unpythonic 0.12.0

We provide missing features for Python, mainly from the list processing tradition, but with some haskellisms mixed in. For the adventurous, we include a set of syntactic macros (using MacroPy) that are designed to work together.

We place a special emphasis on clear, pythonic syntax. Design considerations are simplicity, robustness, and minimal dependencies.

Without macros, our features include tail call optimization (TCO), TCO’d loops in FP style, call/ec, let & letrec, assign-once, multi-expression lambdas, dynamic assignment (a.k.a. parameterize, special variables), memoization (also for generators and iterables), currying, function composition, folds and scans (left and right), unfold, lazy partial unpacking of iterables, functional update for sequences, and pythonic lispy linked lists (cons).

Our curry modifies Python’s reduction rules. It passes any extra arguments through on the right, and calls a callable return value on the remaining arguments, so that we can:

mymap = lambda f: curry(foldr, composerc(cons, f), nil)
myadd = lambda a, b: a + b
assert curry(mymap, myadd, ll(1, 2, 3), ll(2, 4, 6)) == ll(3, 6, 9)

with_n = lambda *args: (partial(f, n) for n, f in args)
look = lambda n1, n2: composel(*with_n((n1, drop), (n2, take)))
assert tuple(curry(look, 5, 10, range(20))) == tuple(range(5, 15))

If MacroPy is installed, unpythonic.syntax becomes available. It provides macros that essentially extend the Python language, adding features that would be either complicated or impossible to provide (and/or use) otherwise.

With macros, we add automatic currying, automatic tail-call optimization, continuations (call/cc for Python), let-syntax (splice code at macro expansion time), lexically scoped let and do with lean syntax, implicit return statements, and easy-to-use multi-expression lambdas with local variables.

The TCO macro has a fairly extensive expression analyzer, so things like and, or, a if p else b and any uses of the do[] and let[] macros are accounted for when performing the tail-call transformation.

The continuation system is based on a semi-automated partial conversion into continuation-passing style (CPS), with continuations represented as closures. It also automatically applies TCO, using the same machinery as the TCO macro. To keep the runtime overhead somewhat reasonable, the continuation is captured only where explicitly requested with call_cc[].

Macro examples:

# let, letseq (let*), letrec with no boilerplate
a = let((x, 17),
        (y, 23))[
          (x, y)]

# alternate haskelly syntax
a = let[((x, 21),(y, 17), (z, 4)) in x + y + z]
a = let[x + y + z, where((x, 21), (y, 17), (z, 4))]

# cond: multi-branch "if" expression
answer = lambda x: cond[x == 2, "two",
                        x == 3, "three",
                        "something else"]
assert answer(42) == "something else"

# do: imperative code in expression position
y = do[local[x << 17],
       print(x),
       x << 23,
       x]
assert y == 23

# autocurry like Haskell
with curry:
    def add3(a, b, c):
        return a + b + c
    assert add3(1)(2)(3) == 6
    # actually partial application so these work, too
    assert add3(1, 2)(3) == 6
    assert add3(1)(2, 3) == 6
    assert add3(1, 2, 3) == 6

    mymap = lambda f: foldr(composerc(cons, f), nil)
    myadd = lambda a, b: a + b
    assert mymap(myadd, ll(1, 2, 3), ll(2, 4, 6)) == ll(3, 6, 9)

# automatic tail-call optimization (TCO) like Scheme, Racket
with tco:
    assert letrec((evenp, lambda x: (x == 0) or oddp(x - 1)),
                  (oddp,  lambda x: (x != 0) and evenp(x - 1)))[
                    evenp(10000)] is True

# lambdas with multiple expressions, local variables, and a name
with multilambda, namedlambda:
    myadd = lambda x, y: [print("myadding", x, y),
                          local[tmp << x + y],
                          print("result is", tmp),
                          tmp]
    assert myadd(2, 3) == 5
    assert myadd.__name__ == "myadd"

# implicit "return" in tail position, like Lisps
with autoreturn:
    def f():
        print("hi")
        "I'll just return this"
    assert f() == "I'll just return this"

    def g(x):
        if x == 1:
            "one"
        elif x == 2:
            "two"
        else:
            "something else"
    assert g(1) == "one"
    assert g(2) == "two"
    assert g(42) == "something else"

# splice code at macro expansion time
with let_syntax:
    with block(a) as twice:
        a
        a
    with block(x, y, z) as appendxyz:
        lst += [x, y, z]
    lst = []
    twice(appendxyz(7, 8, 9))
    assert lst == [7, 8, 9]*2

# lispy prefix syntax for function calls
with prefix:
    (print, "hello world")

# the LisThEll programming language
with prefix, curry:
    mymap = lambda f: (foldr, (compose, cons, f), nil)
    double = lambda x: 2 * x
    (print, (mymap, double, (q, 1, 2, 3)))
    assert (mymap, double, (q, 1, 2, 3)) == ll(2, 4, 6)

# call/cc for Python
with continuations:
    stack = []
    def amb(lst, cc):  # McCarthy's amb operator
        if not lst:
            return fail()
        first, *rest = lst
        if rest:
            ourcc = cc
            stack.append(lambda: amb(rest, cc=ourcc))
        return first
    def fail():
        if stack:
            f = stack.pop()
            return f()

    def pythagorean_triples(maxn):
        z = call_cc[amb(tuple(range(1, maxn+1)))]
        y = call_cc[amb(tuple(range(1, z+1)))]
        x = call_cc[amb(tuple(range(1, y+1)))]
        if x*x + y*y != z*z:
            return fail()
        return x, y, z
    x = pythagorean_triples(20)
    while x:
        print(x)
        x = fail()

# if Python didn't already have generators, we could add them with call/cc:
with continuations:
    @dlet((k, None))
    def g():
        if k:
            return k()
        def my_yield(value, cc):
            k << cc        # rebind the k in the @dlet env
            cc = identity  # override current continuation
            return value
        # generator body
        call_cc[my_yield(1)]
        call_cc[my_yield(2)]
        call_cc[my_yield(3)]
    out = []
    x = g()
    while x is not None:
        out.append(x)
        x = g()
    assert out == [1, 2, 3]

For documentation and full examples, see the project’s GitHub homepage, and the docstrings of the individual features. For even more examples, see the unit tests included in the source distribution.