一、问题的思考
1、对于大整数，我们只需要提供构造，显示，加减法，乘法，除法，模除，指数这些用户接口就可以了。
2、符号位怎么解决？首先考虑的办法是定义一个类，然后里面有sign_, std::vector<char> i_这两个成员变量，我在实现的过程中发现这样写代码，思路不够清晰，代码也不好写。因为当我们编写代码实现加减乘除，比较运算的时候，就不得不考虑符号位，这种设计会使代码的逻辑复杂，不够清晰。
3、正确的办法是先编写一个UnsignedInteger类实现所有的运算，然后再实现有符号的大整数运算。
4、如何实现无符号大整数加减乘除（减法只支持大减小）。加减都很简单，乘法只需提供基本UnsignedInteger * （0–9），UnsignedInteger * （10 ^K)，然后利用加法运算，就可以实现乘法。除法则需要利用UnsignedInteger * （0–9），UnsignedInteger * （10 ^K)，比较运算，减法运算，就可以实现除法。
5、无符号大整数的存储采用std::vector<char>,每一个char表示一个十进制位，并且第一个char表示最低位，最后一个char表示最高位
1、无符号整数的基础运算
代码如下：
#include <iostream>
#include <vector>
#include <assert.h>
#include <cmath>
#include <utility>
class UnsignBigInteger {
private:
std::vector< char > i_;
void compact() {
int count = 0;
for( int i = i_.size() - 1; i >= 0; i-- )
if( i_[i] == 0 )
count++;
else
break;
//resize dose not change storage capacity directly
i_.resize( i_.size() - count );
}
void expand10( int k ) {
int curSize = i_.size();
int expandSize = curSize + k;
i_.resize( expandSize, 0 );
copy_backward( i_.begin(), i_.begin() + curSize,
i_.end() );
fill_n( i_.begin(), k, 0 );
}
void add( const UnsignBigInteger &other ) {
int max = std::max( i_.size(), other.i_.size() );
//to reserve one more place, so it is enough to hold sum
i_.resize( max + 1, 0 );
int c = 0;
for( unsigned i = 0; i < i_.size(); i++ ) {
int value = i_[i] + c ;
if( i < other.i_.size() )
value += other.i_[i];
i_[i] = value % 10;
c = value / 10;
}
}
void sub( const UnsignBigInteger &o ) {
if( o.bigThan(*this) ) {
std::cout << "the other must small then current" << std::endl;
return ;
}
int b = 0;
for( unsigned i = 0; i < i_.size(); i++ ) {
i_[i] -= b;
if( i < o.i_.size() )
i_[i] -= o.i_[i];
if( i_[i] < 0 ) {
i_[i] += 10;
b = 1;
} else {
b = 0;
}
}
compact();
}
void mul( int m ) {
assert( m >= 0 && m < 10 );
i_.resize( i_.size() + 1, 0 );
int c = 0;
for( unsigned i = 0; i < i_.size(); i++ ) {
int value = i_[i] * m + c;
i_[i] = value % 10;
c = value / 10;
}
}
const char & operator[]( int i ) const {
return i_[i];
}
int size() const { return i_.size(); }
public:
bool bigThan( const UnsignBigInteger &o ) const {
if( i_.size() != o.i_.size() )
return i_.size() > o.i_.size();
for( unsigned i = i_.size() - 1; i >= 0; i-- )
if( i_[i] != o.i_[i] )
return i_[i] > o.i_[i];
return false;
}
void print(std::ostream &out) const {
for( int i = i_.size() - 1; i >= 0; i-- )
out << static_cast< int >( i_[i] );
}
//we must stop the implicit conversion
explicit UnsignBigInteger( const std::string & nums ) :
i_(nums.size(), 0){
for( unsigned i = 0; i < nums.size(); i++ )
i_[i_.size() - 1 - i] = nums[i] - '0';
}
explicit UnsignBigInteger( int n ) :
i_(0){
int v = n;
while( v != 0 ) {
i_.push_back( v % 10 );
v /= 10;
}
}
UnsignBigInteger() : i_(0) { }
friend UnsignBigInteger& operator+=( UnsignBigInteger &lhs, const UnsignBigInteger &rhs );
