facets.html revision 1.5
1<?xml version="1.0" encoding="UTF-8" standalone="no"?> 2<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Facets</title><meta name="generator" content="DocBook XSL-NS Stylesheets V1.78.1" /><meta name="keywords" content="ISO C++, library" /><meta name="keywords" content="ISO C++, runtime, library" /><link rel="home" href="../index.html" title="The GNU C++ Library" /><link rel="up" href="localization.html" title="Chapter��8.�� Localization" /><link rel="prev" href="localization.html" title="Chapter��8.�� Localization" /><link rel="next" href="containers.html" title="Chapter��9.�� Containers" /></head><body><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Facets</th></tr><tr><td width="20%" align="left"><a accesskey="p" href="localization.html">Prev</a>��</td><th width="60%" align="center">Chapter��8.�� 3 Localization 4 5</th><td width="20%" align="right">��<a accesskey="n" href="containers.html">Next</a></td></tr></table><hr /></div><div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="std.localization.facet"></a>Facets</h2></div></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="std.localization.facet.ctype"></a>ctype</h3></div></div></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.ctype.impl"></a>Implementation</h4></div></div></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="facet.ctype.impl.spec"></a>Specializations</h5></div></div></div><p> 6For the required specialization <code class="classname">codecvt<wchar_t, char, mbstate_t></code>, 7conversions are made between the internal character set (always UCS4 8on GNU/Linux) and whatever the currently selected locale for the 9<code class="code">LC_CTYPE</code> category implements. 10</p><p> 11The two required specializations are implemented as follows: 12</p><p> 13<code class="code"> 14ctype<char> 15</code> 16</p><p> 17This is simple specialization. Implementing this was a piece of cake. 18</p><p> 19<code class="code"> 20ctype<wchar_t> 21</code> 22</p><p> 23This specialization, by specifying all the template parameters, pretty 24much ties the hands of implementors. As such, the implementation is 25straightforward, involving <code class="function">mcsrtombs</code> for the 26conversions between <span class="type">char</span> to <span class="type">wchar_t</span> and 27<code class="function">wcsrtombs</code> for conversions between <span class="type">wchar_t</span> 28and <span class="type">char</span>. 29</p><p> 30Neither of these two required specializations deals with Unicode 31characters. 32</p></div></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.ctype.future"></a>Future</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p> 33 How to deal with the global locale issue? 34 </p></li><li class="listitem"><p> 35 How to deal with types other than <span class="type">char</span>, <span class="type">wchar_t</span>? 36 </p></li><li class="listitem"><p> 37 Overlap between codecvt/ctype: narrow/widen 38 </p></li><li class="listitem"><p> 39 <span class="type">mask</span> typedef in <code class="classname">codecvt_base</code>, 40 argument types in <span class="type">codecvt</span>. what is know about this type? 41 </p></li><li class="listitem"><p> 42 Why mask* argument in codecvt? 43 </p></li><li class="listitem"><p> 44 Can this be made (more) generic? is there a simple way to 45 straighten out the configure-time mess that is a by-product of 46 this class? 47 </p></li><li class="listitem"><p> 48 Get the <span class="type">ctype<wchar_t>::mask</span> stuff under control. 49 Need to make some kind of static table, and not do lookup every time 50 somebody hits the <code class="code">do_is...</code> functions. Too bad we can't 51 just redefine <span class="type">mask</span> for 52 <code class="classname">ctype<wchar_t></code> 53 </p></li><li class="listitem"><p> 54 Rename abstract base class. See if just smash-overriding is a 55 better approach. Clarify, add sanity to naming. 56 </p></li></ul></div></div><div class="bibliography"><div class="titlepage"><div><div><h4 class="title"><a id="facet.ctype.biblio"></a>Bibliography</h4></div></div></div><div class="biblioentry"><a id="id-1.3.4.6.3.2.4.2"></a><p><span class="citetitle"><em class="citetitle"> 57 The GNU C Library 58 </em>. </span><span class="author"><span class="firstname">Roland</span> <span class="surname">McGrath</span>. </span><span class="author"><span class="firstname">Ulrich</span> <span class="surname">Drepper</span>. </span><span class="copyright">Copyright �� 2007 FSF. </span><span class="pagenums">Chapters 6 Character Set Handling and 7 Locales and Internationalization. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.2.4.3"></a><p><span class="citetitle"><em class="citetitle"> 59 Correspondence 60 </em>. </span><span class="author"><span class="firstname">Ulrich</span> <span class="surname">Drepper</span>. </span><span class="copyright">Copyright �� 2002 . </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.2.4.4"></a><p><span class="citetitle"><em class="citetitle"> 61 ISO/IEC 14882:1998 Programming languages - C++ 62 </em>. </span><span class="copyright">Copyright �� 1998 ISO. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.2.4.5"></a><p><span class="citetitle"><em class="citetitle"> 63 ISO/IEC 9899:1999 Programming languages - C 64 </em>. </span><span class="copyright">Copyright �� 1999 ISO. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.2.4.6"></a><p><span class="title"><em> 65 <a class="link" href="http://www.unix.org/version3/ieee_std.html" target="_top"> 66 The Open Group Base Specifications, Issue 6 (IEEE Std. 1003.1-2004) 67 </a> 68 </em>. </span><span class="copyright">Copyright �� 1999 69 The Open Group/The Institute of Electrical and Electronics Engineers, Inc.. