Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Flatten a list/tuple and Call a function with tuples

75 views
Skip to first unread message

beginner

unread,
Jul 25, 2007, 10:50:18 AM7/25/07
to
Hi,

I am wondering how do I 'flatten' a list or a tuple? For example, I'd
like to transform[1, 2, (3,4)] or [1,2,[3,4]] to [1,2,3,4].

Another question is how do I pass a tuple or list of all the
aurgements of a function to the function. For example, I have all the
arguments of a function in a tuple a=(1,2,3). Then I want to pass each
item in the tuple to a function f so that I make a function call
f(1,2,3). In perl it is a given, but in python, I haven't figured out
a way to do it. (Maybe apply? but it is deprecated?)

Thanks,
cg

kyos...@gmail.com

unread,
Jul 25, 2007, 11:00:39 AM7/25/07
to

I'm not sure about the first question, but as for the second, you
could do a few different things. You could just pass the tuple itself
into the function and have the tuple unpacked inside the function.

<code>

f(a)

def f (tuple):
x,y,z = tuple

</code>

You could also pass the elements of the tuple in:

f(a[0], a[1], a[2])

That would do it the way you describe in your post.

Mike

Stargaming

unread,
Jul 25, 2007, 11:19:40 AM7/25/07
to
On Wed, 25 Jul 2007 14:50:18 +0000, beginner wrote:

> Hi,
>
> I am wondering how do I 'flatten' a list or a tuple? For example, I'd
> like to transform[1, 2, (3,4)] or [1,2,[3,4]] to [1,2,3,4].

A recursive function, always yielding the first element of the list,
could do the job. See the ASPN Python Cookbook for a few implementations.
http://aspn.activestate.com/ASPN/search?
query=flatten&section=PYTHONCKBK&type=Subsection

> Another question is how do I pass a tuple or list of all the aurgements
> of a function to the function. For example, I have all the arguments of
> a function in a tuple a=(1,2,3). Then I want to pass each item in the
> tuple to a function f so that I make a function call f(1,2,3). In perl
> it is a given, but in python, I haven't figured out a way to do it.
> (Maybe apply? but it is deprecated?)

>>> def foo(a, b, c): print a, b, c
...
>>> t = (1, 2, 3)
>>> foo(*t)
1 2 3

Have a look at the official tutorial, 4.7.4 http://www.python.org/doc/
current/tut/node6.html#SECTION006740000000000000000

> Thanks,
> cg

HTH,
Stargaming

beginner

unread,
Jul 25, 2007, 11:46:58 AM7/25/07
to
On Jul 25, 10:19 am, Stargaming <stargam...@gmail.com> wrote:
> On Wed, 25 Jul 2007 14:50:18 +0000, beginner wrote:
> > Hi,
>
> > I am wondering how do I 'flatten' a list or a tuple? For example, I'd
> > like to transform[1, 2, (3,4)] or [1,2,[3,4]] to [1,2,3,4].
>
> A recursive function, always yielding the first element of the list,
> could do the job. See the ASPN Python Cookbook for a few implementations.http://aspn.activestate.com/ASPN/search?

> query=flatten&section=PYTHONCKBK&type=Subsection
>
> > Another question is how do I pass a tuple or list of all the aurgements
> > of a function to the function. For example, I have all the arguments of
> > a function in a tuple a=(1,2,3). Then I want to pass each item in the
> > tuple to a function f so that I make a function call f(1,2,3). In perl
> > it is a given, but in python, I haven't figured out a way to do it.
> > (Maybe apply? but it is deprecated?)
> >>> def foo(a, b, c): print a, b, c
> ...
> >>> t = (1, 2, 3)
> >>> foo(*t)
>
> 1 2 3
>
> Have a look at the official tutorial, 4.7.4http://www.python.org/doc/

> current/tut/node6.html#SECTION006740000000000000000
>
> > Thanks,
> > cg
>
> HTH,
> Stargaming

Hi Stargaming,

I know the * operator. However, a 'partial unpack' does not seem to
work.

def g():
return (1,2)

def f(a,b,c):
return a+b+c

f(*g(),10) will return an error.

