SQLAlchemy Adapter for PyCasbin

Warnings

• breaking Version 1.5.0 switched its core dependency and internal implementation from the older `casbin` library to `pycasbin`. Users migrating from `casbin-sqlalchemy-adapter` or older versions of `sqlalchemy-adapter` that implicitly relied on `casbin` must now ensure `pycasbin` is installed and update their imports to `from sqlalchemy_adapter import Adapter`.
  Fix: Upgrade to `pycasbin` and ensure your application code uses `from sqlalchemy_adapter import Adapter`.
• gotcha When initializing the `Adapter` with `create_table=False`, the library will not automatically create the `casbin_rule` table. It is then the user's responsibility to ensure the necessary policy tables exist in the database, typically via a separate database migration system like Alembic.
  Fix: If `create_table=False` is used, manually create the `casbin_rule` table and its schema before the adapter is used. For automatic table creation, omit `create_table` or set it to `True` (default behavior).
• gotcha Older versions of the adapter (prior to 1.9.0) might produce SQLAlchemy warnings related to duplicate `CasbinRule` class names if `db_class` was customized or specific SQLAlchemy configurations were in place. This has been addressed in v1.9.0.
  Fix: Upgrade to version 1.9.0 or newer to resolve potential SQLAlchemy warnings regarding `CasbinRule` class definitions.
• gotcha Prior to version 1.7.0, using a custom `db_class` with the adapter could inadvertently alter the global SQLAlchemy metadata state. This bug was fixed in v1.7.0.
  Fix: Upgrade to version 1.7.0 or newer to prevent unintended global metadata state alterations when using custom `db_class` models.

Install

• pip install sqlalchemy-adapter Install current version

Imports

• Adapter

  from sqlalchemy_adapter import Adapter

  The library package name changed from `casbin-sqlalchemy-adapter` to `sqlalchemy-adapter`.

Quickstart

This quickstart demonstrates how to initialize the `SQLAlchemyAdapter` and integrate it with `pycasbin.Enforcer`. It uses an in-memory SQLite database for a self-contained example. Policy rules are added, saved to the database via the adapter, and then enforced. Ensure you have `pycasbin` installed alongside `sqlalchemy-adapter`.

import os
import casbin
import tempfile
from sqlalchemy_adapter import Adapter

# Create a dummy model.conf for the Enforcer
model_conf_content = '''
[request_definition]
r = sub, obj, act

[policy_definition]
p = sub, obj, act

[policy_effect]
e = some(where (p.eft == allow))

[matchers]
m = r.sub == p.sub && r.obj == p.obj && r.act == p.act
'''

# Write the model content to a temporary file
with tempfile.NamedTemporaryFile(mode='w+', delete=False, suffix=".conf") as model_file:
    model_file.write(model_conf_content)
    model_path = model_file.name

# Use an in-memory SQLite database for a simple, runnable example.
# For persistent storage, provide a file path or a full database URL
# e.g., 'sqlite:///test.db' or from an environment variable.
database_url = os.environ.get('SQLALCHEMY_DATABASE_URL', 'sqlite:///:memory:')

try:
    # 1. Initialize the SQLAlchemy Adapter
    # By default, create_table=True, so tables will be created if they don't exist.
    adapter = Adapter(database_url)

    # 2. Initialize Casbin Enforcer with the adapter
    e = casbin.Enforcer(model_path, adapter)

    # 3. Add policies
    e.add_policy("alice", "data1", "read")
    e.add_policy("bob", "data2", "write")

    # 4. Save policies to the database (persists changes through the adapter)
    e.save_policy()

    print("Policies added and saved.")

    # 5. Enforce policies
    print(f"Alice can read data1: {e.enforce('alice', 'data1', 'read')}") # Expected: True
    print(f"Bob can write data2: {e.enforce('bob', 'data2', 'write')}") # Expected: True
    print(f"Alice can write data1: {e.enforce('alice', 'data1', 'write')}") # Expected: False

    # 6. Load policies (useful if Enforcer was re-initialized or policy was changed elsewhere)
    e.load_policy()
    print(f"After reloading, Alice can read data1: {e.enforce('alice', 'data1', 'read')}") # Expected: True

except Exception as ex:
    print(f"An error occurred: {ex}")
finally:
    # Clean up the temporary model file
    os.remove(model_path)