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.2.4.7"></a><p><span class="citetitle"><em class="citetitle"> 70 The C++ Programming Language, Special Edition 71 </em>. </span><span class="author"><span class="firstname">Bjarne</span> <span class="surname">Stroustrup</span>. </span><span class="copyright">Copyright �� 2000 Addison Wesley, Inc.. </span><span class="pagenums">Appendix D. </span><span class="publisher"><span class="publishername"> 72 Addison Wesley 73 . </span></span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.2.4.8"></a><p><span class="citetitle"><em class="citetitle"> 74 Standard C++ IOStreams and Locales 75 </em>. </span><span class="subtitle"> 76 Advanced Programmer's Guide and Reference 77 . </span><span class="author"><span class="firstname">Angelika</span> <span class="surname">Langer</span>. </span><span class="author"><span class="firstname">Klaus</span> <span class="surname">Kreft</span>. </span><span class="copyright">Copyright �� 2000 Addison Wesley Longman, Inc.. </span><span class="publisher"><span class="publishername"> 78 Addison Wesley Longman 79 . </span></span></p></div></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="std.localization.facet.codecvt"></a>codecvt</h3></div></div></div><p> 80The standard class codecvt attempts to address conversions between 81different character encoding schemes. In particular, the standard 82attempts to detail conversions between the implementation-defined wide 83characters (hereafter referred to as <span class="type">wchar_t</span>) and the standard 84type <span class="type">char</span> that is so beloved in classic <span class="quote">���<span class="quote">C</span>���</span> 85(which can now be referred to as narrow characters.) This document attempts 86to describe how the GNU libstdc++ implementation deals with the conversion 87between wide and narrow characters, and also presents a framework for dealing 88with the huge number of other encodings that iconv can convert, 89including Unicode and UTF8. Design issues and requirements are 90addressed, and examples of correct usage for both the required 91specializations for wide and narrow characters and the 92implementation-provided extended functionality are given. 93</p><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.codecvt.req"></a>Requirements</h4></div></div></div><p> 94Around page 425 of the C++ Standard, this charming heading comes into view: 95</p><div class="blockquote"><blockquote class="blockquote"><p> 9622.2.1.5 - Template class codecvt 97</p></blockquote></div><p> 98The text around the codecvt definition gives some clues: 99</p><div class="blockquote"><blockquote class="blockquote"><p> 100<span class="emphasis"><em> 101-1- The class <code class="code">codecvt<internT,externT,stateT></code> is for use 102when converting from one codeset to another, such as from wide characters 103to multibyte characters, between wide character encodings such as 104Unicode and EUC. 105</em></span> 106</p></blockquote></div><p> 107Hmm. So, in some unspecified way, Unicode encodings and 108translations between other character sets should be handled by this 109class. 110</p><div class="blockquote"><blockquote class="blockquote"><p> 111<span class="emphasis"><em> 112-2- The <span class="type">stateT</span> argument selects the pair of codesets being mapped between. 113</em></span> 114</p></blockquote></div><p> 115Ah ha! Another clue... 116</p><div class="blockquote"><blockquote class="blockquote"><p> 117<span class="emphasis"><em> 118-3- The instantiations required in the Table 51 (lib.locale.category), namely 119<code class="classname">codecvt<wchar_t,char,mbstate_t></code> and 120<code class="classname">codecvt<char,char,mbstate_t></code>, convert the 121implementation-defined native character set. 122<code class="classname">codecvt<char,char,mbstate_t></code> implements a 123degenerate conversion; it does not convert at all. 124<code class="classname">codecvt<wchar_t,char,mbstate_t></code> converts between 125the native character sets for tiny and wide characters. Instantiations on 126<span class="type">mbstate_t</span> perform conversion between encodings known to the library 127implementor. Other encodings can be converted by specializing on a 128user-defined <span class="type">stateT</span> type. The <span class="type">stateT</span> object can 129contain any state that is useful to communicate to or from the specialized 130<code class="function">do_convert</code> member. 131</em></span> 132</p></blockquote></div><p> 133At this point, a couple points become clear: 134</p><p> 135One: The standard clearly implies that attempts to add non-required 136(yet useful and widely used) conversions need to do so through the 137third template parameter, <span class="type">stateT</span>.</p><p> 138Two: The required conversions, by specifying <span class="type">mbstate_t</span> as the 139third template parameter, imply an implementation strategy that is mostly 140(or wholly) based on the underlying C library, and the functions 141<code class="function">mcsrtombs</code> and <code class="function">wcsrtombs</code> in 142particular.</p></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.codecvt.design"></a>Design</h4></div></div></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="codecvt.design.wchar_t_size"></a><span class="type">wchar_t</span> Size</h5></div></div></div><p> 143 The simple implementation detail of <span class="type">wchar_t</span>'s size seems to 144 repeatedly confound people. Many systems use a two byte, 145 unsigned integral type to represent wide characters, and use an 146 internal encoding of Unicode or UCS2. (See AIX, Microsoft NT, 147 Java, others.) Other systems, use a four byte, unsigned integral 148 type to represent wide characters, and use an internal encoding 149 of UCS4. (GNU/Linux systems using glibc, in particular.) The C 150 programming language (and thus C++) does not specify a specific 151 size for the type <span class="type">wchar_t</span>. 152 </p><p> 153 Thus, portable C++ code cannot assume a byte size (or endianness) either. 