Merge sort code debugging - c++

Question

I am trying to write a code for merge sort. I am not getting the correct output. I am following this pseudocode link Following is my code. I pass my unsorted array into merge_sort function and call merge function recursively to sort and combine the sub arrays.I know there are more simpler and efficient ways to write code for merge sort but I want to try on my own otherwise I won't learn. Thanks in advance.
int* merge_sort(int* a,int size)
{
//cout<<size;
//cout<<"hi";
if(size == 1)
{
//cout<<"less";
//cout<<a[0];
return a;
}
int* left;
int* right;
int middle = ceil(size/2);
left = new int(middle);
right = new int(middle);
for(int i=0;i<middle;i++)
{
left[i]=a[i];
//cout<<left[i];
}
cout<<"\t";
for(int j=middle;j<size;j++)
{
right[j]=a[j];
//cout<<right[j];
}
cout<<"\t";
left = merge_sort(left,middle);
//if(size==2)
//cout<<left[0];
right = merge_sort(right,middle);
//if(size==2)
//cout<<right[0];
return merge(left,right,middle);
}
int* merge(int* l,int* r,int m)
{
int* result;
result = new int(2*m); //to store the output
int lsize=m; // to keep track of left sub list
int rsize=m; // to keep track of right sub list
int counter = 0; // will use to index result
//cout<<m;
while(lsize>0 || rsize>0)
{
if(lsize>0 && rsize>0)
{
if(l[0]<=r[0])
{
result[counter]=l[0];
counter++; //to store next value in result
lsize--;
l=&l[1]; //decrementing the size of left array
}
else
{
result[counter]=r[0];
counter++;
rsize--;
r=&r[1]; //dec. size of right array
}
}
else if(lsize>0)
{
result[counter]=l[0];
counter++;
lsize--;
l=&l[1];
}
else if(rsize>0)
{
result[counter]=l[0];
counter++;
lsize--;
l=&l[1];
}
}
return result;
}

Answer 1

Your code:
int *left = new int(middle);
allocates a single integer initialized to middle. You need:
int *left = new int [middle];
which allocates an array of middle integers. Rinse and repeat for int *right. Actually, you need to use:
int *right = new int [size - middle];
This gets the correct size for the right array. You then have to modify the recursive call to merge_sort() for the right sub-array:
merge_sort(right, size - middle);
Finally, you have to rewrite merge() to take the size of the left array and the size of the right array independently, because they may be of different sizes. For example, if you sort 10 elements,
you then end up with a call to merge two arrays of 5 (which is fine), but at the next level you need to merge an array of 2 and an array of 3 elements (and you're hosed).
The allocation of result also has the () vs [] allocation problem. And there are some other as yet unresolved problems. But these are important steps in the right direction.
As mentioned in a comment to the question, you have a monumental memory leakage problem, too. What's more, it is not trivial to fix because merge_sort() does an early exit without allocating new memory, so it isn't as simple as 'delete the memory returned by merge_sort()'.
Copy and paste is wonderful until you forget to edit the pasted copy correctly:
else if (lsize > 0)
{
result[counter] = l[0];
counter++;
lsize--;
l = &l[1];
}
else if (rsize > 0)
{
result[counter] = l[0];
counter++;
lsize--;
l = &l[1];
}
Methinks you should be using r and rsize in the second of these blocks.
This still isn't the whole story...
And the residual problem (apart from memory management, which is still 100% leaky and problematic) is:
for(int j=middle;j<size;j++)
{
right[j]=a[j];
//cout<<right[j];
}
You're copying into parts of right that you've not allocated. You need something more like:
for(int j = 0; j < size - middle; j++)
{
right[j] = a[j + middle];
//cout<<right[j];
}
This code works as long as you always sort at least two items at the top level (you crash freeing unallocated space if you sort 1 item — that's part of the memory management problem).
