Boost regex: [:alpha:] and accented characters

Question

I am trying to replace every non alpha character in a string with " " using Boost:

std::string sanitize(std::string &str)
{
    boost::regex re;
    re.imbue(std::locale("fr_FR.UTF-8"));
    re.assign("[^[:alpha:]]");
    str = boost::regex_replace(str, re, " ");
    return str;
}


int main ()
{
    std::string test = "(ça) /.2424,@ va très bien ?";
    cout << sanitize(test) << endl;
    return 0;
}

The result is a va tr s bien but I would like to get ça va très bien.

What am I missing?

Answer 1

boost::regex::imbue doesn't do what you are hoping for here - in particular, it won't make boost::regex work with UTF-8. (You could probably make it work this way with ISO 8859-1 or a similar single-byte character encoding, but that doesn't seem to be what you want here).

For UTF-8 support, you will need to use one of the boost::regex classes which will deal with Unicode - see http://www.boost.org/doc/libs/1_55_0/libs/regex/doc/html/boost_regex/unicode.html.

Here is some code which I think does what you want:

#include <string>
#include <boost/regex/icu.hpp>

std::string sanitize(std::string &str)
{
    boost::u32regex re = boost::make_u32regex("[^[:alpha:]]");
    str = boost::u32regex_replace(str, re, " ");
    return str;
}


int main ()
{
    std::string test = "(ça) /.2424,@ va très bien ?";
    std::cout << test << "\n" << sanitize(test) << std::endl;
    return 0;
}

See http://www.boost.org/doc/libs/1_55_0/libs/regex/doc/html/boost_regex/ref/non_std_strings/icu/unicode_algo.html for more examples.