154 </p></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="codecvt.design.unicode"></a>Support for Unicode</h5></div></div></div><p> 155 Probably the most frequently asked question about code conversion 156 is: "So dudes, what's the deal with Unicode strings?" 157 The dude part is optional, but apparently the usefulness of 158 Unicode strings is pretty widely appreciated. The Unicode character 159 set (and useful encodings like UTF-8, UCS-4, ISO 8859-10, 160 etc etc etc) were not mentioned in the first C++ standard. (The 2011 161 standard added support for string literals with different encodings 162 and some library facilities for converting between encodings, but the 163 notes below have not been updated to reflect that.) 164 </p><p> 165 A couple of comments: 166 </p><p> 167 The thought that all one needs to convert between two arbitrary 168 codesets is two types and some kind of state argument is 169 unfortunate. In particular, encodings may be stateless. The naming 170 of the third parameter as <span class="type">stateT</span> is unfortunate, as what is 171 really needed is some kind of generalized type that accounts for the 172 issues that abstract encodings will need. The minimum information 173 that is required includes: 174 </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p> 175 Identifiers for each of the codesets involved in the 176 conversion. For example, using the iconv family of functions 177 from the Single Unix Specification (what used to be called 178 X/Open) hosted on the GNU/Linux operating system allows 179 bi-directional mapping between far more than the following 180 tantalizing possibilities: 181 </p><p> 182 (An edited list taken from <code class="code">`iconv --list`</code> on a 183 Red Hat 6.2/Intel system: 184 </p><div class="blockquote"><blockquote class="blockquote"><pre class="programlisting"> 1858859_1, 8859_9, 10646-1:1993, 10646-1:1993/UCS4, ARABIC, ARABIC7, 186ASCII, EUC-CN, EUC-JP, EUC-KR, EUC-TW, GREEK-CCIcode, GREEK, GREEK7-OLD, 187GREEK7, GREEK8, HEBREW, ISO-8859-1, ISO-8859-2, ISO-8859-3, 188ISO-8859-4, ISO-8859-5, ISO-8859-6, ISO-8859-7, ISO-8859-8, 189ISO-8859-9, ISO-8859-10, ISO-8859-11, ISO-8859-13, ISO-8859-14, 190ISO-8859-15, ISO-10646, ISO-10646/UCS2, ISO-10646/UCS4, 191ISO-10646/UTF-8, ISO-10646/UTF8, SHIFT-JIS, SHIFT_JIS, UCS-2, UCS-4, 192UCS2, UCS4, UNICODE, UNICODEBIG, UNICODELIcodeLE, US-ASCII, US, UTF-8, 193UTF-16, UTF8, UTF16). 194</pre></blockquote></div><p> 195For iconv-based implementations, string literals for each of the 196encodings (i.e. "UCS-2" and "UTF-8") are necessary, 197although for other, 198non-iconv implementations a table of enumerated values or some other 199mechanism may be required. 200</p></li><li class="listitem"><p> 201 Maximum length of the identifying string literal. 202</p></li><li class="listitem"><p> 203 Some encodings require explicit endian-ness. As such, some kind 204 of endian marker or other byte-order marker will be necessary. See 205 "Footnotes for C/C++ developers" in Haible for more information on 206 UCS-2/Unicode endian issues. (Summary: big endian seems most likely, 207 however implementations, most notably Microsoft, vary.) 208</p></li><li class="listitem"><p> 209 Types representing the conversion state, for conversions involving 210 the machinery in the "C" library, or the conversion descriptor, for 211 conversions using iconv (such as the type iconv_t.) Note that the 212 conversion descriptor encodes more information than a simple encoding 213 state type. 214</p></li><li class="listitem"><p> 215 Conversion descriptors for both directions of encoding. (i.e., both 216 UCS-2 to UTF-8 and UTF-8 to UCS-2.) 217</p></li><li class="listitem"><p> 218 Something to indicate if the conversion requested if valid. 219</p></li><li class="listitem"><p> 220 Something to represent if the conversion descriptors are valid. 221</p></li><li class="listitem"><p> 222 Some way to enforce strict type checking on the internal and 223 external types. As part of this, the size of the internal and 224 external types will need to be known. 225</p></li></ul></div></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="codecvt.design.issues"></a>Other Issues</h5></div></div></div><p> 226In addition, multi-threaded and multi-locale environments also impact 227the design and requirements for code conversions. In particular, they 228affect the required specialization 229<code class="classname">codecvt<wchar_t, char, mbstate_t></code> 230when implemented using standard "C" functions. 231</p><p> 232Three problems arise, one big, one of medium importance, and one small. 233</p><p> 234First, the small: <code class="function">mcsrtombs</code> and 235<code class="function">wcsrtombs</code> may not be multithread-safe 236on all systems required by the GNU tools. For GNU/Linux and glibc, 237this is not an issue. 238</p><p> 239Of medium concern, in the grand scope of things, is that the functions 240used to implement this specialization work on null-terminated 241strings. Buffers, especially file buffers, may not be null-terminated, 242thus giving conversions that end prematurely or are otherwise 243incorrect. Yikes! 244</p><p> 245The last, and fundamental problem, is the assumption of a global 246locale for all the "C" functions referenced above. For something like 247C++ iostreams (where codecvt is explicitly used) the notion of 248multiple locales is fundamental. In practice, most users may not run 249into this limitation. However, as a quality of implementation issue, 250the GNU C++ library would like to offer a solution that allows 251multiple locales and or simultaneous usage with computationally 252correct results. In short, libstdc++ is trying to offer, as an 253option, a high-quality implementation, damn the additional complexity! 254</p><p> 255For the required specialization 256<code class="classname">codecvt<wchar_t, char, mbstate_t></code>, 257conversions are made between the internal character set (always UCS4 258on GNU/Linux) and whatever the currently selected locale for the 259LC_CTYPE category implements. 