反向字符串find_first_not_of

Question

I have a std::string and I want to find the position of the first character that:

• Is different from all the following characters: ' ' , '\\n' and '\\t' .
• Has lower position from that indicated by me.

So, for example if I have the following string and position:

string str("AAA BBB=CCC DDD");
size_t pos = 7;

I want to have the possibility to use a method like this:

size_t res = find_first_of_not_reverse(str, pos, " \n\t");
// now res = 4, because 4 is the position of the space character + 1

How can I do?

Answer 1

As Bo commented, templatetypedef's answer was 99% of the way there; we just need std::string::find_last_of rather than std::string::find_last_not_of :

#include <cassert>
#include <string>

std::string::size_type find_first_of_not_reverse(
    std::string const& str,
    std::string::size_type const pos,
    std::string const& chars)
{
    assert(pos > 1);
    assert(pos < str.size());

    std::string::size_type const res = str.find_last_of(chars, pos - 1) + 1;
    return res == pos ? find_first_of_not_reverse(str, pos - 1, chars)
         : res ? res
         : std::string::npos;
}

int main()
{
    std::string const str = "AAA BBB=CCC DDD";
    std::string const chars = " \n\t";
    std::string::size_type res = find_first_of_not_reverse(str, 7, chars); // res == 4
    res = find_first_of_not_reverse(str, 2, chars); // res == npos
}

Answer 2

I was curious why basic_string does not define rfind_first_of and friends myself. I think it should. Regardless here is a non-recursive (see ildjarn's answer) implementation that should fulfill the requirements of this question. It compiles but I've not tested it.

std::string delims = " \n\t";
reverse_iterator start = rend()-pos-1, found = 
std::find_first_of(start,rend(),delims.begin(),delims.end());
return found==rend()?npos:pos-(found-start);

To be like rfind pos needs to be set to size() if it's npos or greater than size().

PS: I think this question could benefit from some editing. For one "find_first_of_not_reverse" is pretty misleading. It should be rfind_first_of I think (and then add 1 to the result.)