Do you know how to get that to work?

Thanks,
cg


kyos...@gmail.com

unread,
Jul 25, 2007, 12:00:18 PM7/25/07
to

As I mentioned, you can access the elements individually using square
brackets. The following works:

f(g()[0], g()[1], 10)

But it's not clear. Unfortunately, I'm not seeing much else for tuple
unpacking except the obvious:

a,b=g()
f(a,b,10)


Mike

beginner

unread,
Jul 25, 2007, 12:12:18 PM7/25/07
to
> Mike- Hide quoted text -
>
> - Show quoted text -

Unfortunately f(g()[0], g()[1], 10) is calling g() twice. Sometimes
this is not a good idea.

> a,b=g()
> f(a,b,10)

would work until you want it to be an expression.

Diez B. Roggisch

unread,
Jul 25, 2007, 12:18:57 PM7/25/07
to
beginner wrote:

f(*(g() + (10,))

Not the most beautiful solution, but it works.

Diez

Stargaming

unread,
Jul 25, 2007, 12:23:04 PM7/25/07
to
On Wed, 25 Jul 2007 15:46:58 +0000, beginner wrote:

> On Jul 25, 10:19 am, Stargaming <stargam...@gmail.com> wrote:
>> On Wed, 25 Jul 2007 14:50:18 +0000, beginner wrote:

[snip]

http://docs.python.org/ref/calls.html

> Do you know how to get that to work?

Well, there are several ways to solve this. You could either invoke f(*g
() + (10,)). Might be a bit nasty and unreadable, though. Or you could
just change your function f to accept them in reversed order (f(10, *g)
should work) or convert g() to return a dictionary like {'b': 1, 'c': 2}
and use f(10, **g). But if your function f's hugest use case is being
called with g(), changing f to accept something and g's result (tuple) --
unpacking it inside f -- might be better.

Paul Rubin

unread,
Jul 25, 2007, 12:33:26 PM7/25/07
to
beginner <zyzh...@gmail.com> writes:
> I know the * operator. However, a 'partial unpack' does not seem to work.

A few other posters have mentioned ways around this, but you might ask
yourself what coding situation makes you want to do this stuff in the
first place. I won't say there's never a reason for it, but a lot of
times, a list containing a mixture of scalars and lists/tuples is a
sign that your underlying data representation is contorted. Things
are logically single values or they are logically lists of values, and
that mixed representation is often a sign that the item logically
should be a list, and you're hairing up the program with special
treatment of the case where the list has exactly one element.

I.e. instead of [[1,2,], 3, [5,6,]] maybe you really want
[[1,2,], [3,], [5,6]] without the special treatment and flattening.


James Stroud

unread,
Jul 25, 2007, 12:49:22 PM7/25/07
to

Were this not hypothetical, I would make use of the commutative property
of addition:

f(10, *g())

Proof:

1+2+10 = 10+1+2


Also, this has not been suggested:

py> def g():
... return (1,2)
...
py> def f(a,b,c):
... return a+b+c
...
py> f(c=10, *g())
13


James

Wildemar Wildenburger

unread,
Jul 25, 2007, 1:08:37 PM7/25/07
to pytho...@python.org
kyos...@gmail.com wrote:
> On Jul 25, 9:50 am, beginner <zyzhu2...@gmail.com> wrote:
>
>> Another question is how do I pass a tuple or list of all the
>> aurgements of a function to the function. For example, I have all the
>> arguments of a function in a tuple a=(1,2,3). Then I want to pass each
>> item in the tuple to a function f so that I make a function call
>> f(1,2,3). In perl it is a given, but in python, I haven't figured out
>> a way to do it. (Maybe apply? but it is deprecated?)
>>
> I'm not sure about the first question, but as for the second, you
> could do a few different things. You could just pass the tuple itself
> into the function and have the tuple unpacked inside the function.
>
OR you could use the syntax invented for just that purpose ;).