260</p></div></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.codecvt.impl"></a>Implementation</h4></div></div></div><p> 261The two required specializations are implemented as follows: 262</p><p> 263<code class="code"> 264codecvt<char, char, mbstate_t> 265</code> 266</p><p> 267This is a degenerate (i.e., does nothing) specialization. Implementing 268this was a piece of cake. 269</p><p> 270<code class="code"> 271codecvt<char, wchar_t, mbstate_t> 272</code> 273</p><p> 274This specialization, by specifying all the template parameters, pretty 275much ties the hands of implementors. As such, the implementation is 276straightforward, involving <code class="function">mcsrtombs</code> for the conversions 277between <span class="type">char</span> to <span class="type">wchar_t</span> and 278<code class="function">wcsrtombs</code> for conversions between <span class="type">wchar_t</span> 279and <span class="type">char</span>. 280</p><p> 281Neither of these two required specializations deals with Unicode 282characters. As such, libstdc++ implements a partial specialization 283of the <span class="type">codecvt</span> class with an iconv wrapper class, 284<code class="classname">encoding_state</code> as the third template parameter. 285</p><p> 286This implementation should be standards conformant. First of all, the 287standard explicitly points out that instantiations on the third 288template parameter, <span class="type">stateT</span>, are the proper way to implement 289non-required conversions. Second of all, the standard says (in Chapter 29017) that partial specializations of required classes are A-OK. Third 291of all, the requirements for the <span class="type">stateT</span> type elsewhere in the 292standard (see 21.1.2 traits typedefs) only indicate that this type be copy 293constructible. 294</p><p> 295As such, the type <span class="type">encoding_state</span> is defined as a non-templatized, 296POD type to be used as the third type of a <span class="type">codecvt</span> instantiation. 297This type is just a wrapper class for iconv, and provides an easy interface 298to iconv functionality. 299</p><p> 300There are two constructors for <span class="type">encoding_state</span>: 301</p><p> 302<code class="code"> 303encoding_state() : __in_desc(0), __out_desc(0) 304</code> 305</p><p> 306This default constructor sets the internal encoding to some default 307(currently UCS4) and the external encoding to whatever is returned by 308<code class="code">nl_langinfo(CODESET)</code>. 309</p><p> 310<code class="code"> 311encoding_state(const char* __int, const char* __ext) 312</code> 313</p><p> 314This constructor takes as parameters string literals that indicate the 315desired internal and external encoding. There are no defaults for 316either argument. 317</p><p> 318One of the issues with iconv is that the string literals identifying 319conversions are not standardized. Because of this, the thought of 320mandating and/or enforcing some set of pre-determined valid 321identifiers seems iffy: thus, a more practical (and non-migraine 322inducing) strategy was implemented: end-users can specify any string 323(subject to a pre-determined length qualifier, currently 32 bytes) for 324encodings. It is up to the user to make sure that these strings are 325valid on the target system. 326</p><p> 327<code class="code"> 328void 329_M_init() 330</code> 331</p><p> 332Strangely enough, this member function attempts to open conversion 333descriptors for a given encoding_state object. If the conversion 334descriptors are not valid, the conversion descriptors returned will 335not be valid and the resulting calls to the codecvt conversion 336functions will return error. 337</p><p> 338<code class="code"> 339bool 340_M_good() 341</code> 342</p><p> 343Provides a way to see if the given <span class="type">encoding_state</span> object has been 344properly initialized. If the string literals describing the desired 345internal and external encoding are not valid, initialization will 346fail, and this will return false. If the internal and external 347encodings are valid, but <code class="function">iconv_open</code> could not allocate 348conversion descriptors, this will also return false. Otherwise, the object is 349ready to convert and will return true. 350</p><p> 351<code class="code"> 352encoding_state(const encoding_state&) 353</code> 354</p><p> 355As iconv allocates memory and sets up conversion descriptors, the copy 356constructor can only copy the member data pertaining to the internal 357and external code conversions, and not the conversion descriptors 358themselves. 359</p><p> 360Definitions for all the required codecvt member functions are provided 361for this specialization, and usage of <code class="code">codecvt<<em class="replaceable"><code>internal 362character type</code></em>, <em class="replaceable"><code>external character type</code></em>, <em class="replaceable"><code>encoding_state</code></em>></code> is consistent with other 363codecvt usage. 364</p></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.codecvt.use"></a>Use</h4></div></div></div><p>A conversion involving a string literal.</p><pre class="programlisting"> 365 typedef codecvt_base::result result; 366 typedef unsigned short unicode_t; 367 typedef unicode_t int_type; 368 typedef char ext_type; 369 typedef encoding_state state_type; 370 typedef codecvt<int_type, ext_type, state_type> unicode_codecvt; 371 372 const ext_type* e_lit = "black pearl jasmine tea"; 373 int size = strlen(e_lit); 374 int_type i_lit_base[24] = 375 { 25088, 27648, 24832, 25344, 27392, 8192, 28672, 25856, 24832, 29184, 376 27648, 8192, 27136, 24832, 29440, 27904, 26880, 28160, 25856, 8192, 29696, 377 25856, 24832, 2560 378 }; 379 const int_type* i_lit = i_lit_base; 380 const ext_type* efrom_next; 381 const int_type* ifrom_next; 382 ext_type* e_arr = new ext_type[size + 1]; 383 ext_type* eto_next; 384 int_type* i_arr = new int_type[size + 1]; 385 int_type* ito_next; 386 387 // construct a locale object with the specialized facet. 