473,396 Members | 2,013 Online

# why does this unpacking work

I'm a little confused, but I'm sure this is something trivial. I'm
>>t = (('hello', 'goodbye'),
('more', 'less'),
('something', 'nothing'),
>>t
(('hello', 'goodbye'), ('more', 'less'), ('something', 'nothing'),
>>for x in t:
print x
('hello', 'goodbye')
('more', 'less')
('something', 'nothing')
>>for x,y in t:
print x,y
hello goodbye
more less
something nothing
>>>
I understand that t returns a single tuple that contains other tuples.
Then 'for x in t' returns the nested tuples themselves.

But what I don't understand is why you can use 'for x,y in t' when t
really only returns one thing. I see that this works, but I can't quite
conceptualize how. I thought 'for x,y in t' would only work if t
returned a two-tuple, which it doesn't.

What seems to be happening is that 'for x,y in t' is acting like:

for x in t:
for y,z in x:
#then it does it correctly

But if so, why is this? It doesn't seem like very intuitive behavior.

Thanks.
Oct 20 '06 #1
16 1315
It's just sequence unpacking. Did you know that this works?:

pair = ("California","San Francisco")
state, city = pair
print city
# 'San Francisco'
print state
# 'California'

John Salerno wrote:
I'm a little confused, but I'm sure this is something trivial. I'm
>>t = (('hello', 'goodbye'),
('more', 'less'),
('something', 'nothing'),
>>t
(('hello', 'goodbye'), ('more', 'less'), ('something', 'nothing'),
>>for x in t:
print x
('hello', 'goodbye')
('more', 'less')
('something', 'nothing')
>>for x,y in t:
print x,y
hello goodbye
more less
something nothing
>>>

I understand that t returns a single tuple that contains other tuples.
Then 'for x in t' returns the nested tuples themselves.

But what I don't understand is why you can use 'for x,y in t' when t
really only returns one thing. I see that this works, but I can't quite
conceptualize how. I thought 'for x,y in t' would only work if t
returned a two-tuple, which it doesn't.

What seems to be happening is that 'for x,y in t' is acting like:

for x in t:
for y,z in x:
#then it does it correctly

But if so, why is this? It doesn't seem like very intuitive behavior.

Thanks.
Oct 20 '06 #2
On Fri, 2006-10-20 at 15:14, John Salerno wrote:
I'm a little confused, but I'm sure this is something trivial. I'm
>>t = (('hello', 'goodbye'),
('more', 'less'),
('something', 'nothing'),
>>t
(('hello', 'goodbye'), ('more', 'less'), ('something', 'nothing'),
>>for x in t:
print x
('hello', 'goodbye')
('more', 'less')
('something', 'nothing')
>>for x,y in t:
print x,y
hello goodbye
more less
something nothing
>>>

I understand that t returns a single tuple that contains other tuples.
t doesn't "return" anything, t *is* a nested tuple.
Then 'for x in t' returns the nested tuples themselves.
It again doesn't "return" anything. It assigns each element of tuple t
to x, one by one, executing the loop body for each element.
But what I don't understand is why you can use 'for x,y in t' when t
really only returns one thing. I see that this works, but I can't quite
conceptualize how. I thought 'for x,y in t' would only work if t
returned a two-tuple, which it doesn't.
You're thinking of "x,y = t".
What seems to be happening is that 'for x,y in t' is acting like:

for x in t:
for y,z in x:
#then it does it correctly
No, it's actually behaving like

for x in t:
y,z = t
# do something with y and z

You seem to have difficulty distinguishing the concept of looping over a
tuple from the concept of unpacking a tuple. This difficulty is
compounded by the fact that, in your example above, you are looping over
a tuple of tuples and unpacking each inner tuple on the fly.

Hope this helps,

Carsten.
Oct 20 '06 #3
On Fri, 2006-10-20 at 15:37, Carsten Haese wrote:
for x in t:
y,z = t
# do something with y and z
Typo here, of course I mean y,z = x.

