An acronym for “Atomicity, Consistency, Isolation, Durability”; a set of properties that guarantee that database transactions are processed reliably. (via Wikipedia)
Annotations are a concept used internally by SQLAlchemy in order to store
additional information along with ClauseElement objects. A Python
dictionary is associated with a copy of the object, which contains key/value
pairs significant to various internal systems, mostly within the ORM:
some_column = Column('some_column', Integer)
some_column_annotated = some_column._annotate({"entity": User})The annotation system differs from the public dictionary Column.info
in that the above annotation operation creates a copy of the new Column,
rather than considering all annotation values to be part of a single
unit. The ORM creates copies of expression objects in order to
apply annotations that are specific to their context, such as to differentiate
columns that should render themselves as relative to a joined-inheritance
entity versus those which should render relative to their immediate parent
table alone, as well as to differentiate columns within the “join condition”
of a relationship where the column in some cases needs to be expressed
in terms of one particular table alias or another, based on its position
within the join expression.
A two-tiered relationship which links two tables
together using an association table in the middle. The
association relationship differs from a many to many
relationship in that the many-to-many table is mapped
by a full class, rather than invisibly handled by the
sqlalchemy.orm.relationship() construct as in the case
with many-to-many, so that additional attributes are
explicitly available.
For example, if we wanted to associate employees with projects, also storing the specific role for that employee with the project, the relational schema might look like:
CREATE TABLE employee (
id INTEGER PRIMARY KEY,
name VARCHAR(30)
)
CREATE TABLE project (
id INTEGER PRIMARY KEY,
name VARCHAR(30)
)
CREATE TABLE employee_project (
employee_id INTEGER PRIMARY KEY,
project_id INTEGER PRIMARY KEY,
role_name VARCHAR(30),
FOREIGN KEY employee_id REFERENCES employee(id),
FOREIGN KEY project_id REFERENCES project(id)
)A SQLAlchemy declarative mapping for the above might look like:
class Employee(Base):
__tablename__ = 'employee'
id = Column(Integer, primary_key)
name = Column(String(30))
class Project(Base):
__tablename__ = 'project'
id = Column(Integer, primary_key)
name = Column(String(30))
class EmployeeProject(Base):
__tablename__ = 'employee_project'
employee_id = Column(Integer, ForeignKey('employee.id'), primary_key=True)
project_id = Column(Integer, ForeignKey('project.id'), primary_key=True)
role_name = Column(String(30))
project = relationship("Project", backref="project_employees")
employee = relationship("Employee", backref="employee_projects")Employees can be added to a project given a role name:
proj = Project(name="Client A")
emp1 = Employee(name="emp1")
emp2 = Employee(name="emp2")
proj.project_employees.extend([
EmployeeProject(employee=emp1, role="tech lead"),
EmployeeProject(employee=emp2, role="account executive")
])See also
Atomicity is one of the components of the ACID model, and requires that each transaction is “all or nothing”: if one part of the transaction fails, the entire transaction fails, and the database state is left unchanged. An atomic system must guarantee atomicity in each and every situation, including power failures, errors, and crashes. (via Wikipedia)
An extension to the relationship system whereby two
distinct relationship() objects can be
mutually associated with each other, such that they coordinate
in memory as changes occur to either side. The most common
way these two relationships are constructed is by using
the relationship() function explicitly
for one side and specifying the backref keyword to it so that
the other relationship() is created
automatically. We can illustrate this against the example we’ve
used in one to many as follows:
class Department(Base):
__tablename__ = 'department'
id = Column(Integer, primary_key=True)
name = Column(String(30))
employees = relationship("Employee", backref="department")
class Employee(Base):
__tablename__ = 'employee'
id = Column(Integer, primary_key=True)
name = Column(String(30))
dep_id = Column(Integer, ForeignKey('department.id'))A backref can be applied to any relationship, including one to many, many to one, and many to many.
A relational algebra term referring to an attribute or set of attributes that form a uniquely identifying key for a row. A row may have more than one candidate key, each of which is suitable for use as the primary key of that row. The primary key of a table is always a candidate key.
