For the required specialization codecvt<wchar_t, char, mbstate_t> , conversions are made between the internal character set (always UCS4 on GNU/Linux) and whatever the currently selected locale for the LC_CTYPE category implements.
The two required specializations are implemented as follows:
ctype<char>
This is simple specialization. Implementing this was a piece of cake.
ctype<wchar_t>
This specialization, by specifying all the template parameters, pretty much ties the hands of implementors. As such, the implementation is straightforward, involving mcsrtombs for the conversions between char to wchar_t and wcsrtombs for conversions between wchar_t and char.
Neither of these two required specializations deals with Unicode characters.
How to deal with the global locale issue?
How to deal with different types than char, wchar_t?
Overlap between codecvt/ctype: narrow/widen
Mask typedef in codecvt_base, argument types in codecvt. what is know about this type?
Why mask* argument in codecvt?
Can this be made (more) generic? is there a simple way to straighten out the configure-time mess that is a by-product of this class?
Get the ctype<wchar_t>::mask stuff under control. Need to make some kind of static table, and not do lookup every time somebody hits the do_is... functions. Too bad we can't just redefine mask for ctype<wchar_t>
Rename abstract base class. See if just smash-overriding is a better approach. Clarify, add sanity to naming.
The GNU C Library . Copyright © 2007 FSF. Chapters 6 Character Set Handling and 7 Locales and Internationalization.
The Open Group Base Specifications, Issue 6 (IEEE Std. 1003.1-2004) . Copyright © 1999 The Open Group/The Institute of Electrical and Electronics Engineers, Inc..
The standard class codecvt attempts to address conversions between different character encoding schemes. In particular, the standard attempts to detail conversions between the implementation-defined wide characters (hereafter referred to as wchar_t) and the standard type char that is so beloved in classic “C” (which can now be referred to as narrow characters.) This document attempts to describe how the GNU libstdc++ implementation deals with the conversion between wide and narrow characters, and also presents a framework for dealing with the huge number of other encodings that iconv can convert, including Unicode and UTF8. Design issues and requirements are addressed, and examples of correct usage for both the required specializations for wide and narrow characters and the implementation-provided extended functionality are given.
Around page 425 of the C++ Standard, this charming heading comes into view:
22.2.1.5 - Template class codecvt
The text around the codecvt definition gives some clues:
-1- The class codecvt<internT,externT,stateT> is for use when converting from one codeset to another, such as from wide characters to multibyte characters, between wide character encodings such as Unicode and EUC.
Hmm. So, in some unspecified way, Unicode encodings and translations between other character sets should be handled by this class.
-2- The stateT argument selects the pair of codesets being mapped between.
Ah ha! Another clue...
-3- The instantiations required in the Table ?? (lib.locale.category), namely codecvt<wchar_t,char,mbstate_t> and codecvt<char,char,mbstate_t>, convert the implementation-defined native character set. codecvt<char,char,mbstate_t> implements a degenerate conversion; it does not convert at all. codecvt<wchar_t,char,mbstate_t> converts between the native character sets for tiny and wide characters. Instantiations on mbstate_t perform conversion between encodings known to the library implementor. Other encodings can be converted by specializing on a user-defined stateT type. The stateT object can contain any state that is useful to communicate to or from the specialized do_convert member.
At this point, a couple points become clear:
One: The standard clearly implies that attempts to add non-required (yet useful and widely used) conversions need to do so through the third template parameter, stateT.
Two: The required conversions, by specifying mbstate_t as the third template parameter, imply an implementation strategy that is mostly (or wholly) based on the underlying C library, and the functions mcsrtombs and wcsrtombs in particular.
The simple implementation detail of wchar_t's size seems to repeatedly confound people. Many systems use a two byte, unsigned integral type to represent wide characters, and use an internal encoding of Unicode or UCS2. (See AIX, Microsoft NT, Java, others.) Other systems, use a four byte, unsigned integral type to represent wide characters, and use an internal encoding of UCS4. (GNU/Linux systems using glibc, in particular.) The C programming language (and thus C++) does not specify a specific size for the type wchar_t.
Thus, portable C++ code cannot assume a byte size (or endianness) either.
Probably the most frequently asked question about code conversion is: "So dudes, what's the deal with Unicode strings?" The dude part is optional, but apparently the usefulness of Unicode strings is pretty widely appreciated. Sadly, this specific encoding (And other useful encodings like UTF8, UCS4, ISO 8859-10, etc etc etc) are not mentioned in the C++ standard.