-Carsten
Oct 20 '06 #4

John Salerno wrote:
I'm a little confused, but I'm sure this is something trivial. I'm
>>t = (('hello', 'goodbye'),
('more', 'less'),
('something', 'nothing'),
>>t
(('hello', 'goodbye'), ('more', 'less'), ('something', 'nothing'),
>>for x in t:
print x
('hello', 'goodbye')
('more', 'less')
('something', 'nothing')
>>for x,y in t:
print x,y
hello goodbye
more less
something nothing
>>>

I understand that t returns a single tuple that contains other tuples.
Then 'for x in t' returns the nested tuples themselves.

But what I don't understand is why you can use 'for x,y in t' when t
really only returns one thing. I see that this works, but I can't quite
conceptualize how. I thought 'for x,y in t' would only work if t
returned a two-tuple, which it doesn't.

What seems to be happening is that 'for x,y in t' is acting like:

for x in t:
for y,z in x:
#then it does it correctly

But if so, why is this? It doesn't seem like very intuitive behavior.
It makes perfect sense: in fact, you have kind of explained it
yourself!

Think of the for statement as returning the next element of some
sequence; in this case it's a tuple. Then on the left side, the
unpacking occurs. Using "for x in t", means that effectively no
unpackig occurs, so you get the tuple. However, since the in is
returning a tuple, using "for x,y in t", the tuple returned gets
unpacked.

Hope that helps.

Jon.

Oct 20 '06 #5
At Friday 20/10/2006 16:14, John Salerno wrote:
>I'm a little confused, but I'm sure this is something trivial. I'm
>>t = (('hello', 'goodbye'),
('more', 'less'),
('something', 'nothing'),
I understand that t returns a single tuple that contains other tuples.
Then 'for x in t' returns the nested tuples themselves.

But what I don't understand is why you can use 'for x,y in t' when t
really only returns one thing. I see that this works, but I can't quite
conceptualize how. I thought 'for x,y in t' would only work if t
returned a two-tuple, which it doesn't.
You can think of

for x in t:
whatever

as meaning "for each element contained in t, name it x and do whatever"

The other concept involved is unpacking:
>>w = (1,2,3)
x,y,z = w
x
1

When you say "for x,y in t:" there is an implicit unpacking, it means
"for each element contained in t, unpack it into x and y and do whatever"
>What seems to be happening is that 'for x,y in t' is acting like:

for x in t:
for y,z in x:
#then it does it correctly
No, it acts like:

for w in t:
x,y = w
...
--
Gabriel Genellina
Softlab SRL

__________________________________________________
Correo Yahoo!
Espacio para todos tus mensajes, antivirus y antispam ¡gratis!
¡Abrí tu cuenta ya! - http://correo.yahoo.com.ar
Oct 20 '06 #6
jo********@gmail.com wrote:
It's just sequence unpacking. Did you know that this works?:

pair = ("California","San Francisco")
state, city = pair
print city
# 'San Francisco'
print state
# 'California'
Yes, I understand that. What confused me was if it had been written like
this:

pair = (("California","San Francisco"))
Oct 20 '06 #7
Carsten Haese wrote:
You seem to have difficulty distinguishing the concept of looping over a
tuple from the concept of unpacking a tuple.
I think you're right. It's starting to make more sense now. I think when
I saw:

for x,y in t

I was expecting 't' to be a two-tuple for it to work. Maybe writing it as:

for (x,y) in t

sort of helps to show that '(x,y)' is equivalent to one object in 't'.
That makes it look a little more cohesive in my mind, I guess, or helps
me to see it map out against 't' better.
Oct 20 '06 #8
In <%b****************@news.tufts.edu>, John Salerno wrote:
jo********@gmail.com wrote:
>It's just sequence unpacking. Did you know that this works?:

pair = ("California","San Francisco")
state, city = pair
print city
# 'San Francisco'
print state
# 'California'

Yes, I understand that. What confused me was if it had been written like
this:

pair = (("California","San Francisco"))
Uhm, you mean::

pair = (("California","San Francisco"),)

Note the extra comma to make that "a tuple in a tuple".