A check constraint is a condition that defines valid data when adding or updating an entry in a table of a relational database. A check constraint is applied to each row in the table.
(via Wikipedia)
A check constraint can be added to a table in standard SQL using DDL like the following:
ALTER TABLE distributors ADD CONSTRAINT zipchk CHECK (char_length(zipcode) = 5);The portion of the SELECT statement which enumerates the
SQL expressions to be returned in the result set. The expressions
follow the SELECT keyword directly and are a comma-separated
list of individual expressions.
E.g.:
SELECT user_account.name, user_account.email
FROM user_account WHERE user_account.name = 'fred'Above, the list of columns user_acount.name,
user_account.email is the columns clause of the SELECT.
Consistency is one of the compoments of the ACID model, and ensures that any transaction will bring the database from one valid state to another. Any data written to the database must be valid according to all defined rules, including but not limited to constraints, cascades, triggers, and any combination thereof. (via Wikipedia)
A subquery is correlated if it depends on data in the
enclosing SELECT.
Below, a subquery selects the aggregate value MIN(a.id)
from the email_address table, such that
it will be invoked for each value of user_account.id, correlating
the value of this column against the email_address.user_account_id
column:
SELECT user_account.name, email_address.email
FROM user_account
JOIN email_address ON user_account.id=email_address.user_account_id
WHERE email_address.id = (
SELECT MIN(a.id) FROM email_address AS a
WHERE a.user_account_id=user_account.id
)The above subquery refers to the user_account table, which is not itself
in the FROM clause of this nested query. Instead, the user_account
table is received from the enclosing query, where each row selected from
user_account results in a distinct execution of the subquery.
A correlated subquery is in most cases present in the WHERE clause
or columns clause of the immediately enclosing SELECT
statement, as well as in the ORDER BY or HAVING clause.
In less common cases, a correlated subquery may be present in the
FROM clause of an enclosing SELECT; in these cases the
correlation is typically due to the enclosing SELECT itself being
enclosed in the WHERE,
ORDER BY, columns or HAVING clause of another SELECT, such as:
SELECT parent.id FROM parent
WHERE EXISTS (
SELECT * FROM (
SELECT child.id AS id, child.parent_id AS parent_id, child.pos AS pos
FROM child
WHERE child.parent_id = parent.id ORDER BY child.pos
LIMIT 3)
WHERE id = 7)Correlation from one SELECT directly to one which encloses the correlated
query via its FROM
clause is not possible, because the correlation can only proceed once the
original source rows from the enclosing statement’s FROM clause are available.
INSERT,
UPDATE, and DELETE statements.DBAPI is shorthand for the phrase “Python Database API
Specification”. This is a widely used specification
within Python to define common usage patterns for all
database connection packages. The DBAPI is a “low level”
API which is typically the lowest level system used
in a Python application to talk to a database. SQLAlchemy’s
dialect system is constructed around the
operation of the DBAPI, providing individual dialect
classes which service a specific DBAPI on top of a
specific database engine; for example, the create_engine()
URL postgresql+psycopg2://@localhost/test
refers to the psycopg2
DBAPI/dialect combination, whereas the URL mysql+mysqldb://@localhost/test
refers to the MySQL for Python
DBAPI DBAPI/dialect combination.
An acronym for Data Definition Language. DDL is the subset of SQL that relational databases use to configure tables, constraints, and other permanent objects within a database schema. SQLAlchemy provides a rich API for constructing and emitting DDL expressions.
This describes one of the major object states which an object can have within a session; a deleted object is an object that was formerly persistent and has had a DELETE statement emitted to the database within a flush to delete its row. The object will move to the detached state once the session’s transaction is committed; alternatively, if the session’s transaction is rolled back, the DELETE is reverted and the object moves back to the persistent state.
See also
In Python, a descriptor is an object attribute with “binding behavior”, one whose attribute access has been overridden by methods in the descriptor protocol. Those methods are __get__(), __set__(), and __delete__(). If any of those methods are defined for an object, it is said to be a descriptor.
