ormantism 0.6.0

A lightweight ORM built on Pydantic for simple CRUD operations with minimal code

Ormantism

A tiny, simple ORM built on top of Pydantic.

When you need to perform simple CRUD operations with minimal code.

Offers support for PostgreSQL, MySQL, SQLite (database URL syntax is the same as in SQLAlchemy).

Features

• Simple Model Declaration: Define your models using familiar Pydantic syntax
• Automatic Table Creation: Tables are created automatically when first accessed
• Lazy Loading: Relationships are loaded on-demand for optimal performance
• Transaction Support: Built-in transaction management with automatic rollback
• Preloading: Efficiently load related data with JOIN queries
• Optional Timestamps: Add created_at, updated_at, deleted_at fields automatically
• Load-or-create: Find by criteria or create in one call
• Versioning: Optional versioning along specified fields

Installation

pip install ormantism

Quick Start

1. Connect to Database

import ormantism

# Connect to a file database
ormantism.connect("sqlite:///my_app.db")

# Or use in-memory database for testing
ormantism.connect("sqlite://:memory:")

# MySQL
ormantism.connect("mysql://login:password@host:port/database")

# PostgreSQL
ormantism.connect("postgresql://login:password@host:port/database")

2. Define Models

from ormantism import Table
from typing import Optional

class User(Table):
    name: str
    email: str
    age: Optional[int] = None

class Post(Table, with_timestamps=True):
    title: str
    content: str
    author: User

3. Create and Save Records

# Create a user
user = User(name="Alice", email="[e-mail]", age=30)
# The record is automatically saved to the database

# Create a post linked to the user
post = Post(title="My First Post", content="Hello World!", author=user)

4. Query Records

# Load by ID
user = User.load(id=1)

# Load by criteria
user = User.load(name="Alice")
user = User.load(email="[e-mail]")

# Load latest post by this author
latest_post = Post.load(author=user, last_created=True)

# Load all records
users = User.load_all()

# Load with criteria
users_named_alice = User.load_all(name="Alice")

5. Update Records

user = User.load(id=1)
user.age = 31  # Automatically saved to database
# or
user.update(age=31, email="[e-mail]")

6. Delete Records

user = User.load(id=1)
user.delete()

Advanced Features

Load or create

Find a row by criteria or create it in one call. Use _search_fields to limit which fields are used for the lookup; other fields are then used to update the row if it already exists, or to populate a new row if not.

# Create or reuse by name; update value if name already exists
user = User.load_or_create(_search_fields=("name",), name="Alice", email="[e-mail]")
user2 = User.load_or_create(_search_fields=("name",), name="Alice", email="[e-mail]")  # same row, email updated

Timestamps

Add automatic timestamp tracking to your models:

class Post(Table, with_timestamps=True):
    title: str
    content: str

This adds created_at, updated_at, and deleted_at fields. Soft deletes are used when timestamps are enabled.

Versioning

Use versioning_along=(...) so that updates that change those fields create a new row instead of updating in place (useful for history).

class Document(Table, versioning_along=("name",)):
    name: str
    content: str

d1 = Document(name="foo", content="v1")
d2 = Document(name="foo", content="v2")  # New row; same name, different content

Relationships and Lazy Loading

class Author(Table):
    name: str

class Book(Table):
    title: str
    author: Author

# Create records
author = Author(name="Jane Doe")
book = Book(title="My Book", author=author)

# Lazy loading - author is loaded from DB when accessed
book = Book.load(id=1)
print(book.author.name)  # Database query happens here

Preloading (Eager Loading)

Avoid N+1 queries by preloading relationships:

# Load book with author in a single query
book = Book.load(id=1, preload="author")
print(book.author.name)  # No additional database query

# Preload nested relationships
book = Book.load(id=1, preload="author.publisher")

# Preload multiple relationships
book = Book.load(id=1, preload=["author", "category"])

Transactions

from ormantism import transaction

try:
    with transaction() as t:
        user1 = User(name="Alice", email="[e-mail]")
        user2 = User(name="Bob", email="[e-mail]")
        # Both users are saved automatically
        # Transaction commits when exiting the context
except Exception:
    # Transaction is automatically rolled back on any exception
    pass

