[RESOLVED] Leave only numeric characters in std::string

[Genotypek]

Hello.
I'm trying to remove all non numeric characters from std::string, but I have no idea how to do it...
I tried to using boost:regex etc. from google search, but it doesn't work.
Do you have any solution?
Greetings, genotypek.

[wolle]

You can scan the string once and use the non-numeric characters to build a new string, like

#include <string>
#include <cctype>
#include <iostream>

std::string source = "hfiukqeoi347cno3p80380709jklhdw5k";
std::string target = "";
for (char c : source) {
	if (!std::isdigit(c)) target += c;
}
std::cout << source << " / " << target << std::endl;

[laserlight]

If you really want to "Leave only numeric characters", then your test for characters should be the opposite of wolle's example. You could also #include <algorithm> and do the removal in-place, e.g.,

source.erase(std::remove_if(source.begin(), source.end(),
                            [](char c) { return !std::isdigit(c); }),
             source.end());

[AvDav]

Another way of doing this:

#include <algorithm>
#include <iostream>
#include <iterator>

bool isnonnum(char c) {
	return !(c >= '0' && c <= '9');
}

int main()
{
	std::string src = "hfiukqeoi347cno3p80380709jklhdw5k";
	src.erase(std::remove_if(src.begin(), src.end(), isnonnum), src.end());
	std::copy(src.begin(), src.end(), std::ostream_iterator<char>(std::cout, ""));
        return 0;
}

[wolle]

If you really want to "Leave only numeric characters", then your test for characters should be the opposite of wolle's example. You could also #include <algorithm> and do the removal in-place, e.g.,

source.erase(std::remove_if(source.begin(), source.end(),
                            [](char c) { return !std::isdigit(c); }),
             source.end());

Okay, to keep the digits instead of discarding them it's just to remove the ! (not) sign in the code in my post #2.

First I suspected your solution was inefficient but then I realized it's not. It's even an established idiom,

https://en.wikipedia.org/wiki/Erase%...93remove_idiom

In-place manipulation of strings feels awkward to me but it's probably just my Java background . Of course one can always modify a copy rather than the original string but then on the other hand one is almost back to the iterative approach I suggested so I think I'll stick to it since it's more transparent to me.