icu translit sketch

translit.cpp

#include <unicode/translit.h>

class TranslitConverter
{
    static const UnicodeString TRANSLITERATION_RULES;

public:
    TranslitConverter()
    {
        UErrorCode status = U_ZERO_ERROR;
        UParseError parserError;

        m_transliterator.reset(Transliterator::createFromRules ("TranslitConverter",
                                           TRANSLITERATION_RULES,
                                           UTRANS_FORWARD, parserError, status));
        if(!m_transliterator || U_FAILURE (status))
        {
            throw std::runtime_error("Failed to create translit convereter");
        }
    }

    ~TranslitConverter() {}

    std::string translit(const std::string &utf8string)
    {
        UnicodeString data = icu::UnicodeString::fromUTF8(utf8string);
        m_transliterator->transliterate(data);
        std::string result;
        data.toUTF8String(result);
        return result;
    }

private:
    boost::scoped_ptr<icu::Transliterator> m_transliterator;

};

const UnicodeString TranslitConverter::TRANSLITERATION_RULES(
        "::NFKD;"
        // fix for old icu lin
        "х > kh;"
        "Х > Kh;"
        "\\/ > \\-;"
        //
        "::Russian-Latin/BGN;"
        "::[:Nonspacing Mark:] Remove;"
        "::NFC;"
        "::lower;"
        "::[^a-z0-9[:separator:]-] Remove;"
        "[[:separator:]-]+ > \\-;"
        );

std::string translit(const std::string &s)
{
    static boost::thread_specific_ptr<TranslitConverter> converterPtr;
    if(!converterPtr.get())
    {
        converterPtr.reset(new TranslitConverter);
    }
    return converterPtr->translit(s);
}