In SQLAlchemy, descriptors are used heavily in order to provide attribute behavior on mapped classes. When a class is mapped as such:
class MyClass(Base):
__tablename__ = 'foo'
id = Column(Integer, primary_key=True)
data = Column(String)The MyClass class will be mapped when its definition
is complete, at which point the id and data attributes,
starting out as Column objects, will be replaced
by the instrumentation system with instances
of InstrumentedAttribute, which are descriptors that
provide the above mentioned __get__(), __set__() and
__delete__() methods. The InstrumentedAttribute
will generate a SQL expression when used at the class level:
>>> print(MyClass.data == 5)
data = :data_1and at the instance level, keeps track of changes to values, and also lazy loads unloaded attributes from the database:
>>> m1 = MyClass()
>>> m1.id = 5
>>> m1.data = "some data"
>>> from sqlalchemy import inspect
>>> inspect(m1).attrs.data.history.added
"some data"This describes one of the major object states which an object can have within a session; a detached object is an object that has a database identity (i.e. a primary key) but is not associated with any session. An object that was previously persistent and was removed from its session either because it was expunged, or the owning session was closed, moves into the detached state. The detached state is generally used when objects are being moved between sessions or when being moved to/from an external object cache.
See also
A result-set column which is used during polymorphic loading to determine what kind of mapped class should be applied to a particular incoming result row. In SQLAlchemy, the classes are always part of a hierarchy mapping using inheritance mapping.
A domain model in problem solving and software engineering is a conceptual model of all the topics related to a specific problem. It describes the various entities, their attributes, roles, and relationships, plus the constraints that govern the problem domain.
(via Wikipedia)
See also
Durability is a property of the ACID model which means that once a transaction has been committed, it will remain so, even in the event of power loss, crashes, or errors. In a relational database, for instance, once a group of SQL statements execute, the results need to be stored permanently (even if the database crashes immediately thereafter). (via Wikipedia)
In the SQLAlchemy ORM, refers to when the data in a persistent or sometimes detached object is erased, such that when the object’s attributes are next accessed, a lazy load SQL query will be emitted in order to refresh the data for this object as stored in the current ongoing transaction.
See also
A referential constraint between two tables. A foreign key is a field or set of fields in a relational table that matches a candidate key of another table. The foreign key can be used to cross-reference tables. (via Wikipedia)
A foreign key constraint can be added to a table in standard SQL using DDL like the following:
ALTER TABLE employee ADD CONSTRAINT dep_id_fk
FOREIGN KEY (employee) REFERENCES department (dep_id)The portion of the SELECT statement which indicates the initial
source of rows.
A simple SELECT will feature one or more table names in its
FROM clause. Multiple sources are separated by a comma:
SELECT user.name, address.email_address
FROM user, address
WHERE user.id=address.user_idThe FROM clause is also where explicit joins are specified. We can
rewrite the above SELECT using a single FROM element which consists
of a JOIN of the two tables:
SELECT user.name, address.email_address
FROM user JOIN address ON user.id=address.user_idA mapping between Python objects and their database identities. The identity map is a collection that’s associated with an ORM session object, and maintains a single instance of every database object keyed to its identity. The advantage to this pattern is that all operations which occur for a particular database identity are transparently coordinated onto a single object instance. When using an identity map in conjunction with an isolated transaction, having a reference to an object that’s known to have a particular primary key can be considered from a practical standpoint to be a proxy to the actual database row.
See also
Martin Fowler - Identity Map - http://martinfowler.com/eaaCatalog/identityMap.html
The isolation property of the ACID model ensures that the concurrent execution of transactions results in a system state that would be obtained if transactions were executed serially, i.e. one after the other. Each transaction must execute in total isolation i.e. if T1 and T2 execute concurrently then each should remain independent of the other. (via Wikipedia)
In object relational mapping, a “lazy load” refers to an attribute that does not contain its database-side value for some period of time, typically when the object is first loaded. Instead, the attribute receives a memoization that causes it to go out to the database and load its data when it’s first used. Using this pattern, the complexity and time spent within object fetches can sometimes be reduced, in that attributes for related tables don’t need to be addressed immediately.
