psqlmodel 0.9.22

A modern, lightweight async PostgreSQL ORM with QueryBuilder, migrations, and trigger support

PSQLModel

PSQLModel is a modern, lightweight asynchronous ORM/Framework for PostgreSQL. It combines the speed of asyncpg with a Pythonic API for generic relationship handling, complex query generation, and robust schema migrations.

[!WARNING] Project in Active Development (Beta)

This package is currently in BETA. APIs may change slightly between versions as we refine the architecture.

We strongly encourage the community to contribute! If you find bugs, see opportunities for optimization, or want to help structure the code better, please submit a PR or open an issue. Let's build the best Python async ORM/Framework together.

---

🏗️ Core Modules

The framework is structured into several key modules:

• psqlmodel.orm: Core Object-Relational Mapping logic. Defines PSQLModel, Column, Table, and relationship descriptors.
• psqlmodel.core: Engine and Session management (create_engine, Session, AsyncSession). Handles connection pooling and transaction lifecycles.
• psqlmodel.migrations: A complete migration system (Alembic-style) supporting auto-generation, DAG dependency validation, and atomic rollbacks.
• psqlmodel.query_builder: A fluent API for constructing complex SQL queries (SELECT, INSERT, CTEs) programmatically.
• psqlmodel.db.triggers: A DSL for defining reactive database triggers in pure Python.

---

🚀 Quick Start

1. Define Models

Use Python type hints and the Column descriptor to define your schema.

from psqlmodel import PSQLModel, Column, table, Relationship, Relation
from psqlmodel.types import serial, varchar, timestamp
from datetime import datetime

@table("users")
class User(PSQLModel):
    id: serial = Column(primary_key=True)
    username: str = Column(max_len=50, unique=True, nullable=False)
    created_at: timestamp = Column(default=datetime.now)
    
    # Relationships
    posts: Relation[list["Post"]] = Relationship("Post")

@table("posts")
class Post(PSQLModel):
    id: serial = Column(primary_key=True)
    title: str = Column(max_len=200)
    user_id: int = Column(foreign_key="users.id")
    
    # Inverse relationship
    author: Relation["User"] = Relationship("User")

2. Async Usage

Perform operations using AsyncSession.

from psqlmodel import create_engine, Select, Session

async def main():
    engine = create_engine("postgresql://user:pass@localhost/db", async_=True)
    
    async with Session(engine) as session:
        # Create
        new_user = User(username="hashdown")
        await session.add(new_user)
        
        # Query with Eager Loading (JOIN)
        stmt = Select(User).Where(User.username == "hashdown").Include(Posts) #<- Use the Model name instead of the table name or User.posts
        user = await session.exec(stmt).first()

---

🔍 Query Structure

PSQLModel provides a powerful Query Builder for complex SQL generation.

Select & Join

# SELECT * FROM users 
# JOIN posts ON posts.user_id = users.id 
# WHERE users.age > 18
stmt = (
    Select(User)
    .Where(User.age > 18)
    .Include(Post) #<- Use the Model name instead of the table name or User.posts
)

CTEs and Chained Inserts

Build Common Table Expressions (WITH clauses) easily.

from psqlmodel import With, Insert

# WITH new_org AS (INSERT ...) 
# INSERT INTO users ... FROM new_org
stmt_org = Insert(Organization).Values(name="MyCorp").Returning(Organization.id)

stmt_user = (
    Insert(User)
    .Select("email", "org_id")
    .From("new_org")
)

query = With("new_org", stmt_org).Then(stmt_user)

UPSERT (On Conflict)

Insert(User).Values(...).OnConflict(
    User.email,
    do_update={"last_login": Now()}
)

---

📦 Migration System

PSQLModel includes a CLI for managing schema changes.

Command	Description
migrate init	Initialize migrations directory.
migrate generate -m "msg"	Auto-generate migration from model changes.
migrate upgrade head	Apply pending migrations.
migrate downgrade -1	Revert the last migration.
migrate history	Show migration history (DDL & Data).
migrate failures	View failed migration attempts.

Data Migrations

For complex data transformations, inherit from DataMigration and use iter_batches.

from psqlmodel.migrations import DataMigration

class Migration_Backfill(DataMigration):
    timeout_seconds = 600
    
    async def up_async(self, ctx):
        async for batch in self.iter_batches(ctx, "users"):
            # Transform batch...
            pass

---

⚡ Types & Triggers

Types

Import types from psqlmodel.types for clarity in definitions:

• serial, bigserial, uuid
• varchar, text, jsonb
• timestamp, date, boolean

Trigger DSL

Define reactive database logic purely in Python:

from psqlmodel import trigger, Trigger, Old, New

def log_change(old, new):
    print(f"User changed from {old.email} to {new.email}")

@trigger(Trigger().BeforeUpdate().Exec(log_change))
@table("users")
class User(PSQLModel):
    ...

---

🔌 Integrations & Middlewares

PSQLModel is designed to be extensible. We provide built-in integrations for common patterns:

Middlewares

Plug these into your engine to enhance query execution:

• ValidationMiddleware: Enforces schema constraints (max_len, min_value) at runtime.
• AuditMiddleware: Records detailed logs of executed queries.
• MetricsMiddleware: Collects performance stats (latency, query counts).
• RetryMiddleware: Automatically retries failed queries (e.g., deadlocks, connection loss).

External Integrations

Triggers can interact with external systems (requires plpython3u):

• Kafka: Produce events directly from database triggers.
• Redis: Publish messages or cache invalidations from triggers.

from psqlmodel.integrations.middlewares import ValidationMiddleware
engine.add_middleware_sync(ValidationMiddleware(model=User).sync)

---

📚 Documentation & Contributing

This documentation is a living document. As the framework evolves, we will continue adding more guides, API references, and best practices.

We strongly encourage the community to contribute!

• Documentation: Help us improve this README, write tutorials, or add docstrings.
• Examples: Have a cool use case? Add it to examples/.
• Code: Fix bugs, add features, or optimize performance.

Please check the examples/ directory for advanced usage patterns (replication, middlewares, complex queries).

Repo Structure:

• /psqlmodel: Source code.
• /examples: Comprehensive usage examples.
• /tests: Pytest suite.

License: MIT