trim matching characters from start or end of string c++

Question

I need to trim the beginning or end of a string, given a matching character.

My function definition looks:

void trim(std::string &s, char c, bool reverse = false);

The bool reverse flags whether to trim the beginning (false) or end (true) of the string.

For example:

s = "--myarg--";
trim(s, '-', false); // should set s to "myarg--"
trim(s, '-', true);  // should set s to "--myarg"

To trim the beginning (ie reverse=false ), this works fine:

bool ok = true;
auto mayberemove = [&ok](std::string::value_type ch){if (ch != '-') ok = false; return ok;};
s.erase(std::remove_if(s.begin(), s.end(), mayberemove), s.end());

The lambda just returns true for each character matching '-', up until the first occurrence of a non-matching character, and continues to return false thereafter. Here I've hard-coded the matching char to be '-', to make the code easier to read.

The trouble I am having is with trimming in reverse. This doesn't work - same as above, but with reverse iterators and ::base():

s.erase(std::remove_if(s.rbegin(), s.rend(), mayberemove).base(), s.end());

Instead, the line above trims all of the ending characters except for the first two.

Any ideas? Thanks

Answer 1

std::string& trim( std::string& s, char c, bool reverse = false )
{
  return reverse
    ? s.erase( s.find_last_not_of( c ) + 1 )
    : s.erase( 0, s.find_first_not_of( c ) );
}

Answer 2

Well, actually I just figured it out. It looks like I needed to reverse the logic of the mayberemove lambda for the reverse case. So, this appears to work fine:

if (reverse) {
    bool notok = false;
    auto mayberemove = [&notok](std::string::value_type ch){if (ch != '-') notok = true; return notok;};
    s.erase(std::remove_if(s.rbegin(), s.rend(), mayberemove).base(), s.end());
}
else {
    bool ok = true;
    auto mayberemove = [&ok](std::string::value_type ch){if (ch != '-') ok = false; return ok;};
    s.erase(std::remove_if(s.begin(), s.end(), mayberemove),s.end());
}

And that works. Now I just need to understand why.