>>> t = 1, 2, 3
>>> f(*t)

bam! :)

This works with dicts as well (for giving keyword arguments). There you
prepend ** (two asterisk to your dict).
Simple :)
/W

attn.st...@gmail.com

unread,
Jul 25, 2007, 1:16:31 PM7/25/07
to

You can use the "partial" method from functools:

import functools

sum_of_three = functools.partial(f, *g())(10)


--
Hope this helps,
Steven


George Sakkis

unread,
Jul 25, 2007, 1:25:06 PM7/25/07
to

Or if you'd rather write it in one line:

f(*(g() + (10,)))

George

Jeff

unread,
Jul 25, 2007, 1:33:13 PM7/25/07
to
Here's a quick flatten() function:

def flatten(obj):
if type(obj) not in (list, tuple, str):
raise TypeError("String, list, or tuple expected in
flatten().")
if len(obj) == 1:
if type(obj[0]) in (tuple, list):
return flatten(obj[0])
else:
return [obj[0]]
else:
return [obj[0]] + flatten(obj[1:])

x = [1, 2, (3, 4)]
y = (1, 2, [3, 4])
z = "It even works with strings!"
d = {"foo": "bar", "baz": "bat"}

print flatten(x)
print flatten(y)
print flatten(z)
print flatten(d)

Aneesh Goel

unread,
Jul 25, 2007, 2:25:18 PM7/25/07
to
On Jul 25, 10:33 am, Jeff <jeffo...@gmail.com> wrote:
> def flatten(obj):
> if type(obj) not in (list, tuple, str):
> raise TypeError("String, list, or tuple expected in
> flatten().")
> if len(obj) == 1:
> if type(obj[0]) in (tuple, list):
> return flatten(obj[0])
> else:
> return [obj[0]]
> else:
> return [obj[0]] + flatten(obj[1:])

This seems to work fine only if the last object is the only one with
the tuple or list. For example:
>>> y = [(1,2),3,4]
>>> y
[(1, 2), 3, 4]
>>> print flatten(y)
[(1, 2), 3, 4]

if the last line is changed to

return flatten([obj[0]]) + flatten(obj[1:])

then it will unpack tuples/lists anywhere in the main collection being
flattened:
>>> y
[(1, 2), 3, 4]
>>> flatten(y)
[1, 2, 3, 4]
>>> z = [1,(2,3),4]
>>> flatten(z)
[1, 2, 3, 4]
>>> x
[1, 2, (3, 4)]
>>> flatten(x)
[1, 2, 3, 4]
>>> k = [(1,2),(3,4)]
>>> flatten(k)
[1, 2, 3, 4]

Neil Cerutti

unread,
Jul 25, 2007, 2:47:45 PM7/25/07
to
On 2007-07-25, Jeff <jeff...@gmail.com> wrote:
> Here's a quick flatten() function:
>
> def flatten(obj):
> if type(obj) not in (list, tuple, str):
> raise TypeError("String, list, or tuple expected in
> flatten().")
> if len(obj) == 1:
> if type(obj[0]) in (tuple, list):
> return flatten(obj[0])
> else:
> return [obj[0]]
> else:
> return [obj[0]] + flatten(obj[1:])
>
> x = [1, 2, (3, 4)]
> y = (1, 2, [3, 4])
> z = "It even works with strings!"
> d = {"foo": "bar", "baz": "bat"}

e = [[1], 2, 3, , 4]
f = [1, 2, 3, 4, []]

--
Neil Cerutti

Neil Cerutti

unread,
Jul 25, 2007, 2:49:10 PM7/25/07
to

Please excuse my bad typography. The above should've been

e = [[1], 2, 3, 4].

