¡@
OpenStack Study: schema.py**** CubicPower OpenStack Study **** def from_migration_import(module_name, fromlist): """ Import a migration file and return the module :param module_name: name of migration module to import from (ex: 001_add_images_table) :param fromlist: list of items to import (ex: define_images_table) :retval: module object This bit of ugliness warrants an explanation: As you're writing migrations, you'll frequently want to refer to tables defined in previous migrations. In the interest of not repeating yourself, you need a way of importing that table into a 'future' migration. However, tables are bound to metadata, so what you need to import is really a table factory, which you can late-bind to your current metadata object. Moreover, migrations begin with a number (001...), which means they aren't valid Python identifiers. This means we can't perform a 'normal' import on them (the Python lexer will 'splode). Instead, we need to use __import__ magic to bring the table-factory into our namespace. Example Usage: (define_images_table,) = from_migration_import( '001_add_images_table', ['define_images_table']) images = define_images_table(meta) # Refer to images table """ module_path = 'glance.db.sqlalchemy.migrate_repo.versions.%s' % module_name module = __import__(module_path, globals(), locals(), fromlist, -1) return [getattr(module, item) for item in fromlist]
**** CubicPower OpenStack Study **** def create_tables(tables): for table in tables: LOG.info(_("creating table %(table)s") % {'table': table}) table.create()
**** CubicPower OpenStack Study **** def drop_tables(tables): for table in tables: LOG.info(_("dropping table %(table)s") % {'table': table}) table.drop()
|