#include <iostream>
using namespace std;
namespace {
int *merge(int *l, int m, int *r, int n);
void dump_array(int *a, int size)
{
int i;
cout << size << ": ";
for (i = 0; i < size; i++)
{
cout << ' ' << a[i];
if (i % 10 == 9)
cout << '\n';
}
if (i % 10 != 0)
cout << '\n';
}
};
int *merge_sort(int *a, int size)
{
cout << "-->> merge_sort:\n";
dump_array(a, size);
if (size <= 1)
{
cout << "<<-- merge_sort: early return\n";
return a;
}
int middle = size/2;
int *left = new int[middle];
int *right = new int[size - middle];
cout << middle << ": ";
for (int i = 0; i < middle; i++)
{
left[i] = a[i];
cout << ' ' << left[i];
}
cout << "\n";
cout << (size - middle) << ": ";
for (int j = 0; j < size - middle; j++)
{
right[j] = a[j + middle];
cout << ' ' << right[j];
}
cout << "\n";
cout << "MSL:\n";
int *nleft = merge_sort(left, middle);
cout << "NL: ";
dump_array(nleft, middle);
cout << "OL: ";
dump_array(left, middle);
cout << "OR: ";
dump_array(right, size - middle);
cout << "MSR:\n";
int *nright = merge_sort(right, size - middle);
cout << "NR: ";
dump_array(nright, size - middle);
cout << "NL: ";
dump_array(nleft, middle);
cout << "OL: ";
dump_array(left, middle);
cout << "OR: ";
dump_array(right, size - middle);
int *result = merge(nleft, middle, nright, size - middle);
cout << "<<-- merge_sort:\n";
dump_array(result, size);
return result;
}
namespace {
int *merge(int *l, int m, int *r, int n)
{
int *result = new int[m + n];
int lsize = m;
int rsize = n;
int counter = 0;
cout << "-->> merge: (" << m << "," << n << ")\n";
dump_array(l, m);
dump_array(r, n);
while (lsize > 0 || rsize > 0)
{
if (lsize > 0 && rsize > 0)
{
if (l[0] <= r[0])
{
result[counter] = l[0];
cout << "C: " << counter << "; L = " << l[0] << "; LS = " << lsize << '\n';
counter++;
lsize--;
l++;
}
else
{
result[counter] = r[0];
cout << "C: " << counter << "; R = " << r[0] << "; RS = " << rsize << '\n';
counter++;
rsize--;
r++;
}
}
else if (lsize > 0)
{
result[counter] = l[0];
cout << "C: " << counter << "; L = " << l[0] << "; LS = " << lsize << '\n';
counter++;
lsize--;
l++;
}
else if (rsize > 0)
{
result[counter] = r[0];
cout << "C: " << counter << "; R = " << r[0] << "; RS = " << rsize << '\n';
counter++;
rsize--;
r++;
}
}
cout << "<<-- merge:\n";
dump_array(result, m+n);
return result;
}
};
int main()
{
for (int i = 2; i <= 10; i++)
{
int array1[] = { 9, 3, 5, 7, 1, 8, 0, 6, 2, 4 };
cout << "\nMerge array of size " << i << "\n\n";
int *result = merge_sort(array1, i);
delete[] result;
}
return 0;
}
This is the debug-laden code. It's the level to which I went to get the result. I could perhaps have used a debugger. Were I on a machine where valgrind works, it might have helped too (but it does not work on Mac OS X 10.8.x, sadly).
There are still many, many ways to improve the code — including the memory management. You'd probably find it easiest to pass the input array to merge() for use as the result array (avoiding the memory allocation in that code). This would reduce the memory management burden.
When you remove the debug code, you'll need to call the dump_array() function in the main() program to get the before and after sorting array images.
Code converted to template functions and leak-free
I've simplified the code a fair bit, especially in the merge() function. Also, more as a matter of curiosity than anything else, converted it to a set of template functions, and then used them with 4 different array types (int, double, std::string, char). The amount of debugging has been dramatically reduced, and the main debugging is conditional on being compiled with -DTRACE_ENABLED now.
