ORM Examples

The SQLAlchemy distribution includes a variety of code examples illustrating a select set of patterns, some typical and some not so typical. All are runnable and can be found in the /examples directory of the distribution. Descriptions and source code for all can be found here.

Additional SQLAlchemy examples, some user contributed, are available on the wiki at http://www.sqlalchemy.org/trac/wiki/UsageRecipes.

Mapping Recipes

Adjacency List

An example of a dictionary-of-dictionaries structure mapped using an adjacency list model.

E.g.:

  1. node = TreeNode('rootnode')
  2. node.append('node1')
  3. node.append('node3')
  4. session.add(node)
  5. session.commit()
  6. dump_tree(node)

Listing of files:

Associations

Examples illustrating the usage of the “association object” pattern, where an intermediary class mediates the relationship between two classes that are associated in a many-to-many pattern.

Listing of files:

  • proxied_association.py - Same example as basic_association, adding in usage of sqlalchemy.ext.associationproxy to make explicit references to OrderItem optional.

  • basic_association.py - Illustrate a many-to-many relationship between an “Order” and a collection of “Item” objects, associating a purchase price with each via an association object called “OrderItem”

  • dict_of_sets_with_default.py - An advanced association proxy example which illustrates nesting of association proxies to produce multi-level Python collections, in this case a dictionary with string keys and sets of integers as values, which conceal the underlying mapped classes.

Asyncio Integration

Examples illustrating the asyncio engine feature of SQLAlchemy.

Listing of files:

  • async_orm.py - Illustrates use of the sqlalchemy.ext.asyncio.AsyncSession object for asynchronous ORM use.

  • basic.py - Illustrates the asyncio engine / connection interface.

  • greenlet_orm.py - Illustrates use of the sqlalchemy.ext.asyncio.AsyncSession object for asynchronous ORM use, including the optional run_sync() method.

Directed Graphs

An example of persistence for a directed graph structure. The graph is stored as a collection of edges, each referencing both a “lower” and an “upper” node in a table of nodes. Basic persistence and querying for lower- and upper- neighbors are illustrated:

  1. n2 = Node(2)
  2. n5 = Node(5)
  3. n2.add_neighbor(n5)
  4. print(n2.higher_neighbors())

Listing of files:

Dynamic Relations as Dictionaries

Illustrates how to place a dictionary-like facade on top of a “dynamic” relation, so that dictionary operations (assuming simple string keys) can operate upon a large collection without loading the full collection at once.

Listing of files:

Generic Associations

Illustrates various methods of associating multiple types of parents with a particular child object.

The examples all use the declarative extension along with declarative mixins. Each one presents the identical use case at the end - two classes, Customer and Supplier, both subclassing the HasAddresses mixin, which ensures that the parent class is provided with an addresses collection which contains Address objects.

The discriminator_on_association.py and generic_fk.py scripts are modernized versions of recipes presented in the 2007 blog post Polymorphic Associations with SQLAlchemy.

Listing of files:

  • table_per_association.py - Illustrates a mixin which provides a generic association via a individually generated association tables for each parent class. The associated objects themselves are persisted in a single table shared among all parents.

  • table_per_related.py - Illustrates a generic association which persists association objects within individual tables, each one generated to persist those objects on behalf of a particular parent class.

  • discriminator_on_association.py - Illustrates a mixin which provides a generic association using a single target table and a single association table, referred to by all parent tables. The association table contains a “discriminator” column which determines what type of parent object associates to each particular row in the association table.

  • generic_fk.py - Illustrates a so-called “generic foreign key”, in a similar fashion to that of popular frameworks such as Django, ROR, etc. This approach bypasses standard referential integrity practices, in that the “foreign key” column is not actually constrained to refer to any particular table; instead, in-application logic is used to determine which table is referenced.

Large Collections

Large collection example.

Illustrates the options to use with relationship() when the list of related objects is very large, including:

  • “dynamic” relationships which query slices of data as accessed

  • how to use ON DELETE CASCADE in conjunction with passive_deletes=True to greatly improve the performance of related collection deletion.

Listing of files:

Materialized Paths

Illustrates the “materialized paths” pattern for hierarchical data using the SQLAlchemy ORM.