388 locale loc(locale::classic(), new unicode_codecvt); 389 // sanity check the constructed locale has the specialized facet. 390 VERIFY( has_facet<unicode_codecvt>(loc) ); 391 const unicode_codecvt& cvt = use_facet<unicode_codecvt>(loc); 392 // convert between const char* and unicode strings 393 unicode_codecvt::state_type state01("UNICODE", "ISO_8859-1"); 394 initialize_state(state01); 395 result r1 = cvt.in(state01, e_lit, e_lit + size, efrom_next, 396 i_arr, i_arr + size, ito_next); 397 VERIFY( r1 == codecvt_base::ok ); 398 VERIFY( !int_traits::compare(i_arr, i_lit, size) ); 399 VERIFY( efrom_next == e_lit + size ); 400 VERIFY( ito_next == i_arr + size ); 401</pre></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.codecvt.future"></a>Future</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p> 402 a. things that are sketchy, or remain unimplemented: 403 do_encoding, max_length and length member functions 404 are only weakly implemented. I have no idea how to do 405 this correctly, and in a generic manner. Nathan? 406</p></li><li class="listitem"><p> 407 b. conversions involving <span class="type">std::string</span> 408 </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: circle; "><li class="listitem"><p> 409 how should operators != and == work for string of 410 different/same encoding? 411 </p></li><li class="listitem"><p> 412 what is equal? A byte by byte comparison or an 413 encoding then byte comparison? 414 </p></li><li class="listitem"><p> 415 conversions between narrow, wide, and unicode strings 416 </p></li></ul></div></li><li class="listitem"><p> 417 c. conversions involving std::filebuf and std::ostream 418</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: circle; "><li class="listitem"><p> 419 how to initialize the state object in a 420 standards-conformant manner? 421 </p></li><li class="listitem"><p> 422 how to synchronize the "C" and "C++" 423 conversion information? 424 </p></li><li class="listitem"><p> 425 wchar_t/char internal buffers and conversions between 426 internal/external buffers? 427 </p></li></ul></div></li></ul></div></div><div class="bibliography"><div class="titlepage"><div><div><h4 class="title"><a id="facet.codecvt.biblio"></a>Bibliography</h4></div></div></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.2"></a><p><span class="citetitle"><em class="citetitle"> 428 The GNU C Library 429 </em>. </span><span class="author"><span class="firstname">Roland</span> <span class="surname">McGrath</span>. </span><span class="author"><span class="firstname">Ulrich</span> <span class="surname">Drepper</span>. </span><span class="copyright">Copyright �� 2007 FSF. </span><span class="pagenums"> 430 Chapters 6 Character Set Handling and 7 Locales and Internationalization 431 . </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.3"></a><p><span class="citetitle"><em class="citetitle"> 432 Correspondence 433 </em>. </span><span class="author"><span class="firstname">Ulrich</span> <span class="surname">Drepper</span>. </span><span class="copyright">Copyright �� 2002 . </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.4"></a><p><span class="citetitle"><em class="citetitle"> 434 ISO/IEC 14882:1998 Programming languages - C++ 435 </em>. </span><span class="copyright">Copyright �� 1998 ISO. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.5"></a><p><span class="citetitle"><em class="citetitle"> 436 ISO/IEC 9899:1999 Programming languages - C 437 </em>. </span><span class="copyright">Copyright �� 1999 ISO. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.6"></a><p><span class="title"><em> 438 <a class="link" href="http://pubs.opengroup.org/onlinepubs/9699919799/" target="_top"> 439 System Interface Definitions, Issue 7 (IEEE Std. 1003.1-2008) 440 </a> 441 </em>. </span><span class="copyright">Copyright �� 2008 442 The Open Group/The Institute of Electrical and Electronics 443 Engineers, Inc. 444 . </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.7"></a><p><span class="citetitle"><em class="citetitle"> 445 The C++ Programming Language, Special Edition 446 </em>. </span><span class="author"><span class="firstname">Bjarne</span> <span class="surname">Stroustrup</span>. </span><span class="copyright">Copyright �� 2000 Addison Wesley, Inc.. </span><span class="pagenums">Appendix D. </span><span class="publisher"><span class="publishername"> 447 Addison Wesley 448 . </span></span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.8"></a><p><span class="citetitle"><em class="citetitle"> 449 Standard C++ IOStreams and Locales 450 </em>. </span><span class="subtitle"> 451 Advanced Programmer's Guide and Reference 452 . </span><span class="author"><span class="firstname">Angelika</span> <span class="surname">Langer</span>. </span><span class="author"><span class="firstname">Klaus</span> <span class="surname">Kreft</span>. </span><span class="copyright">Copyright �� 2000 Addison Wesley Longman, Inc.. </span><span class="publisher"><span class="publishername"> 453 Addison Wesley Longman 454 . </span></span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.9"></a><p><span class="title"><em> 455 <a class="link" href="http://www.lysator.liu.se/c/na1.html" target="_top"> 456 A brief description of Normative Addendum 1 457 </a> 458 </em>. </span><span class="author"><span class="firstname">Clive</span> <span class="surname">Feather</span>. </span><span class="pagenums">Extended Character Sets. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.10"></a><p><span class="title"><em> 459 <a class="link" href="http://tldp.org/HOWTO/Unicode-HOWTO.html" target="_top"> 460 The Unicode HOWTO 461 </a> 462 </em>. </span><span class="author"><span class="firstname">Bruno</span> <span class="surname">Haible</span>. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.3.8.11"></a><p><span class="title"><em> 463 <a class="link" href="http://www.cl.cam.ac.uk/~mgk25/unicode.html" target="_top"> 464 UTF-8 and Unicode FAQ for Unix/Linux 465 </a> 466 </em>. </span><span class="author"><span class="firstname">Markus</span> <span class="surname">Khun</span>. </span></p></div></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="std.localization.facet.messages"></a>messages</h3></div></div></div><p> 467The <code class="classname">std::messages</code> facet implements message retrieval functionality 468equivalent to Java's <code class="classname">java.text.MessageFormat</code> using either GNU <code class="function">gettext</code> 469or IEEE 1003.1-200 functions. 470</p><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.messages.req"></a>Requirements</h4></div></div></div><p> 471The <code class="classname">std::messages</code> facet is probably the most vaguely defined facet in 472the standard library. It's assumed that this facility was built into 473the standard library in order to convert string literals from one 474locale to the other. For instance, converting the "C" locale's 475<code class="code">const char* c = "please"</code> to a German-localized <code class="code">"bitte"</code> 476during program execution. 477</p><div class="blockquote"><blockquote class="blockquote"><p> 47822.2.7.1 - Template class messages [lib.locale.messages] 479</p></blockquote></div><p> 480This class has three public member functions, which directly 481correspond to three protected virtual member functions. 482</p><p> 483The public member functions are: 484</p><p> 485<code class="code">catalog open(const string&, const locale&) const</code> 486</p><p> 487<code class="code">string_type get(catalog, int, int, const string_type&) const</code> 488</p><p> 489<code class="code">void close(catalog) const</code> 490</p><p> 491While the virtual functions are: 492</p><p> 493<code class="code">catalog do_open(const string& name, const locale& loc) const</code> 494</p><div class="blockquote"><blockquote class="blockquote"><p> 495<span class="emphasis"><em> 496-1- Returns: A value that may be passed to <code class="code">get()</code> to retrieve a 497message, from the message catalog identified by the string <code class="code">name</code> 498according to an implementation-defined mapping. The result can be used 499until it is passed to <code class="code">close()</code>. Returns a value less than 0 if no such 500catalog can be opened. 501</em></span> 502</p></blockquote></div><p> 503<code class="code">string_type do_get(catalog cat, int set , int msgid, const string_type& dfault) const</code> 504</p><div class="blockquote"><blockquote class="blockquote"><p> 505<span class="emphasis"><em> 506-3- Requires: A catalog <code class="code">cat</code> obtained from <code class="code">open()</code> and not yet closed. 507-4- Returns: A message identified by arguments <code class="code">set</code>, <code class="code">msgid</code>, and <code class="code">dfault</code>, 508according to an implementation-defined mapping. If no such message can 509be found, returns <code class="code">dfault</code>. 510</em></span> 511</p></blockquote></div><p> 512<code class="code">void do_close(catalog cat) const</code> 513</p><div class="blockquote"><blockquote class="blockquote"><p> 514<span class="emphasis"><em> 515-5- Requires: A catalog cat obtained from <code class="code">open()</code> and not yet closed. 516-6- Effects: Releases unspecified resources associated with <code class="code">cat</code>. 517-7- Notes: The limit on such resources, if any, is implementation-defined. 518</em></span> 519</p></blockquote></div></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.messages.design"></a>Design</h4></div></div></div><p> 520A couple of notes on the standard. 521</p><p> 522First, why is <code class="code">messages_base::catalog</code> specified as a typedef 523to int? This makes sense for implementations that use 524<code class="code">catopen</code> and define <code class="code">nl_catd</code> as int, but not for 525others. Fortunately, it's not heavily used and so only a minor irritant. 526This has been reported as a possible defect in the standard (LWG 2028). 527</p><p> 528Second, by making the member functions <code class="code">const</code>, it is 529impossible to save state in them. Thus, storing away information used 530in the 'open' member function for use in 'get' is impossible. This is 531unfortunate. 532</p><p> 533The 'open' member function in particular seems to be oddly 534designed. The signature seems quite peculiar. Why specify a <code class="code">const 535string& </code> argument, for instance, instead of just <code class="code">const 536char*</code>? Or, why specify a <code class="code">const locale&</code> argument that is 537to be used in the 'get' member function? How, exactly, is this locale 538argument useful? What was the intent? It might make sense if a locale 539argument was associated with a given default message string in the 540'open' member function, for instance. Quite murky and unclear, on 541reflection. 542</p><p> 543Lastly, it seems odd that messages, which explicitly require code 544conversion, don't use the codecvt facet. Because the messages facet 545has only one template parameter, it is assumed that ctype, and not 546codecvt, is to be used to convert between character sets. 547</p><p> 548It is implicitly assumed that the locale for the default message 549string in 'get' is in the "C" locale. Thus, all source code is assumed 550to be written in English, so translations are always from "en_US" to 551other, explicitly named locales. 552</p></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.messages.impl"></a>Implementation</h4></div></div></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="messages.impl.models"></a>Models</h5></div></div></div><p> 553 This is a relatively simple class, on the face of it. The standard 554 specifies very little in concrete terms, so generic 555 implementations that are conforming yet do very little are the 556 norm. Adding functionality that would be useful to programmers and 557 comparable to Java's java.text.MessageFormat takes a bit of work, 558 and is highly dependent on the capabilities of the underlying 559 operating system. 