Goal

Raw-sql-migrate is tool for managing your raw SQL migrations.

Config

To configure database connection you should create a config file in folder, where you will run the migration tool. The config file can be of .yaml or .py type and by default named rsm, but it is not necessary condition, because every command of rsm accepts –config parameter where you can pass a path to config file. If path to file is not specified, then rsm.py and rsm.yaml will be searched in the run directory.

Examples of config files: rsm.yaml:

database:
    engine: engine backend module
    host: database host
    port: database port
    name: database name
    user: user name
    password: user password
    additional_driver_param1: value
    additional_driver_param2: value
history_table_name: migration history table name
packages:
    - package_a
    - package_b
    - package_c.package_d

rsm.py:

RSM_CONFIG = {
    'database': {
        'engine': engine backend module,
        'host': database host,
        'port': database port,
        'name': database name,
        'user': user name,
        'password': user password,
    },
    'history_table_name': migration history table name,
    'packages': [
        'package_a',
        'package_b',
        'package_c.package_d',
    ],
}

Available options of ‘engine’ are:

• raw_sql_migrate.engines.postgresql_psycopg2 (requires psycopg2 package)
• raw_sql_migrate.engines.mysql (requires MySQLdb-python package)

Also you can pass specific params to driver connect method, just add them to config database section. Packages param is a list of packages where to search for new migrations.

Usage

Creating new migration

In order to create new migration just call create command

rsm create package_a.package_b initial

Calling it will create new migrations history table, migrations directory in the package_b and 0001_initial.py migration file

Migrating forward

In order to migrate forward call

rsm migrate package 0003

Note: to migrate all not applied migrations you should skip migration_number param.

Migrating backward

In order to migrate backward call

rsm migrate package 0001

Note: to migrate to initial state you should pass migration_number as 0.

Migrations status

To get latest migration data for tracked packages call method:

rsm status package_a.package_b

To get info for all tracked packages just omit package name attribute.

package                                  name                                     processed_at
------------------------------------------------------------------------------------------------------------------------
package_a.package_b                      0001_initial                             2015-06-25 23:06:56.698562

Squashing migrations

Sometimes there can be situation when you want to merge your dev migrations before publishing them to repository. In this case use squash method:

rsm squash package_a.package_b 0042 squashed_migration

This example does next things: It searches for not applied migration in package begining from number 42, reads their content and appends it to result forward and backward functions. After all migrations were processed command writes new migration file with ‘begin_from’ number and renames squashed migrations with ‘_squashed’ prefix. Note that command can’t squash already applied migrations.

Transaction Control

Each migration runs in separate transaction, which will be started when first sql is executed and committed when all the code in forwardbackward functions is executed. If there is an exception during migrate function all changes will be rolled back.