--
Neil Cerutti
It isn't pollution that is hurting the environment; it's the impurities in our
air and water that are doing it. --Dan Quayle

Eduardo "EdCrypt" O. Padoan

unread,
Jul 25, 2007, 3:05:49 PM7/25/07
to Neil Cerutti, pytho...@python.org
def flatten(listOfLists):
return list(chain(*listOfLists))

>From http://www.python.org/doc/2.4/lib/itertools-recipes.html
--
EduardoOPadoan (eopadoan->altavix::com)
Bookmarks: http://del.icio.us/edcrypt

Jeff

unread,
Jul 25, 2007, 3:31:12 PM7/25/07
to
Sorry about that. Hopefully, this should work ;)

def flatten(obj):
if type(obj) not in (list, tuple, str):
raise TypeError("String, list, or tuple expected in
flatten().")
if len(obj) == 1:
if type(obj[0]) in (tuple, list):
return flatten(obj[0])
else:
return [obj[0]]
else:

if type(obj[0]) in (list, tuple):
return flatten(obj[0]) + flatten(obj[1:])


else:
return [obj[0]] + flatten(obj[1:])


x = (1, 2, [3, 4, (5, 6)])
y = ([1, 2, (3, 4)], 5, 6)
z = (1, [2, 3, (4, 5)], 6)

Jeff

unread,
Jul 25, 2007, 3:34:47 PM7/25/07
to
On Jul 25, 3:05 pm, "Eduardo \"EdCrypt\" O. Padoan"
<eopad...@altavix.com> wrote:
> def flatten(listOfLists):
> return list(chain(*listOfLists))
>
> >Fromhttp://www.python.org/doc/2.4/lib/itertools-recipes.html

>
> --
> EduardoOPadoan (eopadoan->altavix::com)
> Bookmarks:http://del.icio.us/edcrypt

That doesn't necessarily work:

import itertools

x = (1, 2, [3, 4, (5, 6)])
y = ([1, 2, (3, 4)], 5, 6)
z = (1, [2, 3, (4, 5)], 6)

def flatten(listOfLists):
return list(itertools.chain(*listOfLists))

print flatten(x)
print flatten(y)
print flatten(z)

==> TypeError: chain argument #1 must support iteration

beginner

unread,
Jul 25, 2007, 8:58:10 PM7/25/07
to
On Jul 25, 11:33 am, Paul Rubin <http://phr...@NOSPAM.invalid> wrote:

Very good question. It is well possible that the problem is my
programming style. I am new to python and am still developing a style
that works for me. A lot of things that work in perl does not seem to
work. Unpacking and flattening are just two examples.

I need nested lists to represent nested records in a script. Since the
structure of the underlying data is nested, I think it is probably
reasonable to represent them as nested lists. For example, if I have
the below structure:

Big Record
Small Record Type A
Many Small Record Type B
Small Record Type C

It is pretty natural to use lists, although after a while it is
difficult to figure out the meaning of the fields in the lists. If
only there were a way to 'attach' names to members of the list.

For the unpacking question, I encountered it when working with list
comprehensions. For example:

[ f(*x,1,2) for x in list] is difficult to do if I don't want to
expand *x to x[0]..x[n]. There are usually 7-10 items in the list and
it is very tedious and error prone.

The second problem is from a nested list comprehension. I just needed
something to flatten the list at the moment.

I am still forming my way to do things in python via trial and error.
It is well possible that this is not the natural way to do things.


beginner

unread,
Jul 25, 2007, 9:00:08 PM7/25/07
to
> Also, this has not been suggested:
>
> py> def g():
> ... return (1,2)
> ...
> py> def f(a,b,c):
> ... return a+b+c
> ...
> py> f(c=10, *g())
> 13
>
> James- Hide quoted text -

>
> - Show quoted text -

Great idea.

beginner

unread,
Jul 25, 2007, 9:01:40 PM7/25/07
to

> Well, there are several ways to solve this. You could either invoke f(*g
> () + (10,)). Might be a bit nasty and unreadable, though. Or you could
> just change your function f to accept them in reversed order (f(10, *g)
> should work) or convert g() to return a dictionary like {'b': 1, 'c': 2}
> and use f(10, **g). But if your function f's hugest use case is being
> called with g(), changing f to accept something and g's result (tuple) --
> unpacking it inside f -- might be better.- Hide quoted text -

>
> - Show quoted text -

These all work. Thanks.

beginner

unread,
Jul 25, 2007, 9:02:32 PM7/25/07
to

> Not the most beautiful solution, but it works.
>
> Diez- Hide quoted text -

>
> - Show quoted text -

Yeah it works! Thanks.