The code is now leak-free; valgrind on a Linux box (virtual machine) gives it a clean bill of health when there are no exceptions. It is not guaranteed exception-safe, though. In fact, given the naked uses of new and delete, it is pretty much guaranteed not to be exception-safe. I've left the namespace control in place, but I'm far from convinced it is really correct — indeed, I'd lay odds on it not being good. (I'm also curious if anyone has any views on how to layout code within a namespace { … }; block; it seems odd not indenting everything inside a set of braces, but …)
#include <iostream>
using namespace std;
namespace {
#if !defined(TRACE_ENABLED)
#define TRACE_ENABLED 0
#endif
enum { ENABLE_TRACE = TRACE_ENABLED };
template <typename T>
void merge(T *l, int m, T *r, int n, T *result);
template <typename T>
void dump_array(const char *tag, T *a, int size)
{
int i;
cout << tag << ": (" << size << ") ";
for (i = 0; i < size; i++)
{
cout << " " << a[i];
if (i % 10 == 9)
cout << '\n';
}
if (i % 10 != 0)
cout << '\n';
}
};
template <typename T>
void merge_sort(T *a, int size)
{
if (size <= 1)
return;
if (ENABLE_TRACE)
dump_array("-->> merge_sort", a, size);
int middle = size/2;
T *left = new T[middle];
T *right = new T[size - middle];
for (int i = 0; i < middle; i++)
left[i] = a[i];
for (int j = 0; j < size - middle; j++)
right[j] = a[j + middle];
merge_sort(left, middle);
merge_sort(right, size - middle);
merge(left, middle, right, size - middle, a);
delete [] left;
delete [] right;
if (ENABLE_TRACE)
dump_array("<<-- merge_sort", a, size);
}
namespace {
template <typename T>
void merge(T *l, int m, T *r, int n, T *result)
{
T *l_end = l + m;
T *r_end = r + n;
T *out = result;
if (ENABLE_TRACE)
{
cout << "-->> merge: (" << m << "," << n << ")\n";
dump_array("L", l, m);
dump_array("R", r, n);
}
while (l < l_end && r < r_end)
{
if (*l <= *r)
*out++ = *l++;
else
*out++ = *r++;
}
while (l < l_end)
*out++ = *l++;
while (r < r_end)
*out++ = *r++;
if (ENABLE_TRACE)
dump_array("<<-- merge", result, m+n);
}
};
#include <string>
int main()
{
for (size_t i = 1; i <= 10; i++)
{
int array1[] = { 9, 3, 5, 7, 1, 8, 0, 6, 2, 4 };
if (i <= sizeof(array1)/sizeof(array1[0]))
{
cout << "\nMerge array of type int of size " << i << "\n\n";
dump_array("Original", array1, i);
merge_sort(array1, i);
dump_array("PostSort", array1, i);
}
}
for (size_t i = 1; i <= 10; i++)
{
double array2[] = { 9.9, 3.1, 5.2, 7.3, 1.4, 8.5, 0.6, 6.7, 2.8, 4.9 };
if (i <= sizeof(array2)/sizeof(array2[0]))
{
cout << "\nMerge array of type double of size " << i << "\n\n";
dump_array("Original", array2, i);
merge_sort(array2, i);
dump_array("PostSort", array2, i);
}
}
for (size_t i = 1; i <= 10; i++)
{
std::string array3[] = { "nine", "three", "five", "seven", "one", "eight", "zero", "six", "two", "four" };
if (i <= sizeof(array3)/sizeof(array3[0]))
{
cout << "\nMerge array type std::string of size " << i << "\n\n";
dump_array("Original", array3, i);
merge_sort(array3, i);
dump_array("PostSort", array3, i);
}
}
for (size_t i = 1; i <= 10; i++)
{
char array4[] = "jdfhbiagce";
if (i <= sizeof(array4)/sizeof(array4[0]))
{
cout << "\nMerge array type char of size " << i << "\n\n";
dump_array("Original", array4, i);
merge_sort(array4, i);
dump_array("PostSort", array4, i);
}
}
return 0;
}