a few extensions for the itertools

[Mathias Panzenboeck]

Steven D'Aprano wrote:
> On Sun, 19 Nov 2006 21:35:24 +0100, Mathias Panzenboeck wrote:
> 
>> I wrote a few functions which IMHO are missing in python(s itertools).
>>
>> You can download them here:
>> http://sourceforge.net/project/showfiles.php?group_id=165721&package_id=212104
>>
>> A short description to all the functions:
>>
>> icmp(iterable1, iterable2) -> integer
>> 	Return negative if iterable1 < iterable2,
>> 	zero if iterable1 == iterable1,
>> 	positive if iterable1 > iterable1.
> 
> 
> What does it mean for an iterable to be less than another iterable? That
> it has fewer items? How do these two iterables compare?
> 
> iter([1, 2, None, "foo", 3+2j])
> 
> def ones():
>     while 1:
>         yield 1
> 
> Which is smaller?
> 
> 

it's like cmp on lists, but on iterables.

[1,2,3] < [1,2,4]
[1,2,3] < [1,2,3,0]
...

> 
>> isum(iterable, start=0) -> value
>> 	Returns the sum of the elements of a iterable
>> 	plus the value of parameter 'start'.  When the
>> 	iterable is empty, returns start.
> 
> 
> You mean just like the built-in sum()?
> 

No, because the builtin sum can't handle iterables other than lists. Or dose it? Hmm, maby it dose 
since any new version and I didn't mention it.

>>>> sum(xrange(12), 1000)
> 1066
> 
> 
>> iproduct(iterable, start=0) -> value
>> 	Returns the product of the elements of a iterable
>> 	times the value of parameter 'start'.  When the
>> 	iterable is empty, returns start.
> 
> If I recall, product() was requested about the same time that sum() was
> introduced, and Guido rejected it as a built-in because it was really only
> useful for calculating geometric means, and it is easy to do if you need
> it:
> 
> def product(it, start=1):
>     # default value of 1 is more sensible than 0
>     # 1 is the multiplicative identity
>     p = start
>     for x in it:
>         p *= x
>     return p
> 
> 
>> forall(predicate, iterable, default=True) -> bool
>> 	Returns True, when for all elements x in iterable
>> 	predicate(x) is True. When the iterable is empty,
>> 	returns default.
>>
>>
>> forany(predicate, iterable, default=False) -> bool
>> 	Returns True, when for any element x in iterable
>> 	predicate(x) is True. When the iterable is empty,
>> 	returns default.
> 
> 
> I vaguely recall plans for all() and any() builtins -- perhaps for Python
> 2.5?
> 

all() and any() don't get predicate functions as arguments.

> 
>> take(n,iterable) -> iterator
>> 	returns a iterator over the first n
>> 	elements of the iterator
> 
> Just like itertools.islice(iterable, n).
> 
>>>> list(itertools.islice(xrange(10), 5))
> [0, 1, 2, 3, 4]
> 

ok, ok, ok. I have overseen that.

> 
>> drop(n,iterable) -> iterable
>> 	drops the first n elemetns of iterable and
>> 	return a iterator over the rest
> 
> Just like itertools.islice(iterable, n, None)
> 
>>>> list(itertools.islice(xrange(20), 15, None))
> [15, 16, 17, 18, 19]
> 
> (Aside: I think islice would be so much cleaner if it took keyword
> arguments.)
> 
> 
> 
>> heads(iterable) -> iterator over all heads
>> tails(iterable) -> iterator over all tails
> 
> What would you use these for?
> 
> 
>> fcain(funct,*functs) -> function(...,***)
>> 	fcain(f1,f2,...,fn)(*args,*kwargs) equals f1(f2(...fn(*args,*kwargs)))
> 
> 
> The usual term for this is function composition.
> 
>