Listing of files:

Nested Sets

Illustrates a rudimentary way to implement the “nested sets” pattern for hierarchical data using the SQLAlchemy ORM.

Listing of files:

Performance

A performance profiling suite for a variety of SQLAlchemy use cases.

Each suite focuses on a specific use case with a particular performance profile and associated implications:

  • bulk inserts

  • individual inserts, with or without transactions

  • fetching large numbers of rows

  • running lots of short queries

All suites include a variety of use patterns illustrating both Core and ORM use, and are generally sorted in order of performance from worst to greatest, inversely based on amount of functionality provided by SQLAlchemy, greatest to least (these two things generally correspond perfectly).

A command line tool is presented at the package level which allows individual suites to be run:

  1. $ python -m examples.performance --help
  2. usage: python -m examples.performance [-h] [--test TEST] [--dburl DBURL]
  3. [--num NUM] [--profile] [--dump]
  4. [--echo]
  5. {bulk_inserts,large_resultsets,single_inserts}
  6. positional arguments:
  7. {bulk_inserts,large_resultsets,single_inserts}
  8. suite to run
  9. optional arguments:
  10. -h, --help show this help message and exit
  11. --test TEST run specific test name
  12. --dburl DBURL database URL, default sqlite:///profile.db
  13. --num NUM Number of iterations/items/etc for tests;
  14. default is module-specific
  15. --profile run profiling and dump call counts
  16. --dump dump full call profile (implies --profile)
  17. --echo Echo SQL output

An example run looks like:

  1. $ python -m examples.performance bulk_inserts

Or with options:

  1. $ python -m examples.performance bulk_inserts \
  2. --dburl mysql+mysqldb://scott:tiger@localhost/test \
  3. --profile --num 1000

See also

How can I profile a SQLAlchemy powered application?

File Listing

Listing of files:

  • bulk_updates.py - This series of tests illustrates different ways to UPDATE a large number of rows in bulk.

  • large_resultsets.py - In this series of tests, we are looking at time to load a large number of very small and simple rows.

  • bulk_inserts.py - This series of tests illustrates different ways to INSERT a large number of rows in bulk.

  • short_selects.py - This series of tests illustrates different ways to SELECT a single record by primary key

  • single_inserts.py - In this series of tests, we’re looking at a method that inserts a row within a distinct transaction, and afterwards returns to essentially a “closed” state. This would be analogous to an API call that starts up a database connection, inserts the row, commits and closes.

  • __main__.py - Allows the examples/performance package to be run as a script.

Running all tests with time

This is the default form of run:

  1. $ python -m examples.performance single_inserts
  2. Tests to run: test_orm_commit, test_bulk_save,
  3. test_bulk_insert_dictionaries, test_core,
  4. test_core_query_caching, test_dbapi_raw_w_connect,
  5. test_dbapi_raw_w_pool
  6. test_orm_commit : Individual INSERT/COMMIT pairs via the
  7. ORM (10000 iterations); total time 13.690218 sec
  8. test_bulk_save : Individual INSERT/COMMIT pairs using
  9. the "bulk" API (10000 iterations); total time 11.290371 sec
  10. test_bulk_insert_dictionaries : Individual INSERT/COMMIT pairs using
  11. the "bulk" API with dictionaries (10000 iterations);
  12. total time 10.814626 sec
  13. test_core : Individual INSERT/COMMIT pairs using Core.
  14. (10000 iterations); total time 9.665620 sec
  15. test_core_query_caching : Individual INSERT/COMMIT pairs using Core
  16. with query caching (10000 iterations); total time 9.209010 sec
  17. test_dbapi_raw_w_connect : Individual INSERT/COMMIT pairs w/ DBAPI +
  18. connection each time (10000 iterations); total time 9.551103 sec
  19. test_dbapi_raw_w_pool : Individual INSERT/COMMIT pairs w/ DBAPI +
  20. connection pool (10000 iterations); total time 8.001813 sec

Dumping Profiles for Individual Tests

A Python profile output can be dumped for all tests, or more commonly individual tests:

