SQLAlchemy

SQLAlchemy (source code) is a well-regarded database toolkit and object-relational mapper (ORM) implementation written in Python. SQLAlchemy provides a generalized interface for creating and executing database-agnostic code without needing to write SQL statements.

SQLAlchemy logo.

Why is SQLAlchemy a good ORM choice?

SQLAlchemy isn't just an ORM- it also provides SQLAlchemy Core for performing database work that is abstracted from the implementation differences between PostgreSQL, SQLite, etc. In some ways, the ORM is a bonus to Core that automates commonly-required create, read, update and delete operations.

SQLAlchemy can be used with or without the ORM features. Any given project can choose to just use SQLAlchemy Core or both Core and the ORM. The following diagram shows a few example configurations with various application software stacks and backend databases. Any of these configurations can be a valid option depending on what type of application you are coding.

Example SQLAlchemy configurations with different web frameworks.

A benefit many developers enjoy with SQLAlchemy is that it allows them to write Python code in their project to map from the database schema to the applications' Python objects. No SQL is required to create, maintain and query the database. The mapping allows SQLAlchemy to handle the underlying database so developers can work with their Python objects instead of writing bridge code to get data in and out of relational tables.

SQLAlchemy is an implementation of the object-relational mapping (ORM) concept. Learn more in the data chapter or view all topics.

How does SQLAlchemy code compare to raw SQL?

Below is an example of a SQLAlchemy model definition from the open source compare-python-web-frameworks project that uses SQLAlchemy with Flask and Flask-SQLAlchemy.

class Contact(db.Model):
    __tablename__ = 'contacts'
    id = db.Column(db.Integer, primary_key=True)
    first_name = db.Column(db.String(100))
    last_name = db.Column(db.String(100))
    phone_number = db.Column(db.String(32))

    def __repr__(self):
        return '<Contact {0} {1}: {2}>'.format(self.first_name,
                                               self.last_name,
                                               self.phone_number)

SQLAlchemy handles the table creation that otherwise we would have had to write a create table statement like this one to do the work:

CREATE TABLE CONTACTS(
   ID INT PRIMARY KEY        NOT NULL,
   FIRST_NAME     CHAR(100)  NOT NULL,
   LAST_NAME      CHAR(100)  NOT NULL,
   PHONE_NUMBER   CHAR(32)   NOT NULL,
);

By using SQLAlchemy in our Python code, all records can be obtained with a line like contacts = Contact.query.all() instead of a plain SQL such as SELECT * FROM contacts. That may not look like much of a difference in syntax but writing the queries in Python is often faster and easier for many Python developers once multiple tables and specific filtering on fields for queries have to be written. In addition, SQLAlchemy abstracts away idiosyncratic differences between database implementations in SQLite, MySQL and PostgreSQL.

SQLAlchemy Extensions, Plug-ins and Related Libraries

Take a look at the SQLAlchemy extensions, plug-ins and related libraries page for a curated list of useful code libraries to use with SQLAlchemy.

Using SQLAlchemy with Web Frameworks

There is no reason why you cannot use the SQLAlchemy library in any application that requires a database backend. However, if you are building a web app with Flask, Bottle or another web framework then take a look at the following extensions. They provide some glue code along with helper functions that can reduce the boilerplate code needed to connect your application's code with the SQLAlchemy library.

SQLAlchemy resources

The best way to get comfortable with SQLAlchemy is to dig in and write a database-driven application. The following resources can be helpful if you are having trouble getting started or are starting to run into some edge cases.

SQLAlchemy compared to other ORMs

SQLAlchemy is one of many Python object-relational mapper (ORM) implementations. Several open source projects and articles are listed here to make it a bit easier to understand the differences between these implementations.

What would you like to learn about building Python web apps?

Tell me about standard relational databases.

What're these NoSQL data stores hipster developers keep talking about?

I want to know about working with data in Python.


Matt Makai 2012-2020