560 </p><p> 561 Three different mechanisms have been provided, selectable via 562 configure flags: 563 </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p> 564 generic 565 </p><p> 566 This model does very little, and is what is used by default. 567 </p></li><li class="listitem"><p> 568 gnu 569 </p><p> 570 The gnu model is complete and fully tested. It's based on the 571 GNU gettext package, which is part of glibc. It uses the 572 functions <code class="code">textdomain, bindtextdomain, gettext</code> to 573 implement full functionality. Creating message catalogs is a 574 relatively straight-forward process and is lightly documented 575 below, and fully documented in gettext's distributed 576 documentation. 577 </p></li><li class="listitem"><p> 578 ieee_1003.1-200x 579 </p><p> 580 This is a complete, though untested, implementation based on 581 the IEEE standard. The functions <code class="code">catopen, catgets, 582 catclose</code> are used to retrieve locale-specific messages 583 given the appropriate message catalogs that have been 584 constructed for their use. Note, the script <code class="code"> 585 po2msg.sed</code> that is part of the gettext distribution can 586 convert gettext catalogs into catalogs that 587 <code class="code">catopen</code> can use. 588 </p></li></ul></div><p> 589A new, standards-conformant non-virtual member function signature was 590added for 'open' so that a directory could be specified with a given 591message catalog. This simplifies calling conventions for the gnu 592model. 593</p></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="messages.impl.gnu"></a>The GNU Model</h5></div></div></div><p> 594 The messages facet, because it is retrieving and converting 595 between characters sets, depends on the ctype and perhaps the 596 codecvt facet in a given locale. In addition, underlying "C" 597 library locale support is necessary for more than just the 598 <code class="code">LC_MESSAGES</code> mask: <code class="code">LC_CTYPE</code> is also 599 necessary. To avoid any unpleasantness, all bits of the "C" mask 600 (i.e. <code class="code">LC_ALL</code>) are set before retrieving messages. 601 </p><p> 602 Making the message catalogs can be initially tricky, but become 603 quite simple with practice. For complete info, see the gettext 604 documentation. Here's an idea of what is required: 605 </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p> 606 Make a source file with the required string literals that need 607 to be translated. See <code class="code">intl/string_literals.cc</code> for 608 an example. 609 </p></li><li class="listitem"><p> 610 Make initial catalog (see "4 Making the PO Template File" from 611 the gettext docs).</p><p> 612 <code class="code"> xgettext --c++ --debug string_literals.cc -o libstdc++.pot </code> 613 </p></li><li class="listitem"><p>Make language and country-specific locale catalogs.</p><p> 614 <code class="code">cp libstdc++.pot fr_FR.po</code> 615 </p><p> 616 <code class="code">cp libstdc++.pot de_DE.po</code> 617 </p></li><li class="listitem"><p> 618 Edit localized catalogs in emacs so that strings are 619 translated. 620 </p><p> 621 <code class="code">emacs fr_FR.po</code> 622 </p></li><li class="listitem"><p>Make the binary mo files.</p><p> 623 <code class="code">msgfmt fr_FR.po -o fr_FR.mo</code> 624 </p><p> 625 <code class="code">msgfmt de_DE.po -o de_DE.mo</code> 626 </p></li><li class="listitem"><p>Copy the binary files into the correct directory structure.</p><p> 627 <code class="code">cp fr_FR.mo (dir)/fr_FR/LC_MESSAGES/libstdc++.mo</code> 628 </p><p> 629 <code class="code">cp de_DE.mo (dir)/de_DE/LC_MESSAGES/libstdc++.mo</code> 630 </p></li><li class="listitem"><p>Use the new message catalogs.</p><p> 631 <code class="code">locale loc_de("de_DE");</code> 632 </p><p> 633 <code class="code"> 634 use_facet<messages<char> >(loc_de).open("libstdc++", locale(), dir); 635 </code> 636 </p></li></ul></div></div></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.messages.use"></a>Use</h4></div></div></div><p> 637 A simple example using the GNU model of message conversion. 638 </p><pre class="programlisting"> 639#include <iostream> 640#include <locale> 641using namespace std; 642 643void test01() 644{ 645 typedef messages<char>::catalog catalog; 646 const char* dir = 647 "/mnt/egcs/build/i686-pc-linux-gnu/libstdc++/po/share/locale"; 648 const locale loc_de("de_DE"); 649 const messages<char>& mssg_de = use_facet<messages<char> >(loc_de); 650 651 catalog cat_de = mssg_de.open("libstdc++", loc_de, dir); 652 string s01 = mssg_de.get(cat_de, 0, 0, "please"); 653 string s02 = mssg_de.get(cat_de, 0, 0, "thank you"); 654 cout << "please in german:" << s01 << '\n'; 655 cout << "thank you in german:" << s02 << '\n'; 656 mssg_de.close(cat_de); 657} 658</pre></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="facet.messages.future"></a>Future</h4></div></div></div><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p> 659 Things that are sketchy, or remain unimplemented: 660 </p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: circle; "><li class="listitem"><p> 661 _M_convert_from_char, _M_convert_to_char are in flux, 662 depending on how the library ends up doing character set 663 conversions. It might not be possible to do a real character 664 set based conversion, due to the fact that the template 665 parameter for messages is not enough to instantiate the 666 codecvt facet (1 supplied, need at least 2 but would prefer 667 3). 668 </p></li><li class="listitem"><p> 669 There are issues with gettext needing the global locale set 670 to extract a message. This dependence on the global locale 671 makes the current "gnu" model non MT-safe. Future versions 672 of glibc, i.e. glibc 2.3.x will fix this, and the C++ library 673 bits are already in place. 