是否首先使用 std::map::find 进行搜索，然后使用 [] 检索值，通常进行优化？

Question

I find often this pattern in some codebase I work with:

   std::map<std::string, std::string> mymap;

   ...

   if (mymap.find(key) != mymap.end())
   {
       return mymap[key];
   }

I know this can be replaced by

   std::map<std::string, std::string> mymap;

   ...

   auto it = mymap.find(key) 
   if (it != mymap.end())
   {
       return it->second;
   }

But I was wondering if, in practice, the first version would get optimised to not search twice anyway?

Answer 1

No it will not be optimized. The pattern is too high level.

You could preferentially just do this instead:

 std::map<std::string, std::string> mymap;

   ...

   auto it = mymap.find(key);
   if ( it != mymap.end())
   {
       return it->second;
   }

You can check on the benchmark below:

int findOptim( int key )
{
   auto it = mymap.find(key); 
   if (it != mymap.end())
   {
       return it->second;
   }
   return -1;
}

int findTrivial( int key ) {
    if ( mymap.find(key) != mymap.end() ) {
        return mymap[key];
    }
    return -1;
}

Benchmark: https://quick-bench.com/q/7mFWe8jlXD7J9GTDMNSiZdcGCgY