lazr.enum 1.2.1

Enums with zope.schema vocabulary support and database-friendly conveniences.

Enumerated types are used primarily in two distinct places in the Launchpad code: selector types; and database types.

Simple enumerated types do not have values, whereas database enumerated types are a mapping from an integer value to something meaningful in the code.

>>> from lazr.enum import (
...     EnumeratedType, DBEnumeratedType, Item, DBItem, use_template)

The enum values of EnumeratedTypes are instances of Item.

>>> class Fruit(EnumeratedType):
...     "A choice of fruit."
...     APPLE = Item('Apple')
...     PEAR = Item('Pear')
...     ORANGE = Item('Orange')

IVocabulary support

Enumerated types support IVocabularyTokenized.

>>> from zope.interface.verify import verifyObject
>>> from zope.schema.interfaces import (
...     ITitledTokenizedTerm, IVocabularyTokenized)
>>> verifyObject(IVocabularyTokenized, Fruit)
True

The items themselves do not support any interface. Items returned by the methods for vocabularies return wrapped items that support the ITitledTokenizedTerm interface.

The token used to identify terms in the vocabulary is the name of the Item variable.

>>> item = Fruit.getTermByToken('APPLE')
>>> type(item)
<class 'lazr.enum...TokenizedItem'>
>>> verifyObject(ITitledTokenizedTerm, item)
True

TokenizedItems have three attributes (in order to support ITitledTokenizedTerm):

>>> item.value
<Item Fruit.APPLE, Apple>
>>> item.token
'APPLE'
>>> item.title
'Apple'

>>> Fruit.getTermByToken('apple').value
<Item Fruit.APPLE, Apple>

The length of an EnumeratedType returns the number of items it has.

>>> print(len(Fruit))
3

The EnumeratedType registry

All enumerated types that are created are added to the enumerated_type_registry.

>>> from lazr.enum import enumerated_type_registry

The enumerated_type_registry maps the name of the enumerated type to the type itself.

>>> 'Fruit' in enumerated_type_registry
True
>>> enumerated_type_registry['Fruit']
<EnumeratedType 'Fruit'>

You cannot redefine an existing enumerated type, nor create another enumerated type with the same name as an existing type.

>>> class BranchType(EnumeratedType):
...     BAR = Item('Bar')
...
>>> BranchType.name = 'AltBranchType'
>>> class BranchType(EnumeratedType):
...     FOO = Item('Foo')
Traceback (most recent call last):
...
TypeError: An enumerated type already exists with the name BranchType
(...AltBranchType).

Enumerated Type basics

An EnumeratedType has a name and a description. The name is the same as the class name, and the description is the docstring for the class.

>>> print(Fruit.name)
Fruit
>>> print(Fruit.description)
A choice of fruit.

If you do not specify an explicit sort_order for the items of the EnumeratedType one is created for you. This is tuple of the tokens.

>>> print(Fruit.sort_order)
('APPLE', 'PEAR', 'ORANGE')

The items of an enumerated type can be iterated over. However the type that is returned by the iteration is the TokenizedItem, not the item itself.

>>> for item in Fruit:
...     print(item.token, item.title)
APPLE Apple
PEAR Pear
ORANGE Orange

Items can also optionally have a url associated with them.

>>> class Guitar(EnumeratedType):
...     FENDER = Item('Fender', url='http://www.fender.com')
...     RICK = Item('Rickenbacker', url='http://www.rickenbacker.com')
...     GIBSON = Item('Gibson', url='http://www.gibson.com')
...     FRANKENBASS = Item('Home built')

>>> print(Guitar.FENDER.url)
http://www.fender.com
>>> print(Guitar.FRANKENBASS.url)
None

Items in an enumerator support comparison and equality checks. Comparison is based on the sort order of the items.

>>> apple = Fruit.APPLE
>>> pear = Fruit.PEAR
>>> orange = Fruit.ORANGE
>>> apple < pear
True
>>> apple == pear
False
>>> apple == apple
True
>>> apple != pear
True
>>> apple > pear
False
>>> pear < orange
True
>>> apple < orange
True

Items which are not in an enumerator always compare as False.

>>> import warnings
>>> from lazr.enum import Item
>>> with warnings.catch_warnings():
...     warnings.simplefilter('ignore', category=UserWarning)
...     Item('a') == Item('b')
False

The string representation of an Item is the title, and the representation also shows the enumeration that the Item is from.

