nr.types 2.0.0

Provides a number os useful data types for day-to-day Python programming.

nr.types

The nr.types package provides a number os useful data types for day-to-day Python programming. It is compatible with Python 2.7 and modern versions of Python 3.

Installation

pip install nr.types

Run Tests

pip install -e .[test]
pytest --cov=./src/nr

API

nr.types.NotSet

The NotSet singleton is useful in cases where None is an acceptable value for a parameter so there needs to be an additional state that defines the parameter as "not set".

nr.types.abc

An alias for collections.abc or collections (the six module does not provide a move for these modules).

nr.types.functools

Tools to help with Python function internals such as closures, code and function objects.

Example:

import nr.types.functools as ft
def test(value):
  def x():
    return value
  return x
x = test(42)
assert x() == 42
y = ft.copy_function(x, closure={'value': 99})
assert y() == 99

nr.types.generic

Allows you to implement generic types, ie. classes with parameters.

Example:

from nr.types import generic
class HashDict(generic.Generic['key_hash']):
  def __init__(self):
    generic.assert_initialized(self)
    self.data = {}
  def __getitem__(self, key):
    return self.data[self.key_hash(key)]
  def __setitem__(self, key, value):
    self.data[self.key_hash(key)] = value
UnsafeHashDict = HashDict[hash]

nr.types.interface

Similar to zope.interface, but Python 3 compatible and less magical.

Example:

from nr.types.interface import Interface, Implementation, implements, attr

class IFoo(Interface):
  """ The foo interface. """

  x = attr("""Some attribute.""")

  def bar(self, q, r=None):
    """ The bar function. """

assert set(IFoo) == set(['x', 'bar'])
assert not hasattr(IFoo, 'x')
assert not hasattr(IFoo, 'bar')
assert IFoo['x'].name == 'x'
assert IFoo['bar'].name == 'bar'

@implements(IFoo)
class Foo(object):

  def __init__(self, x=None):
    self.x = x

  def bar(self, q, r=None):
    return q, r, self.x

assert issubclass(Foo, Implementation)
assert IFoo.implemented_by(Foo)
assert IFoo.provided_by(Foo())
assert list(IFoo.implementations()) == [Foo]
assert Foo(42).x == 42

nr.types.maps

Provides the following mapping (and mapping-related) implementations:

• OrderedDict
• ObjectAsDict
• ObjectFromDict
• ChainDict (use maps.chain())
• HashDict[hash_func]
• ValueIterableDict

nr.types.meta

Provides useful metaclasses, such as InlineMetaclass.

Example:

from nr.types.meta import InlineMetaclassBase
class MyClass(InlineMetaclassBase):
  def __metainit__(self, name, bases, attr):
    print('MyClass constructed!')
    self.value = 'foo'
assert MyClass.value == 'foo'

nr.types.moduletools

Provides some tools for working with modules. Currently only provides the make_inheritable() function which can be used from within your module to make the module object itself usable as a parent class.

# myclass.py
class MyClass(object): pass
make_inheritable(__name__)

# test.py
import myclass
class MySubclass(myclass): pass
assert issubclass(MySubclass, myclass.MyClass)

nr.types.record

Similar to namedtuple but mutable, with support for keyword arguments, type declarations and default values. Supports multiple forms of declaring a record, eg. via Python 3.6+ class-level annotations, specifying a class-level __fields__ member or declaring attributes by creating record.Field() objects.

Example:

import random
from nr.types import record

class Person(record.Record):
  name: str
  mail: str = None
  age: int = lambda: random.randint(10, 50)

p = Person('John Smith')
assert p.name == 'John Smith'
assert p.mail is None
assert 10 <= p.age <= 50

Alternatives:

import random
from nr.types import record

class Person(record.Record):
  name = record.Field(str)
  mail = record.Field(str, None)
  age = record.Field(str, lambda: random.randint(10, 50))

class Person(record.Record):
  __fields__ = [
    ('name', str),
    ('mail', str, None),
    ('age', str, lambda: random.randint(10, 50)),
  ]

Person = record.create_record('Person', [
  ('name', str),
  record.Field.with_name('mail', str, None),
  ('age', str, lambda: random.randint(10, 50))
])

Person = record.create_record('Person', {
  'name': record.Field(str),
  'mail': record.Field(str, None),
  'age': record.Field(str, lambda: random.randint(10, 50))
})

assert list(Person.__fields__.keys()) == ['name', 'mail', 'age']

nr.types.sets

Currently only provides an OrderedSet implementation.

nr.types.stream

Example:

from nr.types import stream
stream(range(10)).map(lambda x: x*2)
stream.map(range(10), lambda x: x*2)

nr.types.sumtype

Example:

from nr.types import record, sumtype

class Filter(sumtype):
  Date = record.create_record('Date', 'min,max')
  Keyword = sumtype.constructor('text')
  class Duration(record.Record):
    value = record.Field(int, lambda: 3600)

f = Filter.Keyword('building')
assert isinstance(f, Filter)
assert f.is_keyword()
assert f.text == 'building'

f = Filter.Date(10, 42)
assert isinstance(f, Filter)
assert f.is_date()
assert (f.min, f.max) == (10, 42)

f = Filter.Duration()
assert isinstance(f, Filter)
assert f.is_duration()
assert f.value == 3600

---

Copyright © Niklas Rosenstein 2019