Adding two bignumber - c++

Question

I implemented a class named bignumber that takes a big number as a string and stores it in an array.
I made a friend operation + for it to add two bignumbers. After running I get an error that the program doesn't respond. What is the problem?
.h file:
class bignumber
{
private:
int *number;
int size;
string num;
public:
bignumber(int);
bignumber(string ,int);
~bignumber();
friend bignumber operator+(bignumber,bignumber);
};
definitions:
bignumber :: bignumber(int numsize)
{
this->size= numsize;
this->number= new int[size];
};
bignumber :: bignumber(string inputnum,int numsize)
{
int *number = new int[numsize];
size = numsize;
num = inputnum;
for(int i=0;i<numsize;i++){
number[i] = int(num[i])-48;
};
};
bignumber :: ~bignumber()
{
delete [] number;
};
bignumber operator+(bignumber num1,bignumber num2)
{
if(num2.size>num1.size){
int e = num2.size - num1.size - 1;
int *tempnum = new int [num2.size];
for(int i=0;i<num1.size;i++){
tempnum[e+i] = num1.number[i];
}
delete [] num1.number;
num1.number = new int[num2.size];
for(int i=0;i<num2.size;i++){
num1.number[i] = tempnum[i];
}
delete [] tempnum;
}
else if(num1.size>num2.size){
int e = num1.size - num2.size - 1;
int *tempnum = new int [num1.size];
for(int i=0;i<num2.size;i++){
tempnum[e+i] = num2.number[i];
}
delete [] num2.number;
num2.number = new int[num1.size];
for(int i=0;i<num1.size;i++){
num2.number[i] = tempnum[i];
}
delete [] tempnum;
}
bignumber temp(max(num1.size,num2.size));
int carry = 0;
for(int i = 0;i < temp.size;i++){
temp.number[i] = num1.number[i] + num2.number[i] + carry;
if (temp.number[i] > 10){
temp.number[i] -= 10;
int carry = 1;
}
};
if(carry = 1){
int *temp2 = new int[temp.size+1];
temp2[0] = carry;
for(int j = 1;j < temp.size+1;j++){
temp2[j] = temp.number[j-1];
};
temp.size += 1;
delete [] temp.number;
temp.number = new int[temp.size];
for(int i=0;i<temp.size;i++){
temp.number[i] = temp2[i];
}
delete [] temp2;
};
};
Also I don't know how to define operator >> to input the number. I wrote this as a friend but it doesn't work:
istream& operator>>(string s,int size)
{
bignumber(s,size);
};

Answer 1

1) You haven't provided the copy constructor, so compiler makes the copy constructor with shallow copy. The same goes for operator=. That's very wrong when you manually allocate memory.
Call to operator+ where you pass parameters by value will result in creating two shallow temporary copies of operands. Then you modify these temporaries deleting the pointers and setting them to new values. But operands know nothing about it, their number pointers will still point to deleted memory, so the operands will become unusable and crash your program on any access to them or on destruction (when their destructors will try to delete the already deleted).
You do not return anything in your operator+, but you have to, and there the lack of copy constructor will result in another allocation error.
2) To pass complex objects by value is rarely a good practice, do it only if you have real reasons. Otherwise, use const myclass& param instead. For your operator+, the signature would be bignumber operator+(const bignumber& num1, const bignumber& num2). Yes, since you cannot modify num1 and num2, you will still need a local copy of the number that needs to grow, but this is one copying instead of two that you've got.
3) implementing operator+ is best done when you've already implemented more simple MyClass& MyClass::operator+=(const MyClass& that); After that, you can use
MyClass operator+(const MyClass& first, const MyClass& second)
{
MyClass retval(first);
retval+=second;
return retval;
}
Your operator+ will still be more complex than needed because you haven't provided some essential functions. You resize your number several times -- why don't you make it a member function resize(int newsize)? You could test and debug it separately, then your operators will be much simpler.
...which all leads to an obvious question: why don't you use vector<int> number? It will solve all the problems above: the compiler-generated constructors and operator= will work fine, no allocation nightmares, resize() and lots of other useful functions are already provided and thoroughly tested. Or, you can use string only: ints are actually wasted in your code, since you store only numbers 0..9 in each of them. Define member functions like int2character and character2int, and you've got rid of most of your problems.
4) operator>> for streams should have a signature
istream& operator>>(istream& is, bignumber& num);
and be something like that:
istream& operator>>(istream& is, bignumber& num)
{
string strTmp;
is>>strTmp;
//checks for istream state etc
//...
//calculate size of number from the extracted string
//...
//then construct a temporary
bignumber tmp(strTmp, calcsize);
num=tmp;//don't forget assignment operator
//or
//just provide a method to reset value of bignumber from string
//it can be a member function
num.assign(strTmp);
//or operator=(const string& str);
num=strTmp;
return is;
}
That is, usually to make your stream operator, you use already defined stream operators for other types.
5) you do not reset carry flag in cycle after you've used it.