Querying Examples

# Load single record
user = User.load(name="Alice")
latest_user = User.load(last_created=True)

# Load multiple records
all_users = User.load_all()
users_named_alice = User.load_all(name="Alice")

# Include soft-deleted records (when using timestamps)
all_including_deleted = User.load_all(with_deleted=True)

Model Definition

Basic Model

class User(Table):
    name: str
    email: str
    age: int = 25  # Default value
    bio: Optional[str] = None  # Nullable field

With Timestamps

class Post(Table, with_timestamps=True):
    title: str
    content: str
    # Automatically adds: created_at, updated_at, deleted_at

Supported Field Types

• int, float, str
• Optional[T] for nullable fields
• list, dict (stored as JSON)
• ormantism.JSON — arbitrary JSON (dict, list, primitives); stored as a JSON column
• datetime.datetime
• enum.Enum
• Pydantic models (stored as JSON)
• References to other Table models (single or list[...]); optional and self-referential refs supported

Relationships

class Category(Table):
    name: str

class Post(Table):
    title: str
    category: Category  # Foreign key relationship
    tags: Optional[Category] = None  # Nullable relationship

Self-referential and list relationships

from typing import Optional
from pydantic import Field

class Node(Table):
    parent: Optional["Node"] = None  # Self-reference
    name: str

class Parent(Table):
    name: str
    children: list["Parent"] = Field(default_factory=list)  # List of references

JSON and generic reference fields

from ormantism import Table, JSON

class WithJSON(Table):
    j: JSON  # Arbitrary JSON (dict, list, primitives); stored as JSON column

# Generic reference (any Table subclass); cannot be preloaded
class Ptr(Table):
    ref: Table

API Reference

Table class methods

Creating and loading

• Model(**data) - Create and automatically save a new record
• Model.load_or_create(_search_fields=("name",), **data) - Load one matching the given fields, or create; only _search_fields are used for the lookup; other fields can update the row if it already exists
• Model.load(**criteria) - Load single record
• Model.load(last_created=True) - Load most recently created record
• Model.load(as_collection=True, **criteria) - Load as list (no LIMIT 1)
• Model.load_all(**criteria) - Load multiple records
• Model.load(preload="relationship") or preload=["a", "b"] - Eager load relationships (not supported for generic Table references)
• Model.load(with_deleted=True) - Include soft-deleted records

Updating

• instance.update(**kwargs) - Update multiple fields
• instance.field = value - Update single field (auto-saves)

Deleting

• instance.delete() - Delete record (soft delete if timestamps enabled)

Database and transaction

• ormantism.connect(database_url) - Connect to database
• ormantism.transaction(connection_name=...) - Get transaction context manager (optional connection_name for multi-connection setups)

Table class options

• Table(..., with_timestamps=True) - Add created_at, updated_at, deleted_at and soft deletes
• Table(..., with_created_at_timestamp=True, with_timestamps=False) - Only created_at
• Table(..., with_updated_at_timestamp=True, with_timestamps=False) - Only updated_at
• Table(..., versioning_along=("name",)) - Version rows by these fields (creates new row on update when these change)
• Table(..., connection_name="custom_conn") - Use a named connection (inherited by subclasses)

Code reference

For a full code reference of the library (all classes and methods with definition and usage locations), see ormantism/REFERENCE.md. It lists every public class (e.g. Table, Field, JoinInfo, Transaction, SuperModel) and method with their source file and line numbers and where they are used—useful when navigating the codebase or contributing.

Limitations

• Simple Queries: Complex queries may require raw SQL
• No Migrations: New columns are added automatically when the model gains fields; dropping or renaming columns, or changing column types, requires manual handling
• Basic Relationships: Only supports simple foreign key relationships

Requirements

• Python 3.12+
• Pydantic

License

MIT License

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.