  1. $ python -m examples.performance single_inserts --test test_core --num 1000 --dump
  2. Tests to run: test_core
  3. test_core : Individual INSERT/COMMIT pairs using Core. (1000 iterations); total fn calls 186109
  4. 186109 function calls (186102 primitive calls) in 1.089 seconds
  5. Ordered by: internal time, call count
  6. ncalls tottime percall cumtime percall filename:lineno(function)
  7. 1000 0.634 0.001 0.634 0.001 {method 'commit' of 'sqlite3.Connection' objects}
  8. 1000 0.154 0.000 0.154 0.000 {method 'execute' of 'sqlite3.Cursor' objects}
  9. 1000 0.021 0.000 0.074 0.000 /Users/classic/dev/sqlalchemy/lib/sqlalchemy/sql/compiler.py:1950(_get_colparams)
  10. 1000 0.015 0.000 0.034 0.000 /Users/classic/dev/sqlalchemy/lib/sqlalchemy/engine/default.py:503(_init_compiled)
  11. 1 0.012 0.012 1.091 1.091 examples/performance/single_inserts.py:79(test_core)
  12. ...

Writing your Own Suites

The profiler suite system is extensible, and can be applied to your own set of tests. This is a valuable technique to use in deciding upon the proper approach for some performance-critical set of routines. For example, if we wanted to profile the difference between several kinds of loading, we can create a file test_loads.py, with the following content:

  1. from examples.performance import Profiler
  2. from sqlalchemy import Integer, Column, create_engine, ForeignKey
  3. from sqlalchemy.orm import relationship, joinedload, subqueryload, Session
  4. from sqlalchemy.ext.declarative import declarative_base
  5. Base = declarative_base()
  6. engine = None
  7. session = None
  8. class Parent(Base):
  9. __tablename__ = 'parent'
  10. id = Column(Integer, primary_key=True)
  11. children = relationship("Child")
  12. class Child(Base):
  13. __tablename__ = 'child'
  14. id = Column(Integer, primary_key=True)
  15. parent_id = Column(Integer, ForeignKey('parent.id'))
  16. # Init with name of file, default number of items
  17. Profiler.init("test_loads", 1000)
  18. @Profiler.setup_once
  19. def setup_once(dburl, echo, num):
  20. "setup once. create an engine, insert fixture data"
  21. global engine
  22. engine = create_engine(dburl, echo=echo)
  23. Base.metadata.drop_all(engine)
  24. Base.metadata.create_all(engine)
  25. sess = Session(engine)
  26. sess.add_all([
  27. Parent(children=[Child() for j in range(100)])
  28. for i in range(num)
  29. ])
  30. sess.commit()
  31. @Profiler.setup
  32. def setup(dburl, echo, num):
  33. "setup per test. create a new Session."
  34. global session
  35. session = Session(engine)
  36. # pre-connect so this part isn't profiled (if we choose)
  37. session.connection()
  38. @Profiler.profile
  39. def test_lazyload(n):
  40. "load everything, no eager loading."
  41. for parent in session.query(Parent):
  42. parent.children
  43. @Profiler.profile
  44. def test_joinedload(n):
  45. "load everything, joined eager loading."
  46. for parent in session.query(Parent).options(joinedload("children")):
  47. parent.children
  48. @Profiler.profile
  49. def test_subqueryload(n):
  50. "load everything, subquery eager loading."
  51. for parent in session.query(Parent).options(subqueryload("children")):
  52. parent.children
  53. if __name__ == '__main__':
  54. Profiler.main()

We can run our new script directly:

  1. $ python test_loads.py --dburl postgresql+psycopg2://scott:tiger@localhost/test
  2. Running setup once...
  3. Tests to run: test_lazyload, test_joinedload, test_subqueryload
  4. test_lazyload : load everything, no eager loading. (1000 iterations); total time 11.971159 sec
  5. test_joinedload : load everything, joined eager loading. (1000 iterations); total time 2.754592 sec
  6. test_subqueryload : load everything, subquery eager loading. (1000 iterations); total time 2.977696 sec

Relationship Join Conditions

Examples of various relationship() configurations, which make use of the primaryjoin argument to compose special types of join conditions.

