I assume that classes are implemented in da package.
1. In __init__.py file I instantiate sqlalchemy dependency objects.
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy import create_engine
from sqlalchemy.orm import sessionmaker
BaseEntity = declarative_base()
strDbURL = "oracle://db_tes:xxx@xe"
objDbEngine = create_engine(strDbURL, echo=False)
Session = sessionmaker(bind=objDbEngine)
Session.configure(bind=objDbEngine)
def getConnection():
return Session()
In this moment, we can start implementing the entities. Every entity can extend BaseEntity. Every time we need a database connection we use getConnection() method. A SQLAlchemy session is similar to a Hibernate session.
2. I define User class which has a composite chei(username and password ---- don't do this in practice: you can safely choose to use serial number as primary key). In this example, I want to show how composite keys are implemented.
from da import BaseEntity
from sqlalchemy import Column,Integer,String,Numeric,DateTime,ForeignKey,Sequence
from sqlalchemy.orm import relation, backref
class User(BaseEntity):
username = Column("username", String, primary_key=True)
password = Column("password", String, primary_key=True)
descriere = Column("descriere", String)
def __init__(self, strUsername, strPassword, strDesc):
self.username = strUsername
self.password = strPassword
self.descriere = strDesc
3. I implement the Address class.
from da import BaseEntity
from sqlalchemy import Column,Integer,String,Numeric,DateTime,ForeignKey,Sequence
from sqlalchemy.orm import relation, backref
class Adress(BaseEntity):
id = Column("id", Integer, Sequence("seq_xxx_id"), primary_key=True)
fk_username = Column("fk_username", String, ForeignKey("utilizatori.username"))
fk_password = Column("fk_password", String, ForeignKey("utilizatori.password"))
street = Column("street", String)
number = Column("number", String)
user = relation("User",
foreign_keys=[fk_username, fk_password],
primaryjoin="User.username == Address.fk_username and "
"User.password == Address.fk_password",
backref="adresses")
def __init__(self, strUser, strPasswd, strStreet, strNumber):
self.fk_username = strUser
self.fk_password = strPassword
self.street = strStreet
self.number = strNumber
Comments: in this example, I try to show the power and flexibility provided by SQLAlchemy. First of all, in every SQLAlchemy entity we'll define the mapped table structure to attributes of the class. Using relation function, we can implement links(foreign keys) to other entities:
- one-to-one relation
- one-to-many relation
- many-to-many relation
4. After we implemented all the entities, we have to use them. In the folloing example, I show several lines of code that prove the entities functionality:
import da
from da import User, Address
# I select all user from the database
objSes = da.getConnection()
for objUsr, in objSes.query(User).all():
print(objUsr.addresses)
# filter that returns the user with a specific address
for objUsr in objSes.query(User, Address).filter(User.username == Address.fk_username and User.password == Address.fk_password).filter(Address.id == 1).all():
print(objUsr.username)
In conclusion, SQLAlchemy easily maps relational logic to object logic(the scope of any ORM) but it doesn't require a configuration file or annotations(like Hibernate). In addition, writting a query is extremely easy. What was not covered in this tutorial but it is intuitive is entity saving(persist). We use add method of an opened session for adding and updating an entity.