Ciao,
Marc 'BlackJack' Rintsch
Oct 20 '06 #9
At Friday 20/10/2006 17:29, John Salerno wrote:
>I was expecting 't' to be a two-tuple for it to work. Maybe writing it as:

for (x,y) in t

sort of helps to show that '(x,y)' is equivalent to one object in 't'.
That makes it look a little more cohesive in my mind, I guess, or helps
me to see it map out against 't' better.
Note that it's the *comma* in an expression list what creates a
tuple, *not* the parens. A similar rule applies on the target side of
an assignment:

x,y = (1,2)
(x,y) = [1,2]
[x,y] = 1,2

and all variations are all equivalent.
The left part of a for statement is like an assignment.
With this in mind, it's not surprise that

for (x,y) in t: pass
for x,y in t: pass

are exactly the same.

<http://docs.python.org/ref/exprlists.html>
<http://docs.python.org/ref/assignment.html>
--
Gabriel Genellina
Softlab SRL

__________________________________________________
Correo Yahoo!
Espacio para todos tus mensajes, antivirus y antispam ¡gratis!
¡Abrí tu cuenta ya! - http://correo.yahoo.com.ar
Oct 20 '06 #10
Marc 'BlackJack' Rintsch wrote:
Uhm, you mean::

pair = (("California","San Francisco"),)

Note the extra comma to make that "a tuple in a tuple".

Ciao,
Marc 'BlackJack' Rintsch
You're right!
Oct 20 '06 #11
John Salerno wrote:
I'm a little confused, but I'm sure this is something trivial. I'm
>>t = (('hello', 'goodbye'),
('more', 'less'),
('something', 'nothing'),
>>t
(('hello', 'goodbye'), ('more', 'less'), ('something', 'nothing'),
>>for x in t:
print x
('hello', 'goodbye')
('more', 'less')
('something', 'nothing')
>>for x,y in t:
print x,y
hello goodbye
more less
something nothing
>>>

I understand that t returns a single tuple that contains other tuples.
Then 'for x in t' returns the nested tuples themselves.

But what I don't understand is why you can use 'for x,y in t' when t
really only returns one thing. I see that this works, but I can't quite
conceptualize how. I thought 'for x,y in t' would only work if t
returned a two-tuple, which it doesn't.
Hi John,

Thats the point were you go astray.
iterating over t *does* produce a 2-tuple that can be unpacked
immediately.

maybe this will help, (notice that element is always one of the inner
tuples):
>>tpl = ((00,01), (10,11), (20,21))
for element in tpl:
.... print "tpl provides this when iterated over:", element
....
tpl provides this when iterated over: (0, 1)
tpl provides this when iterated over: (10, 11)
tpl provides this when iterated over: (20, 21)
>>for element in tpl:
.... print "tpl provides this when iterated over:", element
.... sub0, sub1 = element
.... print "each element unpacks to:", sub0,"and:", sub1
....
tpl provides this when iterated over: (0, 1)
each element unpacks to: 0 and: 1
tpl provides this when iterated over: (10, 11)
each element unpacks to: 10 and: 11
tpl provides this when iterated over: (20, 21)
each element unpacks to: 20 and: 21
>>for sub0, sub1 in tpl:
.... print "each element of tuple unpacked immediately to:",
sub0,"and:", sub1
....
each element of tuple unpacked immediately to: 0 and: 1
each element of tuple unpacked immediately to: 10 and: 11
each element of tuple unpacked immediately to: 20 and: 21
>>>

Oct 21 '06 #12
Hi,
In answering Johns question I had to approach it from a standpoint of
originally not seeing that their could be a difficulty; railing myself
in; then re-reading Johns question and trying even harder to put myself
in his shoes.
What I am interested in is if John and others might just take time out
to critique the replies. I'm interested in what the group think makes a
good comp.lang.python reply: too short, too long; too cryptic, too
simplistic, too polite (is their such a thing), too nasty; too
self-effacing, too self-promoting; too long a sig ;-) , too anonymous
....

John Salerno wrote:
I'm a little confused, but I'm sure this is something trivial. I'm
>>t = (('hello', 'goodbye'),
('more', 'less'),
('something', 'nothing'),
>>t
(('hello', 'goodbye'), ('more', 'less'), ('something', 'nothing'),
>>for x in t:
print x
('hello', 'goodbye')
('more', 'less')
('something', 'nothing')
>>for x,y in t:
print x,y
hello goodbye
more less
something nothing
>>>
I understand that t returns a single tuple that contains other tuples.
Then 'for x in t' returns the nested tuples themselves.

