如何比较 C++ 中的长字符串？

Question

I know how to compare two strings with "==" or "compare", but if the string is very long, should we use a hash function and then compare with hash code?

static int n = 100000;

bool TestCompare(const string& a, const string& b) {
    return a == b;
}

bool TestCompareHash(const string& a, const string& b) {
    std::hash<std::string> hash_fn;

    std::size_t str_hash_a = hash_fn(a);
    std::size_t str_hash_b = hash_fn(b);
    return str_hash_a == str_hash_b;
}


int main()
{
    string a(100, 'a');
    string b(100, 'c');
    std::chrono::time_point<std::chrono::system_clock> now = std::chrono::system_clock::now();
    for (int i = 0; i < n; i++) {
        TestCompare(a, b);
    }
    std::chrono::duration<float> difference = std::chrono::system_clock::now() - now;
    cout << "difference.count() 1: " << difference.count() << endl;
    
    now = std::chrono::system_clock::now();
    for (int i = 0; i < n; i++) {    
        TestCompareHash(a, b);
    }
    difference = std::chrono::system_clock::now() - now;
    cout << "difference.count() 2: " << difference.count() << endl;    
    
    return 0;
}

I tested such a code and found that the hash_test will slow down when the string becomes longer, why?

when string length is 100

difference.count() 1: 0.00263665 
difference.count() 2: 0.00713478   //hash

when string length is 10000

difference.count() 1: 0.00322366  
difference.count() 2: 1.99765    //hash

I made some improvements to the test from the comments like "make both strings exact matches except for the last character".

---

It seems that doing hashing does not save the amount of calculations. It may be possible to do these operations in the database to avoid a single point of problem, but it may not make much sense in comparing strings?

Answer 1

In your case the main issue is that you need to compute those hashes first and that costs more than comparison of strings (which "compares chars until they don't match", O(n) complexity at worst). You didn't provide hash_fn() but it generally must "go over all chars" (O(n) complexity).

Hashes would help if you compute and store them once and then expect to compare the strings many times.

Note the hashes can be used only to compare for equality (eg no > or <).