Make C++ std::string end at specified character?

Question

Let's say I am traversing a string of length n. I want it to end at a specific character that fulfils some conditions. I know that C style strings can be terminated at the i'th position by simply assigning the character '\0' at position i in the character array.

Is there any way to achieve the same result in an std::string (C++ style string)? I can think of substr, erase, etc. but all of them are linear in their complexity, which I cannot afford to use.

TL;DR, is there any "end" character for an std::string? Can I make the end iterator point to the current character somehow?

Answer 1

You can use resize:

std::string s = /* ... */;

if (auto n = s.find(c); n != s.npos) {
  s.resize(n);
}

Answer 2

The logical answer here is basic_string::resize. What the standard says about this function is:

Effects: Alters the length of the string designated by *this as follows:

• If n <= size(), the function replaces the string designated by *this with a string of length n whose elements are a copy of the initial elements of the original string designated by *this.
• If n > size(), the function replaces the string designated by *this with a string of length n whose first size() elements are a copy of the original string designated by *this, and whose remaining elements are all initialized to c.

Now, that looks very much like linear time. However, the standard does not specifically state that things will happen this way. They only state that it will be "as if" things happen this way. Therefore, an implementation is completely free to implement the shrinking version of resize by shifting one pointer and writing a NUL character. Nothing in the standard would forbid such an implementation.

So the real question is... are standard library implementations written by complete morons? It's certainly possible that they are. But it's probably wise not to assume so.

Personally, I'd just use resize on the assumption that the library implementers know what they're doing. After all, if they can't write an optimization as simple as that, then who knows what other things they're doing wrong? If you can't trust your standard library implementation not to do stupid things, then you shouldn't be using it in performance-critical code.