This article contains content that is written like an advertisement. (December 2017) (Learn how and when to remove this template message)
|Original author(s)||Michael Bayer|
|Initial release||February 14, 2006|
1.2.8 / May 28, 2018
SQLAlchemy provides "a full suite of well known enterprise-level persistence patterns, designed for efficient and high-performing database access, adapted into a simple and Pythonic domain language". SQLAlchemy's philosophy is that relational databases behave less like object collections the more that scale and performance are concerns, while conversely, object collections behave less like tables and rows the more abstraction is a primary factor in the design. For this reason it has adopted the data mapper pattern (similar to Hibernate for Java) rather than the active record pattern used by a number of other object-relational mappers. However, optional plugins allow users to develop using declarative syntax.
The following example represents an n-to-1 relationship between movies and their directors. It is shown how user-defined Python classes create corresponding database tables, how instances with relationships are created from either side of the relationship, and finally how the data can be queried—illustrating automatically-generated SQL queries for both lazy and eager loading.
Creating two Python classes and according database tables in the DBMS:
from sqlalchemy import * from sqlalchemy.ext.declarative import declarative_base from sqlalchemy.orm import relation, sessionmaker Base = declarative_base() class Movie(Base): __tablename__ = 'movies' id = Column(Integer, primary_key=True) title = Column(String(255), nullable=False) year = Column(Integer) directed_by = Column(Integer, ForeignKey('directors.id')) director = relation("Director", backref='movies', lazy=False) def __init__(self, title=None, year=None): self.title = title self.year = year def __repr__(self): return "Movie(%r, %r, %r)" % (self.title, self.year, self.director) class Director(Base): __tablename__ = 'directors' id = Column(Integer, primary_key=True) name = Column(String(50), nullable=False, unique=True) def __init__(self, name=None): self.name = name def __repr__(self): return "Director(%r)" % (self.name) engine = create_engine('dbms://user:pwd@host/dbname') Base.metadata.create_all(engine)
One can insert a director-movie relationship via either entity:
Session = sessionmaker(bind=engine) session = Session() m1 = Movie("Robocop", 1987) m1.director = Director("Paul Verhoeven") d2 = Director("George Lucas") d2.movies = [Movie("Star Wars", 1977), Movie("THX 1138", 1971)] try: session.add(m1) session.add(d2) session.commit() except: session.rollback()
alldata = session.query(Movie).all() for somedata in alldata: print somedata
SQLAlchemy issues the following query to the DBMS (omitting aliases):
SELECT movies.id, movies.title, movies.year, movies.directed_by, directors.id, directors.name FROM movies LEFT OUTER JOIN directors ON directors.id = movies.directed_by
Movie('Robocop', 1987L, Director('Paul Verhoeven')) Movie('Star Wars', 1977L, Director('George Lucas')) Movie('THX 1138', 1971L, Director('George Lucas'))
lazy=True (default) instead, SQLAlchemy would first issue a query to get the list of movies and only when needed (lazy) for each director a query to get the name of the according director:
SELECT movies.id, movies.title, movies.year, movies.directed_by FROM movies SELECT directors.id, directors.name FROM directors WHERE directors.id = %s
- Mike Bayer is the creator of SQLAlchemy and Mako Templates for Python.
- "Download - SQLAlchemy". SQLAlchemy. Retrieved 21 February 2015.
- "1.2 Changelog". Retrieved 19 June 2018.
- "zzzeek / sqlalchemy / source / LICENSE". BitBucket. Retrieved 21 February 2015.
- in The architecture of open source applications