c++: how to remove surrogate unicode values from string?

Question

how do you remove surrogate values from a std::string in c++? looking for regular expression like this:

string pattern = u8"[\uD800-\uDFFF]";
regex regx(pattern);
name = regex_replace(name, regx, "_");

how do you do it in a c++ multiplatform project (e.g. cmake).

Answer 1

First off, you can't store UTF-16 surrogates in a std::string (char-based), you would need std::u16string (char16_t-based), or std::wstring (wchar_t-based) on Windows only. Javascript strings are UTF-16 strings.

For those string types, you can use either:

• std::remove_if() + std::basic_string::erase():

  #include <string>
  #include <algorithm>
  
  std::u16string str; // or std::wstring on Windows
  ...
  str.erase(
      std::remove_if(str.begin(), str.end(),
          [](char16_t ch){ return (ch >= 0xd800) && (ch <= 0xdfff); }
      ),
      str.end()
  );
  

• std::erase_if() (C++20 and later only):

  #include <string>
  
  std::u16string str; // or std::wstring on Windows
  ...
  std::erase_if(str,
      [](char16_t ch){ return (ch >= 0xd800) && (ch <= 0xdfff); }
  );
  

---

UPDATE: You edited your question to change its semantics. Originally, you asked how to remove surrogates, now you are asking how to replace them instead. You can use std::replace_if() for that task, eg:

#include <string>
#include <algorithm>

std::u16string str; // or std::wstring on Windows
...
std::replace_if(str.begin(), str.end(),
    [](char16_t ch){ return (ch >= 0xd800) && (ch <= 0xdfff); },
    u'_'
);

Or, if you really want a regex-based approach, you can use std::regex_replace(), eg:

#include <string>
#include <regex>

std::wstring str; // std::basic_regex does not support char16_t strings!
...
std::wstring newstr = std::regex_replace(
    str,
    std::wregex(L"[\\uD800-\\uDFFF]"),
    L"_"
);