Setup alembic to work properly with PostgreSQL schemas

env.py

from __future__ import with_statement
from alembic import context
from sqlalchemy import engine_from_config, pool
from logging.config import fileConfig

from models import Base

config = context.config

fileConfig(config.config_file_name)

"""
Load models metadata. We should define schema in this class firstly, 
or set schema implicit with `__table_args__ = {'schema' : 'test'}` in model class
"""
target_metadata = Base.metadata


def run_migrations_offline():
    url = config.get_main_option("sqlalchemy.url")
    context.configure(
        url=url, target_metadata=target_metadata, literal_binds=True)

    with context.begin_transaction():
        context.run_migrations()


def run_migrations_online():
    connectable = engine_from_config(
        config.get_section(config.config_ini_section),
        prefix='sqlalchemy.',
        poolclass=pool.NullPool)

    with connectable.connect() as connection:
        """
        Configure migration context
        1. Pass our models metadata
        2. Set schema for alembic_version table
        3. Load all available schemas
        """
        context.configure(
            connection=connection,
            target_metadata=target_metadata,
            version_table_schema=target_metadata.schema,
            include_schemas=True
        )

        with context.begin_transaction():
            """
            By default search_path is setted to "$user",public 
            that why alembic can't create foreign keys correctly
            """
            context.execute('SET search_path TO public')
            context.run_migrations()


if context.is_offline_mode():
    run_migrations_offline()
else:
    run_migrations_online()

        connection.execute(text('set search_path to "%s"' % settings.postgres_db_schema)) #  <-- The magic line

        with context.begin_transaction():
            context.run_migrations()

Hope this helps

Thank you, @Ilyes-git! This definitely helps a lot! 🙇🏼

This solution, in case your app uses a single custom schema, allows you to avoid adding schema configuration to each model in the application.

However, there's a minor thing that can be improved here. Although it works perfectly fine with psycopg, when you use asyncpg the whole migration rolls back since, I assume, connection.execute implicitly starts a transaction. So it's worth moving the magic line under context.run_migrations() just like this:

with context.begin_transaction():
    context.execute(text(f'SET search_path TO {settings.db_schema}'))
    context.run_migrations()

And it fixes the issue for asyncpg.