A style of sqlalchemy.orm.relationship() which links two tables together
via an intermediary table in the middle. Using this configuration,
any number of rows on the left side may refer to any number of
rows on the right, and vice versa.
A schema where employees can be associated with projects:
CREATE TABLE employee (
id INTEGER PRIMARY KEY,
name VARCHAR(30)
)
CREATE TABLE project (
id INTEGER PRIMARY KEY,
name VARCHAR(30)
)
CREATE TABLE employee_project (
employee_id INTEGER PRIMARY KEY,
project_id INTEGER PRIMARY KEY,
FOREIGN KEY employee_id REFERENCES employee(id),
FOREIGN KEY project_id REFERENCES project(id)
)Above, the employee_project table is the many-to-many table,
which naturally forms a composite primary key consisting
of the primary key from each related table.
In SQLAlchemy, the sqlalchemy.orm.relationship() function
can represent this style of relationship in a mostly
transparent fashion, where the many-to-many table is
specified using plain table metadata:
class Employee(Base):
__tablename__ = 'employee'
id = Column(Integer, primary_key)
name = Column(String(30))
projects = relationship(
"Project",
secondary=Table('employee_project', Base.metadata,
Column("employee_id", Integer, ForeignKey('employee.id'),
primary_key=True),
Column("project_id", Integer, ForeignKey('project.id'),
primary_key=True)
),
backref="employees"
)
class Project(Base):
__tablename__ = 'project'
id = Column(Integer, primary_key)
name = Column(String(30))Above, the Employee.projects and back-referencing Project.employees
collections are defined:
proj = Project(name="Client A")
emp1 = Employee(name="emp1")
emp2 = Employee(name="emp2")
proj.employees.extend([emp1, emp2])A style of relationship() which links
a foreign key in the parent mapper’s table to the primary
key of a related table. Each parent object can
then refer to exactly zero or one related object.
The related objects in turn will have an implicit or explicit one to many relationship to any number of parent objects that refer to them.
An example many to one schema (which, note, is identical to the one to many schema):
CREATE TABLE department (
id INTEGER PRIMARY KEY,
name VARCHAR(30)
)
CREATE TABLE employee (
id INTEGER PRIMARY KEY,
name VARCHAR(30),
dep_id INTEGER REFERENCES department(id)
)The relationship from employee to department is
many to one, since many employee records can be associated with a
single department. A SQLAlchemy mapping might look like:
class Department(Base):
__tablename__ = 'department'
id = Column(Integer, primary_key=True)
name = Column(String(30))
class Employee(Base):
__tablename__ = 'employee'
id = Column(Integer, primary_key=True)
name = Column(String(30))
dep_id = Column(Integer, ForeignKey('department.id'))
department = relationship("Department")orm.mapper() function. This process associates the
class with a database table or other selectable
construct, so that instances of it can be persisted
using a Session as well as loaded using a
Query.An object-oriented technique whereby the state of an object is constructed by calling methods on the object. The object features any number of methods, each of which return a new object (or in some cases the same object) with additional state added to the object.
The two SQLAlchemy objects that make the most use of
method chaining are the Select
object and the Query object.
For example, a Select object can
be assigned two expressions to its WHERE clause as well
as an ORDER BY clause by calling upon the where()
and order_by() methods:
stmt = select([user.c.name]).\
where(user.c.id > 5).\
where(user.c.name.like('e%').\
order_by(user.c.name)Each method call above returns a copy of the original
Select object with additional qualifiers
added.
See also
The N plus one problem is a common side effect of the lazy load pattern, whereby an application wishes to iterate through a related attribute or collection on each member of a result set of objects, where that attribute or collection is set to be loaded via the lazy load pattern. The net result is that a SELECT statement is emitted to load the initial result set of parent objects; then, as the application iterates through each member, an additional SELECT statement is emitted for each member in order to load the related attribute or collection for that member. The end result is that for a result set of N parent objects, there will be N + 1 SELECT statements emitted.
The N plus one problem is alleviated using eager loading.
See also
A style of relationship() which links
the primary key of the parent mapper’s table to the foreign
key of a related table. Each unique parent object can
then refer to zero or more unique related objects.
