Cross-platform iteration of Unicode string (counting Graphemes using ICU)

Question

I want to iterate each character of a Unicode string, treating each surrogate pair and combining character sequence as a single unit (one grapheme).

Example

The text "नमस्ते" is comprised of the code points: U+0928, U+092E, U+0938, U+094D, U+0924, U+0947, of which, U+0938 and U+0947 are combining marks.

static void Main(string[] args)
{
    const string s = "नमस्ते";

    Console.WriteLine(s.Length); // Ouptuts "6"

    var l = 0;
    var e = System.Globalization.StringInfo.GetTextElementEnumerator(s);
    while(e.MoveNext()) l++;
    Console.WriteLine(l); // Outputs "4"
}

So there we have it in .NET. We also have Win32's CharNextW()

#include <Windows.h>
#include <iostream>
#include <string>

int main()
{
    const wchar_t * s = L"नमस्ते";

    std::cout << std::wstring(s).length() << std::endl; // Gives "6"

    int l = 0;
    while(CharNextW(s) != s)
    {
        s = CharNextW(s);
        ++l;
    }

    std::cout << l << std::endl; // Gives "4"

    return 0;
}

Question

Both ways I know of are specific to Microsoft. Are there portable ways to do it?

• I heard about ICU but I couldn't find something related quickly (UnicodeString(s).length() still gives 6). Would be an acceptable answer to point to the related function/module in ICU.
• C++ doesn't have a notion of Unicode, so a lightweight cross-platform library for dealing with these issues would make an acceptable answer.

Edit: Correct answer using ICU

@McDowell gave the hint to use BreakIterator from ICU, which I think can be regarded as the de-facto cross-platform standard to deal with Unicode. Here's an example code to demonstrate its use (since examples are surprisingly rare):

#include <unicode/schriter.h>
#include <unicode/brkiter.h>

#include <iostream>
#include <cassert>
#include <memory>

int main()
{
    const UnicodeString str(L"नमस्ते");

    {
        // StringCharacterIterator doesn't seem to recognize graphemes
        StringCharacterIterator iter(str);
        int count = 0;
        while(iter.hasNext())
        {
            ++count;
            iter.next();
        }
        std::cout << count << std::endl; // Gives "6"
    }

    {
        // BreakIterator works!!
        UErrorCode err = U_ZERO_ERROR;
        std::unique_ptr<BreakIterator> iter(
            BreakIterator::createCharacterInstance(Locale::getDefault(), err));
        assert(U_SUCCESS(err));
        iter->setText(str);

        int count = 0;
        while(iter->next() != BreakIterator::DONE) ++count;
        std::cout << count << std::endl; // Gives "4"
    }

    return 0;
}

Answer 1

You should be able to use the ICU BreakIterator for this (the character instance assuming it is feature-equivalent to the Java version).

Answer 2

Glib's ustring class gives you utf-8 strings, if using utf-8 is ok for you. It is designed to be similar to std::string. Since utf-8 is native for Linux, your task is quite easy:

int main()
{
    Glib::ustring s = L"नमस्ते";
    cout << s.size();
}

you can also iterate on string's characters as usual with Glib::ustring::iterator

Answer 3

ICU has a very old interface, Boost.Locale is much better:

#include <iostream>
#include <string_view>

#include <boost/locale.hpp>

using namespace std::string_view_literals;

int main()
{
    boost::locale::generator gen;
    auto string = "noël 😸😾"sv;
    boost::locale::boundary::csegment_index map{
        boost::locale::boundary::character, std::begin(string),
        std::end(string), gen("")};
    for (const auto& i : map)
    {
        std::cout << i << '\n';
    }
}

Text is from here