Bug 6835 – Code pattern: uniq on an array

Status
RESOLVED
Resolution
WONTFIX
Severity
enhancement
Priority
P2
Component
phobos
Product
D
Version
D2
Platform
All
OS
All
Creation time
2011-10-20T18:32:00Z
Last change time
2016-10-15T01:07:07Z
Assigned to
nobody
Creator
bearophile_hugs

Comments

Comment #0 by bearophile_hugs — 2011-10-20T18:32:34Z
"Given an unsorted array, remove the duplicate items" is a common need. In std.algorithm there are tools to perform this task with a very limited amount of code: import std.algorithm; void main() { auto items = [3, 1, 5, 2, 3, 1]; items.length -= copy(uniq(items.sort()), items).length; assert(items == [1, 2, 3, 5]); } (If you see better solutions feel free to add a comment.) It's a single line of code, but in my opinion it's not obvious code, and it repeats the name "items" three times (this makes it a bit bug prone). Removing the duplicated items from an unsorted array is a common operation, so maybe it's worth adding this pattern as a function, named like std.array.sortedUnique or something similar. (Other ways to create an array of unique items is to use hashing, but this is better left to other functions.)
Comment #1 by bearophile_hugs — 2012-11-28T17:33:25Z
I meant something like this: T[] unique(T)(T[] items) /*pure nothrow*/ { items.length -= items.sort().uniq().copy(items).length; return items; }