A couple of comments:
The thought that all one needs to convert between two arbitrary codesets is two types and some kind of state argument is unfortunate. In particular, encodings may be stateless. The naming of the third parameter as stateT is unfortunate, as what is really needed is some kind of generalized type that accounts for the issues that abstract encodings will need. The minimum information that is required includes:
Identifiers for each of the codesets involved in the conversion. For example, using the iconv family of functions from the Single Unix Specification (what used to be called X/Open) hosted on the GNU/Linux operating system allows bi-directional mapping between far more than the following tantalizing possibilities:
(An edited list taken from `iconv --list` on a
Red Hat 6.2/Intel system:
8859_1, 8859_9, 10646-1:1993, 10646-1:1993/UCS4, ARABIC, ARABIC7, ASCII, EUC-CN, EUC-JP, EUC-KR, EUC-TW, GREEK-CCIcode, GREEK, GREEK7-OLD, GREEK7, GREEK8, HEBREW, ISO-8859-1, ISO-8859-2, ISO-8859-3, ISO-8859-4, ISO-8859-5, ISO-8859-6, ISO-8859-7, ISO-8859-8, ISO-8859-9, ISO-8859-10, ISO-8859-11, ISO-8859-13, ISO-8859-14, ISO-8859-15, ISO-10646, ISO-10646/UCS2, ISO-10646/UCS4, ISO-10646/UTF-8, ISO-10646/UTF8, SHIFT-JIS, SHIFT_JIS, UCS-2, UCS-4, UCS2, UCS4, UNICODE, UNICODEBIG, UNICODELIcodeLE, US-ASCII, US, UTF-8, UTF-16, UTF8, UTF16).
For iconv-based implementations, string literals for each of the encodings (i.e. "UCS-2" and "UTF-8") are necessary, although for other, non-iconv implementations a table of enumerated values or some other mechanism may be required.
Maximum length of the identifying string literal.
Some encodings require explicit endian-ness. As such, some kind of endian marker or other byte-order marker will be necessary. See "Footnotes for C/C++ developers" in Haible for more information on UCS-2/Unicode endian issues. (Summary: big endian seems most likely, however implementations, most notably Microsoft, vary.)
Types representing the conversion state, for conversions involving the machinery in the "C" library, or the conversion descriptor, for conversions using iconv (such as the type iconv_t.) Note that the conversion descriptor encodes more information than a simple encoding state type.
Conversion descriptors for both directions of encoding. (i.e., both UCS-2 to UTF-8 and UTF-8 to UCS-2.)
Something to indicate if the conversion requested if valid.
Something to represent if the conversion descriptors are valid.
Some way to enforce strict type checking on the internal and external types. As part of this, the size of the internal and external types will need to be known.
In addition, multi-threaded and multi-locale environments also impact the design and requirements for code conversions. In particular, they affect the required specialization codecvt<wchar_t, char, mbstate_t> when implemented using standard "C" functions.
Three problems arise, one big, one of medium importance, and one small.
First, the small: mcsrtombs and wcsrtombs may not be multithread-safe on all systems required by the GNU tools. For GNU/Linux and glibc, this is not an issue.
Of medium concern, in the grand scope of things, is that the functions used to implement this specialization work on null-terminated strings. Buffers, especially file buffers, may not be null-terminated, thus giving conversions that end prematurely or are otherwise incorrect. Yikes!
The last, and fundamental problem, is the assumption of a global locale for all the "C" functions referenced above. For something like C++ iostreams (where codecvt is explicitly used) the notion of multiple locales is fundamental. In practice, most users may not run into this limitation. However, as a quality of implementation issue, the GNU C++ library would like to offer a solution that allows multiple locales and or simultaneous usage with computationally correct results. In short, libstdc++ is trying to offer, as an option, a high-quality implementation, damn the additional complexity!
For the required specialization codecvt<wchar_t, char, mbstate_t> , conversions are made between the internal character set (always UCS4 on GNU/Linux) and whatever the currently selected locale for the LC_CTYPE category implements.
The two required specializations are implemented as follows:
codecvt<char, char, mbstate_t>
This is a degenerate (i.e., does nothing) specialization. Implementing this was a piece of cake.
codecvt<char, wchar_t, mbstate_t>
This specialization, by specifying all the template parameters, pretty much ties the hands of implementors. As such, the implementation is straightforward, involving mcsrtombs for the conversions between char to wchar_t and wcsrtombs for conversions between wchar_t and char.
Neither of these two required specializations deals with Unicode characters. As such, libstdc++ implements a partial specialization of the codecvt class with and iconv wrapper class, encoding_state as the third template parameter.