674 </p></li></ul></div></li><li class="listitem"><p> 675 Development versions of the GNU "C" library, glibc 2.3 will allow 676 a more efficient, MT implementation of std::messages, and will 677 allow the removal of the _M_name_messages data member. If this is 678 done, it will change the library ABI. The C++ parts to support 679 glibc 2.3 have already been coded, but are not in use: once this 680 version of the "C" library is released, the marked parts of the 681 messages implementation can be switched over to the new "C" 682 library functionality. 683 </p></li><li class="listitem"><p> 684 At some point in the near future, std::numpunct will probably use 685 std::messages facilities to implement truename/falsename 686 correctly. This is currently not done, but entries in 687 libstdc++.pot have already been made for "true" and "false" string 688 literals, so all that remains is the std::numpunct coding and the 689 configure/make hassles to make the installed library search its 690 own catalog. Currently the libstdc++.mo catalog is only searched 691 for the testsuite cases involving messages members. 692 </p></li><li class="listitem"><p> The following member functions:</p><p> 693 <code class="code"> 694 catalog 695 open(const basic_string<char>& __s, const locale& __loc) const 696 </code> 697 </p><p> 698 <code class="code"> 699 catalog 700 open(const basic_string<char>&, const locale&, const char*) const; 701 </code> 702 </p><p> 703 Don't actually return a "value less than 0 if no such catalog 704 can be opened" as required by the standard in the "gnu" 705 model. As of this writing, it is unknown how to query to see 706 if a specified message catalog exists using the gettext 707 package. 708 </p></li></ul></div></div><div class="bibliography"><div class="titlepage"><div><div><h4 class="title"><a id="facet.messages.biblio"></a>Bibliography</h4></div></div></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.2"></a><p><span class="citetitle"><em class="citetitle"> 709 The GNU C Library 710 </em>. </span><span class="author"><span class="firstname">Roland</span> <span class="surname">McGrath</span>. </span><span class="author"><span class="firstname">Ulrich</span> <span class="surname">Drepper</span>. </span><span class="copyright">Copyright �� 2007 FSF. </span><span class="pagenums">Chapters 6 Character Set Handling, and 7 Locales and Internationalization 711 . </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.3"></a><p><span class="citetitle"><em class="citetitle"> 712 Correspondence 713 </em>. </span><span class="author"><span class="firstname">Ulrich</span> <span class="surname">Drepper</span>. </span><span class="copyright">Copyright �� 2002 . </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.4"></a><p><span class="citetitle"><em class="citetitle"> 714 ISO/IEC 14882:1998 Programming languages - C++ 715 </em>. </span><span class="copyright">Copyright �� 1998 ISO. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.5"></a><p><span class="citetitle"><em class="citetitle"> 716 ISO/IEC 9899:1999 Programming languages - C 717 </em>. </span><span class="copyright">Copyright �� 1999 ISO. </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.6"></a><p><span class="title"><em> 718 <a class="link" href="http://pubs.opengroup.org/onlinepubs/9699919799/" target="_top"> 719 System Interface Definitions, Issue 7 (IEEE Std. 1003.1-2008) 720 </a> 721 </em>. </span><span class="copyright">Copyright �� 2008 722 The Open Group/The Institute of Electrical and Electronics 723 Engineers, Inc. 724 . </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.7"></a><p><span class="citetitle"><em class="citetitle"> 725 The C++ Programming Language, Special Edition 726 </em>. </span><span class="author"><span class="firstname">Bjarne</span> <span class="surname">Stroustrup</span>. </span><span class="copyright">Copyright �� 2000 Addison Wesley, Inc.. </span><span class="pagenums">Appendix D. </span><span class="publisher"><span class="publishername"> 727 Addison Wesley 728 . </span></span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.8"></a><p><span class="citetitle"><em class="citetitle"> 729 Standard C++ IOStreams and Locales 730 </em>. </span><span class="subtitle"> 731 Advanced Programmer's Guide and Reference 732 . </span><span class="author"><span class="firstname">Angelika</span> <span class="surname">Langer</span>. </span><span class="author"><span class="firstname">Klaus</span> <span class="surname">Kreft</span>. </span><span class="copyright">Copyright �� 2000 Addison Wesley Longman, Inc.. </span><span class="publisher"><span class="publishername"> 733 Addison Wesley Longman 734 . </span></span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.9"></a><p><span class="title"><em> 735 <a class="link" href="http://www.oracle.com/technetwork/java/api/index.html" target="_top"> 736 API Specifications, Java Platform 737 </a> 738 </em>. </span><span class="pagenums">java.util.Properties, java.text.MessageFormat, 739java.util.Locale, java.util.ResourceBundle 740 . </span></p></div><div class="biblioentry"><a id="id-1.3.4.6.3.4.8.10"></a><p><span class="title"><em> 741 <a class="link" href="https://www.gnu.org/software/gettext/" target="_top"> 742 GNU gettext tools, version 0.10.38, Native Language Support 743 Library and Tools. 744 </a> 745 </em>. </span></p></div></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="localization.html">Prev</a>��</td><td width="20%" align="center"><a accesskey="u" href="localization.html">Up</a></td><td width="40%" align="right">��<a accesskey="n" href="containers.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">Chapter��8.�� 746 Localization 747 748��</td><td width="20%" align="center"><a accesskey="h" href="../index.html">Home</a></td><td width="40%" align="right" valign="top">��Chapter��9.�� 749 Containers 750 751</td></tr></table></div></body></html>