C++ ctype facet for UTF-8 in mingw

Question

In a project all internal strings are kept in utf-8 encoding. The project is ported to Linux and Windows. There is a need for a to_lower functionality now.

On POSIX OS I could use std::ctype_byname("ru_RU.UTF-8"). But with g++ (Debian 4.3.4-1), ctype::tolower() don't recognize Russian UTF-8 characters (latin text is lowercased fine).

On Windows, mingw's standard library throws exception "std::runtime_error: locale::facet::_S_create_c_locale name not valid" when I try to construct std::ctype_byname with "ru_RU.UTF-8" argument.

How do I implement/find std::ctype for utf-8 on Windows? The project already depends on libiconv (codecvt facet is based on it), but I don't see an obvious way to implement to_lower with it.

Answer 1

If all you need is to_lower for Cyrillic characters you can write a function by yourself.

АБВГДЕЖ in UTF8  D0 90 D0 91 D0 92 D0 93 D0 94 D0 95 D0 96 0A
абвгдеж in UTF8  D0 B0 D0 B1 D0 B2 D0 B3 D0 B4 D0 B5 D0 B6 0A

But don't forget that UTF8 is multibyte encoding.

Also you can try to convert a string from UTF8 to wchar_t (using libiconv) and use Windows specific function to implement to_lower.

Answer 2

Try to use STLport

  Here is a description of how you can use STLport to read/write utf8 files.
utf8 is a way of encoding wide characters. As so, management of encoding in
the C++ Standard library is handle by the codecvt locale facet which is part
of the ctype category. However utf8 only describe how encoding must be
performed, it cannot be used to classify characters so it is not enough info
to know how to generate the whole ctype category facets of a locale
instance.

In C++ it means that the following code will throw an exception to
signal that creation failed:

#include 
// Will throw a std::runtime_error exception.
std::locale loc(".utf8");

For the same reason building a locale with the ctype facets based on
UTF8 is also wrong:

// Will throw a std::runtime_error exception:
std::locale loc(locale::classic(), ".utf8", std::locale::ctype);

The only solution to get a locale instance that will handle utf8 encoding
is to specifically signal that the codecvt facet should be based on utf8
encoding:

// Will succeed if there is necessary platform support.
locale loc(locale::classic(), new codecvt_byname(".utf8"));

  Once you have obtain a locale instance you can inject it in a file stream to
read/write utf8 files:

std::fstream fstr("file.utf8");
fstr.imbue(loc);

You can also access the facet directly to perform utf8 encoding/decoding operations:

typedef std::codecvt codecvt_t;
const codecvt_t& encoding = use_facet(loc);

Notes:

1. The dot ('.') is mandatory in front of utf8. This is a POSIX convention, locale
names have the following format:
language[_country[.encoding]]

Ex: 'fr_FR'
    'french'
    'ru_RU.koi8r'

2. utf8 encoding is only supported for the moment under Windows. The less common
utf7 encoding is also supported. 

Answer 3

There is some STL (like the one from Apache - STDCXX, for example) that comes with several locales. But on other situations the locale is dependent only on the system.

If you could use name "ru_RU.UTF-8" on one operating the system, it doesn't mean that other systems have the same name for this locale. Debian and windows have probably other names and this is the reason you have a runtime exception.

You should install the locales you want on the system before. Or use an STL that already have this locale.

My cents...