Erase duplicate element from a vector

Question

I create a vector inside with several elements in c++ and I want to remove the elements of vector with the same values. Basically, I want to remove the whole index of the vector that is found a duplicate element. My vector is called person. I am trying to do something like:

for(int i=0; i < person.size(); i++){
    if(i>0 && person.at(i) == person.at(0:i-1)) { // matlab operator
        continue;
    }
    writeToFile( perason.at(i) );   
}

How is it possible to create the operator 0:i-1 to check all possible combinations of indexes?

Edit: I am trying GarMan solution but I got issues in for each:

        set<string> myset;
        vector<string> outputvector;

        for (string element:person)
        {
            if (myset.find(element) != myset.end())
            {
                myset.insert(element);
                outputvector.emplace_back(element);
            }
        }

possible duplicate of [Remove elements from a c++ vector where the removal condition is dependent on other elements](http://stackoverflow.com/questions/9412434/remove-elements-from-a-c-vector-where-the-removal-condition-is-dependent-on-ot) — stijn, Jan 27 '14 at 08:20
also http://stackoverflow.com/questions/12200486/how-to-remove-duplicates-from-unsorted-stdvector-while-keeping-the-original-or — stijn, Jan 27 '14 at 08:20

score 3 · Accepted Answer · answered Jan 27 '14 at 09:18

Here is an "in-place" version (no second vector required) that should work with older compilers:

std::set<std::string> seen_so_far;
for (std::vector<std::string>::iterator it = person.begin(); it != person.end();)
{
    bool was_inserted = seen_so_far.insert(*it).second;
    if (was_inserted)
    {
        ++it;
    }
    else
    {
        swap(*it, person.back());
        person.pop_back();
    }
}

Let me know if this works for you. Note that the order of elements is not guaranteed to stay the same.

score 2 · Answer 2 · edited Jan 27 '14 at 09:08

2

If you can sort your vector, you can simply call std::unique.

#include <algorithm>    

std::sort(person.begin(), person.end());
person.erase(std::unique(person.begin(), person.end()), person.end());

If you cannot sort, you can use a hash-table instead by scanning the vector and update the hash-table accordingly. On the same time, you can easily check if one element is already existent or not in O(1) (and O(n) in total). You don't need to check all other elements for each one, which will be time-costly O(n^2).

edited Jan 27 '14 at 09:08

fredoverflow

256,549
94
388
662

answered Jan 27 '14 at 08:20

herohuyongtao

49,413
29
133
174

It seems OP cannot sort the vector, so another approach is needed. – juanchopanza Jan 27 '14 at 08:21

score 2 · Answer 3 · answered Jan 27 '14 at 08:28

2

Something like this will work

unordered_set<same_type_as_vector> myset;
vector<same_type_as_vector> outputvector;
for (auto&& element: myvector)
{
    if (myset.find(element) != myset.end())
    {
        myset.insert(element);
        outputvector.emplace_back(element); 
    }
}
myvector.swap(outputvector);

Code written into reply box, so might need tweaking.

answered Jan 27 '14 at 08:28

GarMan

1,034
7
10

Unexpected token : in for (auto&& element: myvector) statement. – snake plissken Jan 27 '14 at 08:31
Like I said, the code might require tweaking. Are you using a moderm compiler? Did you rename the myvector to your vector? – GarMan Jan 27 '14 at 08:39

Erase duplicate element from a vector

3 Answers3