Listing of files:

  • threeway.py - Illustrate a “three way join” - where a primary table joins to a remote table via an association table, but then the primary table also needs to refer to some columns in the remote table directly.

  • cast.py - Illustrate a relationship() that joins two columns where those columns are not of the same type, and a CAST must be used on the SQL side in order to match them.

Space Invaders

A Space Invaders game using SQLite as the state machine.

Originally developed in 2012. Adapted to work in Python 3.

Runs in a textual console using ASCII art.

../_images/space_invaders.jpg

To run:

  1. python -m examples.space_invaders.space_invaders

While it runs, watch the SQL output in the log:

  1. tail -f space_invaders.log

enjoy!

Listing of files:

XML Persistence

Illustrates three strategies for persisting and querying XML documents as represented by ElementTree in a relational database. The techniques do not apply any mappings to the ElementTree objects directly, so are compatible with the native cElementTree as well as lxml, and can be adapted to suit any kind of DOM representation system. Querying along xpath-like strings is illustrated as well.

E.g.:

  1. # parse an XML file and persist in the database
  2. doc = ElementTree.parse("test.xml")
  3. session.add(Document(file, doc))
  4. session.commit()
  5. # locate documents with a certain path/attribute structure
  6. for document in find_document('/somefile/header/field2[@attr=foo]'):
  7. # dump the XML
  8. print(document)

Listing of files:

  • pickle_type.py - illustrates a quick and dirty way to persist an XML document expressed using ElementTree and pickle.

  • adjacency_list.py - Illustrates an explicit way to persist an XML document expressed using ElementTree.

  • optimized_al.py - Uses the same strategy as adjacency_list.py, but associates each DOM row with its owning document row, so that a full document of DOM nodes can be loaded using O(1) queries - the construction of the “hierarchy” is performed after the load in a non-recursive fashion and is more efficient.

Versioning Objects

Versioning with a History Table

Illustrates an extension which creates version tables for entities and stores records for each change. The given extensions generate an anonymous “history” class which represents historical versions of the target object.

Compare to the Versioning using Temporal Rows examples which write updates as new rows in the same table, without using a separate history table.

Usage is illustrated via a unit test module test_versioning.py, which can be run via pytest:

  1. # assume SQLAlchemy is installed where pytest is
  2. cd examples/versioned_history
  3. pytest test_versioning.py

A fragment of example usage, using declarative:

  1. from history_meta import Versioned, versioned_session
  2. Base = declarative_base()
  3. class SomeClass(Versioned, Base):
  4. __tablename__ = 'sometable'
  5. id = Column(Integer, primary_key=True)
  6. name = Column(String(50))
  7. def __eq__(self, other):
  8. assert type(other) is SomeClass and other.id == self.id
  9. Session = sessionmaker(bind=engine)
  10. versioned_session(Session)
  11. sess = Session()
  12. sc = SomeClass(name='sc1')
  13. sess.add(sc)
  14. sess.commit()
  15. sc.name = 'sc1modified'
  16. sess.commit()
  17. assert sc.version == 2
  18. SomeClassHistory = SomeClass.__history_mapper__.class_
  19. assert sess.query(SomeClassHistory).\
  20. filter(SomeClassHistory.version == 1).\
  21. all() \
  22. == [SomeClassHistory(version=1, name='sc1')]

The Versioned mixin is designed to work with declarative. To use the extension with classical mappers, the _history_mapper function can be applied:

  1. from history_meta import _history_mapper
  2. m = mapper(SomeClass, sometable)
  3. _history_mapper(m)
  4. SomeHistoryClass = SomeClass.__history_mapper__.class_

The versioning example also integrates with the ORM optimistic concurrency feature documented at Configuring a Version Counter. To enable this feature, set the flag Versioned.use_mapper_versioning to True:

  1. class SomeClass(Versioned, Base):
  2. __tablename__ = 'sometable'
  3. use_mapper_versioning = True
  4. id = Column(Integer, primary_key=True)
  5. name = Column(String(50))
  6. def __eq__(self, other):
  7. assert type(other) is SomeClass and other.id == self.id

Above, if two instance of SomeClass with the same version identifier are updated and sent to the database for UPDATE concurrently, if the database isolation level allows the two UPDATE statements to proceed, one will fail because it no longer is against the last known version identifier.

