My code is
void utf8_append(UChar32 cp, std::string& str) {
size_t offset = str.size();
str.resize(offset + U8_LENGTH(cp));
auto ptr = reinterpret_cast<uint8_t*>(&str[0]);
U8_APPEND_UNSAFE(ptr, offset, static_cast<uint32_t>(cp));
}
This works but seems ugly. Maybe I am overlooking a simpler approach?
Relevant documentation: https://unicode-org.github.io/icu/userguide/strings/utf-8.html and https://unicode-org.github.io/icu-docs/apidoc/released/icu4c/utf8_8h.html.
Beauty is in the eye of the beholder. I say it is perfectly valid and correct code! The only thing you might get rid of is the static_cast<uint32_t>, as an UChar32, which is an alias forint32_t, will implicitly cast to uint32_t without warnings.
You could also use append() instead of resize(), avoiding the addition, and remove the temporary ptr, to finally get:
void utf8_append(UChar32 cp, std::string& str) {
auto offset = str.size();
str.append(U8_LENGTH(cp), {});
U8_APPEND_UNSAFE(reinterpret_cast<uint8_t *>(&str[0]), offset, cp);
}
If you can use C++17, str.data() is slightly nicer than &str[0] in my opinion. Or you could write &str.front().
static_cast of course. You could consider using std::basic_string<uint8_t> to get rid of both casts, but it will probably open up a can of worms elsewhere in your codebase.
\$\endgroup\$