Python Distribution Utilities
Sometimes you write a function over and over again; sometimes you look up at the ceiling and ask “why, Guido, why isn’t this included in the standard library?”
Well, we perhaps can’t answer that question. But we can collect those functions into a centralized place!
Utils is broken up into broad swathes of functionality, to ease the task of remembering where exactly something lives.
Python doesn’t have a built-in way to define an enum, so this module provides (what I think) is a pretty clean way to go about them.
from utils import enum class Colors(enum.Enum): RED = 0 GREEN = 1 # Defining an Enum class allows you to specify a few # things about the way it's going to behave. class Options: frozen = True # can't change attributes strict = True # can only compare to itself; i.e., Colors.RED == Animals.COW # will raise an exception. # or use the enum factory (no Options, though) ColorsAlso = enum.enum("RED", "GREEN")
Once defined, use is straightforward:
>>> Colors <class 'blahblah.Colors'> >>> Colors.RED <EnumItem: RED > >>> Colors.RED == 0 True >>> Colors.RED == Colors.RED True >>> Colors.RED = 2 Traceback (most recent call last): File "<stdin>", line 1, in <module> File "utils/enum.py", line 114, in __setattr__ raise TypeError("can't set attributes on a frozen enum") TypeError: can't set attributes on a frozen enum
Currently only has the multiplicative analogue of the built-in sum.
intersections, differences, winnowing, a few specialized dicts…
flatten and unlisting. also flat_map!
currently only provides an xor function.
Mostly cool for the TimePeriod classes:
>>> from datetime import date # will also work with datetimes >>> time_period = TimePeriod(date(2013, 5, 10), date(2013, 8, 11)) >>> time_period <TimePeriod: 2013-05-10 00:00:00-2013-08-11 23:59:59> >>> date(2013, 6, 12) in time_period True >>> other_time_period = TimePeriod(date(2013, 6, 1), date(2013, 6, 30)) >>> other_time_period in time_period True >>> another_time_period = TimePeriod(date(2013, 8, 1), date(2013, 8, 30)) >>> time_period.overlaps(another_time_period) True >>> TimePeriod.get_containing_period(time_period, another_time_period) <TimePeriod: 2013-05-08 00:00:00-2013-08-30 23:59:59>
and so on and so forth. There’s also a DiscontinousTimePeriod class, which stores a collection of TimePeriods.
There’s also helper functions for common operations like days_ahead and days_ago, which pretty much do what they say on the tin.
provides get_attr, which is really just a convenient way to do deep getattr chaining:
>>> get_attr(complicated, 'this.is.a.deep.string', default=None) "the deep string" # or None, if anything in the lookup chain didn't exist
There’s also an immutable utility, which will wrap an object and preven all attribute changes, recursively by default. Any attempt to set attributes on the wrapped object will raise an AttributeError:
>>> imm = immutable(something) >>> imm <Immutable Something: <Something>> >>> imm.red <Immutable SomethingElse: <SomethingElse: red>> >>> imm.red = SomethingElse('blue') # ... AttributeError: This object has been marked as immutable; you cannot set its attributes. >>> something.red = SomethingElse('blue') >>> imm.red <Immutable SomethingElse: <SomethingElse: blue>>
You can toggle the recursive immutability by specifying the ‘recursive’ flag.