This implementation should be standards conformant. First of all, the standard explicitly points out that instantiations on the third template parameter, stateT, are the proper way to implement non-required conversions. Second of all, the standard says (in Chapter 17) that partial specializations of required classes are a-ok. Third of all, the requirements for the stateT type elsewhere in the standard (see 21.1.2 traits typedefs) only indicate that this type be copy constructible.
As such, the type encoding_state is defined as a non-templatized, POD type to be used as the third type of a codecvt instantiation. This type is just a wrapper class for iconv, and provides an easy interface to iconv functionality.
There are two constructors for encoding_state:
encoding_state() : __in_desc(0), __out_desc(0)
This default constructor sets the internal encoding to some default (currently UCS4) and the external encoding to whatever is returned by nl_langinfo(CODESET).
encoding_state(const char* __int, const char* __ext)
This constructor takes as parameters string literals that indicate the desired internal and external encoding. There are no defaults for either argument.
One of the issues with iconv is that the string literals identifying conversions are not standardized. Because of this, the thought of mandating and or enforcing some set of pre-determined valid identifiers seems iffy: thus, a more practical (and non-migraine inducing) strategy was implemented: end-users can specify any string (subject to a pre-determined length qualifier, currently 32 bytes) for encodings. It is up to the user to make sure that these strings are valid on the target system.
void
_M_init()
Strangely enough, this member function attempts to open conversion descriptors for a given encoding_state object. If the conversion descriptors are not valid, the conversion descriptors returned will not be valid and the resulting calls to the codecvt conversion functions will return error.
bool
_M_good()
Provides a way to see if the given encoding_state object has been properly initialized. If the string literals describing the desired internal and external encoding are not valid, initialization will fail, and this will return false. If the internal and external encodings are valid, but iconv_open could not allocate conversion descriptors, this will also return false. Otherwise, the object is ready to convert and will return true.
encoding_state(const encoding_state&)
As iconv allocates memory and sets up conversion descriptors, the copy constructor can only copy the member data pertaining to the internal and external code conversions, and not the conversion descriptors themselves.
Definitions for all the required codecvt member functions are provided for this specialization, and usage of codecvt<internal character type, external character type, encoding_state> is consistent with other codecvt usage.
A conversions involving string literal.
typedef codecvt_base::result result;
typedef unsigned short unicode_t;
typedef unicode_t int_type;
typedef char ext_type;
typedef encoding_state state_type;
typedef codecvt<int_type, ext_type, state_type> unicode_codecvt;
const ext_type* e_lit = "black pearl jasmine tea";
int size = strlen(e_lit);
int_type i_lit_base[24] =
{ 25088, 27648, 24832, 25344, 27392, 8192, 28672, 25856, 24832, 29184,
27648, 8192, 27136, 24832, 29440, 27904, 26880, 28160, 25856, 8192, 29696,
25856, 24832, 2560
};
const int_type* i_lit = i_lit_base;
const ext_type* efrom_next;
const int_type* ifrom_next;
ext_type* e_arr = new ext_type[size + 1];
ext_type* eto_next;
int_type* i_arr = new int_type[size + 1];
int_type* ito_next;
// construct a locale object with the specialized facet.
locale loc(locale::classic(), new unicode_codecvt);
// sanity check the constructed locale has the specialized facet.
VERIFY( has_facet<unicode_codecvt>(loc) );
const unicode_codecvt& cvt = use_facet<unicode_codecvt>(loc);
// convert between const char* and unicode strings
unicode_codecvt::state_type state01("UNICODE", "ISO_8859-1");
initialize_state(state01);
result r1 = cvt.in(state01, e_lit, e_lit + size, efrom_next,
i_arr, i_arr + size, ito_next);
VERIFY( r1 == codecvt_base::ok );
VERIFY( !int_traits::compare(i_arr, i_lit, size) );
VERIFY( efrom_next == e_lit + size );
VERIFY( ito_next == i_arr + size );
a. things that are sketchy, or remain unimplemented: do_encoding, max_length and length member functions are only weakly implemented. I have no idea how to do this correctly, and in a generic manner. Nathan?
b. conversions involving std::string
how should operators != and == work for string of different/same encoding?
what is equal? A byte by byte comparison or an encoding then byte comparison?
conversions between narrow, wide, and unicode strings
c. conversions involving std::filebuf and std::ostream
how to initialize the state object in a standards-conformant manner?
how to synchronize the "C" and "C++" conversion information?
wchar_t/char internal buffers and conversions between internal/external buffers?
