Character Set Conversions

Convenience Interface

Boost.Locale provides to_utf, from_utf and utf_to_utf functions in the boost::locale::conv namespace. They are simple and convenient functions to convert a string to and from UTF-8/16/32 strings and strings using other encodings.

For example:

std::string utf8_string = to_utf<char>(latin1_string,"Latin1");
std::wstring wide_string = to_utf<wchar_t>(latin1_string,"Latin1");
std::string latin1_string = from_utf(wide_string,"Latin1");
std::string utf8_string2 = utf_to_utf<char>(wide_string);

This function may use an explicit encoding name like "Latin1" or "ISO-8859-8", or use std::locale as a parameter to fetch this information from it. It also receives a policy parameter that tells it how to behave if the conversion can't be performed (i.e. an illegal or unsupported character is found). By default this function skips all illegal characters and tries to do the best it can, however, it is possible ask it to throw a conversion_error exception by passing the stop flag to it:

std::wstring s=to_utf<wchar_t>("\xFF\xFF","UTF-8",stop); 
// Throws because this string is illegal in UTF-8

std::codecvt facet

Boost.Locale provides stream codepage conversion facets based on the std::codecvt facet. This allows conversion between wide-character encodings and 8-bit encodings like UTF-8, ISO-8859 or Shift-JIS.

Most of compilers provide such facets, but:

  • Under Windows MSVC does not support UTF-8 encodings at all.
  • Under Linux the encodings are supported only if the required locales are generated. For example it may be impossible to create a he_IL.CP1255 locale even when the he_IL locale is available.

Thus Boost.Locale provides an option to generate code-page conversion facets for use with Boost.Iostreams filters or std::wfstream. For example:

    std::locale loc= generator().generate("he_IL.UTF-8");
    std::wofstream file.
    file << L"שלום!" << endl;

Would create a file hello.txt encoded as UTF-8 with "שלום!" (shalom) in it.

Integration with Boost.Iostreams

You can use the std::codecvt facet directly, but this is quite tricky and requires accurate buffer and error management.

You can use the boost::iostreams::code_converter class for stream-oriented conversions between the wide-character set and narrow locale character set.

This is a sample program that converts wide to narrow characters for an arbitrary stream:

#include <boost/iostreams/stream.hpp>
#include <boost/iostreams/categories.hpp> 
#include <boost/iostreams/code_converter.hpp>

#include <boost/locale.hpp>
#include <iostream>

namespace io = boost::iostreams;

// Device that consumes the converted text,
// In our case it just writes to standard output
class consumer {
    typedef char char_type;
    typedef io::sink_tag category;
    std::streamsize write(const char* s, std::streamsize n)
        return n;

int main()
    // the device that converts wide characters
    // to narrow
    typedef io::code_converter<consumer> converter_device;
    // the stream that uses this device
    typedef io::stream<converter_device> converter_stream;

    consumer cons;
    // setup out converter to work
    // with he_IL.UTF-8 locale 
    converter_device dev;
    boost::locale::generator gen;
    converter_stream stream;;
    // Now wide characters that are written
    // to the stream would be given to
    // our consumer as narrow characters 
    // in UTF-8 encoding
    stream << L"שלום" << std::flush;

Limitations of std::codecvt

The Standard does not provide any information about std::mbstate_t that could be used to save intermediate code-page conversion states. It leaves the definition up to the compiler implementation, making it impossible to reimplement std::codecvt<wchar_t,char,mbstate_t> for stateful encodings. Thus, Boost.Locale's codecvt facet implementation may be used with stateless encodings like UTF-8, ISO-8859, and Shift-JIS, but not with stateful encodings like UTF-7 or SCSU.

Recommendation: Prefer the Unicode UTF-8 encoding for char based strings and files in your application.


The implementation of codecvt for single byte encodings like ISO-8859-X and for UTF-8 is very efficent and would allow fast conversion of the content, however its performance may be sub-optimal for double-width encodings like Shift-JIS, due to the stateless problem described above.