lambdak

Full anonymous functions for Python

Lambdak is pronounced 'lambda-k'.

The goal is to wrap up all Python statements inside functions so that they can be run in places which require expressions, e.g. the body of a lambda expression. And, using continuations lets us build expressions which are effectively composed of multiple statements.

This lets us create the Holy Grail of Python, multi-line lambdas. Some examples:

actions = {
  "hello":
    print_("Hello, World!", lambda:

    import_("math", lambda m:
    print_("The area of my circle is:", lambda:
    print_(m.pi * 5 * 5)))),

  "goodbye":
    print_("Goodbye, Cruel World!", lambda:

    try_(lambda: 1 / 0,
      except_ = const_(print_("Danger, Will Robinson!")))) }

actions["hello"]()
actions["goodbye"]()

circumference = given_(lambda r:
  import_("math", lambda m:
  2 * m.pi * r))

print circumference(5)

(Note: I highly recommend using an editor extension like vim-cute-python to prettify the code for readability. For example, it would represent the keyword lambda with the symbol 'λ', without changing the underlying source code.)

News
Overview
Reference

Anyway, what's with all the lambdas?

We can't escape the fact that Python allows only one expression inside a lambda block. So, we use more lambdas inside that one expression to 'continue' our computations for as long as we want. Internally, the functions are designed to conserve stack space and avoid stack overflow.

News

2015-01-17: just published a page gathering tips and tricks for Pythonistas using Vim.

Overview

The central concept in this module is the lambdak. This is a callable type which behaves like a normal Python lambda, except that it can be composed with more lambdaks (or with other values) so that it can execute an unlimited number of statements and expressions. And it does so in a way that preserves memory.

The lambdak functions that you'll see below mostly follow this pattern:

def something_(..., k = None): ...

They mostly take some arguments up front that they need to carry out some action; and then they take a final optional argument (k, the 'continuation'), which can be either:

A function that returns another lambdak, to indicate that we want to continue the computation; or
Anything else, to indicate that this should be the final result of the computation.

You can imagine a chain of lambdaks, each one calling the next one in the chain and passing on any relevant values from its computation. Sometimes there is no relevant value; for example if you print two things one after the other there's no useful value from the first print action that can be passed on the the next print action. Lambdaks automatically handle all of that. You just compose them with the required values, and lots of lambdas.

Reference

The lambdak module is designed to be 'star-imported' (from lambdak import *): the functions below have all been named with an underscore character ('_') as the last character.

`call_`

Call the given function with optional arguments.

Arguments

k. The function to call. It will be called with any arguments as specified below.
*args. If any arguments are given here, they will be exploded and passed to k.

Returns

The result returned by k, or None if k is None.

`return_`

Simple return the value that is passed in to it. This is the identity function for any type.

Arguments

x. Any value.

Returns

x itself.

Example

See a useful example in the documentation for the cond_ function.

`const_`

A convenience function that takes a value and returns a function which always returns that value regardless of its input. Think of it as 'constantly' returning that value.

Arguments

x. Any value.

Returns

A function which takes a single argument and always returns x no matter what.

Example

See the example for the try_ lambdak.

`given_`

Receive arguments at the beginning of a lambdak chain so you can call the lambdak with those arguments. These arguments can also have default values, which effectively lets you bind names to values for the duration of the k closure's scope.

Arguments

k. A callable (usually a lambda) that takes any number of arguments and returns either a final value or another lambdak (to continue the computation).

Returns

A lambdak that can be called with the same number of arguments that are accepted by the k parameter. If the lambdak is called with no arguments, it will call k with no arguments.

Example

This example shows both uses of given_: providing a way to pass in arguments to call the outermost lambdak with, and also providing a way to bind names to values inside a lambdak.

f = given_(lambda x:
  given_(lambda y = 2 * x:
    print_(y)))

f(2)

Output:

You can think of this as, 'Given x, let y be twice x; print y'.

`do_`

Meant to be used when you just want to evaluate an expression which has a side effect, and then optionally carry on with other actions or calculations.