Steven D'Aprano

unread,
Jul 25, 2007, 9:21:22 PM7/25/07
to
On Wed, 25 Jul 2007 15:46:58 +0000, beginner wrote:

> I know the * operator. However, a 'partial unpack' does not seem to
> work.
>
> def g():
> return (1,2)
>
> def f(a,b,c):
> return a+b+c
>
> f(*g(),10) will return an error.


No it doesn't, it _raises_ an exception. This is a function that returns
an error:

def function():
"""Returns an error."""
return Error() # defined elsewhere

But to answer your question:

> Do you know how to get that to work?

It's a little bit messy, but this works:

>>> f(*(g() + (10,)))
13

This is probably easier to read:

>>> t = g() + (10,); f(*t)
13


--
Steven.

Steven D'Aprano

unread,
Jul 25, 2007, 9:27:22 PM7/25/07
to
On Wed, 25 Jul 2007 09:33:26 -0700, Paul Rubin wrote:

> Things
> are logically single values or they are logically lists of values

Except for strings, and string-like objects.

And files.

And records/structs, and tuples.

And lists. And sets.

And bit strings. And tree-like structures.

And, well, just about everything really.

But apart from those minor exceptions, I agree completely with your
recommendation.

(Ha ha only serious.)


--
Steven.

Jeff

unread,
Jul 25, 2007, 9:16:18 PM7/25/07
to
> For example, if I have
> the below structure:
>
> Big Record
> Small Record Type A
> Many Small Record Type B
> Small Record Type C
>
> It is pretty natural to use lists, although after a while it is
> difficult to figure out the meaning of the fields in the lists. If
> only there were a way to 'attach' names to members of the list.

You could use dictionaries:

big_record = {
"small_record_a": { ... },
"many_small_record_b": {
"sub_record_of_b": { ... },
"sub_record_of_b2": { ... },
},
"small_record_c": { ... },
}

Marc 'BlackJack' Rintsch

unread,
Jul 26, 2007, 2:51:33 AM7/26/07
to
On Thu, 26 Jul 2007 00:58:10 +0000, beginner wrote:

> I need nested lists to represent nested records in a script. Since the
> structure of the underlying data is nested, I think it is probably
> reasonable to represent them as nested lists. For example, if I have
> the below structure:
>
> Big Record
> Small Record Type A
> Many Small Record Type B
> Small Record Type C
>
> It is pretty natural to use lists, although after a while it is
> difficult to figure out the meaning of the fields in the lists. If
> only there were a way to 'attach' names to members of the list.

That's where you may start looking into classes. The simplest one is for
just holding attributes seems to be the "bunch":

In [15]: class Bunch(object):
....: def __init__(self, **kwargs):
....: self.__dict__ = kwargs
....:

In [16]: small_a = Bunch(foo=42, bar=23)

In [17]: many_b = [1, 2, 3]

In [18]: small_c = Bunch(name='eric', profession='viking')

In [19]: big_record = Bunch(small_a=small_a, many_b=many_b, small_c=small_c)

In [20]: big_record.small_a.bar
Out[20]: 23

In [21]: big_record.many_b[1]
Out[21]: 2


> For the unpacking question, I encountered it when working with list
> comprehensions. For example:
>
> [ f(*x,1,2) for x in list] is difficult to do if I don't want to
> expand *x to x[0]..x[n]. There are usually 7-10 items in the list and
> it is very tedious and error prone.

If you are the designer of `f` then just receive the whole `x` as *one*
argument.

Ciao,
Marc 'BlackJack' Rintsch

0 new messages