friend UnsignBigInteger& operator-=( UnsignBigInteger &lhs, const UnsignBigInteger &rhs );
friend UnsignBigInteger operator-( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs );
friend UnsignBigInteger operator+( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs );
friend UnsignBigInteger operator*( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs );
friend std::ostream & operator<<( std::ostream &out, const UnsignBigInteger &b );
friend UnsignBigInteger power( const UnsignBigInteger &b, int e );
friend std::pair<UnsignBigInteger, UnsignBigInteger> div( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs);
friend UnsignBigInteger operator/( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs );
friend UnsignBigInteger mod( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs );
};
该算法有一个非常重要的技巧是加法多预留一个字符，计算的时候就不会溢出，所有计算完compact, 这样就不会出问题
2、无符号整数运算的实现
UnsignBigInteger实现了大整数的基本运算，然后将需要的大整数运算定义为友元，下面是这些运算的实现
UnsignBigInteger& operator+=( UnsignBigInteger &lhs, const UnsignBigInteger &rhs ) {
lhs.add( rhs );
lhs.compact();
return lhs;
}
UnsignBigInteger operator+( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs ) {
UnsignBigInteger tmp;
tmp += lhs;
tmp += rhs;
tmp.compact();
return tmp;
}
UnsignBigInteger& operator-=( UnsignBigInteger &lhs, const UnsignBigInteger &rhs ) {
lhs.sub( rhs );
return lhs;
}
UnsignBigInteger operator-( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs ) {
UnsignBigInteger ret( lhs );
ret.sub( rhs );
return ret;
}
UnsignBigInteger operator*( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs ) {
UnsignBigInteger ret;
for( int i = 0; i < rhs.size(); i++ ) {
UnsignBigInteger tmp( lhs );
tmp.mul( rhs[i] );
tmp.expand10( i );
ret += tmp;
}
ret.compact();
return ret;
}
std::pair<UnsignBigInteger, UnsignBigInteger> div( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs) {
UnsignBigInteger u1(lhs), u2(rhs), ret;
while( !u2.bigThan(u1) ) {
unsigned s1 = u1.size(), s2 = u2.size();
unsigned diff = s1 - s2;
UnsignBigInteger temp(u2);
temp.expand10(diff);
if( temp.bigThan(u1) )
diff--;
UnsignBigInteger temp2(u2);
temp2.expand10(diff);
//i use a very naive way to calculate the most significant of the result
int count = 0;
while( !temp2.bigThan(u1) ) {
u1 -= temp2;
count++;
}
//accumulate the result
UnsignBigInteger cur( count );
cur.expand10( diff );
ret += cur;
u1.compact();
}
return std::pair<UnsignBigInteger, UnsignBigInteger>(ret, u1);
}
UnsignBigInteger operator/( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs ) {
return div( lhs, rhs ).first;
}
std::ostream & operator<<( std::ostream &out, const UnsignBigInteger &b ) {
b.print( out );
return out;
}
UnsignBigInteger mod( const UnsignBigInteger &lhs, const UnsignBigInteger &rhs ) {
return div( lhs, rhs ).second;
}
//tail recursion, how to eliminate?
