GitHub - wey-gu/py-pglite: PGlite wrapper in Python for testing. Test your app with Postgres just as lite as SQLite.

def test_users(pglite_session):
    user = User(name="Alice")
    pglite_session.add(user)
    pglite_session.commit()
    assert user.id == 1  # It's real PostgreSQL!

# ❌ Traditional testing
def test_old_way():
    # 1. Install PostgreSQL
    # 2. Configure connection
    # 3. Manage test databases
    # 4. Handle cleanup
    # 5. Docker containers...
    pass

# ✅ py-pglite way
def test_new_way(pglite_session):
    User.objects.create(name="Alice")  # Just works!

# Core (framework-agnostic)
pip install py-pglite

# With your stack
pip install py-pglite[sqlalchemy]  # SQLAlchemy + SQLModel
pip install py-pglite[django]      # Django + pytest-django
pip install py-pglite[asyncpg]     # Pure async client
pip install py-pglite[all]         # Everything

# Extra Features
pip install py-pglite[extensions]  # pglite extensions, like pgvector, fuzzystrmatch etc.

def test_sqlalchemy_just_works(pglite_session):
    user = User(name="Alice", email="alice@test.com")
    pglite_session.add(user)
    pglite_session.commit()

    assert user.id is not None
    assert User.query.count() == 1  # Real PostgreSQL!

def test_django_socket_pattern(configured_django):
    Post.objects.create(title="Hello", content="World")
    assert Post.objects.count() == 1  # Real PostgreSQL via socket!

def test_django_backend_pattern(django_pglite_db):
    Post.objects.create(title="Hello", content="World", metadata={"tags": ["test"]})
    assert Post.objects.count() == 1  # Custom backend with JSON support!

def test_any_client_works(pglite_manager):
    # Extract connection details
    engine = pglite_manager.get_engine()
    host, port, database = str(engine.url.host), engine.url.port, engine.url.database

    # Use with any PostgreSQL client
    # conn = psycopg.connect(host=host, port=port, dbname=database)
    # engine = create_async_engine(f"postgresql+asyncpg://{host}:{port}/{database}")

# For asyncpg specifically, use TCP mode with proper configuration:
async def test_asyncpg_works(pglite_tcp_manager):
    config = pglite_tcp_manager.config
    conn = await asyncpg.connect(
        host=config.tcp_host,
        port=config.tcp_port,
        user="postgres",
        password="postgres",
        database="postgres",
        ssl=False,
        server_settings={}  # CRITICAL: Required for PGlite compatibility
    )
    result = await conn.fetchval("SELECT 1")
    await conn.close()

from fastapi.testclient import TestClient

def test_api_endpoint(client: TestClient):
    response = client.post("/users/", json={"name": "Alice"})
    assert response.status_code == 201

    response = client.get("/users/")
    assert len(response.json()) == 1

def test_postgresql_power(pglite_session):
    pglite_session.execute(text("""
        CREATE TABLE analytics (
            data JSONB,
            tags TEXT[],
            created TIMESTAMP DEFAULT NOW()
        )
    """))

    pglite_session.execute(text("""
        INSERT INTO analytics (data, tags) VALUES
        ('{"clicks": 100}', ARRAY['web', 'mobile'])
    """))

    result = pglite_session.execute(text("""
        SELECT data->>'clicks' as clicks,
               array_length(tags, 1) as tag_count
        FROM analytics
        WHERE data->>'clicks' > '50'
    """)).fetchone()

    assert result.clicks == '100'

pip install 'py-pglite[extensions]'

from py_pglite import PGliteConfig, PGliteManager
from pgvector.psycopg import register_vector
import psycopg
import numpy as np

# Enable the extension
config = PGliteConfig(extensions=["pgvector"])

with PGliteManager(config=config) as db:
    with psycopg.connect(db.get_dsn(), autocommit=True) as conn:
        # Create the extension and register the type
        conn.execute("CREATE EXTENSION IF NOT EXISTS vector")
        register_vector(conn)

        # Create a table and insert a vector
        conn.execute("CREATE TABLE items (embedding vector(3))")
        conn.execute("INSERT INTO items (embedding) VALUES (%s)", (np.array([1, 2, 3]),))

        # Perform a similarity search
        result = conn.execute("SELECT * FROM items ORDER BY embedding <-> %s LIMIT 1", (np.array([1, 1, 1]),)).fetchone()
        assert np.array_equal(result[0], np.array([1, 2, 3]))