But what I don't understand is why you can use 'for x,y in t' when t
really only returns one thing. I see that this works, but I can't quite
conceptualize how. I thought 'for x,y in t' would only work if t
returned a two-tuple, which it doesn't.

Hi John,

Thats the point were you go astray.
iterating over t *does* produce a 2-tuple that can be unpacked
immediately.

maybe this will help, (notice that element is always one of the inner
tuples):
>tpl = ((00,01), (10,11), (20,21))
for element in tpl:
... print "tpl provides this when iterated over:", element
...
tpl provides this when iterated over: (0, 1)
tpl provides this when iterated over: (10, 11)
tpl provides this when iterated over: (20, 21)
>for element in tpl:
... print "tpl provides this when iterated over:", element
... sub0, sub1 = element
... print "each element unpacks to:", sub0,"and:", sub1
...
tpl provides this when iterated over: (0, 1)
each element unpacks to: 0 and: 1
tpl provides this when iterated over: (10, 11)
each element unpacks to: 10 and: 11
tpl provides this when iterated over: (20, 21)
each element unpacks to: 20 and: 21
>for sub0, sub1 in tpl:
... print "each element of tuple unpacked immediately to:",
sub0,"and:", sub1
...
each element of tuple unpacked immediately to: 0 and: 1
each element of tuple unpacked immediately to: 10 and: 11
each element of tuple unpacked immediately to: 20 and: 21
>>

Oct 21 '06 #13
John Salerno wrote:
I understand that t returns a single tuple that contains other tuples.
t *is* a single tuple that contains other tuples.
Then 'for x in t' returns the nested tuples themselves.

But what I don't understand is why you can use 'for x,y in t' when t
really only returns one thing. I see that this works, but I can't quite
conceptualize how.
hint:
>>t = (('hello', 'goodbye'),
.... ('more', 'less'),
.... ('something', 'nothing'),
>>t[0]
('hello', 'goodbye')
>>t[1]
('more', 'less')
>>t[2]
('something', 'nothing')
>>t[3]

</F>

Oct 21 '06 #14
John Salerno a écrit :
jo********@gmail.com wrote:
>It's just sequence unpacking. Did you know that this works?:

pair = ("California","San Francisco")
state, city = pair
print city
# 'San Francisco'
print state
# 'California'

Yes, I understand that. What confused me was if it had been written like
this:

pair = (("California","San Francisco"))

Python 2.4.1 (#1, Jul 23 2005, 00:37:37)
[GCC 3.3.4 20040623 (Gentoo Linux 3.3.4-r1, ssp-3.3.2-2, pie-8.7.6)] on
linux2
>>(("California","San Francisco")) == ("California","San Francisco")
True
pair = "California","San Francisco"
>>pair
('California', 'San Francisco')
>>t = 1,
t
(1,)
>>t = 1, 2
t
(1, 2)
>>>
HTH

Oct 21 '06 #15
What I am interested in is if John and others might just take time out
to critique the replies. I'm interested in what the group think makes a
good comp.lang.python reply: too short, too long; too cryptic, too
simplistic, too polite (is their such a thing), too nasty; too
self-effacing, too self-promoting; too long a sig ;-) , too anonymous
this point, because I'm starting to understand my problem now. But what
I liked about your help was that you took it in steps and show each
process. What's always a little confusing is when people immediately
resort to examples for help instead of explaining it in words first,
because often times this just duplicates the confusion for the OP. But
examples definitely help!
Oct 23 '06 #16
>>>>Carsten Haese <ca*****@uniqsys.com(CH) wrote:
>CHOn Fri, 2006-10-20 at 15:14, John Salerno wrote:
>>What seems to be happening is that 'for x,y in t' is acting like:

for x in t:
for y,z in x:
#then it does it correctly
>CHNo, it's actually behaving like
>CHfor x in t:
CH y,z = t
y,z = x
>CH # do something with y and z
--
Piet van Oostrum <pi**@cs.uu.nl>
URL: http://www.cs.uu.nl/~piet [PGP 8DAE142BE17999C4]
Private email: pi**@vanoostrum.org
Nov 7 '06 #17

This thread has been closed and replies have been disabled. Please start a new discussion.