Best way to split a vector into two smaller arrays

2019-03-17 13:05发布

问题:

What I'm trying to do:

I am trying to split a vector into two separate arrays. The current int vector contains an element per line in a text file. The text file is a list of random integers.

How I'm planning to do it:

My current idea is to create two regular int arrays, then iterate over the entire vector and copy n/2 elements to each of the arrays.

What I would like to know:

What is the most elegant way of accomplishing my task? I have a feeling that I can do this without iterating over the vector multiple times.

Code:

#include <vector>
#include <fstream>
#include <iterator>
#include <iostream>
using namespace std;

vector<int> ifstream_lines(ifstream& fs)
{
  vector<int> out;
  int temp;
  while(fs >> temp)
  {
    out.push_back(temp);
  }
  return out;
}

vector<int> MergeSort(vector<int>& lines)
{
  int split = lines.size() / 2;
  int arrayA[split];
  int arrayB[split];
}

int main(void) 
{
  ifstream fs("textfile.txt");
  vector<int> lines;
  lines = ifstream_lines(fs);

  return 0;
}

Thank you :)

回答1:

Use iterators.

std::vector<int> lines;
// fill
std::size_t const half_size = lines.size() / 2;
std::vector<int> split_lo(lines.begin(), lines.begin() + half_size);
std::vector<int> split_hi(lines.begin() + half_size, lines.end());

Since iterator ranges represent half open ranges [begin, end), you don't need to add 1 to the second begin iterator: lines.begin() + half_size isn't copied to the first vector.


Note that things like

int split = lines.size() / 2;
int arrayA[split];
int arrayB[split];

Are not standard C++ (and as such not portable). These are so-called variable-length arrays (VLAs for short) and are a C99 thing. Some compilers have them as an extension while compiling C++ code (GCC, Clang). Always compile with -pedantic to get a warning. These VLAs act funky for non-POD types and aren't generally useful, since you can't even return them.



回答2:

If you only need a reference to the numbers without manipulating them, then you can do:

int *array_1 = &lines[0];
int *array_2 = &lines[lines.size() / 2];

array_1 and array_2 are, actually, pointers to the start and middle of the vector. This works since STL guarantees that vectors store their elements within a continuous memory. Note that referring to lines.begin() can't be used for this.



回答3:

If you can't use code from Xeo answer due to strict compiler rules or you want more generic way, try std::advance:

#include <vector>
#include <iterator>

size_t middle = input.size()/2;
std::vector<int>::const_iterator middleIter(input.cbegin());
std::advance(middleIter, middle);

std::vector<int> leftHalf(input.begin(), middleIter);
std::vector<int> rightHalf(middleIter, input.end());


回答4:

Solution to split vector to variable count parts using iterator.

#include <iostream>
#include <vector>

int main()
{
   // Original vector of data
   std::vector<double> mainVec{1.2, 2.3, 3.4, 4.5, 5.6, 6.7, 7.8, 8.9, 9.0};
   // Result vectors
   std::vector<std::vector<double>> subVecs{};
   // Start iterator
   auto itr = mainVec.begin();
   // Variable to control size of non divided elements
   unsigned fullSize = mainVec.size();
   // To regulate count of parts
   unsigned partsCount = 4U;
   for(unsigned i = 0; i < partsCount; ++i)
   {
       // Variable controls the size of a part
       auto partSize = fullSize / (partsCount - i);
       fullSize -= partSize;
       // 
       subVecs.emplace_back(std::vector<double>{itr, itr+partSize});
       itr += partSize;
   }
   // Print out result
   for (const auto& elemOuter : subVecs)
   {
       std::cout << std::fixed;
       for (const auto& elemInner : elemOuter)
       {
           std::cout << elemInner << " ";
       }
       std::cout << "\n";
   }
}