python find unique values between sets

[lavinog]

I am working on a python project that involves finding numbers in sets that don't exist in any other set.

The code I have looks like it works, but it seems like there should be an easier way to do this:
(make_unique is the function in question)



def make_unique(setdata):
'''Remove values in each set that exist in another set'''
keys=setdata.keys()

keys_checked=list() # list of keys that have been checked

notunique=set() # set of values that exist in more than one set

for key1 in keys:
keys_checked.append(key1)

for key2 in keys:
if key2 in keys_checked : continue # skip already iterated key

for n in setdata[key1] :

if n in notunique : continue # already know it isn't unique

if n in setdata[key2] :
notunique.add(n) # n not unique
continue

for key in keys:
# remove non-unique numbers
setdata[ key ] = setdata[ key ] - notunique




def makedata():
'''Create dictionary of sets for testing'''
import random
setdata=dict()

for key in 'abcdef':
count=random.randrange(1000,10000)
nlist=set()
for n in range(count):
nlist.add(random.randrange(10000) )

setdata[key]=nlist

return setdata




def printset(setdata):
for key in setdata.keys():
print( key,len( setdata[ key ] ) )


setdata=makedata()
printset(setdata)

make_unique(setdata)
print('-'*30)
printset(setdata)

[Can+~]

I am working on a python project that involves finding numbers in sets that don't exist in any other set.

A intersect B = Things that are on A and in B.

A intersect B^c = Things that are on A but not on B. Set theory says that this is equal to difference.


a = set((1, 2, 3, 4, 5))
b = set((3, 4, 5, 6, 7))

print a - b

Returns set([1, 2]), which are the values on set A that are not on set B.

You can repeat this with the result


a = set((1, 2, 3, 4, 5))
b = set((3, 4, 5, 6, 7))
c = set((2, 3, 5, 6, 7))

print a - b - c

Returns set([1]), the only value on set A that is not in B neither C.

(Link: Set operations (http://docs.python.org/library/stdtypes.html#set))

[Can+~]

I kinda feel bad for the OP, it sucks when you go all out on coding something to realize that there was a one-line solution to it.

[lavinog]

I kinda feel bad for the OP, it sucks when you go all out on coding something to realize that there was a one-line solution to it.
EDIT: at first I didn't think your approach would work, because I was thinking that I needed to return set([1,6]), but I forgot what the original goal was.
I guess it helps to keep the problem straight...thanks

[lavinog]

I figured out my problem, I originally worked out something similar, but it wasn't working because I wasn't copying the sets:
I guess it helps to simplify the problem first.


s=[]
s.append( set((1, 2, 3, 4, 5)) )
s.append( set((3, 4, 5, 7)) )
s.append( set((2, 3, 5, 6, 7)) )

s2=[x.copy() for x in s] #need to copy each set not the list
for i in range(len(s)):
for j in range(len(s)):
if i==j:continue
s2[i] -= s[j]
print(s2)

[CptPicard]

In the general case, using reduce:



def uniques(my_set, others):
return reduce(lambda x,y: x-y, [my_set]+others)

[lavinog]

In the general case, using reduce:



def uniques(my_set, others):
return reduce(lambda x,y: x-y, [my_set]+others)


Neat.
It looks like reduce is not standard in python3.

[unutbu]

Gah! No reduce?



from functools import reduce

Phew. :)

See http://diveintopython3.org/porting-code-to-python-3-with-2to3.html#reduce

[lavinog]

I updated my original example:


def make_unique(setdata):
'''Remove values in each set that exist in another set'''
data={}

for key1 in setdata.keys():

data[ key1 ] = setdata[ key1 ].copy() # make copy of set

for key2 in setdata.keys():

if key2 == key1 : continue # skip already iterated key

# remove original set in key2 from new key1
data[ key1 ] -= setdata[ key2 ]

return data


It looks like to use the reduce method I have to iterate through with each key as my_set and all others in others.


def uniques(my_set, others):
return reduce(lambda x,y: x-y, [my_set]+others)

s=[]
s.append( set((1, 2, 3, 4, 5)) )
s.append( set((3, 4, 5, 7)) )
s.append( set((2, 3, 5, 6, 7)) )

s2=[]
for a in s:
b=s[:]
b.remove(a)
s2.append( uniques (a,b) )
print(s2)

What is the advantage to using reduce?

[lisati]

I kinda feel bad for the OP, it sucks when you go all out on coding something to realize that there was a one-line solution to it.

True: but sometimes it can be helpful to have at least a basic appreciation of the "behind the scenes" stuff. It can help immensely when troubleshooting or designing a routine which is "almost but not quite like" something that comes as standard.

[lavinog]

def uniques(my_set, others):
return reduce(lambda x,y: x-y, [my_set]+others)

def make_unique_2(setdata):
'''Remove values in each set that exist in another set'''
data={}

for key in setdata.keys():
otherdata = [ setdata[ k ] for k in setdata.keys() if k != key ]
data[ key ] = uniques( setdata[ key ] , otherdata )

return data