The GNU C Library . Copyright © 2007 FSF. Chapters 6 Character Set Handling and 7 Locales and Internationalization .
System Interface Definitions, Issue 7 (IEEE Std. 1003.1-2008) . Copyright © 2008 The Open Group/The Institute of Electrical and Electronics Engineers, Inc. .
The C++ Programming Language, Special Edition . Copyright © 2000 Addison Wesley, Inc.. Appendix D. Addison Wesley .
Standard C++ IOStreams and Locales . Advanced Programmer's Guide and Reference . Copyright © 2000 Addison Wesley Longman, Inc.. Addison Wesley Longman .
A brief description of Normative Addendum 1 . Extended Character Sets.
The std::messages facet implements message retrieval functionality equivalent to Java's java.text.MessageFormat .using either GNU gettext or IEEE 1003.1-200 functions.
The std::messages facet is probably the most vaguely defined facet in
the standard library. It's assumed that this facility was built into
the standard library in order to convert string literals from one
locale to the other. For instance, converting the "C" locale's
const char* c = "please" to a German-localized "bitte"
during program execution.
22.2.7.1 - Template class messages [lib.locale.messages]
This class has three public member functions, which directly correspond to three protected virtual member functions.
The public member functions are:
catalog open(const string&, const locale&) const
string_type get(catalog, int, int, const string_type&) const
void close(catalog) const
While the virtual functions are:
catalog do_open(const string&, const locale&) const
-1- Returns: A value that may be passed to get() to retrieve a message, from the message catalog identified by the string name according to an implementation-defined mapping. The result can be used until it is passed to close(). Returns a value less than 0 if no such catalog can be opened.
string_type do_get(catalog, int, int, const string_type&) const
-3- Requires: A catalog cat obtained from open() and not yet closed. -4- Returns: A message identified by arguments set, msgid, and dfault, according to an implementation-defined mapping. If no such message can be found, returns dfault.
void do_close(catalog) const
-5- Requires: A catalog cat obtained from open() and not yet closed. -6- Effects: Releases unspecified resources associated with cat. -7- Notes: The limit on such resources, if any, is implementation-defined.
A couple of notes on the standard.
First, why is messages_base::catalog specified as a typedef
to int? This makes sense for implementations that use
catopen and define nl_catd as int, but not for
others. Fortunately, it's not heavily used and so only a minor irritant.
This has been reported as a possible defect in the standard (LWG 2028).
Second, by making the member functions const, it is
impossible to save state in them. Thus, storing away information used
in the 'open' member function for use in 'get' is impossible. This is
unfortunate.
The 'open' member function in particular seems to be oddly
designed. The signature seems quite peculiar. Why specify a const
string& argument, for instance, instead of just const
char*? Or, why specify a const locale& argument that is
to be used in the 'get' member function? How, exactly, is this locale
argument useful? What was the intent? It might make sense if a locale
argument was associated with a given default message string in the
'open' member function, for instance. Quite murky and unclear, on
reflection.
Lastly, it seems odd that messages, which explicitly require code conversion, don't use the codecvt facet. Because the messages facet has only one template parameter, it is assumed that ctype, and not codecvt, is to be used to convert between character sets.
It is implicitly assumed that the locale for the default message string in 'get' is in the "C" locale. Thus, all source code is assumed to be written in English, so translations are always from "en_US" to other, explicitly named locales.
This is a relatively simple class, on the face of it. The standard specifies very little in concrete terms, so generic implementations that are conforming yet do very little are the norm. Adding functionality that would be useful to programmers and comparable to Java's java.text.MessageFormat takes a bit of work, and is highly dependent on the capabilities of the underlying operating system.
Three different mechanisms have been provided, selectable via configure flags:
generic
This model does very little, and is what is used by default.
gnu
The gnu model is complete and fully tested. It's based on the
GNU gettext package, which is part of glibc. It uses the
functions textdomain, bindtextdomain, gettext to
implement full functionality. Creating message catalogs is a
relatively straight-forward process and is lightly documented
below, and fully documented in gettext's distributed
documentation.
ieee_1003.1-200x
This is a complete, though untested, implementation based on
the IEEE standard. The functions catopen, catgets,
catclose are used to retrieve locale-specific messages
given the appropriate message catalogs that have been
constructed for their use. Note, the script
po2msg.sed that is part of the gettext distribution can
convert gettext catalogs into catalogs that
catopen can use.
A new, standards-conformant non-virtual member function signature was added for 'open' so that a directory could be specified with a given message catalog. This simplifies calling conventions for the gnu model.
