Given a nonempty finite list of integers, output a truthy value if there are exactly two equal entries and all other entries are distinct, and a falsey value otherwise.
truthy:
[1,1]
[1,2,1]
[1,6,3,4,4,7,9]
falsey:
[0]
[1,1,1]
[1,1,1,2]
[1,1,2,2]
[2,1,2,1,2]
[1,2,3,4,5]
lambda m:len({*m})+1==len(m)
{*m} casts the list to a set object, an unordered list of items without duplicates. Doing this will always decrease the length of the list by the number of duplicates in it. By computing how much the length has changed, we can easily tell if the list had a single duplicate and return the result of the test.
-2 bytes thanks to ovs.
{*m} shortcut instead of set, well golfed!
\$\endgroup\$
lambda m:~-len(m[len({*m}):])
\$\endgroup\$
εṠ-u
εṠ-u Implicit input.
u Unique elements.
Ṡ- Delete them from input, counting multiplicities.
ε Is the result a singleton list?
QL‘=L
QL‘=L - Main link, argument L (a list) e.g [1,6,3,4,4,7,9]
Q - Deduplicated elements [1,6,3,4,7,9]
L - Length 6
‘ - Increment 7
L - Length of the input 7 ([1,6,3,4,4,7,9])
= - Are they equal? 1
If the output values can be any consistent values, then QL_L works, which outputs -1 for truthy and any other non-positive number for falsey (thanks @JonathanAllan)
QL_L would output -1 for truthy and some number less than -1 or 0 for falsey (e.g. [1,6,3,4,4,7,9,9,9] would return -3, while [1,6,3,4,7,9] would return 0).
\$\endgroup\$
Commented
Oct 11, 2017 at 21:01
-2.
\$\endgroup\$
Commented
Oct 11, 2017 at 21:06
&=sp4=
One byte saved thanks to @Guiseppe!
Explanation:
&= % Table of pair-wise equality comparisons
%
% [1 0 0 0 0 0 0
% 0 1 0 0 0 0 0
% 0 0 1 0 0 0 0
% 0 0 0 1 1 0 0
% 0 0 0 1 1 0 0
% 0 0 0 0 0 1 0
% 0 0 0 0 0 0 1]
%
s % Sum each Column. Stack:
%
% [1 1 1 2 2 1 1]
%
p % Product of the array. Stack:
%
% [4]
%
4= % Compare the stack to '4'
s is sum and sum sums along the first non-singleton dimension (columns), and the matrix is symmetric, couldn't this just be s instead of Xs?
\$\endgroup\$
Simple implementation of checking whether len(set(list)) == len(list)-1:
LtvuL^=#
Explanation:
\ Implicit: Put all input on stack
Ltv \ Get the stack length - 1, save in auxiliary stack
u \ Remove non-unique elements
L \ Get the new length
^= \ Compare with the previously saved length
# \ Print result
This works as the length will only decrease by 1 if there was only exactly 1 non-distinct integer in the initial list.
This is not using a group or unique approach as many of the other answers, but rather the "cartesian product" of all possible comparisions.
@(x)nnz(triu(x==x',1))==1
x==x' %create a matrix where the entry at (i,j) compares whether x(i) == x(ju)
triu(x==x',1) %only consider the strict upper triangular matrix
nnz(triu(x==x',1)) %count the number of nonzero entries
@(x)nnz(triu(x==x',1))==1 %check whether this number is actually 1
And because no program would be complete without a convolution (thanks @LuisMendo for fixing a mistake):
@(x)nnz(~conv(sort(x),-1:2:1,'same'))==1
=�,=
= check every element for equality with every unique element, creates a matrix with m rows for m unique elements.
0, add an empty row on top.
=&# does the number of rows equal the length of input?
.~ with =
\$\endgroup\$
Thanks to Neil for saving 1 byte.
D`
Mm2`^$
1
Input is linefeed-separated. (The test suite uses comma-separation for convenience.)
D`
Deduplicate the lines in the input, which removes any integer that has appeared before (but leaves the surrounding linefeed(s)).
Mm2`^$
Count the number of empty lines, which is equal to the number of duplicates we removed, but only consider the first two matches. So the output will only be 0 (no duplicates), 1 (one duplicate), 2 (two or more duplicates).
1
Make sure that exactly one duplicate was removed.
A`. to count newlines, because it drops the last one.)
\$\endgroup\$
A`., but the problem is rather that you can't distinguish a single empty line from having no lines at all. Maybe I should consider terminating A and G output with a linefeed if there are any lines. Although that should probably be an option since I can imagine that linefeed being annoying in other scenarios.
\$\endgroup\$
Commented
Oct 12, 2017 at 10:38
^$¶.
\$\endgroup\$
A is identical regardless of whether it retains a single empty line or discards all lines.
\$\endgroup\$
Commented
Oct 12, 2017 at 11:59
-1 byte thanks to @JarkoDubbeldam
cat(sum(duplicated(scan()))==1)
Reads from stdin, writes to stdout.
duplicated iterates through the list, replacing the values of l with TRUE if that value occurs earlier in the list, and FALSE otherwise. If there's a unique pair of soulmates, there should be exactly one TRUE value, so the sum should be 1.
scan() approach.
\$\endgroup\$
Commented
Oct 12, 2017 at 23:15
1=-/#'1?:\
A tacit version of {1=(#x)-#?x}.
