Installation

pip install hawkweed

Usage

hawkweed is roughly divided into three different parts: datatypes, monads and functions. All the functions are exhaustively documented with pydoc, all the parameters, the functions’ time complexity (if applicable) and the return value are given.

Datatypes

Most of the datatypes implemented in hawkweed are mere wrappers around Python standard datatypes. If the function does not return anything in the Python datatype, this implementation will return self to allow for chaining.

A notable exception is the largely unstable and undocumented Future class.

from hawkweed import List, Dict, Set

List([1]).append(2).extend([3, None, 4]).remove_empty() # => List([1, 2, 3, 4])
List(range(10)).take(5) # => generator from 0 to 4
List(range(10)).drop(5) # => generator from 5 to 9
List(range(10)).take_while(lambda x: x < 5) # => generator from 0 to 4
List(range(10)).drop_while(lambda x: x < 5) # => generator from 4 to 9
List(range(10)).nth(3) # => generator yielding 0, 3, 6 and 9 (lazily); works with any subclass of Iterable
List(range(10)).reset(range(5)) # => List([0, 1, 2, 3, 4])

Dict({1: 2, 3: 4}).reverse() # => Dict({2: 1, 4: 3})
Dict({1: 2, 3: 4, 2: None}).remove_empty() # => Dict({1: 2, 3: 4})
Dict({1: 2, 3: 4, None: "go away"}).remove_empty(filter_keys=True) # => Dict({1: 2, 3: 4})
Dict({1: 2, 3: 4, 2: 3}).remove_empty(fun=lambda x: x!=2) # => Dict({1: 2, 3: 4})
Dict({1: 2, 3: 4}).reduce(fun=lambda acc, k, v: acc + k + v, acc=0) # => 10
Dict({1: 2, 3: 4}).reduce(fun=lambda acc, k, v: acc + (k, v)) # => (1, 2, 3, 4)
Dict({1: 2, 3: 4, 5: 6}).pick(1, 5) # => Dict({1: 2, 5: 6})

Set({1, 2, 3, 4}).remove_empty(fun=lambda x: x!=3) # => Set({1, 2, 4})

# And now for something completely different
Dict({
  "foo": List([1, 2, 3, Dict({"bar": "baz"})])
}).get_in("foo", 3, "bar") # => "baz"
Dict({
  "foo": List([1, 2, 3, Dict({"bar": "baz"})])
}).get_in("foo", 100, "bar") # => None
Dict({
  "foo": List([1, 2, 3, Dict({"bar": "baz"})])
}).get_in("foo", 100, "bar", dflt="i am a default value") # => "i am a default value"
Dict({
  "foo": List([1, 2, 3, Dict({"bar": "baz"})])
}).update_in("foo", 1, "bar", to="update") # => Dict({"foo": List([1, 2, 3, Dict({"bar": "update"})])})
# if you want to insert your own datatype, just inherit from hawkweed.Collection
# and implement get(key, dflt=None) and __setitem__

Functions

All of the functions are standalone and curried whenever possible. They do not depend on hawkweeds datatypes in any way.

from hawkweed import map, reduce, List, all, any, constantly, delay

map(inc, range(100)) # => range(1, 101)
incrementor = map(inc)
incrementor(List(range(100))) # => range(1, 101)
summator = reduce(add)
summator(range(5)) # => 10
all(lambda x: x > 100, [101, 102, 103]) # => True
any(lambda x: x > 10, [3, 5, 8]) # => False
constantly(10) # => an infinite generator of 10
delayed = delay(print, 'Hello, World!') # => this will return a variable that, when called, will compute the result of print with the argument 'Hello, World!'
# it will cache the result instead of recomputing it upon reevaluation, i.e. `delayed() or delayed()` will only print 'Hello, World!' once

A few other functions that you might expect from a functional programming library (compose, pipe, identity, apply, flip, curry and the like) are also implemented. They should be intuitive and work as expected. If they do not or are not consider it a bug.

Monads

The implemented monads are: Identity, Maybe (Just/Nothing), Continuation, Either, IO, CachedIO, and List (called ListM). do notation is also supported.

from hawkweed import doM, wrapM, Just

def doMe():
  res1 = yield Just(1)
  res2 = yield Just(10)
  yield Just(res1+ res2)

doM(doMe()) # => Just(11)

wrapM(Just(10)).real
# => 10; the wrapper will try to call the wrapped values' function whenever it does not exist in the monad

There is a callcc function and all of the functions in Haskell’s Data.Maybe and Data.Either are implemented.

Have fun!