migration number
self.create_new_migration_record()
self.printer(u"done.\n")
self.populate_table_foundations()
self.set_current_migration()
return u'inited'
# Run migrations, if appropriate.
migrations_to_run = self.migrations_to_run()
if migrations_to_run:
self.printer(
u'-> Updating %s:\n' % self.name_for_printing())
for migration_number, migration_func in migrations_to_run:
self.printer(
u' + Running migration %s, "%s"... ' % (
migration_number, migration_func.__name__))
migration_func(self.session)
self.set_current_migration(migration_number)
self.printer('done.\n')
return u'migrated'
# Otherwise return None. Well it would do this anyway, but
# for clarity... ;)
return None
class RegisterMigration(object):
"""
Tool for registering migrations
Call like:
@RegisterMigration(33)
def update_dwarves(database):
[...]
This will register your migration with the default migration
registry. Alternately, to specify a very specific
migration_registry, you can pass in that as the second argument.
Note, the number of your migration should NEVER be 0 or less than
0. 0 is the default "no migrations" state!
"""
def __init__(self, migration_number, migration_registry):
assert migration_number > 0, "Migration number must be > 0!"
assert migration_number not in migration_registry, \
"Duplicate migration numbers detected! That's not allowed!"
assert migration_number <= 44, ('Alembic should be used for '
'new migrations')
self.migration_number = migration_number
self.migration_registry = migration_registry
def __call__(self, migration):
self.migration_registry[self.migration_number] = migration
return migration
def assure_migrations_table_setup(db):
"""
Make sure the migrations table is set up in the database.
"""
from mediagoblin.db.models import MigrationData
if not MigrationData.__table__.exists(db.bind):
MigrationData.metadata.create_all(
db.bind, tables=[MigrationData.__table__])
def inspect_table(metadata, table_name):
"""Simple helper to get a ref to an already existing table"""
return Table(table_name, metadata, autoload=True,
autoload_with=metadata.bind)
def replace_table_hack(db, old_table, replacement_table):
"""
A function to fully replace a current table with a new one for migrati-
-ons. This is necessary because some changes are made tricky in some situa-
-tion, for example, dropping a boolean column in sqlite is impossible w/o
this method
:param old_table A ref to the old table, gotten through
inspect_table
:param replacement_table A ref to the new table, gotten through
inspect_table
Users are encouraged to sqlalchemy-migrate replace table solutions, unless
that is not possible... in which case, this solution works,
at least for sqlite.
"""
surviving_columns = replacement_table.columns.keys()
old_table_name = old_table.name
for row in db.execute(select(
[column for column in old_table.columns
if column.name in surviving_columns])):
db.execute(replacement_table.insert().values(**row))
db.commit()
old_table.drop()
db.commit()
replacement_table.rename(old_table_name)
db.commit()
def model_iteration_hack(db, query):
"""
This will return either the query you gave if it's postgres or in the case
of sqlite it will return a list with all the results. This is because in
migrations it seems sqlite can't deal with concurrent quries so if you're
iterating over models and doing a commit inside the loop, you will run into
an exception which says you've closed the connection on your iteration
query. This fixes it.
NB: This loads all of the query reuslts into memeory, there isn't a good
way around this, we're assuming sqlite users have small databases.
"""
# If it's SQLite just return all the objects
if db.bind.url.drivername == "sqlite":
return [obj for obj in db.execute(query)]
# Postgres return the query as it knows how to deal with it.
return db.execute(query)