tabernacleorm 2.1.4

A unified, multi-engine async ORM for Python (MongoDB, PostgreSQL, MySQL, SQLite)

TabernacleORM 2.1.2 🐍⚡

The Async, Pydantic-powered ORM for Python.

TabernacleORM is a modern, lightweight, and fully async Object-Relational Mapper (ORM) designed for FastAPI and modern Python applications. It supports PostgreSQL, MySQL, SQLite, and MongoDB with a unified, expressive API.

Features

• Pydantic V2 Native: Models are Pydantic models. Automatic validation, JSON serialization, and OpenAPI schema generation.
• Fully Async: Built from the ground up for asyncio. No blocking IO.
• Unified API: Switch between SQL and NoSQL databases without changing your business logic.
• Relationships: OneToMany, ManyToMany, ForeignKey with lazy loading.
• Fluent Query Builder: Write expressive queries like User.filter(User.age > 18).
• Lifecycle Hooks: before_save, after_create, etc., for powerful automation.
• Stateless Sessions: Thread-safe and concurrency-safe session management.
• Read/Write Splitting: Native support for replicas and high availability.

Installation

pip install tabernacleorm
# For specific engines:
pip install tabernacleorm[postgresql]
pip install tabernacleorm[mysql]
pip install tabernacleorm[mongodb]

Quickstart

1. Define your Models

Since Tabernacle models are Pydantic models, defining schemas is intuitive:

from typing import Optional
from tabernacleorm import Model, connect
from tabernacleorm.fields import StringField, IntegerField, ForeignKey, OneToMany

# Connect to DB with auto table creation
connect("sqlite:///:memory:", auto_create=True)

class User(Model):
    name: str = StringField(max_length=50)
    age: int = IntegerField(default=18)
    
    # Relationships
    posts: list["Post"] = OneToMany("Post", back_populates="author_id")

    # Lifecycle Hooks
    async def before_save(self):
        self.name = self.name.capitalize()

class Post(Model):
    title: str = StringField()
    content: str = StringField()
    author_id: int = ForeignKey("User")

2. Create and Query

import asyncio

async def main():
    # Create Tables (Handled by auto_create=True, or use migrations in prod)
    # await User.get_engine().executeRaw(User.get_create_table_sql())

    # Create Records
    user = await User.create(name="alice", age=30)
    await Post.create(title="Hello World", content="My first post", author_id=user.id)

    # Fluent Querying (Async)
    users = await User.filter(User.age > 20).all()
    
    # Eager Loading (New!)
    # Fetch users with their 'posts' in one go
    users_with_posts = await User.filter(User.age > 20).include("posts").all()

    # Lazy Loading (New!)
    my_posts = await user.fetch_related("posts")

    # Complex Filtering
    posts = await Post.filter(Post.title.startswith("Hello")).limit(5).all()

asyncio.run(main())

Documentation

• Feature Guide (FEATURES.md) - Detailed Mongoose-style features, advanced populate/include.
• Read Replica Quickstart - Guide for using Read Replicas.
• MongoDB Replicas - High Availability guide.
• Replica Control - granular control over reads.

New in 2.1.3: Weighted Load Balancing ⚖️

You can now assign weights to read replicas to distribute traffic according to server capacity.

db = connect(
    engine="postgresql",
    write={"url": "postgresql://master:5432/db"},
    read=[
        # Heavy server gets 70% of traffic
        {"url": "postgresql://huge-replica:5432/db", "weight": 70},
        # Smaller server gets 30%
        {"url": "postgresql://small-replica:5432/db", "weight": 30}
    ]
)

Advanced Features

Transactions

Use the session context manager for safe transactions:

engine = User.get_engine()
async with engine.transaction() as session:
    user = await User.create(name="Bob", age=25)
    # Pass session to save() to participate in transaction
    await Post.create(title="Intro", content="...", author_id=user.id, session=session)

Hooks

Available hooks:

• before_save / after_save
• before_create / after_create
• before_delete / after_delete

class user(Model):
    password: str
    
    async def before_save(self):
        if not self.password.startswith("hash:"):
            self.password = f"hash:{self.password}"

Relationships

Define relationships using OneToMany, ManyToMany, or standard ForeignKey. Use await instance.fetch_related("field_name") to load them efficiently.

---

Built with ❤️ by the Tabernacle Team.