The related objects in turn will have an implicit or explicit many to one relationship to their parent object.
An example one to many schema (which, note, is identical to the many to one schema):
CREATE TABLE department (
id INTEGER PRIMARY KEY,
name VARCHAR(30)
)
CREATE TABLE employee (
id INTEGER PRIMARY KEY,
name VARCHAR(30),
dep_id INTEGER REFERENCES department(id)
)The relationship from department to employee is
one to many, since many employee records can be associated with a
single department. A SQLAlchemy mapping might look like:
class Department(Base):
__tablename__ = 'department'
id = Column(Integer, primary_key=True)
name = Column(String(30))
employees = relationship("Employee")
class Employee(Base):
__tablename__ = 'employee'
id = Column(Integer, primary_key=True)
name = Column(String(30))
dep_id = Column(Integer, ForeignKey('department.id'))This describes one of the major object states which an object can have within a session; a pending object is a new object that doesn’t have any database identity, but has been recently associated with a session. When the session emits a flush and the row is inserted, the object moves to the persistent state.
See also
This describes one of the major object states which an object can have within a session; a persistent object is an object that has a database identity (i.e. a primary key) and is currently associated with a session. Any object that was previously pending and has now been inserted is in the persistent state, as is any object that’s been loaded by the session from the database. When a persistent object is removed from a session, it is known as detached.
See also
Refers to a function that handles several types at once. In SQLAlchemy, the term is usually applied to the concept of an ORM mapped class whereby a query operation will return different subclasses based on information in the result set, typically by checking the value of a particular column in the result known as the discriminator.
Polymorphic loading in SQLAlchemy implies that a one or a combination of three different schemes are used to map a hierarchy of classes; “joined”, “single”, and “concrete”. The section Mapping Class Inheritance Hierarchies describes inheritance mapping fully.
A constraint that uniquely defines the characteristics of each row. The primary key has to consist of characteristics that cannot be duplicated by any other row. The primary key may consist of a single attribute or multiple attributes in combination. (via Wikipedia)
The primary key of a table is typically, though not always,
defined within the CREATE TABLE DDL:
CREATE TABLE employee (
emp_id INTEGER,
emp_name VARCHAR(30),
dep_id INTEGER,
PRIMARY KEY (emp_id)
)A connecting unit between two mapped classes, corresponding to some relationship between the two tables in the database.
The relationship is defined using the SQLAlchemy function
relationship(). Once created, SQLAlchemy
inspects the arguments and underlying mappings involved
in order to classify the relationship as one of three types:
one to many, many to one, or many to many.
With this classification, the relationship construct
handles the task of persisting the appropriate linkages
in the database in response to in-memory object associations,
as well as the job of loading object references and collections
into memory based on the current linkages in the
database.
See also
In the context of SQLAlchemy, the term “released”
refers to the process of ending the usage of a particular
database connection. SQLAlchemy features the usage
of connection pools, which allows configurability as to
the lifespan of database connections. When using a pooled
connection, the process of “closing” it, i.e. invoking
a statement like connection.close(), may have the effect
of the connection being returned to an existing pool,
or it may have the effect of actually shutting down the
underlying TCP/IP connection referred to by that connection -
which one takes place depends on configuration as well
as the current state of the pool. So we used the term
released instead, to mean “do whatever it is you do
with connections when we’re done using them”.
The term will sometimes be used in the phrase, “release
transactional resources”, to indicate more explicitly that
what we are actually “releasing” is any transactional
state which as accumulated upon the connection. In most
situations, the process of selecting from tables, emitting
updates, etc. acquires isolated state upon
that connection as well as potential row or table locks.
This state is all local to a particular transaction
on the connection, and is released when we emit a rollback.
An important feature of the connection pool is that when
we return a connection to the pool, the connection.rollback()
method of the DBAPI is called as well, so that as the
connection is set up to be used again, it’s in a “clean”
state with no references held to the previous series
of operations.
See also
This is a non-SQL standard clause provided in various forms by certain backends, which provides the service of returning a result set upon execution of an INSERT, UPDATE or DELETE statement. Any set of columns from the matched rows can be returned, as though they were produced from a SELECT statement.
