nORM

Replace ORM complexity with SQL-first codegen.

One tool for your database schema, queries, and typed data-layer code.

Documentation · Python tutorial · GitHub

nORM (no ORM) is a SQL-first toolkit for managing your database and the code that talks to it. You define schema and repository SQL in one place; nORM generates typed data access code for your stack. Python is available today. Rust, Go, and TypeScript generators are in progress, and a migrations workflow is planned, so the same norm CLI can eventually cover schema changes and repositories across languages.

If you like sqlc, the workflow will feel familiar: you keep writing real queries, and nORM removes the repetitive mapping code. It also adds helpers for dynamic filters, sorting, partial updates, and join embedding that are painful to maintain by hand.

Example (Python)

Schema (norm_in/schema.sql):

CREATE TABLE users (
    id SERIAL PRIMARY KEY,
    name text NOT NULL,
    blocked bool DEFAULT false
);

Repository SQL (norm_in/repositories/users_repo.sql):

-- repo_name: UsersRepo

-- name: get_user :one
SELECT * FROM users WHERE id = :id;

-- name: list_users :many
SELECT * FROM users ORDER BY name;

Generated usage:

from norm_out.users_repo import UsersRepo

async with get_db() as db:
    repo = UsersRepo(db)
    user = await repo.get_user(id=42)       # User | None
    users = await repo.list_users()         # list[User]

Run norm init and norm generate to produce norm_out/ with Pydantic models and async repository methods. The same SQL inputs will drive other language targets as they land. See the Python tutorial for config, CRUD, and project layout.

Why nORM?

• One workflow for schema and data layer. Keep database definitions and repository SQL together instead of splitting schema management, query authoring, and hand-written mapping code across different tools.
• SQL stays in charge. Queries and execution plans stay visible; no ORM query builder in the middle.
• Typed APIs from your SQL. Parameters, return types, and row mapping are generated to match what you wrote.
• Dynamic queries without string assembly. Optional filters, sort orders, patch updates, and nested join shapes are expressed in SQL and compiled into safe, typed code.

Dynamic filtering

Prefix a parameter with _ to make it optional. One query can cover many filter combinations:

-- name: search_authors :many
SELECT * FROM authors
WHERE id = :_id OR rating > :rating;

# rating is required; id is optional (defaults to UNSET)
await repo.search_authors(rating=4)
await repo.search_authors(rating=4, id=7)

Dynamic filtering guide · Dynamic sorting · Partial updates · Embedding models

Who this is for

Good fit if you want visible SQL, less boilerplate, strongly typed data access, and runtime filters without hand-built SQL strings.

Not a fit if you prefer designing queries through ORM method chains. Use SQLAlchemy or similar instead.

Install

The CLI is a Python package today (requires 3.12+). Install norm-cli from PyPI; the command on your PATH is norm:

pipx install norm-cli
norm --version

From a checkout: pip install . (same norm command).

Quick start

norm init
# edit norm_in/schema.sql and norm_in/repositories/*.sql
norm generate

Other useful commands: norm schema pull (introspect Postgres), norm check (CI-friendly validation). See commands. Migrations are planned.

Language & database support

nORM is language-agnostic at the core: the same schema and repository SQL feed each code generator.

Language	Status	Databases
Python	Available	Postgres, SQLite, MySQL, ClickHouse, DuckDB
Rust, Go, TypeScript	In progress	TBD per target

Full support matrix

Contributing

Bug reports, ideas, and PRs are welcome. See CONTRIBUTING.md for local setup and docs workflow.

License

MIT