UnsignBigInteger power( const UnsignBigInteger &b, int e ) {
if( e == 0 )
return UnsignBigInteger( 1 );
else if( (e & 1) == 1 )
return b * power( b, e - 1 );
else {
UnsignBigInteger tmp = power( b, e / 2 );
return tmp * tmp;
}
}3、有符号整数运算的实现
因为无符号的和有符号的大整数之间不是继承的关系，我倾向于组合，因此将无符号的大整数作为有符号的一个成员。然后需要做的就是基于无符号的大整数运算实现有符号的。
利用运算的基本规则，有符号的加：如果同号，符号位为两个数的共同符号，如果不同号，则利用无符号的较大减去较小，然后符号位和较大的相同，其他运算和小学的运算规则一样。
class SignBigInteger {
private:
bool sign_;
UnsignBigInteger unbi_;
public:
SignBigInteger( const std::string &s, bool p = true) : sign_(p), unbi_(s) { };
SignBigInteger(): sign_(true) { }
friend SignBigInteger operator-( const SignBigInteger &s );
friend SignBigInteger& operator+=( SignBigInteger &lhs, const SignBigInteger &rhs );
friend SignBigInteger& operator-=( SignBigInteger &lhs, const SignBigInteger &rhs );
friend SignBigInteger operator-( const SignBigInteger &lhs, const SignBigInteger &rhs );
friend SignBigInteger operator+( const SignBigInteger &lhs, const SignBigInteger &rhs );
friend SignBigInteger operator*( const SignBigInteger &lhs, const SignBigInteger &rhs );
friend std::ostream & operator<<( std::ostream &out, const SignBigInteger &b );
friend SignBigInteger power( const SignBigInteger &b, int e );
friend std::pair<SignBigInteger, SignBigInteger> div( const SignBigInteger &lhs, const SignBigInteger &rhs);
friend SignBigInteger operator/( const SignBigInteger &lhs, const SignBigInteger &rhs );
friend SignBigInteger mod( const SignBigInteger &lhs, const SignBigInteger &rhs );
};
SignBigInteger operator-( const SignBigInteger &s ) {
SignBigInteger temp( s );
temp.sign_ = !temp.sign_;
return temp;
}
SignBigInteger& operator+=( SignBigInteger &lhs, const SignBigInteger &rhs ) {
if( lhs.sign_ == rhs.sign_ ) {
lhs.unbi_ += rhs.unbi_;
} else {
if( lhs.unbi_.bigThan( rhs.unbi_ ) ) {
lhs.unbi_ -= rhs.unbi_;
} else {
lhs.unbi_ = rhs.unbi_ - lhs.unbi_;
lhs.sign_ = rhs.sign_;
}
}
return lhs;
}
SignBigInteger& operator-=( SignBigInteger &lhs, const SignBigInteger &rhs ) {
return lhs += (- rhs);
}
SignBigInteger operator-( const SignBigInteger &lhs, const SignBigInteger &rhs ) {
SignBigInteger temp(lhs);
return temp -= rhs;
}
SignBigInteger operator+( const SignBigInteger &lhs, const SignBigInteger &rhs ) {
SignBigInteger temp( lhs );
return temp += rhs;
}
SignBigInteger operator*( const SignBigInteger &lhs, const SignBigInteger &rhs ) {
SignBigInteger sb;
if( lhs.sign_ == rhs.sign_ )
sb.sign_ = true;
else
sb.sign_ = false;
sb.unbi_ = lhs.unbi_ * rhs.unbi_;
return sb;
}
std::ostream & operator<<( std::ostream &out, const SignBigInteger &b ) {
if( b.sign_ == false )
out << "-";
out << b.unbi_;
return out;
}
SignBigInteger power( const SignBigInteger &b, int e ) {
SignBigInteger sb;
sb.sign_ = b.sign_;
if( (b.sign_ == false) && ((e & 1) == 0) )
sb.sign_ = true;
sb.unbi_ = power( b.unbi_, e );
return sb;
} 当然，代码写到这里并没有结束，我们还需要实现无符号大整数和有符号大整数的相互转换，以及和其他一些数字的转换，不然支持的运算太有限了。
二、测试
最后需要做的就是测试，下面是测试代码：
int main() {
UnsignBigInteger b1("12312345678901023456");
UnsignBigInteger b2("4561233123123123123123");
UnsignBigInteger b3(123);
UnsignBigInteger b4(321);
std::cout << b1 + b2 << std::endl;
std::cout << b1 * b2 << std::endl;
std::cout << b3 << std::endl;
std::cout << b1 * b3 << std::endl;
std::cout << power( b3, 2 ) << std::endl;
std::cout << power( b3, 1100 ) << std::endl;
std::cout << b4 - b3 << std::endl;
std::cout << b2 - b1 << std::endl;
std::cout << b4 / b3 << std::endl;
std::cout << b2 / b1 << std::endl;
std::cout << mod( b2, b1 ) << std::endl;
std::cout << mod( b4, b3 ) << std::endl;
SignBigInteger sb1("12345", false), sb2("123");
std::cout << power(sb1, 20) << std::endl;
std::cout << sb1 * sb2 << std::endl;
return 0;
}
下面是执行结果，因为lisp里面支持大整数，我在lisp里面也执行相同的运算，结果是一样的。
扩展
实现整数到大整数，大整数到整数的转换，这样可以丰富运算的数据类型
采用std::vector<int>实现底层，速度至少可以提高8-9倍
