How to identify generator/iterator objects?

[Kenneth McDonald]

I'm trying to write a 'flatten' generator which, when give a
generator/iterator that can yield iterators, generators, and other data
types, will 'flatten' everything so that it in turns yields stuff by
simply yielding the instances of other types, and recursively yields the
stuff yielded by the gen/iter objects.

To do this, I need to determine (as fair as I can see), what are
generator and iterator objects. Unfortunately:

>>iter("abc")

<iterator object at 0x61d90>

>>def f(x):

.... for s in x: yield s
....

>>f

<function f at 0x58230>

>>f.__class__

<type 'function'>

So while I can identify iterators, I can't identify generators by class.

Is there a way to do this? Or perhaps another (better) way to achieve
this flattening effect? itertools doesn't seem to have anything that
will do it.

Thanks,
Ken

[Paddy]

Kenneth McDonald wrote:

I'm trying to write a 'flatten' generator which, when give a
generator/iterator that can yield iterators, generators, and other data
types, will 'flatten' everything so that it in turns yields stuff by
simply yielding the instances of other types, and recursively yields the
stuff yielded by the gen/iter objects.

To do this, I need to determine (as fair as I can see), what are
generator and iterator objects. Unfortunately:

>>iter("abc")

<iterator object at 0x61d90>

>>def f(x):

... for s in x: yield s
...

>>f

<function f at 0x58230>

>>f.__class__

<type 'function'>

So while I can identify iterators, I can't identify generators by class.

Is there a way to do this? Or perhaps another (better) way to achieve
this flattening effect? itertools doesn't seem to have anything that
will do it.

Thanks,
Ken

>>def f(x):

.... for s in x: yield s
....

>>f([1])

<generator object at 0x01388E18>

?
- Paddy.

[Leo Kislov]

Kenneth McDonald wrote:

I'm trying to write a 'flatten' generator which, when give a
generator/iterator that can yield iterators, generators, and other data
types, will 'flatten' everything so that it in turns yields stuff by
simply yielding the instances of other types, and recursively yields the
stuff yielded by the gen/iter objects.

To do this, I need to determine (as fair as I can see), what are
generator and iterator objects. Unfortunately:

>>iter("abc")

<iterator object at 0x61d90>

>>def f(x):

... for s in x: yield s
...

>>f

<function f at 0x58230>

>>f.__class__

<type 'function'>

So while I can identify iterators, I can't identify generators by class.

But f is not a generator, it's a function returning generator:

>>def f():

.... print "Hello"
.... yield 1
....

>>iter(f)

Traceback (most recent call last):
File "<input>", line 1, in ?
TypeError: iteration over non-sequence

>>iter(f())

<generator object at 0x016C7238>

>>type(f())

<type 'generator'>

>>>

Notice, there is no side effect of calling f function.

-- Leo

[Martin v. Löwis]

Kenneth McDonald schrieb:

To do this, I need to determine (as fair as I can see), what are
Is there a way to do this? Or perhaps another (better) way to achieve
this flattening effect? itertools doesn't seem to have anything that
will do it.

As others have pointed out, there is a proper test for generator
objects; you are apparently interested in finding out whether a
function will produce a generator when called.

To do that, use the following code

def is_generator_function(f):
return (f.func_code.co_flags & 0x20) != 0

Here, 0x20 is the numeric value of CO_GENERATOR (also available
through compiler.consts.CO_GENERATOR).

Regards,
Martin

[Paddy]

Kenneth McDonald wrote:

I'm trying to write a 'flatten' generator which, when give a
generator/iterator that can yield iterators, generators, and other data
types, will 'flatten' everything so that it in turns yields stuff by
simply yielding the instances of other types, and recursively yields the
stuff yielded by the gen/iter objects.

To do this, I need to determine (as fair as I can see), what are
generator and iterator objects. Unfortunately:

>>iter("abc")

<iterator object at 0x61d90>

>>def f(x):

... for s in x: yield s
...

>>f

<function f at 0x58230>

>>f.__class__

<type 'function'>

So while I can identify iterators, I can't identify generators by class.

Is there a way to do this? Or perhaps another (better) way to achieve
this flattening effect? itertools doesn't seem to have anything that
will do it.

Thanks,
Ken

Unfortunately, nothing is as easy as it may seem:

>>def is_generator(f):

.... return f.func_code.co_flags & CO_GENERATOR != 0
....

>>def f(x):

.... for s in x: yield s
....

>>is_generator(f)

True

>># But look at the following:
def f2(x):

.... def g(y):
.... for s in y: yield s
.... return g(x)
....

>>f2([1,2,3])

<generator object at 0x013DEC88>

>>is_generator(f2)

False

>>>

;-)

- Paddy.