>>> print(apple)
Apple
>>> print(repr(apple))
<Item Fruit.APPLE, Apple>

The items attribute of an enumerated type is not a list, but a class that provides iteration over the items, and access to the Item attributes through either the name of the Item, or the database value if there is one.

The primary use of this is to provide a backwards compatible accessor for items, but it also provides a suitable alternative to getattr.

>>> name = 'APPLE'
>>> Fruit.items[name]
<Item Fruit.APPLE, Apple>
>>> getattr(Fruit, name)
<Item Fruit.APPLE, Apple>

Database Enumerated Types

A very common use of enumerated types are to give semantic meaning to integer values stored in database columns. EnumeratedType Items themselves don’t have any integer values.

The DBEnumeratedType provides the semantic framework for a type that is used to map integer values to python enumerated values.

>>> # Remove the existing reference to BranchType from the registry
>>> del enumerated_type_registry['BranchType']
>>> class BranchType(DBEnumeratedType):
...     HOSTED = DBItem(1, """
...         Hosted
...
...         Hosted braches use the supermirror as the main repository
...         for the branch.""")
...
...     MIRRORED = DBItem(2, """
...         Mirrored
...
...         Mirrored branches are "pulled" from a remote location.""")
...
...     IMPORTED = DBItem(3, """
...         Imported
...
...         Imported branches are natively maintained in CVS or SVN""")

Note carefully that the value of a DBItem is the integer representation. But the value of the TokenizedItem is the DBItem itself.

>>> hosted = BranchType.HOSTED
>>> hosted.value
1
>>> hosted == BranchType.HOSTED
True
>>> tokenized_item = BranchType.getTermByToken('HOSTED')
>>> tokenized_item.value
<DBItem BranchType.HOSTED, (1) Hosted>

DBEnumeratedTypes also support IVocabularyTokenized

>>> verifyObject(IVocabularyTokenized, BranchType)
True

The items attribute of DBEnumeratedTypes provide a mapping from the database values to the DBItems.

>>> BranchType.items[3]
<DBItem BranchType.IMPORTED, (3) Imported>

The items also support the url field.

>>> class Bass(DBEnumeratedType):
...     FENDER = DBItem(10, 'Fender', url='http://www.fender.com')
...     RICK = DBItem(20, 'Rickenbacker',
...                   url='http://www.rickenbacker.com')
...     GIBSON = DBItem(30, 'Gibson', url='http://www.gibson.com')
...     FRANKENBASS = DBItem(40, 'Home built')

>>> print(Bass.FENDER.url)
http://www.fender.com
>>> print(Bass.FRANKENBASS.url)
None

Items in a DBEnumeratedType class must be of type DBItem.

>>> class BadItemType(DBEnumeratedType):
...     TESTING = Item("Testing")
Traceback (most recent call last):
...
TypeError: Items must be of the appropriate type for the DBEnumeratedType,
...builtin...BadItemType.TESTING

You are not able to define a DBEnumeratedType that has two different DBItems that map to the same numeric value.

>>> class TwoMapping(DBEnumeratedType):
...     FIRST = DBItem(42, 'First')
...     SECOND = DBItem(42, 'Second')
Traceback (most recent call last):
...
TypeError: Two DBItems with the same value 42 (FIRST, SECOND)

Overriding the sort order

By default the sort order of the items in an enumerated type is defined by the order in which the Items are declared. This my be overridden by specifying a sort_order attribute in the class.

If a sort_order is specified, it has to specify every item in the enumeration.

>>> class AnimalClassification(EnumeratedType):
...     sort_order = "REPTILE", "INSECT", "MAMMAL"
...     INSECT = Item("Insect")
...     MAMMAL = Item("Mammal")
...     FISH = Item("Fish")
...     REPTILE = Item("Reptile")
Traceback (most recent call last):
...
TypeError: sort_order for EnumeratedType must contain all and only Item instances ...

The sort_order may also appear anywhere in the definition of the class, although convention has it that it appears first, before the Item instances.

>>> class AnimalClassification(EnumeratedType):
...     sort_order = "REPTILE", "FISH", "INSECT", "MAMMAL"
...     INSECT = Item("Insect")
...     MAMMAL = Item("Mammal")
...     FISH = Item("Fish")
...     REPTILE = Item("Reptile")

The items attribute of the enumerated type is ordered based on the sort_order. The items attribute is also used to control iteration using __iter__.