Listing of files:

Versioning using Temporal Rows

Several examples that illustrate the technique of intercepting changes that would be first interpreted as an UPDATE on a row, and instead turning it into an INSERT of a new row, leaving the previous row intact as a historical version.

Compare to the Versioning with a History Table example which writes a history row to a separate history table.

Listing of files:

  • versioned_rows.py - Illustrates a method to intercept changes on objects, turning an UPDATE statement on a single row into an INSERT statement, so that a new row is inserted with the new data, keeping the old row intact.

  • versioned_rows_w_versionid.py - Illustrates a method to intercept changes on objects, turning an UPDATE statement on a single row into an INSERT statement, so that a new row is inserted with the new data, keeping the old row intact.

  • versioned_map.py - A variant of the versioned_rows example built around the concept of a “vertical table” structure, like those illustrated in Vertical Attribute Mapping examples.

  • versioned_update_old_row.py - Illustrates the same UPDATE into INSERT technique of versioned_rows.py, but also emits an UPDATE on the old row to affect a change in timestamp. Also includes a QueryEvents.before_compile() hook to limit queries to only the most recent version.

Vertical Attribute Mapping

Illustrates “vertical table” mappings.

A “vertical table” refers to a technique where individual attributes of an object are stored as distinct rows in a table. The “vertical table” technique is used to persist objects which can have a varied set of attributes, at the expense of simple query control and brevity. It is commonly found in content/document management systems in order to represent user-created structures flexibly.

Two variants on the approach are given. In the second, each row references a “datatype” which contains information about the type of information stored in the attribute, such as integer, string, or date.

Example:

  1. shrew = Animal(u'shrew')
  2. shrew[u'cuteness'] = 5
  3. shrew[u'weasel-like'] = False
  4. shrew[u'poisonous'] = True
  5. session.add(shrew)
  6. session.flush()
  7. q = (session.query(Animal).
  8. filter(Animal.facts.any(
  9. and_(AnimalFact.key == u'weasel-like',
  10. AnimalFact.value == True))))
  11. print('weasel-like animals', q.all())

Listing of files:

Inheritance Mapping Recipes

Basic Inheritance Mappings

Working examples of single-table, joined-table, and concrete-table inheritance as described in Mapping Class Inheritance Hierarchies.

Listing of files:

  • joined.py - Joined-table (table-per-subclass) inheritance example.

  • concrete.py - Concrete-table (table-per-class) inheritance example.

  • single.py - Single-table (table-per-hierarchy) inheritance example.

Special APIs

Attribute Instrumentation

Examples illustrating modifications to SQLAlchemy’s attribute management system.

Listing of files:

Horizontal Sharding

A basic example of using the SQLAlchemy Sharding API. Sharding refers to horizontally scaling data across multiple databases.

The basic components of a “sharded” mapping are:

  • multiple Engine instances, each assigned a “shard id”. These Engine instances may refer to different databases, or different schemas / accounts within the same database, or they can even be differentiated only by options that will cause them to access different schemas or tables when used.

  • a function which can return a single shard id, given an instance to be saved; this is called “shard_chooser”

  • a function which can return a list of shard ids which apply to a particular instance identifier; this is called “id_chooser”.If it returns all shard ids, all shards will be searched.

  • a function which can return a list of shard ids to try, given a particular Query (“query_chooser”). If it returns all shard ids, all shards will be queried and the results joined together.

In these examples, different kinds of shards are used against the same basic example which accommodates weather data on a per-continent basis. We provide example shard_chooser, id_chooser and query_chooser functions. The query_chooser illustrates inspection of the SQL expression element in order to attempt to determine a single shard being requested.

The construction of generic sharding routines is an ambitious approach to the issue of organizing instances among multiple databases. For a more plain-spoken alternative, the “distinct entity” approach is a simple method of assigning objects to different tables (and potentially database nodes) in an explicit way - described on the wiki at EntityName.

Listing of files:

  • separate_databases.py - Illustrates sharding using distinct SQLite databases.

  • separate_tables.py - Illustrates sharding using a single SQLite database, that will however have multiple tables using a naming convention.

