alembic-offline is an extension for Alembic to enrich offline functionality of migrations using SQLAlchemy
Contents
alembic-offline introduces a helper which allows to implement phased migrations, e.g. those which steps are divided into logical phases. For example, you can have steps to be executed before code deploy and those after.
In your alembic config file (main section):
phases = before-deploy after-deploy final default-phase = after-deploy
In your version file:
from sqlalchemy import INTEGER, VARCHAR, NVARCHAR, TIMESTAMP, Column, func
from alembic import op
from alembic_offline import phased, execute_script
from tests.migrations.scripts import script
revision = '1'
down_revision = None
@phased
def upgrade():
op.create_table(
'account',
Column('id', INTEGER, primary_key=True),
Column('name', VARCHAR(50), nullable=False),
Column('description', NVARCHAR(200)),
Column('timestamp', TIMESTAMP, server_default=func.now())
)
yield
op.execute("update account set name='some'")
yield
execute_script(script.__file__)
def downgrade():
passWill give the sql output (for sqlite):
-- Running upgrade -> 1
-- PHASE::before-deploy::;
CREATE TABLE account (
id INTEGER NOT NULL,
name VARCHAR(50) NOT NULL,
description NVARCHAR(200),
timestamp TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
PRIMARY KEY (id)
);
-- PHASE::after-deploy::;
update account set name='some';
-- PHASE::final::;
-- SCRIPT::scripts/script.py::;
INSERT INTO alembic_version (version_num) VALUES ('1');As you see, phases are rendered as SQL comments to divide migration steps, so those who execute migration can see which phase's step it is. However, if migration procedure is highly customized, you can use alembic-offline API described below. get_migration_data returns migration phases in special form so you can automate their execution.
For complex migrations, it's not enough to execute sql, you might need some script to be executed instead. For that, there's special operation:
from alembic_offline import execute_script
def upgrade():
execute_script('scripts/script.py')If you'll get migration sql, it will be rendered as SQL comment:
-- SCRIPT::scripts/script.py::;For those who execute migrations it will be visible and they can execute the script manually. However, if migration procedure is highly customized, you can use alembic-offline API described below. get_migration_data returns script migration steps in special form so you can automate their execution. For online mode, the script will be executed as subprocess via python subprocess module.
alembic-offline provides specialized API to get certain migration data as dictionary:
from alembic_offline import get_migration_data
from alembic.config import Config
config = Config('path to alembic.ini')
data = get_migration_data(config, 'your-revision')
assert data == {
'revision': 'your-revision',
'phases': {
'after-deploy': [
{
'type': 'mysql',
'script': 'alter table account add column name VARCHAR(255)'
},
{
'type': 'python',
'script': 'from app.models import Session, Account; Session.add(Account()); Session.commit()',
'path': 'scripts/my_script.py'
},
]
}
}get_migration_data requires both phases and default-phase configuration options to be set. default-phase is needed to be able to get migration data even for simple migrations without phases.
alembic-offline provides an API call to get migration data for all revisions:
from alembic_offline import get_migrations_data
from alembic.config import Config
config = Config('path to alembic.ini')
data = get_migrations_data(config)
assert data == [
{
'revision': 'your-revision',
'phases': {
'after-deploy': [
{
'type': 'mysql',
'script': 'alter table account add column name VARCHAR(255)'
},
{
'type': 'python',
'script': 'from app.models import Session, Account; Session.add(Account()); Session.commit()',
'path': 'scripts/my_script.py'
},
]
}
}
]If you have questions, bug reports, suggestions, etc. please create an issue on the GitHub project page.
This software is licensed under the MIT license
© 2015 Anatoly Bubenkov, Paylogic International and others.