Intersection of lists/sets -- with a catch

[James Stroud]

Hello All,

I find myself in this situation from time to time: I want to compare two lists
of arbitrary objects and (1) find those unique to the first list, (2) find
those unique to the second list, (3) find those that overlap. But here is the
catch: comparison is not straight-forward. For example, I will want to
compare 2 objects based on a set of common attributes. These two objects need
not be members of the same class, etc. A function might help to illustrate:

def test_elements(e lement1, element2):
"""
Returns bool.
"""
# any evaluation can follow
return (element1.att_a == element2.att_a) and \
(element1.att_b == element2.att_b)

So my very naieve solution usually bears resemblance to the following code
(made inelegant by the irritating necessity to keep track of a counter):

for some_element in some_list:
i = 0
while i < len(another_lis t):
another_element = another_list[i]
overlap = test_elements(s ome_element, another_element )
if overlap:
store_overlap(s ome_element, another_element )
# speed-up, same as i+=1
another_list.po p(i)
break
else:
i += 1
if not overlap:
store_unique_so me(some_element )

After this loop, "identical" elements (according to the evaluation criteria)
from the two lists are stored somewhere, elements unique to 'some_list' are
stored somewhere else, and 'another_list' has only elements originally unique
to itself according to the evaluation criteria remaining. Of course this code
assumes that both original lists contained only unique elements within
themselves and were previously sorted based on the evaluation criteria (to
speed-up the loop).

So, one improvement I can think of is to use an iterator for the inner loop:

for some_element in some_list:
for another_element in another_list:
overlap = test_elements(s ome_element, another_element )
if overlap:
store_overlap(s ome_element,ano ther_element)
# big problem here -- iterator gets lost
another_list.re move(another_el ement)
break
if not overlap:
unique_some.app end(some_elemen t)

But the problem is obvious, the iterator becomes out of sync with
'another_list'. An example can illustrate:

py> alist = [1,2,3,4]
py> for el in alist:
.... print el
.... if el == 3:
.... alist.remove(el )
....
1
2
3
py> alist
[1, 2, 4]

So, a question would be how one might "correct" the iterator in this
situation. Inspecting dir(iter) provides no clues as to the identity of an
iterator's internal counter.

Its probably obvious to everyone that this type of task seems perfect for
sets. However, it does not seem that sets can be used in the following way,
using a hypothetical "comparator " function. The "comparator " would be
analagous to a function passed to the list.sort() method. Such a device would
crush the previous code to the following very straight-forward statements:

some_set = Set(some_list, comparator=test _elements)
another_set = Set(another_lis t, comparator=test _elements)
overlaps = some_set.inters ection(another_ set)
unique_some = some_set.differ ence(another_se t)
unique_another = another_set.dif ference(some_se t)

I am under the personal opinion that such a modification to the set type would
make it vastly more flexible, if it does not already have this ability.

Any thoughts on how I might accomplish either technique or any thoughts on how
to make my code more straightforward would be greatly appreciated.

James

--
James Stroud
UCLA-DOE Institute for Genomics and Proteomics
Box 951570
Los Angeles, CA 90095

http://www.jamesstroud.com/

[Carl Banks]

James Stroud wrote:

Hello All,

I find myself in this situation from time to time: I want to compare two lists
of arbitrary objects and (1) find those unique to the first list, (2) find
those unique to the second list, (3) find those that overlap. But here is the
catch: comparison is not straight-forward. For example, I will want to
compare 2 objects based on a set of common attributes. These two objects need
not be members of the same class, etc. A function might help to illustrate:

def test_elements(e lement1, element2):
"""
Returns bool.
"""
# any evaluation can follow
return (element1.att_a == element2.att_a) and \
(element1.att_b == element2.att_b)

[snip]
Its probably obvious to everyone that this type of task seems perfect for
sets. However, it does not seem that sets can be used in the following way,
using a hypothetical "comparator " function. The "comparator " would be
analagous to a function passed to the list.sort() method. Such a device would
crush the previous code to the following very straight-forward statements:

some_set = Set(some_list, comparator=test _elements)
another_set = Set(another_lis t, comparator=test _elements)
overlaps = some_set.inters ection(another_ set)
unique_some = some_set.differ ence(another_se t)
unique_another = another_set.dif ference(some_se t)

I am under the personal opinion that such a modification to the set type would
make it vastly more flexible, if it does not already have this ability.

Any thoughts on how I might accomplish either technique or any thoughts on how
to make my code more straightforward would be greatly appreciated.

Howabout something like this (untested):

class CmpProxy(object ):
def __init__(self,o bj):
self.obj = obj
def __eq__(self,oth er):
return (self.obj.att_a == other.obj.att_b
and self.obj.att_b == other.obj.att_b )
def __hash__(self):
return hash((self.obj. att_a,self.obj. att_b))

set_a = set(CmpProxy(x) for x in list_a)
set_b = set(CmpProxy(y) for y in list_b)
overlaps = [ z.obj for z in set_a.intersect ion(set.b) ]
Carl Banks

[George Sakkis]

"Carl Banks" <in**********@a erojockey.com> wrote:

Howabout something like this (untested):

class CmpProxy(object ):
def __init__(self,o bj):
self.obj = obj
def __eq__(self,oth er):
return (self.obj.att_a == other.obj.att_b
and self.obj.att_b == other.obj.att_b )
def __hash__(self):
return hash((self.obj. att_a,self.obj. att_b))

set_a = set(CmpProxy(x) for x in list_a)
set_b = set(CmpProxy(y) for y in list_b)
overlaps = [ z.obj for z in set_a.intersect ion(set.b) ]

Or more generally:

class Comparable(obje ct):
def __init__(self, obj, key):
self.obj,self._ key = obj,key
def __hash__(self):
return hash(self._key( self.obj))
def __eq__(self, other):
return self._key(self. obj) == self._key(other .obj)
if __name__ == '__main__':
some_list = ["hello", "world"]
another_list = ["HeLlO", "word"]
key = str.lower
some_set = set(Comparable( x,key) for x in some_list)
another_set = set(Comparable( x,key) for x in another_list)
print "overlaps:" , [x.obj for x in some_set.inters ection(another_ set)]
print "unique_som e:", [x.obj for x in some_set.differ ence(another_se t)]
print "unique_another :", [x.obj for x in another_set.dif ference(some_se t)]
George