The messages facet, because it is retrieving and converting
between characters sets, depends on the ctype and perhaps the
codecvt facet in a given locale. In addition, underlying "C"
library locale support is necessary for more than just the
LC_MESSAGES mask: LC_CTYPE is also
necessary. To avoid any unpleasantness, all bits of the "C" mask
(i.e. LC_ALL) are set before retrieving messages.
Making the message catalogs can be initially tricky, but become quite simple with practice. For complete info, see the gettext documentation. Here's an idea of what is required:
Make a source file with the required string literals that need
to be translated. See intl/string_literals.cc for
an example.
Make initial catalog (see "4 Making the PO Template File" from the gettext docs).
xgettext --c++ --debug string_literals.cc -o libstdc++.pot
Make language and country-specific locale catalogs.
cp libstdc++.pot fr_FR.po
cp libstdc++.pot de_DE.po
Edit localized catalogs in emacs so that strings are translated.
emacs fr_FR.po
Make the binary mo files.
msgfmt fr_FR.po -o fr_FR.mo
msgfmt de_DE.po -o de_DE.mo
Copy the binary files into the correct directory structure.
cp fr_FR.mo (dir)/fr_FR/LC_MESSAGES/libstdc++.mo
cp de_DE.mo (dir)/de_DE/LC_MESSAGES/libstdc++.mo
Use the new message catalogs.
locale loc_de("de_DE");
use_facet<messages<char> >(loc_de).open("libstdc++", locale(), dir);
A simple example using the GNU model of message conversion.
#include <iostream>
#include <locale>
using namespace std;
void test01()
{
typedef messages<char>::catalog catalog;
const char* dir =
"/mnt/egcs/build/i686-pc-linux-gnu/libstdc++/po/share/locale";
const locale loc_de("de_DE");
const messages<char>& mssg_de = use_facet<messages<char> >(loc_de);
catalog cat_de = mssg_de.open("libstdc++", loc_de, dir);
string s01 = mssg_de.get(cat_de, 0, 0, "please");
string s02 = mssg_de.get(cat_de, 0, 0, "thank you");
cout << "please in german:" << s01 << '\n';
cout << "thank you in german:" << s02 << '\n';
mssg_de.close(cat_de);
}
Things that are sketchy, or remain unimplemented:
_M_convert_from_char, _M_convert_to_char are in flux, depending on how the library ends up doing character set conversions. It might not be possible to do a real character set based conversion, due to the fact that the template parameter for messages is not enough to instantiate the codecvt facet (1 supplied, need at least 2 but would prefer 3).
There are issues with gettext needing the global locale set to extract a message. This dependence on the global locale makes the current "gnu" model non MT-safe. Future versions of glibc, i.e. glibc 2.3.x will fix this, and the C++ library bits are already in place.
Development versions of the GNU "C" library, glibc 2.3 will allow a more efficient, MT implementation of std::messages, and will allow the removal of the _M_name_messages data member. If this is done, it will change the library ABI. The C++ parts to support glibc 2.3 have already been coded, but are not in use: once this version of the "C" library is released, the marked parts of the messages implementation can be switched over to the new "C" library functionality.
At some point in the near future, std::numpunct will probably use std::messages facilities to implement truename/falsename correctly. This is currently not done, but entries in libstdc++.pot have already been made for "true" and "false" string literals, so all that remains is the std::numpunct coding and the configure/make hassles to make the installed library search its own catalog. Currently the libstdc++.mo catalog is only searched for the testsuite cases involving messages members.
The following member functions:
catalog
open(const basic_string<char>& __s, const locale& __loc) const
catalog
open(const basic_string<char>&, const locale&, const char*) const;
Don't actually return a "value less than 0 if no such catalog can be opened" as required by the standard in the "gnu" model. As of this writing, it is unknown how to query to see if a specified message catalog exists using the gettext package.
The GNU C Library . Copyright © 2007 FSF. Chapters 6 Character Set Handling, and 7 Locales and Internationalization .
System Interface Definitions, Issue 7 (IEEE Std. 1003.1-2008) . Copyright © 2008 The Open Group/The Institute of Electrical and Electronics Engineers, Inc. .
The C++ Programming Language, Special Edition . Copyright © 2000 Addison Wesley, Inc.. Appendix D. Addison Wesley .
Standard C++ IOStreams and Locales . Advanced Programmer's Guide and Reference . Copyright © 2000 Addison Wesley Longman, Inc.. Addison Wesley Longman .
API Specifications, Java Platform . java.util.Properties, java.text.MessageFormat, java.util.Locale, java.util.ResourceBundle .