>>> for item in AnimalClassification.items:
...     print(item.title)
Reptile
Fish
Insect
Mammal

The sort order also drives the comparison operations.

>>> reptile, fish, insect, mammal = AnimalClassification.items
>>> reptile < fish < insect < mammal
True

Extending enumerated types

The simplest way to extend a class is to derive from it.

>>> class AnimalClassificationExtended(AnimalClassification):
...     INVERTEBRATE = Item("Invertebrate")

>>> for item in AnimalClassificationExtended:
...     print(item.title)
Reptile
Fish
Insect
Mammal
Invertebrate

The use_template function inserts the items from the specified enumerated type into the new enumerated type. The default case is to take all the enumerated items.

>>> class UIBranchType(EnumeratedType):
...     use_template(BranchType)
>>> for item in UIBranchType:
...     print(item.title)
Hosted
Mirrored
Imported

You can also specify items to be excluded by referring to the attribute name in the exclude parameter. This can be either a string referring to one name or a tuple or list that refers to multiple attribute names.

>>> class UIBranchType2(EnumeratedType):
...     use_template(BranchType, exclude='IMPORTED')
>>> for item in UIBranchType2:
...     print(item.title)
Hosted
Mirrored

Or limit the items to those specified:

>>> class UIBranchType3(EnumeratedType):
...     use_template(BranchType, include=('HOSTED', 'MIRRORED'))
>>> for item in UIBranchType3:
...     print(item.title)
Hosted
Mirrored

Getting from an item back to the enumerated type

Each Item in an EnumeratedType has a reference back to the EnumeratedType.

>>> print(repr(apple))
<Item Fruit.APPLE, Apple>
>>> print(repr(apple.enum))
<EnumeratedType 'Fruit'>
>>> for item in apple.enum:
...     print(item.title)
Apple
Pear
Orange

Item.sortkey

The sortkey attribute of the Items are defined by the sort_order that is defined for the enumerated type. The value is often used as a hidden value in columns in order to ensure appropriate sorting.

>>> for item in Fruit.items:
...     print(item.title, item.sortkey)
Apple  0
Pear   1
Orange 2

>>> for item in BranchType.items:
...     print(item.title, item.sortkey)
Hosted   0
Mirrored 1
Imported 2

JSON Support

Enumerated types instances can be serialised to/from json. This library provides the necessary encode and decode classes which can be used directly or as part of the lazr.json package where they are registered as default handlers for lazr enums.

A enum instance is serialised as a dict containing: - the enumerated type name as per the enumerated_type_registry - the enum instance name

>>> import json
>>> from lazr.enum import EnumJSONEncoder

>>> json_enum = json.dumps(
...     Fruit.APPLE, cls=EnumJSONEncoder, sort_keys=True)
>>> print(json_enum)
{"name": "APPLE", "type": "Fruit"}

To deserialse, we can specify a json object_hook as follows. This is done transparently when using the lazr.json package.

>>> def fruit_enum_decoder(value_dict):
...      return EnumJSONDecoder.from_dict(Fruit, value_dict)

>>> from lazr.enum import EnumJSONDecoder
>>> json.loads(json_enum, object_hook=fruit_enum_decoder)
<Item Fruit.APPLE, Apple>

NEWS for lazr.enum

1.2.1 (2021-09-13)

• Adjust versioning strategy to avoid importing pkg_resources, which is slow in large environments.

1.2 (2019-11-24)

• Switch from buildout to tox.

• Add Python 3 support.

1.1.4 (2012-04-18)

• Support for serialising enums to/from json (lp:984549)

• Items which are not in an enumerator always compare as False (lp:524259)

• Fix the licence statement in _enum.py to be LGPLv3 not LGPLv3+ (lp:526484)

1.1.3 (2011-04-20)

• added case insensitivity to getting the term by the token value (lp:154556)

1.1.2 (2009-08-31)

• removed unnecessary build dependencies

1.1.1 (2009-08-06)

• Removed sys.path hack from setup.py.

1.1 (2009-06-08)

• Added url argument to the BaseItem and DBItem constructors.

1.0 (2009-03-24)

• Initial release on PyPI