// Copyright (c) 2006-2008 The Chromium Authors. All rights reserved. // // Redistribution and use in source and binary forms, with or without // modification, are permitted provided that the following conditions are // met: // // * Redistributions of source code must retain the above copyright // notice, this list of conditions and the following disclaimer. // * Redistributions in binary form must reproduce the above // copyright notice, this list of conditions and the following disclaimer // in the documentation and/or other materials provided with the // distribution. // * Neither the name of Google Inc. nor the names of its // contributors may be used to endorse or promote products derived from // this software without specific prior written permission. // // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. #ifndef BASE_STRING16_H_ #define BASE_STRING16_H_ // WHAT: // A version of std::basic_string that provides 2-byte characters even when // wchar_t is not implemented as a 2-byte type. You can access this class as // string16. We also define char16, which string16 is based upon. // // WHY: // On Windows, wchar_t is 2 bytes, and it can conveniently handle UTF-16/UCS-2 // data. Plenty of existing code operates on strings encoded as UTF-16. // // On many other platforms, sizeof(wchar_t) is 4 bytes by default. We can make // it 2 bytes by using the GCC flag -fshort-wchar. But then std::wstring fails // at run time, because it calls some functions (like wcslen) that come from // the system's native C library -- which was built with a 4-byte wchar_t! // It's wasteful to use 4-byte wchar_t strings to carry UTF-16 data, and it's // entirely improper on those systems where the encoding of wchar_t is defined // as UTF-32. // // Here, we define string16, which is similar to std::wstring but replaces all // libc functions with custom, 2-byte-char compatible routines. It is capable // of carrying UTF-16-encoded data. #include <string> #include "base/basictypes.h" #ifdef WIN32 typedef wchar_t char16; typedef std::wstring string16; #else // !WIN32 typedef uint16 char16; namespace base { // char16 versions of the functions required by string16_char_traits; these // are based on the wide character functions of similar names ("w" or "wcs" // instead of "c16"). int c16memcmp(const char16* s1, const char16* s2, size_t n); size_t c16len(const char16* s); const char16* c16memchr(const char16* s, char16 c, size_t n); char16* c16memmove(char16* s1, const char16* s2, size_t n); char16* c16memcpy(char16* s1, const char16* s2, size_t n); char16* c16memset(char16* s, char16 c, size_t n); struct string16_char_traits { typedef char16 char_type; typedef int int_type; typedef std::streamoff off_type; typedef mbstate_t state_type; typedef std::fpos<state_type> pos_type; static void assign(char_type& c1, const char_type& c2) { c1 = c2; } static bool eq(const char_type& c1, const char_type& c2) { return c1 == c2; } static bool lt(const char_type& c1, const char_type& c2) { return c1 < c2; } static int compare(const char_type* s1, const char_type* s2, size_t n) { return c16memcmp(s1, s2, n); } static size_t length(const char_type* s) { return c16len(s); } static const char_type* find(const char_type* s, size_t n, const char_type& a) { return c16memchr(s, a, n); } static char_type* move(char_type* s1, const char_type* s2, int_type n) { return c16memmove(s1, s2, n); } static char_type* copy(char_type* s1, const char_type* s2, size_t n) { return c16memcpy(s1, s2, n); } static char_type* assign(char_type* s, size_t n, char_type a) { return c16memset(s, a, n); } static int_type not_eof(const int_type& c) { return eq_int_type(c, eof()) ? 0 : c; } static char_type to_char_type(const int_type& c) { return char_type(c); } static int_type to_int_type(const char_type& c) { return int_type(c); } static bool eq_int_type(const int_type& c1, const int_type& c2) { return c1 == c2; } static int_type eof() { return static_cast<int_type>(EOF); } }; } // namespace base // The string class will be explicitly instantiated only once, in string16.cc. // // std::basic_string<> in GNU libstdc++ contains a static data member, // _S_empty_rep_storage, to represent empty strings. When an operation such // as assignment or destruction is performed on a string, causing its existing // data member to be invalidated, it must not be freed if this static data // member is being used. Otherwise, it counts as an attempt to free static // (and not allocated) data, which is a memory error. // // Generally, due to C++ template magic, _S_empty_rep_storage will be marked // as a coalesced symbol, meaning that the linker will combine multiple // instances into a single one when generating output. // // If a string class is used by multiple shared libraries, a problem occurs. // Each library will get its own copy of _S_empty_rep_storage. When strings // are passed across a library boundary for alteration or destruction, memory // errors will result. GNU libstdc++ contains a configuration option, // --enable-fully-dynamic-string (_GLIBCXX_FULLY_DYNAMIC_STRING), which // disables the static data member optimization, but it's a good optimization // and non-STL code is generally at the mercy of the system's STL // configuration. Fully-dynamic strings are not the default for GNU libstdc++ // libstdc++ itself or for the libstdc++ installations on the systems we care // about, such as Mac OS X and relevant flavors of Linux. // // See also http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24196 . // // To avoid problems, string classes need to be explicitly instantiated only // once, in exactly one library. All other string users see it via an "extern" // declaration. This is precisely how GNU libstdc++ handles // std::basic_string<char> (string) and std::basic_string<wchar_t> (wstring). // // This also works around a Mac OS X linker bug in ld64-85.2.1 (Xcode 3.1.2), // in which the linker does not fully coalesce symbols when dead code // stripping is enabled. This bug causes the memory errors described above // to occur even when a std::basic_string<> does not cross shared library // boundaries, such as in statically-linked executables. // // TODO(mark): File this bug with Apple and update this note with a bug number. extern template class std::basic_string<char16, base::string16_char_traits>; typedef std::basic_string<char16, base::string16_char_traits> string16; extern std::ostream& operator<<(std::ostream& out, const string16& str); #endif // !WIN32 #endif // BASE_STRING16_H_