1?:\ run a distinct-do-scan for one iteration, returning a two-item list containing the original input and the distinct items of the input#' take the count/length of each item in that list1=-/ is the difference between the counts one?
2=*/#'=:
Checks if the product of the counts of unique elements is 2. This works because the only way to get the product of 2 is (1, 1, 1, ..., 1, 2) in some order.
{¥_O
Outputs 1 as truthy, any other non-negative integer as falsy. In 05AB1E, 1 is the only truthy number (thanks @Emigna for the insight!).
{ Implicit input. Sort
¥ Consecutive differences
_ Boolean negate
O Sum. Implicit display
->(s){s.uniq.length==s.length-1}
->s{s.uniq.size==s.size-1}
\$\endgroup\$
Commented
May 25, 2018 at 2:35
qtlQl{
l{ - Gets the number of unique elements.
tlQ - Gets the length of the input list, decremented.
q - Checks equality.
q1l.-Q{
+18 for using System.Linq.
n=>n.Distinct().Count()==n.Length-1
67 byte alternative without Linq:
n=>new System.Collections.Generic.HashSet<int>(n).Count==n.Length-1
Danish language version
=TÆLV(A:A)=SUM(--(FREKVENS(A:A,A:A)>0))+1
Assumes each integer from the list in separate cell in column A.
If we were allowed for inconsistent falsey values, we could save 3 bytes:
=TÆLV(A:A)+SUM(-(FREKVENS(A:A,A:A)>0))
English language version (44 bytes)
=COUNTA(A:A)=SUM(--(FREQUENCY(A:A,A:A)>0))+1
($args|sort -u).count-eq$args.count-1
The Sort-Object command (alias sort) with the -unique flag pulls out only the unique components of the input. For example, for input @(1,3,3,2), this will result in @(1,2,3).
Thus, we just need to make sure that the .count of this object (i.e., how many elements it has) is -equal to the .count of our input array -1 (i.e., we have exactly one duplicate entry).
Saved 3 bytes thanks to Sinusoid.
Fixed bug thanks to TessellatingHeckler.
1,2,1 - get-unique only works on pre-sorted input. How about ($args|sort -u).count-eq$args.count-1 which is also 37 but does work for all the test cases, if you call it like f 1 2 1 instead of f 1,2,1 ?
\$\endgroup\$
Commented
Oct 15, 2017 at 23:48
@k{@F}++;say@F==1+keys%k
\$\endgroup\$
Commented
Oct 12, 2017 at 12:47
gIÙg-
g # Get number of elements in input
IÙg # Get number of unique elements in input
- # Subtract
In 05AB1E 1 is the only truthy value, so for a truthy result there must be exactly 1 duplicate element removed by the uniquify.
@(x)nnz(~pdist(x))==1
Anonymous function. Input is a column vector. Output is true (displayed as 1) or false (displayed as0).
pdist(x) computes a vector of Euclidean distances between all pairs of rows from x. It considers each pair only once (order of the two rows doesn't matter), and doesn't consider pairs formed by the same row twice.
In our case x is a column vector, so Euclidean distance between two rows is just absolute difference between the two numbers.
~ is logical (Boolean) negation, nnz is number of nonzeros, and ==1 compares to 1. So the result is true if and only if there is only one pair that gives zero distance.
length-(unique|length)==1
Inspired by Riley's answer and much shorter then my original solution.
!a=sum(a.==a')==endof(a)+2
The code generates a 2-dimensional table of booleans, which is then collected using the sum function, counting the number of same-element pairs in the cartesian square of A. Then this is compared to the length of the string plus two, and the quantities are equal only when there is exactly one repeat character.
This code redefines the NOT operator.
!a=sum(a.==a')==endof(a)+2 saves a few bytes. Try it online!
\$\endgroup\$
@(x)prod(sum(x==x'))==4
The x==x' part was inspired by flawr's answer. This is longer than Luis' answer, but it doesn't use any toolboxes.
This is an anonymous function that takes a vector x as input, and compares it to itself transposed. This will give a matrix where all diagonal elements are 1, and any off diagonal elements signals that there are duplicates elements.
The sum along any given column shows how many duplicates there are of that number. We want two of the numbers to have a duplicate, so we two values equal to two, and the rest unequal to two.
If we take the product of this matrix, we'll get 4 if there are only two equal elements (2*2*1*1*1*1*...), and something other than 4 if there are no duplicates, or more than two.
<?=count(array_unique($argv))==count($argv)-1;
Counts the number of entries in $argv and compares it to the number of unique entries. If the former is higher than the latter by 1 then truthy, else falsey.
$argv is the list of command line parameters. So: no, he cannot just use $a.
\$\endgroup\$
{¥>ΘO
{¥>ΘO # example input: [1, 6, 3, 4, 4, 7, 9]
{ # sort -> [1, 3, 4, 4, 6, 7, 9]
¥ # get deltas -> [ 2, 1, 0, 2, 1, 2 ]
> # increment -> [ 3, 2, 1, 3, 2, 3 ]
Θ # truthify (only 1 gives 1) -> [ 0, 0, 1, 0, 0, 0 ]
O # sum -> 1
1 being the only truthy value in 05AB1E, we can stop here. (Thanks @Emigna for pointing that out.)
To get only two distinct values, we can optionally add:
Θ # equals 1? -> 1
Θ, as 1 is the only truthy value in 05AB1E.
\$\endgroup\$
¢ will not work. It would count [19,4,4,9] as false and [19,9] as true since it finds the 0 in 10.
\$\endgroup\$