The RETURNING clause provides both a dramatic performance boost to common update/select scenarios, including retrieval of inline- or default- generated primary key values and defaults at the moment they were created, as well as a way to get at server-generated default values in an atomic way.
An example of RETURNING, idiomatic to PostgreSQL, looks like:
INSERT INTO user_account (name) VALUES ('new name') RETURNING id, timestampAbove, the INSERT statement will provide upon execution a result set
which includes the values of the columns user_account.id and
user_account.timestamp, which above should have been generated as default
values as they are not included otherwise (but note any series of columns
or SQL expressions can be placed into RETURNING, not just default-value columns).
The backends that currently support RETURNING or a similar construct are PostgreSQL, SQL Server, Oracle, and Firebird. The PostgreSQL and Firebird implementations are generally full featured, whereas the implementations of SQL Server and Oracle have caveats. On SQL Server, the clause is known as “OUTPUT INSERTED” for INSERT and UPDATE statements and “OUTPUT DELETED” for DELETE statements; the key caveat is that triggers are not supported in conjunction with this keyword. On Oracle, it is known as “RETURNING…INTO”, and requires that the value be placed into an OUT parameter, meaning not only is the syntax awkward, but it can also only be used for one row at a time.
SQLAlchemy’s UpdateBase.returning() system provides a layer of abstraction
on top of the RETURNING systems of these backends to provide a consistent
interface for returning columns. The ORM also includes many optimizations
that make use of RETURNING when available.
The container or scope for ORM database operations. Sessions load instances from the database, track changes to mapped instances and persist changes in a single unit of work when flushed.
See also
Refers to a SELECT statement that is embedded within an enclosing
SELECT.
A subquery comes in two general flavors, one known as a “scalar select”
which specifically must return exactly one row and one column, and the
other form which acts as a “derived table” and serves as a source of
rows for the FROM clause of another select. A scalar select is eligible
to be placed in the WHERE clause, columns clause,
ORDER BY clause or HAVING clause of the enclosing select, whereas the
derived table form is eligible to be placed in the FROM clause of the
enclosing SELECT.
Examples:
a scalar subquery placed in the columns clause of an enclosing
SELECT. The subquery in this example is a correlated subquery because part
of the rows which it selects from are given via the enclosing statement.
SELECT id, (SELECT name FROM address WHERE address.user_id=user.id)
FROM usera scalar subquery placed in the WHERE clause of an enclosing
SELECT. This subquery in this example is not correlated as it selects a fixed result.
SELECT id, name FROM user
WHERE status=(SELECT status_id FROM status_code WHERE code='C')a derived table subquery placed in the FROM clause of an enclosing
SELECT. Such a subquery is almost always given an alias name.
SELECT user.id, user.name, ad_subq.email_address
FROM
user JOIN
(select user_id, email_address FROM address WHERE address_type='Q') AS ad_subq
ON user.id = ad_subq.user_idThis describes one of the major object states which an object can have within a session; a transient object is a new object that doesn’t have any database identity and has not been associated with a session yet. When the object is added to the session, it moves to the pending state.
See also
A unique key index can uniquely identify each row of data values in a database table. A unique key index comprises a single column or a set of columns in a single database table. No two distinct rows or data records in a database table can have the same data value (or combination of data values) in those unique key index columns if NULL values are not used. Depending on its design, a database table may have many unique key indexes but at most one primary key index.
(via Wikipedia)
This pattern is where the system transparently keeps track of changes to objects and periodically flushes all those pending changes out to the database. SQLAlchemy’s Session implements this pattern fully in a manner similar to that of Hibernate.
The portion of the SELECT statement which indicates criteria
by which rows should be filtered. It is a single SQL expression
which follows the keyword WHERE.
SELECT user_account.name, user_account.email
FROM user_account
WHERE user_account.name = 'fred' AND user_account.status = 'E'Above, the phrase WHERE user_account.name = 'fred' AND user_account.status = 'E'
comprises the WHERE clause of the SELECT.