yieldfrom

A backport of the `yield from` semantic from Python 3.x to Python 2.7

View the Project on GitHub Nurdok/yieldfrom

yieldfrom is a backport of the yield from semantic from Python 3.x to Python 2.7. It also supports Python 3.x so it can be used as compatibility library for code that supports both major versions of Python.

If you want to nest generators in Python 3.x, you can use the yield from keywords. This allows you to automatically iterate over sub-generators and transparently pass exceptions and return values from the top level caller to the lowest generator.

def subgen():
    yield 2
    yield 3

def gen():
    yield 1
    yield from subgen()  # Python 3.x only
    yield 4

def main():
    for i in gen():
        print i,

>>> main()
... 1 2 3 4

This functionality is not available in Python 2.x, and we emulate it using the yieldfrom decorator and the helper From class:

from yieldfrom import yieldfrom, From
def subgen():
    yield 2
    yield 3

@yieldfrom
def gen():
    yield 1
    yield From(subgen())
    yield 4

def main():
    for i in gen():
        print i,

>>> main()
... 1 2 3 4

Advanced usage allows returning a value from the subgenerator using StopIteration. Using Return does this conveniently:

from yieldfrom import yieldfrom, From, Return

def subgen():
    yield 2
    yield 3
    Return(100)  # Raises `StopIteration(100)`

@yieldfrom
def gen():
    yield 1
    ret = (yield From(subgen()))
    yield 4
    yield ret

def main():
    for i in gen():
        print i,

>>> main()
... 1 2 3 4 100

Subgenerators can be nested on multiple levels, each one requiring additional decoration by yieldfrom:

def subsubgen():
    yield 2

@yieldfrom
def subgen():
    yield From(subsubgen())
    yield 3

@yieldfrom
def gen():
    yield 1
    yield From(subgen())
    yield 4

def main():
    for i in gen():
        print i,

>>> main()
... 1 2 3 4

Exceptions thrown into the top-level generator can be handled in relevant subgenerators:

def subsubgen():
    try:
        yield 2
    except ValueError:
        yield 200

@yieldfrom
def subgen():
    yield From(subsubgen())
    yield 3

@yieldfrom
def gen():
    yield 1
    yield From(subgen())
    yield 4

def main():
    try:
        g = gen()
        while True:
            i = g.next()
            if i == 2:
                i = g.throw(ValueError())
        print i,
    except StopIteration:
        pass

>>> main()
... 1 200 3 4

Note that if you use yield From() on a simple iterable (list, tuple, etc) then the individual members of the iterator will be yielded on each iteration (perhaps in that case you need the usual yield).

@yieldfrom
def gen():
    yield From([1, 2, 3])
    yield [1, 2, 3]

def main():
    for i in gen():
        print i

>>> main()
... 1
... 2
... 3
... [1, 2, 3]

Passing non-iterable objects to From will result in an empty generator that does nothing.

@yieldfrom
def gen():
    yield From(None)
    yield 1

def main():
    for i in gen():
        print i

>>> main()
... 1

This module is an adaptation of the following Python recipe: http://code.activestate.com/recipes/576727
Modifications include bug fixes in exception handling, naming, documentation, handling of empty generators, etc.