Arguments

expr_k. Any expression that we want to evaluate for its side effects. Must be a callable (usually a lambda expression wrapping a value or function call). Will be called (evaluated) when the do_ lambdak is eventually run. We want to delay evaluation because the expression may have side effects, so we want to keep them in sequence.
k. Optional (default None). If supplied, must be a callable which does not accept any arguments and returns any value. This contains the rest of the computation (i.e. the rest of the lambdak chain).

Returns

The same as given_ would return.

Example

def hello(): print "Hello!"
def hi(): print "Hi!"

test = (
  do_(lambda: hello(), lambda:
  do_(lambda: hi())))

Output: nothing yet.

test()

Output:

Hello!
Hi!

`print_`

Print a single expression and optionally carry on the computation.

Arguments

x. The expression to print. This is passed to Python's print statement.
k. Optional (default None). The same as do_.

Returns

The same as given_.

`assert_`

Arguments

expr. The expression to assert. This is passed to Python's assert statement.
k. Optional. The same as do_.

Returns

The same as given_.

Example

f = (
  assert_(True, lambda:
  print_("OK!")))

f()

Output:

OK!

`for_`

Wraps Python's for statement. Fully functional, including the ability to break out of the loop and 'continue' (skip) to the next iteration.

To break out of the loop, simply return the type break_ from the act_k function. To continue (skip), return the type continue_. These types can be referred to by their names; think of them as 'symbols' or as 'tags'. See examples below.

Arguments

seq. An iterable.
act_k. A function that takes one argument and returns the same thing as k (below). It will be called on each iteration of the loop with the value obtained from the iterable.
k. Optional (default None). A function that takes no arguments and returns either a lambdak or a final value. It will be called after the loop is finished to let you continue with something else or stop this chain of the computation.

Returns

The same as given_.

Example

Print some numbers:

for_(range(1, 6), lambda i:
  print_("Number: %s" % i),
lambda: print_("Finished!"))()

Output:

Number: 1
Number: 2
Number: 3
Number: 4
Number: 5
Finished!

Break out of the loop:

for_(range(1, 6), lambda i:
  break_ if i == 3
  else print_("Number: %s" % i),
lambda: print_("Finished!"))()

Output:

Number: 1
Number: 2
Finished!

Continue to the next iteration:

for_(range(1, 6), lambda i:
  continue_ if i == 3
  else print_("Number: %s" % i), lambda:
print_("Finished!"))()

    Number: 1
    Number: 2
    Number: 4
    Number: 5
    Finished!

Notice how nothing is printed when i is 3!

`for_else_`

The same as for_, except with an extra required argument that is run if you don't break out of the loop; see below.

The reason this is a separate function is that the else clause is rarely used in code; and it has the effect of making the function call as a whole a little less readable.

Arguments

seq. Same as for_.
act_k. Same as for_.
else_. A function that is called if you don't break out of the loop. This models Python's native for statement's else clause. For readability, you can write this as a named parameter when you call the for_else_ function; but if you do, you will also have to name the k parameter according to Python's function call syntax.
k. Same as for_.

Returns

Same as for_.

Example

for_else_(range(5),
  const_(None),
else_ = lambda: print_("Didn't break out of loop!"),
k = lambda: print_("Done!"))()

Output:

Didn't break out of loop!
Done!

To verify what happens if you do break out of the loop, replace the second line (const_(None)) with const_(break_).

`while_`

Wraps Python's while statement. Like for_, fully supports breaking and continuing.

Arguments

expr_k. A function that takes no arguments and returns a boolean value. This function is called at each iteration of the loop and the resulting value is used as the test for whether to carry out the loop action.
act_k. A function that takes no arguments and returns either a lambdak to continue the computation, or a final value to finish this iteration of the loop. This is the loop action. As with the for_ lambdak, you can break or continue by returning values of type break_ or continue_. See the examples in the for_ lambdak (above).
k. Optional (default None). The same as for_.

Returns

The same as for_.

Example

Simple loop:

d = { 1: 1 }
inc = lambda x: x + 1

while_(lambda: d[1] < 5, lambda:
  print_(d[1], lambda:
  mod_(1, inc, d)))()

Output:

`while_else_`

This is included for the same reason as for_else_ and works in the same way.

`with_`

Wraps Python's with statement, but is limited to only a single context binding at a time.

Arguments

expr_k. Something that can be called with no arguments to get the context manager.
act_k. Something that will be called with either one argument, or none, depending on whether context manager binds a value or not.

If the context manager doesn't bind a value, i.e. if the equivalent with block in normal Python would have been with x: ... instead of with x as y: ..., then act_k will be called without any arguments.

If the context manager does bind a value, then act_k will be called with that value as the argument.
k. Optional (default None). The same as for do_.

Returns

The same as for given_.

Example

First, the imports.

from contextlib import closing, contextmanager
from lambdak import *
import StringIO

The example of a context manager not binding a value. Here we define a spurious context manager that just prints some text at the start and finish.

@contextmanager
def ctx():
  print "Start!"
  yield
  print "End!"

with_(ctx, lambda: None)()

Output:

Start!
End!

The example of a context manager binding a value. Here we show a more real-world scenario, of opening a resource within the context manager, doing something to it, and then getting its final value before the context manager automatically closes it.

with_(lambda: closing(StringIO.StringIO()), lambda s:
  do_(lambda: s.write("Hello, World!"), lambda:
  print_(s.getvalue())))()

Output:

Hello, World!

`cond_`

Evaluate a list of tests one by one until one of them evaluates to True, and evaluate and return its corresponding value expression.

Short-circuiting: if one of the condition tests evaluates to True, it won't try to evaluate any of the other tests and value expressions after that one.

Should be used in the same way as Python's if: ... elif: ... else: statement would be, or a switch statement in some other language.

Arguments

test_pairs. Must be a sequence (i.e. iterable, like a list) of tuples of (test_expr, then_expr).

test_expr will be called with no arguments to get a boolean value to be tested.

If the value is True, then_expr will be called with no arguments and the result will be passed on to the k function (see below).

If the value is False, the next test_expr in the sequence will be called, and so on.
default_expr. If none of the test_exprs evaluated to True, then this expression will be called with no arguments and the result passed on to the k function. This argument is required as a way to force the developer to think about all possible cases. But as a convenience, you can pass in None and it will do the right thing.
k. Optional (default None). This must be a function which takes no arguments and returns either a lambdak (to indicate a continuing computation), or anything else (to indicate stopping).

If you have effectful code in your then_exprs, you won't necessarily return a meaningful value from them; rather you will be returning the actions (lambdaks) themselves. As a convenience, you can use the handy return_ function as this argument in those cases to just pass along that lambdak and carry out the action. See the second example below.

Returns

The same as given_. You can think of this as a let binding where one binding will ultimately be chosen out of multiple possible bindings.

Example

A 'pure' value calculated and returned:

cond_(
  [ (lambda: False, lambda: 0),
    (lambda: True, lambda: 1) ],
  None, # Default
  lambda val: # The computed value.
print_(val))()

Output:

An 'effectful' action (lambdak) computed and returned:

cond_(
  [ (lambda: False, lambda: print_(0)),
    (lambda: True, lambda: print_(1)) ],
  None,
return_)()

Output:

The return_ function works because it's the exact same thing as lambda x: x, which is what we need as the last argument of cond_ to pass on the computed lambdak (action).

`import_`

Import a module and bind the module object to the parameter name of the k function (see below).

Arguments

mod_name. A string containing the name of the module to import.
k. A function that takes one argument and returns a lambdak (to continue the computation) or a final result (to stop). It will be called with the value of the imported module. Thus, the argument will be bound to the module object.

Returns

The same as given_.

Example

To print the area of a circle with radius 5 units:

import_("math", lambda m:
print_(m.pi * 5 * 5))()

Output:

78.5398...

`try_`

Wrapper for Python's try statement. The last three arguments are meant to be named when try_ is called; this is different from the other lambdak functions' usual call style but because of the different permutations of the arguments, it's more unambiguous to name the arguments and also it looks more naturally like Python's normal try statement. See the example below.

Unlike Python's native try statement, the try_ function only supports one except clause (the except_ parameter, see below).

Also, it 'returns' the value of whichever expression succeeded: the 'try' expression, or the 'except' expression. This means you can do things like 'return' a value from this lambdak and assign it to something, and the value will depend on whether or not an exception was raised. This fulfills the purpose of PEP 463 (Exception-catching expressions). See example below.

Arguments

expr_k. A function that takes no arguments and returns an expression to try.
except_. A function that takes no arguments and does anything. You can think of this as the same as the do_ lambdak's k parameter. It is only run if the expr_k function raises an exceptio.
else_. Optional (default None). A function that takes no argument and does anything. It is only called if the expr_k function does not raise an exception.
finally_. Optional (default None). A function that takes one argument and returns either a lambdak or a final result. Doubles as the finally block of the try statement and as the continuation function, because the finally block is always executed whether or not an exception occurred (for proof, see lambdak_spec.py, class test_try_, method test_python_try_finally_always).

The argument that it takes is either the value from evaluating the expr_k parameter (if it succeeded) or of evaluating the except_ parameter (if the former did not succeed). See example below.

Returns

The same as given_.

Example

try_(lambda: 1 / 0,
  except_ = lambda: print_("Error!"),
  else_ = lambda: print_("No error."),
  finally_ = const_(print_("Cleanup.")))()

Output:

Error!
Cleanup.

Returning a value:

d = { 1: "a" }

x = try_(lambda: d[2],
  except_ = lambda: "b",
  finally_ = return_)()

print x

Output:

Remember, the return_ function takes a value and immediately returns it, so the finally_ parameter above does get an argument which takes a value.

`raise_`

Behaves the same way as Python's raise statement. For details on the arguments below, see the documentation linked.

Arguments

ex_type. Optional.
ex_val. Optional.
tb_val. Optional.

Returns

Theoretically None, but actually never returns because the raise statement jumps control flow to whichever except: ... block is closest, or failing that it crashes the program.

`assign_`

Assign a value to a dict object given its key. This function can be used to manipulate the module's global variables. Example shown below.

Arguments

nm. The key to look up in the dict.
v. The value to assign to the corresponding object.
d. The dict to look in.
k. Optional (default None). Same as for do_.

Returns

The same as given_.

Example

To potentially change a global variable x to some value:

test = assign_("x", 1, globals())

To actually change the value:

test()

`get_`

Get the value of an object in a dict given its key. Note that this function doesn't have a continuation. It's a pure function with no side effects (unless someone has changed the very mechanism of dict lookups themselves), so it doesn't need one.

Arguments

nm. The key to look for.
d. The dict to look in.

Returns

The value corresponding to the given key nm.

Example

x = get_("x", { "x": 1 })
print x

Output:

`mod_`

Modify in-place the value of an object in a dict given its key.

Arguments

nm. The key.
f. The modification function. For example, if the function is lambda x: x + 1, then the value will be incremented by one.
d. The dict to look in.
k. Optional (default None). The same as assign_.

Returns

The same as assign_.

Example

See the example in while_.

`del_`

Delete a key-value pair from a dict. This can also be used to delete a global variable. See example below.

Arguments

nm. The key to look for.
d. The dict to look in.
k. Optional (default None). Same as for do_.

Returns

The same as given_.

Example

To immediately delete a global variable:

# In global scope
x = 1

del_("x", globals())()
print x

Output:

NameError: name 'x' is not defined

Made with λ by yawaramin.

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

yawaramin / lambdak

Programming Languages

lambdak

Contents

News

Overview

Reference

call_

Arguments

Returns

return_

Arguments

Returns

Example

const_

Arguments

Returns

Example

given_

Arguments

Returns

Example

do_

Arguments

Returns

Example

print_

Arguments

Returns

assert_

Arguments

Returns

Example

for_

Arguments

Returns

Example

for_else_

Arguments

Returns

Example

while_

Arguments

Returns

Example

while_else_

with_

Arguments

Returns

Example

cond_

Arguments

Returns

Example

import_

Arguments

Returns

Example

try_

Arguments

Returns

Example

raise_

Arguments

Returns

assign_

Arguments

Returns

Example

get_

Arguments

Returns

Example

mod_

Arguments

Returns

Example

del_

Arguments

`call_`

`return_`

`const_`

`given_`

`do_`

`print_`

`assert_`

`for_`

`for_else_`

`while_`

`while_else_`

`with_`

`cond_`

`import_`

`try_`

`raise_`

`assign_`

`get_`

`mod_`

`del_`