Extending the ORM

ORM Query Events

Recipes which illustrate augmentation of ORM SELECT behavior as used by Session.execute() with 2.0 style use of select(), as well as the 1.x style Query object.

Examples include demonstrations of the with_loader_criteria() option as well as the SessionEvents.do_orm_execute() hook.

As of SQLAlchemy 1.4, the Query construct is unified with the Select construct, so that these two objects are mostly the same.

Listing of files:

  • temporal_range.py - Illustrates a custom per-query criteria that will be applied to selected entities.

  • filter_public.py - Illustrates a global criteria applied to entities of a particular type.

Dogpile Caching

Illustrates how to embed dogpile.cache functionality with ORM queries, allowing full cache control as well as the ability to pull “lazy loaded” attributes from long term cache.

In this demo, the following techniques are illustrated:

  • Using the SessionEvents.do_orm_execute() event hook

  • Basic technique of circumventing Session.execute() to pull from a custom cache source instead of the database.

  • Rudimental caching with dogpile.cache, using “regions” which allow global control over a fixed set of configurations.

  • Using custom UserDefinedOption objects to configure options in a statement object.

See also

Re-Executing Statements - includes a general example of the technique presented here.

E.g.:

  1. # query for Person objects, specifying cache
  2. stmt = select(Person).options(FromCache("default"))
  3. # specify that each Person's "addresses" collection comes from
  4. # cache too
  5. stmt = stmt.options(RelationshipCache(Person.addresses, "default"))
  6. # execute and results
  7. result = session.execute(stmt)
  8. print(result.scalars.all())

To run, both SQLAlchemy and dogpile.cache must be installed or on the current PYTHONPATH. The demo will create a local directory for datafiles, insert initial data, and run. Running the demo a second time will utilize the cache files already present, and exactly one SQL statement against two tables will be emitted - the displayed result however will utilize dozens of lazyloads that all pull from cache.

The demo scripts themselves, in order of complexity, are run as Python modules so that relative imports work:

  1. python -m examples.dogpile_caching.helloworld
  2. python -m examples.dogpile_caching.relationship_caching
  3. python -m examples.dogpile_caching.advanced
  4. python -m examples.dogpile_caching.local_session_caching

Listing of files:

  • environment.py - Establish data / cache file paths, and configurations, bootstrap fixture data if necessary.

  • caching_query.py - Represent functions and classes which allow the usage of Dogpile caching with SQLAlchemy. Introduces a query option called FromCache.

  • model.py - The datamodel, which represents Person that has multiple Address objects, each with PostalCode, City, Country.

  • fixture_data.py - Installs some sample data. Here we have a handful of postal codes for a few US/Canadian cities. Then, 100 Person records are installed, each with a randomly selected postal code.

  • helloworld.py - Illustrate how to load some data, and cache the results.

  • relationship_caching.py - Illustrates how to add cache options on relationship endpoints, so that lazyloads load from cache.

  • advanced.py - Illustrate usage of Query combined with the FromCache option, including front-end loading, cache invalidation and collection caching.

  • local_session_caching.py - This example creates a new dogpile.cache backend that will persist data in a dictionary which is local to the current session. remove() the session and the cache is gone.

PostGIS Integration

A naive example illustrating techniques to help embed PostGIS functionality.

This example was originally developed in the hopes that it would be extrapolated into a comprehensive PostGIS integration layer. We are pleased to announce that this has come to fruition as GeoAlchemy.

The example illustrates:

  • a DDL extension which allows CREATE/DROP to work in conjunction with AddGeometryColumn/DropGeometryColumn

  • a Geometry type, as well as a few subtypes, which convert result row values to a GIS-aware object, and also integrates with the DDL extension.

  • a GIS-aware object which stores a raw geometry value and provides a factory for functions such as AsText().

  • an ORM comparator which can override standard column methods on mapped objects to produce GIS operators.

  • an attribute event listener that intercepts strings and converts to GeomFromText().

  • a standalone operator example.

The implementation is limited to only public, well known and simple to use extension points.

E.g.:

  1. print(session.query(Road).filter(
  2. Road.road_geom.intersects(r1.road_geom)).all())

Listing of files: